1 /* 2 ********************************************************************** 3 * Copyright (C) 1996-2013, International Business Machines 4 * Corporation and others. All Rights Reserved. 5 ********************************************************************** 6 * 7 * Provides functionality for mapping between 8 * LCID and Posix IDs or ICU locale to codepage 9 * 10 * Note: All classes and code in this file are 11 * intended for internal use only. 12 * 13 * Methods of interest: 14 * unsigned long convertToLCID(const char*); 15 * const char* convertToPosix(unsigned long); 16 * 17 * Kathleen Wilson, 4/30/96 18 * 19 * Date Name Description 20 * 3/11/97 aliu Fixed off-by-one bug in assignment operator. Added 21 * setId() method and safety check against 22 * MAX_ID_LENGTH. 23 * 04/23/99 stephen Added C wrapper for convertToPosix. 24 * 09/18/00 george Removed the memory leaks. 25 * 08/23/01 george Convert to C 26 */ 27 28 #include "locmap.h" 29 #include "unicode/uloc.h" 30 #include "cstring.h" 31 #include "cmemory.h" 32 33 #if U_PLATFORM == U_PF_WINDOWS && defined(_MSC_VER) && (_MSC_VER >= 1500) 34 /* 35 * TODO: It seems like we should widen this to 36 * either U_PLATFORM_USES_ONLY_WIN32_API (includes MinGW) 37 * or U_PLATFORM_HAS_WIN32_API (includes MinGW and Cygwin) 38 * but those use gcc and won't have defined(_MSC_VER). 39 * We might need to #include some Windows header and test for some version macro from there. 40 * Or call some Windows function and see what it returns. 41 */ 42 #define USE_WINDOWS_LOCALE_API 43 #endif 44 45 #ifdef USE_WINDOWS_LOCALE_API 46 #include <windows.h> 47 #include <winnls.h> 48 #endif 49 50 /* 51 * Note: 52 * The mapping from Win32 locale ID numbers to POSIX locale strings should 53 * be the faster one. 54 * 55 * Many LCID values come from winnt.h 56 * Some also come from http://www.microsoft.com/globaldev/reference/lcid-all.mspx 57 */ 58 59 /* 60 //////////////////////////////////////////////// 61 // 62 // Internal Classes for LCID <--> POSIX Mapping 63 // 64 ///////////////////////////////////////////////// 65 */ 66 67 typedef struct ILcidPosixElement 68 { 69 const uint32_t hostID; 70 const char * const posixID; 71 } ILcidPosixElement; 72 73 typedef struct ILcidPosixMap 74 { 75 const uint32_t numRegions; 76 const struct ILcidPosixElement* const regionMaps; 77 } ILcidPosixMap; 78 79 80 /* 81 ///////////////////////////////////////////////// 82 // 83 // Easy macros to make the LCID <--> POSIX Mapping 84 // 85 ///////////////////////////////////////////////// 86 */ 87 88 /** 89 * The standard one language/one country mapping for LCID. 90 * The first element must be the language, and the following 91 * elements are the language with the country. 92 * @param hostID LCID in host format such as 0x044d 93 * @param languageID posix ID of just the language such as 'de' 94 * @param posixID posix ID of the language_TERRITORY such as 'de_CH' 95 */ 96 #define ILCID_POSIX_ELEMENT_ARRAY(hostID, languageID, posixID) \ 97 static const ILcidPosixElement locmap_ ## languageID [] = { \ 98 {LANGUAGE_LCID(hostID), #languageID}, /* parent locale */ \ 99 {hostID, #posixID}, \ 100 }; 101 102 /** 103 * Define a subtable by ID 104 * @param id the POSIX ID, either a language or language_TERRITORY 105 */ 106 #define ILCID_POSIX_SUBTABLE(id) \ 107 static const ILcidPosixElement locmap_ ## id [] = 108 109 110 /** 111 * Create the map for the posixID. This macro supposes that the language string 112 * name is the same as the global variable name, and that the first element 113 * in the ILcidPosixElement is just the language. 114 * @param _posixID the full POSIX ID for this entry. 115 */ 116 #define ILCID_POSIX_MAP(_posixID) \ 117 {sizeof(locmap_ ## _posixID)/sizeof(ILcidPosixElement), locmap_ ## _posixID} 118 119 /* 120 //////////////////////////////////////////// 121 // 122 // Create the table of LCID to POSIX Mapping 123 // None of it should be dynamically created. 124 // 125 // Keep static locale variables inside the function so that 126 // it can be created properly during static init. 127 // 128 // Note: This table should be updated periodically. Check the National Lanaguage Support API Reference Website. 129 // Microsoft is moving away from LCID in favor of locale name as of Vista. This table needs to be 130 // maintained for support of older Windows version. 