audio_codebuddy/server/src/services/tts-lang-router.service.ts

/**
 * tts-lang-router.service.ts — 按语言找 TTS vendor / model / voiceId
 *
 * 输入: langKey (BCP-47 风格,如 zh-CMN、zh-YUE、en-US、ja-JP)
 * 输出: { vendor, model, voice } 三元组
 *
 * fallback 链(优先):
 *   1. 矩阵中匹配的"最佳"音色 (RECOMMENDED 字段)
 *   2. 同语言其他音色 (langVoices 列表)
 *   3. 同 vendor 其他模型 (matrix 数据里所有 model)
 *   4. 跨 vendor → Qwen-TTS (支持 11 语种)
 *   5. Edge 同语种 Neural (zh-HK / en-US / en-GB / ja-JP / ko-KR / id-ID)
 *   6. → 422 NotFound (绝不静默跨语言退化)
 */
import fs from 'fs';
import path from 'path';

export type Vendor = 'bailian' | 'edge';
export interface LangRoute {
  vendor: Vendor;
  model: string;
  voice: string;
  preferredName: string;  // provider registry key, 例 'bailian-tts' / 'edge-tts'
  reason: string;         // 选这条的 log 用
}

interface LangDef {
  label: string;
  base: string;
  dialect: string;
  supportedModels: string[];
  voiceCount: number;
  recommended: LangRoute | null;
  fallbackChain: string[];
  sampleVoices: Array<{ voice_id: string; name: string; model: string }>;
}

interface MatrixFile {
  version: string;
  recommended: string[];
  languages: Record<string, LangDef>;
}

// 启动时一次性加载(缓存到 hot path)
let MATRIX: MatrixFile | null = null;
const LANG_MAP: Map<string, LangDef> = new Map();

function ensureLoaded(): void {
  if (MATRIX) return;
  // 编译后 dist/ 目录或源码 src/ 目录
  const candidates = [
    path.join(__dirname, '..', 'config', 'alicloud-tts-lang-voice-matrix.json'),
    path.join(__dirname, '..', '..', 'src', 'config', 'alicloud-tts-lang-voice-matrix.json'),
    path.join(process.cwd(), 'src', 'config', 'alicloud-tts-lang-voice-matrix.json'),
  ];
  let raw: string | null = null;
  for (const p of candidates) {
    try { raw = fs.readFileSync(p, 'utf8'); break; } catch { /* continue */ }
  }
  if (!raw) {
    console.warn('[tts-lang-router] 矩阵 JSON 加载失败,所有语言查询都会 fallback');
    MATRIX = { version: '0', recommended: [], languages: {} };
    return;
  }
  MATRIX = JSON.parse(raw);
  for (const [key, def] of Object.entries(MATRIX.languages || {})) {
    LANG_MAP.set(key, def);
  }
  console.log(`[tts-lang-router] 加载 ${LANG_MAP.size} 个 lang_key,推荐 ${MATRIX.recommended?.length || 0} 个`);
}

// Edge TTS 同语种 fallback(用在矩阵都没匹配时)
function edgeFallbackVoice(langKey: string): { vendor: Vendor; model: string; voice: string; reason: string } | null {
  const map: Record<string, string> = {
    'zh-CMN': 'zh-CN-XiaoxiaoNeural',
    'zh-YUE': 'zh-HK-HiuMaanNeural',
    'zh-NEN': 'zh-CN-liaoning-XiaobeiNeural',  // 东北方言
    'en-US': 'en-US-JennyNeural',
    'en-GB': 'en-GB-RyanNeural',
    'ja-JP': 'ja-JP-NanamiNeural',
    'ko-KR': 'ko-KR-SunHiNeural',
    'id-ID': 'id-ID-ArdiNeural',
  };
  const voice = map[langKey];
  if (!voice) return null;
  return { vendor: 'edge', model: 'edge-tts', voice, reason: `Edge fallback for ${langKey}` };
}

