Coverage Report - org.apache.commons.codec.language.DoubleMetaphone
 
Classes in this File Line Coverage Branch Coverage Complexity
DoubleMetaphone
98%
365/372
90%
399/443
6.08
DoubleMetaphone$DoubleMetaphoneResult
100%
36/36
100%
12/12
6.08
 
 1   
 /*
 2   
  * Licensed to the Apache Software Foundation (ASF) under one or more
 3   
  * contributor license agreements.  See the NOTICE file distributed with
 4   
  * this work for additional information regarding copyright ownership.
 5   
  * The ASF licenses this file to You under the Apache License, Version 2.0
 6   
  * (the "License"); you may not use this file except in compliance with
 7   
  * the License.  You may obtain a copy of the License at
 8   
  *
 9   
  *      http://www.apache.org/licenses/LICENSE-2.0
 10   
  *
 11   
  * Unless required by applicable law or agreed to in writing, software
 12   
  * distributed under the License is distributed on an "AS IS" BASIS,
 13   
  * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 14   
  * See the License for the specific language governing permissions and
 15   
  * limitations under the License.
 16   
  */
 17   
 
 18   
 package org.apache.commons.codec.language;
 19   
 
 20   
 import org.apache.commons.codec.EncoderException;
 21   
 import org.apache.commons.codec.StringEncoder;
 22   
 
 23   
 /**
 24   
  * Encodes a string into a double metaphone value. This Implementation is based on the algorithm by <CITE>Lawrence
 25   
  * Philips</CITE>.
 26   
  * <p>
 27   
  * This class is conditionally thread-safe. The instance field {@link #maxCodeLen} is mutable
 28   
  * {@link #setMaxCodeLen(int)} but is not volatile, and accesses are not synchronized. If an instance of the class is
 29   
  * shared between threads, the caller needs to ensure that suitable synchronization is used to ensure safe publication
 30   
  * of the value between threads, and must not invoke {@link #setMaxCodeLen(int)} after initial setup.
 31   
  *
 32   
  * @see <a href="http://drdobbs.com/184401251?pgno=2">Original Article</a>
 33   
  * @see <a href="http://en.wikipedia.org/wiki/Metaphone">http://en.wikipedia.org/wiki/Metaphone</a>
 34   
  *
 35   
  * @version $Id: DoubleMetaphone.java 1435550 2013-01-19 14:09:52Z tn $
 36   
  */
 37   
 public class DoubleMetaphone implements StringEncoder {
 38   
 
 39   
     /**
 40   
      * "Vowels" to test for
 41   
      */
 42   
     private static final String VOWELS = "AEIOUY";
 43   
 
 44   
     /**
 45   
      * Prefixes when present which are not pronounced
 46   
      */
 47  1 
     private static final String[] SILENT_START =
 48   
         { "GN", "KN", "PN", "WR", "PS" };
 49  1 
     private static final String[] L_R_N_M_B_H_F_V_W_SPACE =
 50   
         { "L", "R", "N", "M", "B", "H", "F", "V", "W", " " };
 51  1 
     private static final String[] ES_EP_EB_EL_EY_IB_IL_IN_IE_EI_ER =
 52   
         { "ES", "EP", "EB", "EL", "EY", "IB", "IL", "IN", "IE", "EI", "ER" };
 53  1 
     private static final String[] L_T_K_S_N_M_B_Z =
 54   
         { "L", "T", "K", "S", "N", "M", "B", "Z" };
 55   
 
 56   
     /**
 57   
      * Maximum length of an encoding, default is 4
 58   
      */
 59  24 
     private int maxCodeLen = 4;
 60   
 
 61   
     /**
 62   
      * Creates an instance of this DoubleMetaphone encoder
 63   
      */
 64   
     public DoubleMetaphone() {
 65  24 
         super();
 66  24 
     }
 67   
 
 68   
     /**
 69   
      * Encode a value with Double Metaphone.
 70   
      *
 71   
      * @param value String to encode
 72   
      * @return an encoded string
 73   
      */
 74   
     public String doubleMetaphone(final String value) {
 75  84 
         return doubleMetaphone(value, false);
 76   
     }
 77   
 
 78   
     /**
 79   
      * Encode a value with Double Metaphone, optionally using the alternate encoding.
 80   
      *
 81   
      * @param value String to encode
 82   
      * @param alternate use alternate encode
 83   
      * @return an encoded string
 84   
      */
 85   
     public String doubleMetaphone(String value, final boolean alternate) {
 86  6506 
         value = cleanInput(value);
 87  6506 
         if (value == null) {
 88  12 
             return null;
 89   
         }
 90   
 
 91  6494 
         final boolean slavoGermanic = isSlavoGermanic(value);
 92  6494 
         int index = isSilentStart(value) ? 1 : 0;
 93   
 
 94  6494 
         final DoubleMetaphoneResult result = new DoubleMetaphoneResult(this.getMaxCodeLen());
 95   
 
