update

app.py

@@ -79,26 +79,38 @@ def main():
     
     # Reinitialize NVIDIA client after environment setup
     print("🔄 Reinitializing NVIDIA API client...")
-    from demo import client
-    if client is None and os.getenv("NVIDIA_API_KEY"):
+    
+    # Import demo module and reinitialize client
+    import demo
+    print(f"🔍 Initial client status: {'✅ Available' if demo.client is not None else '❌ None'}")
+    print(f"🔍 API key in environment: {'✅ Present' if os.getenv('NVIDIA_API_KEY') else '❌ Missing'}")
+    
+    if demo.client is None and os.getenv("NVIDIA_API_KEY"):
         print("🔄 Attempting to reinitialize client with environment variables...")
-        # Reinitialize the client
-        import demo
+        # Reinitialize the client and update the module variable
         demo.client = initialize_nvidia_client()
+        print(f"🔄 Client reinitialized: {'✅ Success' if demo.client is not None else '❌ Failed'}")
+    elif demo.client is not None:
+        print("✅ Client already initialized")
+    else:
+        print("❌ No API key available for client initialization")
+    
+    # Final client status check
+    print(f"🔍 Final client status: {'✅ Available' if demo.client is not None else '❌ None'}")
     
     # Test API connection
-    if os.getenv("NVIDIA_API_KEY"):
+    if os.getenv("NVIDIA_API_KEY") and demo.client is not None:
         print("🧪 Testing NVIDIA API connection...")
         test_nvidia_api_connection()
     
     # Create the Gradio interface
     print("🎯 Creating RoutePilot interface...")
-    demo = create_interface()
+    demo_interface = create_interface()
     
     # Launch the application
     # For Hugging Face Spaces, we don't need to specify server_name and port
     # The platform handles this automatically
-    demo.launch(
+    demo_interface.launch(
         show_error=True,
         debug=False,  # Set to False for production
         show_api=False

demo.py

@@ -184,14 +184,15 @@ client = initialize_nvidia_client()
 
 def test_nvidia_api_connection():
     """Test the NVIDIA API connection to verify authentication"""
-    if client is None:
+    current_client = ensure_client_available()
+    if current_client is None:
         print("❌ Cannot test API connection - client not initialized")
         return False
     
     try:
         print("🧪 Testing NVIDIA API connection...")
         # Make a simple test call
-        test_response = client.chat.completions.create(
+        test_response = current_client.chat.completions.create(
             model="meta/llama-3.1-8b-instruct",
             messages=[{"role": "user", "content": "Hello"}],
             max_tokens=10,
@@ -204,6 +205,28 @@ def test_nvidia_api_connection():
         print(f"❌ NVIDIA API connection test failed: {e}")
         return False
 
+def ensure_client_available():
+    """Ensure the NVIDIA API client is available, reinitialize if needed"""
+    global client
+    
+    if client is not None:
+        return client
+    
+    # Try to reinitialize the client
+    print("🔄 Client not available, attempting to reinitialize...")
+    api_key = os.getenv("NVIDIA_API_KEY")
+    if api_key:
+        client = initialize_nvidia_client()
+        if client is not None:
+            print("✅ Client successfully reinitialized")
+            return client
+        else:
+            print("❌ Failed to reinitialize client")
+            return None
+    else:
+        print("❌ No API key available for client initialization")
+        return None
+
 def model_prompting(
     llm_model: str,
     prompt: str,
@@ -226,7 +249,9 @@ def model_prompting(
     Returns:
         Generated text response
     """
-    if client is None:
+    # Ensure client is available
+    current_client = ensure_client_available()
+    if current_client is None:
         raise Exception("NVIDIA API client not initialized. Please check your .env file contains NVIDIA_API_KEY")
     
     # Debug information
@@ -248,7 +273,7 @@ def model_prompting(
     
     try:
         print(f"🚀 Making API call to model: {llm_model}")
-        completion = client.chat.completions.create(
+        completion = current_client.chat.completions.create(
             model=llm_model,
             messages=[{"role": "user", "content": prompt}],
             max_tokens=max_token_num,
@@ -1038,6 +1063,13 @@ def process_query(query):
     try:
         print(f"Processing query: {query[:50]}...")
         
+        # Check client availability before processing
+        print("🔍 Checking client availability...")
+        current_client = ensure_client_available()
+        if current_client is None:
+            raise Exception("NVIDIA API client not available. Please check your API key configuration.")
+        print("✅ Client available for processing")
+        
         # Use Graph Router to select the best LLM
         routed_llm_name, task_description, selection_info = get_routed_llm(query)
         print(f"Graph router selected: {routed_llm_name}")