app.py

import streamlit as st
import os
from langgraph.graph import MessagesState, StateGraph, START, END
from typing_extensions import TypedDict
from langchain_core.messages import AnyMessage, HumanMessage, SystemMessage, AIMessage
from typing import Annotated
from langgraph.graph.message import add_messages
from langgraph.checkpoint.memory import MemorySaver
from langchain_groq import ChatGroq

# Define the state
class MessagesState(TypedDict):
    messages: Annotated[list[AnyMessage], add_messages]

# Create graph function
def create_chat_graph(system_prompt, model_name):
    # Initialize LLM
    llm = ChatGroq(model=model_name)
    
    # Create system message
    system_message = SystemMessage(content=system_prompt)
    
    # Define the assistant function
    def assistant(state: MessagesState):
        # Get all messages including the system message
        messages = [system_message] + state["messages"]
        # Generate response
        response = llm.invoke(messages)
        # Return the response
        return {"messages": [response]}
    
    # Initialize the graph builder
    builder = StateGraph(MessagesState)
    
    # Add the assistant node
    builder.add_node("assistant", assistant)
    
    # Define edges
    builder.add_edge(START, "assistant")
    builder.add_edge("assistant", END)
    
    # Create memory saver for persistence
    memory = MemorySaver()
    
    # Compile the graph with memory
    graph = builder.compile(checkpointer=memory)
    
    return graph

# Set up Streamlit page
st.set_page_config(page_title="Conversational AI Assistant", page_icon="💬")
st.title("AI Chatbot with Memory")

# Sidebar configuration
st.sidebar.header("Configuration")

# API Key input (using st.secrets in production)
if "GROQ_API_KEY" not in os.environ:
    api_key = st.sidebar.text_input("Enter your Groq API Key:", type="password")
    if api_key:
        os.environ["GROQ_API_KEY"] = api_key
    else:
        st.sidebar.warning("Please enter your Groq API key to continue.")

# Model selection
model_options = ["llama3-70b-8192", "mixtral-8x7b-32768", "gemma-7b-it"]
selected_model = st.sidebar.selectbox("Select Model:", model_options)

# System prompt
default_prompt = "You are a helpful and friendly assistant. Maintain a conversational tone and remember previous interactions with the user."
system_prompt = st.sidebar.text_area("System Prompt:", value=default_prompt, height=150)

# Session ID for this conversation
if "session_id" not in st.session_state:
    import uuid
    st.session_state.session_id = str(uuid.uuid4())

# Initialize or get chat history
if "messages" not in st.session_state:
    st.session_state.messages = []

# Initialize the graph on first run or when config changes
if "chat_graph" not in st.session_state or st.sidebar.button("Reset Conversation"):
    if "GROQ_API_KEY" in os.environ:
        with st.spinner("Initializing chatbot..."):
            st.session_state.chat_graph = create_chat_graph(system_prompt, selected_model)
            st.session_state.messages = []  # Clear messages on reset
            st.success("Chatbot initialized!")
    else:
        st.sidebar.error("API key required to initialize chatbot.")

# Display chat history
for message in st.session_state.messages:
    if isinstance(message, dict):  # Handle dict format
        role = message.get("role", "")
        content = message.get("content", "")
    else:  # Handle direct string format
        role = "user" if message.startswith("User: ") else "assistant"
        content = message.replace("User: ", "").replace("Assistant: ", "")
    
    with st.chat_message(role):
        st.write(content)

# Input for new message
if "chat_graph" in st.session_state and "GROQ_API_KEY" in os.environ:
    user_input = st.chat_input("Type your message here...")
    
    if user_input:
        # Display user message
        with st.chat_message("user"):
            st.write(user_input)
        
        # Add to history
        st.session_state.messages.append({"role": "user", "content": user_input})
        
        # Get response from the chatbot
        with st.spinner("Thinking..."):
            # Call the graph with the user's message
            config = {"configurable": {"thread_id": st.session_state.session_id}}
            user_message = [HumanMessage(content=user_input)]
            result = st.session_state.chat_graph.invoke({"messages": user_message}, config)
            
            # Extract response
            response = result["messages"][-1].content
        
        # Display assistant response
        with st.chat_message("assistant"):
            st.write(response)
        
        # Add to history
        st.session_state.messages.append({"role": "assistant", "content": response})

# Add some additional info in the sidebar
st.sidebar.markdown("---")
st.sidebar.subheader("About")
st.sidebar.info(
    """
    This chatbot uses LangGraph for maintaining conversation context and 
    ChatGroq's language models for generating responses. Each conversation 
    has a unique session ID to maintain history.
    """
)

# Download chat history
if st.sidebar.button("Download Chat History"):
    import json
    from datetime import datetime
    
    # Convert chat history to downloadable format
    chat_export = "\n".join([f"{m['role']}: {m['content']}" for m in st.session_state.messages])
    
    # Create download button
    st.sidebar.download_button(
        label="Download as Text",
        data=chat_export,
        file_name=f"chat_history_{datetime.now().strftime('%Y%m%d_%H%M%S')}.txt",
        mime="text/plain"
    )