Function test_unit_split

scripts/test_stringzilla.py:315–373

()

def test_unit_split():
    native = "line1\nline2\nline3"
    big = Str(native)

    # Splitting using a string
    lines = sz.split(big, "\n")
    assert lines == ["line1", "line2", "line3"]

    lines = sz.rsplit(big, "\n")
    assert lines == ["line1", "line2", "line3"]

    lines = sz.split(big, "\n", keepseparator=True)
    assert lines == ["line1\n", "line2\n", "line3"]

    letters = sz.split("a b c d")
    assert letters == ["a", "b", "c", "d"]

    # Splitting using character sets
    letters = sz.split_byteset("a b_c d", " _")
    assert letters == ["a", "b", "c", "d"]

    letters = sz.rsplit_byteset("a b_c d", " _")
    assert letters == ["a", "b", "c", "d"]

    # Check for equivalence with native Python strings for newline separators
    assert native.splitlines() == list(big.splitlines())
    assert native.splitlines(True) == list(big.splitlines(keeplinebreaks=True))

    # Check for equivalence with native Python strings, including SWAR boundary conditions
    assert native.split("l") == list(big.split("l"))
    assert native.split("li") == list(big.split("li"))
    assert native.split("lin") == list(big.split("lin"))
    assert native.split("line") == list(big.split("line"))
    assert native.split("line1") == list(big.split("line1"))
    assert native.split("line3") == list(big.split("line3"))
    assert native.split("\n", maxsplit=0) == list(big.split("\n", maxsplit=0))
    assert native.split("\n", maxsplit=1) == list(big.split("\n", maxsplit=1))
    assert native.split("\n", maxsplit=2) == list(big.split("\n", maxsplit=2))
    assert native.split("\n", maxsplit=3) == list(big.split("\n", maxsplit=3))
    assert native.split("\n", maxsplit=4) == list(big.split("\n", maxsplit=4))

    # Check for equivalence with native Python strings in reverse order, including boundary conditions
    assert native.rsplit("line1") == list(big.rsplit("line1"))
    assert native.rsplit("line3") == list(big.rsplit("line3"))
    assert native.rsplit("\n", maxsplit=0) == list(big.rsplit("\n", maxsplit=0))
    assert native.rsplit("\n", maxsplit=1) == list(big.rsplit("\n", maxsplit=1))
    assert native.rsplit("\n", maxsplit=2) == list(big.rsplit("\n", maxsplit=2))
    assert native.rsplit("\n", maxsplit=3) == list(big.rsplit("\n", maxsplit=3))
    assert native.rsplit("\n", maxsplit=4) == list(big.rsplit("\n", maxsplit=4))

    # If the passed separator is an empty string, the library must raise a `ValueError`
    with pytest.raises(ValueError):
        sz.split(big, "")
    with pytest.raises(ValueError):
        sz.rsplit(big, "")
    with pytest.raises(ValueError):
        sz.split_byteset(big, "")
    with pytest.raises(ValueError):