app.py

import os
import streamlit as st
import pickle
import time
from langchain import OpenAI
from langchain.chains import RetrievalQAWithSourcesChain
from langchain.text_splitter import RecursiveCharacterTextSplitter
from langchain.document_loaders import UnstructuredURLLoader
from langchain.embeddings import FakeEmbeddings
from langchain.llms import HuggingFaceHub
from langchain.chains import LLMChain
from langchain.vectorstores import FAISS

from dotenv import load_dotenv
load_dotenv()  # take environment variables from .env (especially openai api key)
os.environ["HUGGINGFACEHUB_API_TOKEN"] = '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'


class RockyBot:
    def __init__(self, llm):
        self.llm = llm
        self.vectorstore = None

    def process_urls(self, urls):
        """Processes the given URLs and saves the FAISS index to a pickle file."""

        # load data
        loader = UnstructuredURLLoader(urls=urls)

        # split data
        text_splitter = RecursiveCharacterTextSplitter(
            separators=['\n\n', '\n', '.', ','],
            chunk_size=1000
        )
        docs = text_splitter.split_documents(loader.load())

        # create embeddings and save it to FAISS index
        embeddings = FakeEmbeddings(size=1352)
        self.vectorstore = FAISS.from_documents(docs, embeddings)

        # Save the FAISS index to a pickle file
        with open("faiss_store_openai.pkl", "wb") as f:
            pickle.dump(self.vectorstore, f)

    def answer_question(self, question):
        """Answers the given question using the LLM and retriever."""

        chain = RetrievalQAWithSourcesChain.from_llm(llm=self.llm, retriever=self.vectorstore.as_retriever())
        result = chain({"question": question}, return_only_outputs=True)

        return result["answer"], result.get("sources", "")


if __name__ == '__main__':
    llm = HuggingFaceHub(repo_id="google/flan-t5-xxl", model_kwargs={"temperature": 0.5, "max_length": 64})
    rockybot = RockyBot(llm)

    # Process URLs if the button is clicked
    if st.sidebar.button("Process URLs"):
        rockybot.process_urls(st.sidebar.text_input("URL 1"), st.sidebar.text_input("URL 2"), st.sidebar.text_input("URL 3"))
        st.progress(100.0)

    # Answer the question if it is not empty
    query = st.text_input("Question: ")
    if query:
        answer, sources = rockybot.answer_question(query)

        st.header("Answer")
        st.write(answer)

        # Display sources, if available
        if sources:
            st.subheader("Sources:")
            for source in sources.split("\n"):
                st.write(source)