regex2bitset.py

import re

def regex_to_bitset(r):
    '''Convert character class into bitset. Assumes ASCII range (0<=ch<=127).'''
    bitset = []
    for i in range(0, 128, 32):
        word = 0
        for j in range(i, i+32):
            if r.match(chr(j)):
                word |= 1 << (j - i)
        bitset.append(word)
    return bitset

def format_bitset(name, bitset):
    size = len(bitset)
    hex_words = ', '.join(map(hex, bitset))
    return f'static constexpr uint32_t {name}[{size}] = {{{hex_words}}};'

def format_bitset_func(name, bitset_name):
    return f'static constexpr bool {name}(int ch) {{ return ({bitset_name}[ch >> 5 & 0x3] & 1 << (ch & 0x1f)) != 0; }}'

if __name__ == '__main__':
    ident0 = regex_to_bitset(re.compile(r'[A-Za-z!$%^&*\-_+=@~#<>?/\\|]'))
    identn = regex_to_bitset(re.compile(r'[A-Za-z0-9!$%^&*\-_+=@~#<>?/\\|]'))
    special = regex_to_bitset(re.compile(r'[()\[\]{}"\'.,;:]'))
    print(format_bitset('ident0', ident0))
    print(format_bitset('identn', identn))
    print(format_bitset('special', special))
    print(format_bitset_func('is_ident0', 'ident0'))
    print(format_bitset_func('is_identn', 'identn'))
    print(format_bitset_func('is_special', 'special'))