def create_direct_ingest_process_job_task(
self,
region: Region,
ingest_instance: DirectIngestInstance,
ingest_args: GcsfsIngestArgs,
) -> None:
task_id = _build_task_id(
region.region_code,
ingest_instance,
ingest_args.task_id_tag(),
prefix_only=False,
)
params = {
"region": region.region_code.lower(),
"file_path": ingest_args.file_path.abs_path(),
}
relative_uri = f"/direct/process_job?{urlencode(params)}"
body = self._get_body_from_args(ingest_args)
self._get_process_job_queue_manager(region,
ingest_instance).create_task(
task_id=task_id,
relative_uri=relative_uri,
body=body,
)
def test_is_task_queued_has_tasks(self):
# Arrange
file_path = to_normalized_unprocessed_file_path(
'bucket/file_path.csv', GcsfsDirectIngestFileType.INGEST_VIEW)
gcsfs_args = \
GcsfsIngestArgs(
ingest_time=datetime.datetime.now(),
file_path=GcsfsFilePath.from_absolute_path(file_path))
full_task_name = \
_build_task_id(_REGION.region_code, gcsfs_args.task_id_tag())
info = ProcessIngestJobCloudTaskQueueInfo(
queue_name='queue_name',
task_names=[
'projects/path/to/random_task',
f'projects/path/to/{full_task_name}'
])
file_path = to_normalized_unprocessed_file_path(
'bucket/file_path.csv', GcsfsDirectIngestFileType.INGEST_VIEW)
gcsfs_args = \
GcsfsIngestArgs(
ingest_time=datetime.datetime.now(),
file_path=GcsfsFilePath.from_absolute_path(file_path))
# Act
gcsfs_args_queued = info.is_task_queued(_REGION, gcsfs_args)
# Assert
self.assertTrue(gcsfs_args_queued)
def test_info_single_task(self) -> None:
# Arrange
gcsfs_args = GcsfsIngestArgs(
ingest_time=datetime.datetime.now(),
file_path=self.ingest_view_file_path,
)
full_task_name = _build_task_id(_REGION.region_code,
DirectIngestInstance.PRIMARY,
gcsfs_args.task_id_tag())
info = ProcessIngestJobCloudTaskQueueInfo(
queue_name="queue_name",
task_names=[
"projects/path/to/random_task",
f"projects/path/to/{full_task_name}",
],
)
gcsfs_args = GcsfsIngestArgs(
ingest_time=datetime.datetime.now(),
file_path=self.ingest_view_file_path,
)
# Act
gcsfs_args_queued = info.is_task_queued(_REGION, gcsfs_args)
# Assert
self.assertTrue(gcsfs_args_queued)
self.assertTrue(
info.tasks_for_instance(_REGION.region_code,
DirectIngestInstance.PRIMARY))
self.assertFalse(
info.tasks_for_instance(_REGION.region_code,
DirectIngestInstance.SECONDARY))
def is_task_queued(self, region: Region,
ingest_args: GcsfsIngestArgs) -> bool:
"""Returns true if the ingest_args correspond to a task currently in
the queue.
"""
task_id_prefix = _build_task_id(
region.region_code,
DirectIngestInstance.for_ingest_bucket(
ingest_args.file_path.bucket_path),
ingest_args.task_id_tag(),
prefix_only=True,
)
return bool(next(self._tasks_for_prefix(task_id_prefix), None))
def create_direct_ingest_process_job_task(
self,
region: Region,
ingest_instance: DirectIngestInstance,
ingest_args: GcsfsIngestArgs,
) -> 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_instance,
ingest_args.task_id_tag(),
)
self.process_job_tasks.append(
(f"projects/path/to/{task_id}", ingest_args))
def test_is_task_queued_has_tasks(self):
# Arrange
file_path = to_normalized_unprocessed_file_path('file_path.csv')
gcsfs_args = GcsfsIngestArgs(ingest_time=datetime.datetime.now(),
file_path=file_path)
full_task_name = \
_build_task_id(_REGION.region_code, gcsfs_args.task_id_tag())
info = CloudTaskQueueInfo(queue_name='queue_name',
task_names=[
f'projects/path/to/random_task',
f'projects/path/to/{full_task_name}'
])
file_path = to_normalized_unprocessed_file_path('file_path.csv')
gcsfs_args = GcsfsIngestArgs(ingest_time=datetime.datetime.now(),
file_path=file_path)
# Act
gcsfs_args_queued = info.is_task_queued(_REGION, gcsfs_args)
# Assert
self.assertTrue(gcsfs_args_queued)
def test_info_tasks_both_instances(self) -> None:
# Arrange
gcsfs_args = GcsfsIngestArgs(
ingest_time=datetime.datetime.now(),
file_path=self.ingest_view_file_path,
)
full_task_names = [
_build_task_id(
_REGION.region_code,
ingest_instance,
gcsfs_args.task_id_tag(),
) for ingest_instance in DirectIngestInstance
]
info = ProcessIngestJobCloudTaskQueueInfo(
queue_name="queue_name",
task_names=[
"projects/path/to/random_task",
] + [
f"projects/path/to/{full_task_name}"
for full_task_name in full_task_names
],
)
gcsfs_args = GcsfsIngestArgs(
ingest_time=datetime.datetime.now(),
file_path=self.ingest_view_file_path,
)
# Act
gcsfs_args_queued = info.is_task_queued(_REGION, gcsfs_args)
# Assert
self.assertTrue(gcsfs_args_queued)
for ingest_instance in DirectIngestInstance:
self.assertTrue(
info.tasks_for_instance(_REGION.region_code, ingest_instance))
