From e55a904e1d0b534a133cd998cc0918720fff0f5f Mon Sep 17 00:00:00 2001 From: Yichen Jiang Date: Thu, 30 Jul 2026 20:26:33 +0800 Subject: [PATCH 01/48] feat(web): render remote Markdown images --- ...026-07-23-web-assistant-markdown.i18n.yaml | 6 +- .../2026-07-23-web-assistant-markdown.md | 4 +- .../2026-07-23-web-assistant-markdown.zh.md | 4 +- ...07-30-web-remote-markdown-images.i18n.yaml | 6 + .../2026-07-30-web-remote-markdown-images.md | 29 +++ ...026-07-30-web-remote-markdown-images.zh.md | 29 +++ apps/web/tests/markdown-images.e2e.ts | 205 ++++++++++++++++++ .../snapshots/markdown-images/ui.expected.md | 33 +++ apps/web/tsconfig.json | 1 + .../client/ui-primitives/README.i18n.yaml | 4 +- packages/client/ui-primitives/README.md | 2 +- packages/client/ui-primitives/README.zh.md | 3 +- .../src/markdown/MarkdownText.module.css | 11 + .../src/markdown/MarkdownText.tsx | 27 ++- .../ui-primitives/tests/markdown.spec.tsx | 32 ++- tsconfig.host.json | 1 + 16 files changed, 381 insertions(+), 16 deletions(-) create mode 100644 .agents/notes/implemented/feature/2026-07-30-web-remote-markdown-images.i18n.yaml create mode 100644 .agents/notes/implemented/feature/2026-07-30-web-remote-markdown-images.md create mode 100644 .agents/notes/implemented/feature/2026-07-30-web-remote-markdown-images.zh.md create mode 100644 apps/web/tests/markdown-images.e2e.ts create mode 100644 apps/web/tests/snapshots/markdown-images/ui.expected.md diff --git a/.agents/notes/implemented/feature/2026-07-23-web-assistant-markdown.i18n.yaml b/.agents/notes/implemented/feature/2026-07-23-web-assistant-markdown.i18n.yaml index 1f52492649..656a52d300 100644 --- a/.agents/notes/implemented/feature/2026-07-23-web-assistant-markdown.i18n.yaml +++ b/.agents/notes/implemented/feature/2026-07-23-web-assistant-markdown.i18n.yaml @@ -1,6 +1,6 @@ # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each # side as of the last confirmed-consistent state. Both languages carry equal authority; # after editing either side, bring the other along and re-record with: -# pnpm run verify-translation-pairing --write -2026-07-23-web-assistant-markdown.md: 38d193271d88b3a8f32ba1b191e8a6d432176281 -2026-07-23-web-assistant-markdown.zh.md: be3cd041c6012af142fc27934fda125dfc4cf6de +# pnpm run verify-translation-pairing --write .agents/notes/implemented/feature/2026-07-23-web-assistant-markdown.md +2026-07-23-web-assistant-markdown.md: d5074e6090699229f5c43dd93eef0fdfbfedab76 +2026-07-23-web-assistant-markdown.zh.md: 31f0fd6835c9921f544f4b6217a0c834dff79859 diff --git a/.agents/notes/implemented/feature/2026-07-23-web-assistant-markdown.md b/.agents/notes/implemented/feature/2026-07-23-web-assistant-markdown.md index 38d193271d..d5074e6090 100644 --- a/.agents/notes/implemented/feature/2026-07-23-web-assistant-markdown.md +++ b/.agents/notes/implemented/feature/2026-07-23-web-assistant-markdown.md @@ -20,7 +20,7 @@ The dependency is explicit in `ui-primitives`; because that pure library is seed ## Untrusted output policy -Assistant-authored destinations are restricted to absolute HTTP, HTTPS, and mailto URLs. HTTP(S) links open in a new tab with `rel="noopener noreferrer"`; relative destinations and other protocols render as non-navigable text. Markdown images render only their alt text, so model output cannot initiate a remote image request. Raw HTML remains inert source text because no HTML parser enters the pipeline. Shiki output is a static span tree generated from the fence text (no scripts or user HTML). +Assistant-authored link destinations are restricted to absolute HTTP, HTTPS, and mailto URLs. HTTP(S) links open in a new tab with `rel="noopener noreferrer"`; relative destinations and other protocols render as non-navigable text. Markdown images follow the separate [remote-image policy](2026-07-30-web-remote-markdown-images.md). Raw HTML remains inert source text because no HTML parser enters the pipeline. Shiki output is a static span tree generated from the fence text (no scripts or user HTML). Fenced code and GFM tables own horizontal overflow so long content cannot widen the conversation column. @@ -32,7 +32,7 @@ Fenced code and GFM tables own horizontal overflow so long content cannot widen **Parse Markdown into session snapshots.** This would make React nodes or presentation ASTs durable runtime state and reintroduce a final-versus-streaming mode boundary. Parsing stays at the presentation leaf instead. -**Enable raw HTML or remote images with sanitization.** Neither capability has a current product need, while both enlarge the executable or network privacy boundary. They remain disabled rather than adding sanitizer and image-policy dependencies. +**Enable raw HTML with sanitization.** Raw HTML has no current product need and would enlarge the executable-content boundary, so it remains disabled rather than adding a sanitizer dependency. Remote images are governed by the later [image policy](2026-07-30-web-remote-markdown-images.md). **Port deepsuite Prism `highlight.css` and the mdast pipeline.** Appearance parity is owned by CSS Modules and shared `--dsw-*` tokens; highlighting stays on the existing shiki allowlist so the client does not take a second highlighter or Prism class contract. diff --git a/.agents/notes/implemented/feature/2026-07-23-web-assistant-markdown.zh.md b/.agents/notes/implemented/feature/2026-07-23-web-assistant-markdown.zh.md index be3cd041c6..31f0fd6835 100644 --- a/.agents/notes/implemented/feature/2026-07-23-web-assistant-markdown.zh.md +++ b/.agents/notes/implemented/feature/2026-07-23-web-assistant-markdown.zh.md @@ -20,7 +20,7 @@ Web 对话通过会话事件、历史回放与流式累积保留 assistant Markd ## 不受信任输出策略 -assistant 生成的目标地址仅限绝对 HTTP、HTTPS 与 mailto URL。HTTP(S) 链接会在新标签页中打开，并带有 `rel="noopener noreferrer"`；相对目标地址与其他协议会渲染为不可导航的文本。Markdown 图片仅渲染替代文本，因此模型输出无法发起远程图片请求。由于管线中未引入 HTML 解析器，原始 HTML 仍是不会生效的源文本。Shiki 输出是由围栏文本生成的静态 span 树（不含脚本或用户 HTML）。 +assistant 生成的链接目标地址仅限绝对 HTTP、HTTPS 与 mailto URL。HTTP(S) 链接会在新标签页中打开，并带有 `rel="noopener noreferrer"`；相对目标地址与其他协议会渲染为不可导航的文本。Markdown 图片遵循独立的[远程图片策略](2026-07-30-web-remote-markdown-images.md)。由于流水线中未引入 HTML 解析器，原始 HTML 仍是不会生效的源文本。Shiki 输出是由围栏文本生成的静态 span 树（不含脚本或用户 HTML）。 围栏代码与 GFM 表格各自处理横向溢出，因此较长内容无法撑宽对话栏。 @@ -32,7 +32,7 @@ assistant 生成的目标地址仅限绝对 HTTP、HTTPS 与 mailto URL。HTTP(S **将 Markdown 解析为会话快照。**这会让 React 节点或呈现层 AST 成为持久的运行时状态，并重新引入最终输出与流式输出之间的模式边界。解析仍留在呈现层的叶节点中。 -**通过净化启用原始 HTML 或远程图片。**当前产品并不需要这两项功能，但二者都会扩大可执行行为或网络隐私边界。因此它们保持禁用，无需增加净化器与图片策略依赖。 +**通过净化启用原始 HTML。** 原始 HTML 当前没有产品需求，并且会扩大可执行内容边界，因此保持禁用，无需增加净化器依赖。远程图片由后续的[图片策略](2026-07-30-web-remote-markdown-images.md)约束。 **移植 deepsuite 的 Prism `highlight.css` 与 mdast 管线。**外观一致性由 CSS Modules 与共享的 `--dsw-*` token 负责；高亮仍走现有的 shiki 允许列表，使客户端不必引入第二套高亮器或 Prism class 契约。 diff --git a/.agents/notes/implemented/feature/2026-07-30-web-remote-markdown-images.i18n.yaml b/.agents/notes/implemented/feature/2026-07-30-web-remote-markdown-images.i18n.yaml new file mode 100644 index 0000000000..afe776d402 --- /dev/null +++ b/.agents/notes/implemented/feature/2026-07-30-web-remote-markdown-images.i18n.yaml @@ -0,0 +1,6 @@ +# Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each +# side as of the last confirmed-consistent state. Both languages carry equal authority; +# after editing either side, bring the other along and re-record with: +# pnpm run verify-translation-pairing --write .agents/notes/implemented/feature/2026-07-30-web-remote-markdown-images.md +2026-07-30-web-remote-markdown-images.md: 23dc699aab87597b7b5bfee83d0d745d4798303e +2026-07-30-web-remote-markdown-images.zh.md: 54db9b25e9f558d77c84bb67c05fadea2c9086ac diff --git a/.agents/notes/implemented/feature/2026-07-30-web-remote-markdown-images.md b/.agents/notes/implemented/feature/2026-07-30-web-remote-markdown-images.md new file mode 100644 index 0000000000..23dc699aab --- /dev/null +++ b/.agents/notes/implemented/feature/2026-07-30-web-remote-markdown-images.md @@ -0,0 +1,29 @@ +# Agent Note: Remote Web Markdown images + +Status: implemented + +English | [中文](2026-07-30-web-remote-markdown-images.zh.md) + +## Problem + +Assistant Markdown can name diagrams and screenshots with standard image syntax, but the Web renderer replaces every image with italic alt text. Even absolute HTTP(S) destinations therefore lose ordinary Markdown behavior. + +## Decision + +`MarkdownText` renders absolute HTTP(S) image destinations as lazy, responsive `` elements with asynchronous decoding and `referrerPolicy="no-referrer"`. Relative paths, absolute local paths, `file:` URLs, and unsupported schemes retain the existing alt-text fallback. Raw HTML stays disabled, so an assistant cannot bypass the Markdown image component with a hand-authored ``. + +The image component reuses the renderer's absolute-URL policy without adding a host proxy, local-file route, Session dependency, sanitizer, or image fetcher. Finalized history, streaming output, interrupted partials, and every other `MarkdownText` consumer receive the same behavior. + +## Alternatives considered + +**Keep all images as alt text.** This preserves the smallest network boundary but defeats the product need to inspect network-hosted visual artifacts inline. + +**Proxy remote images through the host.** A proxy could hide the browser's network address from the image origin, but it would make the host perform arbitrary outbound fetches and require a separate redirect, DNS, size, and content policy. Direct HTTP(S) loading keeps that request visible to browser controls; omitting the referrer limits conversation-origin disclosure. + +**Support local paths in the same change.** Web origins cannot directly load host files. A safe implementation needs a separately reviewed authority boundary, so relative paths, absolute local paths, and `file:` URLs remain disabled. + +**Allow `data:` images.** Large data URLs duplicate binary content into durable transcript text. The HTTP(S)-only policy covers the current need without expanding session logs. + +## Consequences + +Assistant replies display remote images during streaming and replay without changing session events or host protocols. Remote origins still observe the image request, client network address, and any credentials that browser policy permits for that origin. Local and unsupported destinations remain inert alt text. diff --git a/.agents/notes/implemented/feature/2026-07-30-web-remote-markdown-images.zh.md b/.agents/notes/implemented/feature/2026-07-30-web-remote-markdown-images.zh.md new file mode 100644 index 0000000000..54db9b25e9 --- /dev/null +++ b/.agents/notes/implemented/feature/2026-07-30-web-remote-markdown-images.zh.md @@ -0,0 +1,29 @@ +# Agent Note: Web 中的远程 Markdown 图片 + +Status: implemented + +[English](2026-07-30-web-remote-markdown-images.md) | 中文 + +## 问题 + +assistant Markdown 可以使用标准图片语法引用图表和截图，但 Web 渲染器会把每张图片替换为斜体替代文本。因此，即使目标地址是绝对 HTTP(S) URL，也无法获得普通的 Markdown 图片行为。 + +## 决策 + +`MarkdownText` 将绝对 HTTP(S) 图片目标地址渲染为延迟加载的响应式 `` 元素，并使用异步解码与 `referrerPolicy="no-referrer"`。相对路径、绝对本地路径、`file:` URL 与不支持的协议继续沿用现有的替代文本回退。原始 HTML 保持禁用，因此 assistant 无法通过手写 `` 绕过 Markdown 图片组件。 + +图片组件复用渲染器的绝对 URL 策略，不新增主机代理、本地文件路由、Session 依赖、净化器或图片抓取器。已完成的历史消息、流式输出、被中断的部分输出以及其他所有 `MarkdownText` 消费方均获得同一行为。 + +## 考虑过的替代方案 + +**将所有图片都保留为替代文本。** 这种方案维持了最小的网络边界，但无法满足在行内查看网络托管的视觉产物这一产品需求。 + +**通过主机代理远程图片。** 代理可以向图片源站隐藏浏览器的网络地址，但这会让主机执行任意出站请求，并且需要单独制定重定向、DNS、大小与内容策略。直接加载 HTTP(S) 图片可让浏览器控制机制继续观察该请求；不发送 referrer 可减少对话来源信息的暴露。 + +**在同一变更中支持本地路径。** Web 源无法直接加载主机文件。安全的实现需要单独评审的权限边界，因此相对路径、绝对本地路径与 `file:` URL 保持禁用。 + +**允许 `data:` 图片。** 大型 data URL 会将二进制内容以文本形式重复写入持久化的 transcript（文本记录）。仅允许 HTTP(S) 的策略足以满足当前需求，且不会扩大会话日志。 + +## 后果 + +assistant 回复会在流式输出与回放期间显示远程图片，且不改变会话事件或主机协议。远程源站仍可观察到图片请求、客户端网络地址，以及浏览器策略允许发送给该源站的任何凭据。本地及不支持的目标地址仍只显示不会发起请求的替代文本。 diff --git a/apps/web/tests/markdown-images.e2e.ts b/apps/web/tests/markdown-images.e2e.ts new file mode 100644 index 0000000000..8dce7f405b --- /dev/null +++ b/apps/web/tests/markdown-images.e2e.ts @@ -0,0 +1,205 @@ +// Web e2e scenario: absolute HTTP(S) Markdown images. A validated session +// assembled through the Session API is seeded cold into the real web +// composition, then a separate image origin proves that the browser receives +// a real network image while local-path Markdown remains inert alt text. +import { createServer, type Server } from 'node:http' +import { fileURLToPath } from 'node:url' +import type { Browser, Page } from 'playwright' +import { chromium } from 'playwright' +import { afterAll, beforeAll, describe, expect, it, onTestFailed } from 'vitest' +import { createMessage, createUserMessage } from '@deepseek-ai/dsh-llm' +import { + SESSION_FORMAT_VERSION, + Session, + SessionId, +} from '@deepseek-ai/dsh-session' +import type {} from '@deepseek-ai/dsh-session-title' +import { + assertFixtureInventory, + captureStableAria, + compareOrRefreshGolden, + launchWebScaffold, + seedSession, + watchConsole, + webSnapshotMode, + type WebScaffold, +} from './scaffold.ts' +import { newEnglishPage, saveFailureShot } from './support.ts' + +const SNAPSHOT_DIR = fileURLToPath(new URL('./snapshots/markdown-images', import.meta.url)) +const UI_EXPECTED = fileURLToPath(new URL('./snapshots/markdown-images/ui.expected.md', import.meta.url)) +const MODE = webSnapshotMode() +const SEED_ID = 'markdown-images-web-e2e' +const REMOTE_ALT = 'Remote test image' +const LOCAL_ALT = 'Local test image' +const PNG = Buffer.from( + 'iVBORw0KGgoAAAANSUhEUgAAAAEAAAABCAQAAAC1HAwCAAAAC0lEQVR42mNk+A8AAQUBAScY42YAAAAASUVORK5CYII=', + 'base64', +) + +interface ImageOrigin { + server: Server + url: string + requests: Array<{ path: string | undefined; referer: string | undefined }> +} + +/** Start the deterministic remote image origin used by this browser scenario. */ +async function startImageOrigin(): Promise { + const requests: ImageOrigin['requests'] = [] + const server = createServer((request, response) => { + requests.push({ path: request.url, referer: request.headers.referer }) + response.writeHead(200, { + 'cache-control': 'no-store', + 'content-length': PNG.length, + 'content-type': 'image/png', + }) + response.end(PNG) + }) + await new Promise((resolve, reject) => { + server.once('error', reject) + server.listen(0, '127.0.0.1', resolve) + }) + const address = server.address() + if (address === null || typeof address === 'string') { + throw new Error('image origin did not expose an IP socket') + } + return { + server, + url: `http://127.0.0.1:${String(address.port)}/image.png`, + requests, + } +} + +/** Stop one image origin after the browser and host release their requests. */ +async function stopServer(server: Server): Promise { + await new Promise((resolve, reject) => { + server.close((error) => { + if (error === undefined) resolve() + else reject(error) + }) + }) +} + +/** Build one closed, invariant-checked session fixture with remote and local image Markdown. */ +function markdownImageFixture(remoteUrl: string): string { + const session = new Session(SessionId('markdown-image-source')) + session.append('turn/start', { + turn: 1, + trigger: { kind: 'message', source: { kind: 'user' } }, + }) + const user = session.append('user/message', createUserMessage({ + content: [{ type: 'text', text: 'Show the Markdown image policy.' }], + source: { kind: 'user' }, + }), { surfaceOp: 'append' }) + session.append('session/title', { + title: 'Markdown image policy', + messageSeqs: [user.seq], + source: { kind: 'fallback' }, + }) + session.append('step/start', { turn: 1, step: 1 }) + session.append('assistant/message', { + turn: 1, + step: 1, + message: createMessage({ + role: 'assistant', + content: [{ + type: 'text', + text: [ + '## Markdown images', + '', + `![${REMOTE_ALT}](${remoteUrl})`, + '', + `![${LOCAL_ALT}](./local-image.png)`, + '', + 'REMOTE_IMAGE_DONE', + ].join('\n'), + }], + source: { kind: 'model', provider: 'fixture', model: 'fixture' }, + }), + }, { surfaceOp: 'append' }) + session.append('step/end', { turn: 1, step: 1 }) + session.append('turn/end', { turn: 1, reason: { kind: 'completed' } }) + + const header = { + type: 'session', + version: SESSION_FORMAT_VERSION, + id: '{{sessionId}}', + createdAt: 0, + cwd: '{{cwd}}', + } + return [ + JSON.stringify(header), + ...session.events.map(event => JSON.stringify(event)), + '', + ].join('\n') +} + +describe('web e2e: remote Markdown image rendering', () => { + let scaffold: WebScaffold + let imageOrigin: ImageOrigin + let browser: Browser + let page: Page + let tripwire: ReturnType + + beforeAll(async () => { + imageOrigin = await startImageOrigin() + scaffold = await launchWebScaffold({}) + await seedSession(scaffold, markdownImageFixture(imageOrigin.url), SEED_ID) + browser = await chromium.launch() + page = await newEnglishPage(browser) + tripwire = watchConsole(page) + await page.goto(scaffold.baseUrl, { waitUntil: 'load' }) + await page.waitForSelector('[class*="frame"]', { timeout: 30_000 }) + }, 120_000) + + afterAll(async () => { + await browser?.close() + await scaffold?.close() + await stopServer(imageOrigin.server) + }) + + it.skipIf(MODE === 'record')('loads only the remote image and matches the conversation golden', async () => { + onTestFailed(() => saveFailureShot(page, 'web-e2e-markdown-images')) + const groupRow = page.locator('[role="treeitem"]').first() + await groupRow.waitFor({ timeout: 15_000 }) + await groupRow.click() + const sessionRow = page.locator('[role="treeitem"]').nth(1) + await sessionRow.waitFor({ timeout: 10_000 }) + await sessionRow.click() + await expect.poll(() => page.getByText('REMOTE_IMAGE_DONE', { exact: true }).count(), { + timeout: 15_000, + }).toBe(1) + + const image = page.getByRole('img', { name: REMOTE_ALT }) + await image.waitFor({ timeout: 10_000 }) + await expect.poll(() => image.evaluate(element => (element as HTMLImageElement).naturalWidth), { + timeout: 10_000, + }).toBeGreaterThan(0) + expect(await image.evaluate((element) => { + const computed = getComputedStyle(element) + return { + borderRadius: computed.borderRadius, + decoding: element.getAttribute('decoding'), + loading: element.getAttribute('loading'), + maxWidth: computed.maxWidth, + referrerPolicy: element.getAttribute('referrerpolicy'), + } + })).toEqual({ + borderRadius: '8px', + decoding: 'async', + loading: 'lazy', + maxWidth: '100%', + referrerPolicy: 'no-referrer', + }) + expect(await page.getByRole('img', { name: LOCAL_ALT }).count()).toBe(0) + expect(await page.getByText(LOCAL_ALT, { exact: true }).count()).toBe(1) + expect(imageOrigin.requests).toEqual([{ path: '/image.png', referer: undefined }]) + + const snapshot = (await captureStableAria(page, '[class*="centerCol"]', scaffold.workspaceCwd)) + .split(SEED_ID).join('{{seededId}}') + await compareOrRefreshGolden(UI_EXPECTED, snapshot, MODE) + expect(tripwire.pageErrors).toEqual([]) + expect(tripwire.warnings).toEqual([]) + await assertFixtureInventory(SNAPSHOT_DIR, ['ui.expected.md']) + }, 60_000) +}) diff --git a/apps/web/tests/snapshots/markdown-images/ui.expected.md b/apps/web/tests/snapshots/markdown-images/ui.expected.md new file mode 100644 index 0000000000..cd33a09eb9 --- /dev/null +++ b/apps/web/tests/snapshots/markdown-images/ui.expected.md @@ -0,0 +1,33 @@ +- banner: + - navigation "Session hierarchy": + - button "Markdown image policy" [disabled] + - tablist: + - tab "Chat" [selected] + - tab "Trajectory" +- text: Show the Markdown image policy. {{clock}} +- button "复制": + - img +- button "在新对话中分支": + - img +- button "编辑": + - img +- heading "Markdown images" [level=2] +- paragraph: + - img "Remote test image" +- paragraph: Local test image +- paragraph: REMOTE_IMAGE_DONE +- button "复制": + - img +- button "在新对话中分支": + - img +- text: {{clock}} +- textbox "Message the agent" +- button "Add attachment": + - img +- 'button "Access mode, current: Danger Full Access"': Danger Full Access +- button "Plan mode off, press to turn on": Plan off +- button "Select model, current deepseek-v4-flash": + - text: deepseek-v4-flash + - img +- button "Send message" [disabled] +- text: 1 turns · 1 steps Input 0 tok · Output 0 tok diff --git a/apps/web/tsconfig.json b/apps/web/tsconfig.json index 7a7f228fb0..ce1f2105af 100644 --- a/apps/web/tsconfig.json +++ b/apps/web/tsconfig.json @@ -37,6 +37,7 @@ "tests/code-mode-round.e2e.ts", "tests/cordis-tool-round.e2e.ts", "tests/message-actions.e2e.ts", + "tests/markdown-images.e2e.ts", "tests/queue-actions.e2e.ts", "tests/skill-invocation-policy.e2e.ts" ], diff --git a/packages/client/ui-primitives/README.i18n.yaml b/packages/client/ui-primitives/README.i18n.yaml index b5e4b5c078..e24142ebcf 100644 --- a/packages/client/ui-primitives/README.i18n.yaml +++ b/packages/client/ui-primitives/README.i18n.yaml @@ -2,5 +2,5 @@ # side as of the last confirmed-consistent state. Both languages carry equal authority; # after editing either side, bring the other along and re-record with: # pnpm run verify-translation-pairing --write packages/client/ui-primitives/README.md -README.md: 0ef3c20f848b3d331c007911d0837f11cd72c024 -README.zh.md: af94551bfb9e12dbadcef6a96a54f9bf7ea71299 +README.md: 929d2edd9ddbd4610f84dd901486f74700cbf4e2 +README.zh.md: 5c6df2ce58e274a53b7792597259749f9a45d4ca diff --git a/packages/client/ui-primitives/README.md b/packages/client/ui-primitives/README.md index 0ef3c20f84..929d2edd9d 100644 --- a/packages/client/ui-primitives/README.md +++ b/packages/client/ui-primitives/README.md @@ -6,7 +6,7 @@ Pure React atoms (zero cordis): StateDot, ic_ds_* icons, Button/Pill/Menu/Modal/ ## Markdown rendering -`MarkdownText` renders GFM from untrusted assistant output through React elements. It omits raw HTML, neutralizes relative and non-HTTP(S)/mailto links, opens HTTP(S) links with safe external-link attributes, and renders image alt text without loading remote resources; `MessageText` remains the literal-text primitive for user-authored content. `extractMarkdownPlainText` removes Markdown presentation markup for compact labels while preserving raw HTML as literal text. Element spacing, tables, links, and inline code use the same `--dsw-alias-markdown-*` / `--dsw-font-markdown-*` tokens as deepsuite `@deepseek/md`. Fenced blocks render through `CodeBlock` (language banner, copy control, shiki for the registered grammars). +`MarkdownText` renders GFM from untrusted assistant output through React elements. It omits raw HTML, neutralizes relative and non-HTTP(S)/mailto links, opens HTTP(S) links with safe external-link attributes, and renders absolute HTTP(S) images without a referrer; relative paths, absolute local paths, `file:` URLs, and unsupported schemes retain their alt text. `MessageText` remains the literal-text primitive for user-authored content. `extractMarkdownPlainText` removes Markdown presentation markup for compact labels while preserving raw HTML as literal text. Element spacing, responsive images, tables, links, and inline code use the same `--dsw-alias-markdown-*` / `--dsw-font-markdown-*` tokens as deepsuite `@deepseek/md`. Fenced blocks render through `CodeBlock` (language banner, copy control, shiki for the registered grammars). ## Terminal output diff --git a/packages/client/ui-primitives/README.zh.md b/packages/client/ui-primitives/README.zh.md index af94551bfb..5c6df2ce58 100644 --- a/packages/client/ui-primitives/README.zh.md +++ b/packages/client/ui-primitives/README.zh.md @@ -6,7 +6,8 @@ ## Markdown 渲染 -`MarkdownText` 通过 React 元素渲染来自不受信任 assistant 输出的 GFM。它会省略原始 HTML，使相对链接及非 HTTP(S)/mailto 链接失效，以安全的外部链接属性打开 HTTP(S) 链接，并只渲染图片 alt 文本而不加载远程资源；`MessageText` 仍是用户创作内容使用的字面文本原语。`extractMarkdownPlainText` 会移除 Markdown 呈现标记以用于紧凑标签，同时将原始 HTML 保留为字面文本。元素间距、表格、链接与行内代码使用与 deepsuite `@deepseek/md` 相同的 `--dsw-alias-markdown-*` / `--dsw-font-markdown-*` token。围栏代码块通过 `CodeBlock` 渲染（语言横幅、复制控件，以及对已注册语法使用 shiki）。 +`MarkdownText` 通过 React 元素渲染来自不受信任 assistant 输出的 GFM。它会省略原始 HTML，使相对链接及非 HTTP(S)/mailto 链接失效，以安全的外部链接属性打开 HTTP(S) 链接，并在不携带 referrer 的情况下渲染绝对 HTTP(S) 图片；相对路径、绝对本地路径、`file:` URL 与不支持的协议仍保留其替代文本。`MessageText` 仍是用户创作内容使用的字面文本原语。`extractMarkdownPlainText` 会移除 Markdown 呈现标记以用于紧凑标签，同时将原始 HTML 保留为字面文本。元素间距、响应式图片、表格、链接与行内代码使用与 deepsuite `@deepseek/md` 相同的 `--dsw-alias-markdown-*` / `--dsw-font-markdown-*` token。围栏代码块通过 `CodeBlock` 渲染（语言横幅、复制控件，以及对已注册语法使用 shiki）。 + ## 终端输出 `TerminalBlock` 将一条 shell 命令渲染为终端表层：命令的每一行各占一个提示行（缩短后的 `cwd` 标签只出现在第一行，因为视图只知道一个工作目录，而一个 `cd` 就会让后面的行去到别处，标签之后是该行）、命令输出、非零退出码或终止信号对应的状态胶囊，以及写入原始 `output` prop 的复制控件。一枚运行状态 `StateDot` 为整次调用标记一次，位于第一行，以脱离文档流的方式落在卡片以自身左内边距预留的落区中，因此它位于卡片盒之内、提示文字之左。它用到 `StateDot` 的三种状态——`running` 期间为追逐动画，与渲染状态胶囊相同的退出状态为红色，其余为绿色——因此卡片直接陈述其命令是否仍在运行，而不是让人从有无输出中推断；由于 `StateDot` 是 `aria-hidden`，它携带一处视觉隐藏的文本标签。无论多少行都只有一枚状态点是有意为之：退出状态属于整次调用，因此每行一枚就会声称一个视图并不携带的逐行结果。命令文本使用 `white-space: pre`，因此重复空格、制表符与缩进续行都原样呈现，同时该行仍保持单行并以省略号截断。ANSI 转义序列通过运行时依赖 `anser` 解析为 React span；光标移动在剥除无显示意义控制符之前先重放进逐行的列缓冲，因为回车与退格**只移动**光标：单是 `100%` 加回车再加 `OK` 显示为 `OK0%`，而 spinner 随重绘写出的 `\x1b[K` 会擦掉尾巴，因此 `100%\r\x1b[KOK` 显示为 `OK`。行内擦除的三种参数形式都被遵循，光标按终端列推进（8 列制表位；emoji 与 CJK 占两列；组合标记不占列），SGR 状态按单元格归一化存储，与终端一致，并跨行延续、在行结束时的状态处收束；基础 16 色前景色映射到 `--dsw-*` token，而 256 色板与真彩色值按字面 rgb 透传。输出保持 `white-space: pre` 并支持横向滚动，因此按列对齐的输出保留其对齐而不会软换行；超过 `maxLines`（默认 16，与 TUI 转录相同的切分算法）时折叠为头部切片加尾部切片，由展开按钮控制。原理：[Web 终端卡片笔记](../../../.agents/notes/implemented/feature/2026-07-28-web-terminal-card.md)。 diff --git a/packages/client/ui-primitives/src/markdown/MarkdownText.module.css b/packages/client/ui-primitives/src/markdown/MarkdownText.module.css index a189528bc9..03bec019bd 100644 --- a/packages/client/ui-primitives/src/markdown/MarkdownText.module.css +++ b/packages/client/ui-primitives/src/markdown/MarkdownText.module.css @@ -224,3 +224,14 @@ color: var(--dsw-alias-label-tertiary); font-style: italic; } + +.image { + display: block; + width: auto; + max-width: 100%; + height: auto; + margin: 0; + border-radius: 8px; + background: var(--dsw-alias-bg-base); + object-fit: contain; +} diff --git a/packages/client/ui-primitives/src/markdown/MarkdownText.tsx b/packages/client/ui-primitives/src/markdown/MarkdownText.tsx index 639f53dbb1..ff6f543ebc 100644 --- a/packages/client/ui-primitives/src/markdown/MarkdownText.tsx +++ b/packages/client/ui-primitives/src/markdown/MarkdownText.tsx @@ -24,6 +24,15 @@ function sanitizeUrl(url: string): string { const safeUrl: UrlTransform = url => sanitizeUrl(url) +function remoteImageUrl(url: string): string | undefined { + try { + const protocol = new URL(url).protocol + return protocol === 'http:' || protocol === 'https:' ? url : undefined + } catch { + return undefined + } +} + /** Build the component table; while `streaming`, fences render the plain arm (see CodeBlock). */ function buildComponents(streaming: boolean): Components { return { @@ -40,7 +49,20 @@ function buildComponents(streaming: boolean): Components { ) }, - img: ({ alt = '' }) => {alt}, + img: ({ alt = '', src = '' }) => { + const imageSrc = remoteImageUrl(src) + if (imageSrc === undefined) return {alt} + return ( + {alt} + ) + }, table: ({ children }) => (
{children}
@@ -74,7 +96,8 @@ const streamingComponents = buildComponents(true) * Render untrusted assistant-authored Markdown as semantic React elements. * @param props - Markdown source text preserved by the session projection; * `streaming` renders fences plain (highlighting lands on the finalize swap). - * @returns A GFM document with raw HTML, relative links, unsafe protocols, and remote images disabled. + * @returns A GFM document with raw HTML, relative destinations, and unsafe + * protocols disabled; absolute HTTP(S) images render directly. */ export function MarkdownText({ text, streaming = false }: { text: string; streaming?: boolean }) { return ( diff --git a/packages/client/ui-primitives/tests/markdown.spec.tsx b/packages/client/ui-primitives/tests/markdown.spec.tsx index b7f665c78a..40295e16cb 100644 --- a/packages/client/ui-primitives/tests/markdown.spec.tsx +++ b/packages/client/ui-primitives/tests/markdown.spec.tsx @@ -94,13 +94,35 @@ describe('MarkdownText', () => { expect(done.container.querySelector('pre.shiki')).not.toBeNull() }) - it('neutralizes raw HTML, unsafe or relative links, and remote images', () => { + it('renders absolute HTTP(S) images with bounded presentation', () => { + const markdown = [ + '![secure diagram](https://example.com/secure.png)', + '![plain diagram](http://example.com/plain.png)', + ].join('\n\n') + const { container } = render() + const images = [...container.querySelectorAll('img')] + expect(images.map(image => image.getAttribute('src'))).toEqual([ + 'https://example.com/secure.png', + 'http://example.com/plain.png', + ]) + for (const image of images) { + expect(image.getAttribute('loading')).toBe('lazy') + expect(image.getAttribute('decoding')).toBe('async') + expect(image.getAttribute('referrerpolicy')).toBe('no-referrer') + } + }) + + it('neutralizes raw HTML, unsafe or relative links, and unsupported images', () => { const markdown = [ '', '', '[script](javascript:alert(1)) [relative](/settings)', '[mail](mailto:dev@example.com) [web](http://example.com) [upper](HTTPS://example.com)', - '![remote diagram](https://example.com/private.png)', + '![relative diagram](private.png)', + '![absolute diagram](/workspace/private.png)', + '![file diagram](file:///workspace/private.png)', + '![script diagram](javascript:alert(1))', + '![mail diagram](mailto:dev@example.com)', ].join('\n\n') const { container } = render() @@ -112,7 +134,11 @@ describe('MarkdownText', () => { expect(screen.getByRole('link', { name: 'mail' }).getAttribute('target')).toBeNull() expect(screen.getByRole('link', { name: 'web' }).getAttribute('rel')).toBe('noopener noreferrer') expect(screen.getByRole('link', { name: 'upper' }).getAttribute('target')).toBe('_blank') - expect(screen.getByText('remote diagram')).toBeTruthy() + expect(screen.getByText('relative diagram')).toBeTruthy() + expect(screen.getByText('absolute diagram')).toBeTruthy() + expect(screen.getByText('file diagram')).toBeTruthy() + expect(screen.getByText('script diagram')).toBeTruthy() + expect(screen.getByText('mail diagram')).toBeTruthy() }) it('keeps incomplete streaming Markdown renderable', () => { diff --git a/tsconfig.host.json b/tsconfig.host.json index b32c410961..795f5ae8d5 100644 --- a/tsconfig.host.json +++ b/tsconfig.host.json @@ -24,6 +24,7 @@ "apps/web/tests/code-mode-round.e2e.ts", "apps/web/tests/cordis-tool-round.e2e.ts", "apps/web/tests/message-actions.e2e.ts", + "apps/web/tests/markdown-images.e2e.ts", "apps/web/tests/queue-actions.e2e.ts", "apps/web/tests/skill-invocation-policy.e2e.ts", "apps/cli/tests/**/*.ts", From 0c9e529f74ef6bffde7e03acf33a28f8abbd6d1d Mon Sep 17 00:00:00 2001 From: Turtle Date: Wed, 29 Jul 2026 22:38:21 +0800 Subject: [PATCH 02/48] feat(tui): hidden mode folds a turn's assistant steps into one message The Ctrl+O hidden phase keeps one Assistant header per turn: the first step with visible text/reasoning owns it, later steps render as headerless continuations, and bodiless (tool-only) steps render nothing. Leaving hidden restores per-step headers. Pure TUI presentation; the session log is unchanged. --- ...28-consolidated-tui-presentation.i18n.yaml | 4 +- ...026-07-28-consolidated-tui-presentation.md | 2 +- ...-07-28-consolidated-tui-presentation.zh.md | 2 +- ...9-tui-hidden-mode-assistant-fold.i18n.yaml | 6 + ...26-07-29-tui-hidden-mode-assistant-fold.md | 25 ++++ ...07-29-tui-hidden-mode-assistant-fold.zh.md | 25 ++++ packages/ui/tui/README.i18n.yaml | 4 +- packages/ui/tui/README.md | 2 +- packages/ui/tui/README.zh.md | 2 +- packages/ui/tui/src/components/transcript.ts | 67 +++++++-- packages/ui/tui/src/index.ts | 60 +++++++- .../tool-cards-hidden-folded.expected.txt | 46 +++++++ packages/ui/tui/tests/tui.snapshot.ts | 26 ++++ packages/ui/tui/tests/tui.spec.ts | 128 ++++++++++++++++++ 14 files changed, 376 insertions(+), 23 deletions(-) create mode 100644 .agents/notes/implemented/feature/2026-07-29-tui-hidden-mode-assistant-fold.i18n.yaml create mode 100644 .agents/notes/implemented/feature/2026-07-29-tui-hidden-mode-assistant-fold.md create mode 100644 .agents/notes/implemented/feature/2026-07-29-tui-hidden-mode-assistant-fold.zh.md create mode 100644 packages/ui/tui/tests/snapshots/tool-cards-hidden-folded.expected.txt diff --git a/.agents/notes/implemented/architecture/2026-07-28-consolidated-tui-presentation.i18n.yaml b/.agents/notes/implemented/architecture/2026-07-28-consolidated-tui-presentation.i18n.yaml index 629476cf8e..43675cbd56 100644 --- a/.agents/notes/implemented/architecture/2026-07-28-consolidated-tui-presentation.i18n.yaml +++ b/.agents/notes/implemented/architecture/2026-07-28-consolidated-tui-presentation.i18n.yaml @@ -2,5 +2,5 @@ # side as of the last confirmed-consistent state. Both languages carry equal authority; # after editing either side, bring the other along and re-record with: # pnpm run verify-translation-pairing --write .agents/notes/implemented/architecture/2026-07-28-consolidated-tui-presentation.md -2026-07-28-consolidated-tui-presentation.md: f87d543a698d6e77abf9120c6579100df4b60b64 -2026-07-28-consolidated-tui-presentation.zh.md: 005e408f0e75207027315546942f9eab57d595d1 +2026-07-28-consolidated-tui-presentation.md: 8200c8e96cdea7cf5ae54c1328623bb03f841d00 +2026-07-28-consolidated-tui-presentation.zh.md: 852bd413b25f93f3b9096b78d69b21ab420df37b diff --git a/.agents/notes/implemented/architecture/2026-07-28-consolidated-tui-presentation.md b/.agents/notes/implemented/architecture/2026-07-28-consolidated-tui-presentation.md index f87d543a69..8200c8e96c 100644 --- a/.agents/notes/implemented/architecture/2026-07-28-consolidated-tui-presentation.md +++ b/.agents/notes/implemented/architecture/2026-07-28-consolidated-tui-presentation.md @@ -26,7 +26,7 @@ A tool card has one colored `Tool / ` status header over one dim body. Pre Injected context renders as prose in `ContextCardComponent`, not through the XML tree renderer. Exact matched outer `` lines are stripped, but mismatched, unpaired, or inline tag-like text remains verbatim. Model-facing content is unchanged. Folding uses the shared `preview` helper after body assembly, so it depends only on row count, never parser success or payload characters. -`Ctrl+O` cycles collapsed, expanded, and hidden. Tool cards disappear in the hidden state together with their card-owned leading gap. Context cards participate in collapsed and expanded states but fall back to collapsed while tools are hidden, because injected instructions are not disposable tool traffic. +`Ctrl+O` cycles collapsed, expanded, and hidden. Tool cards disappear in the hidden state together with their card-owned leading gap. Context cards participate in collapsed and expanded states but fall back to collapsed while tools are hidden, because injected instructions are not disposable tool traffic. The hidden phase additionally folds each turn's assistant steps into one message; the [hidden-mode assistant fold Agent Note](../feature/2026-07-29-tui-hidden-mode-assistant-fold.md) owns that rule. ### Cross-workspace resume diff --git a/.agents/notes/implemented/architecture/2026-07-28-consolidated-tui-presentation.zh.md b/.agents/notes/implemented/architecture/2026-07-28-consolidated-tui-presentation.zh.md index 005e408f0e..852bd413b2 100644 --- a/.agents/notes/implemented/architecture/2026-07-28-consolidated-tui-presentation.zh.md +++ b/.agents/notes/implemented/architecture/2026-07-28-consolidated-tui-presentation.zh.md @@ -26,7 +26,7 @@ Status: implemented 注入上下文由 `ContextCardComponent` 按普通文本呈现，不经过 XML 树渲染器。仅移除精确配对的外层 `` 行；不匹配、单边或正文内类似标签的文本都原样保留。面向模型的内容不变。折叠在正文组装完成后使用共享 `preview` 辅助函数，因此只取决于行数，不依赖解析是否成功或载荷包含哪些字符。 -`Ctrl+O` 在折叠、展开和隐藏之间循环。隐藏状态会连同卡片自有的前导间距一起移除工具卡片。上下文卡片参与折叠和展开状态，但工具隐藏时回到折叠状态，因为注入指令不是可丢弃的工具流量。 +`Ctrl+O` 在折叠、展开和隐藏之间循环。隐藏状态会连同卡片自有的前导间距一起移除工具卡片。上下文卡片参与折叠和展开状态，但工具隐藏时回到折叠状态，因为注入指令不是可丢弃的工具流量。隐藏阶段还会把每个轮次的 assistant 步骤折叠为一条消息；该规则由[隐藏模式 assistant 折叠 Agent Note](../feature/2026-07-29-tui-hidden-mode-assistant-fold.md)负责。 ### 跨工作区恢复 diff --git a/.agents/notes/implemented/feature/2026-07-29-tui-hidden-mode-assistant-fold.i18n.yaml b/.agents/notes/implemented/feature/2026-07-29-tui-hidden-mode-assistant-fold.i18n.yaml new file mode 100644 index 0000000000..338f90664d --- /dev/null +++ b/.agents/notes/implemented/feature/2026-07-29-tui-hidden-mode-assistant-fold.i18n.yaml @@ -0,0 +1,6 @@ +# Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each +# side as of the last confirmed-consistent state. Both languages carry equal authority; +# after editing either side, bring the other along and re-record with: +# pnpm run verify-translation-pairing --write .agents/notes/implemented/feature/2026-07-29-tui-hidden-mode-assistant-fold.md +2026-07-29-tui-hidden-mode-assistant-fold.md: e2bae1d4669fd0a4c8f9278c58f641704b425109 +2026-07-29-tui-hidden-mode-assistant-fold.zh.md: 583d6099f9cd5edfc592a9d510d260d7a4172013 diff --git a/.agents/notes/implemented/feature/2026-07-29-tui-hidden-mode-assistant-fold.md b/.agents/notes/implemented/feature/2026-07-29-tui-hidden-mode-assistant-fold.md new file mode 100644 index 0000000000..e2bae1d466 --- /dev/null +++ b/.agents/notes/implemented/feature/2026-07-29-tui-hidden-mode-assistant-fold.md @@ -0,0 +1,25 @@ +# Agent Note: TUI hidden mode folds a turn's assistant steps into one message + +Status: implemented + +English | [中文](2026-07-29-tui-hidden-mode-assistant-fold.zh.md) + +## Problem + +Ctrl+O's hidden phase ([consolidated TUI presentation](../architecture/2026-07-28-consolidated-tui-presentation.md)) drops tool cards so the transcript reads as a conversation, but each model step still rendered its own `Assistant` header. A multi-step turn (text → tools → text) therefore showed several consecutive `Assistant` blocks with nothing between them — the removed tool cards were the only thing that had justified the repeated headers. Codex-style conversation-only reading wants one assistant message per turn. + +## Decision + +Hidden mode is also a fold rule, applied purely as TUI presentation: per turn, the first step whose rendered content is visible (text, or reasoning while reasoning display is on) owns the turn's single `Assistant` header; every other step renders as a headerless continuation, and a step with no visible body renders nothing at all — a tool-only step neither consumes the header nor leaves a blank segment. Collapsed and expanded phases keep per-step headers; leaving hidden restores them. + +Mechanics: `StreamingAssistantComponent` carries its `StepPosition` and a `setFoldedContinuation` presentation flag; `createTuiChat` keeps a per-turn list of step components and re-derives the fold on Ctrl+O, on each streamed text/reasoning chunk, on message settle, and on retraction of a failed stream (which may hand the header to the next step). Transcript rebuild clears the map and replays the log, so resume, compaction replacement, resize, and theme swaps converge on the same fold. Step timing footers keep their per-step ownership and are unaffected. + +## Alternatives considered + +- **Merge steps into one component** — collides with per-step streaming lifecycle, retry retraction, and timing footers; the flag on existing components changes only the header/spacer. +- **Fold in the session log or `deriveMessages`** — mutates durable/model-visible history for a UI reading mode; the log stays step-shaped. +- **Always fold (all visibility phases)** — collapsed/expanded interleave tool cards between steps, where per-step headers delimit which output belongs to which step. + +## Consequences + +Hidden mode now reads as one assistant message per turn; turns stay separated by their headers. The fold is recomputed state, never stored, so no session or persistence format changes. Coverage: TUI unit specs for the Ctrl+O cycle header counts, tool-only first step header handoff, per-turn separation, and live streaming + rebuild convergence; keyless snapshot `tool-cards-hidden-folded` pins the folded frame. diff --git a/.agents/notes/implemented/feature/2026-07-29-tui-hidden-mode-assistant-fold.zh.md b/.agents/notes/implemented/feature/2026-07-29-tui-hidden-mode-assistant-fold.zh.md new file mode 100644 index 0000000000..583d6099f9 --- /dev/null +++ b/.agents/notes/implemented/feature/2026-07-29-tui-hidden-mode-assistant-fold.zh.md @@ -0,0 +1,25 @@ +# Agent Note: TUI 隐藏模式把一个轮次的 assistant 步骤折叠为一条消息 + +Status: implemented + +[English](2026-07-29-tui-hidden-mode-assistant-fold.md) | 中文 + +## 问题 + +Ctrl+O 的隐藏阶段（[整合的 TUI 展示](../architecture/2026-07-28-consolidated-tui-presentation.md)）去掉工具卡片，让 transcript（文本记录）读作一段对话，但每个模型步骤仍渲染自己的 `Assistant` 标题。因此一个多步骤轮次（文本 → 工具 → 文本）会显示多个连续、之间空无一物的 `Assistant` 区块——被移除的工具卡片正是重复标题曾经的唯一理由。Codex 风格的纯对话阅读需要每轮次一条 assistant 消息。 + +## 决定 + +隐藏模式同时也是一条折叠规则，且纯粹作为 TUI 展示实现：在每个轮次内，第一个渲染内容可见（有文本，或在 reasoning 显示开启时有 reasoning）的步骤拥有该轮次唯一的 `Assistant` 标题；其余步骤渲染为无标题的续段，没有可见正文的步骤则完全不渲染——仅有工具调用的步骤既不占用标题，也不留下空白段。折叠与展开阶段保留每步各自的标题；离开隐藏阶段会恢复它们。 + +机制：`StreamingAssistantComponent` 携带自己的 `StepPosition` 和一个 `setFoldedContinuation` 展示标志；`createTuiChat` 维护每轮次的步骤组件列表，并在 Ctrl+O、每个流式 text/reasoning chunk、消息结算，以及失败流被撤回（可能把标题移交给下一个步骤）时重新推导折叠。transcript 重建会清空该映射并重放日志，因此恢复、压缩替换、调整尺寸和主题切换收敛到同一折叠结果。步骤计时页脚保持按步骤归属，不受影响。 + +## 考虑过的替代方案 + +- **把多个步骤合并为一个组件**——与按步骤的流式生命周期、重试撤回和计时页脚冲突；在现有组件上加标志只改变标题与前导间距。 +- **在会话日志或 `deriveMessages` 中折叠**——为一种 UI 阅读模式改变持久 / 模型可见的历史；日志保持按步骤的形状。 +- **所有可见性阶段都折叠**——折叠 / 展开阶段在步骤之间穿插工具卡片，此时每步的标题用来划分哪段输出属于哪个步骤。 + +## 后果 + +隐藏模式现在每轮次读作一条 assistant 消息；轮次之间仍由各自的标题分隔。折叠是重新计算的状态，从不存储，因此会话与持久化格式没有变化。覆盖：TUI 单元测试覆盖 Ctrl+O 循环的标题计数、仅工具的首步骤标题移交、按轮次分隔，以及实时流式 + 重建收敛；无密钥快照 `tool-cards-hidden-folded` 固定折叠后的帧。 diff --git a/packages/ui/tui/README.i18n.yaml b/packages/ui/tui/README.i18n.yaml index e94be1a857..6587bc7624 100644 --- a/packages/ui/tui/README.i18n.yaml +++ b/packages/ui/tui/README.i18n.yaml @@ -2,5 +2,5 @@ # side as of the last confirmed-consistent state. Both languages carry equal authority; # after editing either side, bring the other along and re-record with: # pnpm run verify-translation-pairing --write packages/ui/tui/README.md -README.md: 63c888b1d51c02fa85a8f0cc1617874debd87c4e -README.zh.md: ca5efc9ae26a9833d271991f73a21c607d8fb09d +README.md: a65d72b3b80b992fabcb33d4b4345942b58e9147 +README.zh.md: fd056da59335af5cb9fc63fd2fb681266fadeb62 diff --git a/packages/ui/tui/README.md b/packages/ui/tui/README.md index 63c888b1d5..55e7eb3280 100644 --- a/packages/ui/tui/README.md +++ b/packages/ui/tui/README.md @@ -22,7 +22,7 @@ Typing `@` at a token boundary searches files and directories under the session When optional `ctx.sessionReferences` is mounted, the same `@` menu also offers metadata-only session candidates, inserts `@[label](dsh-session:)`, and prepares the selected snapshots before dispatch. Session references remain structured because the model has no filesystem-like tool for retrieving session snapshots later. Preparation disables duplicate submission and restores the editor input on failure. The TUI chooses `agent.steer()` or `agent.followup()` from the status after that asynchronous preparation, so idle follow-ups still dispatch `agent/prompt-submit` while in-turn steering joins at a checkpoint without that hook. -While the agent is running, ordinary editor submissions call `agent.steer()`; otherwise they call `agent.followup()`. A slash at the start of the submitted line enters `ctx.commands` instead: known commands execute directly, unknown commands produce a warning, and neither path automatically reaches the model. A command producer may explicitly schedule agent work; [`dsh-plan-mode`](../../plan/plan-mode/README.md#model-and-human-surfaces) uses that contract for `/plan [message]`. The TUI registers `/help`, `/model`, `/clear`, `/palette`, `/reload`, `/resume`, `/status`, and `/exit` as agent-scoped definitions; every other effective command joins autocomplete and `/help` dynamically, as do `/skill:` completions. A status line above the editor reports the turn phase the TUI derives from session events — waiting for the first token, thinking, responding, or executing tools — with the elapsed time in that phase and the running step total, refreshed each second, and ends with the `Enter sends steering, Esc cancels` hint; while steering messages wait to reach the model it inserts a `N queued ·` badge before the hint that clears as each drains. Ctrl+C or Escape cancels a running turn. Tool and injected-context cards collapse long bodies into a configurable head/tail preview; Ctrl+O cycles tool cards through collapsed preview, full output, and hidden — the hidden phase drops tool cards from the transcript entirely while context cards stay at their preview, since injected instructions are not tool traffic. An injected-context card renders its message as prose with the producer's outer reminder frame stripped, so neither the fold nor the frame stripping depends on the payload's syntax. Ctrl+R toggles reasoning, Ctrl+L redraws, and Ctrl+D exits while idle. +While the agent is running, ordinary editor submissions call `agent.steer()`; otherwise they call `agent.followup()`. A slash at the start of the submitted line enters `ctx.commands` instead: known commands execute directly, unknown commands produce a warning, and neither path automatically reaches the model. A command producer may explicitly schedule agent work; [`dsh-plan-mode`](../../plan/plan-mode/README.md#model-and-human-surfaces) uses that contract for `/plan [message]`. The TUI registers `/help`, `/model`, `/clear`, `/palette`, `/reload`, `/resume`, `/status`, and `/exit` as agent-scoped definitions; every other effective command joins autocomplete and `/help` dynamically, as do `/skill:` completions. A status line above the editor reports the turn phase the TUI derives from session events — waiting for the first token, thinking, responding, or executing tools — with the elapsed time in that phase and the running step total, refreshed each second, and ends with the `Enter sends steering, Esc cancels` hint; while steering messages wait to reach the model it inserts a `N queued ·` badge before the hint that clears as each drains. Ctrl+C or Escape cancels a running turn. Tool and injected-context cards collapse long bodies into a configurable head/tail preview; Ctrl+O cycles tool cards through collapsed preview, full output, and hidden — the hidden phase drops tool cards from the transcript entirely while context cards stay at their preview, since injected instructions are not tool traffic. The hidden phase also folds each turn's assistant steps into one message: the first step with visible text or reasoning keeps the turn's single `Assistant` header, later steps render as headerless continuations, and a step without a visible body renders nothing; leaving the hidden phase restores the per-step headers. An injected-context card renders its message as prose with the producer's outer reminder frame stripped, so neither the fold nor the frame stripping depends on the payload's syntax. Ctrl+R toggles reasoning, Ctrl+L redraws, and Ctrl+D exits while idle. `/model` opens the advisory `ctx.llm` catalog as a keyboard selector: a filter box above the list narrows rows by a case-insensitive substring over each row's `provider/model` label, model name, and description, keeping the highlighted row selected when it survives the filter; Up/Down moves, Shift+Tab cycles the focused model's adapter-advertised reasoning efforts in display order, Enter selects the model and effort, and Escape clears a non-empty filter before a second Escape closes it. When an adapter does not advertise a default effort, the cycle also includes `Default`, which clears an explicit selection and preserves the provider default; models without selectable effort metadata ignore Shift+Tab. The selector renders the exact advertised effort list—including `off` when present—and does not synthesize, clamp, or transfer an effort between models. `/model ` still selects an unambiguous model id directly, while `/model /` selects an exact target and uses its adapter default when one exists. The configured target or latest logged request header initializes the selector, and an unlisted current model remains visible because catalogs are advisory. Selection is local to this TUI session. Prompt assembly snapshots the target for one step, replaces `{{provider}}` and `{{model}}`, and applies the same provider/model/reasoning-effort target through `agent/request`; a switch during assembly therefore starts with a later step. The request header durably records targets that reach the model, while an unused selection remains process-local. diff --git a/packages/ui/tui/README.zh.md b/packages/ui/tui/README.zh.md index ca5efc9ae2..9683f6292e 100644 --- a/packages/ui/tui/README.zh.md +++ b/packages/ui/tui/README.zh.md @@ -22,7 +22,7 @@ TUI 从追加来源的会话事件重建已恢复历史，渲染 Markdown 响应 挂载可选的 `ctx.sessionReferences` 后，同一个 `@` 菜单还会提供仅含元数据的会话候选项，插入 `@[label](dsh-session:)`，并在分派前准备所选快照。会话引用保持结构化，因为模型没有类似文件系统的工具可在稍后检索会话快照。准备期间会禁止重复提交，并在失败时恢复编辑器输入。TUI 会在异步准备后根据状态选择 `agent.steer()` 或 `agent.followup()`，因此空闲 followup 仍会分派 `agent/prompt-submit`，而轮次中的 steering 会在检查点加入且不触发该 hook。 -Agent 运行时，普通编辑器提交会调用 `agent.steer()`；其他时候调用 `agent.followup()`。提交行以斜杠开头时会改为进入 `ctx.commands`：已知命令直接执行，未知命令产生警告，两条路径都不会自动到达模型。命令生产方可以显式调度 agent 工作；[`dsh-plan-mode`](../../plan/plan-mode/README.md#model-and-human-surfaces) 使用该契约实现 `/plan [message]`。TUI 将 `/help`、`/model`、`/clear`、`/palette`、`/reload`、`/resume`、`/status` 和 `/exit` 注册为 agent 作用域定义；其他所有有效命令都会动态加入自动补全与 `/help`，`/skill:` 补全也相同。编辑器上方的状态行会报告 TUI 从会话事件派生的轮次阶段，包括等待首个 token、思考、响应或执行工具；它显示该阶段已经过时间和运行中的步骤总数，每秒刷新，并以 `Enter sends steering, Esc cancels` 提示结尾。Steering 消息等待到达模型期间，会在提示前插入 `N queued ·` 徽标，每条消息排空后随即清除。Ctrl+C 或 Escape 会取消运行中的轮次。工具卡片与注入上下文卡片都把长主体折叠为可配置的头尾预览；Ctrl+O 让工具卡片在折叠预览、完整输出、隐藏三种状态间循环——隐藏阶段把工具卡片从 transcript 中完全去掉，而上下文卡片保持预览，因为注入的指令不属于工具流量。注入上下文卡片把消息渲染为文本，并去掉生产方的外层提醒外框，因此折叠与去外框都不依赖载荷的语法。Ctrl+R 切换 reasoning，Ctrl+L 重绘，Ctrl+D 在空闲时退出。 +Agent 运行时，普通编辑器提交会调用 `agent.steer()`；其他时候调用 `agent.followup()`。提交行以斜杠开头时会改为进入 `ctx.commands`：已知命令直接执行，未知命令产生警告，两条路径都不会自动到达模型。命令生产方可以显式调度 agent 工作；[`dsh-plan-mode`](../../plan/plan-mode/README.md#model-and-human-surfaces) 使用该契约实现 `/plan [message]`。TUI 将 `/help`、`/model`、`/clear`、`/palette`、`/reload`、`/resume`、`/status` 和 `/exit` 注册为 agent 作用域定义；其他所有有效命令都会动态加入自动补全与 `/help`，`/skill:` 补全也相同。编辑器上方的状态行会报告 TUI 从会话事件派生的轮次阶段，包括等待首个 token、思考、响应或执行工具；它显示该阶段已经过时间和运行中的步骤总数，每秒刷新，并以 `Enter sends steering, Esc cancels` 提示结尾。Steering 消息等待到达模型期间，会在提示前插入 `N queued ·` 徽标，每条消息排空后随即清除。Ctrl+C 或 Escape 会取消运行中的轮次。工具卡片与注入上下文卡片都把长主体折叠为可配置的头尾预览；Ctrl+O 让工具卡片在折叠预览、完整输出、隐藏三种状态间循环——隐藏阶段把工具卡片从 transcript 中完全去掉，而上下文卡片保持预览，因为注入的指令不属于工具流量。隐藏阶段还会把每个轮次的 assistant 步骤折叠为一条消息：第一个有可见文本或 reasoning 的步骤保留该轮次唯一的 `Assistant` 标题，之后的步骤渲染为无标题的续段，没有可见正文的步骤则不渲染任何内容；离开隐藏阶段会恢复每步各自的标题。注入上下文卡片把消息渲染为文本，并去掉生产方的外层提醒外框，因此折叠与去外框都不依赖载荷的语法。Ctrl+R 切换 reasoning，Ctrl+L 重绘，Ctrl+D 在空闲时退出。 `/model` 将建议性的 `ctx.llm` catalog 打开为键盘选择器：列表上方设有一个过滤框，按对每行 `provider/model` 标签、模型名称和描述的大小写不敏感子串匹配来缩小行集，并在高亮行仍通过过滤时保持其选中状态；Up/Down 移动，Shift+Tab 按显示顺序循环切换适配器为焦点模型公布的推理强度，Enter 选择模型和推理强度，Escape 会先清除非空过滤内容，再次按下才关闭选择器。适配器未公布默认推理强度时，循环还会包含 `Default`，该项会清除显式选择并保留提供方默认行为；没有可选推理强度元数据的模型会忽略 Shift+Tab。选择器会原样呈现公布的推理强度列表（包括存在时的 `off`），不会合成、自动调整或在模型之间转移推理强度。`/model ` 仍可直接选择无歧义的模型 id，`/model /` 则选择精确目标，并在存在时使用其适配器默认值。已配置目标或最新记录的请求 header 会初始化选择器；由于 catalog 仅提供建议，未列出的当前模型仍会显示。选择仅对本 TUI 会话有效。提示词组装会为一个步骤建立目标快照，替换 `{{provider}}` 和 `{{model}}`，并通过 `agent/request` 应用同一个提供方／模型／推理强度目标；因此组装期间的切换会从后续步骤开始生效。请求 header 会持久记录真正到达模型的目标，未使用的选择则只存在于进程本地。 diff --git a/packages/ui/tui/src/components/transcript.ts b/packages/ui/tui/src/components/transcript.ts index c8ba061166..300b39fe08 100644 --- a/packages/ui/tui/src/components/transcript.ts +++ b/packages/ui/tui/src/components/transcript.ts @@ -149,20 +149,28 @@ export class UserMessageComponent extends Container { } } -/** Children of a settled assistant message: optional reasoning block then the response text. */ +/** + * Children of a settled assistant message: optional reasoning block then the + * response text. A folded continuation (a later step of a turn while tool cards + * are hidden) drops the `Assistant` header and renders nothing when it has no + * visible body, so tool-only steps leave no blank segment behind. + */ function assistantMessageChildren( content: readonly ContentBlock[], showReasoning: boolean, + foldedContinuation: boolean, palette: Palette, mdTheme: MarkdownTheme, ): Component[] { const reasoning = displayText(textBlocks(content, 'reasoning').trim()) const text = displayText(textBlocks(content, 'text').trim()) - const children: Component[] = [ - new Spacer(1), - new Text(messageHeader('Assistant', palette.accent, palette), 0, 0), - ] - if (reasoning && showReasoning) { + const showsReasoning = reasoning !== '' && showReasoning + if (foldedContinuation && !showsReasoning && text === '') return [] + const children: Component[] = [new Spacer(1)] + if (!foldedContinuation) { + children.push(new Text(messageHeader('Assistant', palette.accent, palette), 0, 0)) + } + if (showsReasoning) { children.push( new Text(palette.italic(palette.dim('Reasoning')), 0, 0), new Markdown(reasoning, 0, 0, mdTheme, { color: value => palette.dim(value), italic: true }), @@ -220,6 +228,7 @@ interface StreamingBlock { export class StreamingAssistantComponent extends Container { private readonly blocks = new Map() private settledContent: readonly ContentBlock[] | undefined + private foldedContinuation = false /** * The step's timing footer. The renderer keeps it at the tail of the chat so * it trails any tool cards the step appends after this assistant message; it @@ -228,7 +237,8 @@ export class StreamingAssistantComponent extends Container { readonly timing: StepTimingComponent constructor( - position: StepPosition, + /** The step's turn/step coordinates, used to group steps into their turn. */ + readonly position: StepPosition, events: () => readonly SessionEvent[], now: () => number, private showReasoning: boolean, @@ -299,18 +309,49 @@ export class StreamingAssistantComponent extends Container { this.rebuild() } - private rebuild(): void { - this.clear() - const content: readonly ContentBlock[] = this.settledContent ?? [...this.blocks.entries()] + /** + * Mark this step as a folded continuation of its turn: no `Assistant` header, + * and no output at all while the step has no visible body. Used while tool + * cards are hidden so a turn reads as one assistant message. + * @param folded - Whether to render as a headerless continuation. + */ + setFoldedContinuation(folded: boolean): void { + if (this.foldedContinuation === folded) return + this.foldedContinuation = folded + this.rebuild() + } + + /** + * Whether the step currently renders visible reasoning or text. + * @returns `true` when a header-owning render would show a body. + */ + hasVisibleBody(): boolean { + const content = this.presentedContent() + return textBlocks(content, 'text').trim() !== '' + || (this.showReasoning && textBlocks(content, 'reasoning').trim() !== '') + } + + /** The settled content when available, otherwise the streamed blocks in model order. */ + private presentedContent(): readonly ContentBlock[] { + return this.settledContent ?? [...this.blocks.entries()] .sort(([left], [right]) => left - right) .flatMap(([, block]) => { if (block.type === 'text') return [{ type: 'text', text: block.text }] if (block.type === 'reasoning') return [{ type: 'reasoning', text: block.text }] return [] }) - for (const child of assistantMessageChildren(content, this.showReasoning, this.palette, this.mdTheme)) { - this.addChild(child) - } + } + + private rebuild(): void { + this.clear() + const children = assistantMessageChildren( + this.presentedContent(), + this.showReasoning, + this.foldedContinuation, + this.palette, + this.mdTheme, + ) + for (const child of children) this.addChild(child) } } diff --git a/packages/ui/tui/src/index.ts b/packages/ui/tui/src/index.ts index e2eaf89986..3d39c53957 100644 --- a/packages/ui/tui/src/index.ts +++ b/packages/ui/tui/src/index.ts @@ -335,6 +335,10 @@ export function createTuiChat( let toolsVisibility: ToolCardVisibility = 'collapsed' let streaming: StreamingAssistantComponent | undefined let completedStreaming: StreamingAssistantComponent | undefined + // Assistant step components in model order per turn, for hidden-mode folding: + // with tool cards hidden, a turn keeps one Assistant header and later steps + // render as headerless continuations (see applyTurnFolding). + const assistantSteps = new Map() let runningStatus: RunningStatus | undefined let fadingStatus: FadingStatus | undefined // TUI steering submissions that the inbox has not yet claimed or discarded. @@ -614,6 +618,35 @@ export function createTuiChat( return card } + /** + * Re-derive hidden-mode folding for one turn: the first step with a visible + * body owns the turn's single Assistant header, every other step renders as a + * headerless continuation (empty ones render nothing). Any other visibility + * restores the per-step headers. + */ + const applyTurnFolding = (turn: number): void => { + const steps = assistantSteps.get(turn) + if (steps === undefined) return + let headerSeen = false + for (const step of steps) { + if (toolsVisibility !== 'hidden') { + step.setFoldedContinuation(false) + } else if (!headerSeen && step.hasVisibleBody()) { + headerSeen = true + step.setFoldedContinuation(false) + } else { + step.setFoldedContinuation(true) + } + } + } + + const registerAssistantStep = (component: StreamingAssistantComponent): void => { + const steps = assistantSteps.get(component.position.turn) ?? [] + steps.push(component) + assistantSteps.set(component.position.turn, steps) + applyTurnFolding(component.position.turn) + } + const removeStreaming = (current: StreamingAssistantComponent | undefined): void => { if (current === undefined) return for (const child of [current, current.timing]) { @@ -621,6 +654,15 @@ export function createTuiChat( /* v8 ignore next -- streaming components and their timing footers are retained only while attached to the chat. */ if (index >= 0) chat.children.splice(index, 1) } + const steps = assistantSteps.get(current.position.turn) + /* v8 ignore next -- every attached streaming component is registered in the fold map. */ + if (steps === undefined) return + const index = steps.indexOf(current) + /* v8 ignore next -- registration precedes attachment, so the component is present until this removal. */ + if (index < 0) return + steps.splice(index, 1) + // A retracted step may have owned the turn's hidden-mode header. + applyTurnFolding(current.position.turn) } /** @@ -659,6 +701,7 @@ export function createTuiChat( palette, mdTheme, ) + registerAssistantStep(streaming) chat.addChild(streaming) chat.addChild(streaming.timing) } @@ -722,12 +765,20 @@ export function createTuiChat( startAssistantStep(event.data) break case 'assistant/chunk': - if (options.renderChunks) streaming?.update(event.data.chunk) + if (options.renderChunks && streaming !== undefined) { + streaming.update(event.data.chunk) + // The first streamed text/reasoning may make this step the turn's + // hidden-mode header owner (or a continuation with a visible body). + applyTurnFolding(streaming.position.turn) + } break case 'assistant/message': completedStreaming = undefined if (streaming === undefined || !chat.children.includes(streaming)) startAssistantStep(event.data) - streaming?.settle(event.data.message.content) + if (streaming !== undefined) { + streaming.settle(event.data.message.content) + applyTurnFolding(streaming.position.turn) + } break case 'llm/retry': { retractFailedStreaming() @@ -838,6 +889,7 @@ export function createTuiChat( toolCards.clear() allToolCards.clear() contextCards.clear() + assistantSteps.clear() streaming = undefined todo.update([]) const transcriptCalls = transcriptToolCallIds(agent.session) @@ -956,6 +1008,9 @@ export function createTuiChat( // Context cards carry injected instructions rather than tool traffic, so // they never hide: the hidden phase reads as their collapsed preview. for (const card of contextCards) card.setExpanded(toolsVisibility === 'expanded') + // Hidden mode folds each turn's steps into one assistant message; other + // modes restore the per-step Assistant headers. + for (const turn of assistantSteps.keys()) applyTurnFolding(turn) appendNotice(toolsVisibility === 'hidden' ? 'Tool cards hidden.' : `Tool and context cards ${toolsVisibility}.`) } @@ -967,6 +1022,7 @@ export function createTuiChat( if (activeStreaming !== undefined) { streaming = activeStreaming streaming.setShowReasoning(showReasoning) + registerAssistantStep(activeStreaming) chat.addChild(activeStreaming) chat.addChild(activeStreaming.timing) } diff --git a/packages/ui/tui/tests/snapshots/tool-cards-hidden-folded.expected.txt b/packages/ui/tui/tests/snapshots/tool-cards-hidden-folded.expected.txt new file mode 100644 index 0000000000..2aacdef876 --- /dev/null +++ b/packages/ui/tui/tests/snapshots/tool-cards-hidden-folded.expected.txt @@ -0,0 +1,46 @@ +terminal 100x40 buffer=normal length=40 base=0 viewport=0 +lifecycle started=1 stopped=0 progress=inactive +title "DSH snapshot" +cursor hidden column=7 viewportRow=20 bufferRow=20 +buffer +0| " DEEPSEEK HARNESS" + style 1-8 fg=bright-magenta bold + style 10-16 bold +1| " Snapshot agent ready." + style 1-21 dim +2| " main-session" + style 1-12 dim +3| +4| "Assistant " + style 0-8 fg=bright-magenta bold underline +5| "Inspecting the renderer first. " +6| "Model wait 0.0s " + style 0-14 dim +7| +8| "You " + style 0-2 fg=bright-magenta bold underline +9| "Refactor the renderer. " +10| "Model wait 0.0s · Completed 2026-07-29 22:23:17 " + style 0-46 dim +11| +12| "The renderer is sound; no refactor needed. " +13| "Model wait 0.0s · Completed 2026-07-29 22:23:17 " + style 0-46 dim +14| +15| "Tool and context cards expanded. " + style 0-31 dim +16| +17| "Tool cards hidden. " + style 0-17 dim +18| +19| "/workspace/project (tui-staging) deepseek-v4-flash ↑0 ↓0 0% context" + style 0-17 fg=bright-magenta bold + style 18-31 dim + style 34-50 dim + style 53-57 dim + style 60-69 dim +20| " dsh > " + style 1-3 fg=bright-magenta bold + style 5-6 dim + style 7-7 inverse +21-39| diff --git a/packages/ui/tui/tests/tui.snapshot.ts b/packages/ui/tui/tests/tui.snapshot.ts index 18f0a9a793..3f6c72ab18 100644 --- a/packages/ui/tui/tests/tui.snapshot.ts +++ b/packages/ui/tui/tests/tui.snapshot.ts @@ -44,6 +44,7 @@ const CHECKPOINTS = [ 'cordis-tools-pending', 'advanced-cards-collapsed', 'advanced-cards-expanded', + 'tool-cards-hidden-folded', 'untrusted-controls', 'question-dialog', 'question-dialog-single-option', @@ -609,6 +610,31 @@ describe('TUI terminal-state snapshots', () => { await disposeSnapshot(harness) }) + it('pins the hidden phase folding a multi-step turn into one assistant message', async () => { + const harness = await setupSnapshot({ + tools: ADVANCED_CARD_TOOLS, + config: { maxToolOutputLines: 3 }, + }, { columns: 100, rows: 40 }) + await renderAfter(harness, () => { + appendUser(harness.session, 'Refactor the renderer.') + appendAssistant(harness.session, [{ type: 'text', text: 'Inspecting the renderer first.' }]) + appendToolCalls(harness.session, [ + { id: 'fold-1', name: 'bash', arguments: { command: 'pnpm run test' } }, + ]) + appendToolResult(harness.session, 'fold-1', [{ type: 'text', text: 'all tests pass' }]) + harness.session.append('step/end', { turn: 1, step: 1 }) + harness.session.append('step/start', { turn: 1, step: 2 }) + appendAssistant(harness.session, [{ type: 'text', text: 'The renderer is sound; no refactor needed.' }], undefined, { turn: 1, step: 2 }) + harness.session.append('step/end', { turn: 1, step: 2 }) + harness.session.append('turn/end', { turn: 1, reason: { kind: 'completed' } }) + }) + // collapsed -> expanded -> hidden: one Assistant header, no tool card. + await renderAfter(harness, () => { harness.terminal.send('\x0f') }) + await renderAfter(harness, () => { harness.terminal.send('\x0f') }) + await checkpoint('tool-cards-hidden-folded', harness.terminal, { includeScrollback: true }) + await disposeSnapshot(harness) + }) + it('renders terminal controls as inert text across transcripts, tools, dialogs, diagnostics, and title', async () => { const nowSpy = vi.spyOn(Date, 'now').mockReturnValue(new Date(2026, 6, 21, 15, 0, 0).getTime()) const tools = { diff --git a/packages/ui/tui/tests/tui.spec.ts b/packages/ui/tui/tests/tui.spec.ts index 8231dc34d1..66540d2cc2 100644 --- a/packages/ui/tui/tests/tui.spec.ts +++ b/packages/ui/tui/tests/tui.spec.ts @@ -4989,6 +4989,134 @@ describe('tool cards and surface replay', () => { expect(mounted).not.toContain('stored model-only payload') await dispose(result) }) + + /** The last repainted frame, with CSI/OSC escapes and carriage returns stripped. */ + const lastFrame = (terminal: FakeTerminal): string => terminal.output + .slice(terminal.output.lastIndexOf('\x1b[2J')) + .replaceAll(/\x1b\[[0-9;]*[A-Za-z]|\x1b\][^\x07]*\x07|\r/g, '') + + const countAssistantHeaders = (frame: string): number => frame.split('\n') + .filter(row => row.trim() === 'Assistant').length + + /** One turn with text -> tool call/result -> text across two steps. */ + const appendTwoStepTurn = (session: Awaited>['session']): void => { + appendUser(session, 'fold me') + appendAssistant(session, [{ type: 'text', text: 'first step text' }]) + session.append('tool/call', { turn: 1, step: 1, callId: 'fold-1' as never, name: 'bash', arguments: '{}' }) + session.append('tool/result', { + turn: 1, step: 1, + message: createToolResultMessage({ + callId: 'fold-1' as never, content: [{ type: 'text', text: 'tool body' }], isError: false, + }), + }, { surfaceOp: 'append' }) + session.append('step/end', { turn: 1, step: 1 }) + session.append('step/start', { turn: 1, step: 2 }) + appendAssistant(session, [{ type: 'text', text: 'second step text' }], undefined, { turn: 1, step: 2 }) + session.append('step/end', { turn: 1, step: 2 }) + session.append('turn/end', { turn: 1, reason: { kind: 'completed' } }) + } + + it('folds a turn to one Assistant header in hidden mode and restores headers on cycle', async () => { + const result = await setup({ tools }) + appendTwoStepTurn(result.session) + await tick() + + // Collapsed (default): each step keeps its own header. + result.terminal.send('\x0c') + await tick() + expect(countAssistantHeaders(lastFrame(result.terminal))).toBe(2) + + // collapsed -> expanded -> hidden. + result.terminal.send('\x0f') + result.terminal.send('\x0f') + await tick() + result.terminal.send('\x0c') + await tick() + const hidden = lastFrame(result.terminal) + expect(countAssistantHeaders(hidden)).toBe(1) + expect(hidden).toContain('first step text') + expect(hidden).toContain('second step text') + expect(hidden).not.toContain('Tool / bash') + // The fold keeps model order: header text precedes the continuation. + expect(hidden.indexOf('first step text')).toBeLessThan(hidden.indexOf('second step text')) + + // hidden -> collapsed restores per-step headers. + result.terminal.send('\x0f') + await tick() + result.terminal.send('\x0c') + await tick() + expect(countAssistantHeaders(lastFrame(result.terminal))).toBe(2) + await dispose(result) + }) + + it('gives the hidden-mode header to the first step with a visible body and keeps turns separate', async () => { + const result = await setup({ tools }) + // Turn 1, step 1 is tool-only; step 2 carries the turn's text. + appendUser(result.session, 'tool-only first step') + appendAssistant(result.session, [{ type: 'tool-call', id: 'only-1' as never, name: 'bash', arguments: '{}' }]) + result.session.append('tool/call', { turn: 1, step: 1, callId: 'only-1' as never, name: 'bash', arguments: '{}' }) + result.session.append('tool/result', { + turn: 1, step: 1, + message: createToolResultMessage({ + callId: 'only-1' as never, content: [{ type: 'text', text: 'tool body' }], isError: false, + }), + }, { surfaceOp: 'append' }) + result.session.append('step/end', { turn: 1, step: 1 }) + result.session.append('step/start', { turn: 1, step: 2 }) + appendAssistant(result.session, [{ type: 'text', text: 'late turn-one text' }], undefined, { turn: 1, step: 2 }) + result.session.append('step/end', { turn: 1, step: 2 }) + result.session.append('turn/end', { turn: 1, reason: { kind: 'completed' } }) + // Turn 2 keeps its own header. + result.session.append('turn/start', { turn: 2, trigger: { kind: 'message', source: { kind: 'user' } } }) + appendUser(result.session, 'next turn') + result.session.append('step/start', { turn: 2, step: 1 }) + appendAssistant(result.session, [{ type: 'text', text: 'turn-two text' }], undefined, { turn: 2, step: 1 }) + result.session.append('step/end', { turn: 2, step: 1 }) + result.session.append('turn/end', { turn: 2, reason: { kind: 'completed' } }) + await tick() + + result.terminal.send('\x0f') + result.terminal.send('\x0f') + await tick() + result.terminal.send('\x0c') + await tick() + const hidden = lastFrame(result.terminal) + // One header per turn: the tool-only step neither renders a blank segment + // nor consumes turn one's header, which the late text step owns. + expect(countAssistantHeaders(hidden)).toBe(2) + expect(hidden).toContain('late turn-one text') + expect(hidden).toContain('turn-two text') + const rows = hidden.split('\n').map(row => row.trim()) + const turnOneHeader = rows.indexOf('Assistant') + expect(rows[turnOneHeader + 1]).toBe('late turn-one text') + await dispose(result) + }) + + it('folds live hidden-mode streaming once a later step shows text', async () => { + const result = await setup({ tools, status: 'running' }) + result.terminal.send('\x0f') + result.terminal.send('\x0f') + await tick() + result.session.append('assistant/chunk', { turn: 1, step: 1, chunk: { type: 'text-delta', index: 0, text: 'live first' } }) + result.session.append('step/end', { turn: 1, step: 1 }) + result.session.append('step/start', { turn: 1, step: 2 }) + result.session.append('assistant/chunk', { turn: 1, step: 2, chunk: { type: 'text-delta', index: 0, text: 'live second' } }) + await tick() + result.terminal.send('\x0c') + await tick() + const hidden = lastFrame(result.terminal) + expect(countAssistantHeaders(hidden)).toBe(1) + expect(hidden).toContain('live first') + expect(hidden).toContain('live second') + + // A transcript rebuild (resize) recomputes the same fold from the log. + result.terminal.resize(89) + await tick() + const rebuilt = lastFrame(result.terminal) + expect(countAssistantHeaders(rebuilt)).toBe(1) + expect(rebuilt).toContain('live second') + await dispose(result) + }) }) describe('TUI user-interaction dialogs', () => { From 651417e01caf4e0161ac8c97f0844067225d8b1d Mon Sep 17 00:00:00 2001 From: Turtle Date: Wed, 29 Jul 2026 22:40:59 +0800 Subject: [PATCH 03/48] test(tui): pin the folded-snapshot clock so replay is keyless-deterministic --- .../tui/tests/snapshots/tool-cards-hidden-folded.expected.txt | 4 ++-- packages/ui/tui/tests/tui.snapshot.ts | 2 ++ 2 files changed, 4 insertions(+), 2 deletions(-) diff --git a/packages/ui/tui/tests/snapshots/tool-cards-hidden-folded.expected.txt b/packages/ui/tui/tests/snapshots/tool-cards-hidden-folded.expected.txt index 2aacdef876..d4da5ab39c 100644 --- a/packages/ui/tui/tests/snapshots/tool-cards-hidden-folded.expected.txt +++ b/packages/ui/tui/tests/snapshots/tool-cards-hidden-folded.expected.txt @@ -20,11 +20,11 @@ buffer 8| "You " style 0-2 fg=bright-magenta bold underline 9| "Refactor the renderer. " -10| "Model wait 0.0s · Completed 2026-07-29 22:23:17 " +10| "Model wait 0.0s · Completed 2026-07-29 22:30:00 " style 0-46 dim 11| 12| "The renderer is sound; no refactor needed. " -13| "Model wait 0.0s · Completed 2026-07-29 22:23:17 " +13| "Model wait 0.0s · Completed 2026-07-29 22:30:00 " style 0-46 dim 14| 15| "Tool and context cards expanded. " diff --git a/packages/ui/tui/tests/tui.snapshot.ts b/packages/ui/tui/tests/tui.snapshot.ts index 3f6c72ab18..d56db82f14 100644 --- a/packages/ui/tui/tests/tui.snapshot.ts +++ b/packages/ui/tui/tests/tui.snapshot.ts @@ -611,6 +611,7 @@ describe('TUI terminal-state snapshots', () => { }) it('pins the hidden phase folding a multi-step turn into one assistant message', async () => { + const nowSpy = vi.spyOn(Date, 'now').mockReturnValue(new Date(2026, 6, 29, 22, 30, 0).getTime()) const harness = await setupSnapshot({ tools: ADVANCED_CARD_TOOLS, config: { maxToolOutputLines: 3 }, @@ -632,6 +633,7 @@ describe('TUI terminal-state snapshots', () => { await renderAfter(harness, () => { harness.terminal.send('\x0f') }) await renderAfter(harness, () => { harness.terminal.send('\x0f') }) await checkpoint('tool-cards-hidden-folded', harness.terminal, { includeScrollback: true }) + nowSpy.mockRestore() await disposeSnapshot(harness) }) From 3899a26c87bffcf1b89cdb03ed12f29967d0f076 Mon Sep 17 00:00:00 2001 From: Turtle Date: Thu, 30 Jul 2026 18:26:26 +0800 Subject: [PATCH 04/48] feat(tui): /details command sets card visibility and reasoning display /details reports the transcript detail state bare, jumps tool cards to collapsed|expanded|hidden, and sets or toggles reasoning blocks, sharing the closure state behind Ctrl+O/Ctrl+R via setToolsVisibility/setReasoning. Also fixes a replay defect the reasoning rebuild exposed: rebuildTranscript reused a settled StreamingAssistantComponent for a later assistant/message of the same step, overwriting the earlier content; the settled check now lives in renderEvent for both live and replay paths (untrusted-controls re-recorded with the previously dropped content present). --- .../2026-07-30-tui-details-command.i18n.yaml | 6 ++ .../feature/2026-07-30-tui-details-command.md | 31 ++++++++++ .../2026-07-30-tui-details-command.zh.md | 31 ++++++++++ packages/ui/tui/README.md | 2 +- packages/ui/tui/README.zh.md | 2 +- packages/ui/tui/src/index.ts | 62 ++++++++++++++++--- .../snapshots/details-command.expected.txt | 45 ++++++++++++++ .../snapshots/disposed-terminal.expected.txt | 54 ++++++++-------- .../snapshots/errors-and-help.expected.txt | 52 +++++++++------- .../snapshots/untrusted-controls.expected.txt | 55 ++++++++-------- packages/ui/tui/tests/tui.snapshot.ts | 33 ++++++++++ packages/ui/tui/tests/tui.spec.ts | 33 ++++++++++ 12 files changed, 316 insertions(+), 90 deletions(-) create mode 100644 .agents/notes/implemented/feature/2026-07-30-tui-details-command.i18n.yaml create mode 100644 .agents/notes/implemented/feature/2026-07-30-tui-details-command.md create mode 100644 .agents/notes/implemented/feature/2026-07-30-tui-details-command.zh.md create mode 100644 packages/ui/tui/tests/snapshots/details-command.expected.txt diff --git a/.agents/notes/implemented/feature/2026-07-30-tui-details-command.i18n.yaml b/.agents/notes/implemented/feature/2026-07-30-tui-details-command.i18n.yaml new file mode 100644 index 0000000000..df7079c9b8 --- /dev/null +++ b/.agents/notes/implemented/feature/2026-07-30-tui-details-command.i18n.yaml @@ -0,0 +1,6 @@ +# Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each +# side as of the last confirmed-consistent state. Both languages carry equal authority; +# after editing either side, bring the other along and re-record with: +# pnpm run verify-translation-pairing --write .agents/notes/implemented/feature/2026-07-30-tui-details-command.md +2026-07-30-tui-details-command.md: f17c168cb87d546ca158eddef7dd96fb0ab8be2d +2026-07-30-tui-details-command.zh.md: 48e9a8d3f18c51982b8678e39d9315ffe08b015f diff --git a/.agents/notes/implemented/feature/2026-07-30-tui-details-command.md b/.agents/notes/implemented/feature/2026-07-30-tui-details-command.md new file mode 100644 index 0000000000..f17c168cb8 --- /dev/null +++ b/.agents/notes/implemented/feature/2026-07-30-tui-details-command.md @@ -0,0 +1,31 @@ +# Agent Note: /details command for transcript detail state + +Status: implemented + +English | [中文](2026-07-30-tui-details-command.zh.md) + +## Problem + +The TUI's transcript detail state — tool-card visibility (`collapsed`/`expanded`/`hidden`, per the [consolidated TUI presentation](../architecture/2026-07-28-consolidated-tui-presentation.md)) and reasoning-block display — was reachable only through the Ctrl+O cycle and the Ctrl+R toggle. A user who wants a specific mode must cycle through the others, cannot set both dimensions in one action, and has no way to query the current state; a terminal that swallows those control keys has no fallback at all. + +## Decision + +`dsh-tui` registers `/details` beside its other agent-scoped commands. Bare `/details` reports the current state in one notice. Arguments name target states directly: `collapsed|expanded|hidden` jumps tool cards to that phase, `reasoning on|off` sets reasoning display, bare `reasoning` toggles it, and directives combine in one invocation. An unknown token returns a command error carrying the usage line. The command mutates the same closure state as the shortcuts, refactored so the cycle and toggle are thin wrappers over `setToolsVisibility`/`setReasoning`; the shortcuts and their notices are unchanged. + +A combined invocation applies reasoning before visibility because `setReasoning` rebuilds the transcript from session events, which drops non-durable notice components; applying it last would erase the just-appended visibility notice. + +The reasoning rebuild exposed a replay defect that this change fixes in `renderEvent`: the live path cleared a settled `StreamingAssistantComponent` before a later `assistant/message` of the same step (so the second message got a fresh component), but `rebuildTranscript` replay reused the settled component and `settle()` overwrote its content, silently dropping the earlier message's text. The settled check now lives in `renderEvent`'s `assistant/message` case — one home for both paths — and the previously wrong `untrusted-controls` snapshot (an empty `Assistant` header where reasoning and text had been dropped) was re-recorded with the content present. + +## Alternatives considered + +**Cycle on bare `/details`, mirroring Ctrl+O.** Rejected: the command's value over the shortcut is naming an absolute state; a cycling command is the shortcut with more keystrokes, and bare invocation is more useful as a state report. + +**Separate `/tools` and `/reasoning` commands.** Rejected: both dimensions are one presentation concern ("how much detail does the transcript show"), and a single command keeps the registry and `/help` list small while allowing one combined invocation. + +**Config-key defaults per mode.** Out of scope: `showReasoning` already exists as config; the command is runtime state on top of it, matching the shortcuts. + +## Consequences + +- A user can jump to any detail mode, set both dimensions at once, and query the state — including on terminals that intercept Ctrl+O/Ctrl+R. +- The parser accepts order-free tokens, so `/details reasoning expanded` toggles reasoning and expands cards; last directive wins per dimension. This leniency is deliberate and documented in the README. +- Transcript rebuilds no longer lose assistant messages when a step carries more than one `assistant/message` event; the `details-command` snapshot pins the command surface and the fixed replay together. diff --git a/.agents/notes/implemented/feature/2026-07-30-tui-details-command.zh.md b/.agents/notes/implemented/feature/2026-07-30-tui-details-command.zh.md new file mode 100644 index 0000000000..48e9a8d3f1 --- /dev/null +++ b/.agents/notes/implemented/feature/2026-07-30-tui-details-command.zh.md @@ -0,0 +1,31 @@ +# Agent Note: 用于 transcript 细节状态的 /details 命令 + +Status: implemented + +[English](2026-07-30-tui-details-command.md) | 中文 + +## Problem + +TUI 的 transcript（文本记录）细节状态——工具卡片可见性（`collapsed`/`expanded`/`hidden`，见[整合的 TUI 展示](../architecture/2026-07-28-consolidated-tui-presentation.md)）与 reasoning 块显示——过去只能通过 Ctrl+O 循环和 Ctrl+R 切换来触达。想要某个特定模式的用户必须循环经过其他模式，无法一次操作同时设置两个维度，也无法查询当前状态；吞掉这些控制键的终端更是完全没有替代途径。 + +## Decision + +`dsh-tui` 在其他 agent 作用域命令旁注册 `/details`。裸 `/details` 用一条通知报告当前状态。参数直接命名目标状态：`collapsed|expanded|hidden` 让工具卡片跳到该阶段，`reasoning on|off` 设置 reasoning 显示，裸 `reasoning` 切换它，且指令可在一次调用中组合。未知 token 返回携带用法行的命令错误。命令改动的是与快捷键相同的闭包状态，重构后循环与切换成为 `setToolsVisibility`/`setReasoning` 之上的薄封装；快捷键及其通知保持不变。 + +组合调用先应用 reasoning 再应用可见性，因为 `setReasoning` 会从会话事件重建 transcript，而重建会丢弃非持久的通知组件；若最后才应用它，会抹掉刚追加的可见性通知。 + +reasoning 重建暴露了一个重放缺陷，本变更在 `renderEvent` 中修复：实时路径会在同一步骤的后续 `assistant/message` 之前清除已结算的 `StreamingAssistantComponent`（因此第二条消息获得新组件），但 `rebuildTranscript` 重放复用了已结算组件，`settle()` 覆盖其内容，静默丢掉了前一条消息的文本。已结算检查现在位于 `renderEvent` 的 `assistant/message` 分支——两条路径共用一个归属地——此前错误的 `untrusted-controls` 快照（reasoning 与文本被丢弃后只剩空 `Assistant` 标题）已重录为包含内容的版本。 + +## Alternatives considered + +**裸 `/details` 像 Ctrl+O 一样循环。** 否决：命令相对快捷键的价值在于命名绝对状态；循环命令只是按键更多的快捷键，裸调用作为状态报告更有用。 + +**拆分 `/tools` 与 `/reasoning` 两个命令。** 否决：两个维度同属一个展示关注点（“transcript 显示多少细节”），单一命令让注册表与 `/help` 列表更小，同时允许一次组合调用。 + +**按模式提供配置键默认值。** 超出范围：`showReasoning` 已作为配置存在；命令是其上的运行时状态，与快捷键一致。 + +## Consequences + +- 用户可以跳到任意细节模式、一次设置两个维度并查询状态——包括在拦截 Ctrl+O/Ctrl+R 的终端上。 +- 解析器接受无序 token，因此 `/details reasoning expanded` 会切换 reasoning 并展开卡片；每个维度以最后一个指令为准。这一宽松是刻意的，并记录在 README 中。 +- 当一个步骤携带多条 `assistant/message` 事件时，transcript 重建不再丢失 assistant 消息；`details-command` 快照同时固定命令表面与修复后的重放。 diff --git a/packages/ui/tui/README.md b/packages/ui/tui/README.md index 55e7eb3280..23ca8343ec 100644 --- a/packages/ui/tui/README.md +++ b/packages/ui/tui/README.md @@ -22,7 +22,7 @@ Typing `@` at a token boundary searches files and directories under the session When optional `ctx.sessionReferences` is mounted, the same `@` menu also offers metadata-only session candidates, inserts `@[label](dsh-session:)`, and prepares the selected snapshots before dispatch. Session references remain structured because the model has no filesystem-like tool for retrieving session snapshots later. Preparation disables duplicate submission and restores the editor input on failure. The TUI chooses `agent.steer()` or `agent.followup()` from the status after that asynchronous preparation, so idle follow-ups still dispatch `agent/prompt-submit` while in-turn steering joins at a checkpoint without that hook. -While the agent is running, ordinary editor submissions call `agent.steer()`; otherwise they call `agent.followup()`. A slash at the start of the submitted line enters `ctx.commands` instead: known commands execute directly, unknown commands produce a warning, and neither path automatically reaches the model. A command producer may explicitly schedule agent work; [`dsh-plan-mode`](../../plan/plan-mode/README.md#model-and-human-surfaces) uses that contract for `/plan [message]`. The TUI registers `/help`, `/model`, `/clear`, `/palette`, `/reload`, `/resume`, `/status`, and `/exit` as agent-scoped definitions; every other effective command joins autocomplete and `/help` dynamically, as do `/skill:` completions. A status line above the editor reports the turn phase the TUI derives from session events — waiting for the first token, thinking, responding, or executing tools — with the elapsed time in that phase and the running step total, refreshed each second, and ends with the `Enter sends steering, Esc cancels` hint; while steering messages wait to reach the model it inserts a `N queued ·` badge before the hint that clears as each drains. Ctrl+C or Escape cancels a running turn. Tool and injected-context cards collapse long bodies into a configurable head/tail preview; Ctrl+O cycles tool cards through collapsed preview, full output, and hidden — the hidden phase drops tool cards from the transcript entirely while context cards stay at their preview, since injected instructions are not tool traffic. The hidden phase also folds each turn's assistant steps into one message: the first step with visible text or reasoning keeps the turn's single `Assistant` header, later steps render as headerless continuations, and a step without a visible body renders nothing; leaving the hidden phase restores the per-step headers. An injected-context card renders its message as prose with the producer's outer reminder frame stripped, so neither the fold nor the frame stripping depends on the payload's syntax. Ctrl+R toggles reasoning, Ctrl+L redraws, and Ctrl+D exits while idle. +While the agent is running, ordinary editor submissions call `agent.steer()`; otherwise they call `agent.followup()`. A slash at the start of the submitted line enters `ctx.commands` instead: known commands execute directly, unknown commands produce a warning, and neither path automatically reaches the model. A command producer may explicitly schedule agent work; [`dsh-plan-mode`](../../plan/plan-mode/README.md#model-and-human-surfaces) uses that contract for `/plan [message]`. The TUI registers `/help`, `/model`, `/clear`, `/details`, `/palette`, `/reload`, `/resume`, `/status`, and `/exit` as agent-scoped definitions; every other effective command joins autocomplete and `/help` dynamically, as do `/skill:` completions. A status line above the editor reports the turn phase the TUI derives from session events — waiting for the first token, thinking, responding, or executing tools — with the elapsed time in that phase and the running step total, refreshed each second, and ends with the `Enter sends steering, Esc cancels` hint; while steering messages wait to reach the model it inserts a `N queued ·` badge before the hint that clears as each drains. Ctrl+C or Escape cancels a running turn. Tool and injected-context cards collapse long bodies into a configurable head/tail preview; Ctrl+O cycles tool cards through collapsed preview, full output, and hidden — the hidden phase drops tool cards from the transcript entirely while context cards stay at their preview, since injected instructions are not tool traffic. The hidden phase also folds each turn's assistant steps into one message: the first step with visible text or reasoning keeps the turn's single `Assistant` header, later steps render as headerless continuations, and a step without a visible body renders nothing; leaving the hidden phase restores the per-step headers. An injected-context card renders its message as prose with the producer's outer reminder frame stripped, so neither the fold nor the frame stripping depends on the payload's syntax. Ctrl+R toggles reasoning, Ctrl+L redraws, and Ctrl+D exits while idle. `/details` names the same state those two shortcuts cycle: bare it opens a centered keyboard toggle with one entry per dimension — `Tool cards` and `Reasoning` — showing the live values, where Tab cycles the highlighted entry and applies the change immediately (the transcript behind the dialog is the preview), and Enter, Esc, or Ctrl+C closes; `/details collapsed|expanded|hidden` jumps tool cards to that phase directly, and `/details reasoning [on|off]` sets — or bare `reasoning` toggles — reasoning-block display; arguments combine in one invocation, an unknown argument fails with the usage line, and a combined invocation applies reasoning first so its transcript rebuild never drops the card notice. `/model` opens the advisory `ctx.llm` catalog as a keyboard selector: a filter box above the list narrows rows by a case-insensitive substring over each row's `provider/model` label, model name, and description, keeping the highlighted row selected when it survives the filter; Up/Down moves, Shift+Tab cycles the focused model's adapter-advertised reasoning efforts in display order, Enter selects the model and effort, and Escape clears a non-empty filter before a second Escape closes it. When an adapter does not advertise a default effort, the cycle also includes `Default`, which clears an explicit selection and preserves the provider default; models without selectable effort metadata ignore Shift+Tab. The selector renders the exact advertised effort list—including `off` when present—and does not synthesize, clamp, or transfer an effort between models. `/model ` still selects an unambiguous model id directly, while `/model /` selects an exact target and uses its adapter default when one exists. The configured target or latest logged request header initializes the selector, and an unlisted current model remains visible because catalogs are advisory. Selection is local to this TUI session. Prompt assembly snapshots the target for one step, replaces `{{provider}}` and `{{model}}`, and applies the same provider/model/reasoning-effort target through `agent/request`; a switch during assembly therefore starts with a later step. The request header durably records targets that reach the model, while an unused selection remains process-local. diff --git a/packages/ui/tui/README.zh.md b/packages/ui/tui/README.zh.md index 9683f6292e..40aa849838 100644 --- a/packages/ui/tui/README.zh.md +++ b/packages/ui/tui/README.zh.md @@ -22,7 +22,7 @@ TUI 从追加来源的会话事件重建已恢复历史，渲染 Markdown 响应 挂载可选的 `ctx.sessionReferences` 后，同一个 `@` 菜单还会提供仅含元数据的会话候选项，插入 `@[label](dsh-session:)`，并在分派前准备所选快照。会话引用保持结构化，因为模型没有类似文件系统的工具可在稍后检索会话快照。准备期间会禁止重复提交，并在失败时恢复编辑器输入。TUI 会在异步准备后根据状态选择 `agent.steer()` 或 `agent.followup()`，因此空闲 followup 仍会分派 `agent/prompt-submit`，而轮次中的 steering 会在检查点加入且不触发该 hook。 -Agent 运行时，普通编辑器提交会调用 `agent.steer()`；其他时候调用 `agent.followup()`。提交行以斜杠开头时会改为进入 `ctx.commands`：已知命令直接执行，未知命令产生警告，两条路径都不会自动到达模型。命令生产方可以显式调度 agent 工作；[`dsh-plan-mode`](../../plan/plan-mode/README.md#model-and-human-surfaces) 使用该契约实现 `/plan [message]`。TUI 将 `/help`、`/model`、`/clear`、`/palette`、`/reload`、`/resume`、`/status` 和 `/exit` 注册为 agent 作用域定义；其他所有有效命令都会动态加入自动补全与 `/help`，`/skill:` 补全也相同。编辑器上方的状态行会报告 TUI 从会话事件派生的轮次阶段，包括等待首个 token、思考、响应或执行工具；它显示该阶段已经过时间和运行中的步骤总数，每秒刷新，并以 `Enter sends steering, Esc cancels` 提示结尾。Steering 消息等待到达模型期间，会在提示前插入 `N queued ·` 徽标，每条消息排空后随即清除。Ctrl+C 或 Escape 会取消运行中的轮次。工具卡片与注入上下文卡片都把长主体折叠为可配置的头尾预览；Ctrl+O 让工具卡片在折叠预览、完整输出、隐藏三种状态间循环——隐藏阶段把工具卡片从 transcript 中完全去掉，而上下文卡片保持预览，因为注入的指令不属于工具流量。隐藏阶段还会把每个轮次的 assistant 步骤折叠为一条消息：第一个有可见文本或 reasoning 的步骤保留该轮次唯一的 `Assistant` 标题，之后的步骤渲染为无标题的续段，没有可见正文的步骤则不渲染任何内容；离开隐藏阶段会恢复每步各自的标题。注入上下文卡片把消息渲染为文本，并去掉生产方的外层提醒外框，因此折叠与去外框都不依赖载荷的语法。Ctrl+R 切换 reasoning，Ctrl+L 重绘，Ctrl+D 在空闲时退出。 +Agent 运行时，普通编辑器提交会调用 `agent.steer()`；其他时候调用 `agent.followup()`。提交行以斜杠开头时会改为进入 `ctx.commands`：已知命令直接执行，未知命令产生警告，两条路径都不会自动到达模型。命令生产方可以显式调度 agent 工作；[`dsh-plan-mode`](../../plan/plan-mode/README.md#model-and-human-surfaces) 使用该契约实现 `/plan [message]`。TUI 将 `/help`、`/model`、`/clear`、`/details`、`/palette`、`/reload`、`/resume`、`/status` 和 `/exit` 注册为 agent 作用域定义；其他所有有效命令都会动态加入自动补全与 `/help`，`/skill:` 补全也相同。编辑器上方的状态行会报告 TUI 从会话事件派生的轮次阶段，包括等待首个 token、思考、响应或执行工具；它显示该阶段已经过时间和运行中的步骤总数，每秒刷新，并以 `Enter sends steering, Esc cancels` 提示结尾。Steering 消息等待到达模型期间，会在提示前插入 `N queued ·` 徽标，每条消息排空后随即清除。Ctrl+C 或 Escape 会取消运行中的轮次。工具卡片与注入上下文卡片都把长主体折叠为可配置的头尾预览；Ctrl+O 让工具卡片在折叠预览、完整输出、隐藏三种状态间循环——隐藏阶段把工具卡片从 transcript 中完全去掉，而上下文卡片保持预览，因为注入的指令不属于工具流量。隐藏阶段还会把每个轮次的 assistant 步骤折叠为一条消息：第一个有可见文本或 reasoning 的步骤保留该轮次唯一的 `Assistant` 标题，之后的步骤渲染为无标题的续段，没有可见正文的步骤则不渲染任何内容；离开隐藏阶段会恢复每步各自的标题。注入上下文卡片把消息渲染为文本，并去掉生产方的外层提醒外框，因此折叠与去外框都不依赖载荷的语法。Ctrl+R 切换 reasoning，Ctrl+L 重绘，Ctrl+D 在空闲时退出。`/details` 命名的正是这两个快捷键循环的同一份状态：不带参数时打开一个居中的键盘开关，每个维度一个条目——`Tool cards` 与 `Reasoning`——显示实时值，Tab 循环高亮条目并立即应用变更（对话框背后的 transcript 即是预览），Enter、Esc 或 Ctrl+C 关闭；`/details collapsed|expanded|hidden` 让工具卡片直接跳到该阶段，`/details reasoning [on|off]` 设置——或裸 `reasoning` 切换——reasoning 块显示；参数可在一次调用中组合，未知参数会以用法行报错，组合调用先应用 reasoning，使其 transcript 重建不会丢掉卡片通知。 `/model` 将建议性的 `ctx.llm` catalog 打开为键盘选择器：列表上方设有一个过滤框，按对每行 `provider/model` 标签、模型名称和描述的大小写不敏感子串匹配来缩小行集，并在高亮行仍通过过滤时保持其选中状态；Up/Down 移动，Shift+Tab 按显示顺序循环切换适配器为焦点模型公布的推理强度，Enter 选择模型和推理强度，Escape 会先清除非空过滤内容，再次按下才关闭选择器。适配器未公布默认推理强度时，循环还会包含 `Default`，该项会清除显式选择并保留提供方默认行为；没有可选推理强度元数据的模型会忽略 Shift+Tab。选择器会原样呈现公布的推理强度列表（包括存在时的 `off`），不会合成、自动调整或在模型之间转移推理强度。`/model ` 仍可直接选择无歧义的模型 id，`/model /` 则选择精确目标，并在存在时使用其适配器默认值。已配置目标或最新记录的请求 header 会初始化选择器；由于 catalog 仅提供建议，未列出的当前模型仍会显示。选择仅对本 TUI 会话有效。提示词组装会为一个步骤建立目标快照，替换 `{{provider}}` 和 `{{model}}`，并通过 `agent/request` 应用同一个提供方／模型／推理强度目标；因此组装期间的切换会从后续步骤开始生效。请求 header 会持久记录真正到达模型的目标，未使用的选择则只存在于进程本地。 diff --git a/packages/ui/tui/src/index.ts b/packages/ui/tui/src/index.ts index 3d39c53957..6470e70114 100644 --- a/packages/ui/tui/src/index.ts +++ b/packages/ui/tui/src/index.ts @@ -774,7 +774,9 @@ export function createTuiChat( break case 'assistant/message': completedStreaming = undefined - if (streaming === undefined || !chat.children.includes(streaming)) startAssistantStep(event.data) + // A settled component stays attached but never absorbs a later message + // of the same step; both the live and replay paths start a new one. + if (streaming === undefined || streaming.isSettled() || !chat.children.includes(streaming)) startAssistantStep(event.data) if (streaming !== undefined) { streaming.settle(event.data.message.content) applyTurnFolding(streaming.position.turn) @@ -999,11 +1001,8 @@ export function createTuiChat( // same reason. ui.queryTerminalColorScheme({ timeoutMs: 2000 }).catch(() => {}) - const toggleTools = (): void => { - // The cycle order puts the two common reading modes adjacent: preview -> - // full detail -> conversation-only, then back to the preview default. - toolsVisibility = toolsVisibility === 'collapsed' ? 'expanded' - : toolsVisibility === 'expanded' ? 'hidden' : 'collapsed' + const setToolsVisibility = (next: ToolCardVisibility): void => { + toolsVisibility = next for (const card of allToolCards) card.setVisibility(toolsVisibility) // Context cards carry injected instructions rather than tool traffic, so // they never hide: the hidden phase reads as their collapsed preview. @@ -1014,8 +1013,15 @@ export function createTuiChat( appendNotice(toolsVisibility === 'hidden' ? 'Tool cards hidden.' : `Tool and context cards ${toolsVisibility}.`) } - const toggleReasoning = (): void => { - showReasoning = !showReasoning + const toggleTools = (): void => { + // The cycle order puts the two common reading modes adjacent: preview -> + // full detail -> conversation-only, then back to the preview default. + setToolsVisibility(toolsVisibility === 'collapsed' ? 'expanded' + : toolsVisibility === 'expanded' ? 'hidden' : 'collapsed') + } + + const setReasoning = (show: boolean): void => { + showReasoning = show const activeStreaming = streaming rebuildTranscript(false) /* v8 ignore next -- the non-streaming command path is covered; this branch preserves an active stream across rebuild. */ @@ -1029,6 +1035,39 @@ export function createTuiChat( appendNotice(`Reasoning blocks ${showReasoning ? 'shown' : 'hidden'}.`) } + const toggleReasoning = (): void => { setReasoning(!showReasoning) } + + // `/details` names the same transcript-detail state the Ctrl+O cycle and + // Ctrl+R toggle mutate, so a user can jump to a mode without cycling. + const runDetails = (rawInput: string): CommandResult => { + const tokens = rawInput.split(/\s+/u).filter(token => token !== '') + if (tokens.length === 0) { + appendNotice(`Tool and context cards ${toolsVisibility}; reasoning blocks ${showReasoning ? 'shown' : 'hidden'}.`) + return { kind: 'success' } + } + let visibility: ToolCardVisibility | undefined + let reasoning: boolean | undefined + for (let token = tokens.shift(); token !== undefined; token = tokens.shift()) { + if (token === 'collapsed' || token === 'expanded' || token === 'hidden') { + visibility = token + } else if (token === 'reasoning') { + const value = tokens[0] + if (value === 'on' || value === 'off') { + tokens.shift() + reasoning = value === 'on' + } else { + reasoning = !showReasoning + } + } else { + return { kind: 'error', text: `Unknown /details argument "${token}". Usage: /details [collapsed|expanded|hidden] [reasoning [on|off]]` } + } + } + // Reasoning first: its transcript rebuild would drop the visibility notice. + if (reasoning !== undefined) setReasoning(reasoning) + if (visibility !== undefined) setToolsVisibility(visibility) + return { kind: 'success' } + } + const showHelp = (): void => { const commandLines = ctx.commands.list(agent).map((command) => { const input = command.input === undefined ? '' : ` ${command.input.hint}` @@ -1214,6 +1253,12 @@ export function createTuiChat( description: 'Clear the transcript view (session history is unchanged)', handler: () => { chat.clear(); requestRender(); return { kind: 'success' } }, }) + commandCtx.commands.register({ + name: 'details', + description: 'Show or set tool-card visibility and reasoning display', + input: { hint: '[collapsed|expanded|hidden] [reasoning [on|off]]' }, + handler: ({ rawInput }) => runDetails(rawInput), + }) commandCtx.commands.register({ name: 'palette', description: 'Show every color and attribute role this terminal renders', @@ -1553,7 +1598,6 @@ export function createTuiChat( if (event.type === 'tool/result') fileSearch.invalidate() recordEventUsage(tokens, event) if (event.type === 'turn/start' && runningStatus !== undefined) runningStatus.turn = event.data.turn - if (event.type === 'assistant/message' && streaming?.isSettled()) streaming = undefined // A replacement mutates only the model surface, so the rendered transcript // keeps what it already showed; a landed summary checkpoint adds its marker. if (isReplacementSurfaceEvent(event)) { diff --git a/packages/ui/tui/tests/snapshots/details-command.expected.txt b/packages/ui/tui/tests/snapshots/details-command.expected.txt new file mode 100644 index 0000000000..4e84695616 --- /dev/null +++ b/packages/ui/tui/tests/snapshots/details-command.expected.txt @@ -0,0 +1,45 @@ +terminal 100x40 buffer=normal length=40 base=0 viewport=0 +lifecycle started=1 stopped=0 progress=inactive +title "DSH snapshot" +cursor hidden column=7 viewportRow=19 bufferRow=19 +buffer +0| " DEEPSEEK HARNESS" + style 1-8 fg=bright-magenta bold + style 10-16 bold +1| " Snapshot agent ready." + style 1-21 dim +2| " main-session" + style 1-12 dim +3| +4| "Assistant " + style 0-8 fg=bright-magenta bold underline +5| "Running the check now. " +6| "Model wait 0.0s " + style 0-14 dim +7| +8| "You " + style 0-2 fg=bright-magenta bold underline +9| "Inspect the renderer. " +10| "Model wait 0.0s · Completed 2026-07-30 18:00:00 " + style 0-46 dim +11| +12| "Reasoning blocks hidden. " + style 0-23 dim +13| +14| "Tool cards hidden. " + style 0-17 dim +15| +16| "Tool and context cards hidden; reasoning blocks hidden. " + style 0-54 dim +17| +18| "/workspace/project (tui-staging) deepseek-v4-flash ↑0 ↓0 0% context" + style 0-17 fg=bright-magenta bold + style 18-31 dim + style 34-50 dim + style 53-57 dim + style 60-69 dim +19| " dsh > " + style 1-3 fg=bright-magenta bold + style 5-6 dim + style 7-7 inverse +20-39| diff --git a/packages/ui/tui/tests/snapshots/disposed-terminal.expected.txt b/packages/ui/tui/tests/snapshots/disposed-terminal.expected.txt index a6a73566d6..aa41ca137d 100644 --- a/packages/ui/tui/tests/snapshots/disposed-terminal.expected.txt +++ b/packages/ui/tui/tests/snapshots/disposed-terminal.expected.txt @@ -1,7 +1,7 @@ -terminal 92x32 buffer=normal length=37 base=5 viewport=5 +terminal 92x32 buffer=normal length=39 base=7 viewport=7 lifecycle started=1 stopped=1 progress=inactive title "DSH snapshot" -cursor visible column=0 viewportRow=31 bufferRow=36 +cursor visible column=0 viewportRow=31 bufferRow=38 buffer 0| " DEEPSEEK HARNESS" style 1-8 fg=bright-magenta bold @@ -29,48 +29,52 @@ buffer 12| " " 13| "/clear — Clear the transcript view (session history is unchanged) " style 0-64 dim -14| "/exit — Exit after the active turn reaches idle " +14| "/details [collapsed|expanded|hidden] [reasoning [on|off]] — Show or set tool-card visibility" + style 0-91 dim +15| "and reasoning display " + style 0-20 dim +16| "/exit — Exit after the active turn reaches idle " style 0-46 dim -15| "/help — Show keyboard shortcuts and commands " +17| "/help — Show keyboard shortcuts and commands " style 0-43 dim -16| "/model [[provider/]model] — Show or switch this session's model " +18| "/model [[provider/]model] — Show or switch this session's model " style 0-62 dim -17| "/palette — Show every color and attribute role this terminal renders " +19| "/palette — Show every color and attribute role this terminal renders " style 0-67 dim -18| "/quit — Exit after the active turn reaches idle " +20| "/quit — Exit after the active turn reaches idle " style 0-46 dim -19| "/reload — EXPERIMENTAL (dev): re-read loader config files and apply the diff (idle only) " +21| "/reload — EXPERIMENTAL (dev): re-read loader config files and apply the diff (idle only) " style 0-87 dim -20| "/resume — List this workspace's resumable sessions " +22| "/resume — List this workspace's resumable sessions " style 0-49 dim -21| "/status — Show session diagnostics, system prompt, and registered tools " +23| "/status — Show session diagnostics, system prompt, and registered tools " style 0-70 dim -22| "/skill: [instructions] — load a skill into the conversation " +24| "/skill: [instructions] — load a skill into the conversation " style 0-64 dim -23| -24| "provider stream failed after partial output " - style 0-42 fg=red 25| -26| "The previous process ended during this turn. " - style 0-43 fg=yellow +26| "provider stream failed after partial output " + style 0-42 fg=red 27| -28| "Turn stopped: the agent was disposed. " - style 0-36 fg=yellow +28| "The previous process ended during this turn. " + style 0-43 fg=yellow 29| -30| "Turn ended: plugin-policy. " - style 0-25 fg=yellow +30| "Turn stopped: the agent was disposed. " + style 0-36 fg=yellow 31| -32| "Unknown command: /unknown-advanced-command " - style 0-41 fg=yellow +32| "Turn ended: plugin-policy. " + style 0-25 fg=yellow 33| -34| "/workspace/project (tui-staging) deepseek-v4-flash ↑0 ↓0 0% context" +34| "Unknown command: /unknown-advanced-command " + style 0-41 fg=yellow +35| +36| "/workspace/project (tui-staging) deepseek-v4-flash ↑0 ↓0 0% context" style 0-17 fg=bright-magenta bold style 18-31 dim style 34-50 dim style 53-57 dim style 60-69 dim -35| " dsh > " +37| " dsh > " style 1-3 fg=bright-magenta bold style 5-6 dim style 7-7 inverse -36| +38| diff --git a/packages/ui/tui/tests/snapshots/errors-and-help.expected.txt b/packages/ui/tui/tests/snapshots/errors-and-help.expected.txt index b8b62bca11..48c25cf8fa 100644 --- a/packages/ui/tui/tests/snapshots/errors-and-help.expected.txt +++ b/packages/ui/tui/tests/snapshots/errors-and-help.expected.txt @@ -1,7 +1,7 @@ -terminal 92x32 buffer=normal length=36 base=4 viewport=4 +terminal 92x32 buffer=normal length=38 base=6 viewport=6 lifecycle started=1 stopped=0 progress=inactive title "DSH snapshot" -cursor hidden column=7 viewportRow=31 bufferRow=35 +cursor hidden column=7 viewportRow=31 bufferRow=37 buffer 0| " DEEPSEEK HARNESS" style 1-8 fg=bright-magenta bold @@ -29,47 +29,51 @@ buffer 12| " " 13| "/clear — Clear the transcript view (session history is unchanged) " style 0-64 dim -14| "/exit — Exit after the active turn reaches idle " +14| "/details [collapsed|expanded|hidden] [reasoning [on|off]] — Show or set tool-card visibility" + style 0-91 dim +15| "and reasoning display " + style 0-20 dim +16| "/exit — Exit after the active turn reaches idle " style 0-46 dim -15| "/help — Show keyboard shortcuts and commands " +17| "/help — Show keyboard shortcuts and commands " style 0-43 dim -16| "/model [[provider/]model] — Show or switch this session's model " +18| "/model [[provider/]model] — Show or switch this session's model " style 0-62 dim -17| "/palette — Show every color and attribute role this terminal renders " +19| "/palette — Show every color and attribute role this terminal renders " style 0-67 dim -18| "/quit — Exit after the active turn reaches idle " +20| "/quit — Exit after the active turn reaches idle " style 0-46 dim -19| "/reload — EXPERIMENTAL (dev): re-read loader config files and apply the diff (idle only) " +21| "/reload — EXPERIMENTAL (dev): re-read loader config files and apply the diff (idle only) " style 0-87 dim -20| "/resume — List this workspace's resumable sessions " +22| "/resume — List this workspace's resumable sessions " style 0-49 dim -21| "/status — Show session diagnostics, system prompt, and registered tools " +23| "/status — Show session diagnostics, system prompt, and registered tools " style 0-70 dim -22| "/skill: [instructions] — load a skill into the conversation " +24| "/skill: [instructions] — load a skill into the conversation " style 0-64 dim -23| -24| "provider stream failed after partial output " - style 0-42 fg=red 25| -26| "The previous process ended during this turn. " - style 0-43 fg=yellow +26| "provider stream failed after partial output " + style 0-42 fg=red 27| -28| "Turn stopped: the agent was disposed. " - style 0-36 fg=yellow +28| "The previous process ended during this turn. " + style 0-43 fg=yellow 29| -30| "Turn ended: plugin-policy. " - style 0-25 fg=yellow +30| "Turn stopped: the agent was disposed. " + style 0-36 fg=yellow 31| -32| "Unknown command: /unknown-advanced-command " - style 0-41 fg=yellow +32| "Turn ended: plugin-policy. " + style 0-25 fg=yellow 33| -34| "/workspace/project (tui-staging) deepseek-v4-flash ↑0 ↓0 0% context" +34| "Unknown command: /unknown-advanced-command " + style 0-41 fg=yellow +35| +36| "/workspace/project (tui-staging) deepseek-v4-flash ↑0 ↓0 0% context" style 0-17 fg=bright-magenta bold style 18-31 dim style 34-50 dim style 53-57 dim style 60-69 dim -35| " dsh > " +37| " dsh > " style 1-3 fg=bright-magenta bold style 5-6 dim style 7-7 inverse diff --git a/packages/ui/tui/tests/snapshots/untrusted-controls.expected.txt b/packages/ui/tui/tests/snapshots/untrusted-controls.expected.txt index 74cdd416d9..e5a625afc0 100644 --- a/packages/ui/tui/tests/snapshots/untrusted-controls.expected.txt +++ b/packages/ui/tui/tests/snapshots/untrusted-controls.expected.txt @@ -13,46 +13,41 @@ buffer 3| 4| "Assistant " style 0-8 fg=bright-magenta bold underline -5| -6| "You " +5| "Reasoning " + style 0-8 dim italic +6| "Unsafe reasoning \\x1b]2;snapshot-controlled\\x07\\x09\\x7f\\x9b31m " + style 0-61 dim italic +7| "Unsafe assistant \\x1b]2;snapshot-controlled\\x07\\x09\\x7f\\x9b31m " +8| "Model wait 0.0s " + style 0-14 dim +9| +10| "You " style 0-2 fg=bright-magenta bold underline -7| "Unsafe user \\x1b]2;snapshot-controlled\\x07\\x09\\x7f\\x9b31m " -8| -9| "● Tool / unsafe / Unsafe description \\x1b]2;snapshot-controlled\\x07\\x09\\x7f\\x9b31m" +11| "Unsafe user \\x1b]2;snapshot-controlled\\x07\\x09\\x7f\\x9b31m " +12| +13| "Assistant " + style 0-8 fg=bright-magenta bold underline +14| +15| "● Tool / unsafe / Unsafe description \\x1b]2;snapshot-controlled\\x07\\x09\\x7f\\x9b31m" style 0-81 fg=green -10| "$ Unsafe title \\x1b]2;snapshot-controlled\\x07\\x09\\x7f\\x9b31m " +16| "$ Unsafe title \\x1b]2;snapshot-controlled\\x07\\x09\\x7f\\x9b31m " style 0-59 dim -11| "/unsafe/\\x1b]2;snapshot-controlled\\x07\\x09\\x7f\\x9b31m " +17| "/unsafe/\\x1b]2;snapshot-controlled\\x07\\x09\\x7f\\x9b31m " style 0-52 dim -12| "Unsafe output \\x1b]2;snapshot-controlled\\x07\\x09\\x7f\\x9b31m " +18| "Unsafe output \\x1b]2;snapshot-controlled\\x07\\x09\\x7f\\x9b31m " style 0-58 dim -13| "[signal SIG\\x1b]2;snapshot-controlled\\x07\\x09\\x7f\\x9b31m] " +19| "[signal SIG\\x1b]2;snapshot-controlled\\x07\\x09\\x7f\\x9b31m] " style 0-56 fg=red -14| "Model wait 0.0s · Completed 2026-07-21 15:00:00 " +20| "Model wait 0.0s · Completed 2026-07-21 15:00:00 " style 0-46 dim -15| -16| "Context · unsafe-\\x1b]2;snapshot-controlled\\x07\\x09\\x7f\\x9b31m" +21| +22| "Context · unsafe-\\x1b]2;snapshot-controlled\\x07\\x09\\x7f\\x9b31m" style 0-61 dim -17| "Unsafe context \\x1b]2;snapshot-controlled\\x07\\x09\\x7f\\x9b31m " +23| "Unsafe context \\x1b]2;snapshot-controlled\\x07\\x09\\x7f\\x9b31m " style 0-59 dim -18| -19| "Unsafe turn error \\x1b]2;snapshot-controlled\\x07\\x09\\x7f\\x9b31m " +24| +25| "Unsafe turn error \\x1b]2;snapshot-controlled\\x07\\x09\\x7f\\x9b31m " style 0-62 fg=red -20-21| -22| "Plan" - style 0-3 fg=bright-magenta bold -23| " ● Unsafe todo \\x1b]2;snapshot-controlled\\x07\\x09\\x7f\\x9b31m" - style 2-2 fg=yellow -24| "/workspace/project (tui-staging) deepseek-v4-flash ↑0 ↓0 0% context" - style 0-17 fg=bright-magenta bold - style 18-31 dim - style 34-50 dim - style 53-57 dim - style 60-69 dim -25| " dsh > " - style 1-3 fg=bright-magenta bold - style 5-6 dim - style 7-7 inverse 26| " " 27| " Question 1/1 (1 unanswered) · Unsafe header \\x1b]2;snapshot-controlled\\x07\\x09\\x7f\\x9b31m " style 2-90 dim diff --git a/packages/ui/tui/tests/tui.snapshot.ts b/packages/ui/tui/tests/tui.snapshot.ts index d56db82f14..abea27e69e 100644 --- a/packages/ui/tui/tests/tui.snapshot.ts +++ b/packages/ui/tui/tests/tui.snapshot.ts @@ -45,6 +45,7 @@ const CHECKPOINTS = [ 'advanced-cards-collapsed', 'advanced-cards-expanded', 'tool-cards-hidden-folded', + 'details-command', 'untrusted-controls', 'question-dialog', 'question-dialog-single-option', @@ -637,6 +638,38 @@ describe('TUI terminal-state snapshots', () => { await disposeSnapshot(harness) }) + it('pins /details jumping card visibility and reasoning display to named states', async () => { + const nowSpy = vi.spyOn(Date, 'now').mockReturnValue(new Date(2026, 6, 30, 18, 0, 0).getTime()) + const harness = await setupSnapshot({ + tools: ADVANCED_CARD_TOOLS, + config: { maxToolOutputLines: 3 }, + }, { columns: 100, rows: 40 }) + await renderAfter(harness, () => { + appendUser(harness.session, 'Inspect the renderer.') + appendAssistant(harness.session, [ + { type: 'reasoning', text: 'The tool card and this block vanish under /details hidden reasoning off.' }, + { type: 'text', text: 'Running the check now.' }, + ]) + appendToolCalls(harness.session, [ + { id: 'details-1', name: 'bash', arguments: { command: 'pnpm run test' } }, + ]) + appendToolResult(harness.session, 'details-1', [{ type: 'text', text: 'all tests pass' }]) + harness.session.append('step/end', { turn: 1, step: 1 }) + harness.session.append('turn/end', { turn: 1, reason: { kind: 'completed' } }) + }) + await renderAfter(harness, () => { + harness.terminal.send('/details hidden reasoning off') + harness.terminal.send('\r') + }) + await renderAfter(harness, () => { + harness.terminal.send('/details') + harness.terminal.send('\r') + }) + await checkpoint('details-command', harness.terminal, { includeScrollback: true }) + nowSpy.mockRestore() + await disposeSnapshot(harness) + }) + it('renders terminal controls as inert text across transcripts, tools, dialogs, diagnostics, and title', async () => { const nowSpy = vi.spyOn(Date, 'now').mockReturnValue(new Date(2026, 6, 21, 15, 0, 0).getTime()) const tools = { diff --git a/packages/ui/tui/tests/tui.spec.ts b/packages/ui/tui/tests/tui.spec.ts index 66540d2cc2..c532c5a67d 100644 --- a/packages/ui/tui/tests/tui.spec.ts +++ b/packages/ui/tui/tests/tui.spec.ts @@ -2510,6 +2510,39 @@ describe('pi-tui chat lifecycle and transcript', () => { await dispose(result) }) + it('/details reports and sets card visibility and reasoning display', async () => { + const result = await setup() + const run = async (line: string): Promise => { + result.terminal.send(line) + result.terminal.send('\r') + await tick() + } + + await run('/details') + expect(result.terminal.output).toContain('Tool and context cards collapsed; reasoning blocks shown.') + + await run('/details hidden') + expect(result.terminal.output).toContain('Tool cards hidden.') + + await run('/details expanded reasoning off') + expect(result.terminal.output).toContain('Tool and context cards expanded.') + expect(result.terminal.output).toContain('Reasoning blocks hidden.') + + await run('/details reasoning on') + expect(result.terminal.output).toContain('Reasoning blocks shown.') + + // Bare `reasoning` toggles: shown -> hidden, confirmed by the status line. + await run('/details reasoning') + await run('/details collapsed') + await run('/details') + expect(result.terminal.output).toContain('Tool and context cards collapsed; reasoning blocks hidden.') + + await run('/details bogus') + expect(result.terminal.output).toContain('Unknown /details argument "bogus"') + + await dispose(result) + }) + it('sends, steers, handles commands, global keys, and disposed-agent input', async () => { const result = await setup() From 2c98f35a86e70ef943b05c11cfdf42d5fc4897bb Mon Sep 17 00:00:00 2001 From: Turtle Date: Thu, 30 Jul 2026 19:52:58 +0800 Subject: [PATCH 05/48] feat(tui): bare /details opens a keyboard selector DetailsDialog is a centered SelectList over the five transcript-detail states (three tool-card phases, reasoning shown/hidden); it preselects the current phase, marks both current values, applies on Enter, and cancels on Esc/Ctrl+C. Width is the new detailsDialogWidth config key. The argument grammar is unchanged and shares the same setters. --- .../2026-07-30-tui-details-command.i18n.yaml | 4 +- .../feature/2026-07-30-tui-details-command.md | 11 +-- .../2026-07-30-tui-details-command.zh.md | 11 +-- docs/config-catalog.md | 4 +- packages/ui/tui/src/components/dialogs.ts | 58 ++++++++++++++++ packages/ui/tui/src/config.ts | 7 ++ packages/ui/tui/src/index.ts | 32 ++++++++- .../snapshots/details-command.expected.txt | 11 ++- .../snapshots/details-selector.expected.txt | 66 ++++++++++++++++++ .../snapshots/disposed-terminal.expected.txt | 6 +- .../snapshots/errors-and-help.expected.txt | 6 +- packages/ui/tui/tests/tui.snapshot.ts | 6 +- packages/ui/tui/tests/tui.spec.ts | 68 +++++++++++++++++-- 13 files changed, 256 insertions(+), 34 deletions(-) create mode 100644 packages/ui/tui/tests/snapshots/details-selector.expected.txt diff --git a/.agents/notes/implemented/feature/2026-07-30-tui-details-command.i18n.yaml b/.agents/notes/implemented/feature/2026-07-30-tui-details-command.i18n.yaml index df7079c9b8..33278b66d2 100644 --- a/.agents/notes/implemented/feature/2026-07-30-tui-details-command.i18n.yaml +++ b/.agents/notes/implemented/feature/2026-07-30-tui-details-command.i18n.yaml @@ -2,5 +2,5 @@ # side as of the last confirmed-consistent state. Both languages carry equal authority; # after editing either side, bring the other along and re-record with: # pnpm run verify-translation-pairing --write .agents/notes/implemented/feature/2026-07-30-tui-details-command.md -2026-07-30-tui-details-command.md: f17c168cb87d546ca158eddef7dd96fb0ab8be2d -2026-07-30-tui-details-command.zh.md: 48e9a8d3f18c51982b8678e39d9315ffe08b015f +2026-07-30-tui-details-command.md: c14461aa08790033026eead38f53d4b8e686bbbb +2026-07-30-tui-details-command.zh.md: e9c9787a4911f29f49e7709d1ba673dc1a25e8a1 diff --git a/.agents/notes/implemented/feature/2026-07-30-tui-details-command.md b/.agents/notes/implemented/feature/2026-07-30-tui-details-command.md index f17c168cb8..c14461aa08 100644 --- a/.agents/notes/implemented/feature/2026-07-30-tui-details-command.md +++ b/.agents/notes/implemented/feature/2026-07-30-tui-details-command.md @@ -10,7 +10,7 @@ The TUI's transcript detail state — tool-card visibility (`collapsed`/`expande ## Decision -`dsh-tui` registers `/details` beside its other agent-scoped commands. Bare `/details` reports the current state in one notice. Arguments name target states directly: `collapsed|expanded|hidden` jumps tool cards to that phase, `reasoning on|off` sets reasoning display, bare `reasoning` toggles it, and directives combine in one invocation. An unknown token returns a command error carrying the usage line. The command mutates the same closure state as the shortcuts, refactored so the cycle and toggle are thin wrappers over `setToolsVisibility`/`setReasoning`; the shortcuts and their notices are unchanged. +`dsh-tui` registers `/details` beside its other agent-scoped commands. Bare `/details` opens `DetailsDialog`, a centered keyboard selector over the five detail states — the three tool-card phases and reasoning shown/hidden — that preselects the current phase, marks both current values, applies the highlighted state on Enter, and cancels on Esc or Ctrl+C; its width is the `detailsDialogWidth` config key and a second `/details` replaces an open selector, mirroring the `/model` overlay. Arguments name target states directly: `collapsed|expanded|hidden` jumps tool cards to that phase, `reasoning on|off` sets reasoning display, bare `reasoning` toggles it, and directives combine in one invocation. An unknown token returns a command error carrying the usage line. Every entry mutates the same closure state as the shortcuts, refactored so the cycle and toggle are thin wrappers over `setToolsVisibility`/`setReasoning`; the shortcuts and their notices are unchanged. A combined invocation applies reasoning before visibility because `setReasoning` rebuilds the transcript from session events, which drops non-durable notice components; applying it last would erase the just-appended visibility notice. @@ -18,7 +18,9 @@ The reasoning rebuild exposed a replay defect that this change fixes in `renderE ## Alternatives considered -**Cycle on bare `/details`, mirroring Ctrl+O.** Rejected: the command's value over the shortcut is naming an absolute state; a cycling command is the shortcut with more keystrokes, and bare invocation is more useful as a state report. +**Cycle on bare `/details`, mirroring Ctrl+O.** Rejected: the command's value over the shortcut is naming an absolute state; a cycling command is the shortcut with more keystrokes, and bare invocation is more useful as the selector, which shows the current state while offering every target. + +**Bare `/details` as a text-only state report.** Shipped first, replaced by the selector: the report answered "where am I" but still required a second, argument-spelling invocation to change anything, while the selector shows the same state and applies a change in one interaction. The textual grammar remains for scripts, muscle memory, and combined two-dimension changes. **Separate `/tools` and `/reasoning` commands.** Rejected: both dimensions are one presentation concern ("how much detail does the transcript show"), and a single command keeps the registry and `/help` list small while allowing one combined invocation. @@ -26,6 +28,7 @@ The reasoning rebuild exposed a replay defect that this change fixes in `renderE ## Consequences -- A user can jump to any detail mode, set both dimensions at once, and query the state — including on terminals that intercept Ctrl+O/Ctrl+R. +- A user can jump to any detail mode, set both dimensions at once, and see the current state in the selector — including on terminals that intercept Ctrl+O/Ctrl+R. - The parser accepts order-free tokens, so `/details reasoning expanded` toggles reasoning and expands cards; last directive wins per dimension. This leniency is deliberate and documented in the README. -- Transcript rebuilds no longer lose assistant messages when a step carries more than one `assistant/message` event; the `details-command` snapshot pins the command surface and the fixed replay together. +- The selector applies one dimension per confirm; a combined change still needs the argument form. Enter on the already-current row re-applies it idempotently and repeats its notice. +- Transcript rebuilds no longer lose assistant messages when a step carries more than one `assistant/message` event; the `details-command` snapshot pins the argument surface and the fixed replay, and `details-selector` pins the open selector with its current-state markers. diff --git a/.agents/notes/implemented/feature/2026-07-30-tui-details-command.zh.md b/.agents/notes/implemented/feature/2026-07-30-tui-details-command.zh.md index 48e9a8d3f1..e9c9787a49 100644 --- a/.agents/notes/implemented/feature/2026-07-30-tui-details-command.zh.md +++ b/.agents/notes/implemented/feature/2026-07-30-tui-details-command.zh.md @@ -10,7 +10,7 @@ TUI 的 transcript（文本记录）细节状态——工具卡片可见性（`c ## Decision -`dsh-tui` 在其他 agent 作用域命令旁注册 `/details`。裸 `/details` 用一条通知报告当前状态。参数直接命名目标状态：`collapsed|expanded|hidden` 让工具卡片跳到该阶段，`reasoning on|off` 设置 reasoning 显示，裸 `reasoning` 切换它，且指令可在一次调用中组合。未知 token 返回携带用法行的命令错误。命令改动的是与快捷键相同的闭包状态，重构后循环与切换成为 `setToolsVisibility`/`setReasoning` 之上的薄封装；快捷键及其通知保持不变。 +`dsh-tui` 在其他 agent 作用域命令旁注册 `/details`。裸 `/details` 打开 `DetailsDialog`：一个居中的键盘选择器，列出五个细节状态——三个工具卡片阶段与 reasoning 显示/隐藏——预选当前阶段并标记两个当前值，Enter 应用高亮状态并关闭，Esc 或 Ctrl+C 取消；其宽度由配置键 `detailsDialogWidth` 决定，选择器打开时再次执行 `/details` 会替换它，与 `/model` 浮层一致。参数直接命名目标状态：`collapsed|expanded|hidden` 让工具卡片跳到该阶段，`reasoning on|off` 设置 reasoning 显示，裸 `reasoning` 切换它，且指令可在一次调用中组合。未知 token 返回携带用法行的命令错误。每个入口改动的都是与快捷键相同的闭包状态，重构后循环与切换成为 `setToolsVisibility`/`setReasoning` 之上的薄封装；快捷键及其通知保持不变。 组合调用先应用 reasoning 再应用可见性，因为 `setReasoning` 会从会话事件重建 transcript，而重建会丢弃非持久的通知组件；若最后才应用它，会抹掉刚追加的可见性通知。 @@ -18,7 +18,9 @@ reasoning 重建暴露了一个重放缺陷，本变更在 `renderEvent` 中修 ## Alternatives considered -**裸 `/details` 像 Ctrl+O 一样循环。** 否决：命令相对快捷键的价值在于命名绝对状态；循环命令只是按键更多的快捷键，裸调用作为状态报告更有用。 +**裸 `/details` 像 Ctrl+O 一样循环。** 否决：命令相对快捷键的价值在于命名绝对状态；循环命令只是按键更多的快捷键，裸调用作为选择器更有用——它在展示当前状态的同时提供所有目标。 + +**裸 `/details` 仅输出文本状态报告。** 首版如此实现，后被选择器取代：报告回答了“我在哪”，但改变任何东西仍需第二次、拼写参数的调用；选择器展示同样的状态并在一次交互中应用变更。文本语法保留给脚本、肌肉记忆和两维组合变更。 **拆分 `/tools` 与 `/reasoning` 两个命令。** 否决：两个维度同属一个展示关注点（“transcript 显示多少细节”），单一命令让注册表与 `/help` 列表更小，同时允许一次组合调用。 @@ -26,6 +28,7 @@ reasoning 重建暴露了一个重放缺陷，本变更在 `renderEvent` 中修 ## Consequences -- 用户可以跳到任意细节模式、一次设置两个维度并查询状态——包括在拦截 Ctrl+O/Ctrl+R 的终端上。 +- 用户可以跳到任意细节模式、一次设置两个维度，并在选择器中看到当前状态——包括在拦截 Ctrl+O/Ctrl+R 的终端上。 - 解析器接受无序 token，因此 `/details reasoning expanded` 会切换 reasoning 并展开卡片；每个维度以最后一个指令为准。这一宽松是刻意的，并记录在 README 中。 -- 当一个步骤携带多条 `assistant/message` 事件时，transcript 重建不再丢失 assistant 消息；`details-command` 快照同时固定命令表面与修复后的重放。 +- 选择器每次确认只应用一个维度；组合变更仍需参数形式。在已是当前值的行上按 Enter 会幂等地重新应用并重复其通知。 +- 当一个步骤携带多条 `assistant/message` 事件时，transcript 重建不再丢失 assistant 消息；`details-command` 快照固定参数表面与修复后的重放，`details-selector` 固定带当前值标记的打开选择器。 