Benchmark of the String#gsub! based Inflector.camelize method vs. a StringScanner based method

camelize_benchmark.rb

#!/usr/bin/env ruby

require 'benchmark'
require 'strscan'

module Inflector
  def self.gsub_camelize(name)
    name = name.to_s.dup

    # sourced from: https://github.com/dry-rb/dry-inflector/blob/c918f967ff82611da374eb0847a77b7e012d3fa8/lib/dry/inflector.rb#L329-L334
    name.sub!(/^[a-z\d]*/,&:capitalize)
    name.gsub!(%r{(?:[_-]|(/))([a-z\d]*)}i) do |match|
      slash = Regexp.last_match(1)
      word  = Regexp.last_match(2)

      "#{slash}#{word.capitalize}"
    end

    name.gsub!('/','::')
    name
  end

  def self.strscan_camelize(name)
    scanner = StringScanner.new(name.to_s)
    new_string = String.new

    until scanner.eos?
      if (word = scanner.scan(/[a-z\d]+/i))
        new_string << word.capitalize
      elsif scanner.scan(/[_-]+/)
        next
      elsif scanner.scan(/\//)
        new_string << '::'
      end
    end

    new_string
  end
end

Benchmark.bm(12) do |b|
  n = 1_009_000

  underscored = "foo_bar/baz_quix"

  b.report('Inflector.camelize + String#gsub!') do
    n.times do
      Inflector.gsub_camelize(underscored)
    end
  end

  b.report('Inflector.camelize + StringScanner') do
    n.times do
      Inflector.strscan_camelize(underscored)
    end
  end
end

underscore_benchmark.rb

#!/usr/bin/env ruby

require 'benchmark'
require 'strscan'

module Inflector
  def self.gsub_underscore(name)
    # sourced from: https://github.com/dry-rb/dry-inflector/blob/c918f967ff82611da374eb0847a77b7e012d3fa8/lib/dry/inflector.rb#L286-L287
    name = name.to_s.dup

    name.gsub!(/([A-Z\d]+)([A-Z][a-z])/, '\1_\2')
    name.gsub!(/([a-z\d])([A-Z])/, '\1_\2')
    name.tr!('-','_')
    name.downcase!

    name
  end

  def self.strscan_underscore(name)
    scanner = StringScanner.new(name.to_s)
    words   = []

    until scanner.eos?
      if (capitalized = scanner.scan(/[A-Z][a-z\d]+/))
        words << capitalized
      elsif (uppercase = scanner.scan(/[A-Z][A-Z\d]*(?=[A-Z]|$)/))
        words << uppercase
      elsif (lowercase = scanner.scan(/[a-z][a-z\d]*/))
        words << lowercase
      elsif scanner.scan(/[_-]+/)
        next
      else
        raise(ArgumentError,"string contains invalid characters: #{scanner.string}")
      end
    end

    words.join('_').downcase
  end
end

Benchmark.bm(12) do |b|
  n = 1_009_000

  underscored = "FooBarBazQux"

  b.report('Inflector.underscore + String#gsub!') do
    n.times do
      Inflector.gsub_underscore(underscored)
    end
  end

  b.report('Inflector.underscore + StringScanner') do
    n.times do
      Inflector.strscan_underscore(underscored)
    end
  end
end

Output:

                   user     system      total        real
Inflector.camelize + String#gsub!  7.023820   0.001943   7.025763 (  7.066990)
Inflector.camelize + StringScanner  4.646641   0.000993   4.647634 (  4.681039)

                   user     system      total        real
Inflector.underscore + String#gsub!  5.733684   0.002956   5.736640 (  5.791869)
Inflector.underscore + StringScanner  2.842114   0.000000   2.842114 (  2.862411)