The Tor Browser / annotate

intl/icu/source/common/locmap.c@6474c204b198 (annotated)

intl/icu/source/common/locmap.c

Wed, 31 Dec 2014 06:09:35 +0100

author
Michael Schloh von Bennewitz <michael@schloh.com>
date
Wed, 31 Dec 2014 06:09:35 +0100
changeset 0
6474c204b198
permissions
-rw-r--r--

Cloned upstream origin tor-browser at tor-browser-31.3.0esr-4.5-1-build1
revision ID fc1c9ff7c1b2defdbc039f12214767608f46423f for hacking purpose.

michael@0 1 /*
michael@0 2 **********************************************************************
michael@0 3 * Copyright (C) 1996-2013, International Business Machines
michael@0 4 * Corporation and others. All Rights Reserved.
michael@0 5 **********************************************************************
michael@0 6 *
michael@0 7 * Provides functionality for mapping between
michael@0 8 * LCID and Posix IDs or ICU locale to codepage
michael@0 9 *
michael@0 10 * Note: All classes and code in this file are
michael@0 11 * intended for internal use only.
michael@0 12 *
michael@0 13 * Methods of interest:
michael@0 14 * unsigned long convertToLCID(const char*);
michael@0 15 * const char* convertToPosix(unsigned long);
michael@0 16 *
michael@0 17 * Kathleen Wilson, 4/30/96
michael@0 18 *
michael@0 19 * Date Name Description
michael@0 20 * 3/11/97 aliu Fixed off-by-one bug in assignment operator. Added
michael@0 21 * setId() method and safety check against
michael@0 22 * MAX_ID_LENGTH.
michael@0 23 * 04/23/99 stephen Added C wrapper for convertToPosix.
michael@0 24 * 09/18/00 george Removed the memory leaks.
michael@0 25 * 08/23/01 george Convert to C
michael@0 26 */
michael@0 27
michael@0 28 #include "locmap.h"
michael@0 29 #include "cstring.h"
michael@0 30 #include "cmemory.h"
michael@0 31
michael@0 32 #if U_PLATFORM == U_PF_WINDOWS && defined(_MSC_VER) && (_MSC_VER >= 1500)
michael@0 33 /*
michael@0 34 * TODO: It seems like we should widen this to
michael@0 35 * either U_PLATFORM_USES_ONLY_WIN32_API (includes MinGW)
michael@0 36 * or U_PLATFORM_HAS_WIN32_API (includes MinGW and Cygwin)
michael@0 37 * but those use gcc and won't have defined(_MSC_VER).
michael@0 38 * We might need to #include some Windows header and test for some version macro from there.
michael@0 39 * Or call some Windows function and see what it returns.
michael@0 40 */
michael@0 41 #define USE_WINDOWS_LOCALE_API
michael@0 42 #endif
michael@0 43
michael@0 44 #ifdef USE_WINDOWS_LOCALE_API
michael@0 45 #include <windows.h>
michael@0 46 #include <winnls.h>
michael@0 47 #endif
michael@0 48
michael@0 49 /*
michael@0 50 * Note:
michael@0 51 * The mapping from Win32 locale ID numbers to POSIX locale strings should
michael@0 52 * be the faster one.
michael@0 53 *
michael@0 54 * Many LCID values come from winnt.h
michael@0 55 * Some also come from http://www.microsoft.com/globaldev/reference/lcid-all.mspx
michael@0 56 */
michael@0 57
michael@0 58 /*
michael@0 59 ////////////////////////////////////////////////
michael@0 60 //
michael@0 61 // Internal Classes for LCID <--> POSIX Mapping
michael@0 62 //
michael@0 63 /////////////////////////////////////////////////
michael@0 64 */
michael@0 65
michael@0 66 typedef struct ILcidPosixElement
michael@0 67 {
michael@0 68 const uint32_t hostID;
michael@0 69 const char * const posixID;
michael@0 70 } ILcidPosixElement;
michael@0 71
michael@0 72 typedef struct ILcidPosixMap
michael@0 73 {
michael@0 74 const uint32_t numRegions;
michael@0 75 const struct ILcidPosixElement* const regionMaps;
michael@0 76 } ILcidPosixMap;
michael@0 77
michael@0 78
michael@0 79 /*
michael@0 80 /////////////////////////////////////////////////
michael@0 81 //
michael@0 82 // Easy macros to make the LCID <--> POSIX Mapping
michael@0 83 //
michael@0 84 /////////////////////////////////////////////////
michael@0 85 */
michael@0 86
michael@0 87 /**
michael@0 88 * The standard one language/one country mapping for LCID.
michael@0 89 * The first element must be the language, and the following
michael@0 90 * elements are the language with the country.
michael@0 91 * @param hostID LCID in host format such as 0x044d
michael@0 92 * @param languageID posix ID of just the language such as 'de'
michael@0 93 * @param posixID posix ID of the language_TERRITORY such as 'de_CH'
michael@0 94 */
michael@0 95 #define ILCID_POSIX_ELEMENT_ARRAY(hostID, languageID, posixID) \
michael@0 96 static const ILcidPosixElement locmap_ ## languageID [] = { \
michael@0 97 {LANGUAGE_LCID(hostID), #languageID}, /* parent locale */ \
michael@0 98 {hostID, #posixID}, \
michael@0 99 };
michael@0 100
michael@0 101 /**
michael@0 102 * Define a subtable by ID
michael@0 103 * @param id the POSIX ID, either a language or language_TERRITORY
michael@0 104 */
michael@0 105 #define ILCID_POSIX_SUBTABLE(id) \
michael@0 106 static const ILcidPosixElement locmap_ ## id [] =
michael@0 107
michael@0 108
michael@0 109 /**
michael@0 110 * Create the map for the posixID. This macro supposes that the language string
michael@0 111 * name is the same as the global variable name, and that the first element
michael@0 112 * in the ILcidPosixElement is just the language.
michael@0 113 * @param _posixID the full POSIX ID for this entry.
michael@0 114 */
michael@0 115 #define ILCID_POSIX_MAP(_posixID) \
michael@0 116 {sizeof(locmap_ ## _posixID)/sizeof(ILcidPosixElement), locmap_ ## _posixID}
michael@0 117
michael@0 118 /*
michael@0 119 ////////////////////////////////////////////
michael@0 120 //
michael@0 121 // Create the table of LCID to POSIX Mapping
michael@0 122 // None of it should be dynamically created.
michael@0 123 //
michael@0 124 // Keep static locale variables inside the function so that
michael@0 125 // it can be created properly during static init.
michael@0 126 //
michael@0 127 // Note: This table should be updated periodically. Check the National Lanaguage Support API Reference Website.
michael@0 128 // Microsoft is moving away from LCID in favor of locale name as of Vista. This table needs to be
michael@0 129 // maintained for support of older Windows version.