diff --git a/docs/config-catalog.md b/docs/config-catalog.md index 2288d7ffbe..0f04e0a955 100644 --- a/docs/config-catalog.md +++ b/docs/config-catalog.md @@ -2034,6 +2034,8 @@ export interface TuiConfig { questionDialogMaxHeight?: number /** Model-selector width in terminal columns. */ modelDialogWidth?: number + /** Transcript-details selector width in terminal columns. */ + detailsDialogWidth?: number /** Model-selector maximum height in terminal rows. */ modelDialogMaxHeight?: number /** Maximum fuzzy file candidates displayed for one `@` query. */ @@ -2067,7 +2069,7 @@ export interface TuiThemeConfig { } ``` -Source: [`packages/ui/tui/src/config.ts:117`](../packages/ui/tui/src/config.ts) +Source: [`packages/ui/tui/src/config.ts:121`](../packages/ui/tui/src/config.ts) ## `@deepseek-ai/dsh-typert-loader` diff --git a/packages/ui/tui/src/components/dialogs.ts b/packages/ui/tui/src/components/dialogs.ts index bad6a5458b..f59ff747be 100644 --- a/packages/ui/tui/src/components/dialogs.ts +++ b/packages/ui/tui/src/components/dialogs.ts @@ -34,6 +34,7 @@ import type { import type { AskUserQuestionItem } from '@deepseek-ai/dsh-user-interaction' import { BRACKETED_PASTE_END, BRACKETED_PASTE_START, displayText, sanitizePastedText } from './text.ts' import { dialogSelectTheme, type Palette } from './theme.ts' +import type { ToolCardVisibility } from './transcript.ts' import { renderTuiPromptTemplate, type TuiPromptTemplateToken, @@ -432,6 +433,63 @@ export class ModelDialog implements Component { } } +/** One transcript-detail state the details selector applies on confirm. */ +export type DetailsSelection = + | { readonly kind: 'tools'; readonly visibility: ToolCardVisibility } + | { readonly kind: 'reasoning'; readonly show: boolean } + +/** + * Keyboard selector over the transcript detail states: the three tool-card + * visibility phases and reasoning-block display. Enter applies the highlighted + * state and closes; Esc or Ctrl+C closes without changing anything. + */ +export class DetailsDialog implements Component { + private readonly list: SelectList + + constructor( + visibility: ToolCardVisibility, + showReasoning: boolean, + private readonly palette: Palette, + done: (selection: DetailsSelection) => void, + private readonly cancel: () => void, + ) { + const current = (isCurrent: boolean): string => isCurrent ? ' — current' : '' + const items: SelectItem[] = [ + { value: 'collapsed', label: 'Tool cards · collapsed', description: `head/tail preview${current(visibility === 'collapsed')}` }, + { value: 'expanded', label: 'Tool cards · expanded', description: `full bodies${current(visibility === 'expanded')}` }, + { value: 'hidden', label: 'Tool cards · hidden', description: `conversation only${current(visibility === 'hidden')}` }, + { value: 'reasoning-shown', label: 'Reasoning · shown', description: `show reasoning blocks${current(showReasoning)}` }, + { value: 'reasoning-hidden', label: 'Reasoning · hidden', description: `omit reasoning blocks${current(!showReasoning)}` }, + ] + this.list = new SelectList(items, items.length, dialogSelectTheme(palette)) + this.list.setSelectedIndex(items.findIndex(item => item.value === visibility)) + this.list.onSelect = (item) => { + done(item.value === 'reasoning-shown' || item.value === 'reasoning-hidden' + ? { kind: 'reasoning', show: item.value === 'reasoning-shown' } + : { kind: 'tools', visibility: item.value as ToolCardVisibility }) + } + } + + invalidate(): void { + this.list.invalidate() + } + + handleInput(data: string): void { + if (matchesKey(data, Key.escape) || matchesKey(data, Key.ctrl('c'))) this.cancel() + else this.list.handleInput(data) + this.invalidate() + } + + render(width: number): string[] { + const innerWidth = Math.max(1, width - 4) + return renderDialog('Transcript details', [ + ...this.list.render(innerWidth), + '', + this.palette.dim('↑/↓ move • Enter apply • Esc cancel'), + ], width, this.palette) + } +} + /** The provider/model route recovered from a resume candidate's log. */ export interface ResumeRoute { provider: string diff --git a/packages/ui/tui/src/config.ts b/packages/ui/tui/src/config.ts index def548861f..59a12c0b96 100644 --- a/packages/ui/tui/src/config.ts +++ b/packages/ui/tui/src/config.ts @@ -46,6 +46,8 @@ export interface TuiConfig { questionDialogMaxHeight?: number /** Model-selector width in terminal columns. */ modelDialogWidth?: number + /** Transcript-details selector width in terminal columns. */ + detailsDialogWidth?: number /** Model-selector maximum height in terminal rows. */ modelDialogMaxHeight?: number /** Maximum fuzzy file candidates displayed for one `@` query. */ @@ -70,6 +72,7 @@ const maxResumeOptionsSchema = z.number().step(1).min(1).default(8) const questionDialogWidthSchema = z.number().step(1).min(20).default(200) const questionDialogMaxHeightSchema = z.number().step(1).min(6).default(20) const modelDialogWidthSchema = z.number().step(1).min(20).default(76) +const detailsDialogWidthSchema = z.number().step(1).min(20).default(72) const modelDialogMaxHeightSchema = z.number().step(1).min(6).default(20) const fileSearchMaxResultsSchema = z.number().step(1).min(1).default(DEFAULT_FILE_SEARCH_MAX_RESULTS) const fileSearchMaxEntriesSchema = z.number().step(1).min(1).default(DEFAULT_FILE_SEARCH_MAX_ENTRIES) @@ -102,6 +105,7 @@ const tuiConfigSchemaFields = { questionDialogMaxHeight: questionDialogMaxHeightSchema, modelDialogWidth: modelDialogWidthSchema, modelDialogMaxHeight: modelDialogMaxHeightSchema, + detailsDialogWidth: detailsDialogWidthSchema, fileSearchMaxResults: fileSearchMaxResultsSchema, fileSearchMaxEntries: fileSearchMaxEntriesSchema, fileSearchExcludedDirectories: fileSearchExcludedDirectoriesSchema, @@ -142,6 +146,7 @@ export const Config: z = z.object({ questionDialogMaxHeight: tuiConfigSchemaFields.questionDialogMaxHeight, modelDialogWidth: tuiConfigSchemaFields.modelDialogWidth, modelDialogMaxHeight: tuiConfigSchemaFields.modelDialogMaxHeight, + detailsDialogWidth: tuiConfigSchemaFields.detailsDialogWidth, fileSearchMaxResults: tuiConfigSchemaFields.fileSearchMaxResults, fileSearchMaxEntries: tuiConfigSchemaFields.fileSearchMaxEntries, fileSearchExcludedDirectories: tuiConfigSchemaFields.fileSearchExcludedDirectories, @@ -171,6 +176,7 @@ export interface ResolvedTuiConfig { questionDialogMaxHeight: number modelDialogWidth: number modelDialogMaxHeight: number + detailsDialogWidth: number fileSearchMaxResults: number fileSearchMaxEntries: number fileSearchExcludedDirectories: string[] @@ -196,6 +202,7 @@ export function resolveTuiConfig(config: TuiConfig | undefined): ResolvedTuiConf questionDialogMaxHeight: config?.questionDialogMaxHeight ?? 20, modelDialogWidth: config?.modelDialogWidth ?? 76, modelDialogMaxHeight: config?.modelDialogMaxHeight ?? 20, + detailsDialogWidth: config?.detailsDialogWidth ?? 72, fileSearchMaxResults: config?.fileSearchMaxResults ?? DEFAULT_FILE_SEARCH_MAX_RESULTS, fileSearchMaxEntries: config?.fileSearchMaxEntries ?? DEFAULT_FILE_SEARCH_MAX_ENTRIES, fileSearchExcludedDirectories: [...(config?.fileSearchExcludedDirectories ?? DEFAULT_FILE_SEARCH_EXCLUDED_DIRECTORIES)], diff --git a/packages/ui/tui/src/index.ts b/packages/ui/tui/src/index.ts index 6470e70114..7fd76a1a6f 100644 --- a/packages/ui/tui/src/index.ts +++ b/packages/ui/tui/src/index.ts @@ -104,6 +104,7 @@ import { } from './components/transcript.ts' import { compactTargetLabel, + DetailsDialog, diagnosticMeter, formatDiagnosticCount, formatDiagnosticNumber, @@ -112,6 +113,7 @@ import { StatusCardComponent, PromptContextComponent, targetLabel, + type DetailsSelection, type StatusCardRow, } from './components/dialogs.ts' import { @@ -1037,12 +1039,38 @@ export function createTuiChat( const toggleReasoning = (): void => { setReasoning(!showReasoning) } + // The selector and the argument grammar mutate the same closure state the + // Ctrl+O cycle and Ctrl+R toggle drive, so every entry converges. + let detailsOverlay: TuiOverlaySession | undefined + const showDetailsSelector = (): void => { + void detailsOverlay?.close() + const session = overlayManager.open({ + create: () => new DetailsDialog( + toolsVisibility, + showReasoning, + palette, + (selection: DetailsSelection) => { + void session.close() + if (selection.kind === 'reasoning') setReasoning(selection.show) + else setToolsVisibility(selection.visibility) + }, + () => { void session.close() }, + ), + options: { width: resolved.detailsDialogWidth, anchor: 'center', margin: 1 }, + }) + detailsOverlay = session + void session.closed.then(() => { + if (detailsOverlay === session) detailsOverlay = undefined + }) + requestRender() + } + // `/details` names the same transcript-detail state the Ctrl+O cycle and // Ctrl+R toggle mutate, so a user can jump to a mode without cycling. const runDetails = (rawInput: string): CommandResult => { const tokens = rawInput.split(/\s+/u).filter(token => token !== '') if (tokens.length === 0) { - appendNotice(`Tool and context cards ${toolsVisibility}; reasoning blocks ${showReasoning ? 'shown' : 'hidden'}.`) + showDetailsSelector() return { kind: 'success' } } let visibility: ToolCardVisibility | undefined @@ -1255,7 +1283,7 @@ export function createTuiChat( }) commandCtx.commands.register({ name: 'details', - description: 'Show or set tool-card visibility and reasoning display', + description: 'Select tool-card visibility and reasoning display', input: { hint: '[collapsed|expanded|hidden] [reasoning [on|off]]' }, handler: ({ rawInput }) => runDetails(rawInput), }) diff --git a/packages/ui/tui/tests/snapshots/details-command.expected.txt b/packages/ui/tui/tests/snapshots/details-command.expected.txt index 4e84695616..4d971c88df 100644 --- a/packages/ui/tui/tests/snapshots/details-command.expected.txt +++ b/packages/ui/tui/tests/snapshots/details-command.expected.txt @@ -1,7 +1,7 @@ terminal 100x40 buffer=normal length=40 base=0 viewport=0 lifecycle started=1 stopped=0 progress=inactive title "DSH snapshot" -cursor hidden column=7 viewportRow=19 bufferRow=19 +cursor hidden column=7 viewportRow=17 bufferRow=17 buffer 0| " DEEPSEEK HARNESS" style 1-8 fg=bright-magenta bold @@ -29,17 +29,14 @@ buffer 14| "Tool cards hidden. " style 0-17 dim 15| -16| "Tool and context cards hidden; reasoning blocks hidden. " - style 0-54 dim -17| -18| "/workspace/project (tui-staging) deepseek-v4-flash ↑0 ↓0 0% context" +16| "/workspace/project (tui-staging) deepseek-v4-flash ↑0 ↓0 0% context" style 0-17 fg=bright-magenta bold style 18-31 dim style 34-50 dim style 53-57 dim style 60-69 dim -19| " dsh > " +17| " dsh > " style 1-3 fg=bright-magenta bold style 5-6 dim style 7-7 inverse -20-39| +18-39| diff --git a/packages/ui/tui/tests/snapshots/details-selector.expected.txt b/packages/ui/tui/tests/snapshots/details-selector.expected.txt new file mode 100644 index 0000000000..ad6514d4e9 --- /dev/null +++ b/packages/ui/tui/tests/snapshots/details-selector.expected.txt @@ -0,0 +1,66 @@ +terminal 100x40 buffer=normal length=40 base=0 viewport=0 +lifecycle started=1 stopped=0 progress=inactive +title "DSH snapshot" +cursor hidden column=0 viewportRow=39 bufferRow=39 +buffer +0| " DEEPSEEK HARNESS" + style 1-8 fg=bright-magenta bold + style 10-16 bold +1| " Snapshot agent ready." + style 1-21 dim +2| " main-session" + style 1-12 dim +3| +4| "Assistant " + style 0-8 fg=bright-magenta bold underline +5| "Running the check now. " +6| "Model wait 0.0s " + style 0-14 dim +7| +8| "You " + style 0-2 fg=bright-magenta bold underline +9| "Inspect the renderer. " +10| "Model wait 0.0s · Completed 2026-07-30 18:00:00 " + style 0-46 dim +11| +12| "Reasoning blocks hidden. " + style 0-23 dim +13| +14| "Tool cards hidden. " + style 0-17 dim +15| " ╭ Transcript details ──────────────────────────────────────────────────╮ " + style 14-85 fg=bright-magenta +16| "/workspace/pro│ Tool cards · collapsed head/tail preview │ " + style 0-13 fg=bright-magenta bold + style 14-14 fg=bright-magenta + style 40-66 dim + style 85-85 fg=bright-magenta +17| " dsh > │ Tool cards · expanded full bodies │ " + style 1-3 fg=bright-magenta bold + style 5-6 dim + style 7-7 inverse + style 14-14 fg=bright-magenta + style 39-60 dim + style 85-85 fg=bright-magenta +18| " │ → Tool cards · hidden conversation only — current │ " + style 14-14 fg=bright-magenta + style 16-76 fg=bright-magenta inverse + style 85-85 fg=bright-magenta +19| " │ Reasoning · shown show reasoning blocks │ " + style 14-14 fg=bright-magenta + style 35-70 dim + style 85-85 fg=bright-magenta +20| " │ Reasoning · hidden omit reasoning blocks — current │ " + style 14-14 fg=bright-magenta + style 36-80 dim + style 85-85 fg=bright-magenta +21| " │ │ " + style 14-14 fg=bright-magenta + style 85-85 fg=bright-magenta +22| " │ ↑/↓ move • Enter apply • Esc cancel │ " + style 14-14 fg=bright-magenta + style 16-50 dim + style 85-85 fg=bright-magenta +23| " ╰──────────────────────────────────────────────────────────────────────╯ " + style 14-85 fg=bright-magenta +24-39| diff --git a/packages/ui/tui/tests/snapshots/disposed-terminal.expected.txt b/packages/ui/tui/tests/snapshots/disposed-terminal.expected.txt index aa41ca137d..1ade76a5aa 100644 --- a/packages/ui/tui/tests/snapshots/disposed-terminal.expected.txt +++ b/packages/ui/tui/tests/snapshots/disposed-terminal.expected.txt @@ -29,10 +29,10 @@ buffer 12| " " 13| "/clear — Clear the transcript view (session history is unchanged) " style 0-64 dim -14| "/details [collapsed|expanded|hidden] [reasoning [on|off]] — Show or set tool-card visibility" +14| "/details [collapsed|expanded|hidden] [reasoning [on|off]] — Select tool-card visibility and " style 0-91 dim -15| "and reasoning display " - style 0-20 dim +15| "reasoning display " + style 0-16 dim 16| "/exit — Exit after the active turn reaches idle " style 0-46 dim 17| "/help — Show keyboard shortcuts and commands " diff --git a/packages/ui/tui/tests/snapshots/errors-and-help.expected.txt b/packages/ui/tui/tests/snapshots/errors-and-help.expected.txt index 48c25cf8fa..7da75336e3 100644 --- a/packages/ui/tui/tests/snapshots/errors-and-help.expected.txt +++ b/packages/ui/tui/tests/snapshots/errors-and-help.expected.txt @@ -29,10 +29,10 @@ buffer 12| " " 13| "/clear — Clear the transcript view (session history is unchanged) " style 0-64 dim -14| "/details [collapsed|expanded|hidden] [reasoning [on|off]] — Show or set tool-card visibility" +14| "/details [collapsed|expanded|hidden] [reasoning [on|off]] — Select tool-card visibility and " style 0-91 dim -15| "and reasoning display " - style 0-20 dim +15| "reasoning display " + style 0-16 dim 16| "/exit — Exit after the active turn reaches idle " style 0-46 dim 17| "/help — Show keyboard shortcuts and commands " diff --git a/packages/ui/tui/tests/tui.snapshot.ts b/packages/ui/tui/tests/tui.snapshot.ts index abea27e69e..c0b6d44f88 100644 --- a/packages/ui/tui/tests/tui.snapshot.ts +++ b/packages/ui/tui/tests/tui.snapshot.ts @@ -46,6 +46,7 @@ const CHECKPOINTS = [ 'advanced-cards-expanded', 'tool-cards-hidden-folded', 'details-command', + 'details-selector', 'untrusted-controls', 'question-dialog', 'question-dialog-single-option', @@ -661,11 +662,14 @@ describe('TUI terminal-state snapshots', () => { harness.terminal.send('/details hidden reasoning off') harness.terminal.send('\r') }) + await checkpoint('details-command', harness.terminal, { includeScrollback: true }) + // Bare /details opens the selector, preselecting and marking the current + // hidden/reasoning-off state. await renderAfter(harness, () => { harness.terminal.send('/details') harness.terminal.send('\r') }) - await checkpoint('details-command', harness.terminal, { includeScrollback: true }) + await checkpoint('details-selector', harness.terminal, { includeScrollback: true }) nowSpy.mockRestore() await disposeSnapshot(harness) }) diff --git a/packages/ui/tui/tests/tui.spec.ts b/packages/ui/tui/tests/tui.spec.ts index c532c5a67d..093d8ad285 100644 --- a/packages/ui/tui/tests/tui.spec.ts +++ b/packages/ui/tui/tests/tui.spec.ts @@ -186,6 +186,7 @@ describe('TUI config', () => { questionDialogMaxHeight: 20, modelDialogWidth: 76, modelDialogMaxHeight: 20, + detailsDialogWidth: 72, fileSearchMaxResults: 20, fileSearchMaxEntries: 10_000, fileSearchExcludedDirectories: ['.git', 'node_modules'], @@ -210,6 +211,7 @@ describe('TUI config', () => { questionDialogMaxHeight: 14, modelDialogWidth: 64, modelDialogMaxHeight: 16, + detailsDialogWidth: 44, fileSearchMaxResults: 7, fileSearchMaxEntries: 123, fileSearchExcludedDirectories: ['.git', 'generated'], @@ -226,6 +228,7 @@ describe('TUI config', () => { questionDialogMaxHeight: 14, modelDialogWidth: 64, modelDialogMaxHeight: 16, + detailsDialogWidth: 44, fileSearchMaxResults: 7, fileSearchMaxEntries: 123, fileSearchExcludedDirectories: ['.git', 'generated'], @@ -2510,7 +2513,7 @@ describe('pi-tui chat lifecycle and transcript', () => { await dispose(result) }) - it('/details reports and sets card visibility and reasoning display', async () => { + it('/details sets card visibility and reasoning display from arguments', async () => { const result = await setup() const run = async (line: string): Promise => { result.terminal.send(line) @@ -2518,9 +2521,6 @@ describe('pi-tui chat lifecycle and transcript', () => { await tick() } - await run('/details') - expect(result.terminal.output).toContain('Tool and context cards collapsed; reasoning blocks shown.') - await run('/details hidden') expect(result.terminal.output).toContain('Tool cards hidden.') @@ -2531,11 +2531,12 @@ describe('pi-tui chat lifecycle and transcript', () => { await run('/details reasoning on') expect(result.terminal.output).toContain('Reasoning blocks shown.') - // Bare `reasoning` toggles: shown -> hidden, confirmed by the status line. + // Bare `reasoning` toggles: shown -> hidden. + const toggleOutput = result.terminal.output.length await run('/details reasoning') + expect(result.terminal.output.slice(toggleOutput)).toContain('Reasoning blocks hidden.') await run('/details collapsed') - await run('/details') - expect(result.terminal.output).toContain('Tool and context cards collapsed; reasoning blocks hidden.') + expect(result.terminal.output.slice(toggleOutput)).toContain('Tool and context cards collapsed.') await run('/details bogus') expect(result.terminal.output).toContain('Unknown /details argument "bogus"') @@ -2543,6 +2544,59 @@ describe('pi-tui chat lifecycle and transcript', () => { await dispose(result) }) + it('bare /details opens the transcript-details selector and applies the confirmed state', async () => { + const result = await setup() + const open = async (): Promise => { + const from = result.terminal.output.length + result.terminal.send('/details') + result.terminal.send('\r') + await vi.waitFor(() => { expect(result.terminal.output.slice(from)).toContain('Transcript details') }) + return from + } + + await open() + expect(result.terminal.output).toContain('Tool cards · collapsed') + expect(result.terminal.output).toContain('head/tail preview — current') + expect(result.terminal.output).toContain('show reasoning blocks — current') + + // A second /details while the selector is open replaces the overlay + // instead of stacking a second one behind it. + await result.ctx.commands.execute(result.agent, '/details', new AbortController().signal) + await tick() + + // Esc cancels without touching the state. + const cancelOutput = result.terminal.output.length + result.terminal.send('\x1b') + await tick() + expect(result.terminal.output.slice(cancelOutput)).not.toContain('Tool and context cards') + + // Enter on the next visibility row applies it and closes. + await open() + result.terminal.send('\x1b[B') + result.terminal.send('\r') + await tick() + expect(result.terminal.output).toContain('Tool and context cards expanded.') + + // The reopened selector preselects the current phase and marks it. + const reopened = await open() + expect(result.terminal.output.slice(reopened)).toContain('full bodies — current') + result.terminal.send('\x1b[B') + result.terminal.send('\x1b[B') + result.terminal.send('\x1b[B') + result.terminal.send('\r') + await tick() + expect(result.terminal.output).toContain('Reasoning blocks hidden.') + + // Ctrl+C also cancels. + const ctrlCOutput = result.terminal.output.length + await open() + result.terminal.send('\x03') + await tick() + expect(result.terminal.output.slice(ctrlCOutput)).not.toContain('Reasoning blocks shown.') + + await dispose(result) + }) + it('sends, steers, handles commands, global keys, and disposed-agent input', async () => { const result = await setup() From f13003df757125176692740cac16898251457e84 Mon Sep 17 00:00:00 2001 From: Turtle Date: Thu, 30 Jul 2026 23:45:50 +0800 Subject: [PATCH 06/48] feat(tui): two-entry /details selector with Tab cycling DetailsDialog now shows one entry per dimension (Tool cards, Reasoning) seeded with the current values; Tab cycles the highlighted entry's pending value (rendered as current -> pending), Enter applies every changed dimension in one confirm, Esc/Ctrl+C cancels. --- .../2026-07-30-tui-details-command.i18n.yaml | 4 +- .../feature/2026-07-30-tui-details-command.md | 6 +- .../2026-07-30-tui-details-command.zh.md | 6 +- packages/ui/tui/README.i18n.yaml | 4 +- packages/ui/tui/README.md | 6 +- packages/ui/tui/README.zh.md | 6 +- packages/ui/tui/src/components/dialogs.ts | 79 +++++++++++++------ packages/ui/tui/src/index.ts | 5 +- .../snapshots/details-selector.expected.txt | 58 ++++++-------- packages/ui/tui/tests/tui.snapshot.ts | 6 +- packages/ui/tui/tests/tui.spec.ts | 36 ++++++--- 11 files changed, 131 insertions(+), 85 deletions(-) diff --git a/.agents/notes/implemented/feature/2026-07-30-tui-details-command.i18n.yaml b/.agents/notes/implemented/feature/2026-07-30-tui-details-command.i18n.yaml index 33278b66d2..8e6c202a5e 100644 --- a/.agents/notes/implemented/feature/2026-07-30-tui-details-command.i18n.yaml +++ b/.agents/notes/implemented/feature/2026-07-30-tui-details-command.i18n.yaml @@ -2,5 +2,5 @@ # side as of the last confirmed-consistent state. Both languages carry equal authority; # after editing either side, bring the other along and re-record with: # pnpm run verify-translation-pairing --write .agents/notes/implemented/feature/2026-07-30-tui-details-command.md -2026-07-30-tui-details-command.md: c14461aa08790033026eead38f53d4b8e686bbbb -2026-07-30-tui-details-command.zh.md: e9c9787a4911f29f49e7709d1ba673dc1a25e8a1 +2026-07-30-tui-details-command.md: 5de80ceb3ad78a949268c31e9c1d1a50c956b90f +2026-07-30-tui-details-command.zh.md: f74d0658cb776752bf3da1682291de882638310a diff --git a/.agents/notes/implemented/feature/2026-07-30-tui-details-command.md b/.agents/notes/implemented/feature/2026-07-30-tui-details-command.md index c14461aa08..5de80ceb3a 100644 --- a/.agents/notes/implemented/feature/2026-07-30-tui-details-command.md +++ b/.agents/notes/implemented/feature/2026-07-30-tui-details-command.md @@ -10,7 +10,7 @@ The TUI's transcript detail state — tool-card visibility (`collapsed`/`expande ## Decision -`dsh-tui` registers `/details` beside its other agent-scoped commands. Bare `/details` opens `DetailsDialog`, a centered keyboard selector over the five detail states — the three tool-card phases and reasoning shown/hidden — that preselects the current phase, marks both current values, applies the highlighted state on Enter, and cancels on Esc or Ctrl+C; its width is the `detailsDialogWidth` config key and a second `/details` replaces an open selector, mirroring the `/model` overlay. Arguments name target states directly: `collapsed|expanded|hidden` jumps tool cards to that phase, `reasoning on|off` sets reasoning display, bare `reasoning` toggles it, and directives combine in one invocation. An unknown token returns a command error carrying the usage line. Every entry mutates the same closure state as the shortcuts, refactored so the cycle and toggle are thin wrappers over `setToolsVisibility`/`setReasoning`; the shortcuts and their notices are unchanged. +`dsh-tui` registers `/details` beside its other agent-scoped commands. Bare `/details` opens `DetailsDialog`, a centered keyboard selector with one entry per dimension — `Tool cards` and `Reasoning` — seeded with the current values: Tab cycles the highlighted entry's pending value (rendered as `current → pending`), Enter applies every changed dimension in one confirm and closes, and Esc or Ctrl+C cancels; its width is the `detailsDialogWidth` config key and a second `/details` replaces an open selector, mirroring the `/model` overlay. Arguments name target states directly: `collapsed|expanded|hidden` jumps tool cards to that phase, `reasoning on|off` sets reasoning display, bare `reasoning` toggles it, and directives combine in one invocation. An unknown token returns a command error carrying the usage line. Every entry mutates the same closure state as the shortcuts, refactored so the cycle and toggle are thin wrappers over `setToolsVisibility`/`setReasoning`; the shortcuts and their notices are unchanged. A combined invocation applies reasoning before visibility because `setReasoning` rebuilds the transcript from session events, which drops non-durable notice components; applying it last would erase the just-appended visibility notice. @@ -30,5 +30,5 @@ The reasoning rebuild exposed a replay defect that this change fixes in `renderE - A user can jump to any detail mode, set both dimensions at once, and see the current state in the selector — including on terminals that intercept Ctrl+O/Ctrl+R. - The parser accepts order-free tokens, so `/details reasoning expanded` toggles reasoning and expands cards; last directive wins per dimension. This leniency is deliberate and documented in the README. -- The selector applies one dimension per confirm; a combined change still needs the argument form. Enter on the already-current row re-applies it idempotently and repeats its notice. -- Transcript rebuilds no longer lose assistant messages when a step carries more than one `assistant/message` event; the `details-command` snapshot pins the argument surface and the fixed replay, and `details-selector` pins the open selector with its current-state markers. +- The selector applies only changed dimensions on confirm, so Enter without a Tab closes silently; a two-dimension change is one open-Tab-Enter interaction. +- Transcript rebuilds no longer lose assistant messages when a step carries more than one `assistant/message` event; the `details-command` snapshot pins the argument surface and the fixed replay, and `details-selector` pins the open selector with a Tab-cycled `hidden → collapsed` pending value. diff --git a/.agents/notes/implemented/feature/2026-07-30-tui-details-command.zh.md b/.agents/notes/implemented/feature/2026-07-30-tui-details-command.zh.md index e9c9787a49..f74d0658cb 100644 --- a/.agents/notes/implemented/feature/2026-07-30-tui-details-command.zh.md +++ b/.agents/notes/implemented/feature/2026-07-30-tui-details-command.zh.md @@ -10,7 +10,7 @@ TUI 的 transcript（文本记录）细节状态——工具卡片可见性（`c ## Decision -`dsh-tui` 在其他 agent 作用域命令旁注册 `/details`。裸 `/details` 打开 `DetailsDialog`：一个居中的键盘选择器，列出五个细节状态——三个工具卡片阶段与 reasoning 显示/隐藏——预选当前阶段并标记两个当前值，Enter 应用高亮状态并关闭，Esc 或 Ctrl+C 取消；其宽度由配置键 `detailsDialogWidth` 决定，选择器打开时再次执行 `/details` 会替换它，与 `/model` 浮层一致。参数直接命名目标状态：`collapsed|expanded|hidden` 让工具卡片跳到该阶段，`reasoning on|off` 设置 reasoning 显示，裸 `reasoning` 切换它，且指令可在一次调用中组合。未知 token 返回携带用法行的命令错误。每个入口改动的都是与快捷键相同的闭包状态，重构后循环与切换成为 `setToolsVisibility`/`setReasoning` 之上的薄封装；快捷键及其通知保持不变。 +`dsh-tui` 在其他 agent 作用域命令旁注册 `/details`。裸 `/details` 打开 `DetailsDialog`：一个居中的键盘选择器，每个维度一个条目——`Tool cards` 与 `Reasoning`——以当前值为初始：Tab 循环高亮条目的待定值（渲染为 `current → pending`），Enter 一次确认应用所有已改变的维度并关闭，Esc 或 Ctrl+C 取消；其宽度由配置键 `detailsDialogWidth` 决定，选择器打开时再次执行 `/details` 会替换它，与 `/model` 浮层一致。参数直接命名目标状态：`collapsed|expanded|hidden` 让工具卡片跳到该阶段，`reasoning on|off` 设置 reasoning 显示，裸 `reasoning` 切换它，且指令可在一次调用中组合。未知 token 返回携带用法行的命令错误。每个入口改动的都是与快捷键相同的闭包状态，重构后循环与切换成为 `setToolsVisibility`/`setReasoning` 之上的薄封装；快捷键及其通知保持不变。 组合调用先应用 reasoning 再应用可见性，因为 `setReasoning` 会从会话事件重建 transcript，而重建会丢弃非持久的通知组件；若最后才应用它，会抹掉刚追加的可见性通知。 @@ -30,5 +30,5 @@ reasoning 重建暴露了一个重放缺陷，本变更在 `renderEvent` 中修 - 用户可以跳到任意细节模式、一次设置两个维度，并在选择器中看到当前状态——包括在拦截 Ctrl+O/Ctrl+R 的终端上。 - 解析器接受无序 token，因此 `/details reasoning expanded` 会切换 reasoning 并展开卡片；每个维度以最后一个指令为准。这一宽松是刻意的，并记录在 README 中。 -- 选择器每次确认只应用一个维度；组合变更仍需参数形式。在已是当前值的行上按 Enter 会幂等地重新应用并重复其通知。 -- 当一个步骤携带多条 `assistant/message` 事件时，transcript 重建不再丢失 assistant 消息；`details-command` 快照固定参数表面与修复后的重放，`details-selector` 固定带当前值标记的打开选择器。 +- 选择器确认时只应用已改变的维度，因此未按 Tab 直接 Enter 会静默关闭；两个维度的变更是一次打开-Tab-Enter 交互。 +- 当一个步骤携带多条 `assistant/message` 事件时，transcript 重建不再丢失 assistant 消息；`details-command` 快照固定参数表面与修复后的重放，`details-selector` 固定经 Tab 循环出 `hidden → collapsed` 待定值的打开选择器。 diff --git a/packages/ui/tui/README.i18n.yaml b/packages/ui/tui/README.i18n.yaml index 6587bc7624..af9ba4a67a 100644 --- a/packages/ui/tui/README.i18n.yaml +++ b/packages/ui/tui/README.i18n.yaml @@ -2,5 +2,5 @@ # side as of the last confirmed-consistent state. Both languages carry equal authority; # after editing either side, bring the other along and re-record with: # pnpm run verify-translation-pairing --write packages/ui/tui/README.md -README.md: a65d72b3b80b992fabcb33d4b4345942b58e9147 -README.zh.md: fd056da59335af5cb9fc63fd2fb681266fadeb62 +README.md: 7230e75535ce5710e12040e026bc77a1782ed6dd +README.zh.md: 3079b797b670de64cfdc1faa9d002ee04a6e28dd diff --git a/packages/ui/tui/README.md b/packages/ui/tui/README.md index 23ca8343ec..8ef5f20ca0 100644 --- a/packages/ui/tui/README.md +++ b/packages/ui/tui/README.md @@ -12,7 +12,7 @@ This package owns interactive terminal presentation and input only. It injects ` After terminal startup succeeds, the package provides the terminal-local `ctx.tui` extension service. A plugin that injects it can call `openOverlay()` with a component factory and constrained layout options; the host exposes the viewport, semantic theme, display-text escaping, redraw, close, and a lifetime signal, but not the pi-tui tree, terminal, focus controller, or overlay handle. Plugin overlays, the model selector, and user questions share one FIFO modal queue. Each request is an effect of the calling plugin fiber, so unload removes queued work or closes visible work before cleanup settles; terminal shutdown unloads dependents before stopping pi-tui. Overlay state is not logged or replayed. Component code is trusted and may render ANSI styling, but must pass untrusted text through `host.display()`. The [interactive-extension Agent Note](../../../.agents/notes/implemented/architecture/2026-07-22-tui-interactive-extension-service.md) owns the boundary and rejected alternatives. -The TUI rebuilds resumed history from the append-origin session events, renders Markdown responses and reasoning, applies each tool's `presentCall` / `presentResult` intent to terminal, diff, or generic cards, keeps the standing `todo/write` plan above the editor (cleared on the next `turn/start`), and presents `ctx.userInteraction` questions in a wide bottom-left keyboard panel with progress, numbered options, and aligned descriptions. The latest logged session title becomes the header subtitle, with `welcome` before a title exists, and the terminal window title becomes ` — `. A durable `llm/retry` event retracts the failed step's live chunks and renders the scheduled retry count, delay, and failure in the transcript; success, exhaustion, and cancellation then settle through ordinary session events. The footer totals each logged model step's usage once, including failed attempts, while treating committed-message usage as a fallback for logs without a usage chunk. Its idle view compares token-meter pressure with `ctx.llm.resolveModelInfo()` context for the current route, displays `context unknown` when the adapter has no capacity metadata, and also shows tool-card mode plus the current model and any explicitly selected reasoning effort; while the agent runs, an elapsed working indicator and `esc interrupt` replace that summary. A surface replacement never rewrites the rendered transcript: the conversation it shadows stays readable, and a landed compaction checkpoint adds one dim `… earlier context was compacted …` marker at its log position, so the terminal reports where the model stopped seeing that history instead of erasing it. Model-only replacement copies — a pruned tool result, a regenerated assistant message — render nothing. +The TUI rebuilds resumed history from the active session surface, renders Markdown responses and reasoning, applies each tool's `presentCall` / `presentResult` intent to terminal, diff, or generic cards, keeps the standing `todo/write` plan above the editor (cleared on the next `turn/start`), and presents `ctx.userInteraction` questions in a wide bottom-left keyboard panel with progress, numbered options, and aligned descriptions. The latest logged session title becomes the header subtitle, with `welcome` before a title exists, and the terminal window title becomes ` — `. A durable `llm/retry` event retracts the failed step's live chunks and renders the scheduled retry count, delay, and failure in the transcript; success, exhaustion, and cancellation then settle through ordinary session events. The footer totals each logged model step's usage once, including failed attempts, while treating committed-message usage as a fallback for logs without a usage chunk. Its idle view compares token-meter pressure with `ctx.llm.resolveModelInfo()` context for the current route, displays `context unknown` when the adapter has no capacity metadata, and also shows tool-card mode plus the current model and any explicitly selected reasoning effort; while the agent runs, an elapsed working indicator and `esc interrupt` replace that summary. Surface replacement events rebuild the transcript so compacted history does not reappear. An embedding may provide `TuiRuntime.formatCwd` when its logical workspace label differs from the session's host directory. The override changes only the footer label; tools continue to use the session `cwd`. @@ -22,7 +22,7 @@ Typing `@` at a token boundary searches files and directories under the session When optional `ctx.sessionReferences` is mounted, the same `@` menu also offers metadata-only session candidates, inserts `@[label](dsh-session:)`, and prepares the selected snapshots before dispatch. Session references remain structured because the model has no filesystem-like tool for retrieving session snapshots later. Preparation disables duplicate submission and restores the editor input on failure. The TUI chooses `agent.steer()` or `agent.followup()` from the status after that asynchronous preparation, so idle follow-ups still dispatch `agent/prompt-submit` while in-turn steering joins at a checkpoint without that hook. -While the agent is running, ordinary editor submissions call `agent.steer()`; otherwise they call `agent.followup()`. A slash at the start of the submitted line enters `ctx.commands` instead: known commands execute directly, unknown commands produce a warning, and neither path automatically reaches the model. A command producer may explicitly schedule agent work; [`dsh-plan-mode`](../../plan/plan-mode/README.md#model-and-human-surfaces) uses that contract for `/plan [message]`. The TUI registers `/help`, `/model`, `/clear`, `/details`, `/palette`, `/reload`, `/resume`, `/status`, and `/exit` as agent-scoped definitions; every other effective command joins autocomplete and `/help` dynamically, as do `/skill:` completions. A status line above the editor reports the turn phase the TUI derives from session events — waiting for the first token, thinking, responding, or executing tools — with the elapsed time in that phase and the running step total, refreshed each second, and ends with the `Enter sends steering, Esc cancels` hint; while steering messages wait to reach the model it inserts a `N queued ·` badge before the hint that clears as each drains. Ctrl+C or Escape cancels a running turn. Tool and injected-context cards collapse long bodies into a configurable head/tail preview; Ctrl+O cycles tool cards through collapsed preview, full output, and hidden — the hidden phase drops tool cards from the transcript entirely while context cards stay at their preview, since injected instructions are not tool traffic. The hidden phase also folds each turn's assistant steps into one message: the first step with visible text or reasoning keeps the turn's single `Assistant` header, later steps render as headerless continuations, and a step without a visible body renders nothing; leaving the hidden phase restores the per-step headers. An injected-context card renders its message as prose with the producer's outer reminder frame stripped, so neither the fold nor the frame stripping depends on the payload's syntax. Ctrl+R toggles reasoning, Ctrl+L redraws, and Ctrl+D exits while idle. `/details` names the same state those two shortcuts cycle: bare it opens a centered keyboard toggle with one entry per dimension — `Tool cards` and `Reasoning` — showing the live values, where Tab cycles the highlighted entry and applies the change immediately (the transcript behind the dialog is the preview), and Enter, Esc, or Ctrl+C closes; `/details collapsed|expanded|hidden` jumps tool cards to that phase directly, and `/details reasoning [on|off]` sets — or bare `reasoning` toggles — reasoning-block display; arguments combine in one invocation, an unknown argument fails with the usage line, and a combined invocation applies reasoning first so its transcript rebuild never drops the card notice. +While the agent is running, ordinary editor submissions call `agent.steer()`; otherwise they call `agent.followup()`. A slash at the start of the submitted line enters `ctx.commands` instead: known commands execute directly, unknown commands produce a warning, and neither path automatically reaches the model. A command producer may explicitly schedule agent work; [`dsh-plan-mode`](../../plan/plan-mode/README.md#model-and-human-surfaces) uses that contract for `/plan [message]`. The TUI registers `/help`, `/model`, `/clear`, `/details`, `/palette`, `/reload`, `/resume`, `/status`, and `/exit` as agent-scoped definitions; every other effective command joins autocomplete and `/help` dynamically, as do `/skill:` completions. A status line above the editor reports the turn phase the TUI derives from session events — waiting for the first token, thinking, responding, or executing tools — with the elapsed time in that phase and the running step total, refreshed each second, and ends with the `Enter sends steering, Esc cancels` hint; while steering messages wait to reach the model it inserts a `N queued ·` badge before the hint that clears as each drains. Ctrl+C or Escape cancels a running turn. Tool and injected-context cards collapse long bodies into a configurable head/tail preview; Ctrl+O cycles tool cards through collapsed preview, full output, and hidden — the hidden phase drops tool cards from the transcript entirely while context cards stay at their preview, since injected instructions are not tool traffic. The hidden phase also folds each turn's assistant steps into one message: the first step with visible text or reasoning keeps the turn's single `Assistant` header, later steps render as headerless continuations, and a step without a visible body renders nothing; leaving the hidden phase restores the per-step headers. An injected-context card renders its message as prose with the producer's outer reminder frame stripped, so neither the fold nor the frame stripping depends on the payload's syntax. Ctrl+R toggles reasoning, Ctrl+L redraws, and Ctrl+D exits while idle. `/details` names the same state those two shortcuts cycle: bare it opens a centered keyboard selector with one entry per dimension — `Tool cards` and `Reasoning` — seeded with the current values, where Tab cycles the highlighted entry's pending value (rendered as `current → pending`), Enter applies every changed dimension and closes, and Esc or Ctrl+C cancels without changing anything; `/details collapsed|expanded|hidden` jumps tool cards to that phase directly, and `/details reasoning [on|off]` sets — or bare `reasoning` toggles — reasoning-block display; arguments combine in one invocation, an unknown argument fails with the usage line, and a combined invocation applies reasoning first so its transcript rebuild never drops the card notice. `/model` opens the advisory `ctx.llm` catalog as a keyboard selector: a filter box above the list narrows rows by a case-insensitive substring over each row's `provider/model` label, model name, and description, keeping the highlighted row selected when it survives the filter; Up/Down moves, Shift+Tab cycles the focused model's adapter-advertised reasoning efforts in display order, Enter selects the model and effort, and Escape clears a non-empty filter before a second Escape closes it. When an adapter does not advertise a default effort, the cycle also includes `Default`, which clears an explicit selection and preserves the provider default; models without selectable effort metadata ignore Shift+Tab. The selector renders the exact advertised effort list—including `off` when present—and does not synthesize, clamp, or transfer an effort between models. `/model ` still selects an unambiguous model id directly, while `/model /` selects an exact target and uses its adapter default when one exists. The configured target or latest logged request header initializes the selector, and an unlisted current model remains visible because catalogs are advisory. Selection is local to this TUI session. Prompt assembly snapshots the target for one step, replaces `{{provider}}` and `{{model}}`, and applies the same provider/model/reasoning-effort target through `agent/request`; a switch during assembly therefore starts with a later step. The request header durably records targets that reach the model, while an unused selection remains process-local. @@ -32,6 +32,7 @@ The footer sums the session's reported usage as `↑ ↓ `/status` adds a point-in-time diagnostics card to the transcript and remains available while the agent runs. It reports the session id, title, working directory, selected provider/model, selected reasoning effort or default behavior, reasoning-block visibility, agent state, event/turn/step/tool-call counts, exact input/output/cache token buckets, KV-cache hit rate, token-meter context use and capacity, creation time, and latest event time. Missing titles, models, cache input, or context capacity are labeled instead of inferred. The card is terminal-only and does not duplicate the compact footer. + `/resume` opens a full-viewport keyboard selector instead of a centered dialog. Two scopes cover the same candidate set: the current workspace, which it opens on, and all workspaces, which Tab toggles to. The scope line under the search field names the active scope and the count the other holds, and each row in the all-workspaces scope also reports its own workspace. Toggling clears the search and selection so the highlighted row always belongs to the visible list. Its focused search field starts immediately after the search glyph and emits pi-tui's cursor marker, so terminal IME composition remains anchored inside the field. Candidates are sorted by last logged activity and searchable by log-backed title or session id, and by workspace label in the all-workspaces scope; each row reports current/live/persisted state, last turn outcome, recent provider/model, and durable goal phase when present. Up/Down and Page Up/Page Down navigate, Enter resumes, Escape clears a non-empty search before a second Escape cancels, and Ctrl+C cancels directly. The current session, a session already live in this runtime, an unreadable log, a session with no recorded workspace to run in, or a session whose logged provider has no current adapter remains visible but disabled; a workspace other than the current one is a scope rather than a disabled reason, because resume enters that directory. @@ -57,6 +58,7 @@ A launcher can seed a fresh session's first turn by providing `INITIAL_SKILL_KEY | `questionDialogMaxHeight` | `20` | Question-panel maximum rows | | `modelDialogWidth` | `76` | Model-selector width in columns | | `modelDialogMaxHeight` | `20` | Model-selector maximum rows | +| `detailsDialogWidth` | `72` | Transcript-details selector width in columns | | `fileSearchMaxResults` | `20` | Maximum file and directory candidates shown for one `@` query | | `fileSearchMaxEntries` | `10000` | Maximum paths retained in the bounded workspace index used by bare fuzzy queries | | `fileSearchExcludedDirectories` | `['.git', 'node_modules']` | Directory basenames omitted from traversal and direct completion | diff --git a/packages/ui/tui/README.zh.md b/packages/ui/tui/README.zh.md index 40aa849838..48cb25fb94 100644 --- a/packages/ui/tui/README.zh.md +++ b/packages/ui/tui/README.zh.md @@ -12,7 +12,7 @@ DeepSeek Harness agent（智能体）的交互式终端入口，基于 [`@earend 终端成功启动后，本包会提供终端本地的 `ctx.tui` 扩展服务。注入该服务的插件可以使用组件工厂和受限布局选项调用 `openOverlay()`；宿主会公开 viewport、语义化主题、显示文本转义、重绘、关闭和生命周期信号，但不公开 pi-tui 树、终端、焦点控制器或 overlay 句柄。插件 overlay、模型选择器和用户问题共用一个 FIFO 模态队列。每个请求都是调用方插件 fiber 的 effect，因此卸载会移除排队工作，或在清理结算前关闭可见工作；终端关闭会先卸载依赖项，再停止 pi-tui。Overlay 状态不会记录或回放。组件代码受信任，可以渲染 ANSI 样式，但必须通过 `host.display()` 处理不受信任文本。[交互式扩展 Agent Note](../../../.agents/notes/implemented/architecture/2026-07-22-tui-interactive-extension-service.md)持有该边界和未采用的替代方案。 -TUI 从追加来源的会话事件重建已恢复历史，渲染 Markdown 响应与 reasoning，将每个工具的 `presentCall` / `presentResult` 意图应用到终端、diff 或通用卡片，把站立的 `todo/write` 计划保留在编辑器上方（下一个 `turn/start` 时清空），并在左下方宽键盘面板中展示 `ctx.userInteraction` 问题，包含进度、编号选项和对齐说明。最新记录的会话标题成为 header 副标题；标题不存在时使用 `welcome`，终端窗口标题则变为 ` — `。持久 `llm/retry` 事件会撤回失败步骤的实时 chunk，并在 transcript（文本记录）中渲染计划重试次数、延迟和失败；成功、耗尽与取消随后通过普通会话事件结算。Footer 会对每个已记录模型步骤的用量只计一次，包括失败尝试；对于没有用量 chunk 的日志，以已提交消息的用量回退。其空闲视图会将 token-meter 压力与 `ctx.llm.resolveModelInfo()` 为当前路由返回的上下文容量进行比较；适配器没有容量元数据时显示 `context unknown`，并显示工具卡片模式、当前模型，以及任何显式选择的推理强度。Agent 运行时，这些摘要会替换为已经过工作时间指示器和 `esc interrupt`。表层替换从不重写已渲染的 transcript：被它遮蔽的对话仍可阅读，而已落地的压缩（compaction）检查点会在其日志位置添加一行暗色 `… earlier context was compacted …` 标记，因此终端报告的是模型从何处起不再看到那段历史，而不是把它抹掉。仅供模型使用的替换副本——被裁剪的工具结果、重新生成的 assistant 消息——不渲染任何内容。 +TUI 从活跃会话表层重建已恢复历史，渲染 Markdown 响应与 reasoning，将每个工具的 `presentCall` / `presentResult` 意图应用到终端、diff 或通用卡片，把站立的 `todo/write` 计划保留在编辑器上方（下一个 `turn/start` 时清空），并在左下方宽键盘面板中展示 `ctx.userInteraction` 问题，包含进度、编号选项和对齐说明。最新记录的会话标题成为 header 副标题；标题不存在时使用 `welcome`，终端窗口标题则变为 ` — `。持久 `llm/retry` 事件会撤回失败步骤的实时 chunk，并在 transcript（文本记录）中渲染计划重试次数、延迟和失败；成功、耗尽与取消随后通过普通会话事件结算。Footer 会对每个已记录模型步骤的用量只计一次，包括失败尝试；对于没有用量 chunk 的日志，以已提交消息的用量回退。其空闲视图会将 token-meter 压力与 `ctx.llm.resolveModelInfo()` 为当前路由返回的上下文容量进行比较；适配器没有容量元数据时显示 `context unknown`，并显示工具卡片模式、当前模型，以及任何显式选择的推理强度。Agent 运行时，这些摘要会替换为已经过工作时间指示器和 `esc interrupt`。表层替换事件会重建 transcript，使经过压缩（compaction）的历史不会再次出现。 如果逻辑工作区标签与会话宿主目录不同，嵌入方可以提供 `TuiRuntime.formatCwd`。该覆盖只改变 footer 标签；工具仍使用会话 `cwd`。 @@ -22,7 +22,7 @@ TUI 从追加来源的会话事件重建已恢复历史，渲染 Markdown 响应 挂载可选的 `ctx.sessionReferences` 后，同一个 `@` 菜单还会提供仅含元数据的会话候选项，插入 `@[label](dsh-session:)`，并在分派前准备所选快照。会话引用保持结构化，因为模型没有类似文件系统的工具可在稍后检索会话快照。准备期间会禁止重复提交，并在失败时恢复编辑器输入。TUI 会在异步准备后根据状态选择 `agent.steer()` 或 `agent.followup()`，因此空闲 followup 仍会分派 `agent/prompt-submit`，而轮次中的 steering 会在检查点加入且不触发该 hook。 -Agent 运行时，普通编辑器提交会调用 `agent.steer()`；其他时候调用 `agent.followup()`。提交行以斜杠开头时会改为进入 `ctx.commands`：已知命令直接执行，未知命令产生警告，两条路径都不会自动到达模型。命令生产方可以显式调度 agent 工作；[`dsh-plan-mode`](../../plan/plan-mode/README.md#model-and-human-surfaces) 使用该契约实现 `/plan [message]`。TUI 将 `/help`、`/model`、`/clear`、`/details`、`/palette`、`/reload`、`/resume`、`/status` 和 `/exit` 注册为 agent 作用域定义；其他所有有效命令都会动态加入自动补全与 `/help`，`/skill:` 补全也相同。编辑器上方的状态行会报告 TUI 从会话事件派生的轮次阶段，包括等待首个 token、思考、响应或执行工具；它显示该阶段已经过时间和运行中的步骤总数，每秒刷新，并以 `Enter sends steering, Esc cancels` 提示结尾。Steering 消息等待到达模型期间，会在提示前插入 `N queued ·` 徽标，每条消息排空后随即清除。Ctrl+C 或 Escape 会取消运行中的轮次。工具卡片与注入上下文卡片都把长主体折叠为可配置的头尾预览；Ctrl+O 让工具卡片在折叠预览、完整输出、隐藏三种状态间循环——隐藏阶段把工具卡片从 transcript 中完全去掉，而上下文卡片保持预览，因为注入的指令不属于工具流量。隐藏阶段还会把每个轮次的 assistant 步骤折叠为一条消息：第一个有可见文本或 reasoning 的步骤保留该轮次唯一的 `Assistant` 标题，之后的步骤渲染为无标题的续段，没有可见正文的步骤则不渲染任何内容；离开隐藏阶段会恢复每步各自的标题。注入上下文卡片把消息渲染为文本，并去掉生产方的外层提醒外框，因此折叠与去外框都不依赖载荷的语法。Ctrl+R 切换 reasoning，Ctrl+L 重绘，Ctrl+D 在空闲时退出。`/details` 命名的正是这两个快捷键循环的同一份状态：不带参数时打开一个居中的键盘开关，每个维度一个条目——`Tool cards` 与 `Reasoning`——显示实时值，Tab 循环高亮条目并立即应用变更（对话框背后的 transcript 即是预览），Enter、Esc 或 Ctrl+C 关闭；`/details collapsed|expanded|hidden` 让工具卡片直接跳到该阶段，`/details reasoning [on|off]` 设置——或裸 `reasoning` 切换——reasoning 块显示；参数可在一次调用中组合，未知参数会以用法行报错，组合调用先应用 reasoning，使其 transcript 重建不会丢掉卡片通知。 +Agent 运行时，普通编辑器提交会调用 `agent.steer()`；其他时候调用 `agent.followup()`。提交行以斜杠开头时会改为进入 `ctx.commands`：已知命令直接执行，未知命令产生警告，两条路径都不会自动到达模型。命令生产方可以显式调度 agent 工作；[`dsh-plan-mode`](../../plan/plan-mode/README.md#model-and-human-surfaces) 使用该契约实现 `/plan [message]`。TUI 将 `/help`、`/model`、`/clear`、`/details`、`/palette`、`/reload`、`/resume`、`/status` 和 `/exit` 注册为 agent 作用域定义；其他所有有效命令都会动态加入自动补全与 `/help`，`/skill:` 补全也相同。编辑器上方的状态行会报告 TUI 从会话事件派生的轮次阶段，包括等待首个 token、思考、响应或执行工具；它显示该阶段已经过时间和运行中的步骤总数，每秒刷新，并以 `Enter sends steering, Esc cancels` 提示结尾。Steering 消息等待到达模型期间，会在提示前插入 `N queued ·` 徽标，每条消息排空后随即清除。Ctrl+C 或 Escape 会取消运行中的轮次。工具卡片与注入上下文卡片都把长主体折叠为可配置的头尾预览；Ctrl+O 让工具卡片在折叠预览、完整输出、隐藏三种状态间循环——隐藏阶段把工具卡片从 transcript 中完全去掉，而上下文卡片保持预览，因为注入的指令不属于工具流量。隐藏阶段还会把每个轮次的 assistant 步骤折叠为一条消息：第一个有可见文本或 reasoning 的步骤保留该轮次唯一的 `Assistant` 标题，之后的步骤渲染为无标题的续段，没有可见正文的步骤则不渲染任何内容；离开隐藏阶段会恢复每步各自的标题。注入上下文卡片把消息渲染为文本，并去掉生产方的外层提醒外框，因此折叠与去外框都不依赖载荷的语法。Ctrl+R 切换 reasoning，Ctrl+L 重绘，Ctrl+D 在空闲时退出。`/details` 命名的正是这两个快捷键循环的同一份状态：不带参数时打开一个居中的键盘选择器，每个维度一个条目——`Tool cards` 与 `Reasoning`——以当前值为初始，Tab 循环高亮条目的待定值（渲染为 `current → pending`），Enter 应用所有已改变的维度并关闭，Esc 或 Ctrl+C 取消且不改变任何东西；`/details collapsed|expanded|hidden` 让工具卡片直接跳到该阶段，`/details reasoning [on|off]` 设置——或裸 `reasoning` 切换——reasoning 块显示；参数可在一次调用中组合，未知参数会以用法行报错，组合调用先应用 reasoning，使其 transcript 重建不会丢掉卡片通知。 `/model` 将建议性的 `ctx.llm` catalog 打开为键盘选择器：列表上方设有一个过滤框，按对每行 `provider/model` 标签、模型名称和描述的大小写不敏感子串匹配来缩小行集，并在高亮行仍通过过滤时保持其选中状态；Up/Down 移动，Shift+Tab 按显示顺序循环切换适配器为焦点模型公布的推理强度，Enter 选择模型和推理强度，Escape 会先清除非空过滤内容，再次按下才关闭选择器。适配器未公布默认推理强度时，循环还会包含 `Default`，该项会清除显式选择并保留提供方默认行为；没有可选推理强度元数据的模型会忽略 Shift+Tab。选择器会原样呈现公布的推理强度列表（包括存在时的 `off`），不会合成、自动调整或在模型之间转移推理强度。`/model ` 仍可直接选择无歧义的模型 id，`/model /` 则选择精确目标，并在存在时使用其适配器默认值。已配置目标或最新记录的请求 header 会初始化选择器；由于 catalog 仅提供建议，未列出的当前模型仍会显示。选择仅对本 TUI 会话有效。提示词组装会为一个步骤建立目标快照，替换 `{{provider}}` 和 `{{model}}`，并通过 `agent/request` 应用同一个提供方／模型／推理强度目标；因此组装期间的切换会从后续步骤开始生效。请求 header 会持久记录真正到达模型的目标，未使用的选择则只存在于进程本地。 @@ -32,6 +32,7 @@ Footer 将会话报告的用量汇总为 `↑ ↓`；任 `/status` 会向 transcript 添加一张时间点诊断卡片，并在 agent 运行时保持可用。它报告会话 id、标题、工作目录、所选提供方／模型、所选推理强度或默认行为、reasoning 块可见性、agent 状态、事件／轮次／步骤／工具调用计数、精确输入／输出／缓存 token bucket、KV-cache 命中率、token-meter 上下文用量与容量、创建时间和最新事件时间。缺失标题、模型、缓存输入或上下文容量时会明确标记，而非推断。该卡片只存在于终端，不会重复紧凑 footer。 + `/resume` 会打开全 viewport 键盘选择器，而非居中对话框。两个作用域覆盖同一候选项集合：打开时所处的当前工作区，以及按 Tab 切换到的所有工作区。搜索字段下方的作用域行会给出当前作用域的名称以及另一个作用域包含的数量，且在所有工作区作用域中每行还会报告自身所属的工作区。切换会清除搜索与选择，使高亮行始终属于可见列表。 获得焦点的搜索字段紧跟搜索 glyph 开始，并发出 pi-tui 的 cursor marker，使终端 IME 组合保持锚定在字段内。候选项按最近记录的活动排序，可按日志支持的标题或会话 id 搜索，在所有工作区作用域中还可按工作区标签搜索；每行报告 current/live/persisted 状态、上一轮次结果、近期提供方／模型，以及存在时的持久目标阶段。Up/Down 与 Page Up/Page Down 导航，Enter 恢复，Escape 会先清除非空搜索，再次按下才取消，Ctrl+C 则直接取消。当前会话、已在本运行时中活跃的会话、不可读日志、没有可运行的已记录工作区的会话，或日志所记提供方没有当前适配器的会话仍会显示，但不可选择；不同于当前工作区的工作区属于作用域而非禁用原因，因为恢复会进入该目录。 @@ -57,6 +58,7 @@ Footer 将会话报告的用量汇总为 `↑ ↓`；任 | `questionDialogMaxHeight` | `20` | 问题面板最大行数 | | `modelDialogWidth` | `76` | 模型选择器宽度（列数） | | `modelDialogMaxHeight` | `20` | 模型选择器最大行数 | +| `detailsDialogWidth` | `72` | transcript 细节选择器宽度（列数） | | `fileSearchMaxResults` | `20` | 一次 `@` 查询显示的最大文件和目录候选数 | | `fileSearchMaxEntries` | `10000` | 无路径模糊查询使用的有界工作区索引最多保留的路径数 | | `fileSearchExcludedDirectories` | `['.git', 'node_modules']` | 遍历和直接补全时忽略的目录 basename | diff --git a/packages/ui/tui/src/components/dialogs.ts b/packages/ui/tui/src/components/dialogs.ts index f59ff747be..93c416b230 100644 --- a/packages/ui/tui/src/components/dialogs.ts +++ b/packages/ui/tui/src/components/dialogs.ts @@ -433,40 +433,70 @@ export class ModelDialog implements Component { } } -/** One transcript-detail state the details selector applies on confirm. */ -export type DetailsSelection = - | { readonly kind: 'tools'; readonly visibility: ToolCardVisibility } - | { readonly kind: 'reasoning'; readonly show: boolean } +/** Both transcript-detail dimensions the details selector applies on confirm. */ +export interface DetailsSelection { + readonly visibility: ToolCardVisibility + readonly showReasoning: boolean +} + +const TOOL_CARD_PHASES: readonly ToolCardVisibility[] = ['collapsed', 'expanded', 'hidden'] /** - * Keyboard selector over the transcript detail states: the three tool-card - * visibility phases and reasoning-block display. Enter applies the highlighted - * state and closes; Esc or Ctrl+C closes without changing anything. + * Keyboard selector over the two transcript-detail entries — tool-card + * visibility and reasoning display. Tab cycles the highlighted entry's pending + * value, Enter applies both pending values and closes, Esc or Ctrl+C closes + * without changing anything. A pending value renders as `current → pending`. */ export class DetailsDialog implements Component { private readonly list: SelectList + private readonly toolsItem: SelectItem + private readonly reasoningItem: SelectItem + private pendingVisibility: ToolCardVisibility + private pendingReasoning: boolean constructor( - visibility: ToolCardVisibility, - showReasoning: boolean, + private readonly visibility: ToolCardVisibility, + private readonly showReasoning: boolean, private readonly palette: Palette, done: (selection: DetailsSelection) => void, private readonly cancel: () => void, ) { - const current = (isCurrent: boolean): string => isCurrent ? ' — current' : '' - const items: SelectItem[] = [ - { value: 'collapsed', label: 'Tool cards · collapsed', description: `head/tail preview${current(visibility === 'collapsed')}` }, - { value: 'expanded', label: 'Tool cards · expanded', description: `full bodies${current(visibility === 'expanded')}` }, - { value: 'hidden', label: 'Tool cards · hidden', description: `conversation only${current(visibility === 'hidden')}` }, - { value: 'reasoning-shown', label: 'Reasoning · shown', description: `show reasoning blocks${current(showReasoning)}` }, - { value: 'reasoning-hidden', label: 'Reasoning · hidden', description: `omit reasoning blocks${current(!showReasoning)}` }, - ] - this.list = new SelectList(items, items.length, dialogSelectTheme(palette)) - this.list.setSelectedIndex(items.findIndex(item => item.value === visibility)) - this.list.onSelect = (item) => { - done(item.value === 'reasoning-shown' || item.value === 'reasoning-hidden' - ? { kind: 'reasoning', show: item.value === 'reasoning-shown' } - : { kind: 'tools', visibility: item.value as ToolCardVisibility }) + this.pendingVisibility = visibility + this.pendingReasoning = showReasoning + this.toolsItem = { value: 'tools', label: 'Tool cards', description: this.describeTools() } + this.reasoningItem = { value: 'reasoning', label: 'Reasoning', description: this.describeReasoning() } + this.list = new SelectList([this.toolsItem, this.reasoningItem], 2, dialogSelectTheme(palette)) + this.list.onSelect = () => { + done({ visibility: this.pendingVisibility, showReasoning: this.pendingReasoning }) + } + } + + /** `current → pending` when Tab moved the value, otherwise the current value. */ + private static pendingLabel(current: string, pending: string): string { + return pending === current ? current : `${current} → ${pending}` + } + + private describeTools(): string { + return DetailsDialog.pendingLabel(this.visibility, this.pendingVisibility) + } + + private describeReasoning(): string { + const label = (show: boolean): string => show ? 'shown' : 'hidden' + return DetailsDialog.pendingLabel(label(this.showReasoning), label(this.pendingReasoning)) + } + + /** Cycle the highlighted entry's pending value one step. */ + private cyclePending(): void { + const selected = this.list.getSelectedItem() + /* v8 ignore next -- the two-entry list always has a selection. */ + if (selected === null) return + if (selected.value === 'tools') { + const index = TOOL_CARD_PHASES.indexOf(this.pendingVisibility) + this.pendingVisibility = TOOL_CARD_PHASES[(index + 1) % TOOL_CARD_PHASES.length] as ToolCardVisibility + this.toolsItem.description = this.describeTools() + } else { + this.pendingReasoning = !this.pendingReasoning + this.reasoningItem.description = this.describeReasoning() } } @@ -476,6 +506,7 @@ export class DetailsDialog implements Component { handleInput(data: string): void { if (matchesKey(data, Key.escape) || matchesKey(data, Key.ctrl('c'))) this.cancel() + else if (matchesKey(data, Key.tab)) this.cyclePending() else this.list.handleInput(data) this.invalidate() } @@ -485,7 +516,7 @@ export class DetailsDialog implements Component { return renderDialog('Transcript details', [ ...this.list.render(innerWidth), '', - this.palette.dim('↑/↓ move • Enter apply • Esc cancel'), + this.palette.dim('↑/↓ move • Tab cycle • Enter apply • Esc cancel'), ], width, this.palette) } } diff --git a/packages/ui/tui/src/index.ts b/packages/ui/tui/src/index.ts index 7fd76a1a6f..9a07b1f7c4 100644 --- a/packages/ui/tui/src/index.ts +++ b/packages/ui/tui/src/index.ts @@ -1051,8 +1051,9 @@ export function createTuiChat( palette, (selection: DetailsSelection) => { void session.close() - if (selection.kind === 'reasoning') setReasoning(selection.show) - else setToolsVisibility(selection.visibility) + // Reasoning first: its transcript rebuild would drop the card notice. + if (selection.showReasoning !== showReasoning) setReasoning(selection.showReasoning) + if (selection.visibility !== toolsVisibility) setToolsVisibility(selection.visibility) }, () => { void session.close() }, ), diff --git a/packages/ui/tui/tests/snapshots/details-selector.expected.txt b/packages/ui/tui/tests/snapshots/details-selector.expected.txt index ad6514d4e9..d6cacf8cef 100644 --- a/packages/ui/tui/tests/snapshots/details-selector.expected.txt +++ b/packages/ui/tui/tests/snapshots/details-selector.expected.txt @@ -28,39 +28,33 @@ buffer 13| 14| "Tool cards hidden. " style 0-17 dim -15| " ╭ Transcript details ──────────────────────────────────────────────────╮ " - style 14-85 fg=bright-magenta -16| "/workspace/pro│ Tool cards · collapsed head/tail preview │ " - style 0-13 fg=bright-magenta bold - style 14-14 fg=bright-magenta - style 40-66 dim - style 85-85 fg=bright-magenta -17| " dsh > │ Tool cards · expanded full bodies │ " +15| +16| "/workspace/project (tui-staging) deepseek-v4-flash ↑0 ↓0 0% context" + style 0-17 fg=bright-magenta bold + style 18-31 dim + style 34-50 dim + style 53-57 dim + style 60-69 dim +17| " dsh > ╭ Transcript details ──────────────────────────────────────────────────╮ " style 1-3 fg=bright-magenta bold style 5-6 dim style 7-7 inverse - style 14-14 fg=bright-magenta - style 39-60 dim - style 85-85 fg=bright-magenta -18| " │ → Tool cards · hidden conversation only — current │ " - style 14-14 fg=bright-magenta - style 16-76 fg=bright-magenta inverse - style 85-85 fg=bright-magenta -19| " │ Reasoning · shown show reasoning blocks │ " - style 14-14 fg=bright-magenta - style 35-70 dim - style 85-85 fg=bright-magenta -20| " │ Reasoning · hidden omit reasoning blocks — current │ " - style 14-14 fg=bright-magenta - style 36-80 dim - style 85-85 fg=bright-magenta -21| " │ │ " - style 14-14 fg=bright-magenta - style 85-85 fg=bright-magenta -22| " │ ↑/↓ move • Enter apply • Esc cancel │ " - style 14-14 fg=bright-magenta - style 16-50 dim - style 85-85 fg=bright-magenta -23| " ╰──────────────────────────────────────────────────────────────────────╯ " style 14-85 fg=bright-magenta -24-39| +18| " │ → Tool cards hidden → collapsed │ " + style 14-14 fg=bright-magenta + style 16-67 fg=bright-magenta inverse + style 85-85 fg=bright-magenta +19| " │ Reasoning hidden │ " + style 14-14 fg=bright-magenta + style 27-55 dim + style 85-85 fg=bright-magenta +20| " │ │ " + style 14-14 fg=bright-magenta + style 85-85 fg=bright-magenta +21| " │ ↑/↓ move • Tab cycle • Enter apply • Esc cancel │ " + style 14-14 fg=bright-magenta + style 16-62 dim + style 85-85 fg=bright-magenta +22| " ╰──────────────────────────────────────────────────────────────────────╯ " + style 14-85 fg=bright-magenta +23-39| diff --git a/packages/ui/tui/tests/tui.snapshot.ts b/packages/ui/tui/tests/tui.snapshot.ts index c0b6d44f88..6931df723f 100644 --- a/packages/ui/tui/tests/tui.snapshot.ts +++ b/packages/ui/tui/tests/tui.snapshot.ts @@ -663,11 +663,13 @@ describe('TUI terminal-state snapshots', () => { harness.terminal.send('\r') }) await checkpoint('details-command', harness.terminal, { includeScrollback: true }) - // Bare /details opens the selector, preselecting and marking the current - // hidden/reasoning-off state. + // Bare /details opens the two-entry selector seeded with the current + // hidden/reasoning-off state; one Tab renders the tool-card entry's + // pending cycle as `hidden → collapsed`. await renderAfter(harness, () => { harness.terminal.send('/details') harness.terminal.send('\r') + harness.terminal.send('\t') }) await checkpoint('details-selector', harness.terminal, { includeScrollback: true }) nowSpy.mockRestore() diff --git a/packages/ui/tui/tests/tui.spec.ts b/packages/ui/tui/tests/tui.spec.ts index 093d8ad285..1f4272bb6b 100644 --- a/packages/ui/tui/tests/tui.spec.ts +++ b/packages/ui/tui/tests/tui.spec.ts @@ -2554,10 +2554,9 @@ describe('pi-tui chat lifecycle and transcript', () => { return from } - await open() - expect(result.terminal.output).toContain('Tool cards · collapsed') - expect(result.terminal.output).toContain('head/tail preview — current') - expect(result.terminal.output).toContain('show reasoning blocks — current') + const opened = await open() + expect(result.terminal.output.slice(opened)).toContain('Tool cards') + expect(result.terminal.output.slice(opened)).toContain('Reasoning') // A second /details while the selector is open replaces the overlay // instead of stacking a second one behind it. @@ -2570,22 +2569,37 @@ describe('pi-tui chat lifecycle and transcript', () => { await tick() expect(result.terminal.output.slice(cancelOutput)).not.toContain('Tool and context cards') - // Enter on the next visibility row applies it and closes. - await open() - result.terminal.send('\x1b[B') + // Tab cycles the highlighted entry's pending value; Enter applies it. + const cycled = await open() + result.terminal.send('\t') + await tick() + expect(result.terminal.output.slice(cycled)).toContain('collapsed → expanded') result.terminal.send('\r') await tick() expect(result.terminal.output).toContain('Tool and context cards expanded.') - // The reopened selector preselects the current phase and marks it. + // Both entries apply in one confirm: cycle tool cards through the + // wraparound back to collapsed and toggle reasoning off. const reopened = await open() - expect(result.terminal.output.slice(reopened)).toContain('full bodies — current') - result.terminal.send('\x1b[B') - result.terminal.send('\x1b[B') + result.terminal.send('\t') + result.terminal.send('\t') + await tick() + expect(result.terminal.output.slice(reopened)).toContain('expanded → collapsed') result.terminal.send('\x1b[B') + result.terminal.send('\t') + await tick() + expect(result.terminal.output.slice(reopened)).toContain('shown → hidden') result.terminal.send('\r') await tick() expect(result.terminal.output).toContain('Reasoning blocks hidden.') + expect(result.terminal.output).toContain('Tool and context cards collapsed.') + + // Enter with no pending change closes without a notice. + const unchanged = await open() + result.terminal.send('\r') + await tick() + expect(result.terminal.output.slice(unchanged)).not.toContain('Tool and context cards') + expect(result.terminal.output.slice(unchanged)).not.toContain('Reasoning blocks') // Ctrl+C also cancels. const ctrlCOutput = result.terminal.output.length From 5b144238a28352d204926d444bd8f7cbbb20cf9e Mon Sep 17 00:00:00 2001 From: Turtle Date: Thu, 30 Jul 2026 23:55:33 +0800 Subject: [PATCH 07/48] feat(tui): /details Tab toggles apply immediately Drop the pending state and confirm step: Tab cycles the highlighted entry and applies at once, so the transcript behind the dialog is the live preview; Enter/Esc/Ctrl+C just close. --- .../2026-07-30-tui-details-command.i18n.yaml | 4 +- .../feature/2026-07-30-tui-details-command.md | 6 +- .../2026-07-30-tui-details-command.zh.md | 6 +- packages/ui/tui/README.i18n.yaml | 4 +- packages/ui/tui/README.md | 2 +- packages/ui/tui/README.zh.md | 2 +- packages/ui/tui/src/components/dialogs.ts | 65 +++++++------------ packages/ui/tui/src/index.ts | 3 +- .../snapshots/details-selector.expected.txt | 60 ++++++++++------- packages/ui/tui/tests/tui.snapshot.ts | 7 +- packages/ui/tui/tests/tui.spec.ts | 50 ++++++-------- 11 files changed, 98 insertions(+), 111 deletions(-) diff --git a/.agents/notes/implemented/feature/2026-07-30-tui-details-command.i18n.yaml b/.agents/notes/implemented/feature/2026-07-30-tui-details-command.i18n.yaml index 8e6c202a5e..5a1b64f6b3 100644 --- a/.agents/notes/implemented/feature/2026-07-30-tui-details-command.i18n.yaml +++ b/.agents/notes/implemented/feature/2026-07-30-tui-details-command.i18n.yaml @@ -2,5 +2,5 @@ # side as of the last confirmed-consistent state. Both languages carry equal authority; # after editing either side, bring the other along and re-record with: # pnpm run verify-translation-pairing --write .agents/notes/implemented/feature/2026-07-30-tui-details-command.md -2026-07-30-tui-details-command.md: 5de80ceb3ad78a949268c31e9c1d1a50c956b90f -2026-07-30-tui-details-command.zh.md: f74d0658cb776752bf3da1682291de882638310a +2026-07-30-tui-details-command.md: fb7c4dfaedeff27c9cafd0ba82daf4739665f19c +2026-07-30-tui-details-command.zh.md: 5f9e033311d1999998ef08b8f340d71f751b11f6 diff --git a/.agents/notes/implemented/feature/2026-07-30-tui-details-command.md b/.agents/notes/implemented/feature/2026-07-30-tui-details-command.md index 5de80ceb3a..fb7c4dfaed 100644 --- a/.agents/notes/implemented/feature/2026-07-30-tui-details-command.md +++ b/.agents/notes/implemented/feature/2026-07-30-tui-details-command.md @@ -10,7 +10,7 @@ The TUI's transcript detail state — tool-card visibility (`collapsed`/`expande ## Decision -`dsh-tui` registers `/details` beside its other agent-scoped commands. Bare `/details` opens `DetailsDialog`, a centered keyboard selector with one entry per dimension — `Tool cards` and `Reasoning` — seeded with the current values: Tab cycles the highlighted entry's pending value (rendered as `current → pending`), Enter applies every changed dimension in one confirm and closes, and Esc or Ctrl+C cancels; its width is the `detailsDialogWidth` config key and a second `/details` replaces an open selector, mirroring the `/model` overlay. Arguments name target states directly: `collapsed|expanded|hidden` jumps tool cards to that phase, `reasoning on|off` sets reasoning display, bare `reasoning` toggles it, and directives combine in one invocation. An unknown token returns a command error carrying the usage line. Every entry mutates the same closure state as the shortcuts, refactored so the cycle and toggle are thin wrappers over `setToolsVisibility`/`setReasoning`; the shortcuts and their notices are unchanged. +`dsh-tui` registers `/details` beside its other agent-scoped commands. Bare `/details` opens `DetailsDialog`, a centered keyboard toggle with one entry per dimension — `Tool cards` and `Reasoning` — showing the live values: Tab cycles the highlighted entry and applies the change immediately, so the transcript behind the dialog is the preview, and Enter, Esc, or Ctrl+C closes; its width is the `detailsDialogWidth` config key and a second `/details` replaces an open selector, mirroring the `/model` overlay. Arguments name target states directly: `collapsed|expanded|hidden` jumps tool cards to that phase, `reasoning on|off` sets reasoning display, bare `reasoning` toggles it, and directives combine in one invocation. An unknown token returns a command error carrying the usage line. Every entry mutates the same closure state as the shortcuts, refactored so the cycle and toggle are thin wrappers over `setToolsVisibility`/`setReasoning`; the shortcuts and their notices are unchanged. A combined invocation applies reasoning before visibility because `setReasoning` rebuilds the transcript from session events, which drops non-durable notice components; applying it last would erase the just-appended visibility notice. @@ -30,5 +30,5 @@ The reasoning rebuild exposed a replay defect that this change fixes in `renderE - A user can jump to any detail mode, set both dimensions at once, and see the current state in the selector — including on terminals that intercept Ctrl+O/Ctrl+R. - The parser accepts order-free tokens, so `/details reasoning expanded` toggles reasoning and expands cards; last directive wins per dimension. This leniency is deliberate and documented in the README. -- The selector applies only changed dimensions on confirm, so Enter without a Tab closes silently; a two-dimension change is one open-Tab-Enter interaction. -- Transcript rebuilds no longer lose assistant messages when a step carries more than one `assistant/message` event; the `details-command` snapshot pins the argument surface and the fixed replay, and `details-selector` pins the open selector with a Tab-cycled `hidden → collapsed` pending value. +- The selector has no pending state or cancel: every Tab is a real, already-notified change, and closing never reverts. A user who over-cycles simply Tabs on to the wanted value. +- Transcript rebuilds no longer lose assistant messages when a step carries more than one `assistant/message` event; the `details-command` snapshot pins the argument surface and the fixed replay, and `details-selector` pins the open toggle right after a Tab applied `hidden` -> `collapsed`, including the restored tool card behind it. diff --git a/.agents/notes/implemented/feature/2026-07-30-tui-details-command.zh.md b/.agents/notes/implemented/feature/2026-07-30-tui-details-command.zh.md index f74d0658cb..5f9e033311 100644 --- a/.agents/notes/implemented/feature/2026-07-30-tui-details-command.zh.md +++ b/.agents/notes/implemented/feature/2026-07-30-tui-details-command.zh.md @@ -10,7 +10,7 @@ TUI 的 transcript（文本记录）细节状态——工具卡片可见性（`c ## Decision -`dsh-tui` 在其他 agent 作用域命令旁注册 `/details`。裸 `/details` 打开 `DetailsDialog`：一个居中的键盘选择器，每个维度一个条目——`Tool cards` 与 `Reasoning`——以当前值为初始：Tab 循环高亮条目的待定值（渲染为 `current → pending`），Enter 一次确认应用所有已改变的维度并关闭，Esc 或 Ctrl+C 取消；其宽度由配置键 `detailsDialogWidth` 决定，选择器打开时再次执行 `/details` 会替换它，与 `/model` 浮层一致。参数直接命名目标状态：`collapsed|expanded|hidden` 让工具卡片跳到该阶段，`reasoning on|off` 设置 reasoning 显示，裸 `reasoning` 切换它，且指令可在一次调用中组合。未知 token 返回携带用法行的命令错误。每个入口改动的都是与快捷键相同的闭包状态，重构后循环与切换成为 `setToolsVisibility`/`setReasoning` 之上的薄封装；快捷键及其通知保持不变。 +`dsh-tui` 在其他 agent 作用域命令旁注册 `/details`。裸 `/details` 打开 `DetailsDialog`：一个居中的键盘开关，每个维度一个条目——`Tool cards` 与 `Reasoning`——显示实时值：Tab 循环高亮条目并立即应用变更，对话框背后的 transcript 即是预览，Enter、Esc 或 Ctrl+C 关闭；其宽度由配置键 `detailsDialogWidth` 决定，选择器打开时再次执行 `/details` 会替换它，与 `/model` 浮层一致。参数直接命名目标状态：`collapsed|expanded|hidden` 让工具卡片跳到该阶段，`reasoning on|off` 设置 reasoning 显示，裸 `reasoning` 切换它，且指令可在一次调用中组合。未知 token 返回携带用法行的命令错误。每个入口改动的都是与快捷键相同的闭包状态，重构后循环与切换成为 `setToolsVisibility`/`setReasoning` 之上的薄封装；快捷键及其通知保持不变。 组合调用先应用 reasoning 再应用可见性，因为 `setReasoning` 会从会话事件重建 transcript，而重建会丢弃非持久的通知组件；若最后才应用它，会抹掉刚追加的可见性通知。 @@ -30,5 +30,5 @@ reasoning 重建暴露了一个重放缺陷，本变更在 `renderEvent` 中修 - 用户可以跳到任意细节模式、一次设置两个维度，并在选择器中看到当前状态——包括在拦截 Ctrl+O/Ctrl+R 的终端上。 - 解析器接受无序 token，因此 `/details reasoning expanded` 会切换 reasoning 并展开卡片；每个维度以最后一个指令为准。这一宽松是刻意的，并记录在 README 中。 -- 选择器确认时只应用已改变的维度，因此未按 Tab 直接 Enter 会静默关闭；两个维度的变更是一次打开-Tab-Enter 交互。 -- 当一个步骤携带多条 `assistant/message` 事件时，transcript 重建不再丢失 assistant 消息；`details-command` 快照固定参数表面与修复后的重放，`details-selector` 固定经 Tab 循环出 `hidden → collapsed` 待定值的打开选择器。 +- 选择器没有待定状态与取消：每次 Tab 都是已生效、已通知的真实变更，关闭从不回退。循环过头的用户继续 Tab 到想要的值即可。 +- 当一个步骤携带多条 `assistant/message` 事件时，transcript 重建不再丢失 assistant 消息；`details-command` 快照固定参数表面与修复后的重放，`details-selector` 固定 Tab 将 `hidden` 应用为 `collapsed` 后仍打开的开关，包括其背后恢复显示的工具卡片。 diff --git a/packages/ui/tui/README.i18n.yaml b/packages/ui/tui/README.i18n.yaml index af9ba4a67a..1bef78a96e 100644 --- a/packages/ui/tui/README.i18n.yaml +++ b/packages/ui/tui/README.i18n.yaml @@ -2,5 +2,5 @@ # side as of the last confirmed-consistent state. Both languages carry equal authority; # after editing either side, bring the other along and re-record with: # pnpm run verify-translation-pairing --write packages/ui/tui/README.md -README.md: 7230e75535ce5710e12040e026bc77a1782ed6dd -README.zh.md: 3079b797b670de64cfdc1faa9d002ee04a6e28dd +README.md: b6d0ab8e143f14ea2826054f72cf5a778ffa9bbb +README.zh.md: addce9f3b55c081208b0849bc00522fa0a22e295 diff --git a/packages/ui/tui/README.md b/packages/ui/tui/README.md index 8ef5f20ca0..66d7141994 100644 --- a/packages/ui/tui/README.md +++ b/packages/ui/tui/README.md @@ -22,7 +22,7 @@ Typing `@` at a token boundary searches files and directories under the session When optional `ctx.sessionReferences` is mounted, the same `@` menu also offers metadata-only session candidates, inserts `@[label](dsh-session:)`, and prepares the selected snapshots before dispatch. Session references remain structured because the model has no filesystem-like tool for retrieving session snapshots later. Preparation disables duplicate submission and restores the editor input on failure. The TUI chooses `agent.steer()` or `agent.followup()` from the status after that asynchronous preparation, so idle follow-ups still dispatch `agent/prompt-submit` while in-turn steering joins at a checkpoint without that hook. -While the agent is running, ordinary editor submissions call `agent.steer()`; otherwise they call `agent.followup()`. A slash at the start of the submitted line enters `ctx.commands` instead: known commands execute directly, unknown commands produce a warning, and neither path automatically reaches the model. A command producer may explicitly schedule agent work; [`dsh-plan-mode`](../../plan/plan-mode/README.md#model-and-human-surfaces) uses that contract for `/plan [message]`. The TUI registers `/help`, `/model`, `/clear`, `/details`, `/palette`, `/reload`, `/resume`, `/status`, and `/exit` as agent-scoped definitions; every other effective command joins autocomplete and `/help` dynamically, as do `/skill:` completions. A status line above the editor reports the turn phase the TUI derives from session events — waiting for the first token, thinking, responding, or executing tools — with the elapsed time in that phase and the running step total, refreshed each second, and ends with the `Enter sends steering, Esc cancels` hint; while steering messages wait to reach the model it inserts a `N queued ·` badge before the hint that clears as each drains. Ctrl+C or Escape cancels a running turn. Tool and injected-context cards collapse long bodies into a configurable head/tail preview; Ctrl+O cycles tool cards through collapsed preview, full output, and hidden — the hidden phase drops tool cards from the transcript entirely while context cards stay at their preview, since injected instructions are not tool traffic. The hidden phase also folds each turn's assistant steps into one message: the first step with visible text or reasoning keeps the turn's single `Assistant` header, later steps render as headerless continuations, and a step without a visible body renders nothing; leaving the hidden phase restores the per-step headers. An injected-context card renders its message as prose with the producer's outer reminder frame stripped, so neither the fold nor the frame stripping depends on the payload's syntax. Ctrl+R toggles reasoning, Ctrl+L redraws, and Ctrl+D exits while idle. `/details` names the same state those two shortcuts cycle: bare it opens a centered keyboard selector with one entry per dimension — `Tool cards` and `Reasoning` — seeded with the current values, where Tab cycles the highlighted entry's pending value (rendered as `current → pending`), Enter applies every changed dimension and closes, and Esc or Ctrl+C cancels without changing anything; `/details collapsed|expanded|hidden` jumps tool cards to that phase directly, and `/details reasoning [on|off]` sets — or bare `reasoning` toggles — reasoning-block display; arguments combine in one invocation, an unknown argument fails with the usage line, and a combined invocation applies reasoning first so its transcript rebuild never drops the card notice. +While the agent is running, ordinary editor submissions call `agent.steer()`; otherwise they call `agent.followup()`. A slash at the start of the submitted line enters `ctx.commands` instead: known commands execute directly, unknown commands produce a warning, and neither path automatically reaches the model. A command producer may explicitly schedule agent work; [`dsh-plan-mode`](../../plan/plan-mode/README.md#model-and-human-surfaces) uses that contract for `/plan [message]`. The TUI registers `/help`, `/model`, `/clear`, `/details`, `/palette`, `/reload`, `/resume`, `/status`, and `/exit` as agent-scoped definitions; every other effective command joins autocomplete and `/help` dynamically, as do `/skill:` completions. A status line above the editor reports the turn phase the TUI derives from session events — waiting for the first token, thinking, responding, or executing tools — with the elapsed time in that phase and the running step total, refreshed each second, and ends with the `Enter sends steering, Esc cancels` hint; while steering messages wait to reach the model it inserts a `N queued ·` badge before the hint that clears as each drains. Ctrl+C or Escape cancels a running turn. Tool and injected-context cards collapse long bodies into a configurable head/tail preview; Ctrl+O cycles tool cards through collapsed preview, full output, and hidden — the hidden phase drops tool cards from the transcript entirely while context cards stay at their preview, since injected instructions are not tool traffic. The hidden phase also folds each turn's assistant steps into one message: the first step with visible text or reasoning keeps the turn's single `Assistant` header, later steps render as headerless continuations, and a step without a visible body renders nothing; leaving the hidden phase restores the per-step headers. An injected-context card renders its message as prose with the producer's outer reminder frame stripped, so neither the fold nor the frame stripping depends on the payload's syntax. Ctrl+R toggles reasoning, Ctrl+L redraws, and Ctrl+D exits while idle. `/details` names the same state those two shortcuts cycle: bare it opens a centered keyboard toggle with one entry per dimension — `Tool cards` and `Reasoning` — showing the live values, where Tab cycles the highlighted entry and applies the change immediately (the transcript behind the dialog is the preview), and Enter, Esc, or Ctrl+C closes; `/details collapsed|expanded|hidden` jumps tool cards to that phase directly, and `/details reasoning [on|off]` sets — or bare `reasoning` toggles — reasoning-block display; arguments combine in one invocation, an unknown argument fails with the usage line, and a combined invocation applies reasoning first so its transcript rebuild never drops the card notice. `/model` opens the advisory `ctx.llm` catalog as a keyboard selector: a filter box above the list narrows rows by a case-insensitive substring over each row's `provider/model` label, model name, and description, keeping the highlighted row selected when it survives the filter; Up/Down moves, Shift+Tab cycles the focused model's adapter-advertised reasoning efforts in display order, Enter selects the model and effort, and Escape clears a non-empty filter before a second Escape closes it. When an adapter does not advertise a default effort, the cycle also includes `Default`, which clears an explicit selection and preserves the provider default; models without selectable effort metadata ignore Shift+Tab. The selector renders the exact advertised effort list—including `off` when present—and does not synthesize, clamp, or transfer an effort between models. `/model ` still selects an unambiguous model id directly, while `/model /` selects an exact target and uses its adapter default when one exists. The configured target or latest logged request header initializes the selector, and an unlisted current model remains visible because catalogs are advisory. Selection is local to this TUI session. Prompt assembly snapshots the target for one step, replaces `{{provider}}` and `{{model}}`, and applies the same provider/model/reasoning-effort target through `agent/request`; a switch during assembly therefore starts with a later step. The request header durably records targets that reach the model, while an unused selection remains process-local. diff --git a/packages/ui/tui/README.zh.md b/packages/ui/tui/README.zh.md index 48cb25fb94..9b67d1a4a2 100644 --- a/packages/ui/tui/README.zh.md +++ b/packages/ui/tui/README.zh.md @@ -22,7 +22,7 @@ TUI 从活跃会话表层重建已恢复历史，渲染 Markdown 响应与 reaso 挂载可选的 `ctx.sessionReferences` 后，同一个 `@` 菜单还会提供仅含元数据的会话候选项，插入 `@[label](dsh-session:)`，并在分派前准备所选快照。会话引用保持结构化，因为模型没有类似文件系统的工具可在稍后检索会话快照。准备期间会禁止重复提交，并在失败时恢复编辑器输入。TUI 会在异步准备后根据状态选择 `agent.steer()` 或 `agent.followup()`，因此空闲 followup 仍会分派 `agent/prompt-submit`，而轮次中的 steering 会在检查点加入且不触发该 hook。 -Agent 运行时，普通编辑器提交会调用 `agent.steer()`；其他时候调用 `agent.followup()`。提交行以斜杠开头时会改为进入 `ctx.commands`：已知命令直接执行，未知命令产生警告，两条路径都不会自动到达模型。命令生产方可以显式调度 agent 工作；[`dsh-plan-mode`](../../plan/plan-mode/README.md#model-and-human-surfaces) 使用该契约实现 `/plan [message]`。TUI 将 `/help`、`/model`、`/clear`、`/details`、`/palette`、`/reload`、`/resume`、`/status` 和 `/exit` 注册为 agent 作用域定义；其他所有有效命令都会动态加入自动补全与 `/help`，`/skill:` 补全也相同。编辑器上方的状态行会报告 TUI 从会话事件派生的轮次阶段，包括等待首个 token、思考、响应或执行工具；它显示该阶段已经过时间和运行中的步骤总数，每秒刷新，并以 `Enter sends steering, Esc cancels` 提示结尾。Steering 消息等待到达模型期间，会在提示前插入 `N queued ·` 徽标，每条消息排空后随即清除。Ctrl+C 或 Escape 会取消运行中的轮次。工具卡片与注入上下文卡片都把长主体折叠为可配置的头尾预览；Ctrl+O 让工具卡片在折叠预览、完整输出、隐藏三种状态间循环——隐藏阶段把工具卡片从 transcript 中完全去掉，而上下文卡片保持预览，因为注入的指令不属于工具流量。隐藏阶段还会把每个轮次的 assistant 步骤折叠为一条消息：第一个有可见文本或 reasoning 的步骤保留该轮次唯一的 `Assistant` 标题，之后的步骤渲染为无标题的续段，没有可见正文的步骤则不渲染任何内容；离开隐藏阶段会恢复每步各自的标题。注入上下文卡片把消息渲染为文本，并去掉生产方的外层提醒外框，因此折叠与去外框都不依赖载荷的语法。Ctrl+R 切换 reasoning，Ctrl+L 重绘，Ctrl+D 在空闲时退出。`/details` 命名的正是这两个快捷键循环的同一份状态：不带参数时打开一个居中的键盘选择器，每个维度一个条目——`Tool cards` 与 `Reasoning`——以当前值为初始，Tab 循环高亮条目的待定值（渲染为 `current → pending`），Enter 应用所有已改变的维度并关闭，Esc 或 Ctrl+C 取消且不改变任何东西；`/details collapsed|expanded|hidden` 让工具卡片直接跳到该阶段，`/details reasoning [on|off]` 设置——或裸 `reasoning` 切换——reasoning 块显示；参数可在一次调用中组合，未知参数会以用法行报错，组合调用先应用 reasoning，使其 transcript 重建不会丢掉卡片通知。 +Agent 运行时，普通编辑器提交会调用 `agent.steer()`；其他时候调用 `agent.followup()`。提交行以斜杠开头时会改为进入 `ctx.commands`：已知命令直接执行，未知命令产生警告，两条路径都不会自动到达模型。命令生产方可以显式调度 agent 工作；[`dsh-plan-mode`](../../plan/plan-mode/README.md#model-and-human-surfaces) 使用该契约实现 `/plan [message]`。TUI 将 `/help`、`/model`、`/clear`、`/details`、`/palette`、`/reload`、`/resume`、`/status` 和 `/exit` 注册为 agent 作用域定义；其他所有有效命令都会动态加入自动补全与 `/help`，`/skill:` 补全也相同。编辑器上方的状态行会报告 TUI 从会话事件派生的轮次阶段，包括等待首个 token、思考、响应或执行工具；它显示该阶段已经过时间和运行中的步骤总数，每秒刷新，并以 `Enter sends steering, Esc cancels` 提示结尾。Steering 消息等待到达模型期间，会在提示前插入 `N queued ·` 徽标，每条消息排空后随即清除。Ctrl+C 或 Escape 会取消运行中的轮次。工具卡片与注入上下文卡片都把长主体折叠为可配置的头尾预览；Ctrl+O 让工具卡片在折叠预览、完整输出、隐藏三种状态间循环——隐藏阶段把工具卡片从 transcript 中完全去掉，而上下文卡片保持预览，因为注入的指令不属于工具流量。隐藏阶段还会把每个轮次的 assistant 步骤折叠为一条消息：第一个有可见文本或 reasoning 的步骤保留该轮次唯一的 `Assistant` 标题，之后的步骤渲染为无标题的续段，没有可见正文的步骤则不渲染任何内容；离开隐藏阶段会恢复每步各自的标题。注入上下文卡片把消息渲染为文本，并去掉生产方的外层提醒外框，因此折叠与去外框都不依赖载荷的语法。Ctrl+R 切换 reasoning，Ctrl+L 重绘，Ctrl+D 在空闲时退出。`/details` 命名的正是这两个快捷键循环的同一份状态：不带参数时打开一个居中的键盘开关，每个维度一个条目——`Tool cards` 与 `Reasoning`——显示实时值，Tab 循环高亮条目并立即应用变更（对话框背后的 transcript 即是预览），Enter、Esc 或 Ctrl+C 关闭；`/details collapsed|expanded|hidden` 让工具卡片直接跳到该阶段，`/details reasoning [on|off]` 设置——或裸 `reasoning` 切换——reasoning 块显示；参数可在一次调用中组合，未知参数会以用法行报错，组合调用先应用 reasoning，使其 transcript 重建不会丢掉卡片通知。 `/model` 将建议性的 `ctx.llm` catalog 打开为键盘选择器：列表上方设有一个过滤框，按对每行 `provider/model` 标签、模型名称和描述的大小写不敏感子串匹配来缩小行集，并在高亮行仍通过过滤时保持其选中状态；Up/Down 移动，Shift+Tab 按显示顺序循环切换适配器为焦点模型公布的推理强度，Enter 选择模型和推理强度，Escape 会先清除非空过滤内容，再次按下才关闭选择器。适配器未公布默认推理强度时，循环还会包含 `Default`，该项会清除显式选择并保留提供方默认行为；没有可选推理强度元数据的模型会忽略 Shift+Tab。选择器会原样呈现公布的推理强度列表（包括存在时的 `off`），不会合成、自动调整或在模型之间转移推理强度。`/model ` 仍可直接选择无歧义的模型 id，`/model /` 则选择精确目标，并在存在时使用其适配器默认值。已配置目标或最新记录的请求 header 会初始化选择器；由于 catalog 仅提供建议，未列出的当前模型仍会显示。选择仅对本 TUI 会话有效。提示词组装会为一个步骤建立目标快照，替换 `{{provider}}` 和 `{{model}}`，并通过 `agent/request` 应用同一个提供方／模型／推理强度目标；因此组装期间的切换会从后续步骤开始生效。请求 header 会持久记录真正到达模型的目标，未使用的选择则只存在于进程本地。 diff --git a/packages/ui/tui/src/components/dialogs.ts b/packages/ui/tui/src/components/dialogs.ts index 93c416b230..946055f1d8 100644 --- a/packages/ui/tui/src/components/dialogs.ts +++ b/packages/ui/tui/src/components/dialogs.ts @@ -433,7 +433,7 @@ export class ModelDialog implements Component { } } -/** Both transcript-detail dimensions the details selector applies on confirm. */ +/** Both transcript-detail dimensions, applied immediately on each Tab. */ export interface DetailsSelection { readonly visibility: ToolCardVisibility readonly showReasoning: boolean @@ -442,62 +442,47 @@ export interface DetailsSelection { const TOOL_CARD_PHASES: readonly ToolCardVisibility[] = ['collapsed', 'expanded', 'hidden'] /** - * Keyboard selector over the two transcript-detail entries — tool-card - * visibility and reasoning display. Tab cycles the highlighted entry's pending - * value, Enter applies both pending values and closes, Esc or Ctrl+C closes - * without changing anything. A pending value renders as `current → pending`. + * Keyboard toggle over the two transcript-detail entries — tool-card + * visibility and reasoning display. Tab cycles the highlighted entry's value + * and applies it immediately, so the transcript behind the dialog is the live + * preview; Enter, Esc, or Ctrl+C closes. */ export class DetailsDialog implements Component { private readonly list: SelectList private readonly toolsItem: SelectItem private readonly reasoningItem: SelectItem - private pendingVisibility: ToolCardVisibility - private pendingReasoning: boolean constructor( - private readonly visibility: ToolCardVisibility, - private readonly showReasoning: boolean, + private visibility: ToolCardVisibility, + private showReasoning: boolean, private readonly palette: Palette, - done: (selection: DetailsSelection) => void, - private readonly cancel: () => void, + private readonly apply: (selection: DetailsSelection) => void, + private readonly close: () => void, ) { - this.pendingVisibility = visibility - this.pendingReasoning = showReasoning - this.toolsItem = { value: 'tools', label: 'Tool cards', description: this.describeTools() } - this.reasoningItem = { value: 'reasoning', label: 'Reasoning', description: this.describeReasoning() } + this.toolsItem = { value: 'tools', label: 'Tool cards', description: visibility } + this.reasoningItem = { value: 'reasoning', label: 'Reasoning', description: this.reasoningLabel() } this.list = new SelectList([this.toolsItem, this.reasoningItem], 2, dialogSelectTheme(palette)) - this.list.onSelect = () => { - done({ visibility: this.pendingVisibility, showReasoning: this.pendingReasoning }) - } + this.list.onSelect = close } - /** `current → pending` when Tab moved the value, otherwise the current value. */ - private static pendingLabel(current: string, pending: string): string { - return pending === current ? current : `${current} → ${pending}` + private reasoningLabel(): string { + return this.showReasoning ? 'shown' : 'hidden' } - private describeTools(): string { - return DetailsDialog.pendingLabel(this.visibility, this.pendingVisibility) - } - - private describeReasoning(): string { - const label = (show: boolean): string => show ? 'shown' : 'hidden' - return DetailsDialog.pendingLabel(label(this.showReasoning), label(this.pendingReasoning)) - } - - /** Cycle the highlighted entry's pending value one step. */ - private cyclePending(): void { + /** Cycle the highlighted entry one step and apply the new state. */ + private cycle(): void { const selected = this.list.getSelectedItem() /* v8 ignore next -- the two-entry list always has a selection. */ if (selected === null) return if (selected.value === 'tools') { - const index = TOOL_CARD_PHASES.indexOf(this.pendingVisibility) - this.pendingVisibility = TOOL_CARD_PHASES[(index + 1) % TOOL_CARD_PHASES.length] as ToolCardVisibility - this.toolsItem.description = this.describeTools() + const index = TOOL_CARD_PHASES.indexOf(this.visibility) + this.visibility = TOOL_CARD_PHASES[(index + 1) % TOOL_CARD_PHASES.length] as ToolCardVisibility + this.toolsItem.description = this.visibility } else { - this.pendingReasoning = !this.pendingReasoning - this.reasoningItem.description = this.describeReasoning() + this.showReasoning = !this.showReasoning + this.reasoningItem.description = this.reasoningLabel() } + this.apply({ visibility: this.visibility, showReasoning: this.showReasoning }) } invalidate(): void { @@ -505,8 +490,8 @@ export class DetailsDialog implements Component { } handleInput(data: string): void { - if (matchesKey(data, Key.escape) || matchesKey(data, Key.ctrl('c'))) this.cancel() - else if (matchesKey(data, Key.tab)) this.cyclePending() + if (matchesKey(data, Key.escape) || matchesKey(data, Key.ctrl('c'))) this.close() + else if (matchesKey(data, Key.tab)) this.cycle() else this.list.handleInput(data) this.invalidate() } @@ -516,7 +501,7 @@ export class DetailsDialog implements Component { return renderDialog('Transcript details', [ ...this.list.render(innerWidth), '', - this.palette.dim('↑/↓ move • Tab cycle • Enter apply • Esc cancel'), + this.palette.dim('↑/↓ move • Tab toggle • Enter/Esc close'), ], width, this.palette) } } diff --git a/packages/ui/tui/src/index.ts b/packages/ui/tui/src/index.ts index 9a07b1f7c4..0e3e42c6a1 100644 --- a/packages/ui/tui/src/index.ts +++ b/packages/ui/tui/src/index.ts @@ -1049,9 +1049,8 @@ export function createTuiChat( toolsVisibility, showReasoning, palette, + // Each Tab applies immediately; one dimension changes per call. (selection: DetailsSelection) => { - void session.close() - // Reasoning first: its transcript rebuild would drop the card notice. if (selection.showReasoning !== showReasoning) setReasoning(selection.showReasoning) if (selection.visibility !== toolsVisibility) setToolsVisibility(selection.visibility) }, diff --git a/packages/ui/tui/tests/snapshots/details-selector.expected.txt b/packages/ui/tui/tests/snapshots/details-selector.expected.txt index d6cacf8cef..69ab18bddd 100644 --- a/packages/ui/tui/tests/snapshots/details-selector.expected.txt +++ b/packages/ui/tui/tests/snapshots/details-selector.expected.txt @@ -20,41 +20,53 @@ buffer 8| "You " style 0-2 fg=bright-magenta bold underline 9| "Inspect the renderer. " -10| "Model wait 0.0s · Completed 2026-07-30 18:00:00 " - style 0-46 dim -11| -12| "Reasoning blocks hidden. " +10| +11| "Assistant " + style 0-8 fg=bright-magenta bold underline +12| +13| "● Tool / bash / Run the coverage gate" + style 0-36 fg=green +14| "$ pnpm run test:coverage " style 0-23 dim -13| -14| "Tool cards hidden. " +15| "/workspace/project " style 0-17 dim -15| -16| "/workspace/project (tui-staging) deepseek-v4-flash ↑0 ↓0 0% context" - style 0-17 fg=bright-magenta bold - style 18-31 dim - style 34-50 dim - style 53-57 dim - style 60-69 dim -17| " dsh > ╭ Transcript details ──────────────────────────────────────────────────╮ " - style 1-3 fg=bright-magenta bold - style 5-6 dim - style 7-7 inverse +16| "… +4 lines (Ctrl+O to expand) " + style 0-28 dim +17| "[exit 0] ╭ Transcript details ──────────────────────────────────────────────────╮ " + style 0-7 dim style 14-85 fg=bright-magenta -18| " │ → Tool cards hidden → collapsed │ " +18| "Model wait 0.0│ → Tool cards collapsed │ " + style 0-13 dim style 14-14 fg=bright-magenta - style 16-67 fg=bright-magenta inverse + style 16-58 fg=bright-magenta inverse style 85-85 fg=bright-magenta 19| " │ Reasoning hidden │ " style 14-14 fg=bright-magenta style 27-55 dim style 85-85 fg=bright-magenta -20| " │ │ " +20| "Reasoning bloc│ │ " + style 0-13 dim style 14-14 fg=bright-magenta style 85-85 fg=bright-magenta -21| " │ ↑/↓ move • Tab cycle • Enter apply • Esc cancel │ " +21| " │ ↑/↓ move • Tab toggle • Enter/Esc close │ " style 14-14 fg=bright-magenta - style 16-62 dim + style 16-54 dim style 85-85 fg=bright-magenta -22| " ╰──────────────────────────────────────────────────────────────────────╯ " +22| "Tool cards hid╰──────────────────────────────────────────────────────────────────────╯ " + style 0-13 dim style 14-85 fg=bright-magenta -23-39| +23| +24| "Tool and context cards collapsed. " + style 0-32 dim +25| +26| "/workspace/project (tui-staging) deepseek-v4-flash ↑0 ↓0 0% context" + style 0-17 fg=bright-magenta bold + style 18-31 dim + style 34-50 dim + style 53-57 dim + style 60-69 dim +27| " dsh > " + style 1-3 fg=bright-magenta bold + style 5-6 dim + style 7-7 inverse +28-39| diff --git a/packages/ui/tui/tests/tui.snapshot.ts b/packages/ui/tui/tests/tui.snapshot.ts index 6931df723f..706ed7dce5 100644 --- a/packages/ui/tui/tests/tui.snapshot.ts +++ b/packages/ui/tui/tests/tui.snapshot.ts @@ -663,9 +663,10 @@ describe('TUI terminal-state snapshots', () => { harness.terminal.send('\r') }) await checkpoint('details-command', harness.terminal, { includeScrollback: true }) - // Bare /details opens the two-entry selector seeded with the current - // hidden/reasoning-off state; one Tab renders the tool-card entry's - // pending cycle as `hidden → collapsed`. + // Bare /details opens the two-entry toggle seeded with the current + // hidden/reasoning-off state; one Tab immediately cycles tool cards + // hidden -> collapsed, so the frame pins the applied notice, the restored + // tool card behind the dialog, and the updated entry value together. await renderAfter(harness, () => { harness.terminal.send('/details') harness.terminal.send('\r') diff --git a/packages/ui/tui/tests/tui.spec.ts b/packages/ui/tui/tests/tui.spec.ts index 1f4272bb6b..8fe7981cb7 100644 --- a/packages/ui/tui/tests/tui.spec.ts +++ b/packages/ui/tui/tests/tui.spec.ts @@ -2544,7 +2544,7 @@ describe('pi-tui chat lifecycle and transcript', () => { await dispose(result) }) - it('bare /details opens the transcript-details selector and applies the confirmed state', async () => { + it('bare /details opens the transcript-details toggle and Tab applies immediately', async () => { const result = await setup() const open = async (): Promise => { const from = result.terminal.output.length @@ -2563,50 +2563,40 @@ describe('pi-tui chat lifecycle and transcript', () => { await result.ctx.commands.execute(result.agent, '/details', new AbortController().signal) await tick() - // Esc cancels without touching the state. - const cancelOutput = result.terminal.output.length - result.terminal.send('\x1b') - await tick() - expect(result.terminal.output.slice(cancelOutput)).not.toContain('Tool and context cards') - - // Tab cycles the highlighted entry's pending value; Enter applies it. - const cycled = await open() + // Each Tab applies one step immediately while the dialog stays open: + // collapsed -> expanded -> hidden -> collapsed (wraparound). result.terminal.send('\t') await tick() - expect(result.terminal.output.slice(cycled)).toContain('collapsed → expanded') - result.terminal.send('\r') - await tick() expect(result.terminal.output).toContain('Tool and context cards expanded.') - - // Both entries apply in one confirm: cycle tool cards through the - // wraparound back to collapsed and toggle reasoning off. - const reopened = await open() - result.terminal.send('\t') result.terminal.send('\t') await tick() - expect(result.terminal.output.slice(reopened)).toContain('expanded → collapsed') + expect(result.terminal.output).toContain('Tool cards hidden.') + result.terminal.send('\t') + await tick() + expect(result.terminal.output).toContain('Tool and context cards collapsed.') + + // The reasoning entry toggles the same way. result.terminal.send('\x1b[B') result.terminal.send('\t') await tick() - expect(result.terminal.output.slice(reopened)).toContain('shown → hidden') - result.terminal.send('\r') - await tick() expect(result.terminal.output).toContain('Reasoning blocks hidden.') - expect(result.terminal.output).toContain('Tool and context cards collapsed.') - // Enter with no pending change closes without a notice. - const unchanged = await open() + // Enter closes without further changes. + const entered = result.terminal.output.length result.terminal.send('\r') await tick() - expect(result.terminal.output.slice(unchanged)).not.toContain('Tool and context cards') - expect(result.terminal.output.slice(unchanged)).not.toContain('Reasoning blocks') + expect(result.terminal.output.slice(entered)).not.toContain('Reasoning blocks') - // Ctrl+C also cancels. - const ctrlCOutput = result.terminal.output.length - await open() + // Esc and Ctrl+C also close; the reopened dialog shows the live values. + const reopened = await open() + expect(result.terminal.output.slice(reopened)).toContain('collapsed') + expect(result.terminal.output.slice(reopened)).toContain('hidden') + result.terminal.send('\x1b') + await tick() + const ctrlCOutput = await open() result.terminal.send('\x03') await tick() - expect(result.terminal.output.slice(ctrlCOutput)).not.toContain('Reasoning blocks shown.') + expect(result.terminal.output.slice(ctrlCOutput)).not.toContain('Reasoning blocks') await dispose(result) }) From 27ab6ee05ec36dad17f5d47e389457109f70046d Mon Sep 17 00:00:00 2001 From: Turtle Date: Fri, 31 Jul 2026 16:37:46 +0800 Subject: [PATCH 08/48] docs: regenerate cordis catalog and re-record TUI README pairing --- docs/cordis-catalog/services.md | 2 +- packages/ui/tui/README.i18n.yaml | 4 ++-- 2 files changed, 3 insertions(+), 3 deletions(-) diff --git a/docs/cordis-catalog/services.md b/docs/cordis-catalog/services.md index d92f467469..4101fcb9e6 100644 --- a/docs/cordis-catalog/services.md +++ b/docs/cordis-catalog/services.md @@ -2336,7 +2336,7 @@ The concrete provider retains pi-tui, focus, and terminal lifecycle state. Plugi abstract openOverlay(request: TuiOverlayRequest): TuiOverlaySession ``` -Source: [`packages/ui/tui/src/index.ts:241`](../../packages/ui/tui/src/index.ts) +Source: [`packages/ui/tui/src/index.ts:243`](../../packages/ui/tui/src/index.ts) ## `ctx.typert` — `TypertRegistry` diff --git a/packages/ui/tui/README.i18n.yaml b/packages/ui/tui/README.i18n.yaml index 1bef78a96e..df5af8083c 100644 --- a/packages/ui/tui/README.i18n.yaml +++ b/packages/ui/tui/README.i18n.yaml @@ -2,5 +2,5 @@ # side as of the last confirmed-consistent state. Both languages carry equal authority; # after editing either side, bring the other along and re-record with: # pnpm run verify-translation-pairing --write packages/ui/tui/README.md -README.md: b6d0ab8e143f14ea2826054f72cf5a778ffa9bbb -README.zh.md: addce9f3b55c081208b0849bc00522fa0a22e295 +README.md: 66d71419945edc9d9f5fe10bf9bcd810d05ba20c +README.zh.md: 9b67d1a4a2e4a3d27d263421cc909188ff2e34cc From 9146c97d13628200006701da513bfb9c48787c79 Mon Sep 17 00:00:00 2001 From: Turtle Date: Fri, 31 Jul 2026 17:35:37 +0800 Subject: [PATCH 09/48] review: order detailsDialogWidth consistently across the parallel config lists --- docs/config-catalog.md | 4 ++-- packages/ui/tui/src/config.ts | 6 +++--- 2 files changed, 5 insertions(+), 5 deletions(-) diff --git a/docs/config-catalog.md b/docs/config-catalog.md index 0f04e0a955..2cd6ca2bed 100644 --- a/docs/config-catalog.md +++ b/docs/config-catalog.md @@ -2034,10 +2034,10 @@ export interface TuiConfig { questionDialogMaxHeight?: number /** Model-selector width in terminal columns. */ modelDialogWidth?: number - /** Transcript-details selector width in terminal columns. */ - detailsDialogWidth?: number /** Model-selector maximum height in terminal rows. */ modelDialogMaxHeight?: number + /** Transcript-details selector width in terminal columns. */ + detailsDialogWidth?: number /** Maximum fuzzy file candidates displayed for one `@` query. */ fileSearchMaxResults?: number /** Maximum paths retained in one `@` workspace index. */ diff --git a/packages/ui/tui/src/config.ts b/packages/ui/tui/src/config.ts index 59a12c0b96..e82aa46e08 100644 --- a/packages/ui/tui/src/config.ts +++ b/packages/ui/tui/src/config.ts @@ -46,10 +46,10 @@ export interface TuiConfig { questionDialogMaxHeight?: number /** Model-selector width in terminal columns. */ modelDialogWidth?: number - /** Transcript-details selector width in terminal columns. */ - detailsDialogWidth?: number /** Model-selector maximum height in terminal rows. */ modelDialogMaxHeight?: number + /** Transcript-details selector width in terminal columns. */ + detailsDialogWidth?: number /** Maximum fuzzy file candidates displayed for one `@` query. */ fileSearchMaxResults?: number /** Maximum paths retained in one `@` workspace index. */ @@ -72,8 +72,8 @@ const maxResumeOptionsSchema = z.number().step(1).min(1).default(8) const questionDialogWidthSchema = z.number().step(1).min(20).default(200) const questionDialogMaxHeightSchema = z.number().step(1).min(6).default(20) const modelDialogWidthSchema = z.number().step(1).min(20).default(76) -const detailsDialogWidthSchema = z.number().step(1).min(20).default(72) const modelDialogMaxHeightSchema = z.number().step(1).min(6).default(20) +const detailsDialogWidthSchema = z.number().step(1).min(20).default(72) const fileSearchMaxResultsSchema = z.number().step(1).min(1).default(DEFAULT_FILE_SEARCH_MAX_RESULTS) const fileSearchMaxEntriesSchema = z.number().step(1).min(1).default(DEFAULT_FILE_SEARCH_MAX_ENTRIES) const fileSearchExcludedDirectoriesSchema = z.array(z.string()).default([...DEFAULT_FILE_SEARCH_EXCLUDED_DIRECTORIES]) From 70f37206d2ead01a36217f7fe81ffc1baa451f2e Mon Sep 17 00:00:00 2001 From: Turtle Date: Fri, 31 Jul 2026 19:57:45 +0800 Subject: [PATCH 10/48] fix(app-boot): release the terminal before a fatal load exit A dsh launch whose config failed validation returned the user to a broken shell: typing was invisible and the next command was mangled by a stray Device Attributes reply (1;2;4cecho ...). The Loader mounts entries concurrently, so ui-tui can already hold the terminal (raw mode, bracketed paste, keyboard protocol, plus an in-flight DA query) when a sibling entry rejects on its own config. installFailLoud wrote its diagnostic and exited immediately, so nothing disposed the tree and ProcessTerminal.stop() never ran. Give installFailLoud an optional release teardown, awaited between the diagnostic and the exit and bounded by FAIL_LOUD_RELEASE_TIMEOUT_MS. The TUI launcher passes one that disposes the root context, reaching the same shutdown() the /exit path already uses (drainInput() + ui.stop()). The context is captured in boot()'s prepare hook because the rejection arrives while boot() is still in flight. Bins that pass no release keep the previous behavior exactly. --- ...-fail-loud-releases-the-terminal.i18n.yaml | 6 ++ ...6-07-31-fail-loud-releases-the-terminal.md | 57 +++++++++++++++++++ ...7-31-fail-loud-releases-the-terminal.zh.md | 57 +++++++++++++++++++ apps/cli/src/tui.ts | 16 +++++- packages/ui/app-boot/README.i18n.yaml | 4 +- packages/ui/app-boot/README.md | 5 +- packages/ui/app-boot/README.zh.md | 5 +- packages/ui/app-boot/src/index.ts | 51 ++++++++++++++++- packages/ui/app-boot/tests/app-boot.spec.ts | 53 ++++++++++++++++- 9 files changed, 245 insertions(+), 9 deletions(-) create mode 100644 .agents/notes/implemented/bug-fix/2026-07-31-fail-loud-releases-the-terminal.i18n.yaml create mode 100644 .agents/notes/implemented/bug-fix/2026-07-31-fail-loud-releases-the-terminal.md create mode 100644 .agents/notes/implemented/bug-fix/2026-07-31-fail-loud-releases-the-terminal.zh.md diff --git a/.agents/notes/implemented/bug-fix/2026-07-31-fail-loud-releases-the-terminal.i18n.yaml b/.agents/notes/implemented/bug-fix/2026-07-31-fail-loud-releases-the-terminal.i18n.yaml new file mode 100644 index 0000000000..13949d3b73 --- /dev/null +++ b/.agents/notes/implemented/bug-fix/2026-07-31-fail-loud-releases-the-terminal.i18n.yaml @@ -0,0 +1,6 @@ +# Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each +# side as of the last confirmed-consistent state. Both languages carry equal authority; +# after editing either side, bring the other along and re-record with: +# pnpm run verify-translation-pairing --write .agents/notes/implemented/bug-fix/2026-07-31-fail-loud-releases-the-terminal.md +2026-07-31-fail-loud-releases-the-terminal.md: 410e89a1f172f2c7a37016aa6ac023e9cb80d153 +2026-07-31-fail-loud-releases-the-terminal.zh.md: 678834d8705eb6ce7ad52560a0ec255b4ea518a1 diff --git a/.agents/notes/implemented/bug-fix/2026-07-31-fail-loud-releases-the-terminal.md b/.agents/notes/implemented/bug-fix/2026-07-31-fail-loud-releases-the-terminal.md new file mode 100644 index 0000000000..410e89a1f1 --- /dev/null +++ b/.agents/notes/implemented/bug-fix/2026-07-31-fail-loud-releases-the-terminal.md @@ -0,0 +1,57 @@ +# Agent Note: fail-loud releases the terminal before exiting + +Status: implemented + +English | [中文](2026-07-31-fail-loud-releases-the-terminal.zh.md) + +## Problem + +A `dsh` launch whose config failed validation printed its diagnostic and returned the user to a broken shell. Typing was invisible, and the next command was mangled by stray text: + +``` +dsh: fatal load failure: ValidationError: invalid config: + - $.providers expected object but got [object Object] (at providers) +$ 1;2;4cecho hello +zsh: command not found: 4cecho +``` + +The Loader mounts entries concurrently, so entry failure order is not startup order. `ui-tui` activates and calls pi-tui's `ProcessTerminal.start()`, which puts stdin in raw mode, enables bracketed paste, and writes the Kitty keyboard-protocol probe — a sequence ending in a Device Attributes query (`ESC [ c`). A sibling entry (here `llm-pi-ai`) then rejects on its own config. That rejection surfaces as an unhandled rejection, and `installFailLoud` wrote one stderr line and called `process.exit(1)` immediately. + +Nothing disposed the tree, so `ProcessTerminal.stop()` never ran: raw mode, bracketed paste, and the keyboard protocol stayed set on the shell that outlived the process. The terminal's answer to the Device Attributes query (`1;2;4c`) arrived after exit and was read by the shell as typed input — the literal text above. + +The `/exit` path was never affected, because it disposes the tree and reaches the TUI's own `shutdown()`, which calls `drainInput()` (absorbing the pending reply) and then `ui.stop()`. The defect was that a *failed boot* had no path to that same teardown. + +## Decision + +`installFailLoud` takes an optional `release` teardown, awaited between the diagnostic and the exit: + +- The diagnostic is written **before** the release, so the reason survives a disposer that repaints or clears the screen. +- The handler uninstalls itself before releasing. Teardown runs plugin disposers that may themselves reject, and a re-entered handler would report a cleanup failure as a second fatal load failure, burying the real one. +- The release is bounded by `FAIL_LOUD_RELEASE_TIMEOUT_MS` (2s) and its rejection is swallowed. A wedged or failing disposer delays the fatal exit; it never cancels it. +- Omitting `release` keeps the previous behavior exactly, so the ACP, JSON-RPC, and demo bins are unchanged. + +`dsh`'s TUI launcher passes a release that disposes the root context, which runs the TUI's existing `shutdown()` and hands the terminal back. + +The launcher captures the root context in `boot()`'s `prepare` hook rather than from its return value. The rejection arrives while `boot()` is still in flight, so `app.current` assigned after the `await` would still be `undefined` at exactly the moment the hook needs it. `prepare` runs after the Loader installs and before any config-tree entry mounts, which covers the whole window in which an entry can reject. + +## Alternatives considered + +**Reset the terminal from the fail-loud handler** (write `ESC [ ? 2004 l`, pop the keyboard protocol, clear raw mode). This duplicates pi-tui's teardown in a package that owns no terminal, and would drift as pi-tui's startup sequence changes. It also cannot absorb the in-flight Device Attributes reply, which is what corrupts the next prompt — only draining stdin while it is still raw does that. + +**Register a `process.on('exit')` terminal reset in the TUI.** Exit handlers are synchronous, so they cannot await `drainInput()`; the stray reply would still land. It also puts teardown on a global hook rather than the disposal path that already exists. + +**Have the TUI refuse to start until the tree settles.** This serializes a deliberately concurrent Loader and delays first paint for every healthy launch to fix a failure path. + +**Reorder config entries so `llm-pi-ai` mounts before `ui-tui`.** Ordering is not a guarantee the Loader makes, and any future entry could fail after the TUI mounts. + +## Consequences + +A failed boot now costs one tree disposal (bounded at 2s) before exit, and the exit code stays 1. In exchange, a misconfigured `dsh` returns a usable shell instead of one needing `stty sane` or `reset`. + +The guarantee belongs to whichever bin owns the terminal: a surface that grabs terminal state and does not pass `release` reintroduces this defect. `installFailLoud` cannot detect that on its own, since it has no view of what a mounted plugin did to the process. + +## Testing + +`packages/ui/app-boot/tests/app-boot.spec.ts` covers the release contract: the hook is awaited before the exit commits, a rejecting hook still exits 1, a never-settling hook exits after `FAIL_LOUD_RELEASE_TIMEOUT_MS` under fake timers, and the handler is uninstalled before releasing so teardown cannot re-enter it. + +The end-to-end symptom is terminal state after process exit — what the *shell* sees once `dsh` is gone — which no in-process assertion observes. It was verified manually in tmux against a config with a list-shaped `providers` value: before the change the next command was mangled (`zsh: command not found: 4cecho`); after it, the diagnostic is intact, the exit code is 1, and the next command runs normally. The `/exit` path was re-checked to confirm the goodbye line and exit code 0 are unchanged. diff --git a/.agents/notes/implemented/bug-fix/2026-07-31-fail-loud-releases-the-terminal.zh.md b/.agents/notes/implemented/bug-fix/2026-07-31-fail-loud-releases-the-terminal.zh.md new file mode 100644 index 0000000000..678834d870 --- /dev/null +++ b/.agents/notes/implemented/bug-fix/2026-07-31-fail-loud-releases-the-terminal.zh.md @@ -0,0 +1,57 @@ +# Agent Note：fail-loud 在退出前释放终端 + +Status: implemented + +[English](2026-07-31-fail-loud-releases-the-terminal.md) | 中文 + +## Problem + +配置校验失败的 `dsh` 启动会打印诊断信息，然后把用户丢回一个损坏的 shell：输入不可见，下一条命令还会被残留文本弄乱： + +``` +dsh: fatal load failure: ValidationError: invalid config: + - $.providers expected object but got [object Object] (at providers) +$ 1;2;4cecho hello +zsh: command not found: 4cecho +``` + +Loader 并发挂载各个条目，因此条目失败的顺序并不等于启动顺序。`ui-tui` 会先激活并调用 pi-tui 的 `ProcessTerminal.start()`，它把 stdin 置为 raw 模式、启用 bracketed paste，并写出 Kitty 键盘协议探测序列——该序列以一个 Device Attributes 查询（`ESC [ c`）结尾。随后某个同级条目（这里是 `llm-pi-ai`）因自身配置而 rejection。 + +该 rejection 以未处理 rejection 的形式浮现，而 `installFailLoud` 只写一行 stderr 就立即调用 `process.exit(1)`。没有任何环节释放这棵树，因此 `ProcessTerminal.stop()` 从未执行：raw 模式、bracketed paste 和键盘协议都残留在比进程活得更久的 shell 上。终端对 Device Attributes 查询的回应（`1;2;4c`）在进程退出之后才到达，被 shell 当作用户输入读入——也就是上面那段字面文本。 + +`/exit` 路径从不受影响，因为它会释放整棵树，从而进入 TUI 自身的 `shutdown()`：先 `drainInput()`（吸收尚未返回的响应），再 `ui.stop()`。缺陷在于**启动失败**没有通往这同一套拆卸流程的路径。 + +## Decision + +`installFailLoud` 新增可选的 `release` 拆卸回调，在诊断信息与退出之间被等待： + +- 诊断信息在 release **之前**写出，因此即使 disposer 重绘或清屏，失败原因也不会丢失。 +- 处理函数在 release 之前先卸载自己。拆卸会执行插件 disposer，其自身可能 rejection；若处理函数被重入，就会把清理失败报告成第二次致命加载失败，从而掩盖真正的原因。 +- release 以 `FAIL_LOUD_RELEASE_TIMEOUT_MS`（2 秒）为上限，且其 rejection 被吞掉。卡住或失败的 disposer 只会延迟致命退出，绝不会取消它。 +- 不传 `release` 时行为与此前完全一致，因此 ACP、JSON-RPC 和各 demo bin 均无变化。 + +`dsh` 的 TUI 启动器传入的 release 会释放根上下文，从而执行 TUI 已有的 `shutdown()` 并把终端交还。 + +启动器在 `boot()` 的 `prepare` 回调中捕获根上下文，而不是取其返回值。rejection 到达时 `boot()` 尚未结算，因此在 `await` 之后赋值的 `app.current` 恰好在回调需要它的那一刻仍是 `undefined`。`prepare` 在 Loader 安装之后、任何配置树条目挂载之前运行，覆盖了条目可能 rejection 的整个窗口。 + +## Alternatives considered + +**在 fail-loud 处理函数里直接重置终端**（写 `ESC [ ? 2004 l`、弹出键盘协议、清除 raw 模式）。这会在一个并不拥有终端的包里重复 pi-tui 的拆卸逻辑，并随 pi-tui 启动序列的变化而漂移。它同样无法吸收尚未返回的 Device Attributes 响应——而这正是弄乱下一个提示符的原因，只有在 stdin 仍处于 raw 模式时排空它才能解决。 + +**在 TUI 中注册 `process.on('exit')` 终端重置。** exit 处理函数是同步的，无法等待 `drainInput()`，残留响应依旧会落到 shell；而且这把拆卸挂到全局钩子上，而非已经存在的释放路径。 + +**让 TUI 等整棵树结算后再启动。** 这会把刻意并发的 Loader 串行化，并为修复一条失败路径而拖慢每一次正常启动的首次绘制。 + +**调整配置顺序，让 `llm-pi-ai` 先于 `ui-tui` 挂载。** 顺序并不是 Loader 提供的保证，而且未来任何条目都可能在 TUI 挂载之后失败。 + +## Consequences + +启动失败现在会在退出前多付出一次树释放的代价（上限 2 秒），退出码仍为 1。作为交换，配置错误的 `dsh` 会交还一个可用的 shell，而不是需要 `stty sane` 或 `reset` 才能恢复的终端。 + +这项保证属于**拥有终端的那个 bin**：任何抢占终端状态却不传 `release` 的界面都会重新引入该缺陷。`installFailLoud` 自身无法察觉这一点，因为它看不到已挂载的插件对进程做了什么。 + +## Testing + +`packages/ui/app-boot/tests/app-boot.spec.ts` 覆盖 release 契约：退出提交前会等待该回调；回调 rejection 时仍退出 1；在 fake timers 下，永不结算的回调会在 `FAIL_LOUD_RELEASE_TIMEOUT_MS` 后退出；以及处理函数在 release 之前已卸载，使拆卸无法重入它。 + +端到端症状是**进程退出之后**的终端状态——即 `dsh` 消失后 shell 所看到的东西——没有任何进程内断言能观测到它。该症状在 tmux 中针对 `providers` 为列表形状的配置手工验证：修复前下一条命令会被弄乱（`zsh: command not found: 4cecho`），修复后诊断信息完整、退出码为 1、下一条命令正常执行。同时复查了 `/exit` 路径，确认告别行与退出码 0 均未改变。 diff --git a/apps/cli/src/tui.ts b/apps/cli/src/tui.ts index 15e7d6f77b..5e19dc9c6d 100644 --- a/apps/cli/src/tui.ts +++ b/apps/cli/src/tui.ts @@ -113,7 +113,6 @@ export async function runTui( ) process.exit(1) } - installFailLoud(NAME) // The bin already loaded the invoking directory's .env, and that is the // whole environment: $DSH_HOME/.env is credentials-local's writable store, // and hoisting it would make every stored key read as a read-only ambient @@ -140,6 +139,17 @@ export async function runTui( const entry = process.argv[1] const execve = process.execve?.bind(process) const app: { current?: Context } = {} + // The Loader mounts entries concurrently, so `ui-tui` can already hold the + // terminal (raw mode, bracketed paste, keyboard protocol) when a sibling + // entry rejects — and that rejection arrives while `boot` is still in + // flight. Disposing the tree runs the TUI's own shutdown, which stops the + // terminal and hands the shell back; without it a failed boot returns to a + // corrupted prompt. `app.current` is captured from boot's `prepare` hook, so + // it holds the root context for the whole mounting window rather than only + // after boot resolves. + installFailLoud(NAME, process, async () => { + await app.current?.fiber.dispose() + }) // Resume always enters the default surface because experimental-meta rejects // parent options, including `--resume`. The resumed session already persists // its cwd. @@ -216,6 +226,10 @@ export async function runTui( bootConfig, patches, (hostCtx) => { + // Runs after the Loader installs and before any config-tree entry mounts, + // so the fail-loud release hook can reach the tree for the whole window in + // which an entry may reject. + app.current = hostCtx // The launcher owns session identity and the exit line: a config-mounted // app bundle reads both from these slots, so no cordis.yml key can drop // resume. diff --git a/packages/ui/app-boot/README.i18n.yaml b/packages/ui/app-boot/README.i18n.yaml index 75b1ec16b2..c0e13b0cf2 100644 --- a/packages/ui/app-boot/README.i18n.yaml +++ b/packages/ui/app-boot/README.i18n.yaml @@ -2,5 +2,5 @@ # side as of the last confirmed-consistent state. Both languages carry equal authority; # after editing either side, bring the other along and re-record with: # pnpm run verify-translation-pairing --write packages/ui/app-boot/README.md -README.md: ebd8e0842b934f6887e3c122e781c1d0f13bb5d3 -README.zh.md: ccd897d48178482aa74d0eb73505e26ec3a08d6c +README.md: ba5cf9a05b456e2d72abe1e2a65b64825ceef1a5 +README.zh.md: d2f2b2d2c93b1ecb9fb4fad085d4abd663440108 diff --git a/packages/ui/app-boot/README.md b/packages/ui/app-boot/README.md index ebd8e0842b..ba5cf9a05b 100644 --- a/packages/ui/app-boot/README.md +++ b/packages/ui/app-boot/README.md @@ -8,7 +8,8 @@ Shared boot glue for the app bins ([`dsh`](../../../apps/cli/README.md), [`dsh-c |---|---| | `resolveConfigPath(path, snapshotMode, cwd?)` | Absolute config path; `snapshotMode === 'replay'` swaps a `cordis.yml`/`.yaml` basename for its sibling `cordis.snapshot.yml` | | `loadEnv(binName, dir?, warn?)` | Load the gitignored `.env` (Node `process.loadEnvFile`); absent file is fine, an unloadable one warns a single labelled line (default: stderr) | -| `installFailLoud(binName, proc?)` | Turn a post-`boot()` unhandled Loader rejection into one labelled stderr line + `exit(1)`; returns the uninstaller (for tests) | +| `installFailLoud(binName, proc?, release?)` | Turn a post-`boot()` unhandled Loader rejection into one labelled stderr line + `exit(1)`; the optional `release` teardown is awaited between the two (bounded by `FAIL_LOUD_RELEASE_TIMEOUT_MS`) so a terminal-owning surface restores the terminal before exit; returns the uninstaller (for tests) | +| `FAIL_LOUD_RELEASE_TIMEOUT_MS` | How long `installFailLoud` waits for its `release` hook; a wedged disposer delays the fatal exit, never cancels it | | `assertEntriesLoaded(ctx, binName)` | Throw when a settled tree holds an enabled entry with no fiber, reporting every unresolved plugin name as a Cordis startup failure | | `assertEntriesActivated(ctx, binName)` | Include the `assertEntriesLoaded` check, then await every enabled entry after the Loader settles; throw with each failed plugin's original stack or each pending plugin's unresolved services | | `loadPersonalPatches(binName, dir?)` | Parse the optional `config.yaml` in the Harness home (default [`resolveDshHome()`](../../util/paths/README.md): `$DSH_HOME`, else `~/.dsh`) — a top-level YAML array of include `PatchOptions` (id-targeted config overrides, `insert` lists, `!!js` allowed); absent file → `undefined`, an unreadable/unparsable/non-array file throws | @@ -20,6 +21,8 @@ Shared boot glue for the app bins ([`dsh`](../../../apps/cli/README.md), [`dsh-c Two Loader failure classes require separate guards because tree settlement propagates neither to its caller. A failed plugin import leaves a fiber-less entry that `assertEntriesLoaded` turns into a `boot()` rejection naming every unresolved plugin. A plugin callback or config failure leaves a failed fiber because `loader.await()` settles lifecycle tasks without propagating that error; `assertEntriesActivated` awaits the fiber explicitly and includes its original stack in the startup rejection. Before throwing, the audit marks those exact rejection reasons through one process checkpoint so `installFailLoud` coalesces Loader's duplicate notification while every unrelated unhandled rejection remains fatal. +The Loader mounts entries concurrently, so a surface can already own the terminal when a sibling entry rejects: exiting straight from the handler would leave raw mode, bracketed paste, and the keyboard protocol set on the user's shell, and an in-flight terminal query's reply would land as literal text at the next prompt. A terminal-owning bin therefore passes `release` to dispose the tree — running that surface's own shutdown — before the exit commits. `dsh` captures the root context in `boot()`'s `prepare` hook rather than from its return value, because the rejection arrives while `boot()` is still in flight. + Bare plugin specifiers in a config (`@deepseek-ai/dsh-*`, npm packages) resolve through the Cordis Loader's internal module loader. Repository bins install Loader's optional `node-addon-require-builtin` peer; external callers must supply it or install plugins where plain Node import resolution can find them. Relative specifiers resolve against the config directory without the native helper. The built `dsh-app-boot` artifact embeds the statically mounted Include implementation while leaving Loader external, so the include tree and host bind to one Loader peer. The `dsh` source launcher additionally maps manifest-declared workspace packages to their TypeScript source; its configuration gate requires every TUI/Web bare plugin to appear in the resolver manifest's `dependencies`. The bins' subprocess smokes exercise the internal-loader path, while this package's unit suite drives `boot()` in-process against configs with relative specifiers. This package carries no loader hooks and no dev-mode surface. The [`dsh` app](../../../apps/cli/README.md) owns its Node source-launch hook and consumes these helpers for the boot sequence; built consumers continue to use plain Node package resolution. diff --git a/packages/ui/app-boot/README.zh.md b/packages/ui/app-boot/README.zh.md index ccd897d481..d2f2b2d2c9 100644 --- a/packages/ui/app-boot/README.zh.md +++ b/packages/ui/app-boot/README.zh.md @@ -8,7 +8,8 @@ |---|---| | `resolveConfigPath(path, snapshotMode, cwd?)` | 生成绝对配置路径；当 `snapshotMode === 'replay'` 时，把 basename 为 `cordis.yml`/`.yaml` 的文件替换为同级 `cordis.snapshot.yml` | | `loadEnv(binName, dir?, warn?)` | 加载已被 git 忽略的 `.env`（Node `process.loadEnvFile`）；文件不存在不影响启动，文件无法加载时输出一行带标签的警告（默认写入 stderr） | -| `installFailLoud(binName, proc?)` | 将 `boot()` 之后未处理的 Loader rejection 转换为一行带标签的 stderr 消息并执行 `exit(1)`；返回卸载函数（供测试使用） | +| `installFailLoud(binName, proc?, release?)` | 将 `boot()` 之后未处理的 Loader rejection 转换为一行带标签的 stderr 消息并执行 `exit(1)`；两者之间会等待可选的 `release` 拆卸回调（以 `FAIL_LOUD_RELEASE_TIMEOUT_MS` 为上限），使持有终端的界面能在退出前恢复终端；返回卸载函数（供测试使用） | +| `FAIL_LOUD_RELEASE_TIMEOUT_MS` | `installFailLoud` 等待其 `release` 回调的时长；卡住的 disposer 只会延迟致命退出，而不会取消它 | | `assertEntriesLoaded(ctx, binName)` | 树结算后，如果其中存在已启用但没有 fiber 的条目，则抛出异常，并以 Cordis 启动故障的形式报告每个未解析插件的名称 | | `assertEntriesActivated(ctx, binName)` | 先执行 `assertEntriesLoaded` 检查，再在 Loader 结算后等待每个已启用配置项；抛出的错误包含每个失败插件的原始错误堆栈，或每个等待中插件尚未解析的服务 | | `loadPersonalPatches(binName, dir?)` | 解析 Harness home 中可选的 `config.yaml`（默认使用 [`resolveDshHome()`](../../util/paths/README.md)：先取 `$DSH_HOME`，否则取 `~/.dsh`）：其顶层是一个 YAML 数组，内容为 include 的 `PatchOptions`（按 id 定位的配置覆盖、`insert` 列表，允许 `!!js`）；文件不存在时返回 `undefined`，文件不可读、不可解析或内容不是数组时抛出异常 | @@ -20,6 +21,8 @@ Loader 树结算不会向调用方传播两类故障，因此需要分别保护。插件导入失败会留下没有 fiber 的配置项，`assertEntriesLoaded` 将其转换为 `boot()` rejection，并列出每个未解析插件。插件回调或配置失败则会留下失败的 fiber，因为 `loader.await()` 只结算生命周期任务，不传播该错误；`assertEntriesActivated` 会显式等待该 fiber，并把原始错误堆栈写入启动 rejection。抛出错误前，审计会通过一个进程级检查点标记这些 rejection 的确切原因，从而让 `installFailLoud` 将 Loader 的重复通知合并为一次，而所有无关的未处理 rejection 仍然致命。 +Loader 并发挂载各个条目，因此当某个同级条目 rejection 时，某个界面可能已经持有终端：此时直接从处理函数退出，会把 raw 模式、bracketed paste 和键盘协议残留在用户的 shell 上，而尚未返回的终端查询响应会在下一个提示符处显示为字面文本。因此，持有终端的 bin 会传入 `release` 来释放整棵树——执行该界面自身的 shutdown——然后才提交退出。`dsh` 在 `boot()` 的 `prepare` 回调中捕获根上下文，而不是取其返回值，因为 rejection 到达时 `boot()` 尚未结算。 + 配置中的裸插件 specifier（`@deepseek-ai/dsh-*`、npm 包（package））通过 Cordis Loader 的内部模块 loader 解析。仓库 bin 会安装 Loader 的可选 peer `node-addon-require-builtin`；外部调用方必须提供该组件，或者把插件安装到普通 Node import 解析可以找到的位置。相对 specifier 无需原生 helper，并以配置目录为基准解析。构建后的 `dsh-app-boot` 产物内嵌静态挂载的 Include 实现，但仍将 Loader 保持为外部依赖，因此 include 树与 host 会绑定到同一个 Loader peer。`dsh` 源码启动器还会将 manifest（元数据清单）声明的 workspace 包映射到其 TypeScript 源码；其配置门禁要求每个 TUI／Web 裸插件都出现在解析所用 manifest 的 `dependencies` 中。bin 的子进程冒烟测试覆盖内部 loader 路径，而本包的单元测试套件会在进程内使用相对 specifier 配置驱动 `boot()`。 此包不包含 loader 钩子，也不提供开发模式接口。[`dsh` 应用](../../../apps/cli/README.md)持有自己的 Node 源码启动钩子，并在启动序列中使用这些 helper；构建后的消费方仍使用普通 Node 包解析。 diff --git a/packages/ui/app-boot/src/index.ts b/packages/ui/app-boot/src/index.ts index 982bcc59ed..ec8f717f45 100644 --- a/packages/ui/app-boot/src/index.ts +++ b/packages/ui/app-boot/src/index.ts @@ -325,24 +325,69 @@ async function observeLoaderRejectionCheckpoint(reasons: readonly unknown[]): Pr } } +/** + * How long {@link installFailLoud} waits for its `release` hook before exiting + * anyway. A wedged disposer must delay the fatal exit, never cancel it. + */ +export const FAIL_LOUD_RELEASE_TIMEOUT_MS = 2_000 + /** * Install before boot to turn a late unhandled plugin-init rejection into one * labelled stderr diagnostic and `exit(1)`. A rejection already included by * {@link assertEntriesActivated} is ignored during its process checkpoint; * every other rejection remains fatal. Stdout remains untouched for ACP; the * returned function removes the handler. + * + * The Loader mounts entries concurrently, so a surface that owns the terminal + * can already hold it when a sibling entry rejects. Exiting straight from the + * handler would strand raw mode, bracketed paste, and the keyboard protocol on + * the user's shell, and leave an in-flight terminal query's reply to land as + * literal text at the next prompt. `release` is the terminal owner's chance to + * hand it back; it is awaited under {@link FAIL_LOUD_RELEASE_TIMEOUT_MS}. The + * diagnostic is written before the release so the reason survives a disposer + * that repaints or clears the screen, and the handler uninstalls itself before + * releasing so a rejection from teardown cannot re-enter it. * @param binName - the diagnostic prefix on the fatal-failure line. * @param proc - the process slice to register on; tests inject a fake. + * @param release - optional teardown awaited before exit, used by a + * terminal-owning surface to restore the terminal. Its own failure is + * swallowed because the pending fatal exit already owns the outcome. * @returns the uninstaller that removes the rejection handler. */ -export function installFailLoud(binName: string, proc: FailLoudProcess = process): () => void { +export function installFailLoud( + binName: string, + proc: FailLoudProcess = process, + release?: () => Promise | void, +): () => void { const handler = (err: unknown): void => { if (assembledActivationRejections.has(err)) return proc.stderr.write(`${binName}: fatal load failure: ${err instanceof Error ? err.stack ?? err.message : String(err)}\n`) - proc.exit(1) + if (release === undefined) { + proc.exit(1) + return + } + // The release runs plugin disposers, which may themselves reject. Without + // this the handler would re-enter and report a teardown failure as a second + // fatal load failure, hiding the real one. + uninstall() + void (async () => { + try { + await Promise.race([ + (async () => release())(), + new Promise((resolve) => { + setTimeout(resolve, FAIL_LOUD_RELEASE_TIMEOUT_MS).unref() + }), + ]) + } catch { + // The terminal release failed; the fatal exit below is the outcome that + // matters, and no reporter runs after it. + } + proc.exit(1) + })() } + const uninstall = (): void => void proc.off('unhandledRejection', handler) proc.on('unhandledRejection', handler) - return () => void proc.off('unhandledRejection', handler) + return uninstall } /** diff --git a/packages/ui/app-boot/tests/app-boot.spec.ts b/packages/ui/app-boot/tests/app-boot.spec.ts index 7f06016267..c47b16cc78 100644 --- a/packages/ui/app-boot/tests/app-boot.spec.ts +++ b/packages/ui/app-boot/tests/app-boot.spec.ts @@ -5,7 +5,8 @@ import { describe, expect, it, vi } from 'vitest' import { Context } from 'cordis' import SystemPrompt, { renderPrompt } from '@deepseek-ai/dsh-system-prompt' import { - addHarnessSourceSection, assertEntriesActivated, assertEntriesLoaded, boot, HARNESS_SOURCE_SECTION, + addHarnessSourceSection, assertEntriesActivated, assertEntriesLoaded, boot, + FAIL_LOUD_RELEASE_TIMEOUT_MS, HARNESS_SOURCE_SECTION, installFailLoud, loadEnv, loadOverlayPatches, resolveConfigPath, type FailLoudProcess, } from '../src/index.ts' @@ -162,6 +163,56 @@ describe('installFailLoud', () => { proc.handlers[0]!(error) expect(proc.exits).toEqual([1]) }) + + // The Loader mounts entries concurrently, so a terminal-owning surface can + // already hold raw mode when a sibling entry rejects. Exiting without running + // its teardown strands the terminal on the user's shell. + it('awaits the release hook before exiting so the terminal owner can restore it', async () => { + const proc = fakeProc() + const order: string[] = [] + installFailLoud(NAME, proc, async () => { + await Promise.resolve() + order.push('released') + }) + proc.handlers[0]!(new Error('sibling entry rejected')) + expect(proc.written[0]).toContain(`${NAME}: fatal load failure: `) + // The release is in flight, so the exit has not committed yet. + expect(proc.exits).toEqual([]) + await vi.waitFor(() => { expect(proc.exits).toEqual([1]) }) + expect(order).toEqual(['released']) + }) + + it('still exits when the release hook rejects', async () => { + const proc = fakeProc() + installFailLoud(NAME, proc, () => Promise.reject(new Error('terminal stop failed'))) + proc.handlers[0]!(new Error('boom')) + await vi.waitFor(() => { expect(proc.exits).toEqual([1]) }) + }) + + it('exits without waiting when a release hook never settles', async () => { + vi.useFakeTimers() + try { + const proc = fakeProc() + installFailLoud(NAME, proc, () => new Promise(() => {})) + proc.handlers[0]!(new Error('boom')) + expect(proc.exits).toEqual([]) + await vi.advanceTimersByTimeAsync(FAIL_LOUD_RELEASE_TIMEOUT_MS) + expect(proc.exits).toEqual([1]) + } finally { + vi.useRealTimers() + } + }) + + // Teardown runs plugin disposers, whose own rejection must not be reported as + // a second fatal load failure over the real one. + it('uninstalls the handler before releasing, so teardown cannot re-enter it', async () => { + const proc = fakeProc() + installFailLoud(NAME, proc, () => {}) + proc.handlers[0]!(new Error('boom')) + expect(proc.handlers).toHaveLength(0) + await vi.waitFor(() => { expect(proc.exits).toEqual([1]) }) + expect(proc.written).toHaveLength(1) + }) }) describe('assertEntriesLoaded', () => { From b35b06396def4a4a1e5770c1b389cc3f3f9cd4d1 Mon Sep 17 00:00:00 2001 From: Turtle Date: Fri, 31 Jul 2026 20:19:58 +0800 Subject: [PATCH 11/48] fix(app-boot): keep the fail-loud exit fatal while the terminal is released Review of the previous commit found two defects in the release path, both reproduced against the implementation: - The timeout guarding a never-settling release was unref'ed. An unhandledRejection listener suppresses Node's default fatal exit, so with nothing else referenced the process reached an empty event loop and exited 0 on the very failure it was reporting. Keep the timer referenced and clear it once the race settles. - The handler uninstalled itself before awaiting the release. A second concurrent rejection then became uncaught and killed the process mid-teardown, stranding exactly the terminal state this restores. Replace the uninstall with a latch: the first rejection is the reported one, and later rejections (teardown's own included) fall through to the pending exit. Add the PTY regression the fake-process tests cannot express: boot the shipped tree over a fixture whose llm-pi-ai providers value is list-shaped, expect exit 1, and assert the captured bytes carry both the diagnostic and ESC[?2004l. Against the pre-fix source the stream ends at ESC[?2004h ESC[>7u ESC[?u ESC[c with no reset and the case fails, so it pins the actual bug. Split the two-shape formatting test into one install per case; a latched handler reports once by design. --- ...-fail-loud-releases-the-terminal.i18n.yaml | 4 +- ...6-07-31-fail-loud-releases-the-terminal.md | 12 +++--- ...7-31-fail-loud-releases-the-terminal.zh.md | 12 +++--- .../fixtures/tui-invalid-provider.cordis.yml | 10 +++++ apps/cli/tests/tui-keyless-smoke.e2e.ts | 22 ++++++++++ packages/ui/app-boot/README.i18n.yaml | 4 +- packages/ui/app-boot/README.md | 2 +- packages/ui/app-boot/README.zh.md | 2 +- packages/ui/app-boot/src/index.ts | 28 +++++++++---- packages/ui/app-boot/tests/app-boot.spec.ts | 42 ++++++++++++------- 10 files changed, 99 insertions(+), 39 deletions(-) create mode 100644 apps/cli/tests/fixtures/tui-invalid-provider.cordis.yml diff --git a/.agents/notes/implemented/bug-fix/2026-07-31-fail-loud-releases-the-terminal.i18n.yaml b/.agents/notes/implemented/bug-fix/2026-07-31-fail-loud-releases-the-terminal.i18n.yaml index 13949d3b73..97dc84403b 100644 --- a/.agents/notes/implemented/bug-fix/2026-07-31-fail-loud-releases-the-terminal.i18n.yaml +++ b/.agents/notes/implemented/bug-fix/2026-07-31-fail-loud-releases-the-terminal.i18n.yaml @@ -2,5 +2,5 @@ # side as of the last confirmed-consistent state. Both languages carry equal authority; # after editing either side, bring the other along and re-record with: # pnpm run verify-translation-pairing --write .agents/notes/implemented/bug-fix/2026-07-31-fail-loud-releases-the-terminal.md -2026-07-31-fail-loud-releases-the-terminal.md: 410e89a1f172f2c7a37016aa6ac023e9cb80d153 -2026-07-31-fail-loud-releases-the-terminal.zh.md: 678834d8705eb6ce7ad52560a0ec255b4ea518a1 +2026-07-31-fail-loud-releases-the-terminal.md: ccac625171ef5523a4ed27843b543c838bf43ce8 +2026-07-31-fail-loud-releases-the-terminal.zh.md: fe8a3271b26a94b9d986b8d17744893e5f448593 diff --git a/.agents/notes/implemented/bug-fix/2026-07-31-fail-loud-releases-the-terminal.md b/.agents/notes/implemented/bug-fix/2026-07-31-fail-loud-releases-the-terminal.md index 410e89a1f1..ccac625171 100644 --- a/.agents/notes/implemented/bug-fix/2026-07-31-fail-loud-releases-the-terminal.md +++ b/.agents/notes/implemented/bug-fix/2026-07-31-fail-loud-releases-the-terminal.md @@ -25,9 +25,9 @@ The `/exit` path was never affected, because it disposes the tree and reaches th `installFailLoud` takes an optional `release` teardown, awaited between the diagnostic and the exit: -- The diagnostic is written **before** the release, so the reason survives a disposer that repaints or clears the screen. -- The handler uninstalls itself before releasing. Teardown runs plugin disposers that may themselves reject, and a re-entered handler would report a cleanup failure as a second fatal load failure, burying the real one. -- The release is bounded by `FAIL_LOUD_RELEASE_TIMEOUT_MS` (2s) and its rejection is swallowed. A wedged or failing disposer delays the fatal exit; it never cancels it. +- The diagnostic is written **before** the release, so a hanging or failing disposer cannot swallow the reason. +- A latch, not an uninstall, keeps the first rejection the reported one. Removing the listener during teardown would let a second concurrent rejection become uncaught, and Node would kill the process mid-teardown — stranding exactly the terminal state this restores. Later rejections, including the release's own, fall through to the pending exit. +- The release is bounded by `FAIL_LOUD_RELEASE_TIMEOUT_MS` (2s) and its rejection is swallowed. A wedged or failing disposer delays the fatal exit; it never cancels it. That timer stays **referenced**: an `unref()`ed one lets Node reach an empty event loop and exit 0 on the very failure being reported, because an `unhandledRejection` listener suppresses the default fatal exit. - Omitting `release` keeps the previous behavior exactly, so the ACP, JSON-RPC, and demo bins are unchanged. `dsh`'s TUI launcher passes a release that disposes the root context, which runs the TUI's existing `shutdown()` and hands the terminal back. @@ -52,6 +52,8 @@ The guarantee belongs to whichever bin owns the terminal: a surface that grabs t ## Testing -`packages/ui/app-boot/tests/app-boot.spec.ts` covers the release contract: the hook is awaited before the exit commits, a rejecting hook still exits 1, a never-settling hook exits after `FAIL_LOUD_RELEASE_TIMEOUT_MS` under fake timers, and the handler is uninstalled before releasing so teardown cannot re-enter it. +`packages/ui/app-boot/tests/app-boot.spec.ts` covers the release contract: the hook is awaited before the exit commits, a rejecting hook still exits 1, a never-settling hook exits after `FAIL_LOUD_RELEASE_TIMEOUT_MS`, and a burst of rejections reports only the first while the release still completes. -The end-to-end symptom is terminal state after process exit — what the *shell* sees once `dsh` is gone — which no in-process assertion observes. It was verified manually in tmux against a config with a list-shaped `providers` value: before the change the next command was mangled (`zsh: command not found: 4cecho`); after it, the diagnostic is intact, the exit code is 1, and the next command runs normally. The `/exit` path was re-checked to confirm the goodbye line and exit code 0 are unchanged. +Those fake-process tests cannot observe the two failure modes that matter most — process exit code with a real event loop, and terminal state after exit — so the regression lives in `apps/cli/tests/tui-keyless-smoke.e2e.ts`. It boots the shipped tree in a real PTY over `fixtures/tui-invalid-provider.cordis.yml` (a list-shaped `providers`, the mistake users actually make), expects exit 1, and asserts the captured bytes contain both the diagnostic and `ESC[?2004l`. Against the pre-fix source the captured stream ends at `ESC[?2004h ESC[>7u ESC[?u ESC[c` with no reset, and the case fails on that assertion. + +Testing policy requires a PTY case whenever terminal teardown changes, and this is it. The `/exit` path keeps its existing assertion that the same reset appears on a clean exit. diff --git a/.agents/notes/implemented/bug-fix/2026-07-31-fail-loud-releases-the-terminal.zh.md b/.agents/notes/implemented/bug-fix/2026-07-31-fail-loud-releases-the-terminal.zh.md index 678834d870..fe8a3271b2 100644 --- a/.agents/notes/implemented/bug-fix/2026-07-31-fail-loud-releases-the-terminal.zh.md +++ b/.agents/notes/implemented/bug-fix/2026-07-31-fail-loud-releases-the-terminal.zh.md @@ -25,9 +25,9 @@ Loader 并发挂载各个条目，因此条目失败的顺序并不等于启动 `installFailLoud` 新增可选的 `release` 拆卸回调，在诊断信息与退出之间被等待： -- 诊断信息在 release **之前**写出，因此即使 disposer 重绘或清屏，失败原因也不会丢失。 -- 处理函数在 release 之前先卸载自己。拆卸会执行插件 disposer，其自身可能 rejection；若处理函数被重入，就会把清理失败报告成第二次致命加载失败，从而掩盖真正的原因。 -- release 以 `FAIL_LOUD_RELEASE_TIMEOUT_MS`（2 秒）为上限，且其 rejection 被吞掉。卡住或失败的 disposer 只会延迟致命退出，绝不会取消它。 +- 诊断信息在 release **之前**写出，因此卡住或失败的 disposer 无法吞掉失败原因。 +- 使用闩锁（latch）而非卸载监听器，来保证被报告的始终是第一个 rejection。若在拆卸期间移除监听器，第二个并发 rejection 就会变成未捕获错误，Node 会在拆卸中途杀死进程——恰好残留下本次要恢复的终端状态。后续 rejection（包括 release 自身的）都会落入已挂起的退出流程。 +- release 以 `FAIL_LOUD_RELEASE_TIMEOUT_MS`（2 秒）为上限，且其 rejection 被吞掉。卡住或失败的 disposer 只会延迟致命退出，绝不会取消它。该定时器保持 **referenced**：一旦 `unref()`，Node 就会在事件循环清空后、恰恰在报告这次失败时以 0 退出，因为 `unhandledRejection` 监听器抑制了默认的致命退出。 - 不传 `release` 时行为与此前完全一致，因此 ACP、JSON-RPC 和各 demo bin 均无变化。 `dsh` 的 TUI 启动器传入的 release 会释放根上下文，从而执行 TUI 已有的 `shutdown()` 并把终端交还。 @@ -52,6 +52,8 @@ Loader 并发挂载各个条目，因此条目失败的顺序并不等于启动 ## Testing -`packages/ui/app-boot/tests/app-boot.spec.ts` 覆盖 release 契约：退出提交前会等待该回调；回调 rejection 时仍退出 1；在 fake timers 下，永不结算的回调会在 `FAIL_LOUD_RELEASE_TIMEOUT_MS` 后退出；以及处理函数在 release 之前已卸载，使拆卸无法重入它。 +`packages/ui/app-boot/tests/app-boot.spec.ts` 覆盖 release 契约：退出提交前会等待该回调；回调 rejection 时仍退出 1；永不结算的回调会在 `FAIL_LOUD_RELEASE_TIMEOUT_MS` 后退出；以及一连串 rejection 只报告第一个，同时 release 仍能跑完。 -端到端症状是**进程退出之后**的终端状态——即 `dsh` 消失后 shell 所看到的东西——没有任何进程内断言能观测到它。该症状在 tmux 中针对 `providers` 为列表形状的配置手工验证：修复前下一条命令会被弄乱（`zsh: command not found: 4cecho`），修复后诊断信息完整、退出码为 1、下一条命令正常执行。同时复查了 `/exit` 路径，确认告别行与退出码 0 均未改变。 +这些基于假进程的测试无法观测到最关键的两种失败形态——真实事件循环下的进程退出码，以及退出之后的终端状态——因此回归用例放在 `apps/cli/tests/tui-keyless-smoke.e2e.ts`。它在真实 PTY 中以 `fixtures/tui-invalid-provider.cordis.yml`（`providers` 为列表形状，正是用户真实会犯的错误）启动出厂配置树，期望退出码为 1，并断言捕获到的字节流同时包含诊断信息与 `ESC[?2004l`。在修复前的源码上，捕获流止于 `ESC[?2004h ESC[>7u ESC[?u ESC[c` 而没有任何重置，该用例正是在这条断言上失败。 + +测试规范要求：只要改动终端拆卸，就必须有 PTY 用例——这就是它。`/exit` 路径保留其原有断言，确认正常退出时同样会出现该重置序列。 diff --git a/apps/cli/tests/fixtures/tui-invalid-provider.cordis.yml b/apps/cli/tests/fixtures/tui-invalid-provider.cordis.yml new file mode 100644 index 0000000000..f03a58d5d7 --- /dev/null +++ b/apps/cli/tests/fixtures/tui-invalid-provider.cordis.yml @@ -0,0 +1,10 @@ +# An overlay whose `llm-pi-ai` config fails schema validation: `providers` is a +# dict keyed by provider name, and a list is the shape users reach for. The +# entry rejects while `ui-tui` — mounted concurrently by the Loader — already +# holds the terminal, which is the boot failure the fail-loud release hook +# exists for. +- id: llm-pi-ai + config: + providers: + - provider: openai + apiKey: keyless-invalid-shape diff --git a/apps/cli/tests/tui-keyless-smoke.e2e.ts b/apps/cli/tests/tui-keyless-smoke.e2e.ts index 7c7ad10141..8a917517a2 100644 --- a/apps/cli/tests/tui-keyless-smoke.e2e.ts +++ b/apps/cli/tests/tui-keyless-smoke.e2e.ts @@ -24,6 +24,9 @@ const dshBinScript = fileURLToPath(new URL('../src/bin.ts', import.meta.url)) // `--config` layers an overlay over the shared base, so the default surface // needs no config argument at all; these are the overlays under test. const scriptedConfigPath = fileURLToPath(new URL('./fixtures/tui-scripted.cordis.yml', import.meta.url)) +// An overlay whose `llm-pi-ai` config fails validation, so an entry rejects +// while the TUI already holds the terminal. +const invalidProviderConfigPath = fileURLToPath(new URL('./fixtures/tui-invalid-provider.cordis.yml', import.meta.url)) const tsconfigPath = fileURLToPath(new URL('../../../tsconfig.json', import.meta.url)) const firstRunSnapshots = fileURLToPath(new URL('./tui-first-run-snapshots/', import.meta.url)) const synchronizedFrameEnd = '\x1b[?2026l' @@ -380,6 +383,25 @@ describe('dsh TUI keyless smoke (real Loader tree in a PTY)', () => { expect(output).toContain('\u001B[?2004l') }, PTY_SMOKE_TEST_TIMEOUT_MS) + // The Loader mounts entries concurrently, so `ui-tui` can already own the + // terminal when a sibling entry rejects on its config. Exiting straight from + // the fail-loud handler left raw mode and bracketed paste set on the user's + // shell, and the pending Device Attributes reply landed there as literal + // text. The launcher's release hook must reach the TUI's own teardown. + it('restores the terminal when a sibling entry fails to validate during boot', async () => { + const output = await smoke({ + label: 'dsh invalid provider config', + tempDirPrefix: 'dsh-tui-invalid-config-', + configPath: invalidProviderConfigPath, + expectedExitCode: 1, + }) + expect(output).toContain('dsh: fatal load failure:') + expect(output).toContain('$.providers') + // Bracketed paste is disabled again, which only `ProcessTerminal.stop()` + // writes — proof the tree was disposed rather than exited out from under. + expect(output).toContain('\u001B[?2004l') + }, LOADER_SMOKE_TEST_TIMEOUT_MS) + it('switches models, streams a response, answers a user-question dialog, and exits cleanly', async () => { const output = await smoke({ label: 'dsh conversation', diff --git a/packages/ui/app-boot/README.i18n.yaml b/packages/ui/app-boot/README.i18n.yaml index c0e13b0cf2..2e88921efc 100644 --- a/packages/ui/app-boot/README.i18n.yaml +++ b/packages/ui/app-boot/README.i18n.yaml @@ -2,5 +2,5 @@ # side as of the last confirmed-consistent state. Both languages carry equal authority; # after editing either side, bring the other along and re-record with: # pnpm run verify-translation-pairing --write packages/ui/app-boot/README.md -README.md: ba5cf9a05b456e2d72abe1e2a65b64825ceef1a5 -README.zh.md: d2f2b2d2c93b1ecb9fb4fad085d4abd663440108 +README.md: 7107ea20e72a6117f957090e753c126106b26663 +README.zh.md: 1e5b0850d7c92cd365adf441c31ee3432f13f7fa diff --git a/packages/ui/app-boot/README.md b/packages/ui/app-boot/README.md index ba5cf9a05b..7107ea20e7 100644 --- a/packages/ui/app-boot/README.md +++ b/packages/ui/app-boot/README.md @@ -21,7 +21,7 @@ Shared boot glue for the app bins ([`dsh`](../../../apps/cli/README.md), [`dsh-c Two Loader failure classes require separate guards because tree settlement propagates neither to its caller. A failed plugin import leaves a fiber-less entry that `assertEntriesLoaded` turns into a `boot()` rejection naming every unresolved plugin. A plugin callback or config failure leaves a failed fiber because `loader.await()` settles lifecycle tasks without propagating that error; `assertEntriesActivated` awaits the fiber explicitly and includes its original stack in the startup rejection. Before throwing, the audit marks those exact rejection reasons through one process checkpoint so `installFailLoud` coalesces Loader's duplicate notification while every unrelated unhandled rejection remains fatal. -The Loader mounts entries concurrently, so a surface can already own the terminal when a sibling entry rejects: exiting straight from the handler would leave raw mode, bracketed paste, and the keyboard protocol set on the user's shell, and an in-flight terminal query's reply would land as literal text at the next prompt. A terminal-owning bin therefore passes `release` to dispose the tree — running that surface's own shutdown — before the exit commits. `dsh` captures the root context in `boot()`'s `prepare` hook rather than from its return value, because the rejection arrives while `boot()` is still in flight. +The Loader mounts entries concurrently, so a surface can already own the terminal when a sibling entry rejects: exiting straight from the handler would leave raw mode, bracketed paste, and the keyboard protocol set on the user's shell, and an in-flight terminal query's reply would land as literal text at the next prompt. A terminal-owning bin therefore passes `release` to dispose the tree — running that surface's own shutdown — before the exit commits. `dsh` captures the root context in `boot()`'s `prepare` hook rather than from its return value, because the rejection arrives while `boot()` is still in flight. While a release is in flight the handler stays installed and latched: the first rejection is the reported one, and later rejections (teardown's own included) are swallowed rather than becoming uncaught and killing the process mid-teardown. Bare plugin specifiers in a config (`@deepseek-ai/dsh-*`, npm packages) resolve through the Cordis Loader's internal module loader. Repository bins install Loader's optional `node-addon-require-builtin` peer; external callers must supply it or install plugins where plain Node import resolution can find them. Relative specifiers resolve against the config directory without the native helper. The built `dsh-app-boot` artifact embeds the statically mounted Include implementation while leaving Loader external, so the include tree and host bind to one Loader peer. The `dsh` source launcher additionally maps manifest-declared workspace packages to their TypeScript source; its configuration gate requires every TUI/Web bare plugin to appear in the resolver manifest's `dependencies`. The bins' subprocess smokes exercise the internal-loader path, while this package's unit suite drives `boot()` in-process against configs with relative specifiers. diff --git a/packages/ui/app-boot/README.zh.md b/packages/ui/app-boot/README.zh.md index d2f2b2d2c9..1e5b0850d7 100644 --- a/packages/ui/app-boot/README.zh.md +++ b/packages/ui/app-boot/README.zh.md @@ -21,7 +21,7 @@ Loader 树结算不会向调用方传播两类故障，因此需要分别保护。插件导入失败会留下没有 fiber 的配置项，`assertEntriesLoaded` 将其转换为 `boot()` rejection，并列出每个未解析插件。插件回调或配置失败则会留下失败的 fiber，因为 `loader.await()` 只结算生命周期任务，不传播该错误；`assertEntriesActivated` 会显式等待该 fiber，并把原始错误堆栈写入启动 rejection。抛出错误前，审计会通过一个进程级检查点标记这些 rejection 的确切原因，从而让 `installFailLoud` 将 Loader 的重复通知合并为一次，而所有无关的未处理 rejection 仍然致命。 -Loader 并发挂载各个条目，因此当某个同级条目 rejection 时，某个界面可能已经持有终端：此时直接从处理函数退出，会把 raw 模式、bracketed paste 和键盘协议残留在用户的 shell 上，而尚未返回的终端查询响应会在下一个提示符处显示为字面文本。因此，持有终端的 bin 会传入 `release` 来释放整棵树——执行该界面自身的 shutdown——然后才提交退出。`dsh` 在 `boot()` 的 `prepare` 回调中捕获根上下文，而不是取其返回值，因为 rejection 到达时 `boot()` 尚未结算。 +Loader 并发挂载各个条目，因此当某个同级条目 rejection 时，某个界面可能已经持有终端：此时直接从处理函数退出，会把 raw 模式、bracketed paste 和键盘协议残留在用户的 shell 上，而尚未返回的终端查询响应会在下一个提示符处显示为字面文本。因此，持有终端的 bin 会传入 `release` 来释放整棵树——执行该界面自身的 shutdown——然后才提交退出。`dsh` 在 `boot()` 的 `prepare` 回调中捕获根上下文，而不是取其返回值，因为 rejection 到达时 `boot()` 尚未结算。release 执行期间处理函数保持注册并加闩：被报告的始终是第一个 rejection，后续 rejection（包括拆卸自身的）会被吞掉，而不会变成未捕获错误、在拆卸中途杀死进程。 配置中的裸插件 specifier（`@deepseek-ai/dsh-*`、npm 包（package））通过 Cordis Loader 的内部模块 loader 解析。仓库 bin 会安装 Loader 的可选 peer `node-addon-require-builtin`；外部调用方必须提供该组件，或者把插件安装到普通 Node import 解析可以找到的位置。相对 specifier 无需原生 helper，并以配置目录为基准解析。构建后的 `dsh-app-boot` 产物内嵌静态挂载的 Include 实现，但仍将 Loader 保持为外部依赖，因此 include 树与 host 会绑定到同一个 Loader peer。`dsh` 源码启动器还会将 manifest（元数据清单）声明的 workspace 包映射到其 TypeScript 源码；其配置门禁要求每个 TUI／Web 裸插件都出现在解析所用 manifest 的 `dependencies` 中。bin 的子进程冒烟测试覆盖内部 loader 路径，而本包的单元测试套件会在进程内使用相对 specifier 配置驱动 `boot()`。 diff --git a/packages/ui/app-boot/src/index.ts b/packages/ui/app-boot/src/index.ts index ec8f717f45..da824851cc 100644 --- a/packages/ui/app-boot/src/index.ts +++ b/packages/ui/app-boot/src/index.ts @@ -343,10 +343,16 @@ export const FAIL_LOUD_RELEASE_TIMEOUT_MS = 2_000 * handler would strand raw mode, bracketed paste, and the keyboard protocol on * the user's shell, and leave an in-flight terminal query's reply to land as * literal text at the next prompt. `release` is the terminal owner's chance to - * hand it back; it is awaited under {@link FAIL_LOUD_RELEASE_TIMEOUT_MS}. The - * diagnostic is written before the release so the reason survives a disposer - * that repaints or clears the screen, and the handler uninstalls itself before - * releasing so a rejection from teardown cannot re-enter it. + * hand it back; it is awaited under {@link FAIL_LOUD_RELEASE_TIMEOUT_MS}, whose + * timer stays referenced so a never-settling disposer cannot let Node reach an + * empty event loop and exit 0 instead of failing. + * + * The diagnostic is written before the release so a hanging or failing disposer + * cannot swallow the reason. The handler stays installed while the release runs + * — removing it would let a second concurrent rejection become uncaught and kill + * the process mid-teardown, stranding exactly the terminal state this restores — + * so a latch keeps the first rejection the reported one and lets later + * rejections (including the release's own) fall through to the pending exit. * @param binName - the diagnostic prefix on the fatal-failure line. * @param proc - the process slice to register on; tests inject a fake. * @param release - optional teardown awaited before exit, used by a @@ -359,29 +365,33 @@ export function installFailLoud( proc: FailLoudProcess = process, release?: () => Promise | void, ): () => void { + let exiting = false const handler = (err: unknown): void => { if (assembledActivationRejections.has(err)) return + // A release in flight already owns the exit. Swallow later rejections + // (teardown's own included) rather than reporting a second failure over the + // real one or letting Node kill the process before the terminal is back. + if (exiting) return + exiting = true proc.stderr.write(`${binName}: fatal load failure: ${err instanceof Error ? err.stack ?? err.message : String(err)}\n`) if (release === undefined) { proc.exit(1) return } - // The release runs plugin disposers, which may themselves reject. Without - // this the handler would re-enter and report a teardown failure as a second - // fatal load failure, hiding the real one. - uninstall() void (async () => { + let timer: ReturnType | undefined try { await Promise.race([ (async () => release())(), new Promise((resolve) => { - setTimeout(resolve, FAIL_LOUD_RELEASE_TIMEOUT_MS).unref() + timer = setTimeout(resolve, FAIL_LOUD_RELEASE_TIMEOUT_MS) }), ]) } catch { // The terminal release failed; the fatal exit below is the outcome that // matters, and no reporter runs after it. } + if (timer !== undefined) clearTimeout(timer) proc.exit(1) })() } diff --git a/packages/ui/app-boot/tests/app-boot.spec.ts b/packages/ui/app-boot/tests/app-boot.spec.ts index c47b16cc78..f4b2e2a902 100644 --- a/packages/ui/app-boot/tests/app-boot.spec.ts +++ b/packages/ui/app-boot/tests/app-boot.spec.ts @@ -110,16 +110,22 @@ describe('installFailLoud', () => { expect(proc.exits).toEqual([1]) }) + // One rejection is reported per install: the first is the diagnosis, so each + // formatting case needs its own handler rather than reusing a latched one. it('stringifies a non-Error rejection and an Error without a stack falls back to its message', () => { - const proc = fakeProc() - installFailLoud(NAME, proc) - proc.handlers[0]!('plain failure') - expect(proc.written[0]).toContain('plain failure') + const plain = fakeProc() + installFailLoud(NAME, plain) + plain.handlers[0]!('plain failure') + expect(plain.written[0]).toContain('plain failure') + expect(plain.exits).toEqual([1]) + const stackless = new Error('no stack') delete (stackless as { stack?: string }).stack - proc.handlers[0]!(stackless) - expect(proc.written[1]).toContain('no stack') - expect(proc.exits).toEqual([1, 1]) + const bare = fakeProc() + installFailLoud(NAME, bare) + bare.handlers[0]!(stackless) + expect(bare.written[0]).toContain('no stack') + expect(bare.exits).toEqual([1]) }) it('returns an uninstaller that removes the handler (and defaults to the real process)', () => { @@ -203,15 +209,23 @@ describe('installFailLoud', () => { } }) - // Teardown runs plugin disposers, whose own rejection must not be reported as - // a second fatal load failure over the real one. - it('uninstalls the handler before releasing, so teardown cannot re-enter it', async () => { + // Loader failures arrive in bursts, and teardown's own disposers may reject. + // Only the first rejection is the diagnosis; the handler must stay installed + // so a later one cannot become uncaught and kill the process mid-teardown. + it('reports only the first rejection and keeps handling later ones during the release', async () => { const proc = fakeProc() - installFailLoud(NAME, proc, () => {}) - proc.handlers[0]!(new Error('boom')) - expect(proc.handlers).toHaveLength(0) - await vi.waitFor(() => { expect(proc.exits).toEqual([1]) }) + let released = false + installFailLoud(NAME, proc, async () => { + await Promise.resolve() + released = true + }) + proc.handlers[0]!(new Error('first rejection')) + proc.handlers[0]!(new Error('second rejection')) + expect(proc.handlers).toHaveLength(1) expect(proc.written).toHaveLength(1) + expect(proc.written[0]).toContain('first rejection') + await vi.waitFor(() => { expect(proc.exits).toEqual([1]) }) + expect(released).toBe(true) }) }) From b4f1675360f1b36c712eca1a14395707a72bd3dc Mon Sep 17 00:00:00 2001 From: Turtle Date: Fri, 31 Jul 2026 20:29:06 +0800 Subject: [PATCH 12/48] docs(app-boot): correct the pre-fix capture claim and pin the exit seam contract The PTY capture does continue past the terminal-takeover bytes with the fatal diagnostic; only the reset never follows. State that precisely in both notes. Document on FailLoudProcess.exit that callers treat it as the end of the run, matching how the release path already relies on it. --- .../2026-07-31-fail-loud-releases-the-terminal.i18n.yaml | 4 ++-- .../bug-fix/2026-07-31-fail-loud-releases-the-terminal.md | 2 +- .../bug-fix/2026-07-31-fail-loud-releases-the-terminal.zh.md | 2 +- packages/ui/app-boot/src/index.ts | 5 +++++ 4 files changed, 9 insertions(+), 4 deletions(-) diff --git a/.agents/notes/implemented/bug-fix/2026-07-31-fail-loud-releases-the-terminal.i18n.yaml b/.agents/notes/implemented/bug-fix/2026-07-31-fail-loud-releases-the-terminal.i18n.yaml index 97dc84403b..df444bc96d 100644 --- a/.agents/notes/implemented/bug-fix/2026-07-31-fail-loud-releases-the-terminal.i18n.yaml +++ b/.agents/notes/implemented/bug-fix/2026-07-31-fail-loud-releases-the-terminal.i18n.yaml @@ -2,5 +2,5 @@ # side as of the last confirmed-consistent state. Both languages carry equal authority; # after editing either side, bring the other along and re-record with: # pnpm run verify-translation-pairing --write .agents/notes/implemented/bug-fix/2026-07-31-fail-loud-releases-the-terminal.md -2026-07-31-fail-loud-releases-the-terminal.md: ccac625171ef5523a4ed27843b543c838bf43ce8 -2026-07-31-fail-loud-releases-the-terminal.zh.md: fe8a3271b26a94b9d986b8d17744893e5f448593 +2026-07-31-fail-loud-releases-the-terminal.md: 8659c8a72dbb25cceaccbb0fb99b8b0251e1d506 +2026-07-31-fail-loud-releases-the-terminal.zh.md: 19ced1f685c8719a652ebabd27ac199519b09369 diff --git a/.agents/notes/implemented/bug-fix/2026-07-31-fail-loud-releases-the-terminal.md b/.agents/notes/implemented/bug-fix/2026-07-31-fail-loud-releases-the-terminal.md index ccac625171..8659c8a72d 100644 --- a/.agents/notes/implemented/bug-fix/2026-07-31-fail-loud-releases-the-terminal.md +++ b/.agents/notes/implemented/bug-fix/2026-07-31-fail-loud-releases-the-terminal.md @@ -54,6 +54,6 @@ The guarantee belongs to whichever bin owns the terminal: a surface that grabs t `packages/ui/app-boot/tests/app-boot.spec.ts` covers the release contract: the hook is awaited before the exit commits, a rejecting hook still exits 1, a never-settling hook exits after `FAIL_LOUD_RELEASE_TIMEOUT_MS`, and a burst of rejections reports only the first while the release still completes. -Those fake-process tests cannot observe the two failure modes that matter most — process exit code with a real event loop, and terminal state after exit — so the regression lives in `apps/cli/tests/tui-keyless-smoke.e2e.ts`. It boots the shipped tree in a real PTY over `fixtures/tui-invalid-provider.cordis.yml` (a list-shaped `providers`, the mistake users actually make), expects exit 1, and asserts the captured bytes contain both the diagnostic and `ESC[?2004l`. Against the pre-fix source the captured stream ends at `ESC[?2004h ESC[>7u ESC[?u ESC[c` with no reset, and the case fails on that assertion. +Those fake-process tests cannot observe the two failure modes that matter most — process exit code with a real event loop, and terminal state after exit — so the regression lives in `apps/cli/tests/tui-keyless-smoke.e2e.ts`. It boots the shipped tree in a real PTY over `fixtures/tui-invalid-provider.cordis.yml` (a list-shaped `providers`, the mistake users actually make), expects exit 1, and asserts the captured bytes contain both the diagnostic and `ESC[?2004l`. Against the pre-fix source the capture still shows the terminal being taken (`ESC[?2004h ESC[>7u ESC[?u ESC[c`) and the diagnostic printed, but no reset ever follows, and the case fails on the `ESC[?2004l` assertion alone. Testing policy requires a PTY case whenever terminal teardown changes, and this is it. The `/exit` path keeps its existing assertion that the same reset appears on a clean exit. diff --git a/.agents/notes/implemented/bug-fix/2026-07-31-fail-loud-releases-the-terminal.zh.md b/.agents/notes/implemented/bug-fix/2026-07-31-fail-loud-releases-the-terminal.zh.md index fe8a3271b2..19ced1f685 100644 --- a/.agents/notes/implemented/bug-fix/2026-07-31-fail-loud-releases-the-terminal.zh.md +++ b/.agents/notes/implemented/bug-fix/2026-07-31-fail-loud-releases-the-terminal.zh.md @@ -54,6 +54,6 @@ Loader 并发挂载各个条目，因此条目失败的顺序并不等于启动 `packages/ui/app-boot/tests/app-boot.spec.ts` 覆盖 release 契约：退出提交前会等待该回调；回调 rejection 时仍退出 1；永不结算的回调会在 `FAIL_LOUD_RELEASE_TIMEOUT_MS` 后退出；以及一连串 rejection 只报告第一个，同时 release 仍能跑完。 -这些基于假进程的测试无法观测到最关键的两种失败形态——真实事件循环下的进程退出码，以及退出之后的终端状态——因此回归用例放在 `apps/cli/tests/tui-keyless-smoke.e2e.ts`。它在真实 PTY 中以 `fixtures/tui-invalid-provider.cordis.yml`（`providers` 为列表形状，正是用户真实会犯的错误）启动出厂配置树，期望退出码为 1，并断言捕获到的字节流同时包含诊断信息与 `ESC[?2004l`。在修复前的源码上，捕获流止于 `ESC[?2004h ESC[>7u ESC[?u ESC[c` 而没有任何重置，该用例正是在这条断言上失败。 +这些基于假进程的测试无法观测到最关键的两种失败形态——真实事件循环下的进程退出码，以及退出之后的终端状态——因此回归用例放在 `apps/cli/tests/tui-keyless-smoke.e2e.ts`。它在真实 PTY 中以 `fixtures/tui-invalid-provider.cordis.yml`（`providers` 为列表形状，正是用户真实会犯的错误）启动出厂配置树，期望退出码为 1，并断言捕获到的字节流同时包含诊断信息与 `ESC[?2004l`。在修复前的源码上，捕获内容仍能看到终端被接管（`ESC[?2004h ESC[>7u ESC[?u ESC[c`）以及诊断信息被打印，但其后始终没有任何重置序列，该用例仅在 `ESC[?2004l` 这条断言上失败。 测试规范要求：只要改动终端拆卸，就必须有 PTY 用例——这就是它。`/exit` 路径保留其原有断言，确认正常退出时同样会出现该重置序列。 diff --git a/packages/ui/app-boot/src/index.ts b/packages/ui/app-boot/src/index.ts index da824851cc..727c551187 100644 --- a/packages/ui/app-boot/src/index.ts +++ b/packages/ui/app-boot/src/index.ts @@ -295,6 +295,11 @@ export interface FailLoudProcess { on(event: 'unhandledRejection', handler: (err: unknown) => void): unknown off(event: 'unhandledRejection', handler: (err: unknown) => void): unknown stderr: { write(chunk: string): unknown } + /** + * Terminate the process. Callers treat this as the end of the run, as + * `process.exit` is; a fake that returns lets the caller continue, which only + * a test observes. + */ exit(code: number): void } From 54e541d33f76b1fcbfb922d29d80052dd808b0f9 Mon Sep 17 00:00:00 2001 From: Turtle Date: Fri, 31 Jul 2026 22:14:43 +0800 Subject: [PATCH 13/48] fix(app-boot): drop the unreachable timer guard on the fail-loud release path The timeout promise's executor runs synchronously while the race is constructed, so the timer is always assigned; the undefined check was a dead branch the per-file coverage gate rejected. --- packages/ui/app-boot/src/index.ts | 6 ++++-- 1 file changed, 4 insertions(+), 2 deletions(-) diff --git a/packages/ui/app-boot/src/index.ts b/packages/ui/app-boot/src/index.ts index 727c551187..88a1f82736 100644 --- a/packages/ui/app-boot/src/index.ts +++ b/packages/ui/app-boot/src/index.ts @@ -384,7 +384,9 @@ export function installFailLoud( return } void (async () => { - let timer: ReturnType | undefined + // Definitely assigned: the timeout promise's executor runs synchronously + // while the race is being constructed, before the first await. + let timer!: ReturnType try { await Promise.race([ (async () => release())(), @@ -396,7 +398,7 @@ export function installFailLoud( // The terminal release failed; the fatal exit below is the outcome that // matters, and no reporter runs after it. } - if (timer !== undefined) clearTimeout(timer) + clearTimeout(timer) proc.exit(1) })() } From 76547dfe0cc7b1db2b6cead44a7f7f2f78b53550 Mon Sep 17 00:00:00 2001 From: kingwl Date: Sun, 2 Aug 2026 16:25:02 +0800 Subject: [PATCH 14/48] fix(web): restrict message forks to completed turn tails --- ...ions-require-completed-turn-tail.i18n.yaml | 6 ++ ...ork-actions-require-completed-turn-tail.md | 27 ++++++++ ...-actions-require-completed-turn-tail.zh.md | 27 ++++++++ ...6-07-27-web-session-fork-actions.i18n.yaml | 4 +- .../2026-07-27-web-session-fork-actions.md | 8 ++- .../2026-07-27-web-session-fork-actions.zh.md | 8 ++- apps/web/tests/message-actions.e2e.ts | 68 +++++++++++++++---- .../snapshots/message-actions/ui.expected.md | 11 ++- packages/client/runtime/README.i18n.yaml | 4 +- packages/client/runtime/README.md | 2 +- packages/client/runtime/README.zh.md | 2 +- .../src/client/sessions/conversation.ts | 2 + .../runtime/src/client/sessions/session.ts | 13 ++++ packages/client/runtime/tests/session.spec.ts | 2 + packages/client/test-runtime/src/fixtures.ts | 1 + .../client/ui-conversation/README.i18n.yaml | 4 +- packages/client/ui-conversation/README.md | 4 +- packages/client/ui-conversation/README.zh.md | 4 +- .../src/client/chat/AssistantMarkdown.tsx | 9 +-- .../src/client/chat/ChatView.tsx | 8 ++- .../src/client/chat/MessageIconActions.tsx | 14 ++-- .../src/client/chat/MessageItem.tsx | 2 +- .../src/client/chat/chat-flow.ts | 36 +++++++++- .../src/client/contract/slots.ts | 2 +- .../tests/chat-branch-tails.spec.tsx | 9 +++ .../tests/chat-code-subcalls.spec.tsx | 2 +- .../tests/chat-stats-bash-sample.spec.tsx | 2 +- .../ui-conversation/tests/chat-view.spec.tsx | 53 ++++++++++++--- .../ui-conversation/tests/diff-card.spec.tsx | 2 +- .../tests/gate-branch-tails.spec.tsx | 2 +- .../ui-conversation/tests/input-bar.spec.tsx | 2 +- .../tests/input-matrix.spec.tsx | 2 +- .../tests/input-scenarios.spec.tsx | 2 +- .../ui-conversation/tests/queue-dock.spec.tsx | 2 +- .../ui-conversation/tests/read-card.spec.tsx | 2 +- .../tests/search-card.spec.tsx | 2 +- .../ui-conversation/tests/skeleton.spec.tsx | 2 +- .../tests/terminal-card.spec.tsx | 2 +- .../ui-conversation/tests/web-card.spec.tsx | 2 +- 39 files changed, 283 insertions(+), 73 deletions(-) create mode 100644 .agents/notes/implemented/bug-fix/2026-08-02-message-fork-actions-require-completed-turn-tail.i18n.yaml create mode 100644 .agents/notes/implemented/bug-fix/2026-08-02-message-fork-actions-require-completed-turn-tail.md create mode 100644 .agents/notes/implemented/bug-fix/2026-08-02-message-fork-actions-require-completed-turn-tail.zh.md diff --git a/.agents/notes/implemented/bug-fix/2026-08-02-message-fork-actions-require-completed-turn-tail.i18n.yaml b/.agents/notes/implemented/bug-fix/2026-08-02-message-fork-actions-require-completed-turn-tail.i18n.yaml new file mode 100644 index 0000000000..15b7acbeba --- /dev/null +++ b/.agents/notes/implemented/bug-fix/2026-08-02-message-fork-actions-require-completed-turn-tail.i18n.yaml @@ -0,0 +1,6 @@ +# Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each +# side as of the last confirmed-consistent state. Both languages carry equal authority; +# after editing either side, bring the other along and re-record with: +# pnpm run verify-translation-pairing --write .agents/notes/implemented/bug-fix/2026-08-02-message-fork-actions-require-completed-turn-tail.md +2026-08-02-message-fork-actions-require-completed-turn-tail.md: 1f5ee6b7ff37709e07544f3e7c73ba1eee951f8d +2026-08-02-message-fork-actions-require-completed-turn-tail.zh.md: ed897e9af56a601c8fc42379e3907e3551c6389e diff --git a/.agents/notes/implemented/bug-fix/2026-08-02-message-fork-actions-require-completed-turn-tail.md b/.agents/notes/implemented/bug-fix/2026-08-02-message-fork-actions-require-completed-turn-tail.md new file mode 100644 index 0000000000..1f5ee6b7ff --- /dev/null +++ b/.agents/notes/implemented/bug-fix/2026-08-02-message-fork-actions-require-completed-turn-tail.md @@ -0,0 +1,27 @@ +# Agent Note: Message fork actions require a completed turn tail + +Status: implemented + +English | [中文](2026-08-02-message-fork-actions-require-completed-turn-tail.zh.md) + +## Problem + +The Web conversation attached branch to the last assistant node with nonempty text in each turn. A later tool result, interrupted reasoning node, or terminal error did not take ownership because those rows have no content-text IconActions. The branch icon could therefore appear beneath an assistant response while more rows from the same turn remained below it. The Host correctly expanded that message anchor through the containing `turn/end`, but the placement made the action look like a message-level cut and the child visibly inherited the same-turn suffix. + +## Decision + +`ConversationSnapshot.turnEnds` retains the completed turn boundaries present in the raw event window. The conversation view walks transcript nodes through each boundary and exposes branch only when the boundary's last node is a user message or a content-bearing assistant message. Open turns have no eligible message, and a later tool result, reasoning-only interruption, turn error, or other transcript node suppresses branch on earlier messages. Copy and clock remain available under their existing message chrome, and the Host's completed-turn fork semantics remain unchanged. + +This narrows the message eligibility established by the earlier [Web session fork action decision](../feature/2026-07-27-web-session-fork-actions.md). Session-row forking still selects the latest completed turn, and eligible message actions still pass their event seq through the shared client runtime operation. + +## Alternatives considered + +**Cut the event log at the clicked assistant message.** Rejected because an assistant message can sit inside an open step and can contain tool calls whose results occur later. A raw prefix at that seq is not a balanced turn and may not be a valid provider transcript. + +**Infer completion from `running` or the next user message.** Rejected because retry and steering turns need not align with the next visible user bubble, and a paged window may omit that later bubble. The durable `turn/end` event is the authoritative completion fact. + +**Hide branch from every interrupted turn.** Rejected because an aborted turn is durably closed and its final interrupted text can be the true transcript tail. Eligibility depends on the completed boundary and node order, not the outcome kind. + +## Consequences + +A branch icon now denotes the same completed-turn boundary that the Host will copy. In the reported response → tool → interrupted Think shape, the response keeps copy and clock but no longer advertises branch. This change deliberately does not provide same-turn transcript editing or a retry-before-turn operation; the Session-row action remains available when a reader wants to copy the latest completed turn in full. Runtime tests pin boundary projection and reference stability, while conversation tests cover normal assistant tails, user-only tails, and suppression by later tool and interrupted reasoning rows. diff --git a/.agents/notes/implemented/bug-fix/2026-08-02-message-fork-actions-require-completed-turn-tail.zh.md b/.agents/notes/implemented/bug-fix/2026-08-02-message-fork-actions-require-completed-turn-tail.zh.md new file mode 100644 index 0000000000..ed897e9af5 --- /dev/null +++ b/.agents/notes/implemented/bug-fix/2026-08-02-message-fork-actions-require-completed-turn-tail.zh.md @@ -0,0 +1,27 @@ +# Agent Note: 消息 fork 操作要求消息位于已完成轮次尾部 + +Status: implemented + +[English](2026-08-02-message-fork-actions-require-completed-turn-tail.md) | 中文 + +## 问题 + +Web 会话把分支操作挂到每个轮次中最后一个文本非空的 assistant 节点上。如果后面还有工具结果、被中断的推理（reasoning）节点或终态错误，这些行也不会接管操作，因为它们没有内容文本 IconActions。因此，分支图标可能出现在 assistant 响应下方，而同一轮次的更多行仍位于其后。Host 会正确地把该消息锚点扩展到其所在的 `turn/end`，但图标位置使操作看起来像在消息级截断，子会话又会明显继承同轮次的后缀。 + +## 决策 + +`ConversationSnapshot.turnEnds` 保留原始事件窗口中的已完成轮次边界。会话视图按各边界遍历 transcript（文本记录）节点，仅当边界的最后一个节点是用户消息或含内容的 assistant 消息时才暴露分支操作。开放轮次没有符合条件的消息；如果后面还有工具结果、只有推理内容的中断、轮次错误或其他 transcript 节点，较早消息上的分支操作就会被抑制。复制和时钟仍可在既有消息 chrome 下使用，Host 按已完成轮次 fork 的语义保持不变。 + +本决策收紧了较早的 [Web 会话 fork 操作决策](../feature/2026-07-27-web-session-fork-actions.md)所定义的消息资格。Session 行 fork 仍选择最新的已完成轮次；符合条件的消息操作仍通过共享 client 运行时操作传递其事件 seq。 + +## 考虑过的替代方案 + +**在点击的 assistant 消息处截断事件日志。** 不予采纳：assistant 消息可能位于尚未结束的步骤内，也可能包含结果随后才出现的工具调用。以该 seq 截取的原始前缀并不是结构完整的轮次，也可能不是有效的提供方 transcript。 + +**从 `running` 或下一条用户消息推断完成状态。** 不予采纳：重试轮次与 steering（中途引导）轮次不一定和下一个可见用户气泡对齐，分页窗口也可能省略该气泡。持久 `turn/end` 事件才是权威的完成事实。 + +**对每个被中断轮次隐藏分支。** 不予采纳：已中止的轮次会持久关闭，其最终的中断文本可能正是真正的 transcript 尾部。资格取决于已完成边界与节点顺序，而非结果类别。 + +## 后果 + +分支图标现在表示的已完成轮次边界与 Host 实际复制的边界一致。在所报告的「响应 → 工具 → 被中断的 Think」形态中，响应仍保留复制和时钟，但不再显示分支。本变更刻意不提供同轮次 transcript 编辑，也不提供轮次前重试操作；当读者希望完整复制最新的已完成轮次时，仍可使用 Session 行操作。运行时测试固定边界投影和引用稳定性，会话测试则覆盖普通 assistant 尾部、纯用户消息尾部，以及后续工具行和被中断推理行对分支操作的抑制。 diff --git a/.agents/notes/implemented/feature/2026-07-27-web-session-fork-actions.i18n.yaml b/.agents/notes/implemented/feature/2026-07-27-web-session-fork-actions.i18n.yaml index 21eea20254..5268ea62f9 100644 --- a/.agents/notes/implemented/feature/2026-07-27-web-session-fork-actions.i18n.yaml +++ b/.agents/notes/implemented/feature/2026-07-27-web-session-fork-actions.i18n.yaml @@ -2,5 +2,5 @@ # side as of the last confirmed-consistent state. Both languages carry equal authority; # after editing either side, bring the other along and re-record with: # pnpm run verify-translation-pairing --write .agents/notes/implemented/feature/2026-07-27-web-session-fork-actions.md -2026-07-27-web-session-fork-actions.md: 58960169a2e499d953840e5769e7689b5cd48047 -2026-07-27-web-session-fork-actions.zh.md: ea2f9030f672f00fb91bce3546836689a7d41004 +2026-07-27-web-session-fork-actions.md: 578e59ec92e003fe5c8cdfe951a595f3a7371ecf +2026-07-27-web-session-fork-actions.zh.md: d90124f6e6e164b0a1e0fce734f52976630cd848 diff --git a/.agents/notes/implemented/feature/2026-07-27-web-session-fork-actions.md b/.agents/notes/implemented/feature/2026-07-27-web-session-fork-actions.md index 58960169a2..578e59ec92 100644 --- a/.agents/notes/implemented/feature/2026-07-27-web-session-fork-actions.md +++ b/.agents/notes/implemented/feature/2026-07-27-web-session-fork-actions.md @@ -10,7 +10,9 @@ The Session store already provides a fork primitive that creates a child session ## Decision -The Web Session-row menu and message IconActions share the client runtime's `sessions.fork` action. A Session row passes `{ sessionId, increaseTitle: true }`, so it forks at the source session's last completed turn; a user message or settled assistant content message passes `{ sessionId, atSeq: node.seq, increaseTitle: true }`, so it forks at the turn containing that event. Only the client consumes `increaseTitle`: after adding the child session to its local list, the client increments a trailing `(N)` or `（N）` in the source session's persisted title without changing bracket style, appends ` (1)` to an unnumbered title, and skips the rename when no persisted title exists; the Host fork request still contains only `sessionId` and the optional `atSeq`. The caller opens the child only after the rename succeeds; a fork or rename failure leaves the source session and current selection unchanged, while a child created before a rename failure remains in the list. +The message-eligibility portion of this decision is narrowed by the [completed-turn-tail decision](../bug-fix/2026-08-02-message-fork-actions-require-completed-turn-tail.md); the shared runtime action, injection ownership, title handling, and peer-list decisions remain current. + +The Web Session-row menu and message IconActions share the client runtime's `sessions.fork` action. A Session row passes `{ sessionId, increaseTitle: true }`, so it forks at the source session's last completed turn; an eligible completed-turn-tail message passes `{ sessionId, atSeq: node.seq, increaseTitle: true }`, so it forks at the turn ending at that message. Only the client consumes `increaseTitle`: after adding the child session to its local list, the client increments a trailing `(N)` or `（N）` in the source session's persisted title without changing bracket style, appends ` (1)` to an unnumbered title, and skips the rename when no persisted title exists; the Host fork request still contains only `sessionId` and the optional `atSeq`. The caller opens the child only after the rename succeeds; a fork or rename failure leaves the source session and current selection unchanged, while a child created before a rename failure remains in the list. `forkAt(seq)` touches the session service only in ui-conversation's apply injection layer; message components report only the event `seq`. Session rows likewise initiate the operation only through ui-workspace's injected callback. Neither presentation package owns session mutation state or duplicates the host's boundary evaluation. @@ -28,6 +30,6 @@ Session lineage is not projected into a list hierarchy. WorkSpace mode displays ## Consequences -Users can create forks from Session rows, user messages, or settled assistant content messages; all three entry points ultimately use the same runtime/host operation. Message entry points preserve the exact event boundary, while the list entry point preserves the “latest completed turn” shortcut. Successive fork titles increment through `(1)`, `(2)`, and so on instead of repeatedly appending `(1)`; titles with fullwidth parentheses retain that style. Every fork child immediately appears as an ordinary peer row, so the list no longer needs session expansion state, recursive nodes, or twist controls. +Users can create forks from Session rows or eligible completed-turn-tail messages; both entry points ultimately use the same runtime/host operation. Message entry points preserve the exact event boundary, while the list entry point preserves the “latest completed turn” shortcut. Successive fork titles increment through `(1)`, `(2)`, and so on instead of repeatedly appending `(1)`; titles with fullwidth parentheses retain that style. Every fork child immediately appears as an ordinary peer row, so the list no longer needs session expansion state, recursive nodes, or twist controls. -Fork and child-rename failures stay silent and preserve the source selection, preventing a derivation action from disrupting the current reading position; this tradeoff also means the UI does not yet expose a failure reason or retry entry point. Package tests separately pin the two message `seq` paths, title increments, and the peer-list derivation; `apps/web/tests/message-actions.e2e.ts` exercises assistant-message branching and Session-row menu branching through the assembled application. +Fork and child-rename failures stay silent and preserve the source selection, preventing a derivation action from disrupting the current reading position; this tradeoff also means the UI does not yet expose a failure reason or retry entry point. Package tests pin eligible message `seq` forwarding, title increments, and the peer-list derivation; `apps/web/tests/message-actions.e2e.ts` exercises assistant-message branching and Session-row menu branching through the assembled application. diff --git a/.agents/notes/implemented/feature/2026-07-27-web-session-fork-actions.zh.md b/.agents/notes/implemented/feature/2026-07-27-web-session-fork-actions.zh.md index ea2f9030f6..d90124f6e6 100644 --- a/.agents/notes/implemented/feature/2026-07-27-web-session-fork-actions.zh.md +++ b/.agents/notes/implemented/feature/2026-07-27-web-session-fork-actions.zh.md @@ -10,7 +10,9 @@ Session store 已提供按完成轮前缀创建子会话的 fork 原语，但 We ## Decision -Web 的 session 行菜单与消息 IconActions 共用 client runtime 的 `sessions.fork` 操作。Session 行传 `{ sessionId, increaseTitle: true }`，因此在源会话最后一个已完成轮次处分支；用户消息与已定稿 assistant 内容消息传 `{ sessionId, atSeq: node.seq, increaseTitle: true }`，因此在包含该事件的轮次处分支。`increaseTitle` 只由 client 消费：子会话进入本地列表后，client 把源会话持久化标题尾部的 `(N)` 或 `（N）` 递增并保留括号样式，无编号时追加 ` (1)`，没有持久化标题时不改名；Host fork 请求仍只有 `sessionId` 与可选的 `atSeq`。改名成功后调用方才打开子会话；fork 或改名失败时保持源会话与当前选择不变，改名失败时已创建的子会话仍留在列表中。 +本决策中的消息资格部分由[已完成轮次尾部决策](../bug-fix/2026-08-02-message-fork-actions-require-completed-turn-tail.md)收紧；共享运行时操作、注入归属、标题处理和同级列表决策仍然有效。 + +Web 的 session 行菜单与消息 IconActions 共用 client runtime 的 `sessions.fork` 操作。Session 行传 `{ sessionId, increaseTitle: true }`，因此在源会话最后一个已完成轮次处分支；符合条件且位于已完成轮次尾部的消息传 `{ sessionId, atSeq: node.seq, increaseTitle: true }`，因此在以该消息结束的轮次处分支。`increaseTitle` 只由 client 消费：子会话进入本地列表后，client 把源会话持久化标题尾部的 `(N)` 或 `（N）` 递增并保留括号样式，无编号时追加 ` (1)`，没有持久化标题时不改名；Host fork 请求仍只有 `sessionId` 与可选的 `atSeq`。改名成功后调用方才打开子会话；fork 或改名失败时保持源会话与当前选择不变，改名失败时已创建的子会话仍留在列表中。 `forkAt(seq)` 只在 ui-conversation 的 apply 注入层接触 session 服务，消息组件只回传事件 `seq`。Session 行同理只通过 ui-workspace 的注入回调发起操作；两个呈现包都不持有 session mutation 状态，也不复制 host 的边界求值。 @@ -28,6 +30,6 @@ Session lineage 不投影成列表层级。WorkSpace 模式按 `WorkspaceView.se ## Consequences -用户可从 session 行、用户消息或已定稿 assistant 内容消息创建分支，三处最终走同一个 runtime/host 操作；消息点位保留精确事件边界，列表点位保留「最新完成轮」快捷语义。连续 fork 的标题按 `(1)`、`(2)` 递增，而不是重复追加 `(1)`；全角括号标题保持全角样式。所有 fork 子会话立即作为普通同级行出现，列表不再需要 session 展开状态、递归节点或 twist 控件。 +用户可从 session 行或符合条件的已完成轮次尾部消息创建分支，两处最终走同一个 runtime/host 操作；消息点位保留精确事件边界，列表点位保留「最新完成轮」快捷语义。连续 fork 的标题按 `(1)`、`(2)` 递增，而不是重复追加 `(1)`；全角括号标题保持全角样式。所有 fork 子会话立即作为普通同级行出现，列表不再需要 session 展开状态、递归节点或 twist 控件。 -Fork 与子会话改名失败都保持静默并保留源选择，避免一个派生操作破坏当前阅读位置；该取舍也意味着 UI 暂不提供失败原因或重试入口。Package tests 分别钉住两种消息 `seq`、标题递增与同级列表派生，`apps/web/tests/message-actions.e2e.ts` 通过装配后的应用执行 assistant 消息分支与 session 行菜单分支。 +Fork 与子会话改名失败都保持静默并保留源选择，避免一个派生操作破坏当前阅读位置；该取舍也意味着 UI 暂不提供失败原因或重试入口。包级测试固定符合条件的消息 `seq` 转发、标题递增与同级列表派生，`apps/web/tests/message-actions.e2e.ts` 通过装配后的应用执行 assistant 消息分支与 session 行菜单分支。 diff --git a/apps/web/tests/message-actions.e2e.ts b/apps/web/tests/message-actions.e2e.ts index bb58ccefc4..96dd27f393 100644 --- a/apps/web/tests/message-actions.e2e.ts +++ b/apps/web/tests/message-actions.e2e.ts @@ -1,7 +1,7 @@ -// Web e2e scenario: message IconActions + clocks. Cold-seeds the seeded-history -// fixture (zero model calls) and pins the settled conversation aria after the -// user/assistant footers are focus-revealed — the surface package jsdom tests -// cannot substitute for (docs/testing.md snapshot rule). +// Web e2e scenario: message IconActions + clocks. Cold-seeds a deterministic +// completed-turn-tail fork case (zero model calls) and pins the settled +// conversation aria after the footers are focus-revealed — the surface package +// jsdom tests cannot substitute for (docs/testing.md snapshot rule). import { mkdir, readFile, writeFile } from 'node:fs/promises' import { join } from 'node:path' import { fileURLToPath } from 'node:url' @@ -25,6 +25,48 @@ const MODE = webSnapshotMode() const SEED_ID = 'message-actions-web-e2e' const PROMPT = 'Use the read tool twice in one assistant message: read a.txt and b.txt. Then reply with the single word DONE and stop.' +const MID_TURN_TEXT = 'I will read both files before answering.' +const SECOND_PROMPT = 'Now give the final answer.' + +/** + * Adapt the borrowed recording into response -> tools -> interrupted Think, + * followed by one ordinary completed response. The first response keeps + * copy/clock but is not a legal branch point; the second is the real turn tail. + * @param raw - Recorded seeded-history JSONL. + * @returns A contiguous, closed two-turn fixture. + */ +function completedTailFixture(raw: string): string { + const kept: string[] = [] + for (const line of raw.trimEnd().split('\n')) { + const row = JSON.parse(line) as { + type: string + seq?: number + seq0?: number + data?: { content?: unknown[] } + } + const firstSeq = row.seq ?? row.seq0 + if (firstSeq !== undefined && firstSeq >= 101) break + if (row.type === 'assistant/message' && row.seq === 64) { + const content = row.data?.content + if (!Array.isArray(content)) throw new Error('borrowed step-one assistant message has no content') + content.splice(1, 0, { type: 'text', text: MID_TURN_TEXT }) + kept.push(JSON.stringify(row)) + } else { + kept.push(line) + } + } + const tail = [ + { type: 'step/end', seq: 101, time: 1784974102749, data: { turn: 1, step: 2 } }, + { type: 'turn/end', seq: 102, time: 1784974102750, data: { turn: 1, reason: { kind: 'aborted' } } }, + { type: 'turn/start', seq: 103, time: 1784974103000, data: { turn: 2, trigger: { kind: 'message', source: { kind: 'user', rpcId: '{{rpcId}}' } } } }, + { type: 'user/message', seq: 104, time: 1784974103001, data: { content: [{ type: 'text', text: SECOND_PROMPT }], source: { kind: 'user', rpcId: '{{rpcId}}' } }, surfaceOp: 'append' }, + { type: 'step/start', seq: 105, time: 1784974103002, data: { turn: 2, step: 1 } }, + { type: 'assistant/message', seq: 106, time: 1784974103003, data: { turn: 2, step: 1, content: [{ type: 'text', text: 'DONE' }], provenance: { provider: 'deepseek-official', model: 'deepseek-v4-flash' } }, sourceEventSeqs: [], surfaceOp: 'append' }, + { type: 'step/end', seq: 107, time: 1784974103004, data: { turn: 2, step: 1 } }, + { type: 'turn/end', seq: 108, time: 1784974103005, data: { turn: 2, reason: { kind: 'completed' } } }, + ] + return `${[...kept, ...tail.map(row => JSON.stringify(row))].join('\n')}\n` +} describe('web e2e: message IconActions and clocks on settled history', () => { let scaffold: WebScaffold @@ -38,8 +80,8 @@ describe('web e2e: message IconActions and clocks on settled history', () => { await mkdir(sessionCwd, { recursive: true }) await writeFile(join(sessionCwd, 'a.txt'), 'alpha\n') await writeFile(join(sessionCwd, 'b.txt'), 'beta\n') - const raw = await readFile(SEED, 'utf8') - expect(fixtureUserPrompts(raw), 'borrowed seed must carry the drive prompt').toEqual([PROMPT]) + const raw = completedTailFixture(await readFile(SEED, 'utf8')) + expect(fixtureUserPrompts(raw), 'adapted seed must carry both prompts').toEqual([PROMPT, SECOND_PROMPT]) await seedSession(scaffold, raw, SEED_ID) browser = await chromium.launch() page = await newEnglishPage(browser) @@ -53,7 +95,7 @@ describe('web e2e: message IconActions and clocks on settled history', () => { await scaffold?.close() }) - it.skipIf(MODE === 'record')('lists the seeded session and reveals user/assistant IconActions', async () => { + it.skipIf(MODE === 'record')('shows branch only on the completed transcript tail', async () => { onTestFailed(() => saveFailureShot(page, 'web-e2e-message-actions')) const groupRow = page.locator('[role="treeitem"]').first() await groupRow.waitFor({ timeout: 15_000 }) @@ -61,16 +103,17 @@ describe('web e2e: message IconActions and clocks on settled history', () => { const sessionRow = page.locator('[role="treeitem"]').nth(1) await sessionRow.waitFor({ timeout: 10_000 }) await sessionRow.click() + await expect.poll(() => page.getByText(MID_TURN_TEXT, { exact: true }).count(), { timeout: 15_000 }).toBe(1) await expect.poll(() => page.getByText('DONE', { exact: true }).count(), { timeout: 15_000 }).toBe(1) // Focus-reveal the footers (hover:hover keeps them opacity-hidden until - // hover/focus-within). User and each turn's last content assistant both - // have copy + branch. + // hover/focus-within). All message rows keep copy, but only the final + // assistant at a completed transcript tail has branch. const copyButtons = page.getByRole('button', { name: 'Copy' }) - await expect.poll(() => copyButtons.count(), { timeout: 10_000 }).toBeGreaterThanOrEqual(2) + await expect.poll(() => copyButtons.count(), { timeout: 10_000 }).toBeGreaterThanOrEqual(4) await copyButtons.first().focus() await expect.poll(() => page.getByRole('button', { name: 'Branch into a new conversation' }).count(), { timeout: 5_000 }) - .toBeGreaterThanOrEqual(2) + .toBe(1) await expect.poll(() => page.getByRole('button', { name: 'Edit' }).count(), { timeout: 5_000 }).toBe(0) }, 60_000) @@ -89,8 +132,7 @@ describe('web e2e: message IconActions and clocks on settled history', () => { it.skipIf(MODE === 'record')('forks through the settled-message and session-row actions', async () => { onTestFailed(() => saveFailureShot(page, 'web-e2e-message-fork')) - // Exercise the assistant action specifically; package coverage pins the - // user action separately at its own event seq. + // The sole message action belongs to the completed second-turn assistant. await page.getByRole('button', { name: 'Branch into a new conversation' }).last().click() await expect.poll( () => scaffold.ctx.agents.list().find(agent => agent.session.header.parentSession === SessionId(SEED_ID)), diff --git a/apps/web/tests/snapshots/message-actions/ui.expected.md b/apps/web/tests/snapshots/message-actions/ui.expected.md index bf67498178..65152bb4de 100644 --- a/apps/web/tests/snapshots/message-actions/ui.expected.md +++ b/apps/web/tests/snapshots/message-actions/ui.expected.md @@ -8,12 +8,14 @@ - button "Copy": - img - tooltip "Copy" -- button "Branch into a new conversation": - - img - button "Think The user wants me to read a.txt and b.txt, then reply with \"DONE\". Let me do both reads in parallel.": - img - img - text: Think The user wants me to read a.txt and b.txt, then reply with "DONE". Let me do both reads in parallel. +- paragraph: I will read both files before answering. +- button "Copy": + - img +- text: 7/25 {{clock}} - button "Read a.txt": - img - img @@ -28,6 +30,9 @@ - img - img - text: Think Both files have been read. a.txt contains "alpha" and b.txt contains "beta". I'll now reply with DONE as instructed. +- text: Stopped Now give the final answer. 7/25 {{clock}} +- button "Copy": + - img - paragraph: DONE - button "Copy": - img @@ -42,4 +47,4 @@ - text: deepseek-v4-flash - img - button "Send message" [disabled] -- text: 1 turns · 2 steps Tool call {{duration}} Cache hit 98% Input 15.8K tok · Output 135 tok +- text: 2 turns · 3 steps Tool call {{duration}} Cache hit 98% Input 7.8K tok · Output 103 tok diff --git a/packages/client/runtime/README.i18n.yaml b/packages/client/runtime/README.i18n.yaml index f20967ee2f..dd36762aeb 100644 --- a/packages/client/runtime/README.i18n.yaml +++ b/packages/client/runtime/README.i18n.yaml @@ -2,5 +2,5 @@ # side as of the last confirmed-consistent state. Both languages carry equal authority; # after editing either side, bring the other along and re-record with: # pnpm run verify-translation-pairing --write packages/client/runtime/README.md -README.md: f956be22384a42e9ed30e8aa5f25fe8173cc9f9c -README.zh.md: 49449c51d89d957b5bd39798c9167607f72a5c3a +README.md: 91844016ce58e172161343022a1824be91b7ff2f +README.zh.md: 0ea77774c6181858748a8adbc737de7e213be31b diff --git a/packages/client/runtime/README.md b/packages/client/runtime/README.md index f956be2238..91844016ce 100644 --- a/packages/client/runtime/README.md +++ b/packages/client/runtime/README.md @@ -26,7 +26,7 @@ SlotsService gives the renderer separate bare observables for `useSessions` and ## The human transcript -`ConversationSnapshot.nodes` is the human transcript, not the model surface. `TranscriptAdapter` projects the raw window in log order — every append-origin surface event (`isAppendSurfaceEvent`) at its own log position, plus one `CompactionSummaryNode` marker per landed compaction checkpoint — and never consults surface order. A landed compaction therefore keeps the conversation it shadowed on the model side: the marker reports where the model stopped seeing that history instead of erasing it. Model-only replacement copies stay out: a pruned `tool/result` and a regenerated `assistant/message` rewrite one node for the model and mark no boundary. A checkpoint is a `user/message` carrying the compaction seam's plugin source that **replaced** a surface range; an appending plugin-sourced `user/message` is injected context, not a compaction. The adapter's plugin literal is pinned to the seam's own declaration by a type-only import of the cordis-free [`dsh-compact/checkpoint`](../../compact/compact/README.md) leaf, so renaming it there fails `tsc` here; a **value** import of the package would fail the client purity gate, and the package **root** is unreachable even as a type (it reaches `dsh-session`'s root, whose `Context` merge collides the host `sessions` with this program's). `tests/compact-checkpoint-pin.spec.ts` covers the same drift behaviorally. +`ConversationSnapshot.nodes` is the human transcript, not the model surface. `TranscriptAdapter` projects the raw window in log order — every append-origin surface event (`isAppendSurfaceEvent`) at its own log position, plus one `CompactionSummaryNode` marker per landed compaction checkpoint — and never consults surface order. `ConversationSnapshot.turnEnds` maps each completed turn in that window to its `turn/end` seq, retaining turn completion independently from the transcript so presentation can require a real boundary before exposing an action. A landed compaction therefore keeps the conversation it shadowed on the model side: the marker reports where the model stopped seeing that history instead of erasing it. Model-only replacement copies stay out: a pruned `tool/result` and a regenerated `assistant/message` rewrite one node for the model and mark no boundary. A checkpoint is a `user/message` carrying the compaction seam's plugin source that **replaced** a surface range; an appending plugin-sourced `user/message` is injected context, not a compaction. The adapter's plugin literal is pinned to the seam's own declaration by a type-only import of the cordis-free [`dsh-compact/checkpoint`](../../compact/compact/README.md) leaf, so renaming it there fails `tsc` here; a **value** import of the package would fail the client purity gate, and the package **root** is unreachable even as a type (it reaches `dsh-session`'s root, whose `Context` merge collides the host `sessions` with this program's). `tests/compact-checkpoint-pin.spec.ts` covers the same drift behaviorally. Because the projection is log-ordered, the node array is seq-monotonic by construction: log-only `command/run` / `command/done` nodes splice in by seq, `Session` merges interrupted frozen nodes by their fractional seqs, and a window whose checkpoint cites a shadowed range outside it renders the marker with nothing logged. The marker's summary text comes from the checkpoint's `compact/summary` provenance; a window cut that left the provenance outside makes the row non-expandable rather than empty, and a later page that supplies it resolves the text. Performance contract: one append materializes at most one node and copies the projection only when it adds that node; an event that changes no node keeps the previous array reference (a chunk storm costs nothing), and unchanged nodes keep their object identity. diff --git a/packages/client/runtime/README.zh.md b/packages/client/runtime/README.zh.md index 49449c51d8..0ea77774c6 100644 --- a/packages/client/runtime/README.zh.md +++ b/packages/client/runtime/README.zh.md @@ -26,7 +26,7 @@ SlotsService 分别为 renderer 提供 `useSessions` 与 `useWorkspaces` 的裸 ## 面向人的 transcript（文本记录） -`ConversationSnapshot.nodes` 是面向人的 transcript，不是模型 surface。`TranscriptAdapter` 按日志顺序投影原始窗口——每个 append 来源的 surface 事件（`isAppendSurfaceEvent`）落在它自己的日志位置上，外加每次落地的压缩（compaction）检查点贡献一个 `CompactionSummaryNode` 标记——且从不查询 surface 顺序。于是一次落地的压缩会保留它在模型侧遮蔽掉的对话：标记报告模型从哪里开始看不见那段历史，而不是把它抹掉。仅模型可见的 replacement 副本不进入记录：被裁剪的 `tool/result` 和重新生成的 `assistant/message` 只为模型重写一个节点，不标记任何边界。检查点是携带压缩 seam 插件来源、且**替换**了一段 surface 范围的 `user/message`；一条 append 的插件来源 `user/message` 是注入上下文，不是压缩。适配器的插件字面量通过对无 cordis 的 [`dsh-compact/checkpoint`](../../compact/compact/README.md) 叶子做仅类型导入，钉在压缩 seam 自己的声明上：在那里改名会让此处 `tsc` 失败；而对该包（package）做**值**导入会被客户端纯度门禁拒绝，包的**根**即便作为类型也无法到达（它会到达 `dsh-session` 的根，其 `Context` 合并会让 host 的 `sessions` 与本程序的冲突）。`tests/compact-checkpoint-pin.spec.ts` 从行为侧覆盖同一漂移。 +`ConversationSnapshot.nodes` 是面向人的 transcript，不是模型 surface。`TranscriptAdapter` 按日志顺序投影原始窗口——每个 append 来源的 surface 事件（`isAppendSurfaceEvent`）落在它自己的日志位置上，外加每次落地的压缩（compaction）检查点贡献一个 `CompactionSummaryNode` 标记——且从不查询 surface 顺序。`ConversationSnapshot.turnEnds` 把该窗口中的每个已完成轮次映射到其 `turn/end` seq；它独立于 transcript 保留轮次完成状态，使呈现层能够在暴露操作前要求存在真实边界。于是一次落地的压缩会保留它在模型侧遮蔽掉的对话：标记报告模型从哪里开始看不见那段历史，而不是把它抹掉。仅模型可见的 replacement 副本不进入记录：被裁剪的 `tool/result` 和重新生成的 `assistant/message` 只为模型重写一个节点，不标记任何边界。检查点是携带压缩 seam 插件来源、且**替换**了一段 surface 范围的 `user/message`；一条 append 的插件来源 `user/message` 是注入上下文，不是压缩。适配器的插件字面量通过对无 cordis 的 [`dsh-compact/checkpoint`](../../compact/compact/README.md) 叶子做仅类型导入，钉在压缩 seam 自己的声明上：在那里改名会让此处 `tsc` 失败；而对该包（package）做**值**导入会被客户端纯度门禁拒绝，包的**根**即便作为类型也无法到达（它会到达 `dsh-session` 的根，其 `Context` 合并会让 host 的 `sessions` 与本程序的冲突）。`tests/compact-checkpoint-pin.spec.ts` 从行为侧覆盖同一漂移。 由于投影按日志顺序，节点数组天然按 seq 单调：仅日志的 `command/run` / `command/done` 节点按 seq 插入，`Session` 按分数 seq 归并被打断的冻结节点，而检查点所引范围落在窗口之外的窗口会渲染出标记且不打印任何日志。标记的摘要文本来自检查点的 `compact/summary` 溯源；窗口切分把溯源留在窗口外时该行不可展开而非空白，后续补上溯源的分页会解析出文本。性能契约：一次追加最多物化一个节点，并且仅在加入该节点时复制投影；不改变任何节点的事件保持上一次的数组引用（分片风暴零成本），未变化的节点保持其对象标识。 diff --git a/packages/client/runtime/src/client/sessions/conversation.ts b/packages/client/runtime/src/client/sessions/conversation.ts index a4c260a068..7752762eae 100644 --- a/packages/client/runtime/src/client/sessions/conversation.ts +++ b/packages/client/runtime/src/client/sessions/conversation.ts @@ -322,6 +322,8 @@ export interface ConversationSnapshot { sessionId: SessionId /** Human transcript plus retry notices and interrupted-turn terminal nodes in event order. */ nodes: readonly ConversationNode[] + /** In-window completed turn number -> its `turn/end` event seq. */ + turnEnds: ReadonlyMap partial: PartialAssistant | null runningCalls: readonly RunningToolCall[] /** diff --git a/packages/client/runtime/src/client/sessions/session.ts b/packages/client/runtime/src/client/sessions/session.ts index 2bcb2ac5fc..dee41b2fe0 100644 --- a/packages/client/runtime/src/client/sessions/session.ts +++ b/packages/client/runtime/src/client/sessions/session.ts @@ -113,6 +113,11 @@ export class Session implements SessionFace { private pendingCache: { rev: number; value: PendingInteraction[] } | null = null private derivedRev = 0 private nodesCache: { projected: readonly ConversationNode[]; derivedRev: number; value: readonly ConversationNode[] } | null = null + /** Completed turn boundaries retained from the raw window so presentation + * actions never infer a safe fork point from transcript content alone. */ + private turnEnds = new Map() + private turnEndsRev = 0 + private turnEndsCache: { rev: number; value: ReadonlyMap } | null = null /** Authoritative stream-only inbox snapshot; pending work never hits history. */ private queued: QueuedMessage[] = [] private queueRev = 0 @@ -807,6 +812,8 @@ export class Session implements SessionFace { return } case 'turn/end': { + this.turnEnds.set(event.data.turn, event.seq) + this.turnEndsRev++ if (event.data.reason.kind === 'aborted' || event.data.reason.kind === 'disposed') { this.settleScheduledRetry('cancelled', event.data.turn) } @@ -897,6 +904,8 @@ export class Session implements SessionFace { this.callsRev++ this.derivedNodes = [] this.derivedRev++ + this.turnEnds = new Map() + this.turnEndsRev++ this.codeDispatches = new Map() this.dispatchesRev++ for (let i = 0; i < this.events.length; i++) { @@ -928,6 +937,9 @@ export class Session implements SessionFace { if (this.callsCache === null || this.callsCache.rev !== this.callsRev) { this.callsCache = { rev: this.callsRev, value: [...this.openCalls.values()] } } + if (this.turnEndsCache === null || this.turnEndsCache.rev !== this.turnEndsRev) { + this.turnEndsCache = { rev: this.turnEndsRev, value: new Map(this.turnEnds) } + } if (this.pendingCache === null || this.pendingCache.rev !== this.pendingRev) { this.pendingCache = { rev: this.pendingRev, value: [...this.pending.values()] } } @@ -941,6 +953,7 @@ export class Session implements SessionFace { return { sessionId: this.sessionId, nodes, + turnEnds: this.turnEndsCache.value, partial, runningCalls: this.callsCache.value, pending: this.pendingCache.value, diff --git a/packages/client/runtime/tests/session.spec.ts b/packages/client/runtime/tests/session.spec.ts index 7f5f793948..6e75593e3a 100644 --- a/packages/client/runtime/tests/session.spec.ts +++ b/packages/client/runtime/tests/session.spec.ts @@ -426,6 +426,7 @@ describe('live event path', () => { feed(ev.turnEnd(10, 1, 'aborted')) // no assistant/message ever arrives const snapshot = session.getSnapshot() expect(snapshot.partial).toBeNull() + expect(snapshot.turnEnds.get(1)).toBe(10) const frozen = snapshot.nodes.at(-1) expect(frozen).toMatchObject({ kind: 'assistant', interrupted: true, blocks: [{ kind: 'text', text: '说到一半' }] }) // Ordered inside the flow: after the user message (seq 7), before any later turn. @@ -1215,6 +1216,7 @@ describe('reference stability (the memo contract)', () => { expect(after).not.toBe(before) expect(after.runningCalls).toBe(before.runningCalls) expect(after.pending).toBe(before.pending) + expect(after.turnEnds).toBe(before.turnEnds) // And a mutation on the tracked domain swaps that array. feed(ev.toolResult(11, 1, 'c1', 'ECHO')) const resolved = session.getSnapshot() diff --git a/packages/client/test-runtime/src/fixtures.ts b/packages/client/test-runtime/src/fixtures.ts index 2544870ace..d033669215 100644 --- a/packages/client/test-runtime/src/fixtures.ts +++ b/packages/client/test-runtime/src/fixtures.ts @@ -46,6 +46,7 @@ export function conversationSnapshot(sessionId: SessionId): ConversationSnapshot return { sessionId, nodes: [], + turnEnds: new Map(), partial: null, runningCalls: [], codeDispatches: new Map(), diff --git a/packages/client/ui-conversation/README.i18n.yaml b/packages/client/ui-conversation/README.i18n.yaml index aec4bc46bc..597f99f7e3 100644 --- a/packages/client/ui-conversation/README.i18n.yaml +++ b/packages/client/ui-conversation/README.i18n.yaml @@ -2,5 +2,5 @@ # side as of the last confirmed-consistent state. Both languages carry equal authority; # after editing either side, bring the other along and re-record with: # pnpm run verify-translation-pairing --write packages/client/ui-conversation/README.md -README.md: 65e4542e622713e2cd120906926fc0601ef280bd -README.zh.md: 4828ad214bb5ec0a9921307dd01dcc282910b330 +README.md: e351880066ac00677ab9d22c96ea75e2244a2ecf +README.zh.md: f7c7e3ab6d341f9b3d0866951ab80a18d5968044 diff --git a/packages/client/ui-conversation/README.md b/packages/client/ui-conversation/README.md index 65e4542e62..e351880066 100644 --- a/packages/client/ui-conversation/README.md +++ b/packages/client/ui-conversation/README.md @@ -57,8 +57,8 @@ None; this package neither assembles nor sends a provider request. - **Compaction markers show no scale** — the row does not yet report how many messages or which range the checkpoint replaced. - **Stats-line durations cover the in-window flow only** — LLM and tool wall times fold the snapshot's assistant `timing` and tool call/result pairs, so nodes outside the loaded event window (older history) are not counted. - **Details panel is the minimal form and currently has no entry point** — selected call args/result raw display; the Input/Output/Metadata switch, Prev/Next stepping, and See-in-trajectory deep link are deferred. Tool rows stopped being details-panel click targets and nothing replaced that gesture, so `ChatViewInjected.openDetails` is implemented but uncalled and the panel (including its terminal card) is unreachable in the assembled application; its rendering stays covered by mounting it with a selection directly. -- **Assistant per-message paging is a reserved slot** — drawn in the design, not implemented. The finalized content IconActions row (copy / branch / clock) ships under the last content-text assistant of each turn only; mid-turn narration and Think-only nodes stay chrome-free. Branch forks through the turn containing that message, increments the inherited title on the client, and then opens the child, while a fork or rename failure leaves the source selected. -- **Sent user messages cannot be edited** — the user bubble's IconActions row carries clock / copy / branch only, and branching from the message is the nearest gesture. The control returns with the capability behind it: a client mutation over a settled user message, plus the host behavior for the turn that already consumed it ([decision](../../../.agents/notes/implemented/simplification/2026-07-31-drop-user-message-edit-stub.md)). +- **Assistant per-message paging is a reserved slot** — drawn in the design, not implemented. The finalized content IconActions row (copy / clock, plus branch when eligible) ships under the last content-text assistant of each turn only; mid-turn narration and Think-only nodes stay chrome-free. Branch appears only when that message is also the last transcript node of a completed turn, then forks through that turn, increments the inherited title on the client, and opens the child; a fork or rename failure leaves the source selected ([decision](../../../.agents/notes/implemented/bug-fix/2026-08-02-message-fork-actions-require-completed-turn-tail.md)). +- **Sent user messages cannot be edited** — user bubbles retain clock and copy, while branch appears only for a completed turn whose transcript ends at that user message. Editing returns with the capability behind it: a client mutation over a settled user message, plus the host behavior for the turn that already consumed it ([decision](../../../.agents/notes/implemented/simplification/2026-07-31-drop-user-message-edit-stub.md)). - **The sparkle icon for the others tool row is a hand-drawn approximation** — the design glyph's vector geometry is not exportable locally; promotion into ui-primitives waits on an exact export. - **The approval panel's "Always allow this type" is deferred** — durable grants need a grant-storage design; only allow-once/reject answer today. - **TodoPanel truncates long item text to one ellipsized line** — the figma strip has no wrap or expand affordance; full text is not readable inline. diff --git a/packages/client/ui-conversation/README.zh.md b/packages/client/ui-conversation/README.zh.md index 4828ad214b..f7c7e3ab6d 100644 --- a/packages/client/ui-conversation/README.zh.md +++ b/packages/client/ui-conversation/README.zh.md @@ -57,8 +57,8 @@ todo 两个面就是在该形状上的两个注册项，都是普通注册方插 - **压缩标记不显示规模**：该行尚不报告检查点替换了多少条消息或哪段范围。 - **统计行的耗时只覆盖窗口内消息流**：LLM 与工具墙钟时间由快照的 assistant `timing` 与工具 call/result 配对折算，落在已加载事件窗口之外的节点（更早的历史）不计入。 - **详情面板是最小形态，且当前没有入口**：以原始形式显示已选择调用的参数／结果；Input/Output/Metadata 切换、Prev/Next 步进与 See-in-trajectory 深链接暂缓实现。工具行已不再是详情面板的点击目标，且没有任何手势接替它，因此 `ChatViewInjected.openDetails` 虽已实现却无人调用，该面板（含其终端卡片）在组装后的应用中不可达；其渲染仍由直接以选中态挂载它来覆盖。 -- **assistant 逐消息分页是预留 slot**：设计中已有图稿，尚未实现。已定稿的内容 IconActions 行（复制／分支／时钟）只挂在每个轮次中最后一条带 text 内容的 assistant 下；轮次中间的叙述与纯 Think 节点不带 chrome。分支会 fork 到包含该消息的轮次末尾，在 client 端递增继承标题后打开子会话，而 fork 或改名失败时源会话保持选中。 -- **已发送的 user 消息无法编辑**：user 气泡的 IconActions 行只有时钟／复制／分支，从该消息分支是最接近的手势。该控件要与其背后的能力一起回归：既需要针对已定稿 user 消息的 client 变更，也需要 host 侧对已经消费过它的轮次给出行为（[决策](../../../.agents/notes/implemented/simplification/2026-07-31-drop-user-message-edit-stub.md)）。 +- **assistant 逐消息分页是预留 slot**：设计中已有图稿，尚未实现。已定稿的内容 IconActions 行（复制／时钟，符合条件时再显示分支）只挂在每个轮次中最后一条带 text 内容的 assistant 下；轮次中间的叙述与纯 Think 节点不带 chrome。只有当该消息同时也是已完成轮次的最后一个 transcript 节点时才显示分支；随后 fork 到该轮次末尾，在 client 端递增继承标题并打开子会话；fork 或改名失败时源会话保持选中（[决策](../../../.agents/notes/implemented/bug-fix/2026-08-02-message-fork-actions-require-completed-turn-tail.md)）。 +- **已发送的 user 消息无法编辑**：user 气泡保留时钟和复制；仅当已完成轮次的 transcript 结束于该 user 消息时才显示分支。编辑功能要与其背后的能力一起回归：既需要针对已定稿 user 消息的 client 变更，也需要 host 侧对已经消费过它的轮次给出行为（[决策](../../../.agents/notes/implemented/simplification/2026-07-31-drop-user-message-edit-stub.md)）。 - **others 工具行的闪光图标是手绘近似版本**：无法在本地导出设计字形的矢量几何；等到存在精确导出后再将其提升到 ui-primitives。 - **审批面板的「始终允许此类」暂缓**：持久授权需要授权存储设计；今天只能回答允许一次／拒绝。 - **TodoPanel 将过长条目截成单行省略号**：figma 条没有换行或展开入口，完整文本无法在行内读完。 diff --git a/packages/client/ui-conversation/src/client/chat/AssistantMarkdown.tsx b/packages/client/ui-conversation/src/client/chat/AssistantMarkdown.tsx index 387a7fd82a..8569051d30 100644 --- a/packages/client/ui-conversation/src/client/chat/AssistantMarkdown.tsx +++ b/packages/client/ui-conversation/src/client/chat/AssistantMarkdown.tsx @@ -4,9 +4,10 @@ // view groups them into tool rows through its keyed toolview slot (figma // step-summary flow). Shared by finalized nodes and the streaming partial; // the turn-level loading dots live in the chat view's tail, not here. -// Finalized turn-tail content (text) nodes append IconActions once streaming -// ends (`time` is omitted for mid-turn narration); Think / tool-head-only -// nodes stay chrome-free. +// Finalized content (text) nodes append IconActions once streaming ends +// (`time` is omitted for mid-turn narration); their branch action is present +// only when the node is also the completed turn's transcript tail. Think / +// tool-head-only nodes stay chrome-free. import { memo, useMemo } from 'react' import type { AssistantBlock } from '@deepseek-ai/dsh-client-runtime/client' @@ -28,7 +29,7 @@ export interface AssistantMarkdownProps { time?: number | undefined /** Event sequence used as the fork boundary; omitted while streaming. */ seq?: number | undefined - /** Fork the session through the turn containing this finalized message. */ + /** Fork the session through this finalized message's completed turn. */ onFork?: ((seq: number) => void) | undefined /** The owning view's locale seat, passed down as a plain prop. */ t: ChatViewSlotProps['t'] diff --git a/packages/client/ui-conversation/src/client/chat/ChatView.tsx b/packages/client/ui-conversation/src/client/chat/ChatView.tsx index 4a8e715bcf..af0e3a1c39 100644 --- a/packages/client/ui-conversation/src/client/chat/ChatView.tsx +++ b/packages/client/ui-conversation/src/client/chat/ChatView.tsx @@ -30,7 +30,7 @@ import type { import type { SnapshotSelectorHook } from '@deepseek-ai/dsh-client-ui-slots' import { IconChevronDownOutline14 } from '@deepseek-ai/dsh-client-ui-primitives' import type { ChatViewSlotProps } from '../contract/slots.ts' -import { assistantActionsSeqs, deriveChatFlow, type ChatFlowItem } from './chat-flow.ts' +import { assistantActionsSeqs, deriveChatFlow, messageBranchSeqs, type ChatFlowItem } from './chat-flow.ts' import { AssistantMarkdown } from './AssistantMarkdown.tsx' import { GenericCommandCard } from './GenericCommandCard.tsx' import { GenericToolCard } from './GenericToolCard.tsx' @@ -236,6 +236,7 @@ export function ChatView({ useSession, useSessions, useStore, renderSlot, sessionId, openFile, loadOlder, inspectCall, chatScroll, forkAt, t, }: ChatViewSlotProps) { const nodes = useSession(s => s.nodes) + const turnEnds = useSession(s => s.turnEnds) // Workspace root off the session list row: path summaries display relative to it. const cwd = useSessions(s => s.byId[sessionId]?.cwd) const running = useSession(s => s.running) @@ -252,6 +253,7 @@ export function ChatView({ // Only the last content assistant of each turn owns IconActions; mid-turn // text (before tools) omits `time` so AssistantMarkdown stays chrome-free. const actionSeqs = useMemo(() => assistantActionsSeqs(nodes), [nodes]) + const branchSeqs = useMemo(() => messageBranchSeqs(nodes, turnEnds), [nodes, turnEnds]) const listRef = useRef(null) const atBottomRef = useRef(true) @@ -401,7 +403,7 @@ export function ChatView({ interrupted={node.interrupted} time={actionSeqs.has(node.seq) ? node.time : undefined} seq={node.seq} - onFork={forkAt} + onFork={branchSeqs.has(node.seq) ? forkAt : undefined} t={t} /> ) @@ -416,7 +418,7 @@ export function ChatView({ key={item.key} node={node} retryActive={node.kind === 'model-retry' && node.seq === activeRetry} - onFork={forkAt} + {...branchSeqs.has(node.seq) ? { onFork: forkAt } : {}} t={t} /> ) diff --git a/packages/client/ui-conversation/src/client/chat/MessageIconActions.tsx b/packages/client/ui-conversation/src/client/chat/MessageIconActions.tsx index cd6b0a36ce..81129b94ac 100644 --- a/packages/client/ui-conversation/src/client/chat/MessageIconActions.tsx +++ b/packages/client/ui-conversation/src/client/chat/MessageIconActions.tsx @@ -17,7 +17,7 @@ export interface MessageIconActionsProps { time: number /** Clock before icons (user) or after (assistant). */ clock: 'start' | 'end' - /** Fork the session at this message. */ + /** Fork the session at this message; omission hides the branch action. */ onBranch?: (() => void) | undefined /** Parent layout class composed onto the actions row. */ className?: string | undefined @@ -50,11 +50,13 @@ export function MessageIconActions({ - - - + {onBranch !== undefined && ( + + + + )} {clock === 'end' ? clockEl : null}
) diff --git a/packages/client/ui-conversation/src/client/chat/MessageItem.tsx b/packages/client/ui-conversation/src/client/chat/MessageItem.tsx index b54604060a..3a27feaef3 100644 --- a/packages/client/ui-conversation/src/client/chat/MessageItem.tsx +++ b/packages/client/ui-conversation/src/client/chat/MessageItem.tsx @@ -26,7 +26,7 @@ export interface MessageItemProps { | TurnErrorNode | UnknownSurfaceNode retryActive?: boolean - /** Fork the session through the turn containing this message (user-bubble branch action). */ + /** Fork through this message's completed turn when it is the transcript tail. */ onFork?: (seq: number) => void /** The owning view's locale seat, passed down as a plain prop. */ t: ChatViewSlotProps['t'] diff --git a/packages/client/ui-conversation/src/client/chat/chat-flow.ts b/packages/client/ui-conversation/src/client/chat/chat-flow.ts index 4958894154..9bd98fd5a5 100644 --- a/packages/client/ui-conversation/src/client/chat/chat-flow.ts +++ b/packages/client/ui-conversation/src/client/chat/chat-flow.ts @@ -5,8 +5,8 @@ * reuse the first notice's row while projecting the latest retry turn. * Item identity keys are stable across snapshots so the list parent can * subscribe to keys only while rows subscribe to content. IconActions ownership - * (last content assistant per turn) is derived here too so ChatView and the - * flow share one gate. + * and completed-turn branch points are derived here too so ChatView and the + * flow share their gates. */ import type { AssistantBlock, ConversationNode, ToolResultNode, @@ -47,6 +47,38 @@ export function assistantActionsSeqs(nodes: readonly ConversationNode[]): Readon return new Set(lastByTurn.values()) } +/** + * Seq set of message rows that may fork: the last transcript node of a + * completed turn, when that node owns message chrome. A later tool, reasoning, + * error, or other transcript node suppresses the earlier message's branch + * action even though the Host would include the whole turn. + * @param nodes - snapshot nodes in event order. + * @param turnEnds - completed turn boundaries retained from the event window. + * @returns Message seq values whose visible position matches the fork boundary. + */ +export function messageBranchSeqs( + nodes: readonly ConversationNode[], + turnEnds: ReadonlyMap, +): ReadonlySet { + const result = new Set() + const boundaries = [...turnEnds].sort((a, b) => a[1] - b[1]) + let nodeIndex = 0 + for (const [turn, endSeq] of boundaries) { + let tail: ConversationNode | undefined + while (nodeIndex < nodes.length) { + const candidate = nodes[nodeIndex] + if (candidate === undefined || candidate.seq > endSeq) break + tail = candidate + nodeIndex++ + } + if (tail?.kind === 'user' + || (tail?.kind === 'assistant' && tail.turn === turn && hasContentText(tail.blocks))) { + result.add(tail.seq) + } + } + return result +} + /** * Group finalized nodes into the step-summary flow. * @param nodes - snapshot nodes in human-transcript and durable-notice order. diff --git a/packages/client/ui-conversation/src/client/contract/slots.ts b/packages/client/ui-conversation/src/client/contract/slots.ts index a6c9d16e61..390a9aeb76 100644 --- a/packages/client/ui-conversation/src/client/contract/slots.ts +++ b/packages/client/ui-conversation/src/client/contract/slots.ts @@ -454,7 +454,7 @@ export interface ChatViewInjected { /** Last recorded offset, or null when pinned or never recorded. */ read: () => number | null } - /** Fork the session through the turn containing the message at `seq`, then open the child. */ + /** Fork through the completed turn ending at the eligible message `seq`, then open the child. */ forkAt: (seq: number) => void } diff --git a/packages/client/ui-conversation/tests/chat-branch-tails.spec.tsx b/packages/client/ui-conversation/tests/chat-branch-tails.spec.tsx index a829f6f04a..964b261b78 100644 --- a/packages/client/ui-conversation/tests/chat-branch-tails.spec.tsx +++ b/packages/client/ui-conversation/tests/chat-branch-tails.spec.tsx @@ -36,12 +36,14 @@ describe('MessageItem arms', () => { // Same-day clock: construct "today at 14:24" so the label stays `HH:mm`. const now = new Date() const time = new Date(now.getFullYear(), now.getMonth(), now.getDate(), 14, 24).getTime() + const onFork = vi.fn() render( , ) expect(screen.getByText('14:24')).toBeTruthy() @@ -50,6 +52,8 @@ describe('MessageItem arms', () => { expect(screen.queryByRole('button', { name: '编辑' })).toBeNull() fireEvent.click(screen.getByRole('button', { name: '复制' })) expect(writeText).toHaveBeenCalledWith('hello bubble') + fireEvent.click(screen.getByRole('button', { name: '在新对话中分支' })) + expect(onFork).toHaveBeenCalledWith(1) }) it('user copy falls back to execCommand when clipboard.writeText is unavailable', () => { @@ -410,12 +414,15 @@ describe('small branch tails', () => { }) const now = new Date() const time = new Date(now.getFullYear(), now.getMonth(), now.getDate(), 14, 24).getTime() + const onFork = vi.fn() const settled = render( , ) expect(settled.getByText('14:24')).toBeTruthy() @@ -423,6 +430,8 @@ describe('small branch tails', () => { expect(settled.getByRole('button', { name: '在新对话中分支' })).toBeTruthy() fireEvent.click(settled.getByRole('button', { name: '复制' })) expect(writeText).toHaveBeenCalledWith('answer body') + fireEvent.click(settled.getByRole('button', { name: '在新对话中分支' })) + expect(onFork).toHaveBeenCalledWith(3) settled.unmount() const thinkOnly = render( diff --git a/packages/client/ui-conversation/tests/chat-code-subcalls.spec.tsx b/packages/client/ui-conversation/tests/chat-code-subcalls.spec.tsx index e1366721c8..6708cdb759 100644 --- a/packages/client/ui-conversation/tests/chat-code-subcalls.spec.tsx +++ b/packages/client/ui-conversation/tests/chat-code-subcalls.spec.tsx @@ -67,7 +67,7 @@ function snapshotWith( runningCalls: RunningToolCall[] = [], ): ConversationSnapshot { return { - sessionId: SID, nodes, partial: null, runningCalls, codeDispatches, + sessionId: SID, nodes, turnEnds: new Map(), partial: null, runningCalls, codeDispatches, pending: [], queue: [], running: runningCalls.length > 0, composerPhase: 'active', removed: false, openState: 'open', openError: null, hasMore: false, loadingOlder: false, promptError: null, blank: false, subagent: null, lastAgentError: null, diff --git a/packages/client/ui-conversation/tests/chat-stats-bash-sample.spec.tsx b/packages/client/ui-conversation/tests/chat-stats-bash-sample.spec.tsx index a02f8ccb6f..95da35f267 100644 --- a/packages/client/ui-conversation/tests/chat-stats-bash-sample.spec.tsx +++ b/packages/client/ui-conversation/tests/chat-stats-bash-sample.spec.tsx @@ -32,7 +32,7 @@ const assistant = (seq: number, turn: number, usage?: unknown): AssistantMessage function snapshotBase(): ConversationSnapshot { return { - sessionId: SID, nodes: [], partial: null, runningCalls: [], codeDispatches: new Map(), + sessionId: SID, nodes: [], turnEnds: new Map(), partial: null, runningCalls: [], codeDispatches: new Map(), pending: [], queue: [], running: false, composerPhase: 'active', removed: false, openState: 'open', openError: null, hasMore: false, loadingOlder: false, promptError: null, blank: false, subagent: null, lastAgentError: null, } diff --git a/packages/client/ui-conversation/tests/chat-view.spec.tsx b/packages/client/ui-conversation/tests/chat-view.spec.tsx index 04f412fb25..86333eb6ec 100644 --- a/packages/client/ui-conversation/tests/chat-view.spec.tsx +++ b/packages/client/ui-conversation/tests/chat-view.spec.tsx @@ -20,7 +20,7 @@ import { zh as commonZh } from '@deepseek-ai/dsh-client-locale/src/locales/zh.ts import { createChatStore } from '../src/client/stores.ts' import { ChatView } from '../src/client/chat/ChatView.tsx' import { zh } from '../src/client/locales.ts' -import { assistantActionsSeqs, deriveChatFlow, flowKeys } from '../src/client/chat/chat-flow.ts' +import { assistantActionsSeqs, deriveChatFlow, flowKeys, messageBranchSeqs } from '../src/client/chat/chat-flow.ts' afterEach(cleanup) // Keyless create() persists under the bare declared key; clear between cases @@ -33,7 +33,7 @@ const SID = 's1' as SessionId function snapshotBase(): ConversationSnapshot { return { - sessionId: SID, nodes: [], partial: null, runningCalls: [], codeDispatches: new Map(), + sessionId: SID, nodes: [], turnEnds: new Map(), partial: null, runningCalls: [], codeDispatches: new Map(), pending: [], queue: [], running: false, composerPhase: 'active', removed: false, openState: 'open', openError: null, hasMore: false, loadingOlder: false, promptError: null, blank: false, subagent: null, lastAgentError: null, } @@ -211,6 +211,24 @@ describe('chat-flow derivation', () => { ]) expect([...seqs].sort((a, b) => a - b)).toEqual([5, 7]) }) + + it('messageBranchSeqs keeps only message rows at completed transcript tails', () => { + const interruptedThink: AssistantMessageNode = { + kind: 'assistant', seq: 4.1, time: 4_100, turn: 1, step: 2, + blocks: [{ kind: 'reasoning', text: 'bad path' }], interrupted: true, + } + const nodes = [ + user(1, 'first'), + assistant(2, 'answer before tools'), + toolResult(3, 'a'), + interruptedThink, + user(6, 'second'), + assistant(7, 'clean tail', 2), + user(10, 'user-only tail'), + ] + const seqs = messageBranchSeqs(nodes, new Map([[1, 5], [2, 8], [3, 11]])) + expect([...seqs]).toEqual([7, 10]) + }) }) describe('ChatView', () => { @@ -323,21 +341,38 @@ describe('ChatView', () => { user(5, 'next'), assistant(6, 'second turn', 2), ], + turnEnds: new Map([[1, 4], [2, 6]]), }) const view = render() - // 2 user + 2 turn-tail assistants; mid-turn text at seq 2 stays chrome-free. + // User rows keep copy/clock, while only the two completed assistant tails may branch. expect(view.getAllByRole('button', { name: '复制' })).toHaveLength(4) - expect(view.getAllByRole('button', { name: '在新对话中分支' })).toHaveLength(4) + expect(view.getAllByRole('button', { name: '在新对话中分支' })).toHaveLength(2) }) - it('forks from both user and finalized assistant message actions at their event seq', () => { - const h = makeHarness({ nodes: [user(1, 'question'), assistant(2, 'answer')] }) + it('forks only from a finalized assistant at the completed transcript tail', () => { + const h = makeHarness({ + nodes: [user(1, 'question'), assistant(2, 'answer')], + turnEnds: new Map([[1, 3]]), + }) const view = render() const buttons = view.getAllByRole('button', { name: '在新对话中分支' }) - expect(buttons).toHaveLength(2) + expect(buttons).toHaveLength(1) fireEvent.click(buttons[0]!) - fireEvent.click(buttons[1]!) - expect(h.forkAt.mock.calls).toEqual([[1], [2]]) + expect(h.forkAt.mock.calls).toEqual([[2]]) + }) + + it('keeps copy chrome but hides branch when tool and interrupted Think follow the response', () => { + const interruptedThink: AssistantMessageNode = { + kind: 'assistant', seq: 4.1, time: 4_100, turn: 1, step: 2, + blocks: [{ kind: 'reasoning', text: 'bad path' }], interrupted: true, + } + const h = makeHarness({ + nodes: [user(1, 'question'), assistant(2, 'answer'), toolResult(3, 'a'), interruptedThink], + turnEnds: new Map([[1, 5]]), + }) + const view = render() + expect(view.getAllByRole('button', { name: '复制' })).toHaveLength(2) + expect(view.queryByRole('button', { name: '在新对话中分支' })).toBeNull() }) it('renders assistant Markdown across history, streaming, final, and interrupted states while user text stays literal', () => { diff --git a/packages/client/ui-conversation/tests/diff-card.spec.tsx b/packages/client/ui-conversation/tests/diff-card.spec.tsx index 1722b9628e..f7de4e45f9 100644 --- a/packages/client/ui-conversation/tests/diff-card.spec.tsx +++ b/packages/client/ui-conversation/tests/diff-card.spec.tsx @@ -335,7 +335,7 @@ describe('DetailsPanel diff Output section', () => { function snapshot(over: Partial = {}): ConversationSnapshot { return { - sessionId: SID, nodes: [], partial: null, runningCalls: [], codeDispatches: new Map(), + sessionId: SID, nodes: [], turnEnds: new Map(), partial: null, runningCalls: [], codeDispatches: new Map(), pending: [], queue: [], running: false, composerPhase: 'active', removed: false, openState: 'open', openError: null, hasMore: false, loadingOlder: false, promptError: null, blank: false, subagent: null, lastAgentError: null, ...over, diff --git a/packages/client/ui-conversation/tests/gate-branch-tails.spec.tsx b/packages/client/ui-conversation/tests/gate-branch-tails.spec.tsx index 4496423efa..8350b318e9 100644 --- a/packages/client/ui-conversation/tests/gate-branch-tails.spec.tsx +++ b/packages/client/ui-conversation/tests/gate-branch-tails.spec.tsx @@ -24,7 +24,7 @@ const SID = 's1' as SessionId function snapshotBase(): ConversationSnapshot { return { - sessionId: SID, nodes: [], partial: null, runningCalls: [], codeDispatches: new Map(), + sessionId: SID, nodes: [], turnEnds: new Map(), partial: null, runningCalls: [], codeDispatches: new Map(), pending: [], queue: [], running: false, composerPhase: 'active', removed: false, openState: 'open', openError: null, hasMore: false, loadingOlder: false, promptError: null, blank: false, subagent: null, lastAgentError: null, } diff --git a/packages/client/ui-conversation/tests/input-bar.spec.tsx b/packages/client/ui-conversation/tests/input-bar.spec.tsx index 2072fd2c90..5b2884c01e 100644 --- a/packages/client/ui-conversation/tests/input-bar.spec.tsx +++ b/packages/client/ui-conversation/tests/input-bar.spec.tsx @@ -23,7 +23,7 @@ const SID = 's1' as SessionId function snapshotOf(overrides: Partial = {}): ConversationSnapshot { return { - sessionId: SID, nodes: [], partial: null, runningCalls: [], codeDispatches: new Map(), + sessionId: SID, nodes: [], turnEnds: new Map(), partial: null, runningCalls: [], codeDispatches: new Map(), pending: [], queue: [], running: false, composerPhase: 'active', removed: false, openState: 'open', openError: null, hasMore: false, loadingOlder: false, promptError: null, blank: false, subagent: null, lastAgentError: null, diff --git a/packages/client/ui-conversation/tests/input-matrix.spec.tsx b/packages/client/ui-conversation/tests/input-matrix.spec.tsx index 45ce41c3b8..c91631e374 100644 --- a/packages/client/ui-conversation/tests/input-matrix.spec.tsx +++ b/packages/client/ui-conversation/tests/input-matrix.spec.tsx @@ -26,7 +26,7 @@ const SID = 's1' as SessionId /** Standard-props InputBar mount over a real shell (the composer-bar entry shape). */ function mountBar(shell: SessionInputShell, over?: { running?: boolean; disabled?: boolean }) { const session = createSnapshotStore({ - sessionId: SID, nodes: [], partial: null, runningCalls: [], codeDispatches: new Map(), + sessionId: SID, nodes: [], turnEnds: new Map(), partial: null, runningCalls: [], codeDispatches: new Map(), pending: [], queue: [], running: over?.running ?? false, composerPhase: 'active', removed: over?.disabled ?? false, openState: 'open', openError: null, hasMore: false, loadingOlder: false, promptError: null, blank: false, subagent: null, lastAgentError: null, diff --git a/packages/client/ui-conversation/tests/input-scenarios.spec.tsx b/packages/client/ui-conversation/tests/input-scenarios.spec.tsx index a4eb2e668f..157419fd79 100644 --- a/packages/client/ui-conversation/tests/input-scenarios.spec.tsx +++ b/packages/client/ui-conversation/tests/input-scenarios.spec.tsx @@ -112,7 +112,7 @@ async function scopedBench(register?: (slash: SlashService) => void) { actx.on('slash/input-consume-token', req => shell.consumeToken(req.guard) ? true : undefined) const wiring = shell const sessionStore = createSnapshotStore({ - sessionId, nodes: [], partial: null, runningCalls: [], codeDispatches: new Map(), + sessionId, nodes: [], turnEnds: new Map(), partial: null, runningCalls: [], codeDispatches: new Map(), pending: [], queue: [], running: false, composerPhase: 'active', removed: false, openState: 'open', openError: null, hasMore: false, loadingOlder: false, promptError: null, blank: false, subagent: null, lastAgentError: null, diff --git a/packages/client/ui-conversation/tests/queue-dock.spec.tsx b/packages/client/ui-conversation/tests/queue-dock.spec.tsx index 6c11119c0a..321d7250fa 100644 --- a/packages/client/ui-conversation/tests/queue-dock.spec.tsx +++ b/packages/client/ui-conversation/tests/queue-dock.spec.tsx @@ -28,7 +28,7 @@ function row(id: string, text: string | null, preview = text ?? '[image]'): Queu function snapshotWith(queue: QueuedMessage[]): ConversationSnapshot { return { - sessionId: SID, nodes: [], partial: null, runningCalls: [], codeDispatches: new Map(), + sessionId: SID, nodes: [], turnEnds: new Map(), partial: null, runningCalls: [], codeDispatches: new Map(), pending: [], queue, running: true, composerPhase: 'active', removed: false, openState: 'open', openError: null, hasMore: false, loadingOlder: false, promptError: null, blank: false, subagent: null, lastAgentError: null, } diff --git a/packages/client/ui-conversation/tests/read-card.spec.tsx b/packages/client/ui-conversation/tests/read-card.spec.tsx index 700c8c3d75..bf9898cff9 100644 --- a/packages/client/ui-conversation/tests/read-card.spec.tsx +++ b/packages/client/ui-conversation/tests/read-card.spec.tsx @@ -283,7 +283,7 @@ describe('DetailsPanel Output section (read)', () => { function snapshot(over: Partial = {}): ConversationSnapshot { return { - sessionId: SID, nodes: [], partial: null, runningCalls: [], codeDispatches: new Map(), + sessionId: SID, nodes: [], turnEnds: new Map(), partial: null, runningCalls: [], codeDispatches: new Map(), pending: [], queue: [], running: false, composerPhase: 'active', removed: false, openState: 'open', openError: null, hasMore: false, loadingOlder: false, promptError: null, blank: false, subagent: null, lastAgentError: null, ...over, diff --git a/packages/client/ui-conversation/tests/search-card.spec.tsx b/packages/client/ui-conversation/tests/search-card.spec.tsx index b4306d064d..c50c308edf 100644 --- a/packages/client/ui-conversation/tests/search-card.spec.tsx +++ b/packages/client/ui-conversation/tests/search-card.spec.tsx @@ -397,7 +397,7 @@ describe('DetailsPanel Output section (search)', () => { function snapshot(over: Partial = {}): ConversationSnapshot { return { - sessionId: SID, nodes: [], partial: null, runningCalls: [], codeDispatches: new Map(), + sessionId: SID, nodes: [], turnEnds: new Map(), partial: null, runningCalls: [], codeDispatches: new Map(), pending: [], queue: [], running: false, composerPhase: 'active', removed: false, openState: 'open', openError: null, hasMore: false, loadingOlder: false, promptError: null, blank: false, subagent: null, lastAgentError: null, ...over, diff --git a/packages/client/ui-conversation/tests/skeleton.spec.tsx b/packages/client/ui-conversation/tests/skeleton.spec.tsx index eff655d229..413adf7a86 100644 --- a/packages/client/ui-conversation/tests/skeleton.spec.tsx +++ b/packages/client/ui-conversation/tests/skeleton.spec.tsx @@ -68,7 +68,7 @@ const workspaceState = (items: readonly WorkspaceView[]): WorkspaceListState => function conversationSnapshot(overrides: Partial = {}): ConversationSnapshot { return { - sessionId: SID, nodes: [], partial: null, runningCalls: [], codeDispatches: new Map(), + sessionId: SID, nodes: [], turnEnds: new Map(), partial: null, runningCalls: [], codeDispatches: new Map(), pending: [], queue: [], running: false, composerPhase: 'active', removed: false, openState: 'open', openError: null, hasMore: false, loadingOlder: false, promptError: null, blank: false, subagent: null, lastAgentError: null, diff --git a/packages/client/ui-conversation/tests/terminal-card.spec.tsx b/packages/client/ui-conversation/tests/terminal-card.spec.tsx index 2ae268debd..babefe16f1 100644 --- a/packages/client/ui-conversation/tests/terminal-card.spec.tsx +++ b/packages/client/ui-conversation/tests/terminal-card.spec.tsx @@ -455,7 +455,7 @@ describe('DetailsPanel Output section', () => { function snapshot(over: Partial = {}): ConversationSnapshot { return { - sessionId: SID, nodes: [], partial: null, runningCalls: [], codeDispatches: new Map(), + sessionId: SID, nodes: [], turnEnds: new Map(), partial: null, runningCalls: [], codeDispatches: new Map(), pending: [], queue: [], running: false, composerPhase: 'active', removed: false, openState: 'open', openError: null, hasMore: false, loadingOlder: false, promptError: null, blank: false, subagent: null, lastAgentError: null, ...over, diff --git a/packages/client/ui-conversation/tests/web-card.spec.tsx b/packages/client/ui-conversation/tests/web-card.spec.tsx index a60c16f77c..974df2264a 100644 --- a/packages/client/ui-conversation/tests/web-card.spec.tsx +++ b/packages/client/ui-conversation/tests/web-card.spec.tsx @@ -232,7 +232,7 @@ describe('DetailsPanel web Output section', () => { function snapshot(over: Partial = {}): ConversationSnapshot { return { - sessionId: SID, nodes: [], partial: null, runningCalls: [], codeDispatches: new Map(), + sessionId: SID, nodes: [], turnEnds: new Map(), partial: null, runningCalls: [], codeDispatches: new Map(), pending: [], queue: [], running: false, composerPhase: 'active', removed: false, openState: 'open', openError: null, hasMore: false, loadingOlder: false, promptError: null, blank: false, subagent: null, lastAgentError: null, ...over, From d9a370e185d52291ae57959fb2ff2e9a373a5ed2 Mon Sep 17 00:00:00 2001 From: kingwl Date: Sun, 2 Aug 2026 17:53:38 +0800 Subject: [PATCH 15/48] test(web): refresh completed-turn-tail snapshots --- apps/web/tests/snapshots/code-mode-round/ui.expected.md | 2 -- apps/web/tests/snapshots/cordis-tool-round/ui.expected.md | 2 -- apps/web/tests/snapshots/fresh-round-trip/ui.expected.md | 2 -- .../web/tests/snapshots/lifecycle-chrome/reloaded.expected.md | 2 -- apps/web/tests/snapshots/live-interactions/cancel.expected.md | 2 -- .../tests/snapshots/live-interactions/error-auth.expected.md | 2 -- .../web/tests/snapshots/live-interactions/loading.expected.md | 2 -- apps/web/tests/snapshots/live-interactions/retry.expected.md | 2 -- apps/web/tests/snapshots/plan-review/approved.expected.md | 2 -- .../tests/snapshots/question-composer/answered.expected.md | 2 -- apps/web/tests/snapshots/queue-actions/collapsed.expected.md | 2 -- apps/web/tests/snapshots/queue-actions/editing.expected.md | 2 -- apps/web/tests/snapshots/queue-actions/preserved.expected.md | 4 ---- apps/web/tests/snapshots/queue-actions/ui.expected.md | 2 -- .../tests/snapshots/seeded-history/command-row.expected.md | 2 -- apps/web/tests/snapshots/seeded-history/ui.expected.md | 2 -- apps/web/tests/snapshots/steering/mid-steer.expected.md | 2 -- apps/web/tests/snapshots/steering/settled.expected.md | 2 -- apps/web/tests/snapshots/subagent-conversation/ui.expected.md | 4 ---- apps/web/tests/snapshots/web-search-round/ui.expected.md | 2 -- 20 files changed, 44 deletions(-) diff --git a/apps/web/tests/snapshots/code-mode-round/ui.expected.md b/apps/web/tests/snapshots/code-mode-round/ui.expected.md index 8f4c7e5bf2..239c4d4548 100644 --- a/apps/web/tests/snapshots/code-mode-round/ui.expected.md +++ b/apps/web/tests/snapshots/code-mode-round/ui.expected.md @@ -7,8 +7,6 @@ - text: "Using ONE run_code program: run bash `echo CODE_ROUND_OK`, then read the file missing.txt catching its error in the program. Return an object with both outcomes. Then reply DONE and stop. {{clock}}" - button "Copy": - img -- button "Branch into a new conversation": - - img - button "Context injection": - img - img diff --git a/apps/web/tests/snapshots/cordis-tool-round/ui.expected.md b/apps/web/tests/snapshots/cordis-tool-round/ui.expected.md index e4d5ac8426..cc502f2b5f 100644 --- a/apps/web/tests/snapshots/cordis-tool-round/ui.expected.md +++ b/apps/web/tests/snapshots/cordis-tool-round/ui.expected.md @@ -7,8 +7,6 @@ - text: "Use only Cordis tools. First call cordis_inspect with what \"temporary\". Then call cordis_mount with this exact code: \"return { name: \\\"snapshot-noop\\\", apply(ctx) {} }\". Read its returned id and call cordis_unmount with that exact id. After all three calls succeed, reply exactly CORDIS_UI_DONE and stop. {{clock}}" - button "Copy": - img -- button "Branch into a new conversation": - - img - button "Context injection": - img - img diff --git a/apps/web/tests/snapshots/fresh-round-trip/ui.expected.md b/apps/web/tests/snapshots/fresh-round-trip/ui.expected.md index facb7b58cc..73817d6c7e 100644 --- a/apps/web/tests/snapshots/fresh-round-trip/ui.expected.md +++ b/apps/web/tests/snapshots/fresh-round-trip/ui.expected.md @@ -7,8 +7,6 @@ - text: "Use the bash tool to run exactly: echo WEB_E2E_OK. Then reply with the single word DONE and stop. {{clock}}" - button "Copy": - img -- button "Branch into a new conversation": - - img - button "Context injection": - img - img diff --git a/apps/web/tests/snapshots/lifecycle-chrome/reloaded.expected.md b/apps/web/tests/snapshots/lifecycle-chrome/reloaded.expected.md index 5965797c69..d961d20c4f 100644 --- a/apps/web/tests/snapshots/lifecycle-chrome/reloaded.expected.md +++ b/apps/web/tests/snapshots/lifecycle-chrome/reloaded.expected.md @@ -7,8 +7,6 @@ - text: Reply with the single word LIGHTHOUSE and stop. {{clock}} - button "Copy": - img -- button "Branch into a new conversation": - - img - button "Context injection": - img - img diff --git a/apps/web/tests/snapshots/live-interactions/cancel.expected.md b/apps/web/tests/snapshots/live-interactions/cancel.expected.md index d1e4d2bbef..57d2ea37dc 100644 --- a/apps/web/tests/snapshots/live-interactions/cancel.expected.md +++ b/apps/web/tests/snapshots/live-interactions/cancel.expected.md @@ -7,8 +7,6 @@ - text: Reply with a one-sentence description of event sourcing, then stop. {{clock}} - button "Copy": - img -- button "Branch into a new conversation": - - img - button "Context injection": - img - img diff --git a/apps/web/tests/snapshots/live-interactions/error-auth.expected.md b/apps/web/tests/snapshots/live-interactions/error-auth.expected.md index fb9337e978..9539eb180f 100644 --- a/apps/web/tests/snapshots/live-interactions/error-auth.expected.md +++ b/apps/web/tests/snapshots/live-interactions/error-auth.expected.md @@ -7,8 +7,6 @@ - text: Reply with a one-sentence description of event sourcing, then stop. {{clock}} - button "Copy": - img -- button "Branch into a new conversation": - - img - button "Context injection": - img - img diff --git a/apps/web/tests/snapshots/live-interactions/loading.expected.md b/apps/web/tests/snapshots/live-interactions/loading.expected.md index a5dfd08fb5..caee0ad6aa 100644 --- a/apps/web/tests/snapshots/live-interactions/loading.expected.md +++ b/apps/web/tests/snapshots/live-interactions/loading.expected.md @@ -7,8 +7,6 @@ - text: Reply with a one-sentence description of event sourcing, then stop. {{clock}} - button "Copy": - img -- button "Branch into a new conversation": - - img - button "Context injection": - img - img diff --git a/apps/web/tests/snapshots/live-interactions/retry.expected.md b/apps/web/tests/snapshots/live-interactions/retry.expected.md index 6380eaf5c6..a46b0966a9 100644 --- a/apps/web/tests/snapshots/live-interactions/retry.expected.md +++ b/apps/web/tests/snapshots/live-interactions/retry.expected.md @@ -7,8 +7,6 @@ - text: Reply with a one-sentence description of event sourcing, then stop. {{clock}} - button "Copy": - img -- button "Branch into a new conversation": - - img - button "Context injection": - img - img diff --git a/apps/web/tests/snapshots/plan-review/approved.expected.md b/apps/web/tests/snapshots/plan-review/approved.expected.md index 5be3f83247..cd0d50c095 100644 --- a/apps/web/tests/snapshots/plan-review/approved.expected.md +++ b/apps/web/tests/snapshots/plan-review/approved.expected.md @@ -8,8 +8,6 @@ - text: "plan Plan mode on. Use /plan off to leave. Plan a small change: add a --greeting flag to a CLI. Do not read or write any files. Call exit_plan_mode with a short plan of at most five bullet points. Once the plan is approved, reply with the single word DONE and stop. {{clock}}" - button "Copy": - img -- button "Branch into a new conversation": - - img - button "Context injection": - img - img diff --git a/apps/web/tests/snapshots/question-composer/answered.expected.md b/apps/web/tests/snapshots/question-composer/answered.expected.md index 28297569ab..e02232a945 100644 --- a/apps/web/tests/snapshots/question-composer/answered.expected.md +++ b/apps/web/tests/snapshots/question-composer/answered.expected.md @@ -7,8 +7,6 @@ - text: "Use the ask_user_question tool to ask me exactly one question with id \"color\", question \"Which color do you prefer?\", header \"Pick one\", and two options: label \"Blue\" with description \"A cool recessive hue that reads as calm and trustworthy in long reading sessions and dense dashboards.\", and label \"Green\" with description \"A restful mid-spectrum hue with the highest perceived brightness, easiest on the eye over long sessions.