 96  42099 
         while (!result.isComplete() && index <= value.length() - 1) {
 97  35605 
             switch (value.charAt(index)) {
 98   
             case 'A':
 99   
             case 'E':
 100   
             case 'I':
 101   
             case 'O':
 102   
             case 'U':
 103   
             case 'Y':
 104  13685 
                 index = handleAEIOUY(result, index);
 105  13685 
                 break;
 106   
             case 'B':
 107  884 
                 result.append('P');
 108  884 
                 index = charAt(value, index + 1) == 'B' ? index + 2 : index + 1;
 109  884 
                 break;
 110   
             case '\u00C7':
 111   
                 // A C with a Cedilla
 112  1 
                 result.append('S');
 113  1 
                 index++;
 114  1 
                 break;
 115   
             case 'C':
 116  1680 
                 index = handleC(value, result, index);
 117  1680 
                 break;
 118   
             case 'D':
 119  1238 
                 index = handleD(value, result, index);
 120  1238 
                 break;
 121   
             case 'F':
 122  646 
                 result.append('F');
 123  646 
                 index = charAt(value, index + 1) == 'F' ? index + 2 : index + 1;
 124  646 
                 break;
 125   
             case 'G':
 126  801 
                 index = handleG(value, result, index, slavoGermanic);
 127  801 
                 break;
 128   
             case 'H':
 129  521 
                 index = handleH(value, result, index);
 130  521 
                 break;
 131   
             case 'J':
 132  87 
                 index = handleJ(value, result, index, slavoGermanic);
 133  87 
                 break;
 134   
             case 'K':
 135  324 
                 result.append('K');
 136  324 
                 index = charAt(value, index + 1) == 'K' ? index + 2 : index + 1;
 137  324 
                 break;
 138   
             case 'L':
 139  1797 
                 index = handleL(value, result, index);
 140  1797 
                 break;
 141   
             case 'M':
 142  1241 
                 result.append('M');
 143  1241 
                 index = conditionM0(value, index) ? index + 2 : index + 1;
 144  1241 
                 break;
 145   
             case 'N':
 146  2777 
                 result.append('N');
 147  2777 
                 index = charAt(value, index + 1) == 'N' ? index + 2 : index + 1;
 148  2777 
                 break;
 149   
             case '\u00D1':
 150   
                 // N with a tilde (spanish ene)
 151  1 
                 result.append('N');
 152  1 
                 index++;
 153  1 
                 break;
 154   
             case 'P':
 155  1145 
                 index = handleP(value, result, index);
 156  1145 
                 break;
 157   
             case 'Q':
 158  80 
                 result.append('K');
 159  80 
                 index = charAt(value, index + 1) == 'Q' ? index + 2 : index + 1;
 160  80 
                 break;
 161   
             case 'R':
 162  2739 
                 index = handleR(value, result, index, slavoGermanic);
 163  2739 
                 break;
 164   
             case 'S':
 165  2153 
                 index = handleS(value, result, index, slavoGermanic);
 166  2153 
                 break;
 167   
             case 'T':
 168  2224 
                 index = handleT(value, result, index);
 169  2224 
                 break;
 170   
             case 'V':
 171  406 
                 result.append('F');
 172  406 
                 index = charAt(value, index + 1) == 'V' ? index + 2 : index + 1;
 173  406 
                 break;
 174   
             case 'W':
 175  519 
                 index = handleW(value, result, index);
 176  519 
                 break;
 177   
             case 'X':
 178  152 
                 index = handleX(value, result, index);
 179  152 
                 break;
 180   
             case 'Z':
 181  97 
                 index = handleZ(value, result, index, slavoGermanic);
 182  97 
                 break;
 183   
             default:
 184  407 
                 index++;
 185  407 
                 break;
 186   
             }
 187   
         }
 188   
 
 189  6494 
         return alternate ? result.getAlternate() : result.getPrimary();
 190   
     }
 191   
 
 192   
     /**
 193   
      * Encode the value using DoubleMetaphone.  It will only work if
 194   
      * <code>obj</code> is a <code>String</code> (like <code>Metaphone</code>).
 195   
      *
 196   
      * @param obj Object to encode (should be of type String)
 197   
      * @return An encoded Object (will be of type String)
 198   
      * @throws EncoderException encode parameter is not of type String
 199   
      */
 200   
     @Override
 201   
     public Object encode(final Object obj) throws EncoderException {
 202  37 
         if (!(obj instanceof String)) {
 203  3 
             throw new EncoderException("DoubleMetaphone encode parameter is not of type String");
 204   
         }
 205  34 
         return doubleMetaphone((String) obj);
 206   
     }
 207   
 
 208   
     /**
 209   
      * Encode the value using DoubleMetaphone.
 210   
      *
 211   
      * @param value String to encode
 212   
      * @return An encoded String
 213   
      */
 214   
     @Override
 215   
     public String encode(final String value) {
 216  30 
         return doubleMetaphone(value);
 217   
     }
 218   
 
 219   
     /**
 220   
      * Check if the Double Metaphone values of two <code>String</code> values
 221   
      * are equal.
 222   
      *
 223   
      * @param value1 The left-hand side of the encoded {@link String#equals(Object)}.
 224   
      * @param value2 The right-hand side of the encoded {@link String#equals(Object)}.
 225   
      * @return {@code true} if the encoded <code>String</code>s are equal;
 226   
      *          {@code false} otherwise.
 227   
      * @see #isDoubleMetaphoneEqual(String,String,boolean)
 228   
      */
 229   
     public boolean isDoubleMetaphoneEqual(final String value1, final String value2) {
 230  22 
         return isDoubleMetaphoneEqual(value1, value2, false);
 231   
     }
 232   
 
 233   
     /**
 234   
      * Check if the Double Metaphone values of two <code>String</code> values
 235   
      * are equal, optionally using the alternate value.
 236   
      *
 237   
      * @param value1 The left-hand side of the encoded {@link String#equals(Object)}.
 238   
      * @param value2 The right-hand side of the encoded {@link String#equals(Object)}.
 239   
      * @param alternate use the alternate value if {@code true}.
 240   
      * @return {@code true} if the encoded <code>String</code>s are equal;
 241   
      *          {@code false} otherwise.
 242   
      */
 243   
     public boolean isDoubleMetaphoneEqual(final String value1, final String value2, final boolean alternate) {
 244  1970 
         return doubleMetaphone(value1, alternate).equals(doubleMetaphone(value2, alternate));
 245   
     }
 246   
 
