dag-factory-hotfix 0.11.1

Dynamically build Airflow DAGs from YAML files

dag-factory

dag-factory is a library for dynamically generating Apache Airflow DAGs from YAML configuration files.

• Installation
• Usage
• Benefits
• Contributing

Installation

To install dag-factory run pip install dag-factory. It requires Python 3.6.0+ and Apache Airflow 1.10+.

Usage

After installing dag-factory in your Airflow environment, there are two steps to creating DAGs. First, we need to create a YAML configuration file. For example:

example_dag1:
  default_args:
    owner: 'example_owner'
    start_date: 2018-01-01  # or '2 days'
    end_date: 2018-01-05
    retries: 1
    retry_delay_sec: 300
  schedule_interval: '0 3 * * *'
  concurrency: 1
  max_active_runs: 1
  dagrun_timeout_sec: 60
  default_view: 'tree'  # or 'graph', 'duration', 'gantt', 'landing_times'
  orientation: 'LR'  # or 'TB', 'RL', 'BT'
  description: 'this is an example dag!'
  on_success_callback_name: print_hello
  on_success_callback_file: /usr/local/airflow/dags/print_hello.py
  on_failure_callback_name: print_hello
  on_failure_callback_file: /usr/local/airflow/dags/print_hello.py
  tasks:
    task_1:
      operator: airflow.operators.bash_operator.BashOperator
      bash_command: 'echo 1'
    task_2:
      operator: airflow.operators.bash_operator.BashOperator
      bash_command: 'echo 2'
      dependencies: [task_1]
    task_3:
      operator: airflow.operators.bash_operator.BashOperator
      bash_command: 'echo 3'
      dependencies: [task_1]

Then in the DAGs folder in your Airflow environment you need to create a python file like this:

from airflow import DAG
import dagfactory

dag_factory = dagfactory.DagFactory("/path/to/dags/config_file.yml")

dag_factory.clean_dags(globals())
dag_factory.generate_dags(globals())

And this DAG will be generated and ready to run in Airflow!

Notes

HttpSensor (since 0.10.0)

The package airflow.sensors.http_sensor works with all supported versions of Airflow. In Airflow 2.0+, the new package name can be used in the operator value: airflow.providers.http.sensors.http

The following example shows response_check logic in a python file:

task_2:
      operator: airflow.sensors.http_sensor.HttpSensor
      http_conn_id: 'test-http'
      method: 'GET'
      response_check_name: check_sensor
      response_check_file: /path/to/example1/http_conn.py
      dependencies: [task_1]

The response_check logic can also be provided as a lambda:

task_2:
      operator: airflow.sensors.http_sensor.HttpSensor
      http_conn_id: 'test-http'
      method: 'GET'
      response_check_lambda: 'lambda response: "ok" in reponse.text'
      dependencies: [task_1]

Benefits

• Construct DAGs without knowing Python
• Construct DAGs without learning Airflow primitives
• Avoid duplicative code
• Everyone loves YAML! ;)

Contributing

Contributions are welcome! Just submit a Pull Request or Github Issue.