\" After I answer, reply with the single word DONE and stop. {{clock}}" - button "Copy": - img -- button "Branch into a new conversation": - - img - button "Context injection": - img - img diff --git a/apps/web/tests/snapshots/queue-actions/collapsed.expected.md b/apps/web/tests/snapshots/queue-actions/collapsed.expected.md index b9dee060ac..f51e7b3a1b 100644 --- a/apps/web/tests/snapshots/queue-actions/collapsed.expected.md +++ b/apps/web/tests/snapshots/queue-actions/collapsed.expected.md @@ -7,8 +7,6 @@ - text: Reply with a one-sentence description of event sourcing, then stop. {{clock}} - button "Copy": - img -- button "Branch into a new conversation": - - img - button "Context injection": - img - img diff --git a/apps/web/tests/snapshots/queue-actions/editing.expected.md b/apps/web/tests/snapshots/queue-actions/editing.expected.md index 3a70840713..c9e6229194 100644 --- a/apps/web/tests/snapshots/queue-actions/editing.expected.md +++ b/apps/web/tests/snapshots/queue-actions/editing.expected.md @@ -7,8 +7,6 @@ - text: Reply with a one-sentence description of event sourcing, then stop. {{clock}} - button "Copy": - img -- button "Branch into a new conversation": - - img - button "Context injection": - img - img diff --git a/apps/web/tests/snapshots/queue-actions/preserved.expected.md b/apps/web/tests/snapshots/queue-actions/preserved.expected.md index 0a43cee68b..4da58812dd 100644 --- a/apps/web/tests/snapshots/queue-actions/preserved.expected.md +++ b/apps/web/tests/snapshots/queue-actions/preserved.expected.md @@ -7,8 +7,6 @@ - text: Reply with a one-sentence description of event sourcing, then stop. {{clock}} - button "Copy": - img -- button "Branch into a new conversation": - - img - button "Context injection": - img - img @@ -22,8 +20,6 @@ - text: {{clock}} Edited queue item {{clock}} - button "Copy": - img -- button "Branch into a new conversation": - - img - paragraph: partial - status: Deep diving... - list: diff --git a/apps/web/tests/snapshots/queue-actions/ui.expected.md b/apps/web/tests/snapshots/queue-actions/ui.expected.md index 24edb57417..0b70af3f64 100644 --- a/apps/web/tests/snapshots/queue-actions/ui.expected.md +++ b/apps/web/tests/snapshots/queue-actions/ui.expected.md @@ -7,8 +7,6 @@ - text: Reply with a one-sentence description of event sourcing, then stop. {{clock}} - button "Copy": - img -- button "Branch into a new conversation": - - img - button "Context injection": - img - img diff --git a/apps/web/tests/snapshots/seeded-history/command-row.expected.md b/apps/web/tests/snapshots/seeded-history/command-row.expected.md index b916a3add2..e3f8cdc925 100644 --- a/apps/web/tests/snapshots/seeded-history/command-row.expected.md +++ b/apps/web/tests/snapshots/seeded-history/command-row.expected.md @@ -7,8 +7,6 @@ - text: "Use the read tool twice in one assistant message: read a.txt and b.txt. Then reply with the single word DONE and stop. 7/25 {{clock}}" - button "Copy": - img -- button "Branch into a new conversation": - - img - button "Think The user wants me to read a.txt and b.txt, then reply with \"DONE\". Let me do both reads in parallel.": - img - img diff --git a/apps/web/tests/snapshots/seeded-history/ui.expected.md b/apps/web/tests/snapshots/seeded-history/ui.expected.md index a168d5e2a3..fee3747f32 100644 --- a/apps/web/tests/snapshots/seeded-history/ui.expected.md +++ b/apps/web/tests/snapshots/seeded-history/ui.expected.md @@ -7,8 +7,6 @@ - text: "Use the read tool twice in one assistant message: read a.txt and b.txt. Then reply with the single word DONE and stop. 7/25 {{clock}}" - button "Copy": - img -- button "Branch into a new conversation": - - img - button "Think The user wants me to read a.txt and b.txt, then reply with \"DONE\". Let me do both reads in parallel.": - img - img diff --git a/apps/web/tests/snapshots/steering/mid-steer.expected.md b/apps/web/tests/snapshots/steering/mid-steer.expected.md index 27b40ef442..857e80e19c 100644 --- a/apps/web/tests/snapshots/steering/mid-steer.expected.md +++ b/apps/web/tests/snapshots/steering/mid-steer.expected.md @@ -7,8 +7,6 @@ - text: Use the ask_user_question tool to ask me exactly one question with id "checkpoint", question "Ready to continue?", header "Checkpoint", and options labeled "Yes" and "No". After I answer, reply with one short sentence acknowledging my answer and stop. {{clock}} - button "Copy": - img -- button "Branch into a new conversation": - - img - button "Context injection": - img - img diff --git a/apps/web/tests/snapshots/steering/settled.expected.md b/apps/web/tests/snapshots/steering/settled.expected.md index 72e28598e5..6eb6600ccd 100644 --- a/apps/web/tests/snapshots/steering/settled.expected.md +++ b/apps/web/tests/snapshots/steering/settled.expected.md @@ -7,8 +7,6 @@ - text: Use the ask_user_question tool to ask me exactly one question with id "checkpoint", question "Ready to continue?", header "Checkpoint", and options labeled "Yes" and "No". After I answer, reply with one short sentence acknowledging my answer and stop. {{clock}} - button "Copy": - img -- button "Branch into a new conversation": - - img - button "Context injection": - img - img diff --git a/apps/web/tests/snapshots/subagent-conversation/ui.expected.md b/apps/web/tests/snapshots/subagent-conversation/ui.expected.md index 6513bb60e3..37588b7756 100644 --- a/apps/web/tests/snapshots/subagent-conversation/ui.expected.md +++ b/apps/web/tests/snapshots/subagent-conversation/ui.expected.md @@ -12,8 +12,6 @@ - text: Explain event sourcing in one sentence. {{clock}} - button "Copy": - img -- button "Branch into a new conversation": - - img - button "Context injection": - img - img @@ -30,8 +28,6 @@ - text: {{clock}} Now give the same explanation to a human reader. {{clock}} - button "Copy": - img -- button "Branch into a new conversation": - - img - button "Think The user is asking for a one-sentence description of event sourcing. This is a straightforward knowledge question that doesn't require any skill loading or tool calls.": - img - img diff --git a/apps/web/tests/snapshots/web-search-round/ui.expected.md b/apps/web/tests/snapshots/web-search-round/ui.expected.md index 4ff674462e..c0e55a5051 100644 --- a/apps/web/tests/snapshots/web-search-round/ui.expected.md +++ b/apps/web/tests/snapshots/web-search-round/ui.expected.md @@ -7,8 +7,6 @@ - text: Use web_search to search exactly "DeepSeek Harness snapshot search". Then reply exactly SEARCH_DONE and stop. {{clock}} - button "Copy": - img -- button "Branch into a new conversation": - - img - button "Context injection": - img - img From b30686d634a3fb3bc5f93f89ca10bfcd631b847c Mon Sep 17 00:00:00 2001 From: kingwl Date: Sun, 2 Aug 2026 23:31:24 +0800 Subject: [PATCH 16/48] show subagent usage and active duration --- ...07-27-web-subagent-conversations.i18n.yaml | 4 +- .../2026-07-27-web-subagent-conversations.md | 14 +-- ...026-07-27-web-subagent-conversations.zh.md | 14 +-- .../subagent-conversation/tree.expected.md | 8 +- apps/web/tests/subagent-conversation.e2e.ts | 6 ++ docs/cordis-catalog/events.md | 8 +- docs/cordis-catalog/services.md | 2 +- docs/event-producer-consumer.md | 8 +- packages/client/runtime/README.i18n.yaml | 4 +- packages/client/runtime/README.md | 2 +- packages/client/runtime/README.zh.md | 2 +- .../runtime/src/client/sessions/lineage.ts | 5 + .../runtime/src/client/sessions/manager.ts | 13 ++- .../src/client/sessions/projection-store.ts | 15 +++ .../runtime/src/client/sessions/service.ts | 6 ++ .../runtime/tests/projection-store.spec.ts | 41 ++++++++ packages/client/ui-subagent/README.i18n.yaml | 4 +- packages/client/ui-subagent/README.md | 4 +- packages/client/ui-subagent/README.zh.md | 4 +- packages/client/ui-subagent/package.json | 4 + .../client/SubagentCatalogAction.module.css | 6 +- .../src/client/SubagentCatalogAction.tsx | 95 +++++++++++++++---- .../tests/conversation-ui.spec.tsx | 63 +++++++----- packages/client/ui-subagent/tsconfig.json | 6 ++ packages/subagent/subagent/README.i18n.yaml | 4 +- packages/subagent/subagent/README.md | 2 + packages/subagent/subagent/README.zh.md | 2 + packages/subagent/subagent/package.json | 13 +++ packages/subagent/subagent/src/client.ts | 7 ++ packages/subagent/subagent/src/index.ts | 5 + .../subagent/subagent/src/projection-types.ts | 20 ++++ packages/subagent/subagent/src/projection.ts | 68 +++++++++++++ .../subagent/tests/timing-projection.spec.ts | 51 ++++++++++ packages/subagent/subagent/tsconfig.json | 3 + pnpm-lock.yaml | 13 +++ 35 files changed, 438 insertions(+), 88 deletions(-) create mode 100644 packages/subagent/subagent/src/client.ts create mode 100644 packages/subagent/subagent/src/projection-types.ts create mode 100644 packages/subagent/subagent/src/projection.ts create mode 100644 packages/subagent/subagent/tests/timing-projection.spec.ts diff --git a/.agents/notes/implemented/feature/2026-07-27-web-subagent-conversations.i18n.yaml b/.agents/notes/implemented/feature/2026-07-27-web-subagent-conversations.i18n.yaml index bd780ca60d..594934ec07 100644 --- a/.agents/notes/implemented/feature/2026-07-27-web-subagent-conversations.i18n.yaml +++ b/.agents/notes/implemented/feature/2026-07-27-web-subagent-conversations.i18n.yaml @@ -2,5 +2,5 @@ # side as of the last confirmed-consistent state. Both languages carry equal authority; # after editing either side, bring the other along and re-record with: # pnpm run verify-translation-pairing --write .agents/notes/implemented/feature/2026-07-27-web-subagent-conversations.md -2026-07-27-web-subagent-conversations.md: 34acb1410cf6316bca2980ed012046ffab9623f6 -2026-07-27-web-subagent-conversations.zh.md: 5dcd7025c5cd03fed34266834795de1f2b630648 +2026-07-27-web-subagent-conversations.md: 6658bcc960ec691f3646f4ff08d8a065d3a8d70a +2026-07-27-web-subagent-conversations.zh.md: 5e89c9a7a420687c53be961245de7adb99bf3c44 diff --git a/.agents/notes/implemented/feature/2026-07-27-web-subagent-conversations.md b/.agents/notes/implemented/feature/2026-07-27-web-subagent-conversations.md index 34acb1410c..6658bcc960 100644 --- a/.agents/notes/implemented/feature/2026-07-27-web-subagent-conversations.md +++ b/.agents/notes/implemented/feature/2026-07-27-web-subagent-conversations.md @@ -33,7 +33,7 @@ The Figma [subagent list](https://www.figma.com/design/jRBBK7zBgcszdVWQ0Fh5J8/Ha | The session header opens a compact child list. | The trigger aggregates the complete subagent-only descendant lineage; the tree shows every direct catalog entry in service order, including disabled diagnostics. | | Selecting a row reuses the conversation UI. | Addressed history never activates the child; only a continuable row with a live parent retains the ordinary composer. | | Nested agents expand progressively. | Each row carries a one-level `hasChildren` snapshot; disclosure reserves known direct-descendant rows immediately, then loads only that row's direct catalog and retains its own parent address. | -| Rows show labels, state, and relative time without duplicating sidebar rows. | Mode and `running`/`inactive` activity are textual as well as visual; optional title and time come from summaries. `SessionHeader.origin` removes duplicate navigation rows but grants no capability. | +| Rows show labels, state, usage, and active duration without duplicating sidebar rows. | Mode and `running`/`inactive` activity are textual as well as visual; optional title, durable token usage, and exact active-turn duration come from the list's retained projection values. `SessionHeader.origin` removes duplicate navigation rows but grants no capability. | ## Product contract @@ -41,6 +41,8 @@ The header action is absent only after a complete empty direct-catalog response. `running` means the exact child Agent driver is draining work at the Host sampling boundary; `inactive` means that driver is idle or absent. The UI does not translate either value into success, failure, cancellation, completeness, or resumability. `subagent.list` supplies the current driver-status baseline, `host/session-status` updates known activity in place, request-local replay prevents an older in-flight list response from overwriting a newer transition, and `host/session-removed` returns a known row to `inactive`; reconnect reads a fresh baseline. A `host/session-added` frame for a direct subagent immediately flips any loaded parent row to `hasChildren: true`, and that positive hint survives an older in-flight catalog response; membership, labels, mode, diagnostics, and the authoritative snapshot still require a debounced `subagent.list` refresh while the affected branch is open. A prompt response remains delivery-time authority. +Healthy rows reuse the standard session projections retained in the list mirror. The token figure sums the four disjoint `tokenUsage` buckets across the durable log. `subagentTiming` resets at every descriptor so an inherited fork seed cannot enter the child's total, accumulates completed `turn/start` → `turn/end` spans, and carries the current turn's `activeSince`. The menu formats whole seconds and advances its local clock only while a known descendant is running; an inactive row uses settled duration, or the summary's last activity to bound an interrupted open turn, so reopening the menu never restarts completed work. Token chunks do not change `subagentTiming` and therefore do not add a per-token list update path. Neither metric implies a durable outcome. + Selecting a row records its exact address before opening the resident client `Session`. History pagination, event folding, tool render intents, titles, and live mux reconciliation reuse the ordinary conversation machinery. Breadcrumbs use catalog labels, follow parent links only through `origin: 'subagent'` rows, include the first ordinary owner, and keep ordinary forks single-level. Forking an addressed subagent creates an ordinary fork with direct source lineage and attaches it to the nearest workspace-owning ancestor. The catalog is an ARIA tree with lazy ArrowRight/ArrowLeft disclosure, linear ArrowUp/ArrowDown navigation, Home/End, Escape, and focus restoration. A one-shot row always replaces the composer with copy explaining that the execution record is read-only. A continuable row does so only while `parentAvailable` is false. When enabled, its Send action admits another FIFO turn even if the child is currently running; it never becomes Stop. Prompt failures retain the draft through the ordinary error behavior. @@ -65,7 +67,7 @@ The adapter stays in `dsh-host-apiproxy`; `dsh-host-webserver` remains a carrier ## Client object layer and presentation -The React-free runtime owns catalogs, single-flight refreshes, retained addresses, availability hints, and transport selection. Re-selecting a known child retains its address so navigation cannot silently switch to ordinary session APIs. A missing intermediate breadcrumb address can be recovered from an already-loaded ancestor catalog, but it is not retained for transport and creates no scope until the user selects that breadcrumb. Restored navigation persists the full mode-bearing address. +The React-free runtime owns catalogs, single-flight refreshes, retained addresses, availability hints, transport selection, and a reference-stable map of each list row's current projection values. Re-selecting a known child retains its address so navigation cannot silently switch to ordinary session APIs. A missing intermediate breadcrumb address can be recovered from an already-loaded ancestor catalog, but it is not retained for transport and creates no scope until the user selects that breadcrumb. Restored navigation persists the full mode-bearing address. Catalogs ride the standard `useSessions` snapshot. Component-local state owns menu visibility, expanded branches, and focus. `ui-conversation` declares the generic header-action list slot and dispatches the current conversation snapshot through its composer chain; it contains no subagent-specific takeover flag. `@deepseek-ai/dsh-client-ui-subagent` registers the catalog action and elects a reason-specific read-only composer from ordinary owner props. Components receive derived props and callbacks, never `ctx`. @@ -102,14 +104,14 @@ The shipped Web composition mounts SQLite session query beside JSONL persistence - Host protocol tests pin schemas including required boolean expandability, id echoing, mode verification, non-activating history, exact-parent enforcement, FIFO admission receipts, cancellation, and sanitized failure mapping. - Generic Host tests pin attached and cold history and forks without Agent publication, cold projection folding, descriptor/origin/runtime-owner denial, explicit-id adoption denial, and the direct queue-control fence. - Client object tests pin retained and restored addresses, one-shot read-only rejection, history routing, continuable prompt routing, no addressed cancellation, suppression of Agent-bound model controls, live activity flips including in-flight response replay and detach fallback, subagent-parent expandability flips, and membership refresh. -- jsdom tests pin the aggregate descendant count and activity, known loading-row shape, mixed-mode rows, pre-click leaf disclosure, diagnostics, lazy descendant disclosure, direct-parent addresses, keyboard behavior, and both read-only reasons. -- The keyless assembled Web snapshot contains an inactive continuable child, an inactive one-shot sibling, and a persisted grandchild; it pins the three-descendant trigger and aggregate running transition, expands without activation, opens persisted history, admits a human FIFO follow-up, reconciles child mux events, and proves one-shot history remains read-only. +- jsdom tests pin the aggregate descendant count and activity, token totals, second-precision running and frozen inactive durations, known loading-row shape, mixed-mode rows, pre-click leaf disclosure, diagnostics, lazy descendant disclosure, direct-parent addresses, keyboard behavior, and both read-only reasons. +- The keyless assembled Web snapshot contains an inactive continuable child with durable usage, an inactive one-shot sibling, and a persisted grandchild; it pins the three-descendant trigger, usage and timing rows, and aggregate running transition, expands without activation, opens persisted history, admits a human FIFO follow-up, reconciles child mux events, and proves one-shot history remains read-only. - Navigation tests pin subagent-only breadcrumbs, workspace placement for forks created from subagents, and `origin: 'subagent'` sidebar filtering without hiding ordinary forks. ## Consequences -- Catalog reads may rescan persisted lineage and each direct candidate's descriptor log, but expandability reuses only descendant headers already present in that trace; the Web activity baseline adds one Agent-registry lookup per healthy row and then uses existing live frames, while membership refresh stays debounced and single-flight. +- Catalog reads may rescan persisted lineage and each direct candidate's descriptor log, but expandability reuses only descendant headers already present in that trace; the Web activity baseline adds one Agent-registry lookup per healthy row and then uses existing live frames, while usage and duration reuse projection baselines and pushes with no per-row log read, and membership refresh stays debounced and single-flight. - Parent availability, child activity, and `hasChildren` are snapshots. Publication, disposal, another sender, or another process may win after listing; typed prompt failure remains expected. - A child may publish between history fetch and mux subscription, so the existing sequence reconciliation also covers the cold-to-live addressed path. - Persisted origin adds one deliberately weak product-classification field to child headers and list projections; it cannot become an authorization shortcut. -- The UI has no child cancellation, durable outcome, activation duration, deletion, or independently interactive offline mode, and its text must not imply those capabilities. +- The UI has no child cancellation, durable outcome, Activation identity, deletion, or independently interactive offline mode, and its text must not imply those capabilities. Active-turn duration measures logged work rather than Activation residency. diff --git a/.agents/notes/implemented/feature/2026-07-27-web-subagent-conversations.zh.md b/.agents/notes/implemented/feature/2026-07-27-web-subagent-conversations.zh.md index 5dcd7025c5..5e89c9a7a4 100644 --- a/.agents/notes/implemented/feature/2026-07-27-web-subagent-conversations.zh.md +++ b/.agents/notes/implemented/feature/2026-07-27-web-subagent-conversations.zh.md @@ -33,7 +33,7 @@ Figma 中的 [subagent 列表](https://www.figma.com/design/jRBBK7zBgcszdVWQ0Fh5 | 会话页头可打开紧凑的 child 列表。 | 触发器会汇总仅含 subagent 的完整后代谱系；树按服务顺序显示每个直接目录条目，包括已禁用的 diagnostic。 | | 选择一行会复用对话 UI。 | 已寻址历史绝不激活 child；只有 parent 存活的可继续行才保留普通输入框。 | | 嵌套 agent 会逐层展开。 | 每行携带一层 `hasChildren` 快照；展开时会立即预留已知直接后代行，随后仍只加载该行的直接目录，并保留其自身的 parent 地址。 | -| 条目显示 label、状态与相对时间，同时避免侧边栏条目重复。 | mode 与 `running`／`inactive` 活动状态会同时以文字和视觉呈现；可选 title 与时间来自摘要。`SessionHeader.origin` 会移除重复的导航条目，但不授予任何功能权限。 | +| 条目显示 label、状态、用量与活跃耗时，同时避免侧边栏条目重复。 | mode 与 `running`／`inactive` 活动状态会同时以文字和视觉呈现；可选 title、持久化 token 用量与精确的活跃轮次耗时来自列表保留的投影值。`SessionHeader.origin` 会移除重复的导航条目，但不授予任何功能权限。 | ## 产品契约 @@ -41,6 +41,8 @@ Figma 中的 [subagent 列表](https://www.figma.com/design/jRBBK7zBgcszdVWQ0Fh5 `running` 表示在 Host 采样边界，确切 child Agent driver 正在处理工作；`inactive` 表示该 driver 空闲或不存在。UI 不会把任一值解释为成功、失败、取消、完成状态或可恢复性。`subagent.list` 提供当前 driver 状态基线，`host/session-status` 会就地更新已知活动状态，请求内回放会阻止更早发起但尚未完成的列表响应覆盖较新的状态转换，`host/session-removed` 则会使已知行恢复为 `inactive`；重连时会读取新的基线。直接 subagent 的 `host/session-added` 帧会立即把任何已加载的 parent 行翻转为 `hasChildren: true`，并使这项正向提示不被更早发起但尚未完成的目录响应覆盖；受影响分支打开期间，成员、label、mode、diagnostic 与权威快照仍需要通过去抖动的 `subagent.list` 刷新来更新。消息投递时仍以提示词响应为权威依据。 +健康行会复用列表镜像中保留的标准会话投影。token 数值会汇总持久化日志中四个互不重叠的 `tokenUsage` 桶。`subagentTiming` 会在每个描述符处重置，使继承的 fork 种子不会计入 child 总量；它会累加已完成的 `turn/start` → `turn/end` 时段，并携带当前轮次的 `activeSince`。菜单会以整秒格式化时间，且仅在有已知后代处于运行状态时才推进其本地时钟；对 inactive 行，菜单使用已结算耗时，或以摘要的最后活动为被中断未结束轮次的上界，因此重新打开菜单绝不会让已完成工作重新计时。token 分片不会改变 `subagentTiming`，因此不会增加按 token 更新列表的路径。这两项指标都不蕴含持久化结果语义。 + 选择一行后，系统会先记录其确切地址，再打开常驻客户端 `Session`。历史分页、事件 fold、工具渲染意图、title 与实时 mux 归并都会复用普通对话机制。面包屑导航使用目录 label，只会沿 `origin: 'subagent'` 行的父链接逐级回溯，包含第一个普通 owner，并让普通 fork 保持单层。从已寻址 subagent 创建 fork 时，会生成具有直接源谱系的普通 fork，并将其附加到最近拥有 Workspace 的祖先。目录是一棵 ARIA 树，支持懒加载式 ArrowRight／ArrowLeft 展开与折叠、线性 ArrowUp／ArrowDown 导航、Home／End、Escape 以及焦点恢复。 one-shot 行始终会用文案替代输入框，说明执行记录为只读。可继续行仅在 `parentAvailable` 为 false 时如此。启用后，即使 child 正在运行，其 Send 操作也会准入另一个 FIFO 轮次，绝不会变成 Stop。提示词失败会通过普通错误行为保留草稿。 @@ -65,7 +67,7 @@ one-shot 行始终会用文案替代输入框，说明执行记录为只读。 ## 客户端对象层与呈现 -不依赖 React 的运行时负责目录、单次并发刷新、保留的地址、可用性提示与传输选择。再次选择已知 child 时会保留其地址，避免导航静默切换到普通会话 API。缺失的中间面包屑地址可以从已加载的祖先目录恢复，但在用户选择该面包屑之前不会保留为传输地址，也不会创建 scope。恢复的导航会持久化包含 mode 的完整地址。 +不依赖 React 的运行时负责目录、单次并发刷新、保留的地址、可用性提示、传输选择，以及每个列表行当前投影值的引用稳定映射。再次选择已知 child 时会保留其地址，避免导航静默切换到普通会话 API。缺失的中间面包屑地址可以从已加载的祖先目录恢复，但在用户选择该面包屑之前不会保留为传输地址，也不会创建 scope。恢复的导航会持久化包含 mode 的完整地址。 目录通过标准 `useSessions` 快照传递。组件局部状态负责菜单可见性、已展开分支与焦点。`ui-conversation` 声明通用页头操作列表 slot，并通过其编辑器链分发当前对话快照；其中没有 subagent 专用的接管标记。`@deepseek-ai/dsh-client-ui-subagent` 注册目录操作，并根据普通 owner props 选择按原因区分的只读编辑器。组件只接收派生 props 与回调，绝不接收 `ctx`。 @@ -102,14 +104,14 @@ one-shot 行始终会用文案替代输入框，说明执行记录为只读。 - 宿主协议测试固定 schema（包括必需的布尔可展开性）、id 回显、mode 校验、非激活式历史、确切 parent 强制要求、FIFO 准入回执、取消与脱敏后的失败映射。 - 通用 Host 测试固定在不发布 Agent 的情况下读取已附加与冷态历史及执行 fork、冷态投影归并、按描述符／origin／运行时 owner 拒绝、拒绝显式 id 接纳，以及直接队列控制栅栏。 - 客户端对象测试固定已保留与已恢复的地址、one-shot 只读拒绝、历史路由、可继续提示词路由、已寻址对话不提供取消、屏蔽绑定到 agent 的模型控件、实时活动状态翻转（包括在途响应回放与 detach 回退）、subagent parent 可展开性翻转与成员刷新。 -- jsdom 测试固定后代聚合计数与活动状态、已知加载行的形态、混合 mode 行、点击前的叶子展开控件、diagnostic、后代懒加载展开、直接 parent 地址、键盘行为与两种只读原因。 -- 无密钥的组装 Web 快照包含一个 inactive 的可继续 child、一个 inactive 的 one-shot sibling 和一个持久化 grandchild；它会固定触发器显示三个后代及聚合 `running` 状态转换，在不激活的情况下展开、打开持久化历史、准入一条用户 FIFO 后续消息、归并 child mux 事件，并证明 one-shot 历史仍然只读。 +- jsdom 测试固定后代聚合计数与活动状态、token 总量、精确到秒的运行中耗时与冻结后 inactive 耗时、已知加载行的形态、混合 mode 行、点击前的叶子展开控件、diagnostic、后代懒加载展开、直接 parent 地址、键盘行为与两种只读原因。 +- 无密钥的组装 Web 快照包含一个具有持久化用量的 inactive 可继续 child、一个 inactive 的 one-shot sibling 和一个持久化 grandchild；它会固定触发器显示三个后代、用量与计时行，以及聚合 `running` 状态转换，在不激活的情况下展开、打开持久化历史、准入一条用户 FIFO 后续消息、归并 child mux 事件，并证明 one-shot 历史仍然只读。 - 导航测试固定仅含 subagent 的面包屑导航、从 subagent 创建 fork 时的 Workspace 归属，以及 `origin: 'subagent'` 侧边栏过滤，同时不隐藏普通 fork。 ## 后果 -- 目录读取可能重新扫描持久化谱系与每个直接候选的描述符日志，但可展开性只复用该追踪中已有的后代 header；Web 活动基线会为每个健康行增加一次 Agent 注册表查找，随后使用现有实时帧，而成员刷新保持去抖动和单次并发。 +- 目录读取可能重新扫描持久化谱系与每个直接候选的描述符日志，但可展开性只复用该追踪中已有的后代 header；Web 活动基线会为每个健康行增加一次 Agent 注册表查找，随后使用现有实时帧，而用量与耗时会复用投影基线和推送，无需按行读取日志，成员刷新则保持去抖动和单次并发。 - parent 可用性、child 活动状态与 `hasChildren` 都是快照。列出之后，发布、dispose、其他发送方或其他进程都可能抢先改变状态；类型化提示词失败仍属预期行为。 - child 可能在历史获取与 mux 订阅之间发布，因此现有序号归并也涵盖从冷态转为存活的已寻址路径。 - 持久化 origin 会为 child header 与列表投影添加一个有意保持弱约束的产品分类字段；它不能变成授权捷径。 -- UI 不提供 child 取消、持久化结果、激活耗时、删除或可独立交互的离线 mode，其文案不得暗示这些功能已经存在。 +- UI 不提供 child 取消、持久化结果、Activation 身份、删除或可独立交互的离线 mode，其文案不得暗示这些功能已经存在。活跃轮次耗时度量的是已记录工作，而非 Activation 驻留时间。 diff --git a/apps/web/tests/snapshots/subagent-conversation/tree.expected.md b/apps/web/tests/snapshots/subagent-conversation/tree.expected.md index 12520a65c4..b1e23e4020 100644 --- a/apps/web/tests/snapshots/subagent-conversation/tree.expected.md +++ b/apps/web/tests/snapshots/subagent-conversation/tree.expected.md @@ -1,8 +1,8 @@ - tree "子代理会话": - - treeitem "event-sourcing researcher Explain event sourcing in one · 可继续 · 当前未运行 刚刚" [expanded] [level=1]: + - treeitem "event-sourcing researcher Explain event sourcing in one · 可继续 · 当前未运行 7.9K tok · 2秒" [expanded] [level=1]: - button "收起 event-sourcing researcher 的下级子代理": - img - - text: event-sourcing researcher Explain event sourcing in one · 可继续 · 当前未运行 刚刚 + - text: event-sourcing researcher Explain event sourcing in one · 可继续 · 当前未运行 7.9K tok · 2秒 - group: - - treeitem "example editor 可继续 · 当前未运行 刚刚" [level=2] - - treeitem "event-sourcing reviewer 一次性 · 当前未运行 刚刚" [level=1] + - treeitem "example editor 可继续 · 当前未运行 0 tok · 0秒" [level=2] + - treeitem "event-sourcing reviewer 一次性 · 当前未运行 0 tok · 0秒" [level=1] diff --git a/apps/web/tests/subagent-conversation.e2e.ts b/apps/web/tests/subagent-conversation.e2e.ts index e71b54e0d9..2dff253b66 100644 --- a/apps/web/tests/subagent-conversation.e2e.ts +++ b/apps/web/tests/subagent-conversation.e2e.ts @@ -149,6 +149,7 @@ describe('web e2e: persisted subagent conversation and human continuation', () = data: { turn: 1, reason: { kind: 'completed' } }, }, ] as SessionEvent[]) + await scaffold.ctx.sessionProjectionCache.coldSnapshot(oneShotId) grandchildId = sessionId('recorded-grandchild') const authoredAt = Date.now() await scaffold.ctx.sessionPersistence.create({ @@ -192,6 +193,7 @@ describe('web e2e: persisted subagent conversation and human continuation', () = data: { turn: 1, reason: { kind: 'completed' } }, }, ] as SessionEvent[]) + await scaffold.ctx.sessionProjectionCache.coldSnapshot(grandchildId) expect(scaffold.ctx.agents.get(childId)).toBeUndefined() expect(scaffold.ctx.agents.get(oneShotId)).toBeUndefined() expect(scaffold.ctx.agents.get(grandchildId)).toBeUndefined() @@ -246,6 +248,10 @@ describe('web e2e: persisted subagent conversation and human continuation', () = name: `展开 ${ONE_SHOT_LABEL} 的下级子代理`, }).count()).toBe(0) await page.getByRole('button', { name: `展开 ${LABEL} 的下级子代理` }).click() + const childRow = page.getByRole('treeitem', { name: new RegExp(LABEL) }) + const childLabel = await childRow.getAttribute('aria-label') + await page.waitForTimeout(1_100) + expect(await childRow.getAttribute('aria-label')).toBe(childLabel) await page.getByRole('treeitem', { name: new RegExp(NESTED_LABEL) }).waitFor({ timeout: 15_000 }) expect(scaffold.ctx.agents.get(childId)).toBeUndefined() expect(scaffold.ctx.agents.get(grandchildId)).toBeUndefined() diff --git a/docs/cordis-catalog/events.md b/docs/cordis-catalog/events.md index 521a2fa735..8f4385938c 100644 --- a/docs/cordis-catalog/events.md +++ b/docs/cordis-catalog/events.md @@ -794,7 +794,7 @@ A published child settled. Scope-filtered dispatch uses the same delegating pare Types: [Scoped](../core-data-structures/scope.md) · [SubagentService](../core-data-structures/subagent.md) -Source: [`packages/subagent/subagent/src/index.ts:158`](../../packages/subagent/subagent/src/index.ts) +Source: [`packages/subagent/subagent/src/index.ts:160`](../../packages/subagent/subagent/src/index.ts) ### `subagent/provider-added` — emit @@ -811,7 +811,7 @@ A provider became resolvable in the registry. Types: [SubagentProvider](../core-data-structures/subagent.md) -Source: [`packages/subagent/subagent/src/index.ts:132`](../../packages/subagent/subagent/src/index.ts) +Source: [`packages/subagent/subagent/src/index.ts:134`](../../packages/subagent/subagent/src/index.ts) ### `subagent/provider-removed` — emit @@ -826,7 +826,7 @@ A provider left the registry. Accepted runs remain holder-owned. 'subagent/provider-removed'(name: string): void ``` -Source: [`packages/subagent/subagent/src/index.ts:138`](../../packages/subagent/subagent/src/index.ts) +Source: [`packages/subagent/subagent/src/index.ts:140`](../../packages/subagent/subagent/src/index.ts) ### `subagent/start` — emit @@ -848,7 +848,7 @@ A provider established a published child. For in-process providers, `ctx.agents. Types: [Scoped](../core-data-structures/scope.md) · [SubagentService](../core-data-structures/subagent.md) -Source: [`packages/subagent/subagent/src/index.ts:149`](../../packages/subagent/subagent/src/index.ts) +Source: [`packages/subagent/subagent/src/index.ts:151`](../../packages/subagent/subagent/src/index.ts) ## `system-prompt/*` diff --git a/docs/cordis-catalog/services.md b/docs/cordis-catalog/services.md index 96b876a0eb..ae426792e1 100644 --- a/docs/cordis-catalog/services.md +++ b/docs/cordis-catalog/services.md @@ -2072,7 +2072,7 @@ async start(name: string, request: SubagentStartRequest): Promise Types: [Agent](../core-data-structures/core.md) · [ContentBlock](../core-data-structures/core.md) · [ContinuableSetupContribution](../core-data-structures/subagent.md) · [ContinuableStart](../core-data-structures/subagent.md) · [ContinuableStartSpec](../core-data-structures/subagent.md) · [MessageId](../core-data-structures/core.md) · [SessionId](../core-data-structures/core.md) · [SubagentFollowupOptions](../core-data-structures/subagent.md) · [SubagentListEntry](../core-data-structures/subagent.md) · [SubagentProvider](../core-data-structures/subagent.md) · [SubagentReportOptions](../core-data-structures/subagent.md) · [SubagentRun](../core-data-structures/subagent.md) · [SubagentStartRequest](../core-data-structures/subagent.md) -Source: [`packages/subagent/subagent/src/index.ts:163`](../../packages/subagent/subagent/src/index.ts) +Source: [`packages/subagent/subagent/src/index.ts:165`](../../packages/subagent/subagent/src/index.ts) ## `ctx.subprocess` — `SubprocessService` (abstract seam) diff --git a/docs/event-producer-consumer.md b/docs/event-producer-consumer.md index d1457a75af..b59b8733d5 100644 --- a/docs/event-producer-consumer.md +++ b/docs/event-producer-consumer.md @@ -41,10 +41,10 @@ This matrix shows which packages dispatch each harness-owned event and which pac | `settings/document-updated` | `emit` | [`packages/settings/settings/src/index.ts:150`](../packages/settings/settings/src/index.ts) | [`settings`](../packages/settings/settings) (`events.dispatch`) | `apiproxy` | | `settings/updated` | `emit` | [`packages/settings/settings/src/index.ts:137`](../packages/settings/settings/src/index.ts) | [`settings`](../packages/settings/settings) (`events.dispatch`) | [`settings`](../packages/settings/settings) | | `skills/change` | `emit` | [`packages/skill/skill/src/index.ts:188`](../packages/skill/skill/src/index.ts) | [`skill`](../packages/skill/skill) (`events.dispatch`) | [`tui`](../packages/ui/tui) | -| `subagent/end` | `emit` | [`packages/subagent/subagent/src/index.ts:158`](../packages/subagent/subagent/src/index.ts) | [`subagent`](../packages/subagent/subagent) (`events.dispatch`) | [`hooks-claude`](../packages/hooks/hooks-claude), [`jsonrpc`](../packages/ui/jsonrpc), [`subagent`](../packages/subagent/subagent) | -| `subagent/provider-added` | `emit` | [`packages/subagent/subagent/src/index.ts:132`](../packages/subagent/subagent/src/index.ts) | [`subagent`](../packages/subagent/subagent) (`emit`) | [`subagent`](../packages/subagent/subagent), [`tool-subagent`](../packages/subagent/tool-subagent) | -| `subagent/provider-removed` | `emit` | [`packages/subagent/subagent/src/index.ts:138`](../packages/subagent/subagent/src/index.ts) | [`subagent`](../packages/subagent/subagent) (`events.dispatch`) | [`subagent`](../packages/subagent/subagent), [`tool-subagent`](../packages/subagent/tool-subagent) | -| `subagent/start` | `emit` | [`packages/subagent/subagent/src/index.ts:149`](../packages/subagent/subagent/src/index.ts) | [`subagent`](../packages/subagent/subagent) (`events.dispatch`) | [`hooks-claude`](../packages/hooks/hooks-claude), [`subagent`](../packages/subagent/subagent) | +| `subagent/end` | `emit` | [`packages/subagent/subagent/src/index.ts:160`](../packages/subagent/subagent/src/index.ts) | [`subagent`](../packages/subagent/subagent) (`events.dispatch`) | [`hooks-claude`](../packages/hooks/hooks-claude), [`jsonrpc`](../packages/ui/jsonrpc), [`subagent`](../packages/subagent/subagent) | +| `subagent/provider-added` | `emit` | [`packages/subagent/subagent/src/index.ts:134`](../packages/subagent/subagent/src/index.ts) | [`subagent`](../packages/subagent/subagent) (`emit`) | [`subagent`](../packages/subagent/subagent), [`tool-subagent`](../packages/subagent/tool-subagent) | +| `subagent/provider-removed` | `emit` | [`packages/subagent/subagent/src/index.ts:140`](../packages/subagent/subagent/src/index.ts) | [`subagent`](../packages/subagent/subagent) (`events.dispatch`) | [`subagent`](../packages/subagent/subagent), [`tool-subagent`](../packages/subagent/tool-subagent) | +| `subagent/start` | `emit` | [`packages/subagent/subagent/src/index.ts:151`](../packages/subagent/subagent/src/index.ts) | [`subagent`](../packages/subagent/subagent) (`events.dispatch`) | [`hooks-claude`](../packages/hooks/hooks-claude), [`subagent`](../packages/subagent/subagent) | | `system-prompt/assemble` | `waterfall` | [`packages/core/system-prompt/src/index.ts:29`](../packages/core/system-prompt/src/index.ts) | [`system-prompt`](../packages/core/system-prompt) (`waterfall`) | [`agent`](../packages/core/agent), [`system-prompt`](../packages/core/system-prompt) | | `system-prompt/change` | `emit` | [`packages/core/system-prompt/src/index.ts:35`](../packages/core/system-prompt/src/index.ts) | [`system-prompt`](../packages/core/system-prompt) (`emit`) | - | | `telemetry/record` | `waterfall` | [`packages/telemetry/session-telemetry/src/index.ts:41`](../packages/telemetry/session-telemetry/src/index.ts) | [`session-telemetry`](../packages/telemetry/session-telemetry) (`waterfall`) | - | diff --git a/packages/client/runtime/README.i18n.yaml b/packages/client/runtime/README.i18n.yaml index 6f31ca3beb..1ab6c96659 100644 --- a/packages/client/runtime/README.i18n.yaml +++ b/packages/client/runtime/README.i18n.yaml @@ -2,5 +2,5 @@ # side as of the last confirmed-consistent state. Both languages carry equal authority; # after editing either side, bring the other along and re-record with: # pnpm run verify-translation-pairing --write packages/client/runtime/README.md -README.md: eca7db1f9b2d5c7e28fa86a363ca4408703b99df -README.zh.md: 6a2e8c6085d06a9f04c1270e5976452b995a7e77 +README.md: 82f1bc95a6128245f88f01a0de0849494cb98359 +README.zh.md: f3aba75d18671fe377ec8305693ea5025d51e0de diff --git a/packages/client/runtime/README.md b/packages/client/runtime/README.md index eca7db1f9b..82f1bc95a6 100644 --- a/packages/client/runtime/README.md +++ b/packages/client/runtime/README.md @@ -2,7 +2,7 @@ English | [中文](README.zh.md) -Client cordis boot and React-free object services: SlotsService wraps SlotCore and supplies renderer data sources; SessionsService owns Session objects and the Chat-facing list, scope, and event-window state; SessionHistoryService lazily owns independent raw-history ledgers for inspection consumers; WorkspacesService depends on SessionsService and owns Workspace objects, list/actions, default-target derivation, and the New Session blank-reuse entry (`connectWorkspace`). The runtime fans the shared Host stream into the Session, Workspace, and activated history owners without routing inspection state through Session or SessionManager, and bridges the registry-invalidation frames to typed ctx events (`commands/changed`, `settings/changed`, `credentials/changed`, `models/changed`) so surface caches refetch without touching the stream. Client sessions are always Host-born (Session+Agent+cwd in one `session.create`); the client holds no pre-entity session state — a session's Agent scope (the client mirror of host dsh-scope, keyed by the shared agent/session id) is born when its row enters the list mirror and dies with the prune. Contract: api-contracts v3 §4. Each `Session` holds a generic `ProjectionValueStore` seeded from the history-tail `projections` block and updated by `session/projection` frames under higher-seq-wins; domain keys (including `todos`) are read via `projections.faceOf` / `useProjection`, not via `ConversationSnapshot`. +Client cordis boot and React-free object services: SlotsService wraps SlotCore and supplies renderer data sources; SessionsService owns Session objects and the Chat-facing list, scope, and event-window state; SessionHistoryService lazily owns independent raw-history ledgers for inspection consumers; WorkspacesService depends on SessionsService and owns Workspace objects, list/actions, default-target derivation, and the New Session blank-reuse entry (`connectWorkspace`). The runtime fans the shared Host stream into the Session, Workspace, and activated history owners without routing inspection state through Session or SessionManager, and bridges the registry-invalidation frames to typed ctx events (`commands/changed`, `settings/changed`, `credentials/changed`, `models/changed`) so surface caches refetch without touching the stream. Client sessions are always Host-born (Session+Agent+cwd in one `session.create`); the client holds no pre-entity session state — a session's Agent scope (the client mirror of host dsh-scope, keyed by the shared agent/session id) is born when its row enters the list mirror and dies with the prune. Contract: api-contracts v3 §4. Each `Session` holds a generic `ProjectionValueStore` seeded from the history-tail `projections` block and updated by `session/projection` frames under higher-seq-wins; domain keys (including `todos`) are read via `projections.faceOf` / `useProjection`, not via `ConversationSnapshot`. The store also publishes one reference-stable whole-value map through `SessionSummary.projectionValues`, allowing global list consumers to reuse the same projections without creating per-session subscriptions. ## Workspace and Session lists diff --git a/packages/client/runtime/README.zh.md b/packages/client/runtime/README.zh.md index 6a2e8c6085..f3aba75d18 100644 --- a/packages/client/runtime/README.zh.md +++ b/packages/client/runtime/README.zh.md @@ -2,7 +2,7 @@ [English](README.md) | 中文 -客户端 cordis 启动与不依赖 React 的对象服务：SlotsService 包装 SlotCore 并提供 renderer 数据源；SessionsService 拥有 Session 对象以及 Chat 所需的列表、scope 和事件窗口状态；SessionHistoryService 为检查类消费方惰性拥有彼此独立的原始历史账本；WorkspacesService 依赖 SessionsService，拥有 Workspace 对象、列表／操作、默认目标派生，以及 New Session 空会话复用入口（`connectWorkspace`）。运行时把共享 Host 流分发给 Session、Workspace 和已激活的历史数据所有者，不让检查状态经过 Session 或 SessionManager，并把注册表失效帧桥接为类型化 ctx 事件（`commands/changed`、`settings/changed`、`credentials/changed`、`models/changed`），使各表面缓存无需触碰流即可重拉。客户端会话一律由 Host 创建（一次 `session.create` 同时产生 Session、agent（智能体）和 cwd）；客户端不持有任何实体化之前的会话状态——agent scope（host dsh-scope 的客户端镜像，以 agent/session 共用 id 为键）在会话行进入列表镜像时创建，并随 prune 销毁。契约：api-contracts v3 §4。每个 `Session` 持有一个通用的 `ProjectionValueStore`，由历史记录尾部的 `projections` 块播种，并经 `session/projection` 帧按 seq 高者胜更新；领域键（含 `todos`）经 `projections.faceOf`／`useProjection` 读取，不经 `ConversationSnapshot`。 +客户端 cordis 启动与不依赖 React 的对象服务：SlotsService 包装 SlotCore 并提供 renderer 数据源；SessionsService 拥有 Session 对象以及 Chat 所需的列表、scope 和事件窗口状态；SessionHistoryService 为检查类消费方惰性拥有彼此独立的原始历史账本；WorkspacesService 依赖 SessionsService，拥有 Workspace 对象、列表／操作、默认目标派生，以及 New Session 空会话复用入口（`connectWorkspace`）。运行时把共享 Host 流分发给 Session、Workspace 和已激活的历史数据所有者，不让检查状态经过 Session 或 SessionManager，并把注册表失效帧桥接为类型化 ctx 事件（`commands/changed`、`settings/changed`、`credentials/changed`、`models/changed`），使各表面缓存无需触碰流即可重拉。客户端会话一律由 Host 创建（一次 `session.create` 同时产生 Session、agent（智能体）和 cwd）；客户端不持有任何实体化之前的会话状态——agent scope（host dsh-scope 的客户端镜像，以 agent/session 共用 id 为键）在会话行进入列表镜像时创建，并随 prune 销毁。契约：api-contracts v3 §4。每个 `Session` 持有一个通用的 `ProjectionValueStore`，由历史记录尾部的 `projections` 块播种，并经 `session/projection` 帧按 seq 高者胜更新；领域键（含 `todos`）经 `projections.faceOf`／`useProjection` 读取，不经 `ConversationSnapshot`。该 store 还会通过 `SessionSummary.projectionValues` 发布一份引用稳定的完整值映射，使全局列表消费方无需为每个会话创建订阅，即可复用同一组投影。 ## Workspace 与 Session 列表 diff --git a/packages/client/runtime/src/client/sessions/lineage.ts b/packages/client/runtime/src/client/sessions/lineage.ts index 08f8361a92..4f26674420 100644 --- a/packages/client/runtime/src/client/sessions/lineage.ts +++ b/packages/client/runtime/src/client/sessions/lineage.ts @@ -3,10 +3,13 @@ // Orphaned lineage degrades to root level; cycles fail soft and emit as roots. import type { SessionId, SessionSummary } from '@deepseek-ai/dsh-client-connection/client' +import type { SessionProjectionMap } from '@deepseek-ai/dsh-session-projection/types' /** Host list summary enriched with the latest mux-projected durable title. */ export interface TitledSessionSummary extends SessionSummary { title?: string + /** Current host-computed projection values for list consumers. */ + projectionValues?: Readonly> } /** One flattened session-list row (summary + lineage indent depth + live pending-approval bit). */ @@ -21,6 +24,8 @@ export interface SessionListEntry { /** Coarse durable origin for navigation filtering; not a continuation capability. */ origin?: 'subagent' cwd?: string + /** Current host-computed projection values for list consumers. */ + projectionValues?: Readonly> /** An approval question is pending on this session (mux-frame derived; the sidebar's amber dot). */ waitingApproval: boolean /** Lineage indent depth: root = 0; the UI just multiplies by the indent width. */ diff --git a/packages/client/runtime/src/client/sessions/manager.ts b/packages/client/runtime/src/client/sessions/manager.ts index 1019327df5..d87ed0a04f 100644 --- a/packages/client/runtime/src/client/sessions/manager.ts +++ b/packages/client/runtime/src/client/sessions/manager.ts @@ -798,10 +798,14 @@ export class SessionManager { const merged: TitledSessionSummary[] = this.summaries.map((summary) => { // List rows read the generic 'title' projection key (host-computed unit // value; the bespoke session/title frame is retired). - const title = this.projectionStores.get(summary.sessionId)?.get('title') - return typeof title === 'string' && title !== '' - ? { ...summary, title } - : summary + const projectionStore = this.projectionStores.get(summary.sessionId) + const title = projectionStore?.get('title') + const projectionValues = projectionStore?.values() + return { + ...summary, + ...(typeof title === 'string' && title !== '' ? { title } : {}), + ...(projectionValues === undefined ? {} : { projectionValues }), + } }) const fresh = flattenLineage(merged, new Set(this.waitingApprovals.keys())) const items = fresh.map((entry) => { @@ -812,6 +816,7 @@ export class SessionManager { && prev.parentSessionId === entry.parentSessionId && prev.cwd === entry.cwd && prev.origin === entry.origin && prev.title === entry.title && prev.depth === entry.depth && prev.waitingApproval === entry.waitingApproval + && prev.projectionValues === entry.projectionValues ) return prev this.entryCache.set(entry.sessionId, entry) return entry diff --git a/packages/client/runtime/src/client/sessions/projection-store.ts b/packages/client/runtime/src/client/sessions/projection-store.ts index 4e6e7dd626..32401146af 100644 --- a/packages/client/runtime/src/client/sessions/projection-store.ts +++ b/packages/client/runtime/src/client/sessions/projection-store.ts @@ -74,6 +74,7 @@ interface Channel { export class ProjectionValueStore { private readonly rows = new Map() private readonly channels = new Map() + private valuesCache: Readonly> | undefined /** Coarse any-key channel (no snapshot cache to rebuild: reads hit rows directly). */ private readonly anyNotifier = new Notifier(() => {}) @@ -98,6 +99,19 @@ export class ProjectionValueStore { return this.rows.get(key)?.value } + /** + * Read every current projection value as one reference-stable snapshot. + * @returns The same frozen value map until a row changes. + */ + values(): Readonly> { + if (this.valuesCache === undefined) { + this.valuesCache = Object.freeze(Object.fromEntries( + [...this.rows].map(([key, row]) => [key, row.value]), + )) + } + return this.valuesCache + } + /** * Subscribe to any-key changes (microtask-batched) — the manager's list * rebuild channel. @@ -160,6 +174,7 @@ export class ProjectionValueStore { } private changed(key: string): void { + this.valuesCache = undefined this.channels.get(key)?.notifier.markDirty() this.anyNotifier.markDirty() } diff --git a/packages/client/runtime/src/client/sessions/service.ts b/packages/client/runtime/src/client/sessions/service.ts index 087a07b02d..47ee641053 100644 --- a/packages/client/runtime/src/client/sessions/service.ts +++ b/packages/client/runtime/src/client/sessions/service.ts @@ -25,6 +25,7 @@ import { SESSION_SEARCH_RESULT_LIMIT } from '@deepseek-ai/dsh-host-apiproxy/api' import type { HostObservable, SessionMaybeProvideInfo, SessionProvideInfo, } from '@deepseek-ai/dsh-client-ui-slots' +import type { SessionProjectionMap } from '@deepseek-ai/dsh-session-projection/types' import type { SnapshotStore } from '../contract/store.ts' import { createSnapshotStore } from '../contract/store.ts' import type { SessionFace } from '../contract/session.ts' @@ -57,6 +58,8 @@ export interface SessionSummary { */ blank: boolean updatedAt: number + /** Current host-computed projection values retained by the object layer. */ + projectionValues?: Readonly> } /** @@ -613,6 +616,9 @@ export class SessionsService implements ISessions { waitingApproval: entry.waitingApproval, blank: entry.blank, updatedAt: entry.updatedAt, + ...(entry.projectionValues === undefined + ? {} + : { projectionValues: entry.projectionValues }), ...(entry.title !== undefined ? { title: entry.title } : {}), ...(entry.cwd !== undefined ? { cwd: entry.cwd } : {}), ...(entry.parentSessionId !== undefined ? { parentId: entry.parentSessionId } : {}), diff --git a/packages/client/runtime/tests/projection-store.spec.ts b/packages/client/runtime/tests/projection-store.spec.ts index eea43b67f3..143da92348 100644 --- a/packages/client/runtime/tests/projection-store.spec.ts +++ b/packages/client/runtime/tests/projection-store.spec.ts @@ -86,6 +86,17 @@ describe('ProjectionValueStore semantics', () => { const store = new ProjectionValueStore() expect(store.faceOf('test/marks')).toBe(store.faceOf('test/marks')) }) + + it('publishes one reference-stable whole-value snapshot until a row changes', () => { + const store = new ProjectionValueStore() + const empty = store.values() + expect(store.values()).toBe(empty) + store.apply('test/marks', { marks: ['a'] }, 1) + const populated = store.values() + expect(populated).toEqual({ 'test/marks': { marks: ['a'] } }) + expect(populated).not.toBe(empty) + expect(store.values()).toBe(populated) + }) }) describe('Session tail-page seeding', () => { @@ -167,6 +178,36 @@ describe('manager frame routing', () => { expect(manager.getListSnapshot().items[0]?.title).toBeUndefined() }) + it('projects every retained value into list rows with stable snapshot identity', async () => { + const api = new FakeApiClient() + const manager = new SessionManager(api) + api.onList = () => Promise.resolve(ok({ + items: [{ + sessionId: sid('s1'), updatedAt: 1, running: false, blank: false, + projections: { + asOfSeq: 2, + values: { 'test/marks': { marks: ['baseline'] } }, + }, + }], + }) as never) + await manager.refreshList() + const baseline = manager.getListSnapshot().items[0]?.projectionValues + expect(baseline).toEqual({ 'test/marks': { marks: ['baseline'] } }) + expect(manager.getListSnapshot().items[0]?.projectionValues).toBe(baseline) + + manager.handleMuxEnvelope({ + rpcId: 'p2' as never, + payload: { + type: 'session/projection', sessionId: sid('s1'), key: 'test/marks', + value: { marks: ['live'] }, seq: 3, + } as never, + }) + await Promise.resolve() + expect(manager.getListSnapshot().items[0]?.projectionValues) + .toEqual({ 'test/marks': { marks: ['live'] } }) + expect(manager.getListSnapshot().items[0]?.projectionValues).not.toBe(baseline) + }) + it('drops the projection store with the removed session', async () => { const api = new FakeApiClient() const manager = new SessionManager(api) diff --git a/packages/client/ui-subagent/README.i18n.yaml b/packages/client/ui-subagent/README.i18n.yaml index 65f99af124..8af7265330 100644 --- a/packages/client/ui-subagent/README.i18n.yaml +++ b/packages/client/ui-subagent/README.i18n.yaml @@ -2,5 +2,5 @@ # side as of the last confirmed-consistent state. Both languages carry equal authority; # after editing either side, bring the other along and re-record with: # pnpm run verify-translation-pairing --write packages/client/ui-subagent/README.md -README.md: f6b3fa2e9cdf1479a739e0b4eab15a5423e878e4 -README.zh.md: fdfba385e9188cd42bd973b6f32bc01fe8d004f2 +README.md: 2d507dde49f3796f3bbd00c239a768cdfd8a561d +README.zh.md: d0143858ff7eac1437fce381378ebcfad2187f2a diff --git a/packages/client/ui-subagent/README.md b/packages/client/ui-subagent/README.md index f6b3fa2e9c..2d507dde49 100644 --- a/packages/client/ui-subagent/README.md +++ b/packages/client/ui-subagent/README.md @@ -4,7 +4,7 @@ English | [中文](README.zh.md) Web subagent feature owner: contributes the lazily expandable catalog tree to `conversation.session.header.actions`, reason-specific read-only replacements to the conversation composer chain, and the existing `@` reference source to `ctx.slash`. -The header action reads `subagentsByParent` and session summaries through the standard `useSessions` hook. After a non-empty direct catalog arrives, its trigger counts the complete subagent-only descendant lineage, stops at ordinary forks, and shows ongoing activity when any counted descendant is running. The compact tree remains direct-catalog authoritative: continuable and one-shot rows display mode plus `running`/`inactive` activity, an optional log-backed title, and session-summary activity time; an unlabeled one-shot row falls back to its session id, while corrupt, unsupported, or unavailable rows remain readable but disabled. Each healthy row's `hasChildren` hint determines disclosure before interaction, so known leaves never show an arrow; expanding a branch immediately reserves one disabled loading row per known direct descendant, then lazily replaces them with that child's authoritative catalog. Every visible branch is reported to the runtime so membership frames cause a debounced refresh only where the tree is being consumed. Selecting any depth calls `SessionsService.openSubagent()` with the row's exact `{parentSessionId, childSessionId, mode}` address. Component-local state owns tree visibility, expanded branches, and keyboard focus. ArrowRight/ArrowLeft expand and collapse branches; ArrowUp/ArrowDown, Home, End, and Escape navigate or close the tree; closing returns focus to the trigger. Styling uses tokens only. +The header action reads `subagentsByParent` and session summaries through the standard `useSessions` hook. After a non-empty direct catalog arrives, its trigger counts the complete subagent-only descendant lineage, stops at ordinary forks, and shows ongoing activity when any counted descendant is running. The compact tree remains direct-catalog authoritative: continuable and one-shot rows display mode plus `running`/`inactive` activity, an optional log-backed title, total durable provider usage, and active-turn duration to the second; an unlabeled one-shot row falls back to its session id, while corrupt, unsupported, or unavailable rows remain readable but disabled. Token totals sum the four disjoint `tokenUsage` buckets. Duration sums completed `subagentTiming` turns, advances once per second only for an open turn on a running child, and freezes after the child becomes inactive; an interrupted open turn is bounded by the session summary's last activity. Each healthy row's `hasChildren` hint determines disclosure before interaction, so known leaves never show an arrow; expanding a branch immediately reserves one disabled loading row per known direct descendant, then lazily replaces them with that child's authoritative catalog. Every visible branch is reported to the runtime so membership frames cause a debounced refresh only where the tree is being consumed. Selecting any depth calls `SessionsService.openSubagent()` with the row's exact `{parentSessionId, childSessionId, mode}` address. Component-local state owns tree visibility, expanded branches, keyboard focus, and the running-duration clock. ArrowRight/ArrowLeft expand and collapse branches; ArrowUp/ArrowDown, Home, End, and Escape navigate or close the tree; closing returns focus to the trigger. Styling uses tokens only. A one-shot child always elects a read-only composer that identifies the transcript as a completed execution record. A continuable child does so only when its exact parent is unavailable, with copy explaining the recovery path. A continuable child with a live parent keeps the ordinary input chrome, whose Session routes through `subagent.prompt`; running input remains Send because every follow-up joins the child's FIFO inbox, and addressed sessions never expose Stop. This package never receives host context or calls a model-facing tool. The catalog and composer behavior are specified by the [Web subagent conversations Agent Note](../../../.agents/notes/implemented/feature/2026-07-27-web-subagent-conversations.md). @@ -30,5 +30,5 @@ Append-only. This package never edits earlier request tokens. ## Known Limitations and Deferred Work -- **The catalog has coarse activity only** — it cannot show durable outcome, elapsed time, Activation identity, or an authority-safe cancel button. +- **The catalog has no durable outcome** — activity and timing do not distinguish completion, failure, or cancellation, and the UI exposes neither Activation identity nor an authority-safe cancel button. - **`@` references remain display-title text** — duplicate or renamed labels are ambiguous, so they intentionally do not acquire continuation semantics. diff --git a/packages/client/ui-subagent/README.zh.md b/packages/client/ui-subagent/README.zh.md index fdfba385e9..d0143858ff 100644 --- a/packages/client/ui-subagent/README.zh.md +++ b/packages/client/ui-subagent/README.zh.md @@ -4,7 +4,7 @@ Web subagent 功能 owner：向 `conversation.session.header.actions` 贡献可懒加载展开的目录树，向会话编辑器链贡献按原因区分的只读替代呈现，并保留注册到 `ctx.slash` 的既有 `@` 引用 source。 -页头操作通过标准 `useSessions` 钩子读取 `subagentsByParent` 与会话摘要。非空直接目录到达后，其触发器会统计仅含 subagent 的完整后代谱系，在普通 fork 处停止，并在任一计入统计的后代处于 `running` 时显示活动仍在进行。紧凑树仍以直接目录为权威依据：可继续和 one-shot 行会显示 mode、`running`／`inactive` 活动状态、由日志支撑的可选 title 与会话摘要中的活动时间；没有 label 的 one-shot 行会回退到其会话 id，而损坏、不受支持或不可用的行仍保持可读但禁用。每个健康行的 `hasChildren` 提示会在交互前决定是否显示展开控件，因此已知叶子节点从不显示箭头；展开分支时，会立即为每个已知直接后代预留一行禁用的加载行，随后再用该 child 的权威目录懒加载结果替换这些占位行。每个可见分支都会上报给运行时，使成员帧只在树正被消费的位置触发去抖动刷新。选择任意深度的条目都会使用该行的确切地址 `{parentSessionId, childSessionId, mode}` 调用 `SessionsService.openSubagent()`。组件局部状态负责树的可见性、已展开分支与键盘焦点。ArrowRight／ArrowLeft 展开和折叠分支；ArrowUp／ArrowDown、Home、End 与 Escape 用于导航或关闭树；关闭后焦点返回触发器。样式只使用 token。 +页头操作通过标准 `useSessions` 钩子读取 `subagentsByParent` 与会话摘要。非空直接目录到达后，其触发器会统计仅含 subagent 的完整后代谱系，在普通 fork 处停止，并在任一计入统计的后代处于 `running` 时显示活动仍在进行。紧凑树仍以直接目录为权威依据：可继续和 one-shot 行会显示 mode、`running`／`inactive` 活动状态、由日志支撑的可选 title、提供方的持久化总用量，以及精确到秒的活跃轮次耗时；没有 label 的 one-shot 行会回退到其会话 id，而损坏、不受支持或不可用的行仍保持可读但禁用。token 总量为四个互不重叠的 `tokenUsage` 桶之和。耗时会累加已完成的 `subagentTiming` 轮次，仅在运行中 child 存在未结束轮次时每秒递增一次，并在 child 变为 inactive 后冻结；被中断的未结束轮次以会话摘要中的最后活动为上界。每个健康行的 `hasChildren` 提示会在交互前决定是否显示展开控件，因此已知叶子节点从不显示箭头；展开分支时，会立即为每个已知直接后代预留一行禁用的加载行，随后再用该 child 的权威目录懒加载结果替换这些占位行。每个可见分支都会上报给运行时，使成员帧只在树正被消费的位置触发去抖动刷新。选择任意深度的条目都会使用该行的确切地址 `{parentSessionId, childSessionId, mode}` 调用 `SessionsService.openSubagent()`。组件局部状态负责树的可见性、已展开分支、键盘焦点与运行中耗时时钟。ArrowRight／ArrowLeft 展开和折叠分支；ArrowUp／ArrowDown、Home、End 与 Escape 用于导航或关闭树；关闭后焦点返回触发器。样式只使用 token。 one-shot child 始终选用只读编辑器，并将 transcript（文本记录）说明为已完成的执行记录。可继续 child 仅在其确切 parent 不可用时选用只读编辑器，并以文案说明恢复路径。确切 parent 存活时，可继续 child 保留普通输入 chrome，其 Session 会通过 `subagent.prompt` 路由；child 运行期间，输入操作仍为 Send，因为每条后续消息都会进入 child 的 FIFO inbox，且已寻址会话绝不公开 Stop。本包绝不接收宿主 context，也不调用面向模型的工具。目录与编辑器行为由 [Web subagent 对话 Agent Note](../../../.agents/notes/implemented/feature/2026-07-27-web-subagent-conversations.md)规定。 @@ -30,5 +30,5 @@ one-shot child 始终选用只读编辑器，并将 transcript（文本记录） ## 已知限制与暂缓事项 -- **目录只有粗粒度活动状态**：它不能显示持久化结果、耗时、Activation 身份或具备安全授权的取消按钮。 +- **目录没有持久化结果**：活动状态与计时无法区分完成、失败或取消，且 UI 既不公开 Activation 身份，也不公开具备安全授权的取消按钮。 - **`@` 引用仍是显示标题文本**：重复或改名后的 label 会有歧义，因此它们刻意不获得继续执行语义。 diff --git a/packages/client/ui-subagent/package.json b/packages/client/ui-subagent/package.json index efec3bc047..296b3d5ebf 100644 --- a/packages/client/ui-subagent/package.json +++ b/packages/client/ui-subagent/package.json @@ -46,6 +46,8 @@ "@deepseek-ai/dsh-client-ui-slash": "^0.0.1", "@deepseek-ai/dsh-client-ui-slots": "^0.0.1", "@deepseek-ai/dsh-invariants": "^0.0.1", + "@deepseek-ai/dsh-subagent": "^0.0.1", + "@deepseek-ai/dsh-token-meter": "^0.0.1", "cordis": "^4.0.0-rc.7" }, "devDependencies": { @@ -55,6 +57,8 @@ "@deepseek-ai/dsh-client-ui-slash": "workspace:^", "@deepseek-ai/dsh-client-ui-slots": "workspace:^", "@deepseek-ai/dsh-invariants": "workspace:^", + "@deepseek-ai/dsh-subagent": "workspace:^", + "@deepseek-ai/dsh-token-meter": "workspace:^", "@types/react": "~18.3.1", "cordis": "^4.0.0-rc.7" }, diff --git a/packages/client/ui-subagent/src/client/SubagentCatalogAction.module.css b/packages/client/ui-subagent/src/client/SubagentCatalogAction.module.css index e2133e6ce9..13004e6a47 100644 --- a/packages/client/ui-subagent/src/client/SubagentCatalogAction.module.css +++ b/packages/client/ui-subagent/src/client/SubagentCatalogAction.module.css @@ -173,15 +173,17 @@ } .summary, -.time { +.metrics { color: var(--dsw-alias-label-tertiary); font-size: 11px; line-height: 16px; } -.time { +.metrics { flex: none; margin-top: 16px; + font-variant-numeric: tabular-nums; + white-space: nowrap; } .children { diff --git a/packages/client/ui-subagent/src/client/SubagentCatalogAction.tsx b/packages/client/ui-subagent/src/client/SubagentCatalogAction.tsx index c23c82d77b..6bceec3105 100644 --- a/packages/client/ui-subagent/src/client/SubagentCatalogAction.tsx +++ b/packages/client/ui-subagent/src/client/SubagentCatalogAction.tsx @@ -2,13 +2,16 @@ import { useEffect, useRef, useState, type KeyboardEvent, type MouseEvent, } from 'react' import type { - SessionId, SessionListState, SessionSummary, SubagentAddress, SubagentCatalogSnapshot, + SessionId, SessionListState, SessionProjectionMap, SessionSummary, SubagentAddress, + SubagentCatalogSnapshot, } from '@deepseek-ai/dsh-client-runtime/client' import { IconChevronDownOutline14, IconChevronRightOutline14, IconRefreshOutline14, StateDot, } from '@deepseek-ai/dsh-client-ui-primitives' import type { PropsRuntime } from '@deepseek-ai/dsh-client-ui-slots' import type {} from '@deepseek-ai/dsh-client-ui-conversation/client' +import type {} from '@deepseek-ai/dsh-subagent/client' +import type {} from '@deepseek-ai/dsh-token-meter/client' import css from './SubagentCatalogAction.module.css' type CatalogEntry = SubagentCatalogSnapshot['entries'][number] @@ -53,19 +56,53 @@ function treeItems(root: HTMLDivElement | null): HTMLElement[] { : Array.from(root.querySelectorAll('[role="treeitem"]:not([aria-disabled="true"])')) } -/** Compact trailing activity time for a catalog row. */ -function relativeTime(updatedAt: number | undefined, now: number): string | undefined { - if (updatedAt === undefined) return undefined - const minute = 60_000 - const hour = 60 * minute - const day = 24 * hour - const diff = Math.max(0, now - updatedAt) - if (diff < minute) return '刚刚' - if (diff < hour) return `${Math.floor(diff / minute)}分钟` - if (diff < day) return `${Math.floor(diff / hour)}小时` - if (diff < 30 * day) return `${Math.floor(diff / day)}天` - if (diff < 365 * day) return `${Math.floor(diff / (30 * day))}个月` - return `${Math.floor(diff / (365 * day))}年` +/** Compact token count shared in shape with the conversation stats strip. */ +function formatTokens(value: number): string { + const scaled = (next: number): string => next >= 100 + ? String(Math.round(next)) + : String(Math.round(next * 10) / 10) + if (value < 1_000) return String(value) + if (value < 1_000_000) return `${scaled(value / 1_000)}K` + return `${scaled(value / 1_000_000)}M` +} + +/** Sum the four disjoint durable provider-usage buckets. */ +function tokenTotal( + usage: SessionProjectionMap['tokenUsage'] | undefined, +): number | undefined { + return usage === undefined + ? undefined + : usage.uncachedInputTokens + usage.outputTokens + + usage.cacheReadTokens + usage.cacheWriteTokens +} + +/** Exact whole-second active-turn duration for one catalog row. */ +function activityDuration( + timing: SessionProjectionMap['subagentTiming'] | undefined, + activity: 'running' | 'inactive', + updatedAt: number | undefined, + now: number, +): number | undefined { + if (timing === undefined) return undefined + if (timing.activeSince === undefined) return timing.settledMs + const end = activity === 'running' ? now : updatedAt ?? timing.activeSince + return timing.settledMs + Math.max(0, end - timing.activeSince) +} + +/** Format a non-negative duration to seconds without dropping larger units. */ +function formatDuration(ms: number): string { + const totalSeconds = Math.floor(Math.max(0, ms) / 1_000) + const seconds = totalSeconds % 60 + const totalMinutes = Math.floor(totalSeconds / 60) + const minutes = totalMinutes % 60 + const hours = Math.floor(totalMinutes / 60) + if (hours > 0) { + return `${hours}小时${String(minutes).padStart(2, '0')}分${String(seconds).padStart(2, '0')}秒` + } + if (totalMinutes > 0) { + return `${totalMinutes}分${String(seconds).padStart(2, '0')}秒` + } + return `${seconds}秒` } /** Aggregate the complete subagent-only descendant subtree from flat summaries. */ @@ -190,7 +227,17 @@ function CatalogRows({ const secondary = [summary?.title, mode, activity] .filter(value => value !== undefined) .join(' · ') - const time = relativeTime(summary?.updatedAt, now) + const totalTokens = tokenTotal(summary?.projectionValues?.tokenUsage) + const durationMs = activityDuration( + summary?.projectionValues?.subagentTiming, + entry.activity, + summary?.updatedAt, + now, + ) + const metrics = [ + totalTokens === undefined ? undefined : `${formatTokens(totalTokens)} tok`, + durationMs === undefined ? undefined : formatDuration(durationMs), + ].filter(value => value !== undefined).join(' · ') const open = (): void => { openChild({ parentSessionId, childSessionId: entry.id, mode: entry.mode }) @@ -222,7 +269,7 @@ function CatalogRows({ role="treeitem" tabIndex={0} aria-level={level} - aria-label={[label, secondary, time].filter(value => value !== undefined).join(' ')} + aria-label={[label, secondary, metrics].filter(value => value !== '').join(' ')} {...knownLeaf ? {} : { 'aria-expanded': isExpanded }} className={css.row} onClick={open} @@ -247,7 +294,7 @@ function CatalogRows({ {label} {secondary} - {time !== undefined && {time}} + {metrics !== '' && {metrics}} {isExpanded && !knownLeaf && ( @@ -300,6 +347,7 @@ export function SubagentCatalogAction({ const summaries = useSessions(state => state.byId) const catalog = catalogs[sessionId] const [open, setOpen] = useState(false) + const [now, setNow] = useState(() => Date.now()) const [expanded, setExpanded] = useState>(() => new Set()) const rootRef = useRef(null) const triggerRef = useRef(null) @@ -328,7 +376,10 @@ export function SubagentCatalogAction({ const changeOpen = (next: boolean, restoreFocus = false): void => { setOpen(next) - if (next) observeCatalog(sessionId, true) + if (next) { + setNow(Date.now()) + observeCatalog(sessionId, true) + } else closeAllCatalogs() if (restoreFocus) queueMicrotask(() => { triggerRef.current?.focus() }) } @@ -368,6 +419,12 @@ export function SubagentCatalogAction({ return () => { document.removeEventListener('pointerdown', closeOutside) } }, [open]) + useEffect(() => { + if (!open || !descendants.running) return + const timer = setInterval(() => { setNow(Date.now()) }, 1_000) + return () => { clearInterval(timer) } + }, [open, descendants.running]) + useEffect(() => () => { for (const parentSessionId of observedCatalogs.current) { setCatalogOpenRef.current(parentSessionId, false) @@ -443,7 +500,7 @@ export function SubagentCatalogAction({ summaries={summaries} expanded={expanded} level={1} - now={Date.now()} + now={now} openChild={openChild} refresh={refresh} toggleBranch={toggleBranch} diff --git a/packages/client/ui-subagent/tests/conversation-ui.spec.tsx b/packages/client/ui-subagent/tests/conversation-ui.spec.tsx index 6eb9f881c0..d7098c1547 100644 --- a/packages/client/ui-subagent/tests/conversation-ui.spec.tsx +++ b/packages/client/ui-subagent/tests/conversation-ui.spec.tsx @@ -11,6 +11,7 @@ import { SubagentReadOnlyComposer } from '../src/client/SubagentReadOnlyComposer afterEach(() => { cleanup() + vi.useRealTimers() vi.restoreAllMocks() }) @@ -217,42 +218,56 @@ describe('SubagentCatalogAction', () => { }) }) - it('renders compact activity times across every unit and clamps future timestamps', () => { + it('shows durable token totals, ticks active duration by seconds, and freezes inactive rows', async () => { const now = 2_000_000_000_000 - vi.spyOn(Date, 'now').mockReturnValue(now) - const minute = 60_000 - const hour = 60 * minute - const day = 24 * hour + vi.useFakeTimers() + vi.setSystemTime(now) const rows = [ - ['future', now + minute], - ['minutes', now - 2 * minute], - ['hours', now - 2 * hour], - ['days', now - 2 * day], - ['months', now - 60 * day], - ['years', now - 2 * 365 * day], + ['running', 'running', 65_000, now - 5_000, now], + ['finished', 'inactive', 3_723_000, undefined, now - 60_000], + ['interrupted', 'inactive', 2_000, now - 7_000, now - 3_000], ] as const - const entries = rows.map(([id]) => ({ + const entries = rows.map(([id, activity]) => ({ kind: 'child' as const, id: id as SessionId, mode: 'continuable' as const, label: id, - activity: 'inactive' as const, + activity, hasChildren: false, })) - const summaries = Object.fromEntries(rows.map(([id, updatedAt]) => [ - id, - summary(id as SessionId, updatedAt), - ])) as Record + const summaries = Object.fromEntries(rows.map(([id, activity, settledMs, activeSince, updatedAt]) => { + const childId = id as SessionId + return [id, { + ...summary(childId, updatedAt), + parentId: PARENT, + origin: 'subagent' as const, + running: activity === 'running', + projectionValues: { + subagentTiming: { + settledMs, + ...(activeSince === undefined ? {} : { activeSince }), + }, + tokenUsage: { + uncachedInputTokens: 1_000, + outputTokens: 200, + cacheReadTokens: 3_000, + cacheWriteTokens: 400, + }, + }, + }] + })) as Record const input = props(catalog({ entries }), {}, summaries) render() - fireEvent.click(screen.getByRole('button', { name: /6 个子代理/ })) + fireEvent.click(screen.getByRole('button', { name: /3 个子代理/ })) - expect(screen.getByRole('treeitem', { name: /future.*刚刚/ })).toBeTruthy() - expect(screen.getByRole('treeitem', { name: /minutes.*2分钟/ })).toBeTruthy() - expect(screen.getByRole('treeitem', { name: /hours.*2小时/ })).toBeTruthy() - expect(screen.getByRole('treeitem', { name: /days.*2天/ })).toBeTruthy() - expect(screen.getByRole('treeitem', { name: /months.*2个月/ })).toBeTruthy() - expect(screen.getByRole('treeitem', { name: /years.*2年/ })).toBeTruthy() + expect(screen.getByRole('treeitem', { name: /running.*4\.6K tok · 1分10秒/ })).toBeTruthy() + expect(screen.getByRole('treeitem', { name: /finished.*4\.6K tok · 1小时02分03秒/ })).toBeTruthy() + expect(screen.getByRole('treeitem', { name: /interrupted.*4\.6K tok · 6秒/ })).toBeTruthy() + + await vi.advanceTimersByTimeAsync(1_000) + expect(screen.getByRole('treeitem', { name: /running.*4\.6K tok · 1分11秒/ })).toBeTruthy() + expect(screen.getByRole('treeitem', { name: /finished.*4\.6K tok · 1小时02分03秒/ })).toBeTruthy() + expect(screen.getByRole('treeitem', { name: /interrupted.*4\.6K tok · 6秒/ })).toBeTruthy() }) it('lazily expands and collapses descendant catalogs with direct-parent navigation', () => { diff --git a/packages/client/ui-subagent/tsconfig.json b/packages/client/ui-subagent/tsconfig.json index 395281ce6d..0cae499e1a 100644 --- a/packages/client/ui-subagent/tsconfig.json +++ b/packages/client/ui-subagent/tsconfig.json @@ -26,6 +26,12 @@ { "path": "../ui-slots" }, + { + "path": "../../llm/token-meter" + }, + { + "path": "../../subagent/subagent" + }, { "path": "../../support/invariants" } diff --git a/packages/subagent/subagent/README.i18n.yaml b/packages/subagent/subagent/README.i18n.yaml index f7043e2403..76740d4506 100644 --- a/packages/subagent/subagent/README.i18n.yaml +++ b/packages/subagent/subagent/README.i18n.yaml @@ -2,5 +2,5 @@ # side as of the last confirmed-consistent state. Both languages carry equal authority; # after editing either side, bring the other along and re-record with: # pnpm run verify-translation-pairing --write packages/subagent/subagent/README.md -README.md: 9aea27a0f150d90a41d9a7cb4cd422a75e6107fe -README.zh.md: 3f0b534deae53b8d5aff2765974050f26b931953 +README.md: ec4af55bcd9374b1abb55d7bb098eef568449684 +README.zh.md: 8323853ff0de2a15da6475fc1433a68f0074ee93 diff --git a/packages/subagent/subagent/README.md b/packages/subagent/subagent/README.md index 9aea27a0f1..ec4af55bcd 100644 --- a/packages/subagent/subagent/README.md +++ b/packages/subagent/subagent/README.md @@ -92,6 +92,8 @@ Provider additions and removals also emit `subagent/provider-added` and `subagen Continuable children do not create `SubagentRun` or Tasks. The continuation manager directly owns one process-local Activation and retained `AgentHandle` per resident child Session, uses the Agent inbox as the only FIFO, and cold-resumes from the durable descriptor. Exact live direct-parent identity authorizes parent-to-child delivery. Exact live child identity authorizes reports; the manager derives the recipient from durable `parentSession`, and `MessageSource` remains provenance rather than authority. +When `ctx.sessionProjections` is available, the service registers `subagentTiming`. The projection resets at each descriptor so a fork seed's ancestor work cannot enter the child's total, then accumulates `turn/start` → `turn/end` active time and retains `activeSince` for an open turn. Only descriptors and turn boundaries change the value, so token chunks do not create timing updates. + `registerContinuableSetup()` lets optional packages add child-scoped capabilities without teaching the continuation manager their names. Contributions install synchronously before Activation publication, roll back with failed setup, and are released with the child scope. New grants wait for the next Activation, while contribution removal revokes every resident installation immediately. ## Collection model diff --git a/packages/subagent/subagent/README.zh.md b/packages/subagent/subagent/README.zh.md index 3f0b534dea..8323853ff0 100644 --- a/packages/subagent/subagent/README.zh.md +++ b/packages/subagent/subagent/README.zh.md @@ -92,6 +92,8 @@ subagent seam 允许一个 agent（智能体）通过具名提供方把工作委 可继续子级不会创建 `SubagentRun` 或 Task。延续管理器为每个驻留子 Session 直接拥有一个仅存在于当前进程的 Activation 和一个留存的 `AgentHandle`，使用 Agent inbox 作为唯一 FIFO，并从持久化描述符冷恢复。父到子投递由准确的实时直接父级身份授权。上报则由准确的实时子级身份授权；管理器根据持久化的 `parentSession` 推导接收方，`MessageSource` 仍只表示来源，不表示权限。 +当 `ctx.sessionProjections` 可用时，服务会注册 `subagentTiming`。该投影会在每个描述符处重置，使 fork 种子中的祖先工作不会计入 child 总量，随后累加 `turn/start` → `turn/end` 活跃时间，并为未结束的轮次保留 `activeSince`。只有描述符和轮次边界会改变该值，因此 token 分片不会产生计时更新。 + `registerContinuableSetup()` 允许可选包添加子级作用域功能，而无需让延续管理器知道这些功能的名称。贡献会在 Activation 发布前同步安装，在设置失败时一并回滚，并随子级作用域释放。新授权须等到下一个 Activation，移除贡献则会立即撤销每个驻留安装项。 ## 收集模型 diff --git a/packages/subagent/subagent/package.json b/packages/subagent/subagent/package.json index 863f04c117..8cd78c5526 100644 --- a/packages/subagent/subagent/package.json +++ b/packages/subagent/subagent/package.json @@ -15,17 +15,25 @@ "types": "./lib/types/invariant.d.ts", "default": "./lib/invariant.js" }, + "./client": { + "types": "./lib/types/client.d.ts", + "default": "./lib/types/client.js" + }, "./src/*": "./src/*", "./package.json": "./package.json" }, "files": [ "lib/index.js", "lib/invariant.js", + "lib/types/**/*.js", "lib/types/**/*.d.ts", "lib/types/**/*.d.ts.map", "src" ], "license": "BSD-3-Clause", + "dependencies": { + "zod": "^4.4.3" + }, "peerDependencies": { "@deepseek-ai/dsh-agent": "^0.0.1", "@deepseek-ai/dsh-brand": "^0.0.1", @@ -35,6 +43,7 @@ "@deepseek-ai/dsh-session": "^0.0.1", "@deepseek-ai/dsh-session-persistence": "^0.0.1", "@deepseek-ai/dsh-session-query": "^0.0.1", + "@deepseek-ai/dsh-session-projection": "^0.0.1", "@deepseek-ai/dsh-tasks": "^0.0.1", "@deepseek-ai/dsh-tools": "^0.0.1", "cordis": "^4.0.0-rc.7" @@ -46,6 +55,9 @@ "@deepseek-ai/dsh-session-query": { "optional": true }, + "@deepseek-ai/dsh-session-projection": { + "optional": true + }, "@deepseek-ai/dsh-tasks": { "optional": true } @@ -59,6 +71,7 @@ "@deepseek-ai/dsh-session": "workspace:^", "@deepseek-ai/dsh-session-persistence": "workspace:^", "@deepseek-ai/dsh-session-query": "workspace:^", + "@deepseek-ai/dsh-session-projection": "workspace:^", "@deepseek-ai/dsh-tasks": "workspace:^", "@deepseek-ai/dsh-tools": "workspace:^", "cordis": "^4.0.0-rc.7" diff --git a/packages/subagent/subagent/src/client.ts b/packages/subagent/subagent/src/client.ts new file mode 100644 index 0000000000..928637dc7a --- /dev/null +++ b/packages/subagent/subagent/src/client.ts @@ -0,0 +1,7 @@ +/** + * Browser-safe subagent projection vocabulary. + * + * @module @deepseek-ai/dsh-subagent/client + */ + +export type { SubagentTimingProjection } from './projection-types.ts' diff --git a/packages/subagent/subagent/src/index.ts b/packages/subagent/subagent/src/index.ts index 19ad74ff3a..975d07ce4c 100644 --- a/packages/subagent/subagent/src/index.ts +++ b/packages/subagent/subagent/src/index.ts @@ -65,6 +65,7 @@ import type { ContinuableSetupContribution } from './activation-setup-registry.t import { listChildren as listSubagentChildren } from './list-children.ts' import type { SubagentListEntry } from './list-children.ts' import { snapshotSubagentDescriptor } from './descriptor.ts' +import { subagentTimingProjectionDefinition } from './projection.ts' export * from './out-of-process.ts' export { SubagentRunId } from './types.ts' @@ -117,6 +118,7 @@ export type { export type { ContinuableSetupContribution } from './activation-setup-registry.ts' export type { SubagentListEntry } from './list-children.ts' export type { SubagentRunEndInfo, SubagentRunInfo } from './types.ts' +export type { SubagentTimingProjection } from './projection-types.ts' declare module 'cordis' { interface Context { @@ -186,6 +188,9 @@ export class SubagentService extends Service { if (this.continuations === manager) this.continuations = undefined }, 'subagents.continuationBinding()') }) + ctx.inject(['sessionProjections'], (projectionCtx) => { + projectionCtx.sessionProjections.register(subagentTimingProjectionDefinition) + }) } /** diff --git a/packages/subagent/subagent/src/projection-types.ts b/packages/subagent/subagent/src/projection-types.ts new file mode 100644 index 0000000000..cefaec3727 --- /dev/null +++ b/packages/subagent/subagent/src/projection-types.ts @@ -0,0 +1,20 @@ +/** + * Pure client-safe subagent projection vocabulary. + * + * @module @deepseek-ai/dsh-subagent/projection-types + */ + +/** Durable active-turn timing for one descriptor-backed child session. */ +export interface SubagentTimingProjection { + /** Milliseconds accumulated across completed turns after the child's own descriptor. */ + settledMs: number + /** Start of the currently open turn, when one has not reached `turn/end`. */ + activeSince?: number +} + +declare module '@deepseek-ai/dsh-session-projection/types' { + interface SessionProjectionMap { + /** Active-turn duration for a descriptor-backed subagent session. */ + subagentTiming: SubagentTimingProjection + } +} diff --git a/packages/subagent/subagent/src/projection.ts b/packages/subagent/subagent/src/projection.ts new file mode 100644 index 0000000000..6b15a66bbf --- /dev/null +++ b/packages/subagent/subagent/src/projection.ts @@ -0,0 +1,68 @@ +/** + * Pure session projection for subagent active-turn duration. + * + * @module @deepseek-ai/dsh-subagent/projection + */ + +import { z } from 'zod' +import type { ProjectionDefinition } from '@deepseek-ai/dsh-session-projection' +import type { SubagentTimingProjection } from './projection-types.ts' + +interface TimingState extends SubagentTimingProjection { + /** Latest pre-descriptor turn start, promoted when the child's own descriptor arrives. */ + pendingTurnStart?: number + /** Whether the fold has crossed a descriptor in this logical log. */ + descriptorSeen: boolean +} + +const projectionSchema = z.object({ + settledMs: z.number().int().nonnegative(), + activeSince: z.number().int().nonnegative().optional(), +}).strict() as unknown as z.ZodType + +/** + * Fold turn boundaries around the child's own durable descriptor. + * + * A fork seed may contain an ancestor descriptor and completed turns. Every + * descriptor therefore resets the accumulated state; the healthy catalog + * admits only a child with exactly one descriptor in its own suffix, making + * the final reset the child's authoritative timing origin. + */ +export const subagentTimingProjectionDefinition: +ProjectionDefinition<'subagentTiming', TimingState> = { + key: 'subagentTiming', + schema: projectionSchema, + init: () => ({ descriptorSeen: false, settledMs: 0 }), + apply: (state, event) => { + if (event.type === 'turn/start') { + return state.descriptorSeen + ? { ...state, activeSince: event.time } + : { ...state, pendingTurnStart: event.time } + } + if (event.type === 'subagent/descriptor') { + const activeSince = state.activeSince ?? state.pendingTurnStart + return { + descriptorSeen: true, + settledMs: 0, + ...(activeSince === undefined ? {} : { activeSince }), + } + } + if (event.type !== 'turn/end') return state + if (!state.descriptorSeen) { + if (state.pendingTurnStart === undefined) return state + const { pendingTurnStart: _closed, ...next } = state + return next + } + if (state.activeSince === undefined) return state + const { activeSince, ...rest } = state + return { + ...rest, + settledMs: state.settledMs + Math.max(0, event.time - activeSince), + } + }, + view: state => ({ + settledMs: state.settledMs, + ...(state.activeSince === undefined ? {} : { activeSince: state.activeSince }), + }), + stateVersion: 1, +} diff --git a/packages/subagent/subagent/tests/timing-projection.spec.ts b/packages/subagent/subagent/tests/timing-projection.spec.ts new file mode 100644 index 0000000000..41c8b8896c --- /dev/null +++ b/packages/subagent/subagent/tests/timing-projection.spec.ts @@ -0,0 +1,51 @@ +import { describe, expect, it } from 'vitest' +import type { SessionEvent } from '@deepseek-ai/dsh-session' +import { subagentTimingProjectionDefinition } from '../src/projection.ts' + +function event(type: SessionEvent['type'], seq: number, time: number): SessionEvent { + return { type, seq, time, data: {} } as SessionEvent +} + +function fold(events: SessionEvent[]) { + let state = subagentTimingProjectionDefinition.init() + for (const item of events) state = subagentTimingProjectionDefinition.apply(state, item) + return subagentTimingProjectionDefinition.view(state) +} + +describe('subagent timing projection', () => { + it('resets inherited seed timing at the child descriptor and sums later completed turns', () => { + expect(fold([ + event('turn/start', 0, 100), + event('subagent/descriptor', 1, 110), + event('turn/end', 2, 300), + event('turn/start', 3, 1_000), + event('subagent/descriptor', 4, 1_100), + event('turn/end', 5, 4_100), + event('turn/start', 6, 10_000), + event('turn/end', 7, 12_000), + ])).toEqual({ settledMs: 5_100 }) + }) + + it('exposes an open turn start and never subtracts time for reversed boundaries', () => { + expect(fold([ + event('turn/start', 0, 1_000), + event('subagent/descriptor', 1, 1_100), + event('turn/end', 2, 900), + event('turn/start', 3, 2_000), + event('assistant/chunk', 4, 2_500), + ])).toEqual({ settledMs: 0, activeSince: 2_000 }) + }) + + it('ignores completed pre-descriptor turns and unrelated events', () => { + const initial = subagentTimingProjectionDefinition.init() + expect(subagentTimingProjectionDefinition.apply( + initial, + event('assistant/chunk', 0, 1), + )).toBe(initial) + expect(fold([ + event('turn/start', 0, 100), + event('turn/end', 1, 200), + event('subagent/descriptor', 2, 300), + ])).toEqual({ settledMs: 0 }) + }) +}) diff --git a/packages/subagent/subagent/tsconfig.json b/packages/subagent/subagent/tsconfig.json index 1c3a5fb6de..612330c646 100644 --- a/packages/subagent/subagent/tsconfig.json +++ b/packages/subagent/subagent/tsconfig.json @@ -32,6 +32,9 @@ { "path": "../../session-query/session-query" }, + { + "path": "../../session-projection/session-projection" + }, { "path": "../../tasks/tasks" }, diff --git a/pnpm-lock.yaml b/pnpm-lock.yaml index e1d8d15c03..2d45f42458 100644 --- a/pnpm-lock.yaml +++ b/pnpm-lock.yaml @@ -1897,6 +1897,12 @@ importers: '@deepseek-ai/dsh-invariants': specifier: workspace:^ version: link:../../support/invariants + '@deepseek-ai/dsh-subagent': + specifier: workspace:^ + version: link:../../subagent/subagent + '@deepseek-ai/dsh-token-meter': + specifier: workspace:^ + version: link:../../llm/token-meter '@types/react': specifier: ~18.3.1 version: 18.3.31 @@ -4882,6 +4888,10 @@ importers: version: link:../../../vendor/cordis packages/subagent/subagent: + dependencies: + zod: + specifier: ^4.4.3 + version: 4.4.3 devDependencies: '@deepseek-ai/dsh-agent': specifier: workspace:^ @@ -4904,6 +4914,9 @@ importers: '@deepseek-ai/dsh-session-persistence': specifier: workspace:^ version: link:../../session-persistence/session-persistence + '@deepseek-ai/dsh-session-projection': + specifier: workspace:^ + version: link:../../session-projection/session-projection '@deepseek-ai/dsh-session-query': specifier: workspace:^ version: link:../../session-query/session-query From 3a920be9c56c5cc653b64f09dafa095adcb6e546 Mon Sep 17 00:00:00 2001 From: kingwl Date: Mon, 3 Aug 2026 00:22:01 +0800 Subject: [PATCH 17/48] stack subagent token and duration metrics --- .../subagent-conversation/tree.expected.md | 6 +++--- packages/client/ui-subagent/README.i18n.yaml | 4 ++-- packages/client/ui-subagent/README.md | 2 +- packages/client/ui-subagent/README.zh.md | 2 +- .../client/SubagentCatalogAction.module.css | 13 +++++++++++- .../src/client/SubagentCatalogAction.tsx | 20 ++++++++++++++----- .../tests/conversation-ui.spec.tsx | 9 +++++++-- 7 files changed, 41 insertions(+), 15 deletions(-) diff --git a/apps/web/tests/snapshots/subagent-conversation/tree.expected.md b/apps/web/tests/snapshots/subagent-conversation/tree.expected.md index 6b695ff029..43ed15c649 100644 --- a/apps/web/tests/snapshots/subagent-conversation/tree.expected.md +++ b/apps/web/tests/snapshots/subagent-conversation/tree.expected.md @@ -2,7 +2,7 @@ - treeitem "event-sourcing researcher Explain event sourcing in one · continuable · not running 7.9K tok · {{duration}}" [expanded] [level=1]: - button "Collapse event-sourcing researcher descendants": - img - - text: event-sourcing researcher Explain event sourcing in one · continuable · not running 7.9K tok · {{duration}} + - text: event-sourcing researcher Explain event sourcing in one · continuable · not running 7.9K tok {{duration}} - group: - - treeitem "example editor continuable · not running 0 tok · {{duration}}" [level=2] - - treeitem "event-sourcing reviewer one-shot · not running 0 tok · {{duration}}" [level=1] + - treeitem "example editor continuable · not running 0 tok · {{duration}}" [level=2]: example editor continuable · not running 0 tok {{duration}} + - treeitem "event-sourcing reviewer one-shot · not running 0 tok · {{duration}}" [level=1]: event-sourcing reviewer one-shot · not running 0 tok {{duration}} diff --git a/packages/client/ui-subagent/README.i18n.yaml b/packages/client/ui-subagent/README.i18n.yaml index 8af7265330..8d1f594300 100644 --- a/packages/client/ui-subagent/README.i18n.yaml +++ b/packages/client/ui-subagent/README.i18n.yaml @@ -2,5 +2,5 @@ # side as of the last confirmed-consistent state. Both languages carry equal authority; # after editing either side, bring the other along and re-record with: # pnpm run verify-translation-pairing --write packages/client/ui-subagent/README.md -README.md: 2d507dde49f3796f3bbd00c239a768cdfd8a561d -README.zh.md: d0143858ff7eac1437fce381378ebcfad2187f2a +README.md: 538daeffb61f642b2e430cc1a6e2b1f3e3d5f55e +README.zh.md: a39b9c2dcec74a06e2074cc60f1c1f79b99b1ec0 diff --git a/packages/client/ui-subagent/README.md b/packages/client/ui-subagent/README.md index 2d507dde49..538daeffb6 100644 --- a/packages/client/ui-subagent/README.md +++ b/packages/client/ui-subagent/README.md @@ -4,7 +4,7 @@ English | [中文](README.zh.md) Web subagent feature owner: contributes the lazily expandable catalog tree to `conversation.session.header.actions`, reason-specific read-only replacements to the conversation composer chain, and the existing `@` reference source to `ctx.slash`. -The header action reads `subagentsByParent` and session summaries through the standard `useSessions` hook. After a non-empty direct catalog arrives, its trigger counts the complete subagent-only descendant lineage, stops at ordinary forks, and shows ongoing activity when any counted descendant is running. The compact tree remains direct-catalog authoritative: continuable and one-shot rows display mode plus `running`/`inactive` activity, an optional log-backed title, total durable provider usage, and active-turn duration to the second; an unlabeled one-shot row falls back to its session id, while corrupt, unsupported, or unavailable rows remain readable but disabled. Token totals sum the four disjoint `tokenUsage` buckets. Duration sums completed `subagentTiming` turns, advances once per second only for an open turn on a running child, and freezes after the child becomes inactive; an interrupted open turn is bounded by the session summary's last activity. Each healthy row's `hasChildren` hint determines disclosure before interaction, so known leaves never show an arrow; expanding a branch immediately reserves one disabled loading row per known direct descendant, then lazily replaces them with that child's authoritative catalog. Every visible branch is reported to the runtime so membership frames cause a debounced refresh only where the tree is being consumed. Selecting any depth calls `SessionsService.openSubagent()` with the row's exact `{parentSessionId, childSessionId, mode}` address. Component-local state owns tree visibility, expanded branches, keyboard focus, and the running-duration clock. ArrowRight/ArrowLeft expand and collapse branches; ArrowUp/ArrowDown, Home, End, and Escape navigate or close the tree; closing returns focus to the trigger. Styling uses tokens only. +The header action reads `subagentsByParent` and session summaries through the standard `useSessions` hook. After a non-empty direct catalog arrives, its trigger counts the complete subagent-only descendant lineage, stops at ordinary forks, and shows ongoing activity when any counted descendant is running. The compact tree remains direct-catalog authoritative: continuable and one-shot rows display mode plus `running`/`inactive` activity and an optional log-backed title, while the trailing column stacks total durable provider usage above active-turn duration to the second; an unlabeled one-shot row falls back to its session id, while corrupt, unsupported, or unavailable rows remain readable but disabled. Token totals sum the four disjoint `tokenUsage` buckets. Duration sums completed `subagentTiming` turns, advances once per second only for an open turn on a running child, and freezes after the child becomes inactive; an interrupted open turn is bounded by the session summary's last activity. Each healthy row's `hasChildren` hint determines disclosure before interaction, so known leaves never show an arrow; expanding a branch immediately reserves one disabled loading row per known direct descendant, then lazily replaces them with that child's authoritative catalog. Every visible branch is reported to the runtime so membership frames cause a debounced refresh only where the tree is being consumed. Selecting any depth calls `SessionsService.openSubagent()` with the row's exact `{parentSessionId, childSessionId, mode}` address. Component-local state owns tree visibility, expanded branches, keyboard focus, and the running-duration clock. ArrowRight/ArrowLeft expand and collapse branches; ArrowUp/ArrowDown, Home, End, and Escape navigate or close the tree; closing returns focus to the trigger. Styling uses tokens only. A one-shot child always elects a read-only composer that identifies the transcript as a completed execution record. A continuable child does so only when its exact parent is unavailable, with copy explaining the recovery path. A continuable child with a live parent keeps the ordinary input chrome, whose Session routes through `subagent.prompt`; running input remains Send because every follow-up joins the child's FIFO inbox, and addressed sessions never expose Stop. This package never receives host context or calls a model-facing tool. The catalog and composer behavior are specified by the [Web subagent conversations Agent Note](../../../.agents/notes/implemented/feature/2026-07-27-web-subagent-conversations.md). diff --git a/packages/client/ui-subagent/README.zh.md b/packages/client/ui-subagent/README.zh.md index d0143858ff..a39b9c2dce 100644 --- a/packages/client/ui-subagent/README.zh.md +++ b/packages/client/ui-subagent/README.zh.md @@ -4,7 +4,7 @@ Web subagent 功能 owner：向 `conversation.session.header.actions` 贡献可懒加载展开的目录树，向会话编辑器链贡献按原因区分的只读替代呈现，并保留注册到 `ctx.slash` 的既有 `@` 引用 source。 -页头操作通过标准 `useSessions` 钩子读取 `subagentsByParent` 与会话摘要。非空直接目录到达后，其触发器会统计仅含 subagent 的完整后代谱系，在普通 fork 处停止，并在任一计入统计的后代处于 `running` 时显示活动仍在进行。紧凑树仍以直接目录为权威依据：可继续和 one-shot 行会显示 mode、`running`／`inactive` 活动状态、由日志支撑的可选 title、提供方的持久化总用量，以及精确到秒的活跃轮次耗时；没有 label 的 one-shot 行会回退到其会话 id，而损坏、不受支持或不可用的行仍保持可读但禁用。token 总量为四个互不重叠的 `tokenUsage` 桶之和。耗时会累加已完成的 `subagentTiming` 轮次，仅在运行中 child 存在未结束轮次时每秒递增一次，并在 child 变为 inactive 后冻结；被中断的未结束轮次以会话摘要中的最后活动为上界。每个健康行的 `hasChildren` 提示会在交互前决定是否显示展开控件，因此已知叶子节点从不显示箭头；展开分支时，会立即为每个已知直接后代预留一行禁用的加载行，随后再用该 child 的权威目录懒加载结果替换这些占位行。每个可见分支都会上报给运行时，使成员帧只在树正被消费的位置触发去抖动刷新。选择任意深度的条目都会使用该行的确切地址 `{parentSessionId, childSessionId, mode}` 调用 `SessionsService.openSubagent()`。组件局部状态负责树的可见性、已展开分支、键盘焦点与运行中耗时时钟。ArrowRight／ArrowLeft 展开和折叠分支；ArrowUp／ArrowDown、Home、End 与 Escape 用于导航或关闭树；关闭后焦点返回触发器。样式只使用 token。 +页头操作通过标准 `useSessions` 钩子读取 `subagentsByParent` 与会话摘要。非空直接目录到达后，其触发器会统计仅含 subagent 的完整后代谱系，在普通 fork 处停止，并在任一计入统计的后代处于 `running` 时显示活动仍在进行。紧凑树仍以直接目录为权威依据：可继续和 one-shot 行会显示 mode、`running`／`inactive` 活动状态和由日志支撑的可选 title，尾随列则将提供方的持久化总用量置于上行，将精确到秒的活跃轮次耗时置于下行；没有 label 的 one-shot 行会回退到其会话 id，而损坏、不受支持或不可用的行仍保持可读但禁用。token 总量为四个互不重叠的 `tokenUsage` 桶之和。耗时会累加已完成的 `subagentTiming` 轮次，仅在运行中 child 存在未结束轮次时每秒递增一次，并在 child 变为 inactive 后冻结；被中断的未结束轮次以会话摘要中的最后活动为上界。每个健康行的 `hasChildren` 提示会在交互前决定是否显示展开控件，因此已知叶子节点从不显示箭头；展开分支时，会立即为每个已知直接后代预留一行禁用的加载行，随后再用该 child 的权威目录懒加载结果替换这些占位行。每个可见分支都会上报给运行时，使成员帧只在树正被消费的位置触发去抖动刷新。选择任意深度的条目都会使用该行的确切地址 `{parentSessionId, childSessionId, mode}` 调用 `SessionsService.openSubagent()`。组件局部状态负责树的可见性、已展开分支、键盘焦点与运行中耗时时钟。ArrowRight／ArrowLeft 展开和折叠分支；ArrowUp／ArrowDown、Home、End 与 Escape 用于导航或关闭树；关闭后焦点返回触发器。样式只使用 token。 one-shot child 始终选用只读编辑器，并将 transcript（文本记录）说明为已完成的执行记录。可继续 child 仅在其确切 parent 不可用时选用只读编辑器，并以文案说明恢复路径。确切 parent 存活时，可继续 child 保留普通输入 chrome，其 Session 会通过 `subagent.prompt` 路由；child 运行期间，输入操作仍为 Send，因为每条后续消息都会进入 child 的 FIFO inbox，且已寻址会话绝不公开 Stop。本包绝不接收宿主 context，也不调用面向模型的工具。目录与编辑器行为由 [Web subagent 对话 Agent Note](../../../.agents/notes/implemented/feature/2026-07-27-web-subagent-conversations.md)规定。 diff --git a/packages/client/ui-subagent/src/client/SubagentCatalogAction.module.css b/packages/client/ui-subagent/src/client/SubagentCatalogAction.module.css index 13004e6a47..239081c59c 100644 --- a/packages/client/ui-subagent/src/client/SubagentCatalogAction.module.css +++ b/packages/client/ui-subagent/src/client/SubagentCatalogAction.module.css @@ -180,12 +180,23 @@ } .metrics { + display: grid; + grid-template-rows: 18px 16px; flex: none; - margin-top: 16px; font-variant-numeric: tabular-nums; + text-align: right; white-space: nowrap; } +.metricToken { + grid-row: 1; + line-height: 18px; +} + +.metricDuration { + grid-row: 2; +} + .children { position: relative; margin-left: 18px; diff --git a/packages/client/ui-subagent/src/client/SubagentCatalogAction.tsx b/packages/client/ui-subagent/src/client/SubagentCatalogAction.tsx index 8b6471fd46..2875e22ed5 100644 --- a/packages/client/ui-subagent/src/client/SubagentCatalogAction.tsx +++ b/packages/client/ui-subagent/src/client/SubagentCatalogAction.tsx @@ -248,10 +248,15 @@ function CatalogRows({ summary?.updatedAt, now, ) - const metrics = [ - totalTokens === undefined ? undefined : `${formatTokens(totalTokens)} tok`, - durationMs === undefined ? undefined : formatDuration(durationMs, t), - ].filter(value => value !== undefined).join(' · ') + const tokenMetric = totalTokens === undefined + ? undefined + : `${formatTokens(totalTokens)} tok` + const durationMetric = durationMs === undefined + ? undefined + : formatDuration(durationMs, t) + const metrics = [tokenMetric, durationMetric] + .filter(value => value !== undefined) + .join(' · ') const open = (): void => { openChild({ parentSessionId, childSessionId: entry.id, mode: entry.mode }) @@ -308,7 +313,12 @@ function CatalogRows({ {label} {secondary} - {metrics !== '' && {metrics}} + {metrics !== '' && ( + + {tokenMetric !== undefined && {tokenMetric}} + {durationMetric !== undefined && {durationMetric}} + + )} {isExpanded && !knownLeaf && ( diff --git a/packages/client/ui-subagent/tests/conversation-ui.spec.tsx b/packages/client/ui-subagent/tests/conversation-ui.spec.tsx index 9716b6bbf3..c43c31148f 100644 --- a/packages/client/ui-subagent/tests/conversation-ui.spec.tsx +++ b/packages/client/ui-subagent/tests/conversation-ui.spec.tsx @@ -1,6 +1,6 @@ // @vitest-environment jsdom import { afterEach, describe, expect, it, vi } from 'vitest' -import { cleanup, fireEvent, render, screen } from '@testing-library/react' +import { cleanup, fireEvent, render, screen, within } from '@testing-library/react' import { makeTranslate } from '@deepseek-ai/dsh-client-test-runtime' import type { SessionId, SessionListState, SessionSummary, SubagentCatalogSnapshot, @@ -282,7 +282,12 @@ describe('SubagentCatalogAction', () => { render() fireEvent.click(screen.getByRole('button', { name: /3 个子代理/ })) - expect(screen.getByRole('treeitem', { name: /running.*4\.6K tok · 1分10秒/ })).toBeTruthy() + const runningRow = screen.getByRole('treeitem', { name: /running.*4\.6K tok · 1分10秒/ }) + const runningMetrics = within(runningRow) + const tokenMetric = runningMetrics.getByText('4.6K tok') + const durationMetric = runningMetrics.getByText('1分10秒') + expect(tokenMetric.parentElement).toBe(durationMetric.parentElement) + expect(tokenMetric.nextElementSibling).toBe(durationMetric) expect(screen.getByRole('treeitem', { name: /finished.*4\.6K tok · 1小时02分03秒/ })).toBeTruthy() expect(screen.getByRole('treeitem', { name: /interrupted.*4\.6K tok · 6秒/ })).toBeTruthy() From c831c99981f280ca098b1883b9200f6b12d056c5 Mon Sep 17 00:00:00 2001 From: kingwl Date: Mon, 3 Aug 2026 01:08:32 +0800 Subject: [PATCH 18/48] keep subagent duration fix focused --- ...07-27-web-subagent-conversations.i18n.yaml | 4 +- .../2026-07-27-web-subagent-conversations.md | 10 ++-- ...026-07-27-web-subagent-conversations.zh.md | 10 ++-- .../subagent-conversation/tree.expected.md | 8 +-- docs/module-graph.md | 20 +++---- packages/client/ui-subagent/README.i18n.yaml | 4 +- packages/client/ui-subagent/README.md | 2 +- packages/client/ui-subagent/README.zh.md | 2 +- packages/client/ui-subagent/package.json | 2 - .../client/SubagentCatalogAction.module.css | 17 ++---- .../src/client/SubagentCatalogAction.tsx | 52 ++++--------------- .../tests/conversation-ui.spec.tsx | 27 +++------- packages/client/ui-subagent/tsconfig.json | 3 -- .../subagent/tests/timing-projection.spec.ts | 26 ++++++++++ pnpm-lock.yaml | 3 -- 15 files changed, 79 insertions(+), 111 deletions(-) diff --git a/.agents/notes/implemented/feature/2026-07-27-web-subagent-conversations.i18n.yaml b/.agents/notes/implemented/feature/2026-07-27-web-subagent-conversations.i18n.yaml index d07d50641b..168d28d13a 100644 --- a/.agents/notes/implemented/feature/2026-07-27-web-subagent-conversations.i18n.yaml +++ b/.agents/notes/implemented/feature/2026-07-27-web-subagent-conversations.i18n.yaml @@ -2,5 +2,5 @@ # side as of the last confirmed-consistent state. Both languages carry equal authority; # after editing either side, bring the other along and re-record with: # pnpm run verify-translation-pairing --write .agents/notes/implemented/feature/2026-07-27-web-subagent-conversations.md -2026-07-27-web-subagent-conversations.md: f4d2035dfc7224cd7a11575449ff79cdae3fce48 -2026-07-27-web-subagent-conversations.zh.md: 79ce2711af32fb67685f29d48fc53d29376907f4 +2026-07-27-web-subagent-conversations.md: 859c6c5c17e830ab55c8513d56741966655eaf7a +2026-07-27-web-subagent-conversations.zh.md: 05d5c0f1d59b0bdebdecb33dc360e937af44d7b6 diff --git a/.agents/notes/implemented/feature/2026-07-27-web-subagent-conversations.md b/.agents/notes/implemented/feature/2026-07-27-web-subagent-conversations.md index f4d2035dfc..859c6c5c17 100644 --- a/.agents/notes/implemented/feature/2026-07-27-web-subagent-conversations.md +++ b/.agents/notes/implemented/feature/2026-07-27-web-subagent-conversations.md @@ -33,7 +33,7 @@ The Figma [subagent list](https://www.figma.com/design/jRBBK7zBgcszdVWQ0Fh5J8/Ha | The session header opens a compact child list. | The trigger aggregates the complete subagent-only descendant lineage; the tree shows every direct catalog entry in service order, including disabled diagnostics. | | Selecting a row reuses the conversation UI. | Addressed history never activates the child; only a continuable row with a live parent retains the ordinary composer. | | Nested agents expand progressively. | Each row carries a one-level `hasChildren` snapshot; disclosure reserves known direct-descendant rows immediately, then loads only that row's direct catalog and retains its own parent address. | -| Rows show labels, state, usage, and active duration without duplicating sidebar rows. | Mode and `running`/`inactive` activity are textual as well as visual; optional title, durable token usage, and exact active-turn duration come from the list's retained projection values. `SessionHeader.origin` removes duplicate navigation rows but grants no capability. | +| Rows show labels, state, and active duration without duplicating sidebar rows. | Mode and `running`/`inactive` activity are textual as well as visual; optional title and exact active-turn duration come from the list's retained projection values. `SessionHeader.origin` removes duplicate navigation rows but grants no capability. | ## Product contract @@ -41,7 +41,7 @@ The header action is absent only when a complete empty direct-catalog response a `running` means the exact child Agent driver is draining work at the Host sampling boundary; `inactive` means that driver is idle or absent. The UI does not translate either value into success, failure, cancellation, completeness, or resumability. `subagent.list` supplies the current driver-status baseline, `host/session-status` updates known activity in place, request-local replay prevents an older in-flight list response from overwriting a newer transition, and `host/session-removed` returns a known row to `inactive`; reconnect reads a fresh baseline. A `host/session-added` frame for a direct subagent immediately flips any loaded parent row to `hasChildren: true`, and that positive hint survives an older in-flight catalog response; membership, labels, mode, diagnostics, and the authoritative snapshot still require a debounced `subagent.list` refresh while the affected branch is open. A prompt response remains delivery-time authority. -Healthy rows reuse the standard session projections retained in the list mirror. The token figure sums the four disjoint `tokenUsage` buckets across the durable log. `subagentTiming` resets at every descriptor so an inherited fork seed cannot enter the child's total, accumulates completed `turn/start` → `turn/end` spans, and carries the current turn's `activeSince`. The menu formats whole seconds and advances its local clock only while a known descendant is running; an inactive row uses settled duration, or the summary's last activity to bound an interrupted open turn, so reopening the menu never restarts completed work. Token chunks do not change `subagentTiming` and therefore do not add a per-token list update path. Neither metric implies a durable outcome. +Healthy rows reuse the standard session projections retained in the list mirror. `subagentTiming` resets at every descriptor so an inherited fork seed cannot enter the child's total, accumulates completed `turn/start` → `turn/end` spans, and carries the current turn's `activeSince`. The menu formats whole seconds and advances its local clock only while a known descendant is running; an inactive row uses settled duration, or the summary's last activity to bound an interrupted open turn, so reopening the menu never restarts completed work. The duration does not imply a durable outcome. Selecting a row records its exact address before opening the resident client `Session`. History pagination, event folding, tool render intents, titles, and live mux reconciliation reuse the ordinary conversation machinery. Breadcrumbs use catalog labels, follow parent links only through `origin: 'subagent'` rows, include the first ordinary owner, and keep ordinary forks single-level. Forking an addressed subagent creates an ordinary fork with direct source lineage and attaches it to the nearest workspace-owning ancestor. The catalog is an ARIA tree with lazy ArrowRight/ArrowLeft disclosure, linear ArrowUp/ArrowDown navigation, Home/End, Escape, and focus restoration. @@ -104,13 +104,13 @@ The shipped Web composition mounts SQLite session query beside JSONL persistence - Host protocol tests pin schemas including required boolean expandability, id echoing, mode verification, non-activating history, exact-parent enforcement, FIFO admission receipts, cancellation, and sanitized failure mapping. - Generic Host tests pin attached and cold history and forks without Agent publication, cold projection folding, descriptor/origin/runtime-owner denial, explicit-id adoption denial, and the direct queue-control fence. - Client object tests pin retained and restored addresses, one-shot read-only rejection, history routing, continuable prompt routing, no addressed cancellation, suppression of Agent-bound model controls, live activity flips including in-flight response replay and detach fallback, subagent-parent expandability flips, and membership refresh. -- jsdom tests pin the aggregate descendant count and activity, token totals, second-precision running and frozen inactive durations, the summary-backed root action across absent and stale-empty catalogs, known loading-row shape, mixed-mode rows, pre-click leaf disclosure, diagnostics, lazy descendant disclosure, direct-parent addresses, keyboard behavior, and both read-only reasons. -- The keyless assembled Web snapshot contains an inactive continuable child with durable usage, an inactive one-shot sibling, and a persisted grandchild; it pins the three-descendant trigger across a stale empty catalog response, usage and timing rows, and aggregate running transition, expands without activation, opens persisted history, admits a human FIFO follow-up, reconciles child mux events, and proves one-shot history remains read-only. +- jsdom tests pin the aggregate descendant count and activity, second-precision running and frozen inactive durations, the summary-backed root action across absent and stale-empty catalogs, known loading-row shape, mixed-mode rows, pre-click leaf disclosure, diagnostics, lazy descendant disclosure, direct-parent addresses, keyboard behavior, and both read-only reasons. +- The keyless assembled Web snapshot contains an inactive continuable child, an inactive one-shot sibling, and a persisted grandchild; it pins the three-descendant trigger across a stale empty catalog response, timing rows, and aggregate running transition, expands without activation, opens persisted history, admits a human FIFO follow-up, reconciles child mux events, and proves one-shot history remains read-only. - Navigation tests pin subagent-only breadcrumbs, workspace placement for forks created from subagents, and `origin: 'subagent'` sidebar filtering without hiding ordinary forks. ## Consequences -- Catalog reads may rescan persisted lineage and each direct candidate's descriptor log, but expandability reuses only descendant headers already present in that trace; the Web activity baseline adds one Agent-registry lookup per healthy row and then uses existing live frames, while usage and duration reuse projection baselines and pushes with no per-row log read, and membership refresh stays debounced and single-flight. +- Catalog reads may rescan persisted lineage and each direct candidate's descriptor log, but expandability reuses only descendant headers already present in that trace; the Web activity baseline adds one Agent-registry lookup per healthy row and then uses existing live frames, while duration reuses projection baselines and pushes with no per-row log read, and membership refresh stays debounced and single-flight. - Parent availability, child activity, and `hasChildren` are snapshots. Publication, disposal, another sender, or another process may win after listing; typed prompt failure remains expected. - A child may publish between history fetch and mux subscription, so the existing sequence reconciliation also covers the cold-to-live addressed path. - Persisted origin adds one deliberately weak product-classification field to child headers and list projections; it cannot become an authorization shortcut. diff --git a/.agents/notes/implemented/feature/2026-07-27-web-subagent-conversations.zh.md b/.agents/notes/implemented/feature/2026-07-27-web-subagent-conversations.zh.md index 79ce2711af..05d5c0f1d5 100644 --- a/.agents/notes/implemented/feature/2026-07-27-web-subagent-conversations.zh.md +++ b/.agents/notes/implemented/feature/2026-07-27-web-subagent-conversations.zh.md @@ -33,7 +33,7 @@ Figma 中的 [subagent 列表](https://www.figma.com/design/jRBBK7zBgcszdVWQ0Fh5 | 会话页头可打开紧凑的 child 列表。 | 触发器会汇总仅含 subagent 的完整后代谱系；树按服务顺序显示每个直接目录条目，包括已禁用的 diagnostic。 | | 选择一行会复用对话 UI。 | 已寻址历史绝不激活 child；只有 parent 存活的可继续行才保留普通输入框。 | | 嵌套 agent 会逐层展开。 | 每行携带一层 `hasChildren` 快照；展开时会立即预留已知直接后代行，随后仍只加载该行的直接目录，并保留其自身的 parent 地址。 | -| 条目显示 label、状态、用量与活跃耗时，同时避免侧边栏条目重复。 | mode 与 `running`／`inactive` 活动状态会同时以文字和视觉呈现；可选 title、持久化 token 用量与精确的活跃轮次耗时来自列表保留的投影值。`SessionHeader.origin` 会移除重复的导航条目，但不授予任何功能权限。 | +| 条目显示 label、状态与活跃耗时，同时避免侧边栏条目重复。 | mode 与 `running`／`inactive` 活动状态会同时以文字和视觉呈现；可选 title 与精确的活跃轮次耗时来自列表保留的投影值。`SessionHeader.origin` 会移除重复的导航条目，但不授予任何功能权限。 | ## 产品契约 @@ -41,7 +41,7 @@ Figma 中的 [subagent 列表](https://www.figma.com/design/jRBBK7zBgcszdVWQ0Fh5 `running` 表示在 Host 采样边界，确切 child Agent driver 正在处理工作；`inactive` 表示该 driver 空闲或不存在。UI 不会把任一值解释为成功、失败、取消、完成状态或可恢复性。`subagent.list` 提供当前 driver 状态基线，`host/session-status` 会就地更新已知活动状态，请求内回放会阻止更早发起但尚未完成的列表响应覆盖较新的状态转换，`host/session-removed` 则会使已知行恢复为 `inactive`；重连时会读取新的基线。直接 subagent 的 `host/session-added` 帧会立即把任何已加载的 parent 行翻转为 `hasChildren: true`，并使这项正向提示不被更早发起但尚未完成的目录响应覆盖；受影响分支打开期间，成员、label、mode、diagnostic 与权威快照仍需要通过去抖动的 `subagent.list` 刷新来更新。消息投递时仍以提示词响应为权威依据。 -健康行会复用列表镜像中保留的标准会话投影。token 数值会汇总持久化日志中四个互不重叠的 `tokenUsage` 桶。`subagentTiming` 会在每个描述符处重置，使继承的 fork 种子不会计入 child 总量；它会累加已完成的 `turn/start` → `turn/end` 时段，并携带当前轮次的 `activeSince`。菜单会以整秒格式化时间，且仅在有已知后代处于运行状态时才推进其本地时钟；对 inactive 行，菜单使用已结算耗时，或以摘要的最后活动为被中断未结束轮次的上界，因此重新打开菜单绝不会让已完成工作重新计时。token 分片不会改变 `subagentTiming`，因此不会增加按 token 更新列表的路径。这两项指标都不蕴含持久化结果语义。 +健康行会复用列表镜像中保留的标准会话投影。`subagentTiming` 会在每个描述符处重置，使继承的 fork 种子不会计入 child 总量；它会累加已完成的 `turn/start` → `turn/end` 时段，并携带当前轮次的 `activeSince`。菜单会以整秒格式化时间，且仅在有已知后代处于运行状态时才推进其本地时钟；对 inactive 行，菜单使用已结算耗时，或以摘要的最后活动为被中断未结束轮次的上界，因此重新打开菜单绝不会让已完成工作重新计时。该耗时不蕴含持久化结果语义。 选择一行后，系统会先记录其确切地址，再打开常驻客户端 `Session`。历史分页、事件 fold、工具渲染意图、title 与实时 mux 归并都会复用普通对话机制。面包屑导航使用目录 label，只会沿 `origin: 'subagent'` 行的父链接逐级回溯，包含第一个普通 owner，并让普通 fork 保持单层。从已寻址 subagent 创建 fork 时，会生成具有直接源谱系的普通 fork，并将其附加到最近拥有 Workspace 的祖先。目录是一棵 ARIA 树，支持懒加载式 ArrowRight／ArrowLeft 展开与折叠、线性 ArrowUp／ArrowDown 导航、Home／End、Escape 以及焦点恢复。 @@ -104,13 +104,13 @@ one-shot 行始终会用文案替代输入框，说明执行记录为只读。 - 宿主协议测试固定 schema（包括必需的布尔可展开性）、id 回显、mode 校验、非激活式历史、确切 parent 强制要求、FIFO 准入回执、取消与脱敏后的失败映射。 - 通用 Host 测试固定在不发布 Agent 的情况下读取已附加与冷态历史及执行 fork、冷态投影归并、按描述符／origin／运行时 owner 拒绝、拒绝显式 id 接纳，以及直接队列控制栅栏。 - 客户端对象测试固定已保留与已恢复的地址、one-shot 只读拒绝、历史路由、可继续提示词路由、已寻址对话不提供取消、屏蔽绑定到 agent 的模型控件、实时活动状态翻转（包括在途响应回放与 detach 回退）、subagent parent 可展开性翻转与成员刷新。 -- jsdom 测试固定后代聚合计数与活动状态、token 总量、精确到秒的运行中耗时与冻结后 inactive 耗时、目录缺失或为陈旧空目录时由摘要支撑的根操作、已知加载行的形态、混合 mode 行、点击前的叶子展开控件、diagnostic、后代懒加载展开、直接 parent 地址、键盘行为与两种只读原因。 -- 无密钥的组装 Web 快照包含一个具有持久化用量的 inactive 可继续 child、一个 inactive 的 one-shot sibling 和一个持久化 grandchild；它会固定触发器在一次陈旧的空目录响应后仍显示三个后代，并固定用量与计时行以及聚合 `running` 状态转换，在不激活的情况下展开、打开持久化历史、准入一条用户 FIFO 后续消息、归并 child mux 事件，并证明 one-shot 历史仍然只读。 +- jsdom 测试固定后代聚合计数与活动状态、精确到秒的运行中耗时与冻结后 inactive 耗时、目录缺失或为陈旧空目录时由摘要支撑的根操作、已知加载行的形态、混合 mode 行、点击前的叶子展开控件、diagnostic、后代懒加载展开、直接 parent 地址、键盘行为与两种只读原因。 +- 无密钥的组装 Web 快照包含一个 inactive 的可继续 child、一个 inactive 的 one-shot sibling 和一个持久化 grandchild；它会固定触发器在一次陈旧的空目录响应后仍显示三个后代，并固定计时行以及聚合 `running` 状态转换，在不激活的情况下展开、打开持久化历史、准入一条用户 FIFO 后续消息、归并 child mux 事件，并证明 one-shot 历史仍然只读。 - 导航测试固定仅含 subagent 的面包屑导航、从 subagent 创建 fork 时的 Workspace 归属，以及 `origin: 'subagent'` 侧边栏过滤，同时不隐藏普通 fork。 ## 后果 -- 目录读取可能重新扫描持久化谱系与每个直接候选的描述符日志，但可展开性只复用该追踪中已有的后代 header；Web 活动基线会为每个健康行增加一次 Agent 注册表查找，随后使用现有实时帧，而用量与耗时会复用投影基线和推送，无需按行读取日志，成员刷新则保持去抖动和单次并发。 +- 目录读取可能重新扫描持久化谱系与每个直接候选的描述符日志，但可展开性只复用该追踪中已有的后代 header；Web 活动基线会为每个健康行增加一次 Agent 注册表查找，随后使用现有实时帧，而耗时会复用投影基线和推送，无需按行读取日志，成员刷新则保持去抖动和单次并发。 - parent 可用性、child 活动状态与 `hasChildren` 都是快照。列出之后，发布、dispose、其他发送方或其他进程都可能抢先改变状态；类型化提示词失败仍属预期行为。 - child 可能在历史获取与 mux 订阅之间发布，因此现有序号归并也涵盖从冷态转为存活的已寻址路径。 - 持久化 origin 会为 child header 与列表投影添加一个有意保持弱约束的产品分类字段；它不能变成授权捷径。 diff --git a/apps/web/tests/snapshots/subagent-conversation/tree.expected.md b/apps/web/tests/snapshots/subagent-conversation/tree.expected.md index 43ed15c649..c1174c042e 100644 --- a/apps/web/tests/snapshots/subagent-conversation/tree.expected.md +++ b/apps/web/tests/snapshots/subagent-conversation/tree.expected.md @@ -1,8 +1,8 @@ - tree "Subagent sessions": - - treeitem "event-sourcing researcher Explain event sourcing in one · continuable · not running 7.9K tok · {{duration}}" [expanded] [level=1]: + - treeitem "event-sourcing researcher Explain event sourcing in one · continuable · not running {{duration}}" [expanded] [level=1]: - button "Collapse event-sourcing researcher descendants": - img - - text: event-sourcing researcher Explain event sourcing in one · continuable · not running 7.9K tok {{duration}} + - text: event-sourcing researcher Explain event sourcing in one · continuable · not running {{duration}} - group: - - treeitem "example editor continuable · not running 0 tok · {{duration}}" [level=2]: example editor continuable · not running 0 tok {{duration}} - - treeitem "event-sourcing reviewer one-shot · not running 0 tok · {{duration}}" [level=1]: event-sourcing reviewer one-shot · not running 0 tok {{duration}} + - treeitem "example editor continuable · not running {{duration}}" [level=2] + - treeitem "event-sourcing reviewer one-shot · not running {{duration}}" [level=1] diff --git a/docs/module-graph.md b/docs/module-graph.md index 6687a583c1..313c58d914 100644 --- a/docs/module-graph.md +++ b/docs/module-graph.md @@ -737,6 +737,7 @@ flowchart TD pkg_subagent --> pkg_scope pkg_subagent --> pkg_session pkg_subagent --> pkg_session_persistence + pkg_subagent --> pkg_session_projection pkg_subagent --> pkg_session_query pkg_subagent --> pkg_tasks pkg_subagent --> pkg_tools @@ -832,13 +833,6 @@ flowchart TD pkg_client_ui_goal --> pkg_client_ui_slots pkg_client_ui_goal --> pkg_goal pkg_client_ui_goal --> pkg_invariants - pkg_client_ui_subagent --> pkg_client_locale - pkg_client_ui_subagent --> pkg_client_runtime - pkg_client_ui_subagent --> pkg_client_ui_conversation - pkg_client_ui_subagent --> pkg_client_ui_primitives - pkg_client_ui_subagent --> pkg_client_ui_slash - pkg_client_ui_subagent --> pkg_client_ui_slots - pkg_client_ui_subagent --> pkg_invariants pkg_session_reference --> pkg_agent pkg_session_reference --> pkg_compact pkg_session_reference --> pkg_invariants @@ -981,6 +975,14 @@ flowchart TD pkg_client_ui_plan --> pkg_client_ui_slots pkg_client_ui_plan --> pkg_invariants pkg_client_ui_plan --> pkg_plan_mode + pkg_client_ui_subagent --> pkg_client_locale + pkg_client_ui_subagent --> pkg_client_runtime + pkg_client_ui_subagent --> pkg_client_ui_conversation + pkg_client_ui_subagent --> pkg_client_ui_primitives + pkg_client_ui_subagent --> pkg_client_ui_slash + pkg_client_ui_subagent --> pkg_client_ui_slots + pkg_client_ui_subagent --> pkg_invariants + pkg_client_ui_subagent --> pkg_subagent pkg_agent_spine_demo --> pkg_agent pkg_agent_spine_demo --> pkg_agent_loop pkg_agent_spine_demo --> pkg_goal @@ -1201,7 +1203,7 @@ flowchart TD | [`tool-fs-search`](../packages/fs/tool-fs-search) | `fs` | [`bash`](../packages/bash/bash), [`invariants`](../packages/support/invariants), [`llm`](../packages/llm/llm), [`retention`](../packages/util/retention), [`session`](../packages/core/session), [`spill`](../packages/spill/spill), [`system-prompt`](../packages/core/system-prompt), [`tools`](../packages/core/tools) | | [`tool-str-replace-editor`](../packages/fs/tool-str-replace-editor) | `fs` | [`fs`](../packages/fs/fs), [`invariants`](../packages/support/invariants), [`sandbox`](../packages/sandbox/sandbox), [`sandbox-policy`](../packages/sandbox/sandbox-policy), [`tools`](../packages/core/tools) | | [`tool-skill`](../packages/skill/tool-skill) | `skill` | [`agent`](../packages/core/agent), [`invariants`](../packages/support/invariants), [`llm`](../packages/llm/llm), [`skill`](../packages/skill/skill), [`tools`](../packages/core/tools) | -| [`subagent`](../packages/subagent/subagent) | `subagent` | [`agent`](../packages/core/agent), [`brand`](../packages/util/brand), [`invariants`](../packages/support/invariants), [`llm`](../packages/llm/llm), [`scope`](../packages/core/scope), [`session`](../packages/core/session), [`session-persistence`](../packages/session-persistence/session-persistence), [`session-query`](../packages/session-query/session-query), [`tasks`](../packages/tasks/tasks), [`tools`](../packages/core/tools) | +| [`subagent`](../packages/subagent/subagent) | `subagent` | [`agent`](../packages/core/agent), [`brand`](../packages/util/brand), [`invariants`](../packages/support/invariants), [`llm`](../packages/llm/llm), [`scope`](../packages/core/scope), [`session`](../packages/core/session), [`session-persistence`](../packages/session-persistence/session-persistence), [`session-projection`](../packages/session-projection/session-projection), [`session-query`](../packages/session-query/session-query), [`tasks`](../packages/tasks/tasks), [`tools`](../packages/core/tools) | | [`tool-web`](../packages/web/tool-web) | `web` | [`invariants`](../packages/support/invariants), [`llm`](../packages/llm/llm), [`system-prompt`](../packages/core/system-prompt), [`tools`](../packages/core/tools), [`web`](../packages/web/web) | | [`spill-policy`](../packages/spill/spill-policy) | `spill` | [`invariants`](../packages/support/invariants), [`llm`](../packages/llm/llm), [`retention`](../packages/util/retention), [`session`](../packages/core/session), [`spill`](../packages/spill/spill), [`tools`](../packages/core/tools) | | [`timeout-policy`](../packages/timeout/timeout-policy) | `timeout` | [`invariants`](../packages/support/invariants), [`llm`](../packages/llm/llm), [`timeout`](../packages/util/timeout), [`tools`](../packages/core/tools) | @@ -1218,7 +1220,6 @@ flowchart TD | [`tool-ask-user`](../packages/ui/tool-ask-user) | `ui` | [`agent`](../packages/core/agent), [`invariants`](../packages/support/invariants), [`tools`](../packages/core/tools), [`user-interaction`](../packages/ui/user-interaction) | | [`client-ui-command`](../packages/client/ui-command) | `client` | [`client-connection`](../packages/client/connection), [`client-locale`](../packages/client/locale), [`client-runtime`](../packages/client/runtime), [`client-ui-conversation`](../packages/client/ui-conversation), [`client-ui-primitives`](../packages/client/ui-primitives), [`client-ui-slash`](../packages/client/ui-slash), [`client-ui-slots`](../packages/client/ui-slots), [`invariants`](../packages/support/invariants) | | [`client-ui-goal`](../packages/client/ui-goal) | `client` | [`client-connection`](../packages/client/connection), [`client-locale`](../packages/client/locale), [`client-runtime`](../packages/client/runtime), [`client-ui-conversation`](../packages/client/ui-conversation), [`client-ui-primitives`](../packages/client/ui-primitives), [`client-ui-slots`](../packages/client/ui-slots), [`goal`](../packages/goal/goal), [`invariants`](../packages/support/invariants) | -| [`client-ui-subagent`](../packages/client/ui-subagent) | `client` | [`client-locale`](../packages/client/locale), [`client-runtime`](../packages/client/runtime), [`client-ui-conversation`](../packages/client/ui-conversation), [`client-ui-primitives`](../packages/client/ui-primitives), [`client-ui-slash`](../packages/client/ui-slash), [`client-ui-slots`](../packages/client/ui-slots), [`invariants`](../packages/support/invariants) | | [`session-reference`](../packages/context/session-reference) | `context` | [`agent`](../packages/core/agent), [`compact`](../packages/compact/compact), [`invariants`](../packages/support/invariants), [`llm`](../packages/llm/llm), [`retention`](../packages/util/retention), [`session`](../packages/core/session), [`session-query`](../packages/session-query/session-query) | | [`workspace-context`](../packages/context/workspace-context) | `context` | [`agent`](../packages/core/agent), [`fs`](../packages/fs/fs), [`invariants`](../packages/support/invariants), [`llm`](../packages/llm/llm), [`paths`](../packages/util/paths), [`session`](../packages/core/session), [`tools`](../packages/core/tools) | | [`repeat-tool-guard`](../packages/guard/repeat-tool-guard) | `guard` | [`agent`](../packages/core/agent), [`invariants`](../packages/support/invariants), [`tools`](../packages/core/tools) | @@ -1239,6 +1240,7 @@ flowchart TD | [`client-ui-model`](../packages/client/ui-model) | `client` | [`client-connection`](../packages/client/connection), [`client-locale`](../packages/client/locale), [`client-runtime`](../packages/client/runtime), [`client-ui-command`](../packages/client/ui-command), [`client-ui-conversation`](../packages/client/ui-conversation), [`client-ui-primitives`](../packages/client/ui-primitives), [`client-ui-slash`](../packages/client/ui-slash), [`client-ui-slots`](../packages/client/ui-slots), [`invariants`](../packages/support/invariants) | | [`client-ui-permission`](../packages/client/ui-permission) | `client` | [`client-connection`](../packages/client/connection), [`client-locale`](../packages/client/locale), [`client-runtime`](../packages/client/runtime), [`client-schema-form`](../packages/client/schema-form), [`client-ui-command`](../packages/client/ui-command), [`client-ui-primitives`](../packages/client/ui-primitives), [`client-ui-slash`](../packages/client/ui-slash), [`client-ui-slots`](../packages/client/ui-slots), [`invariants`](../packages/support/invariants), [`permission`](../packages/ui/permission) | | [`client-ui-plan`](../packages/client/ui-plan) | `client` | [`client-connection`](../packages/client/connection), [`client-locale`](../packages/client/locale), [`client-runtime`](../packages/client/runtime), [`client-ui-conversation`](../packages/client/ui-conversation), [`client-ui-primitives`](../packages/client/ui-primitives), [`client-ui-slots`](../packages/client/ui-slots), [`invariants`](../packages/support/invariants), [`plan-mode`](../packages/plan/plan-mode) | +| [`client-ui-subagent`](../packages/client/ui-subagent) | `client` | [`client-locale`](../packages/client/locale), [`client-runtime`](../packages/client/runtime), [`client-ui-conversation`](../packages/client/ui-conversation), [`client-ui-primitives`](../packages/client/ui-primitives), [`client-ui-slash`](../packages/client/ui-slash), [`client-ui-slots`](../packages/client/ui-slots), [`invariants`](../packages/support/invariants), [`subagent`](../packages/subagent/subagent) | | [`agent-spine-demo`](../packages/examples/agent-spine-demo) | `examples` | [`agent`](../packages/core/agent), [`agent-loop`](../packages/core/agent-loop), [`goal`](../packages/goal/goal), [`goal-session`](../packages/goal/goal-session), [`invariants`](../packages/support/invariants), [`llm`](../packages/llm/llm), [`llm-retry`](../packages/llm/llm-retry), [`paths`](../packages/util/paths), [`scope`](../packages/core/scope), [`session`](../packages/core/session), [`session-title`](../packages/session-title/session-title), [`skill`](../packages/skill/skill), [`skill-local`](../packages/skill/skill-local), [`system-prompt`](../packages/core/system-prompt), [`tasks-local`](../packages/tasks/tasks-local), [`tool-bash`](../packages/bash/tool-bash), [`tool-goal`](../packages/goal/tool-goal), [`tool-skill`](../packages/skill/tool-skill), [`tool-tasks`](../packages/tasks/tool-tasks), [`tools`](../packages/core/tools), [`workspace-context`](../packages/context/workspace-context) | | [`sdk-protocol`](../packages/sdk/sdk-protocol) | `sdk` | [`invariants`](../packages/support/invariants), [`llm`](../packages/llm/llm), [`session`](../packages/core/session), [`subagent`](../packages/subagent/subagent) | | [`tool-ralph`](../packages/workflow/tool-ralph) | `workflow` | [`agent`](../packages/core/agent), [`invariants`](../packages/support/invariants), [`llm`](../packages/llm/llm), [`subagent`](../packages/subagent/subagent), [`system-prompt`](../packages/core/system-prompt), [`tools`](../packages/core/tools), [`workflow`](../packages/workflow/workflow) | diff --git a/packages/client/ui-subagent/README.i18n.yaml b/packages/client/ui-subagent/README.i18n.yaml index 8d1f594300..2b512252f0 100644 --- a/packages/client/ui-subagent/README.i18n.yaml +++ b/packages/client/ui-subagent/README.i18n.yaml @@ -2,5 +2,5 @@ # side as of the last confirmed-consistent state. Both languages carry equal authority; # after editing either side, bring the other along and re-record with: # pnpm run verify-translation-pairing --write packages/client/ui-subagent/README.md -README.md: 538daeffb61f642b2e430cc1a6e2b1f3e3d5f55e -README.zh.md: a39b9c2dcec74a06e2074cc60f1c1f79b99b1ec0 +README.md: 16a54484fb53544c71af0806c1497a18f9141002 +README.zh.md: 166a25d095b3e7f10f7239262f39e27972344529 diff --git a/packages/client/ui-subagent/README.md b/packages/client/ui-subagent/README.md index 538daeffb6..16a54484fb 100644 --- a/packages/client/ui-subagent/README.md +++ b/packages/client/ui-subagent/README.md @@ -4,7 +4,7 @@ English | [中文](README.zh.md) Web subagent feature owner: contributes the lazily expandable catalog tree to `conversation.session.header.actions`, reason-specific read-only replacements to the conversation composer chain, and the existing `@` reference source to `ctx.slash`. -The header action reads `subagentsByParent` and session summaries through the standard `useSessions` hook. After a non-empty direct catalog arrives, its trigger counts the complete subagent-only descendant lineage, stops at ordinary forks, and shows ongoing activity when any counted descendant is running. The compact tree remains direct-catalog authoritative: continuable and one-shot rows display mode plus `running`/`inactive` activity and an optional log-backed title, while the trailing column stacks total durable provider usage above active-turn duration to the second; an unlabeled one-shot row falls back to its session id, while corrupt, unsupported, or unavailable rows remain readable but disabled. Token totals sum the four disjoint `tokenUsage` buckets. Duration sums completed `subagentTiming` turns, advances once per second only for an open turn on a running child, and freezes after the child becomes inactive; an interrupted open turn is bounded by the session summary's last activity. Each healthy row's `hasChildren` hint determines disclosure before interaction, so known leaves never show an arrow; expanding a branch immediately reserves one disabled loading row per known direct descendant, then lazily replaces them with that child's authoritative catalog. Every visible branch is reported to the runtime so membership frames cause a debounced refresh only where the tree is being consumed. Selecting any depth calls `SessionsService.openSubagent()` with the row's exact `{parentSessionId, childSessionId, mode}` address. Component-local state owns tree visibility, expanded branches, keyboard focus, and the running-duration clock. ArrowRight/ArrowLeft expand and collapse branches; ArrowUp/ArrowDown, Home, End, and Escape navigate or close the tree; closing returns focus to the trigger. Styling uses tokens only. +The header action reads `subagentsByParent` and session summaries through the standard `useSessions` hook. After a non-empty direct catalog arrives, its trigger counts the complete subagent-only descendant lineage, stops at ordinary forks, and shows ongoing activity when any counted descendant is running. The compact tree remains direct-catalog authoritative: continuable and one-shot rows display mode plus `running`/`inactive` activity, an optional log-backed title, and active-turn duration to the second; an unlabeled one-shot row falls back to its session id, while corrupt, unsupported, or unavailable rows remain readable but disabled. Duration sums completed `subagentTiming` turns, advances once per second only for an open turn on a running child, and freezes after the child becomes inactive; an interrupted open turn is bounded by the session summary's last activity. Each healthy row's `hasChildren` hint determines disclosure before interaction, so known leaves never show an arrow; expanding a branch immediately reserves one disabled loading row per known direct descendant, then lazily replaces them with that child's authoritative catalog. Every visible branch is reported to the runtime so membership frames cause a debounced refresh only where the tree is being consumed. Selecting any depth calls `SessionsService.openSubagent()` with the row's exact `{parentSessionId, childSessionId, mode}` address. Component-local state owns tree visibility, expanded branches, keyboard focus, and the running-duration clock. ArrowRight/ArrowLeft expand and collapse branches; ArrowUp/ArrowDown, Home, End, and Escape navigate or close the tree; closing returns focus to the trigger. Styling uses tokens only. A one-shot child always elects a read-only composer that identifies the transcript as a completed execution record. A continuable child does so only when its exact parent is unavailable, with copy explaining the recovery path. A continuable child with a live parent keeps the ordinary input chrome, whose Session routes through `subagent.prompt`; running input remains Send because every follow-up joins the child's FIFO inbox, and addressed sessions never expose Stop. This package never receives host context or calls a model-facing tool. The catalog and composer behavior are specified by the [Web subagent conversations Agent Note](../../../.agents/notes/implemented/feature/2026-07-27-web-subagent-conversations.md). diff --git a/packages/client/ui-subagent/README.zh.md b/packages/client/ui-subagent/README.zh.md index a39b9c2dce..166a25d095 100644 --- a/packages/client/ui-subagent/README.zh.md +++ b/packages/client/ui-subagent/README.zh.md @@ -4,7 +4,7 @@ Web subagent 功能 owner：向 `conversation.session.header.actions` 贡献可懒加载展开的目录树，向会话编辑器链贡献按原因区分的只读替代呈现，并保留注册到 `ctx.slash` 的既有 `@` 引用 source。 -页头操作通过标准 `useSessions` 钩子读取 `subagentsByParent` 与会话摘要。非空直接目录到达后，其触发器会统计仅含 subagent 的完整后代谱系，在普通 fork 处停止，并在任一计入统计的后代处于 `running` 时显示活动仍在进行。紧凑树仍以直接目录为权威依据：可继续和 one-shot 行会显示 mode、`running`／`inactive` 活动状态和由日志支撑的可选 title，尾随列则将提供方的持久化总用量置于上行，将精确到秒的活跃轮次耗时置于下行；没有 label 的 one-shot 行会回退到其会话 id，而损坏、不受支持或不可用的行仍保持可读但禁用。token 总量为四个互不重叠的 `tokenUsage` 桶之和。耗时会累加已完成的 `subagentTiming` 轮次，仅在运行中 child 存在未结束轮次时每秒递增一次，并在 child 变为 inactive 后冻结；被中断的未结束轮次以会话摘要中的最后活动为上界。每个健康行的 `hasChildren` 提示会在交互前决定是否显示展开控件，因此已知叶子节点从不显示箭头；展开分支时，会立即为每个已知直接后代预留一行禁用的加载行，随后再用该 child 的权威目录懒加载结果替换这些占位行。每个可见分支都会上报给运行时，使成员帧只在树正被消费的位置触发去抖动刷新。选择任意深度的条目都会使用该行的确切地址 `{parentSessionId, childSessionId, mode}` 调用 `SessionsService.openSubagent()`。组件局部状态负责树的可见性、已展开分支、键盘焦点与运行中耗时时钟。ArrowRight／ArrowLeft 展开和折叠分支；ArrowUp／ArrowDown、Home、End 与 Escape 用于导航或关闭树；关闭后焦点返回触发器。样式只使用 token。 +页头操作通过标准 `useSessions` 钩子读取 `subagentsByParent` 与会话摘要。非空直接目录到达后，其触发器会统计仅含 subagent 的完整后代谱系，在普通 fork 处停止，并在任一计入统计的后代处于 `running` 时显示活动仍在进行。紧凑树仍以直接目录为权威依据：可继续和 one-shot 行会显示 mode、`running`／`inactive` 活动状态、由日志支撑的可选 title，以及精确到秒的活跃轮次耗时；没有 label 的 one-shot 行会回退到其会话 id，而损坏、不受支持或不可用的行仍保持可读但禁用。耗时会累加已完成的 `subagentTiming` 轮次，仅在运行中 child 存在未结束轮次时每秒递增一次，并在 child 变为 inactive 后冻结；被中断的未结束轮次以会话摘要中的最后活动为上界。每个健康行的 `hasChildren` 提示会在交互前决定是否显示展开控件，因此已知叶子节点从不显示箭头；展开分支时，会立即为每个已知直接后代预留一行禁用的加载行，随后再用该 child 的权威目录懒加载结果替换这些占位行。每个可见分支都会上报给运行时，使成员帧只在树正被消费的位置触发去抖动刷新。选择任意深度的条目都会使用该行的确切地址 `{parentSessionId, childSessionId, mode}` 调用 `SessionsService.openSubagent()`。组件局部状态负责树的可见性、已展开分支、键盘焦点与运行中耗时时钟。ArrowRight／ArrowLeft 展开和折叠分支；ArrowUp／ArrowDown、Home、End 与 Escape 用于导航或关闭树；关闭后焦点返回触发器。样式只使用 token。 one-shot child 始终选用只读编辑器，并将 transcript（文本记录）说明为已完成的执行记录。可继续 child 仅在其确切 parent 不可用时选用只读编辑器，并以文案说明恢复路径。确切 parent 存活时，可继续 child 保留普通输入 chrome，其 Session 会通过 `subagent.prompt` 路由；child 运行期间，输入操作仍为 Send，因为每条后续消息都会进入 child 的 FIFO inbox，且已寻址会话绝不公开 Stop。本包绝不接收宿主 context，也不调用面向模型的工具。目录与编辑器行为由 [Web subagent 对话 Agent Note](../../../.agents/notes/implemented/feature/2026-07-27-web-subagent-conversations.md)规定。 diff --git a/packages/client/ui-subagent/package.json b/packages/client/ui-subagent/package.json index 9e6b120c1e..573bf54097 100644 --- a/packages/client/ui-subagent/package.json +++ b/packages/client/ui-subagent/package.json @@ -49,7 +49,6 @@ "@deepseek-ai/dsh-client-ui-slots": "^0.0.1", "@deepseek-ai/dsh-invariants": "^0.0.1", "@deepseek-ai/dsh-subagent": "^0.0.1", - "@deepseek-ai/dsh-token-meter": "^0.0.1", "cordis": "^4.0.0-rc.7" }, "devDependencies": { @@ -62,7 +61,6 @@ "@deepseek-ai/dsh-client-ui-slots": "workspace:^", "@deepseek-ai/dsh-invariants": "workspace:^", "@deepseek-ai/dsh-subagent": "workspace:^", - "@deepseek-ai/dsh-token-meter": "workspace:^", "@types/react": "~18.3.1", "cordis": "^4.0.0-rc.7" }, diff --git a/packages/client/ui-subagent/src/client/SubagentCatalogAction.module.css b/packages/client/ui-subagent/src/client/SubagentCatalogAction.module.css index 239081c59c..d8642879ea 100644 --- a/packages/client/ui-subagent/src/client/SubagentCatalogAction.module.css +++ b/packages/client/ui-subagent/src/client/SubagentCatalogAction.module.css @@ -173,30 +173,19 @@ } .summary, -.metrics { +.time { color: var(--dsw-alias-label-tertiary); font-size: 11px; line-height: 16px; } -.metrics { - display: grid; - grid-template-rows: 18px 16px; +.time { flex: none; + margin-top: 16px; font-variant-numeric: tabular-nums; - text-align: right; white-space: nowrap; } -.metricToken { - grid-row: 1; - line-height: 18px; -} - -.metricDuration { - grid-row: 2; -} - .children { position: relative; margin-left: 18px; diff --git a/packages/client/ui-subagent/src/client/SubagentCatalogAction.tsx b/packages/client/ui-subagent/src/client/SubagentCatalogAction.tsx index 2875e22ed5..d507b30019 100644 --- a/packages/client/ui-subagent/src/client/SubagentCatalogAction.tsx +++ b/packages/client/ui-subagent/src/client/SubagentCatalogAction.tsx @@ -12,7 +12,6 @@ import type { PropsLocale, PropsRuntime, TranslateNS } from '@deepseek-ai/dsh-cl import { NS } from './locales.ts' import type {} from '@deepseek-ai/dsh-client-ui-conversation/client' import type {} from '@deepseek-ai/dsh-subagent/client' -import type {} from '@deepseek-ai/dsh-token-meter/client' import css from './SubagentCatalogAction.module.css' type CatalogEntry = SubagentCatalogSnapshot['entries'][number] @@ -60,36 +59,18 @@ function treeItems(root: HTMLDivElement | null): HTMLElement[] { : Array.from(root.querySelectorAll('[role="treeitem"]:not([aria-disabled="true"])')) } -/** Compact token count shared in shape with the conversation stats strip. */ -function formatTokens(value: number): string { - const scaled = (next: number): string => next >= 100 - ? String(Math.round(next)) - : String(Math.round(next * 10) / 10) - if (value < 1_000) return String(value) - if (value < 1_000_000) return `${scaled(value / 1_000)}K` - return `${scaled(value / 1_000_000)}M` -} - -/** Sum the four disjoint durable provider-usage buckets. */ -function tokenTotal( - usage: SessionProjectionMap['tokenUsage'] | undefined, -): number | undefined { - return usage === undefined - ? undefined - : usage.uncachedInputTokens + usage.outputTokens - + usage.cacheReadTokens + usage.cacheWriteTokens -} - /** Exact whole-second active-turn duration for one catalog row. */ function activityDuration( - timing: SessionProjectionMap['subagentTiming'] | undefined, + summary: SessionSummary | undefined, activity: 'running' | 'inactive', - updatedAt: number | undefined, now: number, ): number | undefined { + if (summary === undefined) return undefined + const timing: SessionProjectionMap['subagentTiming'] | undefined + = summary.projectionValues?.subagentTiming if (timing === undefined) return undefined if (timing.activeSince === undefined) return timing.settledMs - const end = activity === 'running' ? now : updatedAt ?? timing.activeSince + const end = activity === 'running' ? now : summary.updatedAt return timing.settledMs + Math.max(0, end - timing.activeSince) } @@ -241,22 +222,14 @@ function CatalogRows({ const secondary = [summary?.title, mode, activity] .filter(value => value !== undefined) .join(' · ') - const totalTokens = tokenTotal(summary?.projectionValues?.tokenUsage) const durationMs = activityDuration( - summary?.projectionValues?.subagentTiming, + summary, entry.activity, - summary?.updatedAt, now, ) - const tokenMetric = totalTokens === undefined - ? undefined - : `${formatTokens(totalTokens)} tok` - const durationMetric = durationMs === undefined + const duration = durationMs === undefined ? undefined : formatDuration(durationMs, t) - const metrics = [tokenMetric, durationMetric] - .filter(value => value !== undefined) - .join(' · ') const open = (): void => { openChild({ parentSessionId, childSessionId: entry.id, mode: entry.mode }) @@ -288,7 +261,9 @@ function CatalogRows({ role="treeitem" tabIndex={0} aria-level={level} - aria-label={[label, secondary, metrics].filter(value => value !== '').join(' ')} + aria-label={[label, secondary, duration] + .filter(value => value !== undefined) + .join(' ')} {...knownLeaf ? {} : { 'aria-expanded': isExpanded }} className={css.row} onClick={open} @@ -313,12 +288,7 @@ function CatalogRows({ {label} {secondary} - {metrics !== '' && ( - - {tokenMetric !== undefined && {tokenMetric}} - {durationMetric !== undefined && {durationMetric}} - - )} + {duration !== undefined && {duration}} {isExpanded && !knownLeaf && ( diff --git a/packages/client/ui-subagent/tests/conversation-ui.spec.tsx b/packages/client/ui-subagent/tests/conversation-ui.spec.tsx index c43c31148f..487cd14f35 100644 --- a/packages/client/ui-subagent/tests/conversation-ui.spec.tsx +++ b/packages/client/ui-subagent/tests/conversation-ui.spec.tsx @@ -1,6 +1,6 @@ // @vitest-environment jsdom import { afterEach, describe, expect, it, vi } from 'vitest' -import { cleanup, fireEvent, render, screen, within } from '@testing-library/react' +import { cleanup, fireEvent, render, screen } from '@testing-library/react' import { makeTranslate } from '@deepseek-ai/dsh-client-test-runtime' import type { SessionId, SessionListState, SessionSummary, SubagentCatalogSnapshot, @@ -240,7 +240,7 @@ describe('SubagentCatalogAction', () => { }) }) - it('shows durable token totals, ticks active duration by seconds, and freezes inactive rows', async () => { + it('ticks active duration by seconds and freezes inactive rows', async () => { const now = 2_000_000_000_000 vi.useFakeTimers() vi.setSystemTime(now) @@ -269,12 +269,6 @@ describe('SubagentCatalogAction', () => { settledMs, ...(activeSince === undefined ? {} : { activeSince }), }, - tokenUsage: { - uncachedInputTokens: 1_000, - outputTokens: 200, - cacheReadTokens: 3_000, - cacheWriteTokens: 400, - }, }, }] })) as Record @@ -282,19 +276,14 @@ describe('SubagentCatalogAction', () => { render() fireEvent.click(screen.getByRole('button', { name: /3 个子代理/ })) - const runningRow = screen.getByRole('treeitem', { name: /running.*4\.6K tok · 1分10秒/ }) - const runningMetrics = within(runningRow) - const tokenMetric = runningMetrics.getByText('4.6K tok') - const durationMetric = runningMetrics.getByText('1分10秒') - expect(tokenMetric.parentElement).toBe(durationMetric.parentElement) - expect(tokenMetric.nextElementSibling).toBe(durationMetric) - expect(screen.getByRole('treeitem', { name: /finished.*4\.6K tok · 1小时02分03秒/ })).toBeTruthy() - expect(screen.getByRole('treeitem', { name: /interrupted.*4\.6K tok · 6秒/ })).toBeTruthy() + expect(screen.getByRole('treeitem', { name: /running.*1分10秒/ })).toBeTruthy() + expect(screen.getByRole('treeitem', { name: /finished.*1小时02分03秒/ })).toBeTruthy() + expect(screen.getByRole('treeitem', { name: /interrupted.*6秒/ })).toBeTruthy() await vi.advanceTimersByTimeAsync(1_000) - expect(screen.getByRole('treeitem', { name: /running.*4\.6K tok · 1分11秒/ })).toBeTruthy() - expect(screen.getByRole('treeitem', { name: /finished.*4\.6K tok · 1小时02分03秒/ })).toBeTruthy() - expect(screen.getByRole('treeitem', { name: /interrupted.*4\.6K tok · 6秒/ })).toBeTruthy() + expect(screen.getByRole('treeitem', { name: /running.*1分11秒/ })).toBeTruthy() + expect(screen.getByRole('treeitem', { name: /finished.*1小时02分03秒/ })).toBeTruthy() + expect(screen.getByRole('treeitem', { name: /interrupted.*6秒/ })).toBeTruthy() }) it('lazily expands and collapses descendant catalogs with direct-parent navigation', () => { diff --git a/packages/client/ui-subagent/tsconfig.json b/packages/client/ui-subagent/tsconfig.json index 27d06d16c0..e9d59a6fa1 100644 --- a/packages/client/ui-subagent/tsconfig.json +++ b/packages/client/ui-subagent/tsconfig.json @@ -29,9 +29,6 @@ { "path": "../ui-slots" }, - { - "path": "../../llm/token-meter" - }, { "path": "../../subagent/subagent" }, diff --git a/packages/subagent/subagent/tests/timing-projection.spec.ts b/packages/subagent/subagent/tests/timing-projection.spec.ts index 41c8b8896c..ac3cac9ebb 100644 --- a/packages/subagent/subagent/tests/timing-projection.spec.ts +++ b/packages/subagent/subagent/tests/timing-projection.spec.ts @@ -1,5 +1,9 @@ import { describe, expect, it } from 'vitest' +import { Context } from 'cordis' +import SessionStore from '@deepseek-ai/dsh-session' import type { SessionEvent } from '@deepseek-ai/dsh-session' +import SessionProjectionRegistry from '@deepseek-ai/dsh-session-projection' +import SubagentService from '../src/index.ts' import { subagentTimingProjectionDefinition } from '../src/projection.ts' function event(type: SessionEvent['type'], seq: number, time: number): SessionEvent { @@ -13,6 +17,16 @@ function fold(events: SessionEvent[]) { } describe('subagent timing projection', () => { + it('registers with the optional session projection registry', async () => { + const ctx = new Context() + await ctx.plugin(SessionStore) + await ctx.plugin(SessionProjectionRegistry) + await ctx.plugin(SubagentService) + + expect(ctx.sessionProjections.snapshot(ctx.sessions.create()).values.subagentTiming) + .toEqual({ settledMs: 0 }) + }) + it('resets inherited seed timing at the child descriptor and sums later completed turns', () => { expect(fold([ event('turn/start', 0, 100), @@ -42,6 +56,18 @@ describe('subagent timing projection', () => { initial, event('assistant/chunk', 0, 1), )).toBe(initial) + expect(subagentTimingProjectionDefinition.apply( + initial, + event('turn/end', 1, 2), + )).toBe(initial) + const descriptor = subagentTimingProjectionDefinition.apply( + initial, + event('subagent/descriptor', 2, 3), + ) + expect(subagentTimingProjectionDefinition.apply( + descriptor, + event('turn/end', 3, 4), + )).toBe(descriptor) expect(fold([ event('turn/start', 0, 100), event('turn/end', 1, 200), diff --git a/pnpm-lock.yaml b/pnpm-lock.yaml index 0b634898ea..1b0e2f5542 100644 --- a/pnpm-lock.yaml +++ b/pnpm-lock.yaml @@ -1906,9 +1906,6 @@ importers: '@deepseek-ai/dsh-subagent': specifier: workspace:^ version: link:../../subagent/subagent - '@deepseek-ai/dsh-token-meter': - specifier: workspace:^ - version: link:../../llm/token-meter '@types/react': specifier: ~18.3.1 version: 18.3.31 From 57e2433f455b10b37f959de710bda4532fda3118 Mon Sep 17 00:00:00 2001 From: kingwl Date: Mon, 3 Aug 2026 01:19:36 +0800 Subject: [PATCH 19/48] show subagent token metrics --- ...07-27-web-subagent-conversations.i18n.yaml | 4 +- .../2026-07-27-web-subagent-conversations.md | 10 ++--- ...026-07-27-web-subagent-conversations.zh.md | 10 ++--- .../subagent-conversation/tree.expected.md | 8 ++-- docs/module-graph.md | 3 +- packages/client/ui-subagent/README.i18n.yaml | 4 +- packages/client/ui-subagent/README.md | 2 +- packages/client/ui-subagent/README.zh.md | 2 +- packages/client/ui-subagent/package.json | 2 + .../client/SubagentCatalogAction.module.css | 17 ++++++-- .../src/client/SubagentCatalogAction.tsx | 41 +++++++++++++++--- .../tests/conversation-ui.spec.tsx | 42 +++++++++++++++---- packages/client/ui-subagent/tsconfig.json | 3 ++ pnpm-lock.yaml | 3 ++ 14 files changed, 114 insertions(+), 37 deletions(-) diff --git a/.agents/notes/implemented/feature/2026-07-27-web-subagent-conversations.i18n.yaml b/.agents/notes/implemented/feature/2026-07-27-web-subagent-conversations.i18n.yaml index 168d28d13a..aae06cf1d4 100644 --- a/.agents/notes/implemented/feature/2026-07-27-web-subagent-conversations.i18n.yaml +++ b/.agents/notes/implemented/feature/2026-07-27-web-subagent-conversations.i18n.yaml @@ -2,5 +2,5 @@ # side as of the last confirmed-consistent state. Both languages carry equal authority; # after editing either side, bring the other along and re-record with: # pnpm run verify-translation-pairing --write .agents/notes/implemented/feature/2026-07-27-web-subagent-conversations.md -2026-07-27-web-subagent-conversations.md: 859c6c5c17e830ab55c8513d56741966655eaf7a -2026-07-27-web-subagent-conversations.zh.md: 05d5c0f1d59b0bdebdecb33dc360e937af44d7b6 +2026-07-27-web-subagent-conversations.md: f4d2035dfc7224cd7a11575449ff79cdae3fce48 +2026-07-27-web-subagent-conversations.zh.md: 72ede6799de01961d82e93048652a70a7d3d83aa diff --git a/.agents/notes/implemented/feature/2026-07-27-web-subagent-conversations.md b/.agents/notes/implemented/feature/2026-07-27-web-subagent-conversations.md index 859c6c5c17..f4d2035dfc 100644 --- a/.agents/notes/implemented/feature/2026-07-27-web-subagent-conversations.md +++ b/.agents/notes/implemented/feature/2026-07-27-web-subagent-conversations.md @@ -33,7 +33,7 @@ The Figma [subagent list](https://www.figma.com/design/jRBBK7zBgcszdVWQ0Fh5J8/Ha | The session header opens a compact child list. | The trigger aggregates the complete subagent-only descendant lineage; the tree shows every direct catalog entry in service order, including disabled diagnostics. | | Selecting a row reuses the conversation UI. | Addressed history never activates the child; only a continuable row with a live parent retains the ordinary composer. | | Nested agents expand progressively. | Each row carries a one-level `hasChildren` snapshot; disclosure reserves known direct-descendant rows immediately, then loads only that row's direct catalog and retains its own parent address. | -| Rows show labels, state, and active duration without duplicating sidebar rows. | Mode and `running`/`inactive` activity are textual as well as visual; optional title and exact active-turn duration come from the list's retained projection values. `SessionHeader.origin` removes duplicate navigation rows but grants no capability. | +| Rows show labels, state, usage, and active duration without duplicating sidebar rows. | Mode and `running`/`inactive` activity are textual as well as visual; optional title, durable token usage, and exact active-turn duration come from the list's retained projection values. `SessionHeader.origin` removes duplicate navigation rows but grants no capability. | ## Product contract @@ -41,7 +41,7 @@ The header action is absent only when a complete empty direct-catalog response a `running` means the exact child Agent driver is draining work at the Host sampling boundary; `inactive` means that driver is idle or absent. The UI does not translate either value into success, failure, cancellation, completeness, or resumability. `subagent.list` supplies the current driver-status baseline, `host/session-status` updates known activity in place, request-local replay prevents an older in-flight list response from overwriting a newer transition, and `host/session-removed` returns a known row to `inactive`; reconnect reads a fresh baseline. A `host/session-added` frame for a direct subagent immediately flips any loaded parent row to `hasChildren: true`, and that positive hint survives an older in-flight catalog response; membership, labels, mode, diagnostics, and the authoritative snapshot still require a debounced `subagent.list` refresh while the affected branch is open. A prompt response remains delivery-time authority. -Healthy rows reuse the standard session projections retained in the list mirror. `subagentTiming` resets at every descriptor so an inherited fork seed cannot enter the child's total, accumulates completed `turn/start` → `turn/end` spans, and carries the current turn's `activeSince`. The menu formats whole seconds and advances its local clock only while a known descendant is running; an inactive row uses settled duration, or the summary's last activity to bound an interrupted open turn, so reopening the menu never restarts completed work. The duration does not imply a durable outcome. +Healthy rows reuse the standard session projections retained in the list mirror. The token figure sums the four disjoint `tokenUsage` buckets across the durable log. `subagentTiming` resets at every descriptor so an inherited fork seed cannot enter the child's total, accumulates completed `turn/start` → `turn/end` spans, and carries the current turn's `activeSince`. The menu formats whole seconds and advances its local clock only while a known descendant is running; an inactive row uses settled duration, or the summary's last activity to bound an interrupted open turn, so reopening the menu never restarts completed work. Token chunks do not change `subagentTiming` and therefore do not add a per-token list update path. Neither metric implies a durable outcome. Selecting a row records its exact address before opening the resident client `Session`. History pagination, event folding, tool render intents, titles, and live mux reconciliation reuse the ordinary conversation machinery. Breadcrumbs use catalog labels, follow parent links only through `origin: 'subagent'` rows, include the first ordinary owner, and keep ordinary forks single-level. Forking an addressed subagent creates an ordinary fork with direct source lineage and attaches it to the nearest workspace-owning ancestor. The catalog is an ARIA tree with lazy ArrowRight/ArrowLeft disclosure, linear ArrowUp/ArrowDown navigation, Home/End, Escape, and focus restoration. @@ -104,13 +104,13 @@ The shipped Web composition mounts SQLite session query beside JSONL persistence - Host protocol tests pin schemas including required boolean expandability, id echoing, mode verification, non-activating history, exact-parent enforcement, FIFO admission receipts, cancellation, and sanitized failure mapping. - Generic Host tests pin attached and cold history and forks without Agent publication, cold projection folding, descriptor/origin/runtime-owner denial, explicit-id adoption denial, and the direct queue-control fence. - Client object tests pin retained and restored addresses, one-shot read-only rejection, history routing, continuable prompt routing, no addressed cancellation, suppression of Agent-bound model controls, live activity flips including in-flight response replay and detach fallback, subagent-parent expandability flips, and membership refresh. -- jsdom tests pin the aggregate descendant count and activity, second-precision running and frozen inactive durations, the summary-backed root action across absent and stale-empty catalogs, known loading-row shape, mixed-mode rows, pre-click leaf disclosure, diagnostics, lazy descendant disclosure, direct-parent addresses, keyboard behavior, and both read-only reasons. -- The keyless assembled Web snapshot contains an inactive continuable child, an inactive one-shot sibling, and a persisted grandchild; it pins the three-descendant trigger across a stale empty catalog response, timing rows, and aggregate running transition, expands without activation, opens persisted history, admits a human FIFO follow-up, reconciles child mux events, and proves one-shot history remains read-only. +- jsdom tests pin the aggregate descendant count and activity, token totals, second-precision running and frozen inactive durations, the summary-backed root action across absent and stale-empty catalogs, known loading-row shape, mixed-mode rows, pre-click leaf disclosure, diagnostics, lazy descendant disclosure, direct-parent addresses, keyboard behavior, and both read-only reasons. +- The keyless assembled Web snapshot contains an inactive continuable child with durable usage, an inactive one-shot sibling, and a persisted grandchild; it pins the three-descendant trigger across a stale empty catalog response, usage and timing rows, and aggregate running transition, expands without activation, opens persisted history, admits a human FIFO follow-up, reconciles child mux events, and proves one-shot history remains read-only. - Navigation tests pin subagent-only breadcrumbs, workspace placement for forks created from subagents, and `origin: 'subagent'` sidebar filtering without hiding ordinary forks. ## Consequences -- Catalog reads may rescan persisted lineage and each direct candidate's descriptor log, but expandability reuses only descendant headers already present in that trace; the Web activity baseline adds one Agent-registry lookup per healthy row and then uses existing live frames, while duration reuses projection baselines and pushes with no per-row log read, and membership refresh stays debounced and single-flight. +- Catalog reads may rescan persisted lineage and each direct candidate's descriptor log, but expandability reuses only descendant headers already present in that trace; the Web activity baseline adds one Agent-registry lookup per healthy row and then uses existing live frames, while usage and duration reuse projection baselines and pushes with no per-row log read, and membership refresh stays debounced and single-flight. - Parent availability, child activity, and `hasChildren` are snapshots. Publication, disposal, another sender, or another process may win after listing; typed prompt failure remains expected. - A child may publish between history fetch and mux subscription, so the existing sequence reconciliation also covers the cold-to-live addressed path. - Persisted origin adds one deliberately weak product-classification field to child headers and list projections; it cannot become an authorization shortcut. diff --git a/.agents/notes/implemented/feature/2026-07-27-web-subagent-conversations.zh.md b/.agents/notes/implemented/feature/2026-07-27-web-subagent-conversations.zh.md index 05d5c0f1d5..72ede6799d 100644 --- a/.agents/notes/implemented/feature/2026-07-27-web-subagent-conversations.zh.md +++ b/.agents/notes/implemented/feature/2026-07-27-web-subagent-conversations.zh.md @@ -33,7 +33,7 @@ Figma 中的 [subagent 列表](https://www.figma.com/design/jRBBK7zBgcszdVWQ0Fh5 | 会话页头可打开紧凑的 child 列表。 | 触发器会汇总仅含 subagent 的完整后代谱系；树按服务顺序显示每个直接目录条目，包括已禁用的 diagnostic。 | | 选择一行会复用对话 UI。 | 已寻址历史绝不激活 child；只有 parent 存活的可继续行才保留普通输入框。 | | 嵌套 agent 会逐层展开。 | 每行携带一层 `hasChildren` 快照；展开时会立即预留已知直接后代行，随后仍只加载该行的直接目录，并保留其自身的 parent 地址。 | -| 条目显示 label、状态与活跃耗时，同时避免侧边栏条目重复。 | mode 与 `running`／`inactive` 活动状态会同时以文字和视觉呈现；可选 title 与精确的活跃轮次耗时来自列表保留的投影值。`SessionHeader.origin` 会移除重复的导航条目，但不授予任何功能权限。 | +| 条目显示 label、状态、token 用量与活跃耗时，同时避免侧边栏条目重复。 | mode 与 `running`／`inactive` 活动状态会同时以文字和视觉呈现；可选 title、持久化 token 用量与精确的活跃轮次耗时来自列表保留的投影值。`SessionHeader.origin` 会移除重复的导航条目，但不授予任何功能权限。 | ## 产品契约 @@ -41,7 +41,7 @@ Figma 中的 [subagent 列表](https://www.figma.com/design/jRBBK7zBgcszdVWQ0Fh5 `running` 表示在 Host 采样边界，确切 child Agent driver 正在处理工作；`inactive` 表示该 driver 空闲或不存在。UI 不会把任一值解释为成功、失败、取消、完成状态或可恢复性。`subagent.list` 提供当前 driver 状态基线，`host/session-status` 会就地更新已知活动状态，请求内回放会阻止更早发起但尚未完成的列表响应覆盖较新的状态转换，`host/session-removed` 则会使已知行恢复为 `inactive`；重连时会读取新的基线。直接 subagent 的 `host/session-added` 帧会立即把任何已加载的 parent 行翻转为 `hasChildren: true`，并使这项正向提示不被更早发起但尚未完成的目录响应覆盖；受影响分支打开期间，成员、label、mode、diagnostic 与权威快照仍需要通过去抖动的 `subagent.list` 刷新来更新。消息投递时仍以提示词响应为权威依据。 -健康行会复用列表镜像中保留的标准会话投影。`subagentTiming` 会在每个描述符处重置，使继承的 fork 种子不会计入 child 总量；它会累加已完成的 `turn/start` → `turn/end` 时段，并携带当前轮次的 `activeSince`。菜单会以整秒格式化时间，且仅在有已知后代处于运行状态时才推进其本地时钟；对 inactive 行，菜单使用已结算耗时，或以摘要的最后活动为被中断未结束轮次的上界，因此重新打开菜单绝不会让已完成工作重新计时。该耗时不蕴含持久化结果语义。 +健康行会复用列表镜像中保留的标准会话投影。token 用量数值会汇总持久化日志中四个互不重叠的 `tokenUsage` 桶。`subagentTiming` 会在每个描述符处重置，使继承的 fork 种子不会计入 child 总量；它会累加已完成的 `turn/start` → `turn/end` 时段，并携带当前轮次的 `activeSince`。菜单会以整秒格式化时间，且仅在有已知后代处于运行状态时才推进其本地时钟；对 inactive 行，菜单使用已结算耗时，或以摘要的最后活动为被中断未结束轮次的上界，因此重新打开菜单绝不会让已完成工作重新计时。token 分片不会改变 `subagentTiming`，因此不会增加按 token 更新列表的路径。这两项指标都不蕴含持久化结果语义。 选择一行后，系统会先记录其确切地址，再打开常驻客户端 `Session`。历史分页、事件 fold、工具渲染意图、title 与实时 mux 归并都会复用普通对话机制。面包屑导航使用目录 label，只会沿 `origin: 'subagent'` 行的父链接逐级回溯，包含第一个普通 owner，并让普通 fork 保持单层。从已寻址 subagent 创建 fork 时，会生成具有直接源谱系的普通 fork，并将其附加到最近拥有 Workspace 的祖先。目录是一棵 ARIA 树，支持懒加载式 ArrowRight／ArrowLeft 展开与折叠、线性 ArrowUp／ArrowDown 导航、Home／End、Escape 以及焦点恢复。 @@ -104,13 +104,13 @@ one-shot 行始终会用文案替代输入框，说明执行记录为只读。 - 宿主协议测试固定 schema（包括必需的布尔可展开性）、id 回显、mode 校验、非激活式历史、确切 parent 强制要求、FIFO 准入回执、取消与脱敏后的失败映射。 - 通用 Host 测试固定在不发布 Agent 的情况下读取已附加与冷态历史及执行 fork、冷态投影归并、按描述符／origin／运行时 owner 拒绝、拒绝显式 id 接纳，以及直接队列控制栅栏。 - 客户端对象测试固定已保留与已恢复的地址、one-shot 只读拒绝、历史路由、可继续提示词路由、已寻址对话不提供取消、屏蔽绑定到 agent 的模型控件、实时活动状态翻转（包括在途响应回放与 detach 回退）、subagent parent 可展开性翻转与成员刷新。 -- jsdom 测试固定后代聚合计数与活动状态、精确到秒的运行中耗时与冻结后 inactive 耗时、目录缺失或为陈旧空目录时由摘要支撑的根操作、已知加载行的形态、混合 mode 行、点击前的叶子展开控件、diagnostic、后代懒加载展开、直接 parent 地址、键盘行为与两种只读原因。 -- 无密钥的组装 Web 快照包含一个 inactive 的可继续 child、一个 inactive 的 one-shot sibling 和一个持久化 grandchild；它会固定触发器在一次陈旧的空目录响应后仍显示三个后代，并固定计时行以及聚合 `running` 状态转换，在不激活的情况下展开、打开持久化历史、准入一条用户 FIFO 后续消息、归并 child mux 事件，并证明 one-shot 历史仍然只读。 +- jsdom 测试固定后代聚合计数与活动状态、token 用量总计、精确到秒的运行中耗时与冻结后 inactive 耗时、目录缺失或为陈旧空目录时由摘要支撑的根操作、已知加载行的形态、混合 mode 行、点击前的叶子展开控件、diagnostic、后代懒加载展开、直接 parent 地址、键盘行为与两种只读原因。 +- 无密钥的组装 Web 快照包含一个具有持久化 token 用量的 inactive 可继续 child、一个 inactive 的 one-shot sibling 和一个持久化 grandchild；它会固定触发器在一次陈旧的空目录响应后仍显示三个后代，并固定 token 用量与计时行以及聚合 `running` 状态转换，在不激活的情况下展开、打开持久化历史、准入一条用户 FIFO 后续消息、归并 child mux 事件，并证明 one-shot 历史仍然只读。 - 导航测试固定仅含 subagent 的面包屑导航、从 subagent 创建 fork 时的 Workspace 归属，以及 `origin: 'subagent'` 侧边栏过滤，同时不隐藏普通 fork。 ## 后果 -- 目录读取可能重新扫描持久化谱系与每个直接候选的描述符日志，但可展开性只复用该追踪中已有的后代 header；Web 活动基线会为每个健康行增加一次 Agent 注册表查找，随后使用现有实时帧，而耗时会复用投影基线和推送，无需按行读取日志，成员刷新则保持去抖动和单次并发。 +- 目录读取可能重新扫描持久化谱系与每个直接候选的描述符日志，但可展开性只复用该追踪中已有的后代 header；Web 活动基线会为每个健康行增加一次 Agent 注册表查找，随后使用现有实时帧，而 token 用量与耗时会复用投影基线和推送，无需按行读取日志，成员刷新则保持去抖动和单次并发。 - parent 可用性、child 活动状态与 `hasChildren` 都是快照。列出之后，发布、dispose、其他发送方或其他进程都可能抢先改变状态；类型化提示词失败仍属预期行为。 - child 可能在历史获取与 mux 订阅之间发布，因此现有序号归并也涵盖从冷态转为存活的已寻址路径。 - 持久化 origin 会为 child header 与列表投影添加一个有意保持弱约束的产品分类字段；它不能变成授权捷径。 diff --git a/apps/web/tests/snapshots/subagent-conversation/tree.expected.md b/apps/web/tests/snapshots/subagent-conversation/tree.expected.md index c1174c042e..43ed15c649 100644 --- a/apps/web/tests/snapshots/subagent-conversation/tree.expected.md +++ b/apps/web/tests/snapshots/subagent-conversation/tree.expected.md @@ -1,8 +1,8 @@ - tree "Subagent sessions": - - treeitem "event-sourcing researcher Explain event sourcing in one · continuable · not running {{duration}}" [expanded] [level=1]: + - treeitem "event-sourcing researcher Explain event sourcing in one · continuable · not running 7.9K tok · {{duration}}" [expanded] [level=1]: - button "Collapse event-sourcing researcher descendants": - img - - text: event-sourcing researcher Explain event sourcing in one · continuable · not running {{duration}} + - text: event-sourcing researcher Explain event sourcing in one · continuable · not running 7.9K tok {{duration}} - group: - - treeitem "example editor continuable · not running {{duration}}" [level=2] - - treeitem "event-sourcing reviewer one-shot · not running {{duration}}" [level=1] + - treeitem "example editor continuable · not running 0 tok · {{duration}}" [level=2]: example editor continuable · not running 0 tok {{duration}} + - treeitem "event-sourcing reviewer one-shot · not running 0 tok · {{duration}}" [level=1]: event-sourcing reviewer one-shot · not running 0 tok {{duration}} diff --git a/docs/module-graph.md b/docs/module-graph.md index 313c58d914..8419646dca 100644 --- a/docs/module-graph.md +++ b/docs/module-graph.md @@ -983,6 +983,7 @@ flowchart TD pkg_client_ui_subagent --> pkg_client_ui_slots pkg_client_ui_subagent --> pkg_invariants pkg_client_ui_subagent --> pkg_subagent + pkg_client_ui_subagent --> pkg_token_meter pkg_agent_spine_demo --> pkg_agent pkg_agent_spine_demo --> pkg_agent_loop pkg_agent_spine_demo --> pkg_goal @@ -1240,7 +1241,7 @@ flowchart TD | [`client-ui-model`](../packages/client/ui-model) | `client` | [`client-connection`](../packages/client/connection), [`client-locale`](../packages/client/locale), [`client-runtime`](../packages/client/runtime), [`client-ui-command`](../packages/client/ui-command), [`client-ui-conversation`](../packages/client/ui-conversation), [`client-ui-primitives`](../packages/client/ui-primitives), [`client-ui-slash`](../packages/client/ui-slash), [`client-ui-slots`](../packages/client/ui-slots), [`invariants`](../packages/support/invariants) | | [`client-ui-permission`](../packages/client/ui-permission) | `client` | [`client-connection`](../packages/client/connection), [`client-locale`](../packages/client/locale), [`client-runtime`](../packages/client/runtime), [`client-schema-form`](../packages/client/schema-form), [`client-ui-command`](../packages/client/ui-command), [`client-ui-primitives`](../packages/client/ui-primitives), [`client-ui-slash`](../packages/client/ui-slash), [`client-ui-slots`](../packages/client/ui-slots), [`invariants`](../packages/support/invariants), [`permission`](../packages/ui/permission) | | [`client-ui-plan`](../packages/client/ui-plan) | `client` | [`client-connection`](../packages/client/connection), [`client-locale`](../packages/client/locale), [`client-runtime`](../packages/client/runtime), [`client-ui-conversation`](../packages/client/ui-conversation), [`client-ui-primitives`](../packages/client/ui-primitives), [`client-ui-slots`](../packages/client/ui-slots), [`invariants`](../packages/support/invariants), [`plan-mode`](../packages/plan/plan-mode) | -| [`client-ui-subagent`](../packages/client/ui-subagent) | `client` | [`client-locale`](../packages/client/locale), [`client-runtime`](../packages/client/runtime), [`client-ui-conversation`](../packages/client/ui-conversation), [`client-ui-primitives`](../packages/client/ui-primitives), [`client-ui-slash`](../packages/client/ui-slash), [`client-ui-slots`](../packages/client/ui-slots), [`invariants`](../packages/support/invariants), [`subagent`](../packages/subagent/subagent) | +| [`client-ui-subagent`](../packages/client/ui-subagent) | `client` | [`client-locale`](../packages/client/locale), [`client-runtime`](../packages/client/runtime), [`client-ui-conversation`](../packages/client/ui-conversation), [`client-ui-primitives`](../packages/client/ui-primitives), [`client-ui-slash`](../packages/client/ui-slash), [`client-ui-slots`](../packages/client/ui-slots), [`invariants`](../packages/support/invariants), [`subagent`](../packages/subagent/subagent), [`token-meter`](../packages/llm/token-meter) | | [`agent-spine-demo`](../packages/examples/agent-spine-demo) | `examples` | [`agent`](../packages/core/agent), [`agent-loop`](../packages/core/agent-loop), [`goal`](../packages/goal/goal), [`goal-session`](../packages/goal/goal-session), [`invariants`](../packages/support/invariants), [`llm`](../packages/llm/llm), [`llm-retry`](../packages/llm/llm-retry), [`paths`](../packages/util/paths), [`scope`](../packages/core/scope), [`session`](../packages/core/session), [`session-title`](../packages/session-title/session-title), [`skill`](../packages/skill/skill), [`skill-local`](../packages/skill/skill-local), [`system-prompt`](../packages/core/system-prompt), [`tasks-local`](../packages/tasks/tasks-local), [`tool-bash`](../packages/bash/tool-bash), [`tool-goal`](../packages/goal/tool-goal), [`tool-skill`](../packages/skill/tool-skill), [`tool-tasks`](../packages/tasks/tool-tasks), [`tools`](../packages/core/tools), [`workspace-context`](../packages/context/workspace-context) | | [`sdk-protocol`](../packages/sdk/sdk-protocol) | `sdk` | [`invariants`](../packages/support/invariants), [`llm`](../packages/llm/llm), [`session`](../packages/core/session), [`subagent`](../packages/subagent/subagent) | | [`tool-ralph`](../packages/workflow/tool-ralph) | `workflow` | [`agent`](../packages/core/agent), [`invariants`](../packages/support/invariants), [`llm`](../packages/llm/llm), [`subagent`](../packages/subagent/subagent), [`system-prompt`](../packages/core/system-prompt), [`tools`](../packages/core/tools), [`workflow`](../packages/workflow/workflow) | diff --git a/packages/client/ui-subagent/README.i18n.yaml b/packages/client/ui-subagent/README.i18n.yaml index 2b512252f0..76b8cb344e 100644 --- a/packages/client/ui-subagent/README.i18n.yaml +++ b/packages/client/ui-subagent/README.i18n.yaml @@ -2,5 +2,5 @@ # side as of the last confirmed-consistent state. Both languages carry equal authority; # after editing either side, bring the other along and re-record with: # pnpm run verify-translation-pairing --write packages/client/ui-subagent/README.md -README.md: 16a54484fb53544c71af0806c1497a18f9141002 -README.zh.md: 166a25d095b3e7f10f7239262f39e27972344529 +README.md: 538daeffb61f642b2e430cc1a6e2b1f3e3d5f55e +README.zh.md: 10320fdde6fe2a881e0e1dfc218c87562c4d61b0 diff --git a/packages/client/ui-subagent/README.md b/packages/client/ui-subagent/README.md index 16a54484fb..538daeffb6 100644 --- a/packages/client/ui-subagent/README.md +++ b/packages/client/ui-subagent/README.md @@ -4,7 +4,7 @@ English | [中文](README.zh.md) Web subagent feature owner: contributes the lazily expandable catalog tree to `conversation.session.header.actions`, reason-specific read-only replacements to the conversation composer chain, and the existing `@` reference source to `ctx.slash`. -The header action reads `subagentsByParent` and session summaries through the standard `useSessions` hook. After a non-empty direct catalog arrives, its trigger counts the complete subagent-only descendant lineage, stops at ordinary forks, and shows ongoing activity when any counted descendant is running. The compact tree remains direct-catalog authoritative: continuable and one-shot rows display mode plus `running`/`inactive` activity, an optional log-backed title, and active-turn duration to the second; an unlabeled one-shot row falls back to its session id, while corrupt, unsupported, or unavailable rows remain readable but disabled. Duration sums completed `subagentTiming` turns, advances once per second only for an open turn on a running child, and freezes after the child becomes inactive; an interrupted open turn is bounded by the session summary's last activity. Each healthy row's `hasChildren` hint determines disclosure before interaction, so known leaves never show an arrow; expanding a branch immediately reserves one disabled loading row per known direct descendant, then lazily replaces them with that child's authoritative catalog. Every visible branch is reported to the runtime so membership frames cause a debounced refresh only where the tree is being consumed. Selecting any depth calls `SessionsService.openSubagent()` with the row's exact `{parentSessionId, childSessionId, mode}` address. Component-local state owns tree visibility, expanded branches, keyboard focus, and the running-duration clock. ArrowRight/ArrowLeft expand and collapse branches; ArrowUp/ArrowDown, Home, End, and Escape navigate or close the tree; closing returns focus to the trigger. Styling uses tokens only. +The header action reads `subagentsByParent` and session summaries through the standard `useSessions` hook. After a non-empty direct catalog arrives, its trigger counts the complete subagent-only descendant lineage, stops at ordinary forks, and shows ongoing activity when any counted descendant is running. The compact tree remains direct-catalog authoritative: continuable and one-shot rows display mode plus `running`/`inactive` activity and an optional log-backed title, while the trailing column stacks total durable provider usage above active-turn duration to the second; an unlabeled one-shot row falls back to its session id, while corrupt, unsupported, or unavailable rows remain readable but disabled. Token totals sum the four disjoint `tokenUsage` buckets. Duration sums completed `subagentTiming` turns, advances once per second only for an open turn on a running child, and freezes after the child becomes inactive; an interrupted open turn is bounded by the session summary's last activity. Each healthy row's `hasChildren` hint determines disclosure before interaction, so known leaves never show an arrow; expanding a branch immediately reserves one disabled loading row per known direct descendant, then lazily replaces them with that child's authoritative catalog. Every visible branch is reported to the runtime so membership frames cause a debounced refresh only where the tree is being consumed. Selecting any depth calls `SessionsService.openSubagent()` with the row's exact `{parentSessionId, childSessionId, mode}` address. Component-local state owns tree visibility, expanded branches, keyboard focus, and the running-duration clock. ArrowRight/ArrowLeft expand and collapse branches; ArrowUp/ArrowDown, Home, End, and Escape navigate or close the tree; closing returns focus to the trigger. Styling uses tokens only. A one-shot child always elects a read-only composer that identifies the transcript as a completed execution record. A continuable child does so only when its exact parent is unavailable, with copy explaining the recovery path. A continuable child with a live parent keeps the ordinary input chrome, whose Session routes through `subagent.prompt`; running input remains Send because every follow-up joins the child's FIFO inbox, and addressed sessions never expose Stop. This package never receives host context or calls a model-facing tool. The catalog and composer behavior are specified by the [Web subagent conversations Agent Note](../../../.agents/notes/implemented/feature/2026-07-27-web-subagent-conversations.md). diff --git a/packages/client/ui-subagent/README.zh.md b/packages/client/ui-subagent/README.zh.md index 166a25d095..10320fdde6 100644 --- a/packages/client/ui-subagent/README.zh.md +++ b/packages/client/ui-subagent/README.zh.md @@ -4,7 +4,7 @@ Web subagent 功能 owner：向 `conversation.session.header.actions` 贡献可懒加载展开的目录树，向会话编辑器链贡献按原因区分的只读替代呈现，并保留注册到 `ctx.slash` 的既有 `@` 引用 source。 -页头操作通过标准 `useSessions` 钩子读取 `subagentsByParent` 与会话摘要。非空直接目录到达后，其触发器会统计仅含 subagent 的完整后代谱系，在普通 fork 处停止，并在任一计入统计的后代处于 `running` 时显示活动仍在进行。紧凑树仍以直接目录为权威依据：可继续和 one-shot 行会显示 mode、`running`／`inactive` 活动状态、由日志支撑的可选 title，以及精确到秒的活跃轮次耗时；没有 label 的 one-shot 行会回退到其会话 id，而损坏、不受支持或不可用的行仍保持可读但禁用。耗时会累加已完成的 `subagentTiming` 轮次，仅在运行中 child 存在未结束轮次时每秒递增一次，并在 child 变为 inactive 后冻结；被中断的未结束轮次以会话摘要中的最后活动为上界。每个健康行的 `hasChildren` 提示会在交互前决定是否显示展开控件，因此已知叶子节点从不显示箭头；展开分支时，会立即为每个已知直接后代预留一行禁用的加载行，随后再用该 child 的权威目录懒加载结果替换这些占位行。每个可见分支都会上报给运行时，使成员帧只在树正被消费的位置触发去抖动刷新。选择任意深度的条目都会使用该行的确切地址 `{parentSessionId, childSessionId, mode}` 调用 `SessionsService.openSubagent()`。组件局部状态负责树的可见性、已展开分支、键盘焦点与运行中耗时时钟。ArrowRight／ArrowLeft 展开和折叠分支；ArrowUp／ArrowDown、Home、End 与 Escape 用于导航或关闭树；关闭后焦点返回触发器。样式只使用 token。 +页头操作通过标准 `useSessions` 钩子读取 `subagentsByParent` 与会话摘要。非空直接目录到达后，其触发器会统计仅含 subagent 的完整后代谱系，在普通 fork 处停止，并在任一计入统计的后代处于 `running` 时显示活动仍在进行。紧凑树仍以直接目录为权威依据：可继续和 one-shot 行会显示 mode、`running`／`inactive` 活动状态和由日志支撑的可选 title，尾随列则在上行显示提供方的持久化 token 用量总计，在下行显示精确到秒的活跃轮次耗时；没有 label 的 one-shot 行会回退到其会话 id，而损坏、不受支持或不可用的行仍保持可读但禁用。token 用量总计为四个互不重叠的 `tokenUsage` 桶之和。耗时会累加已完成的 `subagentTiming` 轮次，仅在运行中 child 存在未结束轮次时每秒递增一次，并在 child 变为 inactive 后冻结；被中断的未结束轮次以会话摘要中的最后活动为上界。每个健康行的 `hasChildren` 提示会在交互前决定是否显示展开控件，因此已知叶子节点从不显示箭头；展开分支时，会立即为每个已知直接后代预留一行禁用的加载行，随后再用该 child 的权威目录懒加载结果替换这些占位行。每个可见分支都会上报给运行时，使成员帧只在树正被消费的位置触发去抖动刷新。选择任意深度的条目都会使用该行的确切地址 `{parentSessionId, childSessionId, mode}` 调用 `SessionsService.openSubagent()`。组件局部状态负责树的可见性、已展开分支、键盘焦点与运行中耗时时钟。ArrowRight／ArrowLeft 展开和折叠分支；ArrowUp／ArrowDown、Home、End 与 Escape 用于导航或关闭树；关闭后焦点返回触发器。样式只使用 token。 one-shot child 始终选用只读编辑器，并将 transcript（文本记录）说明为已完成的执行记录。可继续 child 仅在其确切 parent 不可用时选用只读编辑器，并以文案说明恢复路径。确切 parent 存活时，可继续 child 保留普通输入 chrome，其 Session 会通过 `subagent.prompt` 路由；child 运行期间，输入操作仍为 Send，因为每条后续消息都会进入 child 的 FIFO inbox，且已寻址会话绝不公开 Stop。本包绝不接收宿主 context，也不调用面向模型的工具。目录与编辑器行为由 [Web subagent 对话 Agent Note](../../../.agents/notes/implemented/feature/2026-07-27-web-subagent-conversations.md)规定。 diff --git a/packages/client/ui-subagent/package.json b/packages/client/ui-subagent/package.json index 573bf54097..9e6b120c1e 100644 --- a/packages/client/ui-subagent/package.json +++ b/packages/client/ui-subagent/package.json @@ -49,6 +49,7 @@ "@deepseek-ai/dsh-client-ui-slots": "^0.0.1", "@deepseek-ai/dsh-invariants": "^0.0.1", "@deepseek-ai/dsh-subagent": "^0.0.1", + "@deepseek-ai/dsh-token-meter": "^0.0.1", "cordis": "^4.0.0-rc.7" }, "devDependencies": { @@ -61,6 +62,7 @@ "@deepseek-ai/dsh-client-ui-slots": "workspace:^", "@deepseek-ai/dsh-invariants": "workspace:^", "@deepseek-ai/dsh-subagent": "workspace:^", + "@deepseek-ai/dsh-token-meter": "workspace:^", "@types/react": "~18.3.1", "cordis": "^4.0.0-rc.7" }, diff --git a/packages/client/ui-subagent/src/client/SubagentCatalogAction.module.css b/packages/client/ui-subagent/src/client/SubagentCatalogAction.module.css index d8642879ea..239081c59c 100644 --- a/packages/client/ui-subagent/src/client/SubagentCatalogAction.module.css +++ b/packages/client/ui-subagent/src/client/SubagentCatalogAction.module.css @@ -173,19 +173,30 @@ } .summary, -.time { +.metrics { color: var(--dsw-alias-label-tertiary); font-size: 11px; line-height: 16px; } -.time { +.metrics { + display: grid; + grid-template-rows: 18px 16px; flex: none; - margin-top: 16px; font-variant-numeric: tabular-nums; + text-align: right; white-space: nowrap; } +.metricToken { + grid-row: 1; + line-height: 18px; +} + +.metricDuration { + grid-row: 2; +} + .children { position: relative; margin-left: 18px; diff --git a/packages/client/ui-subagent/src/client/SubagentCatalogAction.tsx b/packages/client/ui-subagent/src/client/SubagentCatalogAction.tsx index d507b30019..b959786456 100644 --- a/packages/client/ui-subagent/src/client/SubagentCatalogAction.tsx +++ b/packages/client/ui-subagent/src/client/SubagentCatalogAction.tsx @@ -12,6 +12,7 @@ import type { PropsLocale, PropsRuntime, TranslateNS } from '@deepseek-ai/dsh-cl import { NS } from './locales.ts' import type {} from '@deepseek-ai/dsh-client-ui-conversation/client' import type {} from '@deepseek-ai/dsh-subagent/client' +import type {} from '@deepseek-ai/dsh-token-meter/client' import css from './SubagentCatalogAction.module.css' type CatalogEntry = SubagentCatalogSnapshot['entries'][number] @@ -59,6 +60,26 @@ function treeItems(root: HTMLDivElement | null): HTMLElement[] { : Array.from(root.querySelectorAll('[role="treeitem"]:not([aria-disabled="true"])')) } +/** Compact token count shared in shape with the conversation stats strip. */ +function formatTokens(value: number): string { + const scaled = (next: number): string => next >= 100 + ? String(Math.round(next)) + : String(Math.round(next * 10) / 10) + if (value < 1_000) return String(value) + if (value < 1_000_000) return `${scaled(value / 1_000)}K` + return `${scaled(value / 1_000_000)}M` +} + +/** Sum the four disjoint durable provider-usage buckets. */ +function tokenTotal( + usage: SessionProjectionMap['tokenUsage'] | undefined, +): number | undefined { + return usage === undefined + ? undefined + : usage.uncachedInputTokens + usage.outputTokens + + usage.cacheReadTokens + usage.cacheWriteTokens +} + /** Exact whole-second active-turn duration for one catalog row. */ function activityDuration( summary: SessionSummary | undefined, @@ -222,14 +243,21 @@ function CatalogRows({ const secondary = [summary?.title, mode, activity] .filter(value => value !== undefined) .join(' · ') + const totalTokens = tokenTotal(summary?.projectionValues?.tokenUsage) const durationMs = activityDuration( summary, entry.activity, now, ) - const duration = durationMs === undefined + const tokenMetric = totalTokens === undefined + ? undefined + : `${formatTokens(totalTokens)} tok` + const durationMetric = durationMs === undefined ? undefined : formatDuration(durationMs, t) + const metrics = [tokenMetric, durationMetric] + .filter(value => value !== undefined) + .join(' · ') const open = (): void => { openChild({ parentSessionId, childSessionId: entry.id, mode: entry.mode }) @@ -261,9 +289,7 @@ function CatalogRows({ role="treeitem" tabIndex={0} aria-level={level} - aria-label={[label, secondary, duration] - .filter(value => value !== undefined) - .join(' ')} + aria-label={[label, secondary, metrics].filter(value => value !== '').join(' ')} {...knownLeaf ? {} : { 'aria-expanded': isExpanded }} className={css.row} onClick={open} @@ -288,7 +314,12 @@ function CatalogRows({ {label} {secondary} - {duration !== undefined && {duration}} + {metrics !== '' && ( + + {tokenMetric !== undefined && {tokenMetric}} + {durationMetric !== undefined && {durationMetric}} + + )} {isExpanded && !knownLeaf && ( diff --git a/packages/client/ui-subagent/tests/conversation-ui.spec.tsx b/packages/client/ui-subagent/tests/conversation-ui.spec.tsx index 487cd14f35..0862135126 100644 --- a/packages/client/ui-subagent/tests/conversation-ui.spec.tsx +++ b/packages/client/ui-subagent/tests/conversation-ui.spec.tsx @@ -1,6 +1,6 @@ // @vitest-environment jsdom import { afterEach, describe, expect, it, vi } from 'vitest' -import { cleanup, fireEvent, render, screen } from '@testing-library/react' +import { cleanup, fireEvent, render, screen, within } from '@testing-library/react' import { makeTranslate } from '@deepseek-ai/dsh-client-test-runtime' import type { SessionId, SessionListState, SessionSummary, SubagentCatalogSnapshot, @@ -240,7 +240,7 @@ describe('SubagentCatalogAction', () => { }) }) - it('ticks active duration by seconds and freezes inactive rows', async () => { + it('shows durable token totals, ticks active duration by seconds, and freezes inactive rows', async () => { const now = 2_000_000_000_000 vi.useFakeTimers() vi.setSystemTime(now) @@ -249,6 +249,26 @@ describe('SubagentCatalogAction', () => { ['finished', 'inactive', 3_723_000, undefined, now - 60_000], ['interrupted', 'inactive', 2_000, now - 7_000, now - 3_000], ] as const + const usageById = { + running: { + uncachedInputTokens: 1_000, + outputTokens: 200, + cacheReadTokens: 3_000, + cacheWriteTokens: 400, + }, + finished: { + uncachedInputTokens: 123, + outputTokens: 0, + cacheReadTokens: 0, + cacheWriteTokens: 0, + }, + interrupted: { + uncachedInputTokens: 123_000_000, + outputTokens: 0, + cacheReadTokens: 0, + cacheWriteTokens: 0, + }, + } as const const entries = rows.map(([id, activity]) => ({ kind: 'child' as const, id: id as SessionId, @@ -269,6 +289,7 @@ describe('SubagentCatalogAction', () => { settledMs, ...(activeSince === undefined ? {} : { activeSince }), }, + tokenUsage: usageById[id], }, }] })) as Record @@ -276,14 +297,19 @@ describe('SubagentCatalogAction', () => { render() fireEvent.click(screen.getByRole('button', { name: /3 个子代理/ })) - expect(screen.getByRole('treeitem', { name: /running.*1分10秒/ })).toBeTruthy() - expect(screen.getByRole('treeitem', { name: /finished.*1小时02分03秒/ })).toBeTruthy() - expect(screen.getByRole('treeitem', { name: /interrupted.*6秒/ })).toBeTruthy() + const runningRow = screen.getByRole('treeitem', { name: /running.*4\.6K tok · 1分10秒/ }) + const runningMetrics = within(runningRow) + const tokenMetric = runningMetrics.getByText('4.6K tok') + const durationMetric = runningMetrics.getByText('1分10秒') + expect(tokenMetric.parentElement).toBe(durationMetric.parentElement) + expect(tokenMetric.nextElementSibling).toBe(durationMetric) + expect(screen.getByRole('treeitem', { name: /finished.*123 tok · 1小时02分03秒/ })).toBeTruthy() + expect(screen.getByRole('treeitem', { name: /interrupted.*123M tok · 6秒/ })).toBeTruthy() await vi.advanceTimersByTimeAsync(1_000) - expect(screen.getByRole('treeitem', { name: /running.*1分11秒/ })).toBeTruthy() - expect(screen.getByRole('treeitem', { name: /finished.*1小时02分03秒/ })).toBeTruthy() - expect(screen.getByRole('treeitem', { name: /interrupted.*6秒/ })).toBeTruthy() + expect(screen.getByRole('treeitem', { name: /running.*4\.6K tok · 1分11秒/ })).toBeTruthy() + expect(screen.getByRole('treeitem', { name: /finished.*123 tok · 1小时02分03秒/ })).toBeTruthy() + expect(screen.getByRole('treeitem', { name: /interrupted.*123M tok · 6秒/ })).toBeTruthy() }) it('lazily expands and collapses descendant catalogs with direct-parent navigation', () => { diff --git a/packages/client/ui-subagent/tsconfig.json b/packages/client/ui-subagent/tsconfig.json index e9d59a6fa1..27d06d16c0 100644 --- a/packages/client/ui-subagent/tsconfig.json +++ b/packages/client/ui-subagent/tsconfig.json @@ -29,6 +29,9 @@ { "path": "../ui-slots" }, + { + "path": "../../llm/token-meter" + }, { "path": "../../subagent/subagent" }, diff --git a/pnpm-lock.yaml b/pnpm-lock.yaml index 1b0e2f5542..0b634898ea 100644 --- a/pnpm-lock.yaml +++ b/pnpm-lock.yaml @@ -1906,6 +1906,9 @@ importers: '@deepseek-ai/dsh-subagent': specifier: workspace:^ version: link:../../subagent/subagent + '@deepseek-ai/dsh-token-meter': + specifier: workspace:^ + version: link:../../llm/token-meter '@types/react': specifier: ~18.3.1 version: 18.3.31 From 79072e356c3396886c6eb1e769b5ebb8fed721e8 Mon Sep 17 00:00:00 2001 From: creatixchu Date: Mon, 3 Aug 2026 11:24:35 +0800 Subject: [PATCH 20/48] fix(directory-picker-browse): advertise the path editor and walk the panes with the draft MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit The Select Workspace Directory dialog hid its one route into typing a path behind an invisible click target, and once the editor opened the panes stayed on whatever level was listed when it opened — so the typed text and the list under it disagreed for the whole edit. The edit zone now carries a pencil glyph at the bar's right edge and lights in the editor's own footprint on hover/focus (the bar keeps one height across the swap). While editing, the panes follow the draft: a directory part no pane lists is scanned after a 250ms rest and lands in place, so typing deeper descends and erasing segments steps back up without leaving the editor, and a final segment nobody matches releases the prefix filter instead of emptying the pane it is being spelled into. The draft-following scan is speculative and silent on failure; Enter still owns the view from submission until landing and remains the only path that surfaces an error. --- ...directory-picker-capability-seam.i18n.yaml | 4 +- ...-07-28-directory-picker-capability-seam.md | 4 + ...-28-directory-picker-capability-seam.zh.md | 4 + .../directory-browser.expected.md | 3 +- apps/web/tests/workspace-management.e2e.ts | 35 +++- .../directory-picker-browse/README.i18n.yaml | 4 +- .../host/directory-picker-browse/README.md | 2 +- .../host/directory-picker-browse/README.zh.md | 2 +- .../src/client/DirectoryBrowser.module.css | 46 ++++- .../src/client/DirectoryBrowser.tsx | 184 +++++++++++++++--- .../tests/directory-browser.spec.tsx | 102 +++++++++- 11 files changed, 345 insertions(+), 45 deletions(-) diff --git a/.agents/notes/implemented/architecture/2026-07-28-directory-picker-capability-seam.i18n.yaml b/.agents/notes/implemented/architecture/2026-07-28-directory-picker-capability-seam.i18n.yaml index 6855a0af2b..4342f49f2b 100644 --- a/.agents/notes/implemented/architecture/2026-07-28-directory-picker-capability-seam.i18n.yaml +++ b/.agents/notes/implemented/architecture/2026-07-28-directory-picker-capability-seam.i18n.yaml @@ -2,5 +2,5 @@ # side as of the last confirmed-consistent state. Both languages carry equal authority; # after editing either side, bring the other along and re-record with: # pnpm run verify-translation-pairing --write .agents/notes/implemented/architecture/2026-07-28-directory-picker-capability-seam.md -2026-07-28-directory-picker-capability-seam.md: 9884385cf9e0d51604bab9e4fd3c4bee77448331 -2026-07-28-directory-picker-capability-seam.zh.md: 8c229b9fb08d5052ba8a512f2153a89a9e5fd455 +2026-07-28-directory-picker-capability-seam.md: 90aa8bc7cfc0dc0fb3d057b9991682c9b531ea23 +2026-07-28-directory-picker-capability-seam.zh.md: 12917c95456bca9cdd5e20ae97847156af81277e diff --git a/.agents/notes/implemented/architecture/2026-07-28-directory-picker-capability-seam.md b/.agents/notes/implemented/architecture/2026-07-28-directory-picker-capability-seam.md index 9884385cf9..90aa8bc7cf 100644 --- a/.agents/notes/implemented/architecture/2026-07-28-directory-picker-capability-seam.md +++ b/.agents/notes/implemented/architecture/2026-07-28-directory-picker-capability-seam.md @@ -20,6 +20,7 @@ Placement and policy rulings folded into this decision: - **Dependency survey (hand-roll vs adopt).** Node's stdlib *is* the maintained cross-platform OS layer (`readdir(withFileTypes)`, `homedir`, path semantics); surveyed alternatives fail the dependency bar — file-manager packages (`node-file-manager`, `files-and-folders`, Syncfusion's provider) are whole HTTP apps (fit), drive-letter helpers (`drivelist` native addon, `windows-drive-letters` ~7y stale) fail health/proportionality. The browse backend is a thin adapter over stdlib. - **Hidden entries: return-and-flag.** The host stamps `hidden` (POSIX dot convention) and returns everything; the client filters. Display policy stays client-side, and the show-hidden toggle shipped as exactly that client-only change: a fixed-label footer toggle whose state lives in the pressed presentation (`aria-pressed` + check glyph), a dot-led path-draft prefix reveals the hidden entries it names, and the current selection is exempt from both the hidden and the prefix filter (it anchors the two-pane view). Windows' `FILE_ATTRIBUTE_HIDDEN` is not exposed by dirents — documented limitation until a native probe pays for itself. - **Path-editor cancel scope: the dialog card.** The browse client's path editor cancels on Escape and on focus leaving the card, both observed at a card-scope wrapper rather than the input — after Tab parks focus on a filtered row the input is off the event path, yet Escape must collapse the editor (not the dialog) and a later focus departure must still cancel. Non-cancel exemptions: window/tab focus loss, in-card focus moves, and pointer paths (rows and the toggle suppress focus steal on mousedown while editing). Separators for seeding and draft-tail filtering are inferred from `listing.home`; the wire-field alternative below records the deferred authoritative form. Combobox semantics between the editor and the list it filters (`aria-expanded`/`aria-controls`/active-descendant, result announcements) are likewise deferred — today they read to assistive tech as separate widgets. +- **The path editor advertises itself, and the panes follow the draft.** The click-to-edit zone is not invisible: a pencil glyph sits at the bar's right edge and hover/focus lights the zone in the editor's own footprint, so the one route into typing a path is discoverable and the bar does not resize when zone and input swap. While the editor is open the panes track the draft instead of whatever level happened to be listed when it opened — the final segment prefix-filters the level its directory part names, a tail nobody matches releases the filter (a name still being spelled must not empty the pane it is being spelled into), and a directory part no pane lists is scanned after a 250ms rest and lands single-wide in place, so typing deeper descends and erasing segments steps back up without leaving the editor. That scan is speculative — half-typed directories are unreadable most of the time — so a failure keeps the last readable panes and stays silent. Enter remains the authoritative commit: it owns the view from submission until landing (a debounce timer armed by the same keystrokes is held back rather than superseding the navigation, and a rejected submission stays held until the next edit) and it alone surfaces the failure. - **Navigation lands selection-anchored, quiet, and bounded.** Away from the display root (the same collapse the crumb header renders, so crumbs and pane shape never disagree), the landing is two-pane: the target's actual parent-level entry re-selected (platform case folding on Windows), its children on the right, so a crumb jump reads as stepping back one pane rather than collapsing to a single column. Target and parent legs land as **one frame** when the parent leg settles within the 200ms wait bound — the stale view keeps rendering until then, so navigation swaps the panes without an intermediate single-pane flash — and past the bound the target commits alone at once (an Enter-submitted navigation is never held hostage by a stalled parent) with the late parent leg upgrading the landing in place. The parent leg runs under the landing's supersession scope and is aborted on the wire by any newer intent (Escape inside the landing window therefore withdraws the whole navigation); a failed parent leg, or a truncated parent window lacking the target, leaves the single-pane landing — the upgrade must never orphan the selection it exists to anchor. The loading indicator follows the same quiet rule: it floats over the content's bottom-right corner (never a layout-shifting row; the truncated/error rows own the bottom left and keep rendering through a scan) and only once a scan outlives a 300ms silence window, so a local listing swaps with nothing shown at all. Row picks are deliberately exempt from the one-frame rule: a pick's immediate pane split is its selected-state feedback (aria-current, crumbs following), while a navigation has nothing to acknowledge the click but the swap itself. Both timing constants are calibrated for local enumeration; a remote deployment (one RPC per level, commonly 100–400ms) would sit inside the silence window with no pressed state on the crumbs — revisit the window or add pressed feedback when a remote consumer lands. - **Symlinks: follow for enterability.** `stat` probes symlinks (broken/cyclic → skipped); crumbs keep the logical path the operator navigated, and `workspace.create` already canonicalizes via realpath at adoption. - **Listing levels are bounded, and streamed.** One `list` call returns at most `maxEntries` rows (config, default 1000 — GitHub's web-UI directory-listing bound). The level streams via `opendir` into a name-sorted window of `maxEntries + 1` candidates, so memory stays O(maxEntries) and enterability probing touches only windowed candidates; the wire `DirectoryListing` carries a required `truncated` flag so the client states incompleteness instead of silently missing tail entries. A windowed broken symlink is not backfilled from beyond the window — the eviction already marks the level truncated. Window insertion is binary with an O(1) full-window tail rejection (an oversized level must not pay a window scan per dirent), and `list(path, signal)` threads the carrier's request signal so a scan of a stalled network directory cannot outlive a disconnected caller — every await in the scan (open, each read, each symlink probe) races the signal, an aborted exit abandons rather than awaits the close (Node queues close behind in-flight reads), and abandoned settlements are swallowed so cleanup can never surface as an unhandled rejection. An unbounded level is a memory/responsiveness hole for large or adversarial directories. @@ -34,6 +35,9 @@ Placement and policy rulings folded into this decision: - **Adopting a file-manager/drive-enumeration dependency.** Rejected per the survey above; recorded here as the dependency policy requires. - **A flip-label show-hidden toggle ("Hide hidden files").** Rejected: a flipping action label is ambiguous between state and action and doubles the negative; the fixed label with a pressed presentation states both at once. - **Pure relatedTarget blur cancellation (no mousedown suppression).** Rejected: Safari does not focus buttons on pointer down, so a click's focusout carries a null `relatedTarget` and would cancel the editor before the click lands; editing-scoped mousedown suppression plus the card-anchored relatedTarget guard covers pointer and keyboard paths together. +- **A permanently visible path input above the Miller view.** Rejected: the breadcrumb is already the "where am I" reading, and a second always-present field duplicates it while costing a row of a 500px card that the columns need. The glyph plus the hover-lit zone puts the affordance on the bar that already answers the question. +- **Scanning the draft on every keystroke, or only on Enter.** Per keystroke: walking one path segment issues a listing per character, most of them for directories the operator is typing through, not at. Only on Enter (what shipped first): the panes and the typed text disagreed for the whole edit — the complaint this bullet answers. The 250ms rest keeps one scan per directory the typing actually settles on. +- **Emptying a pane on a prefix miss (what shipped first).** Rejected: mid-name the miss is the normal state, so the pane blanked exactly while the operator needed it to confirm the name; releasing the filter keeps the level readable and costs only the transient wideness. - **A wire `separator` field on `DirectoryListing` (host stamps `path.sep`).** Deferred, not rejected: it is the authoritative form — a POSIX home directory containing a backslash defeats the `listing.home` heuristic — but it touches the seam type and every backend; the browse client's `separatorOf` carries a TODO pointing at this alternative until a wire change is next scheduled. ## Consequences diff --git a/.agents/notes/implemented/architecture/2026-07-28-directory-picker-capability-seam.zh.md b/.agents/notes/implemented/architecture/2026-07-28-directory-picker-capability-seam.zh.md index 8c229b9fb0..12917c9545 100644 --- a/.agents/notes/implemented/architecture/2026-07-28-directory-picker-capability-seam.zh.md +++ b/.agents/notes/implemented/architecture/2026-07-28-directory-picker-capability-seam.zh.md @@ -20,6 +20,7 @@ web GUI 的"打开本地文件夹"流程被焊死在一种交互上：`host.pick - **依赖调研（手写 vs 引入）。** Node 标准库本身就是维护中的跨平台 OS 层（`readdir(withFileTypes)`、`homedir`、路径语义）；调研过的替代品都过不了依赖门槛——文件管理器包（`node-file-manager`、`files-and-folders`、Syncfusion 的 provider）是整套 HTTP 应用（契合度不过），盘符工具（原生插件 `drivelist`、约七年未更的 `windows-drive-letters`）健康度／比例失当。browse 后端是标准库上的薄适配。 - **隐藏条目：返回并打标。** 宿主标注 `hidden`（POSIX 点前缀约定）并返回全部条目；客户端过滤。展示策略留在客户端，"显示隐藏"开关正是作为这一纯客户端改动落地：标签固定的 footer 开关，其状态由按下态呈现承载（`aria-pressed` + 勾选符号）；以点开头的路径草稿前缀会显出它所指名的隐藏条目；当前选中项则不受隐藏与前缀两种过滤影响（它锚定着双栏视图）。Windows 的 `FILE_ATTRIBUTE_HIDDEN` 不被 dirent 暴露——记为限制，直到原生探测值回其成本。 - **路径编辑器的取消范围：对话框卡片。** browse 客户端的路径编辑器在按 Escape 与焦点离开卡片时取消，两者都在卡片范围的包装层而非输入框上监听——Tab 把焦点停到某个过滤命中的行之后，输入框已不在事件路径上，但 Escape 仍须收起编辑器（而非对话框），其后的焦点离开也仍须取消。不取消的豁免：窗口／标签页失焦、卡片内焦点移动，以及指针路径（编辑期间行与开关在 mousedown 时抑制焦点夺取）。预填与草稿末段过滤所用的分隔符从 `listing.home` 推断；下文的线上字段替代方案记录了被延期的权威形态。编辑器与其过滤的列表之间的 combobox 语义（`aria-expanded`／`aria-controls`／active-descendant、结果播报）同样被延期——目前二者在辅助技术看来是彼此独立的控件。 +- **路径编辑器自我点明，各栏跟随草稿。** 点击即编辑的区域不再是隐形的：栏右端坐着一枚铅笔图标，悬停／聚焦时该区以编辑器自身的轮廓亮起，于是键入路径这唯一入口可被发现，且区域与输入框互换时栏高不变。编辑器打开期间，各栏跟随草稿，而不是停在它打开那一刻恰好列出的层级——末段对其目录部分所指的层级做前缀过滤，无一匹配的末段解除过滤（还在拼写中的名字不该把正在拼写它的那一栏清空），而任何一栏都未列出的目录部分会在停顿 250ms 后被扫描、以单宽栏就地落地，于是继续键入即下潜、删掉末段即上退，全程不必离开编辑器。该扫描是推测性的——键入到一半的目录多数时候读不出来——因此失败时保留最后一次可读的分栏并保持沉默。Enter 仍是权威提交：自提交至落地由它独占视图（同一批按键武装的防抖计时器会被扣住，而不是顶掉这次导航；提交被拒后仍扣住，直到下一次编辑），也只有它把失败呈现出来。 - **导航以选中项为锚、安静且有界地落地。** 在展示根之外（与 crumb 头部渲染的是同一塌缩，因此 crumb 与分栏形态永不相左），落地即双栏：重新选中目标在父层级中的实际条目（Windows 上按平台惯例折叠大小写），右侧展示其子项，因此 crumb 跳转读作后退一栏，而不是塌缩成单列。父层级这一程在 200ms 等待上限内落定时，目标与父层级两程以**同一帧**落地——在此之前陈旧视图持续渲染，导航换栏时因此没有中间的单栏闪现——超出该上限则目标即刻单独提交（Enter 提交的导航绝不会被滞塞的父层级扣作人质），迟到的父层级这一程再就地升级这次落地。父层级这一程在落地的 supersession 范围下运行，任何较新的意图都会在线上将其中止（因此在落地窗口内按 Escape 即撤回整次导航）；父层级这一程失败，或被截断的父窗口缺少目标时，都保留单栏落地——升级的存在正是为了锚定选中项，绝不能反而让它悬空。加载指示器遵循同一安静规则：它浮于内容右下角（绝不是会挪动布局的一行；截断／错误行占据左下角，并在扫描期间持续渲染），且仅在扫描超出 300ms 静默窗口后才出现，因此本地列举切换时什么也不显示。行选取被刻意豁免于同一帧规则：选取后立即分栏本身就是其选中态反馈（aria-current、crumb 跟随），而导航除了换栏本身没有任何东西可确认这次点击。两个时序常量都按本地列举校准；远程部署（每层级一次 RPC，通常 100–400ms）会落在静默窗口之内、crumb 上却没有按下态——待远程消费方落地时，重新审视该窗口或补上按下反馈。 - **符号链接：为可进入性而跟随。** 用 `stat` 探测符号链接（断链／循环→跳过）；面包屑保留操作者导航的逻辑路径，`workspace.create` 在接纳时本就做 realpath 规范化。 - **列举层级有上限，且流式处理。** 单次 `list` 至多返回 `maxEntries` 行（配置项，默认 1000——GitHub 网页端目录列举的同一上限）。层级经 `opendir` 流入一个按名排序、容量 `maxEntries + 1` 的候选窗口，内存保持 O(maxEntries)，可进入性探测只触及窗口内候选；线上 `DirectoryListing` 携带必填的 `truncated` 标志，让客户端明示不完整而不是静默缺尾。窗口内的断链符号链接不从窗口外回填——发生过驱逐本身已把层级标记为截断。窗口插入为二分查找、满窗尾部单次比较即拒绝（超大层级不能为每个 dirent 付出一次全窗扫描），且 `list(path, signal)` 透传载体的请求信号，滞塞网络目录的扫描不会在调用方断连后继续存活——扫描中的每个 await（打开、每次读取、每次符号链接探测）都与信号赛跑，中止路径放弃而非等待 close（Node 会把 close 排在在飞读取之后），被放弃的 settlement 全部吞掉，清理不会以未处理拒绝的形式冒出。无上限的层级对超大或恶意构造的目录就是内存／响应性漏洞。 @@ -34,6 +35,9 @@ web GUI 的"打开本地文件夹"流程被焊死在一种交互上：`host.pick - **引入文件管理器／盘符枚举依赖。** 按上文调研否决；依赖政策要求记录于此。 - **动作标签随状态翻转的"显示隐藏"开关（"隐藏隐藏文件"）。** 否决：会翻转的动作标签在状态与动作之间有歧义，还把否定叠了两层；固定标签加按下态呈现一次说清两者。 - **纯 relatedTarget 失焦取消（不做 mousedown 抑制）。** 否决：Safari 在指针按下时不给按钮聚焦，点击触发的 focusout 因而携带空 `relatedTarget`，会在点击落地前就取消编辑器；编辑期作用的 mousedown 抑制加上锚定卡片的 relatedTarget 守卫才能同时覆盖指针与键盘路径。 +- **在 Miller 视图上方常驻一个路径输入框。** 否决：面包屑本就在回答"我在哪儿"，再常驻一个字段是重复回答，还要从 500px 卡片里挪走一行——那是列需要的高度。图标加悬停亮起的区域，把这个入口放在了已经回答该问题的那一栏上。 +- **每敲一个键就扫描草稿，或只在 Enter 时扫描。** 每键扫描：走完一段路径就是每个字符一次列举，其中多数目录操作者只是路过而非停留。只在 Enter 时扫描（最初落地的行为）：整个编辑过程中各栏与所键入文本各说各话——正是本条所回应的抱怨。250ms 的停顿把扫描收敛为"键入真正停下来的每个目录一次"。 +- **前缀无一匹配时清空该栏（最初落地的行为）。** 否决：名字敲到一半时"无匹配"才是常态，于是恰恰在操作者需要它确认名字时把栏清空了；解除过滤保住了层级的可读性，代价只是短暂的宽松。 - **在 `DirectoryListing` 上增设线上 `separator` 字段（宿主标注 `path.sep`）。** 延期而非否决：它才是权威形态——含反斜杠的 POSIX 家目录会击穿 `listing.home` 启发式——但它触及 seam 类型与每个后端；browse 客户端的 `separatorOf` 挂着指向本方案的 TODO，直到下次安排线上变更。 ## 后果 diff --git a/apps/web/tests/snapshots/workspace-management/directory-browser.expected.md b/apps/web/tests/snapshots/workspace-management/directory-browser.expected.md index baaaa6f3dc..47957dab82 100644 --- a/apps/web/tests/snapshots/workspace-management/directory-browser.expected.md +++ b/apps/web/tests/snapshots/workspace-management/directory-browser.expected.md @@ -4,7 +4,8 @@ - button "Home" - img - button "browse-golden" - - button "Edit path" + - button "Edit path": + - img - list: - listitem: - button "adopted": diff --git a/apps/web/tests/workspace-management.e2e.ts b/apps/web/tests/workspace-management.e2e.ts index 1ffcf6f490..1a5531f59a 100644 --- a/apps/web/tests/workspace-management.e2e.ts +++ b/apps/web/tests/workspace-management.e2e.ts @@ -1,6 +1,7 @@ // Web e2e scenarios: workspace management — adding a workspace through the // composed directory dialog (its own New folder affordance is the product's -// one creation route), same-basename directory adoption, the rename round +// one creation route), the dialog's path editor walking the panes with the +// typed draft, same-basename directory adoption, the rename round // trip over the real wire (workspace.rename RPC + durable registry), the // duplicate-name pre-check, the // flat "In one list" view with its persisted group-by preference, the session @@ -12,7 +13,7 @@ // seeded-history seed reused verbatim — no new recording). import { mkdir, readFile, stat, writeFile } from 'node:fs/promises' import { fileURLToPath } from 'node:url' -import { join } from 'node:path' +import { join, sep } from 'node:path' import type { Browser, Locator, Page } from 'playwright' import { chromium } from 'playwright' import { afterAll, beforeAll, describe, expect, it, onTestFailed } from 'vitest' @@ -403,6 +404,36 @@ describe('web e2e: workspace management (create / rename / flat view / hover aff expect(tripwire.pageErrors).toEqual([]) }, 60_000) + it('walks the panes with the typed path: deeper past a separator, back up on erase, whole on a miss', async () => { + // The panes must track the draft without leaving the editor, so the + // typed text and what is listed under it never disagree. + const staged = join(scaffold.workspaceCwd, 'browse-golden') + await mkdir(join(staged, 'alpha', 'only-under-alpha'), { recursive: true }) + const dialog = await browseTo(staged) + await expect.poll(() => dialog.getByText('alpha', { exact: true }).count(), { timeout: 10_000 }).toBe(1) + await dialog.getByRole('button', { name: 'Edit path' }).click() + const path = dialog.getByLabel('Edit path') + // A directory part no pane lists: the panes follow it and keep the editor. + await path.fill(`${join(staged, 'alpha')}${sep}`) + await expect.poll(() => dialog.getByText('only-under-alpha', { exact: true }).count(), { timeout: 10_000 }).toBe(1) + // The editor is still up with the draft intact: the panes moved under it. + expect(await path.inputValue()).toBe(`${join(staged, 'alpha')}${sep}`) + // Erasing back past the separator steps the panes up, the tail filtering + // the level it returns to. + await path.fill(`${staged}${sep}al`) + await expect.poll(() => dialog.getByText('alpha', { exact: true }).count(), { timeout: 10_000 }).toBe(1) + expect(await dialog.getByText('beta', { exact: true }).count()).toBe(0) + expect(await dialog.getByText('only-under-alpha', { exact: true }).count()).toBe(0) + // A tail nobody matches is a name still being spelled: the level shows + // whole instead of emptying under it. + await path.fill(`${staged}${sep}zzz`) + await expect.poll(() => dialog.getByText('beta', { exact: true }).count(), { timeout: 10_000 }).toBe(1) + expect(await dialog.getByText('alpha', { exact: true }).count()).toBe(1) + await dialog.getByRole('button', { name: 'Cancel' }).click() + await dialog.waitFor({ state: 'hidden', timeout: 10_000 }) + expect(tripwire.pageErrors).toEqual([]) + }, 60_000) + /** * Expand Ungrouped and return its seeded session row. The only visible child * is the non-blank persisted Session; the blank Session created while diff --git a/packages/host/directory-picker-browse/README.i18n.yaml b/packages/host/directory-picker-browse/README.i18n.yaml index 673d053e3a..4063c7d692 100644 --- a/packages/host/directory-picker-browse/README.i18n.yaml +++ b/packages/host/directory-picker-browse/README.i18n.yaml @@ -2,5 +2,5 @@ # side as of the last confirmed-consistent state. Both languages carry equal authority; # after editing either side, bring the other along and re-record with: # pnpm run verify-translation-pairing --write packages/host/directory-picker-browse/README.md -README.md: 52b5fe7e89f915be3b50324628e9d5c48f1ef94c -README.zh.md: 742da39470083887a71ddba4a7c8012f0ce0ea1f +README.md: 7cb0ec785766e954ff4bb39df6825ee7e8c9d821 +README.zh.md: ec71a90bbcd004ec9f9c0d8a9a236882a7487b73 diff --git a/packages/host/directory-picker-browse/README.md b/packages/host/directory-picker-browse/README.md index 52b5fe7e89..7cb0ec7857 100644 --- a/packages/host/directory-picker-browse/README.md +++ b/packages/host/directory-picker-browse/README.md @@ -6,7 +6,7 @@ The **in-app browsing backend** of the [directory-picker seam](../directory-pick Behavior facts: listings return **directories only**, name-sorted, with symlinks-to-directories followed (broken/cyclic links skipped — the probe `stat` failing means "not enterable") and a host-owned `hidden` flag (POSIX dot convention) left for the client to act on; `crumbs` is the root-to-target ancestor chain, the root crumb labeled by its full path (`/`, `C:\`); an absent `list` path means the host account's home directory. `createDirectory` is non-recursive (a missing parent is a real failure, not a level to invent) and validates the name as a single non-blank segment even when called directly, mirroring the wire schema's fence. Both primitives reject an explicit path that is not fully qualified — relative forms, and on Windows the rooted drive-less forms (`\foo`, `/foo`) and incomplete UNC prefixes (`\\`, `\\server`) that `isAbsolute` accepts — with `directory-unreadable`/`directory-create-failed`, instead of letting `resolve` rebase it under the host process cwd or current drive. One `list` call returns at most `maxEntries` rows (config, default 1000 — the bound GitHub's web UI applies to directory listings), and the level streams through a bounded window so memory stays O(maxEntries) no matter how many children the directory holds: a cut level keeps the name-sorted head, counts hidden rows against the bound, probes only windowed candidates, and reports `truncated: true` so the client can say the level is incomplete (a windowed broken symlink is not backfilled from beyond the window — the eviction already marks the level truncated); window insertion is binary with an O(1) full-window tail rejection, and `list` threads the caller's `AbortSignal` so a disconnect or timeout stops the scan instead of letting it outlive the caller. Failures throw the seam's typed `DirectoryPickerError`. Policy rationale: [the directory-picker capability seam Agent Note](../../../.agents/notes/implemented/architecture/2026-07-28-directory-picker-capability-seam.md). -**Dual-face package**: the browser half (`./client`) fills [ui-workspace's](../../client/ui-workspace/README.md) two directory-flow holes with the in-app **Select Workspace Directory** dialog (figma `Harness` 813-23126 family — Miller two-column view whose navigations land selection-anchored and quiet: the previous view keeps rendering while a crumb jump or a submitted path is scanned (a "Loading…" pill floats over it only once the scan outlives a 300ms silence window, never shifting the columns), then target and parent legs land as one two-pane frame with the target re-selected as its actual parent-level entry — so stepping back never collapses and no intermediate frame flashes (a parent leg outliving its 200ms wait bound lands the target alone and upgrades in place; a failed or truncated parent leg keeps the single-pane landing; the display root keeps the single wide level); breadcrumb with a click-to-edit path zone whose editor seeds a trailing separator, prefix-filters the listed level from the draft's final segment while typing (case-insensitively, over the listed — possibly truncated — rows only; Enter still navigates by the exact text), and cancels on Escape or when focus leaves the dialog card (window/tab switches and in-card focus moves keep the draft); a fixed-label show-hidden footer toggle over the host's `hidden` flags, with a dot-led typed prefix revealing its matches and the current selection exempt from both filters; nested New-folder dialog), driving `host.listDirectory`/`host.createDirectory` and registering its own locale namespace (`directory-browser`, zh default / en). One cordis.yml row therefore composes both sides of the browse interaction; the client carries no capability-kind branching, and mounting a second flow package fails at load (the holes are `single` kind). +**Dual-face package**: the browser half (`./client`) fills [ui-workspace's](../../client/ui-workspace/README.md) two directory-flow holes with the in-app **Select Workspace Directory** dialog (figma `Harness` 813-23126 family — Miller two-column view whose navigations land selection-anchored and quiet: the previous view keeps rendering while a crumb jump or a submitted path is scanned (a "Loading…" pill floats over it only once the scan outlives a 300ms silence window, never shifting the columns), then target and parent legs land as one two-pane frame with the target re-selected as its actual parent-level entry — so stepping back never collapses and no intermediate frame flashes (a parent leg outliving its 200ms wait bound lands the target alone and upgrades in place; a failed or truncated parent leg keeps the single-pane landing; the display root keeps the single wide level); breadcrumb with a click-to-edit path zone, advertised by the pencil glyph at the bar's right edge and lit on hover in the editor's own footprint, whose editor seeds a trailing separator and then keeps the panes under the draft: the final segment prefix-filters the level its directory part names (case-insensitively, over the listed — possibly truncated — rows only; a tail nobody matches releases the filter instead of emptying the pane), while a directory part no pane lists is scanned after a 250ms rest and shown in place, so typing deeper descends and erasing segments steps back up without leaving the editor — a speculative scan is silent when it fails, and Enter still navigates by the exact text, owning the view until it lands; the editor cancels on Escape or when focus leaves the dialog card (window/tab switches and in-card focus moves keep the draft); a fixed-label show-hidden footer toggle over the host's `hidden` flags, with a dot-led typed prefix revealing its matches and the current selection exempt from every filter; nested New-folder dialog), driving `host.listDirectory`/`host.createDirectory` and registering its own locale namespace (`directory-browser`, zh default / en). One cordis.yml row therefore composes both sides of the browse interaction; the client carries no capability-kind branching, and mounting a second flow package fails at load (the holes are `single` kind). ## Model Experience diff --git a/packages/host/directory-picker-browse/README.zh.md b/packages/host/directory-picker-browse/README.zh.md index 742da39470..ec71a90bbc 100644 --- a/packages/host/directory-picker-browse/README.zh.md +++ b/packages/host/directory-picker-browse/README.zh.md @@ -6,7 +6,7 @@ 行为事实：列举**只返回目录**、按名称排序，指向目录的符号链接会被跟随（断链／循环链接被跳过——探测 `stat` 失败即"不可进入"），并携带宿主判定的 `hidden` 标志（POSIX 点前缀约定），展示决策留给客户端；`crumbs` 是从根到目标的祖先链，根 crumb 以完整路径标注（`/`、`C:\`）；`list` 不带路径即列举宿主账户的家目录。`createDirectory` 不递归（父目录缺失是真实失败，不是要补造的层级），且即便被直接调用也把名称校验为单个非空段，与协议 schema 的栅栏一致。两个原语都拒绝非完全限定的显式路径——相对形态，以及 Windows 上 `isAbsolute` 会放行的无盘符有根形态（`\foo`、`/foo`）与不完整的 UNC 前缀（`\\`、`\\server`）——报 `directory-unreadable`／`directory-create-failed`，而不是任由 `resolve` 把它重定位到宿主进程 cwd 或当前盘符之下。单次 `list` 至多返回 `maxEntries` 行（配置项，默认 1000——GitHub 网页端对目录列举采用的同一上限），且层级以流式方式经过一个有界窗口，无论目录有多少子项内存都保持 O(maxEntries)：被截断的层级保留按名排序的头部、隐藏行计入上限、只探测窗口内候选，并报告 `truncated: true`，供客户端提示层级不完整（窗口内的断链符号链接不会从窗口外回填——发生过驱逐本身已把层级标记为截断）；窗口插入为二分查找、满窗尾部单次比较即拒绝，且 `list` 透传调用方的 `AbortSignal`，断连或超时会停止扫描而不是让它在调用方离开后继续。失败抛出 seam 的类型化 `DirectoryPickerError`。策略依据：[目录选择能力 seam Agent Note](../../../.agents/notes/implemented/architecture/2026-07-28-directory-picker-capability-seam.md)。 -**双面包**：browser half（`./client`）以应用内 **选择工作区目录** 对话框（figma `Harness` 813-23126 家族——Miller 双列视图，其导航以选中项为锚、安静落地：扫描 crumb 跳转或提交的路径期间，先前视图持续渲染（"Loading…" 胶囊仅在扫描超出 300ms 静默窗口后才浮于其上，绝不挪动各列），随后目标与父层级两程以单个双栏帧落地，目标被重新选中为其在父层级中的实际条目——因此后退绝不塌缩，也没有中间帧闪现（父层级这一程超出其 200ms 等待上限时，目标单独落地，随后就地升级；父层级这一程失败或被截断时保持单栏落地；展示根保持单个宽层级）；带点击即编辑路径区的面包屑，其编辑器预填尾随分隔符、输入时以草稿末段对所列层级做前缀过滤（不区分大小写，且仅作用于已列出、可能被截断的行；Enter 仍按确切文本导航）、按 Escape 或焦点离开对话框卡片即取消（窗口／标签页切换与卡片内焦点移动保留草稿）；基于宿主 `hidden` 标志、标签固定的"显示隐藏"footer 开关，键入以点开头的前缀会显出其匹配项，且当前选中项不受这两种过滤影响；嵌套新建文件夹对话框）填入 [ui-workspace](../../client/ui-workspace/README.md) 的两个目录流洞，驱动 `host.listDirectory`／`host.createDirectory`，并注册自己的 locale 命名空间（`directory-browser`，zh 默认／en）。因此一行 cordis.yml 同时组合浏览交互的两侧；client 侧不含任何能力 kind 分支，挂载第二个流程包会在加载期失败（洞为 `single` kind）。 +**双面包**：browser half（`./client`）以应用内 **选择工作区目录** 对话框（figma `Harness` 813-23126 家族——Miller 双列视图，其导航以选中项为锚、安静落地：扫描 crumb 跳转或提交的路径期间，先前视图持续渲染（"Loading…" 胶囊仅在扫描超出 300ms 静默窗口后才浮于其上，绝不挪动各列），随后目标与父层级两程以单个双栏帧落地，目标被重新选中为其在父层级中的实际条目——因此后退绝不塌缩，也没有中间帧闪现（父层级这一程超出其 200ms 等待上限时，目标单独落地，随后就地升级；父层级这一程失败或被截断时保持单栏落地；展示根保持单个宽层级）；带点击即编辑路径区的面包屑，该区由栏右端的铅笔图标点明、悬停时以编辑器自身的轮廓亮起，其编辑器预填尾随分隔符，随后让下方各栏跟随草稿：末段对其目录部分所指的层级做前缀过滤（不区分大小写，且仅作用于已列出、可能被截断的行；无一匹配的末段会解除过滤，而不是把该栏清空），而任何一栏都未列出的目录部分会在停顿 250ms 后被扫描并就地展示，于是继续键入即下潜、删掉末段即上退，全程不必离开编辑器——推测性扫描失败时保持沉默，而 Enter 仍按确切文本导航，并在落地前独占视图；编辑器按 Escape 或焦点离开对话框卡片即取消（窗口／标签页切换与卡片内焦点移动保留草稿）；基于宿主 `hidden` 标志、标签固定的"显示隐藏"footer 开关，键入以点开头的前缀会显出其匹配项，且当前选中项不受任何过滤影响；嵌套新建文件夹对话框）填入 [ui-workspace](../../client/ui-workspace/README.md) 的两个目录流洞，驱动 `host.listDirectory`／`host.createDirectory`，并注册自己的 locale 命名空间（`directory-browser`，zh 默认／en）。因此一行 cordis.yml 同时组合浏览交互的两侧；client 侧不含任何能力 kind 分支，挂载第二个流程包会在加载期失败（洞为 `single` kind）。 ## 模型体验 diff --git a/packages/host/directory-picker-browse/src/client/DirectoryBrowser.module.css b/packages/host/directory-picker-browse/src/client/DirectoryBrowser.module.css index 2f207e4195..eaad4d46d8 100644 --- a/packages/host/directory-picker-browse/src/client/DirectoryBrowser.module.css +++ b/packages/host/directory-picker-browse/src/client/DirectoryBrowser.module.css @@ -53,7 +53,8 @@ display: flex; align-items: center; gap: 4px; - min-height: 20px; + /* The path editor's height: crumb mode and edit mode occupy the same bar. */ + min-height: 24px; } /* Deep chains scroll inside the trail (the effect pins the tail into view) @@ -118,17 +119,52 @@ color: var(--dsw-alias-label-tertiary); } -/* The empty remainder of the bar: invisible, but a real click target that - * flips the bar into path-edit mode. */ +/* The empty remainder of the bar: a real click target that flips the bar + * into path-edit mode. The zone itself stays flush with the crumbs; the + * pencil glyph seated at its right edge is the standing affordance, and + * hover/focus lights the zone in the editor's own rounded shape so the + * gesture reads before the click. */ .crumbEditZone { + display: flex; + align-items: center; + justify-content: flex-end; flex: 1 0 34px; min-width: 34px; - align-self: stretch; - border: none; + /* The editor's own height, so hover previews the input's exact footprint + * and the bar does not resize when the two swap. */ + height: 24px; + padding: 0 6px; + border: 1px solid transparent; + border-radius: 8px; background: transparent; cursor: text; } +.crumbEditZone:hover, +.crumbEditZone:focus-visible { + border-color: var(--dsw-alias-border-l2); + outline: none; +} + +.crumbEditGlyph { + flex: none; + color: var(--dsw-alias-label-tertiary); +} + +.crumbEditZone:hover .crumbEditGlyph, +.crumbEditZone:focus-visible .crumbEditGlyph { + color: var(--dsw-alias-label-primary); +} + +.crumbEditZone:disabled { + border-color: transparent; + cursor: default; +} + +.crumbEditZone:disabled .crumbEditGlyph { + color: var(--dsw-alias-label-caption); +} + .pathInput { box-sizing: border-box; flex: 1 1 0; diff --git a/packages/host/directory-picker-browse/src/client/DirectoryBrowser.tsx b/packages/host/directory-picker-browse/src/client/DirectoryBrowser.tsx index f5510fda7c..1f9903952d 100644 --- a/packages/host/directory-picker-browse/src/client/DirectoryBrowser.tsx +++ b/packages/host/directory-picker-browse/src/client/DirectoryBrowser.tsx @@ -18,15 +18,20 @@ * owning flow decides what "Open" means and owns the workspace-creation * error surface. Hidden entries are host-flagged and hidden by default; the * footer's fixed-label "Show hidden files" toggle (aria-pressed, check when - * on) reveals them (client-side only). The path editor opens seeded with a - * trailing separator, and while the draft's directory part names a listed - * level, its final segment prefix-filters that level's rows (a dot-led - * prefix also reveals the hidden entries it names). + * on) reveals them (client-side only). The path editor announces itself with + * a pencil glyph and a hover-lit zone, opens seeded with a trailing + * separator, and keeps the panes under the draft: the final segment + * prefix-filters the level its directory part names (a dot-led prefix also + * reveals the hidden entries it names, and a prefix nobody matches releases + * the filter), while a directory part no pane lists is scanned after a short + * debounce and shown in place — so typing deeper descends and erasing + * segments steps back up without leaving the editor. */ import { useCallback, useEffect, useRef, useState } from 'react' import clsx from 'clsx' import { - Button, IconCheckOutline16, IconChevronRightOutline14, IconFolderClose16, IconFolderOpen16, IconPlusOutline16, Modal, + Button, IconCheckOutline16, IconChevronRightOutline14, IconEditOutline16, IconFolderClose16, IconFolderOpen16, + IconPlusOutline16, Modal, } from '@deepseek-ai/dsh-client-ui-primitives' import type { DirectoryEntry, DirectoryListing } from '@deepseek-ai/dsh-client-runtime/client' import { DirectoryBrowseError } from '@deepseek-ai/dsh-client-runtime/client' @@ -75,6 +80,15 @@ const SLOW_SCAN_DELAY_MS = 300 */ const PARENT_LEG_WAIT_MS = 200 +/** + * How long a typed draft rests before the panes follow it to a directory no + * pane lists. The window absorbs the keystrokes that walk through + * intermediate directory parts (every character of `/usr/lo` past the + * separator would otherwise be its own scan) while staying short enough that + * a pause reads as "the list moved with me". + */ +const DRAFT_PREVIEW_DEBOUNCE_MS = 250 + /** * Breadcrumb rows for display: inside the home subtree the chain starts at a * localized Home crumb; outside it the full ancestry shows, the root labeled @@ -100,21 +114,84 @@ function separatorOf(listing: DirectoryListing): '\\' | '/' { return listing.home.includes('\\') ? '\\' : '/' } +/** The listed level as a directory part: its own path, separator-terminated (the root already is). */ +function levelDirectory(listing: DirectoryListing): string { + const sep = separatorOf(listing) + return listing.path.endsWith(sep) ? listing.path : `${listing.path}${sep}` +} + +/** + * The draft's directory part — everything through its last separator — or + * null while no separator has been typed at all (nothing addresses a + * directory yet). The platform separator comes from `listing`, so the caller + * passes any listing of the host's filesystem. + */ +function draftDirectory(listing: DirectoryListing, draft: string): string | null { + const cut = draft.lastIndexOf(separatorOf(listing)) + return cut === -1 ? null : draft.slice(0, cut + 1) +} + /** * The path draft's final segment, when its directory part is exactly the * level `listing` lists — the segment the level prefix-filters on while the * user types. Any other draft (no separator yet, or naming some other * directory) leaves the level unfiltered. The directory part compares - * exactly (it is the host's own path text, reached by seeding or erasing); - * only the name filter downstream is case-insensitive. + * exactly (it is the host's own path text, reached by seeding, erasing, or a + * draft-following scan); only the name filter downstream is case-insensitive. */ function draftPrefixFor(listing: DirectoryListing, draft: string | null): string | null { if (draft === null) return null - const sep = separatorOf(listing) - const cut = draft.lastIndexOf(sep) - if (cut === -1) return null - const level = listing.path.endsWith(sep) ? listing.path : `${listing.path}${sep}` - return draft.slice(0, cut + 1) === level ? draft.slice(cut + 1) : null + const directory = draftDirectory(listing, draft) + if (directory === null) return null + return directory === levelDirectory(listing) ? draft.slice(directory.length) : null +} + +/** + * The directory a draft addresses that no rendered pane lists — the level the + * editor must scan for the panes to keep following the typed path. Null when + * a pane already lists it (the prefix filter alone answers the draft), when + * no separator has been typed yet, and when no level is listed at all: the + * platform separator is read off a listing, so the editor's + * failed-home-listing recovery path types blind until Enter. + */ +function pendingPreviewDirectory( + parent: DirectoryListing | null, + child: DirectoryListing | null, + draft: string | null, +): string | null { + if (parent === null || draft === null) return null + const directory = draftDirectory(parent, draft) + if (directory === null || directory === levelDirectory(parent)) return null + if (child !== null && directory === levelDirectory(child)) return null + return directory +} + +/** + * The rows one column renders. The selection is exempt from every filter: it + * anchors the two-pane view (crumbs and the child pane point at it), so + * neither the hidden filter after a dot-reveal pick nor a prefix miss may + * orphan it. A prefix narrows the level only while some row matches it — a + * tail nobody matches is a name being spelled, not a demand for an empty + * pane, so the level shows whole (and its hidden rows return to obeying the + * toggle, the dot-led reveal included). + */ +function visibleEntries( + entries: readonly DirectoryEntry[], + selectedPath: string | null, + showHidden: boolean, + filterPrefix: string | null, +): readonly DirectoryEntry[] { + const needle = filterPrefix === null ? '' : filterPrefix.toLowerCase() + const matches = (entry: DirectoryEntry): boolean => entry.name.toLowerCase().startsWith(needle) + const narrowing = needle !== '' && entries.some(matches) + // A dot-led prefix names hidden entries explicitly, so matching ones + // surface even while the toggle keeps the rest hidden. + const revealHidden = narrowing && needle.startsWith('.') + return entries.filter((entry) => { + if (entry.path === selectedPath) return true + if (narrowing && !matches(entry)) return false + return showHidden || !entry.hidden || revealHidden + }) } /** One column of folder rows (the Miller view renders one or two of these). */ @@ -127,16 +204,7 @@ function LevelColumn({ entries, selectedPath, busy, onPick, showHidden, filterPr filterPrefix: string | null pathEditing: boolean }) { - const visible = entries.filter((entry) => { - // The selection is exempt from both filters: it anchors the two-pane - // view (crumbs and the child pane point at it), so neither the hidden - // filter after a dot-reveal pick nor a prefix miss may orphan it. - if (entry.path === selectedPath) return true - if (filterPrefix !== null && !entry.name.toLowerCase().startsWith(filterPrefix.toLowerCase())) return false - // A dot-led prefix names hidden entries explicitly, so matching ones - // surface even while the toggle keeps the rest hidden. - return showHidden || !entry.hidden || filterPrefix?.startsWith('.') === true - }) + const visible = visibleEntries(entries, selectedPath, showHidden, filterPrefix) return (
{visible.map((entry) => { @@ -381,6 +449,41 @@ export function DirectoryBrowser({ open, listDirectory, createDirectory, onOpen, }) }, [launchListing, pathDraft]) + /** + * Enter owns the view from submission until its navigation lands, so the + * debounce timer the same keystrokes armed must not supersede it. Cleared + * by the next edit (and by opening the editor); a failed submission leaves + * it set, so the rejected path is not immediately re-scanned as a preview. + */ + const previewSuspended = useRef(false) + + /** + * List the directory the draft addresses and show it WITHOUT closing the + * editor: the level replaces the panes single-wide (the selection and its + * child preview belonged to the level the draft left), and the draft's + * final segment prefix-filters it from the next render on. Unlike Enter, + * this is speculative — half-typed directories are unreadable most of the + * time — so a failure keeps the last readable panes and stays silent, + * leaving submission to surface the real error. A landing clears a stale + * error for the same reason: it, not the launch, is what makes the message + * obsolete. + */ + const previewDraftLevel = useCallback((directory: string) => { + const { seq, scan } = launchListing(directory) + setLoading(true) + scan.then((level) => { + if (seq !== requestSeq.current) return + setParent(level) + setSelected(null) + setChild(null) + setLoading(false) + setError(null) + }, () => { + if (seq !== requestSeq.current) return + setLoading(false) + }) + }, [launchListing]) + /** Abandon path editing (Escape or clicking away) and restore the crumb view. */ const cancelPathEdit = useCallback(() => { // Cancel also withdraws a navigation the editor already launched: its @@ -499,6 +602,22 @@ export function DirectoryBrowser({ open, listDirectory, createDirectory, onOpen, return () => { window.clearTimeout(timer) } }, [loading, scanWindow]) + // The panes follow the draft: a directory part no pane lists is scanned + // once the typing rests. The dependency is the target STRING, so the + // landing it commits cannot re-arm the timer (a host that answers with a + // differently spelled path leaves the target unchanged, hence unrepeated), + // and every further keystroke replaces the pending timer instead of + // queueing another scan. + const previewDirectory = pendingPreviewDirectory(parent, child, pathDraft) + useEffect(() => { + if (previewDirectory === null) return + const timer = window.setTimeout(() => { + if (previewSuspended.current) return + previewDraftLevel(previewDirectory) + }, DRAFT_PREVIEW_DEBOUNCE_MS) + return () => { window.clearTimeout(timer) } + }, [previewDirectory, previewDraftLevel]) + // After the hooks: a closed dialog renders nothing and evaluates no copy. const crumbSource = child ?? parent const crumbs = crumbSource === null ? [] : displayCrumbs(crumbSource, t('browser.home')) @@ -637,11 +756,17 @@ export function DirectoryBrowser({ open, listDirectory, createDirectory, onOpen, ))} - {/* The empty zone right of the crumbs is the path-edit affordance. */} + {/* The empty zone right of the crumbs is the path-edit + * affordance: the whole remainder of the bar clicks into + * the editor, and the pencil glyph parked at its right + * edge (with the same tooltip) is what says so — an + * invisible target the operator must guess at is the one + * way into typing a path. */} ) : ( @@ -682,6 +810,9 @@ export function DirectoryBrowser({ open, listDirectory, createDirectory, onOpen, // repopulate the view with the older path. supersede() setLoading(false) + // A fresh edit releases the submission hold: the panes + // may follow the new text wherever it points. + previewSuspended.current = false setPathDraft(event.target.value) }} {...compositionGuard} @@ -699,6 +830,11 @@ export function DirectoryBrowser({ open, listDirectory, createDirectory, onOpen, // focus on the returning crumb edit zone (a failure // keeps the editor, so the flag waits until close). refocusEditZone.current = true + // The submitted path owns the view now: a debounce + // timer still pending from these keystrokes would + // otherwise supersede this navigation and land the + // draft's parent directory instead. + previewSuspended.current = true navigate(pathDraft) } } diff --git a/packages/host/directory-picker-browse/tests/directory-browser.spec.tsx b/packages/host/directory-picker-browse/tests/directory-browser.spec.tsx index ce9f03fb0b..0c7e155add 100644 --- a/packages/host/directory-picker-browse/tests/directory-browser.spec.tsx +++ b/packages/host/directory-picker-browse/tests/directory-browser.spec.tsx @@ -11,9 +11,14 @@ const HOME = '/home/u' const DOCS = `${HOME}/Documents` const HARNESS = `${DOCS}/harness` -/** Listing fake over a tiny fixed tree; unknown paths reject like the Host. */ +/** + * Listing fake over a tiny fixed tree; unknown paths reject like the Host. + * A trailing separator is dropped the way the Host's own `resolve` drops it, + * so a directory part typed into the path editor addresses its level. + */ function listingFor(path?: string): DirectoryListing { - const target = path ?? HOME + const asked = path ?? HOME + const target = asked.length > 1 && asked.endsWith('/') ? asked.slice(0, -1) : asked const tree: Record = { [HOME]: { path: HOME, @@ -659,9 +664,14 @@ describe('DirectoryBrowser', () => { // A dot-led prefix names hidden entries, so it reveals the match. fireEvent.change(input, { target: { value: `${HOME}/.co` } }) expect(screen.getByRole('listitem').textContent).toBe('.config') - // A prefix matching nothing empties the level (no stale rows linger). + // A prefix nobody matches releases the filter: the level shows whole + // (hidden rows back under the toggle) instead of emptying under a name + // the operator is still spelling. fireEvent.change(input, { target: { value: `${HOME}/zzz` } }) - expect(screen.queryByRole('listitem')).toBeNull() + expect(screen.getAllByRole('listitem').map(item => item.textContent)).toEqual(['Documents']) + // Its dot-led reveal lapses with it. + fireEvent.change(input, { target: { value: `${HOME}/.zzz` } }) + expect(screen.getAllByRole('listitem').map(item => item.textContent)).toEqual(['Documents']) // A draft naming some other directory (or none) leaves the level whole. fireEvent.change(input, { target: { value: 'no-separator' } }) expect(screen.getByRole('listitem').textContent).toBe('Documents') @@ -679,17 +689,95 @@ describe('DirectoryBrowser', () => { expect(input.value).toBe(`${DOCS}/`) fireEvent.change(input, { target: { value: `${DOCS}/h` } }) expect(within(columns()[1]!).getByText('harness')).toBeTruthy() + // A miss releases the right pane's filter rather than emptying it. fireEvent.change(input, { target: { value: `${DOCS}/zzz` } }) - expect(within(columns()[1]!).queryAllByRole('listitem')).toHaveLength(0) + expect(within(columns()[1]!).getByText('harness')).toBeTruthy() expect(within(columns()[0]!).getByText('Documents')).toBeTruthy() // Erasing back into the parent's own path moves the filter to the LEFT - // pane and releases the right one. The selected row is exempt (it - // anchors the two-pane view), so it alone survives the miss. + // pane and releases the right one — no scan, both levels are on screen. fireEvent.change(input, { target: { value: `${HOME}/zz` } }) expect(within(columns()[0]!).getAllByRole('listitem').map(item => item.textContent)).toEqual(['Documents']) expect(within(columns()[1]!).getByText('harness')).toBeTruthy() }) + it('follows the draft into a directory no pane lists, and back up when segments are erased', async () => { + const b = mount() + await waitFor(() => { expect(screen.getByRole('listitem')).toBeTruthy() }) + fireEvent.click(screen.getByRole('button', { name: 'browser.editPath' })) + const input = screen.getByLabelText('browser.editPath') + // Typing past a separator addresses a level nobody shows: the panes + // follow it once the typing rests, and the tail filters the arrival. + fireEvent.change(input, { target: { value: `${DOCS}/h` } }) + await waitFor(() => { expect(screen.getByText('harness')).toBeTruthy() }) + expect(b.listDirectory.mock.calls.at(-1)?.[0]).toBe(`${DOCS}/`) + // Still editing: the panes moved under the draft, the editor stayed. + expect(screen.getByLabelText('browser.editPath').value).toBe(`${DOCS}/h`) + // Erasing back past the separator steps the panes up a level again. + fireEvent.change(input, { target: { value: `${HOME}/Do` } }) + await waitFor(() => { expect(screen.getByText('Documents')).toBeTruthy() }) + expect(b.listDirectory.mock.calls.at(-1)?.[0]).toBe(`${HOME}/`) + expect(columns()).toHaveLength(1) + }) + + it('keeps the panes and stays silent when a draft-following scan fails', async () => { + const b = mount() + await waitFor(() => { expect(screen.getByRole('listitem')).toBeTruthy() }) + fireEvent.click(screen.getByRole('button', { name: 'browser.editPath' })) + const input = screen.getByLabelText('browser.editPath') + fireEvent.change(input, { target: { value: `${HOME}/nope/x` } }) + await waitFor(() => { expect(b.listDirectory).toHaveBeenCalledWith(`${HOME}/nope/`, expect.anything()) }) + // A half-typed directory is unreadable most of the time: the last + // readable level keeps rendering and no error interrupts the typing. + expect(screen.getByText('Documents')).toBeTruthy() + expect(screen.queryByRole('alert')).toBeNull() + }) + + it('holds the draft-following scan while a submitted path is in flight', async () => { + const listDirectory = vi.fn(async (path?: string) => { + // The submitted leg never settles, so the debounce window elapses with + // the navigation still owning the view. + if (path === HARNESS) return await new Promise(() => {}) + return listingFor(path) + }) + mount({ listDirectory }) + await waitFor(() => { expect(screen.getByRole('listitem')).toBeTruthy() }) + fireEvent.click(screen.getByRole('button', { name: 'browser.editPath' })) + const input = screen.getByLabelText('browser.editPath') + fireEvent.change(input, { target: { value: HARNESS } }) + fireEvent.keyDown(input, { key: 'Enter' }) + await act(async () => { await new Promise((resolve) => { setTimeout(resolve, 400) }) }) + // Only the initial home listing and the submitted path — the draft's + // directory part was never scanned behind the navigation's back. + expect(listDirectory.mock.calls.map(call => call[0])).toEqual([undefined, HARNESS]) + }) + + it('discards draft-following scans that a newer edit superseded', async () => { + let landDocs = (): void => {} + let failRoot = (): void => {} + const listDirectory = vi.fn(async (path?: string) => { + if (path === `${DOCS}/`) return await new Promise((resolve) => { landDocs = () => { resolve(listingFor(DOCS)) } }) + if (path === '/') { + return await new Promise((_, reject) => { + failRoot = () => { reject(new Error('root unreadable')) } + }) + } + return listingFor(path) + }) + mount({ listDirectory }) + await waitFor(() => { expect(screen.getByRole('listitem')).toBeTruthy() }) + fireEvent.click(screen.getByRole('button', { name: 'browser.editPath' })) + const input = screen.getByLabelText('browser.editPath') + fireEvent.change(input, { target: { value: `${DOCS}/h` } }) + await waitFor(() => { expect(listDirectory).toHaveBeenCalledWith(`${DOCS}/`, expect.anything()) }) + fireEvent.change(input, { target: { value: '/x' } }) + await waitFor(() => { expect(listDirectory).toHaveBeenCalledWith('/', expect.anything()) }) + // Back onto the listed level: neither pending scan may still land. + fireEvent.change(input, { target: { value: `${HOME}/D` } }) + await act(async () => { landDocs(); failRoot() }) + expect(screen.getAllByRole('listitem').map(item => item.textContent)).toEqual(['Documents']) + expect(screen.queryByRole('alert')).toBeNull() + }) + it('keeps the draft and filter through window focus loss and in-dialog focus moves', async () => { mount() await waitFor(() => { expect(screen.getByRole('listitem')).toBeTruthy() }) From 0b0de64769a5a57294012ac03c3852be0103e4de Mon Sep 17 00:00:00 2001 From: kingwl Date: Mon, 3 Aug 2026 11:40:33 +0800 Subject: [PATCH 21/48] bound inactive subagent timing to projection cut --- ...07-27-web-subagent-conversations.i18n.yaml | 4 +- .../2026-07-27-web-subagent-conversations.md | 2 +- ...026-07-27-web-subagent-conversations.zh.md | 2 +- .../src/client/SubagentCatalogAction.tsx | 8 +-- .../tests/conversation-ui.spec.tsx | 14 ++++-- packages/subagent/subagent/README.i18n.yaml | 4 +- packages/subagent/subagent/README.md | 2 +- packages/subagent/subagent/README.zh.md | 2 +- .../subagent/subagent/src/projection-types.ts | 9 +++- packages/subagent/subagent/src/projection.ts | 50 ++++++++++++------- .../subagent/tests/timing-projection.spec.ts | 7 ++- 11 files changed, 66 insertions(+), 38 deletions(-) diff --git a/.agents/notes/implemented/feature/2026-07-27-web-subagent-conversations.i18n.yaml b/.agents/notes/implemented/feature/2026-07-27-web-subagent-conversations.i18n.yaml index 168d28d13a..08db3986b2 100644 --- a/.agents/notes/implemented/feature/2026-07-27-web-subagent-conversations.i18n.yaml +++ b/.agents/notes/implemented/feature/2026-07-27-web-subagent-conversations.i18n.yaml @@ -2,5 +2,5 @@ # side as of the last confirmed-consistent state. Both languages carry equal authority; # after editing either side, bring the other along and re-record with: # pnpm run verify-translation-pairing --write .agents/notes/implemented/feature/2026-07-27-web-subagent-conversations.md -2026-07-27-web-subagent-conversations.md: 859c6c5c17e830ab55c8513d56741966655eaf7a -2026-07-27-web-subagent-conversations.zh.md: 05d5c0f1d59b0bdebdecb33dc360e937af44d7b6 +2026-07-27-web-subagent-conversations.md: 09a444dc6c391f3abdfc5bfe0f3367d4bda430c3 +2026-07-27-web-subagent-conversations.zh.md: 87929e8b35212092587ca4566bd10a1242c84e30 diff --git a/.agents/notes/implemented/feature/2026-07-27-web-subagent-conversations.md b/.agents/notes/implemented/feature/2026-07-27-web-subagent-conversations.md index 859c6c5c17..09a444dc6c 100644 --- a/.agents/notes/implemented/feature/2026-07-27-web-subagent-conversations.md +++ b/.agents/notes/implemented/feature/2026-07-27-web-subagent-conversations.md @@ -41,7 +41,7 @@ The header action is absent only when a complete empty direct-catalog response a `running` means the exact child Agent driver is draining work at the Host sampling boundary; `inactive` means that driver is idle or absent. The UI does not translate either value into success, failure, cancellation, completeness, or resumability. `subagent.list` supplies the current driver-status baseline, `host/session-status` updates known activity in place, request-local replay prevents an older in-flight list response from overwriting a newer transition, and `host/session-removed` returns a known row to `inactive`; reconnect reads a fresh baseline. A `host/session-added` frame for a direct subagent immediately flips any loaded parent row to `hasChildren: true`, and that positive hint survives an older in-flight catalog response; membership, labels, mode, diagnostics, and the authoritative snapshot still require a debounced `subagent.list` refresh while the affected branch is open. A prompt response remains delivery-time authority. -Healthy rows reuse the standard session projections retained in the list mirror. `subagentTiming` resets at every descriptor so an inherited fork seed cannot enter the child's total, accumulates completed `turn/start` → `turn/end` spans, and carries the current turn's `activeSince`. The menu formats whole seconds and advances its local clock only while a known descendant is running; an inactive row uses settled duration, or the summary's last activity to bound an interrupted open turn, so reopening the menu never restarts completed work. The duration does not imply a durable outcome. +Healthy rows reuse the standard session projections retained in the list mirror. `subagentTiming` resets at every descriptor so an inherited fork seed cannot enter the child's total, accumulates completed `turn/start` → `turn/end` spans, and carries same-cut `active.since` and `active.through` bounds for an open turn. The menu formats whole seconds and advances its local clock only while a known descendant is running; an inactive row bounds an interrupted open turn with `active.through`, so a stale projection never borrows newer session metadata and reopening the menu never restarts completed work. The duration does not imply a durable outcome. Selecting a row records its exact address before opening the resident client `Session`. History pagination, event folding, tool render intents, titles, and live mux reconciliation reuse the ordinary conversation machinery. Breadcrumbs use catalog labels, follow parent links only through `origin: 'subagent'` rows, include the first ordinary owner, and keep ordinary forks single-level. Forking an addressed subagent creates an ordinary fork with direct source lineage and attaches it to the nearest workspace-owning ancestor. The catalog is an ARIA tree with lazy ArrowRight/ArrowLeft disclosure, linear ArrowUp/ArrowDown navigation, Home/End, Escape, and focus restoration. diff --git a/.agents/notes/implemented/feature/2026-07-27-web-subagent-conversations.zh.md b/.agents/notes/implemented/feature/2026-07-27-web-subagent-conversations.zh.md index 05d5c0f1d5..87929e8b35 100644 --- a/.agents/notes/implemented/feature/2026-07-27-web-subagent-conversations.zh.md +++ b/.agents/notes/implemented/feature/2026-07-27-web-subagent-conversations.zh.md @@ -41,7 +41,7 @@ Figma 中的 [subagent 列表](https://www.figma.com/design/jRBBK7zBgcszdVWQ0Fh5 `running` 表示在 Host 采样边界，确切 child Agent driver 正在处理工作；`inactive` 表示该 driver 空闲或不存在。UI 不会把任一值解释为成功、失败、取消、完成状态或可恢复性。`subagent.list` 提供当前 driver 状态基线，`host/session-status` 会就地更新已知活动状态，请求内回放会阻止更早发起但尚未完成的列表响应覆盖较新的状态转换，`host/session-removed` 则会使已知行恢复为 `inactive`；重连时会读取新的基线。直接 subagent 的 `host/session-added` 帧会立即把任何已加载的 parent 行翻转为 `hasChildren: true`，并使这项正向提示不被更早发起但尚未完成的目录响应覆盖；受影响分支打开期间，成员、label、mode、diagnostic 与权威快照仍需要通过去抖动的 `subagent.list` 刷新来更新。消息投递时仍以提示词响应为权威依据。 -健康行会复用列表镜像中保留的标准会话投影。`subagentTiming` 会在每个描述符处重置，使继承的 fork 种子不会计入 child 总量；它会累加已完成的 `turn/start` → `turn/end` 时段，并携带当前轮次的 `activeSince`。菜单会以整秒格式化时间，且仅在有已知后代处于运行状态时才推进其本地时钟；对 inactive 行，菜单使用已结算耗时，或以摘要的最后活动为被中断未结束轮次的上界，因此重新打开菜单绝不会让已完成工作重新计时。该耗时不蕴含持久化结果语义。 +健康行会复用列表镜像中保留的标准会话投影。`subagentTiming` 会在每个描述符处重置，使继承的 fork 种子不会计入 child 总量；它会累加已完成的 `turn/start` → `turn/end` 时段，并携带未结束轮次同一切面的 `active.since` 和 `active.through` 边界。菜单会以整秒格式化时间，且仅在有已知后代处于运行状态时才推进其本地时钟；对 inactive 行，菜单以 `active.through` 为被中断未结束轮次的上界，因此陈旧投影绝不会借用更新的会话元数据，且重新打开菜单绝不会让已完成工作重新计时。该耗时不蕴含持久化结果语义。 选择一行后，系统会先记录其确切地址，再打开常驻客户端 `Session`。历史分页、事件 fold、工具渲染意图、title 与实时 mux 归并都会复用普通对话机制。面包屑导航使用目录 label，只会沿 `origin: 'subagent'` 行的父链接逐级回溯，包含第一个普通 owner，并让普通 fork 保持单层。从已寻址 subagent 创建 fork 时，会生成具有直接源谱系的普通 fork，并将其附加到最近拥有 Workspace 的祖先。目录是一棵 ARIA 树，支持懒加载式 ArrowRight／ArrowLeft 展开与折叠、线性 ArrowUp／ArrowDown 导航、Home／End、Escape 以及焦点恢复。 diff --git a/packages/client/ui-subagent/src/client/SubagentCatalogAction.tsx b/packages/client/ui-subagent/src/client/SubagentCatalogAction.tsx index d507b30019..eae3addf0e 100644 --- a/packages/client/ui-subagent/src/client/SubagentCatalogAction.tsx +++ b/packages/client/ui-subagent/src/client/SubagentCatalogAction.tsx @@ -69,9 +69,11 @@ function activityDuration( const timing: SessionProjectionMap['subagentTiming'] | undefined = summary.projectionValues?.subagentTiming if (timing === undefined) return undefined - if (timing.activeSince === undefined) return timing.settledMs - const end = activity === 'running' ? now : summary.updatedAt - return timing.settledMs + Math.max(0, end - timing.activeSince) + if (timing.active === undefined) return timing.settledMs + const end = activity === 'running' + ? now + : timing.active.through + return timing.settledMs + Math.max(0, end - timing.active.since) } /** Format a non-negative duration to seconds without dropping larger units. */ diff --git a/packages/client/ui-subagent/tests/conversation-ui.spec.tsx b/packages/client/ui-subagent/tests/conversation-ui.spec.tsx index 487cd14f35..a6cbb22aeb 100644 --- a/packages/client/ui-subagent/tests/conversation-ui.spec.tsx +++ b/packages/client/ui-subagent/tests/conversation-ui.spec.tsx @@ -245,9 +245,9 @@ describe('SubagentCatalogAction', () => { vi.useFakeTimers() vi.setSystemTime(now) const rows = [ - ['running', 'running', 65_000, now - 5_000, now], - ['finished', 'inactive', 3_723_000, undefined, now - 60_000], - ['interrupted', 'inactive', 2_000, now - 7_000, now - 3_000], + ['running', 'running', 65_000, now - 5_000, now - 1_000, now], + ['finished', 'inactive', 3_723_000, undefined, undefined, now - 60_000], + ['interrupted', 'inactive', 2_000, now - 7_000, now - 3_000, now + 60_000], ] as const const entries = rows.map(([id, activity]) => ({ kind: 'child' as const, @@ -257,7 +257,9 @@ describe('SubagentCatalogAction', () => { activity, hasChildren: false, })) - const summaries = Object.fromEntries(rows.map(([id, activity, settledMs, activeSince, updatedAt]) => { + const summaries = Object.fromEntries(rows.map(([ + id, activity, settledMs, activeSince, activeThrough, updatedAt, + ]) => { const childId = id as SessionId return [id, { ...summary(childId, updatedAt), @@ -267,7 +269,9 @@ describe('SubagentCatalogAction', () => { projectionValues: { subagentTiming: { settledMs, - ...(activeSince === undefined ? {} : { activeSince }), + ...(activeSince === undefined || activeThrough === undefined + ? {} + : { active: { since: activeSince, through: activeThrough } }), }, }, }] diff --git a/packages/subagent/subagent/README.i18n.yaml b/packages/subagent/subagent/README.i18n.yaml index 76740d4506..ceac4245a5 100644 --- a/packages/subagent/subagent/README.i18n.yaml +++ b/packages/subagent/subagent/README.i18n.yaml @@ -2,5 +2,5 @@ # side as of the last confirmed-consistent state. Both languages carry equal authority; # after editing either side, bring the other along and re-record with: # pnpm run verify-translation-pairing --write packages/subagent/subagent/README.md -README.md: ec4af55bcd9374b1abb55d7bb098eef568449684 -README.zh.md: 8323853ff0de2a15da6475fc1433a68f0074ee93 +README.md: e54f0b98ec3649cec428a47026e6657a9749608b +README.zh.md: 1624fa59854d9b61770c5ef0f9d89f7882198da4 diff --git a/packages/subagent/subagent/README.md b/packages/subagent/subagent/README.md index ec4af55bcd..e54f0b98ec 100644 --- a/packages/subagent/subagent/README.md +++ b/packages/subagent/subagent/README.md @@ -92,7 +92,7 @@ Provider additions and removals also emit `subagent/provider-added` and `subagen Continuable children do not create `SubagentRun` or Tasks. The continuation manager directly owns one process-local Activation and retained `AgentHandle` per resident child Session, uses the Agent inbox as the only FIFO, and cold-resumes from the durable descriptor. Exact live direct-parent identity authorizes parent-to-child delivery. Exact live child identity authorizes reports; the manager derives the recipient from durable `parentSession`, and `MessageSource` remains provenance rather than authority. -When `ctx.sessionProjections` is available, the service registers `subagentTiming`. The projection resets at each descriptor so a fork seed's ancestor work cannot enter the child's total, then accumulates `turn/start` → `turn/end` active time and retains `activeSince` for an open turn. Only descriptors and turn boundaries change the value, so token chunks do not create timing updates. +When `ctx.sessionProjections` is available, the service registers `subagentTiming`. The projection resets at each descriptor so a fork seed's ancestor work cannot enter the child's total, then accumulates `turn/start` → `turn/end` active time and retains same-cut `active.since` and `active.through` bounds for an open turn. While that turn remains open, `active.through` follows the latest folded event, giving an inactive consumer a conservative crash bound without mixing in newer session metadata. `registerContinuableSetup()` lets optional packages add child-scoped capabilities without teaching the continuation manager their names. Contributions install synchronously before Activation publication, roll back with failed setup, and are released with the child scope. New grants wait for the next Activation, while contribution removal revokes every resident installation immediately. diff --git a/packages/subagent/subagent/README.zh.md b/packages/subagent/subagent/README.zh.md index 8323853ff0..1624fa5985 100644 --- a/packages/subagent/subagent/README.zh.md +++ b/packages/subagent/subagent/README.zh.md @@ -92,7 +92,7 @@ subagent seam 允许一个 agent（智能体）通过具名提供方把工作委 可继续子级不会创建 `SubagentRun` 或 Task。延续管理器为每个驻留子 Session 直接拥有一个仅存在于当前进程的 Activation 和一个留存的 `AgentHandle`，使用 Agent inbox 作为唯一 FIFO，并从持久化描述符冷恢复。父到子投递由准确的实时直接父级身份授权。上报则由准确的实时子级身份授权；管理器根据持久化的 `parentSession` 推导接收方，`MessageSource` 仍只表示来源，不表示权限。 -当 `ctx.sessionProjections` 可用时，服务会注册 `subagentTiming`。该投影会在每个描述符处重置，使 fork 种子中的祖先工作不会计入 child 总量，随后累加 `turn/start` → `turn/end` 活跃时间，并为未结束的轮次保留 `activeSince`。只有描述符和轮次边界会改变该值，因此 token 分片不会产生计时更新。 +当 `ctx.sessionProjections` 可用时，服务会注册 `subagentTiming`。该投影会在每个描述符处重置，使 fork 种子中的祖先工作不会计入 child 总量，随后累加 `turn/start` → `turn/end` 活跃时间，并为未结束的轮次保留同一切面的 `active.since` 和 `active.through` 边界。在该轮次保持未结束期间，`active.through` 会跟随最近折叠的事件，从而为 inactive 消费方提供保守的崩溃上界，又不会混入更新的会话元数据。 `registerContinuableSetup()` 允许可选包添加子级作用域功能，而无需让延续管理器知道这些功能的名称。贡献会在 Activation 发布前同步安装，在设置失败时一并回滚，并随子级作用域释放。新授权须等到下一个 Activation，移除贡献则会立即撤销每个驻留安装项。 diff --git a/packages/subagent/subagent/src/projection-types.ts b/packages/subagent/subagent/src/projection-types.ts index cefaec3727..c5a23b03b8 100644 --- a/packages/subagent/subagent/src/projection-types.ts +++ b/packages/subagent/subagent/src/projection-types.ts @@ -8,8 +8,13 @@ export interface SubagentTimingProjection { /** Milliseconds accumulated across completed turns after the child's own descriptor. */ settledMs: number - /** Start of the currently open turn, when one has not reached `turn/end`. */ - activeSince?: number + /** Same-cut bounds of the currently open turn, when one has not reached `turn/end`. */ + active?: { + /** Start of the open turn. */ + since: number + /** Latest event time folded into this projection cut. */ + through: number + } } declare module '@deepseek-ai/dsh-session-projection/types' { diff --git a/packages/subagent/subagent/src/projection.ts b/packages/subagent/subagent/src/projection.ts index 6b15a66bbf..171ea18eef 100644 --- a/packages/subagent/subagent/src/projection.ts +++ b/packages/subagent/subagent/src/projection.ts @@ -8,16 +8,25 @@ import { z } from 'zod' import type { ProjectionDefinition } from '@deepseek-ai/dsh-session-projection' import type { SubagentTimingProjection } from './projection-types.ts' -interface TimingState extends SubagentTimingProjection { +interface TimingState { + /** Milliseconds accumulated across completed post-descriptor turns. */ + settledMs: number + /** Current open interval kept paired inside the fold. */ + active?: { since: number; through: number } /** Latest pre-descriptor turn start, promoted when the child's own descriptor arrives. */ pendingTurnStart?: number /** Whether the fold has crossed a descriptor in this logical log. */ descriptorSeen: boolean } +// Cast for the optional values: under exactOptionalPropertyTypes zod infers +// `number | undefined` where the interface declares absent-or-number fields. const projectionSchema = z.object({ settledMs: z.number().int().nonnegative(), - activeSince: z.number().int().nonnegative().optional(), + active: z.object({ + since: z.number().int().nonnegative(), + through: z.number().int().nonnegative(), + }).strict().optional(), }).strict() as unknown as z.ZodType /** @@ -36,33 +45,38 @@ ProjectionDefinition<'subagentTiming', TimingState> = { apply: (state, event) => { if (event.type === 'turn/start') { return state.descriptorSeen - ? { ...state, activeSince: event.time } + ? { ...state, active: { since: event.time, through: event.time } } : { ...state, pendingTurnStart: event.time } } if (event.type === 'subagent/descriptor') { - const activeSince = state.activeSince ?? state.pendingTurnStart + const activeSince = state.active?.since ?? state.pendingTurnStart return { descriptorSeen: true, settledMs: 0, - ...(activeSince === undefined ? {} : { activeSince }), + ...(activeSince === undefined + ? {} + : { active: { since: activeSince, through: event.time } }), } } - if (event.type !== 'turn/end') return state - if (!state.descriptorSeen) { - if (state.pendingTurnStart === undefined) return state - const { pendingTurnStart: _closed, ...next } = state - return next - } - if (state.activeSince === undefined) return state - const { activeSince, ...rest } = state - return { - ...rest, - settledMs: state.settledMs + Math.max(0, event.time - activeSince), + if (event.type === 'turn/end') { + if (!state.descriptorSeen) { + if (state.pendingTurnStart === undefined) return state + const { pendingTurnStart: _closed, ...next } = state + return next + } + if (state.active === undefined) return state + const { active, ...rest } = state + return { + ...rest, + settledMs: state.settledMs + Math.max(0, event.time - active.since), + } } + if (state.active === undefined) return state + return { ...state, active: { ...state.active, through: event.time } } }, view: state => ({ settledMs: state.settledMs, - ...(state.activeSince === undefined ? {} : { activeSince: state.activeSince }), + ...