Handle Unicode characters ≈ and ◁

[paperdoorknob] / texfilter.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


TexFilters: List[Tuple[str, Callable[[bytes], bytes]]] = [
    # Work around `Extra }, or forgotten \endgroup.`
    ("FixBareNesting", lambda x: re.sub(
        b'(^|\n)(\\\\(emph|st){)', b'\\1\\\\hspace{0pt}\\2', x)),
    ("UnicodeApprox", lambda x: re.sub('≈'.encode('utf-8'), b'$\\approx$', x)),
    ("UnicodeTriangle", lambda x: re.sub('◁'.encode('utf-8'), b'$\\lhd$', x)),
]


class TexFilterError(Exception):
    pass


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