Commit | Line | Data |
---|---|---|
6dd70280 JL |
1 | /* This class is based on Lawrence Phillips original c++ code for this class.\r\r |
2 | * Found here: http://aspell.sourceforge.net/metaphone/\r\r | |
3 | *\r\r | |
4 | */\r\r | |
5 | \r\r | |
6 | \r\r | |
7 | package com.swabunga.spell.engine;\r\r | |
8 | \r\r | |
9 | \r\r | |
10 | /**\r\r | |
11 | * Double Meta class\r\r | |
12 | * <p>\r\r | |
13 | * Things that were changed:\r\r | |
14 | * The alternate flag could be set to true but was never checked so why bother with it. REMOVED\r\r | |
15 | * Why was this class serializable?\r\r | |
16 | * The primary, in, length and last variables could be initialized and local to the\r\r | |
17 | * process method and references passed arround the appropriate methods. As such there are\r\r | |
18 | * no class variables and this class becomes firstly threadsafe and secondly could be static final.\r\r | |
19 | * The function call SlavoGermaic was called repeatedly in the process function, it is now only called once.\r\r | |
20 | *\r\r | |
21 | */\r\r | |
22 | class DoubleMeta implements Transformator{\r\r | |
23 | \r\r | |
24 | private static final String[] myList = {\r\r | |
25 | "GN", "KN", "PN", "WR", "PS", ""\r\r | |
26 | };\r\r | |
27 | private static final String[] list1 = {\r\r | |
28 | "ACH", ""\r\r | |
29 | };\r\r | |
30 | private static final String[] list2 = {\r\r | |
31 | "BACHER", "MACHER", ""\r\r | |
32 | };\r\r | |
33 | private static final String[] list3 = {\r\r | |
34 | "CAESAR", ""\r\r | |
35 | };\r\r | |
36 | private static final String[] list4 = {\r\r | |
37 | "CHIA", ""\r\r | |
38 | };\r\r | |
39 | private static final String[] list5 = {\r\r | |
40 | "CH", ""\r\r | |
41 | };\r\r | |
42 | private static final String[] list6 = {\r\r | |
43 | "CHAE", ""\r\r | |
44 | };\r\r | |
45 | private static final String[] list7 = {\r\r | |
46 | "HARAC", "HARIS", ""\r\r | |
47 | };\r\r | |
48 | private static final String[] list8 = {\r\r | |
49 | "HOR", "HYM", "HIA", "HEM", ""\r\r | |
50 | };\r\r | |
51 | private static final String[] list9 = {\r\r | |
52 | "CHORE", ""\r\r | |
53 | };\r\r | |
54 | private static final String[] list10 = {\r\r | |
55 | "VAN ", "VON ", ""\r\r | |
56 | };\r\r | |
57 | private static final String[] list11 = {\r\r | |
58 | "SCH", ""\r\r | |
59 | };\r\r | |
60 | private static final String[] list12 = {\r\r | |
61 | "ORCHES", "ARCHIT", "ORCHID", ""\r\r | |
62 | };\r\r | |
63 | private static final String[] list13 = {\r\r | |
64 | "T", "S", ""\r\r | |
65 | };\r\r | |
66 | private static final String[] list14 = {\r\r | |
67 | "A", "O", "U", "E", ""\r\r | |
68 | };\r\r | |
69 | private static final String[] list15 = {\r\r | |
70 | "L", "R", "N", "M", "B", "H", "F", "V", "W", " ", ""\r\r | |
71 | };\r\r | |
72 | private static final String[] list16 = {\r\r | |
73 | "MC", ""\r\r | |
74 | };\r\r | |
75 | private static final String[] list17 = {\r\r | |
76 | "CZ", ""\r\r | |
77 | };\r\r | |
78 | private static final String[] list18 = {\r\r | |
79 | "WICZ", ""\r\r | |
80 | };\r\r | |
81 | private static final String[] list19 = {\r\r | |
82 | "CIA", ""\r\r | |
83 | };\r\r | |
84 | private static final String[] list20 = {\r\r | |
85 | "CC", ""\r\r | |
86 | };\r\r | |
87 | private static final String[] list21 = {\r\r | |
88 | "I", "E", "H", ""\r\r | |
89 | };\r\r | |
90 | private static final String[] list22 = {\r\r | |
91 | "HU", ""\r\r | |
92 | };\r\r | |
93 | private static final String[] list23 = {\r\r | |
94 | "UCCEE", "UCCES", ""\r\r | |
95 | };\r\r | |
96 | private static final String[] list24 = {\r\r | |
97 | "CK", "CG", "CQ", ""\r\r | |
98 | };\r\r | |
99 | private static final String[] list25 = {\r\r | |
100 | "CI", "CE", "CY", ""\r\r | |
101 | };\r\r | |
102 | private static final String[] list26 = {\r\r | |
103 | "CIO", "CIE", "CIA", ""\r\r | |
104 | };\r\r | |
105 | private static final String[] list27 = {\r\r | |
106 | " C", " Q", " G", ""\r\r | |
