app.py

import PyPDF2
import streamlit as st
from dotenv import load_dotenv
from transformers import pipeline


def retrieve_pdf_text(pdf_file):
    pdf_reader = PyPDF2.PdfReader(pdf_file)
    text = ""
    for page in pdf_reader.pages:
        text += page.extract_text()
    return text


def main():
    load_dotenv()
    st.set_page_config(page_title='Document Summarizer', page_icon=':books:')
    st.header("Summarize a PDF")
    hf_name = "pszemraj/led-base-book-summary"

    pdf_file = st.file_uploader("Upload a PDF file with text", type=["pdf"])
    length = st.slider('Max summary length', 0, 3000, 1000)

    # if a pdf file is uploaded
    if pdf_file:
        raw_text = retrieve_pdf_text(pdf_file)
        if st.button("Run"):
            with st.spinner("Summarizing.."):
                summarizer = pipeline("summarization", hf_name)
                result = summarizer(
                    raw_text,
                    min_length=8,
                    max_length=length,
                    no_repeat_ngram_size=3,
                    encoder_no_repeat_ngram_size=3,
                    repetition_penalty=3.5,
                    num_beams=4,
                    do_sample=False,
                    early_stopping=True,
                )
                st.write(result[0]["summary_text"])



if __name__ == '__main__':
    main()