Transliteration map fix

.rspec

@@ -0,0 +1 @@
+--color

Gemfile

@@ -1,4 +1,4 @@
 source "http://rubygems.org"
 
 gem "riot"
-
+gemspec

Gemfile.lock

@@ -1,12 +1,36 @@
+PATH
+  remote: .
+  specs:
+    translit (0.1.5)
+
 GEM
   remote: http://rubygems.org/
   specs:
+    diff-lcs (1.2.5)
     riot (0.12.5)
       rr
     rr (1.0.4)
+    rspec (3.5.0)
+      rspec-core (~> 3.5.0)
+      rspec-expectations (~> 3.5.0)
+      rspec-mocks (~> 3.5.0)
+    rspec-core (3.5.3)
+      rspec-support (~> 3.5.0)
+    rspec-expectations (3.5.0)
+      diff-lcs (>= 1.2.0, < 2.0)
+      rspec-support (~> 3.5.0)
+    rspec-mocks (3.5.0)
+      diff-lcs (>= 1.2.0, < 2.0)
+      rspec-support (~> 3.5.0)
+    rspec-support (3.5.0)
 
 PLATFORMS
   ruby
 
 DEPENDENCIES
   riot
+  rspec
+  translit!
+
+BUNDLED WITH
+   1.15.4

README.markdown

@@ -14,14 +14,14 @@ Or you can translit stdin now via just: <code>translit</code>. To stop stdin inp
 Translit using autodetection:
 
 ```
-Translit.convert("Отличный день") #=> Otlichnyj den'
-Translit.convert("Otlichnyj den'") #=> Отличный день
+Translit.convert("Отличный день") #=> "Otlichnyj den'"
+Translit.convert("Otlichnyj den'") #=> "Отличный день"
 ```
 Translit forcing target language:
 
 ```
-Translit.convert("Отличный den'", :russian) #=> Отличный деньOtlichnyj den'
-Translit.convert("Otlichnyj день", :english) #=> Otlichnyj den'
+Translit.convert("Отличный den'", :russian) #=> "Отличный день"
+Translit.convert("Otlichnyj день", :english) #=> "Otlichnyj den'"
 ```
 