 247   
     /**
 248   
      * Returns the maxCodeLen.
 249   
      * @return int
 250   
      */
 251   
     public int getMaxCodeLen() {
 252  19484 
         return this.maxCodeLen;
 253   
     }
 254   
 
 255   
     /**
 256   
      * Sets the maxCodeLen.
 257   
      * @param maxCodeLen The maxCodeLen to set
 258   
      */
 259   
     public void setMaxCodeLen(final int maxCodeLen) {
 260  1 
         this.maxCodeLen = maxCodeLen;
 261  1 
     }
 262   
 
 263   
     //-- BEGIN HANDLERS --//
 264   
 
 265   
     /**
 266   
      * Handles 'A', 'E', 'I', 'O', 'U', and 'Y' cases.
 267   
      */
 268   
     private int handleAEIOUY(final DoubleMetaphoneResult result, final int index) {
 269  13685 
         if (index == 0) {
 270  1515 
             result.append('A');
 271   
         }
 272  13685 
         return index + 1;
 273   
     }
 274   
 
 275   
     /**
 276   
      * Handles 'C' cases.
 277   
      */
 278   
     private int handleC(final String value, final DoubleMetaphoneResult result, int index) {
 279  1680 
         if (conditionC0(value, index)) {  // very confusing, moved out
 280  16 
             result.append('K');
 281  16 
             index += 2;
 282  1664 
         } else if (index == 0 && contains(value, index, 6, "CAESAR")) {
 283  6 
             result.append('S');
 284  6 
             index += 2;
 285  1658 
         } else if (contains(value, index, 2, "CH")) {
 286  156 
             index = handleCH(value, result, index);
 287  1502 
         } else if (contains(value, index, 2, "CZ") &&
 288   
                    !contains(value, index - 2, 4, "WICZ")) {
 289   
             //-- "Czerny" --//
 290  7 
             result.append('S', 'X');
 291  7 
             index += 2;
 292  1495 
         } else if (contains(value, index + 1, 3, "CIA")) {
 293   
             //-- "focaccia" --//
 294  2 
             result.append('X');
 295  2 
             index += 3;
 296  1493 
         } else if (contains(value, index, 2, "CC") &&
 297   
                    !(index == 1 && charAt(value, 0) == 'M')) {
 298   
             //-- double "cc" but not "McClelland" --//
 299  109 
             return handleCC(value, result, index);
 300  1384 
         } else if (contains(value, index, 2, "CK", "CG", "CQ")) {
 301  111 
             result.append('K');
 302  111 
             index += 2;
 303  1273 
         } else if (contains(value, index, 2, "CI", "CE", "CY")) {
 304   
             //-- Italian vs. English --//
 305  286 
             if (contains(value, index, 3, "CIO", "CIE", "CIA")) {
 306  46 
                 result.append('S', 'X');
 307   
             } else {
 308  240 
                 result.append('S');
 309   
             }
 310  286 
             index += 2;
 311   
         } else {
 312  987 
             result.append('K');
 313  987 
             if (contains(value, index + 1, 2, " C", " Q", " G")) {
 314   
                 //-- Mac Caffrey, Mac Gregor --//
 315  4 
                 index += 3;
 316  983 
             } else if (contains(value, index + 1, 1, "C", "K", "Q") &&
 317   
                        !contains(value, index + 1, 2, "CE", "CI")) {
 318  9 
                 index += 2;
 319   
             } else {
 320  974 
                 index++;
 321   
             }
 322   
         }
 323   
 
 324  1571 
         return index;
 325   
     }
 326   
 
 327   
     /**
 328   
      * Handles 'CC' cases.
 329   
      */
 330   
     private int handleCC(final String value, final DoubleMetaphoneResult result, int index) {
 331  109 
         if (contains(value, index + 2, 1, "I", "E", "H") &&
 332   
             !contains(value, index + 2, 2, "HU")) {
 333   
             //-- "bellocchio" but not "bacchus" --//
 334  22 
             if ((index == 1 && charAt(value, index - 1) == 'A') ||
 335   
                 contains(value, index - 1, 5, "UCCEE", "UCCES")) {
 336   
                 //-- "accident", "accede", "succeed" --//
 337  14 
                 result.append("KS");
 338   
             } else {
 339   
                 //-- "bacci", "bertucci", other Italian --//
 340  8 
                 result.append('X');
 341   
             }
 342  22 
             index += 3;
 343   
         } else {    // Pierce's rule
 344  87 
             result.append('K');
 345  87 
             index += 2;
 346   
         }
 347   
 
 348  109 
         return index;
 349   
     }
 350   
 
 351   
     /**
 352   
      * Handles 'CH' cases.
 353   
      */
 354   
     private int handleCH(final String value, final DoubleMetaphoneResult result, final int index) {
 355  156 
         if (index > 0 && contains(value, index, 4, "CHAE")) {   // Michael
 356  0 
             result.append('K', 'X');
 357  0 
             return index + 2;
 358  156 
         } else if (conditionCH0(value, index)) {
 359   
             //-- Greek roots ("chemistry", "chorus", etc.) --//
 360  4 
             result.append('K');
 361  4 
             return index + 2;
 362  152 
         } else if (conditionCH1(value, index)) {
 363   
             //-- Germanic, Greek, or otherwise 'ch' for 'kh' sound --//
 364  34 
             result.append('K');
 365  34 
             return index + 2;
 366   
         } else {
 367  118 
             if (index > 0) {
 368  82 
                 if (contains(value, 0, 2, "MC")) {
 369  2 
                     result.append('K');
 370   
                 } else {
 371  80 
                     result.append('X', 'K');
 372   
                 }
 373   
             } else {
 374  36 
                 result.append('X');
 375   
             }
 376  118 
             return index + 2;
 377   
         }
 378   
     }
 379   
 
