Function handler

app/api/transcript/route.ts:30–232

(request: NextRequest)

}

async function handler(request: NextRequest) {
  try {
    const { url, lang, expectedDuration } = await request.json();

    if (!url) {
      return respondWithNoCredits({ error: 'YouTube URL is required' }, 400);
    }

    const videoId = extractVideoId(url);

    if (!videoId) {
      return respondWithNoCredits({ error: 'Invalid YouTube URL' }, 400);
    }

    if (shouldUseMockData()) {
      console.log(
        '[TRANSCRIPT] Using mock data (NEXT_PUBLIC_USE_MOCK_DATA=true)'
      );
      const mockData = getMockTranscript();

      const rawSegments = mockData.content.map((item: any) => ({
        text: item.text,
        start: item.offset / 1000, // Convert milliseconds to seconds
        duration: item.duration / 1000 // Convert milliseconds to seconds
      }));

      // Merge segments into complete sentences for better translation
      const mergedSentences = mergeTranscriptSegmentsIntoSentences(rawSegments);
      const transformedTranscript = mergedSentences.map((sentence) => ({
        text: sentence.text,
        start: sentence.segments[0].start, // Use first segment's start time
        duration: sentence.segments.reduce((sum, seg) => sum + seg.duration, 0) // Sum all durations
      }));

      const transcriptDuration = rawSegments.length > 0
        ? rawSegments[rawSegments.length - 1].start + rawSegments[rawSegments.length - 1].duration
        : 0;

      return NextResponse.json({
        videoId,
        transcript: transformedTranscript,
        language: mockData.lang || 'en',
        availableLanguages: mockData.availableLangs || ['en'],
        transcriptDuration: Math.round(transcriptDuration),
        segmentCount: transformedTranscript.length,
        rawSegmentCount: rawSegments.length,
        isPartial: false,
        coverageRatio: undefined,
      });
    }

    // ── Strategy: Try YouTube direct first (free), fall back to Supadata (paid) ──
    // YouTube's InnerTube API is free but gets blocked from datacenter IPs.
    // Supadata is a paid API that handles YouTube's bot detection for us.
    // By trying free first, we only pay for Supadata when YouTube blocks us.

    let rawSegments: { text: string; start: number; duration: number }[] | null = null;
    let language: string | undefined;