geo/apps/desktop-client/src/main/adapters/yuanbao.test.ts

import { describe, expect, it } from 'vitest'

import { __yuanbaoTestUtils } from './yuanbao'

const {
  appendUniqueSourceItems,
  buildOrderedDomSourceItems,
  buildSourceItem,
  extractCitationMarkerIndexes,
  extractQuestionText,
  findUnresolvedCitationIndexes,
  mergeText,
  normalizeSourceUrl,
  parseYuanbaoCaptures,
  resolveCitationsFromMarkers,
} = __yuanbaoTestUtils

describe('yuanbao adapter helpers', () => {
  it('unwraps redirect URLs and reads Yuanbao document URL fields', () => {
    expect(
      normalizeSourceUrl(
        'https://yuanbao.tencent.com/link?target=https%3A%2F%2Fexample.com%2Fa%23ref',
      ),
    ).toBe('https://example.com/a')

    expect(
      buildSourceItem({
        docUrl: 'https://example.com/doc?from=yuanbao#source',
        title: '引用文章',
        siteName: 'Example',
      }),
    ).toMatchObject({
      url: 'https://example.com/doc?from=yuanbao',
      normalized_url: 'https://example.com/doc?from=yuanbao',
      title: '引用文章',
      site_name: 'Example',
      host: 'example.com',
    })
  })

  it('rejects browser and build diagnostics as monitoring questions', () => {
    expect(extractQuestionText({ question_text: '合肥全屋定制推荐' })).toBe('合肥全屋定制推荐')
    expect(() =>
      extractQuestionText({
        question_text:
          "The resource http://localhost:8000/static/js/runtime.c632b2fe.js was preloaded using link preload but not used within a few seconds from the window's load event.",
      }),
    ).toThrow('yuanbao_question_text_runtime_diagnostic')
    expect(() =>
      extractQuestionText({
        question_text:
          'failed to solve: process "/bin/sh -c go build" did not complete successfully: exit code: 1',
      }),
    ).toThrow('yuanbao_question_text_runtime_diagnostic')
  })

  it('merges streaming answer fragments without duplicating a repeated prefix', () => {
    const first =
      '通过三轮搜索，我已经掌握了合肥全屋定制市场的高性价比品牌。一定要合同中的定标准确和责任。'
    const repeated =
      '通过三轮搜索，我已经掌握了合肥全屋定制市场的高性价比品牌。结合你的预算，我建议优先看本土老牌。'

    expect(mergeText(first, repeated)).toBe(repeated)
  })

  it('resolves citation markers against the best available source pool', () => {
    const sources = [
      { url: 'https://example.com/1', normalized_url: 'https://example.com/1' },
      { url: 'https://example.com/2', normalized_url: 'https://example.com/2' },
      { url: 'https://example.com/3', normalized_url: 'https://example.com/3' },
    ]

    expect(extractCitationMarkerIndexes('正文 [citation:1] 和 [citation:3]')).toEqual([1, 3])
    expect(
      resolveCitationsFromMarkers('正文 [citation:1] 和 [citation:3]', [[], sources]).map(
        (item) => item.url,
      ),
    ).toEqual(['https://example.com/1', 'https://example.com/3'])
  })

  it('preserves indexed DOM citation slots even when URLs repeat', () => {
    const domSources = buildOrderedDomSourceItems([
      {
        url: 'https://example.com/repeated',
        title: '来源 1',
        text: null,
        siteName: 'Example',
        sourceIndex: 1,
      },
      {
        url: 'https://example.com/repeated',
        title: '来源 2',
        text: null,
        siteName: 'Example',
        sourceIndex: 2,
      },
      {
        url: 'https://example.com/third',
        title: '来源 3',
        text: null,
        siteName: 'Example',
        sourceIndex: 3,
      },
    ])

    expect(domSources).toHaveLength(3)
    expect(domSources.map((item) => item.title)).toEqual(['来源 1', '来源 2', '来源 3'])
    expect(findUnresolvedCitationIndexes('正文 [citation:2] [citation:3]', domSources)).toEqual([])
  })

  it('keeps primary DOM citation order when appending parsed extras', () => {
    const primary = [
      { url: 'https://example.com/1', normalized_url: 'https://example.com/1' },
      { url: 'https://example.com/2', normalized_url: 'https://example.com/2' },
    ]
    const appended = appendUniqueSourceItems(primary, [
      { url: 'https://example.com/2', normalized_url: 'https://example.com/2', title: 'duplicate' },
      { url: 'https://example.com/3', normalized_url: 'https://example.com/3' },
    ])

    expect(appended.map((item) => item.url)).toEqual([
      'https://example.com/1',
      'https://example.com/2',
      'https://example.com/3',
    ])
  })

  it('does not report incomplete citations when the answer has no markers', () => {
    expect(findUnresolvedCitationIndexes('正文没有引用小图标', [])).toEqual([])
  })

  it('collects source candidates from deep search thinking fields', () => {
    const summary = parseYuanbaoCaptures([
      {
        url: 'https://yuanbao.tencent.com/api/chat',
        status: 200,
        contentType: 'text/event-stream',
        body: [
          'data: {"type":"deepSearch","contents":[{"type":"step","webPages":[{"targetUrl":"https://example.com/a#ref","title":"来源 A","siteName":"Example"}],"references":[{"sourceUrl":"https://example.com/b","title":"来源 B"}]}]}',
          'data: {"type":"text","msg":"正文 [citation:1] [citation:2]"}',
        ].join('\n\n'),
      },
    ])

    expect(summary.searchResults.map((item) => item.url)).toContain('https://example.com/a')
    expect(summary.citations.map((item) => item.url)).toContain('https://example.com/b')
    expect(
      resolveCitationsFromMarkers(summary.answer, [
        summary.citations,
        summary.searchResults,
        [...summary.citations, ...summary.searchResults],
      ]).map((item) => item.url),
    ).toEqual(['https://example.com/b', 'https://example.com/a'])
  })
})