Wrong main.py file inside pdf_reader folder.

[Ahmad-mufied]

Wrong main.py file which should contain all things related to PDF reader but instead contains about pygame

[carolinavc99]

Here is the code. The only change i did was i turned print('-'*10) into a function called divide().

import re
from collections import Counter
from PyPDF2 import PdfReader


def extract_text_from_pdf(pdf_file: str) -> list[str]:
    with open(pdf_file, 'rb') as pdf:
        reader = PdfReader(pdf_file, strict=False)

        print('Pages:', len(reader.pages))
        divide()

        return [page.extract_text() for page in reader.pages]


def divide():
    print('-' * 75)


def count_words(text_list: list[str]) -> Counter:
    all_words: list[str] = []
    for text in text_list:
        split_text: list[str] = re.split(r'\s+|[,;?!.-]\s*', text.lower())
        all_words += [word for word in split_text if word] # exclude empty string
    return Counter(all_words)


def main():
    extracted_text: list[str] = extract_text_from_pdf('sample.pdf')
    counter: Counter = count_words(text_list=extracted_text)

    for page in extracted_text:
        print(page)

    divide()

    for word, mentions in counter.most_common(5):
        print(f'{word:10} : {mentions} uses')


if __name__ == '__main__':
    main()