django-subscriptions

A django package for managing the status and terms of a subscription.

Compatibility

• Django: 2.2 (LTS versions only)
• Python: 3.6+

Other Django or Python versions may work, but that is totally cooincidental and no effort is made to maintain compatibility with versions other than those listed above.

Installation

$ pip install django-subscriptions

Then add the following packages to INSTALLED_APPS in your settings:

INSTALLED_APPS = [
    ...
    "django_fsm_log",
    "subscriptions.apps.SubscriptionsConfig",
    ...
]

And of course, you'll need to run the migrations:

$ python manage.py migrate

You'll also need to setup the triggers, which can be scheduled with celery or run from a management task. See the Triggers section below.

Design

Manages subscriptions in a single table. Pushes events (signals) so that consumers can do the actual work required for that subscription, like billing.

Subscriptions are built around a Finite State Machine model, where states and allowed transitions between states are well defined on the Model. To update from one state to another, the user calls methods on the Subscription instance. This way, all side-effects and actions are contained within the state methods.

Subscription State must not be modified directly.

When a state change is triggered, the subscription will publish relevant signals so that interested parties can, themselves, react to the state changes.

API

There are 3 major API components. State change methods, signals/events, and the triggers used to begin the state changes.

State Methods

Method	Source States	Target State	Signal Emitted
cancel_autorenew()	ACTIVE	EXPIRING	autorenew_canceled
enable_autorenew()	EXPIRING	ACTIVE	autorenew_enabled
renew()	ACTIVE,SUSPENDED	RENEWING	subscription_due
renewed(new_end, new_ref, description=None)	ACTIVE,RENEWING,ERROR	ACTIVE	subscription_renewed
renewal_failed(description=None)	RENEWING,ERROR	SUSPENDED	renewal_failed
end_subscription(description=None)	ACTIVE,SUSPENDED,EXPIRING,ERROR	ENDED	subscription_ended
state_unknown(description=None)	RENEWING	ERROR	subscription_error

Example:

subscription.renew() may only be called if subscription.state is either ACTIVE or SUSPENDED, and will cause subscription.state to move into the RENEWING state.

The description argument is a string that can be used to persist the reason for a state change in the StateLog table (and admin inlines).

Triggers

There are a bunch of triggers that are used to update subscriptions as they become due or expire. Nothing is configured to run these triggers by default. You can either call them as part of your own process, or use celery beat to execute the triggers using the tasks provided in subscriptions.tasks.

Create a new subscription:

Subscription.objects.add_subscription(start_date, end_date, reference) -> Subscription

Trigger subscriptions that are due for renewal:

Subscription.objects.trigger_renewals() -> int  # number of renewals sent

Trigger subscriptions that are due to expire:

Subscription.objects.trigger_expiring() -> int  # number of expirations

Trigger subscriptions that are suspended:

Subscription.objects.trigger_suspended() -> int  # number of renewals

Trigger subscriptions that have been suspended for longer than timeout_hours to end (uses subscription.end date, not subscription.last_updated):

Subscription.objects.trigger_suspended_timeout(timeout_hours=48) -> int  # number of suspensions

Trigger subscriptions that have been stuck in renewing state for longer than timeout_hours to be marked as an error (uses subscription.last_updated to determine the timeout):

Subscription.objects.trigger_stuck(timeout_hours=2) -> int  # number of error subscriptions

If settings.SUBSCRIPTIONS_STUCK_RETRY is True, then subscriptions are moved back into the SUSPENDED state, ready to be retried. This can be useful when you have an offline process that can resolve stuck subscription issues, and there is no issue retrying the subscription.

Tasks

The following tasks are defined but are not scheduled:

subscriptions.tasks.trigger_renewals
subscriptions.tasks.trigger_expiring
subscriptions.tasks.trigger_suspended
subscriptions.tasks.trigger_suspended_timeout
subscriptions.tasks.trigger_stuck

If you'd like to schedule the tasks, do so with a celery beat configuration like this:

# settings.py

CELERYBEAT_SCHEDULE = {
    "subscriptions_renewals": {
        "task": "subscriptions.tasks.trigger_renewals",
        "schedule": crontab(hour=0, minute=10),
    },
    "subscriptions_expiring": {
        "task": "subscriptions.tasks.trigger_expiring",
        "schedule": crontab(hour=0, minute=15),
    },
    "subscriptions_suspended": {
        "task": "subscriptions.tasks.trigger_suspended",
        "schedule": crontab(hour="3,6,9", minute=30),
    },
    "subscriptions_suspended_timeout": {
        "task": "subscriptions.tasks.trigger_suspended_timeout",
        "schedule": crontab(hour=0, minute=40),
        "kwargs": {"hours": 48},
    },
    "subscriptions_stuck": {
        "task": "subscriptions.tasks.trigger_stuck",
        "schedule": crontab(hour="*/2", minute=50),
        "kwargs": {"hours": 2},
    },
}

Contributing

We use pre-commit <https://pre-commit.com/> to enforce our code style rules locally before you commit them into git. Once you install the pre-commit library (locally via pip is fine), just install the hooks::

pre-commit install -f --install-hooks

The same checks are executed on the build server, so skipping the local linting (with git commit --no-verify) will only result in a failed test build.

Current style checking tools:

• flake8: python linting
• isort: python import sorting
• black: python code formatting

Note:

You must have python3.6 available on your path, as it is required for some
of the hooks.

Generating Migrations

After installing all dependencies, you can generate required migration files like so:

$ poetry run ipython migrate.py <nameofmigration>

Publishing a new version

1. Bump the version number in pyproject.toml and src/subscriptions/init.py
2. Commit and push to master
3. From github, create a new release
4. Name the release "v<maj.minor.patch>" using the version number from step 1.
5. Publish the release
6. If the release successfully builds, circleci will publish the new package to pypi