Benchmark sax vs dom parsing for nokogiri

sax_vs_dom.rb

require 'nokogiri'

class MyDoc < Nokogiri::XML::SAX::Document
  attr_reader :result

  def initialize
    @result = ""
  end

  def xmldecl version, encoding, standalone
    @result << version
    @result << encoding
    @result << standalone
  end

  def comment string
    @result << string
  end

  def start_element name, attributes = []
    b = Hash[attributes]

    if name == 'a' || name == 'link'
      url = b['href'] || ""
      if 'http' == url[0..3]
        b['href'] = b['href'].upcase
      end
    end


    attrs = b.collect { |k,v| "#{k}=\"#{v}\"" }.join(' ')

    if attrs.empty?
      @result << "<#{name}>"
    else
      @result << "<#{name} #{attrs}>"
    end
  end

  def end_element name
    @result << "</#{name}>"
  end

  def cdata_block string
    @result << string
  end

  def characters str
    @result << str
  end

  def processing_instruction name, content
  end

  def start_element_namespace name, attrs = [], prefix = nil, uri = nil, ns = []
  end

  def warning(string)
  end

  def error(string)
  end
end


require 'open-uri'
require "benchmark"

include Benchmark

N = 1

diffs = []

File.open("urls.txt").each do |url|
  begin
    str = open(url).read

    Benchmark.benchmark(CAPTION, 7, FORMAT, ">% Change:") do |x|
      t_sax = x.report("SAX") do
         N.times do
          parser = Nokogiri::HTML::SAX::Parser.new(MyDoc.new)
          parser.parse(str)
        end
      end

      t_dom = x.report("DOM") do
        N.times do
          result = Nokogiri::HTML(str)

          attribute_name = 'href'
          result.css("link[#{attribute_name}]").each do |element|
            if 'http' == url[0..3]
              element.attributes[attribute_name].value = element.attributes[attribute_name].value.upcase
            end
          end

          result.to_s
        end
      end

      sax_real = t_sax.to_a.last
      dom_real = t_dom.to_a.last

      diffs << ( (sax_real - dom_real) / dom_real ) * 100

      [( (t_sax - t_dom) / t_dom ) * 100]
    end

  rescue
    puts $!, $@
  end
end

puts 'end'

diffs.each { |diff| puts "#{diff}\n"}

urls.txt

http://www.ruby-doc.org/
http://en.wikipedia.org/wiki/Pickling