michael@0 130 // Update: Windows 7 (091130)
michael@0 131 //
michael@0 132 // Note: Microsoft assign a different LCID if a locale has a sorting variant. POSIX IDs below may contain
michael@0 133 // @collation=XXX, but no other keywords are allowed (at least for now). When uprv_convertToLCID() is
michael@0 134 // called from uloc_getLCID(), keywords other than collation are already removed. If we really need
michael@0 135 // to support other keywords in this mapping data, we must update the implementation.
michael@0 136 ////////////////////////////////////////////
michael@0 137 */
michael@0 138
michael@0 139 ILCID_POSIX_ELEMENT_ARRAY(0x0436, af, af_ZA)
michael@0 140
michael@0 141 ILCID_POSIX_SUBTABLE(ar) {
michael@0 142 {0x01, "ar"},
michael@0 143 {0x3801, "ar_AE"},
michael@0 144 {0x3c01, "ar_BH"},
michael@0 145 {0x1401, "ar_DZ"},
michael@0 146 {0x0c01, "ar_EG"},
michael@0 147 {0x0801, "ar_IQ"},
michael@0 148 {0x2c01, "ar_JO"},
michael@0 149 {0x3401, "ar_KW"},
michael@0 150 {0x3001, "ar_LB"},
michael@0 151 {0x1001, "ar_LY"},
michael@0 152 {0x1801, "ar_MA"},
michael@0 153 {0x1801, "ar_MO"},
michael@0 154 {0x2001, "ar_OM"},
michael@0 155 {0x4001, "ar_QA"},
michael@0 156 {0x0401, "ar_SA"},
michael@0 157 {0x2801, "ar_SY"},
michael@0 158 {0x1c01, "ar_TN"},
michael@0 159 {0x2401, "ar_YE"}
michael@0 160 };
michael@0 161
michael@0 162 ILCID_POSIX_ELEMENT_ARRAY(0x044d, as, as_IN)
michael@0 163 ILCID_POSIX_ELEMENT_ARRAY(0x045e, am, am_ET)
michael@0 164 ILCID_POSIX_ELEMENT_ARRAY(0x047a, arn,arn_CL)
michael@0 165
michael@0 166 ILCID_POSIX_SUBTABLE(az) {
michael@0 167 {0x2c, "az"},
michael@0 168 {0x082c, "az_Cyrl_AZ"}, /* Cyrillic based */
michael@0 169 {0x742c, "az_Cyrl"}, /* Cyrillic based */
michael@0 170 {0x042c, "az_Latn_AZ"}, /* Latin based */
michael@0 171 {0x782c, "az_Latn"}, /* Latin based */
michael@0 172 {0x042c, "az_AZ"} /* Latin based */
michael@0 173 };
michael@0 174
michael@0 175 ILCID_POSIX_ELEMENT_ARRAY(0x046d, ba, ba_RU)
michael@0 176 ILCID_POSIX_ELEMENT_ARRAY(0x0423, be, be_BY)
michael@0 177
michael@0 178 /*ILCID_POSIX_SUBTABLE(ber) {
michael@0 179 {0x5f, "ber"},
michael@0 180 {0x045f, "ber_Arab_DZ"},
michael@0 181 {0x045f, "ber_Arab"},
michael@0 182 {0x085f, "ber_Latn_DZ"},
michael@0 183 {0x085f, "ber_Latn"}
michael@0 184 };*/
michael@0 185
michael@0 186 ILCID_POSIX_ELEMENT_ARRAY(0x0402, bg, bg_BG)
michael@0 187
michael@0 188 ILCID_POSIX_ELEMENT_ARRAY(0x0466, bin, bin_NG)
michael@0 189
michael@0 190 ILCID_POSIX_SUBTABLE(bn) {
michael@0 191 {0x45, "bn"},
michael@0 192 {0x0845, "bn_BD"},
michael@0 193 {0x0445, "bn_IN"}
michael@0 194 };
michael@0 195
michael@0 196 ILCID_POSIX_SUBTABLE(bo) {
michael@0 197 {0x51, "bo"},
michael@0 198 {0x0851, "bo_BT"},
michael@0 199 {0x0451, "bo_CN"}
michael@0 200 };
michael@0 201
michael@0 202 ILCID_POSIX_ELEMENT_ARRAY(0x047e, br, br_FR)
michael@0 203
michael@0 204 ILCID_POSIX_SUBTABLE(ca) {
michael@0 205 {0x03, "ca"},
michael@0 206 {0x0403, "ca_ES"},
michael@0 207 {0x0803, "ca_ES_VALENCIA"}
michael@0 208 };
michael@0 209
michael@0 210 ILCID_POSIX_ELEMENT_ARRAY(0x0483, co, co_FR)
michael@0 211 ILCID_POSIX_ELEMENT_ARRAY(0x045c, chr,chr_US)
michael@0 212
michael@0 213 ILCID_POSIX_SUBTABLE(ckb) {
michael@0 214 {0x92, "ckb"},
michael@0 215 {0x92, "ku"},
michael@0 216 {0x7c92, "ckb_Arab"},
michael@0 217 {0x7c92, "ku_Arab"},
michael@0 218 {0x0492, "ckb_Arab_IQ"},
michael@0 219 {0x0492, "ku_Arab_IQ"}
michael@0 220 };
michael@0 221
michael@0 222 /* Declared as cs_CZ to get around compiler errors on z/OS, which defines cs as a function */
michael@0 223 ILCID_POSIX_ELEMENT_ARRAY(0x0405, cs, cs_CZ)
michael@0 224
michael@0 225 ILCID_POSIX_ELEMENT_ARRAY(0x0452, cy, cy_GB)
michael@0 226 ILCID_POSIX_ELEMENT_ARRAY(0x0406, da, da_DK)
michael@0 227
michael@0 228 ILCID_POSIX_SUBTABLE(de) {
michael@0 229 {0x07, "de"},
michael@0 230 {0x0c07, "de_AT"},
michael@0 231 {0x0807, "de_CH"},
michael@0 232 {0x0407, "de_DE"},
michael@0 233 {0x1407, "de_LI"},
michael@0 234 {0x1007, "de_LU"},
michael@0 235 {0x10407,"de_DE@collation=phonebook"}, /*This is really de_DE_PHONEBOOK on Windows*/
michael@0 236 {0x10407,"de@collation=phonebook"} /*This is really de_DE_PHONEBOOK on Windows*/