// ─────────────────────────────────────────────────────────────
// 公开 API
// ─────────────────────────────────────────────────────────────

/** 主入口: TTS 任务入队时必调 */
export function pickTtsVendor(langKey?: string): LangRoute | null {
  ensureLoaded();
  if (!langKey) return null;

  const def = LANG_MAP.get(langKey);
  // 1. exact match: matrices 推荐音色
  if (def?.recommended) {
    return { ...def.recommended, preferredName: `${def.recommended.vendor}-tts` };
  }

  // 2. langKey 存在但无 recommended(实际数据但没给最佳)→ 取第一个 sampleVoice
  if (def && def.sampleVoices.length > 0) {
    const v = def.sampleVoices[0];
    return {
      vendor: 'bailian',
      model: v.model,
      voice: v.voice_id,
      preferredName: 'bailian-tts',
      reason: `langKey=${langKey} match, 取首个 sampleVoice ${v.voice_id}`,
    };
  }

  // 3. 跨语种 fallback:同 base
  if (def) {
    // 例如 zh-MNN 没有 fallback 链但有 langVoices,降级到 zh-CMN
    const baseKey = langKey.split('-')[0] + '-' + 'CMN';
    const baseDef = LANG_MAP.get(baseKey);
    if (baseDef?.recommended) {
      return { ...baseDef.recommended, preferredName: 'bailian-tts', reason: `langKey=${langKey} 非主流,降级到 ${baseKey}` };
    }
  }

  // 4. 最后 Edge 同语种 fallback
  const edge = edgeFallbackVoice(langKey);
  if (edge) {
    return { ...edge, preferredName: 'edge-tts' };
  }

  // 5. 完全不支持 → 返回 null(让 controller 抛 422)
  return null;
}

/** 给前端"选语言后展示可选音色" */
export function listVoicesByLang(langKey: string, preferCheap = false): Array<{
  vendor: Vendor; model: string; voiceId: string; voiceName: string; desc?: string;
}> {
  ensureLoaded();
  const def = LANG_MAP.get(langKey);
  if (!def) return [];
  const out: Array<{ vendor: Vendor; model: string; voiceId: string; voiceName: string; desc?: string }> = [];
  for (const sv of def.sampleVoices) {
    if (preferCheap && sv.model.includes('cosyvoice-v2')) continue;
    out.push({
      vendor: 'bailian',
      model: sv.model,
      voiceId: sv.voice_id,
      voiceName: sv.name || sv.voice_id,
      desc: def.label,
    });
  }
  return out;
}

/** 给前端"语言选择器"用 */
export function listSupportedLanguages(): Array<{
  key: string; label: string; recommended: boolean; vendorHint?: string;
}> {
  ensureLoaded();
  const recommended = new Set(MATRIX?.recommended || []);
  const out: Array<{ key: string; label: string; recommended: boolean; vendorHint?: string }> = [];
  for (const [key, def] of [...LANG_MAP.entries()].sort()) {
    out.push({
      key,
      label: def.label,
      recommended: recommended.has(key),
      vendorHint: def.recommended ? `${def.recommended.vendor}/${def.recommended.model}` : undefined,
    });
  }
  return out;
}

/** 校验:客户端传来 langKey 是否支持 */
export function isLanguageSupported(langKey: string): boolean {
  ensureLoaded();
  if (LANG_MAP.has(langKey)) return true;
  // Edge fallback 也算"支持"(虽然音色不在矩阵)
  return edgeFallbackVoice(langKey) !== null;
}

/** 反向查询: 这个 voice 能不能说这种语言 */
export function supportsLanguage(voiceId: string, langKey: string): boolean {
  ensureLoaded();
  if (!voiceId) return false;
  const def = LANG_MAP.get(langKey);
  if (!def) return false;
  return def.sampleVoices.some(v => v.voice_id === voiceId);
}

/** 调试用:获取整个矩阵 */
export function _getMatrix(): MatrixFile | null {
  ensureLoaded();
  return MATRIX;
}