Fix chat response issues: st.experimental_rerun(), remove nested chat messages, increase session timeout

app.py

@@ -257,6 +257,7 @@ if user_input and not st.session_state.is_processing:
     if not is_valid:
         st.error(validated_input)
         st.session_state.is_processing = False
+        st.experimental_rerun()  # Fixed: use experimental_rerun
     else:
         st.session_state.is_processing = True
         
@@ -272,187 +273,182 @@ if user_input and not st.session_state.is_processing:
         })
         
         # Process AI response
-        with st.chat_message("assistant"):
-            status_placeholder = st.empty()
-            response_placeholder = st.empty()
+        response_container = st.empty()
+        status_placeholder = st.empty()
+        response_placeholder = st.empty()
+        
+        try:
+            # Get conversation history from session
+            user_session = session_manager.get_session("default_user")
+            conversation_history = user_session.get("conversation", []).copy()
             
-            try:
-                # Get conversation history from session
-                user_session = session_manager.get_session("default_user")
-                conversation_history = user_session.get("conversation", []).copy()
-                
-                # Add the current user message to history for context
-                conversation_history.append({"role": "user", "content": validated_input})
+            # Add the current user message to history for context
+            conversation_history.append({"role": "user", "content": validated_input})
+            
+            # Check if cosmic mode is enabled
+            if st.session_state.cosmic_mode:
+                # Process cosmic cascade response
+                status_placeholder.info("🐱 Cosmic Kitten Responding...")
                 
-                # Check if cosmic mode is enabled
-                if st.session_state.cosmic_mode:
-                    # Process cosmic cascade response
-                    status_placeholder.info("🐱 Cosmic Kitten Responding...")
+                try:
+                    # Get conversation history
+                    user_session = session_manager.get_session("default_user")
+                    conversation_history = user_session.get("conversation", []).copy()
+                    conversation_history.append({"role": "user", "content": validated_input})
                     
-                    try:
-                        # Get conversation history
-                        user_session = session_manager.get_session("default_user")
-                        conversation_history = user_session.get("conversation", []).copy()
-                        conversation_history.append({"role": "user", "content": validated_input})
-                        
-                        # Stage 1: Local Ollama Response
-                        ollama_provider = OllamaProvider(st.session_state.selected_model)
-                        local_response = ollama_provider.generate(validated_input, conversation_history)
-                        
-                        if local_response:
-                            # Display response
-                            st.markdown(f"### 🐱 Cosmic Kitten Says:\n{local_response}")
-                            st.session_state.messages.append({
-                                "role": "assistant",
-                                "content": local_response,
-                                "source": "local_kitty",
-                                "timestamp": datetime.now().strftime("%H:%M:%S")
-                            })
-                        
-                        # Stage 2: HF Endpoint Analysis
-                        status_placeholder.info("🛰️ Beaming Query to Orbital Station...")
-                        if config.hf_token:
-                            # Check HF status first
-                            hf_status = hf_monitor.check_endpoint_status()
-                            if not hf_status['available']:
-                                status_placeholder.info(personality.get_initializing_message())
-                            
-                            hf_provider = HuggingFaceProvider("meta-llama/Llama-2-7b-chat-hf")
-                            hf_response = hf_provider.generate(validated_input, conversation_history)
-                            
-                            if hf_response:
-                                # Display response
-                                st.markdown(f"### 🛰️ Orbital Station Reports:\n{hf_response}")
-                                st.session_state.messages.append({
-                                    "role": "assistant",
-                                    "content": hf_response,
-                                    "source": "orbital_station",
-                                    "timestamp": datetime.now().strftime("%H:%M:%S")
-                                })
-                        
-                        # Stage 3: Local Synthesis
-                        status_placeholder.info("🐱 Cosmic Kitten Synthesizing Wisdom...")
-                        
-                        # Update history with both responses
-                        synthesis_history = conversation_history.copy()
-                        synthesis_history.extend([
-                            {"role": "assistant", "content": local_response},
-                            {"role": "assistant", "content": hf_response, "source": "cloud"}
-                        ])
+                    # Stage 1: Local Ollama Response
+                    ollama_provider = OllamaProvider(st.session_state.selected_model)
+                    local_response = ollama_provider.generate(validated_input, conversation_history)
+                    
+                    if local_response:
+                        # Display response (no nested st.chat_message)
+                        st.markdown(f"### 🐱 Cosmic Kitten Says:\n{local_response}")
+                        st.session_state.messages.append({
+                            "role": "assistant",
+                            "content": local_response,
+                            "source": "local_kitty",
+                            "timestamp": datetime.now().strftime("%H:%M:%S")
+                        })
+                    
+                    # Stage 2: HF Endpoint Analysis
+                    status_placeholder.info("🛰️ Beaming Query to Orbital Station...")
+                    if config.hf_token:
+                        # Check HF status first
+                        hf_status = hf_monitor.check_endpoint_status()
+                        if not hf_status['available']:
+                            status_placeholder.info(personality.get_initializing_message())
                         
