phonetic 1.0.0

data/spec/phonetic/core_ext/string/caverphone2_spec.rb

@@ -0,0 +1,9 @@
+require 'phonetic'
+
+describe String do
+  describe '#caverphone2' do
+    it 'should return Caverphone 2 value of string' do
+      'Deerdre'.caverphone2.should == 'TTA1111111'
+    end
+  end
+end

data/spec/phonetic/core_ext/string/caverphone_spec.rb

@@ -0,0 +1,9 @@
+require 'spec_helper'
+
+describe String do
+  describe '#caverphone' do
+    it 'should return Caverphone value of string' do
+      'Lashaunda'.caverphone.should == 'LSNT11'
+    end
+  end
+end

data/spec/phonetic/core_ext/string/double_metaphone_spec.rb

@@ -0,0 +1,15 @@
+require 'phonetic'
+
+describe String do
+  describe '#double_metaphone' do
+    it 'should return Double Metaphone code of string' do
+      'Gelatin'.double_metaphone.should == ['KLTN', 'JLTN']
+    end
+  end
+
+  describe '#metaphone2' do
+    it 'should return Double Metaphone code of string' do
+      'whirlpool'.metaphone2.should == ['ARLP', 'ARLP']
+    end
+  end
+end

data/spec/phonetic/core_ext/string/metaphone_spec.rb

@@ -0,0 +1,11 @@
+require 'phonetic'
+
+describe String do
+  describe '#metaphone' do
+    it 'should return Metaphone code of string' do
+      %w(Karen Carina Corin Crain Garneau Gorrono Krenn).each do |name|
+        name.metaphone.should == 'KRN'
+      end
+    end
+  end
+end

data/spec/phonetic/core_ext/string/nysiis_spec.rb

@@ -0,0 +1,12 @@
+require 'spec_helper'
+require 'support/nysiis_data'
+
+describe String do
+  describe '#nysiis' do
+    it 'should return NYSIIS code of word' do
+      Phonetic::NYSIIS_TEST_TABLE.each do |word, result|
+        word.nysiis(trim: false).should == result
+      end
+    end
+  end
+end

data/spec/phonetic/core_ext/string/refined_soundex_spec.rb

@@ -0,0 +1,10 @@
+require 'spec_helper'
+
+describe String do
+  describe '#refined_soundex' do
+    it 'should return refined soundex code of word' do
+      'Braz Caren Hayers'.refined_soundex.should == 'B1905 C30908 H093'
+      'Lambard Noulton'.refined_soundex.should == 'L7081096 N807608'
+    end
+  end
+end

data/spec/phonetic/core_ext/string/soundex_spec.rb

@@ -0,0 +1,14 @@
+require 'spec_helper'
+
+describe String do
+  describe '#soundex' do
+    it 'should return soundex code of word' do
+      'Ackerman'.soundex.should == 'A265'
+      'Fusedale'.soundex.should == 'F234'
+      'Grahl'.soundex.should == 'G640'
+      'Hatchard'.soundex.should == 'H326'
+      'implementation'.soundex.should == 'I514'
+      'Prewett'.soundex.should == 'P630'
+    end
+  end
+end

data/spec/phonetic/double_metaphone_spec.rb

@@ -0,0 +1,16 @@
+require 'spec_helper'
+require 'support/double_metaphone_data'
+
+describe Phonetic::DoubleMetaphone do
+  describe '.encode' do
+    it 'should return Double Metaphone codes of string' do
+      Phonetic::DOUBLE_METAPHONE_TEST_TABLE.each do |w, r|
+        Phonetic::DoubleMetaphone.encode(w).should == r
+      end
+    end
+
+    it 'should support max size of the code option' do
+      Phonetic::DoubleMetaphone.encode('accede', size: 3).should == ['AKS', 'AKS']
+    end
+  end
+end

data/spec/phonetic/metaphone2_spec.rb

@@ -0,0 +1,9 @@
+require 'spec_helper'
+
+describe Phonetic::Metaphone2 do
+  describe '.encode' do
+    it 'should return Double Metaphone codes of string' do
+      Phonetic::Metaphone2.encode('biaggi').should == ['PJ', 'PK']
+    end
+  end
+end

