app.py

import streamlit as st
from dotenv import load_dotenv
from PyPDF2 import PdfReader
from langchain.text_splitter import RecursiveCharacterTextSplitter
from langchain.embeddings import HuggingFaceInstructEmbeddings
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
import os
import numpy as np

#EMBEDDINGS_FILE = "embeddings.npy"
INDEX_FILE = "index.faiss"

def save_embeddings_and_index(index):
    #np.save(EMBEDDINGS_FILE, embeddings)
    index.save_local(INDEX_FILE)

def load_embeddings_and_index():
    if os.path.exists(INDEX_FILE):
        embeddings = HuggingFaceInstructEmbeddings(model_name="hkunlp/instructor-xl")
        index = FAISS.load_local(INDEX_FILE, embeddings)
        return index
    return None


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


def get_files(text_doc):
    text = ""
    for file in text_doc:
        if file.type == "text/plain":
            # Read the text directly from the file
            text += file.getvalue().decode("utf-8")
        elif file.type == "application/pdf":
            text += get_pdf_text(file)
    return text


def get_text_chunks(text):
    text_splitter = RecursiveCharacterTextSplitter(
        chunk_size=900,
        chunk_overlap=0,
        separators="\n",
        add_start_index = True,
        length_function= len
    )
    chunks = text_splitter.split_text(text)
    return chunks


def get_vectorstore(text_chunks, index):
    if index is None:
        embeddings = HuggingFaceInstructEmbeddings(model_name="hkunlp/instructor-xl")
        vectorstore = FAISS.from_texts(texts=text_chunks, embedding=embeddings)
        return vectorstore
    else:
        index.add_texts(texts=text_chunks)
        return index
    


def get_conversation_chain(vectorstore):
    llm = HuggingFaceHub(repo_id="mistralai/Mistral-7B-v0.1", model_kwargs={"temperature":0.2, "max_length":1024})

    memory = ConversationBufferMemory(
        memory_key='chat_history', return_messages=True)
    conversation_chain = ConversationalRetrievalChain.from_llm(
        llm=llm,
        retriever=vectorstore.as_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():
    load_dotenv()
    st.set_page_config(page_title="ChatBot")
    st.write(css, unsafe_allow_html=True)

    if "conversation" not in st.session_state:
        index = load_embeddings_and_index()
        if index==None:
            st.session_state.conversation = None
        else:
            st.session_state.conversation = get_conversation_chain(index)
    if "chat_history" not in st.session_state:
        st.session_state.chat_history = None

    st.header("Chat Bot")
    user_question = st.text_input("Ask a question:")
    if user_question:
        handle_userinput(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"):
            with st.spinner("Processing"):
                index = load_embeddings_and_index()
                raw_text = get_files(pdf_docs)
                text_chunks = get_text_chunks(raw_text)
                # Load a new faiss index or append to existing (if it exists)
                index = get_vectorstore(text_chunks, index)
                # save updated faiss index
                save_embeddings_and_index(index)

                # create conversation chain
                st.session_state.conversation = get_conversation_chain(index)


if __name__ == '__main__':
    main()