1 /* 2 * Copyright (C) 2008-2012 OMRON SOFTWARE Co., Ltd. 3 * 4 * Licensed under the Apache License, Version 2.0 (the "License"); 5 * you may not use this file except in compliance with the License. 6 * You may obtain a copy of the License at 7 * 8 * http://www.apache.org/licenses/LICENSE-2.0 9 * 10 * Unless required by applicable law or agreed to in writing, software 11 * distributed under the License is distributed on an "AS IS" BASIS, 12 * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. 13 * See the License for the specific language governing permissions and 14 * limitations under the License. 15 */ 16 17 package jp.co.omronsoft.openwnn.JAJP; 18 19 import java.text.DecimalFormat; 20 import java.util.ArrayList; 21 import java.util.HashMap; 22 import java.util.List; 23 import java.util.Locale; 24 import jp.co.omronsoft.openwnn.WnnDictionary; 25 import jp.co.omronsoft.openwnn.WnnPOS; 26 import jp.co.omronsoft.openwnn.WnnWord; 27 28 /** 29 * The EISU-KANA converter class for Japanese IME. 30 * 31 * @author Copyright (C) 2009 OMRON SOFTWARE CO., LTD. All Rights Reserved. 32 */ 33 public class KanaConverter { 34 35 /** Conversion rule for half-width numeric */ 36 private static final HashMap<String,String> mHalfNumericMap = new HashMap<String,String>() {{ 37 put( "\u3042", "1"); 38 put( "\u3044", "11"); 39 put( "\u3046", "111"); 40 put( "\u3048", "1111"); 41 put( "\u304a", "11111"); 42 put( "\u3041", "111111"); 43 put( "\u3043", "1111111"); 44 put( "\u3045", "11111111"); 45 put( "\u3047", "111111111"); 46 put( "\u3049", "1111111111"); 47 put( "\u304b", "2"); 48 put( "\u304d", "22"); 49 put( "\u304f", "222"); 50 put( "\u3051", "2222"); 51 put( "\u3053", "22222"); 52 put( "\u3055", "3"); 53 put( "\u3057", "33"); 54 put( "\u3059", "333"); 55 put( "\u305b", "3333"); 56 put( "\u305d", "33333"); 57 put( "\u305f", "4"); 58 put( "\u3061", "44"); 59 put( "\u3064", "444"); 60 put( "\u3066", "4444"); 61 put( "\u3068", "44444"); 62 put( "\u3063", "444444"); 63 put( "\u306a", "5"); 64 put( "\u306b", "55"); 65 put( "\u306c", "555"); 66 put( "\u306d", "5555"); 67 put( "\u306e", "55555"); 68 put( "\u306f", "6"); 69 put( "\u3072", "66"); 70 put( "\u3075", "666"); 71 put( "\u3078", "6666"); 72 put( "\u307b", "66666"); 73 put( "\u307e", "7"); 74 put( "\u307f", "77"); 75 put( "\u3080", "777"); 76 put( "\u3081", "7777"); 77 put( "\u3082", "77777"); 78 put( "\u3084", "8"); 79 put( "\u3086", "88"); 80 put( "\u3088", "888"); 81 put( "\u3083", "8888"); 82 put( "\u3085", "88888"); 83 put( "\u3087", "888888"); 84 put( "\u3089", "9"); 85 put( "\u308a", "99"); 86 put( "\u308b", "999"); 87 put( "\u308c", "9999"); 88 put( "\u308d", "99999"); 89 put( "\u308f", "0"); 90 put( "\u3092", "00"); 91 put( "\u3093", "000"); 92 put( "\u308e", "0000"); 93 put( "\u30fc", "00000"); 94 }}; 95 96 /** Conversion rule for full-width numeric */ 97 private static final HashMap<String,String> mFullNumericMap = new HashMap<String,String>() {{ 98 put( "\u3042", "\uff11"); 99 put( "\u3044", "\uff11\uff11"); 100 put( "\u3046", "\uff11\uff11\uff11"); 101 put( "\u3048", "\uff11\uff11\uff11\uff11"); 102 put( "\u304a", "\uff11\uff11\uff11\uff11\uff11"); 103 