131 // Update: Windows 7 (091130) 132 //////////////////////////////////////////// 133 */ 134 135 ILCID_POSIX_ELEMENT_ARRAY(0x0436, af, af_ZA) 136 137 ILCID_POSIX_SUBTABLE(ar) { 138 {0x01, "ar"}, 139 {0x3801, "ar_AE"}, 140 {0x3c01, "ar_BH"}, 141 {0x1401, "ar_DZ"}, 142 {0x0c01, "ar_EG"}, 143 {0x0801, "ar_IQ"}, 144 {0x2c01, "ar_JO"}, 145 {0x3401, "ar_KW"}, 146 {0x3001, "ar_LB"}, 147 {0x1001, "ar_LY"}, 148 {0x1801, "ar_MA"}, 149 {0x1801, "ar_MO"}, 150 {0x2001, "ar_OM"}, 151 {0x4001, "ar_QA"}, 152 {0x0401, "ar_SA"}, 153 {0x2801, "ar_SY"}, 154 {0x1c01, "ar_TN"}, 155 {0x2401, "ar_YE"} 156 }; 157 158 ILCID_POSIX_ELEMENT_ARRAY(0x044d, as, as_IN) 159 ILCID_POSIX_ELEMENT_ARRAY(0x045e, am, am_ET) 160 ILCID_POSIX_ELEMENT_ARRAY(0x047a, arn,arn_CL) 161 162 ILCID_POSIX_SUBTABLE(az) { 163 {0x2c, "az"}, 164 {0x082c, "az_Cyrl_AZ"}, /* Cyrillic based */ 165 {0x742c, "az_Cyrl"}, /* Cyrillic based */ 166 {0x042c, "az_Latn_AZ"}, /* Latin based */ 167 {0x782c, "az_Latn"}, /* Latin based */ 168 {0x042c, "az_AZ"} /* Latin based */ 169 }; 170 171 ILCID_POSIX_ELEMENT_ARRAY(0x046d, ba, ba_RU) 172 ILCID_POSIX_ELEMENT_ARRAY(0x0423, be, be_BY) 173 174 /*ILCID_POSIX_SUBTABLE(ber) { 175 {0x5f, "ber"}, 176 {0x045f, "ber_Arab_DZ"}, 177 {0x045f, "ber_Arab"}, 178 {0x085f, "ber_Latn_DZ"}, 179 {0x085f, "ber_Latn"} 180 };*/ 181 182 ILCID_POSIX_ELEMENT_ARRAY(0x0402, bg, bg_BG) 183 184 ILCID_POSIX_ELEMENT_ARRAY(0x0466, bin, bin_NG) 185 186 ILCID_POSIX_SUBTABLE(bn) { 187 {0x45, "bn"}, 188 {0x0845, "bn_BD"}, 189 {0x0445, "bn_IN"} 190 }; 191 192 ILCID_POSIX_SUBTABLE(bo) { 193 {0x51, "bo"}, 194 {0x0851, "bo_BT"}, 195 {0x0451, "bo_CN"} 196 }; 197 198 ILCID_POSIX_ELEMENT_ARRAY(0x047e, br, br_FR) 199 200 ILCID_POSIX_SUBTABLE(ca) { 201 {0x03, "ca"}, 202 {0x0403, "ca_ES"}, 203 {0x0803, "ca_ES_VALENCIA"} 204 }; 205 206 ILCID_POSIX_ELEMENT_ARRAY(0x0483, co, co_FR) 207 ILCID_POSIX_ELEMENT_ARRAY(0x045c, chr,chr_US) 208 209 ILCID_POSIX_SUBTABLE(ckb) { 210 {0x92, "ckb"}, 211 {0x92, "ku"}, 212 {0x7c92, "ckb_Arab"}, 213 {0x7c92, "ku_Arab"}, 214 {0x0492, "ckb_Arab_IQ"}, 215 {0x0492, "ku_Arab_IQ"} 216 }; 217 218 /* Declared as cs_CZ to get around compiler errors on z/OS, which defines cs as a function */ 219 ILCID_POSIX_ELEMENT_ARRAY(0x0405, cs, cs_CZ) 220 221 ILCID_POSIX_ELEMENT_ARRAY(0x0452, cy, cy_GB) 222 ILCID_POSIX_ELEMENT_ARRAY(0x0406, da, da_DK) 223 224 ILCID_POSIX_SUBTABLE(de) { 225 {0x07, "de"}, 226 {0x0c07, "de_AT"}, 227 {0x0807, "de_CH"}, 228 {0x0407, "de_DE"}, 229 {0x1407, "de_LI"}, 230 {0x1007, "de_LU"}, 231 {0x10407,"de_DE@collation=phonebook"}, /*This is really de_DE_PHONEBOOK on Windows*/ 232 {0x10407,"de@collation=phonebook"} /*This is really de_DE_PHONEBOOK on Windows*/ 233 }; 234 235 ILCID_POSIX_ELEMENT_ARRAY(0x0465, dv, dv_MV) 236 ILCID_POSIX_ELEMENT_ARRAY(0x0408, el, el_GR) 237 238 ILCID_POSIX_SUBTABLE(en) { 239 {0x09, "en"}, 240 {0x0c09, "en_AU"}, 241 {0x2809, "en_BZ"}, 242 {0x1009, "en_CA"}, 243 {0x0809, "en_GB"}, 244 {0x3c09, "en_HK"}, 245 {0x3809, "en_ID"}, 246 {0x1809, "en_IE"}, 247 {0x4009, "en_IN"}, 248 {0x2009, "en_JM"}, 249 {0x4409, "en_MY"}, 250 {0x1409, "en_NZ"}, 251 {0x3409, "en_PH"}, 252 {0x4809, "en_SG"}, 253 {0x2C09, "en_TT"}, 254 {0x0409, "en_US"}, 255 {0x007f, "en_US_POSIX"}, /* duplicate for roundtripping */ 256 {0x2409, "en_VI"}, /* Virgin Islands AKA Caribbean Islands (en_CB). */ 257 {0x1c09, "en_ZA"}, 258 {0x3009, "en_ZW"}, 259 {0x2409, "en_029"}, 260 {0x0409, "en_AS"}, /* Alias for en_US. Leave last. */ 261 {0x0409, "en_GU"}, /* Alias for en_US. Leave last. */ 262 {0x0409, "en_MH"}, /* Alias for en_US. Leave last. */ 263 {0x0409, "en_MP"}, /* Alias for en_US. Leave last. */ 264 {0x0409, "en_UM"} /* Alias for en_US. Leave last. */ 265 }; 266 267 ILCID_POSIX_SUBTABLE(en_US_POSIX) { 268 {0x007f, "en_US_POSIX"} /* duplicate for roundtripping */ 269 }; 270 271 ILCID_POSIX_SUBTABLE(es) { 272 {0x0a, "es"}, 273 {0x2c0a, "es_AR"}, 274 {0x400a, "es_BO"}, 275 {0x340a, "es_CL"}, 276 {0x240a, "es_CO"}, 277 {0x140a, "es_CR"}, 278 {0x1c0a, "es_DO"}, 279 {0x300a, "es_EC"}, 280 {0x0c0a, "es_ES"}, /*Modern sort.