Merge branch 'nanin/develop' into develop

2026-08-17 03:27:00 +08:00 · 2026-08-05 18:44:25 +08:00
parent 55da2e2e67 c4e13ee7d2
commit 3c59fb64e9
37 changed files with 5185 additions and 492 deletions
@@ -25,10 +25,16 @@ describe('export media', () => {
    const html = renderExportPage('脱敏导出')

    expect(EXPORT_PAGE_SIZE).toBe(240)
-    expect(html).toContain('<script src="data/messages.js"></script>')
+    expect(html).toContain("dataScript.src = 'data/messages.js'")
+    expect(html).toContain('id="archive-loading"')
+    expect(html).toContain('正在加载聊天档案')
+    expect(html).toContain('requestAnimationFrame(() => window.setTimeout(loadArchiveData, 0))')
    expect(html).toContain('aria-label="聊天时间轴"')
+    expect(html).toContain('aria-expanded="')
+    expect(html).toContain('setExpandedTimelineYear')
    expect(html).toContain('data-kind="media"')
    expect(html).toContain('placeholder="搜索发送者或消息内容…"')
+    expect(html).toContain('font-size: 16px;')
    expect(html).toContain('filtered.slice(windowStart, windowEnd)')
    expect(html).toContain('windowStart = Math.max(0, windowEnd - PAGE_SIZE)')
    expect(html).toContain('scheduleWindowSlide')
@@ -68,7 +74,7 @@ describe('export media', () => {

    expect(dom.window.document.querySelectorAll('.message')).toHaveLength(EXPORT_PAGE_SIZE)
    expect(dom.window.document.querySelector('#count')?.textContent).toBe(
-      '已显示 240 / 筛选结果 500 / 全部 500'
+      '已显示 240 / 筛选 500 / 全部 500'
    )
    const list = dom.window.document.querySelector('#messages')!
    expect(list.querySelector('.message')?.getAttribute('data-index')).toBe('260')
@@ -90,8 +96,227 @@ describe('export media', () => {
    search.value = 'needle'
    search.dispatchEvent(new dom.window.Event('input'))
    expect(dom.window.document.querySelectorAll('.message')).toHaveLength(5)
-    expect(dom.window.document.querySelector('#count')?.textContent).toBe('筛选结果 5 / 全部 500')
+    expect(dom.window.document.querySelector('#count')?.textContent).toContain('筛选 5')
+    expect(dom.window.document.querySelectorAll('.search-highlight')).toHaveLength(5)
+    expect(dom.window.document.querySelectorAll('.locate-all')).toHaveLength(5)
    expect(dom.window.document.querySelectorAll('.timeline-month').length).toBeGreaterThan(1)
+    expect(
+      dom.window.document.querySelectorAll('.timeline-year[aria-expanded="true"]')
+    ).toHaveLength(1)
+    expect(dom.window.document.querySelectorAll('.timeline-months:not([hidden])')).toHaveLength(1)
+    dom.window.close()
+  })
+
+  it('does not match hidden sender ids when searching visible message text', () => {
+    const html = renderExportPage('搜索测试')
+    const dom = new JSDOM(html, { runScripts: 'outside-only' })
+    const messages: Message[] = [
+      {
+        id: 'hidden-sender-id-match',
+        from: 'user',
+        type: '普通文本',
+        datetime: '',
+        content: '这条消息不应命中',
+        name: 'Jamie',
+        senderId: 'wxid_fixture_member',
+        isSender: false,
+        createTime: 1_767_225_600
+      },
+      {
+        id: 'visible-content-match',
+        from: 'user',
+        type: '普通文本',
+        datetime: '',
+        content: 'https://example.com/xi',
+        name: 'Cherry',
+        senderId: 'wxid_fixture_self',
+        isSender: true,
+        createTime: 1_767_225_601
+      }
+    ]
+    Object.assign(dom.window, {
+      __WECHAT_EXPORT__: {
+        version: 1,
+        sourceId: 'fixture',
+        name: '搜索测试',
+        exportedAt: '2026-08-05T00:00:00.000Z',
+        messages
+      }
+    })
+
+    dom.window.eval(inlineScriptOf(html))
+    const search = dom.window.document.querySelector('#query') as HTMLInputElement
+    search.value = 'xi'
+    search.dispatchEvent(new dom.window.Event('input'))
+
