Compare zipruby memory consumption with others

zipruby_test.rb

#!/user/bin/env ruby
#
# Demonstrates memory behaviour with zipruby vs rubyzip vs exec('unzip').
# Tested with mri 1.8.7/1.9.3/2.1.2, zipruby 0.3.6 and rubyzip 1.1.6.
#
# Usage: ruby zipruby_test.rb <test method>
#
# Test methods are:
#   read_zipruby  read file from zip using zipruby gem
#   read_rubyzip  read file from zip using rubyzip gem
#   read_exec     read files from zip using exec + unzip(1)
#   data_times    (sanity) populate var using String#* to size of test file
#   data_append   (sanity) populate var using String#<< to size of test file
#
# Code assumes you have unzip(1L), curl(1) and BSD-style ps(1).
#
# This script does an action 10,000 times, stopping every 1,000 times to
# record its own memory. At the end of the run it prints all 10 memory
# footprints in the order they were recorded.
#
# Results:
#
# read_zipruby
#   ruby-1.8.7: 56K, 110K, 161K, 208K, 257K, 306K, 357K, 403K, 451K, 499K
#   ruby-1.9.3: 60K, 112K, 167K, 214K, 262K, 312K, 362K, 409K, 456K, 506K
#   ruby-2.1.2: 72K, 136K, 194K, 258K, 311K, 371K, 412K, 473K, 527K, 587K
#
# read_rubyzip
#   ruby-1.8.7: requires >= 1.9.2
#   ruby-1.9.3: 14K, 20K, 20K, 21K, 21K, 21K, 21K, 23K, 23K, 23K
#   ruby-2.1.2: 21K, 26K, 27K, 27K, 28K, 28K, 30K, 30K, 30K, 30K
#
# read_exec
#   ruby-1.8.7: 5K, 5K, 5K, 5K, 5K, 5K, 5K, 5K, 5K, 5K
#   ruby-1.9.3: 7K, 8K, 8K, 8K, 8K, 8K, 8K, 8K, 8K, 8K
#   ruby-2.1.2: 23K, 27K, 30K, 30K, 31K, 31K, 31K, 31K, 31K, 31K
#
# data_times
#   ruby-1.8.7: 11K, 19K, 19K, 19K, 19K, 19K, 19K, 19K, 27K, 27K
#   ruby-1.9.3: 18K, 19K, 25K, 29K, 36K, 41K, 43K, 49K, 49K, 51K
#   ruby-2.1.2: 21K, 33K, 42K, 49K, 55K, 55K, 55K, 62K, 71K, 71K
#
# data_append
#   ruby-1.8.7: 3K, 3K, 3K, 3K, 3K, 3K, 3K, 3K, 3K, 3K
#   ruby-1.9.3: 6K, 6K, 6K, 6K, 7K, 7K, 7K, 7K, 7K, 7K
#   ruby-2.1.2: 18K, 25K, 31K, 36K, 38K, 42K, 45K, 48K, 49K, 50K

require 'rubygems' # for testing under <= 1.8.7

url = 'https://epub-samples.googlecode.com/files/mymedia_lite-20130621.epub'
file = './mymedia_lite-20130621.epub'
page = 'OEBPS/text/book_0002.xhtml' # file from zip above
page_size = 13_103 # size of page file above

unless File.exists?(file)
  puts "Downloading test epub: #{url}"
  `/usr/bin/env curl #{url} > #{file}`
end

class TestClass
  def initialize(file_path, page_path, page_size)
    @file_path = file_path
    @page_path = page_path
    @page_size = page_size
  end
  
  def data_times
    ('x' * @page_size)
  end

  def data_append
    ''.tap { |d| @page_size.times { d << 'x' } }
  end

  def read_zipruby
    Zip::Archive.open(@file_path) {|zip|
      zip.fopen(@page_path).read
    }
  end

  def read_rubyzip
    Zip::File.open(@file_path) do |zip|
      zip.glob(@page_path).first.get_input_stream.read
    end
  end

  def read_exec
    `/usr/bin/env unzip -p #{@file_path} #{@page_path}`
  end

end

method_list = %w[ read_zipruby read_rubyzip read_exec data_times data_append ]

method_name = ARGV[0]

raise(["provide method name: ", method_list.join(', ') ].join) unless
  method_list.include?(method_name)

case method_name
when 'read_zipruby'
  require 'zipruby' # 0.3.6
when 'read_rubyzip'
  require 'zip' # rubyzip 1.1.6
end

memory_stats = []

1.upto(10_000) do |i|
  print '.'

  x = TestClass.new(file, page, page_size).send(method_name.to_sym)

  raise "Size mismatch: #{x.to_s.bytesize} != #{page_size}" unless
    x.to_s.bytesize == page_size

  if i % 1_000 == 0
    print "#{i} "
    sleep(0.5) # Give time for GC to come in.
    mem = (`/usr/bin/env ps -o rss -p #{$$}`.strip.split.last.to_i/1000).to_s + 'K'
    puts mem
    memory_stats << mem
  end
end

puts "\n#{method_name}: #{memory_stats.join(', ')}"