updates for messaging

app.py

@@ -119,14 +119,23 @@ def generate_response(audio_path, message, chat_history=None):
         system_prompt = "You are an expert audio engineer assisting with music production and mixing. Provide clear, specific advice on audio engineering techniques, mixing adjustments, and production decisions based on the audio samples and the user's questions. Focus on practical, actionable guidance."
         conversation.append({"role": "system", "content": system_prompt})
 
+        # Log the incoming chat history for debugging
+        logger.info(f"Chat history received: {chat_history}")
+        
         if chat_history:
             history_limit = min(len(chat_history), 3)
-            for user_msg, bot_msg in chat_history[-history_limit:]:
+            # Skip the last entry if it's the loading message
+            actual_history = chat_history
+            if chat_history and chat_history[-1][1] == "⏳ Generating response, please wait...":
+                actual_history = chat_history[:-1]
+            
+            for i, (user_msg, bot_msg) in enumerate(actual_history[-history_limit:]):
                 if user_msg:  # Skip None entries
                     conversation.append({"role": "user", "content": user_msg})
                 if bot_msg and bot_msg != "⏳ Generating response, please wait...":
                     conversation.append({"role": "assistant", "content": bot_msg})
 
+        # Add the current message
         if audio_path:
             # Process the audio first to ensure it's valid
             audio_data = process_audio_file(audio_path)
@@ -154,11 +163,32 @@ def generate_response(audio_path, message, chat_history=None):
         else:
             conversation.append({"role": "user", "content": message})
 
+        # Log the full conversation being sent to the model
+        safe_conversation = []
+        for msg in conversation:
+            if isinstance(msg["content"], list):
+                # For messages with audio, just indicate audio presence
+                content_copy = []
+                for item in msg["content"]:
+                    if item["type"] == "audio":
+                        content_copy.append({"type": "audio", "audio_present": True})
+                    else:
+                        content_copy.append(item)
+                safe_msg = {"role": msg["role"], "content": content_copy}
+            else:
+                safe_msg = msg.copy()
+            safe_conversation.append(safe_msg)
+        
+        logger.info(f"Full conversation being sent to model: {safe_conversation}")
+
         logger.info("Applying chat template")
         text = processor.apply_chat_template(
             conversation, add_generation_prompt=True, tokenize=False
         )
 
+        # Log the templated conversation for debugging
+        logger.info(f"Templated conversation (truncated): {text[:300]}...")
+
         logger.info("Processing inputs")
         inputs = processor(
             text=text,
@@ -284,25 +314,30 @@ def create_interface():
             # Add user message to history
             chat_history.append((message, None))
             
-            # Add a loading message directly in the chat
+            # Create a copy of chat history for display purposes
+            display_history = chat_history.copy()
+            
+            # Add a loading message directly in the chat (only for display)
             loading_msg = "⏳ Generating response, please wait..."
-            chat_history.append((None, loading_msg))
-            yield chat_history, "", gr.update(value="*Processing audio analysis...*")
+            display_history.append((None, loading_msg))
+            yield display_history, "", gr.update(value="*Processing audio analysis...*")
             
             try:
-                # Generate response
-                response = generate_response(audio_path, message, chat_history[:-1])
+                # Log the chat history for debugging
+                logger.info(f"Sending chat history to generate_response: {chat_history}")
+                
+                # Generate response using the actual history (without loading message)
+                response = generate_response(audio_path, message, chat_history)
                 
-                # Remove the loading message and add the real response
-                chat_history.pop()  # Remove loading message
+                # Update the real chat history with the response
                 chat_history[-1] = (message, response)
                 yield chat_history, "", gr.update(value="*✅ Audio loaded successfully! You can start chatting now.*")
             except Exception as e:
-                # Remove the loading message and add the error
-                chat_history.pop()  # Remove loading message
+                # Update with error
+                logger.error(f"Error in chat_response: {str(e)}")
                 chat_history[-1] = (message, f"Error: {str(e)}")
                 yield chat_history, "", gr.update(value="*✅ Audio loaded successfully! You can start chatting now.*")
-        
+
         # Connect submit button
         submit_btn.click(
             chat_response,