def test_get_by_unique_name_multiple_results(collection, session):
# This really shouldn't happen
# Given
session.set_response([DatasetDataFactory(), DatasetDataFactory()])
# When
with pytest.raises(RuntimeError):
collection.get_by_unique_name("blah")
def test_list_datasets(paginated_collection, paginated_session):
# Given
datasets_data = DatasetDataFactory.create_batch(50)
paginated_session.set_response(datasets_data)
# When
datasets = list(paginated_collection.list(per_page=20))
# Then
assert 3 == paginated_session.num_calls
expected_first_call = FakeCall(method='GET',
path='projects/{}/datasets'.format(
paginated_collection.project_id),
params={'per_page': 20})
expected_last_call = FakeCall(method='GET',
path='projects/{}/datasets'.format(
paginated_collection.project_id),
params={
'page': 3,
'per_page': 20
})
assert expected_first_call == paginated_session.calls[0]
assert expected_last_call == paginated_session.last_call
assert 50 == len(datasets)
expected_uids = [d['id'] for d in datasets_data]
dataset_ids = [str(d.uid) for d in datasets]
assert dataset_ids == expected_uids
def test_list_datasets_infinite_loop_detect(paginated_collection,
paginated_session):
# Given
batch_size = 100
datasets_data = DatasetDataFactory.create_batch(batch_size)
# duplicate the data, this simulates an API that keeps returning the first page
datasets_data.extend(datasets_data)
paginated_session.set_response(datasets_data)
# When
datasets = list(paginated_collection.list())
# Then
assert 2 == paginated_session.num_calls # duplicate UID detected on the second call
expected_first_call = FakeCall(method='GET',
path='projects/{}/datasets'.format(
paginated_collection.project_id),
params={'per_page': batch_size})
expected_last_call = FakeCall(method='GET',
path='projects/{}/datasets'.format(
paginated_collection.project_id),
params={
'page': 2,
'per_page': batch_size
})
assert expected_first_call == paginated_session.calls[0]
assert expected_last_call == paginated_session.last_call
assert len(datasets) == batch_size
expected_uids = [d['id'] for d in datasets_data[0:batch_size]]
dataset_ids = [str(d.uid) for d in datasets]
assert dataset_ids == expected_uids
def test_register_dataset_with_idempotent_put(collection, session):
# Given
name = 'Test Dataset'
summary = 'testing summary'
description = 'testing description'
unique_name = 'foo'
session.set_response(
DatasetDataFactory(name=name,
summary=summary,
description=description,
unique_name=unique_name))
# When
session.use_idempotent_dataset_put = True
dataset = collection.register(
DatasetFactory(name=name,
summary=summary,
description=description,
unique_name=unique_name))
expected_call = FakeCall(method='PUT',
path='projects/{}/datasets'.format(
collection.project_id),
json={
'name': name,
'summary': summary,
'description': description,
'unique_name': unique_name
})
assert session.num_calls == 1
assert expected_call == session.last_call
assert name == dataset.name
def test_register_dataset_with_existing_id(collection, session):
# Given
name = 'Test Dataset'
summary = 'testing summary'
description = 'testing description'
session.set_response(
DatasetDataFactory(name=name, summary=summary,
description=description))
# When
dataset = DatasetFactory(name=name,
summary=summary,
description=description)
ds_uid = UUID('cafebeef-e341-422c-8076-35adc8828545')
dataset.uid = ds_uid
dataset = collection.register(dataset)
expected_call = FakeCall(method='PUT',
path='projects/{}/datasets/{}'.format(
collection.project_id, ds_uid),
json={
'name': name,
'summary': summary,
'description': description,
'id': str(ds_uid)
})
assert session.num_calls == 1
assert expected_call == session.last_call
assert name == dataset.name
def test_list_datasets(collection, session):
# Given
datasets_data = DatasetDataFactory.create_batch(5)
session.set_response(datasets_data)
# When
datasets = list(collection.list())
# Then
assert 1 == session.num_calls
expected_call = FakeCall(method='GET',
path='projects/{}/datasets'.format(
collection.project_id))
assert expected_call == session.last_call
assert 5 == len(datasets)
def test_get_by_unique_name_with_single_result(collection, session):
# Given
name = 'Test Dataset'
unique_name = "foo"
session.set_response(
[DatasetDataFactory(name=name, unique_name=unique_name)])
# When
result_ds = collection.get_by_unique_name(unique_name)
# Then
expected_call = FakeCall(method='GET',
path='projects/{}/datasets?unique_name={}'.format(
collection.project_id, unique_name))
assert session.num_calls == 1
assert expected_call == session.last_call
assert result_ds.name == name
assert result_ds.unique_name == unique_name
def test_register_dataset(collection, session):
# Given
name = 'Test Dataset'
summary = 'testing summary'
description = 'testing description'
session.set_response(
DatasetDataFactory(name=name, summary=summary,
description=description))
# When
dataset = collection.register(
DatasetFactory(name=name, summary=summary, description=description))
expected_call = FakeCall(method='POST',
path='projects/{}/datasets'.format(
collection.project_id),
json={
'name': name,
'summary': summary,
'description': description
})
assert session.num_calls == 1
assert expected_call == session.last_call
assert name == dataset.name