 ## Русский ##
@@ -40,11 +40,11 @@ Translit.convert("Otlichnyj день", :english) #=> Otlichnyj den'
 Транслитирование с автоопределнием языка:
 
 ```
-Translit.convert("Отличный день") #=> Otlichnyj den'
-Translit.convert("Otlichnyj den'") #=> Отличный день
+Translit.convert("Отличный день") #=> "Otlichnyj den'"
+Translit.convert("Otlichnyj den'") #=> "Отличный день"
 ```
 Транслитирование на определенный язык:
 
 ```
-Translit.convert("Отличный den'", :russian) #=> Отличный деньOtlichnyj den'
-Translit.convert("Otlichnyj день", :english) #=> Otlichnyj den'
+Translit.convert("Отличный den'", :russian) #=> "Отличный день"
+Translit.convert("Otlichnyj день", :english) #=> "Otlichnyj den'"

lib/translit.rb

@@ -1,17 +1,24 @@
-# coding: utf-8
+# Resources for Ukrainian transliteration
+# * http://en.ukrlandia.com.ua/cyrillic-alphabets/
+# * http://www.ukrainiansintheuk.info/eng/00/translit-e.htm
 
 module Translit
+  # Ukrainian only chars: Ґ, І, Ї, Є, ‘
+  UKRAINIAN_ONLY_CHARS = %w(Ґ ґ І і Ї ї Є є ‘)
+
+  # Russian only chars: Э
+  RUSSIAN_ONLY_CHARS = %w(Э э)
+
   def self.convert!(text, enforce_language = nil)
     language = if enforce_language
-      enforce_input_language(enforce_language)
+      enforce_input_language(non_empty_line(text), enforce_language)
     else
-      detect_input_language(text.split(/\s+/).first)
+      detect_input_language(non_empty_line(text))
     end
 
-    map = self.send(language.to_s).sort_by {|k,v| v.length <=>  k.length}
+    map = self.send(language.to_s + "_to_" + enforce_language.to_s).sort_by {|k,v| k.length}.reverse
     map.each do |translit_key, translit_value|
-      text.gsub!(translit_key.capitalize, translit_value.first)
-      text.gsub!(translit_key, translit_value.last)
+      text.gsub!(translit_key, translit_value.first)
     end
     text
   end
@@ -21,37 +28,79 @@ def self.convert(text, enforce_language = nil)
   end
 
 private
-  def self.create_russian_map
-    self.english.inject({}) do |acc, tuple|
+
+  def self.non_empty_line(text)
+    text.split(/\s+/).select { |line| !line.empty? }.first
+  end
+
+  def self.invert_character_map(map)
+    map.dup.inject({}) do |acc, tuple|
+      rus = tuple.last.first
+      eng_value       = tuple.first
+      acc[rus] ? acc[rus] << eng_value            : acc[rus] = [eng_value]
+      acc
+    end
+  end
+
+  def self.latin_cases(map)
+    map.dup.inject({}) do |acc, tuple|
       rus_up, rus_low = tuple.last
       eng_value       = tuple.first
-      acc[rus_up]  ? acc[rus_up]  << eng_value.capitalize : acc[rus_up]  = [eng_value.capitalize]
-      acc[rus_low] ? acc[rus_low] << eng_value            : acc[rus_low] = [eng_value]
+      acc[eng_value] = [rus_low]
+      unless eng_value == eng_value.capitalize
+        acc[eng_value.capitalize] = [rus_up]
+      end
       acc
     end
   end
 
   def self.detect_input_language(text)
-    text.scan(/\w+/).empty? ? :russian : :english
+    if text && text.scan(/\w+/).empty?
+      slavic_language(text)
+    else
+      :english
+    end
   end
 
-  def self.enforce_input_language(language)
+  def self.enforce_input_language(text, language)
     if language == :english
-      :russian
+      slavic_language(text)
     else
       :english
     end
   end
 
-  def self.english
-    { "a"=>["А","а"], "b"=>["Б","б"], "v"=>["В","в"], "g"=>["Г","г"], "d"=>["Д","д"], "e"=>["Е","е"], "yo"=>["Ё","ё"], "jo"=>["Ё","ё"], "ö"=>["Ё","ё"], "zh"=>["Ж","ж"],
+  def self.slavic_language(text)
+    # If text contains Ukrainian chars we know it is Ukrainian
+    if UKRAINIAN_ONLY_CHARS.any? { |uk_char| text && text.include?(uk_char) }
+      :ukrainian
+    else
+      :russian
+    end
+  end
+
+  # Unsupported latin: "ä"=>["Э","э"], "ü"=>["Ю","ю"],
+  def self.english_to_ukrainian
+    @english_to_ukrainian ||= latin_cases({ "a"=>["А","а"], "b"=>["Б","б"], "v"=>["В","в"], "h"=>["Г","г"], "g"=>["Ґ","ґ"], "d"=>["Д","д"], "e"=>["Е","е"], "ye"=>["Є","є"], "je"=>["Є","є"], "zh"=>["Ж","ж"],
+      "z"=>["З","з"], "i"=>["І","і"], "yi"=>["Ї","ї"], "j"=>["Й","й"], "k"=>["К","к"], "l"=>["Л","л"], "m"=>["М","м"], "n"=>["Н","н"], "o"=>["О","о"], "p"=>["П","п"], "r"=>["Р","р"],
+      "s"=>["С","с"], "t"=>["Т","т"], "u"=>["У","у"], "f"=>["Ф","ф"], "kh"=>["Х","х"], "x"=>["Кс","кс"], "ts"=>["Ц","ц"], "ch"=>["Ч","ч"], "sh"=>["Ш","ш"], "w"=>["В","в"],
+      "shch"=>["Щ","щ"], "sch"=>["Щ","щ"], "y"=>["И","и"], "'"=>["Ь","ь"], "yu"=>["Ю","ю"], "ju"=>["Ю","ю"],
+      "ü"=>["Ю","ю"], "ya"=>["Я","я"], "ja"=>["Я","я"], "q"=>["К","к"]})
+  end
+
+  def self.english_to_russian
