def test_get_dataset_metadata():
"Test ``get_dataset_metadata``."
name, version = 'gmb', '1.0.2'
gmb_schema = get_dataset_metadata(name, version=version)
assert gmb_schema == export_schemata().schemata['datasets'].export_schema('datasets', name, version)
def test_secure_connections_succeed_load_schemata(
self, dataset_schema_url_or_path):
"Test secure connections that should succeed for :func:`pydax.load_schemata`."
# We use '/' instead of os.path.sep because URLs only accept / not \ as separators, but Windows path accepts
# both. This is not an issue for the purpose of this test.
init(update_only=True, DATASET_SCHEMA_URL=dataset_schema_url_or_path)
load_schemata(force_reload=True, tls_verification=True)
assert export_schemata().schemata[
'datasets'].retrieved_url_or_path == dataset_schema_url_or_path
def test_export_schemata(self, schema_file_absolute_dir, schema_file_https_url):
"Test high-level export-schemata function."
assert export_schemata() is not _get_schemata()
# The two returned schemata should equal
assert (json.dumps(export_schemata().schemata['datasets'].export_schema(),
sort_keys=True, indent=2, default=str) ==
json.dumps(_get_schemata().schemata['datasets'].export_schema(), sort_keys=True, indent=2, default=str))
# Different from https url used by pydax_initialization autouse fixture
new_urls = {
'DATASET_SCHEMA_URL': schema_file_absolute_dir / 'datasets.yaml',
'LICENSE_SCHEMA_URL': schema_file_absolute_dir / 'licenses.yaml'
}
init(update_only=True, **new_urls)
assert export_schemata().schemata['formats'].retrieved_url_or_path == f'{schema_file_https_url}/formats.yaml'
assert export_schemata().schemata['datasets'].retrieved_url_or_path == new_urls['DATASET_SCHEMA_URL']
assert export_schemata().schemata['licenses'].retrieved_url_or_path == new_urls['LICENSE_SCHEMA_URL']
def test_describe_dataset():
"Test ``describe_dataset``."
name, version = 'gmb', '1.0.2'
gmb_description = describe_dataset(name, version=version)
dataset_schema = export_schemata().schemata['datasets'].export_schema('datasets', name, version)
license_schema = export_schemata().schemata['licenses'].export_schema('licenses')
# Check a couple of spots
assert dataset_schema['name'] in gmb_description
assert dataset_schema['estimated_size'] in gmb_description
assert license_schema[dataset_schema["license"]]["name"] in gmb_description
# Instead of copying over the string for testing, we test a couple of important characteristics here
gmb_lines = gmb_description.splitlines()
assert len(gmb_lines) == 6 # number of lines
assert all(line.strip() != '' for line in gmb_lines) # no blank line
assert all(line.strip() == line for line in gmb_lines) # no trailing or leading whitespace
assert all(':' in line for line in gmb_lines) # no missing colons
def test_insecure_connections_load_schemata(self,
remote_dataset_schema_url,
untrust_self_signed_cert):
"Test insecure connections that should fail when ``tls_verification=True`` for ``load_schemata``."
init(update_only=True, DATASET_SCHEMA_URL=remote_dataset_schema_url)
with pytest.raises(InsecureConnectionError) as e:
load_schemata(force_reload=True, tls_verification=True)
assert remote_dataset_schema_url in str(e.value)
# Insecure load succeeds, no exception raised
load_schemata(force_reload=True, tls_verification=False)
assert export_schemata().schemata[
'datasets'].retrieved_url_or_path == remote_dataset_schema_url