Handle Unicode characters ≈ and ◁

[paperdoorknob] / domfilter.py

# paperdoorknob: Print glowfic
#
# This program is free software: you can redistribute it and/or modify it
# under the terms of the GNU General Public License as published by the
# Free Software Foundation, version 3.


from typing import Any, Callable, List, Tuple

from bs4.element import Tag


def _changeTag(tag: Tag, new_name: str) -> Tag:
    tag.name = new_name
    return tag


DOMFilters: List[Tuple[str, Callable[[Tag], Any]]] = [
    ("Strike", lambda x: [_changeTag(span, 's')
                          for span in x.find_all("span", style="text-decoration: line-through;")]),
]


class DOMFilterError(Exception):
    pass


def ApplyDOMFilters(filter_list: str, tag: Tag) -> None:
    for filter_name in filter_list.split(','):
        filters = [f for (name, f) in DOMFilters if name == filter_name]
        if len(filters) == 0:
            raise DOMFilterError(f"Unknown DOM filter: {filter_name}")
        if len(filters) > 1:
            raise DOMFilterError(
                f"Multiple DOM filters with the same name!?: {filter_name}")
        filters[0](tag)