app.py

import streamlit as st
from langchain.document_loaders import PyPDFLoader
from langchain.text_splitter import RecursiveCharacterTextSplitter
from langchain.vectorstores import Chroma
from langchain.embeddings import HuggingFaceEmbeddings
from langchain.chains import RetrievalQA
from langchain_google_genai import ChatGoogleGenerativeAI
import tempfile
import os
from dotenv import load_dotenv
from pydantic import SecretStr


load_dotenv()
GOOGLE_API_KEY = os.getenv("GOOGLE_API_KEY")

embeddings = HuggingFaceEmbeddings(model_name="sentence-transformers/all-MiniLM-L6-v2")

# ---------------------------- SETUP ----------------------------

st.title("📄 LangChain RAG Chatbot")

# Session state
if "chat_history" not in st.session_state:
    st.session_state.chat_history = []

if "qa_chain" not in st.session_state:
    st.session_state.qa_chain = None

# ---------------------------- FILE UPLOAD ----------------------------

st.subheader("Upload your PDF")
pdf_file = st.file_uploader("Upload", type="pdf")

if pdf_file is not None and st.session_state.qa_chain is None:
    with st.spinner("🔍 Processing document..."):
        # Save file temporarily
        with tempfile.NamedTemporaryFile(delete=False, suffix=".pdf") as tmp_file:
            tmp_file.write(pdf_file.read())
            tmp_path = tmp_file.name

        # Load and split PDF
        loader = PyPDFLoader(tmp_path)
        documents = loader.load_and_split()

        splitter = RecursiveCharacterTextSplitter(chunk_size=1000, chunk_overlap=200)
        chunks = splitter.split_documents(documents)

        # Vector store
        
        vectordb = Chroma.from_documents(
            chunks, embeddings, persist_directory="./chroma_db"
        )
        retriever = vectordb.as_retriever()

        # QA Chain
        llm = ChatGoogleGenerativeAI(model="gemini-2.0-flash", api_key=SecretStr(GOOGLE_API_KEY) if GOOGLE_API_KEY else None)
        qa_chain = RetrievalQA.from_chain_type(llm=llm, retriever=retriever)

        # Store in session
        st.session_state.qa_chain = qa_chain
        st.success("✅ Document processed and indexed!")

# ---------------------------- CHAT ----------------------------

if st.session_state.qa_chain:
    st.subheader("💬 Ask a question")

    question = st.text_input("You:", key="user_input")

    if question:
        with st.spinner("🤖 Generating answer..."):
            answer = st.session_state.qa_chain.run(question)
            st.session_state.chat_history.append({"user": question, "bot": answer})

    # Display chat history
    for chat in st.session_state.chat_history:
        st.markdown(f"🧑 **You:** {chat['user']}")
        st.markdown(f"🤖 **Bot:** {chat['bot']}")

    # Reset button
    if st.button("🔄 Reset Chat"):
        st.session_state.chat_history = []
        st.session_state.qa_chain = None
        st.rerun()
else:
    st.info("📂 Please upload a PDF to begin.")