Chatbot unable to access memory when deployed with streamlit

[tanhl30]

Checked other resources

• I added a very descriptive title to this question.
• I searched the LangChain documentation with the integrated search.
• I used the GitHub search to find a similar question and didn't find it.

Commit to Help

• I commit to help with one of those options 👆

Example Code

from langchain_core.messages import SystemMessage
from langgraph.checkpoint import MemorySaver 
from langgraph.prebuilt import create_react_agent
from langchain_community.agent_toolkits import SQLDatabaseToolkit
from langchain_openai import AzureChatOpenAI 
from langchain_community.utilities.sql_database import SQLDatabase 

# Testing locally 
llm = = AzureChatOpenAI(...)
db = SQLDatabase.from_uri(..)
tools = SQLDatabaseToolkit(db=db, llm=llm).get_tools()
system_message = SystemMessage(content='....')
app = create_react_agent(llm, tools, messages_modifier=system_message, checkpointer=MemorySaver())

def response_generator(input):
    config = {"configurable": {"thread_id": "1"}}
    response = app.invoke({'messages':[("user", input)]}, config)['messages'][-1].content
    return response 

#test questions, app manage to read chat history
response_generator('what is the capital of france?')
response_generator('what is the first question')

# deploy it with streamlit 

import streamlit as st

if "messages" not in st.session_state:
    st.session_state.messages = []

for message in st.session_state.messages:
    with st.chat_message(message["role"]):
        st.markdown(message["content"])

def response_generator(input):
    config = {"configurable": {"thread_id": "1"}}
    response = app.invoke({'messages':[("user", input)]}, config)['messages'][-1].content
    st.session_state.messages.append({"role": "assisstant", "content": response})
    return response

if prompt:= st.chat_input('How can I help you?'):
    st.session_state.messages.append({"role": "user", "content": prompt})
    with st.chat_message("user"):
        st.write(prompt)
    with st.chat_message("assistant"):
       st.write(response_generator(prompt))

Description

I am an inexperienced developer and this is my first time raising question so sorry if I did something wrongly.

I am building a LLM chatbot that can query db with langchain and streamlit. One issue that has bother me for days is that the chatbot could not read chat history, even though the chatbot on my IDE with similar code can read chat history. I was wondering if it's my code issue or there is a bug within the new langgraph agent ? I have tried incorporate StreamlitChatMessageHistory into my code as well but it doesn't work.

Any help is much appreciated !

System Info

latest version of LangGraph and Streamlit

[dosubot[bot]]

Hello @tanhl30! I'm here to assist you with any bugs, questions, or contributions. Let's work together to solve the issue you're facing.

To resolve the issue where your LLM chatbot, built with LangChain and Streamlit, is unable to access chat history when deployed with Streamlit, you can use the StreamlitChatMessageHistory class from LangChain. This class allows you to store chat messages in Streamlit's session state, which can be useful for maintaining chat history during a session.

Here is an example of how you can integrate StreamlitChatMessageHistory into your Streamlit application:

import streamlit as st
from langchain_community.chat_message_histories import StreamlitChatMessageHistory

# Initialize chat history
if 'chat_history' not in st.session_state:
    st.session_state.chat_history = StreamlitChatMessageHistory(key="chat_messages")

# Function to add a message to the chat history
def add_message_to_history(user_message, bot_response):
    st.session_state.chat_history.add_message({"type": "user", "content": user_message})
    st.session_state.chat_history.add_message({"type": "ai", "content": bot_response})

# Function to get the chat history
def get_chat_history():
    return st.session_state.chat_history.messages

# Streamlit app layout
st.title("LLM Chatbot with LangChain and Streamlit")

user_input = st.text_input("You: ", key="user_input")

if st.button("Send"):
    if user_input:
        # Here you would call your LLM to get the bot response
        bot_response = "This is a placeholder response from the bot."
        
        # Add messages to chat history
        add_message_to_history(user_input, bot_response)
        
        # Display chat history
        chat_history = get_chat_history()
        for message in chat_history:
            st.write(f"{message['type']}: {message['content']}")

This code initializes a StreamlitChatMessageHistory when the Streamlit app starts and stores it in the session state. It provides functions to add messages to the chat history and retrieve the chat history. When the user sends a message, it adds the user message and bot response to the chat history and displays the entire chat history.

By using StreamlitChatMessageHistory, you can ensure that the chat history is maintained during the user's session in the Streamlit app ^[1][2][3][4].

To continue talking to Dosu, mention @dosu.

[tanhl30]

Turns out it was streamlit caching function

https://docs.streamlit.io/develop/concepts/architecture/caching#stcache_resource
https://discuss.streamlit.io/t/langchain-memory/38264

[tisage]

It works! Thank you!