def test_empty_mult_suppression():
assert conc.parse("[]0\\d").reduce() == charclass.parse("[]")
assert conc(
mult(pattern(), one), # this mult can never actually match anything
mult(charclass("0"), one),
mult(charclass("0123456789"), one),
).reduce() == charclass.parse("[]")
def test_empty_mult_suppression():
assert conc.parse("[]0\d").reduce() == charclass.parse("[]")
assert conc(
mult(pattern(), one), # this mult can never actually match anything
mult(charclass("0"), one),
mult(charclass("0123456789"), one),
).reduce() == charclass.parse("[]")
def test_odd_bug():
# Odd bug with ([bc]*c)?[ab]*
int5A = mult(charclass("bc"), star).to_fsm({"a", "b", "c", fsm.anything_else})
assert int5A.accepts([])
assert int5A.accepts("")
int5B = mult(charclass("c"), one).to_fsm({"a", "b", "c", fsm.anything_else})
assert int5B.accepts("c")
assert int5B.accepts(["c"])
int5C = int5A + int5B
assert int5C.accepts("c")
assert int5C.accepts(["c"])
def test_odd_bug():
# Odd bug with ([bc]*c)?[ab]*
int5A = mult(charclass("bc"), star).to_fsm({"a", "b", "c", fsm.anything_else})
assert int5A.accepts([])
assert int5A.accepts("")
int5B = mult(charclass("c"), one).to_fsm({"a", "b", "c", fsm.anything_else})
assert int5B.accepts("c")
assert int5B.accepts(["c"])
int5C = int5A + int5B
assert int5C.accepts("c")
assert int5C.accepts(["c"])
def test_mult_reduction_easy():
assert mult.parse("a").reduce() == charclass.parse("a")
assert mult.parse("a").reduce() == charclass("a")
assert mult.parse("a?").reduce() == mult(charclass("a"), qm)
assert mult.parse("a{0}").reduce() == emptystring
assert mult.parse("[]").reduce() == nothing
assert mult.parse("[]?").reduce() == emptystring
assert mult.parse("[]{0}").reduce() == emptystring
assert mult.parse("[]{0,5}").reduce() == emptystring
assert mult(pattern(), one).reduce() == nothing
assert mult(pattern(), qm).reduce() == emptystring
assert mult(pattern(), zero).reduce() == emptystring
assert mult(pattern(), multiplier.parse("{0,5}")).reduce() == emptystring
def test_mult_reduction_easy():
assert mult.parse("a").reduce() == charclass.parse("a")
assert mult.parse("a").reduce() == charclass("a")
assert mult.parse("a?").reduce() == mult(charclass("a"), qm)
assert mult.parse("a{0}").reduce() == emptystring
assert mult.parse("[]").reduce() == nothing
assert mult.parse("[]?").reduce() == emptystring
assert mult.parse("[]{0}").reduce() == emptystring
assert mult.parse("[]{0,5}").reduce() == emptystring
assert mult(pattern(), one).reduce() == nothing
assert mult(pattern(), qm).reduce() == emptystring
assert mult(pattern(), zero).reduce() == emptystring
assert mult(pattern(), multiplier.parse("{0,5}")).reduce() == emptystring
def test_conc_str():
assert str(conc(
mult(charclass("a"), one),
mult(charclass("b"), one),
mult(charclass("c"), one),
mult(charclass("d"), one),
mult(charclass("e"), one),
mult(~charclass("fg"), star),
mult(charclass("h"), multiplier(bound(5), bound(5))),
mult(charclass("abcdefghijklmnopqrstuvwxyz"), plus),
)) == "abcde[^fg]*h{5}[a-z]+"
def test_conc_str():
assert str(conc(
mult(charclass("a"), one),
mult(charclass("b"), one),
mult(charclass("c"), one),
mult(charclass("d"), one),
mult(charclass("e"), one),
mult(~charclass("fg"), star),
mult(charclass("h"), multiplier(bound(5), bound(5))),
mult(charclass("abcdefghijklmnopqrstuvwxyz"), plus),
)) == "abcde[^fg]*h{5}[a-z]+"
def test_pattern_equality():
assert pattern(
conc(mult(charclass("a"), one)),
conc(mult(charclass("b"), one)),
) == pattern(
conc(mult(charclass("b"), one)),
conc(mult(charclass("a"), one)),
)
assert pattern(
conc(mult(charclass("a"), one)),
conc(mult(charclass("a"), one)),
) == pattern(
conc(mult(charclass("a"), one)),
)
def test_pattern_equality():
assert pattern(
conc(mult(charclass("a"), one)),
conc(mult(charclass("b"), one)),
) == pattern(
conc(mult(charclass("b"), one)),
conc(mult(charclass("a"), one)),
)
