def test_get_latest_printing(self, client, created_entities, capsys):
dataset = client.set_dataset(type="local")
created_entities.append(dataset)
version = dataset.create_version(path=__file__)
dataset.get_latest_version(ascending=True)
captured = capsys.readouterr()
assert "got existing dataset version: {}".format(version.id) in captured.out
def test_local_file(self, client, created_entities):
filepath = "conftest.py"
dataset = client.set_dataset(type="local")
created_entities.append(dataset)
version = dataset.create_version(filepath)
retrieved = dataset.get_latest_version()
assert version.id == retrieved.id # of course, but just to be sure
base_path = os.path.abspath(filepath)
self.assert_base_path(version, base_path)
self.assert_base_path(retrieved, base_path)
def test_local_dir(self, client, created_datasets):
dirpath = "."
dataset = client.set_dataset(type="local")
created_datasets.append(dataset)
version = dataset.create_version(dirpath)
retrieved = dataset.get_latest_version()
assert version.id == retrieved.id # of course, but just to be sure
base_path = os.path.abspath(dirpath)
self.assert_base_path(version, base_path)
self.assert_base_path(retrieved, base_path)
def test_reincarnation(self, client, created_datasets):
"""Consecutive identical versions are assigned the same ID."""
dataset = client.set_dataset(type="local")
created_datasets.append(dataset)
version1 = dataset.create_version(path=__file__)
version2 = dataset.create_version(path=__file__)
assert version1.id == version2.id
versions = dataset.get_all_versions()
assert len(versions) == 1
version = dataset.get_latest_version(ascending=True)
assert version.id == version1.id
def test_s3_bucket(self, client, created_entities):
bucket_name = "verta-starter"
botocore = pytest.importorskip("botocore")
try:
dataset = client.set_dataset(type="s3")
created_entities.append(dataset)
version = dataset.create_version(bucket_name)
except botocore.exceptions.ClientError:
pytest.skip("insufficient AWS credentials")
retrieved = dataset.get_latest_version()
assert version.id == retrieved.id # of course, but just to be sure
self.assert_base_path(version, bucket_name)
self.assert_base_path(retrieved, bucket_name)
def test_get_versions(self, client, created_entities):
dataset = client.set_dataset(type="local")
created_entities.append(dataset)
version1 = dataset.create_version(path=__file__)
assert version1.id
version2 = dataset.create_version(path=pytest.__file__)
assert version2.id
versions = dataset.get_all_versions()
assert len(versions) == 2
dataset_version1 = client.get_dataset_version(id=version1.id)
assert dataset_version1.id == version1.id
version = dataset.get_latest_version(ascending=True)
assert version.id == version1.id
