streamlit_app.py

# streamlit_app.py
import streamlit as st
import sys
import os

# *** Add these two lines at the very top ***
from dotenv import load_dotenv
load_dotenv() # Load variables from .env file

# Add the directory containing app.py to the Python path
# This assumes app.py is in the same directory as streamlit_app.py
sys.path.append(os.path.dirname(os.path.abspath(__file__)))

# Import your respond function and any necessary global variables from app.py
# Make sure app.py loads the model, tokenizer, etc. when imported
try:
    from app import respond, model_id # Import your main function and model_id
    # You might also need to import other things if respond relies on globals directly
    # from app import model, tokenizer, embedder, nlp, data, descriptions, embeddings, ...
    print("Successfully imported respond function from app.py")
except ImportError as e:
    st.error(f"Error importing core logic from app.py: {e}")
    st.stop() # Stop the app if the core logic can't be loaded

# Set Streamlit page config
st.set_page_config(page_title="Business Q&A Assistant")

st.title(f"Business Q&A Assistant with {model_id}")
st.write("Ask questions about the business (details from Google Sheet) or general knowledge (via search).")

# Initialize chat history in Streamlit's session state
# Session state persists across reruns for a single user session
if "messages" not in st.session_state:
    st.session_state.messages = []

# Display chat messages from history on app rerun
for message in st.session_state.messages:
    with st.chat_message(message["role"]):
        st.markdown(message["content"])

# Accept user input
if prompt := st.chat_input("Your Question"):
    # Add user message to chat history
    st.session_state.messages.append({"role": "user", "content": prompt})
    # Display user message in chat message container
    with st.chat_message("user"):
        st.markdown(prompt)

    # Get the current chat history in the format your respond function expects
    # Gradio's history is [(user, bot), (user, bot), ...]
    # Streamlit's session state is a list of dicts [{"role": "user", "content": "..."}]
    # We need to convert Streamlit's history format to Gradio's format for your respond function
    gradio_chat_history = []
    # Start from the second message if the first was from the system/initial state
    # Or just iterate through pairs, skipping the latest user prompt for history pass
    # The respond function expects history *before* the current turn
    history_for_respond = []
    # Iterate through messages, excluding the very last user prompt which is the current input
    for i in range(len(st.session_state.messages) - 1):
        if st.session_state.messages[i]["role"] == "user" and st.session_state.messages[i+1]["role"] == "assistant":
            history_for_respond.append((st.session_state.messages[i]["content"], st.st