Support lookup for emoji with multiple codepoints

In addition, this code rewrites encode to just defer to encode_io. This
is an attempt to fix #2, though I believe the lookahead code can and
should be refined.

lib/rumoji.rb

@@ -8,11 +8,8 @@ module Rumoji
 
 
   # Transform emoji into its cheat-sheet code
   # Transform emoji into its cheat-sheet code
   def encode(str)
   def encode(str)
-    remapped_codepoints = str.codepoints.flat_map do |codepoint|
-      emoji = Emoji.find_by_codepoint(codepoint)
-      emoji ? emoji.code.codepoints.entries : codepoint
-    end
-    remapped_codepoints.pack("U*")
+    io = StringIO.new(str)
+    encode_io(io).string
   end
   end
 
 
   # Transform a cheat-sheet code into an Emoji
   # Transform a cheat-sheet code into an Emoji
@@ -21,10 +18,17 @@ module Rumoji
   end
   end
 
 
   def encode_io(readable, writeable=StringIO.new(""))
   def encode_io(readable, writeable=StringIO.new(""))
-    readable.each_codepoint do |codepoint|
-      emoji = Emoji.find_by_codepoint(codepoint)
-      emoji_or_character = emoji ? emoji.code : [codepoint].pack("U")
-      writeable.write emoji_or_character
+    codepoints = readable.each_codepoint.entries
+    codepoints.each_with_index do |codepoint, index|
+      possible_emoji = Emoji.select_by_codepoint(codepoint)
+      sequence = if possible_emoji.empty?
+        [codepoint].pack("U")
+      elsif possible_emoji.size.eql?(1)
+        possible_emoji.first.code
+      else
+        find_multipoint_emoji_in_set(possible_emoji, codepoints[index + 1])
+      end
+      writeable.write sequence
     end
     end
     writeable
     writeable
   end
   end
@@ -36,4 +40,13 @@ module Rumoji
     writeable
     writeable
   end
   end
 
 
+private
+
+  def find_multipoint_emoji_in_set(possible_emoji_set, next_codepoint)
+    if next_codepoint
+      emoji_or_nil = possible_emoji_set.find {|emoji| emoji.codepoints.include?(next_codepoint) }
+      emoji_or_nil.nil? ? "" : emoji_or_nil.code
+    end
+  end
+
 end
 end

lib/rumoji/emoji.rb

@@ -23,7 +23,7 @@ module Rumoji
     end
     end
 
 
     def to_s
     def to_s
-      @codepoints.to_a.pack("U*")
+      codepoints.pack("U*")
     end
     end
 
 
     def hash
     def hash
@@ -34,6 +34,10 @@ module Rumoji
       @codepoints.map{|point| point.to_s(16).upcase }.join("-")
       @codepoints.map{|point| point.to_s(16).upcase }.join("-")
     end
     end
 
 
+    def codepoints
+      @codepoints.entries
+    end
+
     autoload :PEOPLE, 'rumoji/emoji/people'
     autoload :PEOPLE, 'rumoji/emoji/people'
     autoload :NATURE, 'rumoji/emoji/nature'
     autoload :NATURE, 'rumoji/emoji/nature'
     autoload :OBJECTS, 'rumoji/emoji/objects'
     autoload :OBJECTS, 'rumoji/emoji/objects'
@@ -50,6 +54,10 @@ module Rumoji
       ALL.find {|emoji| emoji.to_s == string }
       ALL.find {|emoji| emoji.to_s == string }
     end
     end
 