+    expect(dom.window.document.querySelectorAll('.message')).toHaveLength(1)
+    expect(dom.window.document.querySelector('.message')?.textContent).toContain(
+      'https://example.com/xi'
+    )
+    expect(dom.window.document.querySelectorAll('.search-highlight')).toHaveLength(1)
+    expect(dom.window.document.querySelector('#count')?.textContent).toBe(
+      '已显示 1 / 筛选 1 / 全部 2'
+    )
+    dom.window.close()
+  })
+
+  it('filters a v2 merged archive by conversation before search and month counts', () => {
+    const html = renderExportPage('合并档案')
+    const dom = new JSDOM(html, { runScripts: 'outside-only' })
+    Object.assign(dom.window, {
+      __WECHAT_EXPORT__: {
+        version: 2,
+        name: '合并档案',
+        exportedAt: '2026-08-04T00:00:00.000Z',
+        conversations: [
+          { id: 'alpha', name: '聊天 A', type: 'user', messageCount: 2 },
+          { id: 'beta', name: '聊天 B', type: 'group', messageCount: 1 }
+        ],
+        messages: [
+          messageForArchive('alpha-1', 'alpha', '聊天 A', '共同关键词', 1_767_225_600),
+          messageForArchive('beta-1', 'beta', '聊天 B', '共同关键词', 1_769_904_000),
+          messageForArchive('alpha-2', 'alpha', '聊天 A', '仅 A 可见', 1_769_990_400)
+        ]
+      }
+    })
+
+    dom.window.eval(inlineScriptOf(html))
+
+    const filter = dom.window.document.querySelector('#conversation-filter')!
+    const trigger = dom.window.document.querySelector('#conversation-trigger') as HTMLButtonElement
+    const menu = dom.window.document.querySelector('#conversation-menu') as HTMLElement
+    expect(filter.hasAttribute('hidden')).toBe(false)
+    expect(filter.parentElement?.classList.contains('archive-heading')).toBe(true)
+    expect((dom.window.document.querySelector('#archive-title') as HTMLElement).hidden).toBe(true)
+    expect(trigger.textContent).toContain('全部聊天')
+    expect(menu.querySelectorAll('[data-conversation-id]')).toHaveLength(3)
+    expect(dom.window.document.querySelectorAll('.conversation-source')).toHaveLength(3)
+    expect(dom.window.document.querySelector('#count')?.textContent).toBe(
+      '已显示 3 / 筛选 3 / 全部 3'
+    )
+    trigger.click()
+    ;(menu.querySelector('[data-conversation-id="alpha"]') as HTMLButtonElement).click()
+    expect(dom.window.document.querySelectorAll('.message')).toHaveLength(2)
+    expect(dom.window.document.querySelectorAll('.conversation-source')).toHaveLength(0)
+    expect(dom.window.document.querySelector('#count')?.textContent).toBe(
+      '已显示 2 / 筛选 2 / 当前聊天 2'
+    )
+    expect(dom.window.document.querySelectorAll('.timeline-month')).toHaveLength(2)
+
+    const search = dom.window.document.querySelector('#query') as HTMLInputElement
+    search.value = '共同关键词'
+    search.dispatchEvent(new dom.window.Event('input'))
+    expect(dom.window.document.querySelectorAll('.message')).toHaveLength(1)
+    expect(dom.window.document.querySelector('#count')?.textContent).toBe(
+      '已显示 1 / 筛选 1 / 当前聊天 2'
+    )
+    dom.window.close()
+  })
+
+  it('locates every filtered message kind in all messages, including outside the latest window', () => {
+    const html = renderExportPage('定位消息')
+    const dom = new JSDOM(html, { runScripts: 'outside-only' })
+    const categorized: Message[] = [
+      {
+        ...messageForArchive('target-text', 'fixture', '定位消息', '目标文字', 1),
+        type: '普通文本'
+      },
+      {
+        ...messageForArchive('target-media', 'fixture', '定位消息', '', 2),
+        type: '图片',
+        exportMediaType: 'image',
+        exportMediaUrl: 'media/target.jpg'
+      },
+      {
+        ...messageForArchive('target-voice', 'fixture', '定位消息', '', 3),
+        type: '语音',
+        voiceDataUrl: 'voices/target.wav'
+      },
+      {
+        ...messageForArchive('target-file', 'fixture', '定位消息', '', 4),
