Functional testing

Description

Functional testing tool allows you to use scripted browser to load pages from your site and fill in forms automatically.

Introduction

PloneTestCase product provides FunctionalTestCase base class for functional testing. Unlike unit tests, functional tests simulate real HTTP requests with transaction life cycle.

  • Functional tests has different transaction for each browser.open() request
  • Functional tests do traversing and can check e.g. for cookie based permissions
  • Unit test method is executed in a single transaction and this might make impossible to test cache related behavior

Test browser

Plone uses Products.Five.testbrowser as an browser emulator used in functional tests. It is based on zope.testbrowser package. You can find more information in the zope.testbrowser docs home page. The API is described in zope.testbrowser.interfaces (3.4 used by Plone 3).

Warning

There also exists old zc.testbrowser, which is a different package with similar name.

All code assumes here is is executed in unit test context where self.portal is your unit test site instance.

Recording tests

You can record functional tests through the browser. Think it as a Microsoft Word macro recoder kind of thing.

Functional test skeleton

First see collective.testlayer package which does some of the things described below

Example code:

from Products.Five.testbrowser import Browser
from Products.PloneTestCase import PloneTestCase as ptc

class BaseFunctionalTestCase(ptc.FunctionalTestCase):
    """ This is a base class for functional test cases for your custom product.
    """

    def afterSetUp(self):
        """
        Show errors in console by monkey patching site error_log service
        """

        ptc.FunctionalTestCase.afterSetUp(self)

        self.browser = Browser()
        self.browser.handleErrors = False # Don't get HTTP 500 pages


        self.portal.error_log._ignored_exceptions = ()

        def raising(self, info):
            import traceback
            traceback.print_tb(info[2])
            print info[1]

        from Products.SiteErrorLog.SiteErrorLog import SiteErrorLog
        SiteErrorLog.raising = raising


    def loginAsAdmin(self):
        """ Perform through-the-web login.

        Simulate going to the login form and logging in.

        We use username and password provided by PloneTestCase.

        This sets session cookie for testbrowser.
        """
        from Products.PloneTestCase.setup import portal_owner, default_password

        # Go admin
        browser = self.browser
        browser.open(self.portal.absolute_url() + "/login_form")
        browser.getControl(name='__ac_name').value = portal_owner
        browser.getControl(name='__ac_password').value = default_password
        browser.getControl(name='submit').click()

Preparing error logger

Since zope.testbrowser uses normal Plone paging mechanism, you won't get nice tracebacks to your console.

The following snippet allows you to extract traceback data from site.error_log utility and print it to the console. Put it to your afterSetUp():

self.browser.handleErrors = False
self.portal.error_log._ignored_exceptions = ()

def raising(self, info):
    import traceback
    traceback.print_tb(info[2])
    print info[1]

from Products.SiteErrorLog.SiteErrorLog import SiteErrorLog
SiteErrorLog.raising = raising

Opening an URL

Example:

from Products.Five.testbrowser import Browser

self.browser = Browser()

self.browser.open(self.portal.absolute_url())

Logging in

Example:

from Products.PloneTestCase.setup import portal_owner, default_password

 # Go admin
browser.open(self.portal.absolute_url() + "/login_form")
browser.getControl(name='__ac_name').value = portal_owner
browser.getControl(name='__ac_password').value = default_password
browser.getControl(name='submit').click()

Logout

Example:

def logoutWithTestBrowser(self):
    """
    """
    self.browser.open(self.portal.absolute_url() + '/logout')
    html = self.browser.contents
    self.assertTrue("You are now logged out" in html)

Showing the contents from the last request

After test browser has opened an URL its content can be read from browser.contents variable.

Example:

print browser.contents # browser is zope.testbrowser.Browser instance

Getting a form handler

You can use testbrowser getForm() to access different forms on a page.

Form look-up is available by name or index.

Example:

form = browser.getForm(index=2) # Skip login and search form on Plone 4

Listing available form controls

You can do the following to know what content your form has eaten

  • the mechanize browser instance that is used through zope.testbrowser. zope.testbrowser internally uses a testbrowser provided by the mechanize package. The mechanize objects are saved in browser.mech_browser and as attributes on different other instances returned by zope.testbrowser. mechanize has a different, less convenient api, but also provides more options. To see a list of all controls in a for you can do e.g.:

    # get the login form from the zope.testbrowser
    login_form = self.browser.getForm('login_form')
    # get and print all controls
    controls = login_form.mech_form.controls
    for control in controls:
       print "%s: %s" % (control.attrs['name'], control.attrs['type'])
    

... or one-liner ...:

for c in form.mech_form.controls: print c
  • the HTML page source code:

    print browser.contents
    

Filling in a text field on a page

You can manipulate value of various form input controls.

