Python 3.x 在数据库中组织DAG任务依赖项

我有一个虚拟任务和一个使用循环并行运行的任务列表，对于一个任务，我想再多执行一个串行任务

到目前为止，我所尝试的是：

dummy = DummyOperator(task_id='Dummy'.upper(),
                                              dag=dag)


final = DummyOperator(task_id='FinalTask'.upper(),
                                              dag=dag)


for task in ['Task1', 'Task2', 'Task3']:

    if task == 'Task1'
            task1 = DummyOperator(task_id='Task1-a'.upper(),
                                              dag=dag)

        else:
             ...
    else:
        ...

    tasks = DummyOperator(task_id=task),
                                      dag=dag)

    dummy >> tasks
    tasks >> task1
    tasks >> final

---

您没有解释我们如何知道Task1的子任务的逻辑

这将构建您希望的结构：

tasks = ['Task1', 'Task2', 'Task3']

default_args = {
    'owner': 'airflow',
    'start_date': datetime(2020, 12, 17),
}

with DAG(
    dag_id='dummyplay2',
    default_args=default_args,
    schedule_interval=None,
) as dag:
    start_op = DummyOperator(task_id='start')
    final_op = DummyOperator(task_id='final')
    for task in tasks:
        task_op = DummyOperator(task_id=task)
        start_op >> task_op
        if task == 'Task1':
            #This loop creates the sub task logic.
            #You can replace ord('b') with ord('z) and it will create more sub tasks
            for i in range(ord('a'), ord('b')+1):
                sub_task_op = DummyOperator(task_id=f'{task}_{chr(i)}')
                task_op >> sub_task_op >> final_op
        else:
            task_op >> final_op

DAG将是：

---

您没有解释我们如何知道Task1的子任务的逻辑

这将构建您希望的结构：

tasks = ['Task1', 'Task2', 'Task3']

default_args = {
    'owner': 'airflow',
    'start_date': datetime(2020, 12, 17),
}

with DAG(
    dag_id='dummyplay2',
    default_args=default_args,
    schedule_interval=None,
) as dag:
    start_op = DummyOperator(task_id='start')
    final_op = DummyOperator(task_id='final')
    for task in tasks:
        task_op = DummyOperator(task_id=task)
        start_op >> task_op
        if task == 'Task1':
            #This loop creates the sub task logic.
            #You can replace ord('b') with ord('z) and it will create more sub tasks
            for i in range(ord('a'), ord('b')+1):
                sub_task_op = DummyOperator(task_id=f'{task}_{chr(i)}')
                task_op >> sub_task_op >> final_op
        else:
            task_op >> final_op

DAG将是：