*/ 281 {0x100a, "es_GT"}, 282 {0x480a, "es_HN"}, 283 {0x080a, "es_MX"}, 284 {0x4c0a, "es_NI"}, 285 {0x180a, "es_PA"}, 286 {0x280a, "es_PE"}, 287 {0x500a, "es_PR"}, 288 {0x3c0a, "es_PY"}, 289 {0x440a, "es_SV"}, 290 {0x540a, "es_US"}, 291 {0x380a, "es_UY"}, 292 {0x200a, "es_VE"}, 293 {0xe40a, "es_419"}, 294 {0x040a, "es_ES@collation=traditional"}, 295 {0x040a, "es@collation=traditional"} 296 }; 297 298 ILCID_POSIX_ELEMENT_ARRAY(0x0425, et, et_EE) 299 ILCID_POSIX_ELEMENT_ARRAY(0x042d, eu, eu_ES) 300 301 /* ISO-639 doesn't distinguish between Persian and Dari.*/ 302 ILCID_POSIX_SUBTABLE(fa) { 303 {0x29, "fa"}, 304 {0x0429, "fa_IR"}, /* Persian/Farsi (Iran) */ 305 {0x048c, "fa_AF"} /* Persian/Dari (Afghanistan) */ 306 }; 307 308 /* duplicate for roundtripping */ 309 ILCID_POSIX_SUBTABLE(fa_AF) { 310 {0x8c, "fa_AF"}, /* Persian/Dari (Afghanistan) */ 311 {0x048c, "fa_AF"} /* Persian/Dari (Afghanistan) */ 312 }; 313 314 ILCID_POSIX_SUBTABLE(ff) { 315 {0x67, "ff"}, 316 {0x7c67, "ff_Latn"}, 317 {0x0867, "ff_Latn_SN"} 318 }; 319 320 ILCID_POSIX_ELEMENT_ARRAY(0x040b, fi, fi_FI) 321 ILCID_POSIX_ELEMENT_ARRAY(0x0464, fil,fil_PH) 322 ILCID_POSIX_ELEMENT_ARRAY(0x0438, fo, fo_FO) 323 324 ILCID_POSIX_SUBTABLE(fr) { 325 {0x0c, "fr"}, 326 {0x080c, "fr_BE"}, 327 {0x0c0c, "fr_CA"}, 328 {0x240c, "fr_CD"}, 329 {0x240c, "fr_CG"}, 330 {0x100c, "fr_CH"}, 331 {0x300c, "fr_CI"}, 332 {0x2c0c, "fr_CM"}, 333 {0x040c, "fr_FR"}, 334 {0x3c0c, "fr_HT"}, 335 {0x140c, "fr_LU"}, 336 {0x380c, "fr_MA"}, 337 {0x180c, "fr_MC"}, 338 {0x340c, "fr_ML"}, 339 {0x200c, "fr_RE"}, 340 {0x280c, "fr_SN"}, 341 {0xe40c, "fr_015"}, 342 {0x1c0c, "fr_029"} 343 }; 344 345 ILCID_POSIX_ELEMENT_ARRAY(0x0467, fuv, fuv_NG) 346 347 ILCID_POSIX_ELEMENT_ARRAY(0x0462, fy, fy_NL) 348 349 ILCID_POSIX_SUBTABLE(ga) { /* Gaelic (Ireland) */ 350 {0x3c, "ga"}, 351 {0x083c, "ga_IE"}, 352 {0x043c, "gd_GB"} 353 }; 354 355 ILCID_POSIX_SUBTABLE(gd) { /* Gaelic (Scotland) */ 356 {0x91, "gd"}, 357 {0x0491, "gd_GB"} 358 }; 359 360 ILCID_POSIX_ELEMENT_ARRAY(0x0456, gl, gl_ES) 361 ILCID_POSIX_ELEMENT_ARRAY(0x0447, gu, gu_IN) 362 ILCID_POSIX_ELEMENT_ARRAY(0x0474, gn, gn_PY) 363 ILCID_POSIX_ELEMENT_ARRAY(0x0484, gsw,gsw_FR) 364 365 ILCID_POSIX_SUBTABLE(ha) { 366 {0x68, "ha"}, 367 {0x7c68, "ha_Latn"}, 368 {0x0468, "ha_Latn_NG"}, 369 }; 370 371 ILCID_POSIX_ELEMENT_ARRAY(0x0475, haw,haw_US) 372 ILCID_POSIX_ELEMENT_ARRAY(0x040d, he, he_IL) 373 ILCID_POSIX_ELEMENT_ARRAY(0x0439, hi, hi_IN) 374 375 /* This LCID is really four different locales.*/ 376 ILCID_POSIX_SUBTABLE(hr) { 377 {0x1a, "hr"}, 378 {0x141a, "bs_Latn_BA"}, /* Bosnian, Bosnia and Herzegovina */ 379 {0x681a, "bs_Latn"}, /* Bosnian, Bosnia and Herzegovina */ 380 {0x141a, "bs_BA"}, /* Bosnian, Bosnia and Herzegovina */ 381 {0x781a, "bs"}, /* Bosnian */ 382 {0x201a, "bs_Cyrl_BA"}, /* Bosnian, Bosnia and Herzegovina */ 383 {0x641a, "bs_Cyrl"}, /* Bosnian, Bosnia and Herzegovina */ 384 {0x101a, "hr_BA"}, /* Croatian in Bosnia */ 385 {0x041a, "hr_HR"}, /* Croatian*/ 386 {0x2c1a, "sr_Latn_ME"}, 387 {0x241a, "sr_Latn_RS"}, 388 {0x181a, "sr_Latn_BA"}, /* Serbo-Croatian in Bosnia */ 389 {0x081a, "sr_Latn_CS"}, /* Serbo-Croatian*/ 390 {0x701a, "sr_Latn"}, /* It's 0x1a or 0x081a, pick one to make the test program happy. */ 391 {0x1c1a, "sr_Cyrl_BA"}, /* Serbo-Croatian in Bosnia */ 392 {0x0c1a, "sr_Cyrl_CS"}, /* Serbian*/ 393 {0x301a, "sr_Cyrl_ME"}, 394 {0x281a, "sr_Cyrl_RS"}, 395 {0x6c1a, "sr_Cyrl"}, /* It's 0x1a or 0x0c1a, pick one to make the test program happy. */ 396 {0x7c1a, "sr"} /* In CLDR sr is sr_Cyrl. */ 397 }; 398 399 ILCID_POSIX_ELEMENT_ARRAY(0x040e, hu, hu_HU) 400 ILCID_POSIX_ELEMENT_ARRAY(0x042b, hy, hy_AM) 401 ILCID_POSIX_ELEMENT_ARRAY(0x0469, ibb, ibb_NG) 402 ILCID_POSIX_ELEMENT_ARRAY(0x0421, id, id_ID) 403 ILCID_POSIX_ELEMENT_ARRAY(0x0470, ig, ig_NG) 404 ILCID_POSIX_ELEMENT_ARRAY(0x0478, ii, ii_CN) 405 ILCID_POSIX_ELEMENT_ARRAY(0x040f, is, is_IS) 406 407 ILCID_POSIX_SUBTABLE(it) { 408 {0x10, "it"}, 409 {0x0810, "it_CH"}, 410 {0x0410, "it_IT"} 411 }; 412 413 ILCID_POSIX_SUBTABLE(iu) { 414 {0x5d, "iu"}, 415 {0x045d, "iu_Cans_CA"}, 416 {0x785d, "iu_Cans"}, 417 {0x085d, "iu_Latn_CA"}, 418 {0x7c5d, "iu_Latn"} 419 }; 420 421 ILCID_POSIX_ELEMENT_ARRAY(0x040d, iw, iw_IL) /*Left in for compatibility*/ 422 ILCID_POSIX_ELEMENT_ARRAY(0x0411, ja, ja_JP) 423 ILCID_POSIX_ELEMENT_ARRAY(0x0437, ka, ka_GE) 424 ILCID_POSIX_ELEMENT_ARRAY(0x043f, kk, kk_KZ) 425 