utilities.py 文件源码

def unescape(text: str) -> str:
    def fixup(m: typing.Match):
        in_text = m.group(0)
        if in_text[:2] == "&#":
            # character reference
            try:
                if in_text[:3] == "&#x":
                    return chr(int(in_text[3:-1], 16))
                else:
                    return chr(int(in_text[2:-1]))
            except ValueError:
                pass
        else:
            # named entity
            try:
                in_text = chr(html.entities.name2codepoint[in_text[1:-1]])
            except KeyError:
                pass
        return in_text  # leave as is

    return re.sub("&#?\w+;", fixup, text)