Function preprocessQQJson

packages/parser/src/formats/shuakami-qq-preprocessor.ts:211–351

* 预处理 QQ JSON 文件

(inputPath: string, onProgress?: (progress: ParseProgress) => void)

 * 预处理 QQ JSON 文件
 */
async function preprocessQQJson(inputPath: string, onProgress?: (progress: ParseProgress) => void): Promise<string> {
  const totalBytes = getFileSize(inputPath)
  let bytesRead = 0
  let messagesProcessed = 0

  const tempDir = getTempDir()
  ensureDir(tempDir)
  const outputFilename = `slim_${Date.now()}_${path.basename(inputPath)}`
  const outputPath = path.join(tempDir, outputFilename)

  onProgress?.(createProgress('parsing', 0, totalBytes, 0, ''))

  // 先从原文件读取 avatars（因为它在文件末尾，消息处理时可能无法访问）
  const avatarsStr = readAvatarsFromFile(inputPath)

  return new Promise((resolve, reject) => {
    const headChunks: string[] = []
    let headSize = 0
    const maxHeadSize = 100000

    const headStream = fs.createReadStream(inputPath, { encoding: 'utf-8' })
    let chatInfo: Record<string, unknown> = { name: '未知群聊', type: 'group' }
    let metadata: Record<string, unknown> | undefined
    let statistics: Record<string, unknown> | undefined

    headStream.on('data', (chunk: string | Buffer) => {
      const str = typeof chunk === 'string' ? chunk : chunk.toString('utf-8')
      if (headSize < maxHeadSize) {
        headChunks.push(str)
        headSize += str.length
      } else {
        headStream.destroy()
      }
    })

    headStream.on('close', () => {
      const headContent = headChunks.join('')

      try {
        const chatInfoMatch = headContent.match(/"chatInfo"\s*:\s*(\{[^}]+\})/)
        if (chatInfoMatch) {
          chatInfo = JSON.parse(chatInfoMatch[1])
        }
      } catch {
        // 使用默认值
      }

      try {
        const metadataMatch = headContent.match(/"metadata"\s*:\s*(\{[^}]+\})/)
        if (metadataMatch) {
          metadata = JSON.parse(metadataMatch[1])
        }
      } catch {
        // 忽略
      }

      try {
        const statisticsMatch = headContent.match(/"statistics"\s*:\s*(\{[\s\S]*?\})\s*,\s*"messages"/)