 380   
     /**
 381   
      * Handles 'D' cases.
 382   
      */
 383   
     private int handleD(final String value, final DoubleMetaphoneResult result, int index) {
 384  1238 
         if (contains(value, index, 2, "DG")) {
 385   
             //-- "Edge" --//
 386  10 
             if (contains(value, index + 2, 1, "I", "E", "Y")) {
 387  4 
                 result.append('J');
 388  4 
                 index += 3;
 389   
                 //-- "Edgar" --//
 390   
             } else {
 391  6 
                 result.append("TK");
 392  6 
                 index += 2;
 393   
             }
 394  1228 
         } else if (contains(value, index, 2, "DT", "DD")) {
 395  38 
             result.append('T');
 396  38 
             index += 2;
 397   
         } else {
 398  1190 
             result.append('T');
 399  1190 
             index++;
 400   
         }
 401  1238 
         return index;
 402   
     }
 403   
 
 404   
     /**
 405   
      * Handles 'G' cases.
 406   
      */
 407   
     private int handleG(final String value, final DoubleMetaphoneResult result, int index,
 408   
                         final boolean slavoGermanic) {
 409  801 
         if (charAt(value, index + 1) == 'H') {
 410  106 
             index = handleGH(value, result, index);
 411  695 
         } else if (charAt(value, index + 1) == 'N') {
 412  26 
             if (index == 1 && isVowel(charAt(value, 0)) && !slavoGermanic) {
 413  0 
                 result.append("KN", "N");
 414  26 
             } else if (!contains(value, index + 2, 2, "EY") &&
 415   
                        charAt(value, index + 1) != 'Y' && !slavoGermanic) {
 416  22 
                 result.append("N", "KN");
 417   
             } else {
 418  4 
                 result.append("KN");
 419   
             }
 420  26 
             index = index + 2;
 421  669 
         } else if (contains(value, index + 1, 2, "LI") && !slavoGermanic) {
 422  4 
             result.append("KL", "L");
 423  4 
             index += 2;
 424  665 
         } else if (index == 0 &&
 425   
                    (charAt(value, index + 1) == 'Y' ||
 426   
                     contains(value, index + 1, 2, ES_EP_EB_EL_EY_IB_IL_IN_IE_EI_ER))) {
 427   
             //-- -ges-, -gep-, -gel-, -gie- at beginning --//
 428  16 
             result.append('K', 'J');
 429  16 
             index += 2;
 430  649 
         } else if ((contains(value, index + 1, 2, "ER") ||
 431   
                     charAt(value, index + 1) == 'Y') &&
 432   
                    !contains(value, 0, 6, "DANGER", "RANGER", "MANGER") &&
 433   
                    !contains(value, index - 1, 1, "E", "I") &&
 434   
                    !contains(value, index - 1, 3, "RGY", "OGY")) {
 435   
             //-- -ger-, -gy- --//
 436  22 
             result.append('K', 'J');
 437  22 
             index += 2;
 438  627 
         } else if (contains(value, index + 1, 1, "E", "I", "Y") ||
 439   
                    contains(value, index - 1, 4, "AGGI", "OGGI")) {
 440   
             //-- Italian "biaggi" --//
 441  182 
             if (contains(value, 0 ,4, "VAN ", "VON ") ||
 442   
                 contains(value, 0, 3, "SCH") ||
 443   
                 contains(value, index + 1, 2, "ET")) {
 444   
                 //-- obvious germanic --//
 445  2 
                 result.append('K');
 446  180 
             } else if (contains(value, index + 1, 3, "IER")) {
 447  4 
                 result.append('J');
 448   
             } else {
 449  176 
                 result.append('J', 'K');
 450   
             }
 451  182 
             index += 2;
 452  445 
         } else if (charAt(value, index + 1) == 'G') {
 453  34 
             index += 2;
 454  34 
             result.append('K');
 455   
         } else {
 456  411 
             index++;
 457  411 
             result.append('K');
 458   
         }
 459  801 
         return index;
 460   
     }
 461   
 
 462   
     /**
 463   
      * Handles 'GH' cases.
 464   
      */
 465   
     private int handleGH(final String value, final DoubleMetaphoneResult result, int index) {
 466  106 
         if (index > 0 && !isVowel(charAt(value, index - 1))) {
 467  4 
             result.append('K');
 468  4 
             index += 2;
 469  102 
         } else if (index == 0) {
 470  8 
             if (charAt(value, index + 2) == 'I') {
 471  4 
                 result.append('J');
 472   
             } else {
 473  4 
                 result.append('K');
 474   
             }
 475  8 
             index += 2;
 476  94 
         } else if ((index > 1 && contains(value, index - 2, 1, "B", "H", "D")) ||
 477   
                    (index > 2 && contains(value, index - 3, 1, "B", "H", "D")) ||
 478   
                    (index > 3 && contains(value, index - 4, 1, "B", "H"))) {
 479   
             //-- Parker's rule (with some further refinements) - "hugh"
 480  28 
             index += 2;
 481   
         } else {
 482  66 
             if (index > 2 && charAt(value, index - 1) == 'U' &&
 483   
                 contains(value, index - 3, 1, "C", "G", "L", "R", "T")) {
 484   
                 //-- "laugh", "McLaughlin", "cough", "gough", "rough", "tough"
 485  22 
                 result.append('F');
 486  44 
             } else if (index > 0 && charAt(value, index - 1) != 'I') {
 487  8 
                 result.append('K');
 488   
             }
 489  66 
             index += 2;
 490   
         }
 491  106 
         return index;
 492   
     }
 493   
 
