1/* 2 ********************************************************************** 3 * Copyright (C) 1996-2010, 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 0 34#if defined(U_WINDOWS) && defined(_MSC_VER) && (_MSC_VER >= 1500) 35#define USE_WINDOWS_LOCALE_API 36#endif 37#endif 38 39#ifdef USE_WINDOWS_LOCALE_API 40#include <windows.h> 41#include <winnls.h> 42#endif 43 44/* 45 * Note: 46 * The mapping from Win32 locale ID numbers to POSIX locale strings should 47 * be the faster one. 48 * 49 * Many LCID values come from winnt.h 50 * Some also come from http://www.microsoft.com/globaldev/reference/lcid-all.mspx 51 */ 52 53/* 54//////////////////////////////////////////////// 55// 56// Internal Classes for LCID <--> POSIX Mapping 57// 58///////////////////////////////////////////////// 59*/ 60 61typedef struct ILcidPosixElement 62{ 63 const uint32_t hostID; 64 const char * const posixID; 65} ILcidPosixElement; 66 67typedef struct ILcidPosixMap 68{ 69 const uint32_t numRegions; 70 const struct ILcidPosixElement* const regionMaps; 71} ILcidPosixMap; 72 73 74/* 75///////////////////////////////////////////////// 76// 77// Easy macros to make the LCID <--> POSIX Mapping 78// 79///////////////////////////////////////////////// 80*/ 81 82/** 83 * The standard one language/one country mapping for LCID. 84 * The first element must be the language, and the following 85 * elements are the language with the country. 86 * @param hostID LCID in host format such as 0x044d 87 * @param languageID posix ID of just the language such as 'de' 88 * @param posixID posix ID of the language_TERRITORY such as 'de_CH' 89 */ 90#define ILCID_POSIX_ELEMENT_ARRAY(hostID, languageID, posixID) \ 91static const ILcidPosixElement locmap_ ## languageID [] = { \ 92 {LANGUAGE_LCID(hostID), #languageID}, /* parent locale */ \ 93 {hostID, #posixID}, \ 94}; 95 96/** 97 * Define a subtable by ID 98 * @param id the POSIX ID, either a language or language_TERRITORY 99 */ 100#define ILCID_POSIX_SUBTABLE(id) \ 101static const ILcidPosixElement locmap_ ## id [] = 102 103 104/** 105 * Create the map for the posixID. This macro supposes that the language string 106 * name is the same as the global variable name, and that the first element 107 * in the ILcidPosixElement is just the language. 108 * @param _posixID the full POSIX ID for this entry. 109 */ 110#define ILCID_POSIX_MAP(_posixID) \ 111 {sizeof(locmap_ ## _posixID)/sizeof(ILcidPosixElement), locmap_ ## _posixID} 112 113/* 114//////////////////////////////////////////// 115// 116// Create the table of LCID to POSIX Mapping 117// None of it should be dynamically created. 118// 119// Keep static locale variables inside the function so that 120// it can be created properly during static init. 121// 122// Note: This table should be updated periodically. Check the National Lanaguage Support API Reference Website. 123// Microsoft is moving away from LCID in favor of locale name as of Vista. This table needs to be 124// maintained for support of older Windows version. 125// Update: Windows 7 (091130) 126//////////////////////////////////////////// 127*/ 128 129ILCID_POSIX_ELEMENT_ARRAY(0x0436, af, af_ZA) 130 131ILCID_POSIX_SUBTABLE(ar) { 132 {0x01, "ar"}, 133 {0x3801, "ar_AE"}, 134 {0x3c01, "ar_BH"}, 135 {0x1401, "ar_DZ"}, 136 {0x0c01, "ar_EG"}, 137 {0x0801, "ar_IQ"}, 138 {0x2c01, "ar_JO"}, 139 {0x3401, "ar_KW"}, 140 {0x3001, "ar_LB"}, 141 {0x1001, "ar_LY"}, 142 {0x1801, "ar_MA"}, 143 {0x2001, "ar_OM"}, 144 {0x4001, "ar_QA"}, 145 {0x0401, "ar_SA"}, 146 {0x2801, "ar_SY"}, 147 {0x1c01, "ar_TN"}, 148 {0x2401, "ar_YE"} 149}; 150 151ILCID_POSIX_ELEMENT_ARRAY(0x044d, as, as_IN) 152ILCID_POSIX_ELEMENT_ARRAY(0x045e, am, am_ET) 153ILCID_POSIX_ELEMENT_ARRAY(0x047a, arn,arn_CL) 154 155ILCID_POSIX_SUBTABLE(az) { 156 {0x2c, "az"}, 157 {0x082c, "az_Cyrl_AZ"}, /* Cyrillic