def test_get_additional_tokens(self): assert jautils.get_additional_tokens([u'ABC']) == set() assert jautils.get_additional_tokens(set([u'ABC'])) == set() assert jautils.get_additional_tokens([u'ABC', u'ひらがな']) == \ set([u'HIRAGANA']) assert jautils.get_additional_tokens([u'やまだ', u'たろう']) == \ set([u'YAMADA', u'TARO', u'やまだたろう', u'たろうやまだ']) assert jautils.get_additional_tokens([u'はい', u'やまだ', u'たろう']) == \ set([u'HAI', u'YAMADA', u'TARO'])
def get_alternate_name_tokens(person): """Returns alternate name tokens and their variations.""" tokens = set(TextQuery(person.alternate_names).query_words) # This is no-op for non-Japanese. tokens |= set(jautils.get_additional_tokens(tokens)) return tokens