put( "\u3041", "\uff11\uff11\uff11\uff11\uff11\uff11"); 104 put( "\u3043", "\uff11\uff11\uff11\uff11\uff11\uff11\uff11"); 105 put( "\u3045", "\uff11\uff11\uff11\uff11\uff11\uff11\uff11\uff11"); 106 put( "\u3047", "\uff11\uff11\uff11\uff11\uff11\uff11\uff11\uff11\uff11"); 107 put( "\u3049", "\uff11\uff11\uff11\uff11\uff11\uff11\uff11\uff11\uff11\uff11"); 108 put( "\u304b", "\uff12"); 109 put( "\u304d", "\uff12\uff12"); 110 put( "\u304f", "\uff12\uff12\uff12"); 111 put( "\u3051", "\uff12\uff12\uff12\uff12"); 112 put( "\u3053", "\uff12\uff12\uff12\uff12\uff12"); 113 put( "\u3055", "\uff13"); 114 put( "\u3057", "\uff13\uff13"); 115 put( "\u3059", "\uff13\uff13\uff13"); 116 put( "\u305b", "\uff13\uff13\uff13\uff13"); 117 put( "\u305d", "\uff13\uff13\uff13\uff13\uff13"); 118 put( "\u305f", "\uff14"); 119 put( "\u3061", "\uff14\uff14"); 120 put( "\u3064", "\uff14\uff14\uff14"); 121 put( "\u3066", "\uff14\uff14\uff14\uff14"); 122 put( "\u3068", "\uff14\uff14\uff14\uff14\uff14"); 123 put( "\u3063", "\uff14\uff14\uff14\uff14\uff14\uff14"); 124 put( "\u306a", "\uff15"); 125 put( "\u306b", "\uff15\uff15"); 126 put( "\u306c", "\uff15\uff15\uff15"); 127 put( "\u306d", "\uff15\uff15\uff15\uff15"); 128 put( "\u306e", "\uff15\uff15\uff15\uff15\uff15"); 129 put( "\u306f", "\uff16"); 130 put( "\u3072", "\uff16\uff16"); 131 put( "\u3075", "\uff16\uff16\uff16"); 132 put( "\u3078", "\uff16\uff16\uff16\uff16"); 133 put( "\u307b", "\uff16\uff16\uff16\uff16\uff16"); 134 put( "\u307e", "\uff17"); 135 put( "\u307f", "\uff17\uff17"); 136 put( "\u3080", "\uff17\uff17\uff17"); 137 put( "\u3081", "\uff17\uff17\uff17\uff17"); 138 put( "\u3082", "\uff17\uff17\uff17\uff17\uff17"); 139 put( "\u3084", "\uff18"); 140 put( "\u3086", "\uff18\uff18"); 141 put( "\u3088", "\uff18\uff18\uff18"); 142 put( "\u3083", "\uff18\uff18\uff18\uff18"); 143 put( "\u3085", "\uff18\uff18\uff18\uff18\uff18"); 144 put( "\u3087", "\uff18\uff18\uff18\uff18\uff18\uff18"); 145 put( "\u3089", "\uff19"); 146 put( "\u308a", "\uff19\uff19"); 147 put( "\u308b", "\uff19\uff19\uff19"); 148 put( "\u308c", "\uff19\uff19\uff19\uff19"); 149 put( "\u308d", "\uff19\uff19\uff19\uff19\uff19"); 150 put( "\u308f", "\uff10"); 151 put( "\u3092", "\uff10\uff10"); 152 put( "\u3093", "\uff10\uff10\uff10"); 153 put( "\u308e", "\uff10\uff10\uff10\uff10"); 154 put( "\u30fc", "\uff10\uff10\uff10\uff10\uff10"); 155 }}; 156 157 /** Conversion rule for half-width Katakana */ 158 private static final HashMap<String,String> mHalfKatakanaMap = new HashMap<String,String>() {{ 159 put( "\u3042", "\uff71"); 160 put( "\u3044", "\uff72"); 161 put( "\u3046", "\uff73"); 162 put( "\u3048", "\uff74"); 163 put( "\u304a", "\uff75"); 164 put( "\u3041", "\uff67"); 165 put( "\u3043", "\uff68"); 166 put( "\u3045", "\uff69"); 167 put( "\u3047", "\uff6a"); 168 put( "\u3049", "\uff6b"); 169 put( "\u30f4\u3041", "\uff73\uff9e\uff67"); 170 put( "\u30f4\u3043", "\uff73\uff9e\uff68"); 171 put( "\u30f4", "\uff73\uff9e"); 172 put( "\u30f4\u3047", "\uff73\uff9e\uff6a"); 173 put( "\u30f4\u3049", "\uff73\uff9e\uff6b"); 174 put( "\u304b", "\uff76"); 175 put( "\u304d", "\uff77"); 176 put( "\u304f", "\uff78"); 177 