PageRenderTime 57ms CodeModel.GetById 23ms RepoModel.GetById 0ms app.codeStats 1ms

/lucene-3.6.0/contrib/analyzers/kuromoji/src/java/org/apache/lucene/analysis/ja/util/ToStringUtil.java

#
Java | 1039 lines | 989 code | 11 blank | 39 comment | 367 complexity | ba37499ca1f0423cffe90306a2fbca3f MD5 | raw file
Possible License(s): GPL-3.0, LGPL-2.1, MIT, BSD-3-Clause, CPL-1.0, Apache-2.0, AGPL-1.0, GPL-2.0 
    

  1. package org.apache.lucene.analysis.ja.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.io.IOException;
    21. 

  21. import java.util.HashMap;
    22. 

  22. 

  23. /**
    24. 

  24.  * Utility class for english translations of morphological data,
    25. 

  25.  * used only for debugging.
    26. 

  26.  */
    27. 

  27. public class ToStringUtil {
    28. 

  28.   // a translation map for parts of speech, only used for reflectWith
    29. 

  29.   private static final HashMap<String,String> posTranslations = new HashMap<String,String>();
    30. 

  30.   static {
    31. 

  31.     posTranslations.put("??", "noun");
    32. 

  32.     posTranslations.put("??-??", "noun-common");
    33. 

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

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

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

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

  37.     posTranslations.put("??-????-??-?", "noun-proper-person-surname");
    38. 

  38.     posTranslations.put("??-????-??-?", "noun-proper-person-given_name");
    39. 

  39.     posTranslations.put("??-????-??", "noun-proper-organization");
    40. 

  40.     posTranslations.put("??-????-??", "noun-proper-place");
    41. 

  41.     posTranslations.put("??-????-??-??", "noun-proper-place-misc");
    42. 

  42.     posTranslations.put("??-????-??-?", "noun-proper-place-country");
    43. 

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

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

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

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

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

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

  49.     posTranslations.put("??-?", "noun-numeric");
    50. 

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

  51.     posTranslations.put("??-???-??", "noun-affix-misc");
    52. 

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

  53.     posTranslations.put("??-???-?????", "noun-affix-aux");
    54. 

  54.     posTranslations.put("??-???-??????", "noun-affix-adjective-base");
    55. 

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

  56.     posTranslations.put("??-??-?????", "noun-special-aux");
    57. 

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

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

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

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

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

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

  63.     posTranslations.put("??-??-??????", "noun-suffix-adjective-base");
    64. 

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

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

  66.     posTranslations.put("??-??-??", "noun-suffix-special");
    67. 

  67.     posTranslations.put("??-????", "noun-suffix-conjunctive");
    68. 

  68.     posTranslations.put("??-??????", "noun-verbal_aux");
    69. 

  69.     posTranslations.put("??-?????", "noun-quotation");
    70. 

  70.     posTranslations.put("??-???????", "noun-nai_adjective");
    71. 

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

  72.     posTranslations.put("???-????", "prefix-nominal");
    73. 

  73.     posTranslations.put("???-????", "prefix-verbal");
    74. 

  74.     posTranslations.put("???-?????", "prefix-adjectival");
    75. 

  75.     posTranslations.put("???-???", "prefix-numerical");
    76. 

  76.     posTranslations.put("??", "verb");
    77. 

  77.     posTranslations.put("??-??", "verb-main");
    78. 

  78.     posTranslations.put("??-???", "verb-auxiliary");
    79. 

  79.     posTranslations.put("??-??", "verb-suffix");
    80. 

  80.     posTranslations.put("???", "adjective");
    81. 

  81.     posTranslations.put("???-??", "adjective-main");
    82. 

  82.     posTranslations.put("???-???", "adjective-auxiliary");
    83. 

  83.     posTranslations.put("???-??", "adjective-suffix");
    84. 

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

  85.     posTranslations.put("??-??", "adverb-misc");
    86. 

  86.     posTranslations.put("??-?????", "adverb-particle_conjunction");
    87. 

  87.     posTranslations.put("???", "adnominal");
    88. 

  88.     posTranslations.put("???", "conjunction");
    89. 

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

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

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

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

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

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

  95.     posTranslations.put("??-???", "particle-dependency");
    96. 

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

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

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

  99.     posTranslations.put("??-???", "particle-final");
    100. 

  100.     posTranslations.put("??-????????????", "particle-adverbial/conjunctive/final");
    101. 

  101.     posTranslations.put("??-???", "particle-adnominalizer");
    102. 

  102.     posTranslations.put("??-???", "particle-adnominalizer");
    103. 

  103.     posTranslations.put("??-??", "particle-special");
    104. 

  104.     posTranslations.put("???", "auxiliary-verb");
    105. 

  105.     posTranslations.put("???", "interjection");
    106. 

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

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

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

  109.     posTranslations.put("??-??", "symbol-comma");
    110. 

  110.     posTranslations.put("??-??", "symbol-space");
    111. 

  111.     posTranslations.put("??-???", "symbol-open_bracket");
    112. 

  112.     posTranslations.put("??-???", "symbol-close_bracket");
    113. 

  113.     posTranslations.put("??-???????", "symbol-alphabetic");
    114. 

