ai-chatbot/lib/video-canvas/media-processing.ts

import { execFile } from "node:child_process";
import { promisify } from "node:util";
import { toVideoTimeRange, type VideoTimeRange } from "./media-contract";

export type MediaRect = { x: number; y: number; width: number; height: number };
export function toVideoPixelRect(rect: MediaRect, width: number, height: number) {
  if (
    ![rect.x, rect.y, rect.width, rect.height, width, height].every(Number.isFinite) ||
    rect.x < 0 ||
    rect.y < 0 ||
    rect.width <= 0 ||
    rect.height <= 0 ||
    rect.x + rect.width > 1.001 ||
    rect.y + rect.height > 1.001
  )
    throw new Error("裁剪区域无效");
  const even = (n: number) => Math.floor(n / 2) * 2;
  const x = even(rect.x * width),
    y = even(rect.y * height);
  const w = Math.min(even(rect.width * width), even(width - x)),
    h = Math.min(even(rect.height * height), even(height - y));
  if (w < 2 || h < 2) throw new Error("裁剪区域太小");
  return { x, y, width: w, height: h };
}
const run = promisify(execFile);
const inputSafety = [
  "-protocol_whitelist",
  "file",
  "-format_whitelist",
  "mov,matroska,webm,avi,mpeg,mpegts,flv,ogg",
];
export async function processVideoFile(
  input: string,
  output: string,
  operation: "crop" | "extract-audio" | "silent-video" | "trim" | "frame" | "slow",
  rect?: MediaRect,
  range?: VideoTimeRange,
  options?: { kind?: "first" | "last" | "current"; time?: number; factor?: number },
) {
  const { stdout } = await run(
    "ffprobe",
    ["-v", "error", ...inputSafety, "-show_streams", "-show_format", "-of", "json", input],
    { timeout: 15000, maxBuffer: 1024 * 1024 },
  );
  const info = JSON.parse(stdout),
    video = info.streams?.find((s: any) => s.codec_type === "video");
  if (!video) throw new Error("素材中没有视频画面");
  if (
    !(Number(video.width) > 0 && Number(video.height) > 0) ||
    Number(video.width) * Number(video.height) > 33554432
  )
    throw new Error("视频尺寸不受支持，请使用 8K 以内的视频");
  const duration = Number(video.duration || info.format?.duration || 0);
  if (!Number.isFinite(duration) || duration <= 0 || duration > 600)
    throw new Error("本地视频处理支持 10 分钟以内的视频");
  let args: string[];
  let dimensions = { width: Number(video.width), height: Number(video.height) };
  let outputDuration = duration;
  if (operation === "frame") {
    const [numerator, denominator] = String(video.avg_frame_rate || "30/1")
      .split("/")
      .map(Number);
    const fps = numerator / denominator;
    const last = Math.max(0, duration - 1 / (Number.isFinite(fps) && fps > 0 ? fps : 30));
    const time =
      options?.kind === "last"
        ? last
        : options?.kind === "current"
          ? Math.min(last, Math.max(0, Number(options.time) || 0))
          : 0;
    args = ["-ss", String(time), "-map", "0:v:0", "-frames:v", "1", "-an", "-c:v", "png"];
    outputDuration = 0;
  } else if (operation === "slow") {
    const factor = options?.factor;
    if (!factor || ![2, 3, 5].includes(factor)) throw Error("慢放倍数无效");
    outputDuration = duration * factor;
    if (outputDuration > 600) throw Error("慢放后视频不能超过 10 分钟");
    let speed = 1 / factor;
    const filters: string[] = [];
    while (speed < 0.5) {
      filters.push("atempo=0.5");
      speed /= 0.5;
    }
    filters.push(`atempo=${speed}`);
    args = [
      "-map",
      "0:v:0",
      "-map",
      "0:a?",
      "-vf",
      `setpts=${factor}*(PTS-STARTPTS),scale=trunc(iw/2)*2:trunc(ih/2)*2`,
      "-c:v",
      "libx264",
      "-preset",
      "veryfast",
      "-crf",
      "20",
      "-pix_fmt",
      "yuv420p",
    ];
    if (info.streams.some((stream: any) => stream.codec_type === "audio"))
      args.push("-af", filters.join(","), "-c:a", "aac");
    args.push("-movflags", "+faststart");
    dimensions = {
      width: Math.floor(dimensions.width / 2) * 2,
      height: Math.floor(dimensions.height / 2) * 2,
    };
  } else if (operation === "extract-audio") {
    if (!info.streams.some((s: any) => s.codec_type === "audio"))
      throw new Error("原视频不包含音轨");
    args = ["-map", "0:a:0", "-vn", "-c:a", "aac", "-b:a", "192k"];
  } else if (operation === "silent-video") {
    args = ["-map", "0:v:0", "-an", "-c:v", "copy", "-movflags", "+faststart"];
  } else if (operation === "trim") {
    if (!range) throw new Error("请选择剪辑范围");
    const selected = toVideoTimeRange(range, duration);
    outputDuration = selected.duration;
    dimensions = {
      width: Math.floor(dimensions.width / 2) * 2,
      height: Math.floor(dimensions.height / 2) * 2,
    };
    args = [
      "-ss",
      String(selected.start),
      "-t",
      String(selected.duration),
      "-map",
      "0:v:0",
      "-map",
      "0:a?",
      "-vf",
      "scale=trunc(iw/2)*2:trunc(ih/2)*2",
      "-c:v",
      "libx264",
      "-preset",
      "veryfast",
      "-crf",
      "20",
      "-pix_fmt",
      "yuv420p",
      "-c:a",
      "aac",
      "-movflags",
      "+faststart",
    ];
  } else {
    if (!rect) throw new Error("请选择裁剪区域");
    const crop = toVideoPixelRect(rect, dimensions.width, dimensions.height);
    dimensions = crop;
    args = [
      "-map",
      "0:v:0",
      "-map",
      "0:a?",
      "-vf",
      `crop=${crop.width}:${crop.height}:${crop.x}:${crop.y}`,
      "-c:v",
      "libx264",
      "-preset",
      "veryfast",
      "-crf",
      "20",
      "-pix_fmt",
      "yuv420p",
      "-c:a",
      "aac",
      "-movflags",
      "+faststart",
    ];
  }
  await run(
    "ffmpeg",
    [
      "-nostdin",
      "-y",
      "-v",
      "error",
      "-threads",
      "2",
      "-filter_threads",
      "1",
      ...inputSafety,
      "-i",
      input,
      ...args,
      "-threads",
      "2",
      output,
    ],
    { timeout: 180000, maxBuffer: 1024 * 1024 },
  );
  if (operation === "frame") {
    const probe = await run(
      "ffprobe",
      [
        "-v",
        "error",
        "-select_streams",
        "v:0",
        "-show_entries",
        "stream=width,height",
        "-of",
        "json",
        output,
      ],
      { timeout: 15000, maxBuffer: 1024 * 1024 },
    );
    const frame = JSON.parse(probe.stdout).streams?.[0];
    if (!frame?.width || !frame?.height) throw Error("未截取到有效视频帧");
    dimensions = { width: Number(frame.width), height: Number(frame.height) };
  }
  return { duration: outputDuration, ...dimensions };
}