#!/usr/bin/env ruby
require 'rubygems'
gem 'hpricot', '>=0.6.170'
require 'open-uri'
require 'hpricot'

gem 'nokogiri', '>1.0.5'
require 'nokogiri'
require 'benchmark'

xml_content = open("http://railstips.org/assets/2008/8/9/timeline.xml").read
html_content = open("http://slashdot.org/").read

N = 500

hdoc_xml = Hpricot.XML(xml_content)
hdoc_xml2 = Hpricot.scan(xml_content)
hdoc_html = Hpricot(html_content)
hdoc_html2 = Hpricot.scan(html_content)

ndoc_compat = Nokogiri.Hpricot(xml_content)
ndoc_xml = Nokogiri::XML(xml_content)
ndoc_html = Nokogiri::HTML(html_content)


# XML Document Parsing
puts "XML Document parsing benchmark"
Benchmark.bm(25) do |x|
  x.report('hpricot:xml:doc') do
    N.times do
      Hpricot.XML(xml_content)
    end
  end

  x.report('hpricot2:xml:doc') do
    N.times do
      Hpricot.scan(xml_content)
    end
  end

  x.report('nokogiri:compat:doc') do
    N.times do
      Nokogiri.Hpricot(xml_content)
    end
  end

  x.report('nokogiri:xml:doc') do
    N.times do
      Nokogiri::XML(xml_content)
    end
  end
end

xpath1 = '//status/text'
xpath2 = '//user/name'
puts
puts "XML XPath benchmarks (#{xpath1}, #{xpath2})"
# XML XPath benchmarks
Benchmark.bm(25) do |x|
  x.report('hpricot:xml:xpath') do
    N.times do
      info = hdoc_xml.search(xpath1).first.inner_text
      url = hdoc_xml.search(xpath2).first.inner_text
    end
  end

  x.report('hpricot2:xml:xpath') do
    N.times do
      info = hdoc_xml2.search(xpath1).first.inner_text
      url = hdoc_xml2.search(xpath2).first.inner_text
    end
  end

  x.report('nokogiri:compat:xpath') do
    N.times do
      info = ndoc_compat.search(xpath1).first.inner_text
      url = ndoc_compat.search(xpath2).first.inner_text
    end
  end

  x.report('nokogiri:xml:xpath') do
    N.times do
      info = ndoc_xml.search(xpath1).first.inner_text
      url = ndoc_xml.search(xpath2).first.inner_text
    end
  end
end

# XML CSS benchmarks
css1 = 'status text'
css2 = 'user name'
puts
puts "XML CSS selector benchmarks (#{css1}, #{css2})"
Benchmark.bm(25) do |x|
  x.report('hpricot:xml:css') do
    N.times do
      info = hdoc_xml.search(css1).first.inner_text
      url = hdoc_xml.search(css2).first.inner_text
    end
  end

  x.report('hpricot2:xml:css') do
    N.times do
      info = hdoc_xml2.search(css1).first.inner_text
      url = hdoc_xml2.search(css2).first.inner_text
    end
  end

  x.report('nokogiri:compat:css') do
    N.times do
      info = ndoc_compat.search(css1).first.inner_text
      url = ndoc_compat.search(css2).first.inner_text
    end
  end

  x.report('nokogiri:xml:css') do
    N.times do
      info = ndoc_xml.search(css1).first.inner_text
      url = ndoc_xml.search(css2).first.inner_text
    end
  end
end

# HTML Document Parsing
puts
puts "HTML Document parsing benchmark (slashdot.org)"
Benchmark.bm(25) do |x|
  x.report('hpricot:html:doc') do
    N.times do
      Hpricot(html_content)
    end
  end

  x.report('hpricot2:html:doc') do
    N.times do
      Hpricot.scan(html_content)
    end
  end

  x.report('nokogiri:html:doc') do
    N.times do
      Nokogiri::HTML(html_content)
    end
  end
end

# HTML XPath benchmarks
html_xpath = '//h1/a'
puts
puts "HTML XPath benchmarks (#{html_xpath})"
Benchmark.bm(25) do |x|
  x.report('hpricot:html:xpath') do
    N.times do
      info = hdoc_html.search(html_xpath).first.inner_text
    end
  end

  # Fail
  #x.report('hpricot2:html:xpath') do
  #  N.times do
  #    info = hdoc_html2.search(html_xpath).first.inner_text
  #  end
  #end
  #
  x.report('nokogiri:html:xpath') do
    N.times do
      info = ndoc_html.search(html_xpath).first.inner_text
    end
  end
end

# EASY HTML CSS benchmarks
html_css = 'h1 > a'
puts
puts "Easy HTML CSS benchmarks (#{html_css})"
Benchmark.bm(25) do |x|
  x.report('hpricot:html:css') do
    N.times do
      info = hdoc_html.search(html_css).first.inner_text
    end
  end

  # Fail
  #x.report('hpricot2:html:css') do
  #  N.times do
  #    info = hdoc_html2.search(html_css).first.inner_text
  #  end
  #end

  x.report('nokogiri:html:css') do
    N.times do
      info = ndoc_html.search(html_css).first.inner_text
    end
  end
end

# HARD HTML CSS benchmarks
hard_html_css = 'div#logo > h1 > a'
puts
puts "Hard HTML CSS benchmarks (#{hard_html_css})"
Benchmark.bm(25) do |x|
  x.report('hpricot:html:css') do
    N.times do
      info = hdoc_html.search(hard_html_css).first.inner_text
    end
  end

  # Fail
  #x.report('hpricot2:html:css') do
  #  N.times do
  #    info = hdoc_html2.search(hard_html_css).first.inner_text
  #  end
  #end

  x.report('nokogiri:html:css') do
    N.times do
      info = ndoc_html.search(hard_html_css).first.inner_text
    end
  end
end