def test_same_models_are_equal(dbdiskrepo):
"""
Validates that two separately constructed models using the same parameters
hash to the same artifact in provenance terms.
"""
fit1 = fit_model()
fit2 = fit_model()
assert fit1.artifact.id == fit2.artifact.id
assert fit1.artifact.value_id == fit2.artifact.value_id
assert hash(fit1) == hash(fit2)
def test_copied_models_are_equal(dbdiskrepo):
"""
Validates that a copied model (deep or shallow copied) hashes to the same
artifact as the original in provenance terms.
"""
original = fit_model()
shallow = copy(original)
assert original.artifact.id == shallow.artifact.id
assert original.artifact.value_id == shallow.artifact.value_id
assert hash(original) == hash(shallow)
deep = deepcopy(original)
assert original.artifact.id == deep.artifact.id
assert original.artifact.value_id == deep.artifact.value_id
assert hash(original) == hash(deep)
def test_lazy_dict():
lazy_dict = p.lazy_dict({'foo': lambda: 'bar', 'baz': lambda: 'qux'})
initial_hash = hash(lazy_dict)
# check that keys can be fetched
assert lazy_dict['foo'] == 'bar'
assert lazy_dict['baz'] == 'qux'
# check that the hash remains the same as values are realized
assert hash(lazy_dict) == initial_hash
# check that it raises correctly
with pytest.raises(KeyError):
lazy_dict['bar']
del lazy_dict['foo']
with pytest.raises(KeyError):
lazy_dict['foo']
def test_hashing_of_artifacts_and_proxies(repo):
@p.provenance()
def load_data():
return [1, 2, 3]
original_proxy = load_data()
original_artifact = original_proxy.artifact
loaded_artifact = repo.get_by_id(original_artifact.id)
loaded_proxy = loaded_artifact.proxy()
# All artifacts should have the same hash
assert hash(original_artifact) == hash(loaded_artifact)
# All proxies should have the same hash
assert hash(original_proxy) == hash(loaded_proxy)
# All values should have the same hash
assert hash(original_artifact.value) == hash(loaded_artifact.value)
# Artifacts and proxies should not have the same hash
assert hash(original_artifact) != hash(original_proxy)
# Proxies and values should have the same hash
assert hash(original_proxy) == hash(original_artifact.value)
def test_hashing_with_artifact_hasher_also_returns_iter_of_artifacts_preserves_hash(
repo, ):
@p.provenance()
def load_data():
return [1, 2, 3]
@p.provenance()
def create_composite(data):
return {'foo': 'bar', 'data': data}
data = load_data()
original_proxy = create_composite(data)
original_artifact = original_proxy.artifact
loaded_artifact = repo.get_by_id(original_artifact.id)
loaded_proxy = loaded_artifact.proxy()
expected_proxy_ids = frozenset((original_artifact.id, data.artifact.id))
expected_artifact_ids = frozenset((original_artifact.id, ))
original_proxy_hash, artifacts = hash(original_proxy,
hasher=ah.artifact_hasher())
ids = frozenset(a.id for a in artifacts)
assert original_proxy_hash == hash(original_proxy)
assert ids == expected_proxy_ids
original_artifact_hash, artifacts = hash(original_artifact,
hasher=ah.artifact_hasher())
ids = frozenset(a.id for a in artifacts)
assert original_artifact_hash == hash(original_artifact)
assert ids == expected_artifact_ids
loaded_artifact_hash, artifacts = hash(loaded_artifact,
hasher=ah.artifact_hasher())
ids = frozenset(a.id for a in artifacts)
assert loaded_artifact_hash == hash(loaded_artifact)
assert ids == expected_artifact_ids
loaded_proxy_hash, artifacts = hash(loaded_proxy,
hasher=ah.artifact_hasher())
ids = frozenset(a.id for a in artifacts)
