Django Cloud Task Queue

Integrate your Django application with Google Cloud Task from Google Cloud Platform. This package provides a simple and easy to use decorator to push tasks to Cloud Task Queue and automatically handle all income task execution requests from Cloud Task. In a single entry point.

At the moment dj-cloud-task only works with HTTP targets such as Cloud Run, Cloud Functions, Compute Engine, ... (in case you are using Google Cloud Platform infrastructure). See the Cloud Task documentation about targets for more here!

Some Features

• Easily push tasks to Cloud Task using a decorator
• Automatically route all tasks from a single endpoint
• Ease scheduling with native python datetime
• Named task to avoid duplicated
• Local development support with Redis Queue

Installation

Simple dj-cloud-task can be installed with pip:

pip install dj-cloud-task

Or from this repository:

pip install -e git+https://github.com/txiocoder/dj-cloud-task.git@main#egg=cloudtask

Requirements

• Python >= 3.9
• django >= 3.2.*
• google-cloud-tasks >= 2.5.*

These are the officially supported python and packages versions. Other versions will probably work

Quickstart

Configure

As stated above, Django Cloud Task is a Django Application. To configure your project you simply need to add cloudtask to your INSTALLED_APPS and configure the CLOUDTASK variable in the settings.py file. More details about how to configure the CLOUDTASK variable below.

In file settings.py:

INSTALLED_APPS: list = [
    'cloudtask',
]

CLOUDTASK: dict = {
    'PROJECT': 'project',
    'LOCATION': 'europe-west6',
    'SAE': '[email protected]',
    'URL': 'https://handler.com/_tasks/',
    'QUEUE': 'default',
    'SECRET': 'my-very-secrete-key'
}

Then, add cloudtask.urls to your URL configuration to route and handle all task execution requests coming from Cloud Task.

In your project.urls:

from django.urls import path, include

urlpatterns: list = [
    path('_tasks/', include('cloudtask.urls'))
]

Create Tasks

The tasks are simple python functions that could be defined anywhere. But, I suggest you create a file with the name tasks.py in your django module/app and declare there. To create a task, simply decorate a function with cloudtask.decorators.task. See the example below:

from cloudtask import (
    CloudTaskRequest,
    task)

@task(queue='default')
def add(request: CloudTaskRequest, a: int = 5, b: int = 4) -> None:
    print(f'Running task with args {a=} and {b=}')
    print(a + b)

Run Tasks

To send a task to Cloud Task call the task function and then call the delay method inside the returned task instance. This will send a request to Cloud Task to enqueue the task and Cloud Task will request to run it as fast as possible.

from .tasks import add

add(a=10, b=30).delay()
# or use the alias push
add(a=30, b=10).push()

GCP Authentication

This module requires to be authenticated with Google Cloud Platform as a service. The GC Platform provides various ways to authenticate with it. See the GC Platform page about authentication strategies here.

Configurations

In this session you will see how to configure cloudtask. We have required attributes, optional but required in task declaration and only optional attributes. The required attributes are PROJECT, LOCATION and SAE.

Here the details about all attributes accpeted in CLOUDTASK

Attribute	Type	Required	Description
PROJECT	str	True	Project ID from Google Cloud Platform
LOCATION	str	True	Cloud Task Queue Location
SAE	str	True	Service Account Email
URL	str	False	Default URL
QUEUE	srt	False	Default Queue Name
SECRET	str	False	Secret key to authorize task execution
LOCAL_RQ	bool	False	Use Redis Queue to handle tasks locally
LOCAL_RQ_URL	str	False	Optional Redis connection URL
TESTING	bool	False	Testing Mode

URL attribute

The URL attribute is optional, but if you don't set you will need to explicitly pass as a task decorator argument

from cloudtask import task

@task(queue='emails', url='https://mysite.com/emails-tasks')
def send_email(request, to: str):
    pass

You can change the url of task instance in runtime if you need things to be done more automated.

