a few improvements to the "russian" gem

lib/russian.rb

@@ -1,4 +1,4 @@
-# -*- encoding: utf-8 -*- 
+# -*- encoding: utf-8 -*-
 
 $KCODE = 'u' if RUBY_VERSION < "1.9"
 
@@ -9,9 +9,10 @@
 
 module Russian
   extend self
-  
+
   autoload :Transliteration, 'transliteration'
-
+  autoload :Metaphone, 'metaphone'
+
   # Russian locale
   LOCALE = :'ru'
 
@@ -21,11 +22,11 @@ def locale
   end
 
   # Regexp machers for context-based russian month names and day names translation
-  LOCALIZE_ABBR_MONTH_NAMES_MATCH = /(%d|%e)(.*)(%b)/
-  LOCALIZE_MONTH_NAMES_MATCH = /(%d|%e)(.*)(%B)/
+  LOCALIZE_ABBR_MONTH_NAMES_MATCH = /(%[-\d]?d|%e)(.*)(%b)/
+  LOCALIZE_MONTH_NAMES_MATCH = /(%[-\d]?d|%e)(.*)(%B)/
   LOCALIZE_STANDALONE_ABBR_DAY_NAMES_MATCH = /^%a/
   LOCALIZE_STANDALONE_DAY_NAMES_MATCH = /^%A/
-    
+
   # Init Russian i18n: load all translations shipped with library.
   def init_i18n
     I18n::Backend::Simple.send(:include, I18n::Backend::Pluralization)
@@ -39,23 +40,23 @@ def init_i18n
   # See I18n::translate
   def translate(key, options = {})
     I18n.translate(key, options.merge({ :locale => LOCALE }))
-  end        
+  end
   alias :t :translate
-  
+
   # See I18n::localize
   def localize(object, options = {})
     I18n.localize(object, options.merge({ :locale => LOCALE }))
   end
   alias :l :localize
-  
+
   # strftime() proxy with Russian localization
   def strftime(object, format = :default)
     localize(object, { :format => format })
   end
-  
+
   # Simple pluralization proxy
   #
-  # Usage: 
+  # Usage:
   #   Russian.pluralize(1, "вещь", "вещи", "вещей")
   #   Russian.pluralize(3.14, "вещь", "вещи", "вещей", "вещи")
   def pluralize(n, *variants)
@@ -76,13 +77,47 @@ def transliterate(str)
     Russian::Transliteration.transliterate(str)
   end
   alias :translit :transliterate
-
+
+  # De-transliteration for russian language
+  #
+  # Usage:
+  #  Russian.detranslit("rubin")
+  #  Russian.detransliterate("rubin")
+  def detransliterate(str)
+    Russian::Transliteration.detransliterate(str)
+  end
+  alias :detranslit :detransliterate
+
+  # Metaphone code for russian language
+  #
+  # Usage:
+  #  Russian.metaphone("рубин")
+  def metaphone(str)
+    Russian::Metaphone.generate(str)
+  end
+
+  # Change the input string as it would be typed in the standard russian keyboard layout
+  #
+  # Usage:
+  #  Russian.layout_rus("hemby") # рубин
+  def layout_rus(str)
+    Russian::Transliteration.layout_rus(str)
+  end
+
+  # Change the input string as it would be typed in the standard english keyboard layout
+  #
+  # Usage:
+  #  Russian.layout_eng("дум") # lev
+  def layout_eng(str)
+    Russian::Transliteration.layout_eng(str)
+  end
+
   protected
     # Returns all locale files shipped with library
     def locale_files
       Dir[File.join(File.dirname(__FILE__), "russian", "locale", "**/*")]
     end
-    
+
     # Converts an array of pluralization variants to a Hash that can be used
     # with I18n pluralization.
     def pluralization_variants_to_hash(*variants)

