Radiant wiki spam cleaner

spam_link_checker.rb

require 'rubygems'
require 'mechanize'

def process_links links_block
  
  url_regex = /http:\/\/[^"\/\s]+/
  agent = Mechanize.new
  
  clean_links = []
  possible_spam = []

  links_block.split("\n").each do |line|
    begin
      next unless url = url_regex.match(line)
      links_to_radiant = agent.get("#{url}/admin/login").links_with(:href => "http://radiantcms.org")
      suspect =  links_to_radiant.size < 1
      puts "No links to radiant found at #{url}/admin/login" if suspect
    rescue Exception => e
      puts "Error for #{url} - #{e}"
      # probably a 404
      suspect = true
    end
    if suspect
      possible_spam << line.gsub!(/\s*\* (\*\S)?/, "* *spam(auto)? ")
    else
      clean_links << line
    end
  end
  
  puts "\n\n"

  puts "Clean Links:"
  puts clean_links

  puts "\n\n"

  puts "Possible Spam:"
  puts possible_spam
end


links_block = %q{
   * "Edmund Haselwanter":http://edmund.haselwanter.com on "edmund.haselwanter.com":http://edmund.haselwanter.com, "apart-elisabeth.miemingerplateau.at":http://apart-elisabeth.miemingerplateau.at, "miemingerplateau.at":http://www.miemingerplateau.at, "ecoSyn365":http://www.ecosyn365.com/,"The Cakemaker":http://www.the-cakemaker.at
   * "Elastique":http://elastique.com.au &amp; "Otilas":http://otilas.com.au on "Beehive App":http://beehiveapp.com
   * Elevated Rails on "elevatedrails.com":http://www.elevatedrails.com
   * Elliot on "Telephone Answering Services":http://www.telephoneansweringservice.org.uk
   * Emule on "inchirieri masini":http://www.rentngo.ro
   * "Empire: Total War Heaven":http://etw.heavengames.com/ on "etw.heavengames.com":http://etw.heavengames.com
   * Erik Mallinson on "mallinson.us":http://mallinson.us, "coacalina.org":http://coacalina.org, and "malliliprints.com":http://malliliprints.com
   * Erik van Oosten on "theatergroepzout.nl":http://www.theatergroepzout.nl, and "milo.grons.nl":http://milo.grons.nl
   * "Essence of the tree":http://www.essenceofthetree.com Nursery site designed by Will Emerson of "Livelihood Technology":http://www.livelihoodtech.com
   * Expex China on "expexchina.com":http://www.expexchina.com
   * "Fab:IT":http://www.fab-it.dk/ on  "VPSForce":http://vpsforce.eu/
   * "Farmers Market":http://www.farmersmarket.is
}

process_links links_block