Creating your first PyForge Tool

Adding your PyForge Tool to a PyForge Install

Writing a Wiki Tool Part 1: Pages

Writing a wiki Tool Part 2: Links

Writing a wiki Tool Part 3: Revisions

Testing your Tool

Testing the controllers and models of a new Forge tool is fairly straightforward. Generally, you should follow the example of tests in the allura/tests/functional directory for controller tests and allura.tests.model for model tests. For functional tests, the Forge platform provides a convenient "test harness" :class:`allura.controllers.test.TestController` controller class which is used as the application root for the :class:`allura.tests.TestController` class.

In order to test your new tool controllers, you simply need to use the self.app.get() and self.app.post() methods of your test controller. The test harness makes all the tools available in the system available under the URL /entry point name/. So to test the :mod:`allura.ext.project_home` tool, for instance, we need only write the following:

from allura.tests import TestController

class TestProjectHome(TestController):

    def test_home(self):
        r = self.app.get('/home/')

Whenever you use the :class:`allura.tests.TestController` app property, the test harness sets up the context so that c.project is always the projects/test project and whichever tool name you request is mounted at its entry point (so the Wiki tool will be mounted at /Wiki/). c.user is always set to the test-admin user to avoid authentication issues.

The framework used to generate the WSGI environment for testing your tools is provided by the WebTest module, where you can find further documentation for the .get() and .post() methods.

Testing new Forge models is also straightforward, though it usually requires setting the pylons context object c before your test. An example of this technique follows:

import mock
from pylons import c, g

from allura.lib.app_globals import Globals
from allura import model as M

def setUp():
    g._push_object(Globals())
    c._push_object(mock.Mock())
    g.set_project('projects/test')
    g.set_app('hello')
    c.user = M.User.query.get(username='test-admin')

Testing the reactors/auditors is similar to testing models. Generally, you will simply want to call your callback methods directly rather than setting up a full mocking infrastructure for the messaging system provided by RabbitMQ.