-                        synthesis = ollama_provider.generate(
-                            f"Synthesize these two perspectives:\n1. Local: {local_response}\n2. Cloud: {hf_response}",
-                            synthesis_history
-                        )
+                        hf_provider = HuggingFaceProvider("meta-llama/Llama-2-7b-chat-hf")
+                        hf_response = hf_provider.generate(validated_input, conversation_history)
                         
-                        if synthesis:
-                            # Display response
-                            st.markdown(f"### 🌟 Final Cosmic Summary:\n{synthesis}")
+                        if hf_response:
+                            # Display response (no nested st.chat_message)
+                            st.markdown(f"### 🛰️ Orbital Station Reports:\n{hf_response}")
                             st.session_state.messages.append({
                                 "role": "assistant",
-                                "content": synthesis,
-                                "source": "cosmic_summary",
+                                "content": hf_response,
+                                "source": "orbital_station",
                                 "timestamp": datetime.now().strftime("%H:%M:%S")
                             })
-                        
-                        status_placeholder.success("✨ Cosmic Cascade Complete!")
-                        
-                    except Exception as e:
-                        error_msg = f"🌌 Cosmic disturbance: {str(e)}"
-                        st.error(error_msg)
+                    
+                    # Stage 3: Local Synthesis
+                    status_placeholder.info("🐱 Cosmic Kitten Synthesizing Wisdom...")
+                    
+                    # Update history with both responses
+                    synthesis_history = conversation_history.copy()
+                    synthesis_history.extend([
+                        {"role": "assistant", "content": local_response},
+                        {"role": "assistant", "content": hf_response, "source": "cloud"}
+                    ])
+                    
+                    synthesis = ollama_provider.generate(
+                        f"Synthesize these two perspectives:\n1. Local: {local_response}\n2. Cloud: {hf_response}",
+                        synthesis_history
+                    )
+                    
+                    if synthesis:
+                        # Display response (no nested st.chat_message)
+                        st.markdown(f"### 🌟 Final Cosmic Summary:\n{synthesis}")
                         st.session_state.messages.append({
                             "role": "assistant",
-                            "content": error_msg,
-                            "source": "error",
+                            "content": synthesis,
+                            "source": "cosmic_summary",
                             "timestamp": datetime.now().strftime("%H:%M:%S")
                         })
-                else:
-                    # Traditional processing
-                    # Try Ollama first
-                    status_placeholder.info("🦙 Contacting Ollama...")
-                    ai_response = None
                     
-                    try:
-                        # Use the OllamaProvider directly
-                        ollama_provider = OllamaProvider(st.session_state.selected_model)
-                        ai_response = ollama_provider.generate(validated_input, conversation_history)
+                    status_placeholder.success("✨ Cosmic Cascade Complete!")
+                    
+                except Exception as e:
+                    error_msg = f"🌌 Cosmic disturbance: {str(e)}"
+                    st.error(error_msg)
+                    st.session_state.messages.append({
+                        "role": "assistant",
+                        "content": error_msg,
+                        "source": "error",
+                        "timestamp": datetime.now().strftime("%H:%M:%S")
+                    })
+            else:
+                # Traditional processing
+                # Try Ollama first
+                status_placeholder.info("🦙 Contacting Ollama...")
+                ai_response = None
+                
+                try:
+                    # Use the OllamaProvider directly with proper configuration
+                    ollama_provider = OllamaProvider(st.session_state.selected_model)
+                    ai_response = ollama_provider.generate(validated_input, conversation_history)
+                    
+                    if ai_response:
+                        st.markdown(ai_response)  # Use st.markdown instead of response_placeholder
+                        status_placeholder.success("✅ Response received!")
+                    else:
+                        status_placeholder.warning("⚠️ Empty response from Ollama")
                         
-                        if ai_response:
-                            st.markdown(ai_response)  # Use st.markdown instead of response_placeholder
-                            status_placeholder.success("✅ Response received!")
-                        else:
-                            status_placeholder.warning("⚠️ Empty response from Ollama")
-                            
-                    except Exception as ollama_error:
-                        user_msg = translate_error(ollama_error)
-                        status_placeholder.error(f"⚠️ {user_msg}")
+                except Exception as ollama_error:
+                    error_message = str(ollama_error)
+                    status_placeholder.error(f"❌ Ollama error: {error_message[:100]}...")
+                    logger.error(f"Ollama error: {error_message}")
+                    
+                    # Fallback to HF if available
+                    if config.hf_token and not ai_response:
+                        status_placeholder.info("⚡ Initializing HF Endpoint (2–4 minutes)...")
                         
