def test_process_job_unlaunched_region(self, mock_supported, mock_region,
mock_environment):
mock_supported.return_value = ['us_ca', 'us_pa']
region_code = 'us_ca'
mock_environment.return_value = 'production'
mock_controller = create_autospec(GcsfsDirectIngestController)
mock_region.return_value = fake_region(region_code=region_code,
environment='staging',
ingestor=mock_controller)
ingest_args = IngestArgs(datetime.datetime(year=2019, month=7, day=20))
request_args = {
'region': region_code,
}
body = {
'cloud_task_args': ingest_args.to_serializable(),
'args_type': 'IngestArgs',
}
body_encoded = json.dumps(body).encode()
headers = {'X-Appengine-Cron': 'test-cron'}
response = self.client.post('/process_job',
query_string=request_args,
headers=headers,
data=body_encoded)
self.assertEqual(400, response.status_code)
self.assertEqual(response.get_data().decode(),
"Bad environment [production] for region [us_ca].")
def test_process_job(self, mock_supported, mock_region, mock_environment):
mock_supported.return_value = ['us_nd', 'us_pa']
region_code = 'us_nd'
mock_environment.return_value = 'staging'
mock_controller = create_autospec(GcsfsDirectIngestController)
mock_region.return_value = fake_region(region_code=region_code,
environment='staging',
ingestor=mock_controller)
ingest_args = IngestArgs(datetime.datetime(year=2019, month=7, day=20))
request_args = {
'region': region_code,
}
body = {
'cloud_task_args': ingest_args.to_serializable(),
'args_type': 'IngestArgs',
}
body_encoded = json.dumps(body).encode()
headers = {'X-Appengine-Cron': 'test-cron'}
response = self.client.post('/process_job',
query_string=request_args,
headers=headers,
data=body_encoded)
self.assertEqual(200, response.status_code)
mock_controller.run_ingest_job_and_kick_scheduler_on_completion.assert_called_with(
ingest_args)
def json_to_ingest_args(json_data):
if 'ingest_args' in json_data and 'args_type' in json_data:
args_type = json_data['args_type']
ingest_args = json_data['ingest_args']
if args_type == IngestArgs.__name__:
return IngestArgs.from_serializable(ingest_args)
if args_type == GcsfsIngestArgs.__name__:
return GcsfsIngestArgs.from_serializable(ingest_args)
logging.error('Unexpected args_type in json_data: %s', args_type)
return None
def create_direct_ingest_process_job_task(
self, region: Region, ingest_args: IngestArgs
) -> None:
"""Queues *but does not run* a process job task."""
if not self.controller:
raise ValueError("Controller is null - did you call set_controller()?")
task_id = _build_task_id(
self.controller.region.region_code, ingest_args.task_id_tag()
)
self.process_job_tasks.append((f"projects/path/to/{task_id}", ingest_args))
def test_create_direct_ingest_process_job_task(
self, mock_client: mock.MagicMock,
mock_uuid: mock.MagicMock) -> None:
# Arrange
ingest_args = IngestArgs(datetime.datetime(year=2019, month=7, day=20))
body = {
"cloud_task_args": ingest_args.to_serializable(),
"args_type": "IngestArgs",
}
body_encoded = json.dumps(body).encode()
uuid = "random-uuid"
mock_uuid.uuid4.return_value = uuid
date = "2019-07-20"
queue_path = f"{_REGION.shared_queue}-path"
task_name = "{}/{}-{}-{}".format(_REGION.shared_queue,
_REGION.region_code, date, uuid)
task = tasks_v2.types.task_pb2.Task(
name=task_name,
app_engine_http_request={
"http_method": "POST",
"relative_uri":
f"/direct/process_job?region={_REGION.region_code}",
"body": body_encoded,
},
)
mock_client.return_value.task_path.return_value = task_name
mock_client.return_value.queue_path.return_value = queue_path
# Act
DirectIngestCloudTaskManagerImpl(
).create_direct_ingest_process_job_task(_REGION, ingest_args)
# Assert
mock_client.return_value.queue_path.assert_called_with(
self.mock_project_id, QUEUES_REGION, _REGION.shared_queue)
mock_client.return_value.create_task.assert_called_with(
parent=queue_path, task=task)
def create_direct_ingest_process_job_task(self, region: Region,
ingest_args: IngestArgs) -> None:
task_id = _build_task_id(region.region_code,
ingest_args.task_id_tag(),
prefix_only=False)
relative_uri = f"/direct/process_job?region={region.region_code}"
body = self._get_body_from_args(ingest_args)
self._get_process_job_queue_manager(region).create_task(
task_id=task_id,
