Audience Poll: How many people know and use webrat (50%+ using, 25% using)
Webrat is an awesome tool everyone should use, but it doesn’t run on the iPhone, not written in Scala, and there is no Pr0n.
If you have questions that aren’t answered, find him for a beer after the conference
Writing a rails integration test how we test websites, traditional methods are brittle and tedious.
Webrat code is more readable than traditionally written integration tests
It behaves like a browser:
- “visit” method – the equivalent typing in a URL to the browser.
- “click_link” – like clicking a link (duh)
- “fill_in” – fills in a field
- “assert_contain” – perform assertions
Uses nokogiri to parse the response HTML and run assertions on it.
How to use it?
#config/environment.rb config.gem "webrat", :version => ">= 0.4.4" #test/test_helper.rb Webrat.configure do |config| config.mode = :rails end
Webrat’s Core API
- check and uncheck
Webrat works with RSpec, Shoulda and Cucumber
Webrat and cucumber do not have to be used together. You can use Webrat with whatever testing framework you want to use.
Webrat also works with Rails, Sinatra and Merb.
Webrat can do input matching to labels. Webrat encourages you to put good labels on your form fields.
Webrat verifies HTTP Status codes for you behind the scenes.
- Automatically makes sure you don’t get 500 HTTP server errors.
- Also ensures that fields exist
Verify HTML content
response.should contain("Hello, world!") response.should have_selector("li", :class => "new", :count => 2) response.should_not have_xpath(".//meta[@name='robots']") response.should have_selector("#album li:nth-child(3)") do |li| li.should have_selector("img", :src => photo_path(@photo)) li.should contain("Vacation Photo") end
When things go wrong Webrat uses ”’save_and_open_page”’ and writes out the current response body and brings up the page.
- Mechanize (can use to scrape other websites i.e. google)
Evil plot is to write adapters so he can test everything at once.
# selenium doesn't support transactional fixtures class ActiveSupport::TestCase #... self.use_transactional_fixtures = false #... end setup do |session| session_host "localhost:3000" end Webrat.configure do |config| config.mode = :selenium end
Webrat talks to the selenium client gem to talk to the Selenium RC server which then drives a browser (IE, safari, and/or firefox).
Sometimes you’ll need to an action differently when a browser is actually present. These commands let you write code for a specific context:
webrat.simulate – traditional webrat simulation mode
webrate.automate – drive a real browser
# is not run in normal webrat mode webrat.automate do end # is not run in selenium webrat mode webrat.simulate do ... end
The ‘selenium’ object is exposed from the selenium client gem. You can use it directly within your tests.
Automating a real web browser is SLOW
One more thing…
rack::test – gives you a quick way to generate requests to any rack enabled application
Makes it easy to test complicated routing of requests through rack apps. For example, a rails metal that routes to a sinatra app.
post, put, delete head
- Q: How do you deal with links with the same names?
A: Webrat has a method called “within” which allows you to scope your selectors to be inside an HTML tag.
Q: Does it have support for checking HTML validity?
A: No, that is outside of the scope of core webrat. However, we have talked about the ability to capture
Q: Where does the output of save_and_open_page go?
A: Into Rails’ tmp directory
Q: Selenium client has some interesting screen grab is that being exposed?
- A: Webrat has that in Selenium mode now.