michael@0 237 };
michael@0 238
michael@0 239 ILCID_POSIX_ELEMENT_ARRAY(0x0465, dv, dv_MV)
michael@0 240 ILCID_POSIX_ELEMENT_ARRAY(0x0408, el, el_GR)
michael@0 241
michael@0 242 ILCID_POSIX_SUBTABLE(en) {
michael@0 243 {0x09, "en"},
michael@0 244 {0x0c09, "en_AU"},
michael@0 245 {0x2809, "en_BZ"},
michael@0 246 {0x1009, "en_CA"},
michael@0 247 {0x0809, "en_GB"},
michael@0 248 {0x3c09, "en_HK"},
michael@0 249 {0x3809, "en_ID"},
michael@0 250 {0x1809, "en_IE"},
michael@0 251 {0x4009, "en_IN"},
michael@0 252 {0x2009, "en_JM"},
michael@0 253 {0x4409, "en_MY"},
michael@0 254 {0x1409, "en_NZ"},
michael@0 255 {0x3409, "en_PH"},
michael@0 256 {0x4809, "en_SG"},
michael@0 257 {0x2C09, "en_TT"},
michael@0 258 {0x0409, "en_US"},
michael@0 259 {0x007f, "en_US_POSIX"}, /* duplicate for roundtripping */
michael@0 260 {0x2409, "en_VI"}, /* Virgin Islands AKA Caribbean Islands (en_CB). */
michael@0 261 {0x1c09, "en_ZA"},
michael@0 262 {0x3009, "en_ZW"},
michael@0 263 {0x2409, "en_029"},
michael@0 264 {0x0409, "en_AS"}, /* Alias for en_US. Leave last. */
michael@0 265 {0x0409, "en_GU"}, /* Alias for en_US. Leave last. */
michael@0 266 {0x0409, "en_MH"}, /* Alias for en_US. Leave last. */
michael@0 267 {0x0409, "en_MP"}, /* Alias for en_US. Leave last. */
michael@0 268 {0x0409, "en_UM"} /* Alias for en_US. Leave last. */
michael@0 269 };
michael@0 270
michael@0 271 ILCID_POSIX_SUBTABLE(en_US_POSIX) {
michael@0 272 {0x007f, "en_US_POSIX"} /* duplicate for roundtripping */
michael@0 273 };
michael@0 274
michael@0 275 ILCID_POSIX_SUBTABLE(es) {
michael@0 276 {0x0a, "es"},
michael@0 277 {0x2c0a, "es_AR"},
michael@0 278 {0x400a, "es_BO"},
michael@0 279 {0x340a, "es_CL"},
michael@0 280 {0x240a, "es_CO"},
michael@0 281 {0x140a, "es_CR"},
michael@0 282 {0x1c0a, "es_DO"},
michael@0 283 {0x300a, "es_EC"},
michael@0 284 {0x0c0a, "es_ES"}, /*Modern sort.*/
michael@0 285 {0x100a, "es_GT"},
michael@0 286 {0x480a, "es_HN"},
michael@0 287 {0x080a, "es_MX"},
michael@0 288 {0x4c0a, "es_NI"},
michael@0 289 {0x180a, "es_PA"},
michael@0 290 {0x280a, "es_PE"},
michael@0 291 {0x500a, "es_PR"},
michael@0 292 {0x3c0a, "es_PY"},
michael@0 293 {0x440a, "es_SV"},
michael@0 294 {0x540a, "es_US"},
michael@0 295 {0x380a, "es_UY"},
michael@0 296 {0x200a, "es_VE"},
michael@0 297 {0xe40a, "es_419"},
michael@0 298 {0x040a, "es_ES@collation=traditional"},
michael@0 299 {0x040a, "es@collation=traditional"}
michael@0 300 };
michael@0 301
michael@0 302 ILCID_POSIX_ELEMENT_ARRAY(0x0425, et, et_EE)
michael@0 303 ILCID_POSIX_ELEMENT_ARRAY(0x042d, eu, eu_ES)
michael@0 304
michael@0 305 /* ISO-639 doesn't distinguish between Persian and Dari.*/
michael@0 306 ILCID_POSIX_SUBTABLE(fa) {
michael@0 307 {0x29, "fa"},
michael@0 308 {0x0429, "fa_IR"}, /* Persian/Farsi (Iran) */
michael@0 309 {0x048c, "fa_AF"} /* Persian/Dari (Afghanistan) */
michael@0 310 };
michael@0 311
michael@0 312 /* duplicate for roundtripping */
michael@0 313 ILCID_POSIX_SUBTABLE(fa_AF) {
michael@0 314 {0x8c, "fa_AF"}, /* Persian/Dari (Afghanistan) */
michael@0 315 {0x048c, "fa_AF"} /* Persian/Dari (Afghanistan) */
michael@0 316 };
michael@0 317
michael@0 318 ILCID_POSIX_SUBTABLE(ff) {
michael@0 319 {0x67, "ff"},
michael@0 320 {0x7c67, "ff_Latn"},
michael@0 321 {0x0867, "ff_Latn_SN"}
michael@0 322 };
michael@0 323
michael@0 324 ILCID_POSIX_ELEMENT_ARRAY(0x040b, fi, fi_FI)
michael@0 325 ILCID_POSIX_ELEMENT_ARRAY(0x0464, fil,fil_PH)
michael@0 326 ILCID_POSIX_ELEMENT_ARRAY(0x0438, fo, fo_FO)
michael@0 327
michael@0 328 ILCID_POSIX_SUBTABLE(fr) {
michael@0 329 {0x0c, "fr"},
michael@0 330 {0x080c, "fr_BE"},
michael@0 331 {0x0c0c, "fr_CA"},
michael@0 332 {0x240c, "fr_CD"},
michael@0 333 {0x240c, "fr_CG"},
michael@0 334 {0x100c, "fr_CH"},
michael@0 335 {0x300c, "fr_CI"},
michael@0 336 {0x2c0c, "fr_CM"},
michael@0 337 {0x040c, "fr_FR"},
michael@0 338 {0x3c0c, "fr_HT"},
michael@0 339 {0x140c, "fr_LU"},
michael@0 340 {0x380c, "fr_MA"},
michael@0 341 {0x180c, "fr_MC"},
michael@0 342 {0x340c, "fr_ML"},
michael@0 343 {0x200c, "fr_RE"},
michael@0 344 {0x280c, "fr_SN"},
michael@0 345 {0xe40c, "fr_015"},
michael@0 346 {0x1c0c, "fr_029"}