107 | };\r\r | |
108 | private static final String[] list28 = {\r\r | |
109 | "C", "K", "Q", ""\r\r | |
110 | };\r\r | |
111 | private static final String[] list29 = {\r\r | |
112 | "CE", "CI", ""\r\r | |
113 | };\r\r | |
114 | private static final String[] list30 = {\r\r | |
115 | "DG", ""\r\r | |
116 | };\r\r | |
117 | private static final String[] list31 = {\r\r | |
118 | "I", "E", "Y", ""\r\r | |
119 | };\r\r | |
120 | private static final String[] list32 = {\r\r | |
121 | "DT", "DD", ""\r\r | |
122 | };\r\r | |
123 | private static final String[] list33 = {\r\r | |
124 | "B", "H", "D", ""\r\r | |
125 | };\r\r | |
126 | private static final String[] list34 = {\r\r | |
127 | "B", "H", "D", ""\r\r | |
128 | };\r\r | |
129 | private static final String[] list35 = {\r\r | |
130 | "B", "H", ""\r\r | |
131 | };\r\r | |
132 | private static final String[] list36 = {\r\r | |
133 | "C", "G", "L", "R", "T", ""\r\r | |
134 | };\r\r | |
135 | private static final String[] list37 = {\r\r | |
136 | "EY", ""\r\r | |
137 | };\r\r | |
138 | private static final String[] list38 = {\r\r | |
139 | "LI", ""\r\r | |
140 | };\r\r | |
141 | private static final String[] list39 = {\r\r | |
142 | "ES", "EP", "EB", "EL", "EY", "IB", "IL", "IN", "IE", "EI", "ER",\r\r | |
143 | ""\r\r | |
144 | };\r\r | |
145 | private static final String[] list40 = {\r\r | |
146 | "ER", ""\r\r | |
147 | };\r\r | |
148 | private static final String[] list41 = {\r\r | |
149 | "DANGER", "RANGER", "MANGER", ""\r\r | |
150 | };\r\r | |
151 | private static final String[] list42 = {\r\r | |
152 | "E", "I", ""\r\r | |
153 | };\r\r | |
154 | private static final String[] list43 = {\r\r | |
155 | "RGY", "OGY", ""\r\r | |
156 | };\r\r | |
157 | private static final String[] list44 = {\r\r | |
158 | "E", "I", "Y", ""\r\r | |
159 | };\r\r | |
160 | private static final String[] list45 = {\r\r | |
161 | "AGGI", "OGGI", ""\r\r | |
162 | };\r\r | |
163 | private static final String[] list46 = {\r\r | |
164 | "VAN ", "VON ", ""\r\r | |
165 | };\r\r | |
166 | private static final String[] list47 = {\r\r | |
167 | "SCH", ""\r\r | |
168 | };\r\r | |
169 | private static final String[] list48 = {\r\r | |
170 | "ET", ""\r\r | |
171 | };\r\r | |
172 | private static final String[] list49 = {\r\r | |
173 | "IER ", ""\r\r | |
174 | };\r\r | |
175 | private static final String[] list50 = {\r\r | |
176 | "JOSE", ""\r\r | |
177 | };\r\r | |
178 | private static final String[] list51 = {\r\r | |
179 | "SAN ", ""\r\r | |
180 | };\r\r | |
181 | private static final String[] list52 = {\r\r | |
182 | "SAN ", ""\r\r | |
183 | };\r\r | |
184 | private static final String[] list53 = {\r\r | |
185 | "JOSE", ""\r\r | |
186 | };\r\r | |
187 | private static final String[] list54 = {\r\r | |
188 | "L", "T", "K", "S", "N", "M", "B", "Z", ""\r\r | |
189 | };\r\r | |
190 | private static final String[] list55 = {\r\r | |
191 | "S", "K", "L", ""\r\r | |
192 | };\r\r | |
193 | private static final String[] list56 = {\r\r | |
194 | "ILLO", "ILLA", "ALLE", ""\r\r | |
195 | };\r\r | |
196 | private static final String[] list57 = {\r\r | |
197 | "AS", "OS", ""\r\r | |
198 | };\r\r | |
199 | private static final String[] list58 = {\r\r | |
200 | "A", "O", ""\r\r | |
201 | };\r\r | |
202 | private static final String[] list59 = {\r\r | |
203 | "ALLE", ""\r\r | |
204 | };\r\r | |
205 | private static final String[] list60 = {\r\r | |
206 | "UMB", ""\r\r | |
207 | };\r\r | |
208 | private static final String[] list61 = {\r\r | |
209 | "ER", ""\r\r | |
210 | };\r\r | |
211 | private static final String[] list62 = {\r\r | |
212 | "P", "B", ""\r\r | |
213 | };\r\r | |
214 | private static final String[] list63 = {\r\r | |
215 | "IE", ""\r\r | |
216 | };\r\r | |
217 | private static final String[] list64 = {\r\r | |
218 | "ME", "MA", ""\r\r | |
219 | };\r\r | |
220 | private static final String[] list65 = {\r\r | |
221 | "ISL", "YSL", ""\r\r | |