ILCID_POSIX_ELEMENT_ARRAY(0x046f, kl, kl_GL) 426 ILCID_POSIX_ELEMENT_ARRAY(0x0453, km, km_KH) 427 ILCID_POSIX_ELEMENT_ARRAY(0x044b, kn, kn_IN) 428 429 ILCID_POSIX_SUBTABLE(ko) { 430 {0x12, "ko"}, 431 {0x0812, "ko_KP"}, 432 {0x0412, "ko_KR"} 433 }; 434 435 ILCID_POSIX_ELEMENT_ARRAY(0x0457, kok, kok_IN) 436 ILCID_POSIX_ELEMENT_ARRAY(0x0471, kr, kr_NG) 437 438 ILCID_POSIX_SUBTABLE(ks) { /* We could add PK and CN too */ 439 {0x60, "ks"}, 440 {0x0860, "ks_IN"}, /* Documentation doesn't mention script */ 441 {0x0460, "ks_Arab_IN"}, 442 {0x0860, "ks_Deva_IN"} 443 }; 444 445 ILCID_POSIX_ELEMENT_ARRAY(0x0440, ky, ky_KG) /* Kyrgyz is spoken in Kyrgyzstan */ 446 ILCID_POSIX_ELEMENT_ARRAY(0x0476, la, la_IT) /* TODO: Verify the country */ 447 ILCID_POSIX_ELEMENT_ARRAY(0x046e, lb, lb_LU) 448 ILCID_POSIX_ELEMENT_ARRAY(0x0454, lo, lo_LA) 449 ILCID_POSIX_ELEMENT_ARRAY(0x0427, lt, lt_LT) 450 ILCID_POSIX_ELEMENT_ARRAY(0x0426, lv, lv_LV) 451 ILCID_POSIX_ELEMENT_ARRAY(0x0481, mi, mi_NZ) 452 ILCID_POSIX_ELEMENT_ARRAY(0x042f, mk, mk_MK) 453 ILCID_POSIX_ELEMENT_ARRAY(0x044c, ml, ml_IN) 454 455 ILCID_POSIX_SUBTABLE(mn) { 456 {0x50, "mn"}, 457 {0x0450, "mn_MN"}, 458 {0x7c50, "mn_Mong"}, 459 {0x0850, "mn_Mong_CN"}, 460 {0x0850, "mn_CN"}, 461 {0x7850, "mn_Cyrl"} 462 }; 463 464 ILCID_POSIX_ELEMENT_ARRAY(0x0458, mni,mni_IN) 465 ILCID_POSIX_ELEMENT_ARRAY(0x047c, moh,moh_CA) 466 ILCID_POSIX_ELEMENT_ARRAY(0x044e, mr, mr_IN) 467 468 ILCID_POSIX_SUBTABLE(ms) { 469 {0x3e, "ms"}, 470 {0x083e, "ms_BN"}, /* Brunei Darussalam*/ 471 {0x043e, "ms_MY"} /* Malaysia*/ 472 }; 473 474 ILCID_POSIX_ELEMENT_ARRAY(0x043a, mt, mt_MT) 475 ILCID_POSIX_ELEMENT_ARRAY(0x0455, my, my_MM) 476 477 ILCID_POSIX_SUBTABLE(ne) { 478 {0x61, "ne"}, 479 {0x0861, "ne_IN"}, /* India*/ 480 {0x0461, "ne_NP"} /* Nepal*/ 481 }; 482 483 ILCID_POSIX_SUBTABLE(nl) { 484 {0x13, "nl"}, 485 {0x0813, "nl_BE"}, 486 {0x0413, "nl_NL"} 487 }; 488 489 /* The "no" locale split into nb and nn. By default in ICU, "no" is nb.*/ 490 ILCID_POSIX_SUBTABLE(no) { 491 {0x14, "no"}, /* really nb_NO */ 492 {0x7c14, "nb"}, /* really nb */ 493 {0x0414, "nb_NO"}, /* really nb_NO. Keep first in the 414 list. */ 494 {0x0414, "no_NO"}, /* really nb_NO */ 495 {0x0814, "nn_NO"}, /* really nn_NO. Keep first in the 814 list. */ 496 {0x7814, "nn"}, /* It's 0x14 or 0x814, pick one to make the test program happy. */ 497 {0x0814, "no_NO_NY"}/* really nn_NO */ 498 }; 499 500 ILCID_POSIX_ELEMENT_ARRAY(0x046c, nso,nso_ZA) /* TODO: Verify the ISO-639 code */ 501 ILCID_POSIX_ELEMENT_ARRAY(0x0482, oc, oc_FR) 502 503 ILCID_POSIX_SUBTABLE(om) { /* TODO: Verify the country */ 504 {0x72, "om"}, 505 {0x0472, "om_ET"}, 506 {0x0472, "gaz_ET"} 507 }; 508 509 /* Declared as or_IN to get around compiler errors*/ 510 ILCID_POSIX_SUBTABLE(or_IN) { 511 {0x48, "or"}, 512 {0x0448, "or_IN"}, 513 }; 514 515 516 ILCID_POSIX_SUBTABLE(pa) { 517 {0x46, "pa"}, 518 {0x0446, "pa_IN"}, 519 {0x0846, "pa_PK"}, 520 {0x0846, "pa_Arab_PK"} 521 }; 522 523 ILCID_POSIX_ELEMENT_ARRAY(0x0479, pap, pap_AN) 524 ILCID_POSIX_ELEMENT_ARRAY(0x0415, pl, pl_PL) 525 ILCID_POSIX_ELEMENT_ARRAY(0x0463, ps, ps_AF) 526 527 ILCID_POSIX_SUBTABLE(pt) { 528 {0x16, "pt"}, 529 {0x0416, "pt_BR"}, 530 {0x0816, "pt_PT"} 531 }; 532 533 ILCID_POSIX_SUBTABLE(qu) { 534 {0x6b, "qu"}, 535 {0x046b, "qu_BO"}, 536 {0x086b, "qu_EC"}, 537 {0x0C6b, "qu_PE"}, 538 {0x046b, "quz_BO"}, 539 {0x086b, "quz_EC"}, 540 {0x0C6b, "quz_PE"} 541 }; 542 543 ILCID_POSIX_ELEMENT_ARRAY(0x0486, qut, qut_GT) /* qut is an ISO-639-3 code */ 544 ILCID_POSIX_ELEMENT_ARRAY(0x0417, rm, rm_CH) 545 546 ILCID_POSIX_SUBTABLE(ro) { 547 {0x18, "ro"}, 548 {0x0418, "ro_RO"}, 549 {0x0818, "ro_MD"} 550 }; 551 552 ILCID_POSIX_SUBTABLE(root) { 553 {0x00, "root"} 554 }; 555 556 ILCID_POSIX_SUBTABLE(ru) { 557 {0x19, "ru"}, 558 {0x0419, "ru_RU"}, 559 {0x0819, "ru_MD"} 560 }; 561 562 ILCID_POSIX_ELEMENT_ARRAY(0x0487, rw, rw_RW) 563 ILCID_POSIX_ELEMENT_ARRAY(0x044f, sa, sa_IN) 564 ILCID_POSIX_ELEMENT_ARRAY(0x0485, sah,sah_RU) 565 566 ILCID_POSIX_SUBTABLE(sd) { 567 {0x59, "sd"}, 568 {0x0459, "sd_IN"}, 569 {0x0859, "sd_PK"} 570 }; 571 572 ILCID_POSIX_SUBTABLE(se) { 573 {0x3b, "se"}, 574 {0x0c3b, "se_FI"}, 575 {0x043b, "se_NO"}, 576 {0x083b, "se_SE"}, 577 {0x783b, "sma"}, 578 {0x183b, "sma_NO"}, 579 {0x1c3b, "sma_SE"}, 