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

  115.     posTranslations.put("???-??", "other-interjection");
    116. 

  116.     posTranslations.put("????", "filler");
    117. 

  117.     posTranslations.put("????", "non-verbal");
    118. 

  118.     posTranslations.put("???", "fragment");
    119. 

  119.     posTranslations.put("???", "unknown");
    120. 

  120.   }
    121. 

  121.   
    122. 

  122.   /**
    123. 

  123.    * Get the english form of a POS tag
    124. 

  124.    */
    125. 

  125.   public static String getPOSTranslation(String s) {
    126. 

  126.     return posTranslations.get(s);
    127. 

  127.   }
    128. 

  128.   
    129. 

  129.   // a translation map for inflection types, only used for reflectWith
    130. 

  130.   private static final HashMap<String,String> inflTypeTranslations = new HashMap<String,String>();
    131. 

  131.   static {
    132. 

  132.     inflTypeTranslations.put("*", "*");
    133. 

  133.     inflTypeTranslations.put("????????", "adj-group-a-o-u");
    134. 

  134.     inflTypeTranslations.put("??????", "adj-group-i");
    135. 

  135.     inflTypeTranslations.put("??????",  "adj-group-ii");
    136. 

  136.     inflTypeTranslations.put("????", "non-inflectional");
    137. 

  137.     inflTypeTranslations.put("????", "special-da");
    138. 

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

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

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

  141.     inflTypeTranslations.put("?????", "special-nai");
    142. 

  142.     inflTypeTranslations.put("???????", "5-row-cons-r-special");
    143. 

  143.     inflTypeTranslations.put("????", "special-nu");
    144. 

  144.     inflTypeTranslations.put("????", "classical-ki");
    145. 

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

  146.     inflTypeTranslations.put("?????", "classical-beshi");
    147. 

  147.     inflTypeTranslations.put("????", "special-ya");
    148. 

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

  149.     inflTypeTranslations.put("?????", "2-row-lower-cons-t");
    150. 

  150.     inflTypeTranslations.put("?????", "special-desu");
    151. 

  151.     inflTypeTranslations.put("?????", "special-masu");
    152. 

  152.     inflTypeTranslations.put("???????", "5-row-aru");
    153. 

  153.     inflTypeTranslations.put("?????", "classical-nari");
    154. 

  154.     inflTypeTranslations.put("????", "classical-ri");
    155. 

  155.     inflTypeTranslations.put("?????", "classical-keri");
    156. 

  156.     inflTypeTranslations.put("????", "classical-ru");
    157. 

  157.     inflTypeTranslations.put("????????", "5-row-cons-k-i-onbin");
    158. 

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

  159.     inflTypeTranslations.put("??", "1-row");
    160. 

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

  161.     inflTypeTranslations.put("?????", "5-row-cons-m");
    162. 

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

  163.     inflTypeTranslations.put("?????", "5-row-cons-r");
    164. 

  164.     inflTypeTranslations.put("??????", "irregular-suffix-suru");
    165. 

  165.     inflTypeTranslations.put("?????", "5-row-cons-g");
    166. 

  166.     inflTypeTranslations.put("??????", "irregular-suffix-zuru");
    167. 

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

  168.     inflTypeTranslations.put("????????", "5-row-cons-w-u-onbin");
    169. 

  169.     inflTypeTranslations.put("?????", "2-row-lower-cons-d");
    170. 

  170.     inflTypeTranslations.put("??????????", "5-row-cons-k-cons-onbin-yuku");
    171. 

  171.     inflTypeTranslations.put("?????", "2-row-upper-cons-d");
    172. 

  172.     inflTypeTranslations.put("????????", "5-row-cons-k-cons-onbin");
    173. 

  173.     inflTypeTranslations.put("?????", "1-row-eru");
    174. 

  174.     inflTypeTranslations.put("?????", "4-row-cons-t");
    175. 

  175.     inflTypeTranslations.put("?????", "5-row-cons-n");
    176. 

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

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

  178.     inflTypeTranslations.put("?????", "4-row-cons-b");
    179. 

  179.     inflTypeTranslations.put("?????", "irregular-suru");
    180. 

  180.     inflTypeTranslations.put("?????", "2-row-upper-cons-h");
    181. 

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

  182.     inflTypeTranslations.put("?????", "4-row-cons-s");
    183. 

  183.     inflTypeTranslations.put("?????", "2-row-lower-cons-g");
    184. 

  184.     inflTypeTranslations.put("?????", "kuru-kanji");
    185. 

  185.     inflTypeTranslations.put("??????", "1-row-kureru");
    186. 

  186.     inflTypeTranslations.put("????", "2-row-lower-u");
    187. 

  187.     inflTypeTranslations.put("?????", "kuru-kana");
    188. 

  188.     inflTypeTranslations.put("??", "irregular-cons-r");
    189. 

  189.     inflTypeTranslations.put("?????", "2-row-lower-cons-k");
    190. 

  190.   }
    191. 

  191.   
    192. 

  192.   /**
    193. 

  193.    * Get the english form of inflection type
    194. 

  194.    */
    195. 

  195.   public static String getInflectionTypeTranslation(String s) {
    196. 

  196.     return inflTypeTranslations.get(s);
    197. 

