Capybara roundup | The Ruby Rails Aggregator

Capybara roundup

Here is a roundup of recent changes in and around
Capybara. Where they apply to the
Capybara source, they are not included in any release yet, so make sure you are
using the edge version from GitHub (and be prepared for things to break here
and there ^_^).

The Celerity and Culerity drivers have been extracted into the capybara-celerity and capybara-culerity gems (5c4f28, by Gabriel Sobrinho). They will be officially supported until (at least) Capybara 1.0, but Jonas recommends using Akephalos as your HtmlUnit driver.
For RSpec, Capybara now supports :type => :request example groups, as suggested by the rspec-rails gem (558e7b, by yours truly). See the README, section “Using Capybara with RSpec”.
There is now an (optional) acceptance test DSL for RSpec (f4897f, by Jonas Nicklas and Nicklas Ramhöj). See the README (section “Using Capybara with RSpec”) or Jeff Kreeftmeijer’s blog post.
The Selenium driver now waits for Ajax requests (021b87, by Maurizio Giambalvo, Jonas, and Nicklas). Search the README for :resynchronize for more info.
Capybara now prefers visible over hidden elements (408423, by Jonas and Nicklas, configurable with Capybara.prefer_visible_elements).
The RSpec matchers, which are responsible for giving you nice error messages (and allow you to write should_not have_css without breaking asynchronous JavaScript, though that’s not documented yet), have been improved and are now used in Cucumber as well. Jonas says they might still change a bit, so don’t rely on them too much.
John Firebaugh wrote a capybara-firebug gem to run Capybara with Firebug enabled.

Miscellaneous tips and tricks from around the web:

Hugo Baraúna of Plataforma suggests that you improve your tests by using Capybara’s custom selectors.
Carlos Antônio of Plataforma demonstrates how to configure the User-Agent header with Capybara and Selenium.
Sara Trice shows how to test CKEditor in your pages.
Jason Neylon has some example code for selecting from a dropdown when you don’t care about its contents.
Hiroshi Saito suggests a hack to use transactional fixtures with the Akephalos driver.
Jari Bakken shows how to take screenshots with the Selenium driver.
Leonid Shevtsov started a discussion on preloading the browser in Spork for Selenium, but we were not able to iron out all the kinks to make it usable. It would be most awesome to get this working, so if anyone has an idea, please post it!

Anything else? Leave a comment!

P.S. Thanks to Elabs for continuing to put
company resources into Capybara! Did you know you can hire them to do agile
web development for you? (Full disclosure: I’m not affiliated.)

Read more at the source

Details

When reading text nodes from an XML document, the REXML parser can be coerced into allocating extremely large string objects which can consume all of the memory on a machine, causing a denial of service.

Impacted code will look something like this:

require 'rexml/document'

xml = <<XML
<!DOCTYPE root [
  # ENTITY expansion vector
]>
<cd></cd>
XML

p REXML::Document.new(xml)

All users running an affected release should either upgrade or use one of the workarounds immediately.

Workarounds

If you cannot upgrade Ruby, use this monkey patch as a workaround on versions of Ruby 2.1.0+:

class REXML::Entity
  def value
      if @value
        matches = @value.scan(PEREFERENCE_RE)
        rv = @value.clone
        if @parent
          sum = 0
          matches.each do |entity_reference|
            entity_value = @parent.entity( entity_reference[0] )
            if sum + entity_value.bytesize > Security.entity_expansion_text_limit
              raise "entity expansion has grown too large"
            else
              sum += entity_value.bytesize
            end
            rv.gsub!( /%#{entity_reference.join};/um, entity_value )
          end
        end
        return rv
      end
      nil
   end
end

For versions of Ruby older than 2.1.0, you can use the following monkey patch:

class REXML::Entity
  def value
      if @value
        matches = @value.scan(PEREFERENCE_RE)
        rv = @value.clone
        if @parent
          sum = 0
          matches.each do |entity_reference|
            entity_value = @parent.entity( entity_reference[0] )
            if sum + entity_value.bytesize > Document.entity_expansion_text_limit
              raise "entity expansion has grown too large"
            else
              sum += entity_value.bytesize
            end
            rv.gsub!( /%#{entity_reference.join};/um, entity_value )
          end
        end
        return rv
      end
      nil
   end
end