put( "\u3051", "\uff79"); 178 put( "\u3053", "\uff7a"); 179 put( "\u304c", "\uff76\uff9e"); 180 put( "\u304e", "\uff77\uff9e"); 181 put( "\u3050", "\uff78\uff9e"); 182 put( "\u3052", "\uff79\uff9e"); 183 put( "\u3054", "\uff7a\uff9e"); 184 put( "\u3055", "\uff7b"); 185 put( "\u3057", "\uff7c"); 186 put( "\u3059", "\uff7d"); 187 put( "\u305b", "\uff7e"); 188 put( "\u305d", "\uff7f"); 189 put( "\u3056", "\uff7b\uff9e"); 190 put( "\u3058", "\uff7c\uff9e"); 191 put( "\u305a", "\uff7d\uff9e"); 192 put( "\u305c", "\uff7e\uff9e"); 193 put( "\u305e", "\uff7f\uff9e"); 194 put( "\u305f", "\uff80"); 195 put( "\u3061", "\uff81"); 196 put( "\u3064", "\uff82"); 197 put( "\u3066", "\uff83"); 198 put( "\u3068", "\uff84"); 199 put( "\u3063", "\uff6f"); 200 put( "\u3060", "\uff80\uff9e"); 201 put( "\u3062", "\uff81\uff9e"); 202 put( "\u3065", "\uff82\uff9e"); 203 put( "\u3067", "\uff83\uff9e"); 204 put( "\u3069", "\uff84\uff9e"); 205 put( "\u306a", "\uff85"); 206 put( "\u306b", "\uff86"); 207 put( "\u306c", "\uff87"); 208 put( "\u306d", "\uff88"); 209 put( "\u306e", "\uff89"); 210 put( "\u306f", "\uff8a"); 211 put( "\u3072", "\uff8b"); 212 put( "\u3075", "\uff8c"); 213 put( "\u3078", "\uff8d"); 214 put( "\u307b", "\uff8e"); 215 put( "\u3070", "\uff8a\uff9e"); 216 put( "\u3073", "\uff8b\uff9e"); 217 put( "\u3076", "\uff8c\uff9e"); 218 put( "\u3079", "\uff8d\uff9e"); 219 put( "\u307c", "\uff8e\uff9e"); 220 put( "\u3071", "\uff8a\uff9f"); 221 put( "\u3074", "\uff8b\uff9f"); 222 put( "\u3077", "\uff8c\uff9f"); 223 put( "\u307a", "\uff8d\uff9f"); 224 put( "\u307d", "\uff8e\uff9f"); 225 put( "\u307e", "\uff8f"); 226 put( "\u307f", "\uff90"); 227 put( "\u3080", "\uff91"); 228 put( "\u3081", "\uff92"); 229 put( "\u3082", "\uff93"); 230 put( "\u3084", "\uff94"); 231 put( "\u3086", "\uff95"); 232 put( "\u3088", "\uff96"); 233 put( "\u3083", "\uff6c"); 234 put( "\u3085", "\uff6d"); 235 put( "\u3087", "\uff6e"); 236 put( "\u3089", "\uff97"); 237 put( "\u308a", "\uff98"); 238 put( "\u308b", "\uff99"); 239 put( "\u308c", "\uff9a"); 240 put( "\u308d", "\uff9b"); 241 put( "\u308f", "\uff9c"); 242 put( "\u3092", "\uff66"); 243 put( "\u3093", "\uff9d"); 244 put( "\u308e", "\uff9c"); 245 put( "\u30fc", "\uff70"); 246 }}; 247 248 /** Conversion rule for full-width Katakana */ 249 private static final HashMap<String,String> mFullKatakanaMap = new HashMap<String,String>() {{ 250 put( "\u3042", "\u30a2"); 251 put( "\u3044", "\u30a4"); 252 put( "\u3046", "\u30a6"); 253 put( "\u3048", "\u30a8"); 254 put( "\u304a", "\u30aa"); 255 put( "\u3041", "\u30a1"); 256 put( "\u3043", "\u30a3"); 257 put( "\u3045", "\u30a5"); 258 put( "\u3047", "\u30a7"); 259 put( "\u3049", "\u30a9"); 260 put( "\u30f4\u3041", "\u30f4\u30a1"); 261 put( "\u30f4\u3043", "\u30f4\u30a3"); 262 put( "\u30f4", "\u30f4"); 263 put( "\u30f4\u3047", "\u30f4\u30a7"); 264 put( "\u30f4\u3049", "\u30f4\u30a9"); 265 put( "\u304b", "\u30ab"); 266 put( "\u304d", "\u30ad"); 267 put( "\u304f", "\u30af"); 268 put( "\u3051", "\u30b1"); 269 put( "\u3053", "\u30b3"); 270 put( "\u304c", "\u30ac"); 271 put( "\u304e", "\u30ae"); 272 put( "\u3050", "\u30b0"); 273 put( "\u3052", "\u30b2"); 274 put( "\u3054", "\u30b4"); 