lib/russian/action_view_ext/helpers/date_helper.rb

@@ -1,20 +1,20 @@
-# -*- encoding: utf-8 -*- 
+# -*- encoding: utf-8 -*-
 
-# Заменяет хелпер Rails <tt>select_month</tt> и метод <tt>translated_month_names</tt> 
+# Заменяет хелпер Rails <tt>select_month</tt> и метод <tt>translated_month_names</tt>
 # для поддержки функционала "отдельностоящих имен месяцев".
 #
 # Теперь можно использовать и полные, и сокращенные название месяцев в двух вариантах -- контекстном
 # (по умолчанию) и отдельностоящем (если в текущем языке есть соответствующие переводы).
-# Теперь хелперы поддерживают ключ <tt>:use_standalone_month_names</tt>, хелпер <tt>select_month</tt> 
+# Теперь хелперы поддерживают ключ <tt>:use_standalone_month_names</tt>, хелпер <tt>select_month</tt>
 # устанавливает его по умолчанию.
 # Отдельностоящие имена месяцев также используютс когда указан ключ <tt>:discard_day</tt>.
 #
 #
 # Replaces Rails <tt>select_month</tt> helper and <tt>translated_month_names</tt> private method to provide
-# "standalone month names" feature. 
+# "standalone month names" feature.
 #
 # It is now possible to use both abbreviated and full month names in two variants (if current locale provides them).
-# All date helpers support <tt>:use_standalone_month_names</tt> key now, <tt>select_month</tt> helper sets 
+# All date helpers support <tt>:use_standalone_month_names</tt> key now, <tt>select_month</tt> helper sets
 # it to true by default.
 # Standalone month names are also used when <tt>:discard_day</tt> key is provided.
 if defined?(ActionView::Helpers::DateTimeSelector)
@@ -27,7 +27,7 @@ module DateHelper
         # instead of names -- set the <tt>:use_month_numbers</tt> key in +options+ to true for this to happen. If you
         # want both numbers and names, set the <tt>:add_month_numbers</tt> key in +options+ to true. If you would prefer
         # to show month names as abbreviations, set the <tt>:use_short_month</tt> key in +options+ to true. If you want
-        # to use your own month names, set the <tt>:use_month_names</tt> key in +options+ to an array of 12 month names. 
+        # to use your own month names, set the <tt>:use_month_names</tt> key in +options+ to an array of 12 month names.
         # You can also choose if you want to use i18n standalone month names or default month names -- you can
         # force standalone month names usage by using <tt>:use_standalone_month_names</tt> key.
         # Override the field name using the <tt>:field_name</tt> option, 'month' by default.
@@ -66,7 +66,7 @@ def select_month(date, options = {}, html_options = {})
           DateTimeSelector.new(date, options.merge(:use_standalone_month_names => true), html_options).select_month
         end
       end
-      
+
       class DateTimeSelector #:nodoc:
         private
           # Returns translated month names
@@ -108,11 +108,11 @@ def translated_month_names
                 key = :'date.month_names'
               end
             end
-            
+
             I18n.translate(key, :locale => @options[:locale])
           end
-          
+
       end
     end
   end
-end # if defined?
+end # if defined?

lib/russian/active_model_ext/custom_error_message.rb

@@ -1,4 +1,4 @@
-# -*- encoding: utf-8 -*- 
+# -*- encoding: utf-8 -*-
 
 if defined?(ActiveModel::Errors)
   module ActiveModel
@@ -13,19 +13,19 @@ class Errors
       # теперь не имеют префикса с названием атрибута если в сообщении об ошибке первым символом указан "^".
       #
       # Так, например,
-      # 
+      #
       #   validates_acceptance_of :accepted_terms, :message => 'нужно принять соглашение'
-      # 
+      #
       # даст сообщение
-      # 
+      #
       #   Accepted terms нужно принять соглашение
-      # 
+      #
       # однако,
-      # 
+      #
       #   validates_acceptance_of :accepted_terms, :message => '^Нужно принять соглашение'
-      # 
+      #
       # даст сообщение
-      # 
+      #
       #   Нужно принять соглашение
       #
       #