222 | };\r\r | |
223 | private static final String[] list66 = {\r\r | |
224 | "SUGAR", ""\r\r | |
225 | };\r\r | |
226 | private static final String[] list67 = {\r\r | |
227 | "SH", ""\r\r | |
228 | };\r\r | |
229 | private static final String[] list68 = {\r\r | |
230 | "HEIM", "HOEK", "HOLM", "HOLZ", ""\r\r | |
231 | };\r\r | |
232 | private static final String[] list69 = {\r\r | |
233 | "SIO", "SIA", ""\r\r | |
234 | };\r\r | |
235 | private static final String[] list70 = {\r\r | |
236 | "SIAN", ""\r\r | |
237 | };\r\r | |
238 | private static final String[] list71 = {\r\r | |
239 | "M", "N", "L", "W", ""\r\r | |
240 | };\r\r | |
241 | private static final String[] list72 = {\r\r | |
242 | "Z", ""\r\r | |
243 | };\r\r | |
244 | private static final String[] list73 = {\r\r | |
245 | "Z", ""\r\r | |
246 | };\r\r | |
247 | private static final String[] list74 = {\r\r | |
248 | "SC", ""\r\r | |
249 | };\r\r | |
250 | private static final String[] list75 = {\r\r | |
251 | "OO", "ER", "EN", "UY", "ED", "EM", ""\r\r | |
252 | };\r\r | |
253 | private static final String[] list76 = {\r\r | |
254 | "ER", "EN", ""\r\r | |
255 | };\r\r | |
256 | private static final String[] list77 = {\r\r | |
257 | "I", "E", "Y", ""\r\r | |
258 | };\r\r | |
259 | private static final String[] list78 = {\r\r | |
260 | "AI", "OI", ""\r\r | |
261 | };\r\r | |
262 | private static final String[] list79 = {\r\r | |
263 | "S", "Z", ""\r\r | |
264 | };\r\r | |
265 | private static final String[] list80 = {\r\r | |
266 | "TION", ""\r\r | |
267 | };\r\r | |
268 | private static final String[] list81 = {\r\r | |
269 | "TIA", "TCH", ""\r\r | |
270 | };\r\r | |
271 | private static final String[] list82 = {\r\r | |
272 | "TH", ""\r\r | |
273 | };\r\r | |
274 | private static final String[] list83 = {\r\r | |
275 | "TTH", ""\r\r | |
276 | };\r\r | |
277 | private static final String[] list84 = {\r\r | |
278 | "OM", "AM", ""\r\r | |
279 | };\r\r | |
280 | private static final String[] list85 = {\r\r | |
281 | "VAN ", "VON ", ""\r\r | |
282 | };\r\r | |
283 | private static final String[] list86 = {\r\r | |
284 | "SCH", ""\r\r | |
285 | };\r\r | |
286 | private static final String[] list87 = {\r\r | |
287 | "T", "D", ""\r\r | |
288 | };\r\r | |
289 | private static final String[] list88 = {\r\r | |
290 | "WR", ""\r\r | |
291 | };\r\r | |
292 | private static final String[] list89 = {\r\r | |
293 | "WH", ""\r\r | |
294 | };\r\r | |
295 | private static final String[] list90 = {\r\r | |
296 | "EWSKI", "EWSKY", "OWSKI", "OWSKY", ""\r\r | |
297 | };\r\r | |
298 | private static final String[] list91 = {\r\r | |
299 | "SCH", ""\r\r | |
300 | };\r\r | |
301 | private static final String[] list92 = {\r\r | |
302 | "WICZ", "WITZ", ""\r\r | |
303 | };\r\r | |
304 | private static final String[] list93 = {\r\r | |
305 | "IAU", "EAU", ""\r\r | |
306 | };\r\r | |
307 | private static final String[] list94 = {\r\r | |
308 | "AU", "OU", ""\r\r | |
309 | };\r\r | |
310 | private static final String[] list95 = {\r\r | |
311 | "C", "X", ""\r\r | |
312 | };\r\r | |
313 | private static final String[] list96 = {\r\r | |
314 | "ZO", "ZI", "ZA", ""\r\r | |
315 | };\r\r | |
316 | \r\r | |
317 | /**\r\r | |
318 | * put your documentation comment here\r\r | |
319 | * @return\r\r | |
320 | */\r\r | |
321 | private final static boolean SlavoGermanic (String in) {\r\r | |
322 | if ((in.indexOf("W") > -1) || (in.indexOf("K") > -1) || (in.indexOf("CZ") > -1)\r\r | |
323 | || (in.indexOf("WITZ") > -1))\r\r | |
324 | return true;\r\r | |
325 | return false;\r\r | |
326 | }\r\r | |
327 | \r\r | |
328 | /**\r\r | |
329 | * put your documentation comment here\r\r | |
330 | * @param main\r\r | |
331 | */\r\r | |
332 | private final static void MetaphAdd (StringBuffer primary, String main) {\r\r | |
333 | if (main != null) {\r\r | |
334 | primary.append(main);\r\r | |
335 | }\r\r | |
336 | }\r\r | |
337 | \r\r | |