based */ 158 {0x742c, "az_Cyrl"}, /* Cyrillic based */ 159 {0x042c, "az_Latn_AZ"}, /* Latin based */ 160 {0x782c, "az_Latn"}, /* Latin based */ 161 {0x042c, "az_AZ"} /* Latin based */ 162}; 163 164ILCID_POSIX_ELEMENT_ARRAY(0x046d, ba, ba_RU) 165ILCID_POSIX_ELEMENT_ARRAY(0x0423, be, be_BY) 166 167ILCID_POSIX_SUBTABLE(ber) { 168 {0x5f, "ber"}, 169 {0x045f, "ber_Arab_DZ"}, 170 {0x045f, "ber_Arab"}, 171 {0x085f, "ber_Latn_DZ"}, 172 {0x085f, "ber_Latn"} 173}; 174 175ILCID_POSIX_ELEMENT_ARRAY(0x0402, bg, bg_BG) 176 177ILCID_POSIX_SUBTABLE(bn) { 178 {0x45, "bn"}, 179 {0x0845, "bn_BD"}, 180 {0x0445, "bn_IN"} 181}; 182 183ILCID_POSIX_SUBTABLE(bo) { 184 {0x51, "bo"}, 185 {0x0851, "bo_BT"}, 186 {0x0451, "bo_CN"} 187}; 188 189ILCID_POSIX_ELEMENT_ARRAY(0x047e, br, br_FR) 190ILCID_POSIX_ELEMENT_ARRAY(0x0403, ca, ca_ES) 191ILCID_POSIX_ELEMENT_ARRAY(0x0483, co, co_FR) 192ILCID_POSIX_ELEMENT_ARRAY(0x045c, chr,chr_US) 193 194/* Declared as cs_CZ to get around compiler errors on z/OS, which defines cs as a function */ 195ILCID_POSIX_ELEMENT_ARRAY(0x0405, cs, cs_CZ) 196 197ILCID_POSIX_ELEMENT_ARRAY(0x0452, cy, cy_GB) 198ILCID_POSIX_ELEMENT_ARRAY(0x0406, da, da_DK) 199 200ILCID_POSIX_SUBTABLE(de) { 201 {0x07, "de"}, 202 {0x0c07, "de_AT"}, 203 {0x0807, "de_CH"}, 204 {0x0407, "de_DE"}, 205 {0x1407, "de_LI"}, 206 {0x1007, "de_LU"}, 207 {0x10407,"de_DE@collation=phonebook"}, /*This is really de_DE_PHONEBOOK on Windows*/ 208 {0x10407,"de@collation=phonebook"} /*This is really de_DE_PHONEBOOK on Windows*/ 209}; 210 211ILCID_POSIX_ELEMENT_ARRAY(0x0465, dv, dv_MV) 212ILCID_POSIX_ELEMENT_ARRAY(0x0408, el, el_GR) 213 214ILCID_POSIX_SUBTABLE(en) { 215 {0x09, "en"}, 216 {0x0c09, "en_AU"}, 217 {0x2809, "en_BZ"}, 218 {0x1009, "en_CA"}, 219 {0x0809, "en_GB"}, 220 {0x1809, "en_IE"}, 221 {0x4009, "en_IN"}, 222 {0x2009, "en_JM"}, 223 {0x4409, "en_MY"}, 224 {0x1409, "en_NZ"}, 225 {0x3409, "en_PH"}, 226 {0x4809, "en_SG"}, 227 {0x2C09, "en_TT"}, 228 {0x0409, "en_US"}, 229 {0x007f, "en_US_POSIX"}, /* duplicate for roundtripping */ 230 {0x2409, "en_VI"}, /* Virgin Islands AKA Caribbean Islands (en_CB). */ 231 {0x1c09, "en_ZA"}, 232 {0x3009, "en_ZW"}, 233 {0x2409, "en_029"}, 234 {0x0409, "en_AS"}, /* Alias for en_US. Leave last. */ 235 {0x0409, "en_GU"}, /* Alias for en_US. Leave last. */ 236 {0x0409, "en_MH"}, /* Alias for en_US. Leave last. */ 237 {0x0409, "en_MP"}, /* Alias for en_US. Leave last. */ 238 {0x0409, "en_UM"} /* Alias for en_US. Leave last. */ 239}; 240 241ILCID_POSIX_SUBTABLE(en_US_POSIX) { 242 {0x007f, "en_US_POSIX"} /* duplicate for roundtripping */ 243}; 244 245ILCID_POSIX_SUBTABLE(es) { 246 {0x0a, "es"}, 247 {0x2c0a, "es_AR"}, 248 {0x400a, "es_BO"}, 249 {0x340a, "es_CL"}, 250 {0x240a, "es_CO"}, 251 {0x140a, "es_CR"}, 252 {0x1c0a, "es_DO"}, 253 {0x300a, "es_EC"}, 254 {0x0c0a, "es_ES"}, /*Modern sort.*/ 255 {0x100a, "es_GT"}, 256 {0x480a, "es_HN"}, 257 {0x080a, "es_MX"}, 258 {0x4c0a, "es_NI"}, 259 {0x180a, "es_PA"}, 260 {0x280a, "es_PE"}, 261 {0x500a, "es_PR"}, 262 {0x3c0a, "es_PY"}, 263 {0x440a, "es_SV"}, 264 {0x540a, "es_US"}, 265 {0x380a, "es_UY"}, 266 {0x200a, "es_VE"}, 267 {0x040a, "es_ES@collation=traditional"}, 268 {0x040a, "es@collation=traditional"} 269}; 270 271ILCID_POSIX_ELEMENT_ARRAY(0x0425, et, et_EE) 272ILCID_POSIX_ELEMENT_ARRAY(0x042d, eu, eu_ES) 273 274/* ISO-639 doesn't distinguish between Persian and Dari.