Merge cloud-owned teaching policy and bounded new shortcuts
This commit is contained in:
鲨鱼辣椒
@@ -0,0 +1,62 @@
|
||||
# Task: Keep teacher runtime instructions scoped to entry facts and UI protocol
|
||||
|
||||
## Identity
|
||||
|
||||
- Task ID: 20260928-teacher-protocol-boundary-f2c8a601
|
||||
- Mode: Feature
|
||||
- Branch: codex/20260928-teacher-protocol-boundary-f2c8a601-teacher-protocol-boundary
|
||||
- Worktree: /Users/chillishark/Makelore 麦洛/.codex-worktrees/makelore/20260928-teacher-protocol-boundary-f2c8a601
|
||||
- Base commit: 998796d77cba146ae2a007cbebfdaa1255a3413e
|
||||
- Owner: codex
|
||||
- Status: Ready for Integration
|
||||
|
||||
## Scope
|
||||
|
||||
- Continue the accepted teacher/client boundary on top of completed legacy-component cleanup commit `998796d77cba146ae2a007cbebfdaa1255a3413e`.
|
||||
- Keep entry markers factual and the reply protocol technical. Remove locally appended teaching strategy from active help, guided help, background checks, and reply-format instructions.
|
||||
- Enforce the current 0–3 shortcut capacity for newly generated answers while preserving archived replies and exact diagnostics.
|
||||
- Update focused tests and README; preserve card styling and existing interaction/scheduling machinery.
|
||||
|
||||
## Intent And Constraints
|
||||
|
||||
- User assigned main merge and push to another conversation. This task stays in its isolated feature branch and prepares a tested commit for integration.
|
||||
- Cloud configuration owns Alice's teaching purpose, focus, tone, length, and shortcut prefixes; the host owns invocation facts, available context/tools, and interface protocol.
|
||||
- Keep old component cleanup intact. Do not merge the older `89c913b` patch wholesale because it modifies a superseded protocol.
|
||||
- No cloud configuration publishing, live app restart, scheduler redesign, or teacher-to-operator consensus handoff in this task.
|
||||
- Gate passed before edits: task ownership/status verified; entry, task-relevant memory, and peer scopes reviewed. Relevant canonical documents are unchanged between the previously read baseline and `998796d`; shared memory remains read-only.
|
||||
|
||||
## Outcome
|
||||
|
||||
- Active-help and guided-help messages now identify the student's entry intent; background checks identify a program-triggered check without a fabricated student question. They no longer mandate a teaching strategy.
|
||||
- Reply instructions now declare only the two fields, Markdown body support, optional 0–3 nonempty shortcut strings, and delegation of content/style to published configuration.
|
||||
- Main rejects excess newly generated shortcuts without silently truncating or rewriting them: prose is retained, cards are withheld, and the exact raw answer and format diagnostic are saved. 0/1/3-card replies preserve prefixes and duplicates. The shared compatibility parser and historical store remain unchanged.
|
||||
- The UI now shows recorded reply-format errors beside the preserved answer; raw content stays collapsed and inert. Existing colors, layout, click handling and drafts are unchanged.
|
||||
- Regression coverage includes local/cloud service persistence, history read/list/save, all entry intents, current/legacy envelopes, and UI behavior. README reflects the responsibility boundary and current capacity.
|
||||
- No merge, push, running-app restart, cloud config change, or canonical memory edit performed.
|
||||
|
||||
## Verification
|
||||
|
||||
- Dependency installation: pinned pnpm 10.33.4, frozen lockfile, offline cache; no tracked dependency change.
|
||||
- Focused final suite: 269/269 passed across teacher-guidance, teacher-reply, teacher-reply-history, coding-teacher, coding-teacher-cloud, coding-teacher-model and coding-teacher-ui.
|
||||
- `corepack pnpm run typecheck`: passed.
|
||||
- ESLint for all changed TypeScript/TSX files, including the Electron E2E spec: passed.
|
||||
- `corepack pnpm run build:vite`: passed after the final UI change.
|
||||
- `git diff --check` and task-aware project-doc drift check: passed.
|
||||
- Independent read-only review found no remaining teaching-policy directives in the scoped runtime files and confirmed lossless historical handling.
|
||||
- Added a focused Electron E2E case for visible format errors, absent rejected shortcut buttons, collapsed exact raw text, and preserved drafts. Playwright test collection passed using an existing Electron distribution override; the native test was not executed to avoid launching another app window during concurrent work. Component/service assertions covering these behaviors passed.
|
||||
- Real published-model responses were not tested. Existing topics retain their pinned configuration versions; code tests do not establish that an updated cloud prompt is active.
|
||||
|
||||
## Follow-ups
|
||||
|
||||
- Integrate this commit after the legacy-component cleanup; merging/pushing belongs to the user's other conversation.
|
||||
- Validate actual Alice output with a new topic using the published cloud version. Local tests cannot prove model behavior or that existing pinned topics picked up a new prompt.
|
||||
- Three-operation-round checks and shared confirmed context remain separate work.
|
||||
|
||||
## Promotion Candidates
|
||||
|
||||
- Target: canonical teacher decision/current-state documents during integration.
|
||||
- Proposal: runtime help/check-in markers convey entry facts only; teaching strategy is entirely in the selected published agent configuration. Current newly generated replies accept 0–3 shortcuts; historical card counts remain lossless.
|
||||
- Evidence: context/reply code and focused tests in this task; user's accepted client/agent boundary in this conversation.
|
||||
- Future impact: future personas can use the same interface without inheriting Alice's teaching policy; prompt changes are evaluated against the actual published topic version.
|
||||
- Semantic conflicts: supersedes cleanup task's remaining locally mandated single discussion focus and unrestricted new-card count; retains historical compatibility and UI cleanup.
|
||||
- Human confirmation: not additionally required for these already accepted boundaries; no canonical edits in this feature task.
