Merge pull request #22 from mavenlink/faster_encode

Faster encode technique

.gitignore

@@ -15,3 +15,4 @@ spec/reports
 test/tmp
 test/tmp
 test/version_tmp
 test/version_tmp
 tmp
 tmp
+.idea

bin/rumoji

@@ -3,4 +3,8 @@
 
 
 require 'rumoji'
 require 'rumoji'
 
 
-Rumoji.decode_io(STDIN, STDOUT)
+if ARGV.first == 'encode'
+  Rumoji.encode_io(STDIN, STDOUT)
+else
+  Rumoji.decode_io(STDIN, STDOUT)
+end

lib/rumoji.rb

@@ -8,8 +8,7 @@ module Rumoji
 
 
   # Transform emoji into its cheat-sheet code
   # Transform emoji into its cheat-sheet code
   def encode(str)
   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
   end
 
 
   # Transform a cheat-sheet code into an Emoji
   # Transform a cheat-sheet code into an Emoji
@@ -18,39 +17,9 @@ module Rumoji
   end
   end
 
 
   def encode_io(readable, writeable=StringIO.new(""))
   def encode_io(readable, writeable=StringIO.new(""))
-    previous_codepoints = []
-
-    readable.each_codepoint do |codepoint|
-      possible_emoji = Emoji.select_by_codepoint(codepoint)
-
-      sequence =  if possible_emoji.empty?
-                    # If this codepoint is not part of an emoji
-                    # just write it back out, along with any previously stored codepoints.
-                    dump_codepoints_to_string(previous_codepoints << codepoint)
-                  elsif !possible_emoji.any?(&:multiple?)
-                    # If this codepoint is part of an emoji, but not one
-                    # that contains multiple codepoints, write out the
-                    # first possiblity's cheat code and the previously stored codepoints
-                    dump_codepoints_to_string(previous_codepoints) << possible_emoji.first.code
-                  else
-                    # This codepoint is a part of an emoji with multiple
-                    # codepoints, so push it onto the stack.
-                    previous_codepoints.push(codepoint)
-                    # Check and see if this completes the emoji, otherwise,
-                    # write out an empty string.
-                    if found = possible_emoji.find {|e| e.codepoints.eql?(previous_codepoints) }
-                      previous_codepoints.clear and found.code
-                    else
-                      ""
-                    end
-                  end
-
-      writeable.write sequence
+    readable.each_line do |line|
+      writeable.write encode(line)
     end
     end
-
-    # Write any remaining codepoints.
-    writeable.write dump_codepoints_to_string(previous_codepoints) if previous_codepoints.any?
-
     writeable
     writeable
   end
   end
 
 
@@ -60,13 +29,4 @@ module Rumoji
     end
     end
     writeable
     writeable
   end
   end
-
-private
-
-  def dump_codepoints_to_string(codepoints)
-    codepoints.pack("U*").tap do
-      codepoints.clear
-    end
-  end
-
 end
 end

lib/rumoji/emoji.rb

@@ -2,9 +2,10 @@
 module Rumoji
 module Rumoji
   class Emoji
   class Emoji
 
 
-    attr_reader :name
+    attr_reader :name, :string
 
 
     def initialize(string, symbols, name = nil)
     def initialize(string, symbols, name = nil)
+      @string = string
       @codepoints = string.codepoints
       @codepoints = string.codepoints
       @cheat_codes = [symbols].flatten
       @cheat_codes = [symbols].flatten
       @name = name || @cheat_codes.first.to_s.upcase.gsub("_", " ")
       @name = name || @cheat_codes.first.to_s.upcase.gsub("_", " ")
@@ -23,7 +24,7 @@ module Rumoji
     end
     end
 
 
     def to_s
     def to_s
-      codepoints.pack("U*")
+      @string
     end
     end
 
 
     def hash
     def hash
@@ -31,7 +32,7 @@ module Rumoji
     end
     end
 
 
     def hex
     def hex
-      @codepoints.map{|point| point.to_s(16).upcase }.join("-")
+      @hex ||= @codepoints.map{|point| point.to_s(16).upcase }.join("-")
     end
     end
 
