app.py

import os
import streamlit as st
import pickle
import time
from langchain.chains import RetrievalQA
from langchain.prompts import PromptTemplate
from langchain.text_splitter import RecursiveCharacterTextSplitter
from langchain.document_loaders import UnstructuredURLLoader
from langchain.embeddings import HuggingFaceInstructEmbeddings
from langchain.llms import GooglePalm
from langchain.vectorstores import FAISS

from dotenv import load_dotenv
load_dotenv()  # take environment variables from .env (especially openai api key)

st.title("News Research Tool 📈")
st.sidebar.title("News Article URLs")

urls = []
for i in range(3):
    url = st.sidebar.text_input(f"URL {i+1}")
    if url:
        urls.append(url)

process_url_clicked = st.sidebar.button("Process URLs")
file_path = "faiss_store_openai.pkl"

main_placeholder = st.empty()
llm = GooglePalm(temperature=0.9)

if process_url_clicked:
    # load data
    loader = UnstructuredURLLoader(urls=urls)
    main_placeholder.text("Data Loading...Started...✅✅✅")
    data = loader.load()
    # split data
    text_splitter = RecursiveCharacterTextSplitter(
        separators=['\n\n', '\n', '.', ','],
        chunk_size=500
    )
    main_placeholder.text("Text Splitter...Started...✅✅✅")
    docs = text_splitter.split_documents(data)
    # create embeddings and save it to FAISS index
    embeddings = HuggingFaceInstructEmbeddings(model_name="hkunlp/instructor-base")
    main_placeholder.text("Embedding Vector Started Building...✅✅✅")
    vectorstore_openai = FAISS.from_documents(docs, embeddings)

    # Save the FAISS index to a pickle file
    with open(file_path, "wb") as f:
        pickle.dump(vectorstore_openai, f)

query = main_placeholder.text_input("Question: ")
if query:
    if os.path.exists(file_path):
        with open(file_path, "rb") as f:
            vectorstore = pickle.load(f)
            prompt_template = """
                <context>
                {context}
                </context>
                Question: {question}
                Assistant:"""
            prompt = PromptTemplate(
                template=prompt_template, input_variables=["context", "question"]
            )

            chain = RetrievalQA.from_chain_type(
                llm=llm,
                chain_type="stuff",
                retriever=vectorstore.as_retriever(search_type="similarity", search_kwargs={"k": 1}),
                return_source_documents=True,
                chain_type_kwargs={"prompt": prompt}
            )
            result = chain({"query": query})
            # result will be a dictionary of this format --> {"answer": "", "sources": [] }
            st.header("Answer")
            st.write(result["result"])

            # Display sources, if available
            sources = result.get("sources", "")
            if sources:
                st.subheader("Sources:")
                sources_list = sources.split("\n")  # Split the sources by newline
                for source in sources_list:
                    st.write(source)