lami-parse/api/module/asr/iflytek.ts

// 讯飞录音文件转写大模型 WebAPI 客户端
// API文档: https://www.xfyun.cn/doc/spark/asr_llm/Ifasr_llm.html
import crypto from 'crypto';

interface AsrConfig {
  appId: string;
  accessKeyId: string;
  apiSecret: string;
  host: string;
}

interface CreateTaskParams {
  audio: Buffer;
  fileName: string;
  language?: string;
}

interface TaskResult {
  orderId: string;
  status: 'waiting' | 'processing' | 'completed' | 'failed';
  text?: string;
  error?: string;
}

function randomString(len: number): string {
  const chars = 'ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789';
  return Array.from({ length: len }, () => chars[Math.floor(Math.random() * chars.length)]).join('');
}

function formatDateTime(d: Date): string {
  const pad = (n: number) => String(n).padStart(2, '0');
  return `${d.getFullYear()}-${pad(d.getMonth() + 1)}-${pad(d.getDate())}T${pad(d.getHours())}:${pad(d.getMinutes())}:${pad(d.getSeconds())}+0800`;
}

export class IflytekAsr {
  private config: AsrConfig;
  private baseUrl: string;

  constructor(config: AsrConfig) {
    this.config = config;
    this.baseUrl = config.host.replace(/\/$/, '');
  }

  // 标准签名: 对所有URL参数（除signature）排序 → URL编码 → &拼接 → HMAC-SHA1 → Base64
  private sign(params: Record<string, string>): string {
    // 1. 排除 signature，按 key 排序
    const sorted = Object.keys(params)
      .filter(k => k !== 'signature' && params[k] != null && params[k] !== '')
      .sort()
      .map(k => {
        const encodedKey = encodeURIComponent(k);
        const encodedVal = encodeURIComponent(params[k]);
        return `${encodedKey}=${encodedVal}`;
      })
      .join('&');

    console.log('[ASR] sign raw:', sorted);

    // 2. HMAC-SHA1 → Base64
    const hmac = crypto.createHmac('sha1', this.config.apiSecret);
    return hmac.update(sorted).digest('base64');
  }

  // 创建转写任务
  async createTask(params: CreateTaskParams): Promise<string> {
    const dateTime = formatDateTime(new Date());
    const signatureRandom = randomString(16);
    const fileSize = params.audio.length.toString();
    const language = params.language || 'autodialect';

    // 构建所有参数
    const queryParams: Record<string, string> = {
      appId: this.config.appId,
      accessKeyId: this.config.accessKeyId,
      dateTime,
      signatureRandom,
      fileSize,
      fileName: params.fileName,
      language,
      durationCheckDisable: 'true',  // 关闭时长校验
    };

    // 生成签名 (不含 signature 字段)
    const signature = this.sign(queryParams);

    // 构建 URL — 使用与签名相同的编码方式
    const sortedKeys = Object.keys(queryParams).filter(k => queryParams[k] != null && queryParams[k] !== '').sort();
    const qs = sortedKeys.map(k => `${encodeURIComponent(k)}=${encodeURIComponent(queryParams[k])}`).join('&');
    const url = `${this.baseUrl}/v2/upload?${qs}`;

    console.log('[ASR] upload URL:', url);

    const res = await fetch(url, {
      method: 'POST',
      headers: {
        'Content-Type': 'application/octet-stream',
        'signature': signature,
      },
      body: new Uint8Array(params.audio),
    });

    const text = await res.text();
    console.log('[ASR] upload response:', text.slice(0, 500));

    let json: any;
    try { json = JSON.parse(text); } catch { throw new Error(`响应解析失败: ${text.slice(0, 200)}`); }

    if (json.code !== '000000') {
      throw new Error(`讯飞ASR上传失败: [${json.code}] ${json.descInfo || ''}`);
    }

    const orderId = json.content?.orderId;
    if (!orderId) throw new Error('未获取到 orderId');
    return orderId;
  }

  // 查询转写结果
  async getResult(orderId: string): Promise<TaskResult> {
    const dateTime = formatDateTime(new Date());
    const signatureRandom = randomString(16);

    const queryParams: Record<string, string> = {
      appId: this.config.appId,
      accessKeyId: this.config.accessKeyId,
      dateTime,
      signatureRandom,
      orderId,
      resultType: 'transfer',
    };

    const signature = this.sign(queryParams);

    const sortedKeys = Object.keys(queryParams).filter(k => queryParams[k] != null && queryParams[k] !== '').sort();
    const qs = sortedKeys.map(k => `${encodeURIComponent(k)}=${encodeURIComponent(queryParams[k])}`).join('&');
    const url = `${this.baseUrl}/v2/getResult?${qs}`;

    const res = await fetch(url, {
      method: 'POST',
      headers: {
        'Content-Type': 'application/json',
        'signature': signature,
      },
    });

    const text = await res.text();

    let json: any;
    try { json = JSON.parse(text); } catch { throw new Error(`结果解析失败: ${text.slice(0, 200)}`); }

    if (json.code !== '000000') {
      if (json.code === '000001') return { orderId, status: 'processing' };
      return { orderId, status: 'failed', error: `[${json.code}] ${json.descInfo || ''}` };
    }

    const content = json.content || {};
    let textResult = '';

    if (content.orderResult) {
      try {
        const resultObj = typeof content.orderResult === 'string'
          ? JSON.parse(content.orderResult) : content.orderResult;
        if (resultObj.lattice && Array.isArray(resultObj.lattice)) {
          textResult = resultObj.lattice
            .map((seg: any) => {
              const jb = seg.json_1best || '';
              try {
                const b = JSON.parse(jb);
                return b.st?.rt?.[0]?.ws?.map((w: any) => w.cw?.[0]?.w || '').join('') || '';
              } catch { return seg.onebest || ''; }
            })
            .join('');
        }
      } catch {
        textResult = typeof content.orderResult === 'string' ? content.orderResult : '';
      }
    }

    return { orderId, status: 'completed', text: textResult || content.descInfo || '' };
  }
}