338 | private final static void MetaphAdd (StringBuffer primary, char main) {\r\r | |
339 | primary.append(main);\r\r | |
340 | } \r\r | |
341 | \r\r | |
342 | /**\r\r | |
343 | * put your documentation comment here\r\r | |
344 | * @param at\r\r | |
345 | * @return\r\r | |
346 | */\r\r | |
347 | private final static boolean isVowel (String in, int at, int length) {\r\r | |
348 | if ((at < 0) || (at >= length))\r\r | |
349 | return false;\r\r | |
350 | char it = in.charAt(at);\r\r | |
351 | if ((it == 'A') || (it == 'E') || (it == 'I') || (it == 'O') || (it ==\r\r | |
352 | 'U') || (it == 'Y'))\r\r | |
353 | return true;\r\r | |
354 | return false;\r\r | |
355 | }\r\r | |
356 | \r\r | |
357 | /**\r\r | |
358 | * put your documentation comment here\r\r | |
359 | * @param string\r\r | |
360 | * @param start\r\r | |
361 | * @param length\r\r | |
362 | * @param list\r\r | |
363 | * @return\r\r | |
364 | */\r\r | |
365 | private final static boolean stringAt (String string, int start, int length, String[] list) {\r\r | |
366 | if ((start < 0) || (start >= string.length()) || list.length == 0 )\r\r | |
367 | return false;\r\r | |
368 | String substr = string.substring(start, start + length);\r\r | |
369 | for (int i = 0; i < list.length; i++) {\r\r | |
370 | if (list[i].equals(substr))\r\r | |
371 | return true;\r\r | |
372 | }\r\r | |
373 | return false;\r\r | |
374 | }\r\r | |
375 | \r\r | |
376 | /**\r\r | |
377 | * put your documentation comment here\r\r | |
378 | * @param word The word to process.\r\r | |
379 | * @return\r\r | |
380 | */\r\r | |
381 | public final String transform (String word) {\r\r | |
382 | StringBuffer primary = new StringBuffer( word.length() + 5 );\r\r | |
383 | String in = word.toUpperCase() + " ";\r\r | |
384 | int current = 0;\r\r | |
385 | int length = in.length();\r\r | |
386 | if (length < 1)\r\r | |
387 | return "";\r\r | |
388 | int last = length - 1;\r\r | |
389 | boolean isSlavoGermaic = SlavoGermanic(in);\r\r | |
390 | if (stringAt(in, 0, 2, myList))\r\r | |
391 | current += 1;\r\r | |
392 | if (in.charAt(0) == 'X') {\r\r | |
393 | MetaphAdd(primary, 'S');\r\r | |
394 | current += 1;\r\r | |
395 | }\r\r | |
396 | while (current < length ) {\r\r | |
397 | switch (in.charAt(current)) {\r\r | |
398 | case 'A':case 'E':case 'I':case 'O':case 'U':case 'Y':\r\r | |
399 | if (current == 0)\r\r | |
400 | MetaphAdd(primary, 'A');\r\r | |
401 | current += 1;\r\r | |
402 | break;\r\r | |
403 | case 'B':\r\r | |
404 | MetaphAdd(primary, 'P');\r\r | |
405 | if (in.charAt(current + 1) == 'B')\r\r | |
406 | current += 2;\r\r | |
407 | else\r\r | |
408 | current += 1;\r\r | |
409 | break;\r\r | |
410 | case 'Ç':\r\r | |
411 | MetaphAdd(primary, 'S');\r\r | |
412 | current += 1;\r\r | |
413 | break;\r\r | |
414 | case 'C':\r\r | |
415 | if ((current > 1) && !isVowel(in, current - 2, length) && stringAt(in, (current\r\r | |
416 | - 1), 3, list1) && (in.charAt(current + 2) != 'I') && (in.charAt(\r\r | |
417 | current + 2) != 'E') || stringAt(in, (current - 2), 6, list2)) {\r\r | |
418 | MetaphAdd(primary, 'K');\r\r | |
419 | current += 2;\r\r | |
420 | break;\r\r | |
421 | }\r\r | |
422 | if ((current == 0) && stringAt(in, current, 6, list3)) {\r\r | |
423 | MetaphAdd(primary, 'S');\r\r | |
424 | current += 2;\r\r | |
425 | break;\r\r | |
426 | }\r\r | |
427 | if (stringAt(in, current, 4, list4)) {\r\r | |
428 | MetaphAdd(primary, 'K');\r\r | |
429 | current += 2;\r\r | |
430 | break;\r\r | |
431 | }\r\r | |
432 | if (stringAt(in, current, 2, list5)) {\r\r | |
433 | if ((current > 0) && stringAt(in, current, 4, list6)) {\r\r | |
434 | MetaphAdd(primary, 'K' );\r\r | |
435 | current += 2;\r\r | |
436 | break;\r\r | |
437 | }\r\r | |
438 | if ((current == 0) && stringAt(in, (current + 1), 5, list7) ||\r\r | |
439 | stringAt(in, current + 1, 3, list8) && !stringAt(in, 0, 5,\r\r | |
440 | list9)) {\r\r | |
441 | MetaphAdd(primary, 'K' );\r\r | |
442 | current += 2;\r\r | |