assert pattern(
conc(mult(charclass("a"), one)),
conc(mult(charclass("a"), one)),
) == pattern(
conc(mult(charclass("a"), one)),
)
def test_mult_parsing():
assert mult.parse("[a-g]+") == mult(charclass("abcdefg"), plus)
assert mult.parse("[a-g0-8$%]+") == mult(charclass("abcdefg012345678$%"), plus)
assert mult.parse("[a-g0-8$%\\^]+") == mult(charclass("abcdefg012345678$%^"), plus)
assert mult.match("abcde[^fg]*", 5) == (
mult(~charclass("fg"), star),
11
)
assert mult.match("abcde[^fg]*h{5}[a-z]+", 11) == (
mult(charclass("h"), multiplier(bound(5), bound(5))),
15
)
assert mult.match("abcde[^fg]*h{5}[a-z]+T{1,}", 15) == (
mult(charclass("abcdefghijklmnopqrstuvwxyz"), plus),
21
)
assert mult.match("abcde[^fg]*h{5}[a-z]+T{2,}", 21) == (
mult(charclass("T"), multiplier(bound(2), inf)),
26
)
def test_mult_parsing():
assert mult.parse("[a-g]+") == mult(charclass("abcdefg"), plus)
assert mult.parse("[a-g0-8$%]+") == mult(charclass("abcdefg012345678$%"), plus)
assert mult.parse("[a-g0-8$%\\^]+") == mult(charclass("abcdefg012345678$%^"), plus)
assert mult.match("abcde[^fg]*", 5) == (
mult(~charclass("fg"), star),
11
)
assert mult.match("abcde[^fg]*h{5}[a-z]+", 11) == (
mult(charclass("h"), multiplier(bound(5), bound(5))),
15
)
assert mult.match("abcde[^fg]*h{5}[a-z]+T{1,}", 15) == (
mult(charclass("abcdefghijklmnopqrstuvwxyz"), plus),
21
)
assert mult.match("abcde[^fg]*h{5}[a-z]+T{2,}", 21) == (
mult(charclass("T"), multiplier(bound(2), inf)),
26
)
def test_mult_reduction_easy():
# mult -> mult
# mult -> charclass
assert mult(charclass("a"), one).reduce() == charclass("a")
assert mult(charclass("a"), qm).reduce() == mult(charclass("a"), qm)
assert mult(charclass("a"), zero).reduce() == emptystring
assert mult(nothing, one).reduce() == nothing
assert mult(nothing, qm).reduce() == emptystring
assert mult(nothing, zero).reduce() == emptystring
assert mult(nothing, multiplier(bound(0), bound(5))).reduce() == emptystring
assert mult(pattern(), one).reduce() == nothing
assert mult(pattern(), qm).reduce() == emptystring
assert mult(pattern(), zero).reduce() == emptystring
assert mult(pattern(), multiplier(bound(0), bound(5))).reduce() == emptystring
def test_empty_pattern_suppression():
assert mult.parse("[]?").reduce() == conc.parse("")
assert mult(pattern(), qm).reduce() == conc.parse("")
def test_conc_parsing():
assert conc.parse("abcde[^fg]*h{5}[a-z]+") == conc(
mult(charclass("a"), one),
mult(charclass("b"), one),
mult(charclass("c"), one),
mult(charclass("d"), one),
mult(charclass("e"), one),
mult(~charclass("fg"), star),
mult(charclass("h"), multiplier(bound(5), bound(5))),
mult(charclass("abcdefghijklmnopqrstuvwxyz"), plus),
)
assert conc.parse("[bc]*[ab]*") == conc(
mult(charclass("bc"), star),
mult(charclass("ab"), star),
)
assert conc.parse("abc...") == conc(
mult(charclass("a"), one),
mult(charclass("b"), one),
mult(charclass("c"), one),
mult(dot, one),
mult(dot, one),
mult(dot, one),
)
assert conc.parse("\\d{4}-\\d{2}-\\d{2}") == conc(
mult(charclass("0123456789"), multiplier(bound(4), bound(4))),
mult(charclass("-"), one),
mult(charclass("0123456789"), multiplier(bound(2), bound(2))),
mult(charclass("-"), one),
mult(charclass("0123456789"), multiplier(bound(2), bound(2))),
)
def test_conc_equality():
assert conc(mult(charclass("a"), one)) == conc(mult(charclass("a"), one))
assert conc(mult(charclass("a"), one)) != conc(mult(charclass("b"), one))
assert conc(mult(charclass("a"), one)) != conc(mult(charclass("a"), qm))
assert conc(mult(charclass("a"), one)) != conc(mult(charclass("a"), multiplier(bound(1), bound(2))))
assert conc(mult(charclass("a"), one)) != emptystring
def test_pattern_str():
assert str(pattern(
conc(mult(charclass("a"), one)),
conc(mult(charclass("b"), one)),
)) == "a|b"
assert str(pattern(
conc(mult(charclass("a"), one)),
conc(mult(charclass("a"), one)),
)) == "a"
assert str(pattern(
conc(