(state.active === undefined ? {} : { active: state.active }), }), - stateVersion: 1, + stateVersion: 2, } diff --git a/packages/subagent/subagent/tests/timing-projection.spec.ts b/packages/subagent/subagent/tests/timing-projection.spec.ts index ac3cac9ebb..e9a3be43ea 100644 --- a/packages/subagent/subagent/tests/timing-projection.spec.ts +++ b/packages/subagent/subagent/tests/timing-projection.spec.ts @@ -21,10 +21,13 @@ describe('subagent timing projection', () => { const ctx = new Context() await ctx.plugin(SessionStore) await ctx.plugin(SessionProjectionRegistry) - await ctx.plugin(SubagentService) + const serviceFiber = await ctx.plugin(SubagentService) expect(ctx.sessionProjections.snapshot(ctx.sessions.create()).values.subagentTiming) .toEqual({ settledMs: 0 }) + await serviceFiber.dispose() + expect(ctx.sessionProjections.snapshot(ctx.sessions.create()).values.subagentTiming) + .toBeUndefined() }) it('resets inherited seed timing at the child descriptor and sums later completed turns', () => { @@ -47,7 +50,7 @@ describe('subagent timing projection', () => { event('turn/end', 2, 900), event('turn/start', 3, 2_000), event('assistant/chunk', 4, 2_500), - ])).toEqual({ settledMs: 0, activeSince: 2_000 }) + ])).toEqual({ settledMs: 0, active: { since: 2_000, through: 2_500 } }) }) it('ignores completed pre-descriptor turns and unrelated events', () => { From fdf83f2c3ebb5fc14689491e24a8737101e5ac36 Mon Sep 17 00:00:00 2001 From: kingwl Date: Mon, 3 Aug 2026 11:44:11 +0800 Subject: [PATCH 22/48] document projection-cut duration bound --- packages/client/ui-subagent/README.i18n.yaml | 4 ++-- packages/client/ui-subagent/README.md | 2 +- packages/client/ui-subagent/README.zh.md | 2 +- 3 files changed, 4 insertions(+), 4 deletions(-) diff --git a/packages/client/ui-subagent/README.i18n.yaml b/packages/client/ui-subagent/README.i18n.yaml index 2b512252f0..01e2412288 100644 --- a/packages/client/ui-subagent/README.i18n.yaml +++ b/packages/client/ui-subagent/README.i18n.yaml @@ -2,5 +2,5 @@ # side as of the last confirmed-consistent state. Both languages carry equal authority; # after editing either side, bring the other along and re-record with: # pnpm run verify-translation-pairing --write packages/client/ui-subagent/README.md -README.md: 16a54484fb53544c71af0806c1497a18f9141002 -README.zh.md: 166a25d095b3e7f10f7239262f39e27972344529 +README.md: 33a2f2899fc34af52cda6b19f473847927da7ef0 +README.zh.md: cab1edb6b81df4b89d05153c801a4d277037ad9c diff --git a/packages/client/ui-subagent/README.md b/packages/client/ui-subagent/README.md index 16a54484fb..33a2f2899f 100644 --- a/packages/client/ui-subagent/README.md +++ b/packages/client/ui-subagent/README.md @@ -4,7 +4,7 @@ English | [中文](README.zh.md) Web subagent feature owner: contributes the lazily expandable catalog tree to `conversation.session.header.actions`, reason-specific read-only replacements to the conversation composer chain, and the existing `@` reference source to `ctx.slash`. -The header action reads `subagentsByParent` and session summaries through the standard `useSessions` hook. After a non-empty direct catalog arrives, its trigger counts the complete subagent-only descendant lineage, stops at ordinary forks, and shows ongoing activity when any counted descendant is running. The compact tree remains direct-catalog authoritative: continuable and one-shot rows display mode plus `running`/`inactive` activity, an optional log-backed title, and active-turn duration to the second; an unlabeled one-shot row falls back to its session id, while corrupt, unsupported, or unavailable rows remain readable but disabled. Duration sums completed `subagentTiming` turns, advances once per second only for an open turn on a running child, and freezes after the child becomes inactive; an interrupted open turn is bounded by the session summary's last activity. Each healthy row's `hasChildren` hint determines disclosure before interaction, so known leaves never show an arrow; expanding a branch immediately reserves one disabled loading row per known direct descendant, then lazily replaces them with that child's authoritative catalog. Every visible branch is reported to the runtime so membership frames cause a debounced refresh only where the tree is being consumed. Selecting any depth calls `SessionsService.openSubagent()` with the row's exact `{parentSessionId, childSessionId, mode}` address. Component-local state owns tree visibility, expanded branches, keyboard focus, and the running-duration clock. ArrowRight/ArrowLeft expand and collapse branches; ArrowUp/ArrowDown, Home, End, and Escape navigate or close the tree; closing returns focus to the trigger. Styling uses tokens only. +The header action reads `subagentsByParent` and session summaries through the standard `useSessions` hook. After a non-empty direct catalog arrives, its trigger counts the complete subagent-only descendant lineage, stops at ordinary forks, and shows ongoing activity when any counted descendant is running. The compact tree remains direct-catalog authoritative: continuable and one-shot rows display mode plus `running`/`inactive` activity, an optional log-backed title, and active-turn duration to the second; an unlabeled one-shot row falls back to its session id, while corrupt, unsupported, or unavailable rows remain readable but disabled. Duration sums completed `subagentTiming` turns, advances once per second only for an open turn on a running child, and freezes after the child becomes inactive; an interrupted open turn is bounded by its same-cut `active.through`, never by newer session metadata. Each healthy row's `hasChildren` hint determines disclosure before interaction, so known leaves never show an arrow; expanding a branch immediately reserves one disabled loading row per known direct descendant, then lazily replaces them with that child's authoritative catalog. Every visible branch is reported to the runtime so membership frames cause a debounced refresh only where the tree is being consumed. Selecting any depth calls `SessionsService.openSubagent()` with the row's exact `{parentSessionId, childSessionId, mode}` address. Component-local state owns tree visibility, expanded branches, keyboard focus, and the running-duration clock. ArrowRight/ArrowLeft expand and collapse branches; ArrowUp/ArrowDown, Home, End, and Escape navigate or close the tree; closing returns focus to the trigger. Styling uses tokens only. A one-shot child always elects a read-only composer that identifies the transcript as a completed execution record. A continuable child does so only when its exact parent is unavailable, with copy explaining the recovery path. A continuable child with a live parent keeps the ordinary input chrome, whose Session routes through `subagent.prompt`; running input remains Send because every follow-up joins the child's FIFO inbox, and addressed sessions never expose Stop. This package never receives host context or calls a model-facing tool. The catalog and composer behavior are specified by the [Web subagent conversations Agent Note](../../../.agents/notes/implemented/feature/2026-07-27-web-subagent-conversations.md). diff --git a/packages/client/ui-subagent/README.zh.md b/packages/client/ui-subagent/README.zh.md index 166a25d095..cab1edb6b8 100644 --- a/packages/client/ui-subagent/README.zh.md +++ b/packages/client/ui-subagent/README.zh.md @@ -4,7 +4,7 @@ Web subagent 功能 owner：向 `conversation.session.header.actions` 贡献可懒加载展开的目录树，向会话编辑器链贡献按原因区分的只读替代呈现，并保留注册到 `ctx.slash` 的既有 `@` 引用 source。 -页头操作通过标准 `useSessions` 钩子读取 `subagentsByParent` 与会话摘要。非空直接目录到达后，其触发器会统计仅含 subagent 的完整后代谱系，在普通 fork 处停止，并在任一计入统计的后代处于 `running` 时显示活动仍在进行。紧凑树仍以直接目录为权威依据：可继续和 one-shot 行会显示 mode、`running`／`inactive` 活动状态、由日志支撑的可选 title，以及精确到秒的活跃轮次耗时；没有 label 的 one-shot 行会回退到其会话 id，而损坏、不受支持或不可用的行仍保持可读但禁用。耗时会累加已完成的 `subagentTiming` 轮次，仅在运行中 child 存在未结束轮次时每秒递增一次，并在 child 变为 inactive 后冻结；被中断的未结束轮次以会话摘要中的最后活动为上界。每个健康行的 `hasChildren` 提示会在交互前决定是否显示展开控件，因此已知叶子节点从不显示箭头；展开分支时，会立即为每个已知直接后代预留一行禁用的加载行，随后再用该 child 的权威目录懒加载结果替换这些占位行。每个可见分支都会上报给运行时，使成员帧只在树正被消费的位置触发去抖动刷新。选择任意深度的条目都会使用该行的确切地址 `{parentSessionId, childSessionId, mode}` 调用 `SessionsService.openSubagent()`。组件局部状态负责树的可见性、已展开分支、键盘焦点与运行中耗时时钟。ArrowRight／ArrowLeft 展开和折叠分支；ArrowUp／ArrowDown、Home、End 与 Escape 用于导航或关闭树；关闭后焦点返回触发器。样式只使用 token。 +页头操作通过标准 `useSessions` 钩子读取 `subagentsByParent` 与会话摘要。非空直接目录到达后，其触发器会统计仅含 subagent 的完整后代谱系，在普通 fork 处停止，并在任一计入统计的后代处于 `running` 时显示活动仍在进行。紧凑树仍以直接目录为权威依据：可继续和 one-shot 行会显示 mode、`running`／`inactive` 活动状态、由日志支撑的可选 title，以及精确到秒的活跃轮次耗时；没有 label 的 one-shot 行会回退到其会话 id，而损坏、不受支持或不可用的行仍保持可读但禁用。耗时会累加已完成的 `subagentTiming` 轮次，仅在运行中 child 存在未结束轮次时每秒递增一次，并在 child 变为 inactive 后冻结；被中断的未结束轮次以其同一切面的 `active.through` 为上界，绝不使用更新的会话元数据。每个健康行的 `hasChildren` 提示会在交互前决定是否显示展开控件，因此已知叶子节点从不显示箭头；展开分支时，会立即为每个已知直接后代预留一行禁用的加载行，随后再用该 child 的权威目录懒加载结果替换这些占位行。每个可见分支都会上报给运行时，使成员帧只在树正被消费的位置触发去抖动刷新。选择任意深度的条目都会使用该行的确切地址 `{parentSessionId, childSessionId, mode}` 调用 `SessionsService.openSubagent()`。组件局部状态负责树的可见性、已展开分支、键盘焦点与运行中耗时时钟。ArrowRight／ArrowLeft 展开和折叠分支；ArrowUp／ArrowDown、Home、End 与 Escape 用于导航或关闭树；关闭后焦点返回触发器。样式只使用 token。 one-shot child 始终选用只读编辑器，并将 transcript（文本记录）说明为已完成的执行记录。可继续 child 仅在其确切 parent 不可用时选用只读编辑器，并以文案说明恢复路径。确切 parent 存活时，可继续 child 保留普通输入 chrome，其 Session 会通过 `subagent.prompt` 路由；child 运行期间，输入操作仍为 Send，因为每条后续消息都会进入 child 的 FIFO inbox，且已寻址会话绝不公开 Stop。本包绝不接收宿主 context，也不调用面向模型的工具。目录与编辑器行为由 [Web subagent 对话 Agent Note](../../../.agents/notes/implemented/feature/2026-07-27-web-subagent-conversations.md)规定。 From c183fae6f9827e20e1a5e09ea38bc91b3f0b5a76 Mon Sep 17 00:00:00 2001 From: kingwl Date: Mon, 3 Aug 2026 12:04:44 +0800 Subject: [PATCH 23/48] clarify timing projection schema cast --- packages/subagent/subagent/src/projection.ts | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/packages/subagent/subagent/src/projection.ts b/packages/subagent/subagent/src/projection.ts index 171ea18eef..ffdcb4fd09 100644 --- a/packages/subagent/subagent/src/projection.ts +++ b/packages/subagent/subagent/src/projection.ts @@ -19,8 +19,8 @@ interface TimingState { descriptorSeen: boolean } -// Cast for the optional values: under exactOptionalPropertyTypes zod infers -// `number | undefined` where the interface declares absent-or-number fields. +// Zod's optional output includes explicit `undefined`; with +// exactOptionalPropertyTypes the public interface permits omission only. const projectionSchema = z.object({ settledMs: z.number().int().nonnegative(), active: z.object({ From 30f442d42e36ab989596a6120a04236802a0902f Mon Sep 17 00:00:00 2001 From: creatixchu Date: Mon, 3 Aug 2026 12:11:11 +0800 Subject: [PATCH 24/48] review(directory-picker-browse): re-arm the draft-following wait per keystroke ds-review-bot round one. Keying the debounce on the directory part the draft named left two states with no recovery until the operator crossed a separator: a keystroke that superseded an in-flight scan never re-armed one, and an edit after a rejected submission released the hold with no timer left to release. The wait is now keyed on the draft itself and decides its target when it fires, reading the panes through a ref so a landing cannot re-arm it (a host answering with a differently spelled path would otherwise scan forever). A landed scan that unmounts the row a keyboard operator Tabbed onto re-parks focus on the still-open editor; the Modal has no focus trap. That a walked-to level survives closing the editor is now stated in the README, the Agent Note, and the module contract. The new e2e stages its own beta directory so running it alone sees the tree its assertions describe. --- ...directory-picker-capability-seam.i18n.yaml | 4 +- ...-07-28-directory-picker-capability-seam.md | 2 +- ...-28-directory-picker-capability-seam.zh.md | 2 +- apps/web/tests/workspace-management.e2e.ts | 3 + .../directory-picker-browse/README.i18n.yaml | 4 +- .../host/directory-picker-browse/README.md | 2 +- .../host/directory-picker-browse/README.zh.md | 2 +- .../src/client/DirectoryBrowser.tsx | 54 ++++++++++---- .../tests/directory-browser.spec.tsx | 74 ++++++++++++++++++- 9 files changed, 124 insertions(+), 23 deletions(-) diff --git a/.agents/notes/implemented/architecture/2026-07-28-directory-picker-capability-seam.i18n.yaml b/.agents/notes/implemented/architecture/2026-07-28-directory-picker-capability-seam.i18n.yaml index 4342f49f2b..9afbc7ebcd 100644 --- a/.agents/notes/implemented/architecture/2026-07-28-directory-picker-capability-seam.i18n.yaml +++ b/.agents/notes/implemented/architecture/2026-07-28-directory-picker-capability-seam.i18n.yaml @@ -2,5 +2,5 @@ # side as of the last confirmed-consistent state. Both languages carry equal authority; # after editing either side, bring the other along and re-record with: # pnpm run verify-translation-pairing --write .agents/notes/implemented/architecture/2026-07-28-directory-picker-capability-seam.md -2026-07-28-directory-picker-capability-seam.md: 90aa8bc7cfc0dc0fb3d057b9991682c9b531ea23 -2026-07-28-directory-picker-capability-seam.zh.md: 12917c95456bca9cdd5e20ae97847156af81277e +2026-07-28-directory-picker-capability-seam.md: c7833eaee691618bb76e40bf34c2114dda030315 +2026-07-28-directory-picker-capability-seam.zh.md: 9a6d52a3e1e3629ae54dbcf46ee8e785536294cd diff --git a/.agents/notes/implemented/architecture/2026-07-28-directory-picker-capability-seam.md b/.agents/notes/implemented/architecture/2026-07-28-directory-picker-capability-seam.md index 90aa8bc7cf..c7833eaee6 100644 --- a/.agents/notes/implemented/architecture/2026-07-28-directory-picker-capability-seam.md +++ b/.agents/notes/implemented/architecture/2026-07-28-directory-picker-capability-seam.md @@ -20,7 +20,7 @@ Placement and policy rulings folded into this decision: - **Dependency survey (hand-roll vs adopt).** Node's stdlib *is* the maintained cross-platform OS layer (`readdir(withFileTypes)`, `homedir`, path semantics); surveyed alternatives fail the dependency bar — file-manager packages (`node-file-manager`, `files-and-folders`, Syncfusion's provider) are whole HTTP apps (fit), drive-letter helpers (`drivelist` native addon, `windows-drive-letters` ~7y stale) fail health/proportionality. The browse backend is a thin adapter over stdlib. - **Hidden entries: return-and-flag.** The host stamps `hidden` (POSIX dot convention) and returns everything; the client filters. Display policy stays client-side, and the show-hidden toggle shipped as exactly that client-only change: a fixed-label footer toggle whose state lives in the pressed presentation (`aria-pressed` + check glyph), a dot-led path-draft prefix reveals the hidden entries it names, and the current selection is exempt from both the hidden and the prefix filter (it anchors the two-pane view). Windows' `FILE_ATTRIBUTE_HIDDEN` is not exposed by dirents — documented limitation until a native probe pays for itself. - **Path-editor cancel scope: the dialog card.** The browse client's path editor cancels on Escape and on focus leaving the card, both observed at a card-scope wrapper rather than the input — after Tab parks focus on a filtered row the input is off the event path, yet Escape must collapse the editor (not the dialog) and a later focus departure must still cancel. Non-cancel exemptions: window/tab focus loss, in-card focus moves, and pointer paths (rows and the toggle suppress focus steal on mousedown while editing). Separators for seeding and draft-tail filtering are inferred from `listing.home`; the wire-field alternative below records the deferred authoritative form. Combobox semantics between the editor and the list it filters (`aria-expanded`/`aria-controls`/active-descendant, result announcements) are likewise deferred — today they read to assistive tech as separate widgets. -- **The path editor advertises itself, and the panes follow the draft.** The click-to-edit zone is not invisible: a pencil glyph sits at the bar's right edge and hover/focus lights the zone in the editor's own footprint, so the one route into typing a path is discoverable and the bar does not resize when zone and input swap. While the editor is open the panes track the draft instead of whatever level happened to be listed when it opened — the final segment prefix-filters the level its directory part names, a tail nobody matches releases the filter (a name still being spelled must not empty the pane it is being spelled into), and a directory part no pane lists is scanned after a 250ms rest and lands single-wide in place, so typing deeper descends and erasing segments steps back up without leaving the editor. That scan is speculative — half-typed directories are unreadable most of the time — so a failure keeps the last readable panes and stays silent. Enter remains the authoritative commit: it owns the view from submission until landing (a debounce timer armed by the same keystrokes is held back rather than superseding the navigation, and a rejected submission stays held until the next edit) and it alone surfaces the failure. +- **The path editor advertises itself, and the panes follow the draft.** The click-to-edit zone is not invisible: a pencil glyph sits at the bar's right edge and hover/focus lights the zone in the editor's own footprint, so the one route into typing a path is discoverable and the bar does not resize when zone and input swap. While the editor is open the panes track the draft instead of whatever level happened to be listed when it opened — the final segment prefix-filters the level its directory part names, a tail nobody matches releases the filter (a name still being spelled must not empty the pane it is being spelled into), and a directory part no pane lists is scanned after a 250ms rest and lands single-wide in place, so typing deeper descends and erasing segments steps back up without leaving the editor. That scan is speculative — half-typed directories are unreadable most of the time — so a failure keeps the last readable panes and stays silent. Enter remains the authoritative commit: it owns the view from submission until landing (a debounce timer armed by the same keystrokes is held back rather than superseding the navigation, and a rejected submission stays held until the next edit) and it alone surfaces the failure. Two consequences are deliberate. The wait is keyed on the draft, not on the directory part it names, so a keystroke that superseded an in-flight scan re-arms one and an edit after a rejected submission releases the hold; the panes it reads are a ref rather than a dependency, or the landing would re-arm the wait and a host answering with a differently spelled path would scan forever. And a walk is not rewound: closing the editor — cancellation included — leaves the panes where the draft took them, named by the crumbs and followed by Open's fallback target, because the operator watched them move. A landing that unmounts the row a keyboard operator Tabbed onto re-parks focus on the editor, since the Modal has no focus trap. - **Navigation lands selection-anchored, quiet, and bounded.** Away from the display root (the same collapse the crumb header renders, so crumbs and pane shape never disagree), the landing is two-pane: the target's actual parent-level entry re-selected (platform case folding on Windows), its children on the right, so a crumb jump reads as stepping back one pane rather than collapsing to a single column. Target and parent legs land as **one frame** when the parent leg settles within the 200ms wait bound — the stale view keeps rendering until then, so navigation swaps the panes without an intermediate single-pane flash — and past the bound the target commits alone at once (an Enter-submitted navigation is never held hostage by a stalled parent) with the late parent leg upgrading the landing in place. The parent leg runs under the landing's supersession scope and is aborted on the wire by any newer intent (Escape inside the landing window therefore withdraws the whole navigation); a failed parent leg, or a truncated parent window lacking the target, leaves the single-pane landing — the upgrade must never orphan the selection it exists to anchor. The loading indicator follows the same quiet rule: it floats over the content's bottom-right corner (never a layout-shifting row; the truncated/error rows own the bottom left and keep rendering through a scan) and only once a scan outlives a 300ms silence window, so a local listing swaps with nothing shown at all. Row picks are deliberately exempt from the one-frame rule: a pick's immediate pane split is its selected-state feedback (aria-current, crumbs following), while a navigation has nothing to acknowledge the click but the swap itself. Both timing constants are calibrated for local enumeration; a remote deployment (one RPC per level, commonly 100–400ms) would sit inside the silence window with no pressed state on the crumbs — revisit the window or add pressed feedback when a remote consumer lands. - **Symlinks: follow for enterability.** `stat` probes symlinks (broken/cyclic → skipped); crumbs keep the logical path the operator navigated, and `workspace.create` already canonicalizes via realpath at adoption. - **Listing levels are bounded, and streamed.** One `list` call returns at most `maxEntries` rows (config, default 1000 — GitHub's web-UI directory-listing bound). The level streams via `opendir` into a name-sorted window of `maxEntries + 1` candidates, so memory stays O(maxEntries) and enterability probing touches only windowed candidates; the wire `DirectoryListing` carries a required `truncated` flag so the client states incompleteness instead of silently missing tail entries. A windowed broken symlink is not backfilled from beyond the window — the eviction already marks the level truncated. Window insertion is binary with an O(1) full-window tail rejection (an oversized level must not pay a window scan per dirent), and `list(path, signal)` threads the carrier's request signal so a scan of a stalled network directory cannot outlive a disconnected caller — every await in the scan (open, each read, each symlink probe) races the signal, an aborted exit abandons rather than awaits the close (Node queues close behind in-flight reads), and abandoned settlements are swallowed so cleanup can never surface as an unhandled rejection. An unbounded level is a memory/responsiveness hole for large or adversarial directories. diff --git a/.agents/notes/implemented/architecture/2026-07-28-directory-picker-capability-seam.zh.md b/.agents/notes/implemented/architecture/2026-07-28-directory-picker-capability-seam.zh.md index 12917c9545..9a6d52a3e1 100644 --- a/.agents/notes/implemented/architecture/2026-07-28-directory-picker-capability-seam.zh.md +++ b/.agents/notes/implemented/architecture/2026-07-28-directory-picker-capability-seam.zh.md @@ -20,7 +20,7 @@ web GUI 的"打开本地文件夹"流程被焊死在一种交互上：`host.pick - **依赖调研（手写 vs 引入）。** Node 标准库本身就是维护中的跨平台 OS 层（`readdir(withFileTypes)`、`homedir`、路径语义）；调研过的替代品都过不了依赖门槛——文件管理器包（`node-file-manager`、`files-and-folders`、Syncfusion 的 provider）是整套 HTTP 应用（契合度不过），盘符工具（原生插件 `drivelist`、约七年未更的 `windows-drive-letters`）健康度／比例失当。browse 后端是标准库上的薄适配。 - **隐藏条目：返回并打标。** 宿主标注 `hidden`（POSIX 点前缀约定）并返回全部条目；客户端过滤。展示策略留在客户端，"显示隐藏"开关正是作为这一纯客户端改动落地：标签固定的 footer 开关，其状态由按下态呈现承载（`aria-pressed` + 勾选符号）；以点开头的路径草稿前缀会显出它所指名的隐藏条目；当前选中项则不受隐藏与前缀两种过滤影响（它锚定着双栏视图）。Windows 的 `FILE_ATTRIBUTE_HIDDEN` 不被 dirent 暴露——记为限制，直到原生探测值回其成本。 - **路径编辑器的取消范围：对话框卡片。** browse 客户端的路径编辑器在按 Escape 与焦点离开卡片时取消，两者都在卡片范围的包装层而非输入框上监听——Tab 把焦点停到某个过滤命中的行之后，输入框已不在事件路径上，但 Escape 仍须收起编辑器（而非对话框），其后的焦点离开也仍须取消。不取消的豁免：窗口／标签页失焦、卡片内焦点移动，以及指针路径（编辑期间行与开关在 mousedown 时抑制焦点夺取）。预填与草稿末段过滤所用的分隔符从 `listing.home` 推断；下文的线上字段替代方案记录了被延期的权威形态。编辑器与其过滤的列表之间的 combobox 语义（`aria-expanded`／`aria-controls`／active-descendant、结果播报）同样被延期——目前二者在辅助技术看来是彼此独立的控件。 -- **路径编辑器自我点明，各栏跟随草稿。** 点击即编辑的区域不再是隐形的：栏右端坐着一枚铅笔图标，悬停／聚焦时该区以编辑器自身的轮廓亮起，于是键入路径这唯一入口可被发现，且区域与输入框互换时栏高不变。编辑器打开期间，各栏跟随草稿，而不是停在它打开那一刻恰好列出的层级——末段对其目录部分所指的层级做前缀过滤，无一匹配的末段解除过滤（还在拼写中的名字不该把正在拼写它的那一栏清空），而任何一栏都未列出的目录部分会在停顿 250ms 后被扫描、以单宽栏就地落地，于是继续键入即下潜、删掉末段即上退，全程不必离开编辑器。该扫描是推测性的——键入到一半的目录多数时候读不出来——因此失败时保留最后一次可读的分栏并保持沉默。Enter 仍是权威提交：自提交至落地由它独占视图（同一批按键武装的防抖计时器会被扣住，而不是顶掉这次导航；提交被拒后仍扣住，直到下一次编辑），也只有它把失败呈现出来。 +- **路径编辑器自我点明，各栏跟随草稿。** 点击即编辑的区域不再是隐形的：栏右端坐着一枚铅笔图标，悬停／聚焦时该区以编辑器自身的轮廓亮起，于是键入路径这唯一入口可被发现，且区域与输入框互换时栏高不变。编辑器打开期间，各栏跟随草稿，而不是停在它打开那一刻恰好列出的层级——末段对其目录部分所指的层级做前缀过滤，无一匹配的末段解除过滤（还在拼写中的名字不该把正在拼写它的那一栏清空），而任何一栏都未列出的目录部分会在停顿 250ms 后被扫描、以单宽栏就地落地，于是继续键入即下潜、删掉末段即上退，全程不必离开编辑器。该扫描是推测性的——键入到一半的目录多数时候读不出来——因此失败时保留最后一次可读的分栏并保持沉默。Enter 仍是权威提交：自提交至落地由它独占视图（同一批按键武装的防抖计时器会被扣住，而不是顶掉这次导航；提交被拒后仍扣住，直到下一次编辑），也只有它把失败呈现出来。有两点是刻意为之。等待以草稿为键，而非以它指名的目录部分为键，于是顶掉在飞扫描的那次按键会重新武装等待，被拒提交之后的编辑也能释放那道扣留；而它读取的分栏是 ref 而非依赖，否则落地会重新武装等待，遇到以不同拼写作答的宿主便会永远扫描下去。以及，走过的路不回退：关闭编辑器——包括取消——都把分栏留在草稿带到的地方，由面包屑指明、Open 的兜底目标随之而动，因为操作者亲眼看着它们移动。若落地卸载了键盘操作者 Tab 停留的那一行，焦点会被重新停回编辑器——Modal 并没有焦点陷阱。 - **导航以选中项为锚、安静且有界地落地。** 在展示根之外（与 crumb 头部渲染的是同一塌缩，因此 crumb 与分栏形态永不相左），落地即双栏：重新选中目标在父层级中的实际条目（Windows 上按平台惯例折叠大小写），右侧展示其子项，因此 crumb 跳转读作后退一栏，而不是塌缩成单列。父层级这一程在 200ms 等待上限内落定时，目标与父层级两程以**同一帧**落地——在此之前陈旧视图持续渲染，导航换栏时因此没有中间的单栏闪现——超出该上限则目标即刻单独提交（Enter 提交的导航绝不会被滞塞的父层级扣作人质），迟到的父层级这一程再就地升级这次落地。父层级这一程在落地的 supersession 范围下运行，任何较新的意图都会在线上将其中止（因此在落地窗口内按 Escape 即撤回整次导航）；父层级这一程失败，或被截断的父窗口缺少目标时，都保留单栏落地——升级的存在正是为了锚定选中项，绝不能反而让它悬空。加载指示器遵循同一安静规则：它浮于内容右下角（绝不是会挪动布局的一行；截断／错误行占据左下角，并在扫描期间持续渲染），且仅在扫描超出 300ms 静默窗口后才出现，因此本地列举切换时什么也不显示。行选取被刻意豁免于同一帧规则：选取后立即分栏本身就是其选中态反馈（aria-current、crumb 跟随），而导航除了换栏本身没有任何东西可确认这次点击。两个时序常量都按本地列举校准；远程部署（每层级一次 RPC，通常 100–400ms）会落在静默窗口之内、crumb 上却没有按下态——待远程消费方落地时，重新审视该窗口或补上按下反馈。 - **符号链接：为可进入性而跟随。** 用 `stat` 探测符号链接（断链／循环→跳过）；面包屑保留操作者导航的逻辑路径，`workspace.create` 在接纳时本就做 realpath 规范化。 - **列举层级有上限，且流式处理。** 单次 `list` 至多返回 `maxEntries` 行（配置项，默认 1000——GitHub 网页端目录列举的同一上限）。层级经 `opendir` 流入一个按名排序、容量 `maxEntries + 1` 的候选窗口，内存保持 O(maxEntries)，可进入性探测只触及窗口内候选；线上 `DirectoryListing` 携带必填的 `truncated` 标志，让客户端明示不完整而不是静默缺尾。窗口内的断链符号链接不从窗口外回填——发生过驱逐本身已把层级标记为截断。窗口插入为二分查找、满窗尾部单次比较即拒绝（超大层级不能为每个 dirent 付出一次全窗扫描），且 `list(path, signal)` 透传载体的请求信号，滞塞网络目录的扫描不会在调用方断连后继续存活——扫描中的每个 await（打开、每次读取、每次符号链接探测）都与信号赛跑，中止路径放弃而非等待 close（Node 会把 close 排在在飞读取之后），被放弃的 settlement 全部吞掉，清理不会以未处理拒绝的形式冒出。无上限的层级对超大或恶意构造的目录就是内存／响应性漏洞。 diff --git a/apps/web/tests/workspace-management.e2e.ts b/apps/web/tests/workspace-management.e2e.ts index 1a5531f59a..8eec0bc31c 100644 --- a/apps/web/tests/workspace-management.e2e.ts +++ b/apps/web/tests/workspace-management.e2e.ts @@ -407,8 +407,11 @@ describe('web e2e: workspace management (create / rename / flat view / hover aff it('walks the panes with the typed path: deeper past a separator, back up on erase, whole on a miss', async () => { // The panes must track the draft without leaving the editor, so the // typed text and what is listed under it never disagree. + // Staged by this scenario itself (mkdir is recursive and idempotent), so + // running it alone through -t sees the same tree the assertions describe. const staged = join(scaffold.workspaceCwd, 'browse-golden') await mkdir(join(staged, 'alpha', 'only-under-alpha'), { recursive: true }) + await mkdir(join(staged, 'beta'), { recursive: true }) const dialog = await browseTo(staged) await expect.poll(() => dialog.getByText('alpha', { exact: true }).count(), { timeout: 10_000 }).toBe(1) await dialog.getByRole('button', { name: 'Edit path' }).click() diff --git a/packages/host/directory-picker-browse/README.i18n.yaml b/packages/host/directory-picker-browse/README.i18n.yaml index 4063c7d692..06979f1226 100644 --- a/packages/host/directory-picker-browse/README.i18n.yaml +++ b/packages/host/directory-picker-browse/README.i18n.yaml @@ -2,5 +2,5 @@ # side as of the last confirmed-consistent state. Both languages carry equal authority; # after editing either side, bring the other along and re-record with: # pnpm run verify-translation-pairing --write packages/host/directory-picker-browse/README.md -README.md: 7cb0ec785766e954ff4bb39df6825ee7e8c9d821 -README.zh.md: ec71a90bbcd004ec9f9c0d8a9a236882a7487b73 +README.md: a559f7f23b74f694c25fb43e005cbaccd4024308 +README.zh.md: 321e27d7a1dc9eb9bf0d1dab37c1af42e8ad6519 diff --git a/packages/host/directory-picker-browse/README.md b/packages/host/directory-picker-browse/README.md index 7cb0ec7857..a559f7f23b 100644 --- a/packages/host/directory-picker-browse/README.md +++ b/packages/host/directory-picker-browse/README.md @@ -6,7 +6,7 @@ The **in-app browsing backend** of the [directory-picker seam](../directory-pick Behavior facts: listings return **directories only**, name-sorted, with symlinks-to-directories followed (broken/cyclic links skipped — the probe `stat` failing means "not enterable") and a host-owned `hidden` flag (POSIX dot convention) left for the client to act on; `crumbs` is the root-to-target ancestor chain, the root crumb labeled by its full path (`/`, `C:\`); an absent `list` path means the host account's home directory. `createDirectory` is non-recursive (a missing parent is a real failure, not a level to invent) and validates the name as a single non-blank segment even when called directly, mirroring the wire schema's fence. Both primitives reject an explicit path that is not fully qualified — relative forms, and on Windows the rooted drive-less forms (`\foo`, `/foo`) and incomplete UNC prefixes (`\\`, `\\server`) that `isAbsolute` accepts — with `directory-unreadable`/`directory-create-failed`, instead of letting `resolve` rebase it under the host process cwd or current drive. One `list` call returns at most `maxEntries` rows (config, default 1000 — the bound GitHub's web UI applies to directory listings), and the level streams through a bounded window so memory stays O(maxEntries) no matter how many children the directory holds: a cut level keeps the name-sorted head, counts hidden rows against the bound, probes only windowed candidates, and reports `truncated: true` so the client can say the level is incomplete (a windowed broken symlink is not backfilled from beyond the window — the eviction already marks the level truncated); window insertion is binary with an O(1) full-window tail rejection, and `list` threads the caller's `AbortSignal` so a disconnect or timeout stops the scan instead of letting it outlive the caller. Failures throw the seam's typed `DirectoryPickerError`. Policy rationale: [the directory-picker capability seam Agent Note](../../../.agents/notes/implemented/architecture/2026-07-28-directory-picker-capability-seam.md). -**Dual-face package**: the browser half (`./client`) fills [ui-workspace's](../../client/ui-workspace/README.md) two directory-flow holes with the in-app **Select Workspace Directory** dialog (figma `Harness` 813-23126 family — Miller two-column view whose navigations land selection-anchored and quiet: the previous view keeps rendering while a crumb jump or a submitted path is scanned (a "Loading…" pill floats over it only once the scan outlives a 300ms silence window, never shifting the columns), then target and parent legs land as one two-pane frame with the target re-selected as its actual parent-level entry — so stepping back never collapses and no intermediate frame flashes (a parent leg outliving its 200ms wait bound lands the target alone and upgrades in place; a failed or truncated parent leg keeps the single-pane landing; the display root keeps the single wide level); breadcrumb with a click-to-edit path zone, advertised by the pencil glyph at the bar's right edge and lit on hover in the editor's own footprint, whose editor seeds a trailing separator and then keeps the panes under the draft: the final segment prefix-filters the level its directory part names (case-insensitively, over the listed — possibly truncated — rows only; a tail nobody matches releases the filter instead of emptying the pane), while a directory part no pane lists is scanned after a 250ms rest and shown in place, so typing deeper descends and erasing segments steps back up without leaving the editor — a speculative scan is silent when it fails, and Enter still navigates by the exact text, owning the view until it lands; the editor cancels on Escape or when focus leaves the dialog card (window/tab switches and in-card focus moves keep the draft); a fixed-label show-hidden footer toggle over the host's `hidden` flags, with a dot-led typed prefix revealing its matches and the current selection exempt from every filter; nested New-folder dialog), driving `host.listDirectory`/`host.createDirectory` and registering its own locale namespace (`directory-browser`, zh default / en). One cordis.yml row therefore composes both sides of the browse interaction; the client carries no capability-kind branching, and mounting a second flow package fails at load (the holes are `single` kind). +**Dual-face package**: the browser half (`./client`) fills [ui-workspace's](../../client/ui-workspace/README.md) two directory-flow holes with the in-app **Select Workspace Directory** dialog (figma `Harness` 813-23126 family — Miller two-column view whose navigations land selection-anchored and quiet: the previous view keeps rendering while a crumb jump or a submitted path is scanned (a "Loading…" pill floats over it only once the scan outlives a 300ms silence window, never shifting the columns), then target and parent legs land as one two-pane frame with the target re-selected as its actual parent-level entry — so stepping back never collapses and no intermediate frame flashes (a parent leg outliving its 200ms wait bound lands the target alone and upgrades in place; a failed or truncated parent leg keeps the single-pane landing; the display root keeps the single wide level); breadcrumb with a click-to-edit path zone, advertised by the pencil glyph at the bar's right edge and lit on hover in the editor's own footprint, whose editor seeds a trailing separator and then keeps the panes under the draft: the final segment prefix-filters the level its directory part names (case-insensitively, over the listed — possibly truncated — rows only; a tail nobody matches releases the filter instead of emptying the pane), while a directory part no pane lists is scanned after a 250ms rest and shown in place, so typing deeper descends and erasing segments steps back up without leaving the editor — a speculative scan is silent when it fails, and Enter still navigates by the exact text, owning the view until it lands; the editor cancels on Escape or when focus leaves the dialog card (window/tab switches and in-card focus moves keep the draft), and panes the draft walked to stay where the walk ended — the crumbs name that level and Open's fallback target follows them, so cancelling closes the editor rather than rewinding the walk; a fixed-label show-hidden footer toggle over the host's `hidden` flags, with a dot-led typed prefix revealing its matches and the current selection exempt from every filter; nested New-folder dialog), driving `host.listDirectory`/`host.createDirectory` and registering its own locale namespace (`directory-browser`, zh default / en). One cordis.yml row therefore composes both sides of the browse interaction; the client carries no capability-kind branching, and mounting a second flow package fails at load (the holes are `single` kind). ## Model Experience diff --git a/packages/host/directory-picker-browse/README.zh.md b/packages/host/directory-picker-browse/README.zh.md index ec71a90bbc..321e27d7a1 100644 --- a/packages/host/directory-picker-browse/README.zh.md +++ b/packages/host/directory-picker-browse/README.zh.md @@ -6,7 +6,7 @@ 行为事实：列举**只返回目录**、按名称排序，指向目录的符号链接会被跟随（断链／循环链接被跳过——探测 `stat` 失败即"不可进入"），并携带宿主判定的 `hidden` 标志（POSIX 点前缀约定），展示决策留给客户端；`crumbs` 是从根到目标的祖先链，根 crumb 以完整路径标注（`/`、`C:\`）；`list` 不带路径即列举宿主账户的家目录。`createDirectory` 不递归（父目录缺失是真实失败，不是要补造的层级），且即便被直接调用也把名称校验为单个非空段，与协议 schema 的栅栏一致。两个原语都拒绝非完全限定的显式路径——相对形态，以及 Windows 上 `isAbsolute` 会放行的无盘符有根形态（`\foo`、`/foo`）与不完整的 UNC 前缀（`\\`、`\\server`）——报 `directory-unreadable`／`directory-create-failed`，而不是任由 `resolve` 把它重定位到宿主进程 cwd 或当前盘符之下。单次 `list` 至多返回 `maxEntries` 行（配置项，默认 1000——GitHub 网页端对目录列举采用的同一上限），且层级以流式方式经过一个有界窗口，无论目录有多少子项内存都保持 O(maxEntries)：被截断的层级保留按名排序的头部、隐藏行计入上限、只探测窗口内候选，并报告 `truncated: true`，供客户端提示层级不完整（窗口内的断链符号链接不会从窗口外回填——发生过驱逐本身已把层级标记为截断）；窗口插入为二分查找、满窗尾部单次比较即拒绝，且 `list` 透传调用方的 `AbortSignal`，断连或超时会停止扫描而不是让它在调用方离开后继续。失败抛出 seam 的类型化 `DirectoryPickerError`。策略依据：[目录选择能力 seam Agent Note](../../../.agents/notes/implemented/architecture/2026-07-28-directory-picker-capability-seam.md)。 -**双面包**：browser half（`./client`）以应用内 **选择工作区目录** 对话框（figma `Harness` 813-23126 家族——Miller 双列视图，其导航以选中项为锚、安静落地：扫描 crumb 跳转或提交的路径期间，先前视图持续渲染（"Loading…" 胶囊仅在扫描超出 300ms 静默窗口后才浮于其上，绝不挪动各列），随后目标与父层级两程以单个双栏帧落地，目标被重新选中为其在父层级中的实际条目——因此后退绝不塌缩，也没有中间帧闪现（父层级这一程超出其 200ms 等待上限时，目标单独落地，随后就地升级；父层级这一程失败或被截断时保持单栏落地；展示根保持单个宽层级）；带点击即编辑路径区的面包屑，该区由栏右端的铅笔图标点明、悬停时以编辑器自身的轮廓亮起，其编辑器预填尾随分隔符，随后让下方各栏跟随草稿：末段对其目录部分所指的层级做前缀过滤（不区分大小写，且仅作用于已列出、可能被截断的行；无一匹配的末段会解除过滤，而不是把该栏清空），而任何一栏都未列出的目录部分会在停顿 250ms 后被扫描并就地展示，于是继续键入即下潜、删掉末段即上退，全程不必离开编辑器——推测性扫描失败时保持沉默，而 Enter 仍按确切文本导航，并在落地前独占视图；编辑器按 Escape 或焦点离开对话框卡片即取消（窗口／标签页切换与卡片内焦点移动保留草稿）；基于宿主 `hidden` 标志、标签固定的"显示隐藏"footer 开关，键入以点开头的前缀会显出其匹配项，且当前选中项不受任何过滤影响；嵌套新建文件夹对话框）填入 [ui-workspace](../../client/ui-workspace/README.md) 的两个目录流洞，驱动 `host.listDirectory`／`host.createDirectory`，并注册自己的 locale 命名空间（`directory-browser`，zh 默认／en）。因此一行 cordis.yml 同时组合浏览交互的两侧；client 侧不含任何能力 kind 分支，挂载第二个流程包会在加载期失败（洞为 `single` kind）。 +**双面包**：browser half（`./client`）以应用内 **选择工作区目录** 对话框（figma `Harness` 813-23126 家族——Miller 双列视图，其导航以选中项为锚、安静落地：扫描 crumb 跳转或提交的路径期间，先前视图持续渲染（"Loading…" 胶囊仅在扫描超出 300ms 静默窗口后才浮于其上，绝不挪动各列），随后目标与父层级两程以单个双栏帧落地，目标被重新选中为其在父层级中的实际条目——因此后退绝不塌缩，也没有中间帧闪现（父层级这一程超出其 200ms 等待上限时，目标单独落地，随后就地升级；父层级这一程失败或被截断时保持单栏落地；展示根保持单个宽层级）；带点击即编辑路径区的面包屑，该区由栏右端的铅笔图标点明、悬停时以编辑器自身的轮廓亮起，其编辑器预填尾随分隔符，随后让下方各栏跟随草稿：末段对其目录部分所指的层级做前缀过滤（不区分大小写，且仅作用于已列出、可能被截断的行；无一匹配的末段会解除过滤，而不是把该栏清空），而任何一栏都未列出的目录部分会在停顿 250ms 后被扫描并就地展示，于是继续键入即下潜、删掉末段即上退，全程不必离开编辑器——推测性扫描失败时保持沉默，而 Enter 仍按确切文本导航，并在落地前独占视图；编辑器按 Escape 或焦点离开对话框卡片即取消（窗口／标签页切换与卡片内焦点移动保留草稿），而草稿走到的层级会留在原地——面包屑指明该层级、Open 的兜底目标随之而动，因此取消只是关闭编辑器，并不回退这段行走；基于宿主 `hidden` 标志、标签固定的"显示隐藏"footer 开关，键入以点开头的前缀会显出其匹配项，且当前选中项不受任何过滤影响；嵌套新建文件夹对话框）填入 [ui-workspace](../../client/ui-workspace/README.md) 的两个目录流洞，驱动 `host.listDirectory`／`host.createDirectory`，并注册自己的 locale 命名空间（`directory-browser`，zh 默认／en）。因此一行 cordis.yml 同时组合浏览交互的两侧；client 侧不含任何能力 kind 分支，挂载第二个流程包会在加载期失败（洞为 `single` kind）。 ## 模型体验 diff --git a/packages/host/directory-picker-browse/src/client/DirectoryBrowser.tsx b/packages/host/directory-picker-browse/src/client/DirectoryBrowser.tsx index 1f9903952d..e3a13cc54d 100644 --- a/packages/host/directory-picker-browse/src/client/DirectoryBrowser.tsx +++ b/packages/host/directory-picker-browse/src/client/DirectoryBrowser.tsx @@ -25,7 +25,9 @@ * reveals the hidden entries it names, and a prefix nobody matches releases * the filter), while a directory part no pane lists is scanned after a short * debounce and shown in place — so typing deeper descends and erasing - * segments steps back up without leaving the editor. + * segments steps back up without leaving the editor. Panes the draft walked + * to stay put when the editor closes (cancellation included): the crumbs name + * where the walk ended, and Open's fallback target follows them. */ import { useCallback, useEffect, useRef, useState } from 'react' import clsx from 'clsx' @@ -157,9 +159,9 @@ function draftPrefixFor(listing: DirectoryListing, draft: string | null): string function pendingPreviewDirectory( parent: DirectoryListing | null, child: DirectoryListing | null, - draft: string | null, + draft: string, ): string | null { - if (parent === null || draft === null) return null + if (parent === null) return null const directory = draftDirectory(parent, draft) if (directory === null || directory === levelDirectory(parent)) return null if (child !== null && directory === levelDirectory(child)) return null @@ -453,10 +455,23 @@ export function DirectoryBrowser({ open, listDirectory, createDirectory, onOpen, * Enter owns the view from submission until its navigation lands, so the * debounce timer the same keystrokes armed must not supersede it. Cleared * by the next edit (and by opening the editor); a failed submission leaves - * it set, so the rejected path is not immediately re-scanned as a preview. + * it set until the operator edits again, so the rejected path is not + * immediately re-scanned as a preview. */ const previewSuspended = useRef(false) + // The panes as the draft-following scan must read them when its wait + // fires: current, but NOT a dependency of the wait (see the effect below). + const viewRef = useRef<{ parent: DirectoryListing | null; child: DirectoryListing | null }>({ parent: null, child: null }) + useEffect(() => { viewRef.current = { parent, child } }, [parent, child]) + + /** + * A landed preview replaced the pane a keyboard operator may have Tabbed + * onto, so the focus it drops is re-parked on the still-open editor (the + * Modal has no focus trap). Consumed by the refocus effect below. + */ + const refocusPathInput = useRef(false) + /** * List the directory the draft addresses and show it WITHOUT closing the * editor: the level replaces the panes single-wide (the selection and its @@ -478,6 +493,7 @@ export function DirectoryBrowser({ open, listDirectory, createDirectory, onOpen, setChild(null) setLoading(false) setError(null) + refocusPathInput.current = true }, () => { if (seq !== requestSeq.current) return setLoading(false) @@ -602,21 +618,25 @@ export function DirectoryBrowser({ open, listDirectory, createDirectory, onOpen, return () => { window.clearTimeout(timer) } }, [loading, scanWindow]) - // The panes follow the draft: a directory part no pane lists is scanned - // once the typing rests. The dependency is the target STRING, so the - // landing it commits cannot re-arm the timer (a host that answers with a - // differently spelled path leaves the target unchanged, hence unrepeated), - // and every further keystroke replaces the pending timer instead of - // queueing another scan. - const previewDirectory = pendingPreviewDirectory(parent, child, pathDraft) + // The panes follow the draft: EVERY keystroke replaces the pending timer, + // and the target is decided when it fires, off the panes as they stand + // then. Keying the wait on the draft (not on the directory part it names) + // is what makes a keystroke that superseded an in-flight scan re-arm one, + // and what lets an edit after a rejected submission release the hold the + // submission took. The panes are read through a ref for the converse + // reason: were they dependencies, the landing this commits would re-arm the + // wait, and a host answering with a differently spelled path would scan + // forever. useEffect(() => { - if (previewDirectory === null) return + if (pathDraft === null) return const timer = window.setTimeout(() => { if (previewSuspended.current) return - previewDraftLevel(previewDirectory) + const directory = pendingPreviewDirectory(viewRef.current.parent, viewRef.current.child, pathDraft) + if (directory === null) return + previewDraftLevel(directory) }, DRAFT_PREVIEW_DEBOUNCE_MS) return () => { window.clearTimeout(timer) } - }, [previewDirectory, previewDraftLevel]) + }, [pathDraft, previewDraftLevel]) // After the hooks: a closed dialog renders nothing and evaluates no copy. const crumbSource = child ?? parent @@ -642,6 +662,12 @@ export function DirectoryBrowser({ open, listDirectory, createDirectory, onOpen, // replacing the picked button's column — while Enter and an input-focused // Escape land on the crumb edit zone that replaces the input. useEffect(() => { + if (refocusPathInput.current) { + refocusPathInput.current = false + // Only when the swap actually dropped focus to body: focus the operator + // still holds (the input itself, a surviving row) stays theirs. + if (document.activeElement === document.body) pathInputRef.current?.focus() + } if (pathDraft !== null) return if (refocusPick.current) { refocusPick.current = false diff --git a/packages/host/directory-picker-browse/tests/directory-browser.spec.tsx b/packages/host/directory-picker-browse/tests/directory-browser.spec.tsx index 0c7e155add..d94029e3bf 100644 --- a/packages/host/directory-picker-browse/tests/directory-browser.spec.tsx +++ b/packages/host/directory-picker-browse/tests/directory-browser.spec.tsx @@ -678,7 +678,7 @@ describe('DirectoryBrowser', () => { }) it('filters the child pane in two-pane mode and follows the draft back up a level', async () => { - mount() + const b = mount() await waitFor(() => { expect(screen.getByRole('listitem')).toBeTruthy() }) fireEvent.click(rowButton(screen.getByRole('listitem'))) await waitFor(() => { expect(columns()).toHaveLength(2) }) @@ -689,6 +689,12 @@ describe('DirectoryBrowser', () => { expect(input.value).toBe(`${DOCS}/`) fireEvent.change(input, { target: { value: `${DOCS}/h` } }) expect(within(columns()[1]!).getByText('harness')).toBeTruthy() + // The child pane already lists that directory: no scan follows, and both + // panes stay. + const settled = b.listDirectory.mock.calls.length + await act(async () => { await new Promise((resolve) => { setTimeout(resolve, 400) }) }) + expect(b.listDirectory.mock.calls).toHaveLength(settled) + expect(columns()).toHaveLength(2) // A miss releases the right pane's filter rather than emptying it. fireEvent.change(input, { target: { value: `${DOCS}/zzz` } }) expect(within(columns()[1]!).getByText('harness')).toBeTruthy() @@ -712,6 +718,12 @@ describe('DirectoryBrowser', () => { expect(b.listDirectory.mock.calls.at(-1)?.[0]).toBe(`${DOCS}/`) // Still editing: the panes moved under the draft, the editor stayed. expect(screen.getByLabelText('browser.editPath').value).toBe(`${DOCS}/h`) + // Typing on inside the level the panes now list costs no scan at all: + // the prefix filter alone answers the draft. + const settled = b.listDirectory.mock.calls.length + fireEvent.change(input, { target: { value: `${DOCS}/ha` } }) + await act(async () => { await new Promise((resolve) => { setTimeout(resolve, 400) }) }) + expect(b.listDirectory.mock.calls).toHaveLength(settled) // Erasing back past the separator steps the panes up a level again. fireEvent.change(input, { target: { value: `${HOME}/Do` } }) await waitFor(() => { expect(screen.getByText('Documents')).toBeTruthy() }) @@ -719,6 +731,62 @@ describe('DirectoryBrowser', () => { expect(columns()).toHaveLength(1) }) + it('re-arms the draft-following scan after a keystroke superseded one in flight', async () => { + let started = 0 + const listDirectory = vi.fn(async (path?: string) => { + if (path !== `${DOCS}/`) return listingFor(path) + started += 1 + // The first scan never settles: the next keystroke aborts it, and only + // a re-armed wait can still land the level the draft names. + if (started === 1) return await new Promise(() => {}) + return listingFor(path) + }) + mount({ listDirectory }) + await waitFor(() => { expect(screen.getByRole('listitem')).toBeTruthy() }) + fireEvent.click(screen.getByRole('button', { name: 'browser.editPath' })) + const input = screen.getByLabelText('browser.editPath') + fireEvent.change(input, { target: { value: `${DOCS}/h` } }) + await waitFor(() => { expect(started).toBe(1) }) + // A further tail keystroke supersedes the in-flight scan; the panes must + // still follow, not sit on the stale level until a separator is typed. + fireEvent.change(input, { target: { value: `${DOCS}/ha` } }) + await waitFor(() => { expect(screen.getByText('harness')).toBeTruthy() }) + }) + + it('follows the draft again after an edit releases a failed submission hold', async () => { + const listDirectory = vi.fn(async (path?: string) => { + if (path === HARNESS) throw new Error('target unreadable') + return listingFor(path) + }) + mount({ listDirectory }) + await waitFor(() => { expect(screen.getByRole('listitem')).toBeTruthy() }) + fireEvent.click(screen.getByRole('button', { name: 'browser.editPath' })) + const input = screen.getByLabelText('browser.editPath') + // Submitting inside the debounce window holds the pending scan back. + fireEvent.change(input, { target: { value: HARNESS } }) + fireEvent.keyDown(input, { key: 'Enter' }) + await waitFor(() => { expect(screen.getByRole('alert').textContent).toBe('target unreadable') }) + // Correcting only the final segment leaves the directory part unchanged; + // the edit must still release the hold and re-arm the wait. + fireEvent.change(input, { target: { value: `${HARNESS}x` } }) + await waitFor(() => { expect(listDirectory).toHaveBeenCalledWith(`${DOCS}/`, expect.anything()) }) + await waitFor(() => { expect(screen.getByText('harness')).toBeTruthy() }) + }) + + it('re-parks focus on the editor when a landed scan unmounts the focused row', async () => { + mount() + await waitFor(() => { expect(screen.getByRole('listitem')).toBeTruthy() }) + fireEvent.click(screen.getByRole('button', { name: 'browser.editPath' })) + const input = screen.getByLabelText('browser.editPath') + fireEvent.change(input, { target: { value: `${DOCS}/h` } }) + // The keyboard path: focus Tabbed onto a row of the level about to be + // replaced. Without a re-park it would fall to body, outside a Modal that + // has no focus trap. + rowButton(screen.getByRole('listitem')).focus() + await waitFor(() => { expect(screen.getByText('harness')).toBeTruthy() }) + expect(document.activeElement).toBe(screen.getByLabelText('browser.editPath')) + }) + it('keeps the panes and stays silent when a draft-following scan fails', async () => { const b = mount() await waitFor(() => { expect(screen.getByRole('listitem')).toBeTruthy() }) @@ -1065,6 +1133,10 @@ describe('DirectoryBrowser', () => { fireEvent.click(screen.getByRole('button', { name: 'browser.editPath' })) const input = screen.getByLabelText('browser.editPath') fireEvent.change(input, { target: { value: DOCS } }) + // With no level listed there is no platform separator to read, so the + // draft-following wait resolves to nothing and the editor types blind. + await act(async () => { await new Promise((resolve) => { setTimeout(resolve, 400) }) }) + expect(listDirectory).toHaveBeenCalledTimes(1) listDirectory.mockImplementation(async (path?: string) => listingFor(path)) fireEvent.keyDown(input, { key: 'Enter' }) await waitFor(() => { expect(screen.getByText('harness')).toBeTruthy() }) From c58b07833e2c59cece49a90383f3bf8ba4f2a49f Mon Sep 17 00:00:00 2001 From: creatixchu Date: Mon, 3 Aug 2026 13:08:31 +0800 Subject: [PATCH 25/48] fix(directory-picker-browse): land the draft-following walk two-pane MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit The draft-following scan replaced the panes with one wide level, so typing a path collapsed the dialog's Miller view — the thing the dialog is. It now lands through the same selection-anchored landing every navigation uses: target and parent legs as one frame, the target re-selected in its parent level, its children on the right. Typing a path moves the Miller view exactly as a crumb jump does. One landing shape, two callers: `land(path, {closeEditor, announce})` is what `navigate` and the draft-following scan share. A submitted path closes the editor and announces failures; the speculative scan keeps both to itself and re-parks the focus its swap dropped. A level a pane already lists still needs no scan at all — the filter alone answers the draft — so erasing back into the parent's own path keeps both panes and only moves the filter. --- ...directory-picker-capability-seam.i18n.yaml | 4 +- ...-07-28-directory-picker-capability-seam.md | 2 +- ...-28-directory-picker-capability-seam.zh.md | 2 +- apps/web/tests/workspace-management.e2e.ts | 15 +- .../directory-picker-browse/README.i18n.yaml | 4 +- .../host/directory-picker-browse/README.md | 2 +- .../host/directory-picker-browse/README.zh.md | 2 +- .../src/client/DirectoryBrowser.tsx | 138 +++++++++--------- .../tests/directory-browser.spec.tsx | 59 ++++++-- 9 files changed, 131 insertions(+), 97 deletions(-) diff --git a/.agents/notes/implemented/architecture/2026-07-28-directory-picker-capability-seam.i18n.yaml b/.agents/notes/implemented/architecture/2026-07-28-directory-picker-capability-seam.i18n.yaml index 9afbc7ebcd..1690808ee0 100644 --- a/.agents/notes/implemented/architecture/2026-07-28-directory-picker-capability-seam.i18n.yaml +++ b/.agents/notes/implemented/architecture/2026-07-28-directory-picker-capability-seam.i18n.yaml @@ -2,5 +2,5 @@ # side as of the last confirmed-consistent state. Both languages carry equal authority; # after editing either side, bring the other along and re-record with: # pnpm run verify-translation-pairing --write .agents/notes/implemented/architecture/2026-07-28-directory-picker-capability-seam.md -2026-07-28-directory-picker-capability-seam.md: c7833eaee691618bb76e40bf34c2114dda030315 -2026-07-28-directory-picker-capability-seam.zh.md: 9a6d52a3e1e3629ae54dbcf46ee8e785536294cd +2026-07-28-directory-picker-capability-seam.md: 15d0a6ad3fc1e92e0487024c0d7f611380382e2d +2026-07-28-directory-picker-capability-seam.zh.md: 54042cea3bf4a4888855a60765ccc19977e6a061 diff --git a/.agents/notes/implemented/architecture/2026-07-28-directory-picker-capability-seam.md b/.agents/notes/implemented/architecture/2026-07-28-directory-picker-capability-seam.md index c7833eaee6..15d0a6ad3f 100644 --- a/.agents/notes/implemented/architecture/2026-07-28-directory-picker-capability-seam.md +++ b/.agents/notes/implemented/architecture/2026-07-28-directory-picker-capability-seam.md @@ -20,7 +20,7 @@ Placement and policy rulings folded into this decision: - **Dependency survey (hand-roll vs adopt).** Node's stdlib *is* the maintained cross-platform OS layer (`readdir(withFileTypes)`, `homedir`, path semantics); surveyed alternatives fail the dependency bar — file-manager packages (`node-file-manager`, `files-and-folders`, Syncfusion's provider) are whole HTTP apps (fit), drive-letter helpers (`drivelist` native addon, `windows-drive-letters` ~7y stale) fail health/proportionality. The browse backend is a thin adapter over stdlib. - **Hidden entries: return-and-flag.** The host stamps `hidden` (POSIX dot convention) and returns everything; the client filters. Display policy stays client-side, and the show-hidden toggle shipped as exactly that client-only change: a fixed-label footer toggle whose state lives in the pressed presentation (`aria-pressed` + check glyph), a dot-led path-draft prefix reveals the hidden entries it names, and the current selection is exempt from both the hidden and the prefix filter (it anchors the two-pane view). Windows' `FILE_ATTRIBUTE_HIDDEN` is not exposed by dirents — documented limitation until a native probe pays for itself. - **Path-editor cancel scope: the dialog card.** The browse client's path editor cancels on Escape and on focus leaving the card, both observed at a card-scope wrapper rather than the input — after Tab parks focus on a filtered row the input is off the event path, yet Escape must collapse the editor (not the dialog) and a later focus departure must still cancel. Non-cancel exemptions: window/tab focus loss, in-card focus moves, and pointer paths (rows and the toggle suppress focus steal on mousedown while editing). Separators for seeding and draft-tail filtering are inferred from `listing.home`; the wire-field alternative below records the deferred authoritative form. Combobox semantics between the editor and the list it filters (`aria-expanded`/`aria-controls`/active-descendant, result announcements) are likewise deferred — today they read to assistive tech as separate widgets. -- **The path editor advertises itself, and the panes follow the draft.** The click-to-edit zone is not invisible: a pencil glyph sits at the bar's right edge and hover/focus lights the zone in the editor's own footprint, so the one route into typing a path is discoverable and the bar does not resize when zone and input swap. While the editor is open the panes track the draft instead of whatever level happened to be listed when it opened — the final segment prefix-filters the level its directory part names, a tail nobody matches releases the filter (a name still being spelled must not empty the pane it is being spelled into), and a directory part no pane lists is scanned after a 250ms rest and lands single-wide in place, so typing deeper descends and erasing segments steps back up without leaving the editor. That scan is speculative — half-typed directories are unreadable most of the time — so a failure keeps the last readable panes and stays silent. Enter remains the authoritative commit: it owns the view from submission until landing (a debounce timer armed by the same keystrokes is held back rather than superseding the navigation, and a rejected submission stays held until the next edit) and it alone surfaces the failure. Two consequences are deliberate. The wait is keyed on the draft, not on the directory part it names, so a keystroke that superseded an in-flight scan re-arms one and an edit after a rejected submission releases the hold; the panes it reads are a ref rather than a dependency, or the landing would re-arm the wait and a host answering with a differently spelled path would scan forever. And a walk is not rewound: closing the editor — cancellation included — leaves the panes where the draft took them, named by the crumbs and followed by Open's fallback target, because the operator watched them move. A landing that unmounts the row a keyboard operator Tabbed onto re-parks focus on the editor, since the Modal has no focus trap. +- **The path editor advertises itself, and the panes follow the draft.** The click-to-edit zone is not invisible: a pencil glyph sits at the bar's right edge and hover/focus lights the zone in the editor's own footprint, so the one route into typing a path is discoverable and the bar does not resize when zone and input swap. While the editor is open the panes track the draft instead of whatever level happened to be listed when it opened — the final segment prefix-filters the level its directory part names, a tail nobody matches releases the filter (a name still being spelled must not empty the pane it is being spelled into), and a directory part no pane lists is scanned after a 250ms rest and lands through the same selection-anchored, two-pane landing every navigation uses, so typing a path moves the Miller view exactly as a crumb jump does — typing deeper descends, erasing segments walks back up — without leaving the editor. One landing shape, two callers: a submitted path closes the editor and announces failures, the draft-following scan keeps both to itself. That scan is speculative — half-typed directories are unreadable most of the time — so a failure keeps the last readable panes and stays silent. Enter remains the authoritative commit: it owns the view from submission until landing (a debounce timer armed by the same keystrokes is held back rather than superseding the navigation, and a rejected submission stays held until the next edit) and it alone surfaces the failure. Two consequences are deliberate. The wait is keyed on the draft, not on the directory part it names, so a keystroke that superseded an in-flight scan re-arms one and an edit after a rejected submission releases the hold; the panes it reads are a ref rather than a dependency, or the landing would re-arm the wait and a host answering with a differently spelled path would scan forever. And a walk is not rewound: closing the editor — cancellation included — leaves the panes where the draft took them, named by the crumbs and followed by Open's fallback target, because the operator watched them move. A landing that unmounts the row a keyboard operator Tabbed onto re-parks focus on the editor, since the Modal has no focus trap. - **Navigation lands selection-anchored, quiet, and bounded.** Away from the display root (the same collapse the crumb header renders, so crumbs and pane shape never disagree), the landing is two-pane: the target's actual parent-level entry re-selected (platform case folding on Windows), its children on the right, so a crumb jump reads as stepping back one pane rather than collapsing to a single column. Target and parent legs land as **one frame** when the parent leg settles within the 200ms wait bound — the stale view keeps rendering until then, so navigation swaps the panes without an intermediate single-pane flash — and past the bound the target commits alone at once (an Enter-submitted navigation is never held hostage by a stalled parent) with the late parent leg upgrading the landing in place. The parent leg runs under the landing's supersession scope and is aborted on the wire by any newer intent (Escape inside the landing window therefore withdraws the whole navigation); a failed parent leg, or a truncated parent window lacking the target, leaves the single-pane landing — the upgrade must never orphan the selection it exists to anchor. The loading indicator follows the same quiet rule: it floats over the content's bottom-right corner (never a layout-shifting row; the truncated/error rows own the bottom left and keep rendering through a scan) and only once a scan outlives a 300ms silence window, so a local listing swaps with nothing shown at all. Row picks are deliberately exempt from the one-frame rule: a pick's immediate pane split is its selected-state feedback (aria-current, crumbs following), while a navigation has nothing to acknowledge the click but the swap itself. Both timing constants are calibrated for local enumeration; a remote deployment (one RPC per level, commonly 100–400ms) would sit inside the silence window with no pressed state on the crumbs — revisit the window or add pressed feedback when a remote consumer lands. - **Symlinks: follow for enterability.** `stat` probes symlinks (broken/cyclic → skipped); crumbs keep the logical path the operator navigated, and `workspace.create` already canonicalizes via realpath at adoption. - **Listing levels are bounded, and streamed.** One `list` call returns at most `maxEntries` rows (config, default 1000 — GitHub's web-UI directory-listing bound). The level streams via `opendir` into a name-sorted window of `maxEntries + 1` candidates, so memory stays O(maxEntries) and enterability probing touches only windowed candidates; the wire `DirectoryListing` carries a required `truncated` flag so the client states incompleteness instead of silently missing tail entries. A windowed broken symlink is not backfilled from beyond the window — the eviction already marks the level truncated. Window insertion is binary with an O(1) full-window tail rejection (an oversized level must not pay a window scan per dirent), and `list(path, signal)` threads the carrier's request signal so a scan of a stalled network directory cannot outlive a disconnected caller — every await in the scan (open, each read, each symlink probe) races the signal, an aborted exit abandons rather than awaits the close (Node queues close behind in-flight reads), and abandoned settlements are swallowed so cleanup can never surface as an unhandled rejection. An unbounded level is a memory/responsiveness hole for large or adversarial directories. diff --git a/.agents/notes/implemented/architecture/2026-07-28-directory-picker-capability-seam.zh.md b/.agents/notes/implemented/architecture/2026-07-28-directory-picker-capability-seam.zh.md index 9a6d52a3e1..54042cea3b 100644 --- a/.agents/notes/implemented/architecture/2026-07-28-directory-picker-capability-seam.zh.md +++ b/.agents/notes/implemented/architecture/2026-07-28-directory-picker-capability-seam.zh.md @@ -20,7 +20,7 @@ web GUI 的"打开本地文件夹"流程被焊死在一种交互上：`host.pick - **依赖调研（手写 vs 引入）。** Node 标准库本身就是维护中的跨平台 OS 层（`readdir(withFileTypes)`、`homedir`、路径语义）；调研过的替代品都过不了依赖门槛——文件管理器包（`node-file-manager`、`files-and-folders`、Syncfusion 的 provider）是整套 HTTP 应用（契合度不过），盘符工具（原生插件 `drivelist`、约七年未更的 `windows-drive-letters`）健康度／比例失当。browse 后端是标准库上的薄适配。 - **隐藏条目：返回并打标。** 宿主标注 `hidden`（POSIX 点前缀约定）并返回全部条目；客户端过滤。展示策略留在客户端，"显示隐藏"开关正是作为这一纯客户端改动落地：标签固定的 footer 开关，其状态由按下态呈现承载（`aria-pressed` + 勾选符号）；以点开头的路径草稿前缀会显出它所指名的隐藏条目；当前选中项则不受隐藏与前缀两种过滤影响（它锚定着双栏视图）。Windows 的 `FILE_ATTRIBUTE_HIDDEN` 不被 dirent 暴露——记为限制，直到原生探测值回其成本。 - **路径编辑器的取消范围：对话框卡片。** browse 客户端的路径编辑器在按 Escape 与焦点离开卡片时取消，两者都在卡片范围的包装层而非输入框上监听——Tab 把焦点停到某个过滤命中的行之后，输入框已不在事件路径上，但 Escape 仍须收起编辑器（而非对话框），其后的焦点离开也仍须取消。不取消的豁免：窗口／标签页失焦、卡片内焦点移动，以及指针路径（编辑期间行与开关在 mousedown 时抑制焦点夺取）。预填与草稿末段过滤所用的分隔符从 `listing.home` 推断；下文的线上字段替代方案记录了被延期的权威形态。编辑器与其过滤的列表之间的 combobox 语义（`aria-expanded`／`aria-controls`／active-descendant、结果播报）同样被延期——目前二者在辅助技术看来是彼此独立的控件。 -- **路径编辑器自我点明，各栏跟随草稿。** 点击即编辑的区域不再是隐形的：栏右端坐着一枚铅笔图标，悬停／聚焦时该区以编辑器自身的轮廓亮起，于是键入路径这唯一入口可被发现，且区域与输入框互换时栏高不变。编辑器打开期间，各栏跟随草稿，而不是停在它打开那一刻恰好列出的层级——末段对其目录部分所指的层级做前缀过滤，无一匹配的末段解除过滤（还在拼写中的名字不该把正在拼写它的那一栏清空），而任何一栏都未列出的目录部分会在停顿 250ms 后被扫描、以单宽栏就地落地，于是继续键入即下潜、删掉末段即上退，全程不必离开编辑器。该扫描是推测性的——键入到一半的目录多数时候读不出来——因此失败时保留最后一次可读的分栏并保持沉默。Enter 仍是权威提交：自提交至落地由它独占视图（同一批按键武装的防抖计时器会被扣住，而不是顶掉这次导航；提交被拒后仍扣住，直到下一次编辑），也只有它把失败呈现出来。有两点是刻意为之。等待以草稿为键，而非以它指名的目录部分为键，于是顶掉在飞扫描的那次按键会重新武装等待，被拒提交之后的编辑也能释放那道扣留；而它读取的分栏是 ref 而非依赖，否则落地会重新武装等待，遇到以不同拼写作答的宿主便会永远扫描下去。以及，走过的路不回退：关闭编辑器——包括取消——都把分栏留在草稿带到的地方，由面包屑指明、Open 的兜底目标随之而动，因为操作者亲眼看着它们移动。若落地卸载了键盘操作者 Tab 停留的那一行，焦点会被重新停回编辑器——Modal 并没有焦点陷阱。 +- **路径编辑器自我点明，各栏跟随草稿。** 点击即编辑的区域不再是隐形的：栏右端坐着一枚铅笔图标，悬停／聚焦时该区以编辑器自身的轮廓亮起，于是键入路径这唯一入口可被发现，且区域与输入框互换时栏高不变。编辑器打开期间，各栏跟随草稿，而不是停在它打开那一刻恰好列出的层级——末段对其目录部分所指的层级做前缀过滤，无一匹配的末段解除过滤（还在拼写中的名字不该把正在拼写它的那一栏清空），而任何一栏都未列出的目录部分会在停顿 250ms 后被扫描，并经由每次导航共用的那套以选中项为锚的双栏落地落定，于是键入路径移动 Miller 视图的方式与 crumb 跳转完全一致——继续键入即下潜、删掉末段即上退——全程不必离开编辑器。一种落地形态、两个调用方：提交的路径关闭编辑器并呈现失败，草稿跟随扫描则两者都不做。该扫描是推测性的——键入到一半的目录多数时候读不出来——因此失败时保留最后一次可读的分栏并保持沉默。Enter 仍是权威提交：自提交至落地由它独占视图（同一批按键武装的防抖计时器会被扣住，而不是顶掉这次导航；提交被拒后仍扣住，直到下一次编辑），也只有它把失败呈现出来。有两点是刻意为之。等待以草稿为键，而非以它指名的目录部分为键，于是顶掉在飞扫描的那次按键会重新武装等待，被拒提交之后的编辑也能释放那道扣留；而它读取的分栏是 ref 而非依赖，否则落地会重新武装等待，遇到以不同拼写作答的宿主便会永远扫描下去。以及，走过的路不回退：关闭编辑器——包括取消——都把分栏留在草稿带到的地方，由面包屑指明、Open 的兜底目标随之而动，因为操作者亲眼看着它们移动。若落地卸载了键盘操作者 Tab 停留的那一行，焦点会被重新停回编辑器——Modal 并没有焦点陷阱。 - **导航以选中项为锚、安静且有界地落地。** 在展示根之外（与 crumb 头部渲染的是同一塌缩，因此 crumb 与分栏形态永不相左），落地即双栏：重新选中目标在父层级中的实际条目（Windows 上按平台惯例折叠大小写），右侧展示其子项，因此 crumb 跳转读作后退一栏，而不是塌缩成单列。父层级这一程在 200ms 等待上限内落定时，目标与父层级两程以**同一帧**落地——在此之前陈旧视图持续渲染，导航换栏时因此没有中间的单栏闪现——超出该上限则目标即刻单独提交（Enter 提交的导航绝不会被滞塞的父层级扣作人质），迟到的父层级这一程再就地升级这次落地。父层级这一程在落地的 supersession 范围下运行，任何较新的意图都会在线上将其中止（因此在落地窗口内按 Escape 即撤回整次导航）；父层级这一程失败，或被截断的父窗口缺少目标时，都保留单栏落地——升级的存在正是为了锚定选中项，绝不能反而让它悬空。加载指示器遵循同一安静规则：它浮于内容右下角（绝不是会挪动布局的一行；截断／错误行占据左下角，并在扫描期间持续渲染），且仅在扫描超出 300ms 静默窗口后才出现，因此本地列举切换时什么也不显示。行选取被刻意豁免于同一帧规则：选取后立即分栏本身就是其选中态反馈（aria-current、crumb 跟随），而导航除了换栏本身没有任何东西可确认这次点击。两个时序常量都按本地列举校准；远程部署（每层级一次 RPC，通常 100–400ms）会落在静默窗口之内、crumb 上却没有按下态——待远程消费方落地时，重新审视该窗口或补上按下反馈。 - **符号链接：为可进入性而跟随。** 用 `stat` 探测符号链接（断链／循环→跳过）；面包屑保留操作者导航的逻辑路径，`workspace.create` 在接纳时本就做 realpath 规范化。 - **列举层级有上限，且流式处理。** 单次 `list` 至多返回 `maxEntries` 行（配置项，默认 1000——GitHub 网页端目录列举的同一上限）。层级经 `opendir` 流入一个按名排序、容量 `maxEntries + 1` 的候选窗口，内存保持 O(maxEntries)，可进入性探测只触及窗口内候选；线上 `DirectoryListing` 携带必填的 `truncated` 标志，让客户端明示不完整而不是静默缺尾。窗口内的断链符号链接不从窗口外回填——发生过驱逐本身已把层级标记为截断。窗口插入为二分查找、满窗尾部单次比较即拒绝（超大层级不能为每个 dirent 付出一次全窗扫描），且 `list(path, signal)` 透传载体的请求信号，滞塞网络目录的扫描不会在调用方断连后继续存活——扫描中的每个 await（打开、每次读取、每次符号链接探测）都与信号赛跑，中止路径放弃而非等待 close（Node 会把 close 排在在飞读取之后），被放弃的 settlement 全部吞掉，清理不会以未处理拒绝的形式冒出。无上限的层级对超大或恶意构造的目录就是内存／响应性漏洞。 diff --git a/apps/web/tests/workspace-management.e2e.ts b/apps/web/tests/workspace-management.e2e.ts index 8eec0bc31c..b16e45184a 100644 --- a/apps/web/tests/workspace-management.e2e.ts +++ b/apps/web/tests/workspace-management.e2e.ts @@ -416,17 +416,18 @@ describe('web e2e: workspace management (create / rename / flat view / hover aff await expect.poll(() => dialog.getByText('alpha', { exact: true }).count(), { timeout: 10_000 }).toBe(1) await dialog.getByRole('button', { name: 'Edit path' }).click() const path = dialog.getByLabel('Edit path') - // A directory part no pane lists: the panes follow it and keep the editor. + // A directory part no pane lists: the panes walk to it, landing the + // ordinary two-pane Miller view (level | its children) with the editor + // still up and the draft intact. await path.fill(`${join(staged, 'alpha')}${sep}`) await expect.poll(() => dialog.getByText('only-under-alpha', { exact: true }).count(), { timeout: 10_000 }).toBe(1) - // The editor is still up with the draft intact: the panes moved under it. + expect(await dialog.getByRole('list').count()).toBe(2) expect(await path.inputValue()).toBe(`${join(staged, 'alpha')}${sep}`) - // Erasing back past the separator steps the panes up, the tail filtering - // the level it returns to. + // Erasing back past the separator returns to a level already on screen: + // the tail filters it, no scan needed, both panes stay. await path.fill(`${staged}${sep}al`) - await expect.poll(() => dialog.getByText('alpha', { exact: true }).count(), { timeout: 10_000 }).toBe(1) - expect(await dialog.getByText('beta', { exact: true }).count()).toBe(0) - expect(await dialog.getByText('only-under-alpha', { exact: true }).count()).toBe(0) + await expect.poll(() => dialog.getByText('beta', { exact: true }).count(), { timeout: 10_000 }).toBe(0) + expect(await dialog.getByText('alpha', { exact: true }).count()).toBe(1) // A tail nobody matches is a name still being spelled: the level shows // whole instead of emptying under it. await path.fill(`${staged}${sep}zzz`) diff --git a/packages/host/directory-picker-browse/README.i18n.yaml b/packages/host/directory-picker-browse/README.i18n.yaml index 06979f1226..2d66b6593b 100644 --- a/packages/host/directory-picker-browse/README.i18n.yaml +++ b/packages/host/directory-picker-browse/README.i18n.yaml @@ -2,5 +2,5 @@ # side as of the last confirmed-consistent state. Both languages carry equal authority; # after editing either side, bring the other along and re-record with: # pnpm run verify-translation-pairing --write packages/host/directory-picker-browse/README.md -README.md: a559f7f23b74f694c25fb43e005cbaccd4024308 -README.zh.md: 321e27d7a1dc9eb9bf0d1dab37c1af42e8ad6519 +README.md: c0375331e0e82fd6864b2027e7e36e0c6cb9986a +README.zh.md: 91d35de8821414c095db2a7834309864b5df0cd6 diff --git a/packages/host/directory-picker-browse/README.md b/packages/host/directory-picker-browse/README.md index a559f7f23b..c0375331e0 100644 --- a/packages/host/directory-picker-browse/README.md +++ b/packages/host/directory-picker-browse/README.md @@ -6,7 +6,7 @@ The **in-app browsing backend** of the [directory-picker seam](../directory-pick Behavior facts: listings return **directories only**, name-sorted, with symlinks-to-directories followed (broken/cyclic links skipped — the probe `stat` failing means "not enterable") and a host-owned `hidden` flag (POSIX dot convention) left for the client to act on; `crumbs` is the root-to-target ancestor chain, the root crumb labeled by its full path (`/`, `C:\`); an absent `list` path means the host account's home directory. `createDirectory` is non-recursive (a missing parent is a real failure, not a level to invent) and validates the name as a single non-blank segment even when called directly, mirroring the wire schema's fence. Both primitives reject an explicit path that is not fully qualified — relative forms, and on Windows the rooted drive-less forms (`\foo`, `/foo`) and incomplete UNC prefixes (`\\`, `\\server`) that `isAbsolute` accepts — with `directory-unreadable`/`directory-create-failed`, instead of letting `resolve` rebase it under the host process cwd or current drive. One `list` call returns at most `maxEntries` rows (config, default 1000 — the bound GitHub's web UI applies to directory listings), and the level streams through a bounded window so memory stays O(maxEntries) no matter how many children the directory holds: a cut level keeps the name-sorted head, counts hidden rows against the bound, probes only windowed candidates, and reports `truncated: true` so the client can say the level is incomplete (a windowed broken symlink is not backfilled from beyond the window — the eviction already marks the level truncated); window insertion is binary with an O(1) full-window tail rejection, and `list` threads the caller's `AbortSignal` so a disconnect or timeout stops the scan instead of letting it outlive the caller. Failures throw the seam's typed `DirectoryPickerError`. Policy rationale: [the directory-picker capability seam Agent Note](../../../.agents/notes/implemented/architecture/2026-07-28-directory-picker-capability-seam.md). -**Dual-face package**: the browser half (`./client`) fills [ui-workspace's](../../client/ui-workspace/README.md) two directory-flow holes with the in-app **Select Workspace Directory** dialog (figma `Harness` 813-23126 family — Miller two-column view whose navigations land selection-anchored and quiet: the previous view keeps rendering while a crumb jump or a submitted path is scanned (a "Loading…" pill floats over it only once the scan outlives a 300ms silence window, never shifting the columns), then target and parent legs land as one two-pane frame with the target re-selected as its actual parent-level entry — so stepping back never collapses and no intermediate frame flashes (a parent leg outliving its 200ms wait bound lands the target alone and upgrades in place; a failed or truncated parent leg keeps the single-pane landing; the display root keeps the single wide level); breadcrumb with a click-to-edit path zone, advertised by the pencil glyph at the bar's right edge and lit on hover in the editor's own footprint, whose editor seeds a trailing separator and then keeps the panes under the draft: the final segment prefix-filters the level its directory part names (case-insensitively, over the listed — possibly truncated — rows only; a tail nobody matches releases the filter instead of emptying the pane), while a directory part no pane lists is scanned after a 250ms rest and shown in place, so typing deeper descends and erasing segments steps back up without leaving the editor — a speculative scan is silent when it fails, and Enter still navigates by the exact text, owning the view until it lands; the editor cancels on Escape or when focus leaves the dialog card (window/tab switches and in-card focus moves keep the draft), and panes the draft walked to stay where the walk ended — the crumbs name that level and Open's fallback target follows them, so cancelling closes the editor rather than rewinding the walk; a fixed-label show-hidden footer toggle over the host's `hidden` flags, with a dot-led typed prefix revealing its matches and the current selection exempt from every filter; nested New-folder dialog), driving `host.listDirectory`/`host.createDirectory` and registering its own locale namespace (`directory-browser`, zh default / en). One cordis.yml row therefore composes both sides of the browse interaction; the client carries no capability-kind branching, and mounting a second flow package fails at load (the holes are `single` kind). +**Dual-face package**: the browser half (`./client`) fills [ui-workspace's](../../client/ui-workspace/README.md) two directory-flow holes with the in-app **Select Workspace Directory** dialog (figma `Harness` 813-23126 family — Miller two-column view whose navigations land selection-anchored and quiet: the previous view keeps rendering while a crumb jump or a submitted path is scanned (a "Loading…" pill floats over it only once the scan outlives a 300ms silence window, never shifting the columns), then target and parent legs land as one two-pane frame with the target re-selected as its actual parent-level entry — so stepping back never collapses and no intermediate frame flashes (a parent leg outliving its 200ms wait bound lands the target alone and upgrades in place; a failed or truncated parent leg keeps the single-pane landing; the display root keeps the single wide level); breadcrumb with a click-to-edit path zone, advertised by the pencil glyph at the bar's right edge and lit on hover in the editor's own footprint, whose editor seeds a trailing separator and then keeps the panes under the draft: the final segment prefix-filters the level its directory part names (case-insensitively, over the listed — possibly truncated — rows only; a tail nobody matches releases the filter instead of emptying the pane), while a directory part no pane lists is scanned after a 250ms rest and lands like any other navigation — selection-anchored, two-pane away from the display root — so typing deeper descends and erasing segments walks back up, moving the Miller view without leaving the editor (a level a pane already shows needs no scan at all: the filter alone answers the draft) — a speculative scan is silent when it fails, and Enter still navigates by the exact text, owning the view until it lands; the editor cancels on Escape or when focus leaves the dialog card (window/tab switches and in-card focus moves keep the draft), and panes the draft walked to stay where the walk ended — the crumbs name that level and Open's fallback target follows them, so cancelling closes the editor rather than rewinding the walk; a fixed-label show-hidden footer toggle over the host's `hidden` flags, with a dot-led typed prefix revealing its matches and the current selection exempt from every filter; nested New-folder dialog), driving `host.listDirectory`/`host.createDirectory` and registering its own locale namespace (`directory-browser`, zh default / en). One cordis.yml row therefore composes both sides of the browse interaction; the client carries no capability-kind branching, and mounting a second flow package fails at load (the holes are `single` kind). ## Model Experience diff --git a/packages/host/directory-picker-browse/README.zh.md b/packages/host/directory-picker-browse/README.zh.md index 321e27d7a1..91d35de882 100644 --- a/packages/host/directory-picker-browse/README.zh.md +++ b/packages/host/directory-picker-browse/README.zh.md @@ -6,7 +6,7 @@ 行为事实：列举**只返回目录**、按名称排序，指向目录的符号链接会被跟随（断链／循环链接被跳过——探测 `stat` 失败即"不可进入"），并携带宿主判定的 `hidden` 标志（POSIX 点前缀约定），展示决策留给客户端；`crumbs` 是从根到目标的祖先链，根 crumb 以完整路径标注（`/`、`C:\`）；`list` 不带路径即列举宿主账户的家目录。`createDirectory` 不递归（父目录缺失是真实失败，不是要补造的层级），且即便被直接调用也把名称校验为单个非空段，与协议 schema 的栅栏一致。两个原语都拒绝非完全限定的显式路径——相对形态，以及 Windows 上 `isAbsolute` 会放行的无盘符有根形态（`\foo`、`/foo`）与不完整的 UNC 前缀（`\\`、`\\server`）——报 `directory-unreadable`／`directory-create-failed`，而不是任由 `resolve` 把它重定位到宿主进程 cwd 或当前盘符之下。单次 `list` 至多返回 `maxEntries` 行（配置项，默认 1000——GitHub 网页端对目录列举采用的同一上限），且层级以流式方式经过一个有界窗口，无论目录有多少子项内存都保持 O(maxEntries)：被截断的层级保留按名排序的头部、隐藏行计入上限、只探测窗口内候选，并报告 `truncated: true`，供客户端提示层级不完整（窗口内的断链符号链接不会从窗口外回填——发生过驱逐本身已把层级标记为截断）；窗口插入为二分查找、满窗尾部单次比较即拒绝，且 `list` 透传调用方的 `AbortSignal`，断连或超时会停止扫描而不是让它在调用方离开后继续。失败抛出 seam 的类型化 `DirectoryPickerError`。策略依据：[目录选择能力 seam Agent Note](../../../.agents/notes/implemented/architecture/2026-07-28-directory-picker-capability-seam.md)。 -**双面包**：browser half（`./client`）以应用内 **选择工作区目录** 对话框（figma `Harness` 813-23126 家族——Miller 双列视图，其导航以选中项为锚、安静落地：扫描 crumb 跳转或提交的路径期间，先前视图持续渲染（"Loading…" 胶囊仅在扫描超出 300ms 静默窗口后才浮于其上，绝不挪动各列），随后目标与父层级两程以单个双栏帧落地，目标被重新选中为其在父层级中的实际条目——因此后退绝不塌缩，也没有中间帧闪现（父层级这一程超出其 200ms 等待上限时，目标单独落地，随后就地升级；父层级这一程失败或被截断时保持单栏落地；展示根保持单个宽层级）；带点击即编辑路径区的面包屑，该区由栏右端的铅笔图标点明、悬停时以编辑器自身的轮廓亮起，其编辑器预填尾随分隔符，随后让下方各栏跟随草稿：末段对其目录部分所指的层级做前缀过滤（不区分大小写，且仅作用于已列出、可能被截断的行；无一匹配的末段会解除过滤，而不是把该栏清空），而任何一栏都未列出的目录部分会在停顿 250ms 后被扫描并就地展示，于是继续键入即下潜、删掉末段即上退，全程不必离开编辑器——推测性扫描失败时保持沉默，而 Enter 仍按确切文本导航，并在落地前独占视图；编辑器按 Escape 或焦点离开对话框卡片即取消（窗口／标签页切换与卡片内焦点移动保留草稿），而草稿走到的层级会留在原地——面包屑指明该层级、Open 的兜底目标随之而动，因此取消只是关闭编辑器，并不回退这段行走；基于宿主 `hidden` 标志、标签固定的"显示隐藏"footer 开关，键入以点开头的前缀会显出其匹配项，且当前选中项不受任何过滤影响；嵌套新建文件夹对话框）填入 [ui-workspace](../../client/ui-workspace/README.md) 的两个目录流洞，驱动 `host.listDirectory`／`host.createDirectory`，并注册自己的 locale 命名空间（`directory-browser`，zh 默认／en）。因此一行 cordis.yml 同时组合浏览交互的两侧；client 侧不含任何能力 kind 分支，挂载第二个流程包会在加载期失败（洞为 `single` kind）。 +**双面包**：browser half（`./client`）以应用内 **选择工作区目录** 对话框（figma `Harness` 813-23126 家族——Miller 双列视图，其导航以选中项为锚、安静落地：扫描 crumb 跳转或提交的路径期间，先前视图持续渲染（"Loading…" 胶囊仅在扫描超出 300ms 静默窗口后才浮于其上，绝不挪动各列），随后目标与父层级两程以单个双栏帧落地，目标被重新选中为其在父层级中的实际条目——因此后退绝不塌缩，也没有中间帧闪现（父层级这一程超出其 200ms 等待上限时，目标单独落地，随后就地升级；父层级这一程失败或被截断时保持单栏落地；展示根保持单个宽层级）；带点击即编辑路径区的面包屑，该区由栏右端的铅笔图标点明、悬停时以编辑器自身的轮廓亮起，其编辑器预填尾随分隔符，随后让下方各栏跟随草稿：末段对其目录部分所指的层级做前缀过滤（不区分大小写，且仅作用于已列出、可能被截断的行；无一匹配的末段会解除过滤，而不是把该栏清空），而任何一栏都未列出的目录部分会在停顿 250ms 后被扫描，并像其他任何一次导航那样落地——以选中项为锚，在展示根之外即双栏——于是继续键入即下潜、删掉末段即上退，Miller 视图随之移动而不必离开编辑器（某一栏已经展示的层级则根本不需要扫描：过滤本身就答复了草稿）——推测性扫描失败时保持沉默，而 Enter 仍按确切文本导航，并在落地前独占视图；编辑器按 Escape 或焦点离开对话框卡片即取消（窗口／标签页切换与卡片内焦点移动保留草稿），而草稿走到的层级会留在原地——面包屑指明该层级、Open 的兜底目标随之而动，因此取消只是关闭编辑器，并不回退这段行走；基于宿主 `hidden` 标志、标签固定的"显示隐藏"footer 开关，键入以点开头的前缀会显出其匹配项，且当前选中项不受任何过滤影响；嵌套新建文件夹对话框）填入 [ui-workspace](../../client/ui-workspace/README.md) 的两个目录流洞，驱动 `host.listDirectory`／`host.createDirectory`，并注册自己的 locale 命名空间（`directory-browser`，zh 默认／en）。因此一行 cordis.yml 同时组合浏览交互的两侧；client 侧不含任何能力 kind 分支，挂载第二个流程包会在加载期失败（洞为 `single` kind）。 ## 模型体验 diff --git a/packages/host/directory-picker-browse/src/client/DirectoryBrowser.tsx b/packages/host/directory-picker-browse/src/client/DirectoryBrowser.tsx index e3a13cc54d..395793268c 100644 --- a/packages/host/directory-picker-browse/src/client/DirectoryBrowser.tsx +++ b/packages/host/directory-picker-browse/src/client/DirectoryBrowser.tsx @@ -24,10 +24,12 @@ * prefix-filters the level its directory part names (a dot-led prefix also * reveals the hidden entries it names, and a prefix nobody matches releases * the filter), while a directory part no pane lists is scanned after a short - * debounce and shown in place — so typing deeper descends and erasing - * segments steps back up without leaving the editor. Panes the draft walked - * to stay put when the editor closes (cancellation included): the crumbs name - * where the walk ended, and Open's fallback target follows them. + * debounce and lands like any other navigation — selection-anchored and + * two-pane away from the display root — so typing deeper descends and + * erasing segments walks back up, moving the Miller view without leaving the + * editor. Panes the draft walked to stay put when the editor closes + * (cancellation included): the crumbs name where the walk ended, and Open's + * fallback target follows them. */ import { useCallback, useEffect, useRef, useState } from 'react' import clsx from 'clsx' @@ -334,25 +336,65 @@ export function DirectoryBrowser({ open, listDirectory, createDirectory, onOpen, }, [restartSlowScanWindow, listDirectory]) /** - * Replace the whole view with a freshly navigated level. Away from the + * Enter owns the view from submission until its navigation lands, so the + * debounce timer the same keystrokes armed must not supersede it. Cleared + * by the next edit (and by opening the editor); a failed submission leaves + * it set until the operator edits again, so the rejected path is not + * immediately re-scanned as a preview. + */ + const previewSuspended = useRef(false) + + // The panes as the draft-following scan must read them when its wait + // fires: current, but NOT a dependency of the wait (see the effect below). + const viewRef = useRef<{ parent: DirectoryListing | null; child: DirectoryListing | null }>({ parent: null, child: null }) + useEffect(() => { viewRef.current = { parent, child } }, [parent, child]) + + /** + * A landed preview replaced the pane a keyboard operator may have Tabbed + * onto, so the focus it drops is re-parked on the still-open editor (the + * Modal has no focus trap). Consumed by the refocus effect below. + */ + const refocusPathInput = useRef(false) + + /** + * Replace the whole view with a freshly scanned level. Away from the * display root — the same collapse the crumb header renders, so crumbs and * pane shape never disagree — the landing is two-pane: the target's ACTUAL * parent-level entry re-selected (left pane = parent, right pane = the * target), so a crumb jump reads as stepping back one pane. Both legs land * as one frame when the parent leg settles within * {@link PARENT_LEG_WAIT_MS}; past that bound (or at the display root) the - * target commits alone — single wide level, the editor closes, loading - * ends — and a late parent leg still upgrades the landing in place. A - * failed parent leg, or a truncated parent window that lacks the target, - * leaves the single-pane landing — the upgrade must never orphan the - * selection it exists to anchor. Until whichever commit comes first, the - * previous view keeps rendering: navigation swaps the panes, it never - * blanks them. + * target commits alone — single wide level, loading ends — and a late + * parent leg still upgrades the landing in place. A failed parent leg, or a + * truncated parent window that lacks the target, leaves the single-pane + * landing — the upgrade must never orphan the selection it exists to + * anchor. Until whichever commit comes first, the previous view keeps + * rendering: a landing swaps the panes, it never blanks them. + * + * Two callers, one landing shape. A submitted path (Enter, a crumb) closes + * the editor on arrival and announces its failure; the editor's own + * draft-following scan keeps both to itself — it is speculative, so a + * failure leaves the last readable panes standing and says nothing, while + * an arrival clears the stale message and re-parks focus the swap dropped. + * @param path - the level to list; absent lists the Host home directory. + * @param options - `closeEditor` retires the path draft on arrival; + * `announce` surfaces a failure as the dialog's alert. */ - const navigate = useCallback((path?: string) => { + const land = useCallback((path: string | undefined, options: { closeEditor: boolean; announce: boolean }) => { const { seq, scan } = launchListing(path) setLoading(true) - setError(null) + if (options.announce) setError(null) + // What every landing does once its panes are committed, whichever shape + // committed them. + const settle = (): void => { + setLoading(false) + if (options.closeEditor) { + setPathDraft(null) + return + } + setError(null) + refocusPathInput.current = true + } scan.then((target) => { if (seq !== requestSeq.current) return // The single-pane landing; `landed` makes it first-commit-only, while @@ -364,8 +406,7 @@ export function DirectoryBrowser({ open, listDirectory, createDirectory, onOpen, setParent(target) setSelected(null) setChild(null) - setLoading(false) - setPathDraft(null) + settle() } // Arity is label-independent: only the collapsed chain's depth decides. if (displayCrumbs(target, '').length < 2) { landSingle(); return } @@ -386,10 +427,8 @@ export function DirectoryBrowser({ open, listDirectory, createDirectory, onOpen, setChild(target) // Idempotent on a late upgrade of a timed-out landing: reopening the // editor or starting a newer scan supersedes this seq, so reaching - // here means the draft is closed and the loading flag is this - // navigation's own. - setLoading(false) - setPathDraft(null) + // here means the settlement is still this landing's own. + settle() }, () => { // The parent-leg failure (its abort included) never surfaces: the // target listed fine, and nobody asked to see the parent level. @@ -399,10 +438,15 @@ export function DirectoryBrowser({ open, listDirectory, createDirectory, onOpen, }, (reason: unknown) => { if (seq !== requestSeq.current) return setLoading(false) - setError(failureText(reason)) + if (options.announce) setError(failureText(reason)) }) }, [launchListing, continueScan]) + /** Commit a submitted path (Enter, a crumb, the initial home listing): the editor closes, failures surface. */ + const navigate = useCallback((path?: string) => { + land(path, { closeEditor: true, announce: true }) + }, [land]) + // Editor-close focus parking (consumed by the refocus effect below the // miller-row ref): a pick parks on the selection's row, Enter and an // input-focused Escape park on the crumb edit zone that replaces the @@ -452,53 +496,15 @@ export function DirectoryBrowser({ open, listDirectory, createDirectory, onOpen, }, [launchListing, pathDraft]) /** - * Enter owns the view from submission until its navigation lands, so the - * debounce timer the same keystrokes armed must not supersede it. Cleared - * by the next edit (and by opening the editor); a failed submission leaves - * it set until the operator edits again, so the rejected path is not - * immediately re-scanned as a preview. - */ - const previewSuspended = useRef(false) - - // The panes as the draft-following scan must read them when its wait - // fires: current, but NOT a dependency of the wait (see the effect below). - const viewRef = useRef<{ parent: DirectoryListing | null; child: DirectoryListing | null }>({ parent: null, child: null }) - useEffect(() => { viewRef.current = { parent, child } }, [parent, child]) - - /** - * A landed preview replaced the pane a keyboard operator may have Tabbed - * onto, so the focus it drops is re-parked on the still-open editor (the - * Modal has no focus trap). Consumed by the refocus effect below. - */ - const refocusPathInput = useRef(false) - - /** - * List the directory the draft addresses and show it WITHOUT closing the - * editor: the level replaces the panes single-wide (the selection and its - * child preview belonged to the level the draft left), and the draft's - * final segment prefix-filters it from the next render on. Unlike Enter, - * this is speculative — half-typed directories are unreadable most of the - * time — so a failure keeps the last readable panes and stays silent, - * leaving submission to surface the real error. A landing clears a stale - * error for the same reason: it, not the launch, is what makes the message - * obsolete. + * Walk the panes to the directory the draft addresses, WITHOUT closing the + * editor. The landing is an ordinary one — selection-anchored and two-pane + * away from the display root — so typing a path moves the Miller view + * exactly as a crumb jump does, and the draft's final segment + * prefix-filters the arrival from the next render on. */ const previewDraftLevel = useCallback((directory: string) => { - const { seq, scan } = launchListing(directory) - setLoading(true) - scan.then((level) => { - if (seq !== requestSeq.current) return - setParent(level) - setSelected(null) - setChild(null) - setLoading(false) - setError(null) - refocusPathInput.current = true - }, () => { - if (seq !== requestSeq.current) return - setLoading(false) - }) - }, [launchListing]) + land(directory, { closeEditor: false, announce: false }) + }, [land]) /** Abandon path editing (Escape or clicking away) and restore the crumb view. */ const cancelPathEdit = useCallback(() => { diff --git a/packages/host/directory-picker-browse/tests/directory-browser.spec.tsx b/packages/host/directory-picker-browse/tests/directory-browser.spec.tsx index d94029e3bf..d6d6a3903c 100644 --- a/packages/host/directory-picker-browse/tests/directory-browser.spec.tsx +++ b/packages/host/directory-picker-browse/tests/directory-browser.spec.tsx @@ -652,7 +652,7 @@ describe('DirectoryBrowser', () => { }) it('prefix-filters the listed level from the draft tail, dot revealing hidden matches', async () => { - mount() + const b = mount() await waitFor(() => { expect(screen.getByRole('listitem')).toBeTruthy() }) fireEvent.click(screen.getByRole('button', { name: 'browser.editPath' })) const input = screen.getByLabelText('browser.editPath') @@ -672,9 +672,17 @@ describe('DirectoryBrowser', () => { // Its dot-led reveal lapses with it. fireEvent.change(input, { target: { value: `${HOME}/.zzz` } }) expect(screen.getAllByRole('listitem').map(item => item.textContent)).toEqual(['Documents']) - // A draft naming some other directory (or none) leaves the level whole. + // A tail inside the listed level names no level to walk to: the wait + // fires and finds nothing to scan. + const settled = b.listDirectory.mock.calls.length + await act(async () => { await new Promise((resolve) => { setTimeout(resolve, 400) }) }) + expect(b.listDirectory.mock.calls).toHaveLength(settled) + // A draft naming some other directory (or none) leaves the level whole — + // and a draft with no separator at all addresses no directory either. fireEvent.change(input, { target: { value: 'no-separator' } }) expect(screen.getByRole('listitem').textContent).toBe('Documents') + await act(async () => { await new Promise((resolve) => { setTimeout(resolve, 400) }) }) + expect(b.listDirectory.mock.calls).toHaveLength(settled) }) it('filters the child pane in two-pane mode and follows the draft back up a level', async () => { @@ -706,29 +714,46 @@ describe('DirectoryBrowser', () => { expect(within(columns()[1]!).getByText('harness')).toBeTruthy() }) - it('follows the draft into a directory no pane lists, and back up when segments are erased', async () => { + it('follows the draft into a directory no pane lists, landing the two-pane Miller view', async () => { const b = mount() await waitFor(() => { expect(screen.getByRole('listitem')).toBeTruthy() }) + expect(columns()).toHaveLength(1) fireEvent.click(screen.getByRole('button', { name: 'browser.editPath' })) const input = screen.getByLabelText('browser.editPath') - // Typing past a separator addresses a level nobody shows: the panes - // follow it once the typing rests, and the tail filters the arrival. + // Typing past a separator addresses a level nobody shows: the panes walk + // to it once the typing rests, landing the ordinary selection-anchored + // two-pane view (level | its children) with the tail filtering the right + // pane — a typed path moves the Miller view exactly as a crumb jump does. fireEvent.change(input, { target: { value: `${DOCS}/h` } }) - await waitFor(() => { expect(screen.getByText('harness')).toBeTruthy() }) - expect(b.listDirectory.mock.calls.at(-1)?.[0]).toBe(`${DOCS}/`) + await waitFor(() => { expect(columns()).toHaveLength(2) }) + expect(b.listDirectory).toHaveBeenCalledWith(`${DOCS}/`, expect.anything()) + expect(within(columns()[0]!).getByText('Documents')).toBeTruthy() + expect(within(columns()[1]!).getByText('harness')).toBeTruthy() // Still editing: the panes moved under the draft, the editor stayed. expect(screen.getByLabelText('browser.editPath').value).toBe(`${DOCS}/h`) - // Typing on inside the level the panes now list costs no scan at all: - // the prefix filter alone answers the draft. + // Typing on inside a level the panes already list costs no scan at all: + // the prefix filter alone answers the draft, both panes stay. const settled = b.listDirectory.mock.calls.length fireEvent.change(input, { target: { value: `${DOCS}/ha` } }) await act(async () => { await new Promise((resolve) => { setTimeout(resolve, 400) }) }) expect(b.listDirectory.mock.calls).toHaveLength(settled) - // Erasing back past the separator steps the panes up a level again. - fireEvent.change(input, { target: { value: `${HOME}/Do` } }) - await waitFor(() => { expect(screen.getByText('Documents')).toBeTruthy() }) - expect(b.listDirectory.mock.calls.at(-1)?.[0]).toBe(`${HOME}/`) - expect(columns()).toHaveLength(1) + expect(columns()).toHaveLength(2) + }) + + it('walks the panes back up when erased segments leave the listed levels', async () => { + const b = mount() + await waitFor(() => { expect(screen.getByRole('listitem')).toBeTruthy() }) + fireEvent.click(screen.getByRole('button', { name: 'browser.editPath' })) + const input = screen.getByLabelText('browser.editPath') + fireEvent.change(input, { target: { value: `${DOCS}/h` } }) + await waitFor(() => { expect(columns()).toHaveLength(2) }) + // Erasing back to a directory neither pane lists walks up to it; the + // filesystem root is the display root, so it lands the single wide level + // with the tail filtering it. + fireEvent.change(input, { target: { value: '/ho' } }) + await waitFor(() => { expect(columns()).toHaveLength(1) }) + expect(b.listDirectory).toHaveBeenCalledWith('/', expect.anything()) + expect(screen.getAllByRole('listitem').map(item => item.textContent)).toEqual(['home']) }) it('re-arms the draft-following scan after a keystroke superseded one in flight', async () => { @@ -778,12 +803,14 @@ describe('DirectoryBrowser', () => { await waitFor(() => { expect(screen.getByRole('listitem')).toBeTruthy() }) fireEvent.click(screen.getByRole('button', { name: 'browser.editPath' })) const input = screen.getByLabelText('browser.editPath') - fireEvent.change(input, { target: { value: `${DOCS}/h` } }) + // Two levels down, so the walk replaces the LEFT pane the focused row + // lives in (a landing that re-lists the same level reuses its rows). + fireEvent.change(input, { target: { value: `${HARNESS}/` } }) // The keyboard path: focus Tabbed onto a row of the level about to be // replaced. Without a re-park it would fall to body, outside a Modal that // has no focus trap. rowButton(screen.getByRole('listitem')).focus() - await waitFor(() => { expect(screen.getByText('harness')).toBeTruthy() }) + await waitFor(() => { expect(within(columns()[0]!).getByText('harness')).toBeTruthy() }) expect(document.activeElement).toBe(screen.getByLabelText('browser.editPath')) }) From 255cd90c6e04e3fd14ec54df4764106fbc1a1673 Mon Sep 17 00:00:00 2001 From: kingwl Date: Mon, 3 Aug 2026 13:23:40 +0800 Subject: [PATCH 26/48] compact long subagent durations --- ...07-27-web-subagent-conversations.i18n.yaml | 4 +- .../2026-07-27-web-subagent-conversations.md | 6 +- ...026-07-27-web-subagent-conversations.zh.md | 6 +- apps/web/tests/scaffold.ts | 9 ++- .../subagent-conversation/tree.expected.md | 2 +- apps/web/tests/subagent-conversation.e2e.ts | 16 ++-- packages/client/ui-subagent/README.i18n.yaml | 4 +- packages/client/ui-subagent/README.md | 2 +- packages/client/ui-subagent/README.zh.md | 2 +- .../src/client/SubagentCatalogAction.tsx | 81 ++++++++++++++++--- .../client/ui-subagent/src/client/locales.ts | 16 ++++ .../tests/conversation-ui.spec.tsx | 19 ++++- 12 files changed, 136 insertions(+), 31 deletions(-) diff --git a/.agents/notes/implemented/feature/2026-07-27-web-subagent-conversations.i18n.yaml b/.agents/notes/implemented/feature/2026-07-27-web-subagent-conversations.i18n.yaml index 08db3986b2..e0814009e3 100644 --- a/.agents/notes/implemented/feature/2026-07-27-web-subagent-conversations.i18n.yaml +++ b/.agents/notes/implemented/feature/2026-07-27-web-subagent-conversations.i18n.yaml @@ -2,5 +2,5 @@ # side as of the last confirmed-consistent state. Both languages carry equal authority; # after editing either side, bring the other along and re-record with: # pnpm run verify-translation-pairing --write .agents/notes/implemented/feature/2026-07-27-web-subagent-conversations.md -2026-07-27-web-subagent-conversations.md: 09a444dc6c391f3abdfc5bfe0f3367d4bda430c3 -2026-07-27-web-subagent-conversations.zh.md: 87929e8b35212092587ca4566bd10a1242c84e30 +2026-07-27-web-subagent-conversations.md: a8c7c4716fe797f94193b9ee7560394b0e0ff54a +2026-07-27-web-subagent-conversations.zh.md: 09980b30acf84379ae99e8d30122fcd69f57af75 diff --git a/.agents/notes/implemented/feature/2026-07-27-web-subagent-conversations.md b/.agents/notes/implemented/feature/2026-07-27-web-subagent-conversations.md index 09a444dc6c..a8c7c4716f 100644 --- a/.agents/notes/implemented/feature/2026-07-27-web-subagent-conversations.md +++ b/.agents/notes/implemented/feature/2026-07-27-web-subagent-conversations.md @@ -33,7 +33,7 @@ The Figma [subagent list](https://www.figma.com/design/jRBBK7zBgcszdVWQ0Fh5J8/Ha | The session header opens a compact child list. | The trigger aggregates the complete subagent-only descendant lineage; the tree shows every direct catalog entry in service order, including disabled diagnostics. | | Selecting a row reuses the conversation UI. | Addressed history never activates the child; only a continuable row with a live parent retains the ordinary composer. | | Nested agents expand progressively. | Each row carries a one-level `hasChildren` snapshot; disclosure reserves known direct-descendant rows immediately, then loads only that row's direct catalog and retains its own parent address. | -| Rows show labels, state, and active duration without duplicating sidebar rows. | Mode and `running`/`inactive` activity are textual as well as visual; optional title and exact active-turn duration come from the list's retained projection values. `SessionHeader.origin` removes duplicate navigation rows but grants no capability. | +| Rows show labels, state, and active duration without duplicating sidebar rows. | Mode and `running`/`inactive` activity are textual as well as visual; optional title and active-turn duration come from the list's retained projection values. Compact duration loses smaller units above one day, while hover and accessible naming retain exact whole seconds. `SessionHeader.origin` removes duplicate navigation rows but grants no capability. | ## Product contract @@ -41,7 +41,7 @@ The header action is absent only when a complete empty direct-catalog response a `running` means the exact child Agent driver is draining work at the Host sampling boundary; `inactive` means that driver is idle or absent. The UI does not translate either value into success, failure, cancellation, completeness, or resumability. `subagent.list` supplies the current driver-status baseline, `host/session-status` updates known activity in place, request-local replay prevents an older in-flight list response from overwriting a newer transition, and `host/session-removed` returns a known row to `inactive`; reconnect reads a fresh baseline. A `host/session-added` frame for a direct subagent immediately flips any loaded parent row to `hasChildren: true`, and that positive hint survives an older in-flight catalog response; membership, labels, mode, diagnostics, and the authoritative snapshot still require a debounced `subagent.list` refresh while the affected branch is open. A prompt response remains delivery-time authority. -Healthy rows reuse the standard session projections retained in the list mirror. `subagentTiming` resets at every descriptor so an inherited fork seed cannot enter the child's total, accumulates completed `turn/start` → `turn/end` spans, and carries same-cut `active.since` and `active.through` bounds for an open turn. The menu formats whole seconds and advances its local clock only while a known descendant is running; an inactive row bounds an interrupted open turn with `active.through`, so a stale projection never borrows newer session metadata and reopening the menu never restarts completed work. The duration does not imply a durable outcome. +Healthy rows reuse the standard session projections retained in the list mirror. `subagentTiming` resets at every descriptor so an inherited fork seed cannot enter the child's total, accumulates completed `turn/start` → `turn/end` spans, and carries same-cut `active.since` and `active.through` bounds for an open turn. Below one day the menu formats whole seconds; longer visual values retain at most two adjacent units, using approximate 30-day months and 365-day years, while hover and accessible naming preserve the exact day/hour/minute/second duration. The menu advances its local clock only while a known descendant is running; an inactive row bounds an interrupted open turn with `active.through`, so a stale projection never borrows newer session metadata and reopening the menu never restarts completed work. The duration does not imply a durable outcome. Selecting a row records its exact address before opening the resident client `Session`. History pagination, event folding, tool render intents, titles, and live mux reconciliation reuse the ordinary conversation machinery. Breadcrumbs use catalog labels, follow parent links only through `origin: 'subagent'` rows, include the first ordinary owner, and keep ordinary forks single-level. Forking an addressed subagent creates an ordinary fork with direct source lineage and attaches it to the nearest workspace-owning ancestor. The catalog is an ARIA tree with lazy ArrowRight/ArrowLeft disclosure, linear ArrowUp/ArrowDown navigation, Home/End, Escape, and focus restoration. @@ -104,7 +104,7 @@ The shipped Web composition mounts SQLite session query beside JSONL persistence - Host protocol tests pin schemas including required boolean expandability, id echoing, mode verification, non-activating history, exact-parent enforcement, FIFO admission receipts, cancellation, and sanitized failure mapping. - Generic Host tests pin attached and cold history and forks without Agent publication, cold projection folding, descriptor/origin/runtime-owner denial, explicit-id adoption denial, and the direct queue-control fence. - Client object tests pin retained and restored addresses, one-shot read-only rejection, history routing, continuable prompt routing, no addressed cancellation, suppression of Agent-bound model controls, live activity flips including in-flight response replay and detach fallback, subagent-parent expandability flips, and membership refresh. -- jsdom tests pin the aggregate descendant count and activity, second-precision running and frozen inactive durations, the summary-backed root action across absent and stale-empty catalogs, known loading-row shape, mixed-mode rows, pre-click leaf disclosure, diagnostics, lazy descendant disclosure, direct-parent addresses, keyboard behavior, and both read-only reasons. +- jsdom tests pin the aggregate descendant count and activity, second-precision running and frozen inactive durations, adaptive long-duration units with exact accessible text, the summary-backed root action across absent and stale-empty catalogs, known loading-row shape, mixed-mode rows, pre-click leaf disclosure, diagnostics, lazy descendant disclosure, direct-parent addresses, keyboard behavior, and both read-only reasons. - The keyless assembled Web snapshot contains an inactive continuable child, an inactive one-shot sibling, and a persisted grandchild; it pins the three-descendant trigger across a stale empty catalog response, timing rows, and aggregate running transition, expands without activation, opens persisted history, admits a human FIFO follow-up, reconciles child mux events, and proves one-shot history remains read-only. - Navigation tests pin subagent-only breadcrumbs, workspace placement for forks created from subagents, and `origin: 'subagent'` sidebar filtering without hiding ordinary forks. diff --git a/.agents/notes/implemented/feature/2026-07-27-web-subagent-conversations.zh.md b/.agents/notes/implemented/feature/2026-07-27-web-subagent-conversations.zh.md index 87929e8b35..09980b30ac 100644 --- a/.agents/notes/implemented/feature/2026-07-27-web-subagent-conversations.zh.md +++ b/.agents/notes/implemented/feature/2026-07-27-web-subagent-conversations.zh.md @@ -33,7 +33,7 @@ Figma 中的 [subagent 列表](https://www.figma.com/design/jRBBK7zBgcszdVWQ0Fh5 | 会话页头可打开紧凑的 child 列表。 | 触发器会汇总仅含 subagent 的完整后代谱系；树按服务顺序显示每个直接目录条目，包括已禁用的 diagnostic。 | | 选择一行会复用对话 UI。 | 已寻址历史绝不激活 child；只有 parent 存活的可继续行才保留普通输入框。 | | 嵌套 agent 会逐层展开。 | 每行携带一层 `hasChildren` 快照；展开时会立即预留已知直接后代行，随后仍只加载该行的直接目录，并保留其自身的 parent 地址。 | -| 条目显示 label、状态与活跃耗时，同时避免侧边栏条目重复。 | mode 与 `running`／`inactive` 活动状态会同时以文字和视觉呈现；可选 title 与精确的活跃轮次耗时来自列表保留的投影值。`SessionHeader.origin` 会移除重复的导航条目，但不授予任何功能权限。 | +| 条目显示 label、状态与活跃耗时，同时避免侧边栏条目重复。 | mode 与 `running`／`inactive` 活动状态会同时以文字和视觉呈现；可选 title 与活跃轮次耗时来自列表保留的投影值。紧凑耗时从一天起省略更小的单位，而悬停和无障碍名称仍保留精确的整秒数。`SessionHeader.origin` 会移除重复的导航条目，但不授予任何功能权限。 | ## 产品契约 @@ -41,7 +41,7 @@ Figma 中的 [subagent 列表](https://www.figma.com/design/jRBBK7zBgcszdVWQ0Fh5 `running` 表示在 Host 采样边界，确切 child Agent driver 正在处理工作；`inactive` 表示该 driver 空闲或不存在。UI 不会把任一值解释为成功、失败、取消、完成状态或可恢复性。`subagent.list` 提供当前 driver 状态基线，`host/session-status` 会就地更新已知活动状态，请求内回放会阻止更早发起但尚未完成的列表响应覆盖较新的状态转换，`host/session-removed` 则会使已知行恢复为 `inactive`；重连时会读取新的基线。直接 subagent 的 `host/session-added` 帧会立即把任何已加载的 parent 行翻转为 `hasChildren: true`，并使这项正向提示不被更早发起但尚未完成的目录响应覆盖；受影响分支打开期间，成员、label、mode、diagnostic 与权威快照仍需要通过去抖动的 `subagent.list` 刷新来更新。消息投递时仍以提示词响应为权威依据。 -健康行会复用列表镜像中保留的标准会话投影。`subagentTiming` 会在每个描述符处重置，使继承的 fork 种子不会计入 child 总量；它会累加已完成的 `turn/start` → `turn/end` 时段，并携带未结束轮次同一切面的 `active.since` 和 `active.through` 边界。菜单会以整秒格式化时间，且仅在有已知后代处于运行状态时才推进其本地时钟；对 inactive 行，菜单以 `active.through` 为被中断未结束轮次的上界，因此陈旧投影绝不会借用更新的会话元数据，且重新打开菜单绝不会让已完成工作重新计时。该耗时不蕴含持久化结果语义。 +健康行会复用列表镜像中保留的标准会话投影。`subagentTiming` 会在每个描述符处重置，使继承的 fork 种子不会计入 child 总量；它会累加已完成的 `turn/start` → `turn/end` 时段，并携带未结束轮次同一切面的 `active.since` 和 `active.through` 边界。不足一天时，菜单会以整秒格式化时间；达到一天后的视觉值最多保留两个相邻单位，其中月份按近似 30 天计算，年份按近似 365 天计算，而悬停信息与无障碍名称会保留精确的天／小时／分钟／秒耗时。菜单仅在有已知后代处于运行状态时才推进其本地时钟；对 inactive 行，菜单以 `active.through` 为被中断未结束轮次的上界，因此陈旧投影绝不会借用更新的会话元数据，且重新打开菜单绝不会让已完成工作重新计时。该耗时不蕴含持久化结果语义。 选择一行后，系统会先记录其确切地址，再打开常驻客户端 `Session`。历史分页、事件 fold、工具渲染意图、title 与实时 mux 归并都会复用普通对话机制。面包屑导航使用目录 label，只会沿 `origin: 'subagent'` 行的父链接逐级回溯，包含第一个普通 owner，并让普通 fork 保持单层。从已寻址 subagent 创建 fork 时，会生成具有直接源谱系的普通 fork，并将其附加到最近拥有 Workspace 的祖先。目录是一棵 ARIA 树，支持懒加载式 ArrowRight／ArrowLeft 展开与折叠、线性 ArrowUp／ArrowDown 导航、Home／End、Escape 以及焦点恢复。 @@ -104,7 +104,7 @@ one-shot 行始终会用文案替代输入框，说明执行记录为只读。 - 宿主协议测试固定 schema（包括必需的布尔可展开性）、id 回显、mode 校验、非激活式历史、确切 parent 强制要求、FIFO 准入回执、取消与脱敏后的失败映射。 - 通用 Host 测试固定在不发布 Agent 的情况下读取已附加与冷态历史及执行 fork、冷态投影归并、按描述符／origin／运行时 owner 拒绝、拒绝显式 id 接纳，以及直接队列控制栅栏。 - 客户端对象测试固定已保留与已恢复的地址、one-shot 只读拒绝、历史路由、可继续提示词路由、已寻址对话不提供取消、屏蔽绑定到 agent 的模型控件、实时活动状态翻转（包括在途响应回放与 detach 回退）、subagent parent 可展开性翻转与成员刷新。 -- jsdom 测试固定后代聚合计数与活动状态、精确到秒的运行中耗时与冻结后 inactive 耗时、目录缺失或为陈旧空目录时由摘要支撑的根操作、已知加载行的形态、混合 mode 行、点击前的叶子展开控件、diagnostic、后代懒加载展开、直接 parent 地址、键盘行为与两种只读原因。 +- jsdom 测试固定后代聚合计数与活动状态、精确到秒的运行中耗时与冻结后 inactive 耗时、采用自适应单位的长耗时及其精确无障碍文本、目录缺失或为陈旧空目录时由摘要支撑的根操作、已知加载行的形态、混合 mode 行、点击前的叶子展开控件、diagnostic、后代懒加载展开、直接 parent 地址、键盘行为与两种只读原因。 - 无密钥的组装 Web 快照包含一个 inactive 的可继续 child、一个 inactive 的 one-shot sibling 和一个持久化 grandchild；它会固定触发器在一次陈旧的空目录响应后仍显示三个后代，并固定计时行以及聚合 `running` 状态转换，在不激活的情况下展开、打开持久化历史、准入一条用户 FIFO 后续消息、归并 child mux 事件，并证明 one-shot 历史仍然只读。 - 导航测试固定仅含 subagent 的面包屑导航、从 subagent 创建 fork 时的 Workspace 归属，以及 `origin: 'subagent'` 侧边栏过滤，同时不隐藏普通 fork。 diff --git a/apps/web/tests/scaffold.ts b/apps/web/tests/scaffold.ts index 1b8afbb247..11e834ad0d 100644 --- a/apps/web/tests/scaffold.ts +++ b/apps/web/tests/scaffold.ts @@ -511,7 +511,14 @@ function normalizeAria(snapshot: string, workspaceCwd: string): string { .split(workspaceCwd).join('{{cwd}}') .split(base).join('{{workspace}}') .replace(/[0-9a-f]{8}-[0-9a-f]{4}-[0-9a-f]{4}-[0-9a-f]{4}-[0-9a-f]{12}/gi, '{{uuid}}') - .replace(/\b\d+(?:\.\d+)?(?:ms|s|秒)\b/g, '{{duration}}') + .replace( + /~\d+(?:y(?: \d+mo)?|mo(?: \d+d)?)|\b(?:\d+d(?: \d+h(?: \d+m \d+s)?)?|\d+h \d+m \d+s|\d+m \d+s|\d+s|\d+(?:\.\d+)?ms)\b/g, + duration => duration.startsWith('~') ? duration : '{{duration}}', + ) + .replace( + /约\d+(?:年(?:\d+个月)?|个月(?:\d+天)?)|\d+(?:天(?:\d+小时(?:\d+分\d+秒)?)?|小时\d+分\d+秒|分\d+秒|秒)/g, + duration => duration.startsWith('约') ? duration : '{{duration}}', + ) // Message IconActions clocks widen by calendar day/year; collapse every // shape so goldens stay stable across midnight and year boundaries. .replace(/\d{4}年\d{1,2}月\d{1,2}日 \d{2}:\d{2}/g, '{{clock}}') diff --git a/apps/web/tests/snapshots/subagent-conversation/tree.expected.md b/apps/web/tests/snapshots/subagent-conversation/tree.expected.md index c1174c042e..ccbfa608bb 100644 --- a/apps/web/tests/snapshots/subagent-conversation/tree.expected.md +++ b/apps/web/tests/snapshots/subagent-conversation/tree.expected.md @@ -1,8 +1,8 @@ - tree "Subagent sessions": + - treeitem "event-sourcing reviewer one-shot · not running {{duration}}" [level=1]: event-sourcing reviewer one-shot · not running ~6mo 12d - treeitem "event-sourcing researcher Explain event sourcing in one · continuable · not running {{duration}}" [expanded] [level=1]: - button "Collapse event-sourcing researcher descendants": - img - text: event-sourcing researcher Explain event sourcing in one · continuable · not running {{duration}} - group: - treeitem "example editor continuable · not running {{duration}}" [level=2] - - treeitem "event-sourcing reviewer one-shot · not running {{duration}}" [level=1] diff --git a/apps/web/tests/subagent-conversation.e2e.ts b/apps/web/tests/subagent-conversation.e2e.ts index b719037867..ac2e4fe1f2 100644 --- a/apps/web/tests/subagent-conversation.e2e.ts +++ b/apps/web/tests/subagent-conversation.e2e.ts @@ -108,7 +108,8 @@ describe('web e2e: persisted subagent conversation and human continuation', () = childId = started.childId await waitForAgentToSettle(scaffold, childId) oneShotId = sessionId('recorded-one-shot') - const oneShotAt = Date.now() + const oneShotDurationMs = 192 * 24 * 60 * 60 * 1_000 + const oneShotAt = Date.now() - oneShotDurationMs await scaffold.ctx.sessionPersistence.create({ version: SESSION_FORMAT_VERSION, id: oneShotId, @@ -146,7 +147,7 @@ describe('web e2e: persisted subagent conversation and human continuation', () = { type: 'turn/end', seq: 3, - time: oneShotAt + 3, + time: oneShotAt + oneShotDurationMs, data: { turn: 1, reason: { kind: 'completed' } }, }, ] as SessionEvent[]) @@ -199,14 +200,14 @@ describe('web e2e: persisted subagent conversation and human continuation', () = expect(scaffold.ctx.agents.get(oneShotId)).toBeUndefined() expect(scaffold.ctx.agents.get(grandchildId)).toBeUndefined() await expect(scaffold.ctx.subagents.listChildren(parent.id)).resolves.toMatchObject([ - { - kind: 'child', id: childId, mode: 'continuable', label: LABEL, - activity: 'inactive', hasChildren: true, - }, { kind: 'child', id: oneShotId, mode: 'one-shot', label: ONE_SHOT_LABEL, activity: 'inactive', hasChildren: false, }, + { + kind: 'child', id: childId, mode: 'continuable', label: LABEL, + activity: 'inactive', hasChildren: true, + }, ]) await expect(scaffold.ctx.subagents.listChildren(childId)).resolves.toMatchObject([ { @@ -301,6 +302,9 @@ describe('web e2e: persisted subagent conversation and human continuation', () = expect(await page.getByRole('button', { name: `Expand ${ONE_SHOT_LABEL} descendants`, }).count()).toBe(0) + const oneShotRow = page.getByRole('treeitem', { name: new RegExp(ONE_SHOT_LABEL) }) + expect(await oneShotRow.getByText('~6mo 12d', { exact: true }).count()).toBe(1) + expect(await oneShotRow.getAttribute('aria-label')).toContain('192d 00h 00m 00s') await page.getByRole('button', { name: `Expand ${LABEL} descendants` }).click() const childRow = page.getByRole('treeitem', { name: new RegExp(LABEL) }) const childLabel = await childRow.getAttribute('aria-label') diff --git a/packages/client/ui-subagent/README.i18n.yaml b/packages/client/ui-subagent/README.i18n.yaml index 01e2412288..ea1012bc62 100644 --- a/packages/client/ui-subagent/README.i18n.yaml +++ b/packages/client/ui-subagent/README.i18n.yaml @@ -2,5 +2,5 @@ # side as of the last confirmed-consistent state. Both languages carry equal authority; # after editing either side, bring the other along and re-record with: # pnpm run verify-translation-pairing --write packages/client/ui-subagent/README.md -README.md: 33a2f2899fc34af52cda6b19f473847927da7ef0 -README.zh.md: cab1edb6b81df4b89d05153c801a4d277037ad9c +README.md: c2a005a9c9716d246bb4299ab1654314da21643a +README.zh.md: 4fb42b39fd7810f5d60e1b8a00bd9ad7d8752feb diff --git a/packages/client/ui-subagent/README.md b/packages/client/ui-subagent/README.md index 33a2f2899f..c2a005a9c9 100644 --- a/packages/client/ui-subagent/README.md +++ b/packages/client/ui-subagent/README.md @@ -4,7 +4,7 @@ English | [中文](README.zh.md) Web subagent feature owner: contributes the lazily expandable catalog tree to `conversation.session.header.actions`, reason-specific read-only replacements to the conversation composer chain, and the existing `@` reference source to `ctx.slash`. -The header action reads `subagentsByParent` and session summaries through the standard `useSessions` hook. After a non-empty direct catalog arrives, its trigger counts the complete subagent-only descendant lineage, stops at ordinary forks, and shows ongoing activity when any counted descendant is running. The compact tree remains direct-catalog authoritative: continuable and one-shot rows display mode plus `running`/`inactive` activity, an optional log-backed title, and active-turn duration to the second; an unlabeled one-shot row falls back to its session id, while corrupt, unsupported, or unavailable rows remain readable but disabled. Duration sums completed `subagentTiming` turns, advances once per second only for an open turn on a running child, and freezes after the child becomes inactive; an interrupted open turn is bounded by its same-cut `active.through`, never by newer session metadata. Each healthy row's `hasChildren` hint determines disclosure before interaction, so known leaves never show an arrow; expanding a branch immediately reserves one disabled loading row per known direct descendant, then lazily replaces them with that child's authoritative catalog. Every visible branch is reported to the runtime so membership frames cause a debounced refresh only where the tree is being consumed. Selecting any depth calls `SessionsService.openSubagent()` with the row's exact `{parentSessionId, childSessionId, mode}` address. Component-local state owns tree visibility, expanded branches, keyboard focus, and the running-duration clock. ArrowRight/ArrowLeft expand and collapse branches; ArrowUp/ArrowDown, Home, End, and Escape navigate or close the tree; closing returns focus to the trigger. Styling uses tokens only. +The header action reads `subagentsByParent` and session summaries through the standard `useSessions` hook. After a non-empty direct catalog arrives, its trigger counts the complete subagent-only descendant lineage, stops at ordinary forks, and shows ongoing activity when any counted descendant is running. The compact tree remains direct-catalog authoritative: continuable and one-shot rows display mode plus `running`/`inactive` activity, an optional log-backed title, and active-turn duration. Visual duration stays exact to the second below one day, then uses at most two adjacent units—days/hours, approximate months/days, or approximate years/months—while hover and the accessible name retain the exact day/hour/minute/second value. An unlabeled one-shot row falls back to its session id, while corrupt, unsupported, or unavailable rows remain readable but disabled. Duration sums completed `subagentTiming` turns, advances once per second only for an open turn on a running child, and freezes after the child becomes inactive; an interrupted open turn is bounded by its same-cut `active.through`, never by newer session metadata. Each healthy row's `hasChildren` hint determines disclosure before interaction, so known leaves never show an arrow; expanding a branch immediately reserves one disabled loading row per known direct descendant, then lazily replaces them with that child's authoritative catalog. Every visible branch is reported to the runtime so membership frames cause a debounced refresh only where the tree is being consumed. Selecting any depth calls `SessionsService.openSubagent()` with the row's exact `{parentSessionId, childSessionId, mode}` address. Component-local state owns tree visibility, expanded branches, keyboard focus, and the running-duration clock. ArrowRight/ArrowLeft expand and collapse branches; ArrowUp/ArrowDown, Home, End, and Escape navigate or close the tree; closing returns focus to the trigger. Styling uses tokens only. A one-shot child always elects a read-only composer that identifies the transcript as a completed execution record. A continuable child does so only when its exact parent is unavailable, with copy explaining the recovery path. A continuable child with a live parent keeps the ordinary input chrome, whose Session routes through `subagent.prompt`; running input remains Send because every follow-up joins the child's FIFO inbox, and addressed sessions never expose Stop. This package never receives host context or calls a model-facing tool. The catalog and composer behavior are specified by the [Web subagent conversations Agent Note](../../../.agents/notes/implemented/feature/2026-07-27-web-subagent-conversations.md). diff --git a/packages/client/ui-subagent/README.zh.md b/packages/client/ui-subagent/README.zh.md index cab1edb6b8..4fb42b39fd 100644 --- a/packages/client/ui-subagent/README.zh.md +++ b/packages/client/ui-subagent/README.zh.md @@ -4,7 +4,7 @@ Web subagent 功能 owner：向 `conversation.session.header.actions` 贡献可懒加载展开的目录树，向会话编辑器链贡献按原因区分的只读替代呈现，并保留注册到 `ctx.slash` 的既有 `@` 引用 source。 -页头操作通过标准 `useSessions` 钩子读取 `subagentsByParent` 与会话摘要。非空直接目录到达后，其触发器会统计仅含 subagent 的完整后代谱系，在普通 fork 处停止，并在任一计入统计的后代处于 `running` 时显示活动仍在进行。紧凑树仍以直接目录为权威依据：可继续和 one-shot 行会显示 mode、`running`／`inactive` 活动状态、由日志支撑的可选 title，以及精确到秒的活跃轮次耗时；没有 label 的 one-shot 行会回退到其会话 id，而损坏、不受支持或不可用的行仍保持可读但禁用。耗时会累加已完成的 `subagentTiming` 轮次，仅在运行中 child 存在未结束轮次时每秒递增一次，并在 child 变为 inactive 后冻结；被中断的未结束轮次以其同一切面的 `active.through` 为上界，绝不使用更新的会话元数据。每个健康行的 `hasChildren` 提示会在交互前决定是否显示展开控件，因此已知叶子节点从不显示箭头；展开分支时，会立即为每个已知直接后代预留一行禁用的加载行，随后再用该 child 的权威目录懒加载结果替换这些占位行。每个可见分支都会上报给运行时，使成员帧只在树正被消费的位置触发去抖动刷新。选择任意深度的条目都会使用该行的确切地址 `{parentSessionId, childSessionId, mode}` 调用 `SessionsService.openSubagent()`。组件局部状态负责树的可见性、已展开分支、键盘焦点与运行中耗时时钟。ArrowRight／ArrowLeft 展开和折叠分支；ArrowUp／ArrowDown、Home、End 与 Escape 用于导航或关闭树；关闭后焦点返回触发器。样式只使用 token。 +页头操作通过标准 `useSessions` 钩子读取 `subagentsByParent` 与会话摘要。非空直接目录到达后，其触发器会统计仅含 subagent 的完整后代谱系，在普通 fork 处停止，并在任一计入统计的后代处于 `running` 时显示活动仍在进行。紧凑树仍以直接目录为权威依据：可继续和 one-shot 行会显示 mode、`running`／`inactive` 活动状态、由日志支撑的可选 title，以及活跃轮次耗时。视觉耗时在不足一天时精确到秒，达到一天后则最多使用两个相邻单位——天／小时、近似月份／天或近似年份／月份——而悬停信息与无障碍名称会保留精确的天／小时／分钟／秒数值。没有 label 的 one-shot 行会回退到其会话 id，而损坏、不受支持或不可用的行仍保持可读但禁用。耗时会累加已完成的 `subagentTiming` 轮次，仅在运行中 child 存在未结束轮次时每秒递增一次，并在 child 变为 inactive 后冻结；被中断的未结束轮次以其同一切面的 `active.through` 为上界，绝不使用更新的会话元数据。每个健康行的 `hasChildren` 提示会在交互前决定是否显示展开控件，因此已知叶子节点从不显示箭头；展开分支时，会立即为每个已知直接后代预留一行禁用的加载行，随后再用该 child 的权威目录懒加载结果替换这些占位行。每个可见分支都会上报给运行时，使成员帧只在树正被消费的位置触发去抖动刷新。选择任意深度的条目都会使用该行的确切地址 `{parentSessionId, childSessionId, mode}` 调用 `SessionsService.openSubagent()`。组件局部状态负责树的可见性、已展开分支、键盘焦点与运行中耗时时钟。ArrowRight／ArrowLeft 展开和折叠分支；ArrowUp／ArrowDown、Home、End 与 Escape 用于导航或关闭树；关闭后焦点返回触发器。样式只使用 token。 one-shot child 始终选用只读编辑器，并将 transcript（文本记录）说明为已完成的执行记录。可继续 child 仅在其确切 parent 不可用时选用只读编辑器，并以文案说明恢复路径。确切 parent 存活时，可继续 child 保留普通输入 chrome，其 Session 会通过 `subagent.prompt` 路由；child 运行期间，输入操作仍为 Send，因为每条后续消息都会进入 child 的 FIFO inbox，且已寻址会话绝不公开 Stop。本包绝不接收宿主 context，也不调用面向模型的工具。目录与编辑器行为由 [Web subagent 对话 Agent Note](../../../.agents/notes/implemented/feature/2026-07-27-web-subagent-conversations.md)规定。 diff --git a/packages/client/ui-subagent/src/client/SubagentCatalogAction.tsx b/packages/client/ui-subagent/src/client/SubagentCatalogAction.tsx index eae3addf0e..42241245af 100644 --- a/packages/client/ui-subagent/src/client/SubagentCatalogAction.tsx +++ b/packages/client/ui-subagent/src/client/SubagentCatalogAction.tsx @@ -76,16 +76,54 @@ function activityDuration( return timing.settledMs + Math.max(0, end - timing.active.since) } -/** Format a non-negative duration to seconds without dropping larger units. */ -function formatDuration(ms: number, t: TranslateNS): string { +interface DurationParts { + seconds: number + minutes: number + hours: number + days: number + totalMinutes: number + totalHours: number +} + +function splitDuration(ms: number): DurationParts { const totalSeconds = Math.floor(Math.max(0, ms) / 1_000) - const seconds = totalSeconds % 60 const totalMinutes = Math.floor(totalSeconds / 60) - const minutes = totalMinutes % 60 - const hours = Math.floor(totalMinutes / 60) - if (hours > 0) { + const totalHours = Math.floor(totalMinutes / 60) + return { + seconds: totalSeconds % 60, + minutes: totalMinutes % 60, + hours: totalHours % 24, + days: Math.floor(totalHours / 24), + totalMinutes, + totalHours, + } +} + +/** Format a duration with decreasing visual precision at larger scales. */ +function formatDuration(ms: number, t: TranslateNS): string { + const { seconds, minutes, hours, days, totalMinutes, totalHours } = splitDuration(ms) + if (days >= 365) { + const years = Math.floor(days / 365) + const months = Math.floor((days % 365) / 30) + return months === 0 + ? t('duration.years', { years }) + : t('duration.yearsMonths', { years, months }) + } + if (days >= 30) { + const months = Math.floor(days / 30) + const remainingDays = days % 30 + return remainingDays === 0 + ? t('duration.months', { months }) + : t('duration.monthsDays', { months, days: remainingDays }) + } + if (days > 0) { + return hours === 0 + ? t('duration.days', { days }) + : t('duration.daysHours', { days, hours }) + } + if (totalHours > 0) { return t('duration.hours', { - hours, + hours: totalHours, minutes: String(minutes).padStart(2, '0'), seconds: String(seconds).padStart(2, '0'), }) @@ -99,6 +137,19 @@ function formatDuration(ms: number, t: TranslateNS): string { return t('duration.seconds', { seconds }) } +/** Preserve exact whole seconds for hover and accessible naming. */ +function formatExactDuration(ms: number, t: TranslateNS): string { + const { seconds, minutes, hours, days } = splitDuration(ms) + return days === 0 + ? formatDuration(ms, t) + : t('duration.exactDays', { + days, + hours: String(hours).padStart(2, '0'), + minutes: String(minutes).padStart(2, '0'), + seconds: String(seconds).padStart(2, '0'), + }) +} + /** Aggregate the complete subagent-only descendant subtree from flat summaries. */ function summarizeDescendants( sessionId: SessionId, @@ -231,7 +282,10 @@ function CatalogRows({ ) const duration = durationMs === undefined ? undefined - : formatDuration(durationMs, t) + : { + compact: formatDuration(durationMs, t), + exact: formatExactDuration(durationMs, t), + } const open = (): void => { openChild({ parentSessionId, childSessionId: entry.id, mode: entry.mode }) @@ -263,7 +317,7 @@ function CatalogRows({ role="treeitem" tabIndex={0} aria-level={level} - aria-label={[label, secondary, duration] + aria-label={[label, secondary, duration?.exact] .filter(value => value !== undefined) .join(' ')} {...knownLeaf ? {} : { 'aria-expanded': isExpanded }} @@ -290,7 +344,14 @@ function CatalogRows({ {label} {secondary} - {duration !== undefined && {duration}} + {duration !== undefined && ( + + {duration.compact} + + )}
