import OpenAI from 'openai';
import fs from 'fs';

const openai = new OpenAI({ apiKey: process.env.OPENAI_API_KEY });

async function transcribe(audioPath: string): Promise<string> {
  const transcription = await openai.audio.transcriptions.create({
    file: fs.createReadStream(audioPath),
    model: 'whisper-1',
    language: 'ja', // 日本語
    response_format: 'verbose_json', // タイムスタンプ付き
  });
  return transcription.text;
}

interface TranscriptionChunk {
  start: number;
  end: number;
  text: string;
  speaker?: string;
}

async function transcribeChunk(
  audioBlob: Blob,
  chunkIndex: number
): Promise<TranscriptionChunk[]> {
  const formData = new FormData();
  formData.append('file', audioBlob, `chunk-${chunkIndex}.webm`);
  formData.append('model', 'whisper-1');
  formData.append('response_format', 'verbose_json');
  formData.append('timestamp_granularities[]', 'segment');

  const response = await fetch('https://api.openai.com/v1/audio/transcriptions', {
    method: 'POST',
    headers: { Authorization: `Bearer ${process.env.OPENAI_API_KEY}` },
    body: formData,
  });

  const result = await response.json();
  return result.segments.map((seg: any) => ({
    start: seg.start + chunkIndex * 5, // 5秒チャンクのオフセット
    end: seg.end + chunkIndex * 5,
    text: seg.text,
  }));
}

サンプルレート: 16kHz（Whisper推奨）
チャンネル: Mono
フォーマット: WebM/Opus または WAV
ファイルサイズ: 最大25MB

async function transcribeWithRetry(
  audioPath: string,
  maxRetries = 3
): Promise<string> {
  for (let i = 0; i < maxRetries; i++) {
    try {
      return await transcribe(audioPath);
    } catch (error: any) {
      if (error.status === 429) {
        // Rate limit - exponential backoff
        await new Promise(r => setTimeout(r, Math.pow(2, i) * 1000));
        continue;
      }
      throw error;
    }
  }
  throw new Error('Max retries exceeded');
}