+        type: '文件',
+        exportMediaType: 'file',
+        exportMediaUrl: 'files/target.pdf'
+      },
+      {
+        ...messageForArchive('target-share', 'fixture', '定位消息', '', 5),
+        type: '分享',
+        contentData: { type: 'share', typeVal: '5', title: '目标分享' }
+      },
+      {
+        ...messageForArchive('target-system', 'fixture', '定位消息', '目标系统消息', 6),
+        from: 'system',
+        type: '系统消息',
+        contentData: { type: 'system', content: '目标系统消息' }
+      }
+    ]
+    const laterMessages = Array.from({ length: EXPORT_PAGE_SIZE }, (_, index) => ({
+      ...messageForArchive(
+        `later-${index}`,
+        'fixture',
+        '定位消息',
+        `稍后消息-${index}`,
+        100 + index
+      ),
+      type: '普通文本'
+    }))
+    Object.assign(dom.window, {
+      __WECHAT_EXPORT__: {
+        version: 1,
+        sourceId: 'fixture',
+        name: '定位消息',
+        messages: [...categorized, ...laterMessages]
+      }
+    })
+
+    dom.window.eval(inlineScriptOf(html))
+
+    expect(dom.window.document.querySelectorAll('.locate-all')).toHaveLength(0)
+    for (const kind of ['media', 'voice', 'file', 'share', 'system']) {
+      const filterButton = dom.window.document.querySelector(`[data-kind="${kind}"]`) as HTMLElement
+      filterButton.click()
+      const locateButton = dom.window.document.querySelector('.locate-all') as HTMLElement
+      expect(locateButton?.getAttribute('aria-label')).toBe('定位到聊天位置')
+      expect(locateButton?.querySelector('.locate-icon')?.textContent).toBe('⌖')
+      expect(locateButton?.querySelector('.locate-label')?.textContent).toBe('定位到聊天位置')
+      locateButton.click()
+      expect(
+        dom.window.document.querySelector('[data-kind="all"]')?.classList.contains('active')
+      ).toBe(true)
+      expect(
+        dom.window.document.querySelector('.message.located')?.getAttribute('data-index')
+      ).toBe(String(categorized.findIndex((message) => kindOfFixture(message) === kind)))
+    }
+
+    const textFilter = dom.window.document.querySelector('[data-kind="text"]') as HTMLElement
+    textFilter.click()
+    const search = dom.window.document.querySelector('#query') as HTMLInputElement
+    search.value = '目标文字'
+    search.dispatchEvent(new dom.window.Event('input'))
+    expect(dom.window.document.querySelector('.search-highlight')?.textContent).toBe('目标文字')
+    ;(dom.window.document.querySelector('.locate-all') as HTMLElement).click()
+    expect(
+      dom.window.document.querySelector('[data-kind="all"]')?.classList.contains('active')
+    ).toBe(true)
+    expect(dom.window.document.querySelector('.message.located')?.getAttribute('data-index')).toBe(
+      '0'
+    )
+    expect(dom.window.document.querySelector('.message.located')?.textContent).toContain('目标文字')
+
+    search.value = '稍后消息-137'
+    search.dispatchEvent(new dom.window.Event('input'))
+    expect(
+      dom.window.document.querySelector('[data-kind="all"]')?.classList.contains('active')
+    ).toBe(true)
+    expect(dom.window.document.querySelectorAll('.message')).toHaveLength(1)
+    expect(dom.window.document.querySelector('.search-highlight')?.textContent).toBe('稍后消息-137')
+    ;(dom.window.document.querySelector('.locate-all') as HTMLElement).click()
+    expect(search.value).toBe('')
+    expect(dom.window.document.querySelectorAll('.search-highlight')).toHaveLength(0)
+    expect(dom.window.document.querySelectorAll('.locate-all')).toHaveLength(0)
+    expect(dom.window.document.querySelector('.message.located')?.textContent).toContain(
+      '稍后消息-137'
+    )
    dom.window.close()
  })

@@ -158,6 +383,36 @@ describe('export media', () => {
    expect(html).toContain('aria-label="关闭图片预览"')
    expect(html).toContain("closeButton.addEventListener('click', closeLightbox)")
