def should_log_missing_argument_warning(self, validate_schema_mock: MagicMock) -> None:
# given
metadata_url = any_s3_url()
error_message = any_error_message()
validate_schema_mock.side_effect = ValidationError(error_message)
payload = {HTTP_METHOD_KEY: "POST", BODY_KEY: {METADATA_URL_KEY: metadata_url}}
expected_log = dumps({ERROR_KEY: error_message})
with patch.object(self.logger, "warning") as logger_mock:
# when
create_dataset_version(payload)
# then
logger_mock.assert_any_call(expected_log)
def should_log_warning_if_dataset_does_not_exist(self, datasets_model_mock: MagicMock) -> None:
# given
dataset_id = any_dataset_id()
metadata_url = any_s3_url()
error_message = any_error_message()
datasets_model_mock.return_value.get.side_effect = DoesNotExist(error_message)
payload = {METADATA_URL_KEY: metadata_url, DATASET_ID_SHORT_KEY: dataset_id}
expected_log = dumps({ERROR_KEY: error_message})
with patch.object(self.logger, "warning") as logger_mock:
# when
create_dataset_version(payload)
# then
logger_mock.assert_any_call(expected_log)
def should_log_step_function_state_machine_response(
self, start_execution_mock: MagicMock
) -> None:
# Given
start_execution_mock.return_value = step_function_response = {
"executionArn": "Some Response"
}
expected_execution_log = dumps({"response": step_function_response})
with Dataset() as dataset, patch.object(self.logger, "debug") as logger_mock:
event = {METADATA_URL_KEY: any_s3_url(), DATASET_ID_SHORT_KEY: dataset.dataset_id}
# When
create_dataset_version(event)
# Then
logger_mock.assert_any_call(expected_execution_log)
def should_log_payload(self) -> None:
# Given
with patch(
"backend.dataset_versions.create.STEP_FUNCTIONS_CLIENT.start_execution"
), Dataset() as dataset, patch.object(self.logger, "debug") as logger_mock:
event = {
HTTP_METHOD_KEY: "POST",
BODY_KEY: {
METADATA_URL_KEY: any_s3_url(),
DATASET_ID_SHORT_KEY: dataset.dataset_id,
},
}
expected_payload_log = dumps({"event": event})
# When
create_dataset_version(event)
# Then
logger_mock.assert_any_call(expected_payload_log)
def should_return_success_if_dataset_exists(subtests: SubTests) -> None:
# Given a dataset instance
now = datetime(2001, 2, 3, hour=4, minute=5, second=6, microsecond=789876, tzinfo=timezone.utc)
with patch(
"backend.dataset_versions.create.STEP_FUNCTIONS_CLIENT.start_execution"
), Dataset() as dataset:
body = {
DATASET_ID_SHORT_KEY: dataset.dataset_id,
METADATA_URL_KEY: any_s3_url(),
NOW_KEY: now.isoformat(),
}
# When requesting the dataset by ID and type
response = create_dataset_version(body)
logger.info("Response: %s", response)
# Then we should get the dataset in return
with subtests.test(msg="Status code"):
assert response[STATUS_CODE_KEY] == HTTPStatus.CREATED
with subtests.test(msg="ID"):
assert response[BODY_KEY][VERSION_ID_KEY].startswith("2001-02-03T04-05-06-789Z_")