Handle Unicode characters ≈ and ◁

[paperdoorknob] / htmlfilter.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.


import re

from typing import Callable, List, Tuple


HTMLFilters: List[Tuple[str, Callable[[bytes], bytes]]] = [
    ("NoNBSP", lambda x: re.sub(b"( |\xc2\xa0)", b" ", x)),
]


class HTMLFilterError(Exception):
    pass


def ApplyHTMLFilters(filter_list: str, data: bytes) -> bytes:
    for filter_name in filter_list.split(','):
        filters = [f for (name, f) in HTMLFilters if name == filter_name]
        if len(filters) == 0:
            raise HTMLFilterError(f"Unknown HTML filter: {filter_name}")
        if len(filters) > 1:
            raise HTMLFilterError(
                f"Multiple HTML filters with the same name!?: {filter_name}")
        data = filters[0](data)
    return data