data/spec/phonetic/metaphone_spec.rb

@@ -0,0 +1,81 @@
+require 'spec_helper'
+
+describe Phonetic::Metaphone do
+  describe '.encode' do
+    it 'should drop duplicate adjacent letters, except for C' do
+      Phonetic::Metaphone.encode('Accola Nikki Dillon').should == 'AKKL NK TLN'
+    end
+
+    it 'should drop the first letter if the word begins with KN, GN, PN, AE, WR' do
+      Phonetic::Metaphone.encode('Knapp Gnome Phantom Aeneid Wright').should == 'NP NM FNTM ENT RT'
+    end
+
+    it 'should drop B if after M at the end of the word' do
+      Phonetic::Metaphone.encode('Plumb Amber').should == 'PLM AMBR'
+    end
+
+    it 'should transform CIA to XIA' do
+      Phonetic::Metaphone.encode('Ciara Phylicia').should == 'XR FLX'
+    end
+
+    it 'should transform CH to X if it is not the part of -SCH-' do
+      Phonetic::Metaphone.encode('Dratch Heche Christina').should == 'TRX HX KRST'
+    end
+
+    it 'should transform -SCH- to -SKH-' do
+      Phonetic::Metaphone.encode('School Deschanel').should == 'SKL TSKN'
+    end
+
+    it 'should transform C to S if followed by I, E, or Y. And to K otherwise' do
+      Phonetic::Metaphone.encode('Felicity Joyce Nancy Carol C').should == 'FLST JS NNS KRL K'
+    end
+
+    it 'should transform D to J if followed by GE, GY, or GI. And to T otherwise' do
+      Phonetic::Metaphone.encode('Coolidge Lodgy Dixie Pidgin D').should == 'KLJ LJ TKS PJN T'
+    end
+
+    it 'should drop G if followed by H and H is not at the end or before a vowel' do
+      Phonetic::Metaphone.encode('Knight Clayburgh Haughton Monaghan').should == 'NT KLBR HTN MNKN'
+    end
+
+    it 'should transform G to J if before I, E, or Y, and to K other otherwise' do
+      Phonetic::Metaphone.encode('Gigi Gena Peggy Gyllenhaal Madigan G').should == 'JJ JN PK JLNH MTKN K'
+    end
+
+    it 'should drop G if followed by N or NED and is at the end' do
+      Phonetic::Metaphone.encode('GN GNE GNED Agnes Signed').should == 'N N NT ANS SNT'
+    end
+
+    it 'should drop H if after vowel and not before a vowel' do
+      Phonetic::Metaphone.encode('Sarah Shahi Moorehead Poehler').should == 'SR XH MRHT PLR'
+    end
+
+    it 'should transform CK to K, PH to F, Q to K, V to F, Z to S' do
+      Phonetic::Metaphone.encode('Acker Sophia Quinn Victoria Zelda Karen').should == 'AKR SF KN FKTR SLT KRN'
+    end
+
+    it 'should transform SH to XH, SIO to XIO, SIA to XIA' do
+      Phonetic::Metaphone.encode('Ashley Siobhan Siamese').should == 'AXL XBHN XMS'
+    end
+
+    it 'should transform TIA to XIA, TIO to XIO, TH to 0, TCH to CH' do
+      Phonetic::Metaphone.encode('Tia Portia Interaction Catherine Dratch Fletcher').should == 'X PRX INTR K0RN TRX FLXR'
+    end
+
+    it 'should transform WH to W at the beginning, drop W if not followed by a vowel' do
+      Phonetic::Metaphone.encode('Whoopi Goodwin Harlow Hawley').should == 'WP KTWN HRL HL'
+    end
+
+    it 'should transform X to S at the beginning and X to KS otherwise' do
+      Phonetic::Metaphone.encode('Xor Alexis X').should == 'SR ALKS S'
+    end
+
+    it 'should drop Y if not followed by a vowel' do
+      Phonetic::Metaphone.encode('Yasmine Blonsky Blyth').should == 'YSMN BLNS BL0'
+    end
+
+    it 'should drop all vowels unless it is the beginning' do
+      Phonetic::Metaphone.encode('Boardman Mary Ellen').should == 'BRTM MR ELN'
+    end
+  end
+end