from .tasks import send_email

task = send_email(to='[email protected]')
task.url = 'https://mysite.com/tasks'
task.push()

QUEUE attribute

The same as the URL attribute, the QUEUE attribute is optional, and if you don't set it you will need to explicitly pass as a task decorator argument. You can not change the QUEUE of task instance at runtime.

from cloudtask import task

@task(queue='default')
def send_email(request, to: str):
    pass

TESTING attribute

Useful when testing your django application. If True will run all tasks immediately without push to Cloud Task Queue

Working with Tasks

In this session you will see how to play with tasks. Django Cloud Task provides not just one way to create, push and handle tasks. It's flexible! All defined tasks receive the request keyword argument which is an instance of cloudtask.tasks.CloudTaskRequest containing all request information from Cloud Task

Creating and Push

from cloudtask import (
    CloudTaskRequest,
    task)

@task()
def say_yes(rquest) -> None:
    print('Yes')

@task(queue='default', named=True)
def add(request: CloudTaskRequest, a: int = 5, b: int = 4) -> None:
    print(f'Running task with args {a=} and {b=}')
    print(a + b)

# Pushing to Cloud Task

say_yes().delay() # or with push
add().push()

Immediately Execute a Task

Sometimes you will need to execute the task function immediately (without pushing to Cloud Task), to do that, just call the __cal__ or execute method from the returned task instance. The request will have limited information.

from .tasks import add
add(a=5, b=3).__call__()
# or using the alias
add(a=5, b=3)()
# or using the execute method
add(a=5, b=3).execute()

Scheduling

You can schedule a task to be delivered later using native python datetime.

from datetime import timedelta
from django.utils.timezone import now
from .tasks import add

at = now() + timedelta(days=2)
add(a=3, b=6).schedule(at=at)

Named Tasks

from .tasks import add

add(a=3, b=6).delay()
add(a=3, b=6).delay()
add(a=3, b=6).delay()

By default, the above will run normally. Cloud Task by default adds a unique name for each new task. That makes it possible to have duplicated tasks in the queue, even with the same arguments. If you want a task to only be enqueued once at time, you have to set the task as a named task.

Django Cloud Task will give a task name based on the task function name.

from cloudtask import (
    CloudTaskRequest,
    task)

@task(named=True)
def clean_expired(request: CloudTaskRequest):
    pass

clean_expired().delay()
clean_expired().delay() # this line will raise an entity error by Cloud Task

You can also set the name in task decorator or dynamically. This feature is very useful when you want to do some recursive tasks.

from cloudtask.tasks import Task
from cloudtask import (
    CloudTaskRequest,
    task)


@task(named='SOME_UNIQUE_NAME')
def task_do_some(request: CloudTaskRequest):
    pass

# or dynamically

@task()
def delete_article(request: CloudTaskRequest, article_id: int):
    pass

article_id: int = 34
task: Task = delete_article(article_id=article_id)
task.named = f'DELETE_ARTICLE_{article_id}'
task.push()

Local development support

Setup

Use Redis Queue for local development support. To start, first install rq, rq-scheduler and requests with pip. You will need a Redis connection too. Then configure on your CLOUDTASK settings.

pip install rq requests rq-scheduler

Configurations

On CLOUDTASK settings set LOCAL_RQ as True to start handle the tasks locally.

CLOUDTASK: dict = {
    'LOCAL_RQ': True
}

You can use LOCAL_RQ_URL to change the default redis connection string

CLOUDTASK: dict = {
    'LOCAL_RQ': True,
    'LOCAL_RQ_URL': 'redis://localhost:6379' # default by redis
}

That is all, but do not forget to set the right local URL on your CLOUDTASK settings to handle the tasks.

Running Tasks Locally

To start running task locally just start the worker process with the available management command

python manage.py cloudtask-worker

If you have task that are scheduled, start the rqscheduler worker process to support scheduling tasks. In your project root dir:

rqscheduler