phonetic 1.1.0 → 1.2.0

data/lib/phonetic/double_metaphone/code.rb

@@ -0,0 +1,28 @@
+require 'phonetic/algorithm'
+
+module Phonetic
+  class DoubleMetaphone < Algorithm
+    class Code
+      def initialize
+        @codes = ['', '']
+      end
+
+      def add(primary, secondary)
+        @codes[0] += primary
+        @codes[1] += secondary
+      end
+
+      def results(size)
+        [first[0, size], last[0, size]]
+      end
+
+      def first
+        @codes.first
+      end
+
+      def last
+        @codes.last
+      end
+    end
+  end
+end

data/lib/phonetic/metaphone.rb

@@ -28,93 +28,40 @@ module Phonetic
       for n in 0..(l - 1)
         break unless metaph.size < code_size
         symb = w[n]
-        if symb == 'C' || n == 0 || w[n - 1] != symb
-          case
-          when vowel?(symb) && n == 0
-            metaph = symb
-          when symb == 'B'
-            metaph += symb if n != l - 1 || w[n - 1] != 'M'
-          when symb == 'C'
-            if n == 0 || w[n - 1] != 'S' || !front_vowel?(w[n + 1])
-              if w[n + 1, 2] == 'IA'
-                metaph += 'X'
-              elsif front_vowel?(w[n + 1])
-                metaph += 'S'
-              elsif n > 0 && w[n + 1] == 'H' && w[n - 1] == 'S'
-                metaph += 'K'
-              elsif w[n + 1] == 'H'
-                if n == 0 && !vowel?(w[n + 2])
-                  metaph += 'K'
-                else
-                  metaph += 'X'
-                end
-              else
-                metaph += 'K'
-              end
-            end
-          when symb == 'D'
-            if w[n + 1] == 'G' && front_vowel?(w[n + 2])
-              metaph += 'J'
-            else
-              metaph += 'T'
-            end
-          when symb == 'G'
-            silent = (w[n + 1] == 'H' && !vowel?(w[n + 2]))
-            if n > 0 && (w[n + 1] == 'N' || w[n + 1, 3] == 'NED')
-              silent = true
-            end
-            if n > 0 && w[n - 1] == 'D' && front_vowel?(w[n + 1])
-              silent = true
-            end
-            hard = (n > 0 && w[n - 1] == 'G')
-            unless silent
-              if front_vowel?(w[n + 1]) && !hard
-                metaph += 'J'
-              else
-                metaph += 'K'
-              end
-            end
-          when symb == 'H'
-            if !(n == l - 1 || (n > 0 && VARSON[w[n - 1]]))
-              metaph += 'H' if vowel?(w[n + 1])
-            end
-          when symb =~ /[FJLMNR]/
-            metaph += symb
-          when symb == 'K'
-            if n > 0 && w[n - 1] != 'C'
-              metaph += 'K'
-            elsif n == 0
-              metaph = 'K'
-            end
-          when symb == 'P'
-            metaph += w[n + 1] == 'H' ? 'F' : 'P'
-          when symb == 'Q'
-            metaph += 'K'
-          when symb == 'S'
-            if w[n + 1, 2] =~ /I[OA]/
-              metaph += 'X'
-            elsif w[n + 1] == 'H'
-              metaph += 'X'
-            else
-              metaph += 'S'
-            end
-          when symb == 'T'
-            if w[n + 1, 2] =~ /I[OA]/
-              metaph += 'X'
-            elsif w[n + 1] == 'H'
-              metaph += '0' if n == 0 || w[n - 1] != 'T'
-            else
-              metaph += 'T' if w[n + 1, 2] != 'CH'
-            end
-          when symb == 'V'
-            metaph += 'F'
-          when symb =~ /[WY]/
-            metaph += symb if vowel?(w[n + 1])
-          when symb == 'X'
-            metaph += 'KS'
-          when symb == 'Z'
-            metaph += 'S'
-          end
+        next unless symb == 'C' || n == 0 || w[n - 1] != symb
+        case
+        when vowel?(symb) && n == 0
+          metaph = symb
+        when symb == 'B'
+          metaph += symb if n != l - 1 || w[n - 1] != 'M'
+        when symb == 'C'
+          metaph += encode_c(w, n)
+        when symb == 'D'
+          metaph += encode_d(w, n)
+        when symb == 'G'
+          metaph += encode_g(w, n)
+        when symb == 'H'
+          metaph += encode_h(w, n)
+        when symb =~ /[FJLMNR]/
+          metaph += symb
+        when symb == 'K'
+          metaph += encode_k(w, n)
+        when symb == 'P'
+          metaph += w[n + 1] == 'H' ? 'F' : 'P'
+        when symb == 'Q'
+          metaph += 'K'
+        when symb == 'S'
+          metaph += encode_s(w, n)
+        when symb == 'T'
+          metaph += encode_t(w, n)
+        when symb == 'V'
+          metaph += 'F'
+        when symb =~ /[WY]/
+          metaph += symb if vowel?(w[n + 1])
+        when symb == 'X'
+          metaph += 'KS'
+        when symb == 'Z'
+          metaph += 'S'
         end
       end
       metaph
@@ -131,5 +78,94 @@ module Phonetic
       v = FRONT_VOWELS[symbol.to_s]
       !v.nil? && !v.empty?
     end
+
+    def self.encode_c(w, n)
+      metaph = ''
+      if n == 0 || w[n - 1] != 'S' || !front_vowel?(w[n + 1])
+        if w[n + 1, 2] == 'IA'
+          metaph = 'X'
+        elsif front_vowel?(w[n + 1])
+          metaph = 'S'
+        elsif n > 0 && w[n + 1] == 'H' && w[n - 1] == 'S'
+          metaph = 'K'
+        elsif w[n + 1] == 'H'
+          if n == 0 && !vowel?(w[n + 2])
+            metaph = 'K'
+          else
+            metaph = 'X'
+          end
+        else
+          metaph = 'K'
+        end
+      end
+      metaph
+    end
+
+    def self.encode_d(w, n)
+      w[n + 1] == 'G' && front_vowel?(w[n + 2]) ? 'J' : 'T'
+    end
+
+    def self.encode_g(w, n)
+      metaph = ''
+      silent = (w[n + 1] == 'H' && !vowel?(w[n + 2]))
+      if n > 0 && (w[n + 1] == 'N' || w[n + 1, 3] == 'NED')
+        silent = true
+      end
+      if n > 0 && w[n - 1] == 'D' && front_vowel?(w[n + 1])
+        silent = true
+      end
+      hard = (n > 0 && w[n - 1] == 'G')
+      unless silent
+        if front_vowel?(w[n + 1]) && !hard
+          metaph = 'J'
+        else
+          metaph = 'K'
+        end
+      end
+      metaph
+    end
+
+    def self.encode_h(w, n)
+      metaph = ''
+      unless n == w.size - 1 || (n > 0 && VARSON[w[n - 1]])
+        metaph = 'H' if vowel?(w[n + 1])
+      end
+      metaph
+    end
+
+    def self.encode_k(w, n)
+      metaph = ''
+      if n > 0 && w[n - 1] != 'C'
+        metaph = 'K'
+      elsif n == 0
+        metaph = 'K'
+      end
+      metaph
+    end
+
+    def self.encode_s(w, n)
+      metaph = ''
+      if w[n + 1, 2] =~ /I[OA]/
+        metaph = 'X'
+      elsif w[n + 1] == 'H'
+        metaph = 'X'
+      else
+        metaph = 'S'
+      end
+      metaph
+    end
+
+    def self.encode_t(w, n)
+      metaph = ''
+      if w[n + 1, 2] =~ /I[OA]/
+        metaph = 'X'
+      elsif w[n + 1] == 'H'
+        metaph = '0' if n == 0 || w[n - 1] != 'T'
+      else
+        metaph = 'T' if w[n + 1, 2] != 'CH'
+      end
+      metaph
+    end
+
   end
 end