  197.   }
    198. 

  198. 

  199.   // a translation map for inflection forms, only used for reflectWith
    200. 

  200.   private static final HashMap<String,String> inflFormTranslations = new HashMap<String,String>();
    201. 

  201.   static {
    202. 

  202.     inflFormTranslations.put("*", "*");
    203. 

  203.     inflFormTranslations.put("???", "base");
    204. 

  204.     inflFormTranslations.put("?????", "classical-base");
    205. 

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

  206.     inflFormTranslations.put("?????", "imperfective-u-connection");
    207. 

  207.     inflFormTranslations.put("?????", "conjunctive-ta-connection");
    208. 

  208.     inflFormTranslations.put("?????", "conjunctive-te-connection");
    209. 

  209.     inflFormTranslations.put("???????", "conjunctive-gozai-connection");
    210. 

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

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

  212.     inflFormTranslations.put("???", "imperative-e");
    213. 

  213.     inflFormTranslations.put("?????", "conditional-contracted-1");
    214. 

  214.     inflFormTranslations.put("?????", "conditional-contracted-2");
    215. 

  215.     inflFormTranslations.put("????", "garu-connection");
    216. 

  216.     inflFormTranslations.put("???", "imperfective");
    217. 

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

  218.     inflFormTranslations.put("?????", "onbin-base");
    219. 

  219.     inflFormTranslations.put("?????", "conjunctive-de-connection");
    220. 

  220.     inflFormTranslations.put("????", "imperfective-special");
    221. 

  221.     inflFormTranslations.put("???", "imperative-i");
    222. 

  222.     inflFormTranslations.put("?????", "conjunctive-ni-connection");
    223. 

  223.     inflFormTranslations.put("????", "imperative-yo");
    224. 

  224.     inflFormTranslations.put("??????", "adnominal-special");
    225. 

  225.     inflFormTranslations.put("????", "imperative-ro");
    226. 

  226.     inflFormTranslations.put("???????", "uninflected-special-connection-2");
    227. 

  227.     inflFormTranslations.put("??????", "imperfective-reru-connection");
    228. 

  228.     inflFormTranslations.put("?????", "modern-base");
    229. 

  229.     inflFormTranslations.put("???-???", "base-onbin"); // not sure about this
    230. 

  230.   }
    231. 

  231.   
    232. 

  232.   /**
    233. 

  233.    * Get the english form of inflected form
    234. 

  234.    */
    235. 

  235.   public static String getInflectedFormTranslation(String s) {
    236. 

  236.     return inflFormTranslations.get(s);
    237. 

  237.   }
    238. 

  238.   
    239. 

  239.   /**
    240. 

  240.    * Romanize katakana with modified hepburn
    241. 

  241.    */
    242. 

  242.   public static String getRomanization(String s) {
    243. 

  243.     StringBuilder out = new StringBuilder();
    244. 

  244.     try {
    245. 

  245.       getRomanization(out, s);
    246. 

  246.     } catch (IOException bogus) {
    247. 

  247.       throw new RuntimeException(bogus);
    248. 

  248.     }
    249. 

  249.     return out.toString();
    250. 

  250.   }
    251. 

  251.   
    252. 

  252.   /**
    253. 

  253.    * Romanize katakana with modified hepburn
    254. 

  254.    */
    255. 

  255.   public static void getRomanization(Appendable builder, CharSequence s) throws IOException {
    256. 

  256.     final int len = s.length();
    257. 

  257.     for (int i = 0; i < len; i++) {
    258. 

  258.       // maximum lookahead: 3
    259. 

  259.       char ch = s.charAt(i);
    260. 

  260.       char ch2 = (i < len - 1) ? s.charAt(i + 1) : 0;
    261. 

  261.       char ch3 = (i < len - 2) ? s.charAt(i + 2) : 0;
    262. 

  262.       
    263. 

  263.       main: switch (ch) {
    264. 

  264.         case '?':
    265. 

  265.           switch (ch2) {
    266. 

  266.             case '?':
    267. 

  267.             case '?':
    268. 

  268.             case '?':
    269. 

  269.             case '?':
    270. 

  270.             case '?':
    271. 

  271.               builder.append('k');
    272. 

  272.               break main;
    273. 

  273.             case '?':
    274. 

  274.             case '?':
    275. 

  275.             case '?':
    276. 

  276.             case '?':
    277. 

  277.             case '?':
    278. 

  278.               builder.append('s');
    279. 

  279.               break main;
    280. 

  280.             case '?':
    281. 

  281.             case '?':
    282. 

  282.             case '?':
    283. 

  283.             case '?':
    284. 

  284.             case '?':
    285. 

  285.               builder.append('t');
    286. 

  286.               break main;
    287. 

  287.             case '?':
    288. 

  288.             case '?':
    289. 

  289.             case '?':
    290. 

  290.             case '?':
    291. 

  291.             case '?':
    292. 

  292.               builder.append('p');
    293. 

  293.               break main;
    294. 

  294.           }
    295. 

  295.           break;
    296. 

  296.         case '?':
    297. 

  297.           builder.append('a');
    298. 

  298.           break;
    299. 

  299.         case '?':
    300. 

  300.           if (ch2 == '?') {
    301. 