 
+    def self.select_by_codepoint(codepoint)
+      ALL.select {|emoji| emoji.codepoints.include? codepoint }
+    end
+
     def self.find_by_codepoint(codepoint)
     def self.find_by_codepoint(codepoint)
       ALL.find {|emoji| emoji.hex == codepoint.to_s(16).upcase }
       ALL.find {|emoji| emoji.hex == codepoint.to_s(16).upcase }
     end
     end

lib/rumoji/emoji/places.rb

@@ -98,16 +98,16 @@ module Rumoji
       self.new("\u{26a0}" , [:warning]),
       self.new("\u{26a0}" , [:warning]),
       self.new("\u{1f492}", [:wedding]),
       self.new("\u{1f492}", [:wedding]),
       # Regional Indicator Symbols
       # Regional Indicator Symbols
-      self.new("\u{1f1ef}\u{1f1f5}", [:jp], "REGIONAL INDICATOR SYMBOL LETTERS JP"),
-      self.new("\u{1f1f0}\u{1f1f7}", [:kr], "REGIONAL INDICATOR SYMBOL LETTERS KR"),
-      self.new("\u{1f1e8}\u{1f1f3}", [:cn], "REGIONAL INDICATOR SYMBOL LETTERS CN"),
-      self.new("\u{1f1fa}\u{1f1f8}", [:us], "REGIONAL INDICATOR SYMBOL LETTERS US"),
-      self.new("\u{1f1eb}\u{1f1f7}", [:fr], "REGIONAL INDICATOR SYMBOL LETTERS FR"),
-      self.new("\u{1f1ea}\u{1f1f8}", [:es], "REGIONAL INDICATOR SYMBOL LETTERS ES"),
-      self.new("\u{1f1ee}\u{1f1f9}", [:it], "REGIONAL INDICATOR SYMBOL LETTERS IT"),
-      self.new("\u{1f1f7}\u{1f1fa}", [:ru], "REGIONAL INDICATOR SYMBOL LETTERS RU"),
-      self.new("\u{1f1ec}\u{1f1e7}", [:gb, :uk], "REGIONAL INDICATOR SYMBOL LETTERS GB"),
-      self.new("\u{1f1e9}\u{1f1ea}", [:de], "REGIONAL INDICATOR SYMBOL LETTERS DE"),
+      self.new("\u{1f1ef 1f1f5}", [:jp], "REGIONAL INDICATOR SYMBOL LETTERS JP"),
+      self.new("\u{1f1f0 1f1f7}", [:kr], "REGIONAL INDICATOR SYMBOL LETTERS KR"),
+      self.new("\u{1f1e8 1f1f3}", [:cn], "REGIONAL INDICATOR SYMBOL LETTERS CN"),
+      self.new("\u{1f1fa 1f1f8}", [:us], "REGIONAL INDICATOR SYMBOL LETTERS US"),
+      self.new("\u{1f1eb 1f1f7}", [:fr], "REGIONAL INDICATOR SYMBOL LETTERS FR"),
+      self.new("\u{1f1ea 1f1f8}", [:es], "REGIONAL INDICATOR SYMBOL LETTERS ES"),
+      self.new("\u{1f1ee 1f1f9}", [:it], "REGIONAL INDICATOR SYMBOL LETTERS IT"),
+      self.new("\u{1f1f7 1f1fa}", [:ru], "REGIONAL INDICATOR SYMBOL LETTERS RU"),
+      self.new("\u{1f1ec 1f1e7}", [:gb, :uk], "REGIONAL INDICATOR SYMBOL LETTERS GB"),
+      self.new("\u{1f1e9 1f1ea}", [:de], "REGIONAL INDICATOR SYMBOL LETTERS DE"),
     ]
     ]
   end
   end
 end
 end

