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_openai import OpenAIEmbeddings
from langchain_community.embeddings import HuggingFaceInstructEmbeddings
from langchain_community.vectorstores import FAISS
from langchain_openai import ChatOpenAI
from langchain_community.llms import HuggingFaceHub
from langchain.memory import ConversationBufferMemory
from langchain.chains import ConversationalRetrievalChain
from langchain_community.document_loaders import DirectoryLoader
from htmlTemplates import css, bot_template, user_template
from langchain.globals import set_verbose
set_verbose(False)  # Updated function call


def read_files_from_directory(directory):
    files = []
    for filename in os.listdir(directory):
        if filename.endswith(".pdf"):
            files.append(os.path.join(directory, filename))
    return files

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

def get_vector_store(text_chunks):
    # embeddings = HuggingFaceInstructEmbeddings(model_name="hkunlp/instructor-xl")
    embeddings = OpenAIEmbeddings(openai_api_key=os.getenv('OPENAI_API_KEY'))
    vectorstore = FAISS.from_texts(texts=text_chunks, embedding=embeddings)
    return vectorstore

def get_conversation_chain(vectorstore):
    if not os.getenv('OPENAI_API_KEY') and not os.getenv('LAW_GPT_MODEL_URL'):
        raise ValueError("Please provide either OPENAI_API_KEY or LAW_GPT_MODEL_URL in the .env file")

    # Use LAW GPT model if LAW_GPT_MODEL_URL is provided  
    if os.getenv('LAW_GPT_MODEL_URL'):
        llm = HuggingFaceHub(repo_id=os.getenv('LAW_GPT_MODEL_URL'))
    else:
        llm = ChatOpenAI(openai_api_key=os.getenv('OPENAI_API_KEY'))
        
    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

# get handler user input method
def handle_user_input(user_question):
    if st.session_state.conversation is not None:

        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)
    else:
        st.write("No data is loaded for RAG. Please upload a PDFs files to the data/ directory.")

def main():
    load_dotenv()

    st.set_page_config(page_title="EULawGPT - LLM model that can understand and reason about EU public domain data", page_icon=":books:")
    st.write(css, unsafe_allow_html=True)
    
    #load knowledge data PDF
    files = read_files_from_directory('./data')
    raw_knowledge_text = get_pdf_text(files)
    raw_knowledge_chunks = get_text_chunks(raw_knowledge_text)
    vectorstore_knowledge = get_vector_store(raw_knowledge_chunks)

    st.session_state.conversation = get_conversation_chain(vectorstore_knowledge)

    if "conversation" not in st.session_state:
        st.session_state.conversation = None
    if "chat_history" not in st.session_state:
        st.session_state.chat_history = None

    st.title("EU Law GPT")
    st.write("EU Law GPT is a LLM model that can understand and reason about EU public domain data")

    st.subheader('Popular questions:')
    if st.button("What is happening in Equador?"):
        handle_user_input("What is happening in Equador?")

    if st.button("What EU will do with Ecuador crisis?"):
        handle_user_input("What EU will do with Ecuador crisis?")

    st.subheader('Ask anything:')
    user_question = st.text_input("Ask a question about EU Law and Parlament work")

    if user_question:
        handle_user_input(user_question)

if __name__ == '__main__':
    main()