airflow-diagrams

Auto-generated Diagrams from Airflow DAGs.

This project aims to easily visualise your Airflow DAGs on service level from providers like AWS, GCP, Azure, etc. via diagrams.

🚀 Get started

To install it from PyPI run:

pip install airflow-diagrams

NOTE: Make sure you have Graphviz installed.

Then just call it like this:

Usage: airflow-diagrams generate [OPTIONS]

  Generates <airflow-dag-id>_diagrams.py in <output-path> directory which
  contains the definition to create a diagram. Run this file and you will get
  a rendered diagram.

Options:
  -d, --airflow-dag-id TEXT    The dag id from which to generate the diagram.
                               By default it generates for all.
  -h, --airflow-host TEXT      The host of the airflow rest api from where to
                               retrieve the dag tasks information.  [default:
                               http://localhost:8080/api/v1]
  -u, --airflow-username TEXT  The username of the airflow rest api.
                               [default: admin]
  -p, --airflow-password TEXT  The password of the airflow rest api.
                               [default: admin]
  -o, --output-path DIRECTORY  The path to output the diagrams to.  [default:
                               .]
  -m, --mapping-file FILE      The mapping file to use for static mapping from
                               Airflow task to diagram node. By default no
                               mapping file is being used.
  -v, --verbose                Verbose output i.e. useful for debugging
                               purposes.
  -lw, --label-wrap TEXT       Specify either a number for label width or a
                               separator to indicate when to wrap a label. By
                               default it does not wrap labels.
  --help                       Show this message and exit.

Examples of generated diagrams can be found in the examples directory.

🤔 How it Works

ℹ️ At first it connects, by using the official Apache Airflow Python Client, to your Airflow installation to retrieve all DAGs (in case you don't specify any dag_id) and all Tasks for the DAG(s).

🔮 Then it tries to find a diagram node for every DAGs task, by using Fuzzy String Matching, that matches the most. If you are unhappy about the match you can also provide a mapping.yml file to statically map from Airflow task to diagram node.

🪄 Lastly it renders the results into a python file which can then be executed to retrieve the rendered diagram. 🎉

❤️ Contributing

Contributions are very welcome. Please go ahead and raise an issue if you have one or open a PR. Thank you.