Ejemplo n.º 1
0
    def test_index(self):
        index = MinHashIndex(
            redis.clusters.get('default'),
            0xFFFF,
            8,
            2,
            60 * 60,
            12,
        )

        index.record('example', '1', [('index', 'hello world')])
        index.record('example', '2', [('index', 'hello world')])
        index.record('example', '3', [('index', 'jello world')])
        index.record('example', '4', [('index', 'yellow world')])
        index.record('example', '4', [('index', 'mellow world')])
        index.record('example', '5', [('index', 'pizza world')])

        results = index.query('example', '1', ['index'])[0]
        assert results[0] == ('1', 1.0)
        assert results[1] == ('2', 1.0)  # identical contents
        assert results[2][0] in (
            '3', '4')  # equidistant pairs, order doesn't really matter
        assert results[3][0] in ('3', '4')
        assert results[4][0] == '5'

        index.delete('example', [('index', '3')])
        assert [key for key, _ in index.query('example', '1', ['index'])[0]
                ] == ['1', '2', '4', '5']
Ejemplo n.º 2
0
    def test_index(self):
        index = MinHashIndex(
            redis.clusters.get('default'),
            'sim',
            signature_builder,
            16,
            60 * 60,
            12,
        )

        index.record('example', '1', [('index', 'hello world')])
        index.record('example', '2', [('index', 'hello world')])
        index.record('example', '3', [('index', 'jello world')])
        index.record('example', '4', [('index', 'yellow world')])
        index.record('example', '4', [('index', 'mellow world')])
        index.record('example', '5', [('index', 'pizza world')])

        results = index.compare('example', '1', ['index'])[0]
        assert results[0] == ('1', 1.0)
        assert results[1] == ('2', 1.0)  # identical contents
        assert results[2][0] in (
            '3', '4')  # equidistant pairs, order doesn't really matter
        assert results[3][0] in ('3', '4')
        assert results[4][0] == '5'

        results = index.classify('example', [('index', 'hello world')])[0]
        assert results[0:2] == [('1', 1.0), ('2', 1.0)]
        assert results[2][0] in (
            '3', '4')  # equidistant pairs, order doesn't really matter
        assert results[3][0] in ('3', '4')
        assert results[4][0] == '5'

        index.delete('example', [('index', '3')])
        assert [key for key, _ in index.compare('example', '1', ['index'])[0]
                ] == ['1', '2', '4', '5']

        assert MinHashIndex(
            redis.clusters.get('default'),
            'sim2',
            signature_builder,
            8,
            60 * 60,
            12,
        ).compare('example', '1', ['index']) == [[]]