michael@0 347 };
michael@0 348
michael@0 349 ILCID_POSIX_ELEMENT_ARRAY(0x0467, fuv, fuv_NG)
michael@0 350
michael@0 351 ILCID_POSIX_ELEMENT_ARRAY(0x0462, fy, fy_NL)
michael@0 352
michael@0 353 ILCID_POSIX_SUBTABLE(ga) { /* Gaelic (Ireland) */
michael@0 354 {0x3c, "ga"},
michael@0 355 {0x083c, "ga_IE"},
michael@0 356 {0x043c, "gd_GB"}
michael@0 357 };
michael@0 358
michael@0 359 ILCID_POSIX_SUBTABLE(gd) { /* Gaelic (Scotland) */
michael@0 360 {0x91, "gd"},
michael@0 361 {0x0491, "gd_GB"}
michael@0 362 };
michael@0 363
michael@0 364 ILCID_POSIX_ELEMENT_ARRAY(0x0456, gl, gl_ES)
michael@0 365 ILCID_POSIX_ELEMENT_ARRAY(0x0447, gu, gu_IN)
michael@0 366 ILCID_POSIX_ELEMENT_ARRAY(0x0474, gn, gn_PY)
michael@0 367 ILCID_POSIX_ELEMENT_ARRAY(0x0484, gsw,gsw_FR)
michael@0 368
michael@0 369 ILCID_POSIX_SUBTABLE(ha) {
michael@0 370 {0x68, "ha"},
michael@0 371 {0x7c68, "ha_Latn"},
michael@0 372 {0x0468, "ha_Latn_NG"},
michael@0 373 };
michael@0 374
michael@0 375 ILCID_POSIX_ELEMENT_ARRAY(0x0475, haw,haw_US)
michael@0 376 ILCID_POSIX_ELEMENT_ARRAY(0x040d, he, he_IL)
michael@0 377 ILCID_POSIX_ELEMENT_ARRAY(0x0439, hi, hi_IN)
michael@0 378
michael@0 379 /* This LCID is really four different locales.*/
michael@0 380 ILCID_POSIX_SUBTABLE(hr) {
michael@0 381 {0x1a, "hr"},
michael@0 382 {0x141a, "bs_Latn_BA"}, /* Bosnian, Bosnia and Herzegovina */
michael@0 383 {0x681a, "bs_Latn"}, /* Bosnian, Bosnia and Herzegovina */
michael@0 384 {0x141a, "bs_BA"}, /* Bosnian, Bosnia and Herzegovina */
michael@0 385 {0x781a, "bs"}, /* Bosnian */
michael@0 386 {0x201a, "bs_Cyrl_BA"}, /* Bosnian, Bosnia and Herzegovina */
michael@0 387 {0x641a, "bs_Cyrl"}, /* Bosnian, Bosnia and Herzegovina */
michael@0 388 {0x101a, "hr_BA"}, /* Croatian in Bosnia */
michael@0 389 {0x041a, "hr_HR"}, /* Croatian*/
michael@0 390 {0x2c1a, "sr_Latn_ME"},
michael@0 391 {0x241a, "sr_Latn_RS"},
michael@0 392 {0x181a, "sr_Latn_BA"}, /* Serbo-Croatian in Bosnia */
michael@0 393 {0x081a, "sr_Latn_CS"}, /* Serbo-Croatian*/
michael@0 394 {0x701a, "sr_Latn"}, /* It's 0x1a or 0x081a, pick one to make the test program happy. */
michael@0 395 {0x1c1a, "sr_Cyrl_BA"}, /* Serbo-Croatian in Bosnia */
michael@0 396 {0x0c1a, "sr_Cyrl_CS"}, /* Serbian*/
michael@0 397 {0x301a, "sr_Cyrl_ME"},
michael@0 398 {0x281a, "sr_Cyrl_RS"},
michael@0 399 {0x6c1a, "sr_Cyrl"}, /* It's 0x1a or 0x0c1a, pick one to make the test program happy. */
michael@0 400 {0x7c1a, "sr"} /* In CLDR sr is sr_Cyrl. */
michael@0 401 };
michael@0 402
michael@0 403 ILCID_POSIX_ELEMENT_ARRAY(0x040e, hu, hu_HU)
michael@0 404 ILCID_POSIX_ELEMENT_ARRAY(0x042b, hy, hy_AM)
michael@0 405 ILCID_POSIX_ELEMENT_ARRAY(0x0469, ibb, ibb_NG)
michael@0 406 ILCID_POSIX_ELEMENT_ARRAY(0x0421, id, id_ID)
michael@0 407 ILCID_POSIX_ELEMENT_ARRAY(0x0470, ig, ig_NG)
michael@0 408 ILCID_POSIX_ELEMENT_ARRAY(0x0478, ii, ii_CN)
michael@0 409 ILCID_POSIX_ELEMENT_ARRAY(0x040f, is, is_IS)
michael@0 410
michael@0 411 ILCID_POSIX_SUBTABLE(it) {
michael@0 412 {0x10, "it"},
michael@0 413 {0x0810, "it_CH"},
michael@0 414 {0x0410, "it_IT"}
michael@0 415 };
michael@0 416
michael@0 417 ILCID_POSIX_SUBTABLE(iu) {
michael@0 418 {0x5d, "iu"},
michael@0 419 {0x045d, "iu_Cans_CA"},
michael@0 420 {0x785d, "iu_Cans"},
michael@0 421 {0x085d, "iu_Latn_CA"},
michael@0 422 {0x7c5d, "iu_Latn"}
michael@0 423 };
michael@0 424
michael@0 425 ILCID_POSIX_ELEMENT_ARRAY(0x040d, iw, iw_IL) /*Left in for compatibility*/
michael@0 426 ILCID_POSIX_ELEMENT_ARRAY(0x0411, ja, ja_JP)
michael@0 427 ILCID_POSIX_ELEMENT_ARRAY(0x0437, ka, ka_GE)
michael@0 428 ILCID_POSIX_ELEMENT_ARRAY(0x043f, kk, kk_KZ)
michael@0 429 ILCID_POSIX_ELEMENT_ARRAY(0x046f, kl, kl_GL)
michael@0 430 ILCID_POSIX_ELEMENT_ARRAY(0x0453, km, km_KH)
michael@0 431 ILCID_POSIX_ELEMENT_ARRAY(0x044b, kn, kn_IN)
michael@0 432
michael@0 433 ILCID_POSIX_SUBTABLE(ko) {
michael@0 434 {0x12, "ko"},
michael@0 435 {0x0812, "ko_KP"},
michael@0 436 {0x0412, "ko_KR"}
michael@0 437 };
michael@0 438
michael@0 439 ILCID_POSIX_ELEMENT_ARRAY(0x0457, kok, kok_IN)
michael@0 440 ILCID_POSIX_ELEMENT_ARRAY(0x0471, kr, kr_NG)
michael@0 441
michael@0 442 ILCID_POSIX_SUBTABLE(ks) { /* We could add PK and CN too */