data/spec/phonetic/nysiis_spec.rb

@@ -0,0 +1,20 @@
+require 'spec_helper'
+require 'support/nysiis_data'
+
+describe Phonetic::NYSIIS do
+  describe '.encode' do
+    it 'should return NYSIIS code of word' do
+      Phonetic::NYSIIS_TEST_TABLE.each do |word, result|
+        Phonetic::NYSIIS.encode(word, trim: false).should == result
+      end
+    end
+
+    it 'should return empty string for empty word' do
+      Phonetic::NYSIIS.encode('').should == ''
+    end
+
+    it 'should ignore non-english symbols in input' do
+      Phonetic::NYSIIS.encode('1234567890+-= Bess $').should == 'BAS'
+    end
+  end
+end

data/spec/phonetic/refined_soundex_spec.rb

@@ -0,0 +1,13 @@
+require 'spec_helper'
+
+describe Phonetic::RefinedSoundex do
+  describe '.encode' do
+    it 'should return refined soundex code of word' do
+      Phonetic::RefinedSoundex.encode('Braz').should == 'B1905'
+      Phonetic::RefinedSoundex.encode('Caren').should == 'C30908'
+      Phonetic::RefinedSoundex.encode('Hayers').should == 'H093'
+      Phonetic::RefinedSoundex.encode('Lambard').should == 'L7081096'
+      Phonetic::RefinedSoundex.encode('Noulton').should == 'N807608'
+    end
+  end
+end

data/spec/phonetic/soundex_spec.rb

@@ -0,0 +1,24 @@
+require 'spec_helper'
+
+describe Phonetic::Soundex do
+  describe '.encode' do
+    it 'should return soundex code of word' do
+      Phonetic::Soundex.encode('Ackerman').should == 'A265'
+      Phonetic::Soundex.encode('Fusedale').should == 'F234'
+      Phonetic::Soundex.encode('Grahl').should == 'G640'
+      Phonetic::Soundex.encode('Hatchard').should == 'H326'
+      Phonetic::Soundex.encode('implementation').should == 'I514'
+      Phonetic::Soundex.encode('Prewett').should == 'P630'
+    end
+
+    it 'should add zeros if result less then 4 symbols' do
+      Phonetic::Soundex.encode('ammonium').should == 'A500'
+      Phonetic::Soundex.encode('Rubin').should == 'R150'
+      Phonetic::Soundex.encode('H').should == 'H000'
+    end
+
+    it 'should return empty string for empty word' do
+      Phonetic::Soundex.encode('').should == ''
+    end
+  end
+end

data/spec/spec_helper.rb

@@ -0,0 +1,11 @@
+require 'simplecov'
+SimpleCov.start
+
+require 'phonetic'
+
+RSpec.configure do |config|
+  config.treat_symbols_as_metadata_keys_with_true_values = true
+  config.run_all_when_everything_filtered = true
+  config.filter_run :focus
+  config.order = 'random'
+end

