OLD | NEW |
(Empty) | |
| 1 /* |
| 2 ********************************************************************** |
| 3 * Copyright (C) 2005-2009, International Business Machines |
| 4 * Corporation and others. All Rights Reserved. |
| 5 ********************************************************************** |
| 6 */ |
| 7 |
| 8 #ifndef __CSRSBCS_H |
| 9 #define __CSRSBCS_H |
| 10 |
| 11 #include "unicode/uobject.h" |
| 12 |
| 13 #if !UCONFIG_NO_CONVERSION |
| 14 |
| 15 #include "csrecog.h" |
| 16 |
| 17 U_NAMESPACE_BEGIN |
| 18 |
| 19 class NGramParser : public UMemory |
| 20 { |
| 21 private: |
| 22 int32_t byteIndex; |
| 23 int32_t ngram; |
| 24 |
| 25 const int32_t *ngramList; |
| 26 const uint8_t *charMap; |
| 27 |
| 28 int32_t ngramCount; |
| 29 int32_t hitCount; |
| 30 |
| 31 public: |
| 32 NGramParser(const int32_t *theNgramList, const uint8_t *theCharMap); |
| 33 |
| 34 private: |
| 35 /* |
| 36 * Binary search for value in table, which must have exactly 64 entries. |
| 37 */ |
| 38 int32_t search(const int32_t *table, int32_t value); |
| 39 |
| 40 void lookup(int32_t thisNgram); |
| 41 void addByte(int32_t b); |
| 42 int32_t nextByte(InputText *det); |
| 43 |
| 44 public: |
| 45 int32_t parse(InputText *det); |
| 46 |
| 47 }; |
| 48 |
| 49 class CharsetRecog_sbcs : public CharsetRecognizer |
| 50 { |
| 51 protected: |
| 52 UBool haveC1Bytes; |
| 53 |
| 54 public: |
| 55 CharsetRecog_sbcs(); |
| 56 |
| 57 virtual ~CharsetRecog_sbcs(); |
| 58 |
| 59 virtual const char *getName() const = 0; |
| 60 |
| 61 virtual int32_t match(InputText *det) = 0; |
| 62 |
| 63 int32_t match_sbcs(InputText *det, const int32_t ngrams[], const uint8_t cha
rMap[]); |
| 64 }; |
| 65 |
| 66 class CharsetRecog_8859_1 : public CharsetRecog_sbcs |
| 67 { |
| 68 public: |
| 69 virtual ~CharsetRecog_8859_1(); |
| 70 |
| 71 const char *getName() const; |
| 72 }; |
| 73 |
| 74 class CharsetRecog_8859_2 : public CharsetRecog_sbcs |
| 75 { |
| 76 public: |
| 77 virtual ~CharsetRecog_8859_2(); |
| 78 |
| 79 const char *getName() const; |
| 80 }; |
| 81 |
| 82 class CharsetRecog_8859_5 : public CharsetRecog_sbcs |
| 83 { |
| 84 public: |
| 85 virtual ~CharsetRecog_8859_5(); |
| 86 |
| 87 const char *getName() const; |
| 88 }; |
| 89 |
| 90 class CharsetRecog_8859_6 : public CharsetRecog_sbcs |
| 91 { |
| 92 public: |
| 93 virtual ~CharsetRecog_8859_6(); |
| 94 |
| 95 const char *getName() const; |
| 96 }; |
| 97 |
| 98 class CharsetRecog_8859_7 : public CharsetRecog_sbcs |
| 99 { |
| 100 public: |
| 101 virtual ~CharsetRecog_8859_7(); |
| 102 |
| 103 const char *getName() const; |
| 104 }; |
| 105 |
| 106 class CharsetRecog_8859_8 : public CharsetRecog_sbcs |
| 107 { |
| 108 public: |
| 109 virtual ~CharsetRecog_8859_8(); |
| 110 |
| 111 virtual const char *getName() const; |
| 112 }; |
| 113 |
| 114 class CharsetRecog_8859_9 : public CharsetRecog_sbcs |
| 115 { |
| 116 public: |
| 117 virtual ~CharsetRecog_8859_9(); |
| 118 |
| 119 const char *getName() const; |
| 120 }; |
| 121 |
| 122 class CharsetRecog_8859_1_en : public CharsetRecog_8859_1 |
| 123 { |
| 124 public: |
| 125 virtual ~CharsetRecog_8859_1_en(); |
| 126 |
| 127 const char *getLanguage() const; |
| 128 |
| 129 int32_t match(InputText *textIn); |
| 130 }; |
| 131 |
| 132 class CharsetRecog_8859_1_da : public CharsetRecog_8859_1 |
| 133 { |
| 134 public: |
| 135 virtual ~CharsetRecog_8859_1_da(); |
| 136 |
| 137 const char *getLanguage() const; |
| 138 |
| 139 int32_t match(InputText *textIn); |
| 140 }; |
| 141 |
| 142 class CharsetRecog_8859_1_de : public CharsetRecog_8859_1 |
| 143 { |
| 144 public: |
| 145 virtual ~CharsetRecog_8859_1_de(); |
| 146 |
