pytest-celery-utils 0.1.2

Pytest plugin for inspecting Celery task queues in Redis during tests

Inspect Celery Queues with Pytest

I love Celery for background tasks, but testing them can be a headache. Usually, you have to run a worker or mock everything out. But sometimes you just want to know: "Did my task actually get queued in Redis?" and "Is the payload correct?"

That's why I put together pytest-celery-utils. It's a simple plugin that lets you peek directly into your Redis-backed Celery queues during your tests. No workers required—just pure inspection.

Installation

Grab it with uv:

uv add --dev pytest-celery-utils

Usage

Once installed, you get a celery_utils fixture. It's automatically available, so you can just drop it into your tests.

Here's how I use it to verify tasks are queuing up correctly:

def test_task_queued(celery_utils):
    # Fire off a task
    my_task.delay(arg1, arg2)

    # Get all queued tasks
    tasks = celery_utils.get_all_queued_tasks()
    assert len(tasks) > 0

    # Get tasks of specific type
    my_tasks = celery_utils.jobs_of_type("myapp.tasks.my_task")
    assert len(my_tasks) == 1

You'll need a celery_app fixture configured in your conftest.py so the plugin knows where to look:

import pytest
from celery import Celery

@pytest.fixture
def celery_app():
    # Make sure to point to your Redis instance
    return Celery("myapp", broker="redis://localhost:6379/0")

Features

• Direct Redis Inspection: Look at what's actually in your queue without needing a running worker.
• Task Counting: Easily assert that n jobs of a specific type are queued.
• Job Filtering: Grab all jobs or filter them by task name (string or callable).
• Queue Support: Inspects all queues defined in your Celery app.
• Nested Data Access: Includes a handy dig utility for traversing those deeply nested Celery message dicts.

Limitations

Just a heads up: this is strictly for Redis brokers. If you're using RabbitMQ or SQS, this won't work for you. It's also designed for testing, so don't expect it to be blazing fast for production monitoring.

MIT License