*/ 275ILCID_POSIX_SUBTABLE(fa) { 276 {0x29, "fa"}, 277 {0x0429, "fa_IR"}, /* Persian/Farsi (Iran) */ 278 {0x048c, "fa_AF"} /* Persian/Dari (Afghanistan) */ 279}; 280 281/* duplicate for roundtripping */ 282ILCID_POSIX_SUBTABLE(fa_AF) { 283 {0x8c, "fa_AF"}, /* Persian/Dari (Afghanistan) */ 284 {0x048c, "fa_AF"} /* Persian/Dari (Afghanistan) */ 285}; 286 287ILCID_POSIX_ELEMENT_ARRAY(0x040b, fi, fi_FI) 288ILCID_POSIX_ELEMENT_ARRAY(0x0464, fil,fil_PH) 289ILCID_POSIX_ELEMENT_ARRAY(0x0438, fo, fo_FO) 290 291ILCID_POSIX_SUBTABLE(fr) { 292 {0x0c, "fr"}, 293 {0x080c, "fr_BE"}, 294 {0x0c0c, "fr_CA"}, 295 {0x240c, "fr_CD"}, 296 {0x100c, "fr_CH"}, 297 {0x300c, "fr_CI"}, 298 {0x2c0c, "fr_CM"}, 299 {0x040c, "fr_FR"}, 300 {0x3c0c, "fr_HT"}, 301 {0x140c, "fr_LU"}, 302 {0x380c, "fr_MA"}, 303 {0x180c, "fr_MC"}, 304 {0x340c, "fr_ML"}, 305 {0x200c, "fr_RE"}, 306 {0x280c, "fr_SN"} 307}; 308 309ILCID_POSIX_ELEMENT_ARRAY(0x0462, fy, fy_NL) 310 311/* This LCID is really two different locales.*/ 312ILCID_POSIX_ELEMENT_ARRAY(0x083c, ga, ga_IE) /* Gaelic (Ireland) */ 313ILCID_POSIX_ELEMENT_ARRAY(0x0491, gd, gd_GB) /* Gaelic (Scotland) */ 314 315ILCID_POSIX_ELEMENT_ARRAY(0x0456, gl, gl_ES) 316ILCID_POSIX_ELEMENT_ARRAY(0x0447, gu, gu_IN) 317ILCID_POSIX_ELEMENT_ARRAY(0x0474, gn, gn_PY) 318ILCID_POSIX_ELEMENT_ARRAY(0x0484, gsw,gsw_FR) 319 320ILCID_POSIX_SUBTABLE(ha) { 321 {0x68, "ha"}, 322 {0x7c68, "ha_Latn"}, 323 {0x0468, "ha_Latn_NG"}, 324}; 325 326ILCID_POSIX_ELEMENT_ARRAY(0x0475, haw,haw_US) 327ILCID_POSIX_ELEMENT_ARRAY(0x040d, he, he_IL) 328ILCID_POSIX_ELEMENT_ARRAY(0x0439, hi, hi_IN) 329 330/* This LCID is really four different locales.*/ 331ILCID_POSIX_SUBTABLE(hr) { 332 {0x1a, "hr"}, 333 {0x141a, "bs_Latn_BA"}, /* Bosnian, Bosnia and Herzegovina */ 334 {0x681a, "bs_Latn"}, /* Bosnian, Bosnia and Herzegovina */ 335 {0x141a, "bs_BA"}, /* Bosnian, Bosnia and Herzegovina */ 336 {0x781a, "bs"}, /* Bosnian */ 337 {0x201a, "bs_Cyrl_BA"}, /* Bosnian, Bosnia and Herzegovina */ 338 {0x641a, "bs_Cyrl"}, /* Bosnian, Bosnia and Herzegovina */ 339 {0x101a, "hr_BA"}, /* Croatian in Bosnia */ 340 {0x041a, "hr_HR"}, /* Croatian*/ 341 {0x2c1a, "sr_Latn_ME"}, 342 {0x241a, "sr_Latn_RS"}, 343 {0x181a, "sr_Latn_BA"}, /* Serbo-Croatian in Bosnia */ 344 {0x081a, "sr_Latn_CS"}, /* Serbo-Croatian*/ 345 {0x701a, "sr_Latn"}, /* It's 0x1a or 0x081a, pick one to make the test program happy. */ 346 {0x1c1a, "sr_Cyrl_BA"}, /* Serbo-Croatian in Bosnia */ 347 {0x0c1a, "sr_Cyrl_CS"}, /* Serbian*/ 348 {0x301a, "sr_Cyrl_ME"}, 349 {0x281a, "sr_Cyrl_RS"}, 350 {0x6c1a, "sr_Cyrl"}, /* It's 0x1a or 0x0c1a, pick one to make the test program happy. */ 351 {0x7c1a, "sr"} /* In CLDR sr is sr_Cyrl. */ 352}; 353 354ILCID_POSIX_ELEMENT_ARRAY(0x040e, hu, hu_HU) 355ILCID_POSIX_ELEMENT_ARRAY(0x042b, hy, hy_AM) 356ILCID_POSIX_ELEMENT_ARRAY(0x0421, id, id_ID) 357ILCID_POSIX_ELEMENT_ARRAY(0x0470, ig, ig_NG) 358ILCID_POSIX_ELEMENT_ARRAY(0x0478, ii, ii_CN) 359ILCID_POSIX_ELEMENT_ARRAY(0x040f, is, is_IS) 360 361ILCID_POSIX_SUBTABLE(it) { 362 {0x10, "it"}, 363 {0x0810, "it_CH"}, 364 {0x0410, "it_IT"} 365}; 366 367ILCID_POSIX_SUBTABLE(iu) { 368 {0x5d, "iu"}, 369 {0x045d, "iu_Cans_CA"}, 370 {0x785d, "iu_Cans"}, 371 {0x085d, "iu_Latn_CA"}, 372 {0x7c5d, "iu_Latn"} 373}; 374 375ILCID_POSIX_ELEMENT_ARRAY(0x040d, iw, iw_IL) /*Left in for compatibility*/ 376ILCID_POSIX_ELEMENT_ARRAY(0x0411, ja, ja_JP) 377ILCID_POSIX_ELEMENT_ARRAY(0x0437, ka, ka_GE) 378ILCID_POSIX_ELEMENT_ARRAY(0x043f, kk, kk_KZ) 379ILCID_POSIX_ELEMENT_ARRAY(0x046f, kl, kl_GL) 380ILCID_POSIX_ELEMENT_ARRAY(0x0453, km, km_KH) 381ILCID_POSIX_ELEMENT_ARRAY(0x044b, kn, kn_IN) 382 383ILCID_POSIX_SUBTABLE(ko) { 384 {0x12, "ko"}, 385 {0x0812, "ko_KP"}, 386 {0x0412, "ko_KR"} 387}; 388 389ILCID_POSIX_ELEMENT_ARRAY(0x0457, kok, kok_IN) 390ILCID_POSIX_ELEMENT_ARRAY(0x0471, kr, kr_NG) 391 392ILCID_POSIX_SUBTABLE(ks) { /* We could add PK and CN too */ 393 {0x60, "ks"}, 394 {0x0860, "ks_IN"}, /* Documentation doesn't mention script */ 395 {0x0460, "ks_Arab_IN"} 396}; 397 398ILCID_POSIX_ELEMENT_ARRAY(0x0440, ky, ky_KG) /* Kyrgyz is spoken in Kyrgyzstan */ 399ILCID_POSIX_ELEMENT_ARRAY(0x0476, la, la_IT) /* TODO: Verify the country */ 400ILCID_POSIX_ELEMENT_ARRAY(0x046e, lb, lb_LU) 401ILCID_POSIX_ELEMENT_ARRAY(0x0454, lo, lo_LA) 