Source code for tests.system.providers.google.cloud.dataplex.example_dataplex

# Licensed to the Apache Software Foundation (ASF) under one
# or more contributor license agreements.  See the NOTICE file
# distributed with this work for additional information
# regarding copyright ownership.  The ASF licenses this file
# to you under the Apache License, Version 2.0 (the
# "License"); you may not use this file except in compliance
# with the License.  You may obtain a copy of the License at
#
#   http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing,
# software distributed under the License is distributed on an
# "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
# KIND, either express or implied.  See the License for the
# specific language governing permissions and limitations
# under the License.
"""
Example Airflow DAG that shows how to use Dataplex.
"""
from __future__ import annotations

import datetime
import os

from airflow.models.baseoperator import chain
from airflow.models.dag import DAG
from airflow.providers.google.cloud.operators.dataplex import (
    DataplexCreateLakeOperator,
    DataplexCreateTaskOperator,
    DataplexDeleteLakeOperator,
    DataplexDeleteTaskOperator,
    DataplexGetTaskOperator,
    DataplexListTasksOperator,
)
from airflow.providers.google.cloud.operators.gcs import (
    GCSCreateBucketOperator,
    GCSDeleteBucketOperator,
    GCSSynchronizeBucketsOperator,
)
from airflow.providers.google.cloud.sensors.dataplex import DataplexTaskStateSensor
from airflow.utils.trigger_rule import TriggerRule

[docs]ENV_ID = os.environ.get("SYSTEM_TESTS_ENV_ID")
[docs]PROJECT_ID = os.environ.get("SYSTEM_TESTS_GCP_PROJECT", "project_id")
[docs]DAG_ID = "example_dataplex"
[docs]BUCKET_NAME = f"bucket_{DAG_ID}_{ENV_ID}"
[docs]SPARK_FILE_NAME = "spark_example_pi.py"
[docs]RESOURCE_DATA_BUCKET = "airflow-system-tests-resources"
[docs]LAKE_ID = f"test-lake-dataplex-{ENV_ID}"
[docs]REGION = "us-central1"
[docs]SERVICE_ACC = f"{PROJECT_ID}@appspot.gserviceaccount.com"
[docs]SPARK_FILE_FULL_PATH = f"gs://{BUCKET_NAME}/{SPARK_FILE_NAME}"
[docs]DATAPLEX_TASK_ID = f"test-task-{ENV_ID}"
[docs]TRIGGER_SPEC_TYPE = "ON_DEMAND"
# [START howto_dataplex_configuration]
[docs]EXAMPLE_TASK_BODY = { "trigger_spec": {"type_": TRIGGER_SPEC_TYPE}, "execution_spec": {"service_account": SERVICE_ACC}, "spark": {"python_script_file": SPARK_FILE_FULL_PATH}, }
# [END howto_dataplex_configuration] # [START howto_dataplex_lake_configuration]
[docs]EXAMPLE_LAKE_BODY = { "display_name": "test_display_name", "labels": [], "description": "test_description", "metastore": {"service": ""}, }
# [END howto_dataplex_lake_configuration] with DAG( DAG_ID, start_date=datetime.datetime(2021, 1, 1), schedule="@once", tags=["example", "dataplex"], ) as dag:
[docs] create_bucket = GCSCreateBucketOperator( task_id="create_bucket", bucket_name=BUCKET_NAME, project_id=PROJECT_ID )
sync_bucket = GCSSynchronizeBucketsOperator( task_id="sync_bucket", source_bucket=RESOURCE_DATA_BUCKET, source_object=SPARK_FILE_NAME, destination_bucket=BUCKET_NAME, destination_object=SPARK_FILE_NAME, recursive=True, ) # [START howto_dataplex_create_lake_operator] create_lake = DataplexCreateLakeOperator( project_id=PROJECT_ID, region=REGION, body=EXAMPLE_LAKE_BODY, lake_id=LAKE_ID, task_id="create_lake" ) # [END howto_dataplex_create_lake_operator] # [START howto_dataplex_create_task_operator] create_dataplex_task = DataplexCreateTaskOperator( project_id=PROJECT_ID, region=REGION, lake_id=LAKE_ID, body=EXAMPLE_TASK_BODY, dataplex_task_id=DATAPLEX_TASK_ID, task_id="create_dataplex_task", ) # [END howto_dataplex_create_task_operator] # [START howto_dataplex_async_create_task_operator] create_dataplex_task_async = DataplexCreateTaskOperator( project_id=PROJECT_ID, region=REGION, lake_id=LAKE_ID, body=EXAMPLE_TASK_BODY, dataplex_task_id=f"{DATAPLEX_TASK_ID}-1", asynchronous=True, task_id="create_dataplex_task_async", ) # [END howto_dataplex_async_create_task_operator] # [START howto_dataplex_delete_task_operator] delete_dataplex_task_async = DataplexDeleteTaskOperator( project_id=PROJECT_ID, region=REGION, lake_id=LAKE_ID, dataplex_task_id=f"{DATAPLEX_TASK_ID}-1", task_id="delete_dataplex_task_async", ) # [END howto_dataplex_delete_task_operator] # [START howto_dataplex_list_tasks_operator] list_dataplex_task = DataplexListTasksOperator( project_id=PROJECT_ID, region=REGION, lake_id=LAKE_ID, task_id="list_dataplex_task" ) # [END howto_dataplex_list_tasks_operator] # [START howto_dataplex_get_task_operator] get_dataplex_task = DataplexGetTaskOperator( project_id=PROJECT_ID, region=REGION, lake_id=LAKE_ID, dataplex_task_id=DATAPLEX_TASK_ID, task_id="get_dataplex_task", ) # [END howto_dataplex_get_task_operator] # [START howto_dataplex_task_state_sensor] dataplex_task_state = DataplexTaskStateSensor( project_id=PROJECT_ID, region=REGION, lake_id=LAKE_ID, dataplex_task_id=DATAPLEX_TASK_ID, task_id="dataplex_task_state", ) # [END howto_dataplex_task_state_sensor] # [START howto_dataplex_delete_task_operator] delete_dataplex_task = DataplexDeleteTaskOperator( project_id=PROJECT_ID, region=REGION, lake_id=LAKE_ID, dataplex_task_id=DATAPLEX_TASK_ID, task_id="delete_dataplex_task", trigger_rule=TriggerRule.ALL_DONE, ) # [END howto_dataplex_delete_task_operator] # [START howto_dataplex_delete_lake_operator] delete_lake = DataplexDeleteLakeOperator( project_id=PROJECT_ID, region=REGION, lake_id=LAKE_ID, task_id="delete_lake", trigger_rule=TriggerRule.ALL_DONE, ) # [END howto_dataplex_delete_lake_operator] delete_bucket = GCSDeleteBucketOperator( task_id="delete_bucket", bucket_name=BUCKET_NAME, trigger_rule=TriggerRule.ALL_DONE ) chain( # TEST SETUP create_bucket, sync_bucket, # TEST BODY create_lake, create_dataplex_task, get_dataplex_task, list_dataplex_task, create_dataplex_task_async, delete_dataplex_task_async, dataplex_task_state, # TEST TEARDOWN delete_dataplex_task, delete_lake, delete_bucket, ) from tests.system.utils.watcher import watcher # This test needs watcher in order to properly mark success/failure # when "tearDown" task with trigger rule is part of the DAG list(dag.tasks) >> watcher() from tests.system.utils import get_test_run # noqa: E402 # Needed to run the example DAG with pytest (see: tests/system/README.md#run_via_pytest)
[docs]test_run = get_test_run(dag)

Was this entry helpful?