Consolidate approved trial client updates and preserve composer runtime

tests/e2e/cloud-agents.spec.ts

@@ -12,6 +12,36 @@ test('personal cloud Agent identity, configuration and separate published chat i
     await expect(page.getByTestId('ai-module-selection-page')).toBeVisible();
     const applicationUrl = page.url();
     await page.goto('about:blank');
+    // Exercise the shared voice path without requesting a real microphone.
+    await page.addInitScript(() => {
+      class FixtureMediaRecorder {
+        static isTypeSupported() { return true; }
+        state = 'inactive';
+        mimeType = 'audio/webm';
+        ondataavailable: ((event: { data: Blob }) => void) | null = null;
+        onstop: (() => void) | null = null;
+        start() { this.state = 'recording'; }
+        stop() {
+          this.state = 'inactive';
+          this.ondataavailable?.({ data: new Blob(['fixture voice'], { type: 'audio/webm' }) });
+          this.onstop?.();
+        }
+      }
+      class FixtureAudioContext {
+        async decodeAudioData() { return { duration: 0.1 }; }
+        async close() { /* The fixture does not open an audio device. */ }
+      }
+      class FixtureOfflineAudioContext {
+        destination = {};
+        createBufferSource() { return { buffer: null, connect() {}, start() {} }; }
+        async startRendering() { return { getChannelData: () => new Float32Array(1600) }; }
+      }
+      Object.defineProperty(navigator, 'mediaDevices', { configurable: true,
+        value: { getUserMedia: async () => ({ getTracks: () => [{ stop() {} }] }) } });
+      Object.defineProperty(window, 'MediaRecorder', { configurable: true, value: FixtureMediaRecorder });
+      Object.defineProperty(window, 'AudioContext', { configurable: true, value: FixtureAudioContext });
+      Object.defineProperty(window, 'OfflineAudioContext', { configurable: true, value: FixtureOfflineAudioContext });
+    });
     // This UI test owns the Host API boundary; real cloud HTTP/PG is tested separately.
     await app.evaluate(({ ipcMain, BrowserWindow }, uploadedAvatar) => {
       BrowserWindow.getAllWindows()[0].setSize(1366, 768);
@@ -21,7 +51,8 @@ test('personal cloud Agent identity, configuration and separate published chat i
       const creationInputs: Record<string, unknown>[] = [];
       let pending: Record<string, unknown>[] = [];
       const operations: string[] = [];
-      Reflect.set(globalThis, '__cloudAgentUiFixture', { snapshot: () => ({ draft, createInput, creationInputs, operations }) });
+      let speechRequests = 0;
+      Reflect.set(globalThis, '__cloudAgentUiFixture', { snapshot: () => ({ draft, createInput, creationInputs, operations, speechRequests }) });
       const versions: { version: number; draft_revision: number; created_at: string }[] = [];
       let knowledgeCatalogLoads = 0;
       let knowledgeCreated = false;
@@ -46,6 +77,16 @@ test('personal cloud Agent identity, configuration and separate published chat i
         if (path === '/api/works/user/agent-profile') return result({
           success: true, profile: { display_name: '创作者', age: null, gender: null, avatar_url: null, share_age_with_agents: false, share_gender_with_agents: false, analysis_enabled: true, completed: true, version: 1, updated_at: '2026-09-10T00:00:00Z' },
         });
+        if (path === '/api/works/speech/transcriptions') {
+          const speech = JSON.parse(request.body ?? '{}');
+          if (speech.mimeType !== 'audio/wav' || speech.fileName !== 'voice.wav' || speech.language !== 'zh'
+            || typeof speech.audioBase64 !== 'string' || !speech.audioBase64.startsWith('UklGR')) {
+            throw new Error('Unexpected speech transcription fixture payload');
+          }
+          speechRequests++;
+          await new Promise(resolve => setTimeout(resolve, 800));
+          return result({ success: true, transcription: { text: '加上语音描述', model: 'fixture-transcribe' } });
+        }
         if (path.startsWith('/api/cloud-agents/')) {
           const body = typeof request.body === 'string' ? JSON.parse(request.body) : request.body ?? {};
           if (path.endsWith('/avatars/pick')) return result(uploadedAvatar);
@@ -189,9 +230,9 @@ test('personal cloud Agent identity, configuration and separate published chat i
     await page.getByRole('button', { name: '下一步', exact: true }).click();
     await expect(page.getByLabel('名称', { exact: true })).toBeDisabled();
     await expect(page.getByRole('button', { name: '更换头像', exact: true })).toBeDisabled();
-    await expect(page.getByRole('complementary', { name: '未确认的云端操作' })).toBeVisible();