mult(charclass("a"), one),
mult(charclass("b"), one),
mult(charclass("c"), one),
),
conc(
mult(charclass("d"), one),
mult(charclass("e"), one),
mult(charclass("f"), one),
mult(
pattern(
conc(
mult(charclass("g"), one),
mult(charclass("h"), one),
mult(charclass("i"), one),
),
conc(
mult(charclass("j"), one),
mult(charclass("k"), one),
mult(charclass("l"), one),
),
), one
),
),
)) == "abc|def(ghi|jkl)"
def test_pattern_parsing():
assert pattern.parse("abc|def(ghi|jkl)") == pattern(
conc(
mult(charclass("a"), one),
mult(charclass("b"), one),
mult(charclass("c"), one),
),
conc(
mult(charclass("d"), one),
mult(charclass("e"), one),
mult(charclass("f"), one),
mult(
pattern(
conc(
mult(charclass("g"), one),
mult(charclass("h"), one),
mult(charclass("i"), one),
),
conc(
mult(charclass("j"), one),
mult(charclass("k"), one),
mult(charclass("l"), one),
),
), one
),
)
)
# Accept the "non-capturing group" syntax, "(?: ... )" but give it no
# special significance
assert parse("(?:)") == parse("()")
assert parse("(?:abc|def)") == parse("(abc|def)")
parse("(:abc)") # should give no problems
# Named groups
assert pattern.parse("(?P<ng1>abc)") == parse("(abc)")
def test_mult_str():
assert str(bound(2)) == "2"
assert str(inf) == ""
assert str(multiplier(bound(2), inf)) == "{2,}"
a = charclass("a")
assert str(mult(a, one)) == "a"
assert str(mult(a, multiplier(bound(2), bound(2)))) == "a{2}"
assert str(mult(a, multiplier(bound(3), bound(3)))) == "a{3}"
assert str(mult(a, multiplier(bound(4), bound(4)))) == "a{4}"
assert str(mult(a, multiplier(bound(5), bound(5)))) == "a{5}"
assert str(mult(a, qm)) == "a?"
assert str(mult(a, star)) == "a*"
assert str(mult(a, plus)) == "a+"
assert str(mult(a, multiplier(bound(2), bound(5)))) == "a{2,5}"
assert str(mult(a, multiplier(bound(2), inf))) == "a{2,}"
assert str(mult(d, one)) == "\\d"
assert str(mult(d, multiplier(bound(2), bound(2)))) == "\\d{2}"
assert str(mult(d, multiplier(bound(3), bound(3)))) == "\\d{3}"
def test_mult_equality():
assert mult(charclass("a"), one) == mult(charclass("a"), one)
assert mult(charclass("a"), one) != mult(charclass("b"), one)
assert mult(charclass("a"), one) != mult(charclass("a"), qm)
assert mult(charclass("a"), one) != mult(charclass("a"), multiplier(bound(1), bound(2)))
assert mult(charclass("a"), one) != charclass("a")
def test_pattern_parsing():
assert pattern.parse("abc|def(ghi|jkl)") == pattern(
conc(
mult(charclass("a"), one),
mult(charclass("b"), one),
mult(charclass("c"), one),
),
conc(
mult(charclass("d"), one),
mult(charclass("e"), one),
mult(charclass("f"), one),
mult(
pattern(
conc(
mult(charclass("g"), one),
mult(charclass("h"), one),
mult(charclass("i"), one),
),
conc(
mult(charclass("j"), one),
mult(charclass("k"), one),
mult(charclass("l"), one),
),
), one
),
)
)
# Accept the "non-capturing group" syntax, "(?: ... )" but give it no
# special significance
assert parse("(?:)") == parse("()")
assert parse("(?:abc|def)") == parse("(abc|def)")
parse("(:abc)") # should give no problems
# Named groups
assert pattern.parse("(?P<ng1>abc)") == parse("(abc)")
def test_mult_equality():
assert mult(charclass("a"), one) == mult(charclass("a"), one)
assert mult(charclass("a"), one) != mult(charclass("b"), one)
assert mult(charclass("a"), one) != mult(charclass("a"), qm)
assert mult(charclass("a"), one) != mult(charclass("a"), multiplier(bound(1), bound(2)))
assert mult(charclass("a"), one) != charclass("a")
def test_mult_str():
assert str(bound(2)) == "2"
assert str(inf) == ""
assert str(multiplier(bound(2), inf)) == "{2,}"
a = charclass("a")
assert str(mult(a, one)) == "a"
assert str(mult(a, multiplier(bound(2), bound(2)))) == "a{2}"
assert str(mult(a, multiplier(bound(3), bound(3)))) == "a{3}"
assert str(mult(a, multiplier(bound(4), bound(4)))) == "a{4}"
assert str(mult(a, multiplier(bound(5), bound(5)))) == "a{5}"
assert str(mult(a, qm)) == "a?"