{isExpanded && !knownLeaf && ( diff --git a/packages/client/ui-subagent/src/client/locales.ts b/packages/client/ui-subagent/src/client/locales.ts index 5eaee7b0f1..b3897d7216 100644 --- a/packages/client/ui-subagent/src/client/locales.ts +++ b/packages/client/ui-subagent/src/client/locales.ts @@ -11,6 +11,14 @@ export const zh = { 'duration.seconds': '{seconds}秒', 'duration.minutes': '{minutes}分{seconds}秒', 'duration.hours': '{hours}小时{minutes}分{seconds}秒', + 'duration.days': '{days}天', + 'duration.daysHours': '{days}天{hours}小时', + 'duration.months': '约{months}个月', + 'duration.monthsDays': '约{months}个月{days}天', + 'duration.years': '约{years}年', + 'duration.yearsMonths': '约{years}年{months}个月', + 'duration.exactDays': '{days}天{hours}小时{minutes}分{seconds}秒', + 'duration.exactTitle': '总活跃耗时：{duration}', 'loading.label': '正在加载子代理…', 'loading.aria': '正在加载子代理', 'load.error': '无法加载子代理', @@ -40,6 +48,14 @@ export const en: Record = { 'duration.seconds': '{seconds}s', 'duration.minutes': '{minutes}m {seconds}s', 'duration.hours': '{hours}h {minutes}m {seconds}s', + 'duration.days': '{days}d', + 'duration.daysHours': '{days}d {hours}h', + 'duration.months': '~{months}mo', + 'duration.monthsDays': '~{months}mo {days}d', + 'duration.years': '~{years}y', + 'duration.yearsMonths': '~{years}y {months}mo', + 'duration.exactDays': '{days}d {hours}h {minutes}m {seconds}s', + 'duration.exactTitle': 'Total active duration: {duration}', 'loading.label': 'Loading subagents…', 'loading.aria': 'Loading subagents', 'load.error': 'Unable to load subagents', diff --git a/packages/client/ui-subagent/tests/conversation-ui.spec.tsx b/packages/client/ui-subagent/tests/conversation-ui.spec.tsx index a6cbb22aeb..15fd71dbd9 100644 --- a/packages/client/ui-subagent/tests/conversation-ui.spec.tsx +++ b/packages/client/ui-subagent/tests/conversation-ui.spec.tsx @@ -242,12 +242,21 @@ describe('SubagentCatalogAction', () => { it('ticks active duration by seconds and freezes inactive rows', async () => { const now = 2_000_000_000_000 + const minute = 60_000 + const hour = 60 * minute + const day = 24 * hour vi.useFakeTimers() vi.setSystemTime(now) const rows = [ ['running', 'running', 65_000, now - 5_000, now - 1_000, now], ['finished', 'inactive', 3_723_000, undefined, undefined, now - 60_000], ['interrupted', 'inactive', 2_000, now - 7_000, now - 3_000, now + 60_000], + ['days', 'inactive', 12 * day + 5 * hour + 6 * minute + 7_000, undefined, undefined, now], + ['whole-day', 'inactive', day, undefined, undefined, now], + ['months', 'inactive', 192 * day, undefined, undefined, now], + ['whole-month', 'inactive', 30 * day, undefined, undefined, now], + ['years', 'inactive', 832 * day, undefined, undefined, now], + ['whole-year', 'inactive', 365 * day, undefined, undefined, now], ] as const const entries = rows.map(([id, activity]) => ({ kind: 'child' as const, @@ -278,11 +287,19 @@ describe('SubagentCatalogAction', () => { })) as Record const input = props(catalog({ entries }), {}, summaries) render() - fireEvent.click(screen.getByRole('button', { name: /3 个子代理/ })) + fireEvent.click(screen.getByRole('button', { name: /9 个子代理/ })) expect(screen.getByRole('treeitem', { name: /running.*1分10秒/ })).toBeTruthy() expect(screen.getByRole('treeitem', { name: /finished.*1小时02分03秒/ })).toBeTruthy() expect(screen.getByRole('treeitem', { name: /interrupted.*6秒/ })).toBeTruthy() + expect(screen.getByRole('treeitem', { name: /days.*12天05小时06分07秒/ })).toBeTruthy() + expect(screen.getByText('12天5小时').getAttribute('title')) + .toBe('总活跃耗时：12天05小时06分07秒') + expect(screen.getByText('1天')).toBeTruthy() + expect(screen.getByText('约6个月12天')).toBeTruthy() + expect(screen.getByText('约1个月')).toBeTruthy() + expect(screen.getByText('约2年3个月')).toBeTruthy() + expect(screen.getByText('约1年')).toBeTruthy() await vi.advanceTimersByTimeAsync(1_000) expect(screen.getByRole('treeitem', { name: /running.*1分11秒/ })).toBeTruthy() From 2ceed380dd1ecf60bf8317fa9f58dbfbb96bb053 Mon Sep 17 00:00:00 2001 From: creatixchu Date: Mon, 3 Aug 2026 13:27:45 +0800 Subject: [PATCH 27/48] fix(directory-picker-browse): keep the typed level in the last pane MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Skipping the draft-following scan whenever ANY pane happened to list the directory was the cheaper rule and the wrong one: erasing a segment left the level being typed on the LEFT, with its own child pane still standing to its right, so the two panes stopped reading as "where I am, and where I came from". The pane arity is now the invariant the editor maintains — the last pane lists the level the path names, its parent sits beside it, and only a display root lists alone. Only that last pane's own tail costs no scan; every other directory part re-lands. --- ...directory-picker-capability-seam.i18n.yaml | 4 +-- ...-07-28-directory-picker-capability-seam.md | 2 +- ...-28-directory-picker-capability-seam.zh.md | 2 +- apps/web/tests/workspace-management.e2e.ts | 9 +++-- .../directory-picker-browse/README.i18n.yaml | 4 +-- .../host/directory-picker-browse/README.md | 2 +- .../host/directory-picker-browse/README.zh.md | 2 +- .../src/client/DirectoryBrowser.tsx | 34 +++++++++++-------- .../tests/directory-browser.spec.tsx | 30 +++++++++++++--- 9 files changed, 59 insertions(+), 30 deletions(-) diff --git a/.agents/notes/implemented/architecture/2026-07-28-directory-picker-capability-seam.i18n.yaml b/.agents/notes/implemented/architecture/2026-07-28-directory-picker-capability-seam.i18n.yaml index 1690808ee0..a05ed7b37d 100644 --- a/.agents/notes/implemented/architecture/2026-07-28-directory-picker-capability-seam.i18n.yaml +++ b/.agents/notes/implemented/architecture/2026-07-28-directory-picker-capability-seam.i18n.yaml @@ -2,5 +2,5 @@ # side as of the last confirmed-consistent state. Both languages carry equal authority; # after editing either side, bring the other along and re-record with: # pnpm run verify-translation-pairing --write .agents/notes/implemented/architecture/2026-07-28-directory-picker-capability-seam.md -2026-07-28-directory-picker-capability-seam.md: 15d0a6ad3fc1e92e0487024c0d7f611380382e2d -2026-07-28-directory-picker-capability-seam.zh.md: 54042cea3bf4a4888855a60765ccc19977e6a061 +2026-07-28-directory-picker-capability-seam.md: bd1a0bc2f840416f4c939474d1d90c51701c7fac +2026-07-28-directory-picker-capability-seam.zh.md: f836c00443ad460d9589deba859401b305d0d902 diff --git a/.agents/notes/implemented/architecture/2026-07-28-directory-picker-capability-seam.md b/.agents/notes/implemented/architecture/2026-07-28-directory-picker-capability-seam.md index 15d0a6ad3f..bd1a0bc2f8 100644 --- a/.agents/notes/implemented/architecture/2026-07-28-directory-picker-capability-seam.md +++ b/.agents/notes/implemented/architecture/2026-07-28-directory-picker-capability-seam.md @@ -20,7 +20,7 @@ Placement and policy rulings folded into this decision: - **Dependency survey (hand-roll vs adopt).** Node's stdlib *is* the maintained cross-platform OS layer (`readdir(withFileTypes)`, `homedir`, path semantics); surveyed alternatives fail the dependency bar — file-manager packages (`node-file-manager`, `files-and-folders`, Syncfusion's provider) are whole HTTP apps (fit), drive-letter helpers (`drivelist` native addon, `windows-drive-letters` ~7y stale) fail health/proportionality. The browse backend is a thin adapter over stdlib. - **Hidden entries: return-and-flag.** The host stamps `hidden` (POSIX dot convention) and returns everything; the client filters. Display policy stays client-side, and the show-hidden toggle shipped as exactly that client-only change: a fixed-label footer toggle whose state lives in the pressed presentation (`aria-pressed` + check glyph), a dot-led path-draft prefix reveals the hidden entries it names, and the current selection is exempt from both the hidden and the prefix filter (it anchors the two-pane view). Windows' `FILE_ATTRIBUTE_HIDDEN` is not exposed by dirents — documented limitation until a native probe pays for itself. - **Path-editor cancel scope: the dialog card.** The browse client's path editor cancels on Escape and on focus leaving the card, both observed at a card-scope wrapper rather than the input — after Tab parks focus on a filtered row the input is off the event path, yet Escape must collapse the editor (not the dialog) and a later focus departure must still cancel. Non-cancel exemptions: window/tab focus loss, in-card focus moves, and pointer paths (rows and the toggle suppress focus steal on mousedown while editing). Separators for seeding and draft-tail filtering are inferred from `listing.home`; the wire-field alternative below records the deferred authoritative form. Combobox semantics between the editor and the list it filters (`aria-expanded`/`aria-controls`/active-descendant, result announcements) are likewise deferred — today they read to assistive tech as separate widgets. -- **The path editor advertises itself, and the panes follow the draft.** The click-to-edit zone is not invisible: a pencil glyph sits at the bar's right edge and hover/focus lights the zone in the editor's own footprint, so the one route into typing a path is discoverable and the bar does not resize when zone and input swap. While the editor is open the panes track the draft instead of whatever level happened to be listed when it opened — the final segment prefix-filters the level its directory part names, a tail nobody matches releases the filter (a name still being spelled must not empty the pane it is being spelled into), and a directory part no pane lists is scanned after a 250ms rest and lands through the same selection-anchored, two-pane landing every navigation uses, so typing a path moves the Miller view exactly as a crumb jump does — typing deeper descends, erasing segments walks back up — without leaving the editor. One landing shape, two callers: a submitted path closes the editor and announces failures, the draft-following scan keeps both to itself. That scan is speculative — half-typed directories are unreadable most of the time — so a failure keeps the last readable panes and stays silent. Enter remains the authoritative commit: it owns the view from submission until landing (a debounce timer armed by the same keystrokes is held back rather than superseding the navigation, and a rejected submission stays held until the next edit) and it alone surfaces the failure. Two consequences are deliberate. The wait is keyed on the draft, not on the directory part it names, so a keystroke that superseded an in-flight scan re-arms one and an edit after a rejected submission releases the hold; the panes it reads are a ref rather than a dependency, or the landing would re-arm the wait and a host answering with a differently spelled path would scan forever. And a walk is not rewound: closing the editor — cancellation included — leaves the panes where the draft took them, named by the crumbs and followed by Open's fallback target, because the operator watched them move. A landing that unmounts the row a keyboard operator Tabbed onto re-parks focus on the editor, since the Modal has no focus trap. +- **The path editor advertises itself, and the panes follow the draft.** The click-to-edit zone is not invisible: a pencil glyph sits at the bar's right edge and hover/focus lights the zone in the editor's own footprint, so the one route into typing a path is discoverable and the bar does not resize when zone and input swap. While the editor is open the panes track the draft instead of whatever level happened to be listed when it opened — the final segment prefix-filters the level its directory part names, a tail nobody matches releases the filter (a name still being spelled must not empty the pane it is being spelled into), and any other directory part is scanned after a 250ms rest and lands through the same selection-anchored, two-pane landing every navigation uses, so typing a path moves the Miller view exactly as a crumb jump does — typing deeper descends, erasing segments walks back up — without leaving the editor. **The pane arity is the invariant**: the last pane always lists the level the path names, with its parent beside it and nothing but a display root listing alone. Skipping the scan whenever *any* pane happened to list the directory was the cheaper rule and the wrong one — erasing a segment then left the level being typed on the left with its own child pane still standing to its right, so the panes stopped reading as "where I am, and where I came from". Only the last pane's own tail costs no scan. One landing shape, two callers: a submitted path closes the editor and announces failures, the draft-following scan keeps both to itself. That scan is speculative — half-typed directories are unreadable most of the time — so a failure keeps the last readable panes and stays silent. Enter remains the authoritative commit: it owns the view from submission until landing (a debounce timer armed by the same keystrokes is held back rather than superseding the navigation, and a rejected submission stays held until the next edit) and it alone surfaces the failure. Two consequences are deliberate. The wait is keyed on the draft, not on the directory part it names, so a keystroke that superseded an in-flight scan re-arms one and an edit after a rejected submission releases the hold; the panes it reads are a ref rather than a dependency, or the landing would re-arm the wait and a host answering with a differently spelled path would scan forever. And a walk is not rewound: closing the editor — cancellation included — leaves the panes where the draft took them, named by the crumbs and followed by Open's fallback target, because the operator watched them move. A landing that unmounts the row a keyboard operator Tabbed onto re-parks focus on the editor, since the Modal has no focus trap. - **Navigation lands selection-anchored, quiet, and bounded.** Away from the display root (the same collapse the crumb header renders, so crumbs and pane shape never disagree), the landing is two-pane: the target's actual parent-level entry re-selected (platform case folding on Windows), its children on the right, so a crumb jump reads as stepping back one pane rather than collapsing to a single column. Target and parent legs land as **one frame** when the parent leg settles within the 200ms wait bound — the stale view keeps rendering until then, so navigation swaps the panes without an intermediate single-pane flash — and past the bound the target commits alone at once (an Enter-submitted navigation is never held hostage by a stalled parent) with the late parent leg upgrading the landing in place. The parent leg runs under the landing's supersession scope and is aborted on the wire by any newer intent (Escape inside the landing window therefore withdraws the whole navigation); a failed parent leg, or a truncated parent window lacking the target, leaves the single-pane landing — the upgrade must never orphan the selection it exists to anchor. The loading indicator follows the same quiet rule: it floats over the content's bottom-right corner (never a layout-shifting row; the truncated/error rows own the bottom left and keep rendering through a scan) and only once a scan outlives a 300ms silence window, so a local listing swaps with nothing shown at all. Row picks are deliberately exempt from the one-frame rule: a pick's immediate pane split is its selected-state feedback (aria-current, crumbs following), while a navigation has nothing to acknowledge the click but the swap itself. Both timing constants are calibrated for local enumeration; a remote deployment (one RPC per level, commonly 100–400ms) would sit inside the silence window with no pressed state on the crumbs — revisit the window or add pressed feedback when a remote consumer lands. - **Symlinks: follow for enterability.** `stat` probes symlinks (broken/cyclic → skipped); crumbs keep the logical path the operator navigated, and `workspace.create` already canonicalizes via realpath at adoption. - **Listing levels are bounded, and streamed.** One `list` call returns at most `maxEntries` rows (config, default 1000 — GitHub's web-UI directory-listing bound). The level streams via `opendir` into a name-sorted window of `maxEntries + 1` candidates, so memory stays O(maxEntries) and enterability probing touches only windowed candidates; the wire `DirectoryListing` carries a required `truncated` flag so the client states incompleteness instead of silently missing tail entries. A windowed broken symlink is not backfilled from beyond the window — the eviction already marks the level truncated. Window insertion is binary with an O(1) full-window tail rejection (an oversized level must not pay a window scan per dirent), and `list(path, signal)` threads the carrier's request signal so a scan of a stalled network directory cannot outlive a disconnected caller — every await in the scan (open, each read, each symlink probe) races the signal, an aborted exit abandons rather than awaits the close (Node queues close behind in-flight reads), and abandoned settlements are swallowed so cleanup can never surface as an unhandled rejection. An unbounded level is a memory/responsiveness hole for large or adversarial directories. diff --git a/.agents/notes/implemented/architecture/2026-07-28-directory-picker-capability-seam.zh.md b/.agents/notes/implemented/architecture/2026-07-28-directory-picker-capability-seam.zh.md index 54042cea3b..f836c00443 100644 --- a/.agents/notes/implemented/architecture/2026-07-28-directory-picker-capability-seam.zh.md +++ b/.agents/notes/implemented/architecture/2026-07-28-directory-picker-capability-seam.zh.md @@ -20,7 +20,7 @@ web GUI 的"打开本地文件夹"流程被焊死在一种交互上：`host.pick - **依赖调研（手写 vs 引入）。** Node 标准库本身就是维护中的跨平台 OS 层（`readdir(withFileTypes)`、`homedir`、路径语义）；调研过的替代品都过不了依赖门槛——文件管理器包（`node-file-manager`、`files-and-folders`、Syncfusion 的 provider）是整套 HTTP 应用（契合度不过），盘符工具（原生插件 `drivelist`、约七年未更的 `windows-drive-letters`）健康度／比例失当。browse 后端是标准库上的薄适配。 - **隐藏条目：返回并打标。** 宿主标注 `hidden`（POSIX 点前缀约定）并返回全部条目；客户端过滤。展示策略留在客户端，"显示隐藏"开关正是作为这一纯客户端改动落地：标签固定的 footer 开关，其状态由按下态呈现承载（`aria-pressed` + 勾选符号）；以点开头的路径草稿前缀会显出它所指名的隐藏条目；当前选中项则不受隐藏与前缀两种过滤影响（它锚定着双栏视图）。Windows 的 `FILE_ATTRIBUTE_HIDDEN` 不被 dirent 暴露——记为限制，直到原生探测值回其成本。 - **路径编辑器的取消范围：对话框卡片。** browse 客户端的路径编辑器在按 Escape 与焦点离开卡片时取消，两者都在卡片范围的包装层而非输入框上监听——Tab 把焦点停到某个过滤命中的行之后，输入框已不在事件路径上，但 Escape 仍须收起编辑器（而非对话框），其后的焦点离开也仍须取消。不取消的豁免：窗口／标签页失焦、卡片内焦点移动，以及指针路径（编辑期间行与开关在 mousedown 时抑制焦点夺取）。预填与草稿末段过滤所用的分隔符从 `listing.home` 推断；下文的线上字段替代方案记录了被延期的权威形态。编辑器与其过滤的列表之间的 combobox 语义（`aria-expanded`／`aria-controls`／active-descendant、结果播报）同样被延期——目前二者在辅助技术看来是彼此独立的控件。 -- **路径编辑器自我点明，各栏跟随草稿。** 点击即编辑的区域不再是隐形的：栏右端坐着一枚铅笔图标，悬停／聚焦时该区以编辑器自身的轮廓亮起，于是键入路径这唯一入口可被发现，且区域与输入框互换时栏高不变。编辑器打开期间，各栏跟随草稿，而不是停在它打开那一刻恰好列出的层级——末段对其目录部分所指的层级做前缀过滤，无一匹配的末段解除过滤（还在拼写中的名字不该把正在拼写它的那一栏清空），而任何一栏都未列出的目录部分会在停顿 250ms 后被扫描，并经由每次导航共用的那套以选中项为锚的双栏落地落定，于是键入路径移动 Miller 视图的方式与 crumb 跳转完全一致——继续键入即下潜、删掉末段即上退——全程不必离开编辑器。一种落地形态、两个调用方：提交的路径关闭编辑器并呈现失败，草稿跟随扫描则两者都不做。该扫描是推测性的——键入到一半的目录多数时候读不出来——因此失败时保留最后一次可读的分栏并保持沉默。Enter 仍是权威提交：自提交至落地由它独占视图（同一批按键武装的防抖计时器会被扣住，而不是顶掉这次导航；提交被拒后仍扣住，直到下一次编辑），也只有它把失败呈现出来。有两点是刻意为之。等待以草稿为键，而非以它指名的目录部分为键，于是顶掉在飞扫描的那次按键会重新武装等待，被拒提交之后的编辑也能释放那道扣留；而它读取的分栏是 ref 而非依赖，否则落地会重新武装等待，遇到以不同拼写作答的宿主便会永远扫描下去。以及，走过的路不回退：关闭编辑器——包括取消——都把分栏留在草稿带到的地方，由面包屑指明、Open 的兜底目标随之而动，因为操作者亲眼看着它们移动。若落地卸载了键盘操作者 Tab 停留的那一行，焦点会被重新停回编辑器——Modal 并没有焦点陷阱。 +- **路径编辑器自我点明，各栏跟随草稿。** 点击即编辑的区域不再是隐形的：栏右端坐着一枚铅笔图标，悬停／聚焦时该区以编辑器自身的轮廓亮起，于是键入路径这唯一入口可被发现，且区域与输入框互换时栏高不变。编辑器打开期间，各栏跟随草稿，而不是停在它打开那一刻恰好列出的层级——末段对其目录部分所指的层级做前缀过滤，无一匹配的末段解除过滤（还在拼写中的名字不该把正在拼写它的那一栏清空），而其余任何目录部分都会在停顿 250ms 后被扫描，并经由每次导航共用的那套以选中项为锚的双栏落地落定，于是键入路径移动 Miller 视图的方式与 crumb 跳转完全一致——继续键入即下潜、删掉末段即上退——全程不必离开编辑器。**分栏个数才是不变量**：最后一栏永远是路径所指的那一层，其上一层在它旁边，只有展示根会独占一栏。"只要任意一栏碰巧列出了该目录就跳过扫描"是更省事、也是错的规则——删掉一段之后，正在键入的那一层会留在左栏，而它自己的子栏仍立在右边，于是两栏不再读作"我在哪儿、我从哪儿来"。只有最后一栏自己的末段不需要扫描。一种落地形态、两个调用方：提交的路径关闭编辑器并呈现失败，草稿跟随扫描则两者都不做。该扫描是推测性的——键入到一半的目录多数时候读不出来——因此失败时保留最后一次可读的分栏并保持沉默。Enter 仍是权威提交：自提交至落地由它独占视图（同一批按键武装的防抖计时器会被扣住，而不是顶掉这次导航；提交被拒后仍扣住，直到下一次编辑），也只有它把失败呈现出来。有两点是刻意为之。等待以草稿为键，而非以它指名的目录部分为键，于是顶掉在飞扫描的那次按键会重新武装等待，被拒提交之后的编辑也能释放那道扣留；而它读取的分栏是 ref 而非依赖，否则落地会重新武装等待，遇到以不同拼写作答的宿主便会永远扫描下去。以及，走过的路不回退：关闭编辑器——包括取消——都把分栏留在草稿带到的地方，由面包屑指明、Open 的兜底目标随之而动，因为操作者亲眼看着它们移动。若落地卸载了键盘操作者 Tab 停留的那一行，焦点会被重新停回编辑器——Modal 并没有焦点陷阱。 - **导航以选中项为锚、安静且有界地落地。** 在展示根之外（与 crumb 头部渲染的是同一塌缩，因此 crumb 与分栏形态永不相左），落地即双栏：重新选中目标在父层级中的实际条目（Windows 上按平台惯例折叠大小写），右侧展示其子项，因此 crumb 跳转读作后退一栏，而不是塌缩成单列。父层级这一程在 200ms 等待上限内落定时，目标与父层级两程以**同一帧**落地——在此之前陈旧视图持续渲染，导航换栏时因此没有中间的单栏闪现——超出该上限则目标即刻单独提交（Enter 提交的导航绝不会被滞塞的父层级扣作人质），迟到的父层级这一程再就地升级这次落地。父层级这一程在落地的 supersession 范围下运行，任何较新的意图都会在线上将其中止（因此在落地窗口内按 Escape 即撤回整次导航）；父层级这一程失败，或被截断的父窗口缺少目标时，都保留单栏落地——升级的存在正是为了锚定选中项，绝不能反而让它悬空。加载指示器遵循同一安静规则：它浮于内容右下角（绝不是会挪动布局的一行；截断／错误行占据左下角，并在扫描期间持续渲染），且仅在扫描超出 300ms 静默窗口后才出现，因此本地列举切换时什么也不显示。行选取被刻意豁免于同一帧规则：选取后立即分栏本身就是其选中态反馈（aria-current、crumb 跟随），而导航除了换栏本身没有任何东西可确认这次点击。两个时序常量都按本地列举校准；远程部署（每层级一次 RPC，通常 100–400ms）会落在静默窗口之内、crumb 上却没有按下态——待远程消费方落地时，重新审视该窗口或补上按下反馈。 - **符号链接：为可进入性而跟随。** 用 `stat` 探测符号链接（断链／循环→跳过）；面包屑保留操作者导航的逻辑路径，`workspace.create` 在接纳时本就做 realpath 规范化。 - **列举层级有上限，且流式处理。** 单次 `list` 至多返回 `maxEntries` 行（配置项，默认 1000——GitHub 网页端目录列举的同一上限）。层级经 `opendir` 流入一个按名排序、容量 `maxEntries + 1` 的候选窗口，内存保持 O(maxEntries)，可进入性探测只触及窗口内候选；线上 `DirectoryListing` 携带必填的 `truncated` 标志，让客户端明示不完整而不是静默缺尾。窗口内的断链符号链接不从窗口外回填——发生过驱逐本身已把层级标记为截断。窗口插入为二分查找、满窗尾部单次比较即拒绝（超大层级不能为每个 dirent 付出一次全窗扫描），且 `list(path, signal)` 透传载体的请求信号，滞塞网络目录的扫描不会在调用方断连后继续存活——扫描中的每个 await（打开、每次读取、每次符号链接探测）都与信号赛跑，中止路径放弃而非等待 close（Node 会把 close 排在在飞读取之后），被放弃的 settlement 全部吞掉，清理不会以未处理拒绝的形式冒出。无上限的层级对超大或恶意构造的目录就是内存／响应性漏洞。 diff --git a/apps/web/tests/workspace-management.e2e.ts b/apps/web/tests/workspace-management.e2e.ts index b16e45184a..075f5687ef 100644 --- a/apps/web/tests/workspace-management.e2e.ts +++ b/apps/web/tests/workspace-management.e2e.ts @@ -423,11 +423,14 @@ describe('web e2e: workspace management (create / rename / flat view / hover aff await expect.poll(() => dialog.getByText('only-under-alpha', { exact: true }).count(), { timeout: 10_000 }).toBe(1) expect(await dialog.getByRole('list').count()).toBe(2) expect(await path.inputValue()).toBe(`${join(staged, 'alpha')}${sep}`) - // Erasing back past the separator returns to a level already on screen: - // the tail filters it, no scan needed, both panes stay. + // Erasing back past the separator walks the panes up, so the level being + // typed is the last pane again (its children no longer stand to its + // right) and the tail filters it. await path.fill(`${staged}${sep}al`) - await expect.poll(() => dialog.getByText('beta', { exact: true }).count(), { timeout: 10_000 }).toBe(0) + await expect.poll(() => dialog.getByText('only-under-alpha', { exact: true }).count(), { timeout: 10_000 }).toBe(0) expect(await dialog.getByText('alpha', { exact: true }).count()).toBe(1) + expect(await dialog.getByText('beta', { exact: true }).count()).toBe(0) + expect(await dialog.getByRole('list').count()).toBe(2) // A tail nobody matches is a name still being spelled: the level shows // whole instead of emptying under it. await path.fill(`${staged}${sep}zzz`) diff --git a/packages/host/directory-picker-browse/README.i18n.yaml b/packages/host/directory-picker-browse/README.i18n.yaml index 2d66b6593b..4b4f10ee23 100644 --- a/packages/host/directory-picker-browse/README.i18n.yaml +++ b/packages/host/directory-picker-browse/README.i18n.yaml @@ -2,5 +2,5 @@ # side as of the last confirmed-consistent state. Both languages carry equal authority; # after editing either side, bring the other along and re-record with: # pnpm run verify-translation-pairing --write packages/host/directory-picker-browse/README.md -README.md: c0375331e0e82fd6864b2027e7e36e0c6cb9986a -README.zh.md: 91d35de8821414c095db2a7834309864b5df0cd6 +README.md: 04d97adf71ae4a3ea8d89b24098f27cbdee20961 +README.zh.md: 603afaed7e4b4c77e699e009fc612e73c06d73aa diff --git a/packages/host/directory-picker-browse/README.md b/packages/host/directory-picker-browse/README.md index c0375331e0..04d97adf71 100644 --- a/packages/host/directory-picker-browse/README.md +++ b/packages/host/directory-picker-browse/README.md @@ -6,7 +6,7 @@ The **in-app browsing backend** of the [directory-picker seam](../directory-pick Behavior facts: listings return **directories only**, name-sorted, with symlinks-to-directories followed (broken/cyclic links skipped — the probe `stat` failing means "not enterable") and a host-owned `hidden` flag (POSIX dot convention) left for the client to act on; `crumbs` is the root-to-target ancestor chain, the root crumb labeled by its full path (`/`, `C:\`); an absent `list` path means the host account's home directory. `createDirectory` is non-recursive (a missing parent is a real failure, not a level to invent) and validates the name as a single non-blank segment even when called directly, mirroring the wire schema's fence. Both primitives reject an explicit path that is not fully qualified — relative forms, and on Windows the rooted drive-less forms (`\foo`, `/foo`) and incomplete UNC prefixes (`\\`, `\\server`) that `isAbsolute` accepts — with `directory-unreadable`/`directory-create-failed`, instead of letting `resolve` rebase it under the host process cwd or current drive. One `list` call returns at most `maxEntries` rows (config, default 1000 — the bound GitHub's web UI applies to directory listings), and the level streams through a bounded window so memory stays O(maxEntries) no matter how many children the directory holds: a cut level keeps the name-sorted head, counts hidden rows against the bound, probes only windowed candidates, and reports `truncated: true` so the client can say the level is incomplete (a windowed broken symlink is not backfilled from beyond the window — the eviction already marks the level truncated); window insertion is binary with an O(1) full-window tail rejection, and `list` threads the caller's `AbortSignal` so a disconnect or timeout stops the scan instead of letting it outlive the caller. Failures throw the seam's typed `DirectoryPickerError`. Policy rationale: [the directory-picker capability seam Agent Note](../../../.agents/notes/implemented/architecture/2026-07-28-directory-picker-capability-seam.md). -**Dual-face package**: the browser half (`./client`) fills [ui-workspace's](../../client/ui-workspace/README.md) two directory-flow holes with the in-app **Select Workspace Directory** dialog (figma `Harness` 813-23126 family — Miller two-column view whose navigations land selection-anchored and quiet: the previous view keeps rendering while a crumb jump or a submitted path is scanned (a "Loading…" pill floats over it only once the scan outlives a 300ms silence window, never shifting the columns), then target and parent legs land as one two-pane frame with the target re-selected as its actual parent-level entry — so stepping back never collapses and no intermediate frame flashes (a parent leg outliving its 200ms wait bound lands the target alone and upgrades in place; a failed or truncated parent leg keeps the single-pane landing; the display root keeps the single wide level); breadcrumb with a click-to-edit path zone, advertised by the pencil glyph at the bar's right edge and lit on hover in the editor's own footprint, whose editor seeds a trailing separator and then keeps the panes under the draft: the final segment prefix-filters the level its directory part names (case-insensitively, over the listed — possibly truncated — rows only; a tail nobody matches releases the filter instead of emptying the pane), while a directory part no pane lists is scanned after a 250ms rest and lands like any other navigation — selection-anchored, two-pane away from the display root — so typing deeper descends and erasing segments walks back up, moving the Miller view without leaving the editor (a level a pane already shows needs no scan at all: the filter alone answers the draft) — a speculative scan is silent when it fails, and Enter still navigates by the exact text, owning the view until it lands; the editor cancels on Escape or when focus leaves the dialog card (window/tab switches and in-card focus moves keep the draft), and panes the draft walked to stay where the walk ended — the crumbs name that level and Open's fallback target follows them, so cancelling closes the editor rather than rewinding the walk; a fixed-label show-hidden footer toggle over the host's `hidden` flags, with a dot-led typed prefix revealing its matches and the current selection exempt from every filter; nested New-folder dialog), driving `host.listDirectory`/`host.createDirectory` and registering its own locale namespace (`directory-browser`, zh default / en). One cordis.yml row therefore composes both sides of the browse interaction; the client carries no capability-kind branching, and mounting a second flow package fails at load (the holes are `single` kind). +**Dual-face package**: the browser half (`./client`) fills [ui-workspace's](../../client/ui-workspace/README.md) two directory-flow holes with the in-app **Select Workspace Directory** dialog (figma `Harness` 813-23126 family — Miller two-column view whose navigations land selection-anchored and quiet: the previous view keeps rendering while a crumb jump or a submitted path is scanned (a "Loading…" pill floats over it only once the scan outlives a 300ms silence window, never shifting the columns), then target and parent legs land as one two-pane frame with the target re-selected as its actual parent-level entry — so stepping back never collapses and no intermediate frame flashes (a parent leg outliving its 200ms wait bound lands the target alone and upgrades in place; a failed or truncated parent leg keeps the single-pane landing; the display root keeps the single wide level); breadcrumb with a click-to-edit path zone, advertised by the pencil glyph at the bar's right edge and lit on hover in the editor's own footprint, whose editor seeds a trailing separator and then keeps the panes under the draft: the final segment prefix-filters the level its directory part names (case-insensitively, over the listed — possibly truncated — rows only; a tail nobody matches releases the filter instead of emptying the pane), while any other directory part is scanned after a 250ms rest and lands like any other navigation — selection-anchored, two-pane away from the display root — so typing deeper descends and erasing segments walks back up, moving the Miller view without leaving the editor; the pane arity is the invariant, the last pane always listing the level the path names with its parent beside it (only that level's own tail costs no scan, and only a display root lists alone) — a speculative scan is silent when it fails, and Enter still navigates by the exact text, owning the view until it lands; the editor cancels on Escape or when focus leaves the dialog card (window/tab switches and in-card focus moves keep the draft), and panes the draft walked to stay where the walk ended — the crumbs name that level and Open's fallback target follows them, so cancelling closes the editor rather than rewinding the walk; a fixed-label show-hidden footer toggle over the host's `hidden` flags, with a dot-led typed prefix revealing its matches and the current selection exempt from every filter; nested New-folder dialog), driving `host.listDirectory`/`host.createDirectory` and registering its own locale namespace (`directory-browser`, zh default / en). One cordis.yml row therefore composes both sides of the browse interaction; the client carries no capability-kind branching, and mounting a second flow package fails at load (the holes are `single` kind). ## Model Experience diff --git a/packages/host/directory-picker-browse/README.zh.md b/packages/host/directory-picker-browse/README.zh.md index 91d35de882..603afaed7e 100644 --- a/packages/host/directory-picker-browse/README.zh.md +++ b/packages/host/directory-picker-browse/README.zh.md @@ -6,7 +6,7 @@ 行为事实：列举**只返回目录**、按名称排序，指向目录的符号链接会被跟随（断链／循环链接被跳过——探测 `stat` 失败即"不可进入"），并携带宿主判定的 `hidden` 标志（POSIX 点前缀约定），展示决策留给客户端；`crumbs` 是从根到目标的祖先链，根 crumb 以完整路径标注（`/`、`C:\`）；`list` 不带路径即列举宿主账户的家目录。`createDirectory` 不递归（父目录缺失是真实失败，不是要补造的层级），且即便被直接调用也把名称校验为单个非空段，与协议 schema 的栅栏一致。两个原语都拒绝非完全限定的显式路径——相对形态，以及 Windows 上 `isAbsolute` 会放行的无盘符有根形态（`\foo`、`/foo`）与不完整的 UNC 前缀（`\\`、`\\server`）——报 `directory-unreadable`／`directory-create-failed`，而不是任由 `resolve` 把它重定位到宿主进程 cwd 或当前盘符之下。单次 `list` 至多返回 `maxEntries` 行（配置项，默认 1000——GitHub 网页端对目录列举采用的同一上限），且层级以流式方式经过一个有界窗口，无论目录有多少子项内存都保持 O(maxEntries)：被截断的层级保留按名排序的头部、隐藏行计入上限、只探测窗口内候选，并报告 `truncated: true`，供客户端提示层级不完整（窗口内的断链符号链接不会从窗口外回填——发生过驱逐本身已把层级标记为截断）；窗口插入为二分查找、满窗尾部单次比较即拒绝，且 `list` 透传调用方的 `AbortSignal`，断连或超时会停止扫描而不是让它在调用方离开后继续。失败抛出 seam 的类型化 `DirectoryPickerError`。策略依据：[目录选择能力 seam Agent Note](../../../.agents/notes/implemented/architecture/2026-07-28-directory-picker-capability-seam.md)。 -**双面包**：browser half（`./client`）以应用内 **选择工作区目录** 对话框（figma `Harness` 813-23126 家族——Miller 双列视图，其导航以选中项为锚、安静落地：扫描 crumb 跳转或提交的路径期间，先前视图持续渲染（"Loading…" 胶囊仅在扫描超出 300ms 静默窗口后才浮于其上，绝不挪动各列），随后目标与父层级两程以单个双栏帧落地，目标被重新选中为其在父层级中的实际条目——因此后退绝不塌缩，也没有中间帧闪现（父层级这一程超出其 200ms 等待上限时，目标单独落地，随后就地升级；父层级这一程失败或被截断时保持单栏落地；展示根保持单个宽层级）；带点击即编辑路径区的面包屑，该区由栏右端的铅笔图标点明、悬停时以编辑器自身的轮廓亮起，其编辑器预填尾随分隔符，随后让下方各栏跟随草稿：末段对其目录部分所指的层级做前缀过滤（不区分大小写，且仅作用于已列出、可能被截断的行；无一匹配的末段会解除过滤，而不是把该栏清空），而任何一栏都未列出的目录部分会在停顿 250ms 后被扫描，并像其他任何一次导航那样落地——以选中项为锚，在展示根之外即双栏——于是继续键入即下潜、删掉末段即上退，Miller 视图随之移动而不必离开编辑器（某一栏已经展示的层级则根本不需要扫描：过滤本身就答复了草稿）——推测性扫描失败时保持沉默，而 Enter 仍按确切文本导航，并在落地前独占视图；编辑器按 Escape 或焦点离开对话框卡片即取消（窗口／标签页切换与卡片内焦点移动保留草稿），而草稿走到的层级会留在原地——面包屑指明该层级、Open 的兜底目标随之而动，因此取消只是关闭编辑器，并不回退这段行走；基于宿主 `hidden` 标志、标签固定的"显示隐藏"footer 开关，键入以点开头的前缀会显出其匹配项，且当前选中项不受任何过滤影响；嵌套新建文件夹对话框）填入 [ui-workspace](../../client/ui-workspace/README.md) 的两个目录流洞，驱动 `host.listDirectory`／`host.createDirectory`，并注册自己的 locale 命名空间（`directory-browser`，zh 默认／en）。因此一行 cordis.yml 同时组合浏览交互的两侧；client 侧不含任何能力 kind 分支，挂载第二个流程包会在加载期失败（洞为 `single` kind）。 +**双面包**：browser half（`./client`）以应用内 **选择工作区目录** 对话框（figma `Harness` 813-23126 家族——Miller 双列视图，其导航以选中项为锚、安静落地：扫描 crumb 跳转或提交的路径期间，先前视图持续渲染（"Loading…" 胶囊仅在扫描超出 300ms 静默窗口后才浮于其上，绝不挪动各列），随后目标与父层级两程以单个双栏帧落地，目标被重新选中为其在父层级中的实际条目——因此后退绝不塌缩，也没有中间帧闪现（父层级这一程超出其 200ms 等待上限时，目标单独落地，随后就地升级；父层级这一程失败或被截断时保持单栏落地；展示根保持单个宽层级）；带点击即编辑路径区的面包屑，该区由栏右端的铅笔图标点明、悬停时以编辑器自身的轮廓亮起，其编辑器预填尾随分隔符，随后让下方各栏跟随草稿：末段对其目录部分所指的层级做前缀过滤（不区分大小写，且仅作用于已列出、可能被截断的行；无一匹配的末段会解除过滤，而不是把该栏清空），而其余任何目录部分都会在停顿 250ms 后被扫描，并像其他任何一次导航那样落地——以选中项为锚，在展示根之外即双栏——于是继续键入即下潜、删掉末段即上退，Miller 视图随之移动而不必离开编辑器；分栏个数是这里的不变量：最后一栏永远是路径所指的那一层，其上一层在它旁边（只有这一层自己的末段不触发扫描，也只有展示根会独占一栏）——推测性扫描失败时保持沉默，而 Enter 仍按确切文本导航，并在落地前独占视图；编辑器按 Escape 或焦点离开对话框卡片即取消（窗口／标签页切换与卡片内焦点移动保留草稿），而草稿走到的层级会留在原地——面包屑指明该层级、Open 的兜底目标随之而动，因此取消只是关闭编辑器，并不回退这段行走；基于宿主 `hidden` 标志、标签固定的"显示隐藏"footer 开关，键入以点开头的前缀会显出其匹配项，且当前选中项不受任何过滤影响；嵌套新建文件夹对话框）填入 [ui-workspace](../../client/ui-workspace/README.md) 的两个目录流洞，驱动 `host.listDirectory`／`host.createDirectory`，并注册自己的 locale 命名空间（`directory-browser`，zh 默认／en）。因此一行 cordis.yml 同时组合浏览交互的两侧；client 侧不含任何能力 kind 分支，挂载第二个流程包会在加载期失败（洞为 `single` kind）。 ## 模型体验 diff --git a/packages/host/directory-picker-browse/src/client/DirectoryBrowser.tsx b/packages/host/directory-picker-browse/src/client/DirectoryBrowser.tsx index 395793268c..eaada21965 100644 --- a/packages/host/directory-picker-browse/src/client/DirectoryBrowser.tsx +++ b/packages/host/directory-picker-browse/src/client/DirectoryBrowser.tsx @@ -23,13 +23,14 @@ * separator, and keeps the panes under the draft: the final segment * prefix-filters the level its directory part names (a dot-led prefix also * reveals the hidden entries it names, and a prefix nobody matches releases - * the filter), while a directory part no pane lists is scanned after a short + * the filter), while any other directory part is scanned after a short * debounce and lands like any other navigation — selection-anchored and - * two-pane away from the display root — so typing deeper descends and - * erasing segments walks back up, moving the Miller view without leaving the - * editor. Panes the draft walked to stay put when the editor closes - * (cancellation included): the crumbs name where the walk ended, and Open's - * fallback target follows them. + * two-pane away from the display root. The pane arity holds throughout: the + * last pane is the level the path names and the one beside it is its parent, + * so typing deeper descends and erasing segments walks back up, moving the + * Miller view without leaving the editor. Panes the draft walked to stay put + * when the editor closes (cancellation included): the crumbs name where the + * walk ended, and Open's fallback target follows them. */ import { useCallback, useEffect, useRef, useState } from 'react' import clsx from 'clsx' @@ -151,12 +152,16 @@ function draftPrefixFor(listing: DirectoryListing, draft: string | null): string } /** - * The directory a draft addresses that no rendered pane lists — the level the - * editor must scan for the panes to keep following the typed path. Null when - * a pane already lists it (the prefix filter alone answers the draft), when - * no separator has been typed yet, and when no level is listed at all: the - * platform separator is read off a listing, so the editor's - * failed-home-listing recovery path types blind until Enter. + * The directory a draft addresses that the panes are not already presenting + * as the current level — what the editor must scan to keep the view under the + * typed path. The pane arity is the invariant this preserves: the LAST pane + * always lists the level the path names, with its parent beside it (a display + * root lists alone), so a draft naming any other level re-lands rather than + * leaving a deeper level standing to the right of the one being typed. Null + * when that level is already the last pane, when no separator has been typed + * yet, and when no level is listed at all: the platform separator is read off + * a listing, so the editor's failed-home-listing recovery path types blind + * until Enter. */ function pendingPreviewDirectory( parent: DirectoryListing | null, @@ -165,9 +170,8 @@ function pendingPreviewDirectory( ): string | null { if (parent === null) return null const directory = draftDirectory(parent, draft) - if (directory === null || directory === levelDirectory(parent)) return null - if (child !== null && directory === levelDirectory(child)) return null - return directory + if (directory === null) return null + return directory === levelDirectory(child ?? parent) ? null : directory } /** diff --git a/packages/host/directory-picker-browse/tests/directory-browser.spec.tsx b/packages/host/directory-picker-browse/tests/directory-browser.spec.tsx index d6d6a3903c..69c61751eb 100644 --- a/packages/host/directory-picker-browse/tests/directory-browser.spec.tsx +++ b/packages/host/directory-picker-browse/tests/directory-browser.spec.tsx @@ -707,11 +707,13 @@ describe('DirectoryBrowser', () => { fireEvent.change(input, { target: { value: `${DOCS}/zzz` } }) expect(within(columns()[1]!).getByText('harness')).toBeTruthy() expect(within(columns()[0]!).getByText('Documents')).toBeTruthy() - // Erasing back into the parent's own path moves the filter to the LEFT - // pane and releases the right one — no scan, both levels are on screen. + // Erasing back into the parent's own path re-lands on it rather than + // filtering the LEFT pane: the level being typed is always the last pane, + // never a pane with a deeper level standing to its right. Home is the + // display root, so it lands alone. fireEvent.change(input, { target: { value: `${HOME}/zz` } }) - expect(within(columns()[0]!).getAllByRole('listitem').map(item => item.textContent)).toEqual(['Documents']) - expect(within(columns()[1]!).getByText('harness')).toBeTruthy() + await waitFor(() => { expect(columns()).toHaveLength(1) }) + expect(screen.getAllByRole('listitem').map(item => item.textContent)).toEqual(['Documents']) }) it('follows the draft into a directory no pane lists, landing the two-pane Miller view', async () => { @@ -740,6 +742,26 @@ describe('DirectoryBrowser', () => { expect(columns()).toHaveLength(2) }) + it('keeps the typed level in the last pane, its parent beside it, as the draft walks', async () => { + const b = mount() + await waitFor(() => { expect(screen.getByRole('listitem')).toBeTruthy() }) + fireEvent.click(screen.getByRole('button', { name: 'browser.editPath' })) + const input = screen.getByLabelText('browser.editPath') + // Two levels down: the typed level on the right, its parent on the left. + fireEvent.change(input, { target: { value: `${HARNESS}/` } }) + await waitFor(() => { expect(within(columns()[0]!).getByText('harness')).toBeTruthy() }) + expect(columns()).toHaveLength(2) + expect(within(columns()[1]!).queryAllByRole('listitem')).toHaveLength(0) + // Erasing back to the parent's own path re-lands on it: the level being + // typed moves BACK into the last pane instead of staying on the left with + // its own child pane still to the right. + fireEvent.change(input, { target: { value: `${DOCS}/ha` } }) + await waitFor(() => { expect(within(columns()[0]!).getByText('Documents')).toBeTruthy() }) + expect(columns()).toHaveLength(2) + expect(within(columns()[1]!).getAllByRole('listitem').map(item => item.textContent)).toEqual(['harness']) + expect(b.listDirectory).toHaveBeenCalledWith(`${DOCS}/`, expect.anything()) + }) + it('walks the panes back up when erased segments leave the listed levels', async () => { const b = mount() await waitFor(() => { expect(screen.getByRole('listitem')).toBeTruthy() }) From 6c0ce22f59bc818cc9eebd609e8883b87083ad63 Mon Sep 17 00:00:00 2001 From: creatixchu Date: Mon, 3 Aug 2026 13:45:30 +0800 Subject: [PATCH 28/48] fix(directory-picker-browse): light the whole bar, and move the view once per keystroke MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit The hover affordance framed only the strip right of the crumbs. The bar itself now carries the outline and the padding in both modes, so hovering previews exactly the field the click produces and nothing resizes when the two swap. One keystroke moved the view twice: deleting a separator first narrowed the pane the draft had just walked away from, then replaced it with its landing. The tail now filters only the LAST pane — the one whose level the path names — so a pane on its way out holds still until its landing arrives. Also from the review round: the walk waits both legs out instead of taking the submitted-navigation bound (a speculative scan has nothing waiting on it, and a tail keystroke aborting a slow parent leg would otherwise strand the two-pane view); a level keeps answering the directory text that produced it, so `..` segments and Windows forward slashes filter and stop rescanning; the release-on-miss rule counts displayable rows, so it survives `hidden` ever meaning more than dot-prefixed; and the editor's 250ms rest joins the other two constants on the remote-recalibration list. --- ...directory-picker-capability-seam.i18n.yaml | 4 +- ...-07-28-directory-picker-capability-seam.md | 4 +- ...-28-directory-picker-capability-seam.zh.md | 4 +- .../directory-picker-browse/README.i18n.yaml | 4 +- .../host/directory-picker-browse/README.md | 2 +- .../host/directory-picker-browse/README.zh.md | 2 +- .../src/client/DirectoryBrowser.module.css | 53 +++--- .../src/client/DirectoryBrowser.tsx | 171 +++++++++++------- .../tests/directory-browser.spec.tsx | 114 +++++++++++- 9 files changed, 256 insertions(+), 102 deletions(-) diff --git a/.agents/notes/implemented/architecture/2026-07-28-directory-picker-capability-seam.i18n.yaml b/.agents/notes/implemented/architecture/2026-07-28-directory-picker-capability-seam.i18n.yaml index a05ed7b37d..048536527a 100644 --- a/.agents/notes/implemented/architecture/2026-07-28-directory-picker-capability-seam.i18n.yaml +++ b/.agents/notes/implemented/architecture/2026-07-28-directory-picker-capability-seam.i18n.yaml @@ -2,5 +2,5 @@ # side as of the last confirmed-consistent state. Both languages carry equal authority; # after editing either side, bring the other along and re-record with: # pnpm run verify-translation-pairing --write .agents/notes/implemented/architecture/2026-07-28-directory-picker-capability-seam.md -2026-07-28-directory-picker-capability-seam.md: bd1a0bc2f840416f4c939474d1d90c51701c7fac -2026-07-28-directory-picker-capability-seam.zh.md: f836c00443ad460d9589deba859401b305d0d902 +2026-07-28-directory-picker-capability-seam.md: 01968990db81852dbf965a90fc151bab357ecb55 +2026-07-28-directory-picker-capability-seam.zh.md: ffbb939eabcca3e16711a4cadcadad50660a9e04 diff --git a/.agents/notes/implemented/architecture/2026-07-28-directory-picker-capability-seam.md b/.agents/notes/implemented/architecture/2026-07-28-directory-picker-capability-seam.md index bd1a0bc2f8..01968990db 100644 --- a/.agents/notes/implemented/architecture/2026-07-28-directory-picker-capability-seam.md +++ b/.agents/notes/implemented/architecture/2026-07-28-directory-picker-capability-seam.md @@ -20,8 +20,8 @@ Placement and policy rulings folded into this decision: - **Dependency survey (hand-roll vs adopt).** Node's stdlib *is* the maintained cross-platform OS layer (`readdir(withFileTypes)`, `homedir`, path semantics); surveyed alternatives fail the dependency bar — file-manager packages (`node-file-manager`, `files-and-folders`, Syncfusion's provider) are whole HTTP apps (fit), drive-letter helpers (`drivelist` native addon, `windows-drive-letters` ~7y stale) fail health/proportionality. The browse backend is a thin adapter over stdlib. - **Hidden entries: return-and-flag.** The host stamps `hidden` (POSIX dot convention) and returns everything; the client filters. Display policy stays client-side, and the show-hidden toggle shipped as exactly that client-only change: a fixed-label footer toggle whose state lives in the pressed presentation (`aria-pressed` + check glyph), a dot-led path-draft prefix reveals the hidden entries it names, and the current selection is exempt from both the hidden and the prefix filter (it anchors the two-pane view). Windows' `FILE_ATTRIBUTE_HIDDEN` is not exposed by dirents — documented limitation until a native probe pays for itself. - **Path-editor cancel scope: the dialog card.** The browse client's path editor cancels on Escape and on focus leaving the card, both observed at a card-scope wrapper rather than the input — after Tab parks focus on a filtered row the input is off the event path, yet Escape must collapse the editor (not the dialog) and a later focus departure must still cancel. Non-cancel exemptions: window/tab focus loss, in-card focus moves, and pointer paths (rows and the toggle suppress focus steal on mousedown while editing). Separators for seeding and draft-tail filtering are inferred from `listing.home`; the wire-field alternative below records the deferred authoritative form. Combobox semantics between the editor and the list it filters (`aria-expanded`/`aria-controls`/active-descendant, result announcements) are likewise deferred — today they read to assistive tech as separate widgets. -- **The path editor advertises itself, and the panes follow the draft.** The click-to-edit zone is not invisible: a pencil glyph sits at the bar's right edge and hover/focus lights the zone in the editor's own footprint, so the one route into typing a path is discoverable and the bar does not resize when zone and input swap. While the editor is open the panes track the draft instead of whatever level happened to be listed when it opened — the final segment prefix-filters the level its directory part names, a tail nobody matches releases the filter (a name still being spelled must not empty the pane it is being spelled into), and any other directory part is scanned after a 250ms rest and lands through the same selection-anchored, two-pane landing every navigation uses, so typing a path moves the Miller view exactly as a crumb jump does — typing deeper descends, erasing segments walks back up — without leaving the editor. **The pane arity is the invariant**: the last pane always lists the level the path names, with its parent beside it and nothing but a display root listing alone. Skipping the scan whenever *any* pane happened to list the directory was the cheaper rule and the wrong one — erasing a segment then left the level being typed on the left with its own child pane still standing to its right, so the panes stopped reading as "where I am, and where I came from". Only the last pane's own tail costs no scan. One landing shape, two callers: a submitted path closes the editor and announces failures, the draft-following scan keeps both to itself. That scan is speculative — half-typed directories are unreadable most of the time — so a failure keeps the last readable panes and stays silent. Enter remains the authoritative commit: it owns the view from submission until landing (a debounce timer armed by the same keystrokes is held back rather than superseding the navigation, and a rejected submission stays held until the next edit) and it alone surfaces the failure. Two consequences are deliberate. The wait is keyed on the draft, not on the directory part it names, so a keystroke that superseded an in-flight scan re-arms one and an edit after a rejected submission releases the hold; the panes it reads are a ref rather than a dependency, or the landing would re-arm the wait and a host answering with a differently spelled path would scan forever. And a walk is not rewound: closing the editor — cancellation included — leaves the panes where the draft took them, named by the crumbs and followed by Open's fallback target, because the operator watched them move. A landing that unmounts the row a keyboard operator Tabbed onto re-parks focus on the editor, since the Modal has no focus trap. -- **Navigation lands selection-anchored, quiet, and bounded.** Away from the display root (the same collapse the crumb header renders, so crumbs and pane shape never disagree), the landing is two-pane: the target's actual parent-level entry re-selected (platform case folding on Windows), its children on the right, so a crumb jump reads as stepping back one pane rather than collapsing to a single column. Target and parent legs land as **one frame** when the parent leg settles within the 200ms wait bound — the stale view keeps rendering until then, so navigation swaps the panes without an intermediate single-pane flash — and past the bound the target commits alone at once (an Enter-submitted navigation is never held hostage by a stalled parent) with the late parent leg upgrading the landing in place. The parent leg runs under the landing's supersession scope and is aborted on the wire by any newer intent (Escape inside the landing window therefore withdraws the whole navigation); a failed parent leg, or a truncated parent window lacking the target, leaves the single-pane landing — the upgrade must never orphan the selection it exists to anchor. The loading indicator follows the same quiet rule: it floats over the content's bottom-right corner (never a layout-shifting row; the truncated/error rows own the bottom left and keep rendering through a scan) and only once a scan outlives a 300ms silence window, so a local listing swaps with nothing shown at all. Row picks are deliberately exempt from the one-frame rule: a pick's immediate pane split is its selected-state feedback (aria-current, crumbs following), while a navigation has nothing to acknowledge the click but the swap itself. Both timing constants are calibrated for local enumeration; a remote deployment (one RPC per level, commonly 100–400ms) would sit inside the silence window with no pressed state on the crumbs — revisit the window or add pressed feedback when a remote consumer lands. +- **The path editor advertises itself, and the panes follow the draft.** The click-to-edit zone is not invisible: a pencil glyph sits at the bar's right edge and hover/focus lights the WHOLE bar in the editor's own box — the bar carries the outline and padding in both modes, so the hover previews exactly the field the click produces and nothing resizes when zone and input swap. While the editor is open the panes track the draft instead of whatever level happened to be listed when it opened — the final segment prefix-filters the level its directory part names, a tail nobody matches releases the filter (a name still being spelled must not empty the pane it is being spelled into), and any other directory part is scanned after a 250ms rest and lands through the same selection-anchored, two-pane landing every navigation uses, so typing a path moves the Miller view exactly as a crumb jump does — typing deeper descends, erasing segments walks back up — without leaving the editor. **The pane arity is the invariant**: the last pane always lists the level the path names, with its parent beside it and nothing but a display root listing alone. Skipping the scan whenever *any* pane happened to list the directory was the cheaper rule and the wrong one — erasing a segment then left the level being typed on the left with its own child pane still standing to its right, so the panes stopped reading as "where I am, and where I came from". Only the last pane's own tail costs no scan. One landing shape, two callers: a submitted path closes the editor and announces failures, the draft-following scan keeps both to itself. That scan is speculative — half-typed directories are unreadable most of the time — so a failure keeps the last readable panes and stays silent. Enter remains the authoritative commit: it owns the view from submission until landing (a debounce timer armed by the same keystrokes is held back rather than superseding the navigation, and a rejected submission stays held until the next edit) and it alone surfaces the failure. Two consequences are deliberate. The wait is keyed on the draft, not on the directory part it names, so a keystroke that superseded an in-flight scan re-arms one and an edit after a rejected submission releases the hold; the panes it reads are a ref rather than a dependency, or the landing would re-arm the wait and a host answering with a differently spelled path would scan forever. And a walk is not rewound: closing the editor — cancellation included — leaves the panes where the draft took them, named by the crumbs and followed by Open's fallback target, because the operator watched them move. A landing that unmounts the row a keyboard operator Tabbed onto re-parks focus on the editor, since the Modal has no focus trap. Two further rules keep one keystroke to one movement: the walk waits BOTH legs out rather than taking the submitted-navigation wait bound (nothing waits on a speculative scan, so landing single-pane and upgrading would be the very flash this exists to avoid, and it would strand the two-pane view whenever a tail keystroke aborted a slow parent leg), and the tail filters only the LAST pane — narrowing a pane the draft has walked away from would move the view once as it narrows and again as its landing replaces it. A level also keeps answering the directory text that produced it (`scanned`), because the Host resolves what it is given: `..` segments and, on Windows, forward slashes reach a level whose own path spells the request differently, and without the memo those drafts would rescan on every keystroke and never filter. +- **Navigation lands selection-anchored, quiet, and bounded.** Away from the display root (the same collapse the crumb header renders, so crumbs and pane shape never disagree), the landing is two-pane: the target's actual parent-level entry re-selected (platform case folding on Windows), its children on the right, so a crumb jump reads as stepping back one pane rather than collapsing to a single column. Target and parent legs land as **one frame** when the parent leg settles within the 200ms wait bound — the stale view keeps rendering until then, so navigation swaps the panes without an intermediate single-pane flash — and past the bound the target commits alone at once (an Enter-submitted navigation is never held hostage by a stalled parent) with the late parent leg upgrading the landing in place. The parent leg runs under the landing's supersession scope and is aborted on the wire by any newer intent (Escape inside the landing window therefore withdraws the whole navigation); a failed parent leg, or a truncated parent window lacking the target, leaves the single-pane landing — the upgrade must never orphan the selection it exists to anchor. The loading indicator follows the same quiet rule: it floats over the content's bottom-right corner (never a layout-shifting row; the truncated/error rows own the bottom left and keep rendering through a scan) and only once a scan outlives a 300ms silence window, so a local listing swaps with nothing shown at all. Row picks are deliberately exempt from the one-frame rule: a pick's immediate pane split is its selected-state feedback (aria-current, crumbs following), while a navigation has nothing to acknowledge the click but the swap itself. All three timing constants — the 200ms parent-leg bound, the 300ms silence window, and the editor's 250ms draft rest — are calibrated for local enumeration; a remote deployment (one RPC per level, commonly 100–400ms) would sit inside the silence window with no pressed state on the crumbs, and would pay rest plus RPC before the panes follow a typed path — revisit all three together when a remote consumer lands. - **Symlinks: follow for enterability.** `stat` probes symlinks (broken/cyclic → skipped); crumbs keep the logical path the operator navigated, and `workspace.create` already canonicalizes via realpath at adoption. - **Listing levels are bounded, and streamed.** One `list` call returns at most `maxEntries` rows (config, default 1000 — GitHub's web-UI directory-listing bound). The level streams via `opendir` into a name-sorted window of `maxEntries + 1` candidates, so memory stays O(maxEntries) and enterability probing touches only windowed candidates; the wire `DirectoryListing` carries a required `truncated` flag so the client states incompleteness instead of silently missing tail entries. A windowed broken symlink is not backfilled from beyond the window — the eviction already marks the level truncated. Window insertion is binary with an O(1) full-window tail rejection (an oversized level must not pay a window scan per dirent), and `list(path, signal)` threads the carrier's request signal so a scan of a stalled network directory cannot outlive a disconnected caller — every await in the scan (open, each read, each symlink probe) races the signal, an aborted exit abandons rather than awaits the close (Node queues close behind in-flight reads), and abandoned settlements are swallowed so cleanup can never surface as an unhandled rejection. An unbounded level is a memory/responsiveness hole for large or adversarial directories. - **Whole-filesystem scope, no roots config.** `workspace.create` accepts arbitrary paths and the API serves bash-driving methods, so a browse root would be UX scoping, not a boundary; configurability without a consumer fails the evidence bar. Deferred until a deployment needs it. diff --git a/.agents/notes/implemented/architecture/2026-07-28-directory-picker-capability-seam.zh.md b/.agents/notes/implemented/architecture/2026-07-28-directory-picker-capability-seam.zh.md index f836c00443..ffbb939eab 100644 --- a/.agents/notes/implemented/architecture/2026-07-28-directory-picker-capability-seam.zh.md +++ b/.agents/notes/implemented/architecture/2026-07-28-directory-picker-capability-seam.zh.md @@ -20,8 +20,8 @@ web GUI 的"打开本地文件夹"流程被焊死在一种交互上：`host.pick - **依赖调研（手写 vs 引入）。** Node 标准库本身就是维护中的跨平台 OS 层（`readdir(withFileTypes)`、`homedir`、路径语义）；调研过的替代品都过不了依赖门槛——文件管理器包（`node-file-manager`、`files-and-folders`、Syncfusion 的 provider）是整套 HTTP 应用（契合度不过），盘符工具（原生插件 `drivelist`、约七年未更的 `windows-drive-letters`）健康度／比例失当。browse 后端是标准库上的薄适配。 - **隐藏条目：返回并打标。** 宿主标注 `hidden`（POSIX 点前缀约定）并返回全部条目；客户端过滤。展示策略留在客户端，"显示隐藏"开关正是作为这一纯客户端改动落地：标签固定的 footer 开关，其状态由按下态呈现承载（`aria-pressed` + 勾选符号）；以点开头的路径草稿前缀会显出它所指名的隐藏条目；当前选中项则不受隐藏与前缀两种过滤影响（它锚定着双栏视图）。Windows 的 `FILE_ATTRIBUTE_HIDDEN` 不被 dirent 暴露——记为限制，直到原生探测值回其成本。 - **路径编辑器的取消范围：对话框卡片。** browse 客户端的路径编辑器在按 Escape 与焦点离开卡片时取消，两者都在卡片范围的包装层而非输入框上监听——Tab 把焦点停到某个过滤命中的行之后，输入框已不在事件路径上，但 Escape 仍须收起编辑器（而非对话框），其后的焦点离开也仍须取消。不取消的豁免：窗口／标签页失焦、卡片内焦点移动，以及指针路径（编辑期间行与开关在 mousedown 时抑制焦点夺取）。预填与草稿末段过滤所用的分隔符从 `listing.home` 推断；下文的线上字段替代方案记录了被延期的权威形态。编辑器与其过滤的列表之间的 combobox 语义（`aria-expanded`／`aria-controls`／active-descendant、结果播报）同样被延期——目前二者在辅助技术看来是彼此独立的控件。 -- **路径编辑器自我点明，各栏跟随草稿。** 点击即编辑的区域不再是隐形的：栏右端坐着一枚铅笔图标，悬停／聚焦时该区以编辑器自身的轮廓亮起，于是键入路径这唯一入口可被发现，且区域与输入框互换时栏高不变。编辑器打开期间，各栏跟随草稿，而不是停在它打开那一刻恰好列出的层级——末段对其目录部分所指的层级做前缀过滤，无一匹配的末段解除过滤（还在拼写中的名字不该把正在拼写它的那一栏清空），而其余任何目录部分都会在停顿 250ms 后被扫描，并经由每次导航共用的那套以选中项为锚的双栏落地落定，于是键入路径移动 Miller 视图的方式与 crumb 跳转完全一致——继续键入即下潜、删掉末段即上退——全程不必离开编辑器。**分栏个数才是不变量**：最后一栏永远是路径所指的那一层，其上一层在它旁边，只有展示根会独占一栏。"只要任意一栏碰巧列出了该目录就跳过扫描"是更省事、也是错的规则——删掉一段之后，正在键入的那一层会留在左栏，而它自己的子栏仍立在右边，于是两栏不再读作"我在哪儿、我从哪儿来"。只有最后一栏自己的末段不需要扫描。一种落地形态、两个调用方：提交的路径关闭编辑器并呈现失败，草稿跟随扫描则两者都不做。该扫描是推测性的——键入到一半的目录多数时候读不出来——因此失败时保留最后一次可读的分栏并保持沉默。Enter 仍是权威提交：自提交至落地由它独占视图（同一批按键武装的防抖计时器会被扣住，而不是顶掉这次导航；提交被拒后仍扣住，直到下一次编辑），也只有它把失败呈现出来。有两点是刻意为之。等待以草稿为键，而非以它指名的目录部分为键，于是顶掉在飞扫描的那次按键会重新武装等待，被拒提交之后的编辑也能释放那道扣留；而它读取的分栏是 ref 而非依赖，否则落地会重新武装等待，遇到以不同拼写作答的宿主便会永远扫描下去。以及，走过的路不回退：关闭编辑器——包括取消——都把分栏留在草稿带到的地方，由面包屑指明、Open 的兜底目标随之而动，因为操作者亲眼看着它们移动。若落地卸载了键盘操作者 Tab 停留的那一行，焦点会被重新停回编辑器——Modal 并没有焦点陷阱。 -- **导航以选中项为锚、安静且有界地落地。** 在展示根之外（与 crumb 头部渲染的是同一塌缩，因此 crumb 与分栏形态永不相左），落地即双栏：重新选中目标在父层级中的实际条目（Windows 上按平台惯例折叠大小写），右侧展示其子项，因此 crumb 跳转读作后退一栏，而不是塌缩成单列。父层级这一程在 200ms 等待上限内落定时，目标与父层级两程以**同一帧**落地——在此之前陈旧视图持续渲染，导航换栏时因此没有中间的单栏闪现——超出该上限则目标即刻单独提交（Enter 提交的导航绝不会被滞塞的父层级扣作人质），迟到的父层级这一程再就地升级这次落地。父层级这一程在落地的 supersession 范围下运行，任何较新的意图都会在线上将其中止（因此在落地窗口内按 Escape 即撤回整次导航）；父层级这一程失败，或被截断的父窗口缺少目标时，都保留单栏落地——升级的存在正是为了锚定选中项，绝不能反而让它悬空。加载指示器遵循同一安静规则：它浮于内容右下角（绝不是会挪动布局的一行；截断／错误行占据左下角，并在扫描期间持续渲染），且仅在扫描超出 300ms 静默窗口后才出现，因此本地列举切换时什么也不显示。行选取被刻意豁免于同一帧规则：选取后立即分栏本身就是其选中态反馈（aria-current、crumb 跟随），而导航除了换栏本身没有任何东西可确认这次点击。两个时序常量都按本地列举校准；远程部署（每层级一次 RPC，通常 100–400ms）会落在静默窗口之内、crumb 上却没有按下态——待远程消费方落地时，重新审视该窗口或补上按下反馈。 +- **路径编辑器自我点明，各栏跟随草稿。** 点击即编辑的区域不再是隐形的：栏右端坐着一枚铅笔图标，悬停／聚焦时**整条栏**以编辑器自身的那只框亮起——轮廓与内边距在两种模式下都由栏承载，于是悬停预览的正是点击后出现的那只输入框，区域与输入框互换时也没有任何尺寸变化。编辑器打开期间，各栏跟随草稿，而不是停在它打开那一刻恰好列出的层级——末段对其目录部分所指的层级做前缀过滤，无一匹配的末段解除过滤（还在拼写中的名字不该把正在拼写它的那一栏清空），而其余任何目录部分都会在停顿 250ms 后被扫描，并经由每次导航共用的那套以选中项为锚的双栏落地落定，于是键入路径移动 Miller 视图的方式与 crumb 跳转完全一致——继续键入即下潜、删掉末段即上退——全程不必离开编辑器。**分栏个数才是不变量**：最后一栏永远是路径所指的那一层，其上一层在它旁边，只有展示根会独占一栏。"只要任意一栏碰巧列出了该目录就跳过扫描"是更省事、也是错的规则——删掉一段之后，正在键入的那一层会留在左栏，而它自己的子栏仍立在右边，于是两栏不再读作"我在哪儿、我从哪儿来"。只有最后一栏自己的末段不需要扫描。一种落地形态、两个调用方：提交的路径关闭编辑器并呈现失败，草稿跟随扫描则两者都不做。该扫描是推测性的——键入到一半的目录多数时候读不出来——因此失败时保留最后一次可读的分栏并保持沉默。Enter 仍是权威提交：自提交至落地由它独占视图（同一批按键武装的防抖计时器会被扣住，而不是顶掉这次导航；提交被拒后仍扣住，直到下一次编辑），也只有它把失败呈现出来。有两点是刻意为之。等待以草稿为键，而非以它指名的目录部分为键，于是顶掉在飞扫描的那次按键会重新武装等待，被拒提交之后的编辑也能释放那道扣留；而它读取的分栏是 ref 而非依赖，否则落地会重新武装等待，遇到以不同拼写作答的宿主便会永远扫描下去。以及，走过的路不回退：关闭编辑器——包括取消——都把分栏留在草稿带到的地方，由面包屑指明、Open 的兜底目标随之而动，因为操作者亲眼看着它们移动。若落地卸载了键盘操作者 Tab 停留的那一行，焦点会被重新停回编辑器——Modal 并没有焦点陷阱。另有两条规则保证一次按键只让视图移动一次：这段行走会**等齐两程**，而不套用提交导航的等待上限（推测性扫描没有任何东西在等它，先落单栏再升级恰恰就是它要避免的那次闪动，而且一旦末段按键中止了缓慢的父层级这一程，双栏视图就会永久丢失）；末段也只过滤**最后一栏**——去收窄一个草稿已经走开的分栏，会让视图先因收窄动一次、再因它自己的落地动一次。此外，层级会持续应答产生它的那段目录文本（`scanned`），因为宿主会规范化它收到的东西：`..` 段与 Windows 的正斜杠都会抵达一个自身路径拼写不同的层级；没有这份记忆，这类草稿会每敲一键就重扫一次，而且永远过滤不了。 +- **导航以选中项为锚、安静且有界地落地。** 在展示根之外（与 crumb 头部渲染的是同一塌缩，因此 crumb 与分栏形态永不相左），落地即双栏：重新选中目标在父层级中的实际条目（Windows 上按平台惯例折叠大小写），右侧展示其子项，因此 crumb 跳转读作后退一栏，而不是塌缩成单列。父层级这一程在 200ms 等待上限内落定时，目标与父层级两程以**同一帧**落地——在此之前陈旧视图持续渲染，导航换栏时因此没有中间的单栏闪现——超出该上限则目标即刻单独提交（Enter 提交的导航绝不会被滞塞的父层级扣作人质），迟到的父层级这一程再就地升级这次落地。父层级这一程在落地的 supersession 范围下运行，任何较新的意图都会在线上将其中止（因此在落地窗口内按 Escape 即撤回整次导航）；父层级这一程失败，或被截断的父窗口缺少目标时，都保留单栏落地——升级的存在正是为了锚定选中项，绝不能反而让它悬空。加载指示器遵循同一安静规则：它浮于内容右下角（绝不是会挪动布局的一行；截断／错误行占据左下角，并在扫描期间持续渲染），且仅在扫描超出 300ms 静默窗口后才出现，因此本地列举切换时什么也不显示。行选取被刻意豁免于同一帧规则：选取后立即分栏本身就是其选中态反馈（aria-current、crumb 跟随），而导航除了换栏本身没有任何东西可确认这次点击。三个时序常量——200ms 父层级上限、300ms 静默窗口，以及编辑器的 250ms 草稿停顿——都按本地列举校准；远程部署（每层级一次 RPC，通常 100–400ms）会落在静默窗口之内、crumb 上却没有按下态，而且要先付停顿再付 RPC 分栏才跟上——待远程消费方落地时，三者一并重新审视。 - **符号链接：为可进入性而跟随。** 用 `stat` 探测符号链接（断链／循环→跳过）；面包屑保留操作者导航的逻辑路径，`workspace.create` 在接纳时本就做 realpath 规范化。 - **列举层级有上限，且流式处理。** 单次 `list` 至多返回 `maxEntries` 行（配置项，默认 1000——GitHub 网页端目录列举的同一上限）。层级经 `opendir` 流入一个按名排序、容量 `maxEntries + 1` 的候选窗口，内存保持 O(maxEntries)，可进入性探测只触及窗口内候选；线上 `DirectoryListing` 携带必填的 `truncated` 标志，让客户端明示不完整而不是静默缺尾。窗口内的断链符号链接不从窗口外回填——发生过驱逐本身已把层级标记为截断。窗口插入为二分查找、满窗尾部单次比较即拒绝（超大层级不能为每个 dirent 付出一次全窗扫描），且 `list(path, signal)` 透传载体的请求信号，滞塞网络目录的扫描不会在调用方断连后继续存活——扫描中的每个 await（打开、每次读取、每次符号链接探测）都与信号赛跑，中止路径放弃而非等待 close（Node 会把 close 排在在飞读取之后），被放弃的 settlement 全部吞掉，清理不会以未处理拒绝的形式冒出。无上限的层级对超大或恶意构造的目录就是内存／响应性漏洞。 - **全盘可浏览，不做 roots 配置。** `workspace.create` 接受任意路径且 API 本就提供驱动 bash 的方法，浏览根只会是 UX 范围而非边界；没有消费方的可配置性过不了证据门槛。等到有部署需要再做。 diff --git a/packages/host/directory-picker-browse/README.i18n.yaml b/packages/host/directory-picker-browse/README.i18n.yaml index 4b4f10ee23..7c2df43ab2 100644 --- a/packages/host/directory-picker-browse/README.i18n.yaml +++ b/packages/host/directory-picker-browse/README.i18n.yaml @@ -2,5 +2,5 @@ # side as of the last confirmed-consistent state. Both languages carry equal authority; # after editing either side, bring the other along and re-record with: # pnpm run verify-translation-pairing --write packages/host/directory-picker-browse/README.md -README.md: 04d97adf71ae4a3ea8d89b24098f27cbdee20961 -README.zh.md: 603afaed7e4b4c77e699e009fc612e73c06d73aa +README.md: 62384cc0b0e5756e56c1d608252721c506a0915f +README.zh.md: 8f495e1e4d87486d0565eadcbf7df694494c7096 diff --git a/packages/host/directory-picker-browse/README.md b/packages/host/directory-picker-browse/README.md index 04d97adf71..62384cc0b0 100644 --- a/packages/host/directory-picker-browse/README.md +++ b/packages/host/directory-picker-browse/README.md @@ -6,7 +6,7 @@ The **in-app browsing backend** of the [directory-picker seam](../directory-pick Behavior facts: listings return **directories only**, name-sorted, with symlinks-to-directories followed (broken/cyclic links skipped — the probe `stat` failing means "not enterable") and a host-owned `hidden` flag (POSIX dot convention) left for the client to act on; `crumbs` is the root-to-target ancestor chain, the root crumb labeled by its full path (`/`, `C:\`); an absent `list` path means the host account's home directory. `createDirectory` is non-recursive (a missing parent is a real failure, not a level to invent) and validates the name as a single non-blank segment even when called directly, mirroring the wire schema's fence. Both primitives reject an explicit path that is not fully qualified — relative forms, and on Windows the rooted drive-less forms (`\foo`, `/foo`) and incomplete UNC prefixes (`\\`, `\\server`) that `isAbsolute` accepts — with `directory-unreadable`/`directory-create-failed`, instead of letting `resolve` rebase it under the host process cwd or current drive. One `list` call returns at most `maxEntries` rows (config, default 1000 — the bound GitHub's web UI applies to directory listings), and the level streams through a bounded window so memory stays O(maxEntries) no matter how many children the directory holds: a cut level keeps the name-sorted head, counts hidden rows against the bound, probes only windowed candidates, and reports `truncated: true` so the client can say the level is incomplete (a windowed broken symlink is not backfilled from beyond the window — the eviction already marks the level truncated); window insertion is binary with an O(1) full-window tail rejection, and `list` threads the caller's `AbortSignal` so a disconnect or timeout stops the scan instead of letting it outlive the caller. Failures throw the seam's typed `DirectoryPickerError`. Policy rationale: [the directory-picker capability seam Agent Note](../../../.agents/notes/implemented/architecture/2026-07-28-directory-picker-capability-seam.md). -**Dual-face package**: the browser half (`./client`) fills [ui-workspace's](../../client/ui-workspace/README.md) two directory-flow holes with the in-app **Select Workspace Directory** dialog (figma `Harness` 813-23126 family — Miller two-column view whose navigations land selection-anchored and quiet: the previous view keeps rendering while a crumb jump or a submitted path is scanned (a "Loading…" pill floats over it only once the scan outlives a 300ms silence window, never shifting the columns), then target and parent legs land as one two-pane frame with the target re-selected as its actual parent-level entry — so stepping back never collapses and no intermediate frame flashes (a parent leg outliving its 200ms wait bound lands the target alone and upgrades in place; a failed or truncated parent leg keeps the single-pane landing; the display root keeps the single wide level); breadcrumb with a click-to-edit path zone, advertised by the pencil glyph at the bar's right edge and lit on hover in the editor's own footprint, whose editor seeds a trailing separator and then keeps the panes under the draft: the final segment prefix-filters the level its directory part names (case-insensitively, over the listed — possibly truncated — rows only; a tail nobody matches releases the filter instead of emptying the pane), while any other directory part is scanned after a 250ms rest and lands like any other navigation — selection-anchored, two-pane away from the display root — so typing deeper descends and erasing segments walks back up, moving the Miller view without leaving the editor; the pane arity is the invariant, the last pane always listing the level the path names with its parent beside it (only that level's own tail costs no scan, and only a display root lists alone) — a speculative scan is silent when it fails, and Enter still navigates by the exact text, owning the view until it lands; the editor cancels on Escape or when focus leaves the dialog card (window/tab switches and in-card focus moves keep the draft), and panes the draft walked to stay where the walk ended — the crumbs name that level and Open's fallback target follows them, so cancelling closes the editor rather than rewinding the walk; a fixed-label show-hidden footer toggle over the host's `hidden` flags, with a dot-led typed prefix revealing its matches and the current selection exempt from every filter; nested New-folder dialog), driving `host.listDirectory`/`host.createDirectory` and registering its own locale namespace (`directory-browser`, zh default / en). One cordis.yml row therefore composes both sides of the browse interaction; the client carries no capability-kind branching, and mounting a second flow package fails at load (the holes are `single` kind). +**Dual-face package**: the browser half (`./client`) fills [ui-workspace's](../../client/ui-workspace/README.md) two directory-flow holes with the in-app **Select Workspace Directory** dialog (figma `Harness` 813-23126 family — Miller two-column view whose navigations land selection-anchored and quiet: the previous view keeps rendering while a crumb jump or a submitted path is scanned (a "Loading…" pill floats over it only once the scan outlives a 300ms silence window, never shifting the columns), then target and parent legs land as one two-pane frame with the target re-selected as its actual parent-level entry — so stepping back never collapses and no intermediate frame flashes (a parent leg outliving its 200ms wait bound lands the target alone and upgrades in place; a failed or truncated parent leg keeps the single-pane landing; the display root keeps the single wide level); breadcrumb with a click-to-edit path zone, advertised by the pencil glyph at the bar's right edge and lighting the whole bar — the editor's own box — on hover, whose editor seeds a trailing separator and then keeps the panes under the draft: the final segment prefix-filters the LAST pane while that pane lists the level the directory part names (case-insensitively, over the listed — possibly truncated — rows only; a tail nobody matches releases the filter instead of emptying the pane), while any other directory part is scanned after a 250ms rest and lands like any other navigation — selection-anchored, two-pane away from the display root, both legs waited out so one keystroke moves the view once — so typing deeper descends and erasing segments walks back up without leaving the editor; the pane arity is the invariant, the last pane always listing the level the path names with its parent beside it (only that level's own tail costs no scan, and only a display root lists alone), and a level still answers the text that produced it after the Host resolved it (`..` segments, Windows forward slashes) — a speculative scan is silent when it fails, and Enter still navigates by the exact text, owning the view until it lands; the editor cancels on Escape or when focus leaves the dialog card (window/tab switches and in-card focus moves keep the draft), and panes the draft walked to stay where the walk ended — the crumbs name that level and Open's fallback target follows them, so cancelling closes the editor rather than rewinding the walk; a fixed-label show-hidden footer toggle over the host's `hidden` flags, with a dot-led typed prefix revealing its matches and the current selection exempt from every filter; nested New-folder dialog), driving `host.listDirectory`/`host.createDirectory` and registering its own locale namespace (`directory-browser`, zh default / en). One cordis.yml row therefore composes both sides of the browse interaction; the client carries no capability-kind branching, and mounting a second flow package fails at load (the holes are `single` kind). ## Model Experience diff --git a/packages/host/directory-picker-browse/README.zh.md b/packages/host/directory-picker-browse/README.zh.md index 603afaed7e..8f495e1e4d 100644 --- a/packages/host/directory-picker-browse/README.zh.md +++ b/packages/host/directory-picker-browse/README.zh.md @@ -6,7 +6,7 @@ 行为事实：列举**只返回目录**、按名称排序，指向目录的符号链接会被跟随（断链／循环链接被跳过——探测 `stat` 失败即"不可进入"），并携带宿主判定的 `hidden` 标志（POSIX 点前缀约定），展示决策留给客户端；`crumbs` 是从根到目标的祖先链，根 crumb 以完整路径标注（`/`、`C:\`）；`list` 不带路径即列举宿主账户的家目录。`createDirectory` 不递归（父目录缺失是真实失败，不是要补造的层级），且即便被直接调用也把名称校验为单个非空段，与协议 schema 的栅栏一致。两个原语都拒绝非完全限定的显式路径——相对形态，以及 Windows 上 `isAbsolute` 会放行的无盘符有根形态（`\foo`、`/foo`）与不完整的 UNC 前缀（`\\`、`\\server`）——报 `directory-unreadable`／`directory-create-failed`，而不是任由 `resolve` 把它重定位到宿主进程 cwd 或当前盘符之下。单次 `list` 至多返回 `maxEntries` 行（配置项，默认 1000——GitHub 网页端对目录列举采用的同一上限），且层级以流式方式经过一个有界窗口，无论目录有多少子项内存都保持 O(maxEntries)：被截断的层级保留按名排序的头部、隐藏行计入上限、只探测窗口内候选，并报告 `truncated: true`，供客户端提示层级不完整（窗口内的断链符号链接不会从窗口外回填——发生过驱逐本身已把层级标记为截断）；窗口插入为二分查找、满窗尾部单次比较即拒绝，且 `list` 透传调用方的 `AbortSignal`，断连或超时会停止扫描而不是让它在调用方离开后继续。失败抛出 seam 的类型化 `DirectoryPickerError`。策略依据：[目录选择能力 seam Agent Note](../../../.agents/notes/implemented/architecture/2026-07-28-directory-picker-capability-seam.md)。 -**双面包**：browser half（`./client`）以应用内 **选择工作区目录** 对话框（figma `Harness` 813-23126 家族——Miller 双列视图，其导航以选中项为锚、安静落地：扫描 crumb 跳转或提交的路径期间，先前视图持续渲染（"Loading…" 胶囊仅在扫描超出 300ms 静默窗口后才浮于其上，绝不挪动各列），随后目标与父层级两程以单个双栏帧落地，目标被重新选中为其在父层级中的实际条目——因此后退绝不塌缩，也没有中间帧闪现（父层级这一程超出其 200ms 等待上限时，目标单独落地，随后就地升级；父层级这一程失败或被截断时保持单栏落地；展示根保持单个宽层级）；带点击即编辑路径区的面包屑，该区由栏右端的铅笔图标点明、悬停时以编辑器自身的轮廓亮起，其编辑器预填尾随分隔符，随后让下方各栏跟随草稿：末段对其目录部分所指的层级做前缀过滤（不区分大小写，且仅作用于已列出、可能被截断的行；无一匹配的末段会解除过滤，而不是把该栏清空），而其余任何目录部分都会在停顿 250ms 后被扫描，并像其他任何一次导航那样落地——以选中项为锚，在展示根之外即双栏——于是继续键入即下潜、删掉末段即上退，Miller 视图随之移动而不必离开编辑器；分栏个数是这里的不变量：最后一栏永远是路径所指的那一层，其上一层在它旁边（只有这一层自己的末段不触发扫描，也只有展示根会独占一栏）——推测性扫描失败时保持沉默，而 Enter 仍按确切文本导航，并在落地前独占视图；编辑器按 Escape 或焦点离开对话框卡片即取消（窗口／标签页切换与卡片内焦点移动保留草稿），而草稿走到的层级会留在原地——面包屑指明该层级、Open 的兜底目标随之而动，因此取消只是关闭编辑器，并不回退这段行走；基于宿主 `hidden` 标志、标签固定的"显示隐藏"footer 开关，键入以点开头的前缀会显出其匹配项，且当前选中项不受任何过滤影响；嵌套新建文件夹对话框）填入 [ui-workspace](../../client/ui-workspace/README.md) 的两个目录流洞，驱动 `host.listDirectory`／`host.createDirectory`，并注册自己的 locale 命名空间（`directory-browser`，zh 默认／en）。因此一行 cordis.yml 同时组合浏览交互的两侧；client 侧不含任何能力 kind 分支，挂载第二个流程包会在加载期失败（洞为 `single` kind）。 +**双面包**：browser half（`./client`）以应用内 **选择工作区目录** 对话框（figma `Harness` 813-23126 家族——Miller 双列视图，其导航以选中项为锚、安静落地：扫描 crumb 跳转或提交的路径期间，先前视图持续渲染（"Loading…" 胶囊仅在扫描超出 300ms 静默窗口后才浮于其上，绝不挪动各列），随后目标与父层级两程以单个双栏帧落地，目标被重新选中为其在父层级中的实际条目——因此后退绝不塌缩，也没有中间帧闪现（父层级这一程超出其 200ms 等待上限时，目标单独落地，随后就地升级；父层级这一程失败或被截断时保持单栏落地；展示根保持单个宽层级）；带点击即编辑路径区的面包屑，该区由栏右端的铅笔图标点明，悬停时整条栏——也就是编辑器自身的那只框——亮起，其编辑器预填尾随分隔符，随后让下方各栏跟随草稿：当最后一栏正是目录部分所指的层级时，末段对这一栏做前缀过滤（不区分大小写，且仅作用于已列出、可能被截断的行；无一匹配的末段会解除过滤，而不是把该栏清空），而其余任何目录部分都会在停顿 250ms 后被扫描，并像其他任何一次导航那样落地——以选中项为锚，在展示根之外即双栏，且两程都等齐，于是一次按键只让视图移动一次——继续键入即下潜、删掉末段即上退，全程不必离开编辑器；分栏个数是这里的不变量：最后一栏永远是路径所指的那一层，其上一层在它旁边（只有这一层自己的末段不触发扫描，也只有展示根会独占一栏），而宿主规范化过路径之后（`..` 段、Windows 的正斜杠），该层级仍然应答产生它的那段文本——推测性扫描失败时保持沉默，而 Enter 仍按确切文本导航，并在落地前独占视图；编辑器按 Escape 或焦点离开对话框卡片即取消（窗口／标签页切换与卡片内焦点移动保留草稿），而草稿走到的层级会留在原地——面包屑指明该层级、Open 的兜底目标随之而动，因此取消只是关闭编辑器，并不回退这段行走；基于宿主 `hidden` 标志、标签固定的"显示隐藏"footer 开关，键入以点开头的前缀会显出其匹配项，且当前选中项不受任何过滤影响；嵌套新建文件夹对话框）填入 [ui-workspace](../../client/ui-workspace/README.md) 的两个目录流洞，驱动 `host.listDirectory`／`host.createDirectory`，并注册自己的 locale 命名空间（`directory-browser`，zh 默认／en）。因此一行 cordis.yml 同时组合浏览交互的两侧；client 侧不含任何能力 kind 分支，挂载第二个流程包会在加载期失败（洞为 `single` kind）。 ## 模型体验 diff --git a/packages/host/directory-picker-browse/src/client/DirectoryBrowser.module.css b/packages/host/directory-picker-browse/src/client/DirectoryBrowser.module.css index eaad4d46d8..b5fbb611a4 100644 --- a/packages/host/directory-picker-browse/src/client/DirectoryBrowser.module.css +++ b/packages/host/directory-picker-browse/src/client/DirectoryBrowser.module.css @@ -49,12 +49,29 @@ color: var(--dsw-alias-label-primary); } +/* The bar IS the editor's box in both modes: it carries the rounded outline + * and the inner padding, the crumbs and the input sit inside it, and hovering + * the edit zone lights the whole row rather than the remainder right of the + * crumbs. The negative left margin pays back the border and padding, so the + * crumb (and input) text keeps the column the title sits in. */ .crumbBar { display: flex; align-items: center; gap: 4px; - /* The path editor's height: crumb mode and edit mode occupy the same bar. */ + box-sizing: border-box; min-height: 24px; + margin-left: -9px; + padding: 0 8px; + border: 1px solid transparent; + border-radius: 8px; +} + +/* Lit by the affordance the row belongs to, never by a crumb: a crumb's hover + * offers navigation, not path entry. Editing keeps the outline standing. */ +.crumbBar:has(.crumbEditZone:enabled:hover), +.crumbBar:has(.crumbEditZone:focus-visible), +.crumbBar:has(.pathInput) { + border-color: var(--dsw-alias-border-l2); } /* Deep chains scroll inside the trail (the effect pins the tail into view) @@ -119,30 +136,21 @@ color: var(--dsw-alias-label-tertiary); } -/* The empty remainder of the bar: a real click target that flips the bar - * into path-edit mode. The zone itself stays flush with the crumbs; the - * pencil glyph seated at its right edge is the standing affordance, and - * hover/focus lights the zone in the editor's own rounded shape so the - * gesture reads before the click. */ +/* The empty remainder of the bar: a real click target that flips the bar into + * path-edit mode. The pencil glyph seated at its right edge is the standing + * affordance; the outline the gesture lights belongs to the bar, so the whole + * row reads as the box the input will occupy. */ .crumbEditZone { display: flex; align-items: center; justify-content: flex-end; flex: 1 0 34px; min-width: 34px; - /* The editor's own height, so hover previews the input's exact footprint - * and the bar does not resize when the two swap. */ - height: 24px; - padding: 0 6px; - border: 1px solid transparent; - border-radius: 8px; + height: 22px; + padding: 0; + border: none; background: transparent; cursor: text; -} - -.crumbEditZone:hover, -.crumbEditZone:focus-visible { - border-color: var(--dsw-alias-border-l2); outline: none; } @@ -151,13 +159,12 @@ color: var(--dsw-alias-label-tertiary); } -.crumbEditZone:hover .crumbEditGlyph, +.crumbEditZone:enabled:hover .crumbEditGlyph, .crumbEditZone:focus-visible .crumbEditGlyph { color: var(--dsw-alias-label-primary); } .crumbEditZone:disabled { - border-color: transparent; cursor: default; } @@ -165,14 +172,14 @@ color: var(--dsw-alias-label-caption); } +/* Chrome-free: the bar around it draws the box (border, radius, padding). */ .pathInput { box-sizing: border-box; flex: 1 1 0; min-width: 0; - height: 24px; - padding: 0 8px; - border: 1px solid var(--dsw-alias-border-l2); - border-radius: 8px; + height: 22px; + padding: 0; + border: none; outline: none; background: transparent; font-size: 13px; diff --git a/packages/host/directory-picker-browse/src/client/DirectoryBrowser.tsx b/packages/host/directory-picker-browse/src/client/DirectoryBrowser.tsx index eaada21965..fc6dde82f1 100644 --- a/packages/host/directory-picker-browse/src/client/DirectoryBrowser.tsx +++ b/packages/host/directory-picker-browse/src/client/DirectoryBrowser.tsx @@ -19,18 +19,20 @@ * error surface. Hidden entries are host-flagged and hidden by default; the * footer's fixed-label "Show hidden files" toggle (aria-pressed, check when * on) reveals them (client-side only). The path editor announces itself with - * a pencil glyph and a hover-lit zone, opens seeded with a trailing - * separator, and keeps the panes under the draft: the final segment - * prefix-filters the level its directory part names (a dot-led prefix also - * reveals the hidden entries it names, and a prefix nobody matches releases - * the filter), while any other directory part is scanned after a short - * debounce and lands like any other navigation — selection-anchored and - * two-pane away from the display root. The pane arity holds throughout: the - * last pane is the level the path names and the one beside it is its parent, - * so typing deeper descends and erasing segments walks back up, moving the - * Miller view without leaving the editor. Panes the draft walked to stay put - * when the editor closes (cancellation included): the crumbs name where the - * walk ended, and Open's fallback target follows them. + * a pencil glyph and a bar-wide hover-lit outline, opens seeded with a + * trailing separator, and keeps the panes under the draft: the final segment + * prefix-filters the LAST pane while that pane's level is the one the draft's + * directory part names (a dot-led prefix also reveals the hidden entries it + * names, and a prefix nobody matches releases the filter), while any other + * directory part is scanned after a short debounce and lands like any other + * navigation — selection-anchored and two-pane away from the display root, + * both legs waited out so one keystroke moves the view once. The pane arity + * holds throughout: the last pane is the level the path names and the one + * beside it is its parent, so typing deeper descends and erasing segments + * walks back up, moving the Miller view without leaving the editor. Panes the + * draft walked to stay put when the editor closes (cancellation included): + * the crumbs name where the walk ended, and Open's fallback target follows + * them. */ import { useCallback, useEffect, useRef, useState } from 'react' import clsx from 'clsx' @@ -125,63 +127,65 @@ function levelDirectory(listing: DirectoryListing): string { return listing.path.endsWith(sep) ? listing.path : `${listing.path}${sep}` } +/** The directory text a draft-following scan last sent, with the level path the host answered it with. */ +interface ScannedDirectory { + /** The draft's directory part, verbatim as it went to the host. */ + readonly directory: string + /** `path` of the listing that came back. */ + readonly landed: string +} + /** * The draft's directory part — everything through its last separator — or * null while no separator has been typed at all (nothing addresses a - * directory yet). The platform separator comes from `listing`, so the caller - * passes any listing of the host's filesystem. + * directory yet). The platform comes from `listing`: on Windows a forward + * slash separates too (the host's `resolve` accepts either), while on POSIX a + * backslash is a legal name character and never separates. */ function draftDirectory(listing: DirectoryListing, draft: string): string | null { - const cut = draft.lastIndexOf(separatorOf(listing)) + const cut = separatorOf(listing) === '\\' + ? Math.max(draft.lastIndexOf('\\'), draft.lastIndexOf('/')) + : draft.lastIndexOf('/') return cut === -1 ? null : draft.slice(0, cut + 1) } /** - * The path draft's final segment, when its directory part is exactly the - * level `listing` lists — the segment the level prefix-filters on while the - * user types. Any other draft (no separator yet, or naming some other - * directory) leaves the level unfiltered. The directory part compares - * exactly (it is the host's own path text, reached by seeding, erasing, or a - * draft-following scan); only the name filter downstream is case-insensitive. + * How the draft reads against one level: the directory part it names, and — + * when `listing` is the level that directory part addresses — the final + * segment that prefix-filters it while the user types (case-insensitively, + * downstream). A level answers a directory part when its own path is that + * part, or when it is the level that very text just produced (`scanned`): the + * host resolves what it is given, so `..` segments and Windows forward + * slashes reach a level whose path spells the request differently. + * @param listing - the level to read the draft against. + * @param draft - the current path draft. + * @param scanned - the last draft-following scan's directory and landing. + * @returns the draft's directory part (null with no separator typed) and its + * filtering tail (null when this level does not answer that directory). */ -function draftPrefixFor(listing: DirectoryListing, draft: string | null): string | null { - if (draft === null) return null - const directory = draftDirectory(listing, draft) - if (directory === null) return null - return directory === levelDirectory(listing) ? draft.slice(directory.length) : null -} - -/** - * The directory a draft addresses that the panes are not already presenting - * as the current level — what the editor must scan to keep the view under the - * typed path. The pane arity is the invariant this preserves: the LAST pane - * always lists the level the path names, with its parent beside it (a display - * root lists alone), so a draft naming any other level re-lands rather than - * leaving a deeper level standing to the right of the one being typed. Null - * when that level is already the last pane, when no separator has been typed - * yet, and when no level is listed at all: the platform separator is read off - * a listing, so the editor's failed-home-listing recovery path types blind - * until Enter. - */ -function pendingPreviewDirectory( - parent: DirectoryListing | null, - child: DirectoryListing | null, +function readDraft( + listing: DirectoryListing, draft: string, -): string | null { - if (parent === null) return null - const directory = draftDirectory(parent, draft) - if (directory === null) return null - return directory === levelDirectory(child ?? parent) ? null : directory + scanned: ScannedDirectory | null, +): { directory: string | null; tail: string | null } { + const directory = draftDirectory(listing, draft) + if (directory === null) return { directory: null, tail: null } + const answers = directory === levelDirectory(listing) + || (scanned !== null && scanned.directory === directory && scanned.landed === listing.path) + return { directory, tail: answers ? draft.slice(directory.length) : null } } /** * The rows one column renders. The selection is exempt from every filter: it * anchors the two-pane view (crumbs and the child pane point at it), so * neither the hidden filter after a dot-reveal pick nor a prefix miss may - * orphan it. A prefix narrows the level only while some row matches it — a - * tail nobody matches is a name being spelled, not a demand for an empty - * pane, so the level shows whole (and its hidden rows return to obeying the - * toggle, the dot-led reveal included). + * orphan it. A prefix narrows the level only while some row it would actually + * show matches — a tail nobody matches is a name being spelled, not a demand + * for an empty pane, so the level shows whole and its hidden rows return to + * obeying the toggle. Counting only displayable rows is what keeps that true: + * were a hidden row ever to match a prefix that does not reveal it (today + * `hidden` means dot-prefixed, so it cannot), the level would narrow to + * nothing. */ function visibleEntries( entries: readonly DirectoryEntry[], @@ -190,15 +194,15 @@ function visibleEntries( filterPrefix: string | null, ): readonly DirectoryEntry[] { const needle = filterPrefix === null ? '' : filterPrefix.toLowerCase() - const matches = (entry: DirectoryEntry): boolean => entry.name.toLowerCase().startsWith(needle) - const narrowing = needle !== '' && entries.some(matches) // A dot-led prefix names hidden entries explicitly, so matching ones // surface even while the toggle keeps the rest hidden. - const revealHidden = narrowing && needle.startsWith('.') + const displayable = (entry: DirectoryEntry): boolean => showHidden || !entry.hidden || needle.startsWith('.') + const matches = (entry: DirectoryEntry): boolean => displayable(entry) && entry.name.toLowerCase().startsWith(needle) + const narrowing = needle !== '' && entries.some(matches) return entries.filter((entry) => { if (entry.path === selectedPath) return true - if (narrowing && !matches(entry)) return false - return showHidden || !entry.hidden || revealHidden + if (narrowing) return matches(entry) + return showHidden || !entry.hidden }) } @@ -353,6 +357,12 @@ export function DirectoryBrowser({ open, listDirectory, createDirectory, onOpen, const viewRef = useRef<{ parent: DirectoryListing | null; child: DirectoryListing | null }>({ parent: null, child: null }) useEffect(() => { viewRef.current = { parent, child } }, [parent, child]) + // What the last draft-following scan asked for and what came back, so a + // level still answers the text that produced it after the host respelled + // it. Stale entries are harmless: a match needs both the directory text and + // that level's own path, which together already mean the same directory. + const scanned = useRef(null) + /** * A landed preview replaced the pane a keyboard operator may have Tabbed * onto, so the focus it drops is re-parked on the still-open editor (the @@ -376,13 +386,18 @@ export function DirectoryBrowser({ open, listDirectory, createDirectory, onOpen, * rendering: a landing swaps the panes, it never blanks them. * * Two callers, one landing shape. A submitted path (Enter, a crumb) closes - * the editor on arrival and announces its failure; the editor's own - * draft-following scan keeps both to itself — it is speculative, so a - * failure leaves the last readable panes standing and says nothing, while - * an arrival clears the stale message and re-parks focus the swap dropped. + * the editor on arrival, announces its failure, and takes the wait bound — + * it is answering a gesture, so it may not hang on a stalled parent. The + * editor's own draft-following scan keeps all three to itself: it is + * speculative, nothing waits on it, and the stale view keeps rendering, so + * it waits for BOTH legs rather than flashing a single pane it would then + * upgrade — one keystroke must move the view once. A failure leaves the + * last readable panes standing and says nothing, while an arrival clears + * the stale message and re-parks focus the swap dropped. * @param path - the level to list; absent lists the Host home directory. - * @param options - `closeEditor` retires the path draft on arrival; - * `announce` surfaces a failure as the dialog's alert. + * @param options - `closeEditor` retires the path draft on arrival and + * bounds the wait for the parent leg; `announce` surfaces a failure as the + * dialog's alert. */ const land = useCallback((path: string | undefined, options: { closeEditor: boolean; announce: boolean }) => { const { seq, scan } = launchListing(path) @@ -401,6 +416,11 @@ export function DirectoryBrowser({ open, listDirectory, createDirectory, onOpen, } scan.then((target) => { if (seq !== requestSeq.current) return + // The level the panes will present as current answers this exact + // directory text, however the host respelled it (`..`, a Windows + // forward slash): the tail filters, and the same text asks for no + // second scan. + if (!options.closeEditor && path !== undefined) scanned.current = { directory: path, landed: target.path } // The single-pane landing; `landed` makes it first-commit-only, while // the two-pane commit below may still upgrade an already-landed view. let landed = false @@ -438,7 +458,10 @@ export function DirectoryBrowser({ open, listDirectory, createDirectory, onOpen, // target listed fine, and nobody asked to see the parent level. landSingle() }) - window.setTimeout(landSingle, PARENT_LEG_WAIT_MS) + // Only a submitted navigation is bounded: the walk waits both legs out + // (see the contract above), and a keystroke aborts it if the operator + // moves on first. + if (options.closeEditor) window.setTimeout(landSingle, PARENT_LEG_WAIT_MS) }, (reason: unknown) => { if (seq !== requestSeq.current) return setLoading(false) @@ -641,8 +664,12 @@ export function DirectoryBrowser({ open, listDirectory, createDirectory, onOpen, if (pathDraft === null) return const timer = window.setTimeout(() => { if (previewSuspended.current) return - const directory = pendingPreviewDirectory(viewRef.current.parent, viewRef.current.child, pathDraft) - if (directory === null) return + // The level the panes present as current: it alone may answer the + // draft, so anything else it names is a level to walk to. + const current = viewRef.current.child ?? viewRef.current.parent + if (current === null) return + const { directory, tail } = readDraft(current, pathDraft, scanned.current) + if (directory === null || tail !== null) return previewDraftLevel(directory) }, DRAFT_PREVIEW_DEBOUNCE_MS) return () => { window.clearTimeout(timer) } @@ -650,6 +677,14 @@ export function DirectoryBrowser({ open, listDirectory, createDirectory, onOpen, // After the hooks: a closed dialog renders nothing and evaluates no copy. const crumbSource = child ?? parent + // The draft's tail filters the level it names, which by the pane invariant + // is the LAST pane — never a pane the draft has already walked away from. + // Narrowing that stale pane would move the view twice for one keystroke: + // once as it narrows, again as its landing replaces it. It holds still + // instead, and the filter arrives with the level it belongs to. + const typedPrefix = crumbSource === null || pathDraft === null + ? null + : readDraft(crumbSource, pathDraft, scanned.current).tail const crumbs = crumbSource === null ? [] : displayCrumbs(crumbSource, t('browser.home')) const crumbTail = crumbs.at(-1)?.path useEffect(() => { @@ -888,7 +923,7 @@ export function DirectoryBrowser({ open, listDirectory, createDirectory, onOpen, busy={parentInert} onPick={select} showHidden={showHidden} - filterPrefix={draftPrefixFor(parent, pathDraft)} + filterPrefix={child === null ? typedPrefix : null} pathEditing={draftPending} /> )} @@ -900,7 +935,7 @@ export function DirectoryBrowser({ open, listDirectory, createDirectory, onOpen, busy={parentInert} onPick={advance} showHidden={showHidden} - filterPrefix={draftPrefixFor(child, pathDraft)} + filterPrefix={typedPrefix} pathEditing={draftPending} /> )} diff --git a/packages/host/directory-picker-browse/tests/directory-browser.spec.tsx b/packages/host/directory-picker-browse/tests/directory-browser.spec.tsx index 69c61751eb..6329887add 100644 --- a/packages/host/directory-picker-browse/tests/directory-browser.spec.tsx +++ b/packages/host/directory-picker-browse/tests/directory-browser.spec.tsx @@ -762,6 +762,105 @@ describe('DirectoryBrowser', () => { expect(b.listDirectory).toHaveBeenCalledWith(`${DOCS}/`, expect.anything()) }) + it('holds a stale pane still until its landing, instead of narrowing it first', async () => { + // Own three-level tree: the level that goes stale needs two rows for the + // narrowing this pins against to be visible at all. + const ROOT = '/u' + const MID = `${ROOT}/mid` + const LEAF = `${MID}/leaf` + const chain = [{ name: '/', path: '/', hidden: false }, { name: 'u', path: ROOT, hidden: false }] + const tree: Record = { + [ROOT]: { + path: ROOT, + home: ROOT, + crumbs: chain, + entries: [{ name: 'mid', path: MID, hidden: false }, { name: 'other', path: `${ROOT}/other`, hidden: false }], + truncated: false, + }, + [MID]: { + path: MID, + home: ROOT, + crumbs: [...chain, { name: 'mid', path: MID, hidden: false }], + entries: [{ name: 'leaf', path: LEAF, hidden: false }, { name: 'sibling', path: `${MID}/sibling`, hidden: false }], + truncated: false, + }, + [LEAF]: { + path: LEAF, + home: ROOT, + crumbs: [...chain, { name: 'mid', path: MID, hidden: false }, { name: 'leaf', path: LEAF, hidden: false }], + entries: [], + truncated: false, + }, + } + mount({ + listDirectory: vi.fn(async (path?: string) => { + const asked = path ?? ROOT + const found = tree[asked.length > 1 && asked.endsWith('/') ? asked.slice(0, -1) : asked] + if (found === undefined) throw new Error(`cannot list ${asked}`) + return found + }), + }) + await waitFor(() => { expect(screen.getByText('mid')).toBeTruthy() }) + fireEvent.click(screen.getByRole('button', { name: 'browser.editPath' })) + const input = screen.getByLabelText('browser.editPath') + fireEvent.change(input, { target: { value: `${LEAF}/` } }) + await waitFor(() => { expect(columns()).toHaveLength(2) }) + expect(within(columns()[0]!).getAllByRole('listitem').map(item => item.textContent)).toEqual(['leaf', 'sibling']) + // Deleting the separator names the level the LEFT pane lists. That pane + // is stale — its landing will move it right — so it must not narrow to + // the tail first: one deletion, one movement. + fireEvent.change(input, { target: { value: LEAF } }) + expect(within(columns()[0]!).getAllByRole('listitem').map(item => item.textContent)).toEqual(['leaf', 'sibling']) + await waitFor(() => { expect(within(columns()[0]!).getByText('other')).toBeTruthy() }) + expect(within(columns()[1]!).getAllByRole('listitem').map(item => item.textContent)).toEqual(['leaf']) + }) + + it('keeps the walked-to panes when the editor is cancelled, Open adopting where the walk ended', async () => { + const b = mount() + await waitFor(() => { expect(screen.getByRole('listitem')).toBeTruthy() }) + fireEvent.click(screen.getByRole('button', { name: 'browser.editPath' })) + const input = screen.getByLabelText('browser.editPath') + fireEvent.change(input, { target: { value: `${DOCS}/h` } }) + await waitFor(() => { expect(columns()).toHaveLength(2) }) + fireEvent.keyDown(input, { key: 'Escape' }) + // Cancel closes the editor; it does not rewind the walk. The operator + // watched the panes move, so the crumbs, the panes, and Open's target all + // stay where the walk ended. + expect(screen.queryByLabelText('browser.editPath', { selector: 'input' })).toBeNull() + expect(columns()).toHaveLength(2) + expect(within(columns()[0]!).getByText('Documents')).toBeTruthy() + expect(within(columns()[1]!).getByText('harness')).toBeTruthy() + expect(screen.getByRole('navigation').textContent).toContain('Documents') + const open = screen.getByRole('button', { name: 'browser.open' }) + expect(open.disabled).toBe(false) + fireEvent.click(open) + expect(b.onOpen).toHaveBeenCalledWith(DOCS) + }) + + it('waits both legs out for a walk: one keystroke never flashes a single pane', async () => { + let landParent = (): void => {} + const listDirectory = vi.fn(async (path?: string) => { + // The parent leg outlives the submitted-navigation wait bound; a walk + // has nothing waiting on it, so it holds the stale view instead of + // landing single-pane and upgrading. + if (path === HOME) return await new Promise((resolve) => { landParent = () => { resolve(listingFor(HOME)) } }) + return listingFor(path) + }) + mount({ listDirectory }) + await waitFor(() => { expect(screen.getByRole('listitem')).toBeTruthy() }) + fireEvent.click(screen.getByRole('button', { name: 'browser.editPath' })) + const input = screen.getByLabelText('browser.editPath') + fireEvent.change(input, { target: { value: `${DOCS}/h` } }) + await waitFor(() => { expect(listDirectory).toHaveBeenCalledWith(HOME, expect.anything()) }) + await act(async () => { await new Promise((resolve) => { setTimeout(resolve, 400) }) }) + // Well past the submitted-navigation bound: still the pre-walk view. + expect(columns()).toHaveLength(1) + expect(screen.getByText('Documents')).toBeTruthy() + await act(async () => { landParent() }) + await waitFor(() => { expect(columns()).toHaveLength(2) }) + expect(within(columns()[1]!).getByText('harness')).toBeTruthy() + }) + it('walks the panes back up when erased segments leave the listed levels', async () => { const b = mount() await waitFor(() => { expect(screen.getByRole('listitem')).toBeTruthy() }) @@ -1018,7 +1117,8 @@ describe('DirectoryBrowser', () => { ], truncated: false, } - mount({ listDirectory: vi.fn(async () => windowsListing) }) + const listDirectory = vi.fn(async () => windowsListing) + mount({ listDirectory }) await waitFor(() => { expect(screen.getAllByRole('listitem')).toHaveLength(2) }) fireEvent.click(screen.getByRole('button', { name: 'browser.editPath' })) const input = screen.getByLabelText('browser.editPath') @@ -1026,6 +1126,18 @@ describe('DirectoryBrowser', () => { expect(input.value).toBe(ROOT) fireEvent.change(input, { target: { value: `${ROOT}u` } }) expect(screen.getByRole('listitem').textContent).toBe('Users') + // Windows separates on a forward slash too (so does the Host's resolve), + // so a path typed that way names its directory; the level the Host + // answers with spells it back with a backslash, and once that scan lands + // the level answers the typed spelling — the tail filters it. + fireEvent.change(input, { target: { value: 'C:/p' } }) + await waitFor(() => { expect(screen.getByRole('listitem').textContent).toBe('Program Files') }) + // And the same spelling asks for no second scan. + const settled = listDirectory.mock.calls.length + fireEvent.change(input, { target: { value: 'C:/pr' } }) + await act(async () => { await new Promise((resolve) => { setTimeout(resolve, 400) }) }) + expect(listDirectory.mock.calls).toHaveLength(settled) + expect(screen.getByRole('listitem').textContent).toBe('Program Files') }) it('clicking away from the path editor cancels it back to the crumb view', async () => { From 94dabcb7eda4843e14095af1d8e3e13482dc3544 Mon Sep 17 00:00:00 2001 From: _Kerman Date: Mon, 3 Aug 2026 13:47:37 +0800 Subject: [PATCH 29/48] feat(ui-primitives): support delayed multiline tooltips --- .../ui-primitives/src/Tooltip.module.css | 2 +- packages/client/ui-primitives/src/Tooltip.tsx | 40 +++++++++++++++---- .../ui-primitives/tests/tooltip.spec.tsx | 28 ++++++++++++- 3 files changed, 60 insertions(+), 10 deletions(-) diff --git a/packages/client/ui-primitives/src/Tooltip.module.css b/packages/client/ui-primitives/src/Tooltip.module.css index 5853531bd4..4da0eebc2d 100644 --- a/packages/client/ui-primitives/src/Tooltip.module.css +++ b/packages/client/ui-primitives/src/Tooltip.module.css @@ -14,7 +14,7 @@ color: var(--dsw-static-neutral-bluish-00); font-size: 14px; line-height: 22px; - white-space: nowrap; + white-space: pre-line; pointer-events: none; animation: tooltip-in 150ms var(--ds-ease-in-out); } diff --git a/packages/client/ui-primitives/src/Tooltip.tsx b/packages/client/ui-primitives/src/Tooltip.tsx index 30e0c5fe0e..e85583a50c 100644 --- a/packages/client/ui-primitives/src/Tooltip.tsx +++ b/packages/client/ui-primitives/src/Tooltip.tsx @@ -1,6 +1,6 @@ // Hover/focus label bubble (figma tooltip pill: dark plate, white text). -// TODO: interaction is a placeholder (no show delay, no flip on viewport -// collision, no arrow) — visuals and behavior get a proper pass later. +// TODO: interaction is a placeholder (no flip on viewport collision or +// arrow) — visuals and behavior get a proper pass later. // The anchor is the child element itself (cloneElement, no wrapper node), so // attaching a tooltip never changes the anchor's layout context. The bubble is // position:fixed and coordinates come from the anchor's rect at show time, so @@ -27,12 +27,13 @@ interface AnchorProps { * Attach a hover/focus tooltip to an anchor element. * @param props.label - bubble text. * @param props.side - placement relative to the anchor (default 'right'). + * @param props.delayMs - hover delay in milliseconds; keyboard focus remains immediate. * @param props.disabled - suppress the bubble while true; the anchor renders identically so * toggling never remounts it (which would cut its CSS transitions). * @param props.children - a single anchor element; its own ref (callback or object) is forwarded alongside the tooltip's. * @returns the cloned anchor plus a fixed-position bubble while hovered/focused. */ -export function Tooltip({ label, side = 'right', disabled = false, children }: { label: string; side?: TooltipSide; disabled?: boolean; children: ReactElement }) { +export function Tooltip({ label, side = 'right', delayMs = 0, disabled = false, children }: { label: string; side?: TooltipSide; delayMs?: number; disabled?: boolean; children: ReactElement }) { const anchor = useRef(null) // React 18 keeps the element's ref outside props; forward it so wrapping an // anchor in Tooltip never silently severs the owner's ref. @@ -43,15 +44,26 @@ export function Tooltip({ label, side = 'right', disabled = false, children }: { else if (childRef != null) (childRef as MutableRefObject).current = el }, [childRef]) const [pos, setPos] = useState<{ x: number; y: number } | null>(null) + const showTimer = useRef | null>(null) // Hover and focus are independent triggers: the bubble hides only after // BOTH clear (hovering away from a focused anchor must not drop it). const triggers = useRef({ hover: false, focus: false }) // Disabling mid-hover (e.g. clicking a rail control expands the sidebar) // must drop an already-visible bubble: no mouseleave fires. + const cancelShow = useCallback(() => { + if (showTimer.current === null) return + clearTimeout(showTimer.current) + showTimer.current = null + }, []) useEffect(() => { - if (disabled) { triggers.current = { hover: false, focus: false }; setPos(null) } - }, [disabled]) + if (disabled) { + cancelShow() + triggers.current = { hover: false, focus: false } + setPos(null) + } + return cancelShow + }, [cancelShow, disabled]) const show = () => { if (disabled) return @@ -63,7 +75,19 @@ export function Tooltip({ label, side = 'right', disabled = false, children }: { ? { x: r.right + 10, y: r.top + r.height / 2 } : { x: r.left + r.width / 2, y: r.bottom + 8 }) } + const showAfterHoverDelay = () => { + cancelShow() + if (delayMs <= 0) { + show() + return + } + showTimer.current = setTimeout(() => { + showTimer.current = null + show() + }, delayMs) + } const hide = () => { + cancelShow() if (!triggers.current.hover && !triggers.current.focus) setPos(null) } @@ -71,9 +95,9 @@ export function Tooltip({ label, side = 'right', disabled = false, children }: { <> {cloneElement(children, { ref: mergedRef, - onMouseEnter: (e) => { children.props.onMouseEnter?.(e); triggers.current.hover = true; show() }, - onMouseLeave: (e) => { children.props.onMouseLeave?.(e); triggers.current.hover = false; setPos(null) }, - onFocus: (e) => { children.props.onFocus?.(e); triggers.current.focus = true; show() }, + onMouseEnter: (e) => { children.props.onMouseEnter?.(e); triggers.current.hover = true; showAfterHoverDelay() }, + onMouseLeave: (e) => { children.props.onMouseLeave?.(e); triggers.current.hover = false; cancelShow(); setPos(null) }, + onFocus: (e) => { children.props.onFocus?.(e); triggers.current.focus = true; cancelShow(); show() }, onBlur: (e) => { children.props.onBlur?.(e); triggers.current.focus = false; hide() }, })} {pos !== null && ( diff --git a/packages/client/ui-primitives/tests/tooltip.spec.tsx b/packages/client/ui-primitives/tests/tooltip.spec.tsx index 591a5eb67a..5dc9a1a378 100644 --- a/packages/client/ui-primitives/tests/tooltip.spec.tsx +++ b/packages/client/ui-primitives/tests/tooltip.spec.tsx @@ -1,11 +1,37 @@ // @vitest-environment jsdom -import { cleanup, fireEvent, render, screen } from '@testing-library/react' +import { act, cleanup, fireEvent, render, screen } from '@testing-library/react' import { afterEach, describe, expect, it, vi } from 'vitest' import { Tooltip } from '@deepseek-ai/dsh-client-ui-primitives' afterEach(cleanup) describe('Tooltip', () => { + it('can delay pointer hover without delaying keyboard focus', () => { + vi.useFakeTimers() + try { + render( + + + , + ) + const anchor = screen.getByText('anchor') + fireEvent.mouseEnter(anchor) + act(() => { vi.advanceTimersByTime(499) }) + expect(screen.queryByRole('tooltip')).toBeNull() + fireEvent.mouseLeave(anchor) + act(() => { vi.advanceTimersByTime(1) }) + expect(screen.queryByRole('tooltip')).toBeNull() + fireEvent.mouseEnter(anchor) + act(() => { vi.advanceTimersByTime(500) }) + expect(screen.getByRole('tooltip').textContent).toBe('Timing details') + fireEvent.mouseLeave(anchor) + fireEvent.focus(anchor) + expect(screen.getByRole('tooltip').textContent).toBe('Timing details') + } finally { + vi.useRealTimers() + } + }) + it('shows the bubble to the right on hover and hides it on leave', () => { render( From a8478fc03126fbec3c6cf580817e7fedfb497e66 Mon Sep 17 00:00:00 2001 From: _Kerman Date: Mon, 3 Aug 2026 13:47:51 +0800 Subject: [PATCH 30/48] fix(ui-trajectory): animate responsive ledger layout --- .../src/client/TrajectoryTable.module.css | 83 +++++++++++++++++-- 1 file changed, 74 insertions(+), 9 deletions(-) diff --git a/packages/client/ui-trajectory/src/client/TrajectoryTable.module.css b/packages/client/ui-trajectory/src/client/TrajectoryTable.module.css index 86cc8dfd8b..9e606ca5c6 100644 --- a/packages/client/ui-trajectory/src/client/TrajectoryTable.module.css +++ b/packages/client/ui-trajectory/src/client/TrajectoryTable.module.css @@ -277,7 +277,7 @@ z-index: 3; top: 0; left: 0; - display: inline-flex; + display: inline-grid; flex: none; align-items: center; box-sizing: border-box; @@ -292,8 +292,18 @@ white-space: nowrap; } +.turnLabelFull, .turnLabelCompact { - display: none; + grid-area: 1 / 1; + max-width: 64px; + overflow: hidden; + opacity: 1; + white-space: nowrap; +} + +.turnLabelCompact { + max-width: 0; + opacity: 0; } .turnLabelActive { @@ -350,15 +360,23 @@ } .kindTagIcon { - display: none; + display: inline-flex; + flex: none; align-items: center; justify-content: center; - width: 13px; + width: 0; height: 13px; + overflow: hidden; + opacity: 0; + transform: scale(0.8); } .kindTagLabel { - display: inline; + display: inline-block; + max-width: 72px; + overflow: hidden; + opacity: 1; + white-space: nowrap; } .table .kindSlot .message { @@ -393,19 +411,66 @@ } .kindTagIcon { - display: inline-flex; + width: 13px; + opacity: 1; + transform: scale(1); } .kindTagLabel { - display: none; + max-width: 0; + opacity: 0; } .turnLabelFull { - display: none; + max-width: 0; + opacity: 0; } .turnLabelCompact { - display: inline; + max-width: 64px; + opacity: 1; + } +} + +@media (prefers-reduced-motion: no-preference) { + .eventColumn, + .event, + .requestBoundaryControl, + .kindSlot, + .kindTag, + .kindTagIcon, + .kindTagLabel, + .turnLabelFull, + .turnLabelCompact { + transition-duration: 180ms; + transition-timing-function: var(--ds-ease-in-out); + } + + .eventColumn, + .kindSlot { + transition-property: width; + } + + .event { + transition-property: padding-right, padding-left; + } + + .requestBoundaryControl { + transition-property: left; + } + + .kindTag { + transition-property: padding-right, padding-left; + } + + .kindTagIcon { + transition-property: width, opacity, transform; + } + + .kindTagLabel, + .turnLabelFull, + .turnLabelCompact { + transition-property: max-width, opacity; } } From c8ece8325c34a7376f753ef74d46ee18867c73b9 Mon Sep 17 00:00:00 2001 From: _Kerman Date: Mon, 3 Aug 2026 13:47:55 +0800 Subject: [PATCH 31/48] fix(ui-trajectory): follow live ledger tail --- .../src/client/TrajectoryTable.tsx | 26 +++++++++- .../client/ui-trajectory/tests/table.spec.tsx | 47 +++++++++++++++++++ 2 files changed, 72 insertions(+), 1 deletion(-) diff --git a/packages/client/ui-trajectory/src/client/TrajectoryTable.tsx b/packages/client/ui-trajectory/src/client/TrajectoryTable.tsx index 4f88bf0bca..7973649cf5 100644 --- a/packages/client/ui-trajectory/src/client/TrajectoryTable.tsx +++ b/packages/client/ui-trajectory/src/client/TrajectoryTable.tsx @@ -1,6 +1,6 @@ /** Turn-aware trajectory event ledger with a local record inspector. */ -import { useCallback, useEffect, useMemo, useRef, useState } from 'react' +import { useCallback, useEffect, useLayoutEffect, useMemo, useRef, useState } from 'react' import type { CSSProperties, ReactNode } from 'react' import { IconChevronRightOutline14, @@ -22,6 +22,8 @@ import { formatElapsedSeconds } from './trajectory-record.ts' import { trajectoryPreviewText, type TrajectoryTurnModel } from './layout.ts' import css from './TrajectoryTable.module.css' +const BOTTOM_FOLLOW_THRESHOLD_PX = 2 + const KIND_LABEL: Record = { system: 'SYSTEM', user: 'USER', @@ -1711,6 +1713,9 @@ export function TrajectoryTable({ // ledger has rendered. Not-found leaves the request pending (`turns` in the // deps retries as history pages in); the ack clears the store field. const rootRef = useRef(null) + const tablePaneRef = useRef(null) + const followsTableTail = useRef(false) + const tableScrollInitialized = useRef(false) const pendingScrollIndex = useRef(null) const openRecordSummaryRef = useRef(openRecordSummary) openRecordSummaryRef.current = openRecordSummary @@ -1734,11 +1739,30 @@ export function TrajectoryTable({ row.scrollIntoView({ behavior: 'smooth', block: 'center' }) } }) + useLayoutEffect(() => { + const pane = tablePaneRef.current + if (pane === null) return + if (!tableScrollInitialized.current) { + tableScrollInitialized.current = true + followsTableTail.current = + pane.scrollHeight - pane.clientHeight - pane.scrollTop + <= BOTTOM_FOLLOW_THRESHOLD_PX + return + } + if (followsTableTail.current) pane.scrollTop = pane.scrollHeight + }, [turns]) return (