+    @english_to_russian ||= latin_cases({ "a"=>["А","а"], "b"=>["Б","б"], "v"=>["В","в"], "g"=>["Г","г"], "d"=>["Д","д"], "e"=>["Е","е"], "yo"=>["Ё","ё"], "jo"=>["Ё","ё"], "ö"=>["Ё","ё"], "zh"=>["Ж","ж"],
       "z"=>["З","з"], "i"=>["И","и"], "j"=>["Й","й"], "k"=>["К","к"], "l"=>["Л","л"], "m"=>["М","м"], "n"=>["Н","н"], "o"=>["О","о"], "p"=>["П","п"], "r"=>["Р","р"],
       "s"=>["С","с"], "t"=>["Т","т"], "u"=>["У","у"], "f"=>["Ф","ф"], "h"=>["Х","х"], "x"=>["Кс","кс"], "ts"=>["Ц","ц"], "ch"=>["Ч","ч"], "sh"=>["Ш","ш"], "w"=>["В","в"],
-      "shch"=>["Щ","щ"], "sch"=>["Щ","щ"], "#"=>["Ъ","ъ"], "y"=>["Ы","ы"], ""=>["Ь","ь"], "je"=>["Э","э"], "ä"=>["Э","э"], "yu"=>["Ю","ю"], "ju"=>["Ю","ю"],
-      "ü"=>["Ю","ю"], "ya"=>["Я","я"], "ja"=>["Я","я"], "q"=>["Я","я"]}
+      "shch"=>["Щ","щ"], "sch"=>["Щ","щ"], "#"=>["Ъ","ъ"], "y"=>["Ы","ы"], "'"=>["Ь","ь"], "je"=>["Э","э"], "ä"=>["Э","э"], "yu"=>["Ю","ю"], "ju"=>["Ю","ю"],
+      "ü"=>["Ю","ю"], "ya"=>["Я","я"], "ja"=>["Я","я"], "q"=>["Я","я"]})
+  end
+
+  def self.russian_to_english
+    @russian ||= invert_character_map(english_to_russian)
   end
 
-  def self.russian
-    @russian ||= create_russian_map
+  def self.ukrainian_to_english
+    @ukrainian ||= invert_character_map(english_to_ukrainian)
   end
 end

spec/translit_spec.rb

@@ -0,0 +1,60 @@
+require 'translit'
+
+describe "translit" do
+  let(:russian) { "Транслитерация между кириллицей и латиницей с коммандной строки или в твоей программе" }
+  let(:russian_transliteration) { "Transliteratsiya mezhdu kirillitsej i latinitsej s kommandnoj stroki ili v tvoej programme" }
+  let(:english) { "Transliteration between cyrillic <-> latin from command-line or your program" }
+  let(:ukrainian) { "Транслітерація між кирилицею <-> Латинська з командного рядка або вашої програми" }
+  let(:ukrainian_transliteration) { "Transliteratsiya mizh kyrylytseyu <-> Latyns'ka z komandnoho ryadka abo vashoyi prohramy" }
+  # This is broken
+  let(:english_transliteration) { "Транслитератион бетвеен cыриллиc <-> латин фром cомманд-лине ор ёур програм" }
+  let(:ukrainian_english_transliteration) { "Транслітератіон бетвеен cирілліc <-> латін фром cомманд-ліне ор иоур проґрам" }
+
+  it "transliterates from russian to english" do
+    expect(Translit.convert(russian, :english)).to eq(russian_transliteration)
+  end
+
+  it "transliterates from ukrainian to english" do
+    expect(Translit.convert(ukrainian, :english)).to eq(ukrainian_transliteration)
+  end
+
+  it "transliterates from english to russian" do
+    expect(Translit.convert(english, :russian)).to eq(english_transliteration)
+  end
+
+  it "transliterates from english to ukrainian" do
+    expect(Translit.convert(english, :ukrainian)).to eq(ukrainian_english_transliteration)
+  end
+
+  describe "ukrainian to english transliteration" do
+    it "should put yu ahead of ü" do
+      #"yu"=>["Ю","ю"], "ju"=>["Ю","ю"], "ü"=>["Ю","ю"]
+      expect(Translit.convert("Ю", :english)).to eq("Yu")
+    end
+
+    it "should put yu ahead of ü even at the end of a word" do
+      expect(Translit.convert("Біблію", :english)).to eq("Bibliyu")
+    end
+
+    it "should put ye ahead of je" do
+      #"ye"=>["Є","є"], "je"=>["Є","є"]
+      expect(Translit.convert("Є", :english)).to eq("Ye")
+    end
+  end
+
+  describe "english to ukrainian transliteration" do
+    it "should put yu ahead of ü even at the end of a word" do
+      expect(Translit.convert("Bible", :ukrainian)).to eq("Бібле")
+    end
+  end
+
+  describe "edge cases" do
+    it "should handle empty strings" do
+      expect(Translit.convert("", :english)).to eq("")
+    end
+
+    it "should handle strings with empty lines" do
+      expect(Translit.convert("\n\n", :english)).to eq("\n\n")
+    end
+  end
+end

translit.gemspec

@@ -1,7 +1,5 @@
 # coding: utf-8
 
-require 'rake'
-
 Gem::Specification.new do |s|
   s.name = %q{translit}
   s.version = '0.1.5'
@@ -11,12 +9,12 @@ Gem::Specification.new do |s|
   s.date = %q{2010-09-28}
   s.description = %q{Transliteration between cyrillic <-> latin | Транслитерация между кириллицей и латиницей }
   s.email = %q{tjbladez@gmail.com}
-  s.files = FileList['{bin,lib}/**/*', 'README.markdown'].to_a
   s.has_rdoc = false
   s.bindir = 'bin'
   s.executables = %w{translit}
   s.default_executable = 'bin/translit'
   s.homepage = %q{http://github.com/tjbladez/translit}
   s.summary = %q{Transliteration between cyrillic <-> latin from command-line or your program | Транслитерация между кириллицей и латиницей с коммандной строки или в твоей программе}
   s.post_install_message = %q{You are ready to transliterate | Вы готовы к транслитерации}
+  s.add_development_dependency "rspec"
 end