retrotoon-studio/server/ffmpegLocal.ts

import { execFile } from "child_process";
import { promisify } from "util";
import { mkdtemp, readdir, rm, writeFile, stat } from "fs/promises";
import { createWriteStream } from "fs";
import { tmpdir } from "os";
import { join } from "path";

const execFileAsync = promisify(execFile);

export interface ProbeResult {
  duration: number;
  width: number;
  height: number;
  fps: number;
  codec: string;
}

export async function probeVideo(filePath: string): Promise<ProbeResult> {
  const { stdout } = await execFileAsync("ffprobe", [
    "-v", "quiet",
    "-print_format", "json",
    "-show_format",
    "-show_streams",
    filePath,
  ]);

  const info = JSON.parse(stdout);
  const videoStream = info.streams?.find((s: any) => s.codec_type === "video");
  if (!videoStream) throw new Error("No video stream found");

  let fps = 24;
  if (videoStream.r_frame_rate) {
    const [num, den] = videoStream.r_frame_rate.split("/").map(Number);
    if (den > 0) fps = num / den;
  }

  return {
    duration: parseFloat(info.format?.duration || videoStream.duration || "0"),
    width: videoStream.width || 0,
    height: videoStream.height || 0,
    fps: Math.round(fps * 100) / 100,
    codec: videoStream.codec_name || "unknown",
  };
}

export interface ExtractOptions {
  fps: number;
  width: number;
  height: number;
  format: "jpeg" | "png";
  quality?: number;
}

export async function extractFramesToDir(
  videoPath: string,
  options: ExtractOptions,
): Promise<{ dir: string; files: string[] }> {
  const dir = await mkdtemp(join(tmpdir(), "retrotoon-frames-"));
  const ext = options.format === "png" ? "png" : "jpg";
  const pattern = join(dir, `frame_%06d.${ext}`);

  const filters = [`fps=${options.fps}`];
  if (options.width > 0 && options.height > 0) {
    filters.push(`scale=${options.width}:${options.height}`);
  }

  const args = [
    "-i", videoPath,
    "-vf", filters.join(","),
  ];

  if (options.format === "jpeg") {
    args.push("-q:v", String(options.quality ?? 5));
  }

  args.push("-y", pattern);

  await execFileAsync("ffmpeg", args, {
    maxBuffer: 50 * 1024 * 1024,
    timeout: 600_000,
  });

  const allFiles = await readdir(dir);
  const frameFiles = allFiles
    .filter((f) => f.startsWith("frame_") && f.endsWith(`.${ext}`))
    .sort();

  return { dir, files: frameFiles };
}

export async function extractAudioFile(videoPath: string, outputPath: string): Promise<string> {
  await execFileAsync("ffmpeg", [
    "-i", videoPath,
    "-vn",
    "-acodec", "pcm_s16le",
    "-ar", "44100",
    "-ac", "2",
    "-y", outputPath,
  ], {
    maxBuffer: 50 * 1024 * 1024,
    timeout: 300_000,
  });
  return outputPath;
}

export interface AssembleOptions {
  fps: number;
  format: "mp4" | "webm";
  audioPath?: string | null;
}

export async function detectSceneCutsFromVideo(
  videoPath: string,
  threshold: number = 0.3,
): Promise<Array<{ frame: number; score: number; time: number }>> {
  const { stdout } = await execFileAsync("ffprobe", [
    "-v", "quiet",
    "-show_frames",
    "-of", "json",
    "-f", "lavfi",
    `movie=${videoPath},select='gt(scene,${threshold})'`,
  ], {
    maxBuffer: 100 * 1024 * 1024,
    timeout: 300_000,
  }).catch(() => ({ stdout: "" }));

  if (!stdout) {
    const { stdout: stdout2 } = await execFileAsync("ffmpeg", [
      "-i", videoPath,
      "-vf", `select='gt(scene,${threshold})',showinfo`,
      "-vsync", "vfr",
      "-f", "null", "-",
    ], {
      maxBuffer: 100 * 1024 * 1024,
      timeout: 300_000,
    }).catch((err) => ({ stdout: "", stderr: err.stderr || "" }));

    const lines = (stdout2 || "").split("\n");
    const cuts: Array<{ frame: number; score: number; time: number }> = [];
    for (const line of lines) {
      const nMatch = line.match(/n:\s*(\d+)/);
      const ptsMatch = line.match(/pts_time:\s*([\d.]+)/);
      if (nMatch && ptsMatch) {
        cuts.push({
          frame: parseInt(nMatch[1]),
          score: threshold,
          time: parseFloat(ptsMatch[1]),
        });
      }
    }
    return cuts;
  }

  try {
    const data = JSON.parse(stdout);
    return (data.frames || []).map((f: any, i: number) => ({
      frame: i,
      score: parseFloat(f.tags?.lavfi_scene_score || "0"),
      time: parseFloat(f.pts_time || f.best_effort_timestamp_time || "0"),
    }));
  } catch {
    return [];
  }
}

export async function assembleVideo(
  frameDir: string,
  frameFiles: string[],
  outputPath: string,
  options: AssembleOptions,
): Promise<{ path: string; duration: number; size: number }> {
  const ext = frameFiles[0]?.split(".").pop() || "jpg";
  const concatListPath = join(frameDir, "concat.txt");
  const lines = frameFiles.map(f => `file '${join(frameDir, f)}'`);
  lines.forEach((_, i) => { lines[i] += `\nduration ${1 / options.fps}`; });
  await writeFile(concatListPath, lines.join("\n"), "utf-8");

  const args = [
    "-f", "concat",
    "-safe", "0",
    "-i", concatListPath,
  ];

  if (options.audioPath) {
    args.push("-i", options.audioPath);
    args.push("-c:a", "aac", "-b:a", "192k");
    args.push("-shortest");
  }

  if (options.format === "mp4") {
    args.push("-c:v", "libx264", "-pix_fmt", "yuv420p", "-preset", "medium", "-crf", "23");
  } else {
    args.push("-c:v", "libvpx-vp9", "-crf", "30", "-b:v", "0");
  }

  args.push("-r", String(options.fps));
  args.push("-movflags", "+faststart");
  args.push("-y", outputPath);

  await execFileAsync("ffmpeg", args, {
    maxBuffer: 100 * 1024 * 1024,
    timeout: 600_000,
  });

  const info = await stat(outputPath);
  const duration = frameFiles.length / options.fps;
  return { path: outputPath, duration, size: info.size };
}

export async function cleanupDir(dir: string): Promise<void> {
  await rm(dir, { recursive: true, force: true });
}