402ILCID_POSIX_ELEMENT_ARRAY(0x0427, lt, lt_LT) 403ILCID_POSIX_ELEMENT_ARRAY(0x0426, lv, lv_LV) 404ILCID_POSIX_ELEMENT_ARRAY(0x0481, mi, mi_NZ) 405ILCID_POSIX_ELEMENT_ARRAY(0x042f, mk, mk_MK) 406ILCID_POSIX_ELEMENT_ARRAY(0x044c, ml, ml_IN) 407 408ILCID_POSIX_SUBTABLE(mn) { 409 {0x50, "mn"}, 410 {0x0450, "mn_MN"}, 411 {0x7c50, "mn_Mong"}, 412 {0x0850, "mn_Mong_CN"}, 413 {0x0850, "mn_CN"}, 414 {0x7850, "mn_Cyrl"} 415}; 416 417ILCID_POSIX_ELEMENT_ARRAY(0x0458, mni,mni_IN) 418ILCID_POSIX_ELEMENT_ARRAY(0x047c, moh,moh_CA) 419ILCID_POSIX_ELEMENT_ARRAY(0x044e, mr, mr_IN) 420 421ILCID_POSIX_SUBTABLE(ms) { 422 {0x3e, "ms"}, 423 {0x083e, "ms_BN"}, /* Brunei Darussalam*/ 424 {0x043e, "ms_MY"} /* Malaysia*/ 425}; 426 427ILCID_POSIX_ELEMENT_ARRAY(0x043a, mt, mt_MT) 428ILCID_POSIX_ELEMENT_ARRAY(0x0455, my, my_MM) 429 430ILCID_POSIX_SUBTABLE(ne) { 431 {0x61, "ne"}, 432 {0x0861, "ne_IN"}, /* India*/ 433 {0x0461, "ne_NP"} /* Nepal*/ 434}; 435 436ILCID_POSIX_SUBTABLE(nl) { 437 {0x13, "nl"}, 438 {0x0813, "nl_BE"}, 439 {0x0413, "nl_NL"} 440}; 441 442/* The "no" locale split into nb and nn. By default in ICU, "no" is nb.*/ 443ILCID_POSIX_SUBTABLE(no) { 444 {0x14, "no"}, /* really nb_NO */ 445 {0x7c14, "nb"}, /* really nb */ 446 {0x0414, "nb_NO"}, /* really nb_NO. Keep first in the 414 list. */ 447 {0x0414, "no_NO"}, /* really nb_NO */ 448 {0x0814, "nn_NO"}, /* really nn_NO. Keep first in the 814 list. */ 449 {0x7814, "nn"}, /* It's 0x14 or 0x814, pick one to make the test program happy. */ 450 {0x0814, "no_NO_NY"}/* really nn_NO */ 451}; 452 453ILCID_POSIX_ELEMENT_ARRAY(0x046c, nso,nso_ZA) /* TODO: Verify the ISO-639 code */ 454ILCID_POSIX_ELEMENT_ARRAY(0x0482, oc, oc_FR) 455ILCID_POSIX_ELEMENT_ARRAY(0x0472, om, om_ET) /* TODO: Verify the country */ 456 457/* Declared as or_IN to get around compiler errors*/ 458ILCID_POSIX_SUBTABLE(or_IN) { 459 {0x48, "or"}, 460 {0x0448, "or_IN"}, 461}; 462 463 464ILCID_POSIX_SUBTABLE(pa) { 465 {0x46, "pa"}, 466 {0x0446, "pa_IN"}, 467 {0x0846, "pa_PK"} 468}; 469 470ILCID_POSIX_ELEMENT_ARRAY(0x0415, pl, pl_PL) 471ILCID_POSIX_ELEMENT_ARRAY(0x0463, ps, ps_AF) 472 473ILCID_POSIX_SUBTABLE(pt) { 474 {0x16, "pt"}, 475 {0x0416, "pt_BR"}, 476 {0x0816, "pt_PT"} 477}; 478 479ILCID_POSIX_SUBTABLE(qu) { 480 {0x6b, "qu"}, 481 {0x046b, "qu_BO"}, 482 {0x086b, "qu_EC"}, 483 {0x0C6b, "qu_PE"} 484}; 485 486ILCID_POSIX_ELEMENT_ARRAY(0x0486, qut, qut_GT) /* qut is an ISO-639-3 code */ 487ILCID_POSIX_ELEMENT_ARRAY(0x0417, rm, rm_CH) 488ILCID_POSIX_ELEMENT_ARRAY(0x0418, ro, ro_RO) 489 490ILCID_POSIX_SUBTABLE(root) { 491 {0x00, "root"} 492}; 493 494ILCID_POSIX_ELEMENT_ARRAY(0x0419, ru, ru_RU) 495ILCID_POSIX_ELEMENT_ARRAY(0x0487, rw, rw_RW) 496ILCID_POSIX_ELEMENT_ARRAY(0x044f, sa, sa_IN) 497ILCID_POSIX_ELEMENT_ARRAY(0x0485, sah,sah_RU) 498 499ILCID_POSIX_SUBTABLE(sd) { 500 {0x59, "sd"}, 501 {0x0459, "sd_IN"}, 502 {0x0859, "sd_PK"} 503}; 504 505ILCID_POSIX_SUBTABLE(se) { 506 {0x3b, "se"}, 507 {0x0c3b, "se_FI"}, 508 {0x043b, "se_NO"}, 509 {0x083b, "se_SE"}, 510 {0x783b, "sma"}, 511 {0x183b, "sma_NO"}, 512 {0x1c3b, "sma_SE"}, 513 {0x7c3b, "smj"}, 514 {0x703b, "smn"}, 515 {0x743b, "sms"}, 516 {0x103b, "smj_NO"}, 517 {0x143b, "smj_SE"}, 518 {0x243b, "smn_FI"}, 519 {0x203b, "sms_FI"}, 520}; 521 522ILCID_POSIX_ELEMENT_ARRAY(0x045b, si, si_LK) 523ILCID_POSIX_ELEMENT_ARRAY(0x041b, sk, sk_SK) 524ILCID_POSIX_ELEMENT_ARRAY(0x0424, sl, sl_SI) 525ILCID_POSIX_ELEMENT_ARRAY(0x0477, so, so_ET) /* TODO: Verify the country */ 526ILCID_POSIX_ELEMENT_ARRAY(0x041c, sq, sq_AL) 527 528ILCID_POSIX_SUBTABLE(sv) { 529 {0x1d, "sv"}, 530 {0x081d, "sv_FI"}, 531 {0x041d, "sv_SE"} 532}; 533 534ILCID_POSIX_ELEMENT_ARRAY(0x0441, sw, sw_KE) 535ILCID_POSIX_ELEMENT_ARRAY(0x045A, syr, syr_SY) 536ILCID_POSIX_ELEMENT_ARRAY(0x0449, ta, ta_IN) 537ILCID_POSIX_ELEMENT_ARRAY(0x044a, te, te_IN) 538 