Пример #1
0
def test_size_order():
    assert_extracted_features_equal(
        extractor=SuffixExtractor(sizes=[4, 1]),
        word='test',
        features=[
            ('suffix:t', 1),
        ],
    )
Пример #2
0
def test():
    assert_extracted_features_equal(
        extractor=WordExtractor(),
        word='test',
        features=[
            ('word:test', 1),
        ],
    )
Пример #3
0
def test_sizes():
    assert_extracted_features_equal(
        extractor=SuffixExtractor(sizes=[1, 2]),
        word='test',
        features=[
            ('suffix:t', 1),
            ('suffix:st', 1),
        ],
    )
Пример #4
0
def test_cuts():
    assert_extracted_features_equal(
        extractor=StemExtractor(cuts=[1, 2], min_length=1),
        word='tests',
        features=[
            ('stem:test', 1),
            ('stem:tes', 1),
        ],
    )
Пример #5
0
def test_cut_order():
    assert_extracted_features_equal(
        extractor=StemExtractor(cuts=[2, 3, 1], min_length=2),
        word='cuts',
        features=[
            ('stem:cut', 1),
            ('stem:cu', 1),
        ],
    )
Пример #6
0
def test_min_length():
    assert_extracted_features_equal(
        extractor=StemExtractor(cuts=[1, 2], min_length=4),
        word='test',
        features=[],
    )
Пример #7
0
def test_whole_word_is_not_a_suffix():
    assert_extracted_features_equal(
        extractor=SuffixExtractor(sizes=[4]),
        word='test',
        features=[],
    )