 494   
     /**
 495   
      * Handles 'H' cases.
 496   
      */
 497   
     private int handleH(final String value, final DoubleMetaphoneResult result, int index) {
 498   
         //-- only keep if first & before vowel or between 2 vowels --//
 499  521 
         if ((index == 0 || isVowel(charAt(value, index - 1))) &&
 500   
             isVowel(charAt(value, index + 1))) {
 501  387 
             result.append('H');
 502  387 
             index += 2;
 503   
             //-- also takes car of "HH" --//
 504   
         } else {
 505  134 
             index++;
 506   
         }
 507  521 
         return index;
 508   
     }
 509   
 
 510   
     /**
 511   
      * Handles 'J' cases.
 512   
      */
 513   
     private int handleJ(final String value, final DoubleMetaphoneResult result, int index,
 514   
                         final boolean slavoGermanic) {
 515  87 
         if (contains(value, index, 4, "JOSE") || contains(value, 0, 4, "SAN ")) {
 516   
                 //-- obvious Spanish, "Jose", "San Jacinto" --//
 517  11 
                 if ((index == 0 && (charAt(value, index + 4) == ' ') ||
 518   
                      value.length() == 4) || contains(value, 0, 4, "SAN ")) {
 519  9 
                     result.append('H');
 520   
                 } else {
 521  2 
                     result.append('J', 'H');
 522   
                 }
 523  11 
                 index++;
 524   
             } else {
 525  76 
                 if (index == 0 && !contains(value, index, 4, "JOSE")) {
 526  48 
                     result.append('J', 'A');
 527  28 
                 } else if (isVowel(charAt(value, index - 1)) && !slavoGermanic &&
 528   
                            (charAt(value, index + 1) == 'A' || charAt(value, index + 1) == 'O')) {
 529  10 
                     result.append('J', 'H');
 530  18 
                 } else if (index == value.length() - 1) {
 531  0 
                     result.append('J', ' ');
 532  18 
                 } else if (!contains(value, index + 1, 1, L_T_K_S_N_M_B_Z) &&
 533   
                            !contains(value, index - 1, 1, "S", "K", "L")) {
 534  18 
                     result.append('J');
 535   
                 }
 536   
 
 537  76 
                 if (charAt(value, index + 1) == 'J') {
 538  0 
                     index += 2;
 539   
                 } else {
 540  76 
                     index++;
 541   
                 }
 542   
             }
 543  87 
         return index;
 544   
     }
 545   
 
 546   
     /**
 547   
      * Handles 'L' cases.
 548   
      */
 549   
     private int handleL(final String value, final DoubleMetaphoneResult result, int index) {
 550  1797 
         if (charAt(value, index + 1) == 'L') {
 551  353 
             if (conditionL0(value, index)) {
 552  4 
                 result.appendPrimary('L');
 553   
             } else {
 554  349 
                 result.append('L');
 555   
             }
 556  353 
             index += 2;
 557   
         } else {
 558  1444 
             index++;
 559  1444 
             result.append('L');
 560   
         }
 561  1797 
         return index;
 562   
     }
 563   
 
 564   
     /**
 565   
      * Handles 'P' cases.
 566   
      */
 567   
     private int handleP(final String value, final DoubleMetaphoneResult result, int index) {
 568  1145 
         if (charAt(value, index + 1) == 'H') {
 569  82 
             result.append('F');
 570  82 
             index += 2;
 571   
         } else {
 572  1063 
             result.append('P');
 573  1063 
             index = contains(value, index + 1, 1, "P", "B") ? index + 2 : index + 1;
 574   
         }
 575  1145 
         return index;
 576   
     }
 577   
 
 578   
     /**
 579   
      * Handles 'R' cases.
 580   
      */
 581   
     private int handleR(final String value, final DoubleMetaphoneResult result, final int index,
 582   
                         final boolean slavoGermanic) {
 583  2739 
         if (index == value.length() - 1 && !slavoGermanic &&
 584   
             contains(value, index - 2, 2, "IE") &&
 585   
             !contains(value, index - 4, 2, "ME", "MA")) {
 586  12 
             result.appendAlternate('R');
 587   
         } else {
 588  2727 
             result.append('R');
 589   
         }
 590  2739 
         return charAt(value, index + 1) == 'R' ? index + 2 : index + 1;
 591   
     }
 592   
 
 593   
     /**
 594   
      * Handles 'S' cases.
 595   
      */
 596   
     private int handleS(final String value, final DoubleMetaphoneResult result, int index,
 597   
                         final boolean slavoGermanic) {
 598  2153 
         if (contains(value, index - 1, 3, "ISL", "YSL")) {
 599   
             //-- special cases "island", "isle", "carlisle", "carlysle" --//
 600  12 
             index++;
 601  2141 
         } else if (index == 0 && contains(value, index, 5, "SUGAR")) {
 602   
             //-- special case "sugar-" --//
 603  4 
             result.append('X', 'S');
 604  4 
             index++;
 605  2137 
         } else if (contains(value, index, 2, "SH")) {
 606  78 
             if (contains(value, index + 1, 4, "HEIM", "HOEK", "HOLM", "HOLZ")) {
 607   
                 //-- germanic --//
 608  6 
                 result.append('S');
 609   
             } else {
 610  72 
                 result.append('X');
 611   
             }
 612  78 
             index += 2;
 613  2059 
         } else if (contains(value, index, 3, "SIO", "SIA") || contains(value, index, 4, "SIAN")) {
 614   
             //-- Italian and Armenian --//
 615  50 
             if (slavoGermanic) {
 616  0 
                 result.append('S');
 617   
             } else {
 618  50 
                 result.append('S', 'X');
 619   
             }
 620  50 
             index += 3;
 621  2009 
         } else if ((index == 0 && contains(value, index + 1, 1, "M", "N", "L", "W")) ||
 622   
                    contains(value, index + 1, 1, "Z")) {
 623   
             //-- german & anglicisations, e.g. "smith" match "schmidt" //
 624   
             // "snider" match "schneider" --//
 625   
             //-- also, -sz- in slavic language altho in hungarian it //
 626   
             //   is pronounced "s" --//
 627  48 
             result.append('S', 'X');
 628  48 
             index = contains(value, index + 1, 1, "Z") ? index + 2 : index + 1;
 629  1961 
         } else if (contains(value, index, 2, "SC")) {
 630  114 
             index = handleSC(value, result, index);
 631   
         } else {
 632  1847 
             if (index == value.length() - 1 && contains(value, index - 2, 2, "AI", "OI")) {
 633   
                 //-- french e.g. "resnais", "artois" --//
 634  4 
                 result.appendAlternate('S');
 635   
             } else {
 636  1843 
                 result.append('S');
 637   
             }
 638  1847 
             index = contains(value, index + 1, 1, "S", "Z") ? index + 2 : index + 1;
 639   
         }
 640  2153 
         return index;
 641   
     }
 642   
 
