iter_test.rb

require "securerandom"
require "benchmark"
require "google_hash"
require "memory_profiler"
require "pry"

class Item
  def initialize
    @keep = ([0, 1].sample == 1)
    @key = SecureRandom.uuid
  end

  attr_reader :key

  def keep?
    @keep
  end
end

mutable_hash_impl = ->(items) {
  some_hash = Hash.new
  items.each do |item|
    next if item.keep?
    some_hash[item.key] = item
  end
  some_hash
}

pipeline_impl = ->(items) {
  items.select do |item|
    !item.keep?
  end.map do |item|
    [item.key, item]
  end.to_h
}

reduce_impl = ->(items) {
  items.reduce(Hash.new) do |hash, item|
    hash[item.key] = item unless item.keep?
    hash
  end
}

ghash_dense_impl = ->(items) {
  some_hash = GoogleHashDenseRubyToRuby.new
  items.each do |item|
    next if item.keep?
    some_hash[item.key] = item
  end
  some_hash
}

ghash_sparse_impl = ->(items) {
  some_hash = GoogleHashSparseRubyToRuby.new
  items.each do |item|
    next if item.keep?
    some_hash[item.key] = item
  end
  some_hash
}

tests = [
  ["mutable hash ", mutable_hash_impl],
  ["select + map ", pipeline_impl],
  ["reduce       ", reduce_impl],
  ["g-hash dense ", ghash_dense_impl],
  ["g-hash sparse", ghash_sparse_impl],
]

shuffled_tests = tests.shuffle

control_items = 10.times.map{ Item.new }
control = mutable_hash_impl.call(control_items)
controls = tests.map do |name, impl|
  test = impl.call(control_items)

  if control.keys.sort == test.keys.sort
    if control.keys.all?{ |k| control[k] == test[k] }
      nil
    else
      "#{name.strip}: (values)"
    end
  else
    "#{name.strip}: (keys)"
  end
end.compact

raise "FAIL! #{controls.join(', ')} did not match" unless controls.empty?

# size = 1_000_000
size =   500_000
# size =   100_000
# size =    10_000
items = size.times.map{ Item.new }

Benchmark.bmbm do |bm|
  shuffled_tests.each do |name, impl|
    bm.report(name) do
      impl.call(items)
    end
  end
end

puts
puts
puts '*' * 80
puts
puts

mp_rp_options = {
  top: 10,
  allow_files: File.expand_path(__FILE__),
}
mp_pp_options = {
  color_output: true,
  retained_strings: 0,
  allocated_strings: 0,
  detailed_report: false,
  scale_bytes: true,
  normalize_paths: true,
}
shuffled_tests.each do |name, impl|
  puts name
  MemoryProfiler.report(mp_rp_options) do
    impl.call(items)
  end.pretty_print(mp_pp_options)
end