def test_can_learn_to_normalize_the_unnormalized():
with preserving_dfas():
prev = len(dfas.SHRINKING_DFAS)
dfas.normalize(TEST_DFA_NAME,
non_normalized_test_function,
allowed_to_update=True)
assert len(dfas.SHRINKING_DFAS) == prev + 1
def test_will_error_if_does_not_normalise_and_cannot_update():
with pytest.raises(dfas.FailedToNormalise) as excinfo:
dfas.normalize(
"bad",
a_bad_test_function(),
required_successes=10,
allowed_to_update=False,
)
assert "not allowed" in excinfo.value.args[0]
def test_will_error_if_takes_too_long_to_normalize():
with preserving_dfas():
with pytest.raises(dfas.FailedToNormalise) as excinfo:
dfas.normalize(
"bad",
a_bad_test_function(),
required_successes=1000,
allowed_to_update=True,
max_dfas=0,
)
assert "too hard" in excinfo.value.args[0]
def test_makes_no_changes_if_already_normalized():
def test_function(data):
if data.draw_bits(16) >= 1000:
data.mark_interesting()
with preserving_dfas():
before = dict(dfas.SHRINKING_DFAS)
dfas.normalize(TEST_DFA_NAME, test_function, allowed_to_update=True)
after = dict(dfas.SHRINKING_DFAS)
assert after == before
def test_harder_strategies_normalize_to_minimal(strategy, normalize_kwargs):
def test_function(data):
try:
v = data.draw(strategy)
except UnsatisfiedAssumption:
data.mark_invalid()
data.output = repr(v)
data.mark_interesting()
dfas.normalize(repr(strategy),
test_function,
random=Random(0),
**normalize_kwargs)
def test_common_strategies_normalize_small_values(strategy, n, normalize_kwargs):
excluded = list(map(repr, islice(iter_values(strategy, unique_by=repr), n)))
def test_function(data):
try:
v = data.draw(strategy)
except UnsatisfiedAssumption:
data.mark_invalid()
data.output = repr(v)
if repr(v) not in excluded:
data.mark_interesting()
dfas.normalize(repr(strategy), test_function, **normalize_kwargs)
def test_will_error_on_uninteresting_test():
with pytest.raises(AssertionError):
dfas.normalize(TEST_DFA_NAME, lambda data: data.draw_bits(64))
