PageRenderTime 52ms CodeModel.GetById 20ms RepoModel.GetById 0ms app.codeStats 1ms

/modules/analysis/kuromoji/src/java/org/apache/lucene/analysis/kuromoji/util/ToStringUtil.java

http://github.com/apache/lucene-solr
Java | 1023 lines | 981 code | 10 blank | 32 comment | 367 complexity | a5c4e7f943c7c95bc1f1b920a7efb55c MD5 | raw file
Possible License(s): LGPL-2.1, CPL-1.0, MPL-2.0-no-copyleft-exception, JSON, Apache-2.0, AGPL-1.0, GPL-2.0, GPL-3.0, MIT, BSD-3-Clause 
    

  1. package org.apache.lucene.analysis.kuromoji.util;
    2. 

  2. 

  3. /**
    4. 

  4.  * Licensed to the Apache Software Foundation (ASF) under one or more
    5. 

  5.  * contributor license agreements.  See the NOTICE file distributed with
    6. 

  6.  * this work for additional information regarding copyright ownership.
    7. 

  7.  * The ASF licenses this file to You under the Apache License, Version 2.0
    8. 

  8.  * (the "License"); you may not use this file except in compliance with
    9. 

  9.  * the License.  You may obtain a copy of the License at
    10. 

  10.  *
    11. 

  11.  *     http://www.apache.org/licenses/LICENSE-2.0
    12. 

  12.  *
    13. 

  13.  * Unless required by applicable law or agreed to in writing, software
    14. 

  14.  * distributed under the License is distributed on an "AS IS" BASIS,
    15. 

  15.  * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
    16. 

  16.  * See the License for the specific language governing permissions and
    17. 

  17.  * limitations under the License.
    18. 

  18.  */
    19. 

  19. 

  20. import java.util.HashMap;
    21. 

  21. 

  22. public class ToStringUtil {
    23. 

  23.   // a translation map for parts of speech, only used for reflectWith
    24. 

  24.   private static final HashMap<String,String> posTranslations = new HashMap<String,String>();
    25. 

  25.   static {
    26. 

  26.     posTranslations.put("??", "noun");
    27. 

  27.     posTranslations.put("??-??", "noun-common");
    28. 

  28.     posTranslations.put("??-????", "noun-proper");
    29. 

  29.     posTranslations.put("??-????-??", "noun-proper-misc");
    30. 

  30.     posTranslations.put("??-????-??", "noun-proper-person");
    31. 

  31.     posTranslations.put("??-????-??-??", "noun-proper-person-misc");
    32. 

  32.     posTranslations.put("??-????-??-?", "noun-proper-person-surname");
    33. 

  33.     posTranslations.put("??-????-??-?", "noun-proper-person-given_name");
    34. 

  34.     posTranslations.put("??-????-??", "noun-proper-organization");
    35. 

  35.     posTranslations.put("??-????-??", "noun-proper-place");
    36. 

  36.     posTranslations.put("??-????-??-??", "noun-proper-place-misc");
    37. 

  37.     posTranslations.put("??-????-??-?", "noun-proper-place-country");
    38. 

  38.     posTranslations.put("??-???", "noun-pronoun");
    39. 

  39.     posTranslations.put("??-???-??", "noun-pronoun-misc");
    40. 

  40.     posTranslations.put("??-???-??", "noun-pronoun-contraction");
    41. 

  41.     posTranslations.put("??-????", "noun-adverbial");
    42. 

  42.     posTranslations.put("??-????", "noun-verbal");
    43. 

  43.     posTranslations.put("??-??????", "noun-adjective-base");
    44. 

  44.     posTranslations.put("??-?", "noun-numeric");
    45. 

  45.     posTranslations.put("??-???", "noun-affix");
    46. 

  46.     posTranslations.put("??-???-??", "noun-affix-misc");
    47. 

  47.     posTranslations.put("??-???-????", "noun-affix-adverbial");
    48. 

  48.     posTranslations.put("??-???-?????", "noun-affix-aux");
    49. 

  49.     posTranslations.put("??-???-??????", "noun-affix-adjective-base");
    50. 

  50.     posTranslations.put("??-??", "noun-special");
    51. 

  51.     posTranslations.put("??-??-?????", "noun-special-aux");
    52. 

  52.     posTranslations.put("??-??", "noun-suffix");
    53. 

  53.     posTranslations.put("??-??-??", "noun-suffix-misc");
    54. 

  54.     posTranslations.put("??-??-??", "noun-suffix-person");
    55. 

  55.     posTranslations.put("??-??-??", "noun-suffix-place");
    56. 

  56.     posTranslations.put("??-??-????", "noun-suffix-verbal");
    57. 

  57.     posTranslations.put("??-??-?????", "noun-suffix-aux");
    58. 

  58.     posTranslations.put("??-??-??????", "noun-suffix-adjective-base");
    59. 

  59.     posTranslations.put("??-??-????", "noun-suffix-adverbial");
    60. 

  60.     posTranslations.put("??-??-???", "noun-suffix-classifier");
    61. 

  61.     posTranslations.put("??-??-??", "noun-suffix-special");
    62. 

  62.     posTranslations.put("??-????", "noun-suffix-conjunctive");
    63. 

  63.     posTranslations.put("??-??????", "noun-verbal_aux");
    64. 

  64.     posTranslations.put("??-?????", "noun-quotation");
    65. 

  65.     posTranslations.put("??-???????", "noun-nai_adjective");
    66. 

  66.     posTranslations.put("???", "prefix");
    67. 

  67.     posTranslations.put("???-????", "prefix-nominal");
    68. 

  68.     posTranslations.put("???-????", "prefix-verbal");
    69. 

  69.     posTranslations.put("???-?????", "prefix-adjectival");
    70. 

  70.     posTranslations.put("???-???", "prefix-numerical");
    71. 

  71.     posTranslations.put("??", "verb");
    72. 

  72.     posTranslations.put("??-??", "verb-main");
    73. 

  73.     posTranslations.put("??-???", "verb-auxiliary");
    74. 

  74.     posTranslations.put("??-??", "verb-suffix");
    75. 

  75.     posTranslations.put("???", "adjective");
    76. 

  76.     posTranslations.put("???-??", "adjective-main");
    77. 