assert loaded_proxy_hash == hash(loaded_proxy)
assert ids == expected_proxy_ids
def test_shared_values_hashing(base_data):
base_data = base_data.draw(data)
base_copy = lambda: copy.deepcopy(base_data)
shared_dict = {'a': base_data, 'b': base_data}
without_sharing_dict = {'a': base_copy(), 'b': base_copy()}
assert hash(shared_dict) == hash(without_sharing_dict)
shared_tuple = (base_data, base_data)
without_sharing_tuple = (base_copy(), base_copy())
assert hash(shared_tuple) == hash(without_sharing_tuple)
shared_list = [base_data, base_data]
without_sharing_list = [base_copy(), base_copy()]
assert hash(shared_list) == hash(without_sharing_list)
def test_integration_test(repo):
@p.provenance(version=0, name='initial_data')
def load_data(filename):
return [1, 2]
@p.provenance(repo=repo, remove=('to_remove', ))
def process_data_A(data, process_a_inc, to_remove):
return [i + process_a_inc for i in data]
times_called = 0
@p.provenance(ignore=('to_ignore', ))
def process_data_B(data, process_b_inc, to_ignore):
nonlocal times_called
times_called += 1
return [i + process_b_inc for i in data]
@p.provenance(tags=['tag_a'])
def combine_processed_data(inc_a, inc_b):
return [a + b for a, b in zip(inc_a, inc_b)]
def run_pipeline(filename, to_ignore, to_remove):
data = load_data(filename) # [1, 2]
inc_a = process_data_A(data, 1, to_remove) # [2, 3]
inc_b = process_data_B(data, 5, to_ignore) # [6, 7]
res = combine_processed_data(inc_a, inc_b) # [8, 10]
return res
result = run_pipeline('foo-bar.csv', 'something', 'removed')
artifact = result.artifact
inc_a_artifact = artifact.inputs['kargs']['inc_a'].artifact
inc_b_artifact = artifact.inputs['kargs']['inc_b'].artifact
assert result == [8, 10]
# check initial wrapping
assert artifact.value_id == hash([8, 10])
# check for custom_fields and tags in result
assert artifact.custom_fields == {'tags': ['tag_a']}
assert artifact.tags == ['tag_a']
# check that inputs were removed
assert inc_a_artifact.inputs == {
'kargs': {
'data': [1, 2],
'process_a_inc': 1
},
'varargs': (),
}
# check metadata
data_artifact = inc_a_artifact.inputs['kargs']['data'].artifact
assert data_artifact.name == 'initial_data'
assert data_artifact.version == 0
# Check caching
assert times_called == 1
new_res = run_pipeline('foo-bar.csv', 'something-different',
'removed-again')
assert new_res == [8, 10]
assert times_called == 1
# Check that the dependencies can be returned
dependencies = p.dependencies(result.artifact.id)
assert dependencies == [
data_artifact,
inc_a_artifact,
artifact.inputs['kargs']['inc_b'].artifact,
artifact,
]
# Check that the input_artifact_ids were properly stored
assert result.artifact.input_artifact_ids == frozenset(
(inc_a_artifact.id, inc_b_artifact.id))
def test_hashing_of_functions():
def foo(a, b):
return a + b
assert hash(foo) == hash(foo)
def test_hash_of_fortran_array_is_the_same_as_c_array():
c = np.asarray(np.arange(6000).reshape((1000, 2, 3)), order='C')
f = np.asarray(np.arange(6000).reshape((1000, 2, 3)), order='F')
assert hash(c) == hash(f)
def test_hash_of_contiguous_array_is_the_same_as_noncontiguous():
a = np.asarray(np.arange(6000).reshape((1000, 2, 3)), order='F')[:, :1, :]
b = np.ascontiguousarray(a)
assert hash(a) == hash(b)
def test_shallow_and_deep_copies_hashing(o):
original_hash = hash(o)
shallow_copy = copy.copy(o)
deep_copy = copy.deepcopy(o)
assert hash(shallow_copy) == original_hash
assert hash(deep_copy) == original_hash