data/spec/support/double_metaphone_data.rb

@@ -0,0 +1,142 @@
+# encoding: utf-8
+
+module Phonetic
+  DOUBLE_METAPHONE_TEST_TABLE = {
+    'accede'       => ['AKST', 'AKST'],
+    'accident'     => ['AKST', 'AKST'],
+    'avvenente'    => ['AFNN', 'AFNN'],
+    'Acker'        => ['AKR', 'AKR'],
+    'Agnes'        => ['AKNS', 'ANS'],
+    'Akkad'        => ['AKT', 'AKT'],
+    'Alicia'       => ['ALS', 'ALX'],
+    'Allasio'      => ['ALS', 'ALX'],
+    'aqqa'         => ['AK', 'AK'],
+    'arch'         => ['ARX', 'ARK'],
+    'architect'    => ['ARKT', 'ARKT'],
+    'Arnoff'       => ['ARNF', 'ARNF'],
+    'Arnow'        => ['ARN', 'ARNF'],
+    'Assia'        => ['AS', 'AS'],
+    'artois'       => ['ART', 'ARTS'],
+    'bacchus'      => ['PKS', 'PKS'],
+    'bacci'        => ['PX', 'PX'],
+    'bajador'      => ['PJTR', 'PHTR'],
+    'Bajja'        => ['PJ', 'PJ'],
+    'Barzizza'     => ['PRSS', 'PRST'],
+    'bellocchio'   => ['PLX', 'PLX'],
+    'bertucci'     => ['PRTX', 'PRTX'],
+    'biaggi'       => ['PJ', 'PK'],
+    'breaux'       => ['PR', 'PR'],
+    'braço'        => ['PRS', 'PRS'],
+    'cabrillo'     => ['KPRL', 'KPR'],
+    'Cacgia'       => ['KK', 'KK'],
+    'caesar'       => ['SSR', 'SSR'],
+    'cagney'       => ['KKN', 'KKN'],
+    'campbell'     => ['KMPL', 'KMPL'],
+    'carlisle'     => ['KRLL', 'KRLL'],
+    'carlysle'     => ['KRLL', 'KRLL'],
+    'chemistry'    => ['KMST', 'KMST'],
+    'cherry'       => ['XR', 'XR'],
+    'chianti'      => ['KNT', 'KNT'],
+    'chorus'       => ['KRS', 'KRS'],
+    'Cielo'        => ['SL', 'XL'],
+    'Cioccolato'   => ['SKLT', 'XKLT'],
+    'cough'        => ['KF', 'KF'],
+    'czerny'       => ['SRN', 'XRN'],
+    'dumb'         => ['TM', 'TM'],
+    'edgar'        => ['ATKR', 'ATKR'],
+    'edge'         => ['AJ', 'AJ'],
+    'exam'         => ['AKSM', 'AKSM'],
+    'exceed'       => ['AKST', 'AKST'],
+    'filipowicz'   => ['FLPT', 'FLPF'],
+    'focaccia'     => ['FKX', 'FKX'],
+    'gallegos'     => ['KLKS', 'KKS'],
+    'Gebal'        => ['KPL', 'JPL'],
+    'Gelatin'      => ['KLTN', 'JLTN'],
+    'Gepard'       => ['KPRT', 'JPRT'],
+    'Gerald'       => ['KRLT', 'JRLT'],
+    'Gesture'      => ['KSTR', 'JSTR'],
+    'geyser'       => ['KSR', 'JSR'],
+    'ghiradelli'   => ['JRTL', 'JRTL'],
+    'ghislane'     => ['JLN', 'JLN'],
+    'ghost'        => ['KST', 'KST'],
+    'gibbon'       => ['KPN', 'JPN'],
+    'Gilbert'      => ['KLPR', 'JLPR'],
+    'ginger'       => ['KNKR', 'JNJR'],
+    'Giethoorn'    => ['K0RN', 'JTRN'],
+    'gough'        => ['KF', 'KF'],
+    'Grashof'      => ['KRXF', 'KRXF'],
+    'hochmeier'    => ['HKMR', 'HKMR'],
+    'hugh'         => ['H', 'H'],
+    'island'       => ['ALNT', 'ALNT'],
+    'isle'         => ['AL', 'AL'],
+    'Jankelowicz'  => ['JNKL', 'ANKL'],
+    'Jacqueline'   => ['JKLN', 'AKLN'],
+    'dejer'        => ['TJR', 'TJR'],
+    'jose'         => ['HS', 'HS'],
+    'joseph'       => ['JSF', 'HSF'],
+    'Jugnot'       => ['JNT', 'AKNT'],
+    'Katerine'     => ['KTRN', 'KTRN'],
+    'laugh'        => ['LF', 'LF'],
+    'Lawrence'     => ['LRNS', 'LRNS'],
+    'Loretta'      => ['LRT', 'LRT'],
+    'mac caffrey'  => ['MKFR', 'MKFR'],
+    'mac gregor'   => ['MKRK', 'MKRK'],
+    'Maggie'       => ['MJ', 'MK'],
+    'maña'         => ['MN', 'MN'],
+    'McClellan'    => ['MKLL', 'MKLL'],
+    'McHugh'       => ['MK', 'MK'],
+    'McLaughlin'   => ['MKLF', 'MKLF'],
+    'michael'      => ['MKL', 'MXL'],
+    'Monaghan'     => ['MNKN','MNKN'],
+    'Moosholzer'   => ['MSLS', 'MSLS'],
+    'Mosheim'      => ['MSM', 'MSM'],
+    'nation'       => ['NXN', 'NXN'],
+    'orchestra'    => ['ARKS', 'ARKS'],
+    'orchid'       => ['ARKT', 'ARKT'],
+    'quest'        => ['KST', 'KST'],
+    'pizza'        => ['PS', 'PTS'],
+    'Portia'       => ['PRX', 'PRX'],
+    'raspberry'    => ['RSPR', 'RSPR'],
+    'resnais'      => ['RSN', 'RSNS'],
+    'Rogge'        => ['RK', 'RK'],
+    'rogier'       => ['RJ', 'RJR'],
+    'rough'        => ['RF', 'RF'],
+    'raj'          => ['RJ', 'R'],
+    'san jacinto'  => ['SNHS', 'SNHS'],
+    'Scarlett'     => ['SKRL', 'SKRL'],
+    'schenker'     => ['XNKR', 'SKNK'],
+    'schermerhorn' => ['XRMR', 'SKRM'],
+    'schmidt'      => ['XMT', 'SMT'],
+    'schneider'    => ['XNTR', 'SNTR'],
+    'school'       => ['SKL', 'SKL'],
+    'schooner'     => ['SKNR', 'SKNR'],
+    'Sciorra'      => ['SR', 'SR'],
+    'Sholman'      => ['SLMN', 'SLMN'],
+    'Shoeka'       => ['SK', 'SK'],
+    'smith'        => ['SM0', 'XMT'],
+    'snider'       => ['SNTR', 'XNTR'],
+    'succeed'      => ['SKST', 'SKST'],
+    'sugar'        => ['XKR', 'SKR'],
+    'szamos'       => ['SMS', 'XMS'],
+    'tagliaro'     => ['TKLR', 'TLR'],
+    'thames'       => ['TMS', 'TMS'],
+    'thomas'       => ['TMS', 'TMS'],
+    'thumb'        => ['0M', 'TM'],
+    'tichner'      => ['TXNR', 'TKNR'],
+    'tough'        => ['TF', 'TF'],
+    'Uomo'         => ['AM', 'AM'],
+    'van gelder'   => ['FNKL', 'FNKL'],
+    'Vasserman'    => ['FSRM', 'FSRM'],
+    'wachtler'     => ['AKTL', 'FKTL'],
+    'Wasserman'    => ['ASRM', 'FSRM'],
+    'weschsler'    => ['AXSL', 'FXSL'],
+    'Wesia'        => ['AS', 'FS'],
+    'Womo'         => ['AM', 'FM'],
+    'Winningham'   => ['ANNK', 'FNNK'],
+    'whirlpool'    => ['ARLP', 'ARLP'],
+    'wright'       => ['RT', 'RT'],
+    'Xavier'       => ['SF', 'SFR'],
+    'Yankelovich'  => ['ANKL', 'ANKL'],
+    'zhao'         => ['J', 'J']
+  }
+end

