Location via proxy:   [ UP ]  
[Report a bug]   [Manage cookies]                
Skip to content

Reusable Django app for Penn Labs accounts (DLA)

License

Notifications You must be signed in to change notification settings

pennlabs/django-labs-accounts

Repository files navigation

Django Labs Accounts

CircleCI Coverage Status PyPi Package

Requirements

  • Python 3.11+
  • Django 5.0+

Installation

Install with pip pip install django-labs-accounts

Add accounts to INSTALLED_APPS

INSTALLED_APPS = (
    ...
    'accounts.apps.AccountsConfig',
    'identity.apps.IdentityConfig', # If you want to enable B2B IPC
    'analytics.apps.AnalyticsConfig',
    ...
)

Add the new accounts backend to AUTHENTICATION_BACKENDS

AUTHENTICATION_BACKENDS = (
    ...
    'accounts.backends.LabsUserBackend',
    'django.contrib.auth.backends.ModelBackend',
    ...
)

(Optional) Add the new Platform DRF authentication class to rest framework's DEFAULT_AUTHENTICATION_CLASSES. This authentication class should go at the end of the list of authentication classes in most cases.

REST_FRAMEWORK = {
    ...
    'DEFAULT_AUTHENTICATION_CLASSES': [
        'rest_framework.authentication.SessionAuthentication',
        'rest_framework.authentication.BasicAuthentication',
        'accounts.authentication.PlatformAuthentication',
    ]
    ...
}

Add the following to urls.py

urlpatterns = [
    ...
    path('accounts/', include('accounts.urls', namespace='accounts')),
    ...
]

Documentation

All settings are handled within a PLATFORM_ACCOUNTS dictionary.

Example:

PLATFORM_ACCOUNTS = {
    'CLIENT_ID': 'id',
    'CLIENT_SECRET': 'secret',
    'REDIRECT_URI': 'example',
    'ADMIN_PERMISSION': 'example_admin'
    'CUSTOM_ADMIN': True
}

The available settings are:

CLIENT_ID the client ID to connect to platform with. Defaults to LABS_CLIENT_ID environment variable.

CLIENT_SECRET the client secret to connect to platform with. Defaults to LABS_CLIENT_SECRET environment variable.

REDIRECT_URI the redirect uri to send to platform. Defaults to first the LABS_REDIRECT_URI environment variable and then generating the value from the request object.

SCOPE the scope for this applications tokens. Must include introspection. Defaults to ['read', 'introspection'].

PLATFORM_URL URL of platform server to connect to. Should be https://platform(-dev).pennlabs.org (no trailing slash)

ADMIN_PERMISSION The name of the permission on platform to grant admin access. Defaults to example_admin

CUSTOM_ADMIN enable the custom admin login page to log in users through platform. Defaults to True

When developing locally with an http (not https) callback URL, it may be helpful to set the OAUTHLIB_INSECURE_TRANSPORT environment variable.

os.environ['OAUTHLIB_INSECURE_TRANSPORT'] = "1"

Custom post authentication

If you want to customize how DLA saves user information from platform into User objects, you can subclass accounts.backends.LabsUserBackend and redefine the post_authenticate method. This method will be run after the user is logged in. The parameters are:

  • user the user object
  • created a boolean delineating if the user was just created
  • dictionary a dictionary of user information from platform.

Then just set the AUTHENTICATION_BACKENDS setting to be the subclassed backend.

Here is an example of a custom backend that sets every user's first name to "Modified".

from accounts.backends import LabsUserBackend

class CustomBackend(LabsUserBackend):
    def post_authenticate(self, user, created, dictionary):
        user.first_name = 'Modified'
        user.save()

Analytics

DLA provides a wrapper class to submit analytics data from Labs backend servers to the Labs Analytics Server. For local testing, the necessary environment variables are the CLIENT_ID, CLIENT_SECRET, and PLATFORM_URL. Upon loading these variables, you can send data as follows:

def generate_data(self):
    return {
        "product": Product.MOBILE_BACKEND,
        "pennkey": None,
        "data": [{"key": f"{self.rand_int()}", "value": f"{self.rand_int()}"}],
    }

txn = AnalyticsTxn(**self.generate_data())
self.analytics_wrapper.submit(txn)

B2B IPC

DLA also provides an interface for backend to backend IPC requests. With B2B IPC implemented, the backend of a product will—at startup time—request platform for a JWT to verify its identity. Each product will have an allow-list, and this will enable products to make requests to each other.

In order to limit a view to only be available to a B2B IPC request, you can use the included DRF permission:

from identity.permissions import B2BPermission
class TestView(APIView):
    permission_classes = [B2BPermission("urn:pennlabs:example")]

Make sure to define an URN to limit access. Valid URNs are either a specific product (ex. urn:pennlabs:platform) or a wildcard (ex. urn:pennlabs:*)

In order to make an IPC request, use the included helper function:

from identity.identity import authenticated_b2b_request
result = authenticated_b2b_request('GET', 'http://url/path')

Use in Production

DLA and Penn Labs' templates are set up so that no configuration is needed to run in development. However, in production a client ID and client secret need to be set. These values should be set in vault. Contact platform for both credentials and any questions you have.

B2B IPC

DLA also provides an interface for backend to backend IPC requests. In order to limit a view to only be available to a B2B IPC request, you can use the included DRF permission:

from identity.permissions import B2BPermission

class TestView(APIView):
    permission_classes = [B2BPermission("urn:pennlabs:example")]

Make sure to define an URN to limit access. Valid URNs are either a specific product (ex. urn:pennlabs:platform) or a wildcard (ex. urn:pennlabs:*)

In order to make an IPC request, use the included helper function:

from identity.identity import authenticated_b2b_request

result = authenticated_b2b_request('GET', 'http://url/path')

Development Setup

Install poetry:

pipx install poetry

Install Dependencies:

poetry install

Testing:

export DJANGO_SETTINGS_MODULE=tests.settings && poetry run pytest

Linting:

poetry run black . && poetry run isort . && poetry run flake8

Changelog

See CHANGELOG.md

License

See LICENSE