Update app.py

app.py

@@ -1,47 +1,58 @@
+import os
 import gradio as gr
+
+# Suppose 'run_with_chain' is your pipeline function from pipeline.py
 from pipeline import run_with_chain
+
+# Suppose 'memory' and 'restatement_chain' come from my_memory_logic.py
 from my_memory_logic import memory, restatement_chain
 
 def chat_history_fn(user_input, history):
     """
-    Gradio calls this function each time the user submits a message.
-    'history' is a list of (user_msg, ai_msg) pairs from previous turns.
-    We'll add them to memory, restate the user_input if needed,
-    run the pipeline, store the new turn, then return a list of
-    message dicts that Gradio's ChatInterface accepts.
+    Rely on LangChain memory to store the entire conversation across calls.
+    DO NOT re-add old messages from 'history' each time.
+    Also, handle potential None or invalid strings for user_input/answer 
+    to avoid Pydantic validation errors.
     """
-    # 1) Convert existing history into memory
-    for user_msg, ai_msg in history:
-        memory.chat_memory.add_user_message(user_msg)
-        memory.chat_memory.add_ai_message(ai_msg)
-    
-    # 2) Restate the new user question with chat history
+    # -- 0) Sanitize user_input to ensure it's a valid string
+    if not user_input or not isinstance(user_input, str):
+        user_input = "" if user_input is None else str(user_input)
+
+    # -- 1) Restate the new user question using existing LangChain memory
     reformulated_q = restatement_chain.run({
         "chat_history": memory.chat_memory.messages,
         "input": user_input
     })
-    
-    # 3) Pass the reformulated question to your pipeline
+
+    # -- 2) Pass the reformulated question into your pipeline
     answer = run_with_chain(reformulated_q)
-    
-    # 4) Update memory with the new turn
+    # also sanitize if needed
+    if answer is None or not isinstance(answer, str):
+        answer = "" if answer is None else str(answer)
+
+    # -- 3) Add this new user->assistant turn to memory
     memory.chat_memory.add_user_message(user_input)
     memory.chat_memory.add_ai_message(answer)
-    
-    # 5) Update the 'history' list with (user, ai)
+
+    # -- 4) Update Gradio’s 'history' so the UI shows the new turn
     history.append((user_input, answer))
-    
-    # 6) Gradio's ChatInterface expects a list of message dicts:
-    #    [{"role": "user"|"assistant", "content": "..."} ...]
-    #    We'll build that from our (user_msg, ai_msg) pairs in 'history'.
+
+    # -- 5) Convert the entire 'history' to message dictionaries:
+    #       [{"role":"user","content":...},{"role":"assistant","content":...},...]
     message_dicts = []
     for usr_msg, ai_msg in history:
+        if not isinstance(usr_msg, str):
+            usr_msg = str(usr_msg) if usr_msg else ""
+        if not isinstance(ai_msg, str):
+            ai_msg = str(ai_msg) if ai_msg else ""
+
         message_dicts.append({"role": "user", "content": usr_msg})
         message_dicts.append({"role": "assistant", "content": ai_msg})
-    
-    # 7) Return the message dictionary list
+
+    # -- 6) Return the message dictionary list
     return message_dicts
 
+# Build your ChatInterface with the function
 demo = gr.ChatInterface(
     fn=chat_history_fn,
     title="DailyWellnessAI with Memory",