Update app.py

app.py

@@ -130,41 +130,61 @@ async def get_video_url(youtube_url: str):
 @app.get("/script")
 async def get_transcript(youtube_url: str, language: str = None):
     try:
-        # If no specific language is requested, we'll try to get any available subtitle
+        # Set up yt-dlp options
         ydl_opts = {
             'skip_download': True,
             'writesubtitles': True,
             'writeautomaticsub': True,
-            'subtitleslangs': ['all'] if not language else [language],
-            'subtitlesformat': 'best',
             'outtmpl': '%(id)s.%(ext)s',
             'noplaylist': True,
             'cookiefile': "firefox-cookies.txt"
         }
+        
+        # If a language is specified, only download that language
+        # Otherwise, we'll first get video info to determine the original language
+        if language:
+            ydl_opts['subtitleslangs'] = [language]
+        
         env_to_cookies_from_env("firefox-cookies.txt")
-        # Show current directory structure before download
         logger.info(f"Current directory files (before): {os.listdir('.')}")
         
-        with yt_dlp.YoutubeDL(ydl_opts) as ydl:
-            info = ydl.extract_info(youtube_url, download=True)
+        # First, get video info without downloading anything
+        with yt_dlp.YoutubeDL({**ydl_opts, 'skip_download': True, 'writesubtitles': False, 'writeautomaticsub': False}) as ydl:
+            info = ydl.extract_info(youtube_url, download=False)
             video_id = info['id']
             logger.info(f"Video ID: {video_id}")
             
+            # If no language specified, try to use the original language
+            if not language:
+                # Try to determine the original language if available in the info
+                if 'subtitles' in info and info['subtitles']:
+                    # Use the first available subtitle language
+                    available_languages = list(info['subtitles'].keys())
+                    if available_languages:
+                        language = available_languages[0]
+                        logger.info(f"Using detected language: {language}")
+                        ydl_opts['subtitleslangs'] = [language]
+                else:
+                    # Fall back to 'en' if can't determine
+                    language = 'en'
+                    ydl_opts['subtitleslangs'] = [language]
+        
+        # Now download the subtitle in the selected language
+        with yt_dlp.YoutubeDL(ydl_opts) as ydl:
+            ydl.extract_info(youtube_url, download=True)
+            
             # Check actual downloaded files
             logger.info(f"Current directory files (after extraction): {os.listdir('.')}")
             
-            # First see if we can find a subtitle file for the requested language
-            subtitle_files = []
-            if language:
-                subtitle_files = [f for f in os.listdir('.')
-                                if f.startswith(video_id) and (language in f)]
+            # Look for the subtitle file with the specified language
+            subtitle_files = [f for f in os.listdir('.')
+                             if f.startswith(video_id) and any(ext in f for ext in ['.vtt', '.srt', '.ttml', '.json3'])]
             
-            # If no specific language requested or no files found for requested language,
-            # get any subtitle file for this video
-            if not subtitle_files:
-                subtitle_files = [f for f in os.listdir('.')
-                                if f.startswith(video_id) and 
-                                any(f.endswith(ext) for ext in ['.vtt', '.srt', '.ttml', '.json3'])]
+            # If specific language requested, filter for that language
+            if language:
+                lang_subtitle_files = [f for f in subtitle_files if language in f]
+                if lang_subtitle_files:
+                    subtitle_files = lang_subtitle_files
             
             logger.info(f"Potential subtitle files: {subtitle_files}")
             
@@ -197,7 +217,15 @@ async def get_transcript(youtube_url: str, language: str = None):
                 else:
                     text = f"Unsupported format: {subtitle_file}"
                 
-                return {"transcript": text, "language": subtitle_file.split('.')[-2] if '.' in subtitle_file else "unknown"}
+                # Clean up files to avoid cluttering the directory
+                for f in subtitle_files:
+                    try:
+                        os.remove(f)
+                    except:
+                        logger.warning(f"Could not remove file: {f}")
+                
+                detected_language = subtitle_file.split('.')[-2] if '.' in subtitle_file else "unknown"
+                return {"transcript": text, "language": detected_language}
             
             return {"transcript": f"No subtitle files found for {video_id}", "language": "none"}
     except Exception as e: