Upload

app/main.py

@@ -12,6 +12,8 @@ import io
 import google.generativeai as genai
 import time
 from collections import deque
+import yt_dlp
+
 
 logging.basicConfig(level=logging.INFO)
 logger = logging.getLogger(__name__)
@@ -36,6 +38,10 @@ class ChatMessage(BaseModel):
     message: str
     transcript: str
 
+class VideoRequest(BaseModel):
+    url: str
+
+
 RATE_LIMIT = 15
 RATE_WINDOW = 60
 request_timestamps = deque()
@@ -127,46 +133,35 @@ def _patched_execute_request(url, method="GET", headers=None, data=None):
 request_module._execute_request = _patched_execute_request
 
 @app.post("/transcribe")
-async def transcribe_video(video: VideoURL):
+async def transcribe_video(request: VideoRequest):
+    url = request.url
     try:
-        yt = pytubefix.YouTube(str(video.url))
-        audio_stream = yt.streams.filter(only_audio=True, file_extension='mp4').first()
-        if not audio_stream:
-            raise HTTPException(status_code=400, detail="No audio stream found for this video")
-
-        video_description = yt.description
-
-        temp_dir = "/tmp"  # Use /tmp directory for Hugging Face Spaces
-        audio_file = audio_stream.download(output_path=temp_dir)
-        audio_file_mp3 = audio_file + ".mp3"
-        os.rename(audio_file, audio_file_mp3)
-
-        transcript = await transcribe_audio(audio_file_mp3)
-        
-        if not transcript:
-            raise HTTPException(status_code=500, detail="Transcription failed")
-
-        full_text = transcript['results']['channels'][0]['alternatives'][0]['transcript']
-
-        summary = generate_summary(full_text, video_description, video.summary_length)
-        quiz = generate_quiz(full_text, video_description)
-
-        result = {
-            "transcription": full_text,
-            "summary": summary,
-            "video_description": video_description,
-            "quiz": quiz,
-            "detailed_transcript": transcript
+        # Download the audio from the YouTube video
+        ydl_opts = {
+            'format': 'bestaudio/best',
+            'postprocessors': [{
+                'key': 'FFmpegExtractAudio',
+                'preferredcodec': 'mp3',
+                'preferredquality': '192',
+            }],
+            'outtmpl': 'downloads/%(id)s.%(ext)s',
         }
+        
+        with yt_dlp.YoutubeDL(ydl_opts) as ydl:
+            info_dict = ydl.extract_info(url, download=True)
+            audio_file_path = ydl.prepare_filename(info_dict)
 
-        os.remove(audio_file_mp3)
+        # Check if the audio file exists
+        if not os.path.exists(audio_file_path):
+            raise HTTPException(status_code=500, detail="Failed to download the audio")
 
-        return result
+        # Return the path of the downloaded audio file
+        return {"audio_file": audio_file_path}
 
+    except yt_dlp.utils.DownloadError as e:
+        raise HTTPException(status_code=500, detail=f"Download error: {str(e)}")
     except Exception as e:
-        logger.error(f"Error processing video: {str(e)}")
-        logger.exception("Full traceback:")
-        raise HTTPException(status_code=500, detail=str(e))
+        raise HTTPException(status_code=500, detail=f"An error occurred: {str(e)}")
 
 @app.post("/generate_audio_summary")
 async def generate_audio_summary(summary: str):