{ + const pane = event.currentTarget + followsTableTail.current = + pane.scrollHeight - pane.clientHeight - pane.scrollTop + <= BOTTOM_FOLLOW_THRESHOLD_PX + }} onClick={(event) => { if (event.target === event.currentTarget) clearAllSelections() }} diff --git a/packages/client/ui-trajectory/tests/table.spec.tsx b/packages/client/ui-trajectory/tests/table.spec.tsx index f02eb6e0cc..65c3da3255 100644 --- a/packages/client/ui-trajectory/tests/table.spec.tsx +++ b/packages/client/ui-trajectory/tests/table.spec.tsx @@ -160,6 +160,53 @@ describe('TrajectoryTable', () => { expect(onClearSelection).toHaveBeenCalledOnce() }) + it('follows appended records only while the ledger is already at the bottom', () => { + const view = render() + const tablePane = screen.getByRole('table').parentElement as HTMLElement + let scrollHeight = 200 + Object.defineProperties(tablePane, { + clientHeight: { configurable: true, get: () => 100 }, + scrollHeight: { configurable: true, get: () => scrollHeight }, + }) + tablePane.scrollTop = 100 + fireEvent.scroll(tablePane) + + scrollHeight = 260 + view.rerender( + , + ) + expect(tablePane.scrollTop).toBe(260) + + tablePane.scrollTop = 20 + fireEvent.scroll(tablePane) + scrollHeight = 320 + view.rerender( + , + ) + expect(tablePane.scrollTop).toBe(20) + }) + it('keeps running and failure semantics distinct from record roles', () => { const view = render() expect(view.container.querySelector('tr[data-kind="tool"][data-running="true"]')).toBeTruthy() From 74cd2e4bd9538aa5e0bb10a5cfedd8d0466edd7b Mon Sep 17 00:00:00 2001 From: _Kerman Date: Mon, 3 Aug 2026 13:48:03 +0800 Subject: [PATCH 32/48] feat(ui-trajectory): enrich timeline timing interactions --- apps/web/tests/navigation-panes.e2e.ts | 11 + .../navigation-panes/trajectory.expected.md | 3 +- .../src/client/TrajectoryTimeline.module.css | 55 +++-- .../src/client/TrajectoryTimeline.tsx | 226 +++++++++++++++--- .../client/ui-trajectory/tests/views.spec.tsx | 148 +++++++++++- 5 files changed, 388 insertions(+), 55 deletions(-) diff --git a/apps/web/tests/navigation-panes.e2e.ts b/apps/web/tests/navigation-panes.e2e.ts index 35f77b96ec..af4799a759 100644 --- a/apps/web/tests/navigation-panes.e2e.ts +++ b/apps/web/tests/navigation-panes.e2e.ts @@ -162,6 +162,17 @@ describe('web e2e: navigation & panes over a rich seeded session', () => { await page.evaluate(() => { document.body.removeAttribute('data-ds-dark-theme') }) await page.getByRole('tab', { name: 'Result' }).click() await expect.poll(() => page.getByText('NAVIGATION_OK', { exact: false }).count(), { timeout: 10_000 }).toBeGreaterThanOrEqual(1) + const assistantSpan = page.locator('[data-timeline-span="message"][data-assistant-timing="true"]').first() + await assistantSpan.hover() + const timingTooltip = page.getByRole('tooltip') + await timingTooltip.waitFor({ timeout: 5_000 }) + await expect.poll(() => timingTooltip.textContent(), { timeout: 5_000 }).toMatch(/TTFT .* Decoding/) + const assistantTimingStyle = await assistantSpan.evaluate(node => ({ + background: getComputedStyle(node).backgroundImage, + ttft: getComputedStyle(node).getPropertyValue('--trajectory-assistant-ttft'), + })) + expect(assistantTimingStyle.background).toContain('linear-gradient') + expect(assistantTimingStyle.ttft).toMatch(/%$/) const snapshot = (await captureStableAria(page, '[class*="viewArea"]', scaffold.workspaceCwd)) .split(SEED_ID).join('{{seededId}}') await compareOrRefreshGolden(TRAJECTORY_EXPECTED, snapshot, MODE) diff --git a/apps/web/tests/snapshots/navigation-panes/trajectory.expected.md b/apps/web/tests/snapshots/navigation-panes/trajectory.expected.md index 788b8a3233..704190b9f2 100644 --- a/apps/web/tests/snapshots/navigation-panes/trajectory.expected.md +++ b/apps/web/tests/snapshots/navigation-panes/trajectory.expected.md @@ -4,7 +4,8 @@ - button "Collapse calls": Calls - img - searchbox "Search trajectory" -- region "Trajectory timeline" +- region "Trajectory timeline": + - tooltip "ASSISTANT {{clock}}:40.549 AM → {{clock}}:42.091 AM Total 1.5 s · TTFT 368 ms · Decoding 1.2 s" - table: - rowgroup: - row "SYSTEM, Initial System Prompt": diff --git a/packages/client/ui-trajectory/src/client/TrajectoryTimeline.module.css b/packages/client/ui-trajectory/src/client/TrajectoryTimeline.module.css index 1ca5ab2627..4d548e64d8 100644 --- a/packages/client/ui-trajectory/src/client/TrajectoryTimeline.module.css +++ b/packages/client/ui-trajectory/src/client/TrajectoryTimeline.module.css @@ -7,6 +7,10 @@ user-select: none; } +.root :global([role='tooltip']) { + font: var(--dsw-font-xxxs-11); +} + .plot { display: grid; grid-template-columns: 44px minmax(0, 1fr); @@ -53,6 +57,10 @@ touch-action: none; } +.track[data-panning='true'] { + cursor: grabbing; +} + .empty { position: absolute; top: 50%; @@ -105,8 +113,15 @@ .span { position: absolute; top: calc(var(--trajectory-span-lane) * 14px); - left: calc(var(--trajectory-span-left) + 1px); - width: max(2px, calc(var(--trajectory-span-width) - 2px)); + left: calc(var(--trajectory-span-left) + var(--trajectory-span-gap)); + width: max( + 2px, + calc( + var(--trajectory-span-width) + - var(--trajectory-span-gap) + - var(--trajectory-span-gap) + ) + ); height: 8px; min-width: 2px; border-radius: 1px; @@ -127,23 +142,35 @@ } .span[data-timeline-span='message'] { - background: color-mix( + --trajectory-assistant-decoding-color: color-mix( in srgb, var(--dsw-alias-brand-primary-new-colorprimary-new-color) 60%, var(--dsw-alias-state-error-secondary) ); -} - -.span[data-timeline-span='tool'] { - background: var(--dsw-alias-state-warn-label); -} - -.span[data-timeline-span='subtool'] { - background: color-mix( + --trajectory-assistant-ttft-color: color-mix( in srgb, - var(--dsw-alias-state-warn-label) 62%, - var(--dsw-alias-label-tertiary) + var(--trajectory-assistant-decoding-color) 54%, + var(--dsw-alias-bg-layer-2) ); + + background: var(--trajectory-assistant-decoding-color); + opacity: 1; +} + +.span[data-timeline-span='message'][data-assistant-timing='true'] { + background: linear-gradient( + to right, + var(--trajectory-assistant-ttft-color) 0, + var(--trajectory-assistant-ttft-color) var(--trajectory-assistant-ttft), + var(--trajectory-assistant-decoding-color) var(--trajectory-assistant-ttft), + var(--trajectory-assistant-decoding-color) 100% + ); +} + +.span[data-timeline-span='tool'], +.span[data-timeline-span='subtool'] { + background: var(--dsw-alias-state-warn-label); + opacity: 1; } .span[data-error='true'] { @@ -161,7 +188,7 @@ .span[data-hovered='true']:not([data-current='true']) { z-index: 1; - opacity: 0.78; + opacity: 1; box-shadow: 0 0 0 1px var(--dsw-alias-bg-layer-2), 0 0 0 2px color-mix( diff --git a/packages/client/ui-trajectory/src/client/TrajectoryTimeline.tsx b/packages/client/ui-trajectory/src/client/TrajectoryTimeline.tsx index 0fd0825f95..87d7cdcd34 100644 --- a/packages/client/ui-trajectory/src/client/TrajectoryTimeline.tsx +++ b/packages/client/ui-trajectory/src/client/TrajectoryTimeline.tsx @@ -4,7 +4,9 @@ import { memo, useEffect, useMemo, useRef, useState, type CSSProperties, type KeyboardEvent, type PointerEvent, } from 'react' +import { Tooltip } from '@deepseek-ai/dsh-client-ui-primitives' import type { TrajectoryTurnModel } from './layout.ts' +import type { AssistantMetricDetail, TrajectoryCellKind, TrajectoryCellProps } from './trajectory-record.ts' import { deriveTrajectoryTimeline, formatTimelineOffset, @@ -18,6 +20,14 @@ const MINIMUM_ZOOM_OPERATIONS = 4 const EDGE_PAN_ZONE_FRACTION = 0.08 const EDGE_PAN_STEP_FRACTION = 0.025 const MAXIMUM_EDGE_PAN_PX = 32 +const TIMELINE_TOOLTIP_DELAY_MS = 500 + +interface TimelineRecordDetail { + decodingMs?: number + durationMs?: number + startedAt?: number + ttftMs?: number +} interface FractionRange { start: number @@ -29,6 +39,94 @@ interface HoverPoint { recordIndex: number | null } +interface PanGesture { + anchorClientX: number + anchorStart: number + moved: boolean + pannable: boolean + pointerId: number +} + +function assistantTimingDetail( + metrics: AssistantMetricDetail | undefined, +): Pick { + const start = metrics?.stepStartTime + const first = metrics?.firstTokenTime + const completed = metrics?.completedTime + if ( + metrics?.timingRecorded !== true + || typeof start !== 'number' + || typeof first !== 'number' + || typeof completed !== 'number' + || !Number.isFinite(start) + || !Number.isFinite(first) + || !Number.isFinite(completed) + || first < start + || completed < first + ) return {} + return { ttftMs: first - start, decodingMs: completed - first } +} + +function timelineRecordDetail(cell: TrajectoryCellProps): TimelineRecordDetail { + const durationMs = cell.timeSeconds === null || !Number.isFinite(cell.timeSeconds) + ? undefined + : Math.max(0, cell.timeSeconds * 1_000) + const startedAt = cell.startedAt === null || !Number.isFinite(cell.startedAt) + ? undefined + : cell.startedAt + return { + ...(durationMs === undefined ? {} : { durationMs }), + ...(startedAt === undefined ? {} : { startedAt }), + ...assistantTimingDetail(cell.assistantMetrics), + } +} + +function timelineKindLabel(kind: TrajectoryCellKind): string { + switch (kind) { + case 'system': return 'SYSTEM' + case 'user': return 'USER' + case 'context': return 'CONTEXT' + case 'compacted': return 'COMPACTED' + case 'message': return 'ASSISTANT' + case 'tool': return 'TOOL' + case 'subtool': return 'SUBTOOL' + } +} + +function formatRecordedTime(timestamp: number): string { + return new Date(timestamp).toLocaleTimeString(undefined, { + hour: '2-digit', + minute: '2-digit', + second: '2-digit', + fractionalSecondDigits: 3, + }) +} + +function timelineTooltipLabel( + kind: TrajectoryCellKind, + detail: TimelineRecordDetail | undefined, +): string { + const heading = timelineKindLabel(kind) + if (detail === undefined) return heading + const duration = detail.durationMs === undefined + ? null + : `Total ${formatTimelineOffset(detail.durationMs)}` + const range = detail.startedAt === undefined + ? null + : detail.durationMs === undefined + ? `Started ${formatRecordedTime(detail.startedAt)}` + : `${formatRecordedTime(detail.startedAt)} → ${formatRecordedTime( + detail.startedAt + detail.durationMs, + )}` + const segments = detail.ttftMs === undefined || detail.decodingMs === undefined + ? null + : `TTFT ${formatTimelineOffset(detail.ttftMs)} · Decoding ${formatTimelineOffset( + detail.decodingMs, + )}` + const timing = [duration, segments].filter(value => value !== null).join(' · ') + return [heading, range, timing].filter(value => value !== null && value !== '').join('\n') +} + /** Props for the fixed full-domain overview above the trajectory ledger. */ export interface TrajectoryTimelineProps { turns: readonly TrajectoryTurnModel[] @@ -105,14 +203,10 @@ export const TrajectoryTimeline = memo(function TrajectoryTimeline({ onRecordFocus, }: TrajectoryTimelineProps) { const model = useMemo(() => deriveTrajectoryTimeline(turns, mode), [mode, turns]) - const durationByIndex = useMemo( + const detailByIndex = useMemo( () => new Map(turns.flatMap(turn => turn.groups.flatMap(group => - group.cells.flatMap(cell => - cell.timeSeconds === null || !Number.isFinite(cell.timeSeconds) - ? [] - : [[cell.index, Math.max(0, cell.timeSeconds * 1_000)] as const], - ), + group.cells.map(cell => [cell.index, timelineRecordDetail(cell)] as const), ), )), [turns], @@ -123,10 +217,12 @@ export const TrajectoryTimeline = memo(function TrajectoryTimeline({ anchorClientX: number recordIndex: number | null } | null>(null) + const panRef = useRef(null) const rootRef = useRef(null) const trackRef = useRef(null) const [draft, setDraft] = useState(null) const [hover, setHover] = useState(null) + const [panning, setPanning] = useState(false) const [viewport, setViewport] = useState(null) const [animateViewport, setAnimateViewport] = useState(false) useEffect(() => { @@ -267,6 +363,21 @@ export const TrajectoryTimeline = memo(function TrajectoryTimeline({ } const onPointerDown = (event: PointerEvent) => { + if (event.button === 2) { + panRef.current = { + anchorClientX: event.clientX, + anchorStart: domainStart, + moved: false, + pannable: viewport !== null, + pointerId: event.pointerId, + } + if (viewport !== null) setAnimateViewport(false) + setPanning(true) + if (typeof event.currentTarget.setPointerCapture === 'function') { + event.currentTarget.setPointerCapture(event.pointerId) + } + return + } if (event.button !== 0) return const anchor = fractionAt(event) const anchorTime = domainStart + anchor * domainDuration @@ -285,10 +396,24 @@ export const TrajectoryTimeline = memo(function TrajectoryTimeline({ } const onPointerMove = (event: PointerEvent) => { - const drag = dragRef.current const rect = event.currentTarget.getBoundingClientRect() const fraction = fractionAt(event) setHover({ fraction, recordIndex: recordIndexAt(event) }) + const pan = panRef.current + if (pan !== null && pan.pointerId === event.pointerId) { + if (Math.abs(event.clientX - pan.anchorClientX) >= MINIMUM_DRAG_PX) { + pan.moved = true + } + if (!pan.pannable) return + const delta = (event.clientX - pan.anchorClientX) / Math.max(1, rect.width) + const nextStart = Math.min( + Math.max(pan.anchorStart - delta * domainDuration, model.start), + model.end - domainDuration, + ) + setViewport({ start: nextStart, end: nextStart + domainDuration }) + return + } + const drag = dragRef.current if (drag === null || drag.pointerId !== event.pointerId) return let nextDomainStart = domainStart if (viewport !== null) { @@ -326,6 +451,15 @@ export const TrajectoryTimeline = memo(function TrajectoryTimeline({ } const onPointerEnd = (event: PointerEvent) => { + const pan = panRef.current + if (pan !== null && pan.pointerId === event.pointerId) { + const moved = pan.moved + || Math.abs(event.clientX - pan.anchorClientX) >= MINIMUM_DRAG_PX + panRef.current = null + setPanning(false) + if (!moved) onRangeChange(null) + return + } const drag = dragRef.current if (drag === null || drag.pointerId !== event.pointerId) return const pointFraction = fractionAt(event) @@ -375,8 +509,10 @@ export const TrajectoryTimeline = memo(function TrajectoryTimeline({ const onPointerCancel = () => { dragRef.current = null + panRef.current = null setDraft(null) setHover(null) + setPanning(false) } return ( @@ -386,6 +522,7 @@ export const TrajectoryTimeline = memo(function TrajectoryTimeline({
{ - if (dragRef.current === null) setHover(null) + if (dragRef.current === null && panRef.current === null) setHover(null) }} onDoubleClick={(event) => { event.preventDefault() @@ -402,9 +539,6 @@ export const TrajectoryTimeline = memo(function TrajectoryTimeline({ }} onContextMenu={(event) => { event.preventDefault() - setAnimateViewport(false) - onRangeChange(null) - setViewport(null) }} > {hover !== null && hover.recordIndex === null && draft === null && ( @@ -466,7 +600,6 @@ export const TrajectoryTimeline = memo(function TrajectoryTimeline({ className={css.lanes} data-animate-viewport={animateViewport || undefined} data-timeline-domain - aria-hidden="true" style={projectedDomainStyle} > {model.spans @@ -476,34 +609,51 @@ export const TrajectoryTimeline = memo(function TrajectoryTimeline({ .map((span) => { const left = (span.start - model.start) / fullDuration const width = (span.end - span.start) / fullDuration - const durationMs = durationByIndex.get(span.index) + const widthPercent = Math.max(width * 100, 0.35) + const detail = detailByIndex.get(span.index) + const ttftMs = detail?.ttftMs + const decodingMs = detail?.decodingMs + const ttftFraction = ttftMs === undefined + || decodingMs === undefined + || ttftMs + decodingMs <= 0 + ? null + : ttftMs / (ttftMs + decodingMs) return ( - = activeRange.start - ? 'true' - : 'false'} + + label={timelineTooltipLabel(span.kind, detail)} + side="bottom" + delayMs={TIMELINE_TOOLTIP_DELAY_MS} + > + ) })}
diff --git a/packages/client/ui-trajectory/tests/views.spec.tsx b/packages/client/ui-trajectory/tests/views.spec.tsx index fca6608950..6b7be01ec3 100644 --- a/packages/client/ui-trajectory/tests/views.spec.tsx +++ b/packages/client/ui-trajectory/tests/views.spec.tsx @@ -9,7 +9,7 @@ */ import { Context } from 'cordis' import { afterEach, beforeEach, describe, expect, it, vi } from 'vitest' -import { cleanup, fireEvent, render, screen } from '@testing-library/react' +import { act, cleanup, fireEvent, render, screen } from '@testing-library/react' import { createElement, type ComponentProps, type FC, type ReactNode } from 'react' import { bindSnapshotSelector } from '@deepseek-ai/dsh-client-web-react' import { resolveSlotLabel } from '@deepseek-ai/dsh-client-ui-slots' @@ -445,7 +445,7 @@ describe('tab switching in ConversationRoot', () => { .toBe('outside') fireEvent.contextMenu(plot) expect(screen.getByRole('row', { name: /USER/ }).getAttribute('data-timeline-focus')) - .toBeNull() + .toBe('outside') }) it('clicking a timeline block clears the range, selects the record, and opens its inspector', async () => { @@ -528,6 +528,57 @@ describe('timeline projection', () => { }], }] satisfies readonly TrajectoryTurnModel[] + it('splits assistant time into recorded TTFT and decoding proportions with a delayed tooltip', () => { + vi.useFakeTimers() + try { + const view = render( + , + ) + const span = view.container.querySelector( + '[data-timeline-span="message"]', + ) + expect(span?.getAttribute('title')).toBeNull() + expect(span?.getAttribute('data-assistant-timing')).toBe('true') + expect(span?.style.getPropertyValue('--trajectory-assistant-ttft')).toBe('25%') + + fireEvent.mouseEnter(span as HTMLElement) + act(() => { vi.advanceTimersByTime(499) }) + expect(view.container.querySelector('[role="tooltip"]')).toBeNull() + act(() => { vi.advanceTimersByTime(1) }) + const tooltip = view.container.querySelector('[role="tooltip"]') + expect(tooltip?.textContent).toContain('Total 2.0 s') + expect(tooltip?.textContent).toContain('TTFT 500 ms') + expect(tooltip?.textContent).toContain('Decoding 1.5 s') + } finally { + vi.useRealTimers() + } + }) + it('cancels native scrolling across the timeline while zooming', () => { render( { })).toBe(false) }) + it('scales sequence gutters with narrow operation spans', () => { + const view = render( + , + ) + const span = view.container.querySelector('[data-timeline-span]') + expect(span?.style.getPropertyValue('--trajectory-span-width')).toBe('10%') + expect(span?.style.getPropertyValue('--trajectory-span-gap')) + .toBe('clamp(0.25px, 0.8%, 1px)') + }) + + it('clears the selection without changing zoom on a zoomed right click', () => { + const onRangeChange = vi.fn() + const view = render( + , + ) + const plot = screen.getByLabelText('Timeline overview; drag horizontally to focus events') + vi.spyOn(plot, 'getBoundingClientRect').mockReturnValue({ + x: 0, y: 0, left: 0, top: 0, right: 100, bottom: 72, width: 100, height: 72, + toJSON: () => ({}), + }) + fireEvent.wheel(plot, { clientX: 50, deltaY: -1_000 }) + const domain = view.container.querySelector('[data-timeline-domain]') + const domainWidth = domain?.style.getPropertyValue('--trajectory-domain-width') + expect(domainWidth).not.toBe('100%') + + fireEvent.pointerDown(plot, { button: 2, clientX: 50, pointerId: 1 }) + expect(fireEvent.contextMenu(plot)).toBe(false) + fireEvent.pointerUp(plot, { button: 2, clientX: 50, pointerId: 1 }) + + expect(onRangeChange).toHaveBeenCalledOnce() + expect(onRangeChange).toHaveBeenCalledWith(null) + expect(domain?.style.getPropertyValue('--trajectory-domain-width')).toBe(domainWidth) + }) + + it('clears the selection and suppresses the context menu at full zoom', () => { + const onRangeChange = vi.fn() + render( + , + ) + const plot = screen.getByLabelText('Timeline overview; drag horizontally to focus events') + + fireEvent.pointerDown(plot, { button: 2, clientX: 50, pointerId: 1 }) + expect(fireEvent.contextMenu(plot)).toBe(false) + fireEvent.pointerUp(plot, { button: 2, clientX: 50, pointerId: 1 }) + expect(onRangeChange).toHaveBeenCalledOnce() + expect(onRangeChange).toHaveBeenCalledWith(null) + }) + + it('pans the zoomed viewport with a right-button drag without changing the selection', () => { + const onRangeChange = vi.fn() + const view = render( + , + ) + const plot = screen.getByLabelText('Timeline overview; drag horizontally to focus events') + vi.spyOn(plot, 'getBoundingClientRect').mockReturnValue({ + x: 0, y: 0, left: 0, top: 0, right: 100, bottom: 72, width: 100, height: 72, + toJSON: () => ({}), + }) + fireEvent.wheel(plot, { clientX: 50, deltaY: -1_000 }) + const domain = view.container.querySelector('[data-timeline-domain]') + const before = domain?.style.getPropertyValue('--trajectory-domain-left') + + fireEvent.pointerDown(plot, { button: 2, clientX: 50, pointerId: 1 }) + expect(plot.getAttribute('data-panning')).toBe('true') + expect(fireEvent.contextMenu(plot)).toBe(false) + fireEvent.pointerMove(plot, { buttons: 2, clientX: 75, pointerId: 1 }) + fireEvent.pointerUp(plot, { button: 2, clientX: 75, pointerId: 1 }) + + expect(domain?.style.getPropertyValue('--trajectory-domain-left')).not.toBe(before) + expect(onRangeChange).not.toHaveBeenCalled() + expect(plot.getAttribute('data-panning')).toBeNull() + }) + it('pans the zoomed viewport only far enough to reveal a newly selected record', async () => { const onRangeChange = vi.fn() const view = render( From 68e79ed9833978d170ddb45a1b10554269c4b31a Mon Sep 17 00:00:00 2001 From: _Kerman Date: Mon, 3 Aug 2026 13:48:11 +0800 Subject: [PATCH 33/48] docs(ui-trajectory): record timeline interaction contract --- .../2026-07-27-trajectory-inspection-ledger.i18n.yaml | 4 ++-- .../feature/2026-07-27-trajectory-inspection-ledger.md | 5 +++-- .../feature/2026-07-27-trajectory-inspection-ledger.zh.md | 5 +++-- packages/client/ui-trajectory/README.i18n.yaml | 4 ++-- packages/client/ui-trajectory/README.md | 2 +- packages/client/ui-trajectory/README.zh.md | 2 +- 6 files changed, 12 insertions(+), 10 deletions(-) diff --git a/.agents/notes/implemented/feature/2026-07-27-trajectory-inspection-ledger.i18n.yaml b/.agents/notes/implemented/feature/2026-07-27-trajectory-inspection-ledger.i18n.yaml index 6e9974fdce..17b8219739 100644 --- a/.agents/notes/implemented/feature/2026-07-27-trajectory-inspection-ledger.i18n.yaml +++ b/.agents/notes/implemented/feature/2026-07-27-trajectory-inspection-ledger.i18n.yaml @@ -2,5 +2,5 @@ # side as of the last confirmed-consistent state. Both languages carry equal authority; # after editing either side, bring the other along and re-record with: # pnpm run verify-translation-pairing --write .agents/notes/implemented/feature/2026-07-27-trajectory-inspection-ledger.md -2026-07-27-trajectory-inspection-ledger.md: 8c2a7c42b7898776de5d42459b09c0fb1737ec0b -2026-07-27-trajectory-inspection-ledger.zh.md: 6c5733046dc2cfd3fd2bd005f4cf5c2d2bd110af +2026-07-27-trajectory-inspection-ledger.md: fcdbbb30b065b0b128a8e375cd2e8ed5b2702dac +2026-07-27-trajectory-inspection-ledger.zh.md: ea31389d2fb0970caae271e7d64e4ecff215e93c diff --git a/.agents/notes/implemented/feature/2026-07-27-trajectory-inspection-ledger.md b/.agents/notes/implemented/feature/2026-07-27-trajectory-inspection-ledger.md index 8c2a7c42b7..fcdbbb30b0 100644 --- a/.agents/notes/implemented/feature/2026-07-27-trajectory-inspection-ledger.md +++ b/.agents/notes/implemented/feature/2026-07-27-trajectory-inspection-ledger.md @@ -21,7 +21,8 @@ Trajectory has to make prose, machine payloads, token usage, timing, and nested - Call schemas come from the active recorded Request header. Keyless snapshot fixtures deliberately replace that catalog with the non-array `{{tools}}` token, which the durable inspection boundary treats as unavailable instead of attempting to project or fabricate schemas. - Selecting a record or Request opens an inspector inside Trajectory. Tabs and Summary sections follow the selected entity: Markdown messages expose rendered, source, provenance, and hierarchy views; tools add JSON payload/result and schema views; Requests add options, usage, timing, and result navigation. Images render as media rather than serialized data. - Turn folding removes all rows after its first record and replaces them with a compact step/tool-call count; Assistant folding applies the same interaction to its tool-call descendants. Global controls fold or expand both levels. -- The separate Waterfall tab is removed. A fixed Overview above the ledger projects every record with known `startedAt` onto three semantic timing lanes using its own duration. Dragging left or right commits an inclusive interval filter: any record whose active interval overlaps either boundary remains visible, records without known timing leave the focused ledger, and clearing the selection restores the full branch. The Overview keeps the full time domain while focused so the selection can be resized or cleared without losing orientation. +- The separate Waterfall tab is removed. A fixed Overview above the ledger projects every record with known `startedAt` onto three semantic timing lanes using its own duration. Finalized Assistant spans divide the recorded interval at the first non-empty token delta, so distinct TTFT and decoding colors retain their actual ratio; incomplete timing falls back to one Assistant color. Hovering for 500 ms exposes exact start/end, total duration, TTFT, and decoding time without relying on the browser's native tooltip delay. Dragging left or right commits an inclusive interval filter: any record whose active interval overlaps either boundary remains visible, records without known timing leave the focused ledger, and clearing the selection restores the full branch. Wheel gestures zoom the time domain. A right-button click clears the interval selection; dragging instead pans an already zoomed viewport without mutating it. The Overview keeps the full time domain while focused so the selection can be resized or cleared without losing orientation. +- Live history updates retain the ledger's bottom position only while the user is already following its tail. Scrolling upward clears that follow state, so streamed chunks and newly appended records do not interrupt inspection of earlier rows. - This local inspector remains independent from the conversation-wide Chat details column. At narrow widths it overlays the ledger and remains dismissible by keyboard or pointer. ## Alternatives considered @@ -40,4 +41,4 @@ Trajectory has to make prose, machine payloads, token usage, timing, and nested ## Consequences -Trajectory shows more useful records per viewport while retaining Turn and Request orientation. Context rewrites and compactions remain inline with their surrounding history, while a rewind begins a successor branch that inherits only the retained prefix. The main ledger omits token usage and duration so content receives the available width; the local inspector exposes those facts together with full payloads, provenance, schemas, and request timing. The Overview uses recorded start/duration facts without fabricating live elapsed time, and its inclusive focus behavior matches the interaction users already know from Chrome DevTools Network. Focused component tests pin projection, folding, record and interval selection, entity-specific tabs, and running/error semantics; the assembled Web snapshot pins the ledger, Overview, and inspector through the real client composition. +Trajectory shows more useful records per viewport while retaining Turn and Request orientation. Context rewrites and compactions remain inline with their surrounding history, while a rewind begins a successor branch that inherits only the retained prefix. The main ledger omits token usage and duration so content receives the available width; the local inspector exposes those facts together with full payloads, provenance, schemas, and request timing. The Overview uses recorded start/duration and token-boundary facts without fabricating live elapsed time, and its inclusive focus behavior matches the interaction users already know from Chrome DevTools Network. Focused component tests pin tail following, timing projection, delayed detail disclosure, folding, record and interval selection, entity-specific tabs, and running/error semantics; the assembled Web snapshot pins the ledger, Overview timing details, and inspector through the real client composition. diff --git a/.agents/notes/implemented/feature/2026-07-27-trajectory-inspection-ledger.zh.md b/.agents/notes/implemented/feature/2026-07-27-trajectory-inspection-ledger.zh.md index 6c5733046d..ea31389d2f 100644 --- a/.agents/notes/implemented/feature/2026-07-27-trajectory-inspection-ledger.zh.md +++ b/.agents/notes/implemented/feature/2026-07-27-trajectory-inspection-ledger.zh.md @@ -21,7 +21,8 @@ Status: implemented - 调用 schema 来自当前生效且已记录的请求头。无密钥快照 fixture（测试前置数据）有意将该目录替换为非数组 token `{{tools}}`，持久化检查边界会将其视为不可用，而不是尝试投影或虚构 schema。 - 选择记录或请求后，轨迹视图内部会打开检查器，其标签页和概览区域随实体类型变化：Markdown 消息提供渲染、源码、来源和层级视图；工具提供 JSON 载荷／结果和 schema 视图；请求提供选项、用量、计时和结果跳转。图片以媒体形式渲染，而不是显示为序列化数据。 - 折叠轮次时保留其第一条记录，将后续行替换为紧凑的步骤和工具调用数量；折叠助手时对其工具调用后代应用相同交互。全局控件可以分别折叠或展开这两个层级。 -- 移除独立的 waterfall（瀑布式事件）标签页。固定在记录表上方的 Overview 区域将所有 `startedAt` 已知的记录按各自耗时投影到三条语义计时轨道。向左或向右拖动会提交包含边界的区间筛选：任何活动区间与所选区间相交的记录都会保留，计时未知的记录会从聚焦后的记录表中移除，清除选择则恢复完整分支。聚焦后，Overview 区域仍保留完整时间范围，以便在不失去方位的情况下调整或清除选择。 +- 移除独立的 waterfall（瀑布式事件）标签页。固定在记录表上方的 Overview 区域将所有 `startedAt` 已知的记录按各自耗时投影到三条语义计时轨道。已完成的助手时间条以首个非空 token 增量为分界，用不同颜色按真实比例表示 TTFT 与解码时间；计时不完整时退化为单一助手色。悬停 500 ms 后会显示精确起止时刻、总耗时、TTFT 和解码时间，而不依赖浏览器原生 tooltip 的延迟。向左或向右拖动会提交包含边界的区间筛选：任何活动区间与所选区间相交的记录都会保留，计时未知的记录会从聚焦后的记录表中移除，清除选择则恢复完整分支。滚轮手势用于缩放时间域。右键单击会清除区间选择；右键拖动则只会平移已放大的 viewport，不会改变该选区。聚焦后，Overview 区域仍保留完整时间范围，以便在不失去方位的情况下调整或清除选择。 +- 实时历史更新仅在用户已经跟随记录表末尾时保留底部位置。向上滚动会清除跟随状态，因此流式分块和新追加的记录不会打断对旧记录的检查。 - 此局部检查器与会话级 Chat 详情栏相互独立。在窄屏下，检查器会覆盖记录表，并且仍可通过键盘或指针关闭。 ## 曾考虑的替代方案 @@ -40,4 +41,4 @@ Status: implemented ## 后果 -轨迹视图在保留轮次与请求定位的同时，每个视口可以显示更多有效记录。上下文 `rewrite` 与压缩保持在周边历史中的原始位置，`rewind` 则建立仅继承保留前缀的后继分支。主记录表省略 token 用量和耗时，让内容获得可用宽度；局部检查器展示这些数据以及完整载荷、来源、schema 和请求计时。Overview 区域使用记录的开始时间与耗时数据，而不虚构实时流逝时间，其包含边界的聚焦行为与用户熟悉的 Chrome DevTools Network 交互一致。针对性组件测试锁定投影、折叠、记录与区间选择、实体特定标签页和运行／错误语义；组装后的 Web 快照则通过真实客户端组合锁定记录表、Overview 区域与检查器。 +轨迹视图在保留轮次与请求定位的同时，每个视口可以显示更多有效记录。上下文 `rewrite` 与压缩保持在周边历史中的原始位置，`rewind` 则建立仅继承保留前缀的后继分支。主记录表省略 token 用量和耗时，让内容获得可用宽度；局部检查器展示这些数据以及完整载荷、来源、schema 和请求计时。Overview 区域使用记录的开始时间、耗时与 token 边界数据，而不虚构实时流逝时间，其包含边界的聚焦行为与用户熟悉的 Chrome DevTools Network 交互一致。针对性组件测试锁定末尾跟随、计时投影、延迟展示详情、折叠、记录与区间选择、实体特定标签页和运行／错误语义；组装后的 Web 快照则通过真实客户端组合锁定记录表、Overview 计时详情与检查器。 diff --git a/packages/client/ui-trajectory/README.i18n.yaml b/packages/client/ui-trajectory/README.i18n.yaml index 41c954adc0..a62816aad2 100644 --- a/packages/client/ui-trajectory/README.i18n.yaml +++ b/packages/client/ui-trajectory/README.i18n.yaml @@ -2,5 +2,5 @@ # side as of the last confirmed-consistent state. Both languages carry equal authority; # after editing either side, bring the other along and re-record with: # pnpm run verify-translation-pairing --write packages/client/ui-trajectory/README.md -README.md: a65c11aed9dd74f9b0b60795441f876c1d64b3ad -README.zh.md: 6e25d24c6b65673b3d003e624b6e0727be60c0e1 +README.md: 7136532bff6f6b3fb79eb29a5bbb667fe44a6b74 +README.zh.md: 1edc81b2f05acca9aaa21995e2e9db812cbf4f62 diff --git a/packages/client/ui-trajectory/README.md b/packages/client/ui-trajectory/README.md index a65c11aed9..7136532bff 100644 --- a/packages/client/ui-trajectory/README.md +++ b/packages/client/ui-trajectory/README.md @@ -2,7 +2,7 @@ English | [中文](README.zh.md) -Trajectory renders a turn-aware event ledger with selectable User, Assistant, Tool, and nested Subtool records. Thick rules mark Turn boundaries, compact inline markers identify Steps, and the main ledger keeps only index, event, and content; selection opens a local inspector for token usage, duration, Input, Output, and Timing. A standalone compaction request appears chronologically in its own `Between turns` section, while a numbered compaction remains inside its owning turn. A fixed Overview above the ledger projects real record start/duration timing from left to right; dragging an interval focuses the ledger on every record active at any point in that inclusive range, while clearing the selection restores the full branch. The runtime's independent history source supplies raw context lineage and projects cancellation-frozen Assistant and Tool records, so Trajectory neither reads nor changes the Chat conversation snapshot. The package remains a pure-consumer plugin (registers one view tab into the conversation's `'conversation.view'` slot ring, provides no service, declares no Context merge). Contract: api-contracts v3 §8. +Trajectory renders a turn-aware event ledger with selectable User, Assistant, Tool, and nested Subtool records. Thick rules mark Turn boundaries, compact inline markers identify Steps, and the main ledger keeps only index, event, and content; selection opens a local inspector for token usage, duration, Input, Output, and Timing. A standalone compaction request appears chronologically in its own `Between turns` section, while a numbered compaction remains inside its owning turn. A fixed Overview above the ledger projects real record start/duration timing from left to right; Assistant spans divide recorded TTFT from decoding, and a 500 ms hover reveals exact clock and duration details. Dragging an interval focuses the ledger on every record active at any point in that inclusive range, while clearing the selection restores the full branch. Wheel gestures zoom the time domain. A right-button click clears the selected interval, while a right-button drag pans an already zoomed viewport without changing it. Streaming updates keep the ledger pinned only when it was already at the bottom, so reading earlier records suspends tail following. The runtime's independent history source supplies raw context lineage and projects cancellation-frozen Assistant and Tool records, so Trajectory neither reads nor changes the Chat conversation snapshot. The package remains a pure-consumer plugin (registers one view tab into the conversation's `'conversation.view'` slot ring, provides no service, declares no Context merge). Contract: api-contracts v3 §8. ## Model Experience diff --git a/packages/client/ui-trajectory/README.zh.md b/packages/client/ui-trajectory/README.zh.md index 6e25d24c6b..1edc81b2f0 100644 --- a/packages/client/ui-trajectory/README.zh.md +++ b/packages/client/ui-trajectory/README.zh.md @@ -2,7 +2,7 @@ [English](README.md) | 中文 -Trajectory 渲染按轮次组织的事件记录表，其中可选择用户、助手、工具和嵌套子工具记录。较粗的分割线标示轮次边界，紧凑的行内标记标识步骤，主记录表仅保留索引、事件和内容；选择记录则会打开局部检查器，查看 token 用量、耗时、输入、输出和计时。独立运行的压缩（compaction）请求会按时间顺序显示在自己的 `Between turns` 区段中，而带数值所有者的压缩仍位于其所属轮次内。固定在记录表上方的 Overview 区域从左到右投影记录的真实开始时间与耗时；拖选一个区间会将记录表聚焦到活动区间与该闭区间有重叠的所有记录，清除选择则恢复完整分支。运行时的独立历史数据源提供原始上下文谱系，并投影因取消而冻结的助手和工具记录，因此 Trajectory 既不读取也不改变 Chat 会话快照。该包（package）保持为纯消费方插件（向会话的 `'conversation.view'` slot 环注册一个视图标签页，不提供服务，也不声明 Context 合并）。契约：api-contracts v3 §8。 +Trajectory 渲染按轮次组织的事件记录表，其中可选择用户、助手、工具和嵌套子工具记录。较粗的分割线标示轮次边界，紧凑的行内标记标识步骤，主记录表仅保留索引、事件和内容；选择记录则会打开局部检查器，查看 token 用量、耗时、输入、输出和计时。独立运行的压缩（compaction）请求会按时间顺序显示在自己的 `Between turns` 区段中，而带数值所有者的压缩仍位于其所属轮次内。固定在记录表上方的 Overview 区域从左到右投影记录的真实开始时间与耗时；助手时间条会区分记录到的 TTFT 与解码时间，悬停 500 ms 后可查看精确时刻和耗时详情。拖选一个区间会将记录表聚焦到活动区间与该闭区间有重叠的所有记录，清除选择则恢复完整分支。滚轮手势用于缩放时间域。右键单击会清除所选区间；在已放大的 viewport 上按住右键拖动则只会平移视图，不会改变该区间。仅当记录表在流式更新前已经位于底部时，更新才会保持贴底；向上阅读旧记录会暂停跟随。运行时的独立历史数据源提供原始上下文谱系，并投影因取消而冻结的助手和工具记录，因此 Trajectory 既不读取也不改变 Chat 会话快照。该包（package）保持为纯消费方插件（向会话的 `'conversation.view'` slot 环注册一个视图标签页，不提供服务，也不声明 Context 合并）。契约：api-contracts v3 §8。 ## 模型体验 From 44304a458c577d9adc274896892aa82b8fd2a54b Mon Sep 17 00:00:00 2001 From: _Kerman Date: Mon, 3 Aug 2026 13:53:50 +0800 Subject: [PATCH 34/48] fix(ui-trajectory): separate consecutive request markers --- .../ui-trajectory/src/client/TrajectoryTable.module.css | 5 +++++ 1 file changed, 5 insertions(+) diff --git a/packages/client/ui-trajectory/src/client/TrajectoryTable.module.css b/packages/client/ui-trajectory/src/client/TrajectoryTable.module.css index 9e606ca5c6..ca8e55be68 100644 --- a/packages/client/ui-trajectory/src/client/TrajectoryTable.module.css +++ b/packages/client/ui-trajectory/src/client/TrajectoryTable.module.css @@ -104,6 +104,11 @@ border-bottom: 0; } +.table tbody tr[data-request-only='true']:has(+ tr[data-request-only='true']) td { + /* Keep consecutive boundary markers from painting their halos over one another. */ + height: 9px; +} + .table tbody tr[data-request-only='true']:last-child td { /* Retain the lower half of the 16px boundary marker at the table's end. */ height: 9px; From 062a1507f1bc521f62a149de5e74435efd5e7186 Mon Sep 17 00:00:00 2001 From: Yif <877193178@qq.com> Date: Fri, 31 Jul 2026 15:28:22 +0800 Subject: [PATCH 35/48] feat(ui-trajectory): float the composer over the ledger like chat The trajectory host kept the composer as a fixed flex sibling, so the ledger never reached the viewport bottom. Anchor the composer seat absolutely over the ledger (reusing chat's fade treatment) and have the internal scroll panes reserve the composer's live height plus a 16px gap so end rows and detail bodies scroll clear of the overlay. (cherry picked from commit 0058a1f4b6e8c2e23bfde7c827a84838cc5ebffc) --- .../src/client/TrajectoryTable.module.css | 4 +++- .../ui-trajectory/src/client/views.module.css | 20 +++++++++++++++++-- 2 files changed, 21 insertions(+), 3 deletions(-) diff --git a/packages/client/ui-trajectory/src/client/TrajectoryTable.module.css b/packages/client/ui-trajectory/src/client/TrajectoryTable.module.css index ca8e55be68..c494c0b862 100644 --- a/packages/client/ui-trajectory/src/client/TrajectoryTable.module.css +++ b/packages/client/ui-trajectory/src/client/TrajectoryTable.module.css @@ -16,6 +16,7 @@ flex: 1; min-width: 0; overflow: auto; + padding-bottom: var(--dsh-trajectory-bottom-clearance, 0px); container: trajectory-table / inline-size; } @@ -905,6 +906,7 @@ flex: 1; min-height: 0; overflow: auto; + padding-bottom: var(--dsh-trajectory-bottom-clearance, 0px); scrollbar-gutter: stable; } @@ -912,7 +914,7 @@ display: flex; box-sizing: border-box; flex-direction: column; - padding-bottom: 12px; + padding-bottom: calc(12px + var(--dsh-trajectory-bottom-clearance, 0px)); overflow: hidden; } diff --git a/packages/client/ui-trajectory/src/client/views.module.css b/packages/client/ui-trajectory/src/client/views.module.css index 326ac41a99..f23ab48b4d 100644 --- a/packages/client/ui-trajectory/src/client/views.module.css +++ b/packages/client/ui-trajectory/src/client/views.module.css @@ -14,9 +14,11 @@ } /* Trajectory keeps the ledger and details panel inside the remaining - * conversation height. Only the ledger pane scrolls; the composer remains - * the fixed flex sibling below this view. */ + * conversation height; only the internal panes scroll. The composer floats + * over the ledger like chat's sticky seat — absolute, not sticky, because + * this host does not scroll. */ :global([data-conversation-scroll]):has(.root) { + position: relative; overflow: hidden; } @@ -26,6 +28,15 @@ overflow: hidden; } +/* div qualifier outranks ConversationRoot's active-phase sticky rule (equal + * specificity otherwise, and cross-module source order is bundler-defined). */ +:global([data-conversation-scroll]):has(.root) > :global(div[data-composer-seat]) { + position: absolute; + right: 0; + bottom: 0; + left: 0; +} + .ledger { position: relative; z-index: 0; @@ -35,4 +46,9 @@ min-height: 0; min-width: 0; overflow: hidden; + + /* Internal panes reserve the floating composer's live height plus a 16px + * breathing gap so end rows and detail bodies can scroll clear of the + * overlay. */ + --dsh-trajectory-bottom-clearance: calc(var(--dsh-composer-height, 152px) + 16px); } From 95826603cbd18b1f58b3c4c4ed1da7e1ff471ad1 Mon Sep 17 00:00:00 2001 From: creatixchu Date: Mon, 3 Aug 2026 13:57:20 +0800 Subject: [PATCH 36/48] review(web): poll the pane-arity assertions in the path-editor e2e A bare count can observe a landing mid-commit on a loaded runner, and the arity is the invariant this scenario exists to pin. --- apps/web/tests/workspace-management.e2e.ts | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/apps/web/tests/workspace-management.e2e.ts b/apps/web/tests/workspace-management.e2e.ts index 075f5687ef..b25320f488 100644 --- a/apps/web/tests/workspace-management.e2e.ts +++ b/apps/web/tests/workspace-management.e2e.ts @@ -421,7 +421,7 @@ describe('web e2e: workspace management (create / rename / flat view / hover aff // still up and the draft intact. await path.fill(`${join(staged, 'alpha')}${sep}`) await expect.poll(() => dialog.getByText('only-under-alpha', { exact: true }).count(), { timeout: 10_000 }).toBe(1) - expect(await dialog.getByRole('list').count()).toBe(2) + await expect.poll(() => dialog.getByRole('list').count(), { timeout: 10_000 }).toBe(2) expect(await path.inputValue()).toBe(`${join(staged, 'alpha')}${sep}`) // Erasing back past the separator walks the panes up, so the level being // typed is the last pane again (its children no longer stand to its @@ -430,7 +430,7 @@ describe('web e2e: workspace management (create / rename / flat view / hover aff await expect.poll(() => dialog.getByText('only-under-alpha', { exact: true }).count(), { timeout: 10_000 }).toBe(0) expect(await dialog.getByText('alpha', { exact: true }).count()).toBe(1) expect(await dialog.getByText('beta', { exact: true }).count()).toBe(0) - expect(await dialog.getByRole('list').count()).toBe(2) + await expect.poll(() => dialog.getByRole('list').count(), { timeout: 10_000 }).toBe(2) // A tail nobody matches is a name still being spelled: the level shows // whole instead of emptying under it. await path.fill(`${staged}${sep}zzz`) From 5534431d423c3bc47765be1b8d399892590d1e7a Mon Sep 17 00:00:00 2001 From: _Kerman Date: Mon, 3 Aug 2026 14:02:40 +0800 Subject: [PATCH 37/48] fix(ui-trajectory): own composer overlay geometry --- apps/web/tests/navigation-panes.e2e.ts | 19 +++++++++++++ .../skeleton/ConversationRoot.module.css | 20 +++++++++++++ .../src/client/TrajectoryTable.module.css | 8 ++++-- .../src/client/TrajectoryView.tsx | 2 +- .../ui-trajectory/src/client/views.module.css | 28 +------------------ .../client/ui-trajectory/tests/views.spec.tsx | 1 + 6 files changed, 47 insertions(+), 31 deletions(-) diff --git a/apps/web/tests/navigation-panes.e2e.ts b/apps/web/tests/navigation-panes.e2e.ts index af4799a759..4f13ea21cf 100644 --- a/apps/web/tests/navigation-panes.e2e.ts +++ b/apps/web/tests/navigation-panes.e2e.ts @@ -138,6 +138,23 @@ describe('web e2e: navigation & panes over a rich seeded session', () => { onTestFailed(() => saveFailureShot(page, 'web-e2e-navigation-trajectory')) await page.getByRole('tab', { name: 'Trajectory' }).click() await page.waitForTimeout(100) + const overlayLayout = await page.getByRole('table').evaluate((table) => { + const host = table.closest('[data-conversation-scroll]') + const seat = host?.querySelector('[data-composer-seat]') ?? null + const pane = table.parentElement + return { + hostPosition: host === null ? null : getComputedStyle(host).position, + paneOverflowX: pane === null ? null : getComputedStyle(pane).overflowX, + paneScrollableWidth: pane === null ? null : pane.scrollWidth - pane.clientWidth, + seatPosition: seat === null ? null : getComputedStyle(seat).position, + } + }) + expect(overlayLayout).toEqual({ + hostPosition: 'relative', + paneOverflowX: 'hidden', + paneScrollableWidth: 0, + seatPosition: 'absolute', + }) expect({ pageErrors: tripwire.pageErrors, slotErrors, @@ -153,6 +170,8 @@ describe('web e2e: navigation & panes over a rich seeded session', () => { await page.locator('tr[data-kind="tool"]').first().click() const details = page.getByRole('complementary', { name: 'Event details' }) await expect.poll(() => details.count(), { timeout: 10_000 }).toBe(1) + expect(await details.getByRole('tabpanel').evaluate(panel => getComputedStyle(panel).overflowX)) + .toBe('hidden') await page.evaluate(() => { document.body.setAttribute('data-ds-dark-theme', '') }) const darkSummarySurfaces = await details.getByRole('heading', { name: 'Payload' }).evaluate(heading => ({ heading: getComputedStyle(heading).backgroundColor, diff --git a/packages/client/ui-conversation/src/client/skeleton/ConversationRoot.module.css b/packages/client/ui-conversation/src/client/skeleton/ConversationRoot.module.css index be480eaea2..1a83efb551 100644 --- a/packages/client/ui-conversation/src/client/skeleton/ConversationRoot.module.css +++ b/packages/client/ui-conversation/src/client/skeleton/ConversationRoot.module.css @@ -208,6 +208,26 @@ ); } +/* Views may opt into a composer overlay while ConversationRoot retains + ownership of the seat geometry and its active-phase precedence. */ +.scrollBody:has([data-conversation-composer-overlay]) { + position: relative; + overflow: hidden; +} + +.scrollBody:has([data-conversation-composer-overlay]) > .viewArea { + flex: 1 1 0; + min-height: 0; + overflow: hidden; +} + +.scrollBody:has([data-conversation-composer-overlay]) > .composerSeat { + position: absolute; + right: 0; + bottom: 0; + left: 0; +} + /* Hero phase: the composer stack (hero chrome + workspace row + card) is flex-centered in the column; composer phase docks it at the bottom. Flex, NOT absolute+transform: a transform would make this box the containing diff --git a/packages/client/ui-trajectory/src/client/TrajectoryTable.module.css b/packages/client/ui-trajectory/src/client/TrajectoryTable.module.css index c494c0b862..0b1cbf8030 100644 --- a/packages/client/ui-trajectory/src/client/TrajectoryTable.module.css +++ b/packages/client/ui-trajectory/src/client/TrajectoryTable.module.css @@ -15,7 +15,8 @@ .tablePane { flex: 1; min-width: 0; - overflow: auto; + overflow-x: hidden; + overflow-y: auto; padding-bottom: var(--dsh-trajectory-bottom-clearance, 0px); container: trajectory-table / inline-size; } @@ -28,7 +29,7 @@ ); width: 100%; - min-width: 480px; + min-width: 0; border-spacing: 0; table-layout: fixed; color: var(--dsw-alias-label-primary); @@ -905,7 +906,8 @@ .detailBody { flex: 1; min-height: 0; - overflow: auto; + overflow-x: hidden; + overflow-y: auto; padding-bottom: var(--dsh-trajectory-bottom-clearance, 0px); scrollbar-gutter: stable; } diff --git a/packages/client/ui-trajectory/src/client/TrajectoryView.tsx b/packages/client/ui-trajectory/src/client/TrajectoryView.tsx index d62074a26c..9ba75abaac 100644 --- a/packages/client/ui-trajectory/src/client/TrajectoryView.tsx +++ b/packages/client/ui-trajectory/src/client/TrajectoryView.tsx @@ -459,7 +459,7 @@ export function TrajectoryView({ } return ( -
+
{ diff --git a/packages/client/ui-trajectory/src/client/views.module.css b/packages/client/ui-trajectory/src/client/views.module.css index f23ab48b4d..687f4a4657 100644 --- a/packages/client/ui-trajectory/src/client/views.module.css +++ b/packages/client/ui-trajectory/src/client/views.module.css @@ -13,30 +13,6 @@ background: var(--dsw-alias-bg-layer-1); } -/* Trajectory keeps the ledger and details panel inside the remaining - * conversation height; only the internal panes scroll. The composer floats - * over the ledger like chat's sticky seat — absolute, not sticky, because - * this host does not scroll. */ -:global([data-conversation-scroll]):has(.root) { - position: relative; - overflow: hidden; -} - -:global([data-conversation-scroll]):has(.root) > :first-child { - flex: 1 1 0; - min-height: 0; - overflow: hidden; -} - -/* div qualifier outranks ConversationRoot's active-phase sticky rule (equal - * specificity otherwise, and cross-module source order is bundler-defined). */ -:global([data-conversation-scroll]):has(.root) > :global(div[data-composer-seat]) { - position: absolute; - right: 0; - bottom: 0; - left: 0; -} - .ledger { position: relative; z-index: 0; @@ -47,8 +23,6 @@ min-width: 0; overflow: hidden; - /* Internal panes reserve the floating composer's live height plus a 16px - * breathing gap so end rows and detail bodies can scroll clear of the - * overlay. */ + /* ConversationRoot publishes the floating composer's live height. */ --dsh-trajectory-bottom-clearance: calc(var(--dsh-composer-height, 152px) + 16px); } diff --git a/packages/client/ui-trajectory/tests/views.spec.tsx b/packages/client/ui-trajectory/tests/views.spec.tsx index 6b7be01ec3..0650e28746 100644 --- a/packages/client/ui-trajectory/tests/views.spec.tsx +++ b/packages/client/ui-trajectory/tests/views.spec.tsx @@ -288,6 +288,7 @@ describe('tab switching in ConversationRoot', () => { expect(screen.queryByRole('columnheader')).toBeNull() expect(screen.getByRole('toolbar', { name: 'Trajectory toolbar' })).toBeTruthy() expect(screen.getByRole('region', { name: 'Trajectory timeline' })).toBeTruthy() + expect(view.container.querySelector('[data-conversation-composer-overlay]')).toBeTruthy() fireEvent.click(screen.getByRole('button', { name: 'Collapse turns' })) expect(view.container.querySelector('[data-collapsed-summary="turn"]')).toBeTruthy() fireEvent.click(screen.getByRole('button', { name: 'Expand turns' })) From 667960182a2b9bc8694ef98e5b6ced0f62b6dcdb Mon Sep 17 00:00:00 2001 From: _Kerman Date: Mon, 3 Aug 2026 14:02:46 +0800 Subject: [PATCH 38/48] test(web): normalize detailed local clocks --- apps/web/tests/scaffold.ts | 1 + .../web/tests/snapshots/navigation-panes/trajectory.expected.md | 2 +- 2 files changed, 2 insertions(+), 1 deletion(-) diff --git a/apps/web/tests/scaffold.ts b/apps/web/tests/scaffold.ts index 1b8afbb247..7a48c4d8ac 100644 --- a/apps/web/tests/scaffold.ts +++ b/apps/web/tests/scaffold.ts @@ -516,6 +516,7 @@ function normalizeAria(snapshot: string, workspaceCwd: string): string { // shape so goldens stay stable across midnight and year boundaries. .replace(/\d{4}年\d{1,2}月\d{1,2}日 \d{2}:\d{2}/g, '{{clock}}') .replace(/\d{1,2}月\d{1,2}日 \d{2}:\d{2}/g, '{{clock}}') + .replace(/(? Date: Mon, 3 Aug 2026 14:02:54 +0800 Subject: [PATCH 39/48] docs(ui-trajectory): record composer overlay contract --- .../2026-07-27-trajectory-inspection-ledger.i18n.yaml | 4 ++-- .../feature/2026-07-27-trajectory-inspection-ledger.md | 5 ++++- .../feature/2026-07-27-trajectory-inspection-ledger.zh.md | 5 ++++- packages/client/ui-trajectory/README.i18n.yaml | 4 ++-- packages/client/ui-trajectory/README.md | 2 +- packages/client/ui-trajectory/README.zh.md | 2 +- 6 files changed, 14 insertions(+), 8 deletions(-) diff --git a/.agents/notes/implemented/feature/2026-07-27-trajectory-inspection-ledger.i18n.yaml b/.agents/notes/implemented/feature/2026-07-27-trajectory-inspection-ledger.i18n.yaml index 17b8219739..d7e0f75f2d 100644 --- a/.agents/notes/implemented/feature/2026-07-27-trajectory-inspection-ledger.i18n.yaml +++ b/.agents/notes/implemented/feature/2026-07-27-trajectory-inspection-ledger.i18n.yaml @@ -2,5 +2,5 @@ # side as of the last confirmed-consistent state. Both languages carry equal authority; # after editing either side, bring the other along and re-record with: # pnpm run verify-translation-pairing --write .agents/notes/implemented/feature/2026-07-27-trajectory-inspection-ledger.md -2026-07-27-trajectory-inspection-ledger.md: fcdbbb30b065b0b128a8e375cd2e8ed5b2702dac -2026-07-27-trajectory-inspection-ledger.zh.md: ea31389d2fb0970caae271e7d64e4ecff215e93c +2026-07-27-trajectory-inspection-ledger.md: cdeaa30ea64f47b0e0110baf566f747a4591a384 +2026-07-27-trajectory-inspection-ledger.zh.md: df2a3d266161a7c1c4444863971f3d177533af8c diff --git a/.agents/notes/implemented/feature/2026-07-27-trajectory-inspection-ledger.md b/.agents/notes/implemented/feature/2026-07-27-trajectory-inspection-ledger.md index fcdbbb30b0..cdeaa30ea6 100644 --- a/.agents/notes/implemented/feature/2026-07-27-trajectory-inspection-ledger.md +++ b/.agents/notes/implemented/feature/2026-07-27-trajectory-inspection-ledger.md @@ -23,6 +23,7 @@ Trajectory has to make prose, machine payloads, token usage, timing, and nested - Turn folding removes all rows after its first record and replaces them with a compact step/tool-call count; Assistant folding applies the same interaction to its tool-call descendants. Global controls fold or expand both levels. - The separate Waterfall tab is removed. A fixed Overview above the ledger projects every record with known `startedAt` onto three semantic timing lanes using its own duration. Finalized Assistant spans divide the recorded interval at the first non-empty token delta, so distinct TTFT and decoding colors retain their actual ratio; incomplete timing falls back to one Assistant color. Hovering for 500 ms exposes exact start/end, total duration, TTFT, and decoding time without relying on the browser's native tooltip delay. Dragging left or right commits an inclusive interval filter: any record whose active interval overlaps either boundary remains visible, records without known timing leave the focused ledger, and clearing the selection restores the full branch. Wheel gestures zoom the time domain. A right-button click clears the interval selection; dragging instead pans an already zoomed viewport without mutating it. The Overview keeps the full time domain while focused so the selection can be resized or cleared without losing orientation. - Live history updates retain the ledger's bottom position only while the user is already following its tail. Scrolling upward clears that follow state, so streamed chunks and newly appended records do not interrupt inspection of earlier rows. +- Trajectory opts into a conversation-owned composer overlay through `data-conversation-composer-overlay`. `ConversationRoot` positions the composer seat and publishes its live height; Trajectory keeps the ledger at full height and reserves that height plus 16 px inside its vertical table and inspector scrollers. Those panes adapt to the available width instead of exposing horizontal scrollbars beneath the overlay. - This local inspector remains independent from the conversation-wide Chat details column. At narrow widths it overlays the ledger and remains dismissible by keyboard or pointer. ## Alternatives considered @@ -35,10 +36,12 @@ Trajectory has to make prose, machine payloads, token usage, timing, and nested **Reuse the global Chat details column.** Rejected: it would couple local inspection to conversation navigation and make a row click unexpectedly change another view's state. +**Override the composer seat from Trajectory CSS.** Rejected: a cross-package selector would depend on generated class specificity and stylesheet order. An explicit view marker keeps seat geometry and active-phase precedence in `ConversationRoot`, while Trajectory owns only its internal clearance. + **Keep timing in a separate Waterfall tab.** Rejected: the placeholder summarized node counts rather than record timing and forced users to switch away from the rows they wanted to focus. A full-domain Overview keeps timing and filtered records in one visual context. **Change global theme tokens to match the reference.** Rejected: the existing theme already provides paired light and dark semantic layers, and a local redesign does not justify changing unrelated surfaces. ## Consequences -Trajectory shows more useful records per viewport while retaining Turn and Request orientation. Context rewrites and compactions remain inline with their surrounding history, while a rewind begins a successor branch that inherits only the retained prefix. The main ledger omits token usage and duration so content receives the available width; the local inspector exposes those facts together with full payloads, provenance, schemas, and request timing. The Overview uses recorded start/duration and token-boundary facts without fabricating live elapsed time, and its inclusive focus behavior matches the interaction users already know from Chrome DevTools Network. Focused component tests pin tail following, timing projection, delayed detail disclosure, folding, record and interval selection, entity-specific tabs, and running/error semantics; the assembled Web snapshot pins the ledger, Overview timing details, and inspector through the real client composition. +Trajectory shows more useful records per viewport while retaining Turn and Request orientation. Context rewrites and compactions remain inline with their surrounding history, while a rewind begins a successor branch that inherits only the retained prefix. The floating composer leaves the ledger visible to the viewport edge without covering its final rows or hiding horizontal controls. The main ledger omits token usage and duration so content receives the available width; the local inspector exposes those facts together with full payloads, provenance, schemas, and request timing. The Overview uses recorded start/duration and token-boundary facts without fabricating live elapsed time, and its inclusive focus behavior matches the interaction users already know from Chrome DevTools Network. Focused component tests pin tail following, timing projection, delayed detail disclosure, folding, record and interval selection, entity-specific tabs, and running/error semantics; the assembled Web snapshot pins the ledger, Overview timing details, composer overlay geometry, and inspector through the real client composition. diff --git a/.agents/notes/implemented/feature/2026-07-27-trajectory-inspection-ledger.zh.md b/.agents/notes/implemented/feature/2026-07-27-trajectory-inspection-ledger.zh.md index ea31389d2f..df2a3d2661 100644 --- a/.agents/notes/implemented/feature/2026-07-27-trajectory-inspection-ledger.zh.md +++ b/.agents/notes/implemented/feature/2026-07-27-trajectory-inspection-ledger.zh.md @@ -23,6 +23,7 @@ Status: implemented - 折叠轮次时保留其第一条记录，将后续行替换为紧凑的步骤和工具调用数量；折叠助手时对其工具调用后代应用相同交互。全局控件可以分别折叠或展开这两个层级。 - 移除独立的 waterfall（瀑布式事件）标签页。固定在记录表上方的 Overview 区域将所有 `startedAt` 已知的记录按各自耗时投影到三条语义计时轨道。已完成的助手时间条以首个非空 token 增量为分界，用不同颜色按真实比例表示 TTFT 与解码时间；计时不完整时退化为单一助手色。悬停 500 ms 后会显示精确起止时刻、总耗时、TTFT 和解码时间，而不依赖浏览器原生 tooltip 的延迟。向左或向右拖动会提交包含边界的区间筛选：任何活动区间与所选区间相交的记录都会保留，计时未知的记录会从聚焦后的记录表中移除，清除选择则恢复完整分支。滚轮手势用于缩放时间域。右键单击会清除区间选择；右键拖动则只会平移已放大的 viewport，不会改变该选区。聚焦后，Overview 区域仍保留完整时间范围，以便在不失去方位的情况下调整或清除选择。 - 实时历史更新仅在用户已经跟随记录表末尾时保留底部位置。向上滚动会清除跟随状态，因此流式分块和新追加的记录不会打断对旧记录的检查。 +- Trajectory 通过 `data-conversation-composer-overlay` 启用由会话持有的 composer 浮层模式。`ConversationRoot` 负责定位 composer seat 并发布其实时高度；Trajectory 让记录表保持全高，并在记录表与检查器的纵向滚动容器内预留该高度加 16 px。这两个窗格会根据可用宽度自适应，而不会在浮层下方暴露横向滚动条。 - 此局部检查器与会话级 Chat 详情栏相互独立。在窄屏下，检查器会覆盖记录表，并且仍可通过键盘或指针关闭。 ## 曾考虑的替代方案 @@ -35,10 +36,12 @@ Status: implemented **复用全局 Chat 详情栏。** 不予采纳：这会让局部检查与会话导航耦合，还会使行点击意外改变另一个视图的状态。 +**由 Trajectory CSS 覆盖 composer seat。** 不予采纳：跨包（package）选择器会依赖生成类选择器的优先级和样式表顺序。显式视图标记让 seat 几何形状和活跃阶段优先级留在 `ConversationRoot` 中，而 Trajectory 只负责自身内部的避让空间。 + **将计时保留在独立的 waterfall 标签页中。** 不予采纳：占位实现汇总的是节点数而非记录计时，并迫使用户离开想要聚焦的记录。保留完整时间范围的 Overview 区域让计时和筛选后的记录处于同一视觉上下文中。 **修改全局主题 token 以匹配参考设计。** 不予采纳：现有主题已经提供配对的亮色与暗色语义层，局部重新设计不足以成为修改无关表面的理由。 ## 后果 -轨迹视图在保留轮次与请求定位的同时，每个视口可以显示更多有效记录。上下文 `rewrite` 与压缩保持在周边历史中的原始位置，`rewind` 则建立仅继承保留前缀的后继分支。主记录表省略 token 用量和耗时，让内容获得可用宽度；局部检查器展示这些数据以及完整载荷、来源、schema 和请求计时。Overview 区域使用记录的开始时间、耗时与 token 边界数据，而不虚构实时流逝时间，其包含边界的聚焦行为与用户熟悉的 Chrome DevTools Network 交互一致。针对性组件测试锁定末尾跟随、计时投影、延迟展示详情、折叠、记录与区间选择、实体特定标签页和运行／错误语义；组装后的 Web 快照则通过真实客户端组合锁定记录表、Overview 计时详情与检查器。 +轨迹视图在保留轮次与请求定位的同时，每个视口可以显示更多有效记录。上下文 `rewrite` 与压缩保持在周边历史中的原始位置，`rewind` 则建立仅继承保留前缀的后继分支。浮动 composer 让记录表一直显示到视口边缘，同时不会遮住最后几行，也不会隐藏横向控件。主记录表省略 token 用量和耗时，让内容获得可用宽度；局部检查器展示这些数据以及完整载荷、来源、schema 和请求计时。Overview 区域使用记录的开始时间、耗时与 token 边界数据，而不虚构实时流逝时间，其包含边界的聚焦行为与用户熟悉的 Chrome DevTools Network 交互一致。针对性组件测试锁定末尾跟随、计时投影、延迟展示详情、折叠、记录与区间选择、实体特定标签页和运行／错误语义；组装后的 Web 快照则通过真实客户端组合锁定记录表、Overview 计时详情、composer 浮层几何形状与检查器。 diff --git a/packages/client/ui-trajectory/README.i18n.yaml b/packages/client/ui-trajectory/README.i18n.yaml index a62816aad2..36e56c4569 100644 --- a/packages/client/ui-trajectory/README.i18n.yaml +++ b/packages/client/ui-trajectory/README.i18n.yaml @@ -2,5 +2,5 @@ # side as of the last confirmed-consistent state. Both languages carry equal authority; # after editing either side, bring the other along and re-record with: # pnpm run verify-translation-pairing --write packages/client/ui-trajectory/README.md -README.md: 7136532bff6f6b3fb79eb29a5bbb667fe44a6b74 -README.zh.md: 1edc81b2f05acca9aaa21995e2e9db812cbf4f62 +README.md: 5d0ea3bbbbfca2b8c0ee02ed07ca956fbd377e11 +README.zh.md: 1bfff4c18ea2e834781e2c6cb76773595eeed5ad diff --git a/packages/client/ui-trajectory/README.md b/packages/client/ui-trajectory/README.md index 7136532bff..5d0ea3bbbb 100644 --- a/packages/client/ui-trajectory/README.md +++ b/packages/client/ui-trajectory/README.md @@ -2,7 +2,7 @@ English | [中文](README.zh.md) -Trajectory renders a turn-aware event ledger with selectable User, Assistant, Tool, and nested Subtool records. Thick rules mark Turn boundaries, compact inline markers identify Steps, and the main ledger keeps only index, event, and content; selection opens a local inspector for token usage, duration, Input, Output, and Timing. A standalone compaction request appears chronologically in its own `Between turns` section, while a numbered compaction remains inside its owning turn. A fixed Overview above the ledger projects real record start/duration timing from left to right; Assistant spans divide recorded TTFT from decoding, and a 500 ms hover reveals exact clock and duration details. Dragging an interval focuses the ledger on every record active at any point in that inclusive range, while clearing the selection restores the full branch. Wheel gestures zoom the time domain. A right-button click clears the selected interval, while a right-button drag pans an already zoomed viewport without changing it. Streaming updates keep the ledger pinned only when it was already at the bottom, so reading earlier records suspends tail following. The runtime's independent history source supplies raw context lineage and projects cancellation-frozen Assistant and Tool records, so Trajectory neither reads nor changes the Chat conversation snapshot. The package remains a pure-consumer plugin (registers one view tab into the conversation's `'conversation.view'` slot ring, provides no service, declares no Context merge). Contract: api-contracts v3 §8. +Trajectory renders a turn-aware event ledger with selectable User, Assistant, Tool, and nested Subtool records. Thick rules mark Turn boundaries, compact inline markers identify Steps, and the main ledger keeps only index, event, and content; selection opens a local inspector for token usage, duration, Input, Output, and Timing. A standalone compaction request appears chronologically in its own `Between turns` section, while a numbered compaction remains inside its owning turn. A fixed Overview above the ledger projects real record start/duration timing from left to right; Assistant spans divide recorded TTFT from decoding, and a 500 ms hover reveals exact clock and duration details. Dragging an interval focuses the ledger on every record active at any point in that inclusive range, while clearing the selection restores the full branch. Wheel gestures zoom the time domain. A right-button click clears the selected interval, while a right-button drag pans an already zoomed viewport without changing it. Streaming updates keep the ledger pinned only when it was already at the bottom, so reading earlier records suspends tail following. Trajectory asks the conversation shell to float the composer over the full-height ledger, while its responsive vertical scrollers reserve the composer's live height so final rows remain reachable. The runtime's independent history source supplies raw context lineage and projects cancellation-frozen Assistant and Tool records, so Trajectory neither reads nor changes the Chat conversation snapshot. The package remains a pure-consumer plugin (registers one view tab into the conversation's `'conversation.view'` slot ring, provides no service, declares no Context merge). Contract: api-contracts v3 §8. ## Model Experience diff --git a/packages/client/ui-trajectory/README.zh.md b/packages/client/ui-trajectory/README.zh.md index 1edc81b2f0..1bfff4c18e 100644 --- a/packages/client/ui-trajectory/README.zh.md +++ b/packages/client/ui-trajectory/README.zh.md @@ -2,7 +2,7 @@ [English](README.md) | 中文 -Trajectory 渲染按轮次组织的事件记录表，其中可选择用户、助手、工具和嵌套子工具记录。较粗的分割线标示轮次边界，紧凑的行内标记标识步骤，主记录表仅保留索引、事件和内容；选择记录则会打开局部检查器，查看 token 用量、耗时、输入、输出和计时。独立运行的压缩（compaction）请求会按时间顺序显示在自己的 `Between turns` 区段中，而带数值所有者的压缩仍位于其所属轮次内。固定在记录表上方的 Overview 区域从左到右投影记录的真实开始时间与耗时；助手时间条会区分记录到的 TTFT 与解码时间，悬停 500 ms 后可查看精确时刻和耗时详情。拖选一个区间会将记录表聚焦到活动区间与该闭区间有重叠的所有记录，清除选择则恢复完整分支。滚轮手势用于缩放时间域。右键单击会清除所选区间；在已放大的 viewport 上按住右键拖动则只会平移视图，不会改变该区间。仅当记录表在流式更新前已经位于底部时，更新才会保持贴底；向上阅读旧记录会暂停跟随。运行时的独立历史数据源提供原始上下文谱系，并投影因取消而冻结的助手和工具记录，因此 Trajectory 既不读取也不改变 Chat 会话快照。该包（package）保持为纯消费方插件（向会话的 `'conversation.view'` slot 环注册一个视图标签页，不提供服务，也不声明 Context 合并）。契约：api-contracts v3 §8。 +Trajectory 渲染按轮次组织的事件记录表，其中可选择用户、助手、工具和嵌套子工具记录。较粗的分割线标示轮次边界，紧凑的行内标记标识步骤，主记录表仅保留索引、事件和内容；选择记录则会打开局部检查器，查看 token 用量、耗时、输入、输出和计时。独立运行的压缩（compaction）请求会按时间顺序显示在自己的 `Between turns` 区段中，而带数值所有者的压缩仍位于其所属轮次内。固定在记录表上方的 Overview 区域从左到右投影记录的真实开始时间与耗时；助手时间条会区分记录到的 TTFT 与解码时间，悬停 500 ms 后可查看精确时刻和耗时详情。拖选一个区间会将记录表聚焦到活动区间与该闭区间有重叠的所有记录，清除选择则恢复完整分支。滚轮手势用于缩放时间域。右键单击会清除所选区间；在已放大的 viewport 上按住右键拖动则只会平移视图，不会改变该区间。仅当记录表在流式更新前已经位于底部时，更新才会保持贴底；向上阅读旧记录会暂停跟随。Trajectory 要求会话壳将 composer 作为浮层置于全高记录表上方；其响应式纵向滚动容器会预留 composer 的实时高度，确保仍可滚动到最后几行。运行时的独立历史数据源提供原始上下文谱系，并投影因取消而冻结的助手和工具记录，因此 Trajectory 既不读取也不改变 Chat 会话快照。该包（package）保持为纯消费方插件（向会话的 `'conversation.view'` slot 环注册一个视图标签页，不提供服务，也不声明 Context 合并）。契约：api-contracts v3 §8。 ## 模型体验 From 0a4ad25ce7390a3bec55e09a3d32644e68769b8e Mon Sep 17 00:00:00 2001 From: ZiyaZhang <199893125+ZiyaZhang@users.noreply.github.com> Date: Sun, 2 Aug 2026 23:06:00 -0700 Subject: [PATCH 40/48] fix(web): expand aborted Bash rows --- apps/web/tests/bash-abort-row.e2e.ts | 79 +++++++++++++++++++ .../snapshots/bash-abort-row/ui.expected.md | 32 ++++++++ apps/web/tsconfig.json | 3 +- .../client/ui-conversation/README.i18n.yaml | 4 +- packages/client/ui-conversation/README.md | 2 +- packages/client/ui-conversation/README.zh.md | 2 +- .../client/toolviews/bash-sample.module.css | 57 +++++++++++++ .../src/client/toolviews/bash-sample.tsx | 54 ++++++++++--- .../tests/terminal-card.spec.tsx | 22 ++++++ tsconfig.host.json | 1 + 10 files changed, 240 insertions(+), 16 deletions(-) create mode 100644 apps/web/tests/bash-abort-row.e2e.ts create mode 100644 apps/web/tests/snapshots/bash-abort-row/ui.expected.md diff --git a/apps/web/tests/bash-abort-row.e2e.ts b/apps/web/tests/bash-abort-row.e2e.ts new file mode 100644 index 0000000000..657d910e6d --- /dev/null +++ b/apps/web/tests/bash-abort-row.e2e.ts @@ -0,0 +1,79 @@ +// Web e2e scenario: a cancelled Bash call can settle without terminal-card +// material. Borrow the real cancellation fixture and prove the keyed Bash row +// still exposes the recorded command and full error without any model call. +import { readFile } from 'node:fs/promises' +import { fileURLToPath } from 'node:url' +import { join } from 'node:path' +import type { Browser, Page } from 'playwright' +import { chromium } from 'playwright' +import { afterAll, beforeAll, describe, expect, it, onTestFailed } from 'vitest' +import { + assertFixtureInventory, captureStableAria, compareOrRefreshGolden, fixtureUserPrompts, + launchWebScaffold, seedSession, watchConsole, webSnapshotMode, type WebScaffold, +} from './scaffold.ts' +import { newEnglishPage, saveFailureShot } from './support.ts' + +const FIXTURE = fileURLToPath(new URL('../../../examples/acp-agent/tests/snapshots/cancel-tool-calls/session.jsonl', import.meta.url)) +const SNAPSHOT_DIR = fileURLToPath(new URL('./snapshots/bash-abort-row', import.meta.url)) +const UI_EXPECTED = join(SNAPSHOT_DIR, 'ui.expected.md') +const MODE = webSnapshotMode() +const SEED_ID = 'bash-abort-row-web-e2e' +const PROMPT = 'Run two shell commands: wait for cancellation, then write skipped.txt.' + +describe.skipIf(MODE === 'record')('web e2e: cancelled Bash row disclosure', () => { + let scaffold: WebScaffold + let browser: Browser + let page: Page + let tripwire: ReturnType + + beforeAll(async () => { + const fixture = await readFile(FIXTURE, 'utf8') + expect(fixtureUserPrompts(fixture)).toEqual([PROMPT]) + scaffold = await launchWebScaffold({}) + await seedSession(scaffold, fixture, SEED_ID) + browser = await chromium.launch() + page = await newEnglishPage(browser) + tripwire = watchConsole(page) + await page.goto(scaffold.baseUrl, { waitUntil: 'load' }) + await page.waitForSelector('[class*="frame"]', { timeout: 30_000 }) + + const groupRow = page.locator('[role="treeitem"]').first() + await groupRow.waitFor({ timeout: 15_000 }) + await groupRow.click() + const sessionRow = page.locator('[role="treeitem"]').nth(1) + await sessionRow.waitFor({ timeout: 10_000 }) + await sessionRow.click() + await page.locator('[data-sample="bash"]').nth(1).waitFor({ timeout: 15_000 }) + }, 120_000) + + afterAll(async () => { + await browser?.close() + await scaffold?.close() + }) + + it('expands the aborted row to its command and full error', async () => { + onTestFailed(() => saveFailureShot(page, 'web-e2e-bash-abort-row')) + const row = page.locator('[data-sample="bash"]').first() + const call = row.locator('xpath=..') + await expect.poll(() => row.getAttribute('aria-expanded')).toBe('false') + await expect.poll(() => call.getByText('Error: command aborted', { exact: true }).count()).toBe(1) + await row.click() + + await expect.poll(() => row.getAttribute('aria-expanded')).toBe('true') + await call.getByText('IN', { exact: true }).waitFor() + await call.getByText('OUT', { exact: true }).waitFor() + await call.getByText('Wait until cancellation', { exact: false }).waitFor() + await call.getByText('setInterval(() => {}, 1000)', { exact: false }).waitFor() + await expect.poll(() => call.getByText('Error: command aborted', { exact: true }).count()).toBe(2) + + const snapshot = (await captureStableAria(page, '[class*="centerCol"]', scaffold.workspaceCwd)) + .split(SEED_ID).join('{{seededId}}') + await compareOrRefreshGolden(UI_EXPECTED, snapshot, MODE) + expect(tripwire.pageErrors).toEqual([]) + expect(tripwire.warnings).toEqual([]) + }, 60_000) + + it('keeps its snapshot inventory closed', async () => { + await assertFixtureInventory(SNAPSHOT_DIR, ['ui.expected.md']) + }) +}) diff --git a/apps/web/tests/snapshots/bash-abort-row/ui.expected.md b/apps/web/tests/snapshots/bash-abort-row/ui.expected.md new file mode 100644 index 0000000000..798111ee21 --- /dev/null +++ b/apps/web/tests/snapshots/bash-abort-row/ui.expected.md @@ -0,0 +1,32 @@ +- banner: + - navigation "Session hierarchy": + - 'button "Run two shell commands: wait" [disabled]' + - tablist: + - tab "Chat" [selected] + - tab "Trajectory" +- text: "Run two shell commands: wait for cancellation, then write skipped.txt. 7/18 {{clock}}" +- button "Copy": + - img +- button "Branch into a new conversation": + - img +- button "Context injection": + - img + - img + - text: Context injection +- 'button "Failed Bash Error: command aborted" [expanded]': + - img + - text: "Failed Bash Error: command aborted" +- text: "IN { \"command\": \"node -e \\\"require('node:fs').writeFileSync('started.txt', 'started'); setInterval(() => {}, 1000)\\\"\", \"description\": \"Wait until cancellation\" } OUT Error: command aborted" +- button "Inspect" +- 'button "Failed Bash Error: tool call aborted before dispatch"': + - img + - text: "Failed Bash Error: tool call aborted before dispatch" +- textbox "Message the agent" +- button "Commands": + - img +- 'button "Access mode, current: Workspace Write"': Workspace Write +- button "Select model, current deepseek-v4-flash": + - text: deepseek-v4-flash + - img +- button "Send message" [disabled] +- text: 1 turns · 1 steps Tool call {{duration}} Cache hit 0% Input 10 tok · Output 10 tok diff --git a/apps/web/tsconfig.json b/apps/web/tsconfig.json index 9693ed1885..5128d01da8 100644 --- a/apps/web/tsconfig.json +++ b/apps/web/tsconfig.json @@ -53,7 +53,8 @@ "tests/shipped-composition.e2e.ts", "tests/goal-bar.e2e.ts", "tests/startup-auto-selection.e2e.ts", - "tests/subagent-conversation.e2e.ts" + "tests/subagent-conversation.e2e.ts", + "tests/bash-abort-row.e2e.ts" ], "references": [ { diff --git a/packages/client/ui-conversation/README.i18n.yaml b/packages/client/ui-conversation/README.i18n.yaml index d131e31202..1dc3761c5f 100644 --- a/packages/client/ui-conversation/README.i18n.yaml +++ b/packages/client/ui-conversation/README.i18n.yaml @@ -2,5 +2,5 @@ # side as of the last confirmed-consistent state. Both languages carry equal authority; # after editing either side, bring the other along and re-record with: # pnpm run verify-translation-pairing --write packages/client/ui-conversation/README.md -README.md: e610b990dd89204fd7e22e8b86f807d10b8ba439 -README.zh.md: 268e05a806db1468ba689608c178646af132fe25 +README.md: 728d67aaccee609d5f28ee8744c1f11ca74b040a +README.zh.md: eec1d36b9758dddcf63eb75c3a74c51b4afb0a72 diff --git a/packages/client/ui-conversation/README.md b/packages/client/ui-conversation/README.md index e610b990dd..728d67aacc 100644 --- a/packages/client/ui-conversation/README.md +++ b/packages/client/ui-conversation/README.md @@ -20,7 +20,7 @@ A Think row stays collapsed by default and exposes live reasoning throughput wit Generic tool rows classify the built-in bash, read, search, write, edit, and run_code names into dedicated visual variants. The filesystem variants render the edit icon and a path summary; that path is a hover-underline link that opens the file with the host OS default application (`host.openPath`, relative paths resolve against the session cwd). Tool rows are not whole-row click targets and do not open the details panel. The code variant summarizes with the model-authored `description` and expands to the program itself; its logged sub-dispatches render as always-visible nested rows through the SAME keyed toolview hole (custom registrations and the GenericToolCard fallback apply to sub-rows unchanged). Cordis lifecycle tools reuse those generic variants while presenting `Inspect`, `Mount temporary Plugin`, and `Unmount temporary Plugin` with a shared Cordis accent; mount keeps the code variant's expandable source rendering. -A tool call declaring the `terminal` render intent renders its command output inline, at both conversation render sites, through ui-primitives' `TerminalBlock`. `contract/terminal-card-model.ts` is the single derivation from the snapshot's `callView`/`resultView` pair, so the sites cannot disagree about a command, its cwd, or its exit status; it yields null — the generic path — for any other card tag, including one this client version does not know. Both sites therefore also show the card's run-state dot, which is the same `StateDot` semantic a tool row's leading icon carries, so a row and its own card always agree about one command's state. A multi-line command gets one prompt row per line, with the dot marking the call once on the first row — the exit status is the whole call's, so a dot per line would claim a per-line outcome bash does not report. The keyed `BashRow` carries the card resident below its summary row; since tool rows are no longer details-panel click targets, the card's copy and expand controls are the row's only interactions. The render-site fallback row keeps the card behind its existing expand control. Rows cap at `CHAT_TERMINAL_MAX_LINES` (8) against the panel's 16, which is what keeps a summary surface bounded — the panel stays the single-call reading surface. Inline output is licensed per render intent — the terminal and web cards, each with its own bound; a generic tool's content remains panel-only ([decision](../../../.agents/notes/implemented/feature/2026-07-28-web-terminal-card.md)). +A tool call declaring the `terminal` render intent renders its command output inline, at both conversation render sites, through ui-primitives' `TerminalBlock`. `contract/terminal-card-model.ts` is the single derivation from the snapshot's `callView`/`resultView` pair, so the sites cannot disagree about a command, its cwd, or its exit status; it yields null — the generic path — for any other card tag, including one this client version does not know. Both sites therefore also show the card's run-state dot, which is the same `StateDot` semantic a tool row's leading icon carries, so a row and its own card always agree about one command's state. A multi-line command gets one prompt row per line, with the dot marking the call once on the first row — the exit status is the whole call's, so a dot per line would claim a per-line outcome bash does not report. The keyed `BashRow` carries the card resident below its summary row; since tool rows are no longer details-panel click targets, the card's copy and expand controls are the row's only interactions. The render-site fallback row keeps the card behind its existing expand control. Rows cap at `CHAT_TERMINAL_MAX_LINES` (8) against the panel's 16, which is what keeps a summary surface bounded — the panel stays the single-call reading surface. Inline output is licensed per render intent — the terminal and web cards, each with its own bound. A Bash execution failure that settles on the generic path instead exposes its original arguments and full error through the same bounded IN/OUT disclosure, while successful generic results such as a background-start acknowledgement remain summary-only ([decision](../../../.agents/notes/implemented/feature/2026-07-28-web-terminal-card.md)). A tool call declaring the `web` render intent renders its web retrieval inline, at both conversation render sites, through ui-primitives' `WebBlock`. `contract/web-card-model.ts` is the single derivation from the snapshot's `resultView`, mirroring the terminal card, so the sites cannot disagree about what a web call shows; it yields null — the generic path — for a running call, a non-web result view, a generic result view, a `card` tag this client version does not know, or a web card whose `kind` this client version does not know (a newer host's value, which the wire cannot be trusted to be `search` or `fetch`). The keyed `WebRow` registers one component under both `web_search` and `web_fetch`, discriminating on the tool name only for its icon and title; it composes the shared `ToolRow`, feeding the card as ToolRow's `web` body, so the retrieval is the row's collapsed-by-default expanded card (the same unified expand every card row has). A web-declaring tool without a keyed row lands on the `GenericToolCard` fallback, which routes the card through ToolRow the same way, and the details panel renders it at the primitive's full source allowance and, below the card, the flattened model-visible result content — a fetch body is readable only there, since its card carries only the URL and status. Rows cap at `CHAT_WEB_MAX_SOURCES` (8) against the panel's 16, the same summary-versus-reading split the terminal card draws ([decision](../../../.agents/notes/implemented/feature/2026-07-30-web-result-card-frontend.md)). diff --git a/packages/client/ui-conversation/README.zh.md b/packages/client/ui-conversation/README.zh.md index 268e05a806..eec1d36b97 100644 --- a/packages/client/ui-conversation/README.zh.md +++ b/packages/client/ui-conversation/README.zh.md @@ -18,7 +18,7 @@ Think 行默认保持折叠，并在不展开思维链的情况下暴露实时 通用工具行把内置的 bash、read、search、write、edit 和 run_code 名称归入专用视觉变体。文件系统变体会渲染 edit 图标和路径摘要；该路径是悬停下划线链接，点击后通过宿主操作系统的默认应用打开文件（`host.openPath`，相对路径相对会话 cwd 解析）。工具行不再是整行点击目标，也不会打开 details 面板。code 变体以模型撰写的 `description` 作摘要，展开后显示程序本身；其已记录的子调用经由同一个键控 toolview 空位渲染为始终可见的嵌套行（自定义注册和 GenericToolCard fallback 原样适用于子行）。Cordis 生命周期工具复用这些通用变体，同时以统一的 Cordis 强调色呈现 `Inspect`、`Mount temporary Plugin` 和 `Unmount temporary Plugin`；mount 行保留 code 变体的可展开源码渲染。 -声明 `terminal` 渲染意图的工具调用，会在两个对话渲染点上都通过 ui-primitives 的 `TerminalBlock` 内联渲染其命令输出。`contract/terminal-card-model.ts` 是从快照的 `callView`／`resultView` 对推导的唯一位置，因此两个渲染点不可能在命令、cwd 或退出状态上产生分歧；对任何其他 card 标签——包括当前客户端版本不认识的标签——它返回 null，落回通用路径。因此两个渲染点也都显示卡片的运行状态点，它与工具行行首图标承载同一套 `StateDot` 语义，所以一行与其自身的卡片对同一条命令的状态总是一致。多行命令的每一行各占一个提示行，状态点只在第一行为整次调用标记一次——退出状态属于整次调用，因此每行一枚就会声称一个 bash 并不报告的逐行结果。键控的 `BashRow` 把卡片常驻在摘要行下方；由于工具行已不再是详情面板的点击目标，卡片的复制与展开控件就是该行唯一的交互。渲染点兜底行则保持其既有的展开控件。行的上限是 `CHAT_TERMINAL_MAX_LINES`（8），面板为 16，正是这一点让摘要面保持有界——面板仍是单次调用的阅读面。内联输出按渲染意图开放——终端卡片与 web 卡片，各有自己的上限；通用工具的内容仍然只在面板中呈现（[决策](../../../.agents/notes/implemented/feature/2026-07-28-web-terminal-card.md)）。 +声明 `terminal` 渲染意图的工具调用，会在两个对话渲染点上都通过 ui-primitives 的 `TerminalBlock` 内联渲染其命令输出。`contract/terminal-card-model.ts` 是从快照的 `callView`／`resultView` 对推导的唯一位置，因此两个渲染点不可能在命令、cwd 或退出状态上产生分歧；对任何其他 card 标签——包括当前客户端版本不认识的标签——它返回 null，落回通用路径。因此两个渲染点也都显示卡片的运行状态点，它与工具行行首图标承载同一套 `StateDot` 语义，所以一行与其自身的卡片对同一条命令的状态总是一致。多行命令的每一行各占一个提示行，状态点只在第一行为整次调用标记一次——退出状态属于整次调用，因此每行一枚就会声称一个 bash 并不报告的逐行结果。键控的 `BashRow` 把卡片常驻在摘要行下方；由于工具行已不再是详情面板的点击目标，卡片的复制与展开控件就是该行唯一的交互。渲染点兜底行则保持其既有的展开控件。行的上限是 `CHAT_TERMINAL_MAX_LINES`（8），面板为 16，正是这一点让摘要面保持有界——面板仍是单次调用的阅读面。内联输出按渲染意图开放——终端卡片与 web 卡片，各有自己的上限。若 Bash 执行失败时落在通用路径，则改用同样有界的 IN/OUT 展开区暴露原始参数和完整错误；后台启动确认等成功的通用结果仍只显示摘要（[决策](../../../.agents/notes/implemented/feature/2026-07-28-web-terminal-card.md)）。 声明 `web` 渲染意图的工具调用，会在两个对话渲染点上都通过 ui-primitives 的 `WebBlock` 内联渲染其 web 检索。`contract/web-card-model.ts` 是从快照的 `resultView` 推导的唯一位置，镜像终端卡片，因此两个渲染点不可能对一次 web 调用的显示产生分歧；对运行中的调用、非 web 的 result view、generic result view、本客户端版本不认识的 `card` 标签，或本客户端版本不认识 `kind` 的 web 卡片（更新的 host 发来的值，wire 上不可信其为 `search` 或 `fetch`），它返回 null，落回通用路径。键控的 `WebRow` 把一个组件注册在 `web_search` 与 `web_fetch` 两个键下，仅根据工具名判别以选取图标与标题；它组合共享的 `ToolRow`，把卡片作为 ToolRow 的 `web` body 传入，因此检索成为该行默认折叠的展开卡片（与每个卡片行相同的统一展开交互）。没有自己键控行的 web 声明工具落到 `GenericToolCard` 兜底，它以同样方式经 ToolRow 渲染卡片，详情面板则以原语的完整 source 额度渲染它，并在卡片下方渲染摊平的模型可见结果内容——fetch 正文只在此处可读，因为其卡片只携带 URL 和状态。行的上限是 `CHAT_WEB_MAX_SOURCES`（8），面板为 16，与终端卡片所画的摘要面对阅读面的同一划分（[决策](../../../.agents/notes/implemented/feature/2026-07-30-web-result-card-frontend.md)）。 diff --git a/packages/client/ui-conversation/src/client/toolviews/bash-sample.module.css b/packages/client/ui-conversation/src/client/toolviews/bash-sample.module.css index ef9f246dd7..d4323309d4 100644 --- a/packages/client/ui-conversation/src/client/toolviews/bash-sample.module.css +++ b/packages/client/ui-conversation/src/client/toolviews/bash-sample.module.css @@ -20,6 +20,63 @@ border: 1px solid var(--dsw-alias-border-l1); } +/* A bash execution error can settle without terminal-card material (for + example, command cancellation). Preserve ToolRow's bounded IN/OUT fallback + so the original command and full error remain available from this keyed row. */ +.ioCard { + display: flex; + flex-direction: column; + margin: 4px 0 4px 4px; + border: 1px solid var(--dsw-alias-border-l1); + border-radius: 12px; + background: var(--dsw-alias-markdown-code-block); + font: var(--dsw-font-markdown-code-block-small); +} + +.ioSection { + display: grid; + grid-template-columns: max-content 1fr; + column-gap: 14px; + align-items: baseline; + padding: 12px 16px; + max-height: 150px; + overflow-y: auto; +} + +.ioSection::-webkit-scrollbar-thumb { + border: 2px solid transparent; + background-clip: padding-box; + border-radius: 6px; +} + +.ioSection::-webkit-scrollbar-track { + margin: 6px 0; +} + +.ioLabel { + position: sticky; + top: 0; + align-self: start; + color: var(--dsw-alias-label-caption); +} + +.ioDivider { + flex: none; + height: 1px; + background: var(--dsw-alias-border-l2); +} + +.ioText { + min-width: 0; + white-space: pre-wrap; + word-break: break-word; + color: var(--dsw-alias-label-secondary); +} + +.ioText[data-error] { + color: var(--dsw-alias-state-error-primary); +} + /* ToolRow's unified expand interaction, replicated per the registrant posture: pointer on the expandable row (the icon→chevron hover preview is the affordance, no row fill). */ diff --git a/packages/client/ui-conversation/src/client/toolviews/bash-sample.tsx b/packages/client/ui-conversation/src/client/toolviews/bash-sample.tsx index 5a3e3f40fe..fbe05e0a1b 100644 --- a/packages/client/ui-conversation/src/client/toolviews/bash-sample.tsx +++ b/packages/client/ui-conversation/src/client/toolviews/bash-sample.tsx @@ -2,8 +2,10 @@ // (ctx.slots.register + ToolRowProps only — never imports the chat domain). // Product chrome matches ToolRow / Think (figma: Bash · {description}). // -// A bash call declares the terminal render intent, so this row renders the -// command's own output through TerminalBlock — expand-gated exactly like +// A bash call normally declares the terminal render intent, so this row renders +// the command's own output through TerminalBlock. Execution failures that +// settle without terminal material use the bounded generic IN/OUT fallback — +// both are expand-gated exactly like // ToolRow's unified interaction: collapsed by default, the whole summary row // is the toggle (click / Enter / Space, icon→chevron hover preview; the // summary stays inline while open), @@ -48,7 +50,7 @@ function stateStatus(state: ToolRowState, t: BashRowProps['t']): string | null { /** * Bash row: icon + Bash · {description} in the shared ToolRow chrome, the - * whole row toggling the command's terminal card (ToolRow's unified + * whole row toggling the command's terminal or generic error card (ToolRow's unified * expand interaction, replicated locally per the registrant posture). */ export function BashRow({ toolName, block, sessionId, useSessions, inspect, t }: BashRowProps) { @@ -64,7 +66,13 @@ export function BashRow({ toolName, block, sessionId, useSessions, inspect, t }: : model.state const status = stateStatus(state, t) const [expanded, setExpanded] = useState(false) - const expandable = terminal !== null + // Execution failures (for example cancellation before the process reports a + // terminal result) use the generic presenter. Keep their recorded args and + // full error reachable instead of collapsing the row to the first line. + const genericError = terminal === null + && model.state === 'error' + && (model.body !== null || model.output !== null) + const expandable = terminal !== null || genericError const open = expanded && expandable const failureLine = model.state === 'error' ? model.errorSummary : null const toggleExpand = () => { @@ -109,16 +117,40 @@ export function BashRow({ toolName, block, sessionId, useSessions, inspect, t }: {failureLine ?? terminal?.description ?? model.summary}
- {terminal !== null && open && ( + {open && ( /* Same hover-Inspect posture as ToolRow's expanded body, replicated locally per the registrant posture. */
- + {terminal !== null + ? ( + + ) + : ( +
+ {model.body !== null && ( +
+ IN + {model.body} +
+ )} + {model.body !== null && model.output !== null && ( + + )} + {model.output !== null && ( +
+ OUT + + {model.output} + +
+ )} +
+ )} {inspect !== undefined && ( {showBranch && onBranch !== undefined && ( - - )} + {showBranch && onBranch !== undefined && branchUnavailable && ( + {t('message.branchUnavailable')} + )} {clock === 'end' ? clockEl : null}
) diff --git a/packages/client/ui-conversation/src/client/chat/MessageItem.tsx b/packages/client/ui-conversation/src/client/chat/MessageItem.tsx index 7ea870e326..df04ac19c2 100644 --- a/packages/client/ui-conversation/src/client/chat/MessageItem.tsx +++ b/packages/client/ui-conversation/src/client/chat/MessageItem.tsx @@ -26,8 +26,10 @@ export interface MessageItemProps { | TurnErrorNode | UnknownSurfaceNode retryActive?: boolean - /** Fork through this message's completed turn when it is the transcript tail. */ + /** Fork through this message's completed turn when eligible. */ onFork?: (seq: number) => void + /** The message is not the transcript tail of a completed turn. */ + forkUnavailable?: boolean /** The owning view's locale seat, passed down as a plain prop. */ t: ChatViewSlotProps['t'] } @@ -220,7 +222,7 @@ export function PendingSteeringBubble({ content, t }: { } export const MessageItem = memo(function MessageItem({ - node, retryActive = false, onFork, t, + node, retryActive = false, onFork, forkUnavailable = false, t, }: MessageItemProps) { const truncated = (total: number): string => t('json.truncated', { total }) switch (node.kind) { @@ -236,6 +238,7 @@ export const MessageItem = memo(function MessageItem({ time={node.time} clock="start" onBranch={onFork === undefined ? undefined : () => { onFork(node.seq) }} + branchUnavailable={forkUnavailable} className={css.actions} t={t} /> diff --git a/packages/client/ui-conversation/src/client/chat/chat-flow.ts b/packages/client/ui-conversation/src/client/chat/chat-flow.ts index cabcba23dd..9de9876d89 100644 --- a/packages/client/ui-conversation/src/client/chat/chat-flow.ts +++ b/packages/client/ui-conversation/src/client/chat/chat-flow.ts @@ -50,8 +50,8 @@ export function assistantActionsSeqs(nodes: readonly ConversationNode[]): Readon /** * Seq set of message rows that may fork: the last transcript node of a * completed turn, when that node owns message chrome. A later tool, reasoning, - * error, or other transcript node suppresses the earlier message's branch - * action even though the Host would include the whole turn. + * error, or other transcript node leaves the earlier message's branch action + * unavailable because the Host would include the whole turn. * @param nodes - snapshot nodes in event order. * @param turnEnds - completed turn boundaries retained from the event window. * @returns Message seq values whose visible position matches the fork boundary. diff --git a/packages/client/ui-conversation/src/client/locales.ts b/packages/client/ui-conversation/src/client/locales.ts index 87252fbc77..7bc01b4d95 100644 --- a/packages/client/ui-conversation/src/client/locales.ts +++ b/packages/client/ui-conversation/src/client/locales.ts @@ -59,6 +59,7 @@ export const zh = { 'message.unknownBlock': '未知内容块', 'message.stopped': '已停止', 'message.branch': '在新对话中分支', + 'message.branchUnavailable': '仅可从已完成轮次的最后一条消息分支', 'message.retry.active': '正在重试模型请求', 'message.retry.cancelled': '模型请求重试已取消', 'message.retry.started': '已重试模型请求', @@ -166,6 +167,7 @@ export const en = { 'message.unknownBlock': 'Unknown content block', 'message.stopped': 'Stopped', 'message.branch': 'Branch into a new conversation', + 'message.branchUnavailable': 'Available only on the last message of a completed turn', 'message.retry.active': 'Retrying model request', 'message.retry.cancelled': 'Model request retry cancelled', 'message.retry.started': 'Retried model request', diff --git a/packages/client/ui-conversation/tests/chat-branch-tails.spec.tsx b/packages/client/ui-conversation/tests/chat-branch-tails.spec.tsx index 92a804ebfd..89a09ca28d 100644 --- a/packages/client/ui-conversation/tests/chat-branch-tails.spec.tsx +++ b/packages/client/ui-conversation/tests/chat-branch-tails.spec.tsx @@ -78,6 +78,30 @@ describe('MessageItem arms', () => { expect(exec).toHaveBeenCalledWith('copy') }) + it('keeps an unavailable branch focusable and explains why without sending a fork', () => { + const onFork = vi.fn() + render( + , + ) + const branch = screen.getByRole('button', { name: '在新对话中分支' }) as HTMLButtonElement + expect(branch.disabled).toBe(false) + expect(branch.getAttribute('aria-disabled')).toBe('true') + const reasonId = branch.getAttribute('aria-describedby') + expect(reasonId).not.toBeNull() + expect(document.getElementById(reasonId!)?.textContent).toBe('仅可从已完成轮次的最后一条消息分支') + fireEvent.click(branch) + expect(onFork).not.toHaveBeenCalled() + fireEvent.focus(branch) + expect(screen.getByRole('tooltip').textContent).toBe('仅可从已完成轮次的最后一条消息分支') + }) + it('user copy stays quiet when execCommand throws or is absent', () => { Object.defineProperty(navigator, 'clipboard', { configurable: true, diff --git a/packages/client/ui-conversation/tests/chat-view.spec.tsx b/packages/client/ui-conversation/tests/chat-view.spec.tsx index 5a06f588ee..a041aca214 100644 --- a/packages/client/ui-conversation/tests/chat-view.spec.tsx +++ b/packages/client/ui-conversation/tests/chat-view.spec.tsx @@ -325,14 +325,19 @@ describe('ChatView', () => { expect(view.getAllByText('interrupt now')).toHaveLength(1) expect(view.container.querySelector('[data-pending-steering]')).toBeNull() expect(view.getAllByRole('button', { name: '复制' })).toHaveLength(2) - expect(view.queryByRole('button', { name: '在新对话中分支' })).toBeNull() + const durableBubble = view.getByText('interrupt now').closest('[class*="userRow"]') as HTMLElement + const unavailable = within(durableBubble).getByRole('button', { name: '在新对话中分支' }) + expect(unavailable.getAttribute('aria-disabled')).toBe('true') + fireEvent.click(unavailable) + expect(h.forkAt).not.toHaveBeenCalled() act(() => { h.set({ running: false, turnEnds: new Map([[1, 3]]) }) }) const branchButtons = view.getAllByRole('button', { name: '在新对话中分支' }) - expect(branchButtons).toHaveLength(1) - fireEvent.click(branchButtons[0]!) + expect(branchButtons).toHaveLength(2) + expect(branchButtons.map(button => button.getAttribute('aria-disabled'))).toEqual(['true', null]) + fireEvent.click(branchButtons[1]!) expect(h.forkAt).toHaveBeenCalledWith(2) }) @@ -435,24 +440,28 @@ describe('ChatView', () => { turnEnds: new Map([[1, 4], [2, 6]]), }) const view = render() - // User rows keep copy/clock, while only the two completed assistant tails may branch. + // Every message footer keeps branch visible; only completed assistant tails enable it. expect(view.getAllByRole('button', { name: '复制' })).toHaveLength(4) - expect(view.getAllByRole('button', { name: '在新对话中分支' })).toHaveLength(2) + const branchButtons = view.getAllByRole('button', { name: '在新对话中分支' }) + expect(branchButtons).toHaveLength(4) + expect(branchButtons.map(button => button.getAttribute('aria-disabled'))).toEqual(['true', null, 'true', null]) }) - it('forks only from a finalized assistant at the completed transcript tail', () => { + it('enables fork only on the finalized assistant at the completed transcript tail', () => { const h = makeHarness({ nodes: [user(1, 'question'), assistant(2, 'answer')], turnEnds: new Map([[1, 3]]), }) const view = render() const buttons = view.getAllByRole('button', { name: '在新对话中分支' }) - expect(buttons).toHaveLength(1) + expect(buttons).toHaveLength(2) + expect(buttons.map(button => button.getAttribute('aria-disabled'))).toEqual(['true', null]) fireEvent.click(buttons[0]!) + fireEvent.click(buttons[1]!) expect(h.forkAt.mock.calls).toEqual([[2]]) }) - it('keeps copy chrome but hides branch when tool and interrupted Think follow the response', () => { + it('keeps branch visible but unavailable when tool and interrupted Think follow the response', () => { const interruptedThink: AssistantMessageNode = { kind: 'assistant', seq: 4.1, time: 4_100, turn: 1, step: 2, blocks: [{ kind: 'reasoning', text: 'bad path' }], interrupted: true, @@ -463,7 +472,12 @@ describe('ChatView', () => { }) const view = render() expect(view.getAllByRole('button', { name: '复制' })).toHaveLength(2) - expect(view.queryByRole('button', { name: '在新对话中分支' })).toBeNull() + const buttons = view.getAllByRole('button', { name: '在新对话中分支' }) + expect(buttons).toHaveLength(2) + expect(buttons.every(button => button.getAttribute('aria-disabled') === 'true')).toBe(true) + fireEvent.click(buttons[0]!) + fireEvent.click(buttons[1]!) + expect(h.forkAt).not.toHaveBeenCalled() }) it('renders assistant Markdown across history, streaming, final, and interrupted states while user text stays literal', () => { From 886c39d30200ed36dc8196bd14a673292bec038d Mon Sep 17 00:00:00 2001 From: Yichen Jiang Date: Mon, 3 Aug 2026 16:21:02 +0800 Subject: [PATCH 47/48] test(web): refresh Markdown image snapshot --- .../snapshots/markdown-images/ui.expected.md | 15 ++++++--------- 1 file changed, 6 insertions(+), 9 deletions(-) diff --git a/apps/web/tests/snapshots/markdown-images/ui.expected.md b/apps/web/tests/snapshots/markdown-images/ui.expected.md index cd33a09eb9..76e01397c2 100644 --- a/apps/web/tests/snapshots/markdown-images/ui.expected.md +++ b/apps/web/tests/snapshots/markdown-images/ui.expected.md @@ -5,27 +5,24 @@ - tab "Chat" [selected] - tab "Trajectory" - text: Show the Markdown image policy. {{clock}} -- button "复制": +- button "Copy": - img -- button "在新对话中分支": - - img -- button "编辑": +- button "Branch into a new conversation": - img - heading "Markdown images" [level=2] - paragraph: - img "Remote test image" - paragraph: Local test image - paragraph: REMOTE_IMAGE_DONE -- button "复制": +- button "Copy": - img -- button "在新对话中分支": +- button "Branch into a new conversation": - img - text: {{clock}} - textbox "Message the agent" -- button "Add attachment": +- button "Commands": - img -- 'button "Access mode, current: Danger Full Access"': Danger Full Access -- button "Plan mode off, press to turn on": Plan off +- 'button "Access mode, current: Workspace Write"': Workspace Write - button "Select model, current deepseek-v4-flash": - text: deepseek-v4-flash - img From f3fd8e4cd34c919d42057e77d95e2488ff799954 Mon Sep 17 00:00:00 2001 From: Tianyi Cui <53024+tianyicui@users.noreply.github.com> Date: Mon, 3 Aug 2026 16:38:11 +0800 Subject: [PATCH 48/48] chore: add Issue management runtime --- .github/ISSUE_TEMPLATE/bug.md | 22 + .github/ISSUE_TEMPLATE/config.yml | 2 + .github/ISSUE_TEMPLATE/feature.md | 20 + .github/ISSUE_TEMPLATE/idea.md | 20 + .github/ISSUE_TEMPLATE/research.md | 21 + .github/ISSUE_TEMPLATE/task.md | 20 + .github/issue-management/config.json | 17 + .github/issue-management/policy.mjs | 545 +++++++++++++++++++++++ .github/issue-management/policy.test.mjs | 240 ++++++++++ .github/pull_request_template.md | 13 + 10 files changed, 920 insertions(+) create mode 100644 .github/ISSUE_TEMPLATE/bug.md create mode 100644 .github/ISSUE_TEMPLATE/config.yml create mode 100644 .github/ISSUE_TEMPLATE/feature.md create mode 100644 .github/ISSUE_TEMPLATE/idea.md create mode 100644 .github/ISSUE_TEMPLATE/research.md create mode 100644 .github/ISSUE_TEMPLATE/task.md create mode 100644 .github/issue-management/config.json create mode 100644 .github/issue-management/policy.mjs create mode 100644 .github/issue-management/policy.test.mjs create mode 100644 .github/pull_request_template.md diff --git a/.github/ISSUE_TEMPLATE/bug.md b/.github/ISSUE_TEMPLATE/bug.md new file mode 100644 index 0000000000..9427dc1d52 --- /dev/null +++ b/.github/ISSUE_TEMPLATE/bug.md @@ -0,0 +1,22 @@ +--- +name: Bug +about: 记录现有预期行为的失效 +title: '' +labels: '' +assignees: '' +type: Bug +--- + + +一句话说明错误结果。 + +
+复现、预期与验收 + +- 复现步骤： +- 实际结果： +- 预期结果： +- 环境： +- 验收条件： + +
diff --git a/.github/ISSUE_TEMPLATE/config.yml b/.github/ISSUE_TEMPLATE/config.yml new file mode 100644 index 0000000000..8005e32267 --- /dev/null +++ b/.github/ISSUE_TEMPLATE/config.yml @@ -0,0 +1,2 @@ +blank_issues_enabled: false +contact_links: [] diff --git a/.github/ISSUE_TEMPLATE/feature.md b/.github/ISSUE_TEMPLATE/feature.md new file mode 100644 index 0000000000..2c65f5544d --- /dev/null +++ b/.github/ISSUE_TEMPLATE/feature.md @@ -0,0 +1,20 @@ +--- +name: Feature +about: 新增或有意改变可观察行为 +title: '' +labels: '' +assignees: '' +type: Feature +--- + + +一句话说明预期结果。 + +
+验收与细节 + +- 验收条件： +- 用户或模型可见变化： +- 测试证据： + +
diff --git a/.github/ISSUE_TEMPLATE/idea.md b/.github/ISSUE_TEMPLATE/idea.md new file mode 100644 index 0000000000..c8bf80402d --- /dev/null +++ b/.github/ISSUE_TEMPLATE/idea.md @@ -0,0 +1,20 @@ +--- +name: Idea +about: 记录尚未承诺实施、但具有行动可能的想法 +title: '' +labels: '' +assignees: '' +type: Idea +--- + + +一句话说明价值假设。 + +
+价值与细节 + +- 价值假设： +- 需要验证： +- 可能的后续工作： + +
diff --git a/.github/ISSUE_TEMPLATE/research.md b/.github/ISSUE_TEMPLATE/research.md new file mode 100644 index 0000000000..8acfe2cc6e --- /dev/null +++ b/.github/ISSUE_TEMPLATE/research.md @@ -0,0 +1,21 @@ +--- +name: Research +about: 形成结论、证据或决策 +title: '' +labels: '' +assignees: '' +type: Research +--- + + +一句话说明待回答的问题。 + +
+问题与证据标准 + +- 核心问题： +- 证据标准： +- 交付结论： +- 可能的后续工作： + +
diff --git a/.github/ISSUE_TEMPLATE/task.md b/.github/ISSUE_TEMPLATE/task.md new file mode 100644 index 0000000000..376855c27c --- /dev/null +++ b/.github/ISSUE_TEMPLATE/task.md @@ -0,0 +1,20 @@ +--- +name: Task +about: 明确的非 Feature、非 Bug 工作 +title: '' +labels: '' +assignees: '' +type: Task +--- + + +一句话说明要完成的工作。 + +
+验收与细节 + +- 验收条件： +- 交付物： +- 测试证据： + +
diff --git a/.github/issue-management/config.json b/.github/issue-management/config.json new file mode 100644 index 0000000000..41019f0aa2 --- /dev/null +++ b/.github/issue-management/config.json @@ -0,0 +1,17 @@ +{ + "organization": "deepseek-harness", + "repository": "deepseek-harness", + "projectNumber": 1, + "projectTitle": "DSH Issue Management", + "priorityField": "Priority", + "allowUnassignedOwner": true, + "statuses": [ + "Inbox", + "Backlog", + "Ready", + "In progress", + "In review", + "Done", + "No action" + ] +} diff --git a/.github/issue-management/policy.mjs b/.github/issue-management/policy.mjs new file mode 100644 index 0000000000..bc8c881eda --- /dev/null +++ b/.github/issue-management/policy.mjs @@ -0,0 +1,545 @@ +#!/usr/bin/env node + +import fs from 'node:fs' +import process from 'node:process' +import { pathToFileURL } from 'node:url' + +import config from './config.json' with { type: 'json' } + +const API_VERSION = '2026-03-10' +const BODY_LIMIT = 50 +const AUDIT_MARKER = '' +const OWNER_LINE = /^Owner: @([A-Za-z0-9](?:[A-Za-z0-9-]{0,37}[A-Za-z0-9])?)$/ +const TYPES = new Set(['Idea', 'Feature', 'Bug', 'Research', 'Task']) +const PRIORITIES = ['p0', 'p1', 'p2', 'p3'] + +/** + * Return Markdown outside balanced details elements. + * @param {string} body Markdown body. + * @returns {{text: string, balanced: boolean, detailsCount: number, allCollapsed: boolean}} Visible source and details shape. + */ +export function extractOutsideDetails(body) { + const source = body.replace(//g, '') + const tag = /<\/?details\b[^>]*>/gi + let depth = 0 + let cursor = 0 + let balanced = true + let text = '' + let detailsCount = 0 + let allCollapsed = true + + for (const match of source.matchAll(tag)) { + const index = match.index ?? 0 + if (depth === 0) text += source.slice(cursor, index) + if (/^<\//.test(match[0])) { + if (depth === 0) balanced = false + else depth -= 1 + } else { + depth += 1 + detailsCount += 1 + if (/\sopen(?:\s|=|>)/i.test(match[0])) allCollapsed = false + } + cursor = index + match[0].length + } + + if (depth === 0) text += source.slice(cursor) + if (depth !== 0) balanced = false + return { text, balanced, detailsCount, allCollapsed } +} + +/** + * Count Chinese characters and contiguous Latin, numeric, or code tokens. + * @param {string} body Markdown body. + * @returns {{units: number, balanced: boolean, detailsCount: number, allCollapsed: boolean}} Visible unit count and details shape. + */ +export function countVisibleUnits(body) { + const outside = extractOutsideDetails(body) + const visible = outside.text + .replace(/!\[([^\]]*)\]\([^)]*\)/g, '$1') + .replace(/\[([^\]]+)\]\([^)]*\)/g, '$1') + .replace(/\[([^\]]+)\]\[[^\]]*\]/g, '$1') + .replace(/<((?:https?:\/\/|mailto:)[^>]+)>/gi, '$1') + .replace(/<[^>]+>/g, ' ') + .replace(/&(?:[A-Za-z]+|#\d+|#x[0-9A-Fa-f]+);/g, ' ') + .replace(/[\u0060*~\[\]{}()<>#!|]/g, ' ') + const han = visible.match(/\p{Script=Han}/gu)?.length ?? 0 + const tokens = visible.match(/[\p{Script=Latin}\p{Number}_./:@+-]+/gu)?.length ?? 0 + return { + units: han + tokens, + balanced: outside.balanced, + detailsCount: outside.detailsCount, + allCollapsed: outside.allCollapsed, + } +} + +function firstNonblankLine(body) { + return body + .split(/\r?\n/) + .map((line) => line.trim()) + .find(Boolean) +} + +/** + * Validate body shape and Owner against assignees. + * @param {{body: string, assignees: string[], allowUnassignedOwner?: boolean}} input Body input. + * @returns {string[]} Validation errors. + */ +export function validateBody({ + body, + assignees, + allowUnassignedOwner = config.allowUnassignedOwner ?? false, +}) { + const errors = [] + const count = countVisibleUnits(body) + const owner = firstNonblankLine(body)?.match(OWNER_LINE)?.[1] ?? null + const normalized = [...new Set(assignees.map((login) => login.toLowerCase()))] + + if (!count.balanced) errors.push('details 标签必须成对闭合') + if (count.detailsCount === 0) errors.push('正文必须包含默认收起的
区域') + if (!count.allCollapsed) errors.push('details 必须默认收起，不得设置 open') + if (count.units > BODY_LIMIT) { + errors.push(`正文外露部分为 ${count.units} 单位，超过 50 单位`) + } + if (normalized.length >= 2 && !owner) { + errors.push('多个 Assignees 时首个非空行必须是 Owner: @login') + } else if (normalized.length >= 2 && !normalized.includes(owner.toLowerCase())) { + errors.push('Owner 必须属于 Assignees') + } else if ( + normalized.length < 2 && + owner && + !(normalized.length === 0 && allowUnassignedOwner) + ) { + errors.push('零或一个 Assignee 时不得写 Owner 行') + } + return errors +} + +/** + * Decide whether a PR has entered the human-review enforcement boundary. + * @param {{isDraft: boolean, authorType: string, reviewRequestCount: number, reviewCount: number}} input PR state. + * @returns {boolean} Whether the PR policy is mandatory. + */ +export function requiresPullRequestPolicy({ + isDraft, + authorType, + reviewRequestCount, + reviewCount, +}) { + const automated = authorType === 'Bot' || authorType === 'App' + return !isDraft && !automated && (reviewRequestCount > 0 || reviewCount > 0) +} + +function stripIgnoredMarkdown(body) { + const lines = body.replace(//g, '').split(/\r?\n/) + const kept = [] + let fence = null + for (const line of lines) { + const marker = line.match(/^\s*([\u0060~]{3,})/) + if (marker) { + if (fence === null) fence = marker[1][0] + else if (marker[1][0] === fence) fence = null + continue + } + if (fence === null) kept.push(line) + } + return kept.join('\n').replace(/\u0060[^\u0060]*\u0060/g, ' ') +} + +/** + * Parse same-repository resolving and informational references. + * @param {{body: string, repository: string}} input PR body and repository. + * @returns {{all: number[], resolving: number[], related: number[]}} References. + */ +export function parseReferences({ body, repository }) { + const source = stripIgnoredMarkdown(body) + const expected = repository.toLowerCase() + const all = new Set() + const resolving = new Set() + const reference = + /(?:([A-Za-z0-9_.-]+\/[A-Za-z0-9_.-]+)#|#)(\d+)|https:\/\/github\.com\/([A-Za-z0-9_.-]+\/[A-Za-z0-9_.-]+)\/issues\/(\d+)/gi + const closing = + /\b(?:close(?:s|d)?|fix(?:es|ed)?|resolve(?:s|d)?)\s*:?\s+(?:(?:([A-Za-z0-9_.-]+\/[A-Za-z0-9_.-]+)#|#)(\d+)|https:\/\/github\.com\/([A-Za-z0-9_.-]+\/[A-Za-z0-9_.-]+)\/issues\/(\d+))/gi + + for (const match of source.matchAll(reference)) { + const explicit = (match[1] ?? match[3] ?? '').toLowerCase() + const number = Number(match[2] ?? match[4]) + if (!explicit || explicit === expected) all.add(number) + } + for (const match of source.matchAll(closing)) { + const explicit = (match[1] ?? match[3] ?? '').toLowerCase() + const number = Number(match[2] ?? match[4]) + if (!explicit || explicit === expected) { + all.add(number) + resolving.add(number) + } + } + return { + all: [...all].sort((left, right) => left - right), + resolving: [...resolving].sort((left, right) => left - right), + related: [...all].filter((number) => !resolving.has(number)).sort((a, b) => a - b), + } +} + +/** + * Validate one Issue with its Project status. + * @param {{title: string, body: string, assignees: string[], labels: string[], type: string|null, priority: string|null, status: string|null, state: string, stateReason: string|null}} issue Issue snapshot. + * @returns {string[]} Validation errors. + */ +export function validateIssue(issue) { + const errors = validateBody(issue) + const status = issue.status + + if (!/\p{Script=Han}/u.test(issue.title)) errors.push('Issue 标题必须包含中文') + if ( + /^\s*(?:\[(?:Idea|Feature|Bug|Research|Task|P[0-3]|Inbox|Backlog|Ready|In progress|In review|Done|No action|Owner|area\/[^\]]+)[^\]]*\]|(?:Idea|Feature|Bug|Research|Task|P[0-3]|Inbox|Backlog|Ready|In progress|In review|Done|No action|Owner|area\/[^:： ]+)\s*[:：-])/iu.test( + issue.title, + ) + ) { + errors.push('Issue 标题不得带 Type、Priority、Status、area 或 Owner 前缀') + } + if (!TYPES.has(issue.type ?? '')) errors.push('Type 必须是五种原生英文 Type 之一') + if (!status || !config.statuses.includes(status)) errors.push('Issue 必须在 Project 中且具有合法 Status') + if (issue.priority !== null && !PRIORITIES.includes(issue.priority.toLowerCase())) { + errors.push('Priority 必须为空或为 P0–P3') + } + if (status === 'Done' && (issue.state !== 'closed' || issue.stateReason !== 'completed')) { + errors.push('Done 必须对应 Completed 关闭原因') + } + if ( + status === 'No action' && + (issue.state !== 'closed' || issue.stateReason !== 'not_planned') + ) { + errors.push('No action 必须对应 Not planned 关闭原因') + } + if (!['Done', 'No action'].includes(status ?? '') && issue.state !== 'open') { + errors.push(`${status} 必须对应开放 Issue`) + } + return errors +} + +/** + * Validate PR metadata and its referenced Issues. + * @param {{authorType: string, labels: string[], references: ReturnType, issues: Map}} input PR snapshot. + * @returns {string[]} Validation errors. + */ +export function validatePullRequest(input) { + if (!requiresPullRequestPolicy(input)) return [] + const errors = [] + const kinds = input.labels.filter((label) => label.startsWith('kind/')) + const priorities = input.labels.filter((label) => PRIORITIES.includes(label)) + const areas = input.labels.filter((label) => label.startsWith('area/')) + + if (input.references.all.length === 0) errors.push('PR 正文必须引用至少一个同仓库 Issue') + if (kinds.length !== 1) errors.push(`PR 必须恰好有一个 kind/*，当前为 ${kinds.length}`) + if (priorities.length > 1) errors.push(`PR 最多有一个 p0–p3，当前为 ${priorities.length}`) + if (areas.length === 0) errors.push('PR 必须至少有一个 area/*') + for (const number of input.references.all) { + if (!input.issues.has(number)) errors.push(`#${number} 不是同仓库 Issue`) + } + + const resolving = input.references.resolving + .map((number) => [number, input.issues.get(number)]) + .filter((entry) => entry[1]) + if (resolving.length === 0) return errors + + const issuePriorities = resolving + .map(([, issue]) => issue.priority?.toLowerCase()) + .filter((priority) => PRIORITIES.includes(priority)) + if (priorities.length === 0 && issuePriorities.length > 0) { + const highest = issuePriorities.sort( + (left, right) => PRIORITIES.indexOf(left) - PRIORITIES.indexOf(right), + )[0] + errors.push(`PR Priority 应为 ${highest}`) + } else if (priorities.length === 1 && issuePriorities.length !== resolving.length) { + errors.push('有 Priority 的解决型 PR 要求每个被解决 Issue 都设置 Priority') + } else if (priorities.length === 1) { + const highest = issuePriorities.sort( + (left, right) => PRIORITIES.indexOf(left) - PRIORITIES.indexOf(right), + )[0] + if (priorities[0] !== highest) errors.push(`PR Priority 应为 ${highest}`) + } + return errors +} + +function token() { + const value = process.env.GH_TOKEN || process.env.GITHUB_TOKEN + if (!value) throw new Error('GH_TOKEN 或 GITHUB_TOKEN 未设置') + return value +} + +async function api(path, options = {}) { + const response = await fetch(`${process.env.GITHUB_API_URL ?? 'https://api.github.com'}${path}`, { + ...options, + headers: { + Accept: 'application/vnd.github+json', + Authorization: `Bearer ${token()}`, + 'X-GitHub-Api-Version': API_VERSION, + 'User-Agent': 'dsh-issue-policy', + ...options.headers, + }, + }) + if (options.allow404 && response.status === 404) return null + if (!response.ok) { + const body = await response.text() + throw new Error(`${options.method ?? 'GET'} ${path}: ${response.status} ${body}`) + } + if (response.status === 204) return null + return response.json() +} + +async function graphql(query, variables) { + const result = await api('/graphql', { + method: 'POST', + body: JSON.stringify({ query, variables }), + headers: { 'Content-Type': 'application/json' }, + }) + if (result.errors?.length) throw new Error(result.errors.map((error) => error.message).join('; ')) + return result.data +} + +async function issueSnapshot(number, status = undefined) { + const issue = await api(`/repos/${config.organization}/${config.repository}/issues/${number}`) + if (issue.pull_request) return null + const values = await api( + `/repos/${config.organization}/${config.repository}/issues/${number}/issue-field-values?per_page=100`, + ) + const field = (name) => values.find((value) => value.issue_field_name === name) + return { + number, + nodeId: issue.node_id, + title: issue.title, + body: issue.body ?? '', + assignees: issue.assignees.map((assignee) => assignee.login), + labels: issue.labels.map((label) => label.name), + type: issue.type?.name ?? null, + priority: field(config.priorityField)?.single_select_option?.name ?? null, + status: status === undefined ? await projectStatus(number) : status, + state: issue.state, + stateReason: issue.state_reason ?? null, + } +} + +async function projectContext(number) { + const data = await graphql( + `query($organization: String!, $repository: String!, $number: Int!, $project: Int!) { + organization(login: $organization) { + projectV2(number: $project) { + id + title + fields(first: 50) { + nodes { + ... on ProjectV2SingleSelectField { id name options { id name } } + } + } + } + } + repository(owner: $organization, name: $repository) { + issue(number: $number) { + id + projectItems(first: 20, includeArchived: true) { + nodes { + id + project { id } + fieldValueByName(name: "Status") { + ... on ProjectV2ItemFieldSingleSelectValue { name optionId } + } + } + } + } + } + }`, + { + organization: config.organization, + repository: config.repository, + number, + project: config.projectNumber, + }, + ) + const project = data.organization?.projectV2 + const issue = data.repository?.issue + if (!project || project.title !== config.projectTitle) throw new Error('目标 Project 不存在或标题不匹配') + if (!issue) throw new Error(`#${number} 不存在`) + const statusField = project.fields.nodes.find((field) => field?.name === 'Status') + if (!statusField) throw new Error('Project 缺少 Status 字段') + const item = issue.projectItems.nodes.find((candidate) => candidate.project.id === project.id) + return { project, issue, statusField, item } +} + +async function projectStatus(number) { + const context = await projectContext(number) + return context.item?.fieldValueByName?.name ?? null +} + +async function ensureProjectItem(number) { + const context = await projectContext(number) + if (context.item) return context + const data = await graphql( + `mutation($projectId: ID!, $contentId: ID!) { + addProjectV2ItemById(input: {projectId: $projectId, contentId: $contentId}) { + item { id } + } + }`, + { projectId: context.project.id, contentId: context.issue.id }, + ) + return { + ...context, + item: { id: data.addProjectV2ItemById.item.id, fieldValueByName: null }, + } +} + +async function setStatus(number, status) { + const context = await ensureProjectItem(number) + const option = context.statusField.options.find((candidate) => candidate.name === status) + if (!option) throw new Error(`Status 不存在：${status}`) + if (context.item.fieldValueByName?.name === status) return + await graphql( + `mutation($projectId: ID!, $itemId: ID!, $fieldId: ID!, $optionId: String!) { + updateProjectV2ItemFieldValue(input: { + projectId: $projectId, + itemId: $itemId, + fieldId: $fieldId, + value: {singleSelectOptionId: $optionId} + }) { projectV2Item { id } } + }`, + { + projectId: context.project.id, + itemId: context.item.id, + fieldId: context.statusField.id, + optionId: option.id, + }, + ) +} + +async function upsertAudit(number, errors) { + const comments = await api( + `/repos/${config.organization}/${config.repository}/issues/${number}/comments?per_page=100`, + ) + const existing = comments.find( + (comment) => comment.user?.type === 'Bot' && comment.body?.includes(AUDIT_MARKER), + ) + if (errors.length === 0) { + if (existing) { + await api(`/repos/${config.organization}/${config.repository}/issues/comments/${existing.id}`, { + method: 'DELETE', + }) + } + return + } + const body = `${AUDIT_MARKER}\n⚠️ Issue policy 未通过：\n\n${errors.map((error) => `- ${error}`).join('\n')}` + if (existing) { + if (existing.body === body) return + await api(`/repos/${config.organization}/${config.repository}/issues/comments/${existing.id}`, { + method: 'PATCH', + body: JSON.stringify({ body }), + headers: { 'Content-Type': 'application/json' }, + }) + } else { + await api(`/repos/${config.organization}/${config.repository}/issues/${number}/comments`, { + method: 'POST', + body: JSON.stringify({ body }), + headers: { 'Content-Type': 'application/json' }, + }) + } +} + +async function auditIssue(number, extraErrors = [], status = undefined) { + const issue = await issueSnapshot(number, status) + if (!issue) return [] + const errors = [...extraErrors, ...validateIssue(issue)] + await upsertAudit(number, errors) + return errors +} + +async function pullRequestSnapshot(number) { + const pull = await api(`/repos/${config.organization}/${config.repository}/pulls/${number}`) + const [reviewRequests, reviews] = await Promise.all([ + api(`/repos/${config.organization}/${config.repository}/pulls/${number}/requested_reviewers`), + api(`/repos/${config.organization}/${config.repository}/pulls/${number}/reviews?per_page=100`), + ]) + const references = parseReferences({ + body: pull.body ?? '', + repository: `${config.organization}/${config.repository}`, + }) + const issues = new Map() + for (const issueNumber of references.all) { + const issue = await issueSnapshot(issueNumber, null) + if (issue) issues.set(issueNumber, issue) + } + return { + number, + isDraft: pull.draft, + authorType: pull.user?.type ?? 'User', + reviewRequestCount: reviewRequests.users.length + reviewRequests.teams.length, + reviewCount: reviews.length, + labels: pull.labels.map((label) => label.name), + references, + issues, + } +} + +async function moveResolvingIssues(pull, from, to) { + for (const number of pull.references.resolving) { + const current = await issueSnapshot(number) + if (!current || current.status !== from) continue + await setStatus(number, to) + await auditIssue(number) + } +} + +async function runPullRequestCheck(event) { + const pull = await pullRequestSnapshot(event.pull_request.number) + const errors = validatePullRequest(pull) + if (errors.length > 0) { + for (const error of errors) process.stdout.write(`::error::${error}\n`) + throw new Error(`Issue policy 未通过，共 ${errors.length} 项`) + } + process.stdout.write( + requiresPullRequestPolicy(pull) ? 'Issue policy 通过。\n' : 'PR 尚未进入 Issue policy 强制范围。\n', + ) +} + +async function runLifecycle(eventName, event) { + if (eventName === 'issues') { + const number = event.issue.number + if (event.action === 'opened') await setStatus(number, 'Inbox') + if (event.action === 'closed') { + const target = event.issue.state_reason === 'not_planned' ? 'No action' : 'Done' + await setStatus(number, target) + } + if (event.action === 'reopened') { + await setStatus(number, 'Inbox') + } + await ensureProjectItem(number) + await auditIssue(number) + return + } + + if (eventName === 'pull_request' || eventName === 'pull_request_review') { + const pull = await pullRequestSnapshot(event.pull_request.number) + const errors = validatePullRequest(pull) + if (errors.length > 0) return + await moveResolvingIssues(pull, 'Ready', 'In progress') + if (pull.reviewRequestCount > 0 || pull.reviewCount > 0) { + await moveResolvingIssues(pull, 'In progress', 'In review') + } + } +} + +function readEvent() { + if (!process.env.GITHUB_EVENT_PATH) throw new Error('GITHUB_EVENT_PATH 未设置') + return JSON.parse(fs.readFileSync(process.env.GITHUB_EVENT_PATH, 'utf8')) +} + +async function main(argv) { + const [command] = argv + if (command === 'pr') await runPullRequestCheck(readEvent()) + else if (command === 'lifecycle') await runLifecycle(process.env.GITHUB_EVENT_NAME, readEvent()) + else throw new Error('用法：policy.mjs pr|lifecycle') +} + +if (process.argv[1] && import.meta.url === pathToFileURL(process.argv[1]).href) { + main(process.argv.slice(2)).catch((error) => { + process.stderr.write(`${error instanceof Error ? error.message : String(error)}\n`) + process.exitCode = 1 + }) +} diff --git a/.github/issue-management/policy.test.mjs b/.github/issue-management/policy.test.mjs new file mode 100644 index 0000000000..890247db29 --- /dev/null +++ b/.github/issue-management/policy.test.mjs @@ -0,0 +1,240 @@ +import assert from 'node:assert/strict' +import test from 'node:test' + +import { + countVisibleUnits, + parseReferences, + requiresPullRequestPolicy, + validateBody, + validateIssue, + validatePullRequest, +} from './policy.mjs' + +const withDetails = (summary) => + `${summary}\n\n
验收与细节待补充。
` + +const legalIssue = { + title: '完成议题管理校验', + body: withDetails('完成议题管理校验。'), + assignees: [], + labels: [], + type: 'Idea', + priority: null, + status: 'In review', + state: 'open', + stateReason: null, +} + +test('counts only text outside details', () => { + assert.deepEqual(countVisibleUnits('支持 GitHub Project。
隐藏文字
'), { + units: 4, + balanced: true, + detailsCount: 1, + allCollapsed: true, + }) +}) + +test('requires a balanced default-collapsed details region', () => { + assert.deepEqual(validateBody({ body: '完成工作。', assignees: [] }), [ + '正文必须包含默认收起的
区域', + ]) + assert.deepEqual( + validateBody({ + body: '完成工作。\n\n
细节待补充。
', + assignees: [], + }), + ['details 必须默认收起，不得设置 open'], + ) + assert.deepEqual( + validateBody({ body: '完成工作。\n\n
细节', assignees: [] }), + ['details 标签必须成对闭合'], + ) +}) + +test('requires Owner for multiple assignees', () => { + assert.deepEqual( + validateBody({ + body: withDetails('完成工作。'), + assignees: ['tianyicui', 'tianyicui-bot'], + }), + ['多个 Assignees 时首个非空行必须是 Owner: @login'], + ) +}) + +test('accepts an intended Owner while assignment permission is pending', () => { + assert.deepEqual( + validateBody({ + body: withDetails('Owner: @octocat\n\n完成工作。'), + assignees: [], + }), + [], + ) + assert.deepEqual( + validateBody({ + body: withDetails('Owner: @octocat\n\n完成工作。'), + assignees: ['hubot'], + }), + ['零或一个 Assignee 时不得写 Owner 行'], + ) +}) + +test('allows optional metadata in every open Status', () => { + assert.deepEqual(validateIssue(legalIssue), []) + for (const status of ['Inbox', 'Backlog', 'Ready', 'In progress', 'In review']) { + assert.deepEqual(validateIssue({ ...legalIssue, status }), []) + } +}) + +test('rejects metadata prefixes in an Issue title', () => { + const errors = validateIssue({ ...legalIssue, title: '[Bug] 修复恢复错误' }) + assert.ok(errors.includes('Issue 标题不得带 Type、Priority、Status、area 或 Owner 前缀')) +}) + +test('keeps terminal Status aligned with the native close reason', () => { + assert.deepEqual( + validateIssue({ ...legalIssue, status: 'Done', state: 'closed', stateReason: 'completed' }), + [], + ) + assert.deepEqual( + validateIssue({ + ...legalIssue, + status: 'No action', + state: 'closed', + stateReason: 'not_planned', + }), + [], + ) + assert.ok(validateIssue({ ...legalIssue, status: 'Done' }).includes('Done 必须对应 Completed 关闭原因')) +}) + +test('separates resolving and informational references', () => { + assert.deepEqual( + parseReferences({ + body: 'Fixes #12\nRelated to #4\nRefs deepseekharness/dsh-test#7', + repository: 'deepseekharness/dsh-test', + }), + { all: [4, 7, 12], resolving: [12], related: [4, 7] }, + ) +}) + +test('allows informational references without cross-object constraints', () => { + const errors = validatePullRequest({ + isDraft: false, + authorType: 'User', + reviewRequestCount: 1, + reviewCount: 0, + labels: ['kind/cleanup', 'area/infra'], + references: { all: [4], resolving: [], related: [4] }, + issues: new Map([[4, { type: 'Bug', priority: 'P0', labels: ['area/web'] }]]), + }) + assert.deepEqual(errors, []) +}) + +test('enforces highest resolving Priority without Type or area synchronization', () => { + const pull = { + isDraft: false, + authorType: 'User', + reviewRequestCount: 0, + reviewCount: 1, + labels: ['kind/cleanup', 'p0', 'area/web'], + references: { all: [2, 3], resolving: [2, 3], related: [] }, + issues: new Map([ + [2, { type: 'Feature', priority: 'P2', labels: ['area/web'] }], + [3, { type: 'Bug', priority: 'P0', labels: ['area/session'] }], + ]), + } + assert.deepEqual(validatePullRequest(pull), []) + assert.ok( + validatePullRequest({ ...pull, labels: ['kind/cleanup', 'p2', 'area/web'] }).includes( + 'PR Priority 应为 p0', + ), + ) +}) + +test('requires policy only after a human PR enters review', () => { + assert.equal( + requiresPullRequestPolicy({ + isDraft: false, + authorType: 'User', + reviewRequestCount: 1, + reviewCount: 0, + }), + true, + ) + assert.equal( + requiresPullRequestPolicy({ + isDraft: false, + authorType: 'User', + reviewRequestCount: 0, + reviewCount: 0, + }), + false, + ) +}) + +test('exempts Draft, Bot, and App PRs', () => { + const invalid = { + isDraft: false, + labels: [], + references: { all: [], resolving: [], related: [] }, + issues: new Map(), + reviewRequestCount: 1, + reviewCount: 0, + } + assert.deepEqual(validatePullRequest({ ...invalid, authorType: 'Bot' }), []) + assert.deepEqual(validatePullRequest({ ...invalid, authorType: 'App' }), []) + assert.deepEqual(validatePullRequest({ ...invalid, authorType: 'User', isDraft: true }), []) + assert.ok(validatePullRequest({ ...invalid, authorType: 'User' }).length > 0) +}) + +test('requires repository PR labels in the enforcement scope', () => { + const errors = validatePullRequest({ + isDraft: false, + authorType: 'User', + reviewRequestCount: 1, + reviewCount: 0, + labels: [], + references: { all: [2], resolving: [], related: [2] }, + issues: new Map([[2, { priority: null }]]), + }) + assert.ok(errors.includes('PR 必须恰好有一个 kind/*，当前为 0')) + assert.ok(errors.includes('PR 必须至少有一个 area/*')) +}) + +test('accepts repository-extensible kind labels', () => { + assert.deepEqual( + validatePullRequest({ + isDraft: false, + authorType: 'User', + reviewRequestCount: 1, + reviewCount: 0, + labels: ['kind/dependency', 'area/infra'], + references: { all: [2], resolving: [], related: [2] }, + issues: new Map([[2, { priority: null }]]), + }), + [], + ) +}) + +test('allows missing Priority only when resolving Issues are also unprioritized', () => { + const pull = { + isDraft: false, + authorType: 'User', + reviewRequestCount: 1, + reviewCount: 0, + labels: ['kind/feature', 'area/web'], + references: { all: [2], resolving: [2], related: [] }, + issues: new Map([[2, { priority: null }]]), + } + assert.deepEqual(validatePullRequest(pull), []) + assert.ok( + validatePullRequest({ ...pull, issues: new Map([[2, { priority: 'P2' }]]) }).includes( + 'PR Priority 应为 p2', + ), + ) + assert.ok( + validatePullRequest({ ...pull, labels: [...pull.labels, 'p2'] }).includes( + '有 Priority 的解决型 PR 要求每个被解决 Issue 都设置 Priority', + ), + ) +}) diff --git a/.github/pull_request_template.md b/.github/pull_request_template.md new file mode 100644 index 0000000000..e960016ae8 --- /dev/null +++ b/.github/pull_request_template.md @@ -0,0 +1,13 @@ + + + + +关联 Issue： + +
+变更与验证 + +- 变更： +- 验证： + +