relative_uri=relative_uri,
body=body,
)
def test_process_job_unlaunched_region(
self,
mock_supported: mock.MagicMock,
mock_region: mock.MagicMock,
mock_environment: mock.MagicMock,
) -> None:
mock_supported.return_value = ["us_ca", "us_pa"]
region_code = "us_ca"
mock_environment.return_value = "production"
mock_controller = create_autospec(GcsfsDirectIngestController)
mock_region.return_value = fake_region(region_code=region_code,
environment="staging",
ingestor=mock_controller)
ingest_args = IngestArgs(datetime.datetime(year=2019, month=7, day=20))
request_args = {
"region": region_code,
}
body = {
"cloud_task_args": ingest_args.to_serializable(),
"args_type": "IngestArgs",
}
body_encoded = json.dumps(body).encode()
headers = {"X-Appengine-Cron": "test-cron"}
response = self.client.post(
"/process_job",
query_string=request_args,
headers=headers,
data=body_encoded,
)
self.assertEqual(400, response.status_code)
self.assertEqual(
response.get_data().decode(),
"Bad environment [production] for region [us_ca].",
)
def test_create_direct_ingest_process_job_task(self, mock_client,
mock_uuid):
# Arrange
project_id = 'recidiviz-456'
ingest_args = IngestArgs(datetime.datetime(year=2019, month=7, day=20))
body = {
'cloud_task_args': ingest_args.to_serializable(),
'args_type': 'IngestArgs'
}
body_encoded = json.dumps(body).encode()
uuid = 'random-uuid'
mock_uuid.uuid4.return_value = uuid
date = '2019-07-20'
queue_path = _REGION.shared_queue + '-path'
task_name = _REGION.shared_queue + '/{}-{}-{}'.format(
_REGION.region_code, date, uuid)
task = tasks_v2.types.task_pb2.Task(
name=task_name,
app_engine_http_request={
'http_method': 'POST',
'relative_uri':
f'/direct/process_job?region={_REGION.region_code}',
'body': body_encoded
})
mock_client.return_value.task_path.return_value = task_name
mock_client.return_value.queue_path.return_value = queue_path
# Act
DirectIngestCloudTaskManagerImpl(project_id=project_id).\
create_direct_ingest_process_job_task(_REGION, ingest_args)
# Assert
mock_client.return_value.queue_path.assert_called_with(
project_id, QUEUES_REGION, _REGION.shared_queue)
mock_client.return_value.create_task.assert_called_with(
parent=queue_path, task=task)
def create_direct_ingest_process_job_task(self, region: Region,
ingest_args: IngestArgs):
task_id = _build_task_id(region.region_code,
ingest_args.task_id_tag(),
prefix_only=False)
relative_uri = f'/direct/process_job?region={region.region_code}'
body = self._get_body_from_args(ingest_args)
self.cloud_task_client.create_task(
task_id=task_id,
queue_name=region.get_queue_name(),
relative_uri=relative_uri,
body=body,
)
def is_task_queued(self, region: Region, ingest_args: IngestArgs) -> bool:
"""Returns true if the ingest_args correspond to a task currently in
the queue.
"""
task_id_prefix = _build_task_id(region.region_code,
ingest_args.task_id_tag(),
prefix_only=True)
for task_name in self.task_names:
_, task_id = os.path.split(task_name)
if task_id.startswith(task_id_prefix):
return True
return False
def test_process_job(
self,
mock_supported: mock.MagicMock,
mock_region: mock.MagicMock,
mock_environment: mock.MagicMock,
) -> None:
mock_supported.return_value = ["us_nd", "us_pa"]
region_code = "us_nd"
mock_environment.return_value = "staging"
mock_controller = create_autospec(GcsfsDirectIngestController)
mock_region.return_value = fake_region(region_code=region_code,
environment="staging",
ingestor=mock_controller)
ingest_args = IngestArgs(datetime.datetime(year=2019, month=7, day=20))
request_args = {
"region": region_code,
}
body = {
"cloud_task_args": ingest_args.to_serializable(),
"args_type": "IngestArgs",
}
body_encoded = json.dumps(body).encode()
headers = {"X-Appengine-Cron": "test-cron"}
response = self.client.post(
"/process_job",
query_string=request_args,
headers=headers,
data=body_encoded,
)
self.assertEqual(200, response.status_code)
mock_controller.run_ingest_job_and_kick_scheduler_on_completion.assert_called_with(
ingest_args)
def _get_next_job_args(self) -> Optional[IngestArgs]:
df = pd.read_sql_query('SELECT MIN(export_time) FROM booking',
self._create_engine())
ingest_time = df[min][0]
if not ingest_time:
logging.info("No more export times - successfully persisted all "
"data exports.")