  77.     posTranslations.put("???-???", "adjective-auxiliary");
    78. 

  78.     posTranslations.put("???-??", "adjective-suffix");
    79. 

  79.     posTranslations.put("??", "adverb");
    80. 

  80.     posTranslations.put("??-??", "adverb-misc");
    81. 

  81.     posTranslations.put("??-?????", "adverb-particle_conjunction");
    82. 

  82.     posTranslations.put("???", "adnominal");
    83. 

  83.     posTranslations.put("???", "conjunction");
    84. 

  84.     posTranslations.put("??", "particle");
    85. 

  85.     posTranslations.put("??-???", "particle-case");
    86. 

  86.     posTranslations.put("??-???-??", "particle-case-misc");
    87. 

  87.     posTranslations.put("??-???-??", "particle-case-quote");
    88. 

  88.     posTranslations.put("??-???-??", "particle-case-compound");
    89. 

  89.     posTranslations.put("??-????", "particle-conjunctive");
    90. 

  90.     posTranslations.put("??-???", "particle-dependency");
    91. 

  91.     posTranslations.put("??-???", "particle-adverbial");
    92. 

  92.     posTranslations.put("??-????", "particle-interjective");
    93. 

  93.     posTranslations.put("??-????", "particle-coordinate");
    94. 

  94.     posTranslations.put("??-???", "particle-final");
    95. 

  95.     posTranslations.put("??-????????????", "particle-adverbial/conjunctive/final");
    96. 

  96.     posTranslations.put("??-???", "particle-adnominalizer");
    97. 

  97.     posTranslations.put("??-???", "particle-adnominalizer");
    98. 

  98.     posTranslations.put("??-??", "particle-special");
    99. 

  99.     posTranslations.put("???", "auxiliary-verb");
    100. 

  100.     posTranslations.put("???", "interjection");
    101. 

  101.     posTranslations.put("??", "symbol");
    102. 

  102.     posTranslations.put("??-??", "symbol-misc");
    103. 

  103.     posTranslations.put("??-??", "symbol-period");
    104. 

  104.     posTranslations.put("??-??", "symbol-comma");
    105. 

  105.     posTranslations.put("??-??", "symbol-space");
    106. 

  106.     posTranslations.put("??-???", "symbol-open_bracket");
    107. 

  107.     posTranslations.put("??-???", "symbol-close_bracket");
    108. 

  108.     posTranslations.put("??-???????", "symbol-alphabetic");
    109. 

  109.     posTranslations.put("???", "other");
    110. 

  110.     posTranslations.put("???-??", "other-interjection");
    111. 

  111.     posTranslations.put("????", "filler");
    112. 

  112.     posTranslations.put("????", "non-verbal");
    113. 

  113.     posTranslations.put("???", "fragment");
    114. 

  114.     posTranslations.put("???", "unknown");
    115. 

  115.   }
    116. 

  116.   
    117. 

  117.   /**
    118. 

  118.    * Get the english form of a POS tag
    119. 

  119.    */
    120. 

  120.   public static String getPOSTranslation(String s) {
    121. 

  121.     return posTranslations.get(s);
    122. 

  122.   }
    123. 

  123.   
    124. 

  124.   // a translation map for inflection types, only used for reflectWith
    125. 

  125.   private static final HashMap<String,String> inflTypeTranslations = new HashMap<String,String>();
    126. 

  126.   static {
    127. 

  127.     inflTypeTranslations.put("*", "*");
    128. 

  128.     inflTypeTranslations.put("????????", "adj-group-a-o-u");
    129. 

  129.     inflTypeTranslations.put("??????", "adj-group-i");
    130. 

  130.     inflTypeTranslations.put("??????",  "adj-group-ii");
    131. 

  131.     inflTypeTranslations.put("????", "non-inflectional");
    132. 

  132.     inflTypeTranslations.put("????", "special-da");
    133. 

  133.     inflTypeTranslations.put("????", "special-ta");
    134. 

  134.     inflTypeTranslations.put("??????", "classical-gotoshi");
    135. 

  135.     inflTypeTranslations.put("?????", "special-ja");
    136. 

  136.     inflTypeTranslations.put("?????", "special-nai");
    137. 

  137.     inflTypeTranslations.put("???????", "5-row-cons-r-special");
    138. 

  138.     inflTypeTranslations.put("????", "special-nu");
    139. 

  139.     inflTypeTranslations.put("????", "classical-ki");
    140. 

  140.     inflTypeTranslations.put("?????", "special-tai");
    141. 

  141.     inflTypeTranslations.put("?????", "classical-beshi");
    142. 

  142.     inflTypeTranslations.put("????", "special-ya");
    143. 

  143.     inflTypeTranslations.put("?????", "classical-maji");
    144. 

  144.     inflTypeTranslations.put("?????", "2-row-lower-cons-t");
    145. 

  145.     inflTypeTranslations.put("?????", "special-desu");
    146. 

  146.     inflTypeTranslations.put("?????", "special-masu");
    147. 

  147.     inflTypeTranslations.put("???????", "5-row-aru");
    148. 

  148.     inflTypeTranslations.put("?????", "classical-nari");
    149. 

  149.     inflTypeTranslations.put("????", "classical-ri");
    150. 

  150.     inflTypeTranslations.put("?????", "classical-keri");
    151. 

  151.     inflTypeTranslations.put("????", "classical-ru");
    152. 

  152.     inflTypeTranslations.put("????????", "5-row-cons-k-i-onbin");
    153. 

  153.     inflTypeTranslations.put("?????", "5-row-cons-s");
    154. 

  154.     inflTypeTranslations.put("??", "1-row");
    155. 

  155.     inflTypeTranslations.put("????????", "5-row-cons-w-cons-onbin");
    156. 

  156.     inflTypeTranslations.put("?????", "5-row-cons-m");
    157. 

  157.     inflTypeTranslations.put("?????", "5-row-cons-t");
    158. 

  158.     inflTypeTranslations.put("?????", "5-row-cons-r");
    159. 

  159.     inflTypeTranslations.put("??????", "irregular-suffix-suru");
    160. 

  160.     inflTypeTranslations.put("?????", "5-row-cons-g");
    161. 

