ftw.testing

This package provides helpers for writing tests.

Certified: 01/2013

Browser testing with splinter

Splinter is a library which provides a common API for multiple browser. It allows to operate zope.testbrowser, PhantomJS or other Selenium brwowsers such as Firefox or Chrome with the same API.

The ftw.testing package provides integration of Splinter with Plone using Page Objects.

For using the splinter features, use the splinter extras require:

ftw.testing [splinter]

Setting a package up for browser tests

It's easy to setup your package for browser tests:

Add a test-dependency to ftw.testing in your `setup.py`:

tests_require = [
    'ftw.testing',
    ]

setup(name='my.package',
      ...
      tests_require=tests_require,
      extras_require=dict(tests=tests_require),
      )

In your testing.py use the FunctionalSplinterTesting layer wrapper:

from ftw.testing import FunctionalSplinterTesting
from plone.app.testing import PLONE_FIXTURE
from plone.app.testing import PloneSandboxLayer
from plone.app.testing import applyProfile
from zope.configuration import xmlconfig


class MyPackageLayer(PloneSandboxLayer):

    defaultBases = (PLONE_FIXTURE,)

    def setUpZope(self, app, configurationContext):
        import my.package
        xmlconfig.file('configure.zcml', my.package)

    def setUpPloneSite(self, portal):
        applyProfile(portal, 'my.package:default')


MY_PACKAGE_FIXTURE = MyPackageLayer()
MY_PACKAGE_FUNCTIONAL_TESTING = FunctionalSplinterTesting(
    bases=(MY_PACKAGE_FIXTURE, ),
    name="my.package:functional")

Write tests using the Plone Page Objects:

from ftw.testing import browser
from ftw.testing import browser
from ftw.testing.pages import Plone
from my.package.testing import MY_PACKAGE_FUNCTIONAL_TESTING
from plone.app.testing import SITE_OWNER_NAME
from plone.app.testing import SITE_OWNER_PASSWORD
from unittest2 import TestCase


class TestDocument(TestCase):

    layer = MY_PACKAGE_FUNCTIONAL_TESTING

    def test_add_document(self):
        Plone().login(SITE_OWNER_NAME, SITE_OWNER_PASSWORD)
        Plone().visit_portal()
        Plone().create_object('Page', {'Title': 'Foo',
                                       'Body Text': '<b>Hello World</b>'})
        self.assertTrue(browser().is_text_present('Hello World'))

Writing Page Objects

Write your own Page Objects for your views and content types. Put a module pages.py in your tests folder:

from ftw.testing.pages import Plone


class MyContentType(Plone):

    def create_my_content(self, title, text):
        self.create_object('MyContent', {'Title': title,
                                         'Body Text: text})
        return self

The Page Object should have methods for all features of your view.

Switching to a JavaScript browser

The default browser for JavaScript enabled tests is PhantomJS. PhantomJS is fast, headless but runs JS by using Gecko. You should write JavaScript tests as few as possible but as much as necessary, because it will get really slow when you have lots of them.

Switching to the PhantomJS is done by just marking your test with the javascript decorator:

from ftw.testing import javascript


class TestDocument(TestCase):

    @javascript
    def test_add_document(self):
        Plone().login(SITE_OWNER_NAME, SITE_OWNER_PASSWORD)
        Plone().visit_portal()
        Plone().create_object('Page', {'Title': 'Foo',
                                       'Body Text': '<b>Hello World</b>'})
        self.assertTrue(browser().is_text_present('Hello World'))

Using the Plone Page Objects

The Plone page object provided by ftw.testing already has the most important features built in, such as:

portal_url handling (the zope.testbrowser URL is different than the PhantomJS url)
Login
Accessing Headings, <body>-CSS-classes, status messages
Adding content
TinyMCE handling

Currently it's best to just look in the page object code <https://github.com/4teamwork/ftw.testing/blob/jone-splinter/ftw/testing/pages.py>.

