11

scripts/nodejs/subtitle_bgm_generate.js

@@ -0,0 +1,288 @@
+"use strict";
+
+/**
+ * 自动生成字幕并混 BGM（对齐 Electron autoGenerateSubtitleAndBGM）
+ */
+const fs = require("fs");
+const path = require("path");
+const os = require("os");
+const { spawnSync } = require("child_process");
+
+const pipelineNative = require("./pipeline_native.js");
+const desktopConfig = require("./desktop_config.js");
+
+const OUTPUT_DIR = path.join(os.tmpdir(), "aiclient-subtitle-bgm");
+
+function ensureDir() {
+  fs.mkdirSync(OUTPUT_DIR, { recursive: true });
+  return OUTPUT_DIR;
+}
+
+function makeOutputPath(prefix, ext = "mp4") {
+  return path.join(ensureDir(), `${prefix}_${Date.now()}.${ext}`);
+}
+
+function execFfmpeg(args) {
+  const ffmpeg = pipelineNative.locateFfmpeg();
+  if (!ffmpeg) {
+    throw new Error("未找到 ffmpeg，请配置 AICLIENT_FFMPEG_PATH 或 src-tauri/binaries/ffmpeg.exe");
+  }
+  const r = spawnSync(ffmpeg, args, { encoding: "utf8", windowsHide: true });
+  if (r.status !== 0) {
+    throw new Error((r.stderr || r.stdout || "").trim() || `ffmpeg 退出码 ${r.status}`);
+  }
+}
+
+function probeVideoDuration(videoPath) {
+  const ffprobe = pipelineNative.locateFfmpeg()?.replace(/ffmpeg(\.exe)?$/i, "ffprobe$1");
+  if (ffprobe && fs.existsSync(ffprobe)) {
+    const r = spawnSync(
+      ffprobe,
+      [
+        "-v",
+        "error",
+        "-show_entries",
+        "format=duration",
+        "-of",
+        "default=noprint_wrappers=1:nokey=1",
+        videoPath,
+      ],
+      { encoding: "utf8", windowsHide: true },
+    );
+    if (r.status === 0) {
+      const n = parseFloat(String(r.stdout || "").trim());
+      if (Number.isFinite(n) && n > 0) return n;
+    }
+  }
+  const ffmpeg = pipelineNative.locateFfmpeg();
+  const r2 = spawnSync(ffmpeg, ["-i", videoPath], { encoding: "utf8", windowsHide: true });
+  const stderr = r2.stderr || "";
+  const m = stderr.match(/Duration:\s*(\d+):(\d+):(\d+(?:\.\d+)?)/);
+  if (m) {
+    return Number(m[1]) * 3600 + Number(m[2]) * 60 + Number(m[3]);
+  }
+  return 0;
+}
+
+function splitScriptLines(text) {
+  return String(text || "")
+    .split(/(?<=[。！？.!?])\s*|\n+/)
+    .map((s) => s.trim())
+    .filter(Boolean);
+}
+
+function alignScriptToRecords(script, records) {
+  const lines = splitScriptLines(script);
+  if (!lines.length || !records.length) return records;
+
+  if (lines.length === records.length) {
+    return records.map((r, i) => ({ ...r, text: lines[i] }));
+  }
+
+  const startMs = records[0].start;
+  const endMs = records[records.length - 1].end;
+  const total = Math.max(endMs - startMs, lines.length * 2000);
+  const step = total / lines.length;
+  return lines.map((text, i) => ({
+    text,
+    start: Math.round(startMs + i * step),
+    end: Math.round(startMs + (i + 1) * step),
+  }));
+}
+
+function recordsFromTextEvenly(text, durationSec) {
+  const lines = splitScriptLines(text);
+  if (!lines.length) return [];
+  const totalMs = Math.max(Math.round(durationSec * 1000), lines.length * 1500);
+  const step = totalMs / lines.length;
+  return lines.map((line, i) => ({
+    text: line,
+    start: Math.round(i * step),
+    end: Math.round((i + 1) * step),
+  }));
+}
+