539/* Cyrillic based by default */ 540ILCID_POSIX_SUBTABLE(tg) { 541 {0x28, "tg"}, 542 {0x7c28, "tg_Cyrl"}, 543 {0x0428, "tg_Cyrl_TJ"} 544}; 545 546ILCID_POSIX_ELEMENT_ARRAY(0x041e, th, th_TH) 547 548ILCID_POSIX_SUBTABLE(ti) { 549 {0x73, "ti"}, 550 {0x0873, "ti_ER"}, 551 {0x0473, "ti_ET"} 552}; 553 554ILCID_POSIX_ELEMENT_ARRAY(0x0442, tk, tk_TM) 555ILCID_POSIX_ELEMENT_ARRAY(0x0432, tn, tn_BW) 556ILCID_POSIX_ELEMENT_ARRAY(0x041f, tr, tr_TR) 557ILCID_POSIX_ELEMENT_ARRAY(0x0444, tt, tt_RU) 558 559ILCID_POSIX_SUBTABLE(tzm) { 560 {0x5f, "tzm"}, 561 {0x7c5f, "tzm_Latn"}, 562 {0x085f, "tzm_Latn_DZ"} 563}; 564 565ILCID_POSIX_ELEMENT_ARRAY(0x0480, ug, ug_CN) 566ILCID_POSIX_ELEMENT_ARRAY(0x0422, uk, uk_UA) 567 568ILCID_POSIX_SUBTABLE(ur) { 569 {0x20, "ur"}, 570 {0x0820, "ur_IN"}, 571 {0x0420, "ur_PK"} 572}; 573 574ILCID_POSIX_SUBTABLE(uz) { 575 {0x43, "uz"}, 576 {0x0843, "uz_Cyrl_UZ"}, /* Cyrillic based */ 577 {0x7843, "uz_Cyrl"}, /* Cyrillic based */ 578 {0x0843, "uz_UZ"}, /* Cyrillic based */ 579 {0x0443, "uz_Latn_UZ"}, /* Latin based */ 580 {0x7c43, "uz_Latn"} /* Latin based */ 581}; 582 583ILCID_POSIX_ELEMENT_ARRAY(0x0433, ve, ve_ZA) /* TODO: Verify the country */ 584ILCID_POSIX_ELEMENT_ARRAY(0x042a, vi, vi_VN) 585 586ILCID_POSIX_SUBTABLE(wen) { 587 {0x2E, "wen"}, 588 {0x042E, "wen_DE"}, 589 {0x042E, "hsb_DE"}, 590 {0x082E, "dsb_DE"}, 591 {0x7C2E, "dsb"}, 592 {0x2E, "hsb"} 593}; 594 595ILCID_POSIX_ELEMENT_ARRAY(0x0488, wo, wo_SN) 596ILCID_POSIX_ELEMENT_ARRAY(0x0434, xh, xh_ZA) 597ILCID_POSIX_ELEMENT_ARRAY(0x046a, yo, yo_NG) 598 599ILCID_POSIX_SUBTABLE(zh) { 600 {0x0004, "zh_Hans"}, 601 {0x7804, "zh"}, 602 {0x0804, "zh_CN"}, 603 {0x0804, "zh_Hans_CN"}, 604 {0x0c04, "zh_Hant_HK"}, 605 {0x0c04, "zh_HK"}, 606 {0x1404, "zh_Hant_MO"}, 607 {0x1404, "zh_MO"}, 608 {0x1004, "zh_Hans_SG"}, 609 {0x1004, "zh_SG"}, 610 {0x0404, "zh_Hant_TW"}, 611 {0x7c04, "zh_Hant"}, 612 {0x0404, "zh_TW"}, 613 {0x30404,"zh_Hant_TW"}, /* Bopomofo order */ 614 {0x30404,"zh_TW"}, /* Bopomofo order */ 615 {0x20004,"zh@collation=stroke"}, 616 {0x20404,"zh_Hant@collation=stroke"}, 617 {0x20404,"zh_Hant_TW@collation=stroke"}, 618 {0x20404,"zh_TW@collation=stroke"}, 619 {0x20804,"zh_Hans@collation=stroke"}, 620 {0x20804,"zh_Hans_CN@collation=stroke"}, 621 {0x20804,"zh_CN@collation=stroke"} 622}; 623 624ILCID_POSIX_ELEMENT_ARRAY(0x0435, zu, zu_ZA) 625 626/* This must be static and grouped by LCID. */ 627 628/* non-existent ISO-639-2 codes */ 629/* 6300x466 Edo 6310x467 Fulfulde - Nigeria 6320x486 K'iche - Guatemala 6330x430 Sutu 634*/ 635static const ILcidPosixMap gPosixIDmap[] = { 636 ILCID_POSIX_MAP(af), /* af Afrikaans 0x36 */ 637 ILCID_POSIX_MAP(am), /* am Amharic 0x5e */ 638 ILCID_POSIX_MAP(ar), /* ar Arabic 0x01 */ 639 ILCID_POSIX_MAP(arn), /* arn Araucanian/Mapudungun 0x7a */ 640 ILCID_POSIX_MAP(as), /* as Assamese 0x4d */ 641 ILCID_POSIX_MAP(az), /* az Azerbaijani 0x2c */ 642 ILCID_POSIX_MAP(ba), /* ba Bashkir 0x6d */ 643 ILCID_POSIX_MAP(be), /* be Belarusian 0x23 */ 644/* ILCID_POSIX_MAP(ber), ber Berber/Tamazight 0x5f */ 645 ILCID_POSIX_MAP(bg), /* bg Bulgarian 0x02 */ 646 ILCID_POSIX_MAP(bn), /* bn Bengali; Bangla 0x45 */ 647 ILCID_POSIX_MAP(bo), /* bo Tibetan 0x51 */ 648 ILCID_POSIX_MAP(br), /* br Breton 0x7e */ 649 ILCID_POSIX_MAP(ca), /* ca Catalan 0x03 */ 650 ILCID_POSIX_MAP(chr), /* chr Cherokee 0x5c */ 651 ILCID_POSIX_MAP(co), /* co Corsican 0x83 */ 652 ILCID_POSIX_MAP(cs), /* cs Czech 0x05 */ 653 ILCID_POSIX_MAP(cy), /* cy Welsh 0x52 */ 654 ILCID_POSIX_MAP(da), /* da Danish 0x06 */ 655 ILCID_POSIX_MAP(de), /* de German 0x07 */ 656 ILCID_POSIX_MAP(dv), /* dv Divehi 0x65 */ 657 ILCID_POSIX_MAP(el), /* el Greek 0x08 */ 658 ILCID_POSIX_MAP(en), /* en English 0x09 */ 659 ILCID_POSIX_MAP(en_US_POSIX), /* invariant 0x7f */ 660 ILCID_POSIX_MAP(es), /* es Spanish 0x0a */ 661 ILCID_POSIX_MAP(et), /* et