Question

我有以下气流障碍：

    start_task = DummyOperator(task_id='start_task', dag=dag)

    gcs_export_uri_template = 'adstest/2018/08/31/*'
    update_bigquery = GoogleCloudStorageToBigQueryOperator(
        dag=dag,
        task_id='load_ads_to_BigQuery',
        bucket=GCS_BUCKET_ID,
        destination_project_dataset_table=table_name_template,
        source_format='CSV',
        source_objects=[gcs_export_uri_template],
        schema_fields=dc(),
        create_disposition='CREATE_IF_NEEDED',
        write_disposition='WRITE_APPEND',
        skip_leading_rows = 1,
        google_cloud_storage_conn_id=CONNECTION_ID,
        bigquery_conn_id=CONNECTION_ID
    )

start_task >> update_bigquery

此数据从adstest/2018/08/31/*加载到BigQuery，效果很好。

我想将Dag修改为基于execution date运行日期：

Execution date
Execution date - 1 days
Execution date - 2 days

示例，如果执行日期为2018-09-02，我希望DAG转到：

Execution date : adstest/2018/09/02/*
Execution date - 1 days : adstest/2018/09/01/*
Execution date - 2 days : adstest/2018/08/31/*

我该怎么做？

编辑： 这是我更新的代码：

for i in range(5, 0, -1):
    gcs_export_uri_template = ['''adstest/{{ macros.ds_format(macros.ds_add(ds, -{0}), '%Y-%m-%d', '%Y/%m/%d') }}/*'''.format(i)]
    update_bigquery = GoogleCloudStorageToBigQueryOperator(
        dag=dag,
        task_id='load_ads_to_BigQuery-{}'.format(i),
        bucket=GCS_BUCKET_ID,
        destination_project_dataset_table=table_name_template,
        source_format='CSV',
        source_objects=gcs_export_uri_template,
        schema_fields=dc(),
        create_disposition='CREATE_IF_NEEDED',
        write_disposition='WRITE_APPEND',
        skip_leading_rows=1,
        google_cloud_storage_conn_id=CONNECTION_ID,
        bigquery_conn_id=CONNECTION_ID
    )
    start_task >> update_bigquery

编辑2：

我的代码：

for i in range(5, 0, -1):
    gcs_export_uri_template = ['''adstest/{{ macros.ds_format(macros.ds_add(ds, -params.i), '%Y-%m-%d', '%Y/%m/%d') }}/*'''.format(i)]

    update_bigquery = GoogleCloudStorageToBigQueryOperator(
        dag=dag,
        task_id='load_ads_to_BigQuery-{}'.format(i),
        bucket=GCS_BUCKET_ID,
        destination_project_dataset_table=table_name_template,
        source_format='CSV',
        source_objects=gcs_export_uri_template,
        schema_fields=dc(),
        params={'i': i},
        create_disposition='CREATE_IF_NEEDED',
        write_disposition='WRITE_APPEND',
        skip_leading_rows=1,
        google_cloud_storage_conn_id=CONNECTION_ID,
        bigquery_conn_id=CONNECTION_ID
    )

模板：

代码给出此错误：

"Source URI must not contain the ',' character: gs://adstest/{ macros.ds_format(macros.ds_add(ds, -params.i), '%Y-%m-%d', '%Y/%m/%d') }/*">

Answer 1

您可以使用Airflow Macros来实现此目标，如下所示：

gcs_export_uri_template=[
    "adstest/{{ macros.ds_format(ds, '%Y-%m-%d', '%Y/%m/%d') }}/*",
    "adstest/{{ macros.ds_format(prev_ds, '%Y-%m-%d', '%Y/%m/%d') }}/*",
    "adstest/{{ macros.ds_format(macros.ds_add(ds, -2), '%Y-%m-%d', '%Y/%m/%d') }}/*"
]

update_bigquery = GoogleCloudStorageToBigQueryOperator(
    dag=dag,
    task_id='load_ads_to_BigQuery',
    bucket=GCS_BUCKET_ID,
    destination_project_dataset_table=table_name_template,
    source_format='CSV',
    source_objects=gcs_export_uri_template,
    schema_fields=dc(),
    create_disposition='CREATE_IF_NEEDED',
    write_disposition='WRITE_APPEND',
    skip_leading_rows = 1,
    google_cloud_storage_conn_id=CONNECTION_ID,
    bigquery_conn_id=CONNECTION_ID
)

运行以上代码时，可以在Web UI中检入呈现的参数：

对于已编辑的评论：

您将需要在i参数中传递循环变量params的值，并在字符串中将其用作params.i，如下所示：

for i in range(5, 0, -1):
    gcs_export_uri_template = ["adstest/{{ macros.ds_format(macros.ds_add(ds, -params.i), '%Y-%m-%d', '%Y/%m/%d') }}/*"]
    update_bigquery = GoogleCloudStorageToBigQueryOperator(
        dag=dag,
        task_id='load_ads_to_BigQuery-{}'.format(i),
        bucket=GCS_BUCKET_ID,
        destination_project_dataset_table=table_name_template,
        source_format='CSV',
        source_objects=gcs_export_uri_template,
        schema_fields=dc(),
        params={'i': i},
        create_disposition='CREATE_IF_NEEDED',
        write_disposition='WRITE_APPEND',
        skip_leading_rows=1,
        google_cloud_storage_conn_id=CONNECTION_ID,
        bigquery_conn_id=CONNECTION_ID
    )
    start_task >> update_bigquery

如何在Airflow中使用执行日期创建路径？

1 个答案: