Comparing sax-machine and xmlcodec to parse an Atom feed

sax-machine_xmlcodec_comparison.rb

#!/usr/bin/env ruby

ONGOING_URL="https://www.tbray.org/ongoing/ongoing.atom"
# For some reason net/http can't download the feed properly
XML_TEXT = IO.popen("curl -s #{ONGOING_URL}").read

## First with sax-machine
require 'sax-machine'
class SM_AtomContent
  include SAXMachine
  attribute :type
  value :text
end

class SM_AtomEntry
  include SAXMachine
  element :content, :class => SM_AtomContent
end

class SM_Atom
  include SAXMachine
  elements :entry, :as => :entries, :class => SM_AtomEntry
end

feed = SM_Atom.parse(XML_TEXT)

# sax-machine strips away the HTML tags
puts "sax-machine says: "+feed.entries.first.content.text[0..20].inspect

## The same with xmlcodec
require 'xmlcodec'
class XC_Atom < XMLCodec::XMLElement
  xmlformat 'Atom'
end

class XC_Feed < XC_Atom
  elname 'feed'
  xmlsubel_mult 'entry'
end

class XC_Entry < XC_Atom
  elname 'entry'
  xmlsubel :content
end

class XC_Content < XC_Atom
  elname 'content'
  elallvalue #This is what says "process the subtree of this element as if it was text"
end

feed = XC_Atom.import_xml XML_TEXT
#This actually returns the HTML content
puts "   xmlcodec says: "+feed.entry[0].content.value[0..20].inspect

require 'benchmark'
BENCH_ITER = 1000
Benchmark.bm(10) do |x|
  x.report("sax-machine:") do
    BENCH_ITER.times do
      feed = SM_Atom.parse(XML_TEXT)
    end
  end
  x.report("   xmlcodec:") do
    BENCH_ITER.times do
      feed = XC_Atom.import_xml XML_TEXT
    end
  end
end

# On my machine xmlcodec is around 10% faster than sax-machine and none of them 
# seem to use too much memory (not shown here)
#
#                  user     system      total        real
# sax-machine: 34.880000   0.030000  34.910000 ( 35.064932)
#    xmlcodec: 30.970000   0.010000  30.980000 ( 31.024513)