Airflow Programming Model: Difference between revisions

From NovaOrdis Knowledge Base
Jump to navigation Jump to search
Line 95: Line 95:
==Using the <tt>PythonOperator</tt>==
==Using the <tt>PythonOperator</tt>==
==Using the <tt>@task</tt> Decorator==
==Using the <tt>@task</tt> Decorator==
<span id='Decorator_Example'></span><syntaxhighlight lang='py'>
from airflow.decorators import dag, task
from datetime import datetime
@dag(
    [...]
)
def some_dag():
    @task
    def task_a():
        print(">>> A")
    @task
    def task_b():
        print(">>> B")
    task_a() >> task_b()
</syntaxhighlight>


=Module Management=
=Module Management=
<font color=darkkhaki>TO PROCESS: https://airflow.apache.org/docs/apache-airflow/2.3.2/modules_management.html?highlight=import</font>
<font color=darkkhaki>TO PROCESS: https://airflow.apache.org/docs/apache-airflow/2.3.2/modules_management.html?highlight=import</font>

Revision as of 02:12, 16 July 2022

Internal

Overview

Airflow DAG Examples

https://github.com/apache/airflow/tree/main/airflow/example_dags

requirements.txt

apache-airflow == 2.3.3

Declaring a DAG

With @dag Decorator

The @dag decorator turns a Python function into a DAG generator function. @dag is only available in Airflow 2 and newer. The decorator can be configured with the a number of pre-defined parameters. The function includes task declaration and task relationship declaration, as follows:

from airflow.decorators import dag, task
from datetime import datetime


@dag(
    schedule_interval=None,
    start_date=datetime(2022, 7, 13, 0),
    catchup=False
)
def some_dag():
    @task
    def task_a():
        print(">>> A")

    @task
    def task_b():
        print(">>> B")

    @task
    def task_c():
        print(">>> C")

    task_a() >> task_b() >> task_c()


dag = some_dag()

Also see:

Airflow Concepts | Declaring a DAG

TO PARSE:

With DAG() Constructor

Also see:

Airflow Concepts | Declaring a DAG

import time
from datetime import datetime, timedelta
from airflow import DAG
from airflow.operators.empty import EmptyOperator
from airflow.operators.python import PythonOperator

default_args = {
    'owner': 'somebody',
    "depends_on_past": False,
    "email": ["somebody@example.com"],
    "email_on_failure": True,
    "email_on_retry": False
}

dag = DAG('some_dag',
          max_active_runs=1,
          catchup=True,
          start_date=datetime(2022, 7, 13, 0),
          schedule_interval="* * * * *",
          dagrun_timeout=timedelta(minutes=2),
          default_args=default_args
          )


def some_function():
    print("something")


with dag:
    start = EmptyOperator(task_id='start', dag=dag)
    end = EmptyOperator(task_id='end', dag=dag)
    some_function = PythonOperator(
        task_id='some_function',
        python_callable=some_function,
        dag=dag
    )

    start >> some_function >> end

Declaring a Task

Using the PythonOperator

Using the @task Decorator

from airflow.decorators import dag, task
from datetime import datetime


@dag(
    [...]
)
def some_dag():
    @task
    def task_a():
        print(">>> A")

    @task
    def task_b():
        print(">>> B")

    task_a() >> task_b()

Module Management

TO PROCESS: https://airflow.apache.org/docs/apache-airflow/2.3.2/modules_management.html?highlight=import