275 put( "\u3055", "\u30b5"); 276 put( "\u3057", "\u30b7"); 277 put( "\u3059", "\u30b9"); 278 put( "\u305b", "\u30bb"); 279 put( "\u305d", "\u30bd"); 280 put( "\u3056", "\u30b6"); 281 put( "\u3058", "\u30b8"); 282 put( "\u305a", "\u30ba"); 283 put( "\u305c", "\u30bc"); 284 put( "\u305e", "\u30be"); 285 put( "\u305f", "\u30bf"); 286 put( "\u3061", "\u30c1"); 287 put( "\u3064", "\u30c4"); 288 put( "\u3066", "\u30c6"); 289 put( "\u3068", "\u30c8"); 290 put( "\u3063", "\u30c3"); 291 put( "\u3060", "\u30c0"); 292 put( "\u3062", "\u30c2"); 293 put( "\u3065", "\u30c5"); 294 put( "\u3067", "\u30c7"); 295 put( "\u3069", "\u30c9"); 296 put( "\u306a", "\u30ca"); 297 put( "\u306b", "\u30cb"); 298 put( "\u306c", "\u30cc"); 299 put( "\u306d", "\u30cd"); 300 put( "\u306e", "\u30ce"); 301 put( "\u306f", "\u30cf"); 302 put( "\u3072", "\u30d2"); 303 put( "\u3075", "\u30d5"); 304 put( "\u3078", "\u30d8"); 305 put( "\u307b", "\u30db"); 306 put( "\u3070", "\u30d0"); 307 put( "\u3073", "\u30d3"); 308 put( "\u3076", "\u30d6"); 309 put( "\u3079", "\u30d9"); 310 put( "\u307c", "\u30dc"); 311 put( "\u3071", "\u30d1"); 312 put( "\u3074", "\u30d4"); 313 put( "\u3077", "\u30d7"); 314 put( "\u307a", "\u30da"); 315 put( "\u307d", "\u30dd"); 316 put( "\u307e", "\u30de"); 317 put( "\u307f", "\u30df"); 318 put( "\u3080", "\u30e0"); 319 put( "\u3081", "\u30e1"); 320 put( "\u3082", "\u30e2"); 321 put( "\u3084", "\u30e4"); 322 put( "\u3086", "\u30e6"); 323 put( "\u3088", "\u30e8"); 324 put( "\u3083", "\u30e3"); 325 put( "\u3085", "\u30e5"); 326 put( "\u3087", "\u30e7"); 327 put( "\u3089", "\u30e9"); 328 put( "\u308a", "\u30ea"); 329 put( "\u308b", "\u30eb"); 330 put( "\u308c", "\u30ec"); 331 put( "\u308d", "\u30ed"); 332 put( "\u308f", "\u30ef"); 333 put( "\u3092", "\u30f2"); 334 put( "\u3093", "\u30f3"); 335 put( "\u308e", "\u30ee"); 336 put( "\u30fc", "\u30fc"); 337 }}; 338 339 /** Conversion rule for half-width alphabet */ 340 private static final HashMap<String,String> mHalfAlphabetMap = new HashMap<String,String>() {{ 341 put( "\u3042", "."); 342 put( "\u3044", "@"); 343 put( "\u3046", "-"); 344 put( "\u3048", "_"); 345 put( "\u304a", "/"); 346 put( "\u3041", ":"); 347 put( "\u3043", "~"); 348 put( "\u304b", "A"); 349 put( "\u304d", "B"); 350 put( "\u304f", "C"); 351 put( "\u3055", "D"); 352 put( "\u3057", "E"); 353 put( "\u3059", "F"); 354 put( "\u305f", "G"); 355 put( "\u3061", "H"); 356 put( "\u3064", "I"); 357 put( "\u306a", "J"); 358 put( "\u306b", "K"); 359 put( "\u306c", "L"); 360 put( "\u306f", "M"); 361 put( "\u3072", "N"); 362 put( "\u3075", "O"); 363 put( "\u307e", "P"); 364 put( "\u307f", "Q"); 365 put( "\u3080", "R"); 366 put( "\u3081", "S"); 367 put( "\u3084", "T"); 368 put( "\u3086", "U"); 369 put( "\u3088", "V"); 370 put( "\u3089", "W"); 371 put( "\u308a", "X"); 372 put( "\u308b", "Y"); 373 put( "\u308c", "Z"); 374 put( "\u308f", "-"); 375 }}; 376 377 /** Conversion rule for full-width alphabet */ 378 private static final HashMap<String,String> mFullAlphabetMap = new HashMap<String,String>() {{ 379 put( "\u3042", "\uff0e"); 380 put( "\u3044", "\uff20"); 381 put( "\u3046", "\u30fc"); 382 put( "\u3048", "\uff3f"); 383 