580 {0x7c3b, "smj"}, 581 {0x703b, "smn"}, 582 {0x743b, "sms"}, 583 {0x103b, "smj_NO"}, 584 {0x143b, "smj_SE"}, 585 {0x243b, "smn_FI"}, 586 {0x203b, "sms_FI"}, 587 }; 588 589 ILCID_POSIX_ELEMENT_ARRAY(0x045b, si, si_LK) 590 ILCID_POSIX_ELEMENT_ARRAY(0x041b, sk, sk_SK) 591 ILCID_POSIX_ELEMENT_ARRAY(0x0424, sl, sl_SI) 592 593 ILCID_POSIX_SUBTABLE(so) { /* TODO: Verify the country */ 594 {0x77, "so"}, 595 {0x0477, "so_ET"}, 596 {0x0477, "so_SO"} 597 }; 598 599 ILCID_POSIX_ELEMENT_ARRAY(0x041c, sq, sq_AL) 600 ILCID_POSIX_ELEMENT_ARRAY(0x0430, st, st_ZA) 601 602 ILCID_POSIX_SUBTABLE(sv) { 603 {0x1d, "sv"}, 604 {0x081d, "sv_FI"}, 605 {0x041d, "sv_SE"} 606 }; 607 608 ILCID_POSIX_ELEMENT_ARRAY(0x0441, sw, sw_KE) 609 ILCID_POSIX_ELEMENT_ARRAY(0x045A, syr, syr_SY) 610 611 ILCID_POSIX_SUBTABLE(ta) { 612 {0x49, "ta"}, 613 {0x0449, "ta_IN"}, 614 {0x0849, "ta_LK"} 615 }; 616 617 ILCID_POSIX_ELEMENT_ARRAY(0x044a, te, te_IN) 618 619 /* Cyrillic based by default */ 620 ILCID_POSIX_SUBTABLE(tg) { 621 {0x28, "tg"}, 622 {0x7c28, "tg_Cyrl"}, 623 {0x0428, "tg_Cyrl_TJ"} 624 }; 625 626 ILCID_POSIX_ELEMENT_ARRAY(0x041e, th, th_TH) 627 628 ILCID_POSIX_SUBTABLE(ti) { 629 {0x73, "ti"}, 630 {0x0873, "ti_ER"}, 631 {0x0473, "ti_ET"} 632 }; 633 634 ILCID_POSIX_ELEMENT_ARRAY(0x0442, tk, tk_TM) 635 636 ILCID_POSIX_SUBTABLE(tn) { 637 {0x32, "tn"}, 638 {0x0832, "tn_BW"}, 639 {0x0432, "tn_ZA"} 640 }; 641 642 ILCID_POSIX_ELEMENT_ARRAY(0x041f, tr, tr_TR) 643 ILCID_POSIX_ELEMENT_ARRAY(0x0431, ts, ts_ZA) 644 ILCID_POSIX_ELEMENT_ARRAY(0x0444, tt, tt_RU) 645 646 ILCID_POSIX_SUBTABLE(tzm) { 647 {0x5f, "tzm"}, 648 {0x7c5f, "tzm_Latn"}, 649 {0x085f, "tzm_Latn_DZ"}, 650 {0x105f, "tzm_Tfng_MA"}, 651 {0x045f, "tmz"} 652 }; 653 654 ILCID_POSIX_SUBTABLE(ug) { 655 {0x80, "ug"}, 656 {0x0480, "ug_CN"}, 657 {0x0480, "ug_Arab_CN"} 658 }; 659 660 ILCID_POSIX_ELEMENT_ARRAY(0x0422, uk, uk_UA) 661 662 ILCID_POSIX_SUBTABLE(ur) { 663 {0x20, "ur"}, 664 {0x0820, "ur_IN"}, 665 {0x0420, "ur_PK"} 666 }; 667 668 ILCID_POSIX_SUBTABLE(uz) { 669 {0x43, "uz"}, 670 {0x0843, "uz_Cyrl_UZ"}, /* Cyrillic based */ 671 {0x7843, "uz_Cyrl"}, /* Cyrillic based */ 672 {0x0843, "uz_UZ"}, /* Cyrillic based */ 673 {0x0443, "uz_Latn_UZ"}, /* Latin based */ 674 {0x7c43, "uz_Latn"} /* Latin based */ 675 }; 676 677 ILCID_POSIX_SUBTABLE(ve) { /* TODO: Verify the country */ 678 {0x33, "ve"}, 679 {0x0433, "ve_ZA"}, 680 {0x0433, "ven_ZA"} 681 }; 682 683 ILCID_POSIX_ELEMENT_ARRAY(0x042a, vi, vi_VN) 684 685 ILCID_POSIX_SUBTABLE(wen) { 686 {0x2E, "wen"}, 687 {0x042E, "wen_DE"}, 688 {0x042E, "hsb_DE"}, 689 {0x082E, "dsb_DE"}, 690 {0x7C2E, "dsb"}, 691 {0x2E, "hsb"} 692 }; 693 694 ILCID_POSIX_ELEMENT_ARRAY(0x0488, wo, wo_SN) 695 ILCID_POSIX_ELEMENT_ARRAY(0x0434, xh, xh_ZA) 696 ILCID_POSIX_ELEMENT_ARRAY(0x043d, yi, yi) 697 ILCID_POSIX_ELEMENT_ARRAY(0x046a, yo, yo_NG) 698 699 ILCID_POSIX_SUBTABLE(zh) { 700 {0x0004, "zh_Hans"}, 701 {0x7804, "zh"}, 702 {0x0804, "zh_CN"}, 703 {0x0804, "zh_Hans_CN"}, 704 {0x0c04, "zh_Hant_HK"}, 705 {0x0c04, "zh_HK"}, 706 {0x1404, "zh_Hant_MO"}, 707 {0x1404, "zh_MO"}, 708 {0x1004, "zh_Hans_SG"}, 709 {0x1004, "zh_SG"}, 710 {0x0404, "zh_Hant_TW"}, 711 {0x7c04, "zh_Hant"}, 712 {0x0404, "zh_TW"}, 713 {0x30404,"zh_Hant_TW"}, /* Bopomofo order */ 714 {0x30404,"zh_TW"}, /* Bopomofo order */ 715 {0x20004,"zh@collation=stroke"}, 716 {0x20404,"zh_Hant@collation=stroke"}, 717 {0x20404,"zh_Hant_TW@collation=stroke"}, 718 {0x20404,"zh_TW@collation=stroke"}, 719 {0x20804,"zh_Hans@collation=stroke"}, 720 {0x20804,"zh_Hans_CN@collation=stroke"}, 721 {0x20804,"zh_CN@collation=stroke"} 722 }; 723 724 ILCID_POSIX_ELEMENT_ARRAY(0x0435, zu, zu_ZA) 725 726 /* This must be static and grouped by LCID. */ 727 static const ILcidPosixMap gPosixIDmap[] = { 728 ILCID_POSIX_MAP(af), /* af Afrikaans 0x36 */ 729 ILCID_POSIX_MAP(am), /* am Amharic 0x5e */ 730 ILCID_POSIX_MAP(ar), /* ar Arabic 0x01 */ 731 ILCID_POSIX_MAP(arn), /* arn Araucanian/Mapudungun 0x7a */ 732 ILCID_POSIX_MAP(as), /* as Assamese 0x4d */ 733 ILCID_POSIX_MAP(az), /* az Azerbaijani 0x2c */ 734 ILCID_POSIX_MAP(ba), /* ba Bashkir 0x6d */ 735 ILCID_POSIX_MAP(be), /* be Belarusian 0x23 */ 736 /* ILCID_POSIX_MAP(ber), ber Berber/Tamazight 0x5f */ 737 ILCID_POSIX_MAP(bg), /* bg Bulgarian 0x02 */ 738 ILCID_POSIX_MAP(bin), /* bin