Merge pull request #36 from myhobnob/properly-detect-multi-character-emoji

Properly detect multi-character emoji

lib/rumoji/emoji.rb

@@ -47,6 +47,14 @@ module Rumoji
       codepoints.size > 1
       codepoints.size > 1
     end
     end
 
 
+    # Sort by reverse alphabetical order so that longer more complex emoji are
+    # matched in the regex before their simpler components
+    # e.g. :man-man-boy: needs to come before :man: in the regex or else
+    # :man-man-boy: will never be matched
+    def <=>(other)
+      other.symbol <=> symbol
+    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'

lib/rumoji/emoji/nature.rb

@@ -5,7 +5,7 @@ require 'set'
 
 
 module Rumoji
 module Rumoji
   class Emoji
   class Emoji
-    NATURE = Set[
+    NATURE = SortedSet[
       self.new("\u{2600}" , [:sunny], "BLACK SUN WITH RAYS"),
       self.new("\u{2600}" , [:sunny], "BLACK SUN WITH RAYS"),
       self.new("\u{2614}" , [:umbrella], "UMBRELLA WITH RAIN DROPS"),
       self.new("\u{2614}" , [:umbrella], "UMBRELLA WITH RAIN DROPS"),
       self.new("\u{2601}" , [:cloud]),
       self.new("\u{2601}" , [:cloud]),

lib/rumoji/emoji/objects.rb

@@ -5,7 +5,7 @@ require 'set'
 
 
 module Rumoji
 module Rumoji
   class Emoji
   class Emoji
-    OBJECTS = Set[
+    OBJECTS = SortedSet[
       self.new("\u{1F38D}", [:bamboo], "PINE DECORATION"), # "Japanese new year's door decoration
       self.new("\u{1F38D}", [:bamboo], "PINE DECORATION"), # "Japanese new year's door decoration
       self.new("\u{1F49D}", [:gift_heart], "HEART WITH RIBBON"),
       self.new("\u{1F49D}", [:gift_heart], "HEART WITH RIBBON"),
       self.new("\u{1F38E}", [:dolls], "JAPANESE DOLLS"), # "Japanese Hinamatsuri or girls' doll festival"
       self.new("\u{1F38E}", [:dolls], "JAPANESE DOLLS"), # "Japanese Hinamatsuri or girls' doll festival"

lib/rumoji/emoji/people.rb

@@ -5,7 +5,7 @@ require 'set'
 
 
 module Rumoji
 module Rumoji
   class Emoji
   class Emoji
-    PEOPLE = Set[
+    PEOPLE = SortedSet[
       self.new("\u{1F604}", [:smile, :simple_smile], "SMILING FACE WITH OPEN MOUTH AND SMILING EYES"),
       self.new("\u{1F604}", [:smile, :simple_smile], "SMILING FACE WITH OPEN MOUTH AND SMILING EYES"),
       self.new("\u{1F606}", [:laughing], "SMILING FACE WITH OPEN MOUTH AND TIGHTLY-CLOSED EYES"),
       self.new("\u{1F606}", [:laughing], "SMILING FACE WITH OPEN MOUTH AND TIGHTLY-CLOSED EYES"),
       self.new("\u{1F60A}", [:blush], "SMILING FACE WITH SMILING EYES"),
       self.new("\u{1F60A}", [:blush], "SMILING FACE WITH SMILING EYES"),

lib/rumoji/emoji/places.rb

@@ -5,7 +5,7 @@ require 'set'
 
 
 module Rumoji
 module Rumoji
   class Emoji
   class Emoji
-    PLACES = Set[
+    PLACES = SortedSet[
       self.new("\u{1f6a1}", [:aerial_tramway]),
       self.new("\u{1f6a1}", [:aerial_tramway]),
       self.new("\u{2708}" , [:airplane]),
       self.new("\u{2708}" , [:airplane]),
       self.new("\u{1f691}", [:ambulance]),
       self.new("\u{1f691}", [:ambulance]),

lib/rumoji/emoji/symbols.rb

@@ -5,7 +5,7 @@ require 'set'
 
 
 module Rumoji
 module Rumoji
   class Emoji
   class Emoji
-    SYMBOLS = Set[
+    SYMBOLS = SortedSet[
       self.new("\u{1f4af}", [:"100"]),
       self.new("\u{1f4af}", [:"100"]),
       self.new("\u{1f522}", [:"1234"]),
       self.new("\u{1f522}", [:"1234"]),
       self.new("\u{1f170}", [:a]),
       self.new("\u{1f170}", [:a]),

spec/rumoji_spec.rb

@@ -10,6 +10,7 @@ describe Rumoji do
     @zero = "0️⃣"
     @zero = "0️⃣"
     @us = "🇺🇸"
     @us = "🇺🇸"
     @non_potable_water = "🚱"
     @non_potable_water = "🚱"
+    @man_man_boy_boy = "👨‍👨‍👦‍👦"
   end
   end
 
 
   describe "#encode" do
   describe "#encode" do
@@ -23,6 +24,12 @@ describe Rumoji do
       Rumoji.encode(text).must_equal text
       Rumoji.encode(text).must_equal text
     end
     end
 
 
+    it "encodes man_man_boy_boy" do
+      text = "#{@man_man_boy_boy}"
+      Rumoji.encode(text).must_equal ":man-man-boy-boy:"
+      Rumoji.encode(text).wont_equal ":man::man::boy::boy:"
+    end
+
     describe "with multiple codepoints" do
     describe "with multiple codepoints" do
       it "transforms a stream" do
       it "transforms a stream" do
         Rumoji.encode("#{@zero}").must_equal ":zero:"
         Rumoji.encode("#{@zero}").must_equal ":zero:"