    expect(html).toContain('if (event.target === box) closeLightbox()')
-    expect(html).toContain("if (event.key === 'Escape') closeLightbox()")
+    expect(html).toContain("if (event.key === 'Escape')")
+    expect(html).toContain('closeLightbox()')
  })
 })
+
+function messageForArchive(
+  id: string,
+  conversationId: string,
+  conversationName: string,
+  content: string,
+  createTime: number
+): Message {
+  return {
+    id,
+    from: 'user',
+    type: '普通文本',
+    datetime: '',
+    content,
+    isSender: false,
+    createTime,
+    exportConversationId: conversationId,
+    exportConversationName: conversationName
+  }
+}
+
+function kindOfFixture(message: Message): string {
+  if (message.exportMediaType === 'image') return 'media'
+  if (message.voiceDataUrl) return 'voice'
+  if (message.exportMediaType === 'file') return 'file'
+  if (message.contentData?.type === 'share') return 'share'
+  if (message.contentData?.type === 'system') return 'system'
+  return 'text'
+}
@@ -28,6 +28,34 @@ describe('message parser', () => {
    }
  })

+  it.each([
+    ['6', '测试附件.pdf'],
+    ['74', '发送中的附件.zip']
+  ])(
+    'keeps file app message type %s when attachment metadata contains record tags',
+    (typeVal, title) => {
+      const parsed = parseMessageContent(
+        `<appmsg><type>${typeVal}</type><title>${title}</title><des>1 MB</des><appattach><recorditem>legacy metadata</recorditem><dataitem datatype="8"><datatitle>${title}</datatitle></dataitem></appattach></appmsg>`,
+        49
+      )
+
+      expect(parsed).toMatchObject({
+        type: 'share',
+        title,
+        typeVal
+      })
+    }
+  )
+
+  it('does not classify empty incidental record metadata as a merged forward', () => {
+    const parsed = parseMessageContent(
+      '<appmsg><type>5</type><title>普通分享</title><recorditem>legacy metadata</recorditem></appmsg>',
+      49
+    )
+
+    expect(parsed).toMatchObject({ type: 'share', title: '普通分享', typeVal: '5' })
+  })
+
  it('uses the quoted group member id instead of the chatroom id', () => {
    const parsed = parseMessageContent(
      '<appmsg><type>57</type><title>回复内容</title><refermsg><type>1</type><fromusr>123456789@chatroom</fromusr><chatusr>wxid_fixture_member</chatusr><content>被引用内容</content></refermsg></appmsg>',
@@ -0,0 +1,78 @@
+import crypto from 'crypto'
+import { mkdirSync, mkdtempSync, writeFileSync } from 'fs'
+import { tmpdir } from 'os'
+import { join } from 'path'
+import { beforeAll, describe, expect, it, vi } from 'vitest'
+import type { Message } from '../../src/shared/types'
+
+const state = vi.hoisted(() => ({ userData: '' }))
+
+vi.mock('electron', () => ({
+  app: { getPath: () => state.userData }
+}))
+
+const message = (createTime: number, serverId?: string): Message => ({
+  id: `message-${createTime}`,
+  from: 'user',
+  isSender: false,
+  type: '普通文本',
+  datetime: new Date(createTime * 1000).toISOString(),
+  content: String(createTime),
+  img: '',
+  name: 'Jamie',
+  sessionId: 'fixture-user',
+  localId: 1,
+  serverId,
+  createTime
+})
+
+describe('recall archive message identity', () => {
+  beforeAll(() => {
+    state.userData = mkdtempSync(join(tmpdir(), 'wxe-recall-identity-'))
+  })
+
+  it('does not collapse messages whose local ids repeat across database shards', async () => {
+    const accountRoot = '/fixture/account'
+    const sessionMd5 = 'fixture-session'
+    const archiveDir = join(state.userData, 'recall-archive')
+    const archiveName = crypto
+      .createHash('sha1')
+      .update(`${process.platform}:${accountRoot}`)
+      .digest('hex')
+      .slice(0, 16)
+    mkdirSync(archiveDir, { recursive: true })
+    writeFileSync(
+      join(archiveDir, `${archiveName}.json`),
+      JSON.stringify({
+        version: 1,
