app.py

import os
import pdfplumber
import re
import gradio as gr
from transformers import pipeline, AutoModelForQuestionAnswering, AutoTokenizer
from io import BytesIO
import pytorch

"""
Extract the text from a section of a PDF file  between 'wanted_section' and 'next_section'.
Parameters: 
- path (str): The file path to the PDF file.
- wanted_section (str): The section to start extracting text from.
- next_section (str): The section to stop extracting text at.
Returns:
- text (str): The extracted text from the specified section range.
"""


def get_section(path, wanted_section, next_section):
    print(wanted_section)

    # Open the PDF file
    doc = pdfplumber.open(BytesIO(path)) 
    start_page = []
    end_page = []

    # Find the all the pages for the specified sections
    for page in range(len(doc.pages)):
        if len(doc.pages[page].search(wanted_section, return_chars=False, case=False)) > 0:
            start_page.append(page)
        if len(doc.pages[page].search(next_section, return_chars=False, case=False)) > 0:
            end_page.append(page)

    # Extract the text between the start and end page of the wanted section
    text = []
    for page_num in range(max(start_page), max(end_page)+1):
        page = doc.pages[page_num]
        text.append(page.extract_text())
    text = " ".join(text)
    final_text = text.replace("\n", " ")
    return final_text


def extract_between(big_string, start_string, end_string):
    # Use a non-greedy match for content between start_string and end_string
    pattern = re.escape(start_string) + '(.*?)' + re.escape(end_string)
    match = re.search(pattern, big_string, re.DOTALL)

    if match:
        # Return the content without the start and end strings
        return match.group(1)
    else:
        # Return None if the pattern is not found
        return None

def format_section1(section1_text):
    result_section1_dict = {}

    result_section1_dict['TOPIC'] = extract_between(section1_text, "Sektor", "EZ-Programm")
    result_section1_dict['PROGRAM'] = extract_between(section1_text, "Sektor", "EZ-Programm")
    result_section1_dict['PROJECT DESCRIPTION'] = extract_between(section1_text, "EZ-Programmziel", "Datum der letzten BE")
    result_section1_dict['PROJECT NAME'] = extract_between(section1_text, "Modul", "Modulziel")
    result_section1_dict['OBJECTIVE'] = extract_between(section1_text, "Modulziel", "Berichtszeitraum")
    result_section1_dict['PROGRESS'] = extract_between(section1_text, "Zielerreichung des Moduls", "Massnahme im Zeitplan")
    result_section1_dict['STATUS'] = extract_between(section1_text, "Massnahme im Zeitplan", "Risikoeinschätzung")
    result_section1_dict['RECOMMENDATIONS'] = extract_between(section1_text, "Vorschläge zur Modulanpas-", "Voraussichtliche")

    return result_section1_dict

def answer_questions(text,language="de"):
    # Initialize the zero-shot classification pipeline
    model_name = "deepset/gelectra-large-germanquad"
    model = AutoModelForQuestionAnswering.from_pretrained(model_name)
    tokenizer = AutoTokenizer.from_pretrained(model_name)

    # Initialize the QA pipeline
    qa_pipeline = pipeline("question-answering", model=model, tokenizer=tokenizer)
    questions = [
        "Welches ist das Titel des Moduls?",
        "Welches ist das Sektor oder das Kernthema?",
        "Welches ist das Land?",
        "Zu welchem Program oder EZ-Programm gehort das Projekt?"
        #"Welche Durchführungsorganisation aus den 4 Varianten 'giz', 'kfw', 'ptb' und 'bgr' implementiert das Projekt?"
        # "In dem Dokument was steht bei Sektor?",
        # "In dem Dokument was steht von 'EZ-Programm' bis 'EZ-Programmziel'?",
        # "In dem Dokument was steht bei EZ-Programmziel?",
        # "In dem Dokument in dem Abschnitt '1. Kurzbeschreibung' was steht bei Modul?",
        # "In dem Dokument was steht bei Zielerreichung des Moduls?",
        # "In dem Dokument in dem Abschnitt '1. Kurzbeschreibung' was steht bei Maßnahme im Zeitplan?",
        # "In dem Dokument was steht bei Vorschläge zur Modulanpassung?",
        # "In dem Dokument in dem Abschnitt 'Anlage 1: Wirkungsmatrix des Moduls' was steht unter Laufzeit als erstes Datum?",
        # "In dem Dokument in dem Abschnitt 'Anlage 1: Wirkungsmatrix des Moduls' was steht unter Laufzeit als zweites Datum?"
    ]

    # Iterate over each question and get answers
    for question in questions:
        result = qa_pipeline(question=question, context=text)
        # print(f"Question: {question}")
        # print(f"Answer: {result['answer']}\n")
        answers_dict[question] = result['answer']
    return answers_dict


def process_pdf(path):
    results_dict = {}
    results_dict["1. Kurzbeschreibung"] = \
        get_section(path, "1. Kurzbeschreibung", "2. Einordnung des Moduls")
    answers = answer_questions(results_dict["1. Kurzbeschreibung"])
    return result_section1_dict['TOPIC']

def get_first_page_text(file_data):
    doc = pdfplumber.open(BytesIO(file_data)) 
    if len(doc.pages):
        return doc.pages[0].extract_text()

if __name__ == "__main__":
    
    # Define the Gradio interface
    # iface = gr.Interface(fn=process_pdf, 
    demo = gr.Interface(fn=process_pdf, 
                     inputs=gr.File(type="binary", label="Upload PDF"),
                     outputs=gr.Textbox(label="Extracted Text"),
                     title="PDF Text Extractor",
                     description="Upload a PDF file to extract.")
    demo.launch()