Naively spiders a website to extract HTML style attributes and make a CSS stylesheet

css-snowball.rb

#!/usr/bin/env ruby

require "rubygems"
require "spidr"
require "open-uri"

=begin
  = CSS snowball
  Rolls through your site, picking up style attributes and making you 
  a new stylesheet. Once you've got your stylesheet, run it through
  something like http://www.cleancss.com/ to clean it up.

  Outputs a CSS file and an instruction sheet.
=end

if ARGV.empty?
  puts "CSS-Snowball - rolls through your site picking up styles"
  puts "Usage: #{__FILE__} http://www.example > output.css"
  exit
end

count = 1
styles = []

Spidr.site(ARGV[0]) do |spider|
  spider.every_page do |page|

    #url = page.url
    body = page.search('body').first
  
    if body.nil?
      # probably not an HTML file, or a malformed one
      next
    else

      style_attributes = body.search('[style]')

      style_attributes.each do |s|
        styles << s['style'].chomp(' ')
      end
  end
end
end  

styles.uniq.each do |s|
  puts ".snowball#{count} {"
    s.split(';').each do |att|
      puts "  #{att};"
    end
  puts "}\n"
  count = count +  1
end

Great script. Found a link to it on a SO question. Thanks for sharing!