+        accountRoot,
+        updatedAt: Date.now(),
+        sessions: {
+          [sessionMd5]: {
+            username: 'fixture-user',
+            updatedAt: Date.now(),
+            messages: [],
+            recalls: []
+          }
+        }
+      })
+    )
+    const { configureRecallArchive, mergeRecallArchiveMessages, messageIdentity } =
+      await import('../../src/main/services/recall-archive-service')
+    configureRecallArchive(accountRoot)
+    const oldMessage = message(1_731_327_263)
+    const newMessage = message(1_765_000_000)
+
+    expect(messageIdentity(oldMessage)).not.toBe(messageIdentity(newMessage))
+    expect(mergeRecallArchiveMessages(sessionMd5, [oldMessage, newMessage])).toEqual([
+      oldMessage,
+      newMessage
+    ])
+  })
+
+  it('prefers the globally unique server id when one is available', async () => {
+    const { messageIdentity } = await import('../../src/main/services/recall-archive-service')
+
+    expect(messageIdentity(message(1_731_327_263, 'server-2024'))).toBe('server:server-2024')
+  })
+})
@@ -5,9 +5,16 @@ import { dirname, join, resolve } from 'path'
 import { afterAll, describe, expect, it } from 'vitest'

 const nodeRequire = createRequire(import.meta.url)
-const { validateFfmpegRuntime, validateSherpaRuntime, validateSilkWasmRuntime } = nodeRequire(
-  '../../scripts/after-pack.cjs'
-) as {
+const asar = nodeRequire('@electron/asar') as {
+  createPackage: (source: string, destination: string) => Promise<void>
+}
+const {
+  validateAsarRuntimeDependencies,
+  validateFfmpegRuntime,
+  validateSherpaRuntime,
+  validateSilkWasmRuntime
+} = nodeRequire('../../scripts/after-pack.cjs') as {
+  validateAsarRuntimeDependencies: (runtimeResources: string) => void
  validateFfmpegRuntime: (runtimeResources: string, platform?: NodeJS.Platform) => void
  validateSherpaRuntime: (runtimeResources: string, platform: NodeJS.Platform, arch: string) => void
  validateSilkWasmRuntime: (runtimeResources: string) => void
@@ -33,6 +40,19 @@ describe('production runtime packaging', () => {
    expect(config).toContain('node_modules/silk-wasm/**')
  })

+  it('rejects an app archive with missing runtime dependencies', async () => {
+    const resources = join(root, 'asar-resources')
+    const source = join(root, 'asar-source')
+    mkdirSync(source, { recursive: true })
+    writeFileSync(join(source, 'package.json'), '{}')
+    mkdirSync(resources, { recursive: true })
+    await asar.createPackage(source, join(resources, 'app.asar'))
+
+    expect(() => validateAsarRuntimeDependencies(resources)).toThrow(
+      /Missing packaged runtime dependencies:.*@electron-toolkit\/utils/
+    )
+  })
+
  it('requires and unpacks the bundled ffmpeg-static executable', () => {
    const resources = join(root, 'ffmpeg-resources')
    const ffmpegPath = join(
@@ -0,0 +1,140 @@
+import { createHash } from 'crypto'
+import { mkdirSync, mkdtempSync, rmSync, writeFileSync } from 'fs'
+import { tmpdir } from 'os'
+import { join } from 'path'
+import { afterEach, describe, expect, it } from 'vitest'
+import { VideoAssetService } from '../../src/main/video-asset-service'
+
+const temporaryDirectories: string[] = []
+
+const box = (type: string, payload: Buffer): Buffer => {
+  const header = Buffer.alloc(8)
+  header.writeUInt32BE(header.length + payload.length, 0)
+  header.write(type, 4, 4, 'ascii')
+  return Buffer.concat([header, payload])
+}
+
+const mp4Fixture = (durationSeconds: number, marker: string): Buffer => {
+  const movieHeader = Buffer.alloc(20)
+  movieHeader.writeUInt32BE(1000, 12)
+  movieHeader.writeUInt32BE(Math.round(durationSeconds * 1000), 16)
+  return Buffer.concat([
+    box('ftyp', Buffer.from('isom0000', 'ascii')),
+    box('moov', box('mvhd', movieHeader)),
