feat(qwen-adapter): capture chat SSE and expand tab_container cards

Attach a response listener to the `/api/v2/chat` POST stream, parse its SSE payloads, and reconstruct answer content/extra_info/communication from the `multi_load/iframe` and `bar/iframe` messages. Expand embedded `[(tab_container_N)]` references against the SSE cardMap so nested tab content is inlined instead of leaking a raw placeholder. Prefer the API answer, falling back to page state, and record `response_source` plus `api_event_count` in the raw response for diagnosis. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-07-12 14:17:14 +08:00
parent b70dc9dc6d
commit bbc2d89450
2 changed files with 410 additions and 19 deletions
@@ -3,7 +3,13 @@ import { describe, expect, it } from 'vitest'
 import type { AdapterContext } from './base'
 import { __qwenTestUtils, qwenAdapter } from './qwen'

-const { extractQuestionText } = __qwenTestUtils
+const { extractQuestionText, parseQwenSSEResponse, resolveQwenAnswerText } = __qwenTestUtils
+
+function buildSSEBody(messages: Record<string, unknown>[]): string {
+  return messages
+    .map((message) => `data:${JSON.stringify({ data: { messages: [message] } })}`)
+    .join('\n\n')
+}

 describe('qwen adapter helpers', () => {
  it('extracts question text from monitoring payload', () => {
@@ -17,10 +23,55 @@ describe('qwen adapter helpers', () => {
  })
 })

