test_corpora_hashdictionary.py 文件源码

def testBuild(self):
        d = HashDictionary(self.texts, myhash=zlib.adler32)
        expected =  {5232: 2,
                     5798: 3,
                     10608: 2,
                     12466: 2,
                     12736: 3,
                     15001: 2,
                     18451: 3,
                     23844: 3,
                     28591: 2,
                     29104: 2,
                     31002: 2,
                     31049: 2}

        self.assertEqual(d.dfs, expected)
        expected = {'minors': 15001, 'graph': 18451, 'system': 5798, 'trees': 23844, 'eps': 31049, 'computer': 10608, 'survey': 28591, 'user': 12736, 'human': 31002, 'time': 29104, 'interface': 12466, 'response': 5232}

        for ex in expected:
            self.assertEqual(d.token2id[ex], expected[ex])