判断国内电话号码运营商，将多个电话号码按照运营商分组

sp.py

import collections


class groupby2(object):
    """
    [k for k, g in groupby2('AAAABBBCCDAABBB')] --> A B C D
    """

    def __init__(self, iterable, key=lambda x: x):
        self.key_func = key
        self.iterable = iter(iterable)

    def __iter__(self):
        result = collections.defaultdict(list)
        for elem in self.iterable:
            key = self.key_func(elem)
            result[key].append(elem)
        for key, elem in result.iteritems():
            yield key, elem


class ServiceProvider(object):
    """
    reference: https://zh.wikipedia.org/wiki/%E4%B8%AD%E5%9B%BD%E5%86%85%E5%9C%B0%E7%A7%BB%E5%8A%A8%E7%BB%88%E7%AB%AF%E9%80%9A%E8%AE%AF%E5%8F%B7%E7%A0%81
    date: 2017-12-04 13:31:47 CST
    """
    MOBILE = set(range(1340, 1349)) | {
        135, 136, 137, 138, 139, 1440, 147, 148, 150, 151, 152, 157, 158, 159, 178, 182, 183, 184,
        187, 188, 198,
    } | {147} | {1703, 1705, 1706}
    UNICOM = {130, 131, 132, 145, 146, 155, 156, 175, 176, 185, 186, } | {145} | {
        1704, 1707, 1708, 1709, 171
    }
    TELECOM = set(range(17400, 17406)) | {
        133, 1349, 1410, 149, 153, 173, 177, 180, 181, 189, 199
    } | {149}

    def __init__(self):
        self.phone_sp = {}
        for value in self.MOBILE:
            self.phone_sp[value] = 'mobile'
        for value in self.UNICOM:
            self.phone_sp[value] = 'unicom'
        for value in self.TELECOM:
            self.phone_sp[value] = 'telecom'

    def get_sp_by_phone_number(self, number_prefix):
        number_prefix = str(int(number_prefix))
        if int(number_prefix[:3]) in self.phone_sp:
            return self.phone_sp[int(number_prefix[:3])]
        if int(number_prefix[:4]) in self.phone_sp:
            return self.phone_sp[int(number_prefix[:4])]
        else:
            return 'unknown'


if __name__ == '__main__':
    a = [
        130064, 130081, 130182, 2133113, 133164, 133209, 133227, 133309, 133367, 134419, 138992,
        138995, 139067, 139081, 786645, 786668, 786828
    ]
    key_func = ServiceProvider().get_sp_by_phone_number
    for i, k in groupby2(a, key=lambda x: key_func(x)):
        print i, len(k), list(k)
        
# telecom 5 [133164, 133209, 133227, 133309, 133367]
# unknown 4 [2133113, 786645, 786668, 786828]
# unicom 3 [130064, 130081, 130182]
# mobile 5 [134419, 138992, 138995, 139067, 139081]