def test_execute(self, mock_extract_object_id, mock_create_dataset,
mock_xcom):
op = AutoMLCreateDatasetOperator(
dataset=DATASET,
location=GCP_LOCATION,
project_id=GCP_PROJECT_ID,
task_id=TASK_ID,
)
op.execute(context=None)
mock_create_dataset.assert_called_once_with(
dataset=DATASET,
location=GCP_LOCATION,
metadata=None,
project_id=GCP_PROJECT_ID,
retry=None,
timeout=None,
)
mock_xcom.assert_called_once_with(None,
key="dataset_id",
value=DATASET_ID)
IMPORT_INPUT_CONFIG = {"gcs_source": {"input_uris": [GCP_AUTOML_TEXT_BUCKET]}}
default_args = {"start_date": days_ago(1)}
extract_object_id = CloudAutoMLHook.extract_object_id
# Example DAG for AutoML Natural Language Entities Extraction
with models.DAG(
"example_automl_text",
default_args=default_args,
schedule_interval=None, # Override to match your needs
user_defined_macros={"extract_object_id": extract_object_id},
tags=['example'],
) as example_dag:
create_dataset_task = AutoMLCreateDatasetOperator(
task_id="create_dataset_task",
dataset=DATASET,
location=GCP_AUTOML_LOCATION)
dataset_id = (
'{{ task_instance.xcom_pull("create_dataset_task", key="dataset_id") }}'
)
import_dataset_task = AutoMLImportDataOperator(
task_id="import_dataset_task",
dataset_id=dataset_id,
location=GCP_AUTOML_LOCATION,
input_config=IMPORT_INPUT_CONFIG,
)
MODEL["dataset_id"] = dataset_id
# Example DAG to create dataset, train model_id and deploy it.
with models.DAG(
"example_create_and_deploy",
default_args=default_args,
schedule_interval=None, # Override to match your needs
user_defined_macros={
"get_target_column_spec": get_target_column_spec,
"target": TARGET,
"extract_object_id": extract_object_id,
},
) as create_deploy_dag:
# [START howto_operator_automl_create_dataset]
create_dataset_task = AutoMLCreateDatasetOperator(
task_id="create_dataset_task",
dataset=DATASET,
location=GCP_AUTOML_LOCATION,
project_id=GCP_PROJECT_ID,
)
dataset_id = (
"{{ task_instance.xcom_pull('create_dataset_task', key='dataset_id') }}"
)
# [END howto_operator_automl_create_dataset]
MODEL["dataset_id"] = dataset_id
# [START howto_operator_automl_import_data]
import_dataset_task = AutoMLImportDataOperator(
task_id="import_dataset_task",
dataset_id=dataset_id,
location=GCP_AUTOML_LOCATION,