 643   
     /**
 644   
      * Handles 'SC' cases.
 645   
      */
 646   
     private int handleSC(final String value, final DoubleMetaphoneResult result, final int index) {
 647  114 
         if (charAt(value, index + 2) == 'H') {
 648   
             //-- Schlesinger's rule --//
 649  38 
             if (contains(value, index + 3, 2, "OO", "ER", "EN", "UY", "ED", "EM")) {
 650   
                 //-- Dutch origin, e.g. "school", "schooner" --//
 651  10 
                 if (contains(value, index + 3, 2, "ER", "EN")) {
 652   
                     //-- "schermerhorn", "schenker" --//
 653  6 
                     result.append("X", "SK");
 654   
                 } else {
 655  4 
                     result.append("SK");
 656   
                 }
 657   
             } else {
 658  28 
                 if (index == 0 && !isVowel(charAt(value, 3)) && charAt(value, 3) != 'W') {
 659  10 
                     result.append('X', 'S');
 660   
                 } else {
 661  18 
                     result.append('X');
 662   
                 }
 663   
             }
 664  76 
         } else if (contains(value, index + 2, 1, "I", "E", "Y")) {
 665  12 
             result.append('S');
 666   
         } else {
 667  64 
             result.append("SK");
 668   
         }
 669  114 
         return index + 3;
 670   
     }
 671   
 
 672   
     /**
 673   
      * Handles 'T' cases.
 674   
      */
 675   
     private int handleT(final String value, final DoubleMetaphoneResult result, int index) {
 676  2224 
         if (contains(value, index, 4, "TION")) {
 677  52 
             result.append('X');
 678  52 
             index += 3;
 679  2172 
         } else if (contains(value, index, 3, "TIA", "TCH")) {
 680  17 
             result.append('X');
 681  17 
             index += 3;
 682  2155 
         } else if (contains(value, index, 2, "TH") || contains(value, index, 3, "TTH")) {
 683  168 
             if (contains(value, index + 2, 2, "OM", "AM") ||
 684   
                 //-- special case "thomas", "thames" or germanic --//
 685   
                 contains(value, 0, 4, "VAN ", "VON ") ||
 686   
                 contains(value, 0, 3, "SCH")) {
 687  10 
                 result.append('T');
 688   
             } else {
 689  158 
                 result.append('0', 'T');
 690   
             }
 691  168 
             index += 2;
 692   
         } else {
 693  1987 
             result.append('T');
 694  1987 
             index = contains(value, index + 1, 1, "T", "D") ? index + 2 : index + 1;
 695   
         }
 696  2224 
         return index;
 697   
     }
 698   
 
 699   
     /**
 700   
      * Handles 'W' cases.
 701   
      */
 702   
     private int handleW(final String value, final DoubleMetaphoneResult result, int index) {
 703  519 
         if (contains(value, index, 2, "WR")) {
 704   
             //-- can also be in middle of word --//
 705  12 
             result.append('R');
 706  12 
             index += 2;
 707   
         } else {
 708  507 
             if (index == 0 && (isVowel(charAt(value, index + 1)) ||
 709   
                                contains(value, index, 2, "WH"))) {
 710  216 
                 if (isVowel(charAt(value, index + 1))) {
 711   
                     //-- Wasserman should match Vasserman --//
 712  186 
                     result.append('A', 'F');
 713   
                 } else {
 714   
                     //-- need Uomo to match Womo --//
 715  30 
                     result.append('A');
 716   
                 }
 717  216 
                 index++;
 718  291 
             } else if ((index == value.length() - 1 && isVowel(charAt(value, index - 1))) ||
 719   
                        contains(value, index - 1, 5, "EWSKI", "EWSKY", "OWSKI", "OWSKY") ||
 720   
                        contains(value, 0, 3, "SCH")) {
 721   
                 //-- Arnow should match Arnoff --//
 722  46 
                 result.appendAlternate('F');
 723  46 
                 index++;
 724  245 
             } else if (contains(value, index, 4, "WICZ", "WITZ")) {
 725   
                 //-- Polish e.g. "filipowicz" --//
 726  16 
                 result.append("TS", "FX");
 727  16 
                 index += 4;
 728   
             } else {
 729  229 
                 index++;
 730   
             }
 731   
         }
 732  519 
         return index;
 733   
     }
 734   
 
