def regex(draw, state_bound=None):
bases = draw(st.lists(
st.builds(rd.char, st.integers(0, 255),), min_size=1, average_size=20))
while len(bases) > 1:
n, op = draw(
st.sampled_from((
(1, rd.star), (1, rd.nonempty),
(2, rd.union), (2, rd.intersection), (2, rd.concatenate),
(2, rd.subtract), (1, lambda r: rd.bounded(r, 1)),
(1, lambda r: rd.bounded(r, 2)),
(1, lambda r: rd.bounded(r, 3)),
(1, lambda r: rd.bounded(r, 4)),
)),
)
if n > len(bases):
continue
args = [bases.pop() for _ in range(n)]
bases.append(op(*args))
result = bases[0]
assume(result not in (rd.Empty, rd.Epsilon))
assume(not isinstance(result, rd.Characters))
if state_bound is not None:
seen = {result}
threshold = {result}
while threshold:
threshold = {
rd.derivative(u, c) for u in threshold
for c in rd.valid_starts(u)
}
threshold -= seen
seen.update(threshold)
assume(len(seen) <= state_bound)
return result
def test_bounds_are_not_nested():
x = rd.bounded(rd.star(rd.char(0)), 7)
y = rd.bounded(x, 5)
assert x.bound == 7
assert y.bound == 5
assert isinstance(y.child, rd.Star)
def test_bounds_propagate_through_subtraction():
x = rd.star(rd.char(b'\0\1'))
y = rd.literal(b'\0\0\0\1')
z = rd.subtract(x, y)
b = rd.bounded(z, 10)
assert isinstance(b, rd.Subtraction)
assert isinstance(b.left, rd.Bounded)
def test_bounds_propagate_through_intersections():
x = rd.star(rd.char(b'\0\1'))
y = rd.star(rd.char(b'\1\2'))
assert isinstance(
rd.bounded(rd.intersection(x, y), 3),
rd.Intersection
)
def test_count_below_bound_is_the_same(re, m, n):
assume(rd.has_matches(re))
m, n = sorted((m, n))
count1 = rd.LanguageCounter(*rd.build_dfa(re)).count(m)
count2 = rd.LanguageCounter(*rd.build_dfa(rd.bounded(re, n))).count(m)
assert count1 == count2
def test_can_simulate_accurately(regex, seed, max_size, param):
assume(param > 0)
assume(rd.has_matches(regex))
if max_size is not None:
assume(rd.has_matches(rd.bounded(regex, max_size)))
sim = Simulator(regex, seed)
d = sim.draw(param, max_size=max_size)
try:
next(d)
except ParamTooLarge:
assert param > 0
reject()
for _ in range(10):
x = next(d)
assert rd.matches(regex, x)
if max_size is not None:
assert len(x) <= max_size
def test_complex_graphs_may_be_finite():
x = to_basic(rd.bounded(
rd.union(rd.star(rd.char(0)), rd.star(rd.char(1))), 20))
assert not rd.is_infinite(x)
def test_bounded_is_not_infinite():
assert not rd.is_infinite(rd.bounded(rd.star(rd.char(0)), 10 ** 6))
def test_bounded_min_matches_bounds(re, n):
bd = rd.bounded(re, n)
assume(rd.has_matches(bd))
assert len(rd.lexmin(bd)) <= n
def test_bounds_propagate_through_unions():
assert isinstance(
rd.bounded(rd.union(rd.star(rd.char(0)), rd.star(rd.char(1))), 1),
rd.Union
)
def test_basic_impossible_bounds_are_empty():
assert rd.bounded(rd.char(0), -1) is rd.Empty
assert rd.bounded(rd.char(0), 0) is rd.Empty
def test_bounded_does_not_wrap_obviously_bounded(c):
assert rd.bounded(c, 1) is c
assert rd.bounded(rd.Empty, 1) is rd.Empty
count1 = rd.LanguageCounter(*rd.build_dfa(re)).count(m)
count2 = rd.LanguageCounter(*rd.build_dfa(rd.bounded(re, n))).count(m)
assert count1 == count2
def test_clearing_caches_resets_identity():
c1 = rd.char(0)
c2 = rd.char(0)
rd.clear_caches()
c3 = rd.char(0)
assert c1 is c2 is not c3
@pytest.mark.parametrize(
'c',
[rd.Empty, rd.Epsilon, rd.char(0), rd.bounded(rd.star(rd.char(0)), 1)]
)
def test_bounded_does_not_wrap_obviously_bounded(c):
assert rd.bounded(c, 1) is c
assert rd.bounded(rd.Empty, 1) is rd.Empty
def test_basic_impossible_bounds_are_empty():
assert rd.bounded(rd.char(0), -1) is rd.Empty
assert rd.bounded(rd.char(0), 0) is rd.Empty
def test_bounds_are_not_nested():
x = rd.bounded(rd.star(rd.char(0)), 7)
y = rd.bounded(x, 5)