Estonian 0x25 */ 662 ILCID_POSIX_MAP(eu), /* eu Basque 0x2d */ 663 ILCID_POSIX_MAP(fa), /* fa Persian/Farsi 0x29 */ 664 ILCID_POSIX_MAP(fa_AF), /* fa Persian/Dari 0x8c */ 665 ILCID_POSIX_MAP(fi), /* fi Finnish 0x0b */ 666 ILCID_POSIX_MAP(fil), /* fil Filipino 0x64 */ 667 ILCID_POSIX_MAP(fo), /* fo Faroese 0x38 */ 668 ILCID_POSIX_MAP(fr), /* fr French 0x0c */ 669 ILCID_POSIX_MAP(fy), /* fy Frisian 0x62 */ 670 ILCID_POSIX_MAP(ga), /* * Gaelic (Ireland,Scotland) 0x3c */ 671 ILCID_POSIX_MAP(gd), /* gd Gaelic (United Kingdom) 0x91 */ 672 ILCID_POSIX_MAP(gl), /* gl Galician 0x56 */ 673 ILCID_POSIX_MAP(gn), /* gn Guarani 0x74 */ 674 ILCID_POSIX_MAP(gsw), /* gsw Alemanic/Alsatian/Swiss German 0x84 */ 675 ILCID_POSIX_MAP(gu), /* gu Gujarati 0x47 */ 676 ILCID_POSIX_MAP(ha), /* ha Hausa 0x68 */ 677 ILCID_POSIX_MAP(haw), /* haw Hawaiian 0x75 */ 678 ILCID_POSIX_MAP(he), /* he Hebrew (formerly iw) 0x0d */ 679 ILCID_POSIX_MAP(hi), /* hi Hindi 0x39 */ 680 ILCID_POSIX_MAP(hr), /* * Croatian and others 0x1a */ 681 ILCID_POSIX_MAP(hu), /* hu Hungarian 0x0e */ 682 ILCID_POSIX_MAP(hy), /* hy Armenian 0x2b */ 683 ILCID_POSIX_MAP(id), /* id Indonesian (formerly in) 0x21 */ 684 ILCID_POSIX_MAP(ig), /* ig Igbo 0x70 */ 685 ILCID_POSIX_MAP(ii), /* ii Sichuan Yi 0x78 */ 686 ILCID_POSIX_MAP(is), /* is Icelandic 0x0f */ 687 ILCID_POSIX_MAP(it), /* it Italian 0x10 */ 688 ILCID_POSIX_MAP(iu), /* iu Inuktitut 0x5d */ 689 ILCID_POSIX_MAP(iw), /* iw Hebrew 0x0d */ 690 ILCID_POSIX_MAP(ja), /* ja Japanese 0x11 */ 691 ILCID_POSIX_MAP(ka), /* ka Georgian 0x37 */ 692 ILCID_POSIX_MAP(kk), /* kk Kazakh 0x3f */ 693 ILCID_POSIX_MAP(kl), /* kl Kalaallisut 0x6f */ 694 ILCID_POSIX_MAP(km), /* km Khmer 0x53 */ 695 ILCID_POSIX_MAP(kn), /* kn Kannada 0x4b */ 696 ILCID_POSIX_MAP(ko), /* ko Korean 0x12 */ 697 ILCID_POSIX_MAP(kok), /* kok Konkani 0x57 */ 698 ILCID_POSIX_MAP(kr), /* kr Kanuri 0x71 */ 699 ILCID_POSIX_MAP(ks), /* ks Kashmiri 0x60 */ 700 ILCID_POSIX_MAP(ky), /* ky Kyrgyz 0x40 */ 701 ILCID_POSIX_MAP(lb), /* lb Luxembourgish 0x6e */ 702 ILCID_POSIX_MAP(la), /* la Latin 0x76 */ 703 ILCID_POSIX_MAP(lo), /* lo Lao 0x54 */ 704 ILCID_POSIX_MAP(lt), /* lt Lithuanian 0x27 */ 705 ILCID_POSIX_MAP(lv), /* lv Latvian, Lettish 0x26 */ 706 ILCID_POSIX_MAP(mi), /* mi Maori 0x81 */ 707 ILCID_POSIX_MAP(mk), /* mk Macedonian 0x2f */ 708 ILCID_POSIX_MAP(ml), /* ml Malayalam 0x4c */ 709 ILCID_POSIX_MAP(mn), /* mn Mongolian 0x50 */ 710 ILCID_POSIX_MAP(mni), /* mni Manipuri 0x58 */ 711 ILCID_POSIX_MAP(moh), /* moh Mohawk 0x7c */ 712 ILCID_POSIX_MAP(mr), /* mr Marathi 0x4e */ 713 ILCID_POSIX_MAP(ms), /* ms Malay 0x3e */ 714 ILCID_POSIX_MAP(mt), /* mt Maltese 0x3a */ 715 ILCID_POSIX_MAP(my), /* my Burmese 0x55 */ 716/* ILCID_POSIX_MAP(nb), // no Norwegian 0x14 */ 717 ILCID_POSIX_MAP(ne), /* ne Nepali 0x61 */ 718 ILCID_POSIX_MAP(nl), /* nl Dutch 0x13 */ 719/* ILCID_POSIX_MAP(nn), // no Norwegian 0x14 */ 720 ILCID_POSIX_MAP(no), /* * Norwegian 0x14 */ 721 ILCID_POSIX_MAP(nso), /* nso Sotho, Northern (Sepedi dialect) 0x6c */ 722 ILCID_POSIX_MAP(oc), /* oc Occitan 0x82 */ 723 ILCID_POSIX_MAP(om), /* om Oromo 0x72 */ 724 ILCID_POSIX_MAP(or_IN), /* or Oriya 0x48 */ 725 ILCID_POSIX_MAP(pa), /* pa Punjabi 0x46 */ 726 ILCID_POSIX_MAP(pl), /* pl Polish 0x15 */ 727 ILCID_POSIX_MAP(ps), /* ps Pashto 0x63 */ 728 ILCID_POSIX_MAP(pt), /* pt Portuguese 0x16 */ 729 ILCID_POSIX_MAP(qu), /* qu Quechua 0x6B */ 730 ILCID_POSIX_MAP(qut), /* qut K'iche 0x86 */ 731 ILCID_POSIX_MAP(rm), /* rm Raeto-Romance/Romansh 0x17 */ 732 ILCID_POSIX_MAP(ro), /* ro Romanian 0x18 */ 733 ILCID_POSIX_MAP(root), /* root 0x00 */ 734 ILCID_POSIX_MAP(ru), /* ru Russian 0x19 */ 735 ILCID_POSIX_MAP(rw), /* rw Kinyarwanda 0x87 */ 736 ILCID_POSIX_MAP(sa), /* sa Sanskrit 0x4f */ 737 ILCID_POSIX_MAP(sah), /* sah Yakut 0x85 */ 738 ILCID_POSIX_MAP(sd), /* sd