bytesizes.py

import itertools


def readable_size(nbytes, i=True):
    sizes = ("B KiB MiB GiB TiB" if i else "B KB MB GB TB").split()
    last = sizes[-1]
    d = 1024 if i else 1000
    for sz in sizes:
        if nbytes < d or sz is last:
            return '{} {}'.format(format(nbytes, '.1f').rstrip('0').rstrip('.'), sz)
        nbytes /= d


sizes = ['B', 'kB KB KiB', 'MB Mb MiB', 'GB GiB', 'TB TiB']
sizes = sorted(
    (itertools.chain((sz, i) for i, szs in enumerate(sizes) for sz in szs.split())),
    key=lambda kv: len(kv[0]), reverse=True
)

def parse_size(size_string, i=True):
    s = size_string.strip()
    for size, value in sizes:
        if s.endswith(size):
            r = s[:-len(size)].strip()
            r = float(r.replace(',', '.'))
            if i or 'i' in size:
                r *= 1024**value
            else:
                r *= 1000**value
            return round(r)
    else:
        raise ValueError(size_string)


for s in "100MB 23KB 2GB 51.2MB".split():
    print(s, parse_size(s), readable_size(parse_size(s)))