This is a note to make a point on our (mine and my team’s) current use of Selenium to test the ajax behaviour in the Rails webapp we’re currently developing. Ajax replacing of part of the page is growing, and with it we have to face the classical question: “how do we test (I mean automatically :-) the ajax/javascript behaviours in our webapp?”.
This is how we are trying to manage this issue now, after some days of spiking on Selenium, Watir and BlueRidge (I hope to write more on Watir and BlueRidge in some future post, because these two tools are worth speaking).
Actually we are giving a try to the combination of Webrat + Selenium, since we already have a big test suite of integration test using Webrat, and have a good knowledge of the Webrat API.
We added the selenium-client gem to be able to drive Selenium through the Webrat API.
This is extracted from our test environment configuration file:
Then, we defined a class from which all the selenium test cases will inherit.
This class basically is used to
We also added a rake task to be able to launch all the selenium tests
One thing we learned through several repeated mistakes is that the Webrat API is different when called in the “selenium” mode then the one we were used to when using Webrat in the classical “rails” mode.
For example, the “assert_have_selector” method for selenium only takes one argument, that is the CSS selector, while in the classical webrat mode, the same method takes another parameter to specify the expected content to match with (see this rdoc: http://gitrdoc.com/brynary/webrat/tree/master). So we had to define helper methods based on “assert_have_xpath” method using xpath to express the same intent of a method like assert_have_selector(css_selector, expected_content)
Here is our helper method
test.rb
... config.gem 'selenium-client', :lib => 'selenium/client' config.gem "webrat", :version => '>= 0.6.0' ...
- disable the transactional fixtures in Rails, to allow the browser process where Selenium runs to access the data prepared in the tests
- configure Webrat with the “selenium” mode
- be the place to collect helper methods as “login” or “logout”, used in many tests.
selenium_integration_test.rb
class SeleniumIntegrationTest < ActionController::IntegrationTest self.use_transactional_fixtures = false setup :switch_webrat_to_selenium def switch_webrat_to_selenium Webrat.configure do |config| config.mode = :selenium config.application_environment = :test end selenium.set_speed(100) # default is 0 ms selenium.set_timeout(10000) # default is 30000 ms end teardown :delete_cookies def delete_cookies selenium.delete_all_visible_cookies end protected ... [other helper methods here, like login, logout, and so on...] ...
test.rake
namespace :test do ... ... desc "Run Selenium Test" Rake::TestTask.new(:selenium) do |t| t.libs << "test" t.test_files = FileList['test/selenium/*test.rb'] t.verbose = true end end
selenium_integration_test.rb
... def assert_has_id id, text_content assert_have_xpath "//*[@id='#{id}'][1][text()='#{text_content}']" end ...