Deadweight is a CSS coverage tool. Given a set of stylesheets and a set of URLs, it determines which selectors are actually used and reports which can be "safely" deleted.

Ryan Bates has worked his magic once again. Head over here for an excellent introduction to Deadweight.

There are multiple ways to use Deadweight. It's designed to be completely agnostic of whatever framework (or, indeed language) your website is built on, but optimised for Ruby and Rails.

I'm going to tell you about the coolest way to use it first.

Deadweight can hijack your Rails integration tests (both the spartan Test::Unit type and the refreshing Cucumber variety), capturing every page that is returned by your app during testing and saving you the trouble of manually specifying a ton of URLs and login processses. THIS IS PRETTY COOL YOU SHOULD TRY IT. (It's also somewhat new and untested, but you're the kind of person who just loves it out here on the edge, I can tell.)

Here's how it works. First, put this in your Gemfile:

group :test do
  gem 'colored' # optional, in the same way that dressing respectably when you leave the house is 'optional'
  gem 'deadweight', :require => 'deadweight/hijack/rails'
end

Then, run your integration tests with the environment variable DEADWEIGHT set to true:

rake test DEADWEIGHT=true
rake cucumber DEADWEIGHT=true

Let me know how it goes. It's not terribly customisable at the moment (you can't specify what exact stylesheets to look at, or what selectors to ignore). I'm looking for your feedback on how you'd like to be able to do that.

# lib/tasks/deadweight.rake

require 'deadweight'

Deadweight::RakeTask.new do |dw|
  dw.stylesheets = %w( /stylesheets/style.css )
  dw.pages = %w( / /page/1 /about )
end

Running rake deadweight will output all unused rules, one per line. Note that it looks at http://localhost:3000 by default, so you'll need to have script/server (or whatever your server command looks like) running.

$ deadweight -s styles.css -s ie.css index.html about.html
$ deadweight -s http://www.tigerbloodwins.com/index.css http://www.tigerbloodwins.com/
$ deadweight --root http://kottke.org/ -s '/templates/2009/css.php?p=mac' / /everfresh /about

You can pipe in CSS rules from STDIN:

$ cat styles.css | deadweight index.html

And you can use it as an HTTP proxy:

$ deadweight -l deadweight.log -s styles.css -w http://github.com/ -P

require 'deadweight'

dw = Deadweight.new
dw.stylesheets = %w( /stylesheets/style.css )
dw.pages = %w( / /page/1 /about )
puts dw.run

By default, Deadweight uses http://localhost:3000 as the base URL for all paths. To change it, set root:

dw.root = "http://staging.example.com"      # staging server
dw.root = "http://example.com/staging-area" # urls can have paths in
dw.root = "/path/to/some/html"              # local paths work too

Deadweight is completely dumb about any classes, IDs or tags that are only added by your Javascript layer, but you can filter them out by setting ignore_selectors:

dw.ignore_selectors = /hover|lightbox|superimposed_kittens/

The command-line tool also has basic support for Lyndon with the -L flag, which simply pipes all HTML through the lyndon executable.

Set mechanize to true and add a Proc to pages (rather than a String), and Deadweight will execute it using Mechanize:

dw.mechanize = true

# go through the login form to get to a protected URL
dw.pages << proc {
  fetch('/login')
  form = agent.page.forms.first
  form.username = 'username'
  form.password = 'password'
  agent.submit(form)
  fetch('/secret-page')
}

# use HTTP basic auth
dw.pages << proc {
  agent.auth('username', 'password')
  fetch('/other-secret-page')
}

The agent method returns the Mechanize instance. The fetch method is a wrapper around agent.get that will abort in the event of an HTTP error status.

gem install colored

Copyright (c) 2009 Aanand Prasad. See LICENSE for details.