def testDo(self):
source_data_dir = os.path.join(
os.path.dirname(os.path.dirname(__file__)), 'testdata')
eval_stats_artifact = standard_artifacts.ExampleStatistics()
eval_stats_artifact.uri = os.path.join(source_data_dir,
'statistics_gen')
eval_stats_artifact.split_names = artifact_utils.encode_split_names(
['train', 'eval', 'test'])
schema_artifact = standard_artifacts.Schema()
schema_artifact.uri = os.path.join(source_data_dir, 'schema_gen')
output_data_dir = os.path.join(
os.environ.get('TEST_UNDECLARED_OUTPUTS_DIR', self.get_temp_dir()),
self._testMethodName)
validation_output = standard_artifacts.ExampleAnomalies()
validation_output.uri = os.path.join(output_data_dir, 'output')
input_dict = {
STATISTICS_KEY: [eval_stats_artifact],
SCHEMA_KEY: [schema_artifact],
}
exec_properties = {
# List needs to be serialized before being passed into Do function.
EXCLUDE_SPLITS_KEY: json_utils.dumps(['test'])
}
output_dict = {
ANOMALIES_KEY: [validation_output],
}
example_validator_executor = executor.Executor()
example_validator_executor.Do(input_dict, output_dict, exec_properties)
self.assertEqual(artifact_utils.encode_split_names(['train', 'eval']),
validation_output.split_names)
# Check example_validator outputs.
train_anomalies_path = os.path.join(validation_output.uri,
'Split-train', 'SchemaDiff.pb')
eval_anomalies_path = os.path.join(validation_output.uri, 'Split-eval',
'SchemaDiff.pb')
self.assertTrue(fileio.exists(train_anomalies_path))
self.assertTrue(fileio.exists(eval_anomalies_path))
train_anomalies_bytes = io_utils.read_bytes_file(train_anomalies_path)
train_anomalies = anomalies_pb2.Anomalies()
train_anomalies.ParseFromString(train_anomalies_bytes)
eval_anomalies_bytes = io_utils.read_bytes_file(eval_anomalies_path)
eval_anomalies = anomalies_pb2.Anomalies()
eval_anomalies.ParseFromString(eval_anomalies_bytes)
self.assertEqual(0, len(train_anomalies.anomaly_info))
self.assertEqual(0, len(eval_anomalies.anomaly_info))
# Assert 'test' split is excluded.
train_file_path = os.path.join(validation_output.uri, 'Split-test',
'SchemaDiff.pb')
self.assertFalse(fileio.exists(train_file_path))
def testDo(self):
source_data_dir = os.path.join(
os.path.dirname(os.path.dirname(__file__)), 'testdata')
eval_stats_artifact = standard_artifacts.ExampleStatistics()
eval_stats_artifact.uri = os.path.join(source_data_dir,
'statistics_gen')
eval_stats_artifact.split_names = artifact_utils.encode_split_names(
['eval'])
schema_artifact = standard_artifacts.Schema()
schema_artifact.uri = os.path.join(source_data_dir, 'schema_gen')
output_data_dir = os.path.join(
os.environ.get('TEST_UNDECLARED_OUTPUTS_DIR', self.get_temp_dir()),
self._testMethodName)
validation_output = standard_artifacts.ExampleAnomalies()
validation_output.uri = os.path.join(output_data_dir, 'output')
input_dict = {
executor.STATISTICS_KEY: [eval_stats_artifact],
executor.SCHEMA_KEY: [schema_artifact],
}
output_dict = {
executor.ANOMALIES_KEY: [validation_output],
}
exec_properties = {}
example_validator_executor = executor.Executor()
example_validator_executor.Do(input_dict, output_dict, exec_properties)
self.assertEqual(['anomalies.pbtxt'],
tf.io.gfile.listdir(validation_output.uri))
anomalies = io_utils.parse_pbtxt_file(
os.path.join(validation_output.uri, 'anomalies.pbtxt'),
anomalies_pb2.Anomalies())
self.assertNotEqual(0, len(anomalies.anomaly_info))
def test_do(self):
source_data_dir = os.path.join(
os.path.dirname(os.path.dirname(__file__)), 'testdata')
eval_stats_artifact = types.Artifact('ExampleStatsPath', split='eval')
eval_stats_artifact.uri = os.path.join(source_data_dir,
'statistics_gen/eval/')
schema_artifact = standard_artifacts.Schema()
schema_artifact.uri = os.path.join(source_data_dir, 'schema_gen/')
output_data_dir = os.path.join(
os.environ.get('TEST_UNDECLARED_OUTPUTS_DIR', self.get_temp_dir()),
self._testMethodName)
validation_output = standard_artifacts.ExampleValidationResult()
validation_output.uri = os.path.join(output_data_dir, 'output')
input_dict = {
'stats': [eval_stats_artifact],
'schema': [schema_artifact],
}
output_dict = {
'output': [validation_output],
}
exec_properties = {}
example_validator_executor = executor.Executor()
example_validator_executor.Do(input_dict, output_dict, exec_properties)
self.assertEqual(['anomalies.pbtxt'],
tf.gfile.ListDirectory(validation_output.uri))
anomalies = io_utils.parse_pbtxt_file(
os.path.join(validation_output.uri, 'anomalies.pbtxt'),
anomalies_pb2.Anomalies())
self.assertNotEqual(0, len(anomalies.anomaly_info))