lib/rumoji/emoji/symbols.rb

@@ -81,7 +81,7 @@ module Rumoji
       self.new("\u{1f6c3}", [:customs]),
       self.new("\u{1f6c3}", [:customs]),
       self.new("\u{1f4a0}", [:diamond_shape_with_a_dot_inside]),
       self.new("\u{1f4a0}", [:diamond_shape_with_a_dot_inside]),
       self.new("\u{1f6af}", [:do_not_litter]),
       self.new("\u{1f6af}", [:do_not_litter]),
-      self.new("\u{0038}" , [:eight]),
+      self.new("\u{0038 20e3}" , [:eight]),
       self.new("\u{2734}" , [:eight_pointed_black_star]),
       self.new("\u{2734}" , [:eight_pointed_black_star]),
       self.new("\u{2733}" , [:eight_spoked_asterisk]),
       self.new("\u{2733}" , [:eight_spoked_asterisk]),
       self.new("\u{1f51a}", [:end]),
       self.new("\u{1f51a}", [:end]),
@@ -90,7 +90,7 @@ module Rumoji
       self.new("\u{0034}" , [:four]),
       self.new("\u{0034}" , [:four]),
       self.new("\u{1f193}", [:free]),
       self.new("\u{1f193}", [:free]),
       self.new("\u{264a}" , [:gemini]),
       self.new("\u{264a}" , [:gemini]),
-      self.new("\u{0023}" , [:hash]),
+      self.new("\u{0023 20e3}" , [:hash]),
       self.new("\u{1f49f}", [:heart_decoration]),
       self.new("\u{1f49f}", [:heart_decoration]),
       self.new("\u{2714}" , [:heavy_check_mark]),
       self.new("\u{2714}" , [:heavy_check_mark]),
       self.new("\u{2797}" , [:heavy_division_sign]),
       self.new("\u{2797}" , [:heavy_division_sign]),
@@ -120,7 +120,7 @@ module Rumoji
       self.new("\u{274e}" , [:negative_squared_cross_mark]),
       self.new("\u{274e}" , [:negative_squared_cross_mark]),
       self.new("\u{1f195}", [:new]),
       self.new("\u{1f195}", [:new]),
       self.new("\u{1f196}", [:ng]),
       self.new("\u{1f196}", [:ng]),
-      self.new("\u{0039}" , [:nine]),
+      self.new("\u{0039 20e3}" , [:nine]),
       self.new("\u{1f6b3}", [:no_bicycles]),
       self.new("\u{1f6b3}", [:no_bicycles]),
       self.new("\u{26d4}" , [:no_entry]),
       self.new("\u{26d4}" , [:no_entry]),
       self.new("\u{1f6ab}", [:no_entry_sign]),
       self.new("\u{1f6ab}", [:no_entry_sign]),
@@ -132,7 +132,7 @@ module Rumoji
       self.new("\u{1f17e}", [:o2]),
       self.new("\u{1f17e}", [:o2]),
       self.new("\u{1f197}", [:ok]),
       self.new("\u{1f197}", [:ok]),
       self.new("\u{1f51b}", [:on]),
       self.new("\u{1f51b}", [:on]),
-      self.new("\u{0031}" , [:one]),
+      self.new("\u{0031 20e3}" , [:one]),
       self.new("\u{26ce}" , [:ophiuchus]),
       self.new("\u{26ce}" , [:ophiuchus]),
       self.new("\u{1f17f}", [:parking]),
       self.new("\u{1f17f}", [:parking]),
       self.new("\u{303d}" , [:part_alternation_mark]),
       self.new("\u{303d}" , [:part_alternation_mark]),
@@ -152,9 +152,9 @@ module Rumoji
       self.new("\u{2650}" , [:sagittarius]),
       self.new("\u{2650}" , [:sagittarius]),
       self.new("\u{264f}" , [:scorpius]),
       self.new("\u{264f}" , [:scorpius]),
       self.new("\u{3299}" , [:secret]),
       self.new("\u{3299}" , [:secret]),
-      self.new("\u{0037}" , [:seven]),
+      self.new("\u{0037 20e3}" , [:seven]),
       self.new("\u{1f4f6}", [:signal_strength]),
       self.new("\u{1f4f6}", [:signal_strength]),
-      self.new("\u{0036}" , [:six]),
+      self.new("\u{0036 20e3}" , [:six]),
       self.new("\u{1f52f}", [:six_pointed_star]),
       self.new("\u{1f52f}", [:six_pointed_star]),
       self.new("\u{1f539}", [:small_blue_diamond]),
       self.new("\u{1f539}", [:small_blue_diamond]),
       self.new("\u{1f538}", [:small_orange_diamond]),
       self.new("\u{1f538}", [:small_orange_diamond]),
@@ -164,12 +164,12 @@ module Rumoji
       self.new("\u{1f198}", [:sos]),
       self.new("\u{1f198}", [:sos]),
       self.new("\u{1f523}", [:symbols]),
       self.new("\u{1f523}", [:symbols]),
       self.new("\u{2649}" , [:taurus]),
       self.new("\u{2649}" , [:taurus]),
-      self.new("\u{0033}" , [:three]),
+      self.new("\u{0033 20e3}" , [:three]),
       self.new("\u{2122}" , [:tm]),
       self.new("\u{2122}" , [:tm]),
       self.new("\u{1f51d}", [:top]),
       self.new("\u{1f51d}", [:top]),
       self.new("\u{1f531}", [:trident]),
       self.new("\u{1f531}", [:trident]),
       self.new("\u{1f500}", [:twisted_rightwards_arrows]),
       self.new("\u{1f500}", [:twisted_rightwards_arrows]),
-      self.new("\u{0032}" , [:two]),
+      self.new("\u{0032 20e3}" , [:two]),
       self.new("\u{1f239}", [:u5272]),
       self.new("\u{1f239}", [:u5272]),
       self.new("\u{1f234}", [:u5408]),
       self.new("\u{1f234}", [:u5408]),
       self.new("\u{1f23a}", [:u55b6]),
       self.new("\u{1f23a}", [:u55b6]),
@@ -195,7 +195,7 @@ module Rumoji
       self.new("\u{1f533}", [:white_square_button]),
       self.new("\u{1f533}", [:white_square_button]),
       self.new("\u{1f6ba}", [:womens]),
       self.new("\u{1f6ba}", [:womens]),
       self.new("\u{274c}" , [:x]),
       self.new("\u{274c}" , [:x]),
-      self.new("\u{0030}" , [:zero])
+      self.new("\u{0030 20e3}" , [:zero])
     ]
     ]
   end
   end
 end
 end

spec/rumoji_spec.rb

@@ -7,6 +7,8 @@ describe Rumoji do
   before do
   before do
     @poop = "💩"
     @poop = "💩"
     @smile = "😄"
     @smile = "😄"
+    @zero = "0⃣"
+    @us = "🇺🇸"
   end
   end
 
 
   describe "#encode" do
   describe "#encode" do
@@ -27,6 +29,23 @@ describe Rumoji do
       io = StringIO.new("#{@smile}")
       io = StringIO.new("#{@smile}")
       Rumoji.encode_io(io).string.must_equal ":smile:"
       Rumoji.encode_io(io).string.must_equal ":smile:"
     end
     end
+
+    describe "with multiple codepoints" do
+      it "transforms a stream" do
+        io1 = StringIO.new("#{@zero}")
+        io2 = StringIO.new("#{@us}")
+        Rumoji.encode_io(io1).string.must_equal ":zero:"
+        Rumoji.encode_io(io2).string.must_equal ":us:"
+      end
+
+      describe "with leading and trailing characters" do
+        it "is able to pull multipoint emoji out of a sequence" do
+          io = StringIO.new("An example of a multipoint emoji is the #{@us} flag.")
+          Rumoji.encode_io(io).string.must_equal "An example of a multipoint emoji is the :us: flag."
+        end
+      end
+    end
+
   end
   end
 
 
   describe "#decode_io" do
   describe "#decode_io" do
@@ -34,5 +53,12 @@ describe Rumoji do
       io = StringIO.new(":poop:")
       io = StringIO.new(":poop:")
       Rumoji.decode_io(io).string.must_equal @poop
       Rumoji.decode_io(io).string.must_equal @poop
     end
     end
+
+    describe "with multiple codepoints" do
+      it "decodes a stream" do
+        io = StringIO.new(":zero:")
+        Rumoji.decode_io(io).string.must_equal @zero
+      end
+    end
   end
   end
 end
 end