feat: wire favorites read path, AMR ffmpeg decode, voice cache/VoiceTemp walk

2026-10-04 03:01:42 +08:00 · 2026-09-24 21:27:48 +08:00
parent 784bed0ee0
commit 19bf497af0
6 changed files with 247 additions and 8 deletions
@@ -0,0 +1,43 @@
+import type { Message, ParsedContent } from '../shared/types'
+import { favoriteRowToContent } from '../shared/favorites'
+import type { Wcdb4Client } from './wcdb4-client'
+
+/** 收藏只读服务：favorite.db → 可读卡片 / 导出消息。 */
+export class FavoritesService {
+  constructor(private readonly wcdb4Client: Wcdb4Client) {}
+
+  async listContents(limit = 200): Promise<ParsedContent[]> {
+    const rows = await this.wcdb4Client.listFavoriteItems(limit)
+    return rows.map((row) => favoriteRowToContent(row))
+  }
+
+  /** 收藏转成导出/搜索可用的 Message 列表（合成会话「收藏」）。 */
+  async listExportMessages(limit = 200): Promise<Message[]> {
+    const rows = await this.wcdb4Client.listFavoriteItems(limit)
+    return rows.map((row, index) => {
+      const contentData = favoriteRowToContent(row)
+      const createTime = Number(row.update_time) || 0
+      return {
+        id: `fav-${row.local_id ?? index}`,
+        from: 'favorite',
+        type: '收藏',
+        datetime: createTime ? new Date(createTime * 1000).toISOString() : '',
+        content: textOf(contentData),
+        isSender: false,
+        createTime,
+        name: '收藏',
+        contentData
+      } as Message
+    })
+  }
+}
+
+function textOf(content: ParsedContent): string {
+  if (content.type === 'text') return content.content
+  if (content.type === 'system') return content.content
+  if (content.type === 'share') return [content.title, content.des].filter(Boolean).join(' · ')
+  if (content.type === 'location') return content.poiname || content.label || '[位置]'
+  if (content.type === 'miniProgram') return content.title || '[小程序]'
+  if (content.type === 'forwardBundle') return content.title || '[聊天记录]'
+  return '[收藏]'
+}
@@ -126,5 +126,54 @@ export class AudioDecoderRegistry {
 }

 export function createDefaultAudioDecoderRegistry(): AudioDecoderRegistry {
-  return new AudioDecoderRegistry().register(new SilkAudioDecoder())
+  return new AudioDecoderRegistry()
+    .register(new SilkAudioDecoder())
+    .register(new AmrAudioDecoder())
+}
+
+/**
+ * V3 AMR：外部 `ffmpeg` 解码为 PCM（MultiMediaDyn `CAMRDecoder` 未导出）。
+ */
+export class AmrAudioDecoder implements VoiceAudioDecoder {
+  readonly codec = 'amr'
+
+  async decode(source: EncodedVoiceSource): Promise<DecodedVoiceAudio> {
+    const { spawn } = await import('child_process')
+    const { tmpdir } = await import('os')
+    const { promises: fs } = await import('fs')
+    const { randomBytes } = await import('crypto')
+    const stamp = randomBytes(8).toString('hex')
+    const inPath = join(tmpdir(), `tracememo-amr-in-${stamp}.amr`)
+    const outPath = join(tmpdir(), `tracememo-amr-out-${stamp}.wav`)
+    await fs.writeFile(inPath, source.data)
+    try {
+      await new Promise<void>((resolve, reject) => {
+        const child = spawn(
+          'ffmpeg',
+          ['-y', '-i', inPath, '-f', 'wav', '-acodec', 'pcm_s16le', '-ar', '8000', '-ac', '1', outPath],
+          { stdio: ['ignore', 'ignore', 'pipe'] }
+        )
+        let err = ''
+        child.stderr.on('data', (chunk) => {
+          err += String(chunk)
+        })
+        child.on('error', reject)
+        child.on('close', (code) => {
+          if (code === 0) resolve()
+          else reject(new Error(`ffmpeg amr decode failed: ${err.slice(-200)}`))
+        })
+      })
+      const wav = await fs.readFile(outPath)
+      const pcm = wav.length > 44 && wav.toString('ascii', 0, 4) === 'RIFF' ? wav.subarray(44) : wav
+      return {
+        pcm,
+        sampleRate: 8000,
+        channels: 1,
+        sourceHash: source.sourceHash
+      }
+    } finally {
+      await fs.rm(inPath, { force: true })
+      await fs.rm(outPath, { force: true })
+    }
+  }
 }
@@ -328,25 +328,41 @@ const VOICE_DISK_DIRS = [
  'msg/History',
  'Message',
  'MsgAndFiles',
-  'VoiceTemp'
+  'VoiceTemp',
+  // 真机布局（2026-09-24）：cache/YYYY-MM/Message/<md5>/VoiceTemp/
+  'cache'
 ]

-/** 语音磁盘候选路径（扩展名/目录与 GetMsgAudioPath 对齐）。 */
+/** 语音磁盘候选路径（扩展名/目录与 GetMsgAudioPath / VoiceTemp 对齐）。 */
 export async function listVoiceDiskCandidates(accountRoot: string): Promise<string[]> {
  const out: string[] = []
-  for (const rel of VOICE_DISK_DIRS) {
-    const dir = join(accountRoot, rel)
+  const walk = async (dir: string, depth: number): Promise<void> => {
+    if (out.length > 200 || depth > 6) return
    let names: string[]
    try {
      names = await fs.readdir(dir)
    } catch {
-      continue
+      return
    }
    for (const name of names) {
+      const full = join(dir, name)
+      let stat: Awaited<ReturnType<typeof fs.stat>>
+      try {
+        stat = await fs.stat(full)
+      } catch {
+        continue
+      }
+      if (stat.isDirectory()) {
+        await walk(full, depth + 1)
+        continue
+      }
      const ext = name.slice(name.lastIndexOf('.')).toLowerCase()
      if (!VOICE_DISK_EXTS.has(ext)) continue
-      out.push(join(dir, name))
+      out.push(full)
    }
+  }
+  for (const rel of VOICE_DISK_DIRS) {
+    await walk(join(accountRoot, rel), 0)
    if (out.length > 200) break
  }
  return out
@@ -2678,6 +2678,27 @@ export class Wcdb4Client {
    return this.accountRoot
  }

+  /** 只读列收藏（favorite.db / fav_db_item），供导出与搜索。 */
+  async listFavoriteItems(limit = 200): Promise<Record<string, unknown>[]> {
+    if (!this.wcdbExecQuery) return []
+    const sql = `SELECT local_id, server_id, type, update_time, content, fromusr, realchatname FROM fav_db_item ORDER BY update_time DESC LIMIT ${Math.max(
+      1,
+      Math.min(1000, limit)
+    )}`
+    try {
+      const rows = await this.callJsonAsync<Record<string, unknown>[]>(
+        this.wcdbExecQuery as unknown as KoffiAsyncFunction,
+        'favorite',
+        path.join(this.accountRoot, 'db_storage/favorite/favorite.db'),
+        sql
+      )
+      return Array.isArray(rows) ? rows : []
+    } catch (error) {
+      console.warn('[WCDB4] listFavoriteItems failed:', error)
+      return []
+    }
+  }
+
  getKey(): string {
    return this.key
  }
@@ -1,5 +1,79 @@
 import type { ParsedContent } from './types'

