0c9be6826b01fc82f6c57ae40b3e3fbbdfb95fc2,test/unit_test/test_cut_word.py,,test_cut_by_words_overlap,#,36

Before Change

    // even with the prop = 0, the overlap makes it so that there are 2 words
    assert cut_by_words("test test test", 2, 1, .5) == ["test test ",
                                                        "test test"]
    assert cut_by_words("test test test", 1, 2, .5) == ["test ", "test test ",
                                                        "test test test"]
    // again, although we only want 1 word per chunk the overlap makes it so
    // that the 2nd doc has 2 words and the 3rd has 3 words

After Change

    assert cut_by_words("test test test", 2, 1, .5) == ["test test ",
                                                        "test test"]
    try:
        _ = cut_by_words("test test test", 1, 2, .5)
        raise AssertionError("did not throw error")
    except AssertionError as error:
        assert str(error) == OVERLAP_LARGE_MESSAGE


def test_cut_by_words_proportion():
    assert cut_by_words("test test test", 2, 0, 0) == ["test test ", "test"]
    assert cut_by_words("test test test", 2, 0, .5) == ["test test ", "test"]
    assert cut_by_words("test test test", 2, 0, 1) == ["test test test"]

Instances