+describe('qwen official SSE response parsing', () => {
+  it('uses the complete API snapshot instead of an earlier incremental snapshot', () => {
+    const parsed = parseQwenSSEResponse(
+      buildSSEBody([
+        { mime_type: 'multi_load/iframe', status: 'processing', content: '部分' },
+        { mime_type: 'multi_load/iframe', status: 'complete', content: '完整 API 答案' },
+      ]),
+    )
+
+    expect(resolveQwenAnswerText(parsed.content).answerText).toBe('完整 API 答案')
+    expect(parsed.events).toHaveLength(2)
+  })
+
+  it('builds the tab card map from API metadata and expands the selected tab', () => {
+    const parsed = parseQwenSSEResponse(
+      buildSSEBody([
+        {
+          mime_type: 'multi_load/iframe',
+          status: 'complete',
+          content: '[(tab_container_1)]',
+          meta_data: {
+            multi_load: [
+              {
+                source_seq: 'tab_container_1',
+                type: 'tab_container',
+                content: {
+                  default_tab: 'answer_2',
+                  tabs: [
+                    { tab_id: 'answer_1', content: '备选回答' },
+                    { tab_id: 'answer_2', content: 'API 卡片中的完整回答' },
+                  ],
+                },
+              },
+            ],
+          },
+        },
+      ]),
+    )
+
+    expect(resolveQwenAnswerText(parsed.content).answerText).toBe('API 卡片中的完整回答')
+  })
+})
+
 describe('qwen adapter query result validation', () => {
-  it('does not submit an internal tab-container placeholder as a successful answer', async () => {
+  it('expands an internal tab-container reference into the selected answer', async () => {
    const page = {
      waitForLoadState: async () => undefined,
+      on: () => undefined,
+      off: () => undefined,
      url: () => 'https://www.qianwen.com/chat/demo',
      goto: async () => undefined,
      evaluate: async () => ({
@@ -38,8 +89,57 @@ describe('qwen adapter query result validation', () => {
          content: {
            multiLoadIframe: {
              content: '[(tab_container_1)]',
+              cardMap: {
+                tab_container_1: {
+                  type: 'tab_container',
+                  content: {
+                    default_tab: 'answer_2',
+                    tabs: [
+                      { tab_id: 'answer_1', content: '备选答案。' },
+                      { tab_id: 'answer_2', content: '这是千问实际展示的完整答案。' },
+                    ],
+                  },
+                },
+              },
            },
-            searchResults: [{ url: 'https://example.com/source', title: '来源' }],
+          },
+          extraInfo: null,
+          communication: null,
+        },
+      }),
+    }
+    const context = {
+      playwright: { page, browser: {} },
+      reportProgress: () => undefined,
+    } as unknown as AdapterContext
+
+    const result = await qwenAdapter.query(context, { question_text: '测试问题' })
+
+    expect(result.status).toBe('succeeded')
+    expect(result.payload?.answer).toBe('这是千问实际展示的完整答案。')
+  })
+
+  it('does not submit a tab-container reference when its card data is missing', async () => {
+    const page = {
+      waitForLoadState: async () => undefined,
+      on: () => undefined,
+      off: () => undefined,
+      url: () => 'https://www.qianwen.com/chat/demo',
+      goto: async () => undefined,
+      evaluate: async () => ({
+        ok: true as const,
+        url: 'https://www.qianwen.com/chat/demo',
+        question: {
+          model: 'Qwen',
+          deepSearch: '1',
+          enableSearch: true,
+          cardText: '测试问题',
+        },
+        answer: {
+          reqId: 'req-missing-card',
+          status: 'finish',
+          content: {
+            multiLoadIframe: { content: '[(tab_container_1)]', cardMap: {} },
          },
          extraInfo: null,
          communication: null,
@@ -60,6 +160,8 @@ describe('qwen adapter query result validation', () => {
  it('uses a resolved fallback answer when another content field is still a placeholder', async () => {
    const page = {
      waitForLoadState: async () => undefined,
+      on: () => undefined,
+      off: () => undefined,
      url: () => 'https://www.qianwen.com/chat/demo',
      goto: async () => undefined,
      evaluate: async () => ({
@@ -93,4 +195,67 @@ describe('qwen adapter query result validation', () => {
    expect(result.status).toBe('succeeded')
    expect(result.payload?.answer).toBe('这是完整答案。')
  })
+
+  it('prefers the official SSE answer over the page-state fallback', async () => {
+    const apiBody = buildSSEBody([
+      {
+        mime_type: 'multi_load/iframe',
+        status: 'complete',
+        content: '官方 SSE 完整答案',
+      },
+    ])
+    type MockResponse = {
+      url(): string
+      request(): { method(): string }
+      text(): Promise<string>
+    }
+    let responseListener: ((response: MockResponse) => void) | null = null
+    const page = {
+      waitForLoadState: async () => undefined,
+      on: (_event: string, listener: (response: MockResponse) => void) => {
+        responseListener = listener
+      },
+      off: () => {
+        responseListener = null
+      },
+      url: () => 'https://www.qianwen.com/chat/demo',
+      goto: async () => undefined,
+      evaluate: async () => {
+        responseListener?.({
+          url: () => 'https://chat2.qianwen.com/api/v2/chat',
+          request: () => ({ method: () => 'POST' }),
+          text: async () => apiBody,
+        })
+        return {
+          ok: true as const,
+          url: 'https://www.qianwen.com/chat/demo',
+          question: {
+            model: 'Qwen',
+            deepSearch: '1',
+            enableSearch: true,
+            cardText: '测试问题',
+          },
+          answer: {
+            reqId: 'req-api',
+            status: 'finish',
+            content: { multiLoadIframe: { content: '页面 state 兜底答案' } },
+            extraInfo: null,
+            communication: null,
+          },
+        }
+      },
+    }
+    const context = {
+      playwright: { page, browser: {} },
+      reportProgress: () => undefined,
+    } as unknown as AdapterContext
+
+    const result = await qwenAdapter.query(context, { question_text: '测试问题' })
+
+    expect(result.status).toBe('succeeded')
+    expect(result.payload?.answer).toBe('官方 SSE 完整答案')
+    expect(
+      (result.payload?.raw_response_json as Record<string, unknown>).response_source,
+    ).toBe('api_sse')
+  })
 })
@@ -1,4 +1,5 @@
 import type { JsonValue, MonitoringSourceItem } from '@geo/shared-types'
+import type { Response as PlaywrightResponse } from 'playwright-core'

 import type { MonitorAdapter } from './base'
 import { normalizeText } from './common'
@@ -7,6 +8,7 @@ const QWEN_BOOTSTRAP_URL = 'https://www.qianwen.com/'
 const QWEN_PAGE_READY_TIMEOUT_MS = 20_000
 const QWEN_QUERY_TIMEOUT_MS = 90_000
 const QWEN_INTERNAL_TAB_PLACEHOLDER_PATTERN = /^(?:\[\(tab_container_\d+\)\]\s*)+$/i
+const QWEN_INTERNAL_TAB_REFERENCE_PATTERN = /\[\(tab_container_\d+\)\]/i

 type QwenPageQuestionSnapshot = {
  model: string | null
@@ -41,6 +43,13 @@ type QwenPageQueryFailureResult = {

 type QwenPageQueryResult = QwenPageQuerySuccessResult | QwenPageQueryFailureResult