443 | break;\r\r | |
444 | }\r\r | |
445 | if (stringAt(in, 0, 4, list10) || stringAt(in, 0, 3, list11) ||\r\r | |
446 | stringAt(in, current - 2, 6, list12) || stringAt(in, current\r\r | |
447 | + 2, 1, list13) || (stringAt(in, current - 1, 1, list14) ||\r\r | |
448 | (current == 0)) && stringAt(in, current + 2, 1, list15)) {\r\r | |
449 | MetaphAdd(primary, 'K');\r\r | |
450 | }\r\r | |
451 | else {\r\r | |
452 | if (current > 0) {\r\r | |
453 | if (stringAt(in, 0, 2, list16))\r\r | |
454 | MetaphAdd(primary, 'K');\r\r | |
455 | else\r\r | |
456 | MetaphAdd(primary, 'X');\r\r | |
457 | }\r\r | |
458 | else {\r\r | |
459 | MetaphAdd(primary, 'X');\r\r | |
460 | }\r\r | |
461 | }\r\r | |
462 | current += 2;\r\r | |
463 | break;\r\r | |
464 | }\r\r | |
465 | if (stringAt(in, current, 2, list17) && !stringAt(in, current, 4,\r\r | |
466 | list18)) {\r\r | |
467 | MetaphAdd(primary, 'S');\r\r | |
468 | current += 2;\r\r | |
469 | break;\r\r | |
470 | }\r\r | |
471 | if (stringAt(in, current, 2, list19)) {\r\r | |
472 | MetaphAdd(primary, 'X');\r\r | |
473 | current += 2;\r\r | |
474 | break;\r\r | |
475 | }\r\r | |
476 | if (stringAt(in, current, 2, list20) && !((current == 1) && in.charAt(0)\r\r | |
477 | == 'M')) {\r\r | |
478 | if (stringAt(in, current + 2, 1, list21) && !stringAt(in, current\r\r | |
479 | + 2, 2, list22)) {\r\r | |
480 | if (((current == 1) && (in.charAt(current - 1) == 'A')) || stringAt(in,\r\r | |
481 | (current - 1), 5, list23))\r\r | |
482 | MetaphAdd(primary, "KS");\r\r | |
483 | else\r\r | |
484 | MetaphAdd(primary, 'X');\r\r | |
485 | current += 3;\r\r | |
486 | break;\r\r | |
487 | }\r\r | |
488 | else {\r\r | |
489 | MetaphAdd(primary, 'K');\r\r | |
490 | current += 2;\r\r | |
491 | break;\r\r | |
492 | }\r\r | |
493 | }\r\r | |
494 | if (stringAt(in, current, 2, list24)) {\r\r | |
495 | MetaphAdd(primary, 'K');\r\r | |
496 | current += 2;\r\r | |
497 | break;\r\r | |
498 | }\r\r | |
499 | else if (stringAt(in, current, 2, list25)) {\r\r | |
500 | MetaphAdd(primary, 'S');\r\r | |
501 | current += 2;\r\r | |
502 | break;\r\r | |
503 | }\r\r | |
504 | \r\r | |
505 | MetaphAdd(primary, 'K');\r\r | |
506 | if (stringAt(in, current + 1, 2, list27))\r\r | |
507 | current += 3;\r\r | |
508 | else if (stringAt(in, current + 1, 1, list28) && !stringAt(in, current\r\r | |
509 | + 1, 2, list29))\r\r | |
510 | current += 2;\r\r | |
511 | else\r\r | |
512 | current += 1;\r\r | |
513 | break;\r\r | |
514 | case 'D':\r\r | |
515 | if (stringAt(in, current, 2, list30)) {\r\r | |
516 | if (stringAt(in, current + 2, 1, list31)) {\r\r | |
517 | MetaphAdd(primary, 'J');\r\r | |
518 | current += 3;\r\r | |
519 | break;\r\r | |
520 | }\r\r | |
521 | else {\r\r | |
522 | MetaphAdd(primary, "TK");\r\r | |
523 | current += 2;\r\r | |
524 | break;\r\r | |
525 | }\r\r | |
526 | }\r\r | |
527 | MetaphAdd(primary, 'T');\r\r | |
528 | if (stringAt(in, current, 2, list32)) {\r\r | |
529 | current += 2;\r\r | |
530 | } else {\r\r | |
531 | current += 1;\r\r | |
532 | }\r\r | |
533 | break;\r\r | |
534 | case 'F':\r\r | |
535 | if (in.charAt(current + 1) == 'F')\r\r | |
536 | current += 2;\r\r | |
537 | else\r\r | |
538 | current += 1;\r\r | |
539 | MetaphAdd(primary, 'F');\r\r | |
540 | break;\r\r | |
541 | case 'G':\r\r | |
542 | if (in.charAt(current + 1) == 'H') {\r\r | |
543 | if ((current > 0) && !isVowel(in, current - 1, length)) {\r\r | |
544 | MetaphAdd(primary, 'K');\r\r | |
545 | current += 2;\r\r | |
546 | break;\r\r | |
547 | }\r\r | |
548 | if (current < 3) {\r\r | |
549 | if (current == 0) {\r\r | |
550 | if (in.charAt(current + 2) == 'I')\r\r | |
551 | MetaphAdd(primary, 'J');\r\r | |
552 | else\r\r | |
553 | MetaphAdd(primary, 'K');\r\r | |
554 | current += 2;\r\r | |
555 | break;\r\r | |
556 | }\r\r | |
557 | }\r\r | |
558 | if ((current > 1) && stringAt(in, current - 2, 1, list33) || ((current > 2)\r\r | |
559 | && stringAt(in, current - 3, 1, list34)) || ((current > 3) &&\r\r | |