michael@0 443 {0x60, "ks"},
michael@0 444 {0x0860, "ks_IN"}, /* Documentation doesn't mention script */
michael@0 445 {0x0460, "ks_Arab_IN"},
michael@0 446 {0x0860, "ks_Deva_IN"}
michael@0 447 };
michael@0 448
michael@0 449 ILCID_POSIX_ELEMENT_ARRAY(0x0440, ky, ky_KG) /* Kyrgyz is spoken in Kyrgyzstan */
michael@0 450 ILCID_POSIX_ELEMENT_ARRAY(0x0476, la, la_IT) /* TODO: Verify the country */
michael@0 451 ILCID_POSIX_ELEMENT_ARRAY(0x046e, lb, lb_LU)
michael@0 452 ILCID_POSIX_ELEMENT_ARRAY(0x0454, lo, lo_LA)
michael@0 453 ILCID_POSIX_ELEMENT_ARRAY(0x0427, lt, lt_LT)
michael@0 454 ILCID_POSIX_ELEMENT_ARRAY(0x0426, lv, lv_LV)
michael@0 455 ILCID_POSIX_ELEMENT_ARRAY(0x0481, mi, mi_NZ)
michael@0 456 ILCID_POSIX_ELEMENT_ARRAY(0x042f, mk, mk_MK)
michael@0 457 ILCID_POSIX_ELEMENT_ARRAY(0x044c, ml, ml_IN)
michael@0 458
michael@0 459 ILCID_POSIX_SUBTABLE(mn) {
michael@0 460 {0x50, "mn"},
michael@0 461 {0x0450, "mn_MN"},
michael@0 462 {0x7c50, "mn_Mong"},
michael@0 463 {0x0850, "mn_Mong_CN"},
michael@0 464 {0x0850, "mn_CN"},
michael@0 465 {0x7850, "mn_Cyrl"}
michael@0 466 };
michael@0 467
michael@0 468 ILCID_POSIX_ELEMENT_ARRAY(0x0458, mni,mni_IN)
michael@0 469 ILCID_POSIX_ELEMENT_ARRAY(0x047c, moh,moh_CA)
michael@0 470 ILCID_POSIX_ELEMENT_ARRAY(0x044e, mr, mr_IN)
michael@0 471
michael@0 472 ILCID_POSIX_SUBTABLE(ms) {
michael@0 473 {0x3e, "ms"},
michael@0 474 {0x083e, "ms_BN"}, /* Brunei Darussalam*/
michael@0 475 {0x043e, "ms_MY"} /* Malaysia*/
michael@0 476 };
michael@0 477
michael@0 478 ILCID_POSIX_ELEMENT_ARRAY(0x043a, mt, mt_MT)
michael@0 479 ILCID_POSIX_ELEMENT_ARRAY(0x0455, my, my_MM)
michael@0 480
michael@0 481 ILCID_POSIX_SUBTABLE(ne) {
michael@0 482 {0x61, "ne"},
michael@0 483 {0x0861, "ne_IN"}, /* India*/
michael@0 484 {0x0461, "ne_NP"} /* Nepal*/
michael@0 485 };
michael@0 486
michael@0 487 ILCID_POSIX_SUBTABLE(nl) {
michael@0 488 {0x13, "nl"},
michael@0 489 {0x0813, "nl_BE"},
michael@0 490 {0x0413, "nl_NL"}
michael@0 491 };
michael@0 492
michael@0 493 /* The "no" locale split into nb and nn. By default in ICU, "no" is nb.*/
michael@0 494 ILCID_POSIX_SUBTABLE(no) {
michael@0 495 {0x14, "no"}, /* really nb_NO */
michael@0 496 {0x7c14, "nb"}, /* really nb */
michael@0 497 {0x0414, "nb_NO"}, /* really nb_NO. Keep first in the 414 list. */
michael@0 498 {0x0414, "no_NO"}, /* really nb_NO */
michael@0 499 {0x0814, "nn_NO"}, /* really nn_NO. Keep first in the 814 list. */
michael@0 500 {0x7814, "nn"}, /* It's 0x14 or 0x814, pick one to make the test program happy. */
michael@0 501 {0x0814, "no_NO_NY"}/* really nn_NO */
michael@0 502 };
michael@0 503
michael@0 504 ILCID_POSIX_ELEMENT_ARRAY(0x046c, nso,nso_ZA) /* TODO: Verify the ISO-639 code */
michael@0 505 ILCID_POSIX_ELEMENT_ARRAY(0x0482, oc, oc_FR)
michael@0 506
michael@0 507 ILCID_POSIX_SUBTABLE(om) { /* TODO: Verify the country */
michael@0 508 {0x72, "om"},
michael@0 509 {0x0472, "om_ET"},
michael@0 510 {0x0472, "gaz_ET"}
michael@0 511 };
michael@0 512
michael@0 513 /* Declared as or_IN to get around compiler errors*/
michael@0 514 ILCID_POSIX_SUBTABLE(or_IN) {
michael@0 515 {0x48, "or"},
michael@0 516 {0x0448, "or_IN"},
michael@0 517 };
michael@0 518
michael@0 519
michael@0 520 ILCID_POSIX_SUBTABLE(pa) {
michael@0 521 {0x46, "pa"},
michael@0 522 {0x0446, "pa_IN"},
michael@0 523 {0x0846, "pa_PK"},
michael@0 524 {0x0846, "pa_Arab_PK"}
michael@0 525 };
michael@0 526
michael@0 527 ILCID_POSIX_ELEMENT_ARRAY(0x0479, pap, pap_AN)
michael@0 528 ILCID_POSIX_ELEMENT_ARRAY(0x0415, pl, pl_PL)
michael@0 529 ILCID_POSIX_ELEMENT_ARRAY(0x0463, ps, ps_AF)
michael@0 530
michael@0 531 ILCID_POSIX_SUBTABLE(pt) {
michael@0 532 {0x16, "pt"},
michael@0 533 {0x0416, "pt_BR"},
michael@0 534 {0x0816, "pt_PT"}
michael@0 535 };
michael@0 536
michael@0 537 ILCID_POSIX_SUBTABLE(qu) {
michael@0 538 {0x6b, "qu"},
michael@0 539 {0x046b, "qu_BO"},
michael@0 540 {0x086b, "qu_EC"},
michael@0 541 {0x0C6b, "qu_PE"},
michael@0 542 {0x046b, "quz_BO"},
michael@0 543 {0x086b, "quz_EC"},
michael@0 544 {0x0C6b, "quz_PE"}
michael@0 545 };
michael@0 546
michael@0 547 ILCID_POSIX_ELEMENT_ARRAY(0x0486, qut, qut_GT) /* qut is an ISO-639-3 code */
