app.py

import streamlit as st
from langchain.text_splitter import RecursiveCharacterTextSplitter
from langchain_community.embeddings import HuggingFaceEmbeddings
from langchain_community.vectorstores import FAISS
from langchain_community.document_loaders import PyPDFLoader
from langchain.chains import RetrievalQA
from langchain.llms import HuggingFacePipeline
from transformers import pipeline

# ----------------------------
# APP CONFIG
# ----------------------------
st.set_page_config(page_title="📘 PDF Question Answering", layout="wide")
st.title("📘 PDF Question Answering App")
st.markdown("Upload a PDF and ask questions about its content.")

# ----------------------------
# GLOBAL VARIABLE
# ----------------------------
qa_chain = None

# ----------------------------
# FUNCTIONS
# ----------------------------
def load_pdf(pdf_file):
    """Load PDF and split into chunks"""
    loader = PyPDFLoader(pdf_file.name)
    documents = loader.load()
    text_splitter = RecursiveCharacterTextSplitter(chunk_size=800, chunk_overlap=100)
    docs = text_splitter.split_documents(documents)
    return docs

def build_vectorstore(docs):
    """Create FAISS vector store from documents"""
    embeddings = HuggingFaceEmbeddings(model_name="sentence-transformers/all-MiniLM-L6-v2")
    vectorstore = FAISS.from_documents(docs, embeddings)
    return vectorstore

def build_qa_chain(vectorstore):
    """Build QA chain using FLAN-T5"""
    llm = HuggingFacePipeline(
        pipeline=pipeline(
            "text2text-generation",
            model="google/flan-t5-base",
            max_length=512,
            temperature=0
        )
    )
    qa_chain = RetrievalQA.from_chain_type(
        llm=llm,
        retriever=vectorstore.as_retriever(search_kwargs={"k": 3}),
        chain_type="stuff"
    )
    return qa_chain

# ----------------------------
# STREAMLIT UI
# ----------------------------
uploaded_file = st.file_uploader("Upload a PDF", type=["pdf"])

if uploaded_file:
    with st.spinner("Processing PDF..."):
        docs = load_pdf(uploaded_file)
        vectorstore = build_vectorstore(docs)
        qa_chain = build_qa_chain(vectorstore)
    st.success("✅ PDF uploaded & processed. You can now ask questions!")

if qa_chain:
    query = st.text_input("Ask a question about the PDF:")
    if query:
        with st.spinner("Searching..."):
            answer = qa_chain.run(query)
        st.subheader("📌 Answer:")
        st.write(answer)