app.py

import os
import streamlit as st
from dotenv import load_dotenv
from PyPDF2 import PdfReader
from langchain.text_splitter import CharacterTextSplitter
from langchain.embeddings import HuggingFaceInstructEmbeddings, OpenAIEmbeddings
from langchain.vectorstores import FAISS
from langchain.memory import ConversationBufferMemory
from langchain.chains import ConversationalRetrievalChain
from htmlTemplates import css, bot_template, user_template
from langchain.llms import HuggingFaceHub
from langchain.chat_models import ChatOpenAI

load_dotenv()

def update_api_token(model_choice, api_token):
    dotenv_file = '.env'
    if model_choice == "OpenAI":
        with open(dotenv_file, 'r') as file:
            lines = file.readlines()
        with open(dotenv_file, 'w') as file:
            for line in lines:
                if line.startswith("OPENAI_API_KEY"):
                    file.write(f"OPENAI_API_KEY={api_token}\n")
                else:
                    file.write(line)
        os.environ['OPENAI_API_KEY'] = api_token
    elif model_choice == "HuggingFace":
        with open(dotenv_file, 'r') as file:
            lines = file.readlines()
        with open(dotenv_file, 'w') as file:
            for line in lines:
                if line.startswith("HUGGINGFACEHUB_API_TOKEN"):
                    file.write(f"HUGGINGFACEHUB_API_TOKEN={api_token}\n")
                else:
                    file.write(line)
        os.environ['HUGGINGFACEHUB_API_TOKEN'] = api_token

def validate_token(model_choice):
    if 'validation_done' not in st.session_state:
        try:
            if model_choice == "OpenAI":
                st.session_state.EMBEDDINGS = OpenAIEmbeddings()
                st.session_state.LLM = ChatOpenAI()
            else:
                st.session_state.EMBEDDINGS = HuggingFaceInstructEmbeddings(model_name="hkunlp/instructor-xl")
                st.session_state.LLM = HuggingFaceHub(repo_id="google/flan-t5-base", model_kwargs={"temperature": 0.5, "max_length": 512})
            st.session_state.validation_done = True
            return True
        except Exception as e:
            return False
    else:
        return True

def get_pdf_text(pdf_docs):
    text = ""
    if pdf_docs:
        for pdf in pdf_docs:
            pdf_reader = PdfReader(pdf)
            for page in pdf_reader.pages:
                text += page.extract_text()
    return text

def get_text_chunks(text):
    text_splitter = CharacterTextSplitter(
        separator="\n",
        chunk_size=1000,
        chunk_overlap=200,
        length_function=len
    )
    chunks = text_splitter.split_text(text)
    return chunks

def get_vectorstore(text_chunks, embeddings):
    vectorstore = FAISS.from_texts(texts=text_chunks, embedding=embeddings)
    return vectorstore

def get_conversation_chain(llm, embeddings, vectorstore=None):
    if llm is None or embeddings is None:
        raise ValueError("LLM or EMBEDDINGS is not initialized.")
    memory = ConversationBufferMemory(memory_key='chat_history', return_messages=True)

    if vectorstore is None:
        dummy_text = [""]
        vectorstore = FAISS.from_texts(texts=dummy_text, embedding=embeddings)

    retriever = vectorstore.as_retriever()
    conversation_chain = ConversationalRetrievalChain.from_llm(llm=llm, retriever=retriever, memory=memory)
    return conversation_chain

def handle_userinput(user_question):
    response = st.session_state.conversation({'question': user_question})
    st.session_state.chat_history = response['chat_history']
    for i, message in enumerate(st.session_state.chat_history):
        if i % 2 == 0:
            st.write(user_template.replace(
                "{{MSG}}", message.content), unsafe_allow_html=True)
        else:
            st.write(bot_template.replace(
                "{{MSG}}", message.content), unsafe_allow_html=True)

def main():
    global LLM, EMBEDDINGS
    LLM = None
    EMBEDDINGS = None

    st.set_page_config(page_title="MultiDoc_ChatBot", page_icon=":mag:")
    st.write(css, unsafe_allow_html=True)

    st.header("Chat with multiple PDFs :mag:")

    # User options for LLM and Embeddings
    model_choice = st.radio("Choose your model source", ("OpenAI", "HuggingFace"))
    api_token = st.text_input("Enter your API token", type="password")

    if st.button("Save API Token"):
        update_api_token(model_choice, api_token)
        with st.spinner("Validating API Token..."):
            if validate_token(model_choice):
                st.success(f"{model_choice} API token saved and model uploaded!")
            else:
                st.error("Invalid API token. Please try again.")
            
            print("LLM : ", st.session_state.LLM)
            print("EMBEDDINGS : ", st.session_state.EMBEDDINGS)

    if 'LLM' in st.session_state:
        LLM = st.session_state.LLM
    if 'EMBEDDINGS' in st.session_state:
        EMBEDDINGS = st.session_state.EMBEDDINGS

    if "user_question" not in st.session_state:
        st.session_state.user_question = ""

    user_question = st.text_input("Ask a question about your documents:", key="question_input", value=st.session_state.user_question)
    submit_button = st.button("Submit")

    if submit_button and user_question:
        if LLM is None or EMBEDDINGS is None:
            st.error("LLM or EMBEDDINGS is not initialized.")
        else:
            if "conversation" not in st.session_state:
                st.session_state.conversation = get_conversation_chain(LLM, EMBEDDINGS)
            if "chat_history" not in st.session_state:
                st.session_state.chat_history = []
            handle_userinput(user_question)
            st.session_state.user_question = ""

    with st.sidebar:
        st.subheader("Your documents")
        pdf_docs = st.file_uploader("Upload your PDFs here and click on 'Process'", accept_multiple_files=True)
        if st.button("Process"):
            if LLM is None or EMBEDDINGS is None:
                st.error("LLM or EMBEDDINGS is not initialized.")
            else:
                with st.spinner("Processing"):
                    raw_text = get_pdf_text(pdf_docs)
                    text_chunks = get_text_chunks(raw_text)
                    vectorstore = get_vectorstore(text_chunks, EMBEDDINGS)
                    st.session_state.conversation = get_conversation_chain(LLM, EMBEDDINGS, vectorstore=vectorstore)

if __name__ == '__main__':
    main()