spider script for testing site's broken links

spider.rb

#!/usr/bin/env ruby
ENV['TZ'] = 'UTC'

require 'rubygems'
require 'active_support/core_ext/object/blank'
require 'nokogiri'
require 'net/http'
require 'byebug'
require 'logger'
require 'uri'

alias p puts
def h(i); Nokogiri::HTML(i.strip); end

$stdout.sync = true

visited = []
error_counter = 0
urls = [ARGV.first.to_s]

`rm errors.log access.log ignored.log &>/dev/null`

error = Logger.new("errors.log")
access = Logger.new("access.log")
ignored = Logger.new("ignored.log")

while (current = urls.pop)
  visited.push(current)
  uri = URI.parse(current)
  base = "#{uri.scheme}://#{uri.host}:#{uri.port}"
  response = Net::HTTP.get_response(uri)
  if response.code.to_i / 100 == 2
    print "\e[32m.\e[m"
    access.info current
    html = h response.body
    html.css("a").each do |link|
      link = link[:href]
      next if link.blank?
      if link.start_with?('/') and not(link.start_with?('//'))
        link = "#{base}#{link}"
      else
        ignored.info link
        next
      end
      urls.push(link) unless visited.include? link
    end
  else
    if response.code.to_i / 100 == 3
      print "\e[33mR\e[m"
      error.info "\e[33m[R#{response.code}]\e[m > #{current}"
    else
      error_counter += 1
      print "\e[31mE\e[m"
      error.error "\e[31m[E#{response.code}]\e[m > #{current}"
    end
  end
end

puts "\n\nprocessed: #{visited.count} links, \e[31merrors: #{error_counter}\e[m, \e[32msuccess: #{visited.count - error_counter}\e[m"