import re

from breadtube_bot.unidecode_data import UNIDECODE_DICT


DISCORD_PATTERN = re.compile(r'[^a-zA-Z0-9\-_]')


def unidecode(text: str) -> str:
    result: list[str] = []
    for char in text:
        codepoint: int = ord(char)
        if codepoint < 0x80:  # noqa: PLR2004
            result.append(DISCORD_PATTERN.sub('_', char).lower())
        elif codepoint >> 8 in UNIDECODE_DICT:
            result.append(UNIDECODE_DICT[codepoint >> 8][codepoint % 256])
        else:
            result.append('_')
    return ''.join(result)