560 | stringAt(in, current - 4, 1, list35))) {\r\r | |
561 | current += 2;\r\r | |
562 | break;\r\r | |
563 | }\r\r | |
564 | else {\r\r | |
565 | if ((current > 2) && (in.charAt(current - 1) == 'U') && stringAt(in,\r\r | |
566 | current - 3, 1, list36)) {\r\r | |
567 | MetaphAdd(primary, 'F');\r\r | |
568 | }\r\r | |
569 | else {\r\r | |
570 | if ((current > 0) && (in.charAt(current - 1) != 'I'))\r\r | |
571 | MetaphAdd(primary, 'K');\r\r | |
572 | }\r\r | |
573 | current += 2;\r\r | |
574 | break;\r\r | |
575 | }\r\r | |
576 | }\r\r | |
577 | if (in.charAt(current + 1) == 'N') {\r\r | |
578 | if ((current == 1) && isVowel(in, 0, length) && !isSlavoGermaic) {\r\r | |
579 | MetaphAdd(primary, "KN");\r\r | |
580 | }\r\r | |
581 | else {\r\r | |
582 | if (!stringAt(in, current + 2, 2, list37) && (in.charAt(current\r\r | |
583 | + 1) != 'Y') && !isSlavoGermaic) {\r\r | |
584 | MetaphAdd(primary, "N");\r\r | |
585 | }\r\r | |
586 | else {\r\r | |
587 | MetaphAdd(primary, "KN");\r\r | |
588 | }\r\r | |
589 | }\r\r | |
590 | current += 2;\r\r | |
591 | break;\r\r | |
592 | }\r\r | |
593 | if (stringAt(in, current + 1, 2, list38) && !isSlavoGermaic) {\r\r | |
594 | MetaphAdd(primary, "KL");\r\r | |
595 | current += 2;\r\r | |
596 | break;\r\r | |
597 | }\r\r | |
598 | if ((current == 0) && ((in.charAt(current + 1) == 'Y') || stringAt(in,\r\r | |
599 | current + 1, 2, list39))) {\r\r | |
600 | MetaphAdd(primary, 'K');\r\r | |
601 | current += 2;\r\r | |
602 | break;\r\r | |
603 | }\r\r | |
604 | if ((stringAt(in, current + 1, 2, list40) || (in.charAt(current +\r\r | |
605 | 1) == 'Y')) && !stringAt(in, 0, 6, list41) && !stringAt(in, current\r\r | |
606 | - 1, 1, list42) && !stringAt(in, current - 1, 3, list43)) {\r\r | |
607 | MetaphAdd(primary, 'K');\r\r | |
608 | current += 2;\r\r | |
609 | break;\r\r | |
610 | }\r\r | |
611 | if (stringAt(in, current + 1, 1, list44) || stringAt(in, current -\r\r | |
612 | 1, 4, list45)) {\r\r | |
613 | if (stringAt(in, 0, 4, list46) || stringAt(in, 0, 3, list47) ||\r\r | |
614 | stringAt(in, current + 1, 2, list48)) {\r\r | |
615 | MetaphAdd(primary, 'K');\r\r | |
616 | }\r\r | |
617 | else {\r\r | |
618 | MetaphAdd(primary, 'J');\r\r | |
619 | }\r\r | |
620 | current += 2;\r\r | |
621 | break;\r\r | |
622 | }\r\r | |
623 | if (in.charAt(current + 1) == 'G')\r\r | |
624 | current += 2;\r\r | |
625 | else\r\r | |
626 | current += 1;\r\r | |
627 | MetaphAdd(primary, 'K');\r\r | |
628 | break;\r\r | |
629 | case 'H':\r\r | |
630 | if (((current == 0) || isVowel(in, current - 1, length)) && isVowel(in, current +\r\r | |
631 | 1, length)) {\r\r | |
632 | MetaphAdd(primary, 'H');\r\r | |
633 | current += 2;\r\r | |
634 | }\r\r | |
635 | else {\r\r | |
636 | current += 1;\r\r | |
637 | }\r\r | |
638 | break;\r\r | |
639 | case 'J':\r\r | |
640 | if (stringAt(in, current, 4, list50) || stringAt(in, 0, 4, list51)) {\r\r | |
641 | if ((current == 0) && (in.charAt(current + 4) == ' ') || stringAt(in,\r\r | |
642 | 0, 4, list52)) {\r\r | |
643 | MetaphAdd(primary, 'H');\r\r | |
644 | }\r\r | |
645 | else {\r\r | |
646 | MetaphAdd(primary, 'J');\r\r | |
647 | }\r\r | |
648 | current += 1;\r\r | |
649 | break;\r\r | |
650 | }\r\r | |
651 | if ((current == 0) && !stringAt(in, current, 4, list53)) {\r\r | |
652 | MetaphAdd(primary, 'J');\r\r | |
653 | }\r\r | |
654 | else {\r\r | |
655 | if (isVowel(in, current - 1, length) && !isSlavoGermaic && ((in.charAt(current\r\r | |
656 | + 1) == 'A') || in.charAt(current + 1) == 'O')) {\r\r | |
657 | MetaphAdd(primary, 'J' );\r\r | |
658 | }\r\r | |
659 | else {\r\r | |
660 | if (current == last) {\r\r | |
661 | MetaphAdd(primary, 'J' );\r\r | |
662 | }\r\r | |
663 | else {\r\r | |
664 | if (!stringAt(in, current + 1, 1, list54) && !stringAt(in,\r\r | |
665 | current - 1, 1, list55)) {\r\r | |
666 | MetaphAdd(primary, 'J');\r\r | |
667 | }\r\r | |
668 | }\r\r | |