+type QwenParsedSSEResponse = {
+  content: Record<string, unknown> | null
+  extraInfo: Record<string, unknown> | null
+  communication: Record<string, unknown> | null
+  events: Record<string, unknown>[]
+}
+
 function isRecord(value: unknown): value is Record<string, unknown> {
  return typeof value === 'object' && value !== null && !Array.isArray(value)
 }
@@ -206,18 +215,67 @@ function isQwenInternalPlaceholderOnly(value: unknown): boolean {
  return Boolean(text && QWEN_INTERNAL_TAB_PLACEHOLDER_PATTERN.test(text))
 }

+function resolveQwenTabContainerText(card: unknown): string | null {
+  if (!isRecord(card) || card.type !== 'tab_container' || !isRecord(card.content)) {
+    return null
+  }
+
+  const tabs = Array.isArray(card.content.tabs)
+    ? card.content.tabs.filter(isRecord)
+    : []
+  const selectedTabId =
+    normalizeOptionalString(card.content.selected_tab) ??
+    normalizeOptionalString(card.content.default_tab)
+  const selectedTab = selectedTabId
+    ? tabs.find((tab) => normalizeOptionalString(tab.tab_id) === selectedTabId)
+    : null
+  const fallbackTab = tabs.find((tab) => normalizeOptionalString(tab.content))
+
+  return normalizeOptionalString((selectedTab ?? fallbackTab)?.content)
+}
+
+function expandQwenTabContainers(content: unknown, value: string): string {
+  if (!isRecord(content) || !isRecord(content.multiLoadIframe)) {
+    return value
+  }
+
+  const cardMap = content.multiLoadIframe.cardMap
+  if (!isRecord(cardMap)) {
+    return value
+  }
+
+  let expanded = value
+  for (let depth = 0; depth < 4; depth += 1) {
+    const next = expanded.replace(/\[\((tab_container_\d+)\)\]/gi, (reference, sourceSeq) => {
+      return resolveQwenTabContainerText(cardMap[sourceSeq]) ?? reference
+    })
+    if (next === expanded) {
+      break
+    }
+    expanded = next
+  }
+
+  return expanded.trim()
+}
+
 function resolveQwenAnswerText(content: unknown): {
  answerText: string | null
  unresolvedPlaceholder: string | null
 } {
-  const candidates = [
+  const rawCandidates = [
    readNestedString(content, ['multiLoadIframe', 'content']),
    readNestedString(content, ['barIframe', 'content']),
  ]
-  const answerText = candidates.find(
-    (candidate) => candidate && !isQwenInternalPlaceholderOnly(candidate),
+  const candidates = rawCandidates.map((candidate) =>
+    candidate ? expandQwenTabContainers(content, candidate) : null,
  )
-  const unresolvedPlaceholder = candidates.find(isQwenInternalPlaceholderOnly)
+  const answerText = candidates.find(
+    (candidate) => candidate && !QWEN_INTERNAL_TAB_REFERENCE_PATTERN.test(candidate),
+  )
+  const unresolvedPlaceholder =
+    candidates.find(
+      (candidate) => candidate && QWEN_INTERNAL_TAB_REFERENCE_PATTERN.test(candidate),
+    ) ?? rawCandidates.find(isQwenInternalPlaceholderOnly)

  return {
    answerText: answerText ?? null,
@@ -225,6 +283,152 @@ function resolveQwenAnswerText(content: unknown): {
  }
 }