MockTestCase

ftw.testing provides an advanced MockTestCase which provides bases on the plone.mocktestcase MockTestCase.

from ftw.testing import MockTestCase

The following additional methods are available:

self.providing_mock(interfaces, *args, **kwargs): Creates a mock which provides interfaces.
self.mock_interface(interface, provides=None, *args, **kwargs): Creates a mock object implementing interface. The mock does not only provide interface, but also use it as specification and asserts that the mocked methods do exist on the interface.
self.stub(*args, **kwargs): Creates a stub. It acts like a mock but has no assertions.
self.providing_stub(interfaces, *args, **kwargs): Creates a stub which provides interfaces.
self.stub_interface(interface, provides=None, *args, **kwargs): Does the same as mock_interface, but disables counting of expected method calls and attribute access. See "Mocking vs. stubbing" below.
self.set_parent(context, parent_context): Stubs the context so that its acquisition parent is parent_context. Expects at least context to be a mock or a stub. Returns the context.
self.stub_request(interfaces=[], stub_response=True, content_type='text/html', status=200): Returns a request stub which can be used for rendering templates. With the stub_response option, you can define if the request should stub a response by itself. The other optional arguments: content_type: Defines the expected output content type of the response. status: Defines the expected status code of the response.
self.stub_response(request=None, content_type='text/html', status=200)): Returns a stub response with some headers and options. When a request is given the response is also added to the given request. The other optional arguments: content_type: Defines the expected output content type of the response. status: Defines the expected status code of the response.
self.assertRaises(*args, **kwargs): Uses unittest2 implementation of assertRaises instead of unittest implementation.

It also fixes a problem in mock_tool, where the getToolByName mock had assertions which is not very useful in some cases.

Mocking vs. stubbing

A mock is used for testing the communication between two objects. It asserts method calls. This is used when a test should not test if a object has a specific state after doing something (e.g. it has it's attribute xy set to something), but if the object does something with another object. If for example an object Foo sends an email when method bar is called, we could mock the sendmail object and assert on the send-email method call.

On the other hand we often have to test the state of an object (attribute values) after doing something. This can be done without mocks by just calling the method and asserting the attribute values. But then we have to set up an integration test and install plone, which takes very long. For testing an object with dependencies to other parts of plone in a unit test, we can use stubs for faking other (separately tested) parts of plone. Stubs work like mocks: you can "expect" a method call and define a result. The difference between stubs and mocks is that stubs do not assert the expectations, so there will be no errors if something expected does not happen. So when using stubs we can assert the state without asserting the communcation between objects.

Component registry layer

The MockTestCase is able to mock components (adapters, utilities). It cleans up the component registry after every test.

But when we use a ZCML layer, loading the ZCML of the package it should use the same component registry for all tests on the same layer. The ComponentRegistryLayer is a layer superclass for sharing the component registry and speeding up tests.

Usage:

from ftw.testing.layer import ComponentRegistryLayer

class ZCMLLayer(ComponentRegistryLayer):

    def setUp(self):
        super(ZCMLLayer, self).setUp()

        import my.package
        self.load_zcml_file('configure.zcml', my.package)

ZCML_LAYER = ZCMLLayer()

Be aware that ComponentRegistryLayer is a base class for creating your own layer (by subclassing ComponentRegistryLayer) and is not usable with defaultBases directly. This allows us to use the functions load_zcml_file and load_zcml_string.

Robot framework testing

For loading the needed dependencies for robot testing, just add a dependency to ftw.testing[robot]. You may also want plone.act for plone specific keywords.

Translations

Use the LocalizedRobotLayer for using robot framework in another language:

from ftw.testing import LocalizedRobotLayer
from plone.testing import Layer

class MyPackage(Layer):

    defaultBases = (LocalizedRobotLayer(['de']),)

MY_PACKAGE = MyPackage()

Links

Main github project repository: https://github.com/4teamwork/ftw.testing
Issue tracker: https://github.com/4teamwork/ftw.testing/issues
Package on pypi: http://pypi.python.org/pypi/ftw.testing
Continuous integration: https://jenkins.4teamwork.ch/search?q=ftw.testing

Copyright

This package is copyright by 4teamwork.

ftw.testing is licensed under GNU General Public License, version 2.

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

README.rst

README.rst

ftw.testing

Browser testing with splinter

Setting a package up for browser tests

Writing Page Objects

Switching to a JavaScript browser

Using the Plone Page Objects

MockTestCase

Mocking vs. stubbing

Component registry layer

Robot framework testing

Links

Copyright

Files

README.rst

Latest commit

History

README.rst

File metadata and controls

ftw.testing

Browser testing with splinter

Setting a package up for browser tests

Writing Page Objects

Switching to a JavaScript browser

Using the Plone Page Objects

MockTestCase

Mocking vs. stubbing

Component registry layer

Robot framework testing

Links

Copyright