  161.     inflTypeTranslations.put("??????", "irregular-suffix-zuru");
    162. 

  162.     inflTypeTranslations.put("?????", "5-row-cons-b");
    163. 

  163.     inflTypeTranslations.put("????????", "5-row-cons-w-u-onbin");
    164. 

  164.     inflTypeTranslations.put("?????", "2-row-lower-cons-d");
    165. 

  165.     inflTypeTranslations.put("??????????", "5-row-cons-k-cons-onbin-yuku");
    166. 

  166.     inflTypeTranslations.put("?????", "2-row-upper-cons-d");
    167. 

  167.     inflTypeTranslations.put("????????", "5-row-cons-k-cons-onbin");
    168. 

  168.     inflTypeTranslations.put("?????", "1-row-eru");
    169. 

  169.     inflTypeTranslations.put("?????", "4-row-cons-t");
    170. 

  170.     inflTypeTranslations.put("?????", "5-row-cons-n");
    171. 

  171.     inflTypeTranslations.put("?????", "2-row-lower-cons-h");
    172. 

  172.     inflTypeTranslations.put("?????", "4-row-cons-h");
    173. 

  173.     inflTypeTranslations.put("?????", "4-row-cons-b");
    174. 

  174.     inflTypeTranslations.put("?????", "irregular-suru");
    175. 

  175.     inflTypeTranslations.put("?????", "2-row-upper-cons-h");
    176. 

  176.     inflTypeTranslations.put("?????", "2-row-lower-cons-m");
    177. 

  177.     inflTypeTranslations.put("?????", "4-row-cons-s");
    178. 

  178.     inflTypeTranslations.put("?????", "2-row-lower-cons-g");
    179. 

  179.     inflTypeTranslations.put("?????", "kuru-kanji");
    180. 

  180.     inflTypeTranslations.put("??????", "1-row-kureru");
    181. 

  181.     inflTypeTranslations.put("????", "2-row-lower-u");
    182. 

  182.     inflTypeTranslations.put("?????", "kuru-kana");
    183. 

  183.     inflTypeTranslations.put("??", "irregular-cons-r");
    184. 

  184.     inflTypeTranslations.put("?????", "2-row-lower-cons-k");
    185. 

  185.   }
    186. 

  186.   
    187. 

  187.   /**
    188. 

  188.    * Get the english form of inflection type
    189. 

  189.    */
    190. 

  190.   public static String getInflectionTypeTranslation(String s) {
    191. 

  191.     return inflTypeTranslations.get(s);
    192. 

  192.   }
    193. 

  193. 

  194.   // a translation map for inflection forms, only used for reflectWith
    195. 

  195.   private static final HashMap<String,String> inflFormTranslations = new HashMap<String,String>();
    196. 

  196.   static {
    197. 

  197.     inflFormTranslations.put("*", "*");
    198. 

  198.     inflFormTranslations.put("???", "base");
    199. 

  199.     inflFormTranslations.put("?????", "classical-base");
    200. 

  200.     inflFormTranslations.put("?????", "imperfective-nu-connection");
    201. 

  201.     inflFormTranslations.put("?????", "imperfective-u-connection");
    202. 

  202.     inflFormTranslations.put("?????", "conjunctive-ta-connection");
    203. 

  203.     inflFormTranslations.put("?????", "conjunctive-te-connection");
    204. 

  204.     inflFormTranslations.put("???????", "conjunctive-gozai-connection");
    205. 

  205.     inflFormTranslations.put("????", "uninflected-connection");
    206. 

  206.     inflFormTranslations.put("???", "subjunctive");
    207. 

  207.     inflFormTranslations.put("???", "imperative-e");
    208. 

  208.     inflFormTranslations.put("?????", "conditional-contracted-1");
    209. 

  209.     inflFormTranslations.put("?????", "conditional-contracted-2");
    210. 

  210.     inflFormTranslations.put("????", "garu-connection");
    211. 

  211.     inflFormTranslations.put("???", "imperfective");
    212. 

  212.     inflFormTranslations.put("???", "conjunctive");
    213. 

  213.     inflFormTranslations.put("?????", "onbin-base");
    214. 

  214.     inflFormTranslations.put("?????", "conjunctive-de-connection");
    215. 

  215.     inflFormTranslations.put("????", "imperfective-special");
    216. 

  216.     inflFormTranslations.put("???", "imperative-i");
    217. 

  217.     inflFormTranslations.put("?????", "conjunctive-ni-connection");
    218. 

  218.     inflFormTranslations.put("????", "imperative-yo");
    219. 

  219.     inflFormTranslations.put("??????", "adnominal-special");
    220. 

  220.     inflFormTranslations.put("????", "imperative-ro");
    221. 

  221.     inflFormTranslations.put("???????", "uninflected-special-connection-2");
    222. 

  222.     inflFormTranslations.put("??????", "imperfective-reru-connection");
    223. 

  223.     inflFormTranslations.put("?????", "modern-base");
    224. 

  224.     inflFormTranslations.put("???-???", "base-onbin"); // not sure about this
    225. 

  225.   }
    226. 

  226.   
    227. 

  227.   /**
    228. 

  228.    * Get the english form of inflected form
    229. 

  229.    */
    230. 

  230.   public static String getInflectedFormTranslation(String s) {
    231. 

  231.     return inflFormTranslations.get(s);
    232. 

  232.   }
    233. 

  233.   
    234. 

  234.   /**
    235. 

  235.    * Romanize katakana with modified hepburn
    236. 

  236.    */
    237. 

  237.   public static String getRomanization(String s) {
    238. 

  238.     StringBuilder builder = new StringBuilder();
    239. 

  239.     final int len = s.length();
    240. 

  240.     for (int i = 0; i < len; i++) {
    241. 

  241.       // maximum lookahead: 3
    242. 

  242.       char ch = s.charAt(i);
    243. 

  243.       char ch2 = (i < len - 1) ? s.charAt(i + 1) : 0;
    244. 

  244.       char ch3 = (i < len - 2) ? s.charAt(i + 2) : 0;
    245. 

  245.       
    246. 

  246.       main: switch (ch) {
    247. 

  247.         case '?':
    248. 

  248.           switch (ch2) {
    249. 