+    box('mdat', Buffer.from(marker, 'utf8'))
+  ])
+}
+
+const jpegFixture = (width: number, height: number): Buffer =>
+  Buffer.from([
+    0xff,
+    0xd8,
+    0xff,
+    0xc0,
+    0x00,
+    0x11,
+    0x08,
+    (height >> 8) & 0xff,
+    height & 0xff,
+    (width >> 8) & 0xff,
+    width & 0xff,
+    0x03,
+    0x01,
+    0x11,
+    0x00,
+    0x02,
+    0x11,
+    0x00,
+    0x03,
+    0x11,
+    0x00,
+    0xff,
+    0xd9
+  ])
+
+const createService = (): {
+  accountRoot: string
+  service: VideoAssetService
+} => {
+  const accountRoot = mkdtempSync(join(tmpdir(), 'wxe-video-assets-'))
+  temporaryDirectories.push(accountRoot)
+  return {
+    accountRoot,
+    service: new VideoAssetService({
+      getAccountRoot: () => accountRoot,
+      resolveVideoHardlink: () => null
+    } as never)
+  }
+}
+
+const monthTimestamp = (year: number, month: number): number =>
+  Math.floor(new Date(year, month - 1, 15, 12).getTime() / 1000)
+
+afterEach(() => {
+  for (const directory of temporaryDirectories.splice(0)) {
+    rmSync(directory, { recursive: true, force: true })
+  }
+})
+
+describe('VideoAssetService local fallback', () => {
+  it('finds a video by its content MD5 when the hardlink mapping is absent', async () => {
+    const { accountRoot, service } = createService()
+    const month = join(accountRoot, 'msg', 'video', '2026-07')
+    mkdirSync(month, { recursive: true })
+    const content = mp4Fixture(22, 'content-md5-match')
+    const filePath = join(month, `${'2'.repeat(32)}.mp4`)
+    writeFileSync(filePath, content)
+    const contentHash = createHash('md5').update(content).digest('hex')
+
+    const result = await service.resolve([contentHash], {
+      createTime: monthTimestamp(2026, 7)
+    })
+
+    expect(result.success).toBe(true)
+    expect(service.pathForUrl(result.url!)).toBe(filePath)
+  })
+
+  it('finds a uniquely matching video by month, thumbnail size, and duration', async () => {
+    const { accountRoot, service } = createService()
+    const month = join(accountRoot, 'msg', 'video', '2025-11')
+    mkdirSync(month, { recursive: true })
+    const stem = '66cecd68e095d87175fb5ed138de3cef'
+    const filePath = join(month, `${stem}.mp4`)
+    const posterPath = join(month, `${stem}_thumb.jpg`)
+    writeFileSync(filePath, mp4Fixture(68.441, 'metadata-match'))
+    writeFileSync(posterPath, jpegFixture(279, 630))
+
+    const result = await service.resolve(
+      ['c92c54c8eae4471be9cc18396daf8015', '021e8a18a765ce14f4c54f40065db98e'],
+      {
+        createTime: monthTimestamp(2025, 11),
+        duration: 68,
+        width: 279,
+        height: 630
+      }
+    )
+
+    expect(result.success).toBe(true)
+    expect(service.pathForUrl(result.url!)).toBe(filePath)
+    expect(service.pathForUrl(result.poster!)).toBe(posterPath)
+  })
+
+  it('does not guess when multiple files match the same metadata', async () => {
+    const { accountRoot, service } = createService()
+    const month = join(accountRoot, 'msg', 'video', '2026-02')
+    mkdirSync(month, { recursive: true })
+    for (const stem of ['a'.repeat(32), 'b'.repeat(32)]) {
+      writeFileSync(join(month, `${stem}.mp4`), mp4Fixture(12, stem))
+      writeFileSync(join(month, `${stem}_thumb.jpg`), jpegFixture(224, 398))
+    }
+
+    const result = await service.resolve(['c'.repeat(32)], {
+      createTime: monthTimestamp(2026, 2),
+      duration: 12,
+      width: 224,
+      height: 398
+    })
+
+    expect(result).toEqual({ success: false, error: '本地未找到该视频文件' })
+  })
+})
@@ -1,9 +1,9 @@
 import { describe, expect, it, vi } from 'vitest'
 import { Wcdb4Client, type Wcdb4Message } from '../../src/main/wcdb4-client'

-const message = (id: string, year: number): Wcdb4Message => ({
