test_character.py

from sre_parse import parse as sre_parse
from char import Character
from sre import SreOpParser


def from_regex(pattern: str) -> Character:
    (parsed_char,) = sre_parse(pattern)
    char = SreOpParser.parse_op(*parsed_char)
    assert isinstance(char, Character)
    return char


def test_literal_and():
    assert from_regex("[abc]") & from_regex("[bcd]") == from_regex("[bc]")


def test_literal_negated_and():
    assert from_regex("[^abc]") & from_regex("[^bcd]") == from_regex("[^abcd]")


def test_literal_mixed_and():
    assert from_regex("[abcz]") & from_regex("[^bcd]") == from_regex("[az]")
    assert from_regex("[^bcd]") & from_regex("[abcz]") == from_regex("[az]")


def test_category_and():
    assert from_regex(r"[\s\d]") & from_regex(r"[\d\w]") == from_regex(r"\d")


def test_category_negated_and():
    assert from_regex(r"[^\s\d]") & from_regex(r"[^\d\w]") == from_regex(r"[^\s\d\w]")


def test_category_negated_and_simplifies_to_nothing():
    assert (from_regex(r"[^\s\d]") & from_regex(r"[^\D]")) is None


def test_mixed_and():
    assert from_regex(r"[abc123\s]") & from_regex(r"[^\d\s]") == from_regex("[abc]")


def test_mixed_and_none():
    c = from_regex(r"[123]") & from_regex(r"[^\d\s]")
    assert c is None


def test_and_any_none():
    any = Character.ANY()
    for other in (
        from_regex(r"a"),
        from_regex(r"\s"),
        from_regex(r"[a\s\S\d]"),
        from_regex(r"[A-z]"),
        from_regex(r"[^A-z\d]"),
    ):
        assert any & other == other
        assert other & any == other
        assert any & None == None
        assert None & any == None


def test_class():
    assert from_regex("[abc]").exact_character_class() == from_regex("[cba]")


def test_negate_simple():
    assert from_regex("a").negate() == from_regex("[^a]")
    assert from_regex(r"\w").negate() == from_regex(r"[^\w]")
    assert from_regex("[^ab]").negate() == from_regex("[ab]")
    assert from_regex(r"[^\s]").negate() == from_regex(r"\s")


def test_negate_mixed():
    assert from_regex(r"[a\s\w]").negate() == from_regex(r"[^a\s\w]")


def test_or():
    assert from_regex("a") | from_regex("a") == from_regex("a")
    assert from_regex("a") | from_regex("b") == from_regex("[ab]")
    assert from_regex(r"\w") | from_regex("b") == from_regex(r"\w").expand_categories()
    assert (
        from_regex(r"\w") | from_regex("9") == from_regex(r"[9\w]").expand_categories()
    )
    assert from_regex("[^a]") | from_regex("[^b]") == from_regex(".")