Sindhi 0x59 */ 739 ILCID_POSIX_MAP(se), /* se Sami 0x3b */ 740/* ILCID_POSIX_MAP(sh), // sh Serbo-Croatian 0x1a */ 741 ILCID_POSIX_MAP(si), /* si Sinhalese 0x5b */ 742 ILCID_POSIX_MAP(sk), /* sk Slovak 0x1b */ 743 ILCID_POSIX_MAP(sl), /* sl Slovenian 0x24 */ 744 ILCID_POSIX_MAP(so), /* so Somali 0x77 */ 745 ILCID_POSIX_MAP(sq), /* sq Albanian 0x1c */ 746/* ILCID_POSIX_MAP(sr), // sr Serbian 0x1a */ 747 ILCID_POSIX_MAP(sv), /* sv Swedish 0x1d */ 748 ILCID_POSIX_MAP(sw), /* sw Swahili 0x41 */ 749 ILCID_POSIX_MAP(syr), /* syr Syriac 0x5A */ 750 ILCID_POSIX_MAP(ta), /* ta Tamil 0x49 */ 751 ILCID_POSIX_MAP(te), /* te Telugu 0x4a */ 752 ILCID_POSIX_MAP(tg), /* tg Tajik 0x28 */ 753 ILCID_POSIX_MAP(th), /* th Thai 0x1e */ 754 ILCID_POSIX_MAP(ti), /* ti Tigrigna 0x73 */ 755 ILCID_POSIX_MAP(tk), /* tk Turkmen 0x42 */ 756 ILCID_POSIX_MAP(tn), /* tn Tswana 0x32 */ 757 ILCID_POSIX_MAP(tr), /* tr Turkish 0x1f */ 758 ILCID_POSIX_MAP(tt), /* tt Tatar 0x44 */ 759 ILCID_POSIX_MAP(tzm), /* tzm 0x5f */ 760 ILCID_POSIX_MAP(ug), /* ug Uighur 0x80 */ 761 ILCID_POSIX_MAP(uk), /* uk Ukrainian 0x22 */ 762 ILCID_POSIX_MAP(ur), /* ur Urdu 0x20 */ 763 ILCID_POSIX_MAP(uz), /* uz Uzbek 0x43 */ 764 ILCID_POSIX_MAP(ve), /* ve Venda 0x33 */ 765 ILCID_POSIX_MAP(vi), /* vi Vietnamese 0x2a */ 766 ILCID_POSIX_MAP(wen), /* wen Sorbian 0x2e */ 767 ILCID_POSIX_MAP(wo), /* wo Wolof 0x88 */ 768 ILCID_POSIX_MAP(xh), /* xh Xhosa 0x34 */ 769 ILCID_POSIX_MAP(yo), /* yo Yoruba 0x6a */ 770 ILCID_POSIX_MAP(zh), /* zh Chinese 0x04 */ 771 ILCID_POSIX_MAP(zu), /* zu Zulu 0x35 */ 772}; 773 774static const uint32_t gLocaleCount = sizeof(gPosixIDmap)/sizeof(ILcidPosixMap); 775 776/** 777 * Do not call this function. It is called by hostID. 778 * The function is not private because this struct must stay as a C struct, 779 * and this is an internal class. 780 */ 781static int32_t 782idCmp(const char* id1, const char* id2) 783{ 784 int32_t diffIdx = 0; 785 while (*id1 == *id2 && *id1 != 0) { 786 diffIdx++; 787 id1++; 788 id2++; 789 } 790 return diffIdx; 791} 792 793/** 794 * Searches for a Windows LCID 795 * 796 * @param posixid the Posix style locale id. 797 * @param status gets set to U_ILLEGAL_ARGUMENT_ERROR when the Posix ID has 798 * no equivalent Windows LCID. 799 * @return the LCID 800 */ 801static uint32_t 802getHostID(const ILcidPosixMap *this_0, const char* posixID, UErrorCode* status) 803{ 804 int32_t bestIdx = 0; 805 int32_t bestIdxDiff = 0; 806 int32_t posixIDlen = (int32_t)uprv_strlen(posixID); 807 uint32_t idx; 808 809 for (idx = 0; idx < this_0->numRegions; idx++ ) { 810 int32_t sameChars = idCmp(posixID, this_0->regionMaps[idx].posixID); 811 if (sameChars > bestIdxDiff && this_0->regionMaps[idx].posixID[sameChars] == 0) { 812 if (posixIDlen == sameChars) { 813 /* Exact match */ 814 return this_0->regionMaps[idx].hostID; 815 } 816 bestIdxDiff = sameChars; 817 bestIdx = idx; 818 } 819 } 820 /* We asked for something unusual, like en_ZZ, and we try to return the number for the same language. */ 821 /* We also have to make sure that sid and si and similar string subsets don't match. */ 822 if ((posixID[bestIdxDiff] == '_' || posixID[bestIdxDiff] == '@') 823 && this_0->regionMaps[bestIdx].posixID[bestIdxDiff] == 0) 824 { 825 *status = U_USING_FALLBACK_WARNING; 826 return this_0->regionMaps[bestIdx].hostID; 827 } 828 829 /*no match found */ 830 *status = U_ILLEGAL_ARGUMENT_ERROR; 831 return this_0->regionMaps->hostID; 832} 833 834static const char* 835getPosixID(const ILcidPosixMap *this_0, uint32_t hostID) 836{ 837 uint32_t i; 838 for (i = 0; i <= this_0->numRegions; i++) 839 { 840 if (this_0->regionMaps[i].hostID == hostID) 841 { 842 return this_0->regionMaps[i].posixID; 843 } 844 } 845 846 /* If you get here, then no matching region was found, 847 so return the language id with the wild