 735   
     /**
 736   
      * Handles 'X' cases.
 737   
      */
 738   
     private int handleX(final String value, final DoubleMetaphoneResult result, int index) {
 739  152 
         if (index == 0) {
 740  5 
             result.append('S');
 741  5 
             index++;
 742   
         } else {
 743  147 
             if (!((index == value.length() - 1) &&
 744   
                   (contains(value, index - 3, 3, "IAU", "EAU") ||
 745   
                    contains(value, index - 2, 2, "AU", "OU")))) {
 746   
                 //-- French e.g. breaux --//
 747  141 
                 result.append("KS");
 748   
             }
 749  147 
             index = contains(value, index + 1, 1, "C", "X") ? index + 2 : index + 1;
 750   
         }
 751  152 
         return index;
 752   
     }
 753   
 
 754   
     /**
 755   
      * Handles 'Z' cases.
 756   
      */
 757   
     private int handleZ(final String value, final DoubleMetaphoneResult result, int index,
 758   
                         final boolean slavoGermanic) {
 759  97 
         if (charAt(value, index + 1) == 'H') {
 760   
             //-- Chinese pinyin e.g. "zhao" or Angelina "Zhang" --//
 761  2 
             result.append('J');
 762  2 
             index += 2;
 763   
         } else {
 764  95 
             if (contains(value, index + 1, 2, "ZO", "ZI", "ZA") ||
 765   
                 (slavoGermanic && (index > 0 && charAt(value, index - 1) != 'T'))) {
 766  12 
                 result.append("S", "TS");
 767   
             } else {
 768  83 
                 result.append('S');
 769   
             }
 770  95 
             index = charAt(value, index + 1) == 'Z' ? index + 2 : index + 1;
 771   
         }
 772  97 
         return index;
 773   
     }
 774   
 
 775   
     //-- BEGIN CONDITIONS --//
 776   
 
 777   
     /**
 778   
      * Complex condition 0 for 'C'.
 779   
      */
 780   
     private boolean conditionC0(final String value, final int index) {
 781  1680 
         if (contains(value, index, 4, "CHIA")) {
 782  2 
             return true;
 783  1678 
         } else if (index <= 1) {
 784  680 
             return false;
 785  998 
         } else if (isVowel(charAt(value, index - 2))) {
 786  357 
             return false;
 787  641 
         } else if (!contains(value, index - 1, 3, "ACH")) {
 788  621 
             return false;
 789   
         } else {
 790  20 
             final char c = charAt(value, index + 2);
 791  20 
             return (c != 'I' && c != 'E') ||
 792   
                     contains(value, index - 2, 6, "BACHER", "MACHER");
 793   
         }
 794   
     }
 795   
 
 796   
     /**
 797   
      * Complex condition 0 for 'CH'.
 798   
      */
 799   
     private boolean conditionCH0(final String value, final int index) {
 800  156 
         if (index != 0) {
 801  114 
             return false;
 802  42 
         } else if (!contains(value, index + 1, 5, "HARAC", "HARIS") &&
 803   
                    !contains(value, index + 1, 3, "HOR", "HYM", "HIA", "HEM")) {
 804  38 
             return false;
 805  4 
         } else if (contains(value, 0, 5, "CHORE")) {
 806  0 
             return false;
 807   
         } else {
 808  4 
             return true;
 809   
         }
 810   
     }
 811   
 
 812   
     /**
 813   
      * Complex condition 1 for 'CH'.
 814   
      */
 815   
     private boolean conditionCH1(final String value, final int index) {
 816  152 
         return ((contains(value, 0, 4, "VAN ", "VON ") || contains(value, 0, 3, "SCH")) ||
 817   
                 contains(value, index - 2, 6, "ORCHES", "ARCHIT", "ORCHID") ||
 818   
                 contains(value, index + 2, 1, "T", "S") ||
 819   
                 ((contains(value, index - 1, 1, "A", "O", "U", "E") || index == 0) &&
 820   
                  (contains(value, index + 2, 1, L_R_N_M_B_H_F_V_W_SPACE) || index + 1 == value.length() - 1)));
 821   
     }
 822   
 
 823   
     /**
 824   
      * Complex condition 0 for 'L'.
 825   
      */
 826   
     private boolean conditionL0(final String value, final int index) {
 827  353 
         if (index == value.length() - 3 &&
 828   
             contains(value, index - 1, 4, "ILLO", "ILLA", "ALLE")) {
 829  2 
             return true;
 830  351 
         } else if ((contains(value, value.length() - 2, 2, "AS", "OS") ||
 831   
                     contains(value, value.length() - 1, 1, "A", "O")) &&
 832   
                    contains(value, index - 1, 4, "ALLE")) {
 833  2 
             return true;
 834   
         } else {
 835  349 
             return false;
 836   
         }
 837   
     }
 838   
 
 839   
     /**
 840   
      * Complex condition 0 for 'M'.
 841   
      */
 842   
     private boolean conditionM0(final String value, final int index) {
 843  1241 
         if (charAt(value, index + 1) == 'M') {
 844  100 
             return true;
 845   
         }
 846  1141 
         return contains(value, index - 1, 3, "UMB") &&
 847   
                ((index + 1) == value.length() - 1 || contains(value, index + 2, 2, "ER"));
 848   
     }
 849   
 
 850   
     //-- BEGIN HELPER FUNCTIONS --//
 851   
 
 852   
     /**
 853   
      * Determines whether or not a value is of slavo-germanic orgin. A value is
 854   
      * of slavo-germanic origin if it contians any of 'W', 'K', 'CZ', or 'WITZ'.
 855   
      */
 856   
     private boolean isSlavoGermanic(final String value) {
 857  6494 
         return value.indexOf('W') > -1 || value.indexOf('K') > -1 ||
 858   
             value.indexOf("CZ") > -1 || value.indexOf("WITZ") > -1;
 859   
     }
 860   
 
