rld.py

_roots = None
def roots_level_domains():
    global _roots
    if _roots is not None:
        return _roots
    # get https://publicsuffix.org/list/effective_tld_names.dat
    rld = []
    with open('effective_tld_names.dat', 'rb') as f:
        for l in f:
            ul = l.strip().decode('UTF-8')
            if ul and not ul.startswith(u'//'):
                rld += [list(reversed(ul.split('.')))]

    _roots = dict()
    for parts in rld:
        curr = _roots
        for p in parts:
            if p == '*':
                curr['valid'] = True
            if p not in curr:
                curr[p] = dict()
            curr = curr[p]
        curr['valid'] = True

    return _roots


def rld(host):
    parts = list(reversed(host.split('.')))
    roots = roots_level_domains()
    res = []

    curr = roots
    for p in parts:
        subs = curr.get(p, {})
        if subs == {}:
            subs = curr.get('*', {})

        if subs.get('valid', False):
            res += [p]
        else:
            return '.'.join(reversed(res))
        curr = subs