|
||||
@@ -174,10 +174,10 @@ Pi 正式包必须继续运行 `pnpm run verify:artifact:pi`、`pnpm run smoke:p
|
||||
- 主动发言采用紧贴小头像的短气泡；只有存在真实主动消息或运营欢迎语时才显示这组浮层，收起消息时头像和气泡一起消失，顶部入口仍保留。长消息在气泡中最多显示三行，点击接回原智能体对话查看全文。不提供“智能体偶尔来看看”开关或本地模拟巡看控制，既有自动跟进与真实消息接收逻辑保留。作品原生预览对整组气泡和头像测量避让，避免图片被原生页面遮住。
|
||||
- 新的单会话入口停止旧版前端定时跟进派发；项目主动观察由独立功能衔接，不能重新创建可见话题。未下发目录的旧版入口仍保留原跟进合同：前端每 5 分钟请求一次智能体跟进；窗口隐藏、来源归档、主对话正在执行、智能体正在回复或学生正在智能体栏写草稿时延后。Main 再检查在线启停、来源归属、项目级冷却和已完成文本指纹，未变上下文按下述十五分钟冷却处理。跟进沿用运营模型、已发布 Skills 与当前智能体话题，生成符合所选配置的具体建议或引导；以主动智能体发言持久化，不伪造学生消息。模型调用沿用智能体计费规则。
|
||||
- 进入项目时，智能体头像旁先显示运营发布的欢迎语气泡，不调用模型、不声称已检查项目；每个账号／项目主动收起后不重复弹出。真实的未读主动建议优先替换欢迎语。主动发言在头像旁直接显示气泡，不抢焦点、不盖住作品；作品区按气泡与原生预览的实际交叠范围避让，已有工具栏高度计入计算，气泡位于独立智能体栏时不额外留白，窗口／分栏尺寸变化和收起气泡时实时归还空间；“和智能体聊聊”接回同一段咨询，“等会儿聊”收起并保留历史。已有未处理气泡时不堆叠新邀请；咨询已展开时也显示气泡，直到学生主动点击查看或收起，期间的其他回复不会吞掉未读邀请。五分钟检查计时在刷新和项目切换后保留；智能体咨询里的新讨论也算进展，即使操作对话没有变化也会继续跟进；上下文完全未变时放缓到十五分钟一次，不永久停止。事件连接丢失时，每十五秒只读同步未完成回复，收到结果后停止轮询，不重复调用模型。失败如实提示，不以本地文案替代模型回复。已接受请求恢复使用原身份，退出账号和来源删除仍沿用 Main 中止边界。
|
||||
- 智能体支持自由提问；输入框为空时，底部显示暖黄色小按钮“继续看看👀”，输入文字后收起，清空后重现，发送按钮保持原位。只有学生点击才调用模型，结合当前操作对话与咨询历史推荐一个具体切入点，并生成围绕它的快捷回复。点击快捷回复会主动开始讨论；回复下方不再附加固定的二次求助链接；失败或中断后，可点暖黄色小按钮“继续看看👀”重试。这些快捷求助保留已有输入草稿，网络结果不确定时复用请求身份，已确认终态的请求不重复执行；解析失败可重新求助，不用固定问题伪装模型结果。服务端下发的推荐问题只填入草稿，由用户确认发送。界面不提供独立示范页面、“记一下”、共识或自动待办，智能体的回复下不再展示“我去试一试”和“复制”按钮，学生通过已有的操作对话／作品入口继续创作。
|
||||
- 智能体支持自由提问；输入框为空时，底部显示暖黄色小按钮“继续看看👀”，输入文字后收起，清空后重现，发送按钮保持原位。只有学生点击才调用模型，程序说明学生主动求助且尚未提出具体问题，智能体结合可用上下文按云端配置回应。点击快捷回复会主动开始讨论；回复下方不再附加固定的二次求助链接；失败或中断后，可点暖黄色小按钮“继续看看👀”重试。这些快捷求助保留已有输入草稿，网络结果不确定时复用请求身份，已确认终态的请求不重复执行；解析失败可重新求助，不用固定问题伪装模型结果。服务端下发的推荐问题只填入草稿，由用户确认发送。界面不提供独立示范页面、“记一下”、共识或自动待办，智能体的回复下不再展示“我去试一试”和“复制”按钮，学生通过已有的操作对话／作品入口继续创作。
|
||||
- 老师咨询只展示老师正文和可点击的快捷回复卡片，学生也可在“和老师聊聊”输入框自由输入。点击卡片原样发送该回复，保留当前输入草稿；旧话题的数据不会恢复想法板、结构图、流程图或对照表及其操作。
|
||||
- “继续看看👀”由老师结合项目推荐一个具体切入点，快捷回复围绕这个切入点帮助学生接话；客户端不要求多个独立话题，也不固定正文长度、选项数量、前缀或表达风格。老师继续依据已配置职责帮助孩子整理想法、理解关系和承接已确认的共识，区分建议与已确认内容。
|
||||
- Main 对咨询回复统一使用 `{reply, quickReplies}`，本地读取工具的前言和云端非最终片段不当作最终回答；完整解析后展示正文和卡片，主动进展提醒仍直接展示正文。普通 Markdown、JSON 数据和代码示例保持正文。兼容读取旧 `{intro, questions}` 和带 `tool` 的回复，只提取正文与快捷回复，不恢复或更新组件。资源上限只用于防护，不做旧式字数截断；格式错误保留可读正文及原始回答，原文默认折叠、字面显示，不再次送进模型。
|
||||
- 客户端只提供入口事实、可用上下文和界面协议。讨论切入点、教学方式、是否提问、正文长度、前缀和表达风格由所选智能体的云端配置决定；不在“继续看看👀”、引导开口或后台检查时附加教学策略。当前新回复允许 0–3 条快捷回复，没有时用空数组，不要求凑满。
|
||||
- Main 对咨询回复统一使用 `{reply, quickReplies}`，本地读取工具的前言和云端非最终片段不当作最终回答；完整解析后展示正文和卡片，主动进展提醒仍直接展示正文。普通 Markdown、JSON 数据和代码示例保持正文。兼容读取旧 `{intro, questions}` 和带 `tool` 的回复，只提取正文与快捷回复，不恢复或更新组件。新输出超过 3 条快捷回复时保留完整正文、提示格式错误并保留原始回答，不展示超额卡片或静默截取；历史中已有的更多卡片在读取与后续保存时仍原样保留。独立资源上限只用于防护，不做旧式字数截断；格式错误的原文默认折叠、字面显示，不再次送进模型。
|
||||
- 回复中的引号或换行格式有误时，仅在正文边界明确、修复后整体合法的情况下恢复完整文字，不把半句话当成完整回答。附加建议损坏时仍展示完整正文；正文无法确认完整时显示“这次回复未能完整显示”，由学生点击“重新回答”，沿用原问题、引用、来源、话题和发布版本，保留当前草稿，连续点击不重复发送。打开旧话题会从保留的原文中在本地恢复可确认的正文，不调用模型，不因显示修复而改写磁盘记录；原始内容继续折叠保留。不完整的回答和建议不进入后续模型上下文。
