import streamlit as st
from transformers import pipeline
from PyPDF2 import PdfReader
import PyPDF2
import fitz
import os
import nltk


def get_pdf_text(pdf_docs):
    text = ""
    for pdf in pdf_docs:
        pdf_reader = PdfReader(pdf)
        for page in pdf_reader.pages:
            text += page.extract_text()
    return text

def main():
    st.title('Question Generator from PDFs')
    pipe = pipeline(
        task = 'text2text-generation',
        model = 'ramsrigouthamg/t5_squad_v1'
    )
    file = st.file_uploader(label='Upload',accept_multiple_files=True)
    pr = st.button(label='Start')
    if pr:
        st.write('Hi')
        raw_text = get_pdf_text(file)
        sentences = nltk.sent_tokenize(text=raw_text)
        # st.write(sts)
        # for i in sentences:
        #     st.write(i)
        questions = []
        st.subheader("Generated Questions are: ")
        s = pipe(sentences)
        for i in s:
            questions.append(i['generated_text'][10:])
            st.write(i['generated_text'][10:])
        if st.toggle(label='Show Pipeline Output'):
            st.write(s)
        if st.toggle(label='Show Questions list'):
            st.write(questions)
        # for i in sts:
        #     x = pipe(i)
        #     questions.append(x)
        #     st.write(x)

if __name__ == '__main__':
    main()