app.py

import gradio as gr
import transformers
from transformers import pipeline
import PyPDF2
import pdfplumber
from pdfminer.high_level import extract_pages, extract_text
from pdfminer.layout import LTTextContainer, LTChar, LTRect, LTFigure
import re
import torch
from datasets import load_dataset
import soundfile as sf
from IPython.display import Audio
import numpy as np
from datasets import load_dataset
import sentencepiece as spm
import os
import tempfile



def text_extraction(element):
    # Extracting the text from the in-line text element
    line_text = element.get_text()

    # Find the formats of the text
    # Initialize the list with all the formats that appeared in the line of text
    line_formats = []
    for text_line in element:
        if isinstance(text_line, LTTextContainer):
            # Iterating through each character in the line of text
            for character in text_line:
                if isinstance(character, LTChar):
                    # Append the font name of the character
                    line_formats.append(character.fontname)
                    # Append the font size of the character
                    line_formats.append(character.size)
    # Find the unique font sizes and names in the line
    format_per_line = list(set(line_formats))

    # Return a tuple with the text in each line along with its format
    return (line_text, format_per_line)

def read_pdf(pdf_pathy):
  # create a PDF file object
  pdfFileObj = open(pdf_pathy, 'rb')
  # create a PDF reader object
  pdfReaded = PyPDF2.PdfReader(pdfFileObj)

  # Create the dictionary to extract text from each image
  text_per_pagy = {}
  # We extract the pages from the PDF
  for pagenum, page in enumerate(extract_pages(pdf_pathy)):
      print("Elaborating Page_" +str(pagenum))
      # Initialize the variables needed for the text extraction from the page
      pageObj = pdfReaded.pages[pagenum]
      page_text = []
      line_format = []
      page_content = []

      # Open the pdf file
      pdf = pdfplumber.open(pdf_pathy)


      # Find all the elements
      page_elements = [(element.y1, element) for element in page._objs]
      # Sort all the elements as they appear in the page
      page_elements.sort(key=lambda a: a[0], reverse=True)

      # Find the elements that composed a page
      for i,component in enumerate(page_elements):
          # Extract the position of the top side of the element in the PDF
          pos= component[0]
          # Extract the element of the page layout
          element = component[1]

          # Check if the element is a text element
          if isinstance(element, LTTextContainer):
              # Check if the text appeared in a table
                  # Use the function to extract the text and format for each text element
                  (line_text, format_per_line) = text_extraction(element)
                  # Append the text of each line to the page text
                  page_text.append(line_text)
                  # Append the format for each line containing text
                  line_format.append(format_per_line)
                  page_content.append(line_text)
          

      # Create the key of the dictionary
      dctkey = 'Page_'+str(pagenum)
      # Add the list of list as the value of the page key
      text_per_pagy[dctkey]= [page_text, line_format, page_content]

  # Closing the pdf file object
  pdfFileObj.close()


  return text_per_pagy

#performing a cleaning of the contents
import re

def clean_text(text):
    # remove extra spaces
    text = re.sub(r'\s+', ' ', text)

    return text.strip()


def extract_abstract(text_per_pagy):
    abstract_text = ""
    
    #iterate through each page in the extracted text dictionary
    for page_num, page_text in text_per_pagy.items():
        if page_text:
           # Replace hyphens used for line breaks
            page_text = page_text.replace("- ", "")
           
            # Looking for the start of the abstract
            start_index = page_text.find("Abstract")
            if start_index != -1:
                # Adjust the start index to exclude the word "Abstract" itself
                # The length of "Abstract" is 8 characters; we also add 1 to skip the space after it
                start_index += len("Abstract") + 1

                # Searching the possible end markers of the abstract
                end_markers = ["Introduction", "Summary", "Overview", "Background"]
                end_index = -1

                for marker in end_markers:
                    temp_index = page_text.find(marker, start_index)
                    if temp_index != -1:
                        end_index = temp_index
                        break

                # If no end marker found, take entire text after "Abstract"
                if end_index == -1:
                    end_index = len(page_text)

                # Extract the abstract text
                abstract = page_text[start_index:end_index].strip()

                # Add the abstract to the complete text
                abstract_text += " " + abstract

                break

    return abstract_text


def main_function(uploaded_filepath):
    #a control to see if there is a file uploaded
    if uploaded_filepath is None:
        return "No file loaded", None

    #read and process the file
    text_per_pagy = read_pdf(uploaded_filepath)

    #cleaning the text and getting the abstract
    for key, value in text_per_pagy.items():
        cleaned_text = clean_text(' '.join(value[0]))
        text_per_pagy[key] = cleaned_text
    abstract_text = extract_abstract(text_per_pagy)

    #abstract summary
    summarizer = pipeline("summarization", model="pszemraj/long-t5-tglobal-base-sci-simplify")
    summary = summarizer(abstract_text, max_length=50, min_length=30, do_sample=False)[0]['summary_text']

    #generating the audio from the text, with my pipeline and model
    synthesiser = pipeline("text-to-speech", model="microsoft/speecht5_tts")
    embeddings_dataset = load_dataset("Matthijs/cmu-arctic-xvectors", split="validation")
    speaker_embedding = torch.tensor(embeddings_dataset[7306]["xvector"]).unsqueeze(0)
    speech = synthesiser(summary, forward_params={"speaker_embeddings": speaker_embedding})

    #saving the audio in a temp file
    audio_file_path = "summary.wav"
    sf.write(audio_file_path, speech["audio"], samplerate=speech["sampling_rate"])

    #the function returns the 2 pieces we need
    return summary, audio_file_path


iface = gr.Interface(
    fn=main_function,
    inputs=gr.File(type="filepath"),  # Cambiato da "pdf" a "file"
    outputs=[gr.Textbox(label="Summary Text"), gr.Audio(label="Summary Audio", type="filepath")]
)

# Avvia l'app
if __name__ == "__main__":
    iface.launch()