| 147 const char *getLanguage() const; |
| 148 |
| 149 int32_t match(InputText *textIn); |
| 150 }; |
| 151 |
| 152 class CharsetRecog_8859_1_es : public CharsetRecog_8859_1 |
| 153 { |
| 154 public: |
| 155 virtual ~CharsetRecog_8859_1_es(); |
| 156 |
| 157 const char *getLanguage() const; |
| 158 |
| 159 int32_t match(InputText *textIn); |
| 160 }; |
| 161 |
| 162 class CharsetRecog_8859_1_fr : public CharsetRecog_8859_1 |
| 163 { |
| 164 public: |
| 165 virtual ~CharsetRecog_8859_1_fr(); |
| 166 |
| 167 const char *getLanguage() const; |
| 168 |
| 169 int32_t match(InputText *textIn); |
| 170 }; |
| 171 |
| 172 class CharsetRecog_8859_1_it : public CharsetRecog_8859_1 |
| 173 { |
| 174 public: |
| 175 virtual ~CharsetRecog_8859_1_it(); |
| 176 |
| 177 const char *getLanguage() const; |
| 178 |
| 179 int32_t match(InputText *textIn); |
| 180 }; |
| 181 |
| 182 class CharsetRecog_8859_1_nl : public CharsetRecog_8859_1 |
| 183 { |
| 184 public: |
| 185 virtual ~CharsetRecog_8859_1_nl(); |
| 186 |
| 187 const char *getLanguage() const; |
| 188 |
| 189 int32_t match(InputText *textIn); |
| 190 }; |
| 191 |
| 192 class CharsetRecog_8859_1_no : public CharsetRecog_8859_1 |
| 193 { |
| 194 public: |
| 195 virtual ~CharsetRecog_8859_1_no(); |
| 196 |
| 197 const char *getLanguage() const; |
| 198 |
| 199 int32_t match(InputText *textIn); |
| 200 }; |
| 201 |
| 202 class CharsetRecog_8859_1_pt : public CharsetRecog_8859_1 |
| 203 { |
| 204 public: |
| 205 virtual ~CharsetRecog_8859_1_pt(); |
| 206 |
| 207 const char *getLanguage() const; |
| 208 |
| 209 int32_t match(InputText *textIn); |
| 210 }; |
| 211 |
| 212 class CharsetRecog_8859_1_sv : public CharsetRecog_8859_1 |
| 213 { |
| 214 public: |
| 215 virtual ~CharsetRecog_8859_1_sv(); |
| 216 |
| 217 const char *getLanguage() const; |
| 218 |
| 219 int32_t match(InputText *textIn); |
| 220 }; |
| 221 |
| 222 class CharsetRecog_8859_2_cs : public CharsetRecog_8859_2 |
| 223 { |
| 224 public: |
| 225 virtual ~CharsetRecog_8859_2_cs(); |
| 226 |
| 227 const char *getLanguage() const; |
| 228 |
| 229 int32_t match(InputText *textIn); |
| 230 }; |
| 231 |
| 232 class CharsetRecog_8859_2_hu : public CharsetRecog_8859_2 |
| 233 { |
| 234 public: |
| 235 virtual ~CharsetRecog_8859_2_hu(); |
| 236 |
| 237 const char *getLanguage() const; |
| 238 |
| 239 int32_t match(InputText *textIn); |
| 240 }; |
| 241 |
| 242 class CharsetRecog_8859_2_pl : public CharsetRecog_8859_2 |
| 243 { |
| 244 public: |
| 245 virtual ~CharsetRecog_8859_2_pl(); |
| 246 |
| 247 const char *getLanguage() const; |
| 248 |
| 249 int32_t match(InputText *textIn); |
| 250 }; |
| 251 |
| 252 class CharsetRecog_8859_2_ro : public CharsetRecog_8859_2 |
| 253 { |
| 254 public: |
| 255 virtual ~CharsetRecog_8859_2_ro(); |
| 256 |
| 257 const char *getLanguage() const; |
| 258 |
| 259 int32_t match(InputText *textIn); |
| 260 }; |
| 261 |
| 262 class CharsetRecog_8859_5_ru : public CharsetRecog_8859_5 |
| 263 { |
| 264 public: |
| 265 virtual ~CharsetRecog_8859_5_ru(); |
| 266 |
| 267 const char *getLanguage() const; |
| 268 |
| 269 int32_t match(InputText *textIn); |
| 270 }; |
| 271 |
| 272 class CharsetRecog_8859_6_ar : public CharsetRecog_8859_6 |
| 273 { |
| 274 public: |
| 275 virtual ~CharsetRecog_8859_6_ar(); |
| 276 |
| 277 const char *getLanguage() const; |
| 278 |
| 279 int32_t match(InputText *textIn); |
| 280 }; |
| 281 |
| 282 class CharsetRecog_8859_7_el : public CharsetRecog_8859_7 |
| 283 { |
| 284 public: |
| 285 virtual ~CharsetRecog_8859_7_el(); |
| 286 |
| 287 const char *getLanguage() const; |
| 288 |
| 289 int32_t match(InputText *textIn); |
| 290 }; |
| 291 |
