helper.py 文件源码

def pinyinify(string):
    # TODO: Use static file instead of constructing table in real time
    table = dict()
    for i in range(sys.maxunicode):
        if re.match('P|S|Z|C', unicodedata.category(chr(i))) is not None:
            table[i] = '-'
    string = string.translate(table)
    for char in [x for x in string if unicodedata.name(x).startswith('CJK')]:
        string = string.replace(char, pinyin.get(char, format='strip') + '-')
    string = re.sub('\-+', '-', string)
    return pinyin.get(string, delimiter='', format='strip').lower()