Ruby CSS Coverage Tester

css_coverage.rb

#!/opt/local/bin/ruby

# Copyright (c) 2007 Dan Kubb <dan.kubb@gmail.com>

# This tool shows any CSS rules that do not apply to HTML.  Accepts a single
# argument on the command line, may be either a URI or an HTML file.

require 'rubygems'
require 'csspool'
require 'hpricot'
require 'net/https'
require 'open-uri'

STRING_MATCH       = /\s*(?:"(.+?)"|'(.+?)')\s*/
URL_PROPERTY_MATCH = /\s*url\((?:#{STRING_MATCH}|(.+?))\)/
AT_CHARSET_MATCH   = /\s*@charset#{STRING_MATCH};/
AT_IMPORT_MATCH    = /@import\s(?:#{URL_PROPERTY_MATCH}|#{STRING_MATCH}).*?;/

module Net
  class HTTP
    # allow testing of sites without valid certificates
    def verify_mode=(verify_mode)
    end
  end
end

module OpenURI
  class << self
    # allow http and https URIs to redirect to themselves or each other
    def redirectable_with_https?(uri1, uri2)
      redirectable_without_https?(uri1, uri2) || ([ uri1.scheme, uri2.scheme ] - %w[ http https ]).empty?
    end

    alias_method :redirectable_without_https?, :redirectable?
    alias_method :redirectable?, :redirectable_with_https?
  end
end

open(ARGV.shift) do |file|
  base = file.respond_to?(:base_uri) ? file.base_uri : ''

  html_doc = Hpricot file

  # find all linked and embedded stylesheets
  css_source = html_doc.search('link[@rel][@href]|style').map do |tag|
    case tag.name
      when 'link'  : open(base + tag[:href]).read if tag[:rel].downcase == 'stylesheet'  # xpath/css cannot match case insensitively
      when 'style' : tag.inner_html
    end
  end * ''

  # recursively load @import stylesheets and insert them inline
  1 while css_source.gsub!(AT_IMPORT_MATCH) do
    open(base + Regexp.last_match.captures.compact.first).read.gsub(AT_CHARSET_MATCH, '')  # strip embedded @charset at-rules
  end

  css_doc = CSS::SAC::Parser.new.parse css_source

  # Find all CSS rules an HTML page does not use, minus psuedo-class selectors
  # Output from most to least specific rules and show the specificity and name
  (css_doc.rules - css_doc.find_all_rules_matching(html_doc)).sort.reverse.each do |rule|
    next if rule.selector.respond_to?(:condition) && rule.selector.condition.kind_of?(CSS::SAC::Conditions::PseudoClassCondition)
    puts "#{rule.selector.specificity.inspect}: #{rule.selector.to_css}"
  end
end