  249.             case '?':
    250. 

  250.             case '?':
    251. 

  251.             case '?':
    252. 

  252.             case '?':
    253. 

  253.             case '?':
    254. 

  254.               builder.append('k');
    255. 

  255.               break main;
    256. 

  256.             case '?':
    257. 

  257.             case '?':
    258. 

  258.             case '?':
    259. 

  259.             case '?':
    260. 

  260.             case '?':
    261. 

  261.               builder.append('s');
    262. 

  262.               break main;
    263. 

  263.             case '?':
    264. 

  264.             case '?':
    265. 

  265.             case '?':
    266. 

  266.             case '?':
    267. 

  267.             case '?':
    268. 

  268.               builder.append('t');
    269. 

  269.               break main;
    270. 

  270.             case '?':
    271. 

  271.             case '?':
    272. 

  272.             case '?':
    273. 

  273.             case '?':
    274. 

  274.             case '?':
    275. 

  275.               builder.append('p');
    276. 

  276.               break main;
    277. 

  277.           }
    278. 

  278.           break;
    279. 

  279.         case '?':
    280. 

  280.           builder.append('a');
    281. 

  281.           break;
    282. 

  282.         case '?':
    283. 

  283.           if (ch2 == '?') {
    284. 

  284.             builder.append("yi");
    285. 

  285.             i++;
    286. 

  286.           } else if (ch2 == '?') {
    287. 

  287.             builder.append("ye");
    288. 

  288.             i++;
    289. 

  289.           } else {
    290. 

  290.             builder.append('i');
    291. 

  291.           }
    292. 

  292.           break;
    293. 

  293.         case '?':
    294. 

  294.           switch(ch2) {
    295. 

  295.             case '?':
    296. 

  296.               builder.append("wa");
    297. 

  297.               i++;
    298. 

  298.               break;
    299. 

  299.             case '?':
    300. 

  300.               builder.append("wi");
    301. 

  301.               i++;
    302. 

  302.               break;
    303. 

  303.             case '?':
    304. 

  304.               builder.append("wu");
    305. 

  305.               i++;
    306. 

  306.               break;
    307. 

  307.             case '?':
    308. 

  308.               builder.append("we");
    309. 

  309.               i++;
    310. 

  310.               break;
    311. 

  311.             case '?':
    312. 

  312.               builder.append("wo");
    313. 

  313.               i++;
    314. 

  314.               break;
    315. 

  315.             case '?':
    316. 

  316.               builder.append("wyu");
    317. 

  317.               i++;
    318. 

  318.               break;
    319. 

  319.             default:
    320. 

  320.               builder.append('u');
    321. 

  321.               break;
    322. 

  322.           }
    323. 

  323.           break;
    324. 

  324.         case '?':
    325. 

  325.           builder.append('e');
    326. 

  326.           break;
    327. 

  327.         case '?':
    328. 

  328.           if (ch2 == '?') {
    329. 

  329.             builder.append('?');
    330. 

  330.             i++;
    331. 

  331.           } else {
    332. 

  332.             builder.append('o');
    333. 

  333.           }
    334. 

  334.           break;
    335. 

  335.         case '?':
    336. 

  336.           builder.append("ka");
    337. 

  337.           break;
    338. 

  338.         case '?':
    339. 

  339.           if (ch2 == '?' && ch3 == '?') {
    340. 

  340.             builder.append("ky?");
    341. 

  341.             i += 2;
    342. 

  342.           } else if (ch2 == '?' && ch3 == '?') {
    343. 

  343.             builder.append("ky?");
    344. 

  344.             i += 2;
    345. 

  345.           } else if (ch2 == '?') {
    346. 

  346.             builder.append("kya");
    347. 

  347.             i++;
    348. 

  348.           } else if (ch2 == '?') {
    349. 

  349.             builder.append("kyo");
    350. 

  350.             i++;
    351. 

  351.           } else if (ch2 == '?') {
    352. 

  352.             builder.append("kyu");
    353. 

  353.             i++;
    354. 

  354.           } else if (ch2 == '?') {
    355. 

  355.             builder.append("kye");
    356. 

  356.             i++;
    357. 

  357.           } else {
    358. 

  358.             builder.append("ki");
    359. 

  359.           }
    360. 

  360.           break;
    361. 

  361.         case '?':
    362. 

  362.           switch(ch2) {
    363. 

  363.             case '?':
    364. 

  364.               builder.append("kwa");
    365. 

  365.               i++;
    366. 

  366.               break;
    367. 

  367.             case '?':
    368. 

  368.               builder.append("kwi");
    369. 

  369.               i++;
    370. 

  370.               break;
    371. 

  371.             case '?':
    372. 

  372.               builder.append("kwe");
    373. 

  373.               i++;
    374. 

  374.               break;
    375. 

  375.             case '?':
    376. 

  376.               builder.append("kwo");
    377. 

  377.               i++;
    378. 

  378.               break;
    379. 

  379.             case '?':
    380. 

  380.               builder.append("kwa");
    381. 

  381.               i++;
    382. 

  382.               break;
    383. 

  383.             default:
    384. 

  384.               builder.append("ku");
    385. 

  385.               break;
    386. 

  386.           }
    387. 

  387.           break;
    388. 

  388.         case '?':
    389. 

  389.           builder.append("ke");
    390. 

  390.           break;
    391. 

  391.         case '?':
    392. 

  392.           if (ch2 == '?') {
    393. 

  393.             builder.append("k?");
    394. 

  394.             i++;
    395. 

  395.           } else {
    396. 

  396.             builder.append("ko");
    397. 

  397.           }
    398. 

  398.           break;
    399. 

  399.         case '?':
    400. 

  400.           builder.append("sa");
    401. 

  401.           break;
    402. 

  402.         case '?':
    403. 

  403.           if (ch2 == '?' && ch3 == '?') {
    404. 

  404.             builder.append("sh?");
    405. 

  405.             i += 2;
    406. 

  406.           } else if (ch2 == '?' && ch3 == '?') {
    407. 

  407.             builder.append("sh?");
    408. 

  408.             i += 2;
    409. 

  409.           } else if (ch2 == '?') {
    410. 

  410.             builder.append("sha");
    411. 