data/lib/phonetic/refined_nysiis.rb

@@ -0,0 +1,72 @@
+require 'phonetic/algorithm'
+
+module Phonetic
+  # This class implements Refined NYSIIS algorithm.
+  # @see http://www.dropby.com/NYSIIS.html NYSIIS Code
+  # @example
+  #    Phonetic::RefinedNYSIIS.encode('Aumont') # => 'ANAD'
+  #    Phonetic::RefinedNYSIIS.encode('Schmidt') # => 'SNAD'
+  #    Phonetic::RefinedNYSIIS.encode('Phoenix') # => 'FANAC'
+  class RefinedNYSIIS < Algorithm
+    FIRST_MAP = {
+      /[SZ]+$/ => '',
+      /^MAC/ => 'MC',
+      /^PF/ =>  'F',
+      /IX$/  => 'IC',
+      /EX$/  => 'EC',
+      /(YE|EE|IE)$/ => 'Y',
+      /(DT|RT|RD|NT|ND)$/ => 'D',
+      /(.)EV/ => '\1EF'
+    }
+
+    SECOND_MAP = {
+      /([AEIOU])W/ => '\1',
+      /[AEIOU]+/ => 'A',
+      'GHT' => 'GT',
+      'DG' => 'G',
+      'PH' => 'F',
+      'AH' => 'A',
+      /(.)HA/ => '\1A',
+      'KN' => 'N',
+      'K' => 'C',
+      /(.)M/ => '\1N',
+      /(.)Q/ => '\1G',
+      'SH' => 'S',
+      'SCH' => 'S',
+      'YW' => 'Y',
+      /(.)Y(.)/ => '\1A\2',
+      'WR' => 'R',
+      /(.)Z/ => '\1S',
+      /AY$/ => 'Y',
+      /A+$/ => '',
+      /[^\w\s]|(.)(?=\1)/ => ''
+    }
+
+    # Convert string to Refined NYSIIS code
+    def self.encode(str, options = { trim: true })
+      self.encode_word(str, options)
+    end
+
+    # Convert word to its Refined NYSIIS code
+    def self.encode_word(word, options = { trim: true })
+      return '' if !word or word.empty?
+      trim = options[:trim]
+      w = word.upcase.strip
+      w.gsub! /\s([IV]+|[JS]R)$/, ''
+      w.gsub! /[^A-Z]/, ''
+      return if w.empty?
+      FIRST_MAP.each{ |rx, v| w.gsub!(rx, v) }
+      first_char = w[0]
+      SECOND_MAP.each{ |rx, v| w.gsub!(rx, v) }
+      w.gsub! /^A*/, first_char if vowel?(first_char)
+      w = w[0..5] if trim
+      w
+    end
+
+    private
+
+    def self.vowel?(char)
+      char =~ /^[AEIOU]/
+    end
+  end
+end