|
||||
- 咨询正文支持 Markdown 标题、列表、表格、代码围栏、HTTP(S) 链接／图片和数学公式，长代码与表格在栏内横向滚动，不执行 HTML。云端正文维持 string 合同，未知对象不会猜测转成回答。工具活动只从云端主线程的类型化工具事件及 Main 本地读取过程获得，按本次问题／工具调用身份合并名称和状态，跨云端暂停、续接仍只计一次；暂停读取不算失败，状态以实际读取结果或问题终态为准，单独折叠显示；参数、结果原文和内部错误不混入回复，思考内容不作为正文。客户端仅保存自己实际收到的工具状态，不能补回旧历史或断线期间已经过期的事件。
|
||||
- 云端咨询在 local_context 声明 read_protocol=2；需先部署配套 Yuxi API 和 worker，再升级客户端。云端持久累计读取字节及批次，每轮告知模型剩余额度；预算耗尽后消费最后一批结果，并以 tool_choice=none 要求根据现有证据形成答案和说明缺口。Main 限制实际返回量并拒绝第十三批读取，区分读取达到上限、上下文失效及格式无效。服务端对未声明协议的已安装旧客户端保留原工具参数与六批边界。
|
||||
|
||||
@@ -113,7 +113,7 @@ export function compileTeacherContext(
|
||||
const current: TeacherModelMessage = {
|
||||
role: intent === 'check-in' ? 'system' : 'user',
|
||||
content: intent === 'check-in'
|
||||
? '本轮是一次项目进展提醒，不是用户提问。按照已配置的人设和职责，结合来源操作对话中已完成的文字和咨询历史回应，表达方式遵循已配置的要求，不要求用户立即回答。仅依据已有证据，不重复上次提醒，不声称实际运行或试玩过作品。'
|
||||
? '本轮入口：程序触发的项目进展检查，学生没有在本轮主动提问。回应方式遵循当前智能体的云端配置。'
|
||||
: [
|
||||
...references.map(
|
||||
(ref) =>
|
||||
@@ -123,10 +123,10 @@ export function compileTeacherContext(
|
||||
ref.text
|
||||
),
|
||||
'当前问题：\n' + question,
|
||||
...(intent === 'suggestions' || intent === 'guided-help'
|
||||
? [
|
||||
'本轮学生希望你帮忙找到交流起点。依据当前项目、来源操作对话和咨询历史，推荐一个具体切入点，并说明为何值得从这里聊；快捷回复围绕这个切入点帮助学生接话。没有可用上下文时，坦诚从构思切入，不假定学生已完成功能，不编造作品表现或项目进展。',
|
||||
]
|
||||
...(intent === 'suggestions'
|
||||
? ['本轮入口：学生通过“帮我看看”主动请求帮助，尚未提出具体问题。']
|
||||
: intent === 'guided-help'
|
||||
? ['本轮入口：学生主动求助，表示暂时说不清想问什么。']
|
||||
: []),
|
||||
].join('\n\n'),
|
||||
};
|
||||
|
||||
@@ -1,13 +1,21 @@
|
||||
import type { TeacherRequest } from '../../shared/coding-teacher';
|
||||
import { parseTeacherReply } from '../../shared/teacher-reply';
|
||||
|
||||
const MAX_CURRENT_QUICK_REPLIES = 3;
|
||||
|
||||
export function teacherReplyInstructions(): string {
|
||||
return '本轮界面接收一个 JSON 对象 {"reply":string,"quickReplies":string[]}。reply 是回复正文，可使用 Markdown；quickReplies 是可选的后续回复建议，没有建议时用空数组。只使用这两个字段。字符串中的英文双引号、反斜杠和换行必须按 JSON 规则转义，确保整个对象是合法 JSON。依据已配置职责帮助学生整理想法、理解关系、承接已确认的共识；区分学生已确认内容与智能体建议、待定想法。回复的语言、风格、长度、前缀和建议内容遵循当前智能体的云端配置。';
|
||||
return `本轮界面接收一个 JSON 对象 {"reply":string,"quickReplies":string[]}。reply 是回复正文，可使用 Markdown；quickReplies 是可选的快捷回复，最多 ${MAX_CURRENT_QUICK_REPLIES} 条，每条为非空字符串，没有快捷回复时用空数组。只使用这两个字段。字符串中的英文双引号、反斜杠和换行必须按 JSON 规则转义，确保整个对象是合法 JSON。回复的内容、语言、风格、长度和前缀遵循当前智能体的云端配置。`;
|
||||
}
|
||||
|
||||
/** The original response remains available for diagnosis when parsing is incomplete. */
|
||||
export function applyTeacherReply(request: TeacherRequest, raw: string): void {
|
||||
const parsed = parseTeacherReply(raw);
|
||||
// Apply the current UI contract only to new replies; archived replies keep
|
||||
// their original cards through the compatibility parser and topic store.