Edo 0x66 */ 739 ILCID_POSIX_MAP(bn), /* bn Bengali; Bangla 0x45 */ 740 ILCID_POSIX_MAP(bo), /* bo Tibetan 0x51 */ 741 ILCID_POSIX_MAP(br), /* br Breton 0x7e */ 742 ILCID_POSIX_MAP(ca), /* ca Catalan 0x03 */ 743 ILCID_POSIX_MAP(chr), /* chr Cherokee 0x5c */ 744 ILCID_POSIX_MAP(ckb), /* ckb Sorani (Central Kurdish) 0x92 */ 745 ILCID_POSIX_MAP(co), /* co Corsican 0x83 */ 746 ILCID_POSIX_MAP(cs), /* cs Czech 0x05 */ 747 ILCID_POSIX_MAP(cy), /* cy Welsh 0x52 */ 748 ILCID_POSIX_MAP(da), /* da Danish 0x06 */ 749 ILCID_POSIX_MAP(de), /* de German 0x07 */ 750 ILCID_POSIX_MAP(dv), /* dv Divehi 0x65 */ 751 ILCID_POSIX_MAP(el), /* el Greek 0x08 */ 752 ILCID_POSIX_MAP(en), /* en English 0x09 */ 753 ILCID_POSIX_MAP(en_US_POSIX), /* invariant 0x7f */ 754 ILCID_POSIX_MAP(es), /* es Spanish 0x0a */ 755 ILCID_POSIX_MAP(et), /* et Estonian 0x25 */ 756 ILCID_POSIX_MAP(eu), /* eu Basque 0x2d */ 757 ILCID_POSIX_MAP(fa), /* fa Persian/Farsi 0x29 */ 758 ILCID_POSIX_MAP(fa_AF), /* fa Persian/Dari 0x8c */ 759 ILCID_POSIX_MAP(ff), /* ff Fula 0x67 */ 760 ILCID_POSIX_MAP(fi), /* fi Finnish 0x0b */ 761 ILCID_POSIX_MAP(fil), /* fil Filipino 0x64 */ 762 ILCID_POSIX_MAP(fo), /* fo Faroese 0x38 */ 763 ILCID_POSIX_MAP(fr), /* fr French 0x0c */ 764 ILCID_POSIX_MAP(fuv), /* fuv Fulfulde - Nigeria 0x67 */ 765 ILCID_POSIX_MAP(fy), /* fy Frisian 0x62 */ 766 ILCID_POSIX_MAP(ga), /* * Gaelic (Ireland,Scotland) 0x3c */ 767 ILCID_POSIX_MAP(gd), /* gd Gaelic (United Kingdom) 0x91 */ 768 ILCID_POSIX_MAP(gl), /* gl Galician 0x56 */ 769 ILCID_POSIX_MAP(gn), /* gn Guarani 0x74 */ 770 ILCID_POSIX_MAP(gsw), /* gsw Alemanic/Alsatian/Swiss German 0x84 */ 771 ILCID_POSIX_MAP(gu), /* gu Gujarati 0x47 */ 772 ILCID_POSIX_MAP(ha), /* ha Hausa 0x68 */ 773 ILCID_POSIX_MAP(haw), /* haw Hawaiian 0x75 */ 774 ILCID_POSIX_MAP(he), /* he Hebrew (formerly iw) 0x0d */ 775 ILCID_POSIX_MAP(hi), /* hi Hindi 0x39 */ 776 ILCID_POSIX_MAP(hr), /* * Croatian and others 0x1a */ 777 ILCID_POSIX_MAP(hu), /* hu Hungarian 0x0e */ 778 ILCID_POSIX_MAP(hy), /* hy Armenian 0x2b */ 779 ILCID_POSIX_MAP(ibb), /* ibb Ibibio - Nigeria 0x69 */ 780 ILCID_POSIX_MAP(id), /* id Indonesian (formerly in) 0x21 */ 781 ILCID_POSIX_MAP(ig), /* ig Igbo 0x70 */ 782 ILCID_POSIX_MAP(ii), /* ii Sichuan Yi 0x78 */ 783 ILCID_POSIX_MAP(is), /* is Icelandic 0x0f */ 784 ILCID_POSIX_MAP(it), /* it Italian 0x10 */ 785 ILCID_POSIX_MAP(iu), /* iu Inuktitut 0x5d */ 786 ILCID_POSIX_MAP(iw), /* iw Hebrew 0x0d */ 787 ILCID_POSIX_MAP(ja), /* ja Japanese 0x11 */ 788 ILCID_POSIX_MAP(ka), /* ka Georgian 0x37 */ 789 ILCID_POSIX_MAP(kk), /* kk Kazakh 0x3f */ 790 ILCID_POSIX_MAP(kl), /* kl Kalaallisut 0x6f */ 791 ILCID_POSIX_MAP(km), /* km Khmer 0x53 */ 792 ILCID_POSIX_MAP(kn), /* kn Kannada 0x4b */ 793 ILCID_POSIX_MAP(ko), /* ko Korean 0x12 */ 794 ILCID_POSIX_MAP(kok), /* kok Konkani 0x57 */ 795 ILCID_POSIX_MAP(kr), /* kr Kanuri 0x71 */ 796 ILCID_POSIX_MAP(ks), /* ks Kashmiri 0x60 */ 797 ILCID_POSIX_MAP(ky), /* ky Kyrgyz 0x40 */ 798 ILCID_POSIX_MAP(lb), /* lb Luxembourgish 0x6e */ 799 ILCID_POSIX_MAP(la), /* la Latin 0x76 */ 800 ILCID_POSIX_MAP(lo), /* lo Lao 0x54 */ 801 ILCID_POSIX_MAP(lt), /* lt Lithuanian 0x27 */ 802 ILCID_POSIX_MAP(lv), /* lv Latvian, Lettish 0x26 */ 803 ILCID_POSIX_MAP(mi), /* mi Maori 0x81 */ 804 ILCID_POSIX_MAP(mk), /* mk Macedonian 0x2f */ 805 ILCID_POSIX_MAP(ml), /* ml Malayalam 0x4c */ 806 ILCID_POSIX_MAP(mn), /* mn Mongolian 0x50 */ 807 ILCID_POSIX_MAP(mni), /* mni Manipuri 0x58 */ 808 ILCID_POSIX_MAP(moh), /* moh Mohawk 0x7c */ 809 ILCID_POSIX_MAP(mr), /* mr Marathi 0x4e */ 810 ILCID_POSIX_MAP(ms), /* ms Malay 0x3e */ 811 ILCID_POSIX_MAP(mt), /* mt Maltese 0x3a */ 812 ILCID_POSIX_MAP(my), /* my Burmese 0x55 */ 813 /* ILCID_POSIX_MAP(nb), // no Norwegian 0x14 */ 814 ILCID_POSIX_MAP(ne), /* ne Nepali 0x61 */ 815 ILCID_POSIX_MAP(nl), /* nl Dutch 0x13 */ 816 /* ILCID_POSIX_MAP(nn), // no Norwegian 0x14 */ 817 ILCID_POSIX_MAP(no), /* * Norwegian 0x14 */ 818 ILCID_POSIX_MAP(nso), /* nso Sotho, Northern (Sepedi dialect) 0x6c */ 819 ILCID_POSIX_MAP(oc), /* oc Occitan 0x82 */ 820 ILCID_POSIX_MAP(om), /* om Oromo 0x72 */ 821 ILCID_POSIX_MAP(or_IN), /* or Oriya 0x48 */ 822 ILCID_POSIX_MAP(pa), /* pa Punjabi 0x46 */ 823 ILCID_POSIX_MAP(pap), /* pap Papiamentu 0x79 */ 824 