  411.             i++;
    412. 

  412.           } else if (ch2 == '?') {
    413. 

  413.             builder.append("sho");
    414. 

  414.             i++;
    415. 

  415.           } else if (ch2 == '?') {
    416. 

  416.             builder.append("shu");
    417. 

  417.             i++;
    418. 

  418.           } else if (ch2 == '?') {
    419. 

  419.             builder.append("she");
    420. 

  420.             i++;
    421. 

  421.           } else {
    422. 

  422.             builder.append("shi");
    423. 

  423.           }
    424. 

  424.           break;
    425. 

  425.         case '?':
    426. 

  426.           if (ch2 == '?') {
    427. 

  427.             builder.append("si");
    428. 

  428.             i++;
    429. 

  429.           } else {
    430. 

  430.             builder.append("su");
    431. 

  431.           }
    432. 

  432.           break;
    433. 

  433.         case '?':
    434. 

  434.           builder.append("se");
    435. 

  435.           break;
    436. 

  436.         case '?':
    437. 

  437.           if (ch2 == '?') {
    438. 

  438.             builder.append("s?");
    439. 

  439.             i++;
    440. 

  440.           } else {
    441. 

  441.             builder.append("so");
    442. 

  442.           }
    443. 

  443.           break;
    444. 

  444.         case '?':
    445. 

  445.           builder.append("ta");
    446. 

  446.           break;
    447. 

  447.         case '?':
    448. 

  448.           if (ch2 == '?' && ch3 == '?') {
    449. 

  449.             builder.append("ch?");
    450. 

  450.             i += 2;
    451. 

  451.           } else if (ch2 == '?' && ch3 == '?') {
    452. 

  452.             builder.append("ch?");
    453. 

  453.             i += 2;
    454. 

  454.           } else if (ch2 == '?') {
    455. 

  455.             builder.append("cha");
    456. 

  456.             i++;
    457. 

  457.           } else if (ch2 == '?') {
    458. 

  458.             builder.append("cho");
    459. 

  459.             i++;
    460. 

  460.           } else if (ch2 == '?') {
    461. 

  461.             builder.append("chu");
    462. 

  462.             i++;
    463. 

  463.           } else if (ch2 == '?') {
    464. 

  464.             builder.append("che");
    465. 

  465.             i++;
    466. 

  466.           } else {
    467. 

  467.             builder.append("chi");
    468. 

  468.           }
    469. 

  469.           break;
    470. 

  470.         case '?':
    471. 

  471.           if (ch2 == '?') {
    472. 

  472.             builder.append("tsa");
    473. 

  473.             i++;
    474. 

  474.           } else if (ch2 == '?') {
    475. 

  475.             builder.append("tsi");
    476. 

  476.             i++;
    477. 

  477.           } else if (ch2 == '?') {
    478. 

  478.             builder.append("tse");
    479. 

  479.             i++;
    480. 

  480.           } else if (ch2 == '?') {
    481. 

  481.             builder.append("tso");
    482. 

  482.             i++;
    483. 

  483.           } else if (ch2 == '?') {
    484. 

  484.             builder.append("tsyu");
    485. 

  485.             i++;
    486. 

  486.           } else {
    487. 

  487.             builder.append("tsu");
    488. 

  488.           }
    489. 

  489.           break;
    490. 

  490.         case '?':
    491. 

  491.           if (ch2 == '?') {
    492. 

  492.             builder.append("ti");
    493. 

  493.             i++;
    494. 

  494.           } else if (ch2 == '?') {
    495. 

  495.             builder.append("tu");
    496. 

  496.             i++;
    497. 

  497.           } else if (ch2 == '?') {
    498. 

  498.             builder.append("tyu");
    499. 

  499.             i++;
    500. 

  500.           } else {
    501. 

  501.             builder.append("te");
    502. 

  502.           }
    503. 

  503.           break;
    504. 

  504.         case '?':
    505. 

  505.           if (ch2 == '?') {
    506. 

  506.             builder.append("t?");
    507. 

  507.             i++;
    508. 

  508.           } else {
    509. 

  509.             builder.append("to");
    510. 

  510.           }
    511. 

  511.           break;
    512. 

  512.         case '?':
    513. 

  513.           builder.append("na");
    514. 

  514.           break;
    515. 

  515.         case '?':
    516. 

  516.           if (ch2 == '?' && ch3 == '?') {
    517. 

  517.             builder.append("ny?");
    518. 

  518.             i += 2;
    519. 

  519.           } else if (ch2 == '?' && ch3 == '?') {
    520. 

  520.             builder.append("ny?");
    521. 

  521.             i += 2;
    522. 

  522.           } else if (ch2 == '?') {
    523. 

  523.             builder.append("nya");
    524. 

  524.             i++;
    525. 

  525.           } else if (ch2 == '?') {
    526. 

  526.             builder.append("nyo");
    527. 

  527.             i++;
    528. 

  528.           } else if (ch2 == '?') {
    529. 

  529.             builder.append("nyu");
    530. 

  530.             i++;
    531. 

  531.           } else if (ch2 == '?') {
    532. 

  532.             builder.append("nye");
    533. 

  533.             i++;
    534. 

  534.           } else {
    535. 

  535.             builder.append("ni");
    536. 

  536.           }
    537. 

  537.           break;
    538. 

  538.         case '?':
    539. 

  539.           builder.append("nu");
    540. 

  540.           break;
    541. 

  541.         case '?':
    542. 

  542.           builder.append("ne");
    543. 

  543.           break;
    544. 

  544.         case '?':
    545. 

  545.           if (ch2 == '?') {
    546. 

  546.             builder.append("n?");
    547. 

  547.             i++;
    548. 

  548.           } else {
    549. 

  549.             builder.append("no");
    550. 

  550.           }
    551. 

  551.           break;
    552. 

  552.         case '?':
    553. 

  553.           builder.append("ha");
    554. 

  554.           break;
    555. 

  555.         case '?':
    556. 

  556.           if (ch2 == '?' && ch3 == '?') {
    557. 

  557.             builder.append("hy?");
    558. 

  558.             i += 2;
    559. 

  559.           } else if (ch2 == '?' && ch3 == '?') {
    560. 

  560.             builder.append("hy?");
    561. 