|
||||
if (parsed.quickReplies.length > MAX_CURRENT_QUICK_REPLIES) {
|
||||
parsed.quickReplies = [];
|
||||
parsed.parseError ??= '快捷回复超过当前界面最多 3 条的限制，正文已保留，快捷回复未展示。';
|
||||
}
|
||||
request.response = parsed.reply;
|
||||
request.suggestedQuestions = parsed.quickReplies;
|
||||
if (parsed.incomplete) request.replyIncomplete = true;
|
||||
|
||||
@@ -538,7 +538,7 @@ export function TeacherChatPanel({
|
||||
{request.intent !== 'suggestions' && request.status === 'completed' && !request.replyIncomplete && Boolean(request.suggestedQuestions?.length) && <div className="consultation-follow-ups" aria-label="接着聊">
|
||||
{request.suggestedQuestions?.map((question, index) => <button type="button" key={`${request.id}:${index}`} disabled={helpUnavailable} onClick={() => void send({ text: question })} className="consultation-follow-up"><span className="min-w-0 flex-1">{question}</span><ChevronRight className="h-4 w-4 shrink-0 opacity-70" aria-hidden="true" /></button>)}
|
||||
</div>}
|
||||
{request.status === 'completed' && !request.replyIncomplete && request.replyParseError && <p className="text-xs text-muted-foreground">回答正文已保留，部分附加内容未能显示。</p>}
|
||||
{request.status === 'completed' && !request.replyIncomplete && request.replyParseError && <p role="status" className="text-xs text-muted-foreground">回答正文已保留，部分附加内容未能显示。</p>}
|
||||
{request.unparsedResponse && <details className="min-w-0 text-xs text-muted-foreground">
|
||||
<summary className="cursor-pointer">查看收到的原始内容</summary>
|
||||
<pre className="mt-2 max-h-64 overflow-auto whitespace-pre-wrap break-words rounded-lg border p-3">{request.unparsedResponse}</pre>
|
||||
|
||||
@@ -21,6 +21,13 @@ type HostConnection = {
|
||||
token: string;
|
||||
};
|
||||
|
||||
type ConsultationReply = {
|
||||
response: string;
|
||||
suggestedQuestions: string[];
|
||||
replyParseError?: string;
|
||||
unparsedResponse?: string;
|
||||
};
|
||||
|
||||
type TeacherCatalogFixture = {
|
||||
nextTeacherReplyIncomplete?: boolean;
|
||||
teacherCatalog: TeacherCatalog;
|
||||
@@ -97,9 +104,10 @@ async function installCodingFirstChatHost(
|
||||
managedCapabilities = false,
|
||||
removedModel = false,
|
||||
audioPreview?: { executionId: string; path: string; dataUrl: string },
|
||||
consultationReply?: ConsultationReply,
|
||||
): Promise<void> {
|
||||
await electronApp.evaluate(async (_, payload) => {
|
||||
const { connection, featureComplete, managedCapabilities, removedModel, audioPreview } = payload;
|
||||
const { connection, featureComplete, managedCapabilities, removedModel, audioPreview, consultationReply } = payload;
|
||||
const { ipcMain } = process.mainModule!.require('electron') as typeof import('electron');
|
||||
type MainState = TeacherCatalogFixture & {
|
||||
captured: CapturedRequest[];
|
||||
@@ -575,6 +583,13 @@ async function installCodingFirstChatHost(
|
||||
const suggestions = body!.intent === 'suggestions';
|
||||
const response = suggestions ? '我们可以从你最近试过的地方聊起。' : body!.intent === 'guided-help' ? '你最近做的哪一步，让你停下来想了一会儿？' : '先理解状态如何随点击变化，再修改代码。';
|
||||
consultationTopics[agentId] = {...current,revision:Number(current.revision)+1,requests:[...(current.requests as unknown[]),{id:body!.requestId,projectId:body!.projectId,sourceConversationId:body!.sourceConversationId,text:body!.text,intent:body!.intent,references:body!.references??[],createdAt:new Date().toISOString(),sourceCursor:{workerGeneration:1,seq:1},sourceCapturedAt:now,includedSourceMessageIds:[],omittedMessages:0,truncatedMessages:1,status:'completed',response,...(suggestions?{suggestedQuestions:['怎样观察别人玩游戏？','我该先试哪个想法？']}:{})}]};
|
||||
const last = (consultationTopics[agentId].requests as Record<string, unknown>[]).at(-1)!;
|
||||
Object.assign(last, consultationReply);
|
||||
if (state.nextTeacherReplyIncomplete) {
|
||||
state.nextTeacherReplyIncomplete = false;
|
||||
Object.assign(last, { response: '它早就不是', replyIncomplete: true, replyParseError: '格式不完整',
|
||||
unparsedResponse: '{"reply":"它早就不是"刚搭好架子' });
|
||||
}
|
||||
return respond(consultationTopics[agentId],202);
|
||||
}
|
||||
}
|
||||
@@ -843,7 +858,7 @@ async function installCodingFirstChatHost(
|
||||
if (path === '/api/coding/runtime/diagnostics') return respond({ runtime: { revision: { provider: 1, resources: 1 }, workers: [{ conversationId: conversation.id, generation: 1, state: 'running', stage: 'running' }] } });
|
||||
return respond({ success: false, error: `Unhandled E2E route: ${method} ${path}` }, 404);
|
||||
});
|
||||
}, { connection: hostConnection, featureComplete, managedCapabilities, removedModel, audioPreview });
|
||||
}, { connection: hostConnection, featureComplete, managedCapabilities, removedModel, audioPreview, consultationReply });
|
||||
}
|
||||
|
||||
test('saved Game Audio offers click-only playable local preview in Electron', async ({ launchElectronApp }, testInfo) => {
|
||||
@@ -1965,6 +1980,57 @@ test('incomplete consultation replies retry only on click and preserve both draf
|
||||
} finally { await releaseSnapshot(app); }
|
||||
});
|
||||
|
||||
test('project consultations show reply protocol errors without exposing rejected shortcuts', async ({ launchElectronApp }) => {
|
||||
const electronApp = await launchElectronApp({ skipSetup: true });
|
||||
