app.py

import streamlit as st
from PyPDF2 import PdfReader
from langchain.text_splitter import RecursiveCharacterTextSplitter
from langchain_google_genai import GoogleGenerativeAIEmbeddings
import google.generativeai as genai
from langchain.vectorstores import FAISS
from langchain_google_genai import ChatGoogleGenerativeAI
from langchain.chains.question_answering import load_qa_chain
from langchain.prompts import PromptTemplate
import os
import json
from transformers import AutoModelForCausalLM, AutoTokenizer, GenerationConfig, TextStreamer, ConversationalPipeline


####CREDIT#####
#Credit to author (Sri Laxmi) of original code reference: SriLaxmi1993
#Sri LaxmiGithub Link: https://github.com/SriLaxmi1993/Document-Genie-using-RAG-Framwork
#Sri Laxmi Youtube:https://www.youtube.com/watch?v=SkY2u4UUr6M&t=112s
###############
os.system("pip install -r requirements.txt")

#some model
#tokenizer = AutoTokenizer.from_pretrained("bofenghuang/vigogne-2-7b-chat")
model_name_or_path = "bofenghuang/vigogne-2-7b-chat"
revision = "v2.0"

tokenizer = AutoTokenizer.from_pretrained(model_name_or_path, revision=revision, padding_side="right", use_fast=False)
model = AutoModelForCausalLM.from_pretrained(model_name_or_path, revision=revision, torch_dtype=torch.float16, device_map="auto")


st.set_page_config(page_title="Gemini RAG", layout="wide")

# This is the first API key input; no need to repeat it in the main function.
api_key = 'AIzaSyCvXRggpO2yNwIpZmoMy_5Xhm2bDyD-pOo'

#os.mkdir('faiss_index')

#empty faise_index and chat_history.json
def delete_files_in_folder(folder_path):
    try:
        # Iterate over all the files in the folder
        chat_history_file = "chat_history.json"
        if os.path.exists(chat_history_file):
            os.remove(chat_history_file)
        for file_name in os.listdir(folder_path):
            file_path = os.path.join(folder_path, file_name)
            if os.path.isfile(file_path):  # Check if it's a file
                os.remove(file_path)  # Delete the file
                print(f"Deleted file: {file_path}")
        print("All files within the folder have been deleted successfully!")
    except Exception as e:
        print(f"An error occurred: {e}")


with st.sidebar:
    st.title("Menu:")

    if st.button("Reset Files", key="reset_button"):
        folder_path = 'faiss_index'
        delete_files_in_folder(folder_path)

    CH_size = st.slider("Chunk Size", 0, 1000, 450)
    CH_overlap = st.slider("Chunk Overlap", 0, 1000, 50)


def get_pdf_text(pdf_docs):
    text = ""
    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 = RecursiveCharacterTextSplitter(chunk_size=CH_size, chunk_overlap=CH_overlap)
    chunks = text_splitter.split_text(text)
    return chunks

def get_vector_store(text_chunks, api_key):
    embeddings = GoogleGenerativeAIEmbeddings(model="models/embedding-001", google_api_key=api_key)
    vector_store = FAISS.from_texts(text_chunks, embedding=embeddings)
    vector_store.save_local("faiss_index")

def get_conversational_chain():
    prompt_template = """
    Answer the question as detailed as possible from the provided context, make sure to provide all the details, if the answer is not in
    provided context just say, "answer is not available in the context", don't provide the wrong answer. When giving an answer, try to include all mentionings of the subject being asked and include this within your response\n\n
    Context:\n {context}?\n
    Question: \n{question}\n

    Answer:
    """
    model = ChatGoogleGenerativeAI(model="gemini-pro", temperature=0.2, google_api_key=api_key)
    prompt = PromptTemplate(template=prompt_template, input_variables=["context", "question"])
    chain = load_qa_chain(model, chain_type="stuff", prompt=prompt)
    return chain

#chat history functionality
def update_chat_history(question, reply):
    # Check if chat history file exists
    chat_history_file = "chat_history.json"
    if os.path.exists(chat_history_file):
        # If file exists, load existing chat history
        with open(chat_history_file, "r") as file:
            chat_history = json.load(file)
    else:
        # If file doesn't exist, initialize chat history
        chat_history = {"conversations": []}

    # Add current conversation to chat history
    chat_history["conversations"].append({"question": question, "reply": reply})

    # Write updated chat history back to file
    with open(chat_history_file, "w") as file:
        json.dump(chat_history, file, indent=4)
        # Display chat history
    st.subheader("Chat History")
    for conversation in chat_history["conversations"]:
        st.write(f"**Question:** {conversation['question']}")
        st.write(f"**Reply:** {conversation['reply']}")
        st.write("---")


'''
def user_input(user_question, api_key):
    embeddings = GoogleGenerativeAIEmbeddings(model="models/embedding-001", google_api_key=api_key)
    new_db = FAISS.load_local("faiss_index", embeddings)
    docs = new_db.similarity_search(user_question)
    chain = get_conversational_chain()
    response = chain({"input_documents": docs, "question": user_question}, return_only_outputs=True)
    st.write("Reply: ", response["output_text"])

    #chat history
    update_chat_history(user_question, response["output_text"])
'''


def user_input(user_question, api_key):
    embeddings = GoogleGenerativeAIEmbeddings(model="models/embedding-001", google_api_key=api_key)
    new_db = FAISS.load_local("faiss_index", embeddings)
    docs = new_db.similarity_search(user_question)
    chain = get_conversational_chain()
    response_gemini = chain({"input_documents": docs, "question": user_question}, return_only_outputs=True)
    
    # Initialize the Hugging Face conversational pipeline with your custom model
    pipeline = ConversationalPipeline(model_name_or_path="bofenghuang/vigogne-2-7b-chat")

    # Prompt template for making the response more conversational
    prompt_template = f"""
    Transform the following response into a more conversational tone without adding new information:
    
    Response:
    {response_gemini["output_text"]}
    
    Transformed Response:
    """

    # Generate the transformed response using the Hugging Face model
    transformed_response = pipeline(prompt=prompt_template, max_length=100)

    # Display the transformed response
    st.write("Reply: ", transformed_response)

    # Update chat history
    update_chat_history(user_question, transformed_response)



def main():
    st.header("RAG based LLM Application")

    user_question = st.text_input("Ask a Question from the PDF Files", key="user_question")

    if user_question and api_key:
        user_input(user_question, api_key)

    with st.sidebar:
        st.title("Menu:")

        pdf_docs = st.file_uploader("Upload your PDF Files and Click on the Submit & Process Button", accept_multiple_files=True, key="pdf_uploader")
        if st.button("Submit & Process", key="process_button") and api_key:
            with st.spinner("Processing..."):
                raw_text = get_pdf_text(pdf_docs)
                text_chunks = get_text_chunks(raw_text)
                get_vector_store(text_chunks, api_key)
                st.success("Done")

if __name__ == "__main__":
    main()