Deserialization speed of marc-in-json vs marcxml under ruby-marc with fastest available libraries

marc_deserialization_bench.rb

require 'rubygems'
require 'marc'
require 'yajl'
require 'benchmark'

iterations = 5

xmlsourcefile = 'topics.xml' # 18k records as a MARC-XML collection
jsonsourcefile = 'topics.ndj' # Same records as newline-delimited marc-in-json

Benchmark.bm do |x|
  x.report("xml w/libxml") do
    iterations.times do
      reader =  MARC::XMLReader.new(xmlsourcefile, :parser=>'libxml')
      reader.each do |r|
        title = r['245']
      end
    end
  end
  
  x.report("marc-in-json w/yajl") do
    iterations.times do
      File.open(jsonsourcefile).each_line do |json|
        r = MARC::Record.new_from_hash(Yajl::Parser.parse(json))
        title = r['245']
      end
    end
  end
end