-    await page.getByRole('button', { name: '移除提醒', exact: true }).click();
     await expect(page.getByRole('complementary', { name: '未确认的云端操作' })).toHaveCount(0);
+    await expect(page.getByRole('button', { name: '返回列表确认', exact: true })).toBeVisible();
+    await expect(page.getByRole('dialog', { name: '创建记录', exact: true })).toHaveCount(0);
     await page.getByRole('button', { name: '重试创建', exact: true }).click();
     await expect(page.getByRole('alert')).toContainText('创建结果暂未确认');
     await expect(page.getByLabel('名称', { exact: true })).toBeDisabled();
@@ -306,6 +347,41 @@ test('personal cloud Agent identity, configuration and separate published chat i
     expect(await messageOperations()).toEqual([]);
     await expect(formalChat.getByRole('heading', { name: 'Hi，今天想聊些什么？', exact: true })).toBeVisible();
     await expect(formalChat.locator('.agent-chat-welcome').getByRole('img', { name: '我的写作搭档头像', exact: true })).toHaveAttribute('src', avatar.icon_url);
+    const voiceComposer = formalChat.getByTestId('cloud-chat-composer');
+    const voiceDraft = formalChat.getByLabel('消息', { exact: true });
+    await expect(voiceComposer.locator('.agent-attachment-composer')).toHaveCount(1);
+    await expect(voiceComposer.getByRole('button', { name: '语音输入', exact: true })).toHaveClass(/rounded-full/);
+    await expect(voiceDraft).toHaveAttribute('rows', '1');
+    const sendGeometry = await voiceComposer.getByRole('button', { name: '发送', exact: true }).evaluate(element => {
+      const style = getComputedStyle(element);
+      return { width: element.clientWidth, height: element.clientHeight, radius: Number.parseFloat(style.borderTopLeftRadius) };
+    });
+    expect(sendGeometry.width).toBe(32);
+    expect(sendGeometry.height).toBe(32);
+    expect(sendGeometry.radius).toBeGreaterThanOrEqual(sendGeometry.width / 2);
+    const toolbarPositions = await voiceComposer.locator('[aria-label="添加文件或图片"], [aria-label="语音输入"], [aria-label="发送"]').evaluateAll(elements =>
+      Object.fromEntries(elements.map(element => {
+        const bounds = element.getBoundingClientRect();
+        return [element.getAttribute('aria-label'), { left: bounds.left, right: bounds.right }];
+      })));
+    expect(toolbarPositions['语音输入'].left).toBeGreaterThan(toolbarPositions['添加文件或图片'].right);
+    expect(toolbarPositions['发送'].left).toBeGreaterThan(toolbarPositions['语音输入'].right);
+    expect(toolbarPositions['发送'].left - toolbarPositions['语音输入'].right).toBeCloseTo(4, 1);
+    await voiceDraft.fill('原来的想法');
+    await voiceComposer.getByRole('button', { name: '语音输入', exact: true }).click();
+    await expect(voiceComposer.getByRole('button', { name: '停止语音输入' })).toBeVisible();
+    await voiceDraft.fill('录音时继续编辑');
+    await expect(voiceComposer.getByRole('button', { name: '发送', exact: true })).toBeDisabled();
+    await voiceDraft.press('Enter');
+    await voiceComposer.getByRole('button', { name: '停止语音输入' }).click();
+    await expect(voiceComposer.getByRole('button', { name: '正在识别语音' })).toBeDisabled();
+    await voiceDraft.fill('识别时继续编辑');
+    await expect(voiceDraft).toHaveValue('识别时继续编辑\n加上语音描述');
+    await expect(voiceDraft).toHaveAttribute('rows', '2');
+    expect((await fixtureState()).speechRequests).toBe(1);
+    expect(await messageOperations()).toEqual([]);
+    await page.screenshot({ path: testInfo.outputPath('cloud-agent-published-voice-composer.png') });
+    await voiceDraft.fill('');
     const expectCleanChat = async () => {
       await expect(page.getByRole('navigation', { name: '对话来源', exact: true })).toHaveCount(0);
       await expect(page.getByRole('textbox', { name: '名称', exact: true })).toHaveCount(0);
@@ -331,10 +407,10 @@ test('personal cloud Agent identity, configuration and separate published chat i
     const userBubble = formalChat.locator('.agent-chat-message[data-role="user"]');
     await expect(userBubble).toHaveText(/先给我一些创作建议/);
     await expect(userBubble.locator('time')).toHaveAttribute('datetime', '2026-10-13T09:24:30.000Z');
-    await expect(userBubble.locator('time')).toHaveAccessibleName(/^发送时间：/);
+    await expect(userBubble.locator('time')).toHaveAttribute('aria-label', /^发送时间：/);
     const answerTime = formalChat.locator('.agent-chat-message[data-role="assistant"] time');
     await expect(answerTime).toHaveAttribute('datetime', '2026-10-13T09:25:40.000Z');
-    await expect(answerTime).toHaveAccessibleName(/^回复时间：/);
+    await expect(answerTime).toHaveAttribute('aria-label', /^回复时间：/);
     const bubbleBody = userBubble.locator('.agent-chat-message-body');
     const bubbleStyle = await bubbleBody.evaluate(element => {
       const style = getComputedStyle(element);