Print all DAGs except...

dag_ignore.py

import os
import sys


def print_all_dag_files(except_list):
    root_dir = '/Users/alex.levene/workspace/airflow-dags/'

    exclude_dirs = ['.buildkite', '.idea', '.mine', 'common']
    keep_dirs = ['airflow_config']

    ignores = []
    for dir_name, subdir_list, file_list in os.walk(root_dir):
        short_dir_name = dir_name.replace(root_dir, '').split('/')[0]
        if short_dir_name in (exclude_dirs + keep_dirs):
            continue

        for fname in file_list:
            if fname.endswith('.py'):
                full_fname = os.path.join(dir_name.replace(root_dir, ''), fname)
                if full_fname in except_list:
                    continue
                ignores.append(full_fname)

    [print('{}/.*'.format(f)) for f in exclude_dirs]

    for ignore in sorted(ignores):
        print(ignore)


if __name__ == '__main__':
    # pass list of filenames
    # Ex: python3 dag_ignore.py sqooperduper/zenpayroll.py athena_schema/__init__.py
    file_args = sys.argv[1:]
    print_all_dag_files(file_args)