def test_base_model_validator_ptransform(self):
with pipeline.TestPipeline(runner=direct_runner.DirectRunner()) as p:
invalid_id = MockModel(id='123@?!*',
deleted=False,
created_on=self.year_ago,
last_updated=self.now)
invalid_timestamp = MockModel(id='124',
deleted=False,
created_on=self.now,
last_updated=self.year_later)
expired_model = MockModel(id='125',
deleted=True,
created_on=self.year_ago,
last_updated=self.year_ago)
valid_model = MockModel(id='126',
deleted=False,
created_on=self.year_ago,
last_updated=self.now)
pcoll = (p
| beam.Create([
invalid_id, invalid_timestamp, expired_model,
valid_model
]))
output = pcoll | base_model_validator.BaseModelValidator()
beam_testing_util.assert_that(
output,
beam_testing_util.equal_to([
errors.ModelInvalidIdError(invalid_id),
errors.ModelMutatedDuringJobError(invalid_timestamp),
errors.ModelExpiredError(expired_model)
]))
def test_base_model_validator_ptransform(self):
model_with_invalid_id = base_models.BaseModel(id='123@?!*',
deleted=False,
created_on=self.YEAR_AGO,
last_updated=self.NOW)
model_with_invalid_timestamp = base_models.BaseModel(
id='124',
deleted=False,
created_on=self.NOW,
last_updated=self.YEAR_LATER)
expired_model = base_models.BaseModel(id='125',
deleted=True,
created_on=self.YEAR_AGO,
last_updated=self.YEAR_AGO)
valid_model = base_models.BaseModel(id='126',
deleted=False,
created_on=self.YEAR_AGO,
last_updated=self.NOW)
output = (self.pipeline
| beam.Create([
model_with_invalid_id,
model_with_invalid_timestamp,
expired_model,
valid_model,
])
| base_model_validator.BaseModelValidator())
self.assert_pcoll_equal(output, [
errors.InvalidIdError(
model_with_invalid_id,
base_model_validator.DEFAULT_ID_REGEX_STRING),
errors.ModelMutatedDuringJobError(model_with_invalid_timestamp),
errors.ModelExpiredError(expired_model),
])
def test_model_mutated_during_job_error(self):
model = MockModel(id='124',
created_on=self.now,
last_updated=self.year_later)
error = errors.ModelMutatedDuringJobError(model)
msg = ('Entity id %s: The last_updated field has a value %s which '
'is greater than the time when the job was run' %
(model.id, model.last_updated))
self.assertEqual(error.message, msg)
def test_message(self):
model = base_models.BaseModel(id='123',
created_on=self.NOW,
last_updated=self.YEAR_LATER)
error = errors.ModelMutatedDuringJobError(model)
self.assertEqual(
error.message,
'ModelMutatedDuringJobError in BaseModel(id="123"): '
'last_updated=%r is later than the validation job\'s start time' %
(model.last_updated))
def test_process_reports_model_mutated_during_job_error(self):
invalid_timestamp = base_models.BaseModel(id='124',
created_on=self.NOW,
last_updated=self.YEAR_LATER)
output = (self.pipeline
| beam.Create([invalid_timestamp])
| beam.ParDo(base_model_validator.ValidateModelTimestamps()))
self.assert_pcoll_equal(output, [
errors.ModelMutatedDuringJobError(invalid_timestamp),
])
def test_process_reports_model_mutated_during_job_error(self):
with pipeline.TestPipeline(runner=direct_runner.DirectRunner()) as p:
invalid_timestamp = MockModel(id='124',
created_on=self.now,
last_updated=self.year_later)
pcoll = p | beam.Create([invalid_timestamp])
output = (pcoll
| beam.ParDo(
base_model_validator.ValidateModelTimeFields()))
beam_testing_util.assert_that(
output,
beam_testing_util.equal_to(
[errors.ModelMutatedDuringJobError(invalid_timestamp)]))
def process(self, input_model):
"""Function that defines how to process each element in a pipeline of
models.
Args:
input_model: datastore_services.Model. Entity to validate.
Yields:
ModelMutatedDuringJobError. Error for timestamp validation.
ModelTimestampRelationshipError. Error for timestamp validation.
"""
model = jobs_utils.clone_model(input_model)
if model.created_on > (model.last_updated + MAX_CLOCK_SKEW_SECS):
yield errors.ModelTimestampRelationshipError(model)
current_datetime = datetime.datetime.utcnow()
if (model.last_updated - MAX_CLOCK_SKEW_SECS) > current_datetime:
yield errors.ModelMutatedDuringJobError(model)