put( "\u304a", "\uff0f"); 384 put( "\u3041", "\uff1a"); 385 put( "\u3043", "\u301c"); 386 put( "\u304b", "\uff21"); 387 put( "\u304d", "\uff22" ); 388 put( "\u304f", "\uff23"); 389 put( "\u3055", "\uff24" ); 390 put( "\u3057", "\uff25" ); 391 put( "\u3059", "\uff26" ); 392 put( "\u305f", "\uff27"); 393 put( "\u3061", "\uff28" ); 394 put( "\u3064", "\uff29"); 395 put( "\u306a", "\uff2a"); 396 put( "\u306b", "\uff2b" ); 397 put( "\u306c", "\uff2c" ); 398 put( "\u306f", "\uff2d"); 399 put( "\u3072", "\uff2e"); 400 put( "\u3075", "\uff2f"); 401 put( "\u307e", "\uff30"); 402 put( "\u307f", "\uff31"); 403 put( "\u3080", "\uff32"); 404 put( "\u3081", "\uff33" ); 405 put( "\u3084", "\uff34" ); 406 put( "\u3086", "\uff35" ); 407 put( "\u3088", "\uff36" ); 408 put( "\u3089", "\uff37" ); 409 put( "\u308a", "\uff38" ); 410 put( "\u308b", "\uff39"); 411 put( "\u308c", "\uff3a" ); 412 put( "\u308f", "\u30fc" ); 413 }}; 414 415 /** Conversion rule for full-width alphabet (QWERTY mode) */ 416 private static final HashMap<String,String> mFullAlphabetMapQwety = new HashMap<String,String>() {{ 417 put( "a", "\uff41"); 418 put( "b", "\uff42"); 419 put( "c", "\uff43"); 420 put( "d", "\uff44"); 421 put( "e", "\uff45"); 422 put( "f", "\uff46"); 423 put( "g", "\uff47"); 424 put( "h", "\uff48"); 425 put( "i", "\uff49"); 426 put( "j", "\uff4a"); 427 put( "k", "\uff4b"); 428 put( "l", "\uff4c"); 429 put( "m", "\uff4d"); 430 put( "n", "\uff4e"); 431 put( "o", "\uff4f"); 432 put( "p", "\uff50"); 433 put( "q", "\uff51"); 434 put( "r", "\uff52"); 435 put( "s", "\uff53"); 436 put( "t", "\uff54"); 437 put( "u", "\uff55"); 438 put( "v", "\uff56"); 439 put( "w", "\uff57"); 440 put( "x", "\uff58"); 441 put( "y", "\uff59"); 442 put( "z", "\uff5a"); 443 444 put( "A", "\uff21"); 445 put( "B", "\uff22"); 446 put( "C", "\uff23"); 447 put( "D", "\uff24"); 448 put( "E", "\uff25"); 449 put( "F", "\uff26"); 450 put( "G", "\uff27"); 451 put( "H", "\uff28"); 452 put( "I", "\uff29"); 453 put( "J", "\uff2a"); 454 put( "K", "\uff2b"); 455 put( "L", "\uff2c"); 456 put( "M", "\uff2d"); 457 put( "N", "\uff2e"); 458 put( "O", "\uff2f"); 459 put( "P", "\uff30"); 460 put( "Q", "\uff31"); 461 put( "R", "\uff32"); 462 put( "S", "\uff33"); 463 put( "T", "\uff34"); 464 put( "U", "\uff35"); 465 put( "V", "\uff36"); 466 put( "W", "\uff37"); 467 put( "X", "\uff38"); 468 put( "Y", "\uff39"); 469 put( "Z", "\uff3a"); 470 }}; 471 472 /** Decimal format using comma */ 473 private static final DecimalFormat mFormat = new DecimalFormat("###,###"); 474 475 /** List of the generated candidates */ 476 private List<WnnWord> mAddCandidateList; 477 /** Work area for generating string */ 478 private StringBuffer mStringBuff; 479 480 /** part of speech (default) */ 481 private WnnPOS mPosDefault; 482 /** part of speech (number) */ 483 private WnnPOS mPosNumber; 484 /** part of speech (symbol) */ 485 private WnnPOS mPosSymbol; 486 487 /** 488 * Constructor 489 */ KanaConverter()490 public KanaConverter() { 491 mAddCandidateList = new ArrayList<WnnWord>(); 492 mStringBuff = new StringBuffer(); 493 } 494 495 /** 496 * Set The dictionary. 497 * <br> 498 * {@link KanaConverter} gets part-of-speech tags from the dictionary. 