data/lib/phonetic/version.rb

@@ -1,3 +1,3 @@
 module Phonetic
-  VERSION = '1.1.0'
+  VERSION = '1.2.0'
 end

data/phonetic.gemspec

@@ -1,27 +1,29 @@
-# coding: utf-8
-lib = File.expand_path('../lib', __FILE__)
-$LOAD_PATH.unshift(lib) unless $LOAD_PATH.include?(lib)
-require 'phonetic/version'
-
-Gem::Specification.new do |spec|
-  spec.name          = 'phonetic'
-  spec.version       = Phonetic::VERSION
-  spec.authors       = ['n7v']
-  spec.email         = ['novsem@gmail.com']
-  spec.description   = %q{Ruby library for phonetic algorithms. It supports Soundex, Metaphone, Caverphone, NYSIIS and others.}
-  spec.summary       = %q{Ruby library for phonetic algorithms.}
-  spec.homepage      = 'http://github.com/n7v/phonetic'
-  spec.license       = 'MIT'
-
-  spec.files         = `git ls-files`.split($/)
-  spec.executables   = spec.files.grep(%r{^bin/}) { |f| File.basename(f) }
-  spec.test_files    = spec.files.grep(%r{^(test|spec|features)/})
-  spec.require_paths = ['lib']
-
-  spec.add_development_dependency 'bundler', '~> 1.3'
-  spec.add_development_dependency 'rake'
-  spec.add_development_dependency 'rspec'
-  spec.add_development_dependency 'simplecov'
-  spec.add_development_dependency 'coveralls'
-  spec.add_development_dependency 'yard'
-end
+# coding: utf-8
+lib = File.expand_path('../lib', __FILE__)
+$LOAD_PATH.unshift(lib) unless $LOAD_PATH.include?(lib)
+require 'phonetic/version'
+
+Gem::Specification.new do |spec|
+  spec.name          = 'phonetic'
+  spec.version       = Phonetic::VERSION
+  spec.authors       = ['n7v']
+  spec.email         = ['novsem@gmail.com']
+  spec.description   = %q{Ruby library for phonetic algorithms. It supports Soundex, Metaphone, Caverphone, NYSIIS and others.}
+  spec.summary       = %q{Ruby library for phonetic algorithms.}
+  spec.homepage      = 'http://github.com/n7v/phonetic'
+  spec.license       = 'MIT'
+
+  spec.files         = `git ls-files`.split($/)
+  spec.executables   = spec.files.grep(%r{^bin/}) { |f| File.basename(f) }
+  spec.test_files    = spec.files.grep(%r{^(test|spec|features)/})
+  spec.require_paths = ['lib']
+
+  spec.add_development_dependency 'bundler', '~> 1.3'
+  spec.add_development_dependency 'rake'
+  spec.add_development_dependency 'rspec'
+  spec.add_development_dependency 'simplecov'
+  spec.add_development_dependency 'coveralls'
+  spec.add_development_dependency 'yard'
+
+  spec.required_ruby_version = '>= 1.9'
+end

