getninjas / celery-executor

A `concurrent.futures.Executor` implementation using Celery as backend

Celery Executor

A concurrent.futures.Executor implementation using Celery as backend

• Free software: Apache Software License 2.0
• Documentation: https://celery-executor.readthedocs.io.

Features

The package provides a CeleryExecutor implementing the interface of concurrent.futures.Executor

>>> from celery_executor.executors import CeleryExecutor
>>> executor = CeleryExecutor()
>>> for result in executor.map(str.upper, ['one', 'two', 'three']):
...     print(result)
ONE
TWO
THREE

Beware that the Executor.map() interface can yield the results out of order, if later ones got to finish first.

Caveats

This executor frees the developer to the burden of mark every single task function with the Celery decorators, and to import such tasks on the Worker beforehand. But does not frees from sending the code to the Worker.

The function sent to CeleryExecutor.map() should be pickable on the client (caller of .map() or .submit()) and should be unpickable on the Celery Worker handling the "Task" sent. Is not possible to send lambdas for example.

As Celery assumes that is to the developer to put the needed code on the Worker, be sure that the function/partial code sent to CeleryExecutor to exist on the Worker.

To Be Done

• Document the CeleryExecutor.__init__() nonstandard extra options predelay, postdelay and applyasync_kwargs.
• Test behaviours of canceling a Task when canceling a Future
• Test behaviours of shutting down executors and trying to send new tasks
• Find a way to test the RUNNING state of Celery Tasks, as its events are not propagated by the test worker Celery provides

Credits

This package was created with Cookiecutter and the audreyr/cookiecutter-pypackage project template.