Simplify demo creation - remove try/except wrapper for cleaner module-level assignment

app.py

@@ -833,54 +833,35 @@ def main():
 # For Hugging Face Spaces: create demo at module level
 # Spaces will import this module and look for a 'demo' variable
 def create_demo_for_spaces():
-    """Create demo for Hugging Face Spaces"""
-    try:
-        # Initialize with default args for Spaces
-        parser = argparse.ArgumentParser()
-        # Use Llama-3.2-3B as default (will use float16 on CPU to save memory)
-        # For Spaces with limited memory, consider upgrading hardware tier
-        parser.add_argument('--model', type=str, default='meta-llama/Llama-3.2-3B-Instruct')
-        parser.add_argument('--vector-db-dir', default='./chroma_db')
-        parser.add_argument('--data-dir', default='./Data Resources')
-        parser.add_argument('--max-new-tokens', type=int, default=1024)
-        parser.add_argument('--temperature', type=float, default=0.2)
-        parser.add_argument('--top-p', type=float, default=0.9)
-        parser.add_argument('--repetition-penalty', type=float, default=1.1)
-        parser.add_argument('--k', type=int, default=5)
-        parser.add_argument('--skip-indexing', action='store_true', default=True)
-        parser.add_argument('--verbose', action='store_true', default=False)
-        parser.add_argument('--share', action='store_true', default=False)
-        parser.add_argument('--server-name', type=str, default='0.0.0.0')
-        parser.add_argument('--server-port', type=int, default=7860)
-        parser.add_argument('--seed', type=int, default=42)
-        
-        args = parser.parse_args([])  # Empty args for Spaces
-        
-        # Create bot but skip model loading (we'll use Inference API)
-        # We still need the vector database
-        # Set a flag to skip model loading
-        args.skip_model_loading = True
-        
-        # Create bot - it will skip model loading due to the flag
-        bot = RAGBot(args)
-        
-        # Don't load the model - we'll use Inference API
-        # Just verify vector DB is available
-        if bot.vector_retriever is None:
-            raise Exception("Vector database not available")
-        
-        # Use Inference API instead of loading model
-        interface_demo = create_interface(bot, use_inference_api=True)
-        logger.info("Demo interface created successfully")
-        return interface_demo
-    except Exception as e:
-        logger.error(f"Error creating demo for Spaces: {e}", exc_info=True)
-        import traceback
-        logger.error(f"Traceback: {traceback.format_exc()}")
-        # Return a simple error demo
-        with gr.Blocks() as error_demo:
-            gr.Markdown(f"# Error Initializing Chatbot\n\nAn error occurred: {str(e)}\n\nPlease check the logs for details.")
-        return error_demo
+    """Create demo for Hugging Face Spaces - called at module level"""
+    # Initialize with default args for Spaces
+    parser = argparse.ArgumentParser()
+    parser.add_argument('--model', type=str, default='meta-llama/Llama-3.2-3B-Instruct')
+    parser.add_argument('--vector-db-dir', default='./chroma_db')
+    parser.add_argument('--data-dir', default='./Data Resources')
+    parser.add_argument('--max-new-tokens', type=int, default=1024)
+    parser.add_argument('--temperature', type=float, default=0.2)
+    parser.add_argument('--top-p', type=float, default=0.9)
+    parser.add_argument('--repetition-penalty', type=float, default=1.1)
+    parser.add_argument('--k', type=int, default=5)
+    parser.add_argument('--skip-indexing', action='store_true', default=True)
+    parser.add_argument('--verbose', action='store_true', default=False)
+    parser.add_argument('--share', action='store_true', default=False)
+    parser.add_argument('--server-name', type=str, default='0.0.0.0')
+    parser.add_argument('--server-port', type=int, default=7860)
+    parser.add_argument('--seed', type=int, default=42)
+    
+    args = parser.parse_args([])  # Empty args for Spaces
+    args.skip_model_loading = True  # Skip model loading, use Inference API
+    
+    # Create bot (will skip model loading)
+    bot = RAGBot(args)
+    
+    if bot.vector_retriever is None:
+        raise Exception("Vector database not available")
+    
+    # Create and return the demo interface
+    return create_interface(bot, use_inference_api=True)
 
 # For Hugging Face Spaces: create demo at module level
 # Spaces will import this module and look for a 'demo' variable
@@ -891,25 +872,13 @@ IS_SPACES = os.getenv("SPACE_ID") is not None or os.getenv("SYSTEM") == "spaces"
 
 # For Hugging Face Spaces: create demo at module level
 # Spaces will import this module and look for a 'demo' variable
-# Similar to the basic Gradio example: demo = gr.Interface(...)
+# Pattern: demo = gr.Interface(...) or demo = gr.Blocks(...)
+# DO NOT call demo.launch() - Spaces handles that automatically
+
 if IS_SPACES:
-    try:
-        logger.info("Creating demo for Hugging Face Spaces...")
-        demo = create_demo_for_spaces()
-        logger.info(f"Demo created successfully. Type: {type(demo)}")
-        if demo is None:
-            raise Exception("Demo creation returned None")
-        if not isinstance(demo, gr.Blocks):
-            raise Exception(f"Demo is not a Gradio Blocks object, got {type(demo)}")
-        logger.info("✅ Demo is ready for Spaces")
-    except Exception as e:
-        logger.error(f"Error creating demo: {e}", exc_info=True)
-        import traceback
-        logger.error(f"Traceback: {traceback.format_exc()}")
-        # Create error demo
-        with gr.Blocks() as error_demo:
-            gr.Markdown(f"# Error Initializing Chatbot\n\nAn error occurred: {str(e)}\n\nPlease check the logs.")
-        demo = error_demo
+    logger.info("Initializing for Hugging Face Spaces...")
+    demo = create_demo_for_spaces()
+    logger.info(f"Demo created: {type(demo)}")
 else:
     # Local execution: demo will be created in main()
     demo = None