 
     def codepoints
     def codepoints
@@ -50,21 +51,18 @@ module Rumoji
 
 
     ALL = PEOPLE | NATURE | OBJECTS | PLACES | SYMBOLS
     ALL = PEOPLE | NATURE | OBJECTS | PLACES | SYMBOLS
 
 
+    ALL_REGEXP = Regexp.new(ALL.map(&:string).join('|'))
+
     def self.find(symbol)
     def self.find(symbol)
       ALL.find {|emoji| emoji.include? symbol }
       ALL.find {|emoji| emoji.include? symbol }
     end
     end
 
 
-    def self.find_by_string(string)
-      ALL.find {|emoji| emoji.to_s == string }
-    end
-
-    def self.select_by_codepoint(codepoint)
-      ALL.select {|emoji| emoji.codepoints.include? codepoint }
+    STRING_LOOKUP = ALL.each.with_object({}) do |emoji, lookup|
+      lookup[emoji.string] = emoji
     end
     end
 
 
-    def self.find_by_codepoint(codepoint)
-      ALL.find {|emoji| emoji.hex == codepoint.to_s(16).upcase }
+    def self.find_by_string(string)
+      STRING_LOOKUP[string]
     end
     end
-
   end
   end
 end
 end

lib/rumoji/emoji/symbols.rb

@@ -75,7 +75,7 @@ module Rumoji
       self.new("\u{1f564}", [:clock930]),
       self.new("\u{1f564}", [:clock930]),
       self.new("\u{3297}" , [:congratulations]),
       self.new("\u{3297}" , [:congratulations]),
       self.new("\u{1f192}", [:cool]),
       self.new("\u{1f192}", [:cool]),
-      self.new("\u{00a9}" , [:copyright]),
+      # self.new("\u{00a9}" , [:copyright]),
       self.new("\u{27b0}" , [:curly_loop]),
       self.new("\u{27b0}" , [:curly_loop]),
       self.new("\u{1f4b1}", [:currency_exchange]),
       self.new("\u{1f4b1}", [:currency_exchange]),
       self.new("\u{1f6c3}", [:customs]),
       self.new("\u{1f6c3}", [:customs]),
@@ -143,7 +143,7 @@ module Rumoji
       self.new("\u{1f518}", [:radio_button]),
       self.new("\u{1f518}", [:radio_button]),
       self.new("\u{267b}" , [:recycle]),
       self.new("\u{267b}" , [:recycle]),
       self.new("\u{1f534}", [:red_circle]),
       self.new("\u{1f534}", [:red_circle]),
-      self.new("\u{00ae}" , [:registered]),
+      # self.new("\u{00ae}" , [:registered]),
       self.new("\u{1f501}", [:repeat]),
       self.new("\u{1f501}", [:repeat]),
       self.new("\u{1f502}", [:repeat_one]),
       self.new("\u{1f502}", [:repeat_one]),
       self.new("\u{1f6bb}", [:restroom]),
       self.new("\u{1f6bb}", [:restroom]),

spec/rumoji/emoji_spec.rb

@@ -69,12 +69,5 @@ describe Rumoji::Emoji do
     it "finds an emoji from a string" do
     it "finds an emoji from a string" do
       subject.find_by_string(smile_str).symbol.must_equal smile_sym
       subject.find_by_string(smile_str).symbol.must_equal smile_sym
     end
     end
-
-    it "finds an emoji from a codepoint" do
-      smile_str.codepoints.map do |point|
-        subject.find_by_codepoint(point).symbol.must_equal smile_sym
-      end
-    end
   end
   end
-
 end
 end

spec/rumoji_spec.rb

@@ -18,6 +18,44 @@ describe Rumoji do
       Rumoji.encode(@smile).must_equal ":smile:"
       Rumoji.encode(@smile).must_equal ":smile:"
       Rumoji.encode("#{@smile}").must_equal ":smile:"
       Rumoji.encode("#{@smile}").must_equal ":smile:"
     end
     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
   end
 
 
   describe "#decode" do
   describe "#decode" do