+function xmlValue(xml: string, tag: string): string | undefined {
+  const match = new RegExp(`<${tag}[^>]*>([\\s\\S]*?)<\\/${tag}>`, 'i').exec(xml)
+  const raw = match?.[1]?.replace(/^<!\[CDATA\[([\s\S]*?)\]\]>$/, '$1').trim()
+  return raw || undefined
+}
+
+function decodeXmlEntities(value: string): string {
+  return value
+    .replace(/&lt;/g, '<')
+    .replace(/&gt;/g, '>')
+    .replace(/&quot;/g, '"')
+    .replace(/&apos;/g, "'")
+    .replace(/&amp;/g, '&')
+}
+
+/**
+ * 解析 `fav_db_item.content` 的 `<favitem>` XML → `FavoriteRecord`（只读）。
+ * 覆盖真机 2026-09-24 样本形态：desc/title/link/locitem/dataitem/appbranditem/finderFeed。
+ */
+export function parseFavItemXml(content: string | null | undefined): FavoriteRecord {
+  const xml = String(content || '')
+  const typeMatch = /<favitem[^>]*\stype="(\d+)"/i.exec(xml)
+  const type = typeMatch?.[1]
+  const title = decodeXmlEntities(xmlValue(xml, 'title') || '')
+  const desc = decodeXmlEntities(xmlValue(xml, 'desc') || xmlValue(xml, 'datadesc') || '')
+  const datatitle = decodeXmlEntities(xmlValue(xml, 'datatitle') || '')
+  const link = decodeXmlEntities(xmlValue(xml, 'link') || xmlValue(xml, 'datacdnurl') || '')
+  const poiname = decodeXmlEntities(xmlValue(xml, 'poiname') || '')
+  const label = decodeXmlEntities(xmlValue(xml, 'label') || '')
+  const lat = Number(xmlValue(xml, 'lat'))
+  const lng = Number(xmlValue(xml, 'lng'))
+  const dataid = xmlValue(xml, 'dataid')
+  const datatype = xmlValue(xml, 'datatype')
+  const duration = Number(xmlValue(xml, 'duration') || xmlValue(xml, 'voicelength'))
+  const appname =
+    decodeXmlEntities(
+      xmlValue(xml, 'appbrandnick') ||
+        xmlValue(xml, 'appname') ||
+        xmlValue(xml, 'sourcedisplayname') ||
+        ''
+    ) || undefined
+  return {
+    type,
+    dataType: datatype,
+    title: title || datatitle || undefined,
+    description: desc || undefined,
+    url: link || undefined,
+    appName: appname,
+    poiname: poiname || undefined,
+    label: label || undefined,
+    lat: Number.isFinite(lat) ? lat : undefined,
+    lng: Number.isFinite(lng) ? lng : undefined,
+    md5: dataid,
+    duration: Number.isFinite(duration) && duration > 0 ? duration : undefined,
+    raw: xml
+  }
+}
+
+/** `fav_db_item` 行 → 可读卡片（导出/搜索共用）。 */
+export function favoriteRowToContent(row: {
+  local_id?: number | string
+  server_id?: number | string
+  type?: number | string
+  content?: string | null
+  update_time?: number | string
+  fromusr?: string | null
+  realchatname?: string | null
+}): ParsedContent {
+  const record = parseFavItemXml(row.content)
+  if (!record.type && row.type !== undefined) record.type = row.type
+  record.favId = row.local_id ?? row.server_id
+  return favoriteRecordToContent(record)
+}
+
 /**
 * 收藏类型（`MM_FAV_ITEM_TYPE_*` / `MM_FAV_DATA_TYPE_*`）只读映射。
 * 名称来自 wechat.dylib 导出名；数字为枚举顺序推断，未知值原样回退。
@@ -1,5 +1,10 @@
 import { describe, expect, it } from 'vitest'
-import { describeFavoriteType, favoriteRecordToContent } from '../../src/shared/favorites'
+import {
+  describeFavoriteType,
+  favoriteRecordToContent,
+  favoriteRowToContent,
+  parseFavItemXml
+} from '../../src/shared/favorites'

 describe('favorites type map', () => {
  it('labels known MM_FAV names and numeric fallbacks', () => {
@@ -57,4 +62,35 @@ describe('favorites type map', () => {
      typeVal: 'note'
    })
  })
+
+  it('parses favitem XML from favorite.db content', () => {
+    const xml = [
+      '<favitem type="1"><ctrlflag>0</ctrlflag><version>0</version>',
+      '<desc><![CDATA[收藏文字]]></desc>',
+      '<source sourcetype="22"><fromusr>wxid_a</fromusr></source>',
+      '</favitem>'
+    ].join('')
+    expect(parseFavItemXml(xml)).toMatchObject({
+      type: '1',
+      description: '收藏文字'
+    })
+    const web = [
+      '<favitem type="5"><source><link><![CDATA[https://example.com/a]]></link></source>',
+      '<datalist count="1"><dataitem datatype="5"><datatitle><![CDATA[标题]]></datatitle>',
+      '<datadesc><![CDATA[摘要]]></datadesc></dataitem></datalist></favitem>'
+    ].join('')
+    expect(parseFavItemXml(web)).toMatchObject({
+      type: '5',
+      title: '标题',
+      description: '摘要',
+      url: 'https://example.com/a'
+    })
+    expect(
+      favoriteRowToContent({
+        local_id: 9,
+        type: 5,
+        content: web
+      })
+    ).toMatchObject({ type: 'share', title: '标题', typeVal: '5' })
+  })
 })