Retrieves proxies from HideMyAss.com

hma_proxies.rb

# redka => thisredoned@gmail.com
# 24.08.2012

require 'typhoeus'
require 'nokogiri'

def proxies
  front = Typhoeus::Request.get(url = "http://www.hidemyass.com/proxy-list/").body
  hydra = Typhoeus::Hydra.new
  requests = []
  Nokogiri::HTML(front).xpath("//div[@class='pagination']/li")[-2].children[-1].values[0][/\d+/].to_i.downto(2) do |i|
    r = Typhoeus::Request.new "#{url}#{i}"
    requests << r
    hydra.queue r
  end
  hydra.run
  proxies = [front, *requests.map{ |x| x.response.body } ].map do |page|
    Nokogiri::HTML(page).xpath("//table[@id='listtable']/tr").map do |tr|
      styles = tr.css('td style')[0].text.split("\n")[1..-1].map{|x|Hash[*x[1..-2].split(?{)] }.inject(&:merge)
      tr.css("td[2] span")[0].children[1..-1].reject do |c|
        c.values.each do |value|
          styles.each { |c, d| value.sub! c, d }
        end.grep(/display.*none/)[0]
      end.map(&:text).join << ':' << tr.xpath("td")[2].text.strip
    end
  end.inject(&:+)
end