Normalize input file with Hazm Normalizer, replace in-place.

normalize.py

from hazm import Normalizer
from zipfile import ZipFile
import argparse
import binascii
import re
import os
import shutil


parser = argparse.ArgumentParser(description='Normalizes input text files.')
parser.add_argument('file', type=str, help='File path to normalize.')

args = parser.parse_args()
epub_filename = args.file

normalizer = Normalizer()

html_file_pattern = re.compile('\.html?$')

temp_path = '/tmp/epub-' + binascii.hexlify(os.urandom(10)).decode('ascii') + '/'
os.mkdir(temp_path)

with ZipFile(epub_filename) as epub:
    epub.extractall(temp_path)

with ZipFile(epub_filename, 'w') as epub:
    for root, dirs, files in os.walk(temp_path):
        for file in files:
            filename = os.path.join(root, file)
            if html_file_pattern.findall(file):
                with open(filename, 'rt', encoding='utf-8') as f:
                    content = normalizer.normalize(f.read())
                    content = content.replace("\u200F", "\u200C")
                with open(filename, 'wt', encoding='utf-8') as f:
                    f.write(content)

            epub.write(filename, filename.replace(temp_path, ''))

shutil.rmtree(temp_path)