Update app.py

app.py

@@ -132,16 +132,15 @@ async def get_transcript(youtube_url: str):
     try:
         ydl_opts = {
             'skip_download': True,
-            'writesubtitles': True,       # 仅下载手动上传的字幕
-            'writeautomaticsub': False,   # 禁用自动生成字幕
-            'subtitleslangs': ['original'],  # 关键参数：只获取原始语言
+            'writesubtitles': True,
+            'writeautomaticsub': True,  # 启用自动字幕作为备选
+            'subtitleslangs': ['original'],  # 优先原始语言
             'subtitlesformat': 'best',
             'outtmpl': '%(id)s.%(ext)s',
             'noplaylist': True,
             'cookiefile': "firefox-cookies.txt",
-            # 禁用所有翻译功能
-            'postprocessors': [],
-            'compat_opts': {'no-sub-translate'},
+            'compat_opts': {'no-sub-translate'},  # 保持原始语言
+            'retries': 3,
         }
 
         env_to_cookies_from_env("firefox-cookies.txt")
@@ -150,35 +149,55 @@ async def get_transcript(youtube_url: str):
             info = ydl.extract_info(youtube_url, download=True)
             video_id = info['id']
             
-            # 获取原始语言代码（如视频是英语则返回en）
-            original_lang = info.get('original_lang') or 'en'
-            
-            # 查找匹配原始语言的字幕文件
-            subtitle_pattern = f"{video_id}.{original_lang}.*"
-            subtitle_files = [f for f in os.listdir('.') 
-                             if re.match(subtitle_pattern, f)]
+            # 精确获取原始语言（修复YouTube API返回不一致问题）
+            original_lang = info.get('original_lang') 
+            if not original_lang:
+                original_lang = info.get('subtitles', {}).get('original', [{}])[0].get('language')
+            original_lang = original_lang or 'en'
+
+            # 扩展匹配模式（包含自动生成字幕）
+            subtitle_pattern = re.compile(
+                rf"^{re.escape(video_id)}\.(a\.)?{re.escape(original_lang)}\..+$"
+            )
             
-            logger.info(f"原始语言字幕文件: {subtitle_files}")
+            # 查找所有可能匹配的文件
+            subtitle_files = [
+                f for f in os.listdir('.') 
+                if subtitle_pattern.match(f)
+            ]
+
+            logger.info(f"匹配到的字幕文件: {subtitle_files}")
 
             if subtitle_files:
-                # 选择第一个匹配的文件
-                subtitle_file = subtitle_files[0]
-                with open(subtitle_file, 'r', encoding='utf-8') as f:
+                # 优先手动上传的字幕（非自动生成）
+                preferred_files = [f for f in subtitle_files if 'a.' not in f]
+                selected_file = preferred_files[0] if preferred_files else subtitle_files[0]
+
+                with open(selected_file, 'r', encoding='utf-8') as f:
                     content = f.read()
-                
+
+                # 清理字幕内容
+                clean_content = re.sub(
+                    r'^\d+\n\d{2}:\d{2}:\d{2}[\.,]\d{3} --> \d{2}:\d{2}:\d{2}[\.,]\d{3}\n',
+                    '', 
+                    content,
+                    flags=re.MULTILINE
+                ).strip()
+
                 return {
-                    "transcript": content,
+                    "transcript": clean_content,
                     "language": original_lang,
-                    "is_translated": False
+                    "sub_type": "manual" if 'a.' not in selected_file else "auto-generated"
                 }
-            
+
             return {
-                "error": "无原始语言字幕",
-                "available_langs": list(info.get('subtitles', {}).keys())
+                "error": "未找到原始语言字幕",
+                "available_langs": list(info.get('subtitles', {}).keys()) 
+                                 + list(info.get('automatic_captions', {}).keys())
             }
 
     except Exception as e:
-        return {"error": str(e)}
+        return {"error": f"处理失败: {str(e)}"}