def test_same_skip_same_sample(self):
""" Multiple submission skips for one sample collapse by reason. """
# Designate all-but-one of the failure reasons as the observations.
for failures in itertools.combinations(self.SKIP_REASONS,
len(self.SKIP_REASONS) - 1):
# Build up the expectations and the input.
all_skip_reasons = []
# Randomize skip/fail count for each reason.
for skip in failures:
n_skip = nprand.randint(low=2, high=5, size=1)[0]
all_skip_reasons.extend([skip] * n_skip)
# Aggregation is order-agnostic...
random.shuffle(all_skip_reasons)
original_skip_reasons = [(all_skip_reasons, "control-sample")]
# ...and maps each reason to pair of sample and count.
expected_aggregation = {
skip: ["control-sample"]
for skip in set(all_skip_reasons)
}
# Validate.
observed_aggregation = aggregate_exec_skip_reasons(
original_skip_reasons)
assert expected_aggregation == observed_aggregation
def test_many_samples_once_each_few_failures(self):
""" One/few reasons for several/many samples, one skip each. """
# Looping is to boost confidence from randomization.
# We don't really want each case to be a parameterization.
for reasons in itertools.combinations(self.SKIP_REASONS, 2):
original_reasons = []
expected = defaultdict(list)
# Choose one or both reasons as single-fail for this sample.
for sample in self.SAMPLE_NAMES:
this_sample_reasons = nprand.choice(reasons,
size=nprand.choice([1, 2]),
replace=False)
for reason in this_sample_reasons:
expected[reason].append(sample)
original_reasons.append((this_sample_reasons, sample))
observed = aggregate_exec_skip_reasons(original_reasons)
assert expected == observed
def test_no_failures(self, empty_skips):
""" Aggregation step returns empty collection for no-fail case. """
assert defaultdict(list) == aggregate_exec_skip_reasons(empty_skips)