  301.             builder.append("yi");
    302. 

  302.             i++;
    303. 

  303.           } else if (ch2 == '?') {
    304. 

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

  305.             i++;
    306. 

  306.           } else {
    307. 

  307.             builder.append('i');
    308. 

  308.           }
    309. 

  309.           break;
    310. 

  310.         case '?':
    311. 

  311.           switch(ch2) {
    312. 

  312.             case '?':
    313. 

  313.               builder.append("wa");
    314. 

  314.               i++;
    315. 

  315.               break;
    316. 

  316.             case '?':
    317. 

  317.               builder.append("wi");
    318. 

  318.               i++;
    319. 

  319.               break;
    320. 

  320.             case '?':
    321. 

  321.               builder.append("wu");
    322. 

  322.               i++;
    323. 

  323.               break;
    324. 

  324.             case '?':
    325. 

  325.               builder.append("we");
    326. 

  326.               i++;
    327. 

  327.               break;
    328. 

  328.             case '?':
    329. 

  329.               builder.append("wo");
    330. 

  330.               i++;
    331. 

  331.               break;
    332. 

  332.             case '?':
    333. 

  333.               builder.append("wyu");
    334. 

  334.               i++;
    335. 

  335.               break;
    336. 

  336.             default:
    337. 

  337.               builder.append('u');
    338. 

  338.               break;
    339. 

  339.           }
    340. 

  340.           break;
    341. 

  341.         case '?':
    342. 

  342.           builder.append('e');
    343. 

  343.           break;
    344. 

  344.         case '?':
    345. 

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

  346.             builder.append('?');
    347. 

  347.             i++;
    348. 

  348.           } else {
    349. 

  349.             builder.append('o');
    350. 

  350.           }
    351. 

  351.           break;
    352. 

  352.         case '?':
    353. 

  353.           builder.append("ka");
    354. 

  354.           break;
    355. 

  355.         case '?':
    356. 

  356.           if (ch2 == '?' && ch3 == '?') {
    357. 

  357.             builder.append("ky?");
    358. 

  358.             i += 2;
    359. 

  359.           } else if (ch2 == '?' && ch3 == '?') {
    360. 

  360.             builder.append("ky?");
    361. 

  361.             i += 2;
    362. 

  362.           } else if (ch2 == '?') {
    363. 

  363.             builder.append("kya");
    364. 

  364.             i++;
    365. 

  365.           } else if (ch2 == '?') {
    366. 

  366.             builder.append("kyo");
    367. 

  367.             i++;
    368. 

  368.           } else if (ch2 == '?') {
    369. 

  369.             builder.append("kyu");
    370. 

  370.             i++;
    371. 

  371.           } else if (ch2 == '?') {
    372. 

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

  373.             i++;
    374. 

  374.           } else {
    375. 

  375.             builder.append("ki");
    376. 

  376.           }
    377. 

  377.           break;
    378. 

  378.         case '?':
    379. 

  379.           switch(ch2) {
    380. 

  380.             case '?':
    381. 

  381.               builder.append("kwa");
    382. 

  382.               i++;
    383. 

  383.               break;
    384. 

  384.             case '?':
    385. 

  385.               builder.append("kwi");
    386. 

  386.               i++;
    387. 

  387.               break;
    388. 

  388.             case '?':
    389. 

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

  390.               i++;
    391. 

  391.               break;
    392. 

  392.             case '?':
    393. 

  393.               builder.append("kwo");
    394. 

  394.               i++;
    395. 

  395.               break;
    396. 

  396.             case '?':
    397. 

  397.               builder.append("kwa");
    398. 

  398.               i++;
    399. 

  399.               break;
    400. 

  400.             default:
    401. 

  401.               builder.append("ku");
    402. 

  402.               break;
    403. 

  403.           }
    404. 

  404.           break;
    405. 

  405.         case '?':
    406. 

  406.           builder.append("ke");
    407. 

  407.           break;
    408. 

  408.         case '?':
    409. 

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

  410.             builder.append("k?");
    411. 

  411.             i++;
    412. 

  412.           } else {
    413. 

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

  414.           }
    415. 

  415.           break;
    416. 

  416.         case '?':
    417. 

  417.           builder.append("sa");
    418. 

  418.           break;
    419. 

  419.         case '?':
    420. 

  420.           if (ch2 == '?' && ch3 == '?') {
    421. 

  421.             builder.append("sh?");
    422. 

  422.             i += 2;
    423. 

  423.           } else if (ch2 == '?' && ch3 == '?') {
    424. 

  424.             builder.append("sh?");
    425. 

  425.             i += 2;
    426. 

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

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

  428.             i++;
    429. 

  429.           } else if (ch2 == '?') {
    430. 

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

  431.             i++;
    432. 

  432.           } else if (ch2 == '?') {
    433. 

  433.             builder.append("shu");
    434. 

  434.             i++;
    435. 

  435.           } else if (ch2 == '?') {
    436. 

  436.             builder.append("she");
    437. 

  437.             i++;
    438. 

  438.           } else {
    439. 

  439.             builder.append("shi");
    440. 

  440.           }
    441. 

  441.           break;
    442. 

  442.         case '?':
    443. 

