bstidy.py

bstidy.py

from ebooklib.plugins.base import BasePlugin
from bs4 import BeautifulSoup


def tidy_cleanup(content):
    return BeautifulSoup(content, "html.parser", from_encoding="UTF-8").prettify()


class TidyPlugin(BasePlugin):
    NAME = 'Tidy HTML using BeautifulSoup4'

    def html_before_write(self, book, chapter):
        if not chapter.content:
            return None

        return tidy_cleanup(chapter.content)

    def html_after_read(self, book, chapter):
        if not chapter.content:
            return None

        return tidy_cleanup(chapter.content)