app.py

# imports

import gradio as gr
from transformers import pipeline
import torch
import PyPDF2

# function to read the uploaded PDF and extract its PDF when present based on the keyword "abstract" search.
# If the PDF doesn't have the word "abstract" it won't work
# also, I'm trying to limitate to the abstract itself, not to other sections, by adding a pattern of in between headers

def process_pdf(pdf):
    with open(pdf.name, "rb") as f:
        reader = PyPDF2.PdfReader(f)
        text = ""
        for page in reader.pages:
            text += page.extract_text()
        abstract_start = text.lower().find("abstract:")
        if abstract_start != -1:
            abstract_end = text.lower().find("\n\n", abstract_start)
            if abstract_end != -1:
                abstract = text[abstract_start:abstract_end]
            else:
                abstract = text[abstract_start:]
        else:
            abstract = "Abstract not found."

        return abstract

#Now creating the interface to read the PDFs

interface = gr.Interface(fn=process_pdf,
                           inputs=gr.inputs.File(type="file", label="Upload PDF"),
                           outputs="text",
                           title="Summarizing outloud",
                           description="Extract abstracts from PDFs, summarize then in 1 sentence and get an audio of it",
                           examples=[["example_pdf1.pdf"], ["example_pdf2.pdf"]])

   if __name__ == "__main__":
       interface.launch()