replace greater than 3 byte utf8

benchmark.rb

# encoding: utf-8
require 'gemoji'
require 'benchmark'
require 'benchmark/ips'

task :default => :benchmark

task :benchmark do
  regex       = /([^\u0000-\uD7FF\uE000-\uFFFF])/
  replacement = "�"
  test_string = "this is my 😊, there are many like it but this one is mine"

  Benchmark.ips do |x|
    x.report 'String#gsub regex' do
      test_string.gsub(regex, replacement)
    end

    x.report 'String#gsub with gemoji' do
      test_string.gsub(regex) { |e| ":#{Emoji.find_by_unicode(e).aliases.first}:" }
    end

    x.report 'String#chars bytesize' do
      "".tap do |out_str|

        test_string.chars.each do |char|
          if char.bytesize > 3
            out_str << replacement
          else
            out_str << char
          end
        end

      end
    end

    x.report 'Demoji Gem Style' do
      "".tap do |out_str|

        # for instead of split and joins for perf
        for i in (0...test_string.length)
          char = test_string[i]
          char = replacement if char.ord > 65535
          out_str << char
        end

      end
    end

    x.compare!
  end
end

output.sh

rake benchmark
Calculating -------------------------------------
   String#gsub regex    27.135k i/100ms
String#gsub with gemoji
                        21.825k i/100ms
String#chars bytesize
                         6.193k i/100ms
    Demoji Gem Style     5.812k i/100ms
-------------------------------------------------
   String#gsub regex    295.730k (± 4.4%) i/s -      1.492M
String#gsub with gemoji
                        266.580k (± 4.0%) i/s -      1.331M
String#chars bytesize
                         84.961k (± 4.3%) i/s -    427.317k
    Demoji Gem Style     59.164k (± 3.2%) i/s -    296.412k

Comparison:
   String#gsub regex:   295730.2 i/s
String#gsub with gemoji:   266580.3 i/s - 1.11x slower
String#chars bytesize:    84961.4 i/s - 3.48x slower
    Demoji Gem Style:    59164.3 i/s - 5.00x slower