return None
if ingest_time in self.scheduled_ingest_times:
raise DirectIngestError(
msg=f"Received a second job for ingest time [{ingest_time}]. "
"Did the previous job delete this export from the database?",
error_type=DirectIngestErrorType.CLEANUP_ERROR)
return IngestArgs(ingest_time=ingest_time)
def create_direct_ingest_process_job_task(self, region: Region,
ingest_args: IngestArgs):
body = self._get_body_from_args(ingest_args)
task_name = self._build_task_name_for_queue_and_region(
region.get_queue_name(), region.region_code,
ingest_args.task_id_tag())
task = tasks.types.Task(
name=task_name,
app_engine_http_request={
'relative_uri':
f'/direct/process_job?region={region.region_code}',
'body': json.dumps(body).encode()
})
self._queue_task(region.get_queue_name(), task)
def json_to_cloud_task_args(json_data: dict):
if 'cloud_task_args' in json_data and 'args_type' in json_data:
args_type = json_data['args_type']
cloud_task_args_dict = json_data['cloud_task_args']
if args_type == IngestArgs.__name__:
return IngestArgs.from_serializable(cloud_task_args_dict)
if args_type == GcsfsIngestArgs.__name__:
return GcsfsIngestArgs.from_serializable(cloud_task_args_dict)
if args_type == GcsfsRawDataBQImportArgs.__name__:
return GcsfsRawDataBQImportArgs.from_serializable(
cloud_task_args_dict)
if args_type == GcsfsIngestViewExportArgs.__name__:
return GcsfsIngestViewExportArgs.from_serializable(
cloud_task_args_dict)
logging.error('Unexpected args_type in json_data: %s', args_type)
return None
def json_to_cloud_task_args(json_data: dict) -> Optional[CloudTaskArgs]:
if "cloud_task_args" in json_data and "args_type" in json_data:
args_type = json_data["args_type"]
cloud_task_args_dict = json_data["cloud_task_args"]
if args_type == IngestArgs.__name__:
return IngestArgs.from_serializable(cloud_task_args_dict)
if args_type == GcsfsIngestArgs.__name__:
return GcsfsIngestArgs.from_serializable(cloud_task_args_dict)
if args_type == GcsfsRawDataBQImportArgs.__name__:
return GcsfsRawDataBQImportArgs.from_serializable(
cloud_task_args_dict)
if args_type == GcsfsIngestViewExportArgs.__name__:
return GcsfsIngestViewExportArgs.from_serializable(
cloud_task_args_dict)
logging.error("Unexpected args_type in json_data: %s", args_type)
return None
def test_is_task_queued_no_tasks(self):
# Arrange
info = CloudTaskQueueInfo(queue_name='queue_name', task_names=[])
file_path = to_normalized_unprocessed_file_path('file_path.csv')
args = IngestArgs(ingest_time=datetime.datetime.now())
gcsfs_args = GcsfsIngestArgs(ingest_time=datetime.datetime.now(),
file_path=file_path)
# Act
basic_args_queued = info.is_task_queued(_REGION, args)
gcsfs_args_queued = info.is_task_queued(_REGION, gcsfs_args)
# Assert
self.assertFalse(basic_args_queued)
self.assertFalse(gcsfs_args_queued)
self.assertFalse(info.is_task_queued(_REGION, gcsfs_args))
def test_is_task_queued_no_tasks(self):
# Arrange
info = ProcessIngestJobCloudTaskQueueInfo(queue_name='queue_name',
task_names=[])
file_path = to_normalized_unprocessed_file_path(
'bucket/file_path.csv', GcsfsDirectIngestFileType.INGEST_VIEW)
args = IngestArgs(ingest_time=datetime.datetime.now())
gcsfs_args = \
GcsfsIngestArgs(
ingest_time=datetime.datetime.now(),
file_path=GcsfsFilePath.from_absolute_path(file_path))
# Act
basic_args_queued = info.is_task_queued(_REGION, args)
gcsfs_args_queued = info.is_task_queued(_REGION, gcsfs_args)
# Assert
self.assertFalse(basic_args_queued)
self.assertFalse(gcsfs_args_queued)
self.assertFalse(info.is_task_queued(_REGION, gcsfs_args))
def _get_body_from_args(ingest_args: IngestArgs) -> Dict:
body = {
'ingest_args': ingest_args.to_serializable(),
'args_type': ingest_args.__class__.__name__
}
return body