make_charset.py

#!/usr/bin/env python3

import argparse
import sys

def _main():
    ap = argparse.ArgumentParser()
    ap.add_argument('--min-range-length', '-n', type=int, default=3)
    ap.add_argument('input')
    args = ap.parse_args()

    ranges = []
    chars = []
    def add_range(first, last):
        if last - first + 1 >= args.min_range_length:
            ranges.append(f'{chr(first)}{chr(last)}')
        else:
            for i in range(first, last + 1):
                chars.append(chr(i))

    it = iter(sorted(args.input.encode('ascii')))

    try:
        start = next(it)
    except StopIteration:
        pass
    else:
        prev = start
        for c in it:
            if c != prev + 1:
                add_range(start, prev)
                start = c
            prev = c
        else:
            add_range(start, prev)

    print(repr(''.join(chars)))
    print(repr(''.join(ranges)))
    return 0

if __name__ == '__main__':
    sys.exit(_main())