data/spec/support/nysiis_data.rb

@@ -0,0 +1,31 @@
+module Phonetic
+  NYSIIS_TEST_TABLE = {
+    'Alexandra'     => 'ALAXANDR',
+    'Aumont'        => 'AANAD',
+    'Bonnie'        => 'BANY',
+    'Christensen'   => 'CHRASTANSAN',
+    'Cleveland'     => 'CLAFALAD',
+    'Claudia'       => 'CLAD',
+    'Dedee'         => 'DADY',
+    'DeLaurentiis'  => 'DALARANT',
+    'Echikunwoke'   => 'ECACANWAC',
+    'Fahey'         => 'FAHY',
+    'Jacqueline'    => 'JACGALAN',
+    'John'          => 'J',
+    'Hessel'        => 'HASAL',
+    'Hubert'        => 'HABAD',
+    'Howard'        => 'HAD',
+    'Knuth'         => 'NNAT',
+    'Kepler'        => 'CAPLAR',
+    'Marguerite'    => 'MARGARAT',
+    'Smith'         => 'SNAT',
+    'Schelte'       => 'SSALT',
+    'Macdonald'     => 'MCDANALD',
+    'Michael'       => 'MACAL',
+    'Phoenix'       => 'FFANAX',
+    'Pfeiffer'      => 'FFAFAR',
+    'Rebecca'       => 'RABAC',
+    'Rosalind'      => 'RASALAD',
+    'Schmidt'       => 'SSNAD'
+  }
+end