utils/CaptionRender.js

// FULL FIXED VERSION — eliminates concat=n=1 and all copy filters

import { spawn } from 'child_process';
import os from 'os';
import path from 'path';
import fs from 'fs';

const ffmpegLocation = os.platform() === 'win32' ? 'ffmpeg.exe' : 'ffmpeg';

export class CaptionRenderer {
  validateCaption(originalManuscript) {
    let captionFiles = originalManuscript.transcript.map(item => item.audioCaptionFile);
    // make sure the caption files are in ass format and exist
    for (let captionFile of captionFiles) {
      if (!captionFile || !captionFile.endsWith('.ass')) {
        throw new Error('Invalid caption file format. Expected .ass files for item ' + (captionFiles.indexOf(captionFile) + 1) + '. Did you forget to use `caption` plugin?');
      }
    }
    return true;
  }

  async doRender(jobId, originalManuscript, onLog, npmScript, options, controller) {
    const outDir = path.join(process.cwd(), 'out');
    if (!fs.existsSync(outDir)) fs.mkdirSync(outDir, { recursive: true });

    const segmentFiles = [];

    // Helper to clean up intermediate files
    function cleanupFiles(files) {
      for (const f of files) {
        try { fs.unlinkSync(f); } catch (e) { /* ignore */ }
      }
    }

    // Render each transcript section separately
    for (const [i, section] of originalManuscript.transcript.entries()) {
      const segmentFile = path.join(outDir, `${jobId}_segment_${i}.mp4`);
      const inputs = [];
      const filter = [];
      let inputIndex = 0;

      const media = section.mediaAbsPaths || [];
      if (!media.length) continue;

      media.forEach(m => {
        inputs.push('-i', m.path);
        inputIndex++;
      });

      inputs.push('-i', section.audioFullPath);
      inputIndex++;

      let clips = [];
      let elapsed = 0;

      media.forEach((m, j) => {
        let d = m.prompt?.durationSec || m.durationSec;
        d = Math.min(d, section.durationInSeconds - elapsed);
        if (d <= 0) return;

        const label = `v${i}_${j}`;
        filter.push(`[${inputIndex - media.length - 1 + j}:v]trim=duration=${d},setpts=PTS-STARTPTS[${label}]`);
        clips.push(`[${label}]`);
        elapsed += d;
      });

      const vcat = `vc${i}`;
      filter.push(`${clips.join('')}concat=n=${clips.length}:v=1:a=0[${vcat}]`);

      const sub = `vs${i}`;
      const esc = section.audioCaptionFile.replace(/\\/g, '/').replace(/:/g, '\\:');
      filter.push(`[${vcat}]subtitles='${esc}'[${sub}]`);

      const vout = `vout${i}`;
      const aout = `aout${i}`;

      filter.push(`[${sub}]null[${vout}]`);
      filter.push(`[${inputIndex - 1}:a]anull[${aout}]`);

      const args = [
        ...inputs,
        '-filter_complex', filter.join(';'),
        '-map', `[${vout}]`,
        '-map', `[${aout}]`,
        '-c:v', 'libx264',
        '-c:a', 'aac',
        '-y', segmentFile
      ];

      await this.runFFmpeg(args, controller, onLog);
      segmentFiles.push(segmentFile);
    }

    // Concatenate all segments together
    const resultFile = path.join(outDir, `${jobId}_final.mp4`);
    const concatFile = path.join(outDir, `${jobId}_combined_final.mp4`);
    const concatList = path.join(outDir, `${jobId}_concat_list.txt`);

    fs.writeFileSync(concatList, segmentFiles.map(f => `file '${f}'`).join('\n'));

    const concatArgs = [
      '-f', 'concat',
      '-safe', '0',
      '-i', concatList,
      '-c', 'copy',
      '-y', concatFile
    ];
    await this.runFFmpeg(concatArgs, controller, onLog);

    // Add background music if present
    const { bgMusic, bgMusicVolume = 0.25 } = originalManuscript;
    if (bgMusic) {
      const finalWithMusic = resultFile; // Overwrite the _final.mp4 file
      const getDuration = (filePath) => {
        try {
          const ffprobeSync = require('child_process').spawnSync;
          const result = ffprobeSync(ffmpegLocation.replace('ffmpeg', 'ffprobe'), [
            '-v', 'error',
            '-show_entries', 'format=duration',
            '-of', 'default=noprint_wrappers=1:nokey=1',
            filePath
          ]);
          return parseFloat(result.stdout.toString().trim());
        } catch (e) {
          return null;
        }
      };
      const videoDuration = getDuration(concatFile);
      const atrimArg = (videoDuration && !isNaN(videoDuration)) ? `,atrim=end=${videoDuration}` : '';

      const voiceVolume = options?.voiceVolume || 1.5;
      const musicArgs = [
        '-i', concatFile,
        '-i', bgMusic,
        '-filter_complex', `[0:a]volume=${voiceVolume}[voice];[1:a]volume=${bgMusicVolume}${atrimArg}[bgm];[voice][bgm]amix=inputs=2:duration=first:dropout_transition=2:normalize=0[mixeda]`,
        '-map', '0:v',
        '-map', '[mixeda]',
        '-c:v', 'copy',
        '-c:a', 'aac',
        '-y', finalWithMusic
      ];
      await this.runFFmpeg(musicArgs, controller, onLog);
      segmentFiles.push(concatFile);
    }
    else {
      fs.renameSync(concatFile, resultFile);
    }

    cleanupFiles([...segmentFiles, concatList]);
    return resultFile;
  }


  runFFmpeg(args, controller, onLog) {
    console.log('FFMPEG cmd:', args.join(' '));

    return new Promise((resolve, reject) => {
      const p = spawn(ffmpegLocation, args, { detached: true });

      if (controller) controller.stop = () => p.kill('SIGKILL');

      p.stderr.on('data', d => onLog && onLog(d.toString()));
      p.stdout.on('data', d => onLog && onLog(d.toString()));

      p.on('close', code => code === 0 ? resolve() : reject(new Error(`ffmpeg exited ${code}`)));
      p.on('error', reject);
    });
  }
}