ILCID_POSIX_MAP(pl), /* pl Polish 0x15 */ 825 ILCID_POSIX_MAP(ps), /* ps Pashto 0x63 */ 826 ILCID_POSIX_MAP(pt), /* pt Portuguese 0x16 */ 827 ILCID_POSIX_MAP(qu), /* qu Quechua 0x6B */ 828 ILCID_POSIX_MAP(qut), /* qut K'iche 0x86 */ 829 ILCID_POSIX_MAP(rm), /* rm Raeto-Romance/Romansh 0x17 */ 830 ILCID_POSIX_MAP(ro), /* ro Romanian 0x18 */ 831 ILCID_POSIX_MAP(root), /* root 0x00 */ 832 ILCID_POSIX_MAP(ru), /* ru Russian 0x19 */ 833 ILCID_POSIX_MAP(rw), /* rw Kinyarwanda 0x87 */ 834 ILCID_POSIX_MAP(sa), /* sa Sanskrit 0x4f */ 835 ILCID_POSIX_MAP(sah), /* sah Yakut 0x85 */ 836 ILCID_POSIX_MAP(sd), /* sd Sindhi 0x59 */ 837 ILCID_POSIX_MAP(se), /* se Sami 0x3b */ 838 /* ILCID_POSIX_MAP(sh), // sh Serbo-Croatian 0x1a */ 839 ILCID_POSIX_MAP(si), /* si Sinhalese 0x5b */ 840 ILCID_POSIX_MAP(sk), /* sk Slovak 0x1b */ 841 ILCID_POSIX_MAP(sl), /* sl Slovenian 0x24 */ 842 ILCID_POSIX_MAP(so), /* so Somali 0x77 */ 843 ILCID_POSIX_MAP(sq), /* sq Albanian 0x1c */ 844 /* ILCID_POSIX_MAP(sr), // sr Serbian 0x1a */ 845 ILCID_POSIX_MAP(st), /* st Sutu 0x30 */ 846 ILCID_POSIX_MAP(sv), /* sv Swedish 0x1d */ 847 ILCID_POSIX_MAP(sw), /* sw Swahili 0x41 */ 848 ILCID_POSIX_MAP(syr), /* syr Syriac 0x5A */ 849 ILCID_POSIX_MAP(ta), /* ta Tamil 0x49 */ 850 ILCID_POSIX_MAP(te), /* te Telugu 0x4a */ 851 ILCID_POSIX_MAP(tg), /* tg Tajik 0x28 */ 852 ILCID_POSIX_MAP(th), /* th Thai 0x1e */ 853 ILCID_POSIX_MAP(ti), /* ti Tigrigna 0x73 */ 854 ILCID_POSIX_MAP(tk), /* tk Turkmen 0x42 */ 855 ILCID_POSIX_MAP(tn), /* tn Tswana 0x32 */ 856 ILCID_POSIX_MAP(tr), /* tr Turkish 0x1f */ 857 ILCID_POSIX_MAP(ts), /* ts Tsonga 0x31 */ 858 ILCID_POSIX_MAP(tt), /* tt Tatar 0x44 */ 859 ILCID_POSIX_MAP(tzm), /* tzm Tamazight 0x5f */ 860 ILCID_POSIX_MAP(ug), /* ug Uighur 0x80 */ 861 ILCID_POSIX_MAP(uk), /* uk Ukrainian 0x22 */ 862 ILCID_POSIX_MAP(ur), /* ur Urdu 0x20 */ 863 ILCID_POSIX_MAP(uz), /* uz Uzbek 0x43 */ 864 ILCID_POSIX_MAP(ve), /* ve Venda 0x33 */ 865 ILCID_POSIX_MAP(vi), /* vi Vietnamese 0x2a */ 866 ILCID_POSIX_MAP(wen), /* wen Sorbian 0x2e */ 867 ILCID_POSIX_MAP(wo), /* wo Wolof 0x88 */ 868 ILCID_POSIX_MAP(xh), /* xh Xhosa 0x34 */ 869 ILCID_POSIX_MAP(yi), /* yi Yiddish 0x3d */ 870 ILCID_POSIX_MAP(yo), /* yo Yoruba 0x6a */ 871 ILCID_POSIX_MAP(zh), /* zh Chinese 0x04 */ 872 ILCID_POSIX_MAP(zu), /* zu Zulu 0x35 */ 873 }; 874 875 static const uint32_t gLocaleCount = sizeof(gPosixIDmap)/sizeof(ILcidPosixMap); 876 877 /** 878 * Do not call this function. It is called by hostID. 879 * The function is not private because this struct must stay as a C struct, 880 * and this is an internal class. 881 */ 882 static int32_t 883 idCmp(const char* id1, const char* id2) 884 { 885 int32_t diffIdx = 0; 886 while (*id1 == *id2 && *id1 != 0) { 887 diffIdx++; 888 id1++; 889 id2++; 890 } 891 return diffIdx; 892 } 893 894 /** 895 * Searches for a Windows LCID 896 * 897 * @param posixid the Posix style locale id. 898 * @param status gets set to U_ILLEGAL_ARGUMENT_ERROR when the Posix ID has 899 * no equivalent Windows LCID. 900 * @return the LCID 901 */ 902 static uint32_t 903 getHostID(const ILcidPosixMap *this_0, const char* posixID, UErrorCode* status) 904 { 905 int32_t bestIdx = 0; 906 int32_t bestIdxDiff = 0; 907 int32_t posixIDlen = (int32_t)uprv_strlen(posixID); 908 uint32_t idx; 909 910 for (idx = 0; idx < this_0->numRegions; idx++ ) { 911 int32_t sameChars = idCmp(posixID, this_0->regionMaps[idx].posixID); 912 if (sameChars > bestIdxDiff && this_0->regionMaps[idx].posixID[sameChars] == 0) { 913 if (posixIDlen == sameChars) { 914 /* Exact match */ 915 return this_0->regionMaps[idx].hostID; 916 } 917 bestIdxDiff = sameChars; 918 bestIdx = idx; 919 } 920 } 921 /* We asked for something unusual, like en_ZZ, and we try to return the number for the same language. */ 922 /* We also have to make sure that sid and si and similar string subsets don't match. */ 923 if ((posixID[bestIdxDiff] == '_' || posixID[bestIdxDiff] == '@') 924 && this_0->regionMaps[bestIdx].posixID[bestIdxDiff] == 0) 925 { 926 *status = U_USING_FALLBACK_WARNING; 927 return this_0->regionMaps[bestIdx].hostID; 928 } 929 930 /*no match found */ 931 *status = U_ILLEGAL_ARGUMENT_ERROR; 932 return this_0->regionMaps->hostID; 933 } 934 935 static const char* 936 getPosixID(const ILcidPosixMap *this_0, uint32_t hostID) 937 { 938 uint32_t i; 939 for (i = 0; i <= this_0->numRegions; i++) 940 { 941 if (this_0->regionMaps[i].hostID == hostID) 942 { 943 return this_0->regionMaps[i].posixID; 944 } 945 } 946 947 /* If you get here, then no matching region was found, 948 so return the language id with the wild card region. */ 949 return this_0->regionMaps[0].posixID; 950 } 951 952 /* 953 ////////////////////////////////////// 954 // 955 // LCID --> POSIX 956 // 957 ///////////////////////////////////// 958 */ 959 #ifdef USE_WINDOWS_LOCALE_API 960 /* 961 * Change the tag separator from '-' to '_' 962 */ 963 #define FIX_LOCALE_ID_TAG_SEPARATOR(buffer, len, i) \ 964 for(i = 0; i < len; i++) \ 965 if (buffer[i] == '-') buffer[i] = '_'; 966 967 /* 968 * Various language tags needs to be changed: 969 * quz -> qu 970 * prs -> fa 971 */ 972 #define FIX_LANGUAGE_ID_TAG(buffer, len) \ 973 if (len >= 3) { \ 974 if (buffer[0] == 'q' && buffer[1] == 'u' && buffer[2] == 'z') {\ 975 buffer[2] = 0; \ 976 uprv_strcat(buffer, buffer+3); \ 977 } else if (buffer[0] == 'p' && buffer[1] == 'r' && buffer[2] == 's') {\ 978 buffer[0] = 'f'; buffer[1] = 'a'; buffer[2] = 0; \ 979 uprv_strcat(buffer, buffer+3); \ 980 } \ 981 } 982 983 static char gPosixFromLCID[ULOC_FULLNAME_CAPACITY]; 984 #endif 985 U_CAPI const char * 986 uprv_convertToPosix(uint32_t hostid, UErrorCode* status) 987 { 988 uint16_t langID; 989 uint32_t localeIndex; 990 #ifdef USE_WINDOWS_LOCALE_API 991 int32_t ret = 0; 992 993 uprv_memset(gPosixFromLCID, 0, sizeof(gPosixFromLCID)); 994 995 ret = GetLocaleInfoA(hostid, LOCALE_SNAME, (LPSTR)gPosixFromLCID, sizeof(gPosixFromLCID)); 996 if (ret > 1) { 997 FIX_LOCALE_ID_TAG_SEPARATOR(gPosixFromLCID, (uint32_t)ret, localeIndex) 998 FIX_LANGUAGE_ID_TAG(gPosixFromLCID, ret) 999 1000 return gPosixFromLCID; 1001 } 1002 #endif 1003 langID = LANGUAGE_LCID(hostid); 1004 1005 for (localeIndex = 0; localeIndex < gLocaleCount; localeIndex++) 1006 { 1007 if (langID == gPosixIDmap[localeIndex].regionMaps->hostID) 1008 { 1009 return getPosixID(&gPosixIDmap[localeIndex], hostid); 1010 } 1011 } 1012 1013 /* no match found */ 1014 *status = U_ILLEGAL_ARGUMENT_ERROR; 1015 return NULL; 1016 } 1017 1018 /* 1019 ////////////////////////////////////// 1020 // 1021 // POSIX --> LCID 1022 // This should only be called from uloc_getLCID. 1023 // The locale ID must be in canonical form. 1024 // langID is separate so that this file doesn't depend on the uloc_* API. 1025 // 1026 ///////////////////////////////////// 1027 */ 1028 1029 U_CAPI uint32_t 1030 uprv_convertToLCID(const char *langID, const char* posixID, UErrorCode* status) 1031 { 1032 1033 uint32_t low = 0; 1034 uint32_t high = gLocaleCount; 1035 uint32_t mid; 1036 uint32_t oldmid = 0; 1037 int32_t compVal; 1038 1039 uint32_t value = 0; 1040 uint32_t fallbackValue = (uint32_t)-1; 1041 UErrorCode myStatus; 1042 uint32_t idx; 1043 1044 /* Check for incomplete id. */ 1045 if (!langID || !posixID || uprv_strlen(langID) < 2 || uprv_strlen(posixID) < 2) { 1046 return 0; 1047 } 1048 1049 /*Binary search for the map entry for normal cases */ 1050 1051 while (high > low) /*binary search*/{ 1052 1053 mid = (high+low) >> 1; /*Finds median*/ 1054 1055 if (mid == oldmid) 1056 break; 1057 1058 compVal = uprv_strcmp(langID, gPosixIDmap[mid].regionMaps->posixID); 1059 if (compVal < 0){ 1060 high = mid; 1061 } 1062 else if (compVal > 0){ 1063 low = mid; 1064 } 1065 else /*we found it*/{ 1066 return getHostID(&gPosixIDmap[mid], posixID, status); 1067 } 1068 oldmid = mid; 1069 } 1070 1071 /* 1072 * Sometimes we can't do a binary search on posixID because some LCIDs 1073 * go to different locales. We hit one of those special cases. 1074 */ 1075 for (idx = 0; idx < gLocaleCount; idx++ ) { 1076 myStatus = U_ZERO_ERROR; 1077 value = getHostID(&gPosixIDmap[idx], posixID, &myStatus); 1078 if (myStatus == U_ZERO_ERROR) { 1079 return value; 1080 } 1081 else if (myStatus == U_USING_FALLBACK_WARNING) { 1082 fallbackValue = value; 1083 } 1084 } 1085 1086 if (fallbackValue != (uint32_t)-1) { 1087 *status = U_USING_FALLBACK_WARNING; 1088 return fallbackValue; 1089 } 1090 1091 /* no match found */ 1092 *status = U_ILLEGAL_ARGUMENT_ERROR; 1093 return 0; /* return international (root) */ 1094 } 1095 1096