+const message = (id: string, year: number, serverId = `server-${id}`): Wcdb4Message => ({
  mesLocalID: id,
-  serverId: `server-${id}`,
+  serverId,
  mesDes: 0,
  messageType: '1',
  msgCreateTime: String(Math.floor(Date.UTC(year, 0, 1) / 1000)),
@@ -12,6 +12,21 @@ const message = (id: string, year: number): Wcdb4Message => ({
 })

 describe('WCDB message shard pagination', () => {
+  it('keeps messages whose local ids repeat across database shards', async () => {
+    const cursor = vi.fn(async () => [
+      message('1', 2024, 'server-2024'),
+      message('1', 2025, 'server-2025')
+    ])
+    const client = Object.assign(Object.create(Wcdb4Client.prototype), {
+      getMessagesByCursorAsync: cursor
+    }) as Wcdb4Client
+
+    const result = await client.getMessagesAsync('fixture@chatroom')
+
+    expect(result).toHaveLength(2)
+    expect(result.map((item) => item.serverId)).toEqual(['server-2024', 'server-2025'])
+  })
+
  it('merges cursor and all-store rows for a bounded cross-year page', async () => {
    const cursor = vi.fn(async () => [message('2025', 2025)])
    const tableScan = vi.fn(async () => [message('2017', 2017), message('2025', 2025)])
@@ -0,0 +1,71 @@
+import { describe, expect, it, vi } from 'vitest'
+import { WechatDb, type WechatMessage } from '../../src/main/wechat-db'
+
+describe('WechatDb normalized messages', () => {
+  it('keeps normalized identity fields when raw table columns conflict', async () => {
+    const normalized = {
+      mesLocalID: '1',
+      serverId: 'server-1',
+      mesDes: 0,
+      messageType: '1',
+      msgCreateTime: '1731327263',
+      msgContent: 'fixture',
+      sender: 'wxid_self',
+      senderNickname: 'Nanin',
+      raw: {
+        mesDes: 1,
+        sender: '',
+        senderNickname: ''
+      }
+    }
+    const client = { getMessagesAsync: vi.fn(async () => [normalized]) }
+    const db = Object.assign(Object.create(WechatDb.prototype), {
+      wcdb4Client: client,
+      chatMd5ToUsername: new Map([['fixture-md5', 'fixture-user']]),
+      ensureChatTableMapping: vi.fn()
+    }) as WechatDb
+
+    const start = Math.floor(new Date(2024, 10, 11).getTime() / 1000)
+    const end = Math.floor(new Date(2024, 11, 1).getTime() / 1000)
+    const messages = await db.getUserMessagesForExport('fixture-md5', start, end)
+
+    expect(messages[0]).toMatchObject({
+      mesDes: 0,
+      sender: 'wxid_self',
+      senderNickname: 'Nanin'
+    })
+  })
+
+  it('scans without time bounds, then filters, deduplicates and sorts in application code', async () => {
+    const row = (id: string, createTime: number): WechatMessage => ({
+      mesLocalID: id,
+      serverId: `server-${id}`,
+      mesDes: 0,
+      messageType: '1',
+      msgCreateTime: String(createTime),
+      msgContent: id,
+      raw: {}
+    })
+    const start = Math.floor(new Date(2025, 0, 1).getTime() / 1000)
+    const end = Math.floor(new Date(2025, 0, 4).getTime() / 1000)
+    const shardBoundaryMessage = row('jan-2-boundary', start + 32 * 60 * 60)
+    const getMessagesAsync = vi.fn(async () => [
+      row('before-range', start - 1),
+      row('newest', start + 48 * 60 * 60),
+      shardBoundaryMessage,
+      { ...shardBoundaryMessage },
+      row('after-range', end + 1)
+    ])
+    const db = Object.assign(Object.create(WechatDb.prototype), {
+      wcdb4Client: { getMessagesAsync },
+      chatMd5ToUsername: new Map([['fixture-md5', 'fixture-user']]),
+      ensureChatTableMapping: vi.fn()
+    }) as WechatDb
+
+    const messages = await db.getUserMessagesForExport('fixture-md5', start, end)
+
+    expect(getMessagesAsync).toHaveBeenCalledOnce()
+    expect(getMessagesAsync).toHaveBeenCalledWith('fixture-user')
+    expect(messages.map((message) => message.mesLocalID)).toEqual(['jan-2-boundary', 'newest'])
+  })
+})