+function parseQwenSSEPayloads(body: string): Record<string, unknown>[] {
+  const payloads: Record<string, unknown>[] = []
+  for (const block of body.split(/\r?\n\r?\n/)) {
+    const data = block
+      .split(/\r?\n/)
+      .filter((line) => line.startsWith('data:'))
+      .map((line) => line.slice(5).trimStart())
+      .join('\n')
+      .trim()
+    if (!data || data === '[DONE]') {
+      continue
+    }
+
+    try {
+      const parsed = JSON.parse(data) as unknown
+      if (isRecord(parsed)) {
+        payloads.push(parsed)
+      }
+    } catch {
+      continue
+    }
+  }
+  return payloads
+}
+
+function qwenSSEMessages(payload: Record<string, unknown>): Record<string, unknown>[] {
+  const data = isRecord(payload.data) ? payload.data : payload
+  const messages = Array.isArray(data.messages)
+    ? data.messages
+    : Array.isArray(data.response_messages)
+      ? data.response_messages
+      : []
+  return messages.filter(isRecord)
+}
+
+function selectBestQwenSSEMessage(
+  messages: Record<string, unknown>[],
+  mimeType: string,
+): Record<string, unknown> | null {
+  let selected: Record<string, unknown> | null = null
+  let selectedScore = -1
+  for (const message of messages) {
+    if (normalizeOptionalString(message.mime_type) !== mimeType) {
+      continue
+    }
+    const contentLength = normalizeOptionalString(message.content)?.length ?? 0
+    const score = (message.status === 'complete' ? 1_000_000 : 0) + contentLength
+    if (score >= selectedScore) {
+      selected = message
+      selectedScore = score
+    }
+  }
+  return selected
+}
+
+function buildQwenSSECardMap(messages: Record<string, unknown>[]): Record<string, unknown> {
+  const cardMap: Record<string, unknown> = {}
+  for (const message of messages) {
+    if (normalizeOptionalString(message.mime_type) !== 'multi_load/iframe') {
+      continue
+    }
+    const metaData = isRecord(message.meta_data) ? message.meta_data : null
+    const cards = Array.isArray(metaData?.multi_load) ? metaData.multi_load : []
+    for (const card of cards) {
+      if (!isRecord(card)) {
+        continue
+      }
+      const sourceSeq = normalizeOptionalString(card.source_seq ?? card.sourceSeq)
+      if (!sourceSeq) {
+        continue
+      }
+      cardMap[sourceSeq] = {
+        sourceSeq,
+        type: normalizeOptionalString(card.type),
+        content: card.content ?? null,
+        html: card.html ?? null,
+        htmlMeta: card.html_meta ?? card.htmlMeta ?? null,
+      }
+    }
+  }
+  return cardMap
+}
+
+function mergeQwenSSERecord(
+  current: Record<string, unknown> | null,
+  value: unknown,
+): Record<string, unknown> | null {
+  return isRecord(value) ? { ...(current ?? {}), ...value } : current
+}
+
+function parseQwenSSEResponse(body: string): QwenParsedSSEResponse {
+  const payloads = parseQwenSSEPayloads(body)
+  const messages = payloads.flatMap(qwenSSEMessages)
+  const multiLoadMessage = selectBestQwenSSEMessage(messages, 'multi_load/iframe')
+  const barMessage = selectBestQwenSSEMessage(messages, 'bar/iframe')
+  let extraInfo: Record<string, unknown> | null = null
+  let communication: Record<string, unknown> | null = null
+
+  for (const payload of payloads) {
+    const data = isRecord(payload.data) ? payload.data : payload
+    extraInfo = mergeQwenSSERecord(extraInfo, data.extra_info ?? data.extraInfo)
+    communication = mergeQwenSSERecord(
+      communication,
+      data.communication ?? data.communication_info,
+    )
+  }
+
+  const content: Record<string, unknown> = {}
+  if (multiLoadMessage) {
+    content.multiLoadIframe = {
+      status: normalizeOptionalString(multiLoadMessage.status),
+      content: normalizeOptionalString(multiLoadMessage.content),
+      metaData: isRecord(multiLoadMessage.meta_data) ? multiLoadMessage.meta_data : null,
+      cardMap: buildQwenSSECardMap(messages),
+    }
+  }
+  if (barMessage) {
+    content.barIframe = {
+      status: normalizeOptionalString(barMessage.status),
+      content: normalizeOptionalString(barMessage.content),
+      metaData: isRecord(barMessage.meta_data) ? barMessage.meta_data : null,
+    }
+  }
+
+  return {
+    content: Object.keys(content).length ? content : null,
+    extraInfo,
+    communication,
+    events: payloads.map((payload) => (isRecord(payload.data) ? payload.data : payload)),
+  }
+}
+
+function isQwenChatAPIResponse(response: {
+  url(): string
+  request(): { method(): string }
+}): boolean {
+  if (response.request().method() !== 'POST') {
+    return false
+  }
+  try {
+    return new URL(response.url()).pathname === '/api/v2/chat'
+  } catch {
+    return false
+  }
+}
+
 function toJsonValue(value: unknown, depth = 0): JsonValue {
  if (value == null || depth > 16) {
    return null
@@ -636,12 +840,21 @@ export const qwenAdapter: MonitorAdapter = {
    }