  561.             i += 2;
    562. 

  562.           } else if (ch2 == '?') {
    563. 

  563.             builder.append("hya");
    564. 

  564.             i++;
    565. 

  565.           } else if (ch2 == '?') {
    566. 

  566.             builder.append("hyo");
    567. 

  567.             i++;
    568. 

  568.           } else if (ch2 == '?') {
    569. 

  569.             builder.append("hyu");
    570. 

  570.             i++;
    571. 

  571.           } else if (ch2 == '?') {
    572. 

  572.             builder.append("hye");
    573. 

  573.             i++;
    574. 

  574.           } else {
    575. 

  575.             builder.append("hi");
    576. 

  576.           }
    577. 

  577.           break;
    578. 

  578.         case '?':
    579. 

  579.           if (ch2 == '?') {
    580. 

  580.             builder.append("fya");
    581. 

  581.             i++;
    582. 

  582.           } else if (ch2 == '?') {
    583. 

  583.             builder.append("fyu");
    584. 

  584.             i++;
    585. 

  585.           } else if (ch2 == '?' && ch3 == '?') {
    586. 

  586.             builder.append("fye");
    587. 

  587.             i+=2;
    588. 

  588.           } else if (ch2 == '?') {
    589. 

  589.             builder.append("fyo");
    590. 

  590.             i++;
    591. 

  591.           } else if (ch2 == '?') {
    592. 

  592.             builder.append("fa");
    593. 

  593.             i++;
    594. 

  594.           } else if (ch2 == '?') {
    595. 

  595.             builder.append("fi");
    596. 

  596.             i++;
    597. 

  597.           } else if (ch2 == '?') {
    598. 

  598.             builder.append("fe");
    599. 

  599.             i++;
    600. 

  600.           } else if (ch2 == '?') {
    601. 

  601.             builder.append("fo");
    602. 

  602.             i++;
    603. 

  603.           } else {
    604. 

  604.             builder.append("fu");
    605. 

  605.           }
    606. 

  606.           break;
    607. 

  607.         case '?':
    608. 

  608.           builder.append("he");
    609. 

  609.           break;
    610. 

  610.         case '?':
    611. 

  611.           if (ch2 == '?') {
    612. 

  612.             builder.append("h?");
    613. 

  613.             i++;
    614. 

  614.           } else if (ch2 == '?') {
    615. 

  615.             builder.append("hu");
    616. 

  616.             i++;
    617. 

  617.           } else {
    618. 

  618.             builder.append("ho");
    619. 

  619.           }
    620. 

  620.           break;
    621. 

  621.         case '?':
    622. 

  622.           builder.append("ma");
    623. 

  623.           break;
    624. 

  624.         case '?':
    625. 

  625.           if (ch2 == '?' && ch3 == '?') {
    626. 

  626.             builder.append("my?");
    627. 

  627.             i += 2;
    628. 

  628.           } else if (ch2 == '?' && ch3 == '?') {
    629. 

  629.             builder.append("my?");
    630. 

  630.             i += 2;
    631. 

  631.           } else if (ch2 == '?') {
    632. 

  632.             builder.append("mya");
    633. 

  633.             i++;
    634. 

  634.           } else if (ch2 == '?') {
    635. 

  635.             builder.append("myo");
    636. 

  636.             i++;
    637. 

  637.           } else if (ch2 == '?') {
    638. 

  638.             builder.append("myu");
    639. 

  639.             i++;
    640. 

  640.           } else if (ch2 == '?') {
    641. 

  641.             builder.append("mye");
    642. 

  642.             i++;
    643. 

  643.           } else {
    644. 

  644.             builder.append("mi");
    645. 

  645.           }
    646. 

  646.           break;
    647. 

  647.         case '?':
    648. 

  648.           builder.append("mu");
    649. 

  649.           break;
    650. 

  650.         case '?':
    651. 

  651.           builder.append("mi");
    652. 

  652.           break;
    653. 

  653.         case '?':
    654. 

  654.           if (ch2 == '?') {
    655. 

  655.             builder.append("m?");
    656. 

  656.             i++;
    657. 

  657.           } else {
    658. 

  658.             builder.append("mo");
    659. 

  659.           }
    660. 

  660.           break;
    661. 

  661.         case '?':
    662. 

  662.           builder.append("ya");
    663. 

  663.           break;
    664. 

  664.         case '?':
    665. 

  665.           builder.append("yu");
    666. 

  666.           break;
    667. 

  667.         case '?':
    668. 

  668.           if (ch2 == '?') {
    669. 

  669.             builder.append("y?");
    670. 

  670.             i++;
    671. 

  671.           } else {
    672. 

  672.             builder.append("yo");
    673. 

  673.           }
    674. 

  674.           break;
    675. 

  675.         case '?':
    676. 

  676.           builder.append("ra");
    677. 

  677.           break;
    678. 

  678.         case '?':
    679. 

  679.           if (ch2 == '?' && ch3 == '?') {
    680. 

  680.             builder.append("ry?");
    681. 

  681.             i += 2;
    682. 

  682.           } else if (ch2 == '?' && ch3 == '?') {
    683. 

  683.             builder.append("ry?");
    684. 

  684.             i += 2;
    685. 

  685.           } else if (ch2 == '?') {
    686. 

  686.             builder.append("rya");
    687. 

  687.             i++;
    688. 

  688.           } else if (ch2 == '?') {
    689. 

  689.             builder.append("ryo");
    690. 

  690.             i++;
    691. 

  691.           } else if (ch2 == '?') {
    692. 

  692.             builder.append("ryu");
    693. 

  693.             i++;
    694. 

  694.           } else if (ch2 == '?') {
    695. 

  695.             builder.append("rye");
    696. 

  696.             i++;
    697. 

  697.           } else {
    698. 

  698.             builder.append("ri");
    699. 

  699.           }
    700. 

  700.           break;
    701. 

  701.         case '?':
    702. 

  702.           builder.append("ru");
    703. 

  703.           break;
    704. 

  704.         case '?':
    705. 

  705.           builder.append("re");
    706. 

  706.           break;
    707. 

  707.         case '?':
    708. 

  708.           if (ch2 == '?') {
    709. 

