chore(benchmark): enhance the codes for measuring performance between…

… each version

[ci skip]

benchmark/env.rb

@@ -3,19 +3,7 @@
 $LOAD_PATH << File.expand_path('../../lib', __FILE__)
 require 'bundler'
 Bundler.setup(:benchmark)
-
-SAMPLES = {
-  ascii: [
-    %w[al al], %w[martha marhta], %w[jones johnson], %w[abcvwxyz cabvwxyz],
-    %w[dwayne duane], %w[dixon dicksonx], %w[fvie ten]
-  ].freeze,
-  utf8: [
-    %w[馬英九 馬英丸], %w[蔡英文 蔡中文], %w[簡煒航 簡偉航], %w[焦玟綾 焦紋綾],
-    %w[眼球中央電視台 眼球中英電視台], %w[床前明月光 床前日月光],
-    %w[海水退了就知道誰沒穿褲子 海水退了就知道誰沒穿襪子],
-    %w[阿里山的姑娘美如水 阿里山的姑娘沒乳水]
-  ].freeze
-}.freeze
+require 'samples'
 
 def gem_name_with_version(gem)
   "#{gem} (#{Gem.loaded_specs[gem].version})"

benchmark/measure.rb

@@ -1,28 +1,35 @@
 # frozen_string_literal: true
 
-require File.expand_path('../env', __FILE__)
-require 'jaro_winkler'
+require File.expand_path('../samples', __FILE__)
+gem 'jaro_winkler', ENV['JARO_WINKLER_VERSION'] || ARGV[0] || raise('missing ENV["JARO_WINKLER_VERSION"]')
 require 'benchmark'
 require 'csv'
 
 csv = CSV.new($stdout)
 n = 100_000
+version = Gem::Version.new(Gem.loaded_specs["jaro_winkler"].version)
 
-jobs = case JaroWinkler::VERSION
-when /\A1\.1/
-  {
-    ascii: -> { n.times { SAMPLES[:ascii].each { |str1, str2| JaroWinkler.c_distance(str1, str2) } } }
-  }
-when /\A1\.[2-3]/
-  {
-    ascii: -> { n.times { SAMPLES[:ascii].each { |str1, str2| JaroWinkler.c_distance(str1, str2) } } },
-    utf8: -> { n.times { SAMPLES[:utf8].each { |str1, str2| JaroWinkler.c_distance(str1, str2) } } }
-  }
+if version >= Gem::Version.new('1.1.0') && version < Gem::Version.new('1.2.4')
+  require 'jaro_winkler.bundle'
 else
-  {
-    ascii: -> { n.times { SAMPLES[:ascii].each { |str1, str2| JaroWinkler.distance(str1, str2) } } },
-    utf8: -> { n.times { SAMPLES[:utf8].each { |str1, str2| JaroWinkler.distance(str1, str2) } } }
-  }
+  require 'jaro_winkler'
+end
+
+jobs = {
+  ascii: -> { n.times { SAMPLES[:ascii].each { |str1, str2| JaroWinkler.jaro_winkler_distance(str1, str2) } } }
+}
+
+if version >= Gem::Version.new('1.1.0')
+  jobs[:ascii] = -> { n.times { SAMPLES[:ascii].each { |str1, str2| JaroWinkler.c_distance(str1, str2) } } }
+end
+
+if version >= Gem::Version.new('1.2.0')
+  jobs[:utf8] = -> { n.times { SAMPLES[:utf8].each { |str1, str2| JaroWinkler.c_distance(str1, str2) } } }
+end
+
+if version >= Gem::Version.new('1.4.0')
+  jobs[:ascii] = -> { n.times { SAMPLES[:ascii].each { |str1, str2| JaroWinkler.distance(str1, str2) } } }
+  jobs[:utf8] = -> { n.times { SAMPLES[:utf8].each { |str1, str2| JaroWinkler.distance(str1, str2) } } }
 end
 
 # rehearsal
@@ -35,5 +42,5 @@
   GC.start
   tms = Benchmark.measure(label, &job)
   # version,label,utime,stime,cutime,cstime,real
-  csv << [JaroWinkler::VERSION, *tms.to_a]
+  csv << [version, *tms.to_a]
 end

benchmark/samples.rb

@@ -0,0 +1,12 @@
+SAMPLES = {
+  ascii: [
+    %w[al al], %w[martha marhta], %w[jones johnson], %w[abcvwxyz cabvwxyz],
+    %w[dwayne duane], %w[dixon dicksonx], %w[fvie ten]
+  ].freeze,
+  utf8: [
+    %w[馬英九 馬英丸], %w[蔡英文 蔡中文], %w[簡煒航 簡偉航], %w[焦玟綾 焦紋綾],
+    %w[眼球中央電視台 眼球中英電視台], %w[床前明月光 床前日月光],
+    %w[海水退了就知道誰沒穿褲子 海水退了就知道誰沒穿襪子],
+    %w[阿里山的姑娘美如水 阿里山的姑娘沒乳水]
+  ].freeze
+}.freeze

bin/measure

@@ -0,0 +1,9 @@
+#!/bin/sh
+echo version,label,utime,stime,cutime,cstime,real
+gem search -ear jaro_winkler \
+  | grep -o '\((.*)\)$' \
+  | tr -d '() ' \
+  | tr ',' "\n" \
+  | grep -o '\d\.\d\.\d' \
+  | sort \
+  | xargs -I{} ruby "`dirname $0`"/../benchmark/measure.rb '{}'