def _fetch_validation_jobs_to_perform(
region_code_filter: Optional[str] = None,
validation_name_filter: Optional[Pattern] = None,
dataset_overrides: Optional[Dict[str, str]] = None,
) -> List[DataValidationJob]:
"""
Creates and returns validation jobs for all validations meeting the name filter,
for the given region code, and with the dataset overrides if given.
"""
validation_checks = get_all_validations()
region_configs = get_validation_region_configs()
global_config = get_validation_global_config()
validation_jobs: List[DataValidationJob] = []
for check in validation_checks:
if check.validation_name in global_config.disabled:
continue
if validation_name_filter is not None and not re.search(
validation_name_filter, check.validation_name):
continue
for region_code, region_config in region_configs.items():
if region_code_filter and region_code != region_code_filter:
continue
if check.validation_name not in region_config.exclusions:
updated_check = check.updated_for_region(region_config)
validation_jobs.append(
DataValidationJob(
validation=updated_check,
region_code=region_code,
dataset_overrides=dataset_overrides,
))
return validation_jobs
def test_cross_product_states_and_checks_production(
self, _mock_get_environment: MagicMock
) -> None:
all_validations = get_all_validations()
region_configs_to_validate = get_validation_region_configs()
global_config = get_validation_global_config()
state_codes_to_validate = region_configs_to_validate.keys()
# When you promote a state to production, we will start running validations against that state - add it to this
# list to confirm you've updated all relevant external data validation tables in production to include
# validation data for the newly promoted region.
self.assertCountEqual(
state_codes_to_validate, ["US_ID", "US_MO", "US_ND", "US_PA"]
)
num_exclusions = sum(
[len(config.exclusions) for config in region_configs_to_validate.values()]
) + len(global_config.disabled) * len(state_codes_to_validate)
expected_length = (
len(all_validations) * len(state_codes_to_validate) - num_exclusions
)
result = _fetch_validation_jobs_to_perform()
self.assertEqual(expected_length, len(result))
def test_cross_product_states_and_checks_staging(
self, _mock_get_environment: MagicMock
) -> None:
all_validations = get_all_validations()
all_region_configs = get_validation_region_configs()
global_config = get_validation_global_config()
all_regions = all_region_configs.keys()
num_exclusions = sum(
[len(config.exclusions) for config in all_region_configs.values()]
) + len(global_config.disabled) * len(all_regions)
expected_length = len(all_validations) * len(all_regions) - num_exclusions
result = _fetch_validation_jobs_to_perform()
self.assertEqual(expected_length, len(result))
def test_configs_all_reference_real_validations(self) -> None:
validation_names = {
validation.validation_name for validation in get_all_validations()
}
region_configs_to_validate = get_validation_region_configs()
global_config = get_validation_global_config()
global_disabled_names = {
validation.validation_name for validation in global_config.disabled.values()
}
global_names_not_in_validations_list = global_disabled_names.difference(
validation_names
)
self.assertEqual(
set(),
global_names_not_in_validations_list,
f"Found views referenced in global config that do not exist in validations list: "
f"{global_names_not_in_validations_list}",
)
for region_code, region_config in region_configs_to_validate.items():
region_validation_names = {
exclusion.validation_name
for exclusion in region_config.exclusions.values()
}
region_validation_names.update(
{
override.validation_name
for override in region_config.max_allowed_error_overrides.values()
}
)
region_validation_names.update(
{
override.validation_name
for override in region_config.num_allowed_rows_overrides.values()
}
)
region_names_not_in_validations_list = region_validation_names.difference(
validation_names
)
self.assertEqual(
set(),
region_names_not_in_validations_list,
f"Found views referenced in region [{region_code}] config that do not exist in validations"
f" list: {global_names_not_in_validations_list}",
)
def _fetch_validation_jobs_to_perform(
region_code_filter: Optional[str] = None, ) -> List[DataValidationJob]:
validation_checks = get_all_validations()
region_configs = get_validation_region_configs()
global_config = get_validation_global_config()
validation_jobs: List[DataValidationJob] = []
for check in validation_checks:
if check.validation_name in global_config.disabled:
continue
for region_code, region_config in region_configs.items():
if region_code_filter and region_code != region_code_filter:
continue
if check.validation_name not in region_config.exclusions:
updated_check = check.updated_for_region(region_config)
validation_jobs.append(
DataValidationJob(validation=updated_check,
region_code=region_code))
return validation_jobs
def test_parse_global_config_parses(self) -> None:
# Test passes if this parses
self.assertIsNotNone(get_validation_global_config())