Add better logging and direct AI test endpoint

app.py

@@ -140,6 +140,7 @@ Minimum purchase is 1 roll (67-70 yards)."""
     def generate_response(self, user_message: str) -> str:
         """Generate response using HuggingFace Inference API"""
         if not self.client:
+            logger.warning("No HuggingFace client available, using mock response")
             return self.get_mock_response(user_message)
         
         try:
@@ -151,6 +152,8 @@ Minimum purchase is 1 roll (67-70 yards)."""
                 # For other models, use the full prompt
                 prompt = f"<|system|>\n{self.system_prompt}\n<|user|>\n{user_message}\n<|assistant|>\n"
             
+            logger.info(f"Generating response for prompt: {prompt[:100]}...")
+            
             # Generate response
             response = self.client.text_generation(
                 prompt,
@@ -162,6 +165,8 @@ Minimum purchase is 1 roll (67-70 yards)."""
                 stop_sequences=["<|end|>", "<|user|>", "User:", "Assistant:"]
             )
             
+            logger.info(f"Raw AI response: {response[:200]}...")
+            
             # Clean up the response
             if "Assistant:" in response:
                 assistant_response = response.split("Assistant:")[-1].strip()
@@ -173,8 +178,11 @@ Minimum purchase is 1 roll (67-70 yards)."""
             # Remove any remaining special tokens
             assistant_response = assistant_response.replace("<|end|>", "").replace("<|user|>", "").strip()
             
+            logger.info(f"Cleaned AI response: {assistant_response[:100]}...")
+            
             # If response is too short or generic, use mock response
             if len(assistant_response) < 10 or "I don't know" in assistant_response.lower():
+                logger.warning("AI response too short, using mock response")
                 return self.get_mock_response(user_message)
                 
             return assistant_response
@@ -474,6 +482,48 @@ async def test_trained_model():
         "note": "This is a simulated response for HF Spaces demo"
     }
 
+@app.post("/api/test/ai")
+async def test_ai_directly(request: ChatRequest):
+    """Test AI directly without fallback to mock responses"""
+    try:
+        if not ai_assistant.client:
+            return {
+                "success": False,
+                "message": "No HuggingFace client available",
+                "response": None
+            }
+        
+        # Test with a simple prompt
+        test_prompt = f"User: {request.message}\nAssistant:"
+        
+        logger.info(f"Testing AI with prompt: {test_prompt}")
+        
+        response = ai_assistant.client.text_generation(
+            test_prompt,
+            max_new_tokens=100,
+            temperature=0.7,
+            top_p=0.9,
+            top_k=40
+        )
+        
+        logger.info(f"Direct AI response: {response}")
+        
+        return {
+            "success": True,
+            "message": "AI response generated successfully",
+            "raw_response": response,
+            "model": ai_assistant.model,
+            "api_key_available": bool(ai_assistant.api_key)
+        }
+        
+    except Exception as e:
+        logger.error(f"Error in direct AI test: {e}")
+        return {
+            "success": False,
+            "message": f"Error: {str(e)}",
+            "response": None
+        }
+
 if __name__ == "__main__":
     # Get port from environment variable (Hugging Face Spaces uses 7860)
     port = int(os.getenv("PORT", 7860))