let page = await getStableWindow(electronApp);
|
||||
const connection = await page.evaluate(async () => ({ token: await window.electron.ipcRenderer.invoke('hostapi:token') as string, baseUrl: await window.electron.ipcRenderer.invoke('hostapi:base-url') as string }));
|
||||
const response = '可以先观察玩家在哪里停下来，再聊聊你想改的地方。';
|
||||
const quickReplies = ['看懂规则｜怎样观察玩家？', '想想作用｜我想聊聊按钮。', '试试结果｜怎样看出变化？', '<button>原始选项</button> **不要渲染**'];
|
||||
const replyParseError = '快捷回复超过当前界面最多 3 条的限制，正文已保留，快捷回复未展示。';
|
||||
const unparsedResponse = '\n' + JSON.stringify({ reply: response, quickReplies }, null, 2) + '\n';
|
||||
await installCodingFirstChatHost(electronApp, connection, true, false, false, undefined, {
|
||||
response, suggestedQuestions: [], replyParseError, unparsedResponse,
|
||||
});
|
||||
await settleSnapshot(electronApp);
|
||||
await disableCodingEventSource(page);
|
||||
try {
|
||||
await page.reload(); page = await getStableWindow(electronApp);
|
||||
await page.getByTestId('ai-module-option-programming').click();
|
||||
await page.evaluate(() => { window.location.hash = '/chat'; });
|
||||
await expect(page.getByTestId('project-conversations')).toBeVisible();
|
||||
const composer = page.getByTestId('coding-message-composer').getByRole('textbox', { includeHidden: true });
|
||||
await composer.fill('保留操作草稿');
|
||||
await page.getByRole('button', { name: '与朋友聊天', exact: true }).click();
|
||||
const teacher = page.getByTestId('teacher-chat-panel');
|
||||
const teacherComposer = teacher.getByRole('textbox', { name: '向智能体提问' });
|
||||
await teacherComposer.fill('还没说完的困惑');
|
||||
await teacher.getByRole('button', { name: '提问', exact: true }).click();
|
||||
const reply = teacher.locator('.consultation-messages > div').filter({ has: page.getByText(response, { exact: true }) });
|
||||
await expect(reply.getByText(response, { exact: true })).toBeVisible();
|
||||
await expect(reply.getByRole('status')).toHaveText('回答正文已保留，部分附加内容未能显示。');
|
||||
await expect(reply.getByRole('status')).toBeVisible();
|
||||
await expect(teacher.locator('.consultation-follow-up')).toHaveCount(0);
|
||||
for (const question of quickReplies) {
|
||||
await expect(teacher.getByRole('button', { name: question, exact: true })).toHaveCount(0);
|
||||
}
|
||||
const raw = reply.locator('details').filter({ has: page.getByText('查看收到的原始内容', { exact: true }) });
|
||||
await expect(raw).not.toHaveAttribute('open');
|
||||
await expect(raw.locator('pre')).toBeHidden();
|
||||
await raw.locator('summary').click();
|
||||
await expect(raw.locator('pre')).toBeVisible();
|
||||
expect(await raw.locator('pre').textContent()).toBe(unparsedResponse);
|
||||
await expect(raw.locator('button, strong')).toHaveCount(0);
|
||||
await expect(teacherComposer).toHaveValue('');
|
||||
await expect(composer).toHaveValue('保留操作草稿');
|
||||
const requests = (await readState(electronApp)).captured;
|
||||
expect(requests.filter(item => item.path.endsWith('/messages') && item.method === 'POST')).toHaveLength(1);
|
||||
expect(requests.filter(item => item.path.endsWith('/prompt') && item.method === 'POST')).toHaveLength(0);
|
||||
} finally {
|
||||
await releaseSnapshot(electronApp);
|
||||
}
|
||||
});
|
||||
|
||||
|
||||
test('manual agent refresh updates the catalog and retries failures without replacing the ongoing chat or drafts', async ({ launchElectronApp }) => {
|
||||
const electronApp = await launchElectronApp({ skipSetup: true });
|
||||
let page = await getStableWindow(electronApp);
|
||||
|
||||
@@ -418,8 +418,8 @@ it('submits a nonempty proactive check-in without inventing a user message', asy
|
||||
f.access.source.messages = [];
|
||||
const { body, compiled } = await submitCompiledContext(f, 'check-in');
|
||||
expect(compiled.messages.map(message => message.role)).toEqual(['system', 'system']);
|
||||
expect(compiled.messages.at(-1)?.content).toContain('本轮是一次项目进展提醒，不是用户提问');
|
||||
expect(body.query).toContain('仅依据已有证据');
|
||||
expect(compiled.messages.at(-1)?.content).toContain('程序触发的项目进展检查，学生没有在本轮主动提问');
|
||||
expect(body.query).toContain('回应方式遵循当前智能体的云端配置');
|
||||
});
|
||||
|
||||
it('gives a server-defined friend the same scoped read tools', async () => {
|
||||
|
||||
@@ -277,6 +277,29 @@ describe('teacher side chat', () => {
|
||||
expect(api.send).not.toHaveBeenCalled();
|
||||
});
|
||||
|
||||
it('shows the format error alongside the preserved reply while keeping excess shortcuts noninteractive', async () => {
|
||||
const response = '先看看摸头会带来什么变化。';
|
||||
const quickReplies = ['看懂规则｜摸头会改变什么？', '想想作用｜我们聊聊摸头。', '试试结果｜怎样看出变化？', '换个话题｜我想聊别的地方。'];
|
||||
const replyParseError = '快捷回复超过当前界面最多 3 条的限制，正文已保留，快捷回复未展示。';
|
||||
const raw = '\n' + JSON.stringify({ reply: response, quickReplies }, null, 2) + '\n';
|
||||
api.read.mockResolvedValue({ ...first, requests: [suggestionsRequest({
|
||||
presentation: 'reply-v1', response, suggestedQuestions: [], replyParseError, unparsedResponse: raw,
|
||||
})] });
|
||||
render(<TeacherChatPanel projectId="p" sourceId="c" />);
|
||||
await ready();
|
||||
|
||||
expect(screen.getByText(response)).toBeVisible();
|
||||
expect(screen.getByText('回答正文已保留，部分附加内容未能显示。')).toBeVisible();
|
||||
for (const shortcut of quickReplies) {
|
||||
expect(screen.queryByRole('button', { name: shortcut, exact: true })).not.toBeInTheDocument();
