add logging

app.py

@@ -104,21 +104,25 @@ class RAGChatbot:
         
         # Initialize Inference API client
         hf_token = os.environ.get("HF_TOKEN")
+        # Debug: report HF_TOKEN status (masked)
         if not hf_token:
+            print("[DEBUG] HF_TOKEN: not set (empty or missing)")
             print("Warning: HF_TOKEN not set. Inference API calls may fail.")
             print("Set HF_TOKEN environment variable or add it to Space secrets.")
         else:
+            masked = f"{hf_token[:4]}...{hf_token[-4:]}" if len(hf_token) > 8 else "****"
+            print(f"[DEBUG] HF_TOKEN: set (length={len(hf_token)}, masked={masked})")
             print("HF_TOKEN found. Inference API ready.")
         
-        print(f"Initializing Inference API client for model: {model_name}")
+        print(f"[DEBUG] Initializing Inference API client for model: {model_name}")
         try:
             self.inference_client = InferenceClient(
                 model=model_name,
                 token=hf_token
             )
-            print("Inference API client initialized successfully")
+            print("[DEBUG] Inference API client initialized successfully (using chat_completion for this model)")
         except Exception as e:
-            print(f"Error initializing Inference API client: {e}")
+            print(f"[DEBUG] Error initializing Inference API client: {type(e).__name__}: {e}")
             self.inference_client = None
         
         # Initialize document ingestion
@@ -138,18 +142,27 @@ class RAGChatbot:
     
     def _generate_with_chat(self, user_content: str, max_new_tokens: int = 512) -> str:
         """Call the Inference API using chat/comversational endpoint (required for Mistral instruct)."""
-        response = self.inference_client.chat_completion(
-            model=self.model_name,
-            messages=[{"role": "user", "content": user_content}],
-            max_tokens=max_new_tokens,
-            temperature=0.7,
-        )
-        # ChatCompletion has choices[0].message.content
-        if response and response.choices and len(response.choices) > 0:
-            msg = response.choices[0].message
-            if hasattr(msg, "content") and msg.content:
-                return msg.content.strip()
-        return ""
+        print(f"[DEBUG] _generate_with_chat: model={self.model_name}, API=chat_completion, prompt_len={len(user_content)}, max_tokens={max_new_tokens}")
+        try:
+            response = self.inference_client.chat_completion(
+                model=self.model_name,
+                messages=[{"role": "user", "content": user_content}],
+                max_tokens=max_new_tokens,
+                temperature=0.7,
+            )
+            print(f"[DEBUG] chat_completion response type: {type(response).__name__}")
+            # ChatCompletion has choices[0].message.content
+            if response and response.choices and len(response.choices) > 0:
+                msg = response.choices[0].message
+                if hasattr(msg, "content") and msg.content:
+                    return msg.content.strip()
+            print("[DEBUG] chat_completion returned empty or unexpected structure")
+            return ""
+        except Exception as e:
+            print(f"[DEBUG] _generate_with_chat exception: {type(e).__name__}: {e}")
+            import traceback
+            traceback.print_exc()
+            raise
     
     def generate_response(self, query: str, use_rag: bool = True, num_results: int = 5) -> str:
         """
@@ -197,7 +210,7 @@ Answer:"""
                             return response_text
                         raise ValueError("Empty response from model")
                     except Exception as api_error:
-                        print(f"Error calling Inference API: {api_error}")
+                        print(f"[DEBUG] RAG generation failed: {type(api_error).__name__}: {api_error}")
                         # Fallback: return formatted chunks with note
                         response_parts = []
                         response_parts.append("I retrieved relevant information, but couldn't generate a synthesized answer. Here are the relevant chunks:\n\n")