michael@0 548 ILCID_POSIX_ELEMENT_ARRAY(0x0417, rm, rm_CH)
michael@0 549
michael@0 550 ILCID_POSIX_SUBTABLE(ro) {
michael@0 551 {0x18, "ro"},
michael@0 552 {0x0418, "ro_RO"},
michael@0 553 {0x0818, "ro_MD"}
michael@0 554 };
michael@0 555
michael@0 556 ILCID_POSIX_SUBTABLE(root) {
michael@0 557 {0x00, "root"}
michael@0 558 };
michael@0 559
michael@0 560 ILCID_POSIX_SUBTABLE(ru) {
michael@0 561 {0x19, "ru"},
michael@0 562 {0x0419, "ru_RU"},
michael@0 563 {0x0819, "ru_MD"}
michael@0 564 };
michael@0 565
michael@0 566 ILCID_POSIX_ELEMENT_ARRAY(0x0487, rw, rw_RW)
michael@0 567 ILCID_POSIX_ELEMENT_ARRAY(0x044f, sa, sa_IN)
michael@0 568 ILCID_POSIX_ELEMENT_ARRAY(0x0485, sah,sah_RU)
michael@0 569
michael@0 570 ILCID_POSIX_SUBTABLE(sd) {
michael@0 571 {0x59, "sd"},
michael@0 572 {0x0459, "sd_IN"},
michael@0 573 {0x0859, "sd_PK"}
michael@0 574 };
michael@0 575
michael@0 576 ILCID_POSIX_SUBTABLE(se) {
michael@0 577 {0x3b, "se"},
michael@0 578 {0x0c3b, "se_FI"},
michael@0 579 {0x043b, "se_NO"},
michael@0 580 {0x083b, "se_SE"},
michael@0 581 {0x783b, "sma"},
michael@0 582 {0x183b, "sma_NO"},
michael@0 583 {0x1c3b, "sma_SE"},
michael@0 584 {0x7c3b, "smj"},
michael@0 585 {0x703b, "smn"},
michael@0 586 {0x743b, "sms"},
michael@0 587 {0x103b, "smj_NO"},
michael@0 588 {0x143b, "smj_SE"},
michael@0 589 {0x243b, "smn_FI"},
michael@0 590 {0x203b, "sms_FI"},
michael@0 591 };
michael@0 592
michael@0 593 ILCID_POSIX_ELEMENT_ARRAY(0x045b, si, si_LK)
michael@0 594 ILCID_POSIX_ELEMENT_ARRAY(0x041b, sk, sk_SK)
michael@0 595 ILCID_POSIX_ELEMENT_ARRAY(0x0424, sl, sl_SI)
michael@0 596
michael@0 597 ILCID_POSIX_SUBTABLE(so) { /* TODO: Verify the country */
michael@0 598 {0x77, "so"},
michael@0 599 {0x0477, "so_ET"},
michael@0 600 {0x0477, "so_SO"}
michael@0 601 };
michael@0 602
michael@0 603 ILCID_POSIX_ELEMENT_ARRAY(0x041c, sq, sq_AL)
michael@0 604 ILCID_POSIX_ELEMENT_ARRAY(0x0430, st, st_ZA)
michael@0 605
michael@0 606 ILCID_POSIX_SUBTABLE(sv) {
michael@0 607 {0x1d, "sv"},
michael@0 608 {0x081d, "sv_FI"},
michael@0 609 {0x041d, "sv_SE"}
michael@0 610 };
michael@0 611
michael@0 612 ILCID_POSIX_ELEMENT_ARRAY(0x0441, sw, sw_KE)
michael@0 613 ILCID_POSIX_ELEMENT_ARRAY(0x045A, syr, syr_SY)
michael@0 614
michael@0 615 ILCID_POSIX_SUBTABLE(ta) {
michael@0 616 {0x49, "ta"},
michael@0 617 {0x0449, "ta_IN"},
michael@0 618 {0x0849, "ta_LK"}
michael@0 619 };
michael@0 620
michael@0 621 ILCID_POSIX_ELEMENT_ARRAY(0x044a, te, te_IN)
michael@0 622
michael@0 623 /* Cyrillic based by default */
michael@0 624 ILCID_POSIX_SUBTABLE(tg) {
michael@0 625 {0x28, "tg"},
michael@0 626 {0x7c28, "tg_Cyrl"},
michael@0 627 {0x0428, "tg_Cyrl_TJ"}
michael@0 628 };
michael@0 629
michael@0 630 ILCID_POSIX_ELEMENT_ARRAY(0x041e, th, th_TH)
michael@0 631
michael@0 632 ILCID_POSIX_SUBTABLE(ti) {
michael@0 633 {0x73, "ti"},
michael@0 634 {0x0873, "ti_ER"},
michael@0 635 {0x0473, "ti_ET"}
michael@0 636 };
michael@0 637
michael@0 638 ILCID_POSIX_ELEMENT_ARRAY(0x0442, tk, tk_TM)
michael@0 639
michael@0 640 ILCID_POSIX_SUBTABLE(tn) {
michael@0 641 {0x32, "tn"},
michael@0 642 {0x0832, "tn_BW"},
michael@0 643 {0x0432, "tn_ZA"}
michael@0 644 };
michael@0 645
michael@0 646 ILCID_POSIX_ELEMENT_ARRAY(0x041f, tr, tr_TR)
michael@0 647 ILCID_POSIX_ELEMENT_ARRAY(0x0431, ts, ts_ZA)
michael@0 648 ILCID_POSIX_ELEMENT_ARRAY(0x0444, tt, tt_RU)
michael@0 649
michael@0 650 ILCID_POSIX_SUBTABLE(tzm) {
michael@0 651 {0x5f, "tzm"},
michael@0 652 {0x7c5f, "tzm_Latn"},
michael@0 653 {0x085f, "tzm_Latn_DZ"},
michael@0 654 {0x105f, "tzm_Tfng_MA"},
michael@0 655 {0x045f, "tmz"}
michael@0 656 };
michael@0 657
michael@0 658 ILCID_POSIX_SUBTABLE(ug) {
michael@0 659 {0x80, "ug"},
michael@0 660 {0x0480, "ug_CN"},
michael@0 661 {0x0480, "ug_Arab_CN"}
michael@0 662 };
michael@0 663
michael@0 664 ILCID_POSIX_ELEMENT_ARRAY(0x0422, uk, uk_UA)
michael@0 665
michael@0 666 ILCID_POSIX_SUBTABLE(ur) {
michael@0 667 {0x20, "ur"},
michael@0 668 {0x0820, "ur_IN"},
michael@0 669 {0x0420, "ur_PK"}
michael@0 670 };
michael@0 671
michael@0 672 ILCID_POSIX_SUBTABLE(uz) {
michael@0 673 {0x43, "uz"},
michael@0 674 {0x0843, "uz_Cyrl_UZ"}, /* Cyrillic based */