|
||||
}
|
||||
const details = screen.getByText('查看收到的原始内容').closest('details');
|
||||
expect(details).not.toHaveAttribute('open');
|
||||
expect(details?.querySelector('pre')?.textContent).toBe(raw);
|
||||
expect(screen.getByLabelText('向智能体提问')).toBeEnabled();
|
||||
expect(api.send).not.toHaveBeenCalled();
|
||||
});
|
||||
|
||||
it('renders formulas and preserves code escapes while rejecting executable links and HTML', async () => {
|
||||
api.read.mockResolvedValue({ ...first, requests: [request({
|
||||
response: '$E=mc^2$\n\n```js\nconst value = "\\n";\n```\n\n[危险](javascript:alert)\n\n<script>alert(1)</script>\n\n\n\n',
|
||||
@@ -771,6 +794,22 @@ describe('teacher side chat', () => {
|
||||
|
||||
|
||||
describe('legacy discussion history compatibility', () => {
|
||||
it('shows every historical shortcut even when the old reply has more than three', async () => {
|
||||
const shortcuts = ['看懂规则｜摸头会改变什么？', '想想作用｜我们聊聊摸头。', '试试结果｜怎样看出变化？', '换个话题｜我想聊别的地方。'];
|
||||
api.read.mockResolvedValue({ ...first, requests: [request({
|
||||
presentation: 'discussion-v1', response: '这些是之前留下的讨论入口。', suggestedQuestions: shortcuts,
|
||||
})] });
|
||||
render(<TeacherChatPanel projectId="p" sourceId="c" />);
|
||||
await ready();
|
||||
|
||||
expect(screen.getByText('这些是之前留下的讨论入口。')).toBeVisible();
|
||||
for (const shortcut of shortcuts) {
|
||||
expect(screen.getByRole('button', { name: shortcut, exact: true })).toBeEnabled();
|
||||
}
|
||||
expect(screen.queryByText('查看收到的原始内容')).not.toBeInTheDocument();
|
||||
expect(api.send).not.toHaveBeenCalled();
|
||||
});
|
||||
|
||||
const contents = [
|
||||
{ kind: 'ideas', title: '旧想法组件', items: [{ id: 'dog', text: '养只小狗', state: 'kept' }] },
|
||||
{ kind: 'structure', title: '旧结构组件', nodes: [{ id: 'dog', label: '照顾小狗' }] },
|
||||
|
||||
@@ -625,6 +625,28 @@ describe('teacher contextual discussion entry points', () => {
|
||||
});
|
||||
});
|
||||
|
||||
it.each([false, true])('persists prose and exact raw diagnostics instead of excess new shortcuts (cloud=%s)', async (cloudTeacher) => {
|
||||
const f = await fixture(cloudTeacher ? { cloudTeacher: true, mockCloud: true } : {});
|
||||
const scope = newScope(f);
|
||||
const topic = await f.service.create(scope);
|
||||
const reply = '我们先看看规则会怎样影响结局。';
|
||||
const quickReplies = ['看懂规则｜现在是什么规则？', '想想影响｜摸头会改变什么？', '试试结果｜怎样验证？', '换个话题｜先聊别的。'];
|
||||
const raw = JSON.stringify({ reply, quickReplies }, null, 2);
|
||||
f.replyWith(raw);
|
||||
await f.service.send(scope, topic.id, { requestId, intent: 'suggestions', text: '帮我看看' });
|
||||
const completed = await finishRequest(f, scope, topic.id);
|
||||
expect(completed.requests[0]).toMatchObject({
|
||||
status: 'completed', response: reply, suggestedQuestions: [],
|
||||
replyParseError: expect.stringContaining('最多 3 条'), unparsedResponse: raw,
|
||||
});
|
||||
const disk = JSON.parse(await readFile(path.join(
|
||||
f.created.project.path, '.makelore/teacher-conversations', topic.accountId, 'project', topic.id + '.json'
|
||||
), 'utf8'));
|
||||
expect(disk.requests[0]).toMatchObject({
|
||||
response: reply, suggestedQuestions: [], replyParseError: completed.requests[0].replyParseError, unparsedResponse: raw,
|
||||
});
|
||||
});
|
||||
|
||||
it('does not treat a cancelled response as valid suggestions even when its JSON is complete', async () => {
|
||||
const f = await fixture();
|
||||
const scope = newScope(f);
|
||||
@@ -695,7 +717,7 @@ describe('teacher contextual discussion entry points', () => {
|
||||
expect(saved.requests[0].suggestedQuestions).toBeUndefined();
|
||||
});
|
||||
|
||||
it('adds a one-question guided opening for this round without changing the teacher system instructions', async () => {
|
||||
it('marks the guided-help entry without choosing a teaching strategy or changing the system instructions', async () => {
|
||||
const f = await fixture();
|
||||
const scope = newScope(f);
|
||||
const topic = await f.service.create(scope);
|
||||
@@ -707,7 +729,7 @@ describe('teacher contextual discussion entry points', () => {
|
||||
});
|
||||
expect(completed.requests[0].suggestedQuestions).toEqual([]);
|
||||
const messages = f.run.mock.calls[0][0];
|
||||
expect(messages.at(-1).content).toContain('推荐一个具体切入点');
|
||||
expect(messages.at(-1).content).toContain('学生主动求助，表示暂时说不清想问什么');
|
||||
expect(messages[0]).toEqual(compileTeacherContext(definition, context, [], '问题', [], undefined, 'question', undefined, true).messages[0]);
|
||||
});
|
||||
|
||||
@@ -752,26 +774,27 @@ describe('teacher contextual discussion entry points', () => {
|
||||
expect(f.run).not.toHaveBeenCalled();
|
||||
});
|
||||
|
||||
it('budgets the round-specific instructions and honestly starts from ideas with no source', () => {
|
||||
it('budgets the entry facts even when no source conversation is available', () => {
|
||||
const empty = { ...context, messages: [] };
|
||||
const ordinary = compileTeacherContext(definition, empty, [], '老师，帮我看看', []);
|
||||
const suggestions = compileTeacherContext(definition, empty, [], '老师，帮我看看', [], 8000, 'suggestions');
|
||||
expect(suggestions.messages[0]).toEqual(ordinary.messages[0]);
|
||||
expect(suggestions.messages.at(-1)?.content).toContain('没有可用上下文时');
