Create demo unconditionally at module level - ensure Spaces can always detect it

app.py

@@ -836,51 +836,56 @@ def main():
 # Pattern: demo = gr.Interface(...) or demo = gr.Blocks(...)
 # DO NOT call demo.launch() - Spaces handles that automatically
 
-# Check if we're on Spaces
-IS_SPACES = os.getenv("SPACE_ID") is not None or os.getenv("SYSTEM") == "spaces"
+# Check if we're on Spaces (be more permissive - check multiple env vars)
+IS_SPACES = (
+    os.getenv("SPACE_ID") is not None or 
+    os.getenv("SYSTEM") == "spaces" or
+    os.getenv("HF_SPACE_ID") is not None
+)
 
-# Create demo at module level for Spaces (similar to HF docs example)
-# For local execution, main() will create its own demo
-if IS_SPACES:
-    try:
+# Create demo unconditionally at module level (like HF docs example)
+# This ensures Spaces can always find it when importing the module
+# For local execution, we'll still create it here but main() will also create its own
+try:
+    if IS_SPACES:
         logger.info("Initializing for Hugging Face Spaces...")
-        # 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 the demo interface directly at module level (like HF docs example)
-        demo = create_interface(bot, use_inference_api=True)
-        logger.info(f"Demo created successfully: {type(demo)}")
-    except Exception as e:
-        logger.error(f"Error creating demo for Spaces: {e}", exc_info=True)
-        # Create a fallback error demo so Spaces doesn't show blank
-        with gr.Blocks() as demo:
-            gr.Markdown(f"# Error Initializing Chatbot\n\nAn error occurred while initializing the chatbot.\n\nError: {str(e)}\n\nPlease check the logs for details.")
-else:
-    # Local execution: demo will be created in main()
-    demo = None
+    else:
+        logger.info("Initializing for local execution...")
+    
+    # Initialize with default args
+    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
+    args.skip_model_loading = IS_SPACES  # Skip model loading on Spaces, use Inference API
+    
+    # Create bot
+    bot = RAGBot(args)
+    
+    if bot.vector_retriever is None:
+        raise Exception("Vector database not available")
+    
+    # Create the demo interface directly at module level (like HF docs example)
+    demo = create_interface(bot, use_inference_api=IS_SPACES)
+    logger.info(f"Demo created successfully: {type(demo)}")
+except Exception as e:
+    logger.error(f"Error creating demo: {e}", exc_info=True)
+    # Create a fallback error demo so Spaces doesn't show blank
+    with gr.Blocks() as demo:
+        gr.Markdown(f"# Error Initializing Chatbot\n\nAn error occurred while initializing the chatbot.\n\nError: {str(e)}\n\nPlease check the logs for details.")
 
 # For local execution only (not on Spaces)
 if __name__ == "__main__":