assert str(mult(a, star)) == "a*"
assert str(mult(a, plus)) == "a+"
assert str(mult(a, multiplier(bound(2), bound(5)))) == "a{2,5}"
assert str(mult(a, multiplier(bound(2), inf))) == "a{2,}"
assert str(mult(d, one)) == "\\d"
assert str(mult(d, multiplier(bound(2), bound(2)))) == "\\d{2}"
assert str(mult(d, multiplier(bound(3), bound(3)))) == "\\d{3}"
def test_conc_parsing():
assert conc.parse("abcde[^fg]*h{5}[a-z]+") == conc(
mult(charclass("a"), one),
mult(charclass("b"), one),
mult(charclass("c"), one),
mult(charclass("d"), one),
mult(charclass("e"), one),
mult(~charclass("fg"), star),
mult(charclass("h"), multiplier(bound(5), bound(5))),
mult(charclass("abcdefghijklmnopqrstuvwxyz"), plus),
)
assert conc.parse("[bc]*[ab]*") == conc(
mult(charclass("bc"), star),
mult(charclass("ab"), star),
)
assert conc.parse("abc...") == conc(
mult(charclass("a"), one),
mult(charclass("b"), one),
mult(charclass("c"), one),
mult(dot, one),
mult(dot, one),
mult(dot, one),
)
assert conc.parse("\\d{4}-\\d{2}-\\d{2}") == conc(
mult(charclass("0123456789"), multiplier(bound(4), bound(4))),
mult(charclass("-"), one),
mult(charclass("0123456789"), multiplier(bound(2), bound(2))),
mult(charclass("-"), one),
mult(charclass("0123456789"), multiplier(bound(2), bound(2))),
)
def test_pattern_parsing():
assert pattern.parse("abc|def(ghi|jkl)") == pattern(
conc(
mult(charclass("a"), one),
mult(charclass("b"), one),
mult(charclass("c"), one),
),
conc(
mult(charclass("d"), one),
mult(charclass("e"), one),
mult(charclass("f"), one),
mult(
pattern(
conc(
mult(charclass("g"), one),
mult(charclass("h"), one),
mult(charclass("i"), one),
),
conc(
mult(charclass("j"), one),
mult(charclass("k"), one),
mult(charclass("l"), one),
),
), one
),
)
)
def test_pattern_str():
assert str(pattern(
conc(mult(charclass("a"), one)),
conc(mult(charclass("b"), one)),
)) == "a|b"
assert str(pattern(
conc(mult(charclass("a"), one)),
conc(mult(charclass("a"), one)),
)) == "a"
assert str(pattern(
conc(
mult(charclass("a"), one),
mult(charclass("b"), one),
mult(charclass("c"), one),
),
conc(
mult(charclass("d"), one),
mult(charclass("e"), one),
mult(charclass("f"), one),
mult(
pattern(
conc(
mult(charclass("g"), one),
mult(charclass("h"), one),
mult(charclass("i"), one),
),
conc(
mult(charclass("j"), one),
mult(charclass("k"), one),
mult(charclass("l"), one),
),
), one
),
),
)) == "abc|def(ghi|jkl)"
def test_conc_equality():
assert conc(mult(charclass("a"), one)) == conc(mult(charclass("a"), one))
assert conc(mult(charclass("a"), one)) != conc(mult(charclass("b"), one))
assert conc(mult(charclass("a"), one)) != conc(mult(charclass("a"), qm))
assert conc(mult(charclass("a"), one)) != conc(mult(charclass("a"), multiplier(bound(1), bound(2))))
assert conc(mult(charclass("a"), one)) != emptystring