  443.           if (ch2 == '?') {
    444. 

  444.             builder.append("si");
    445. 

  445.             i++;
    446. 

  446.           } else {
    447. 

  447.             builder.append("su");
    448. 

  448.           }
    449. 

  449.           break;
    450. 

  450.         case '?':
    451. 

  451.           builder.append("se");
    452. 

  452.           break;
    453. 

  453.         case '?':
    454. 

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

  455.             builder.append("s?");
    456. 

  456.             i++;
    457. 

  457.           } else {
    458. 

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

  459.           }
    460. 

  460.           break;
    461. 

  461.         case '?':
    462. 

  462.           builder.append("ta");
    463. 

  463.           break;
    464. 

  464.         case '?':
    465. 

  465.           if (ch2 == '?' && ch3 == '?') {
    466. 

  466.             builder.append("ch?");
    467. 

  467.             i += 2;
    468. 

  468.           } else if (ch2 == '?' && ch3 == '?') {
    469. 

  469.             builder.append("ch?");
    470. 

  470.             i += 2;
    471. 

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

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

  473.             i++;
    474. 

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

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

  476.             i++;
    477. 

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

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

  479.             i++;
    480. 

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

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

  482.             i++;
    483. 

  483.           } else {
    484. 

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

  485.           }
    486. 

  486.           break;
    487. 

  487.         case '?':
    488. 

  488.           if (ch2 == '?') {
    489. 

  489.             builder.append("tsa");
    490. 

  490.             i++;
    491. 

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

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

  493.             i++;
    494. 

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

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

  496.             i++;
    497. 

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

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

  499.             i++;
    500. 

  500.           } else if (ch2 == '?') {
    501. 

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

  502.             i++;
    503. 

  503.           } else {
    504. 

  504.             builder.append("tsu");
    505. 

  505.           }
    506. 

  506.           break;
    507. 

  507.         case '?':
    508. 

  508.           if (ch2 == '?') {
    509. 

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

  510.             i++;
    511. 

  511.           } else if (ch2 == '?') {
    512. 

  512.             builder.append("tu");
    513. 

  513.             i++;
    514. 

  514.           } else if (ch2 == '?') {
    515. 

  515.             builder.append("tyu");
    516. 

  516.             i++;
    517. 

  517.           } else {
    518. 

  518.             builder.append("te");
    519. 

  519.           }
    520. 

  520.           break;
    521. 

  521.         case '?':
    522. 

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

  523.             builder.append("t?");
    524. 

  524.             i++;
    525. 

  525.           } else {
    526. 

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

  527.           }
    528. 

  528.           break;
    529. 

  529.         case '?':
    530. 

  530.           builder.append("na");
    531. 

  531.           break;
    532. 

  532.         case '?':
    533. 

  533.           if (ch2 == '?' && ch3 == '?') {
    534. 

  534.             builder.append("ny?");
    535. 

  535.             i += 2;
    536. 

  536.           } else if (ch2 == '?' && ch3 == '?') {
    537. 

  537.             builder.append("ny?");
    538. 

  538.             i += 2;
    539. 

  539.           } else if (ch2 == '?') {
    540. 

  540.             builder.append("nya");
    541. 

  541.             i++;
    542. 

  542.           } else if (ch2 == '?') {
    543. 

  543.             builder.append("nyo");
    544. 

  544.             i++;
    545. 

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

  546.             builder.append("nyu");
    547. 

  547.             i++;
    548. 

  548.           } else if (ch2 == '?') {
    549. 

  549.             builder.append("nye");
    550. 

  550.             i++;
    551. 

  551.           } else {
    552. 

  552.             builder.append("ni");
    553. 

  553.           }
    554. 

  554.           break;
    555. 

  555.         case '?':
    556. 

  556.           builder.append("nu");
    557. 

  557.           break;
    558. 

  558.         case '?':
    559. 

  559.           builder.append("ne");
    560. 

  560.           break;
    561. 

  561.         case '?':
    562. 

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

  563.             builder.append("n?");
    564. 

  564.             i++;
    565. 

  565.           } else {
    566. 

  566.             builder.append("no");
    567. 

  567.           }
    568. 

  568.           break;
    569. 

  569.         case '?':
    570. 

  570.           builder.append("ha");
    571. 

  571.           break;
    572. 

  572.         case '?':
    573. 

  573.           if (ch2 == '?' && ch3 == '?') {
    574. 

  574.             builder.append("hy?");
    575. 

  575.             i += 2;
    576. 

  576.           } else if (ch2 == '?' && ch3 == '?') {
    577. 

  577.             builder.append("hy?");
    578. 

  578.             i += 2;
    579. 

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

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

  581.             i++;
    582. 

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

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

  584.             i++;
    585. 

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

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

  587.             i++;
    588. 

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

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

  590.             i++;
    591. 

  591.           } else {
    592. 

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

  593.           }
    594. 

  594.           break;
    595. 

  595.         case '?':
    596. 

  596.           if (ch2 == '?') {
    597. 

  597.             builder.append("fya");
    598. 

  598.             i++;
    599. 

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

  600.             builder.append("fyu");
    601. 

  601.             i++;
    602. 

  602.           } else if (ch2 == '?' && ch3 == '?') {
    603. 