michael@0 675 {0x7843, "uz_Cyrl"}, /* Cyrillic based */
michael@0 676 {0x0843, "uz_UZ"}, /* Cyrillic based */
michael@0 677 {0x0443, "uz_Latn_UZ"}, /* Latin based */
michael@0 678 {0x7c43, "uz_Latn"} /* Latin based */
michael@0 679 };
michael@0 680
michael@0 681 ILCID_POSIX_SUBTABLE(ve) { /* TODO: Verify the country */
michael@0 682 {0x33, "ve"},
michael@0 683 {0x0433, "ve_ZA"},
michael@0 684 {0x0433, "ven_ZA"}
michael@0 685 };
michael@0 686
michael@0 687 ILCID_POSIX_ELEMENT_ARRAY(0x042a, vi, vi_VN)
michael@0 688
michael@0 689 ILCID_POSIX_SUBTABLE(wen) {
michael@0 690 {0x2E, "wen"},
michael@0 691 {0x042E, "wen_DE"},
michael@0 692 {0x042E, "hsb_DE"},
michael@0 693 {0x082E, "dsb_DE"},
michael@0 694 {0x7C2E, "dsb"},
michael@0 695 {0x2E, "hsb"}
michael@0 696 };
michael@0 697
michael@0 698 ILCID_POSIX_ELEMENT_ARRAY(0x0488, wo, wo_SN)
michael@0 699 ILCID_POSIX_ELEMENT_ARRAY(0x0434, xh, xh_ZA)
michael@0 700 ILCID_POSIX_ELEMENT_ARRAY(0x043d, yi, yi)
michael@0 701 ILCID_POSIX_ELEMENT_ARRAY(0x046a, yo, yo_NG)
michael@0 702
michael@0 703 ILCID_POSIX_SUBTABLE(zh) {
michael@0 704 {0x0004, "zh_Hans"},
michael@0 705 {0x7804, "zh"},
michael@0 706 {0x0804, "zh_CN"},
michael@0 707 {0x0804, "zh_Hans_CN"},
michael@0 708 {0x0c04, "zh_Hant_HK"},
michael@0 709 {0x0c04, "zh_HK"},
michael@0 710 {0x1404, "zh_Hant_MO"},
michael@0 711 {0x1404, "zh_MO"},
michael@0 712 {0x1004, "zh_Hans_SG"},
michael@0 713 {0x1004, "zh_SG"},
michael@0 714 {0x0404, "zh_Hant_TW"},
michael@0 715 {0x7c04, "zh_Hant"},
michael@0 716 {0x0404, "zh_TW"},
michael@0 717 {0x30404,"zh_Hant_TW"}, /* Bopomofo order */
michael@0 718 {0x30404,"zh_TW"}, /* Bopomofo order */
michael@0 719 {0x20004,"zh@collation=stroke"},
michael@0 720 {0x20404,"zh_Hant@collation=stroke"},
michael@0 721 {0x20404,"zh_Hant_TW@collation=stroke"},
michael@0 722 {0x20404,"zh_TW@collation=stroke"},
michael@0 723 {0x20804,"zh_Hans@collation=stroke"},
michael@0 724 {0x20804,"zh_Hans_CN@collation=stroke"},
michael@0 725 {0x20804,"zh_CN@collation=stroke"}
michael@0 726 };
michael@0 727
michael@0 728 ILCID_POSIX_ELEMENT_ARRAY(0x0435, zu, zu_ZA)
michael@0 729
michael@0 730 /* This must be static and grouped by LCID. */
michael@0 731 static const ILcidPosixMap gPosixIDmap[] = {
michael@0 732 ILCID_POSIX_MAP(af), /* af Afrikaans 0x36 */
michael@0 733 ILCID_POSIX_MAP(am), /* am Amharic 0x5e */
michael@0 734 ILCID_POSIX_MAP(ar), /* ar Arabic 0x01 */
michael@0 735 ILCID_POSIX_MAP(arn), /* arn Araucanian/Mapudungun 0x7a */
michael@0 736 ILCID_POSIX_MAP(as), /* as Assamese 0x4d */
michael@0 737 ILCID_POSIX_MAP(az), /* az Azerbaijani 0x2c */
michael@0 738 ILCID_POSIX_MAP(ba), /* ba Bashkir 0x6d */
michael@0 739 ILCID_POSIX_MAP(be), /* be Belarusian 0x23 */
michael@0 740 /* ILCID_POSIX_MAP(ber), ber Berber/Tamazight 0x5f */
michael@0 741 ILCID_POSIX_MAP(bg), /* bg Bulgarian 0x02 */
michael@0 742 ILCID_POSIX_MAP(bin), /* bin Edo 0x66 */
michael@0 743 ILCID_POSIX_MAP(bn), /* bn Bengali; Bangla 0x45 */
michael@0 744 ILCID_POSIX_MAP(bo), /* bo Tibetan 0x51 */
michael@0 745 ILCID_POSIX_MAP(br), /* br Breton 0x7e */
michael@0 746 ILCID_POSIX_MAP(ca), /* ca Catalan 0x03 */
michael@0 747 ILCID_POSIX_MAP(chr), /* chr Cherokee 0x5c */
michael@0 748 ILCID_POSIX_MAP(ckb), /* ckb Sorani (Central Kurdish) 0x92 */
michael@0 749 ILCID_POSIX_MAP(co), /* co Corsican 0x83 */
michael@0 750 ILCID_POSIX_MAP(cs), /* cs Czech 0x05 */
michael@0 751 ILCID_POSIX_MAP(cy), /* cy Welsh 0x52 */
michael@0 752 ILCID_POSIX_MAP(da), /* da Danish 0x06 */
michael@0 753 ILCID_POSIX_MAP(de), /* de German 0x07 */
michael@0 754 ILCID_POSIX_MAP(dv), /* dv Divehi 0x65 */
michael@0 755 ILCID_POSIX_MAP(el), /* el Greek 0x08 */
michael@0 756 ILCID_POSIX_MAP(en), /* en English 0x09 */
michael@0 757 ILCID_POSIX_MAP(en_US_POSIX), /* invariant 0x7f */
michael@0 758 ILCID_POSIX_MAP(es), /* es Spanish 0x0a */
michael@0 759 ILCID_POSIX_MAP(et), /* et Estonian 0x25 */
michael@0 760 ILCID_POSIX_MAP(eu), /* eu Basque 0x2d */
michael@0 761 ILCID_POSIX_MAP(fa), /* fa Persian/Farsi 0x29 */
michael@0 762 ILCID_POSIX_MAP(fa_AF), /* fa Persian/Dari 0x8c */
michael@0 763 ILCID_POSIX_MAP(ff), /* ff Fula 0x67 */