data/spec/phonetic/caverphone2_spec.rb

@@ -1,61 +1,10 @@
 require 'spec_helper'
-
-# Examples are taken from Caversham Project's paper
-# http://caversham.otago.ac.nz/files/working/ctp150804.pdf
-CAVERPHONE2_TEST_TABLE = {
-  'STFNSN1111' => ['Stevenson'],
-  'PTA1111111' => ['Peter'],
-  'AT11111111' => [
-    'add', 'aid', 'at', 'art', 'eat', 'earth', 'head', 'hit', 'hot',
-    'hold', 'hard', 'heart', 'it', 'out', 'old'
-  ],
-  'RTA1111111' => ['rather', 'ready', 'writer'],
-  'SSA1111111' => ['social'],
-  'APA1111111' => ['able', 'appear'],
-  'TTA1111111' => [
-    'Darda', 'Datha', 'Dedie', 'Deedee', 'Deerdre', 'Deidre', 'Deirdre',
-    'Detta', 'Didi', 'Didier', 'Dido', 'Dierdre', 'Dieter', 'Dita',
-    'Ditter', 'Dodi', 'Dodie', 'Dody', 'Doherty', 'Dorthea', 'Dorthy',
-    'Doti', 'Dotti', 'Dottie', 'Dotty', 'Doty', 'Doughty', 'Douty',
-    'Dowdell', 'Duthie', 'Tada', 'Taddeo', 'Tadeo', 'Tadio', 'Tati',
-    'Teador', 'Tedda', 'Tedder', 'Teddi', 'Teddie', 'Teddy', 'Tedi',
-    'Tedie', 'Teeter', 'Teodoor', 'Teodor', 'Terti', 'Theda', 'Theodor',
-    'Theodore', 'Theta', 'Thilda', 'Thordia', 'Tilda', 'Tildi', 'Tildie',
-    'Tildy', 'Tita', 'Tito', 'Tjader', 'Toddie', 'Toddy', 'Torto', 'Tuddor',
-    'Tudor', 'Turtle', 'Tuttle', 'Tutto'
-  ],
-  'KLN1111111' => [
-    'Cailean', 'Calan', 'Calen', 'Callahan', 'Callan', 'Callean',
-    'Carleen', 'Carlen', 'Carlene', 'Carlin', 'Carline', 'Carlyn',
-    'Carlynn', 'Carlynne', 'Charlean', 'Charleen', 'Charlene',
-    'Charline', 'Cherlyn', 'Chirlin', 'Clein', 'Cleon', 'Cline',
-    'Cohleen', 'Colan', 'Coleen', 'Colene', 'Colin', 'Colleen',
-    'Collen', 'Collin', 'Colline', 'Colon', 'Cullan', 'Cullen',
-    'Cullin', 'Gaelan', 'Galan', 'Galen', 'Garlan', 'Garlen',
-    'Gaulin', 'Gayleen', 'Gaylene', 'Giliane', 'Gillan', 'Gillian',
-    'Glen', 'Glenn', 'Glyn', 'Glynn', 'Gollin', 'Gorlin', 'Kalin',
-    'Karlan', 'Karleen', 'Karlen', 'Karlene', 'Karlin', 'Karlyn',
-    'Kaylyn', 'Keelin', 'Kellen', 'Kellene', 'Kellyann', 'Kellyn',
-    'Khalin', 'Kilan', 'Kilian', 'Killen', 'Killian', 'Killion',
-    'Klein', 'Kleon', 'Kline', 'Koerlin', 'Kylen', 'Kylynn', 'Quillan',
-    'Quillon', 'Qulllon', 'Xylon'
-  ],
-  'TN11111111' => [
-    'Dan', 'Dane', 'Dann', 'Darn', 'Daune', 'Dawn', 'Ddene', 'Dean', 'Deane',
-    'Deanne', 'DeeAnn', 'Deeann', 'Deeanne', 'Deeyn', 'Den', 'Dene', 'Denn',
-    'Deonne', 'Diahann', 'Dian', 'Diane', 'Diann', 'Dianne', 'Diannne', 'Dine',
-    'Dion', 'Dione', 'Dionne', 'Doane', 'Doehne', 'Don', 'Donn', 'Doone', 'Dorn',
-    'Down', 'Downe', 'Duane', 'Dun', 'Dunn', 'Duyne', 'Dyan', 'Dyane', 'Dyann',
-    'Dyanne', 'Dyun', 'Tan', 'Tann', 'Teahan', 'Ten', 'Tenn', 'Terhune', 'Thain',
-    'Thaine', 'Thane', 'Thanh', 'Thayne', 'Theone', 'Thin', 'Thorn', 'Thorne',
-    'Thun', 'Thynne', 'Tien', 'Tine', 'Tjon', 'Town', 'Towne', 'Turne', 'Tyne'
-  ]
-}
+require 'support/caverphone2_data'
 
 describe Phonetic::Caverphone2 do
   describe '.encode' do
     it 'should return Caverphone 2 value of string' do
-      CAVERPHONE2_TEST_TABLE.each do |code, words|
+      Phonetic::CAVERPHONE2_TEST_TABLE.each do |code, words|
         words.each do |word|
           res = Phonetic::Caverphone2.encode(word)
           res.should eq(code), "expected: #{code}\ngot: #{res}\nword: #{word}"