669 | }\r\r | |
670 | }\r\r | |
671 | if (in.charAt(current + 1) == 'J')\r\r | |
672 | current += 2;\r\r | |
673 | else\r\r | |
674 | current += 1;\r\r | |
675 | break;\r\r | |
676 | case 'K':\r\r | |
677 | if (in.charAt(current + 1) == 'K')\r\r | |
678 | current += 2;\r\r | |
679 | else\r\r | |
680 | current += 1;\r\r | |
681 | MetaphAdd(primary, 'K');\r\r | |
682 | break;\r\r | |
683 | case 'L':\r\r | |
684 | if (in.charAt(current + 1) == 'L') {\r\r | |
685 | if (((current == (length - 3)) && stringAt(in, current - 1, 4,\r\r | |
686 | list56)) || ((stringAt(in, last - 1, 2, list57) || stringAt(in,\r\r | |
687 | last, 1, list58)) && stringAt(in, current - 1, 4, list59))) {\r\r | |
688 | MetaphAdd(primary, 'L' );\r\r | |
689 | current += 2;\r\r | |
690 | break;\r\r | |
691 | }\r\r | |
692 | current += 2;\r\r | |
693 | }\r\r | |
694 | else\r\r | |
695 | current += 1;\r\r | |
696 | MetaphAdd(primary, 'L');\r\r | |
697 | break;\r\r | |
698 | case 'M':\r\r | |
699 | if ((stringAt(in, current - 1, 3, list60) && (((current + 1) == last)\r\r | |
700 | || stringAt(in, current + 2, 2, list61))) || (in.charAt(current\r\r | |
701 | + 1) == 'M'))\r\r | |
702 | current += 2;\r\r | |
703 | else\r\r | |
704 | current += 1;\r\r | |
705 | MetaphAdd(primary, 'M');\r\r | |
706 | break;\r\r | |
707 | case 'N':\r\r | |
708 | if (in.charAt(current + 1) == 'N')\r\r | |
709 | current += 2;\r\r | |
710 | else\r\r | |
711 | current += 1;\r\r | |
712 | MetaphAdd(primary, 'N');\r\r | |
713 | break;\r\r | |
714 | case 'Ñ':\r\r | |
715 | current += 1;\r\r | |
716 | MetaphAdd(primary, 'N');\r\r | |
717 | break;\r\r | |
718 | case 'P':\r\r | |
719 | if (in.charAt(current + 1) == 'N') {\r\r | |
720 | MetaphAdd(primary, 'F');\r\r | |
721 | current += 2;\r\r | |
722 | break;\r\r | |
723 | }\r\r | |
724 | if (stringAt(in, current + 1, 1, list62))\r\r | |
725 | current += 2;\r\r | |
726 | else\r\r | |
727 | current += 1;\r\r | |
728 | MetaphAdd(primary, 'P');\r\r | |
729 | break;\r\r | |
730 | case 'Q':\r\r | |
731 | if (in.charAt(current + 1) == 'Q')\r\r | |
732 | current += 2;\r\r | |
733 | else\r\r | |
734 | current += 1;\r\r | |
735 | MetaphAdd(primary, 'K');\r\r | |
736 | break;\r\r | |
737 | case 'R':\r\r | |
738 | if ((current == last) && !isSlavoGermaic && stringAt(in, current\r\r | |
739 | - 2, 2, list63) && !stringAt(in, current - 4, 2, list64)) {\r\r | |
740 | // MetaphAdd(primary, "");\r\r | |
741 | } else\r\r | |
742 | MetaphAdd(primary, 'R');\r\r | |
743 | if (in.charAt(current + 1) == 'R')\r\r | |
744 | current += 2;\r\r | |
745 | else\r\r | |
746 | current += 1;\r\r | |
747 | break;\r\r | |
748 | case 'S':\r\r | |
749 | if (stringAt(in, current - 1, 3, list65)) {\r\r | |
750 | current += 1;\r\r | |
751 | break;\r\r | |
752 | }\r\r | |
753 | if ((current == 0) && stringAt(in, current, 5, list66)) {\r\r | |
754 | MetaphAdd(primary, 'X');\r\r | |
755 | current += 1;\r\r | |
756 | break;\r\r | |
757 | }\r\r | |
758 | if (stringAt(in, current, 2, list67)) {\r\r | |
759 | if (stringAt(in, current + 1, 4, list68))\r\r | |
760 | MetaphAdd(primary, 'S');\r\r | |
761 | else\r\r | |
762 | MetaphAdd(primary, 'X');\r\r | |
763 | current += 2;\r\r | |
764 | break;\r\r | |
765 | }\r\r | |
766 | if (stringAt(in, current, 3, list69) || stringAt(in, current, 4,\r\r | |
767 | list70)) {\r\r | |
768 | MetaphAdd(primary, 'S');\r\r | |
769 | current += 3;\r\r | |
770 | break;\r\r | |
771 | }\r\r | |
772 | if (((current == 0) && stringAt(in, current + 1, 1, list71)) || stringAt(in,\r\r | |
773 | current + 1, 1, list72)) {\r\r | |
774 | MetaphAdd(primary, 'S');\r\r | |
775 | if (stringAt(in, current + 1, 1, list73))\r\r | |
776 | current += 2;\r\r | |
777 | else\r\r | |
778 | current += 1;\r\r | |
779 | break;\r\r | |
780 | }\r\r | |
781 | if (stringAt(in, current, 2, list74)) {\r\r | |
782 | if (in.charAt(current + 2) == 'H')\r\r | |
783 | if (stringAt(in, current + 3, 2, list75)) {\r\r | |