Example how to submit Plone search page:

self.browser.open(self.portal.absolute_url() + "/search")

# Input some values to the search that we see we get
# zero hits and at least one hit
for search_terms in [u"Plone", u"youcantfindthis"]:
    form = self.browser.getForm("searchform")

    # Fill in the search field
    input = form.getControl(name="SearchableText")
    input.value = search_terms

    # Submit the search form
    form.submit(u"Search")

Selecting a checkbox

Checkboxes are usually presented as name:list style names:

checkbox = form.getControl(name="myitem.select:list")
checkbox.value = [u"selected"]

Clicking a button

Example:

button = form.getControl(name="mybuttonname")
button.click()

If you have a form instance, you can use the submit action. To click on the Button labeled "Log in" in the login form, you do:

login_form = self.browser.getForm('login_form')
login_form.submit('Log in')

Checking Unauthorized response

Example:

def checkIsUnauthorized(self, url):
    """
    Check whether URL gives Unauthorized response.
    """

    import urllib2

    # Disable redirect on security error
    self.portal.acl_users.credentials_cookie_auth.login_path = ""

    # Unfuse exception tracking for debugging
    # as set up in afterSetUp()
    self.browser.handleErrors = True

    def raising(self, info):
        pass
    self.portal.error_log._ignored_exceptions = ("Unauthorized")
    from Products.SiteErrorLog.SiteErrorLog import SiteErrorLog
    SiteErrorLog.raising = raising

    try:
        self.browser.open(url)
        raise AssertionError("No Unauthorized risen:" + url)
    except urllib2.HTTPError,  e:
        # Mechanize, the engine under testbrowser
        # uses urlllib2 and will raise this exception
        self.assertEqual(e.code, 401, "Got HTTP response code:" + str(e.code))

Another example where test browser / Zope 2 publisher where invalidly handling Unauthorized exception:

def test_anon_access_forum(self):
    """
    Anonymous users should not be able to open the forum page.
    """

    self.portal.error_log._ignored_exceptions = ()
    self.portal.acl_users.credentials_cookie_auth.login_path = ""

    exception = None
    try:
        self.browser.open(self.portal.intranet.forum.absolute_url())
    except:
        # Handle a broken case where
        # test browser spits out an exception without a base class (WTF)
        import sys
        exception = sys.exc_info()[0]

    self.assertFalse(exception is None)

Checking a HTTP response header

Exaple:

self.assertEqual(self.browser.headers["Content-type"], 'application/octet-stream')

Checking HTTP exception

Example how to check for HTTP 500 Internal Server Error:

def test_no_language(self):
    """ Check that language parameter is needed and nothing is executed unless it is given. """

    from urllib2 import HTTPError
    try:
        self.browser.handleErrors = True # Don't get HTTP 500 pages
        url = self.portal.absolute_url() + "/@@mobile_sitemap?mode=mobile"
        self.browser.open(url)
        # should cause HTTPError: HTTP Error 500: Internal Server Error
        raise AssertionError("Should be never reached")
    except HTTPError, e:
        pass

Setting test browser headers

Headers must be passed to underlying PublisherMechanizeBrowser instance and test browser must be constructed based on this instance.

Note

When passing parameters to PublisherMechanizeBrowser.addheaders HTTP prefix will be automatically added to header name.

Add header to browser

>>> from Products.Five.testbrowser import Browser
>>> browser = Browser()
>>> browser.addHeader(key, value)

Setting user agent

Example:

class BaseFunctionalTestCase(ptc.FunctionalTestCase):

    def setUA(self, user_agent):
        """
        Create zope.testbrowser Browser with a specific user agent.
        """

        # Be sure to use Products.Five.testbrowser here
        self.browser = UABrowser(user_agent)
        self.browser.handleErrors = False # Don't get HTTP 500 pages

from zope.testbrowser import browser
from Products.Five.testbrowser import PublisherHTTPHandler
from Products.Five.testbrowser import PublisherMechanizeBrowser

class UABrowser(browser.Browser):
    """A Zope ``testbrowser`` Browser that uses the Zope Publisher.

    The instance must set a custom user agent string.
    """

    def __init__(self, user_agent, url=None):

        mech_browser = PublisherMechanizeBrowser()
        mech_browser.addheaders = [("User-agent", user_agent),]

        # override the http handler class
        mech_browser.handler_classes["http"] = PublisherHTTPHandler
        browser.Browser.__init__(self, url=url, mech_browser=mech_browser)

For more information, see