Skip to content

seedofjoy/darq

Repository files navigation

darq

[ ~ Dependencies scanned by PyUp.io ~ ]

https://github.com/seedofjoy/darq/workflows/Lint%20&%20test/badge.svg?branch=master

Async task manager with Celery-like features. Fork of arq.

Features

  • Celery-like @task decorator, adds .delay() to enqueue job
  • Proper mypy type checking: all arguments passed to .delay() will be checked against the original function signature
  • Graceful shutdown: waits until running tasks are finished

Installation

Darq uses aioredis 1.x as Redis client. Unfortunately, this library has been abandoned, and does not support Python 3.11. I made a fork with compatability fixes: evo-aioredis (https://github.com/evo-company/aioredis-py).

Because of this, aioredis is not currently added as Darq dependency, and you must install it yourself:

For Python<3.11 you can use:

pip install aioredis<2.0.0

For Python 3.11 (and older versions too) you can use fork:

pip install evo-aioredis<2.0.0

Quick start

# some_project/darq_app.py
import asyncio
import darq

darq = darq.Darq(redis_settings=darq.RedisSettings(host='redis'))


@darq.task
async def add_to_42(a: int) -> int:
    return 42 + a


async def main():
    # Before adding tasks to queue we should connect darq instance to redis
    await darq.connect()

    # Direct call job as function:
    result = await add_to_42(5)  # result == 47

    # Celery-like add task to queue:
    await add_to_42.delay(a=5)

    await darq.disconnect()


if __name__ == '__main__':
    asyncio.run(main())

And start worker:

python3 -m darq.cli -A some_project.darq_app.darq worker

Worker output:

15:24:42: Starting worker for 1 functions: some_project.darq_app.add_to_42
15:24:42: redis_version=5.0.7 mem_usage=834.87K clients_connected=1 db_keys=2
15:25:08:   0.22s → 1315f27608e9408392bf5d3310bca38c:darq_app.add_to_42(a=5)
15:25:08:   0.00s ← 1315f27608e9408392bf5d3310bca38c:darq_app.add_to_42 ● 47