|
||||
expect(suggestions.messages.at(-1)?.content).toContain('不编造');
|
||||
expect(suggestions.messages.at(-1)?.content).toContain('学生通过“帮我看看”主动请求帮助');
|
||||
expect(suggestions.messages.some(message => message.content.startsWith('来源编程会话'))).toBe(false);
|
||||
expect(() => compileTeacherContext(definition, empty, [], '老师，帮我看看', [], estimateTeacherTokens(ordinary.messages), 'suggestions'))
|
||||
.toThrow('超过上下文预算');
|
||||
});
|
||||
|
||||
it.each(['suggestions', 'guided-help'] as const)('recommends one concrete starting point for %s without local expression policy', (intent) => {
|
||||
it.each(['suggestions', 'guided-help'] as const)('passes %s entry facts without local teaching or expression policy', (intent) => {
|
||||
const selected = { ...definition, system_prompt: '云端配置：正文长度随内容，选项保留【我想聊】前缀。' };
|
||||
const compiled = compileTeacherContext(selected, context, [], '帮我看看', [], 8000, intent, teacherReplyInstructions());
|
||||
const current = compiled.messages.at(-1)!.content;
|
||||
expect(current).toContain('推荐一个具体切入点');
|
||||
expect(current).toContain('快捷回复围绕这个切入点');
|
||||
expect(current).toContain(intent === 'suggestions'
|
||||
? '学生通过“帮我看看”主动请求帮助' : '学生主动求助，表示暂时说不清想问什么');
|
||||
expect(compiled.messages[0].content).toContain(selected.system_prompt);
|
||||
const local = compiled.messages.slice(1).map(message => message.content).join('\n');
|
||||
for (const removed of ['400 字', '120 字', '2–3', '简短中文', '学生口吻', 'Alice', '组件', '先这些'])
|
||||
for (const removed of ['400 字', '120 字', '2–3', '简短中文', '学生口吻', 'Alice', '组件', '先这些',
|
||||
'推荐一个具体切入点', '说明为何', '快捷回复围绕', '从构思切入', '整理想法', '理解关系', '承接已确认的共识'])
|
||||
expect(local).not.toContain(removed);
|
||||
});
|
||||
|
||||
@@ -819,9 +842,8 @@ describe('project teacher check-ins', () => {
|
||||
expect(JSON.stringify(messages)).toContain('创建计数器');
|
||||
expect(JSON.stringify(messages)).toContain('变量是什么意思');
|
||||
expect(messages.at(-1)).toMatchObject({ role: 'system' });
|
||||
expect(messages.at(-1).content).toContain('表达方式遵循已配置的要求');
|
||||
expect(messages.at(-1).content).toContain('按照已配置的人设和职责');
|
||||
expect(messages.at(-1).content).toContain('不声称实际运行');
|
||||
expect(messages.at(-1).content).toContain('程序触发的项目进展检查，学生没有在本轮主动提问');
|
||||
expect(messages.at(-1).content).toContain('回应方式遵循当前智能体的云端配置');
|
||||
expect(f.prepareModel).toHaveBeenLastCalledWith(expect.anything(), selected.definition, expect.objectContaining({
|
||||
projectPath: f.created.project.path, source: context, assertCurrent: expect.any(Function),
|
||||
}), { finalOnly: false });
|
||||
@@ -1435,7 +1457,7 @@ describe('project consultations with selected cloud teachers', () => {
|
||||
expect(result.topic?.id).toBe(topic.id);
|
||||
expect(result.topic?.requests[0]).toMatchObject({ intent: 'check-in', sourceConversationId: f.scope.sourceId });
|
||||
const messages = f.run.mock.calls[0][0] as Array<{ role: string; content: string }>;
|
||||
expect(messages.some(message => message.role === 'system' && message.content.includes('不是用户提问'))).toBe(true);
|
||||
expect(messages.some(message => message.role === 'system' && message.content.includes('学生没有在本轮主动提问'))).toBe(true);
|
||||
expect(messages[0].content).not.toContain(TEACHER_BEHAVIOR_PROMPT);
|
||||
expect(f.prepareCloud.mock.calls[0][3]).toMatchObject({ projectPath: f.created.project.path, source: context });
|
||||
expect(f.prepareModel).not.toHaveBeenCalled();
|
||||
|
||||
@@ -30,13 +30,14 @@ describe('published guidance and text reply wiring', () => {
|
||||
expect(system).toContain('启用的教学补充');
|
||||
expect(system).not.toContain('不应加载的资料');
|
||||
expect(compiled.messages.map(message => message.content).join('\n'))
|
||||
.not.toMatch(/你是麦洛的创作老师|Alice|最多120字|不超过400字|2–3|学生自己的口吻/);
|
||||
.not.toMatch(/你是麦洛的创作老师|Alice|最多120字|不超过400字|2–3|学生自己的口吻|推荐一个具体切入点|说明为何|快捷回复围绕|整理想法|理解关系|承接已确认的共识/);
|
||||
expect(definition).toEqual(before);
|
||||
expect(compiled.messages.at(-1)?.role).toBe(intent === 'check-in' ? 'system' : 'user');
|
||||
if (instructions) expect(compiled.messages.at(-2)).toEqual({ role: 'system', content: instructions });
|
||||
if (intent === 'suggestions' || intent === 'guided-help') {
|
||||
expect(compiled.messages.at(-1)?.content).toContain('推荐一个具体切入点');
|
||||
}
|
||||
const entry = compiled.messages.at(-1)?.content;
|
||||
if (intent === 'suggestions') expect(entry).toContain('学生通过“帮我看看”主动请求帮助');
|
||||
if (intent === 'guided-help') expect(entry).toContain('学生主动求助，表示暂时说不清想问什么');
|
||||
if (intent === 'check-in') expect(entry).toContain('程序触发的项目进展检查，学生没有在本轮主动提问');
|
||||
}
|
||||
);
|
||||
|
||||
|
||||
@@ -113,6 +113,40 @@ async function fixture(value: TeacherTopic) {
|
||||
}
|
||||
|
||||
describe('passive teacher reply history compatibility', () => {
|
||||
it('retains all four historical shortcuts through read, list and a later save', async () => {
|
||||
const original = topic(archivedCases[4]);
|
||||
const historicalReplies = [
|
||||
'看懂规则｜摸头会改变什么？',
|
||||
'想想作用｜我们聊聊摸头的作用。',
|
||||
'试试结果｜怎样知道变化发生了？',
|
||||
'换个话题｜我想先聊别的地方。',
|
||||
];
|
||||
original.requests[0].suggestedQuestions = historicalReplies;
|
||||
const f = await fixture(original);
|
||||
|
||||
const loaded = await f.store.read(original.id);
|
||||
expect(loaded).toEqual(original);
|
||||
