@ww_nero/media 1.0.3 → 1.0.5

package/index.js

@@ -230,15 +230,8 @@ const runAsrScript = (audioPath, outputPath, uploadUrl, apiKey) => {
     const isWsl = isWslEnvironment();
 
     // 参考 bash.js 的调用方式，通过 bash -ic 执行命令
-    let cmd, args;
-    if (isWsl || process.platform === 'linux') {
-      cmd = 'bash';
-      args = ['-ic', pythonCommand];
-    } else {
-      // Windows 环境，通过 wsl 调用 bash
-      cmd = 'wsl';
-      args = ['-e', 'bash', '-ic', pythonCommand];
-    }
+    cmd = 'bash';
+    args = ['-ic', pythonCommand];
 
     const child = spawn(cmd, args, {
       stdio: ['pipe', 'pipe', 'pipe'],
@@ -279,26 +272,7 @@ const runAsrScript = (audioPath, outputPath, uploadUrl, apiKey) => {
       if (code === 0) {
         resolve(stdout.trim());
       } else {
-        // 尝试解析结构化的 ASR 错误信息
-        const stderrContent = stderr.trim();
-        const asrErrorMatch = stderrContent.match(/ASR_ERROR:\s*(\{.*\})/);
-        if (asrErrorMatch) {
-          try {
-            const errorInfo = JSON.parse(asrErrorMatch[1]);
-            const parts = [errorInfo.error || '识别失败'];
-            if (errorInfo.audio_url) {
-              parts.push(`静态资源地址: ${errorInfo.audio_url}`);
-            }
-            if (errorInfo.details) {
-              parts.push(`服务器错误信息: ${errorInfo.details}`);
-            }
-            reject(new Error(parts.join('\n')));
-            return;
-          } catch {
-            // JSON 解析失败，使用原始错误信息
-          }
-        }
-        const errorMsg = stderrContent || stdout.trim() || `进程退出码: ${code}`;
+        const errorMsg = stderr.trim() || stdout.trim() || `进程退出码: ${code}`;
         reject(new Error(errorMsg));
       }
     });
@@ -343,7 +317,7 @@ const asr = async ({ working_directory, audio_file }) => {
 const server = new Server(
   {
     name: 'media',
-    version: '1.0.3',
+    version: '1.0.5',
   },
   {
     capabilities: {

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@ww_nero/media",
-  "version": "1.0.3",
+  "version": "1.0.5",
   "description": "MCP server for media processing, including ASR speech recognition",
   "main": "index.js",
   "bin": {

package/scripts/asr_srt.py

@@ -97,34 +97,28 @@ def upload_audio(upload_url: str, audio_path: str) -> str:
     if not data.get('success') or 'data' not in data or 'path' not in data['data']:
         raise Exception(f"上传响应格式错误: {data}")
 
-    return Path(data['data']['path']).name
+    # 返回完整的相对路径，用于拼接到服务器地址后面
+    return data['data']['path']
 
 
-def get_static_url(upload_url: str, filename: str) -> str:
+def get_static_url(upload_url: str, file_path: str) -> str:
     """
-    根据上传接口 URL 和文件名构建静态资源 URL
+    根据上传接口 URL 和文件路径构建静态资源 URL
 
     Args:
         upload_url: 上传接口的完整 URL (如 http://server.domain.com/upload)
-        filename: 上传后的文件名
+        file_path: 上传后返回的文件路径 (如 /tmp/xxx.wav)
 
     Returns:
         静态资源的完整 URL
     """
     # 从上传 URL 中提取基础 URL
-    # 例如: http://server.domain.com/upload -> http://server.domain.com/
+    # 例如: http://server.domain.com/upload -> http://server.domain.com
     from urllib.parse import urlparse, urlunparse
     parsed = urlparse(upload_url)
-    base_url = urlunparse((parsed.scheme, parsed.netloc, '/', '', '', ''))
-    return f"{base_url.rstrip('/')}/{filename}"
-
-
-class AsrError(Exception):
-    """ASR 识别错误，包含静态资源 URL 和错误详情"""
-    def __init__(self, message: str, audio_url: str = None, details: str = None):
-        self.audio_url = audio_url
-        self.details = details
-        super().__init__(message)
+    base_url = urlunparse((parsed.scheme, parsed.netloc, '', '', '', ''))
+    # file_path 已经是以 / 开头的完整相对路径，直接拼接
+    return f"{base_url}{file_path}"
 
 
 def transcribe_audio(audio_url: str, api_key: str) -> list:
@@ -141,17 +135,14 @@ def transcribe_audio(audio_url: str, api_key: str) -> list:
     dashscope.api_key = api_key
 