+function formatSrtTime(ms) {
+  const h = Math.floor(ms / 3600000);
+  const m = Math.floor((ms % 3600000) / 60000);
+  const s = Math.floor((ms % 60000) / 1000);
+  const msPart = Math.floor(ms % 1000);
+  return `${String(h).padStart(2, "0")}:${String(m).padStart(2, "0")}:${String(s).padStart(2, "0")},${String(msPart).padStart(3, "0")}`;
+}
+
+function buildSrtContent(records) {
+  return records
+    .map((r, i) => {
+      const start = formatSrtTime(r.start);
+      const end = formatSrtTime(Math.max(r.end, r.start + 200));
+      return `${i + 1}\n${start} --> ${end}\n${r.text}\n`;
+    })
+    .join("\n");
+}
+
+function escapeSubPath(filePath) {
+  return filePath.replace(/\\/g, "/").replace(/:/g, "\\:").replace(/'/g, "\\'");
+}
+
+function addSubtitleToVideo(inputVideo, srtPath, forceStyle) {
+  const outputVideo = makeOutputPath("with_subtitle");
+  const style = forceStyle || "FontName=Microsoft YaHei,FontSize=24,PrimaryColour=&H00FFFFFF,OutlineColour=&H00000000,Outline=2,Alignment=2,MarginV=40";
+  const sub = escapeSubPath(srtPath);
+  const vf = `subtitles='${sub}':force_style='${style}'`;
+  execFfmpeg(["-y", "-i", inputVideo, "-vf", vf, "-c:a", "copy", outputVideo]);
+  return outputVideo;
+}
+
+function addBgmToVideo(inputVideo, bgmPath, bgmVolume = 30) {
+  const outputVideo = makeOutputPath("with_bgm");
+  const ratio = Number(bgmVolume) / 100;
+  const filter = `[0:a]volume=1.0[a0];[1:a]volume=${ratio}[a1];[a0][a1]amix=inputs=2:duration=first`;
+  execFfmpeg([
+    "-y",
+    "-i",
+    inputVideo,
+    "-stream_loop",
+    "-1",
+    "-i",
+    bgmPath,
+    "-filter_complex",
+    filter,
+    "-c:v",
+    "copy",
+    "-shortest",
+    outputVideo,
+  ]);
+  return outputVideo;
+}
+
+async function runAsr(audioPath, modelConfig, scriptContent) {
+  const asrMode = modelConfig.asrMode || "online";
+  if (asrMode === "online") {
+    if (!modelConfig.aliyunApiKey) {
+      throw new Error("在线 ASR 需配置 BAILIAN_API_KEY 或 DASHSCOPE_API_KEY");
+    }
+    if (!modelConfig.ossConfig?.bucket) {
+      throw new Error("在线 ASR 需配置 OSS（上传音频供识别）");
+    }
+    globalThis.__native?.emitProgress?.("正在上传音频到 OSS…");
+    const upload = await pipelineNative.aliyunOssUpload(audioPath, modelConfig.ossConfig);
+    if (!upload?.success || !upload.url) {
+      throw new Error(upload?.error || "OSS 上传失败");
+    }
+    globalThis.__native?.emitProgress?.("正在进行语音识别…");
+    const asr = await pipelineNative.aliyunAsrFiletrans(upload.url, {
+      apiKey: modelConfig.aliyunApiKey,
+      model: "qwen3-asr-flash-filetrans",
+    });
+    if (!asr?.success) {
+      throw new Error(asr?.error || "语音识别失败");
+    }
+    let records = Array.isArray(asr.sentences) ? asr.sentences : [];
+    if (!records.length && asr.text) {
+      records = recordsFromTextEvenly(asr.text, 60);
+    }
+    return { text: asr.text || "", records };
+  }
+
+  const info = modelConfig.asrServerInfo;
+  if (!info) {
+    throw new Error("本地 ASR 需配置 ASR_SERVER_URL 或 ASR_SERVER_INFO");
+  }
+  globalThis.__native?.emitProgress?.("正在调用本地语音识别…");
+  const asr = await pipelineNative.localAsr(audioPath, info);
+  if (!asr?.success) {
+    throw new Error(asr?.error || "本地语音识别失败");