expect(loaded.requests[0].suggestedQuestions).toEqual(historicalReplies);
|
||||
expect((await f.store.list()).items).toEqual([{
|
||||
id: original.id, title: original.requests[0].text, updatedAt: original.updatedAt,
|
||||
version: 31, teacherId: 'selected-agent',
|
||||
}]);
|
||||
expect(await new TeacherTopicStore(f.root).read(original.id)).toEqual(original);
|
||||
await f.unchanged();
|
||||
|
||||
loaded.revision++;
|
||||
loaded.requests.push(request({ id: 'new-turn', presentation: 'reply-v1', suggestedQuestions: [],
|
||||
discussionError: undefined, unparsedResponse: undefined }));
|
||||
await f.store.save(loaded);
|
||||
const expected = JSON.parse(JSON.stringify({ ...original,
|
||||
revision: original.revision + 1, requests: [...original.requests, loaded.requests[1]],
|
||||
}));
|
||||
const saved = JSON.parse(await readFile(f.file, 'utf8'));
|
||||
expect(saved).toEqual(expected);
|
||||
expect(saved.requests[0].suggestedQuestions).toEqual(historicalReplies);
|
||||
expect(await new TeacherTopicStore(f.root).read(original.id)).toEqual(expected);
|
||||
});
|
||||
|
||||
it.each(archivedCases)('keeps $name opaque through read, list and a later save', async archive => {
|
||||
const original = topic(archive);
|
||||
const f = await fixture(original);
|
||||
|
||||
@@ -9,6 +9,13 @@ import {
|
||||
import { applyTeacherReply, teacherReplyInstructions } from '../../electron/coding-teacher/reply';
|
||||
import type { TeacherRequest } from '../../shared/coding-teacher';
|
||||
|
||||
const fourHistoricalReplies = [
|
||||
'看懂规则｜摸头会改变什么？',
|
||||
'想想作用｜我们聊聊摸头的作用。',
|
||||
'试试结果｜怎样知道变化发生了？',
|
||||
'换个话题｜我想先聊别的地方。',
|
||||
];
|
||||
|
||||
describe('teacher text and suggestions transport', () => {
|
||||
const envelope = { reply: '🍎 **先看看重力带来的变化。**', quickReplies: ['继续解释', '比较不同选择'] };
|
||||
|
||||
@@ -66,6 +73,17 @@ describe('teacher text and suggestions transport', () => {
|
||||
.toEqual({ reply: '一个入口。', quickReplies: ['继续'] });
|
||||
});
|
||||
|
||||
it('keeps four historical shortcuts intact when decoding either envelope format', () => {
|
||||
const reply = '这些是之前留下的讨论入口。';
|
||||
for (const historical of [
|
||||
{ reply, quickReplies: fourHistoricalReplies },
|
||||
{ intro: reply, questions: fourHistoricalReplies },
|
||||
]) {
|
||||
expect(parseTeacherReply(JSON.stringify(historical)))
|
||||
.toEqual({ reply, quickReplies: fourHistoricalReplies });
|
||||
}
|
||||
});
|
||||
|
||||
it('preserves long ordinary and structured replies without the old 12000 character truncation', () => {
|
||||
const reply = '🍏 云端前缀\n\n' + '这是一段完整的详细解释。'.repeat(10000);
|
||||
expect(parseTeacherReply(reply)).toEqual({ reply, quickReplies: [] });
|
||||
@@ -245,10 +263,47 @@ describe('Main reply application', () => {
|
||||
expect(current.discussionSnapshot).toEqual({ kind: 'ideas', title: '历史记录' });
|
||||
});
|
||||
|
||||
it.each([
|
||||
{ quickReplies: [] },
|
||||
{ quickReplies: ['看懂规则｜摸头会改变什么？'] },
|
||||
{ quickReplies: ['看懂规则｜摸头会改变什么？', '看懂规则｜摸头会改变什么？', '换个话题｜我想聊别的地方。'] },
|
||||
])('applies zero to three new shortcuts without stripping prefixes or deduplicating: $quickReplies', ({ quickReplies }) => {
|
||||
const current = request();
|
||||
current.replyParseError = '之前的格式错误';
|
||||
current.unparsedResponse = '之前的原始输出';
|
||||
applyTeacherReply(current, JSON.stringify({ reply: '先看看摸头的作用。', quickReplies }));
|
||||
|
||||
expect(current.response).toBe('先看看摸头的作用。');
|
||||
expect(current.suggestedQuestions).toEqual(quickReplies);
|
||||
expect(current.replyParseError).toBeUndefined();
|
||||
expect(current.unparsedResponse).toBeUndefined();
|
||||
});
|
||||
|
||||
it.each(['current', 'legacy'] as const)('rejects four newly generated shortcuts in a %s envelope without losing the reply or raw output', format => {
|
||||
const current = request();
|
||||
current.suggestedQuestions = ['上一轮的旧选项'];
|
||||
const reply = '保留这一轮的完整正文。';
|
||||
const envelope = format === 'current'
|
||||
? { reply, quickReplies: fourHistoricalReplies }
|
||||
: { intro: reply, questions: fourHistoricalReplies };
|
||||
const raw = ' \n```json\n' + JSON.stringify(envelope, null, 2) + '\n```\n ';
|
||||
expect(parseTeacherReply(raw)).toEqual({ reply, quickReplies: fourHistoricalReplies });
|
||||
|
||||
applyTeacherReply(current, raw);
|
||||
|
||||
expect(current.response).toBe(reply);
|
||||
expect(current.suggestedQuestions).toEqual([]);
|
||||
expect(current.replyParseError).toBeTruthy();
|
||||
expect(current.unparsedResponse).toBe(raw);
|
||||
expect(current.discussionSnapshot).toEqual({ kind: 'ideas', title: '历史记录' });
|
||||
});
|
||||
|
||||
it('limits instructions to the text/suggestions shape and defers writing choices to cloud configuration', () => {
|
||||
const instructions = teacherReplyInstructions();
|
||||
expect(instructions).toContain('{"reply":string,"quickReplies":string[]}');
|
||||
expect(instructions).toContain('云端配置');
|
||||
expect(instructions).not.toMatch(/Alice|120|400|2–3|学生口吻|kind|tool|想法板|流程图/);
|
||||
expect(instructions).toMatch(/最多\s*3\s*条|0[–-]3\s*条/);
|
||||
expect(instructions).toContain('空数组');
|
||||
expect(instructions).not.toMatch(/Alice|120|400|2–3|学生口吻|kind|tool|想法板|流程图|整理想法|理解关系|承接|共识|已确认内容|待定想法/);
|
||||
});
|
||||
});
|
||||
|
||||
Reference in New Issue
Block a user