784 | if (stringAt(in, current + 3, 2, list76)) {\r\r | |
785 | MetaphAdd(primary, "X");\r\r | |
786 | }\r\r | |
787 | else {\r\r | |
788 | MetaphAdd(primary, "SK");\r\r | |
789 | }\r\r | |
790 | current += 3;\r\r | |
791 | break;\r\r | |
792 | }\r\r | |
793 | else {\r\r | |
794 | MetaphAdd(primary, 'X');\r\r | |
795 | current += 3;\r\r | |
796 | break;\r\r | |
797 | }\r\r | |
798 | if (stringAt(in, current + 2, 1, list77)) {\r\r | |
799 | MetaphAdd(primary, 'S');\r\r | |
800 | current += 3;\r\r | |
801 | break;\r\r | |
802 | }\r\r | |
803 | MetaphAdd(primary, "SK");\r\r | |
804 | current += 3;\r\r | |
805 | break;\r\r | |
806 | }\r\r | |
807 | if ((current == last) && stringAt(in, current - 2, 2, list78)) {\r\r | |
808 | //MetaphAdd(primary, "");\r\r | |
809 | } else\r\r | |
810 | MetaphAdd(primary, 'S');\r\r | |
811 | if (stringAt(in, current + 1, 1, list79))\r\r | |
812 | current += 2;\r\r | |
813 | else\r\r | |
814 | current += 1;\r\r | |
815 | break;\r\r | |
816 | case 'T':\r\r | |
817 | if (stringAt(in, current, 4, list80)) {\r\r | |
818 | MetaphAdd(primary, 'X');\r\r | |
819 | current += 3;\r\r | |
820 | break;\r\r | |
821 | }\r\r | |
822 | if (stringAt(in, current, 3, list81)) {\r\r | |
823 | MetaphAdd(primary, 'X');\r\r | |
824 | current += 3;\r\r | |
825 | break;\r\r | |
826 | }\r\r | |
827 | if (stringAt(in, current, 2, list82) || stringAt(in, current, 3,\r\r | |
828 | list83)) {\r\r | |
829 | if (stringAt(in, (current + 2), 2, list84) || stringAt(in, 0, 4,\r\r | |
830 | list85) || stringAt(in, 0, 3, list86)) {\r\r | |
831 | MetaphAdd(primary, 'T');\r\r | |
832 | }\r\r | |
833 | else {\r\r | |
834 | MetaphAdd(primary, '0');\r\r | |
835 | }\r\r | |
836 | current += 2;\r\r | |
837 | break;\r\r | |
838 | }\r\r | |
839 | if (stringAt(in, current + 1, 1, list87)) {\r\r | |
840 | current += 2;\r\r | |
841 | }\r\r | |
842 | else\r\r | |
843 | current += 1;\r\r | |
844 | MetaphAdd(primary, 'T');\r\r | |
845 | break;\r\r | |
846 | case 'V':\r\r | |
847 | if (in.charAt(current + 1) == 'V')\r\r | |
848 | current += 2;\r\r | |
849 | else\r\r | |
850 | current += 1;\r\r | |
851 | MetaphAdd(primary, 'F');\r\r | |
852 | break;\r\r | |
853 | case 'W':\r\r | |
854 | if (stringAt(in, current, 2, list88)) {\r\r | |
855 | MetaphAdd(primary, 'R');\r\r | |
856 | current += 2;\r\r | |
857 | break;\r\r | |
858 | }\r\r | |
859 | if ((current == 0) && (isVowel(in, current + 1, length) || stringAt(in, current,\r\r | |
860 | 2, list89))) {\r\r | |
861 | MetaphAdd(primary, 'A');\r\r | |
862 | }\r\r | |
863 | if (((current == last) && isVowel(in, current - 1, length)) || stringAt(in, current\r\r | |
864 | - 1, 5, list90) || stringAt(in, 0, 3, list91)) {\r\r | |
865 | MetaphAdd(primary, 'F');\r\r | |
866 | current += 1;\r\r | |
867 | break;\r\r | |
868 | }\r\r | |
869 | if (stringAt(in, current, 4, list92)) {\r\r | |
870 | MetaphAdd(primary, "TS");\r\r | |
871 | current += 4;\r\r | |
872 | break;\r\r | |
873 | }\r\r | |
874 | current += 1;\r\r | |
875 | break;\r\r | |
876 | case 'X':\r\r | |
877 | if (!((current == last) && (stringAt(in, current - 3, 3, list93) ||\r\r | |
878 | stringAt(in, current - 2, 2, list94))))\r\r | |
879 | MetaphAdd(primary, "KS");\r\r | |
880 | if (stringAt(in, current + 1, 1, list95))\r\r | |
881 | current += 2;\r\r | |
882 | else\r\r | |
883 | current += 1;\r\r | |
884 | break;\r\r | |
885 | case 'Z':\r\r | |
886 | if (in.charAt(current + 1) == 'H') {\r\r | |
887 | MetaphAdd(primary, 'J');\r\r | |
888 | current += 2;\r\r | |
889 | break;\r\r | |
890 | } \r\r | |
891 | else {\r\r | |
892 | MetaphAdd(primary, 'S');\r\r | |
893 | }\r\r | |
894 | if (in.charAt(current + 1) == 'Z')\r\r | |
895 | current += 2;\r\r | |
896 | else\r\r | |
897 | current += 1;\r\r | |
898 | break;\r\r | |
899 | default:\r\r | |
900 | current += 1;\r\r | |
901 | }\r\r | |
902 | }\r\r | |
903 | return primary.toString();\r\r | |
904 | }\r\r | |
905 | }\r\r | |
906 | \r\r | |
907 | \r\r | |
908 | \r\r |