app.py

import gradio as gr
from transformers import pipeline
import PyPDF2

# Load Hugging Face QA model (FLAN-T5)
qa_pipeline = pipeline("text2text-generation", model="google/flan-t5-large")

# Function to extract text from uploaded PDF and generate answer
def answer_question(pdf_file, question):
    # Step 1: Read and extract text from the PDF
    reader = PyPDF2.PdfReader(pdf_file)
    pdf_text = ""
    for page in reader.pages:
        text = page.extract_text()
        if text:
            pdf_text += text.strip() + "\n"

    # Step 2: Build the prompt
    prompt = f"Context: {pdf_text}\nQuestion: {question}"

    # Step 3: Run the model
    result = qa_pipeline(prompt, max_length=512, do_sample=True)[0]['generated_text']
    return result.strip()

# Gradio interface
iface = gr.Interface(
    fn=answer_question,
    inputs=[
        gr.File(label="Upload a PDF file"),
        gr.Textbox(label="Ask your question")
    ],
    outputs="text",
    title="PDF-Based Question Answering",
    description="Upload a PDF and ask questions about its content using FLAN-T5."
)

# Launch the app
if __name__ == "__main__":
    iface.launch()