# app.py

import streamlit as st
import pdfplumber
from transformers import pipeline

# Load QA model pipeline
qa_pipeline = pipeline("question-answering", model="deepset/roberta-base-squad2")

st.title("📄 PDF Question Answering using LLM")

uploaded_pdf = st.file_uploader("Upload a PDF", type=["pdf"])

if uploaded_pdf:
    # Extract text from PDF
    with pdfplumber.open(uploaded_pdf) as pdf:
        extracted_text = ""
        for page in pdf.pages:
            extracted_text += page.extract_text() + "\n"

    # Store extracted text in session
    st.session_state['document_text'] = extracted_text

    st.success("✅ PDF uploaded and text extracted successfully!")

    # Display preview
    with st.expander("📄 Preview Extracted Text"):
        st.write(extracted_text[:1500] + "...")

# Ask questions about the PDF
if 'document_text' in st.session_state:
    st.subheader("💬 Ask a question based on the uploaded PDF")
    user_question = st.text_input("Enter your question")

    if user_question:
        with st.spinner("Generating answer..."):
            result = qa_pipeline({
                "question": user_question,
                "context": st.session_state['document_text']
            })
            st.markdown(f"**Answer:** {result['answer']}")