def test_execute_with_xcom_exceeded_max_xcom_size(
self, mock_sys_getsizeof, mock_json_dumps,
mock_s3_hook_load_string, mock_google_api_hook_query):
context = {'task_instance': Mock()}
xcom_kwargs = {
'google_api_response_via_xcom': 'response',
'google_api_endpoint_params_via_xcom': 'params',
'google_api_endpoint_params_via_xcom_task_ids': 'params',
}
context['task_instance'].xcom_pull.return_value = {}
self.assertRaises(
RuntimeError,
GoogleApiToS3TransferOperator(**self.kwargs,
**xcom_kwargs).execute, context)
mock_google_api_hook_query.assert_called_once_with(
endpoint=self.kwargs['google_api_endpoint_path'],
data=self.kwargs['google_api_endpoint_params'],
paginate=self.kwargs['google_api_pagination'],
num_retries=self.kwargs['google_api_num_retries'])
mock_json_dumps.assert_called_once_with(
mock_google_api_hook_query.return_value)
mock_s3_hook_load_string.assert_called_once_with(
string_data=mock_json_dumps.return_value,
key=self.kwargs['s3_destination_key'],
replace=self.kwargs['s3_overwrite'])
context['task_instance'].xcom_pull.assert_called_once_with(
task_ids=xcom_kwargs[
'google_api_endpoint_params_via_xcom_task_ids'],
key=xcom_kwargs['google_api_endpoint_params_via_xcom'])
context['task_instance'].xcom_push.assert_not_called()
mock_sys_getsizeof.assert_called_once_with(
mock_google_api_hook_query.return_value)
def test_execute(self, mock_json_dumps, mock_s3_hook_load_string,
mock_google_api_hook_query):
context = {'task_instance': Mock()}
GoogleApiToS3TransferOperator(**self.kwargs).execute(context)
mock_google_api_hook_query.assert_called_once_with(
endpoint=self.kwargs['google_api_endpoint_path'],
data=self.kwargs['google_api_endpoint_params'],
paginate=self.kwargs['google_api_pagination'],
num_retries=self.kwargs['google_api_num_retries'])
mock_json_dumps.assert_called_once_with(
mock_google_api_hook_query.return_value)
mock_s3_hook_load_string.assert_called_once_with(
string_data=mock_json_dumps.return_value,
key=self.kwargs['s3_destination_key'],
replace=self.kwargs['s3_overwrite'])
context['task_instance'].xcom_pull.assert_not_called()
context['task_instance'].xcom_push.assert_not_called()
from airflow import DAG
from airflow.providers.amazon.aws.operators.google_api_to_s3_transfer import GoogleApiToS3TransferOperator
from airflow.utils.dates import days_ago
# [START howto_operator_google_api_to_s3_transfer_basic_env_variables]
GOOGLE_SHEET_ID = getenv("GOOGLE_SHEET_ID")
GOOGLE_SHEET_RANGE = getenv("GOOGLE_SHEET_RANGE")
S3_DESTINATION_KEY = getenv("S3_DESTINATION_KEY", "s3://bucket/key.json")
# [END howto_operator_google_api_to_s3_transfer_basic_env_variables]
default_args = {"start_date": days_ago(1)}
with DAG(dag_id="example_google_api_to_s3_transfer_basic",
default_args=default_args,
schedule_interval=None,
tags=['example']) as dag:
# [START howto_operator_google_api_to_s3_transfer_basic_task_1]
task_google_sheets_values_to_s3 = GoogleApiToS3TransferOperator(
google_api_service_name='sheets',
google_api_service_version='v4',
google_api_endpoint_path='sheets.spreadsheets.values.get',
google_api_endpoint_params={
'spreadsheetId': GOOGLE_SHEET_ID,
'range': GOOGLE_SHEET_RANGE
},
s3_destination_key=S3_DESTINATION_KEY,
task_id='google_sheets_values_to_s3',
dag=dag)
# [END howto_operator_google_api_to_s3_transfer_basic_task_1]
s3_file_name, _ = s3_file.rsplit('.', 1)
with DAG(dag_id="example_google_api_to_s3_transfer_advanced",
default_args=default_args,
schedule_interval=None,
tags=['example']) as dag:
# [START howto_operator_google_api_to_s3_transfer_advanced_task_1]
task_video_ids_to_s3 = GoogleApiToS3TransferOperator(
gcp_conn_id=YOUTUBE_CONN_ID,
google_api_service_name='youtube',
google_api_service_version='v3',
google_api_endpoint_path='youtube.search.list',
google_api_endpoint_params={
'part': 'snippet',
'channelId': YOUTUBE_CHANNEL_ID,
'maxResults': 50,
'publishedAfter': YOUTUBE_VIDEO_PUBLISHED_AFTER,
'publishedBefore': YOUTUBE_VIDEO_PUBLISHED_BEFORE,
'type': 'video',
'fields': 'items/id/videoId'
},
google_api_response_via_xcom='video_ids_response',
s3_destination_key=f'{s3_directory}/youtube_search_{s3_file_name}.json',
task_id='video_ids_to_s3')
# [END howto_operator_google_api_to_s3_transfer_advanced_task_1]
# [START howto_operator_google_api_to_s3_transfer_advanced_task_1_1]
task_check_and_transform_video_ids = BranchPythonOperator(
python_callable=_check_and_transform_video_ids,
op_args=[
task_video_ids_to_s3.google_api_response_via_xcom,
task_video_ids_to_s3.task_id