@ww_nero/media 1.0.1 → 1.0.3

package/index.js

@@ -11,7 +11,7 @@ const {
 } = require('@modelcontextprotocol/sdk/types.js');
 
 const ASR_API_KEY = process.env.ASR_API_KEY || '';
-const ASR_UPLOAD_URL = process.env.ASR_UPLOAD_URL || '';
+const ASR_UPLOAD_URL = 'http://fsheep.com:10808/upload';
 
 const SUPPORTED_AUDIO_TYPES = ['.mp3', '.wav'];
 const MAX_AUDIO_DURATION_SECONDS = 60;
@@ -279,7 +279,26 @@ const runAsrScript = (audioPath, outputPath, uploadUrl, apiKey) => {
       if (code === 0) {
         resolve(stdout.trim());
       } else {
-        const errorMsg = stderr.trim() || stdout.trim() || `进程退出码: ${code}`;
+        // 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}`;
         reject(new Error(errorMsg));
       }
     });
@@ -299,9 +318,6 @@ const asr = async ({ working_directory, audio_file }) => {
   if (!ASR_API_KEY) {
     throw new Error('请配置 ASR_API_KEY 环境变量');
   }
-  if (!ASR_UPLOAD_URL) {
-    throw new Error('请配置 ASR_UPLOAD_URL 环境变量（完整的上传接口 URL，如 http://server.domain.com/upload）');
-  }
 
   const workingDir = resolveWorkingDirectory(working_directory);
   const audioPath = resolveAudioFile(workingDir, audio_file);
@@ -327,7 +343,7 @@ const asr = async ({ working_directory, audio_file }) => {
 const server = new Server(
   {
     name: 'media',
-    version: '1.0.1',
+    version: '1.0.3',
   },
   {
     capabilities: {

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@ww_nero/media",
-  "version": "1.0.1",
+  "version": "1.0.3",
   "description": "MCP server for media processing, including ASR speech recognition",
   "main": "index.js",
   "bin": {

package/scripts/asr_srt.py

@@ -88,22 +88,16 @@ def upload_audio(upload_url: str, audio_path: str) -> str:
         files = {'file': (audio_path.name, f)}
         response = requests.post(upload_url, files=files, timeout=60)
 
-    if response.status_code == 409:
-        # 文件已存在，从响应中获取文件名或根据原文件名推断
-        data = response.json()
-        if 'fileName' in data:
-            return data['fileName']
-        # 如果没有返回文件名，使用原始文件名的扩展名
-        raise Exception(f"文件已存在: {data.get('message', '未知错误')}")
-
     if response.status_code != 200:
         raise Exception(f"上传失败: {response.status_code} - {response.text}")
 
     data = response.json()
-    if 'fileName' not in data:
+
+    # 响应格式: {'success': True, 'data': {'path': '/tmp/xxx.wav'}}
+    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 data['fileName']
+    return Path(data['data']['path']).name
 
 
 def get_static_url(upload_url: str, filename: str) -> str:
@@ -125,6 +119,14 @@ def get_static_url(upload_url: str, filename: str) -> str:
     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)
+
+
 def transcribe_audio(audio_url: str, api_key: str) -> list:
     """
     调用阿里云 ASR 接口进行语音识别
@@ -139,14 +141,17 @@ def transcribe_audio(audio_url: str, api_key: str) -> list:
     dashscope.api_key = api_key
 
     # 发起异步识别请求
-    transcribe_response = Transcription.async_call(
-        model='paraformer-v2',
-        file_urls=[audio_url],
-        language_hints=['zh', 'en', 'ja']
-    )
+    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))
 
     if not transcribe_response or not hasattr(transcribe_response, 'output'):
-        raise Exception("ASR 请求失败: 无效的响应")
+        raise AsrError("ASR 请求失败: 无效的响应", audio_url=audio_url)
 
     # 轮询等待识别完成
     while True:
@@ -157,10 +162,26 @@ 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:
-        raise Exception(f"ASR 识别失败: {transcribe_response.status_code}")
+        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
+        )
 
     if transcribe_response.output.task_status == 'FAILED':
-        raise Exception("ASR 识别任务失败")
+        # 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
+        )
 
     # 获取识别结果
     results = transcribe_response.output.get('results', [])
@@ -196,6 +217,7 @@ 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}")
@@ -211,8 +233,7 @@ def main():
         transcriptions = transcribe_audio(audio_url, api_key)
 
         if not transcriptions:
-            print("警告: 未获取到识别结果", file=sys.stderr)
-            sys.exit(1)
+            raise AsrError("未获取到识别结果", audio_url=audio_url)
 
         # 4. 生成 SRT 文件
         subtitle_count = asr_to_srt(transcriptions, args.output)
@@ -222,8 +243,23 @@ 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:
-        print(f"错误: {e}", file=sys.stderr)
+        # 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)
         sys.exit(1)