def test_success_some_errors(self) -> None:
result = SamenessStringsValidationResultDetails(
num_error_rows=5,
total_num_rows=100,
max_allowed_error=0.05,
non_null_counts_per_column_per_partition=[
(tuple(), {"a": 100, "b": 100, "c": 100}),
],
)
self.assertTrue(result.was_successful())
self.assertIsNone(result.failure_description())
def test_sameness_check_strings_different_values_no_allowed_error(self) -> None:
self.mock_client.run_query_async.return_value = [{"a": "a", "b": "b", "c": "c"}]
job = DataValidationJob(
region_code="US_XX",
validation=SamenessDataValidationCheck(
validation_category=ValidationCategory.EXTERNAL_INDIVIDUAL,
validation_type=ValidationCheckType.SAMENESS,
comparison_columns=["a", "b", "c"],
partition_columns=[],
sameness_check_type=SamenessDataValidationCheckType.STRINGS,
view_builder=SimpleBigQueryViewBuilder(
dataset_id="my_dataset",
view_id="test_view",
description="test_view description",
view_query_template="select * from literally_anything",
),
),
)
result = SamenessValidationChecker.run_check(job)
self.assertEqual(
result,
DataValidationJobResult(
validation_job=job,
result_details=SamenessStringsValidationResultDetails(
num_error_rows=1,
total_num_rows=1,
max_allowed_error=0.0,
non_null_counts_per_column_per_partition=[
(tuple(), {"a": 1, "b": 1, "c": 1}),
],
),
),
)
def test_failure_all_errors(self) -> None:
result = SamenessStringsValidationResultDetails(
num_error_rows=1,
total_num_rows=1,
max_allowed_error=0.0,
non_null_counts_per_column_per_partition=[
(tuple(), {"a": 1, "b": 1, "c": 1}),
],
)
self.assertFalse(result.was_successful())
self.assertEqual(
result.failure_description(),
"1 out of 1 row(s) did not contain matching strings. "
"The acceptable margin of error is only 0.0, but the "
"validation returned an error rate of 1.0.",
)
def test_multiple_partitions(self) -> None:
result = SamenessStringsValidationResultDetails(
num_error_rows=3,
total_num_rows=7,
max_allowed_error=0.0,
non_null_counts_per_column_per_partition=[
(("US_XX", "2021-01-31"), {"a": 3, "b": 3}),
(("US_XX", "2020-12-31"), {"a": 2, "b": 3}),
],
)
self.assertFalse(result.was_successful())
self.assertEqual(
result.failure_description(),
"3 out of 7 row(s) did not contain matching strings. "
"The acceptable margin of error is only 0.0, but the "
"validation returned an error rate of 0.4286.",
)
def test_sameness_check_strings_multiple_dates(self) -> None:
self.mock_client.run_query_async.return_value = [
# January 2021
{"region": "US_XX", "date": "2021-01-31", "a": "00", "b": "00"},
{"region": "US_XX", "date": "2021-01-31", "a": "01", "b": None},
{"region": "US_XX", "date": "2021-01-31", "a": "02", "b": "02"},
{"region": "US_XX", "date": "2021-01-31", "a": None, "b": "03"},
# December 2020
{"region": "US_XX", "date": "2020-12-31", "a": "00", "b": "00"},
{"region": "US_XX", "date": "2020-12-31", "a": "02", "b": "02"},
{"region": "US_XX", "date": "2020-12-31", "a": None, "b": "04"},
]
job = DataValidationJob(
region_code="US_XX",
validation=SamenessDataValidationCheck(
validation_category=ValidationCategory.EXTERNAL_INDIVIDUAL,
validation_type=ValidationCheckType.SAMENESS,
comparison_columns=["a", "b"],
partition_columns=["region", "date"],
sameness_check_type=SamenessDataValidationCheckType.STRINGS,
max_allowed_error=0.0,
view_builder=SimpleBigQueryViewBuilder(
dataset_id="my_dataset",
view_id="test_view",
description="test_view description",
view_query_template="select * from literally_anything",
),
),
)
result = SamenessValidationChecker.run_check(job)
self.assertEqual(
result,
DataValidationJobResult(
validation_job=job,
result_details=SamenessStringsValidationResultDetails(
num_error_rows=3,
total_num_rows=7,
max_allowed_error=0.0,
non_null_counts_per_column_per_partition=[
(("US_XX", "2021-01-31"), {"a": 3, "b": 3}),
(("US_XX", "2020-12-31"), {"a": 2, "b": 3}),
],
),
),
)
def test_from_successful_result_strings(self) -> None:
# Arrange
result_details = SamenessStringsValidationResultDetails(
num_error_rows=0,
total_num_rows=5,
max_allowed_error=0.5,
non_null_counts_per_column_per_partition=[
(("US_XX", "2020-12-01"), {"internal": 5, "external": 5})
],
)
job_result = DataValidationJobResult(
validation_job=DataValidationJob(
region_code="US_XX",
validation=SamenessDataValidationCheck(
validation_category=ValidationCategory.EXTERNAL_INDIVIDUAL,
validation_type=ValidationCheckType.SAMENESS,
comparison_columns=["internal", "external"],
partition_columns=["state_code", "date"],
sameness_check_type=SamenessDataValidationCheckType.STRINGS,
view_builder=SimpleBigQueryViewBuilder(
dataset_id="my_dataset",
view_id="test_view",
description="test_view description",
view_query_template="select * from literally_anything",
),
),
),
result_details=result_details,
)
# Act
result = ValidationResultForStorage.from_validation_result(
run_id="abc123",
run_datetime=datetime.datetime(2000, 1, 1, 0, 0, 0),
result=job_result,
)
# Assert
self.assertEqual(
ValidationResultForStorage(
run_id="abc123",
run_date=datetime.date(2000, 1, 1),
run_datetime=datetime.datetime(2000, 1, 1, 0, 0, 0),
system_version="v1.0.0",
check_type=ValidationCheckType.SAMENESS,
validation_name="test_view",
region_code="US_XX",
did_run=True,
was_successful=True,
failure_description=None,
result_details_type="SamenessStringsValidationResultDetails",
result_details=result_details,
validation_category=ValidationCategory.EXTERNAL_INDIVIDUAL,
),
result,
)
self.assertEqual(
{
"run_id": "abc123",
"run_date": "2000-01-01",
"run_datetime": "2000-01-01T00:00:00",
"system_version": "v1.0.0",
"check_type": "SAMENESS",
"validation_name": "test_view",
"region_code": "US_XX",
"did_run": True,
"was_successful": True,
"failure_description": None,
"result_details_type": "SamenessStringsValidationResultDetails",
"result_details": '{"num_error_rows": 0, "total_num_rows": 5, "max_allowed_error": 0.5, "non_null_counts_per_column_per_partition": [[["US_XX", "2020-12-01"], {"internal": 5, "external": 5}]]}',
"validation_category": "EXTERNAL_INDIVIDUAL",
},
result.to_serializable(),
)