499 * 500 * @param dict The dictionary 501 */ setDictionary(WnnDictionary dict)502 public void setDictionary(WnnDictionary dict) { 503 /* get part of speech tags */ 504 mPosDefault = dict.getPOS(WnnDictionary.POS_TYPE_MEISI); 505 mPosNumber = dict.getPOS(WnnDictionary.POS_TYPE_SUUJI); 506 mPosSymbol = dict.getPOS(WnnDictionary.POS_TYPE_KIGOU); 507 } 508 509 /** 510 * Create the pseudo candidate list 511 * <br> 512 * @param inputHiragana The input string (Hiragana) 513 * @param inputRomaji The input string (Romaji) 514 * @param keyBoardMode The mode of keyboard 515 * @return The candidate list 516 */ createPseudoCandidateList(String inputHiragana, String inputRomaji, int keyBoardMode)517 public List<WnnWord> createPseudoCandidateList(String inputHiragana, String inputRomaji, int keyBoardMode) { 518 List<WnnWord> list = mAddCandidateList; 519 520 list.clear(); 521 if (inputHiragana.length() == 0) { 522 return list; 523 } 524 525 /* Create pseudo candidates for all keyboard type */ 526 /* Hiragana(reading) / Full width katakana / Half width katakana */ 527 list.add(new WnnWord(inputHiragana, inputHiragana)); 528 if (createCandidateString(inputHiragana, mFullKatakanaMap, mStringBuff)) { 529 list.add(new WnnWord(mStringBuff.toString(), inputHiragana, mPosDefault)); 530 } 531 if (createCandidateString(inputHiragana, mHalfKatakanaMap, mStringBuff)) { 532 list.add(new WnnWord(mStringBuff.toString(), inputHiragana, mPosDefault)); 533 } 534 535 if (keyBoardMode == OpenWnnEngineJAJP.KEYBOARD_QWERTY) { 536 /* Create pseudo candidates for Qwerty keyboard */ 537 createPseudoCandidateListForQwerty(inputHiragana, inputRomaji); 538 } else { 539 /* Create pseudo candidates for 12key */ 540 541 /* Create pseudo candidates for half width numeric */ 542 if (createCandidateString(inputHiragana, mHalfNumericMap, mStringBuff)) { 543 String convHanSuuji = mStringBuff.toString(); 544 String convNumComma = convertNumber(convHanSuuji); 545 list.add(new WnnWord(convHanSuuji, inputHiragana, mPosNumber)); 546 if (convNumComma != null) { 547 list.add(new WnnWord(convNumComma, inputHiragana, mPosNumber)); 548 } 549 } 550 551 /* Create pseudo candidates for full width numeric */ 552 if (createCandidateString(inputHiragana, mFullNumericMap, mStringBuff)) { 553 list.add(new WnnWord(mStringBuff.toString(), inputHiragana, mPosNumber)); 554 } 555 556 /* Create pseudo candidates for half width alphabet */ 557 if (createCandidateString(inputHiragana, mHalfAlphabetMap, mStringBuff)) { 558 String convHanEiji = mStringBuff.toString(); 559 String convHanEijiLower = convHanEiji.toLowerCase(); 560 list.add(new WnnWord(convHanEijiLower, inputHiragana, mPosSymbol)); 561 list.add(new WnnWord(convertCaps(convHanEijiLower), inputHiragana, mPosSymbol)); 562 list.add(new WnnWord(convHanEiji, inputHiragana, mPosSymbol)); 563 } 564 565 /* Create pseudo candidates for full width alphabet */ 566 if (createCandidateString(inputHiragana, mFullAlphabetMap, mStringBuff)) { 567 String convZenEiji = mStringBuff.toString(); 568 String convZenEijiLower = convZenEiji.toLowerCase(Locale.JAPAN); 569 list.add(new