  709.             builder.append("r?");
    710. 

  710.             i++;
    711. 

  711.           } else {
    712. 

  712.             builder.append("ro");
    713. 

  713.           }
    714. 

  714.           break;
    715. 

  715.         case '?':
    716. 

  716.           builder.append("wa");
    717. 

  717.           break;
    718. 

  718.         case '?':
    719. 

  719.           builder.append("i");
    720. 

  720.           break;
    721. 

  721.         case '?':
    722. 

  722.           builder.append("e");
    723. 

  723.           break;
    724. 

  724.         case '?':
    725. 

  725.           builder.append("o");
    726. 

  726.           break;
    727. 

  727.         case '?':
    728. 

  728.           switch (ch2) {
    729. 

  729.             case '?':
    730. 

  730.             case '?':
    731. 

  731.             case '?':
    732. 

  732.             case '?':
    733. 

  733.             case '?':
    734. 

  734.             case '?':
    735. 

  735.             case '?':
    736. 

  736.             case '?':
    737. 

  737.             case '?':
    738. 

  738.             case '?':
    739. 

  739.             case '?':
    740. 

  740.             case '?':
    741. 

  741.             case '?':
    742. 

  742.             case '?':
    743. 

  743.             case '?':
    744. 

  744.               builder.append('m');
    745. 

  745.               break main;
    746. 

  746.             case '?':
    747. 

  747.             case '?':
    748. 

  748.             case '?':
    749. 

  749.             case '?':
    750. 

  750.             case '?':
    751. 

  751.             case '?':
    752. 

  752.             case '?':
    753. 

  753.             case '?':
    754. 

  754.               builder.append("n'");
    755. 

  755.               break main;
    756. 

  756.             default:
    757. 

  757.               builder.append("n");
    758. 

  758.               break main;
    759. 

  759.           }
    760. 

  760.         case '?':
    761. 

  761.           builder.append("ga");
    762. 

  762.           break;
    763. 

  763.         case '?':
    764. 

  764.           if (ch2 == '?' && ch3 == '?') {
    765. 

  765.             builder.append("gy?");
    766. 

  766.             i += 2;
    767. 

  767.           } else if (ch2 == '?' && ch3 == '?') {
    768. 

  768.             builder.append("gy?");
    769. 

  769.             i += 2;
    770. 

  770.           } else if (ch2 == '?') {
    771. 

  771.             builder.append("gya");
    772. 

  772.             i++;
    773. 

  773.           } else if (ch2 == '?') {
    774. 

  774.             builder.append("gyo");
    775. 

  775.             i++;
    776. 

  776.           } else if (ch2 == '?') {
    777. 

  777.             builder.append("gyu");
    778. 

  778.             i++;
    779. 

  779.           } else if (ch2 == '?') {
    780. 

  780.             builder.append("gye");
    781. 

  781.             i++;
    782. 

  782.           } else {
    783. 

  783.             builder.append("gi");
    784. 

  784.           }
    785. 

  785.           break;
    786. 

  786.         case '?':
    787. 

  787.           switch(ch2) {
    788. 

  788.             case '?':
    789. 

  789.               builder.append("gwa");
    790. 

  790.               i++;
    791. 

  791.               break;
    792. 

  792.             case '?':
    793. 

  793.               builder.append("gwi");
    794. 

  794.               i++;
    795. 

  795.               break;
    796. 

  796.             case '?':
    797. 

  797.               builder.append("gwe");
    798. 

  798.               i++;
    799. 

  799.               break;
    800. 

  800.             case '?':
    801. 

  801.               builder.append("gwo");
    802. 

  802.               i++;
    803. 

  803.               break;
    804. 

  804.             case '?':
    805. 

  805.               builder.append("gwa");
    806. 

  806.               i++;
    807. 

  807.               break;
    808. 

  808.             default:
    809. 

  809.               builder.append("gu");
    810. 

  810.               break;
    811. 

  811.           }
    812. 

  812.           break;
    813. 

  813.         case '?':
    814. 

  814.           builder.append("ge");
    815. 

  815.           break;
    816. 

  816.         case '?':
    817. 

  817.           if (ch2 == '?') {
    818. 

  818.             builder.append("g?");
    819. 

  819.             i++;
    820. 

  820.           } else {
    821. 

  821.             builder.append("go");
    822. 

  822.           }
    823. 

  823.           break;
    824. 

  824.         case '?':
    825. 

  825.           builder.append("za");
    826. 

  826.           break;
    827. 

  827.         case '?':
    828. 

  828.           if (ch2 == '?' && ch3 == '?') {
    829. 

  829.             builder.append("j?");
    830. 

  830.             i += 2;
    831. 

  831.           } else if (ch2 == '?' && ch3 == '?') {
    832. 

  832.             builder.append("j?");
    833. 

  833.             i += 2;
    834. 

  834.           } else if (ch2 == '?') {
    835. 

  835.             builder.append("ja");
    836. 

  836.             i++;
    837. 

  837.           } else if (ch2 == '?') {
    838. 

  838.             builder.append("jo");
    839. 

  839.             i++;
    840. 

  840.           } else if (ch2 == '?') {
    841. 

  841.             builder.append("ju");
    842. 

  842.             i++;
    843. 

  843.           } else if (ch2 == '?') {
    844. 

  844.             builder.append("je");
    845. 

  845.             i++;
    846. 

  846.           } else {
    847. 

  847.             builder.append("ji");
    848. 

  848.           }
    849. 

  849.           break;
    850. 

  850.         case '?':
    851. 

  851.           if (ch2 == '?') {
    852. 

  852.             builder.append("zi");
    853. 

  853.             i++;
    854. 

  854.           } else {
    855. 

  855.             builder.append("zu");
    856. 

  856.           }
    857. 

  857.           break;
    858. 

  858.         case '?':
    859. 

  859.           builder.append("ze");
    860. 

  860.           break;
    861. 

  861.         case '?':
    862. 

  862.           if (ch2 == '?') {
    863. 

  863.             builder.append("z?");
    864. 

  864.             i++;
    865. 

  865.           } else {
    866. 

  866.             builder.append("zo");
    867. 

  867.           }
    868. 

  868.           break;
    869. 

  869.         case '?':
    870. 