lib/russian/locale/actionview.yml

@@ -7,7 +7,7 @@ ru:
     # These are also the defaults for 'currency', 'percentage', 'precision', and 'human'
     format:
       # Sets the separator between the units, for more precision (e.g. 1.0 / 2.0 == 0.5)
-      separator: "."
+      separator: ","
       # Delimets thousands (e.g. 1,000,000 is a million) (always in groups of three)
       delimiter: " "
       # Number of decimals, behind the separator (the number 1 with a precision of 2 gives: 1.00)

lib/russian/locale/activemodel.yml

@@ -6,7 +6,7 @@ ru:
       inclusion: "имеет непредусмотренное значение"
       exclusion: "имеет зарезервированное значение"
       invalid: "имеет неверное значение"
-      confirmation: "не совпадает с подтверждением"
+      confirmation: "не совпадает с подтверждаемым значением"
       accepted: "нужно подтвердить"
       empty: "не может быть пустым"
       blank: "не может быть пустым"

lib/russian/locale/activerecord.yml

@@ -19,7 +19,7 @@ ru:
         inclusion: "имеет непредусмотренное значение"
         exclusion: "имеет зарезервированное значение"
         invalid: "имеет неверное значение"
-        confirmation: "не совпадает с подтверждением"
+        confirmation: "не совпадает с подтверждаемым значением"
         accepted: "нужно подтвердить"
         empty: "не может быть пустым"
         blank: "не может быть пустым"

lib/russian/locale/datetime.rb

@@ -1,4 +1,4 @@
-# -*- encoding: utf-8 -*- 
+# -*- encoding: utf-8 -*-
 
 # Context-based month name and day name switching for Russian
 #

lib/russian/locale/pluralization.rb

@@ -1,4 +1,4 @@
-# -*- encoding: utf-8 -*- 
+# -*- encoding: utf-8 -*-
 
 # Правило плюрализации для русского языка, взято из CLDR, http://unicode.org/cldr/
 #
@@ -19,10 +19,10 @@
   :ru => {
     :'i18n' => {
       :plural => {
-        :rule => lambda { |n| 
-          n % 10 == 1 && n % 100 != 11 ? :one : [2, 3, 4].include?(n % 10) && ![12, 13, 14].include?(n % 100) ? :few : n % 10 == 0 || [5, 6, 7, 8, 9].include?(n % 10) || [11, 12, 13, 14].include?(n % 100) ? :many : :other 
+        :rule => lambda { |n|
+          n % 10 == 1 && n % 100 != 11 ? :one : [2, 3, 4].include?(n % 10) && ![12, 13, 14].include?(n % 100) ? :few : n % 10 == 0 || [5, 6, 7, 8, 9].include?(n % 10) || [11, 12, 13, 14].include?(n % 100) ? :many : :other
         }
       }
     }
   }
-}
+}

lib/russian/locale/transliterator.rb

@@ -1,4 +1,4 @@
-# -*- encoding: utf-8 -*- 
+# -*- encoding: utf-8 -*-
 
 # I18n transliteration delegates to Russian::Transliteration (we're unable
 # to use common I18n transliteration tables with Russian)