     # 发起异步识别请求
-    try:
-        transcribe_response = Transcription.async_call(
-            model='paraformer-v2',
-            file_urls=[audio_url],
-            language_hints=['zh', 'en', 'ja']
-        )
-    except Exception as e:
-        raise AsrError(f"ASR 请求失败: {e}", audio_url=audio_url, details=str(e))
+    transcribe_response = Transcription.async_call(
+        model='paraformer-v2',
+        file_urls=[audio_url],
+        language_hints=['zh', 'en', 'ja']
+    )
 
     if not transcribe_response or not hasattr(transcribe_response, 'output'):
-        raise AsrError("ASR 请求失败: 无效的响应", audio_url=audio_url)
+        raise Exception("ASR 请求失败: 无效的响应")
 
     # 轮询等待识别完成
     while True:
@@ -162,26 +153,10 @@ def transcribe_audio(audio_url: str, api_key: str) -> list:
         transcribe_response = Transcription.fetch(task=transcribe_response.output.task_id)
 
     if transcribe_response.status_code != HTTPStatus.OK:
-        error_msg = getattr(transcribe_response, 'message', str(transcribe_response.status_code))
-        raise AsrError(
-            f"ASR 识别失败: {transcribe_response.status_code}",
-            audio_url=audio_url,
-            details=error_msg
-        )
+        raise Exception(f"ASR 识别失败: {transcribe_response.status_code}")
 
     if transcribe_response.output.task_status == 'FAILED':
-        # 尝试获取更详细的错误信息
-        error_details = None
-        results = transcribe_response.output.get('results', [])
-        for result in results:
-            if result.get('subtask_status') == 'FAILED':
-                error_details = result.get('message', '未知错误')
-                break
-        raise AsrError(
-            "ASR 识别任务失败",
-            audio_url=audio_url,
-            details=error_details
-        )
+        raise Exception("ASR 识别任务失败")
 
     # 获取识别结果
     results = transcribe_response.output.get('results', [])
@@ -217,15 +192,14 @@ def main():
         print("错误: 请通过 --api-key 参数或 ASR_API_KEY 环境变量提供 API Key", file=sys.stderr)
         sys.exit(1)
 
-    audio_url = None
     try:
         # 1. 上传音频文件
         print(f"正在上传音频文件: {args.audio}")
-        filename = upload_audio(args.upload_url, args.audio)
-        print(f"上传成功: {filename}")
+        file_path = upload_audio(args.upload_url, args.audio)
+        print(f"上传成功: {file_path}")
 
         # 2. 构建静态资源 URL
-        audio_url = get_static_url(args.upload_url, filename)
+        audio_url = get_static_url(args.upload_url, file_path)
         print(f"音频 URL: {audio_url}")
 
         # 3. 调用 ASR 识别
@@ -233,7 +207,8 @@ def main():
         transcriptions = transcribe_audio(audio_url, api_key)
 
         if not transcriptions:
-            raise AsrError("未获取到识别结果", audio_url=audio_url)
+            print("警告: 未获取到识别结果", file=sys.stderr)
+            sys.exit(1)
 
         # 4. 生成 SRT 文件
         subtitle_count = asr_to_srt(transcriptions, args.output)
@@ -243,23 +218,8 @@ def main():
     except FileNotFoundError as e:
         print(f"错误: {e}", file=sys.stderr)
         sys.exit(1)
-    except AsrError as e:
-        # 输出结构化的错误信息，包含静态资源 URL
-        error_info = {
-            'error': str(e),
-            'audio_url': e.audio_url or audio_url,
-            'details': e.details
-        }
-        print(f"ASR_ERROR: {json.dumps(error_info, ensure_ascii=False)}", file=sys.stderr)
-        sys.exit(1)
     except Exception as e:
-        # 其他错误也尝试包含静态资源 URL
-        error_info = {
-            'error': str(e),
-            'audio_url': audio_url,
-            'details': None
-        }
-        print(f"ASR_ERROR: {json.dumps(error_info, ensure_ascii=False)}", file=sys.stderr)
+        print(f"错误: {e}", file=sys.stderr)
         sys.exit(1)