  603.             builder.append("fye");
    604. 

  604.             i+=2;
    605. 

  605.           } else if (ch2 == '?') {
    606. 

  606.             builder.append("fyo");
    607. 

  607.             i++;
    608. 

  608.           } else if (ch2 == '?') {
    609. 

  609.             builder.append("fa");
    610. 

  610.             i++;
    611. 

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

  612.             builder.append("fi");
    613. 

  613.             i++;
    614. 

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

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

  616.             i++;
    617. 

  617.           } else if (ch2 == '?') {
    618. 

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

  619.             i++;
    620. 

  620.           } else {
    621. 

  621.             builder.append("fu");
    622. 

  622.           }
    623. 

  623.           break;
    624. 

  624.         case '?':
    625. 

  625.           builder.append("he");
    626. 

  626.           break;
    627. 

  627.         case '?':
    628. 

  628.           if (ch2 == '?') {
    629. 

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

  630.             i++;
    631. 

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

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

  633.             i++;
    634. 

  634.           } else {
    635. 

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

  636.           }
    637. 

  637.           break;
    638. 

  638.         case '?':
    639. 

  639.           builder.append("ma");
    640. 

  640.           break;
    641. 

  641.         case '?':
    642. 

  642.           if (ch2 == '?' && ch3 == '?') {
    643. 

  643.             builder.append("my?");
    644. 

  644.             i += 2;
    645. 

  645.           } else if (ch2 == '?' && ch3 == '?') {
    646. 

  646.             builder.append("my?");
    647. 

  647.             i += 2;
    648. 

  648.           } else if (ch2 == '?') {
    649. 

  649.             builder.append("mya");
    650. 

  650.             i++;
    651. 

  651.           } else if (ch2 == '?') {
    652. 

  652.             builder.append("myo");
    653. 

  653.             i++;
    654. 

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

  655.             builder.append("myu");
    656. 

  656.             i++;
    657. 

  657.           } else if (ch2 == '?') {
    658. 

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

  659.             i++;
    660. 

  660.           } else {
    661. 

  661.             builder.append("mi");
    662. 

  662.           }
    663. 

  663.           break;
    664. 

  664.         case '?':
    665. 

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

  666.           break;
    667. 

  667.         case '?':
    668. 

  668.           builder.append("mi");
    669. 

  669.           break;
    670. 

  670.         case '?':
    671. 

  671.           if (ch2 == '?') {
    672. 

  672.             builder.append("m?");
    673. 

  673.             i++;
    674. 

  674.           } else {
    675. 

  675.             builder.append("mo");
    676. 

  676.           }
    677. 

  677.           break;
    678. 

  678.         case '?':
    679. 

  679.           builder.append("ya");
    680. 

  680.           break;
    681. 

  681.         case '?':
    682. 

  682.           builder.append("yu");
    683. 

  683.           break;
    684. 

  684.         case '?':
    685. 

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

  686.             builder.append("y?");
    687. 

  687.             i++;
    688. 

  688.           } else {
    689. 

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

  690.           }
    691. 

  691.           break;
    692. 

  692.         case '?':
    693. 

  693.           builder.append("ra");
    694. 

  694.           break;
    695. 

  695.         case '?':
    696. 

  696.           if (ch2 == '?' && ch3 == '?') {
    697. 

  697.             builder.append("ry?");
    698. 

  698.             i += 2;
    699. 

  699.           } else if (ch2 == '?' && ch3 == '?') {
    700. 

  700.             builder.append("ry?");
    701. 

  701.             i += 2;
    702. 

  702.           } else if (ch2 == '?') {
    703. 

  703.             builder.append("rya");
    704. 

  704.             i++;
    705. 

  705.           } else if (ch2 == '?') {
    706. 

  706.             builder.append("ryo");
    707. 

  707.             i++;
    708. 

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

  709.             builder.append("ryu");
    710. 

  710.             i++;
    711. 

  711.           } else if (ch2 == '?') {
    712. 

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

  713.             i++;
    714. 

  714.           } else {
    715. 

  715.             builder.append("ri");
    716. 

  716.           }
    717. 

  717.           break;
    718. 

  718.         case '?':
    719. 

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

  720.           break;
    721. 

  721.         case '?':
    722. 

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

  723.           break;
    724. 

  724.         case '?':
    725. 

  725.           if (ch2 == '?') {
    726. 

  726.             builder.append("r?");
    727. 

  727.             i++;
    728. 

  728.           } else {
    729. 

  729.             builder.append("ro");
    730. 

  730.           }
    731. 

  731.           break;
    732. 

  732.         case '?':
    733. 

  733.           builder.append("wa");
    734. 

  734.           break;
    735. 

  735.         case '?':
    736. 

  736.           builder.append("i");
    737. 

  737.           break;
    738. 

  738.         case '?':
    739. 

  739.           builder.append("e");
    740. 

  740.           break;
    741. 

  741.         case '?':
    742. 

  742.           builder.append("o");
    743. 

  743.           break;
    744. 

  744.         case '?':
    745. 

  745.           switch (ch2) {
    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.             case '?':
    755. 

  755.             case '?':
    756. 

  756.             case '?':
    757. 

  757.             case '?':
    758. 

