My first Ruby script

wp_version.rb

#!/usr/bin/ruby

require 'rubygems'
require 'nokogiri'
require 'open-uri'
require 'uri'

# Chrome on Win7
USERAGENT= "Mozilla/5.0 (Windows; U; Windows NT 6.1; en-US) AppleWebKit/532.0 (KHTML, like Gecko) Chrome/3.0.195.10 Safari/532.0"

#todo
# || and ||= to clean up nil checking code
# #{@name} to clean up .to_s
# mechanize has user-agent aliases

# This was my first Ruby script ever. It attempts to identify
# Wordpress versions from a variety of sources. It was used to generate
# statistics about the percentage of Security Twits that regularly
# follow their own advice and patch.

class Blog
  attr_accessor :name, :major, :minor, :revision

  def initialize(name=nil, major=nil, minor=nil, revision=nil)
    if name == nil then @name = "unknown"
    else @name = name end

    if major == nil then @major = 0
    else @major = major end

    if minor == nil then @minor = 0
    else @minor = minor end

    if revision == nil then @revision = 0
    else @revision = revision end
  end

  def to_s
    if @name === "WordPress"
      @name.to_s + " " + @major.to_s + "." + @minor.to_s + "." + @revision.to_s
    else
      @name
    end
  end

  def version()
    @major.to_s + "." + @minor.to_s + "." + @revision.to_s
  end

  # if its not one of these values, chances are its modified
  def legit()
    if @major === 2 && (0..9) === @minor && (0..9) === @revision
      return true #legit
    else
      return false #not legit
    end
  end

  #def <, >, ==, etc

end

def use_metatag(url)
  begin
    html = open(url, "User-Agent" => USERAGENT)
  rescue
    return Blog.new("the website is down")
  rescue Timeout::Error
    return Blog.new("the website is down")
  end

  doc = Nokogiri::HTML(html)
  node = doc.search('//meta[@name=\'generator\']')

    if node.empty? #no meta generator tag
    return Blog.new("unknown")
  else #we got a meta tag
    version_string = node.attr('content')
    if version_string === "WordPress.com" #it said wordpress.com
      return Blog.new("WordPress.com")
    else
      (text, num) = version_string.split(' ') #try splitting it
      if text === "WordPress" && num != nil
        return Blog.new(text, *num.split('.').map { |s| s.to_i } ) #it said wordpress
      else
        return Blog.new(version_string) #it said something else
      end
    end
  end
end

def use_readme(url)
  begin
    doc = Nokogiri::HTML(open(url + "/readme.html", "User-Agent" => USERAGENT))
  rescue
    return Blog.new("readme removed")
  rescue Timeout::Error
    return Blog.new("timeout")
  end

  node = doc.search('//h1[@id=\'logo\']')

  if node.empty?
    return Blog.new("readme modified")
  end

  (text, num) = node.inner_text.strip().split() # "Version 2.8"
  return Blog.new("WordPress", *num.split('.').map { |s| s.to_i } )
end

#def wp_version(url="http://192.168.1.103/wordpress-2.8.1/")
  #metatag_says = use_metatag(url)
  #puts metatag_says

  #if metatag_says.name != "WordPress"
  #  return metatag_says #stop here
  #else
  #  readme_says = use_readme(url)

    #puts "Got " + metatag_says.to_s + " from meta tag"
    #puts "Got " + readme_says.to_s + " from readme.html"
    #puts "Major: " + readme_says.major.to_s
    #puts "Minor: " + readme_says.minor.to_s
    #puts "Revision: " + readme_says.revision.to_s

    #if metatag_says.legit() and metatag_says.version === readme_says.version
    #  #puts "both answers match"
    #  return metatag_says
    #elsif readme_says.legit()
    #  #puts "readme is legit"
    #  return readme_says
    #else
    #  #puts "readme is not legit"
    #  return metatag_says
    #end
  #end
#end

#if ARGV[0] then
#puts wp_version(ARGV[0])

#else

doc = Nokogiri::HTML(open("http://www.security-twits.com/"))
links = doc.css('a').map { |link| link['href'] }

for link in links
  if link.chomp().empty? or link =~ /twitter/i or link =~ /freenode/
   links.delete(link)
  end
end

puts links.count.to_s + " blogs to identify"
puts "url,metatag,readme"

for link in links
  puts link + "," + use_metatag(link).to_s + "," + use_readme(link).to_s
end

#end