card region. */ 848 return this_0->regionMaps[0].posixID; 849} 850 851/* 852////////////////////////////////////// 853// 854// LCID --> POSIX 855// 856///////////////////////////////////// 857*/ 858#ifdef USE_WINDOWS_LOCALE_API 859/* 860 * Change the tag separator from '-' to '_' 861 */ 862#define FIX_LOCALE_ID_TAG_SEPARATOR(buffer, len, i) \ 863 for(i = 0; i < len; i++) \ 864 if (buffer[i] == '-') buffer[i] = '_'; 865 866/* 867 * Various language tags needs to be changed: 868 * quz -> qu 869 * prs -> fa 870 */ 871#define FIX_LANGUAGE_ID_TAG(buffer, len) \ 872 if (len >= 3) { \ 873 if (buffer[0] == 'q' && buffer[1] == 'u' && buffer[2] == 'z') {\ 874 buffer[2] = 0; \ 875 uprv_strcat(buffer, buffer+3); \ 876 } else if (buffer[0] == 'p' && buffer[1] == 'r' && buffer[2] == 's') {\ 877 buffer[0] = 'f'; buffer[1] = 'a'; buffer[2] = 0; \ 878 uprv_strcat(buffer, buffer+3); \ 879 } \ 880 } 881 882static char gPosixFromLCID[ULOC_FULLNAME_CAPACITY]; 883#endif 884U_CAPI const char * 885uprv_convertToPosix(uint32_t hostid, UErrorCode* status) 886{ 887 uint16_t langID; 888 uint32_t localeIndex; 889#ifdef USE_WINDOWS_LOCALE_API 890 int32_t ret = 0; 891 892 uprv_memset(gPosixFromLCID, 0, sizeof(gPosixFromLCID)); 893 894 ret = GetLocaleInfoA(hostid, LOCALE_SNAME, (LPSTR)gPosixFromLCID, sizeof(gPosixFromLCID)); 895 if (ret > 1) { 896 FIX_LOCALE_ID_TAG_SEPARATOR(gPosixFromLCID, ret, localeIndex) 897 FIX_LANGUAGE_ID_TAG(gPosixFromLCID, ret) 898 899 return gPosixFromLCID; 900 } 901#endif 902 langID = LANGUAGE_LCID(hostid); 903 904 for (localeIndex = 0; localeIndex < gLocaleCount; localeIndex++) 905 { 906 if (langID == gPosixIDmap[localeIndex].regionMaps->hostID) 907 { 908 return getPosixID(&gPosixIDmap[localeIndex], hostid); 909 } 910 } 911 912 /* no match found */ 913 *status = U_ILLEGAL_ARGUMENT_ERROR; 914 return NULL; 915} 916 917/* 918////////////////////////////////////// 919// 920// POSIX --> LCID 921// This should only be called from uloc_getLCID. 922// The locale ID must be in canonical form. 923// langID is separate so that this file doesn't depend on the uloc_* API. 924// 925///////////////////////////////////// 926*/ 927 928U_CAPI uint32_t 929uprv_convertToLCID(const char *langID, const char* posixID, UErrorCode* status) 930{ 931 932 uint32_t low = 0; 933 uint32_t high = gLocaleCount; 934 uint32_t mid = high; 935 uint32_t oldmid = 0; 936 int32_t compVal; 937 938 uint32_t value = 0; 939 uint32_t fallbackValue = (uint32_t)-1; 940 UErrorCode myStatus; 941 uint32_t idx; 942 943 /* Check for incomplete id. */ 944 if (!langID || !posixID || uprv_strlen(langID) < 2 || uprv_strlen(posixID) < 2) { 945 return 0; 946 } 947 948 /*Binary search for the map entry for normal cases */ 949 950 while (high > low) /*binary search*/{ 951 952 mid = (high+low) >> 1; /*Finds median*/ 953 954 if (mid == oldmid) 955 break; 956 957 compVal = uprv_strcmp(langID, gPosixIDmap[mid].regionMaps->posixID); 958 if (compVal < 0){ 959 high = mid; 960 } 961 else if (compVal > 0){ 962 low = mid; 963 } 964 else /*we found it*/{ 965 return getHostID(&gPosixIDmap[mid], posixID, status); 966 } 967 oldmid = mid; 968 } 969 970 /* 971 * Sometimes we can't do a binary search on posixID because some LCIDs 972 * go to different locales. We hit one of those special cases. 973 */ 974 for (idx = 0; idx < gLocaleCount; idx++ ) { 975 myStatus = U_ZERO_ERROR; 976 value = getHostID(&gPosixIDmap[idx], posixID, &myStatus); 977 if (myStatus == U_ZERO_ERROR) { 978 return value; 979 } 980 else if (myStatus == U_USING_FALLBACK_WARNING) { 981 fallbackValue = value; 982 } 983 } 984 985 if (fallbackValue != (uint32_t)-1) { 986 *status = U_USING_FALLBACK_WARNING; 987 return fallbackValue; 988 } 989 990 /* no match found */ 991 *status = U_ILLEGAL_ARGUMENT_ERROR; 992 return 0; /* return international (root) */ 993} 994 995