django-openfeature 0.2.0

OpenFeature integration for Django

django-openfeature

⚠️ Caution: this repository is a work-in-progress. ⚠️

Django OpenFeature is a set of utilities to use OpenFeature in your Django applications.

Features

• Django Debug Toolbar integration
• Templatetags for flag evaluation
• Automatic evaluation context from request
• Flag override mechanism for testing

Installation

pip install django-openfeature

Add django_openfeature to your INSTALLED_APPS setting.

INSTALLED_APPS = [
    # ...
    'django_openfeature',
    # ...
]

Usage

Flag Evaluation Helpers

django-openfeature provides a set of helpers to evaluate a feature flag:

from django_openfeature import feature

feature(request, 'my_feature', False)

The feature function will infer the type of the feature flag based on the default value provided and call the appropriate resolver method.

It will also create an evaluation context from the request. This context can be configured via the OPENFEATURE setting (described below).

Inside templates, you can use the feature template tag to evaluate feature flags. Remember to load the openfeature library first.

{% load openfeature %}

{% feature 'my_feature' False as my_feature_enabled %}
{% if my_feature_enabed %}
    <p>Feature is enabled</p>
{% else %}
    <p>Feature is disabled</p>
{% endif %}

As an alternative to evaluating feature flags to a variable you can use the iffeature template tag. It will output the contents of the block if the boolean flag resolves to True. It supports a single {% else %} clause that will be displayed otherwise.

{% load openfeature %}

{% iffeature 'my_feature' %}
    <p>Feature is enabled</p>
{% else %}
    <p>Feature is disabled</p>
{% endif %}

Configuration

OPENFEATURE = {
    "PROVIDER": "myapp.utils.get_openfeature_provider",
    "CONTEXT_EVALUATOR": "myapp.utils.get_evaluation_context",
}

The PROVIDER setting configures the global OpenFeature provider when the app registry loads, saving you from calling openfeature.api.set_provider() yourself at import time. It should point to a provider class, a zero-argument factory function that returns a provider instance, or a provider instance. When omitted (or None), django-openfeature does not touch the provider and you can keep setting it manually.

# myapp/utils.py
from devcycle_python_sdk import DevCycleCloudClient, DevCycleCloudOptions
from django.conf import settings

def get_openfeature_provider():
    client = DevCycleCloudClient(settings.DEVCYCLE_API_KEY, DevCycleCloudOptions())
    return client.get_openfeature_provider()

The provider is re-configured when the OPENFEATURE setting changes, so override_settings(OPENFEATURE={"PROVIDER": ...}) works as expected in tests.

The CONTEXT_EVALUATOR setting should point to a function that receives a request and returns an OpenFeature EvaluationContext as defined in the openfeature-sdk package.

Alternatively, the context evaluator can point to a class that produces callable instances with the same signature. The default context evaluator is django_openfeature.context.OpenFeatureContext and looks roughly like this:

class OpenFeatureContext:
    targeting_key = "id"
    targeting_key_anonymous = "unknown"
    user_attributes = ("username", "email")

    def get_context(self, request):
        user = request.user
        if not user or user.is_anonymous:
            return EvaluationContext(self.targeting_key_anonymous)
        return self.get_user_context(user)

    def get_user_context(self, user):
        attributes = {}
        for attr in self.user_attributes:
            attributes[attr] = getattr(user, attr)
        targeting_key = str(getattr(user, self.targeting_key))
        return EvaluationContext(targeting_key, attributes)

    def __call__(self, request):
        return self.get_context(request)

Use this class as base to declaratively define your own context evaluator. For example, to change the targeting key to the user's username:

from django_openfeature.context import OpenFeatureContext

class MyOpenFeatureContext(OpenFeatureContext):
    targeting_key = "username"

Then in your settings:

OPENFEATURE = {
    "CONTEXT_EVALUATOR": "myapp.utils.MyOpenFeatureContext",
}

Debug Toolbar Panel

django-openfeature comes with a Feature Flags panel for the Django Debug Toolbar. You can activate it by adding openfeature.debug_toolbar.panels.FeatureFlagsPanel to your DEBUG_TOOLBAR_PANELS setting.

DEBUG_TOOLBAR_PANELS = [
    # ...
    'openfeature.debug_toolbar.panels.FeatureFlagsPanel',
    # ...
]

The Feature Flags panel will show you the feature flag evaluations for the current request, the request evaluation context and the configured providers for those evaluations.

Testing Utilities

django-openfeature provides a set of utilities to help you test your feature flags. To use them, you must set the provider to django_openfeature.provider.DjangoTestProvider in your test settings.

# in your test settings
OPENFEATURE = {
    "PROVIDER": "django_openfeature.provider.DjangoTestProvider",
}

override_feature is a function decorator that allows you to override the value of a feature flag for the duration of a test.

from django_openfeature.test import override_feature

@override_feature('my_feature', True)
def test_my_feature_enabled(self):
    # ...

Decorators can be stacked to override multiple feature flags.

You can also use override_feature as a context manager.

from django_openfeature.test import override_feature

def test_my_feature_enabled(self):
    with override_feature('my_feature', True):
        # ...

TODO

• Add support for OpenFeature domains (requires SDK release)

License

MIT License