Function parseInstagram

packages/parser/src/formats/instagram-native.ts:226–349

(options: ParseOptions)

// ==================== 解析器实现 ====================

async function* parseInstagram(options: ParseOptions): AsyncGenerator<ParseEvent, void, unknown> {
  const { filePath, batchSize = 5000, onProgress, onLog } = options

  const totalBytes = getFileSize(filePath)
  let messagesProcessed = 0

  // 发送初始进度
  const initialProgress = createProgress('parsing', 0, totalBytes, 0, '正在解析 Instagram 聊天记录...')
  yield { type: 'progress', data: initialProgress }
  onProgress?.(initialProgress)

  onLog?.('info', `开始解析 Instagram 聊天记录，大小: ${(totalBytes / 1024 / 1024).toFixed(2)} MB`)

  // 读取并解析 JSON 文件
  let data: InstagramData
  try {
    const content = fs.readFileSync(filePath, 'utf-8')
    data = JSON.parse(content)
  } catch (error) {
    const err = new Error(`无法解析 Instagram JSON 文件: ${error}`)
    yield { type: 'error', data: err }
    return
  }

  // 判断聊天类型
  const isGroup = data.participants.length > 2 || !!data.joinable_mode
  const chatType = isGroup ? ChatType.GROUP : ChatType.PRIVATE

  // 判断 Owner
  let ownerId: string | undefined
  if (chatType === ChatType.PRIVATE) {
    // 私聊：title 是对方名字，另一个参与者是 owner
    const owner = data.participants.find((p) => decodeInstagramText(p.name) !== decodeInstagramText(data.title))
    ownerId = owner ? decodeInstagramText(owner.name) : undefined
  } else {
    // 群聊：找 "You created the group." 消息的发送者
    const createMsg = data.messages.find((m) => m.content === 'You created the group.')
    ownerId = createMsg ? decodeInstagramText(createMsg.sender_name) : undefined
  }

  // 发送 meta
  const meta: ParsedMeta = {
    name: decodeInstagramText(data.title) || extractNameFromFilePath(filePath),
    platform: KNOWN_PLATFORMS.INSTAGRAM,
    type: chatType,
    ownerId,
  }
  yield { type: 'meta', data: meta }

  // 收集成员信息
  const memberMap = new Map<string, ParsedMember>()
  for (const participant of data.participants) {
    const name = decodeInstagramText(participant.name)
    memberMap.set(name, {
      platformId: name,
      accountName: name,
    })
  }