+  }
+  let records = Array.isArray(asr.sentences) ? asr.sentences : [];
+  if (!records.length && asr.text) {
+    records = recordsFromTextEvenly(asr.text, 60);
+  }
+  return { text: asr.text || "", records };
+}
+
+globalThis.__nodejsMain = async function main(params) {
+  const p = params || {};
+  const inputVideo = String(p.inputVideo || "").trim();
+  if (!inputVideo || !fs.existsSync(inputVideo)) {
+    return { success: false, error: "缺少或找不到输入视频，请先在步骤 02/03 生成并处理视频" };
+  }
+
+  const autoSubtitle = Boolean(p.autoSubtitle);
+  const bgmEnabled = Boolean(p.bgmEnabled);
+  if (!autoSubtitle && !bgmEnabled) {
+    return { success: false, error: "请至少勾选「自动生成字幕」或「添加背景音乐」" };
+  }
+  if (bgmEnabled) {
+    const bgm = String(p.bgmPath || "").trim();
+    if (!bgm || !fs.existsSync(bgm)) {
+      return { success: false, error: "已启用背景音乐，请先选择音乐文件" };
+    }
+  }
+
+  const modelConfig = desktopConfig.buildModelConfig();
+  const check = desktopConfig.validateModelConfig(modelConfig);
+  if (autoSubtitle && !check.ok) {
+    return { success: false, error: check.error };
+  }
+  if (autoSubtitle && (modelConfig.asrMode === "online" || !modelConfig.asrMode)) {
+    const appCheck = desktopConfig.validateModelConfig(modelConfig);
+    if (!appCheck.ok && modelConfig.asrMode === "online") {
+      return { success: false, error: appCheck.error };
+    }
+  }
+
+  let current = inputVideo;
+  let srtPath = "";
+
+  try {
+    if (autoSubtitle) {
+      globalThis.__native?.emitProgress?.("正在提取音频…");
+      const audioPath = await pipelineNative.ffmpegExtractAudio(current);
+      const scriptContent = String(p.scriptContent || "").trim();
+      const { records: rawRecords } = await runAsr(audioPath, modelConfig, scriptContent);
+
+      let records = rawRecords;
+      if (p.smartSubtitle !== false && scriptContent) {
+        records = alignScriptToRecords(scriptContent, records);
+      }
+      if (!records.length) {
+        const duration = probeVideoDuration(current);
+        const fallbackText = scriptContent || " ";
+        records = recordsFromTextEvenly(fallbackText, duration || 30);
+      }
+      if (!records.length) {
+        return { success: false, error: "未识别到语音内容，无法生成字幕" };
+      }
+
+      srtPath = makeOutputPath("subtitle", "srt");
+      fs.writeFileSync(srtPath, buildSrtContent(records), "utf8");
+
+      globalThis.__native?.emitProgress?.("正在烧录字幕到视频…");
+      const forceStyle = String(p.subtitleForceStyle || "").trim();
+      current = addSubtitleToVideo(current, srtPath, forceStyle || undefined);
+      pipelineNative.deleteFile(audioPath);
+    }
+
+    if (bgmEnabled) {
+      globalThis.__native?.emitProgress?.("正在混合背景音乐…");
+      current = addBgmToVideo(current, p.bgmPath, p.bgmVolume ?? 30);
+    }
+
+    return {
+      success: true,
+      videoPath: current,
+      srtPath: srtPath || null,
+      message: autoSubtitle && bgmEnabled
+        ? "字幕与背景音乐处理完成"
+        : autoSubtitle
+          ? "字幕生成完成"
+          : "背景音乐添加完成",
+    };
+  } catch (e) {
+    return {
+      success: false,
+      error: e.message || String(e),
+      partialVideo: current !== inputVideo ? current : null,
+    };
+  }
+};