| 292 class CharsetRecog_8859_8_I_he : public CharsetRecog_8859_8 |
| 293 { |
| 294 public: |
| 295 virtual ~CharsetRecog_8859_8_I_he(); |
| 296 |
| 297 const char *getName() const; |
| 298 |
| 299 const char *getLanguage() const; |
| 300 |
| 301 int32_t match(InputText *textIn); |
| 302 }; |
| 303 |
| 304 class CharsetRecog_8859_8_he : public CharsetRecog_8859_8 |
| 305 { |
| 306 public: |
| 307 virtual ~CharsetRecog_8859_8_he (); |
| 308 |
| 309 const char *getLanguage() const; |
| 310 |
| 311 int32_t match(InputText *textIn); |
| 312 }; |
| 313 |
| 314 class CharsetRecog_8859_9_tr : public CharsetRecog_8859_9 |
| 315 { |
| 316 public: |
| 317 virtual ~CharsetRecog_8859_9_tr (); |
| 318 |
| 319 const char *getLanguage() const; |
| 320 |
| 321 int32_t match(InputText *textIn); |
| 322 }; |
| 323 |
| 324 class CharsetRecog_windows_1256 : public CharsetRecog_sbcs |
| 325 { |
| 326 public: |
| 327 virtual ~CharsetRecog_windows_1256(); |
| 328 |
| 329 const char *getName() const; |
| 330 |
| 331 const char *getLanguage() const; |
| 332 |
| 333 int32_t match(InputText *textIn); |
| 334 }; |
| 335 |
| 336 class CharsetRecog_windows_1251 : public CharsetRecog_sbcs |
| 337 { |
| 338 public: |
| 339 virtual ~CharsetRecog_windows_1251(); |
| 340 |
| 341 const char *getName() const; |
| 342 |
| 343 const char *getLanguage() const; |
| 344 |
| 345 int32_t match(InputText *textIn); |
| 346 }; |
| 347 |
| 348 |
| 349 class CharsetRecog_KOI8_R : public CharsetRecog_sbcs |
| 350 { |
| 351 public: |
| 352 virtual ~CharsetRecog_KOI8_R(); |
| 353 |
| 354 const char *getName() const; |
| 355 |
| 356 const char *getLanguage() const; |
| 357 |
| 358 int32_t match(InputText *textIn); |
| 359 }; |
| 360 |
| 361 class CharsetRecog_IBM424_he : public CharsetRecog_sbcs |
| 362 { |
| 363 public: |
| 364 virtual ~CharsetRecog_IBM424_he(); |
| 365 |
| 366 const char *getLanguage() const; |
| 367 }; |
| 368 |
| 369 class CharsetRecog_IBM424_he_rtl : public CharsetRecog_IBM424_he { |
| 370 public: |
| 371 virtual ~CharsetRecog_IBM424_he_rtl(); |
| 372 |
| 373 const char *getName() const; |
| 374 |
| 375 int32_t match(InputText *textIn); |
| 376 }; |
| 377 |
| 378 class CharsetRecog_IBM424_he_ltr : public CharsetRecog_IBM424_he { |
| 379 virtual ~CharsetRecog_IBM424_he_ltr(); |
| 380 |
| 381 const char *getName() const; |
| 382 |
| 383 int32_t match(InputText *textIn); |
| 384 }; |
| 385 |
| 386 class CharsetRecog_IBM420_ar : public CharsetRecog_sbcs |
| 387 { |
| 388 public: |
| 389 virtual ~CharsetRecog_IBM420_ar(); |
| 390 |
| 391 const char *getLanguage() const; |
| 392 |
| 393 protected: |
| 394 void matchInit(InputText *textIn); |
| 395 void matchFinish(InputText *textIn); |
| 396 |
| 397 private: |
| 398 uint8_t *prev_fInputBytes; |
| 399 int32_t prev_fInputBytesLength; |
| 400 UBool deleteBuffer; |
| 401 |
| 402 UBool isLamAlef(uint8_t b); |
| 403 uint8_t *unshapeLamAlef(const uint8_t *inputBytes, int32_t inputBytesLength,
int32_t &length); |
| 404 uint8_t *unshape(const uint8_t *inputBytes, int32_t inputBytesLength, int32_
t &length); |
| 405 }; |
| 406 |
| 407 class CharsetRecog_IBM420_ar_rtl : public CharsetRecog_IBM420_ar { |
| 408 public: |
| 409 virtual ~CharsetRecog_IBM420_ar_rtl(); |
| 410 |
| 411 const char *getName() const; |
| 412 |
| 413 int32_t match(InputText *textIn); |
| 414 }; |
| 415 |
| 416 class CharsetRecog_IBM420_ar_ltr : public CharsetRecog_IBM420_ar { |
| 417 virtual ~CharsetRecog_IBM420_ar_ltr(); |
| 418 |
| 419 const char *getName() const; |
| 420 |
| 421 int32_t match(InputText *textIn); |
| 422 }; |
| 423 |
| 424 U_NAMESPACE_END |
| 425 |
| 426 #endif |
| 427 #endif /* __CSRSBCS_H */ |
OLD | NEW |