  758.             case '?':
    759. 

  759.             case '?':
    760. 

  760.             case '?':
    761. 

  761.               builder.append('m');
    762. 

  762.               break main;
    763. 

  763.             case '?':
    764. 

  764.             case '?':
    765. 

  765.             case '?':
    766. 

  766.             case '?':
    767. 

  767.             case '?':
    768. 

  768.             case '?':
    769. 

  769.             case '?':
    770. 

  770.             case '?':
    771. 

  771.               builder.append("n'");
    772. 

  772.               break main;
    773. 

  773.             default:
    774. 

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

  775.               break main;
    776. 

  776.           }
    777. 

  777.         case '?':
    778. 

  778.           builder.append("ga");
    779. 

  779.           break;
    780. 

  780.         case '?':
    781. 

  781.           if (ch2 == '?' && ch3 == '?') {
    782. 

  782.             builder.append("gy?");
    783. 

  783.             i += 2;
    784. 

  784.           } else if (ch2 == '?' && ch3 == '?') {
    785. 

  785.             builder.append("gy?");
    786. 

  786.             i += 2;
    787. 

  787.           } else if (ch2 == '?') {
    788. 

  788.             builder.append("gya");
    789. 

  789.             i++;
    790. 

  790.           } else if (ch2 == '?') {
    791. 

  791.             builder.append("gyo");
    792. 

  792.             i++;
    793. 

  793.           } else if (ch2 == '?') {
    794. 

  794.             builder.append("gyu");
    795. 

  795.             i++;
    796. 

  796.           } else if (ch2 == '?') {
    797. 

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

  798.             i++;
    799. 

  799.           } else {
    800. 

  800.             builder.append("gi");
    801. 

  801.           }
    802. 

  802.           break;
    803. 

  803.         case '?':
    804. 

  804.           switch(ch2) {
    805. 

  805.             case '?':
    806. 

  806.               builder.append("gwa");
    807. 

  807.               i++;
    808. 

  808.               break;
    809. 

  809.             case '?':
    810. 

  810.               builder.append("gwi");
    811. 

  811.               i++;
    812. 

  812.               break;
    813. 

  813.             case '?':
    814. 

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

  815.               i++;
    816. 

  816.               break;
    817. 

  817.             case '?':
    818. 

  818.               builder.append("gwo");
    819. 

  819.               i++;
    820. 

  820.               break;
    821. 

  821.             case '?':
    822. 

  822.               builder.append("gwa");
    823. 

  823.               i++;
    824. 

  824.               break;
    825. 

  825.             default:
    826. 

  826.               builder.append("gu");
    827. 

  827.               break;
    828. 

  828.           }
    829. 

  829.           break;
    830. 

  830.         case '?':
    831. 

  831.           builder.append("ge");
    832. 

  832.           break;
    833. 

  833.         case '?':
    834. 

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

  835.             builder.append("g?");
    836. 

  836.             i++;
    837. 

  837.           } else {
    838. 

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

  839.           }
    840. 

  840.           break;
    841. 

  841.         case '?':
    842. 

  842.           builder.append("za");
    843. 

  843.           break;
    844. 

  844.         case '?':
    845. 

  845.           if (ch2 == '?' && ch3 == '?') {
    846. 

  846.             builder.append("j?");
    847. 

  847.             i += 2;
    848. 

  848.           } else if (ch2 == '?' && ch3 == '?') {
    849. 

  849.             builder.append("j?");
    850. 

  850.             i += 2;
    851. 

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

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

  853.             i++;
    854. 

  854.           } else if (ch2 == '?') {
    855. 

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

  856.             i++;
    857. 

  857.           } else if (ch2 == '?') {
    858. 

  858.             builder.append("ju");
    859. 

  859.             i++;
    860. 

  860.           } else if (ch2 == '?') {
    861. 

  861.             builder.append("je");
    862. 

  862.             i++;
    863. 

  863.           } else {
    864. 

  864.             builder.append("ji");
    865. 

  865.           }
    866. 

  866.           break;
    867. 

  867.         case '?':
    868. 

  868.           if (ch2 == '?') {
    869. 

  869.             builder.append("zi");
    870. 

  870.             i++;
    871. 

  871.           } else {
    872. 

  872.             builder.append("zu");
    873. 

  873.           }
    874. 

  874.           break;
    875. 

  875.         case '?':
    876. 

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

  877.           break;
    878. 

  878.         case '?':
    879. 

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

  880.             builder.append("z?");
    881. 

  881.             i++;
    882. 

  882.           } else {
    883. 

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

  884.           }
    885. 

  885.           break;
    886. 

  886.         case '?':
    887. 

  887.           builder.append("da");
    888. 

  888.           break;
    889. 

  889.         case '?':
    890. 

  890.           builder.append("ji");
    891. 

  891.           break;
    892. 

  892.         case '?':
    893. 

  893.           builder.append("zu");
    894. 

  894.           break;
    895. 

  895.         case '?':
    896. 

  896.           if (ch2 == '?') {
    897. 

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

  898.             i++;
    899. 

  899.           } else if (ch2 == '?') {
    900. 

  900.             builder.append("dyu");
    901. 

  901.             i++;
    902. 

  902.           } else {
    903. 