lib/russian/metaphone.rb

@@ -0,0 +1,86 @@
+# -*- encoding: utf-8 -*-
+
+module Russian
+  # Metaphone code generation for english words and metaphon-like code for russian titles
+  #
+  # Генерация метафон-кодов для английский слов и русских названий
+  # (русская версия заточена и будет дальше дорабатываться именно
+  # в эту сторону - названия и заголовки)
+  module Metaphone
+    extend self
+
+    require "active_support/core_ext/string"
+
+    TRANSFORMATIONS_EN = [
+      [/\A[gkp]n/       ,  'n'],   # gn, kn, or pn at the start turns into 'n'
+      [/\Ax/            ,  's'],   # x at the start turns into 's'
+      [/\Awh/           ,  'w'],   # wh at the start turns into 'w'
+      [/mb\z/           ,  'm'],   # mb at the end turns into 'm'
+      [/sch/            ,  'sk'],  # sch sounds like 'sk'
+      [/x/              ,  'ks'],
+      [/cia/            ,  'xia'], # the 'c' -cia- and -ch- sounds like 'x'
+      [/ch/             ,  'xh'],
+      [/c([iey])/       ,  's\1'], # the 'c' -ce-, -ci-, or -cy- sounds like 's'
+      [/ck/             ,  'k'],
+      [/c/              ,  'k'],
+      [/dg([eiy])/      ,  'j\1'], # the 'dg' in -dge-, -dgi-, or -dgy- sounds like 'j'
+      [/d/              ,  't'],
+      [/gh/             ,  ''],
+      [/gned/           ,  'ned'],
+      [/gn((?![aeiou])|(\z))/ ,  'n'],
+      [/g[eiy]/         ,  'j'],
+      [/ph/             ,  'f'],
+      [/[aeiou]h(?!(?:[aeoiu]|$))/ ,  '\1'], # 'h' is silent after a vowel unless it's between vowels
+      [/q/              ,  'k'],
+      [/s(h|(ia)|(io))/ ,  'x\1'],
+      [/t((ia)|(io))/   ,  'x\1'],
+      [/v/              ,  'f'],
+      [/w(?![aeiou])/   ,  ''],
+      [/y(?![aeiou])/   ,  ''],
+      [/z/              ,  's'],
+#      [/th/             ,  '0'], # <-- zero ?!?
+      [/th/             ,  'z'],  # need only latin letters, no digits or smth else
+    ]
+
+    # english metaphone code was inspired by
+    # author: AndyV  http://snippets.dzone.com/user/AndyV
+    # source: http://snippets.dzone.com/posts/show/4112
+    def generate_en(aWord)
+      word = aWord.downcase
+      TRANSFORMATIONS_EN.each { |transform| word.gsub!(transform[0], transform[1]) }
+      word.squeeze
+      return word.present? ? (word[0] + word[1..-1].gsub(/[aeiou]/, '')).upcase : ""
+    end
+
+    TRANSFORMATIONS_RU = [
+      [ /[дт]ь?с/,'ц' ],      # seems this improves matching
+      [ /зз/, 'цц' ],         # for metaphone(detranslit(pizza)) == metaphone(пицца)
+      [ /[аяоёуюыиэеї]/, '' ],# remove vowels
+      [ /[йъь]/, '' ],        # these also ; this also removes all adjactive endings (-ый, -ая, -ое, ...)
+      [ /сч/,'ш' ],           # seems this improves matching
+#      [ /ч/, 'ц' ],
+      [ /[ґгк]/, 'х' ],
+      [ /б/, 'п' ],           # map pair letters
+      [ /в/, 'ф' ],
+      [ /д/, 'т' ],
+      [ /[жщ]/, 'ш' ],
+      [ /з/, 'с' ],
+    ]
+
+    def generate_ru(aWord)
+      word = aWord.mb_chars.downcase
+      TRANSFORMATIONS_RU.each { |transform| word.gsub!(transform[0], transform[1]) }
+      word.squeeze
+    end
+
+    # Generates a metaphone code for the given string
+    #
+    # Генерирует метафон-код заданной строки (в т.ч. русской, в т.ч. рус+eng)
+    def generate(str)
+      str.split(/\s+/).map do |s|
+        s =~ /[A-Za-z]/ ? generate_en(s =~ /[^A-Za-z]/ ? Russian.translit(s) : s) : generate_ru(s)
+      end .join(' ').mb_chars.upcase.gsub(/[^A-ZА-Я]+/, ' ').gsub(/\s+/, ' ').strip
+    end
+
+  end
+end

lib/russian/russian_rails.rb

@@ -4,5 +4,5 @@
 end
 
 if defined?(ActionView::Helpers)
-  require 'action_view_ext/helpers/date_helper' 
+  require 'action_view_ext/helpers/date_helper'
 end