WnnWord(convZenEijiLower, inputHiragana, mPosSymbol)); 570 list.add(new WnnWord(convertCaps(convZenEijiLower), inputHiragana, mPosSymbol)); 571 list.add(new WnnWord(convZenEiji, inputHiragana, mPosSymbol)); 572 } 573 } 574 return list; 575 } 576 577 /** 578 * Create the pseudo candidate list for Qwerty keyboard 579 * <br> 580 * @param inputHiragana The input string (Hiragana) 581 * @param inputRomaji The input string (Romaji) 582 */ createPseudoCandidateListForQwerty(String inputHiragana, String inputRomaji)583 private void createPseudoCandidateListForQwerty(String inputHiragana, String inputRomaji) { 584 List<WnnWord> list = mAddCandidateList; 585 586 /* Create pseudo candidates for half width alphabet */ 587 String convHanEijiLower = inputRomaji.toLowerCase(); 588 list.add(new WnnWord(inputRomaji, inputHiragana, mPosDefault)); 589 list.add(new WnnWord(convHanEijiLower, inputHiragana, mPosSymbol)); 590 list.add(new WnnWord(convertCaps(convHanEijiLower), inputHiragana, mPosSymbol)); 591 list.add(new WnnWord(inputRomaji.toUpperCase(), inputHiragana, mPosSymbol)); 592 593 /* Create pseudo candidates for the full width alphabet */ 594 if (createCandidateString(inputRomaji, mFullAlphabetMapQwety, mStringBuff)) { 595 String convZenEiji = mStringBuff.toString(); 596 String convZenEijiLower = convZenEiji.toLowerCase(Locale.JAPAN); 597 list.add(new WnnWord(convZenEiji, inputHiragana, mPosSymbol)); 598 list.add(new WnnWord(convZenEijiLower, inputHiragana, mPosSymbol)); 599 list.add(new WnnWord(convertCaps(convZenEijiLower), inputHiragana, mPosSymbol)); 600 list.add(new WnnWord(convZenEiji.toUpperCase(Locale.JAPAN), inputHiragana, mPosSymbol)); 601 } 602 } 603 604 /** 605 * Create the candidate string 606 * <br> 607 * @param input The input string 608 * @param map The hash map 609 * @param outBuf The output string 610 * @return {@code true} if success 611 */ createCandidateString(String input, HashMap<String,String> map, StringBuffer outBuf)612 private boolean createCandidateString(String input, HashMap<String,String> map, StringBuffer outBuf) { 613 if (outBuf.length() > 0) { 614 outBuf.delete(0, outBuf.length()); 615 } 616 for (int index = 0; index < input.length(); index++) { 617 String convChar = map.get(input.substring(index, index + 1)); 618 if (convChar == null) { 619 return false; 620 } 621 outBuf.append(convChar); 622 } 623 return true; 624 } 625 626 /** 627 * Convert into both small and capital letter 628 * <br> 629 * @param moji The input string 630 * @return The converted string 631 */ convertCaps(String moji)632 private String convertCaps(String moji) { 633 String tmp = ""; 634 if (moji != null && moji.length() > 0) { 635 tmp = moji.substring(0, 1).toUpperCase(Locale.JAPAN) 636 + moji.substring(1).toLowerCase(Locale.JAPAN); 637 } 638 return tmp; 639 } 640 641 /** 642 * Convert the numeric into formatted string 643 * <br> 644 * @param numComma The value 645 * @return {@code true} if success 646 */ convertNumber(String numComma)647 private String convertNumber(String numComma) { 648 try { 649 return mFormat.format(Double.parseDouble(numComma)); 650 } catch (NumberFormatException e) { 651 return null; 652 } 653 } 654 } 655