  903.             builder.append("de");
    904. 

  904.           }
    905. 

  905.           break;
    906. 

  906.         case '?':
    907. 

  907.           if (ch2 == '?') {
    908. 

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

  909.             i++;
    910. 

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

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

  912.             i++;
    913. 

  913.           } else {
    914. 

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

  915.           }
    916. 

  916.           break;
    917. 

  917.         case '?':
    918. 

  918.           builder.append("ba");
    919. 

  919.           break;
    920. 

  920.         case '?':
    921. 

  921.           if (ch2 == '?' && ch3 == '?') {
    922. 

  922.             builder.append("by?");
    923. 

  923.             i += 2;
    924. 

  924.           } else if (ch2 == '?' && ch3 == '?') {
    925. 

  925.             builder.append("by?");
    926. 

  926.             i += 2;
    927. 

  927.           } else if (ch2 == '?') {
    928. 

  928.             builder.append("bya");
    929. 

  929.             i++;
    930. 

  930.           } else if (ch2 == '?') {
    931. 

  931.             builder.append("byo");
    932. 

  932.             i++;
    933. 

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

  934.             builder.append("byu");
    935. 

  935.             i++;
    936. 

  936.           } else if (ch2 == '?') {
    937. 

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

  938.             i++;
    939. 

  939.           } else {
    940. 

  940.             builder.append("bi");
    941. 

  941.           }
    942. 

  942.           break;
    943. 

  943.         case '?':
    944. 

  944.           builder.append("bu");
    945. 

  945.           break;
    946. 

  946.         case '?':
    947. 

  947.           builder.append("be");
    948. 

  948.           break;
    949. 

  949.         case '?':
    950. 

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

  951.             builder.append("b?");
    952. 

  952.             i++;
    953. 

  953.           } else {
    954. 

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

  955.           }
    956. 

  956.           break;
    957. 

  957.         case '?':
    958. 

  958.           builder.append("pa");
    959. 

  959.           break;
    960. 

  960.         case '?':
    961. 

  961.           if (ch2 == '?' && ch3 == '?') {
    962. 

  962.             builder.append("py?");
    963. 

  963.             i += 2;
    964. 

  964.           } else if (ch2 == '?' && ch3 == '?') {
    965. 

  965.             builder.append("py?");
    966. 

  966.             i += 2;
    967. 

  967.           } else if (ch2 == '?') {
    968. 

  968.             builder.append("pya");
    969. 

  969.             i++;
    970. 

  970.           } else if (ch2 == '?') {
    971. 

  971.             builder.append("pyo");
    972. 

  972.             i++;
    973. 

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

  974.             builder.append("pyu");
    975. 

  975.             i++;
    976. 

  976.           } else if (ch2 == '?') {
    977. 

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

  978.             i++;
    979. 

  979.           } else {
    980. 

  980.             builder.append("pi");
    981. 

  981.           }
    982. 

  982.           break;
    983. 

  983.         case '?':
    984. 

  984.           builder.append("pu");
    985. 

  985.           break;
    986. 

  986.         case '?':
    987. 

  987.           builder.append("pe");
    988. 

  988.           break;
    989. 

  989.         case '?':
    990. 

  990.           if (ch2 == '?') {
    991. 

  991.             builder.append("p?");
    992. 

  992.             i++;
    993. 

  993.           } else {
    994. 

  994.             builder.append("po");
    995. 

  995.           }
    996. 

  996.           break;
    997. 

  997.         case '?':
    998. 

  998.           if (ch2 == '?' && ch3 == '?') {
    999. 

  999.             builder.append("vye");
    1000. 

  1000.             i+= 2;
    1001. 

  1001.           } else {
    1002. 

  1002.             builder.append('v');
    1003. 

  1003.           }
    1004. 

  1004.           break;
    1005. 

  1005.         case '?':
    1006. 

  1006.           builder.append('a');
    1007. 

  1007.           break;
    1008. 

  1008.         case '?':
    1009. 

  1009.           builder.append('i');
    1010. 

  1010.           break;
    1011. 

  1011.         case '?':
    1012. 

  1012.           builder.append('u');
    1013. 

  1013.           break;
    1014. 

  1014.         case '?':
    1015. 

  1015.           builder.append('e');
    1016. 

  1016.           break;
    1017. 

  1017.         case '?':
    1018. 

  1018.           builder.append('o');
    1019. 

  1019.           break;
    1020. 

  1020.         case '?':
    1021. 

  1021.           builder.append("wa");
    1022. 

  1022.           break;
    1023. 

  1023.         case '?':
    1024. 

  1024.           builder.append("ya");
    1025. 

  1025.           break;
    1026. 

  1026.         case '?':
    1027. 

  1027.           builder.append("yu");
    1028. 

  1028.           break;
    1029. 

  1029.         case '?':
    1030. 

  1030.           builder.append("yo");
    1031. 

  1031.           break;
    1032. 

  1032.         case '?':
    1033. 

  1033.           break;
    1034. 

  1034.         default:
    1035. 

  1035.           builder.append(ch);
    1036. 

  1036.       }
    1037. 

  1037.     }
    1038. 

  1038.   }
    1039. 

  1039. }
    1040. 

  1040.