memoize, build lookup tables, and use regexp for speed

lib/rumoji.rb

@@ -8,8 +8,7 @@ module Rumoji
 
   # Transform emoji into its cheat-sheet code
   def encode(str)
-    io = StringIO.new(str)
-    encode_io(io).string
+    str.gsub(Emoji::ALL_REGEXP) { |match| (emoji = Emoji.find_by_string(match)) && emoji.code || match }
   end
 
   # Transform a cheat-sheet code into an Emoji

lib/rumoji/emoji.rb

@@ -2,9 +2,10 @@
 module Rumoji
   class Emoji
 
-    attr_reader :name
+    attr_reader :name, :string
 
     def initialize(string, symbols, name = nil)
+      @string = string
       @codepoints = string.codepoints
       @cheat_codes = [symbols].flatten
       @name = name || @cheat_codes.first.to_s.upcase.gsub("_", " ")
@@ -23,7 +24,7 @@ module Rumoji
     end
 
     def to_s
-      codepoints.pack("U*")
+      @string
     end
 
     def hash
@@ -31,7 +32,7 @@ module Rumoji
     end
 
     def hex
-      @codepoints.map{|point| point.to_s(16).upcase }.join("-")
+      @hex ||= @codepoints.map{|point| point.to_s(16).upcase }.join("-")
     end
 
     def codepoints
@@ -50,16 +51,29 @@ module Rumoji
 
     ALL = PEOPLE | NATURE | OBJECTS | PLACES | SYMBOLS
 
+    ALL_REGEXP = Regexp.new(ALL.map(&:string).join('|'))
+
     def self.find(symbol)
       ALL.find {|emoji| emoji.include? symbol }
     end
 
+    STRING_LOOKUP = ALL.each.with_object({}) do |emoji, lookup|
+      lookup[emoji.string] = emoji
+    end
+
     def self.find_by_string(string)
-      ALL.find {|emoji| emoji.to_s == string }
+      STRING_LOOKUP[string]
+    end
+
+    CODEPOINT_LOOKUP = ALL.each.with_object({}) do |emoji, lookup|
+      emoji.codepoints.each do |codepoint|
+        lookup[codepoint] ||= []
+        lookup[codepoint] << emoji unless lookup[codepoint].include?(emoji)
+      end
     end
 
     def self.select_by_codepoint(codepoint)
-      ALL.select {|emoji| emoji.codepoints.include? codepoint }
+      CODEPOINT_LOOKUP[codepoint] || []
     end
 
     def self.find_by_codepoint(codepoint)

spec/rumoji_spec.rb

@@ -18,6 +18,44 @@ describe Rumoji do
       Rumoji.encode(@smile).must_equal ":smile:"
       Rumoji.encode("#{@smile}").must_equal ":smile:"
     end
+
+    it "keeps codepoints that match the beginnings of multi-codepoint emoji" do
+      text = "i like #hashtags and 1direction they are the #1 band. end with 9"
+      Rumoji.encode(text).must_equal text
+    end
+
+    describe "with multiple codepoints" do
+      it "transforms a stream" do
+        Rumoji.encode("#{@zero}").must_equal ":zero:"
+        Rumoji.encode("#{@us}").must_equal ":us:"
+      end
+
+      it "transforms a stream of many emoji" do
+        num = ":one: :two: :three: :four: :five: :six: :seven: :eight: :nine: :zero: :hash:"
+        emoji = "1️⃣ 2️⃣ 3️⃣ 4️⃣ 5️⃣ 6️⃣ 7️⃣ 8️⃣ 9️⃣ 0️⃣ #️⃣"
+        Rumoji.encode(emoji).must_equal num
+      end
+
+      it "does not encode double digits" do
+        num = ":zero: :one: :two: :three: :four: :five: :six: :seven: :eight: :nine: :hash:"
+        double_digits = "00 11 22 33 44 55 66 77 88 99 ##"
+        Rumoji.encode(double_digits).wont_equal num
+      end
+
+      describe "with leading and trailing characters" do
+        it "is able to pull multipoint emoji out of a sequence" do
+          string = "An example of a multipoint emoji is the #{@us} flag."
+          Rumoji.encode(string).must_equal "An example of a multipoint emoji is the :us: flag."
+        end
+      end
+
+      describe "with trailing emoji" do
+        it "writes characters that are in a multipoint emoji followed by an emoji" do
+          string = "I would like 0#{@poop}"
+          Rumoji.encode(string).must_equal "I would like 0:poop:"
+        end
+      end
+    end
   end
 
   describe "#decode" do