  870.           builder.append("da");
    871. 

  871.           break;
    872. 

  872.         case '?':
    873. 

  873.           builder.append("ji");
    874. 

  874.           break;
    875. 

  875.         case '?':
    876. 

  876.           builder.append("zu");
    877. 

  877.           break;
    878. 

  878.         case '?':
    879. 

  879.           if (ch2 == '?') {
    880. 

  880.             builder.append("di");
    881. 

  881.             i++;
    882. 

  882.           } else if (ch2 == '?') {
    883. 

  883.             builder.append("dyu");
    884. 

  884.             i++;
    885. 

  885.           } else {
    886. 

  886.             builder.append("de");
    887. 

  887.           }
    888. 

  888.           break;
    889. 

  889.         case '?':
    890. 

  890.           if (ch2 == '?') {
    891. 

  891.             builder.append("d?");
    892. 

  892.             i++;
    893. 

  893.           } else if (ch2 == '?') {
    894. 

  894.             builder.append("du");
    895. 

  895.             i++;
    896. 

  896.           } else {
    897. 

  897.             builder.append("do");
    898. 

  898.           }
    899. 

  899.           break;
    900. 

  900.         case '?':
    901. 

  901.           builder.append("ba");
    902. 

  902.           break;
    903. 

  903.         case '?':
    904. 

  904.           if (ch2 == '?' && ch3 == '?') {
    905. 

  905.             builder.append("by?");
    906. 

  906.             i += 2;
    907. 

  907.           } else if (ch2 == '?' && ch3 == '?') {
    908. 

  908.             builder.append("by?");
    909. 

  909.             i += 2;
    910. 

  910.           } else if (ch2 == '?') {
    911. 

  911.             builder.append("bya");
    912. 

  912.             i++;
    913. 

  913.           } else if (ch2 == '?') {
    914. 

  914.             builder.append("byo");
    915. 

  915.             i++;
    916. 

  916.           } else if (ch2 == '?') {
    917. 

  917.             builder.append("byu");
    918. 

  918.             i++;
    919. 

  919.           } else if (ch2 == '?') {
    920. 

  920.             builder.append("bye");
    921. 

  921.             i++;
    922. 

  922.           } else {
    923. 

  923.             builder.append("bi");
    924. 

  924.           }
    925. 

  925.           break;
    926. 

  926.         case '?':
    927. 

  927.           builder.append("bu");
    928. 

  928.           break;
    929. 

  929.         case '?':
    930. 

  930.           builder.append("be");
    931. 

  931.           break;
    932. 

  932.         case '?':
    933. 

  933.           if (ch2 == '?') {
    934. 

  934.             builder.append("b?");
    935. 

  935.             i++;
    936. 

  936.           } else {
    937. 

  937.             builder.append("bo");
    938. 

  938.           }
    939. 

  939.           break;
    940. 

  940.         case '?':
    941. 

  941.           builder.append("pa");
    942. 

  942.           break;
    943. 

  943.         case '?':
    944. 

  944.           if (ch2 == '?' && ch3 == '?') {
    945. 

  945.             builder.append("py?");
    946. 

  946.             i += 2;
    947. 

  947.           } else if (ch2 == '?' && ch3 == '?') {
    948. 

  948.             builder.append("py?");
    949. 

  949.             i += 2;
    950. 

  950.           } else if (ch2 == '?') {
    951. 

  951.             builder.append("pya");
    952. 

  952.             i++;
    953. 

  953.           } else if (ch2 == '?') {
    954. 

  954.             builder.append("pyo");
    955. 

  955.             i++;
    956. 

  956.           } else if (ch2 == '?') {
    957. 

  957.             builder.append("pyu");
    958. 

  958.             i++;
    959. 

  959.           } else if (ch2 == '?') {
    960. 

  960.             builder.append("pye");
    961. 

  961.             i++;
    962. 

  962.           } else {
    963. 

  963.             builder.append("pi");
    964. 

  964.           }
    965. 

  965.           break;
    966. 

  966.         case '?':
    967. 

  967.           builder.append("pu");
    968. 

  968.           break;
    969. 

  969.         case '?':
    970. 

  970.           builder.append("pe");
    971. 

  971.           break;
    972. 

  972.         case '?':
    973. 

  973.           if (ch2 == '?') {
    974. 

  974.             builder.append("p?");
    975. 

  975.             i++;
    976. 

  976.           } else {
    977. 

  977.             builder.append("po");
    978. 

  978.           }
    979. 

  979.           break;
    980. 

  980.         case '?':
    981. 

  981.           if (ch2 == '?' && ch3 == '?') {
    982. 

  982.             builder.append("vye");
    983. 

  983.             i+= 2;
    984. 

  984.           } else {
    985. 

  985.             builder.append('v');
    986. 

  986.           }
    987. 

  987.           break;
    988. 

  988.         case '?':
    989. 

  989.           builder.append('a');
    990. 

  990.           break;
    991. 

  991.         case '?':
    992. 

  992.           builder.append('i');
    993. 

  993.           break;
    994. 

  994.         case '?':
    995. 

  995.           builder.append('u');
    996. 

  996.           break;
    997. 

  997.         case '?':
    998. 

  998.           builder.append('e');
    999. 

  999.           break;
    1000. 

  1000.         case '?':
    1001. 

  1001.           builder.append('o');
    1002. 

  1002.           break;
    1003. 

  1003.         case '?':
    1004. 

  1004.           builder.append("wa");
    1005. 

  1005.           break;
    1006. 

  1006.         case '?':
    1007. 

  1007.           builder.append("ya");
    1008. 

  1008.           break;
    1009. 

  1009.         case '?':
    1010. 

  1010.           builder.append("yu");
    1011. 

  1011.           break;
    1012. 

  1012.         case '?':
    1013. 

  1013.           builder.append("yo");
    1014. 

  1014.           break;
    1015. 

  1015.         case '?':
    1016. 

  1016.           break;
    1017. 

  1017.         default:
    1018. 

  1018.           builder.append(ch);
    1019. 

  1019.       }
    1020. 

  1020.     }
    1021. 

  1021.     return builder.toString();
    1022. 

  1022.   }
    1023. 

  1023. }
    1024. 

  1024.