Html Test

A plugin for HTML validation and link checking in Rails functional and integration tests. In addition you can run the HTML and link tests against a production or staging server over HTTP.

Installation

ruby script/plugin install git://github.com/peter/html_test.git

Usage

• Add validation to your controller or integration tests:

  get :some_action assert_response :success # The following validates using all three validators: tidy, w3c, xmllint # To use a specific validator, use one of assert_tidy, assert_w3c, and # assert_xmllint. assert_validates

• Validate all requests in your controller and integration tests by adding these lines to your RAILS_ROOT/test/test_helper.rb file:

  ApplicationController.validate_all = true ApplicationController.validators = [:tidy, :w3c]

• Check for broken URLs (URLs that don’t resolve to a route, a controller, and an action) in your links, images, forms, and redirects. Works both in your controller and integration tests. Add these lines to your RAILS_ROOT/test/test_helper.rb file:

  ApplicationController.check_urls = true ApplicationController.check_redirects = true

• Follow links in your controller or integration tests:

  include Html::Test::UrlSelector
  
  def clicks_around
    page_url = request.request_uri
    anchor_urls.reject { |url| external_http?(url) || skip_url?(url) || url == page_url }.
      reject { |url| url =~ /(logout|signout)/ }.uniq.each do |url|
  
      get url
      assert(@response.redirect? || @response.success?,
        "Invalid response code #{@response.response_code} for url #{url}")
    end
  end
  

• Use the link validator to validate your staging or production server over HTTP:

./vendor/plugins/html_test/script/validate my.blog.com –validators tidy,xmllint

• Configuration options for your tests

Here is a list of configuration options to use in test/test_helper.rb:

# # Validates all controller and integration test requests if set to true:
# ApplicationController.validate_all = true # Default: false.
# ApplicationController.validators = [:tidy, :w3c, :xmllint] # Default: [:tidy]. Which validators to use.
# # A list of regular expressions for Tidy warnings to ignore:
# Html::Test::Validator.tidy_ignore_list = [/<table> lacks "summary" attribute/] # Default: [].
# # Set the URL of a locally installed W3C validator here:
# Html::Test::Validator.w3c_url = "http://localhost/validator/htdocs/check" # Default: "http://validator.w3.org/check"
# Html::Test::Validator.w3c_show_source = "0" # Default: "1". Whether to list HTML document in the W3C report
# ApplicationController.check_urls = true # Default: false. Whether to check URLs in all links
# ApplicationController.check_redirects = true # Default: false. Whether to check URLs in all redirects

Validators and Dependencies

To use Tidy for HTML validation you need the rails_tidy plugin. The xmllint validation option requires xmllint to be available on the command line. The xmllint validator is part of libxml2. The W3C validator has no dependencies since we access it via the web. To install the W3C validator locally (highly recommended), see those instructions:

http://validator.w3.org/docs/install.html

On Mac OSX I currently use the following easy to install W3C validator:

http://habilis.net/validator-sac

I recommend using both the tidy and W3C validators (or xmllint if for some reason you can’t use the W3C validator). Tidy gives useful warnings for example about empty tags, but it misses obvious validation failures such as lack of HTML quoting or unclosed tags. The W3C validator offers much better reports for debugging than xmllint. The W3C HTML reports are written to disk so you can bring it up in your browser for debugging.

The xmllint validator will use the XHTML 1.0 strict DTD by default. To specify a different DTD, have the Html::Test::Validator.dtd(document) method return the path to your DTD file. If you prefer to use the DTD referenced in the DOCTYPE declaration (can be very slow) then have the dtd method return the string “doctype”.

If you are using a catch-all route for 404:s, then you need to implement the method Html::Test::UrlChecker#check_not_404(url, params) and throw an exception if that route is invoked.

There are methods such as skip_url?(url) and should_validate? that you can override to control which pages get validated and which URLs get checked.

Tests

If you make changes to this plugin, remember to run the tests with rake. The tests require the rails_tidy and mocha plugins as well as xmllint.

Author

Peter Marklund (marklunds.com)

License

Licensed under the same terms as Ruby on Rails.

Gemified with Jeweler

vi Rakefile
rake version:write

rake version:bump:patch
rake version:bump:minor
rake version:bump:major

rake gemspec

rake install
rake release

Testing

Since upgrading some apps to rails 3, I found that this gem has some issues. So, I’m wrapping the repo code in a rails 3 app solely for testing. During gem creation/installation only the necessary code will be included.

Features untested in rails 3 due to me not using them …

• the tidy validator

• the check_urls option

  • ActionController::Routing::Routes doesn’t exist in Rails 3

• also note, Test::Unit::AssertionFailedError doesn’t exist in Ruby 1.9

  • replaced all with “MiniTest::Assertion”

  • these assert_raises are wrapped around assertion that don’t have a negative

• also, the gem now adds the Gemfile dependencies to it which isn’t really true