Ruby MARC serialization / deserialization benchmark

marc_serialization_bench.rb

require 'benchmark'
require 'marc'
require 'msgpack'
require 'json'

unless defined? JRUBY_VERSION
  require 'oj'
end


jsonfile = 'topics.ndj'
marcfile = 'topics.mrc'
marshalfile = 'topics.msh'
msgpackfile = 'topics.msgpack'
tmpfile = 'tmpfile.delete'

records = MARC::Reader.new(marcfile).each_with_object([]) {|x,a| a << x}
puts "Got #{records.size} records"

puts "\n\n\n" + RUBY_DESCRIPTION

Benchmark.bmbm do |x|
  x.report("load marc21") do
    reader = MARC::Reader.new(marcfile)
    total = 0
    reader.each do |r|
      total += 1
    end
  end

  x.report("load json  ") do
    reader = File.open(jsonfile)
    total = 0
    reader.each do |json|
      r = MARC::Record.new_from_hash(JSON.parse(json))
      total += 1
    end
  end

  unless defined? JRUBY_VERSION
    x.report("load json (oj)") do
      reader = File.open(jsonfile)
      total = 0
      reader.each do |json|
        r = MARC::Record.new_from_hash(Oj.strict_load(json))
        total += 1
      end
    end
  end


  x.report('load msgpack') do
    reader = File.open(msgpackfile)
    total = 0
    MessagePack::Unpacker.new(reader).each do |h|
      r = MARC::Record.new_from_hash(h)
      total += 1
    end
  end

  x.report("load marshal") do
    infile = File.open(marshalfile)
    total = 0
    while !infile.eof?
      r = Marshal.load(infile)
      total += 1
    end
  end



  x.report("write marc21") do
    writer = MARC::Writer.new(tmpfile)
    records.each {|r| writer.write(r)}
    writer.close
  end

  x.report('write json') do
    out = File.open(tmpfile, 'w:utf-8')
    records.each {|r| out.puts JSON.dump(r.to_hash)}
    out.close
  end

  unless defined? JRUBY_VERSION
    x.report('write json (oj)') do
      out = File.open(tmpfile, 'w:utf-8')
      records.each {|r| Oj.to_stream(out, r.to_hash)}
      out.close
    end
  end


  x.report('write msgpack') do
    out = File.open(tmpfile, 'wb')
    records.each {|r| out.write MessagePack.dump(r.to_hash) }
    out.close
  end


  x.report('write marshal') do
    out = File.open(tmpfile, 'wb')
    records.each {|r| Marshal.dump(r, out)}
    out.close
  end

end