-                        # Fallback to HF if available
-                        if config.hf_token and not ai_response:
-                            status_placeholder.info("⚡ Initializing HF Endpoint (2–4 minutes)...")
-                            
-                            try:
-                                # Check HF status first
-                                hf_status = hf_monitor.check_endpoint_status()
-                                if not hf_status['available']:
-                                    status_placeholder.info(personality.get_initializing_message())
-                                    
-                                # Use the HuggingFaceProvider directly
-                                hf_provider = HuggingFaceProvider("meta-llama/Llama-2-7b-chat-hf")
-                                ai_response = hf_provider.generate(validated_input, conversation_history)
+                        try:
+                            # Check HF status first
+                            hf_status = hf_monitor.check_endpoint_status()
+                            if not hf_status['available']:
+                                status_placeholder.info(personality.get_initializing_message())
                                 
-                                if ai_response:
-                                    st.markdown(ai_response)  # Use st.markdown instead of response_placeholder
-                                    status_placeholder.success("✅ HF response received!")
-                                else:
-                                    status_placeholder.error("❌ No response from HF")
-                                    
-                            except Exception as hf_error:
-                                user_msg = translate_error(hf_error)
-                                status_placeholder.error(f"⚠️ {user_msg}")
+                            # Use the HuggingFaceProvider directly
+                            hf_provider = HuggingFaceProvider("meta-llama/Llama-2-7b-chat-hf")
+                            ai_response = hf_provider.generate(validated_input, conversation_history)
+                            
+                            if ai_response:
+                                st.markdown(ai_response)  # Use st.markdown instead of response_placeholder
+                                status_placeholder.success("✅ HF response received!")
+                            else:
+                                status_placeholder.error("❌ No response from HF")
                                 
-                    # Save response if successful
-                    if ai_response:
-                        # Update conversation history
-                        conversation = user_session.get("conversation", [])
-                        conversation.append({"role": "user", "content": validated_input})
-                        conversation.append({"role": "assistant", "content": ai_response})
-                        user_session["conversation"] = conversation
-                        session_manager.update_session("default_user", user_session)
-                        
-                        # Add to message history with proper display
-                        message_data = {
-                            "role": "assistant",
-                            "content": ai_response,
-                            "timestamp": datetime.now().strftime("%H:%M:%S")
-                        }
-                        
-                        st.session_state.messages.append(message_data)
-                        
-                        # Force display update
-                        st.markdown(ai_response)
-                    else:
-                        error_msg = "Sorry, I couldn't process your request. Please try again."
-                        st.session_state.messages.append({
-                            "role": "assistant",
-                            "content": error_msg,
-                            "timestamp": datetime.now().strftime("%H:%M:%S")
-                        })
-                        st.markdown(error_msg)
-                        
-            except Exception as e:
-                user_msg = translate_error(e)
-                error_message = f"⚠️ {user_msg}"
-                st.error(error_message)  # Use st.error instead of response_placeholder.error
-                st.session_state.messages.append({
-                    "role": "assistant", 
-                    "content": error_message,
-                    "timestamp": datetime.now().strftime("%H:%M:%S")
-                })
-            finally:
-                st.session_state.is_processing = False
-                # Force UI update
-                st.rerun()  # Changed from experimental_rerun() to rerun()
+                        except Exception as hf_error:
+                            error_message = str(hf_error)
+                            status_placeholder.error(f"❌ HF also failed: {error_message[:100]}...")
+                            logger.error(f"HF error: {error_message}")
+                            
+                # Save response if successful
+                if ai_response:
+                    # Update conversation history in session
+                    conversation = user_session.get("conversation", []).copy()
+                    conversation.append({"role": "user", "content": validated_input})
+                    conversation.append({"role": "assistant", "content": ai_response})
+                    session_manager.update_session("default_user", {"conversation": conversation})
+                    
+                    # Add to message history
+                    st.session_state.messages.append({
+                        "role": "assistant",
+                        "content": ai_response,
+                        "timestamp": datetime.now().strftime("%H:%M:%S")
+                    })
+                else:
+                    error_msg = "Sorry, I couldn't process your request. Please try again."
+                    st.session_state.messages.append({
+                        "role": "assistant",
+                        "content": error_msg,
+                        "timestamp": datetime.now().strftime("%H:%M:%S")
+                    })
+                    st.markdown(error_msg)
+                    
+        except Exception as e:
+            error_msg = f"System error: {str(e)}"
+            logger.error(f"Chat processing error: {error_msg}")
+            st.error(error_msg)
+            st.session_state.messages.append({
+                "role": "assistant",
+                "content": error_msg,
+                "timestamp": datetime.now().strftime("%H:%M:%S")
+            })
+        finally:
+            st.session_state.is_processing = False
+            st.experimental_rerun()  # Fixed: use experimental_rerun
 
 # Add evaluation dashboard tab (separate from chat interface) - ONLY ABOUT TAB NOW
 st.divider()
@@ -498,4 +494,4 @@ if user_input and user_input.lower().strip() in ["tell me a story", "tell me a c
         "timestamp": datetime.now().strftime("%H:%M:%S")
     })
     st.session_state.is_processing = False
-    st.rerun()
+    st.experimental_rerun()