Update app.py

app.py

@@ -1374,31 +1374,109 @@ def analyze_image(file_path: str, query: str) -> str:
         raise ToolError("analyze_image", e)
 
 
-class YoutubeInput(BaseModel):
+class YoutubeInput(BaseModel:
     video_url: str = Field(description="YouTube URL")
 
 @tool(args_schema=YoutubeInput)
-@retry_with_backoff(max_retries=2)
 def get_youtube_transcript(video_url: str) -> str:
-    """Get YouTube transcript using AssemblyAI"""
+    """Get YouTube transcript using AssemblyAI with proper status handling"""
     start_time = time.time()
     
     try:
         aai.settings.api_key = os.getenv("ASSEMBLYAI_API_KEY")
-        print(f"📺 Transcribing: {video_url}")
+        if not aai.settings.api_key:
+            raise ValueError("ASSEMBLYAI_API_KEY not set in Space secrets")
+        
+        print(f"📺 Transcribing YouTube: {video_url}")
         
+        # Validate URL
+        if not ("youtube.com" in video_url or "youtu.be" in video_url):
+            raise ValueError(f"Invalid YouTube URL: {video_url}")
+        
+        # Submit transcription request
         transcriber = aai.Transcriber()
-        transcript = transcriber.transcribe(video_url)
+        print(f"   Submitting to AssemblyAI...")
+        
+        config = aai.TranscriptionConfig(
+            speech_model=aai.SpeechModel.best,
+        )
+        
+        transcript = transcriber.transcribe(video_url, config=config)
+        
+        # Wait for completion
+        print(f"   Initial status: {transcript.status}")
         
+        # Poll for completion (max 5 minutes)
+        max_wait = 300
+        poll_interval = 5
+        elapsed = 0
+        
+        while transcript.status == aai.TranscriptStatus.queued or transcript.status == aai.TranscriptStatus.processing:
+            if elapsed >= max_wait:
+                raise TimeoutError(f"Transcription timed out after {max_wait}s. Video may be too long.")
+            
+            time.sleep(poll_interval)
+            elapsed += poll_interval
+            
+            # Refresh transcript object
+            try:
+                transcript = transcriber.get_transcript(transcript.id)
+                print(f"   Status after {elapsed}s: {transcript.status}")
+            except Exception as refresh_err:
+                print(f"   Warning: Could not refresh status: {refresh_err}")
+                # Continue anyway, maybe it finished
+                break
+        
+        # Check final status
         if transcript.status == aai.TranscriptStatus.error:
-            raise RuntimeError(transcript.error)
+            error_msg = getattr(transcript, 'error', 'Unknown error')
+            raise RuntimeError(f"AssemblyAI transcription failed: {error_msg}")
+        
+        if transcript.status != aai.TranscriptStatus.completed:
+            raise RuntimeError(f"Unexpected status: {transcript.status}")
+        
+        # Extract text
+        if not hasattr(transcript, 'text'):
+            raise AttributeError("Transcript object has no 'text' attribute")
+        
+        result_text = transcript.text
+        
+        if not result_text or not isinstance(result_text, str):
+            raise ValueError(f"Transcript text is invalid: {type(result_text)}")
+        
+        result_text = result_text.strip()
+        
+        if len(result_text) < 10:
+            raise ValueError(f"Transcript too short ({len(result_text)} chars). Video may have no audio.")
+        
+        print(f"✓ Transcribed {len(result_text)} chars")
         
         telemetry.record_call("get_youtube_transcript", time.time() - start_time, True)
-        return f"Transcript:\n{transcript.text}"
+        return f"YouTube Transcript:\n{truncate_if_needed(result_text)}"
+        
+    except aai.types.TranscriptError as e:
+        telemetry.record_call("get_youtube_transcript", time.time() - start_time, False)
+        error_msg = str(e)
+        
+        suggestions = []
+        if "not found" in error_msg.lower():
+            suggestions.append("Video may be private or deleted")
+        if "quota" in error_msg.lower() or "limit" in error_msg.lower():
+            suggestions.append("AssemblyAI quota exceeded")
+        if "timeout" in error_msg.lower():
+            suggestions.append("Video may be too long (try shorter video)")
+        
+        suggestion_text = " | ".join(suggestions) if suggestions else "Check video URL and API quota"
         
+        raise ToolError("get_youtube_transcript", e, suggestion_text)
+    
+    except TimeoutError as e:
+        telemetry.record_call("get_youtube_transcript", time.time() - start_time, False)
+        raise ToolError("get_youtube_transcript", e, "Video too long or AssemblyAI overloaded. Try shorter video.")
+    
     except Exception as e:
         telemetry.record_call("get_youtube_transcript", time.time() - start_time, False)
-        raise ToolError("get_youtube_transcript", e)
+        raise ToolError("get_youtube_transcript", e, "Check video URL is valid and public")
 
 
 class BrowseInput(BaseModel):