    context.reportProgress('qwen.query')
-    const pageResult = await page.evaluate(qwenQueryInPage, {
-      questionText,
-      timeoutMs: QWEN_QUERY_TIMEOUT_MS,
-      readyTimeoutMs: QWEN_PAGE_READY_TIMEOUT_MS,
-      deepThink: true,
-    })
+    let apiResponse: PlaywrightResponse | null = null
+    const captureAPIResponse = (response: PlaywrightResponse) => {
+      if (!apiResponse && isQwenChatAPIResponse(response)) {
+        apiResponse = response
+      }
+    }
+    page.on('response', captureAPIResponse)
+    const pageResult = await page
+      .evaluate(qwenQueryInPage, {
+        questionText,
+        timeoutMs: QWEN_QUERY_TIMEOUT_MS,
+        readyTimeoutMs: QWEN_PAGE_READY_TIMEOUT_MS,
+        deepThink: true,
+      })
+      .finally(() => page.off('response', captureAPIResponse))

    if (!pageResult?.ok || !pageResult.answer) {
      const message = formatQwenQueryError(pageResult)
@@ -655,9 +868,18 @@ export const qwenAdapter: MonitorAdapter = {
      }
    }

-    const { answerText, unresolvedPlaceholder } = resolveQwenAnswerText(
-      pageResult.answer.content,
-    )
+    const capturedAPIResponse = apiResponse as PlaywrightResponse | null
+    const apiResponseBody = capturedAPIResponse
+      ? await capturedAPIResponse.text().catch(() => null)
+      : null
+    const apiResult = apiResponseBody ? parseQwenSSEResponse(apiResponseBody) : null
+
+    const apiAnswer = resolveQwenAnswerText(apiResult?.content)
+    const pageAnswer = resolveQwenAnswerText(pageResult.answer.content)
+    const answerText = apiAnswer.answerText ?? pageAnswer.answerText
+    const unresolvedPlaceholder =
+      apiAnswer.unresolvedPlaceholder ?? pageAnswer.unresolvedPlaceholder
+    const responseSource = apiAnswer.answerText ? 'api_sse' : 'page_state'
    const requestId =
      normalizeText(pageResult.answer.reqId) ??
      readNestedString(pageResult.answer.communication, ['reqid']) ??
@@ -666,6 +888,7 @@ export const qwenAdapter: MonitorAdapter = {
    const providerModel = resolveQwenProviderModel(pageResult)

    const searchResults: MonitoringSourceItem[] = []
+    collectQwenSources(apiResult?.events, searchResults)
    collectQwenSources(pageResult.answer.content, searchResults)
    const dedupedSearchResults = dedupeSourceItems(searchResults)

@@ -708,6 +931,8 @@ export const qwenAdapter: MonitorAdapter = {
        raw_response_json: {
          platform: 'qwen',
          mode: pageResult.question?.deepSearch === '1' ? 'deep_think' : 'standard',
+          response_source: responseSource,
+          api_event_count: apiResult?.events.length ?? 0,
          page_url: pageResult.url,
          session_id: sessionId,
          provider_request_id: requestId,
@@ -716,9 +941,9 @@ export const qwenAdapter: MonitorAdapter = {
          question: toJsonValue(pageResult.question),
          source_count: dedupedSearchResults.length,
          search_results: toJsonSources(dedupedSearchResults),
-          content: toJsonValue(pageResult.answer.content),
-          extra_info: toJsonValue(pageResult.answer.extraInfo),
-          communication: toJsonValue(pageResult.answer.communication),
+          content: toJsonValue(apiResult?.content ?? pageResult.answer.content),
+          extra_info: toJsonValue(apiResult?.extraInfo ?? pageResult.answer.extraInfo),
+          communication: toJsonValue(apiResult?.communication ?? pageResult.answer.communication),
          status: 'succeeded',
        },
      },
@@ -730,4 +955,5 @@ export const __qwenTestUtils = {
  extractQuestionText,
  isQwenInternalPlaceholderOnly,
  resolveQwenAnswerText,
+  parseQwenSSEResponse,
 }