import streamlit as st from streamlit_chat import message from langchain_google_genai import ChatGoogleGenerativeAI from langchain.chains import ConversationChain from langchain.chains.conversation.memory import ConversationBufferWindowMemory from langchain.prompts import SystemMessagePromptTemplate, HumanMessagePromptTemplate, ChatPromptTemplate, MessagesPlaceholder from utils import get_conversation_string, query_refiner, find_match import pickle import os import concurrent.futures # Load embeddings and documents only once @st.cache_resource def load_embeddings(): embedding = pickle.load(open('embedding.pkl', 'rb')) document = pickle.load(open('document.pkl', 'rb')) return embedding, document embedding, document = load_embeddings() st.subheader("Chatbot with Langchain, Gemini, Chromadb, and Streamlit") if 'responses' not in st.session_state: st.session_state['responses'] = ["How can I assist you?"] if 'requests' not in st.session_state: st.session_state['requests'] = [] if 'buffer_memory' not in st.session_state: st.session_state.buffer_memory = ConversationBufferWindowMemory(k=3, return_messages=True) GOOGLE_API_KEY = os.getenv('Google_API_KEY') llm = ChatGoogleGenerativeAI(model="gemini-pro",google_api_key=GOOGLE_API_KEY) system_msg_template = SystemMessagePromptTemplate.from_template( template="""Answer the question as truthfully as possible using the provided context, and if the answer is not contained within the text below, say 'I don't know'""" ) human_msg_template = HumanMessagePromptTemplate.from_template(template="{input}") prompt_template = ChatPromptTemplate.from_messages([system_msg_template, MessagesPlaceholder(variable_name="history"), human_msg_template]) conversation = ConversationChain(memory=st.session_state.buffer_memory, prompt=prompt_template, llm=llm, verbose=True) response_container = st.container() text_container = st.container() @st.cache_resource def setup_db(): from langchain.vectorstores import Chroma db = Chroma.from_documents(document, embedding) return db db = setup_db() with text_container: query = st.text_input("Query: ", key="input") if query: with st.spinner("typing..."): conversation_string = get_conversation_string() # Use a thread pool to run query_refiner and find_match in parallel with concurrent.futures.ThreadPoolExecutor() as executor: refined_query_future = executor.submit(query_refiner, conversation_string, query) refined_query = refined_query_future.result() st.subheader("Refined Query:") st.write(refined_query) context = find_match(refined_query) response = conversation.predict(input=f"Context:\n {context} \n\n Query:\n{refined_query}") st.session_state.requests.append(query) st.session_state.responses.append(response) with response_container: if st.session_state['responses']: for i in range(len(st.session_state['responses'])): message(st.session_state['responses'][i], key=str(i)) if i < len(st.session_state['requests']): message(st.session_state["requests"][i], is_user=True, key=str(i) + '_user')