 861   
     /**
 862   
      * Determines whether or not a character is a vowel or not
 863   
      */
 864   
     private boolean isVowel(final char ch) {
 865  2201 
         return VOWELS.indexOf(ch) != -1;
 866   
     }
 867   
 
 868   
     /**
 869   
      * Determines whether or not the value starts with a silent letter.  It will
 870   
      * return {@code true} if the value starts with any of 'GN', 'KN',
 871   
      * 'PN', 'WR' or 'PS'.
 872   
      */
 873   
     private boolean isSilentStart(final String value) {
 874  6494 
         boolean result = false;
 875  38816 
         for (final String element : SILENT_START) {
 876  32390 
             if (value.startsWith(element)) {
 877  68 
                 result = true;
 878  68 
                 break;
 879   
             }
 880   
         }
 881  6494 
         return result;
 882   
     }
 883   
 
 884   
     /**
 885   
      * Cleans the input.
 886   
      */
 887   
     private String cleanInput(String input) {
 888  6506 
         if (input == null) {
 889  3 
             return null;
 890   
         }
 891  6503 
         input = input.trim();
 892  6503 
         if (input.length() == 0) {
 893  9 
             return null;
 894   
         }
 895  6494 
         return input.toUpperCase(java.util.Locale.ENGLISH);
 896   
     }
 897   
 
 898   
     /**
 899   
      * Gets the character at index <code>index</code> if available, otherwise
 900   
      * it returns <code>Character.MIN_VALUE</code> so that there is some sort
 901   
      * of a default.
 902   
      */
 903   
     protected char charAt(final String value, final int index) {
 904  17631 
         if (index < 0 || index >= value.length()) {
 905  1503 
             return Character.MIN_VALUE;
 906   
         }
 907  16128 
         return value.charAt(index);
 908   
     }
 909   
 
 910   
     /**
 911   
      * Determines whether <code>value</code> contains any of the criteria starting at index <code>start</code> and
 912   
      * matching up to length <code>length</code>.
 913   
      */
 914   
     protected static boolean contains(final String value, final int start, final int length,
 915   
                                       final String... criteria) {
 916  53227 
         boolean result = false;
 917  53227 
         if (start >= 0 && start + length <= value.length()) {
 918  45052 
             final String target = value.substring(start, start + length);
 919   
 
 920  115152 
             for (final String element : criteria) {
 921  72524 
                 if (target.equals(element)) {
 922  2424 
                     result = true;
 923  2424 
                     break;
 924   
                 }
 925   
             }
 926   
         }
 927  53227 
         return result;
 928   
     }
 929   
 
 930   
     //-- BEGIN INNER CLASSES --//
 931   
 
 932   
     /**
 933   
      * Inner class for storing results, since there is the optional alternate encoding.
 934   
      */
 935   
     public class DoubleMetaphoneResult {
 936   
 
 937  6494 
         private final StringBuilder primary = new StringBuilder(getMaxCodeLen());
 938  6494 
         private final StringBuilder alternate = new StringBuilder(getMaxCodeLen());
 939   
         private final int maxLength;
 940   
 
 941  6494 
         public DoubleMetaphoneResult(final int maxLength) {
 942  6494 
             this.maxLength = maxLength;
 943  6494 
         }
 944   
 
 945   
         public void append(final char value) {
 946  21361 
             appendPrimary(value);
 947  21361 
             appendAlternate(value);
 948  21361 
         }
 949   
 
 950   
         public void append(final char primary, final char alternate) {
 951  863 
             appendPrimary(primary);
 952  863 
             appendAlternate(alternate);
 953  863 
         }
 954   
 
 955   
         public void appendPrimary(final char value) {
 956  22228 
             if (this.primary.length() < this.maxLength) {
 957  22226 
                 this.primary.append(value);
 958   
             }
 959  22228 
         }
 960   
 
 961   
         public void appendAlternate(final char value) {
 962  22286 
             if (this.alternate.length() < this.maxLength) {
 963  22268 
                 this.alternate.append(value);
 964   
             }
 965  22286 
         }
 966   
 
 967   
         public void append(final String value) {
 968  233 
             appendPrimary(value);
 969  233 
             appendAlternate(value);
 970  233 
         }
 971   
 
 972   
         public void append(final String primary, final String alternate) {
 973  60 
             appendPrimary(primary);
 974  60 
             appendAlternate(alternate);
 975  60 
         }
 976   
 
 977   
         public void appendPrimary(final String value) {
 978  293 
             final int addChars = this.maxLength - this.primary.length();
 979  293 
             if (value.length() <= addChars) {
 980  249 
                 this.primary.append(value);
 981   
             } else {
 982  44 
                 this.primary.append(value.substring(0, addChars));
 983   
             }
 984  293 
         }
 985   
 
 986   
         public void appendAlternate(final String value) {
 987  293 
             final int addChars = this.maxLength - this.alternate.length();
 988  293 
             if (value.length() <= addChars) {
 989  237 
                 this.alternate.append(value);
 990   
             } else {
 991  56 
                 this.alternate.append(value.substring(0, addChars));
 992   
             }
 993  293 
         }
 994   
 
 995   
         public String getPrimary() {
 996  3301 
             return this.primary.toString();
 997   
         }
 998   
 
 999   
         public String getAlternate() {
 1000  3193 
             return this.alternate.toString();
 1001   
         }
 1002   
 
 1003   
         public boolean isComplete() {
 1004  42099 
             return this.primary.length() >= this.maxLength &&
 1005   
                    this.alternate.length() >= this.maxLength;
 1006   
         }
 1007   
     }
 1008   
 }
Report generated by Cobertura 1.9.4.1 on 5/2/13 5:34 PM.