michael@0 764 ILCID_POSIX_MAP(fi), /* fi Finnish 0x0b */
michael@0 765 ILCID_POSIX_MAP(fil), /* fil Filipino 0x64 */
michael@0 766 ILCID_POSIX_MAP(fo), /* fo Faroese 0x38 */
michael@0 767 ILCID_POSIX_MAP(fr), /* fr French 0x0c */
michael@0 768 ILCID_POSIX_MAP(fuv), /* fuv Fulfulde - Nigeria 0x67 */
michael@0 769 ILCID_POSIX_MAP(fy), /* fy Frisian 0x62 */
michael@0 770 ILCID_POSIX_MAP(ga), /* * Gaelic (Ireland,Scotland) 0x3c */
michael@0 771 ILCID_POSIX_MAP(gd), /* gd Gaelic (United Kingdom) 0x91 */
michael@0 772 ILCID_POSIX_MAP(gl), /* gl Galician 0x56 */
michael@0 773 ILCID_POSIX_MAP(gn), /* gn Guarani 0x74 */
michael@0 774 ILCID_POSIX_MAP(gsw), /* gsw Alemanic/Alsatian/Swiss German 0x84 */
michael@0 775 ILCID_POSIX_MAP(gu), /* gu Gujarati 0x47 */
michael@0 776 ILCID_POSIX_MAP(ha), /* ha Hausa 0x68 */
michael@0 777 ILCID_POSIX_MAP(haw), /* haw Hawaiian 0x75 */
michael@0 778 ILCID_POSIX_MAP(he), /* he Hebrew (formerly iw) 0x0d */
michael@0 779 ILCID_POSIX_MAP(hi), /* hi Hindi 0x39 */
michael@0 780 ILCID_POSIX_MAP(hr), /* * Croatian and others 0x1a */
michael@0 781 ILCID_POSIX_MAP(hu), /* hu Hungarian 0x0e */
michael@0 782 ILCID_POSIX_MAP(hy), /* hy Armenian 0x2b */
michael@0 783 ILCID_POSIX_MAP(ibb), /* ibb Ibibio - Nigeria 0x69 */
michael@0 784 ILCID_POSIX_MAP(id), /* id Indonesian (formerly in) 0x21 */
michael@0 785 ILCID_POSIX_MAP(ig), /* ig Igbo 0x70 */
michael@0 786 ILCID_POSIX_MAP(ii), /* ii Sichuan Yi 0x78 */
michael@0 787 ILCID_POSIX_MAP(is), /* is Icelandic 0x0f */
michael@0 788 ILCID_POSIX_MAP(it), /* it Italian 0x10 */
michael@0 789 ILCID_POSIX_MAP(iu), /* iu Inuktitut 0x5d */
michael@0 790 ILCID_POSIX_MAP(iw), /* iw Hebrew 0x0d */
michael@0 791 ILCID_POSIX_MAP(ja), /* ja Japanese 0x11 */
michael@0 792 ILCID_POSIX_MAP(ka), /* ka Georgian 0x37 */
michael@0 793 ILCID_POSIX_MAP(kk), /* kk Kazakh 0x3f */
michael@0 794 ILCID_POSIX_MAP(kl), /* kl Kalaallisut 0x6f */
michael@0 795 ILCID_POSIX_MAP(km), /* km Khmer 0x53 */
michael@0 796 ILCID_POSIX_MAP(kn), /* kn Kannada 0x4b */
michael@0 797 ILCID_POSIX_MAP(ko), /* ko Korean 0x12 */
michael@0 798 ILCID_POSIX_MAP(kok), /* kok Konkani 0x57 */
michael@0 799 ILCID_POSIX_MAP(kr), /* kr Kanuri 0x71 */
michael@0 800 ILCID_POSIX_MAP(ks), /* ks Kashmiri 0x60 */
michael@0 801 ILCID_POSIX_MAP(ky), /* ky Kyrgyz 0x40 */
michael@0 802 ILCID_POSIX_MAP(lb), /* lb Luxembourgish 0x6e */
michael@0 803 ILCID_POSIX_MAP(la), /* la Latin 0x76 */
michael@0 804 ILCID_POSIX_MAP(lo), /* lo Lao 0x54 */
michael@0 805 ILCID_POSIX_MAP(lt), /* lt Lithuanian 0x27 */
michael@0 806 ILCID_POSIX_MAP(lv), /* lv Latvian, Lettish 0x26 */
michael@0 807 ILCID_POSIX_MAP(mi), /* mi Maori 0x81 */
michael@0 808 ILCID_POSIX_MAP(mk), /* mk Macedonian 0x2f */
michael@0 809 ILCID_POSIX_MAP(ml), /* ml Malayalam 0x4c */
michael@0 810 ILCID_POSIX_MAP(mn), /* mn Mongolian 0x50 */
michael@0 811 ILCID_POSIX_MAP(mni), /* mni Manipuri 0x58 */
michael@0 812 ILCID_POSIX_MAP(moh), /* moh Mohawk 0x7c */
michael@0 813 ILCID_POSIX_MAP(mr), /* mr Marathi 0x4e */
michael@0 814 ILCID_POSIX_MAP(ms), /* ms Malay 0x3e */
michael@0 815 ILCID_POSIX_MAP(mt), /* mt Maltese 0x3a */
michael@0 816 ILCID_POSIX_MAP(my), /* my Burmese 0x55 */
michael@0 817 /* ILCID_POSIX_MAP(nb), // no Norwegian 0x14 */
michael@0 818 ILCID_POSIX_MAP(ne), /* ne Nepali 0x61 */
michael@0 819 ILCID_POSIX_MAP(nl), /* nl Dutch 0x13 */
michael@0 820 /* ILCID_POSIX_MAP(nn), // no Norwegian 0x14 */
michael@0 821 ILCID_POSIX_MAP(no), /* * Norwegian 0x14 */
michael@0 822 ILCID_POSIX_MAP(nso), /* nso Sotho, Northern (Sepedi dialect) 0x6c */
michael@0 823 ILCID_POSIX_MAP(oc), /* oc Occitan 0x82 */
michael@0 824 ILCID_POSIX_MAP(om), /* om Oromo 0x72 */
michael@0 825 ILCID_POSIX_MAP(or_IN), /* or Oriya 0x48 */
michael@0 826 ILCID_POSIX_MAP(pa), /* pa Punjabi 0x46 */
michael@0 827 ILCID_POSIX_MAP(pap), /* pap Papiamentu 0x79 */
michael@0 828 ILCID_POSIX_MAP(pl), /* pl Polish 0x15 */
michael@0 829 ILCID_POSIX_MAP(ps), /* ps Pashto 0x63 */
michael@0 830 ILCID_POSIX_MAP(pt), /* pt Portuguese 0x16 */
michael@0 831 ILCID_POSIX_MAP(qu), /* qu Quechua 0x6B */
michael@0 832 ILCID_POSIX_MAP(qut), /* qut K'iche 0x86 */
michael@0 833 ILCID_POSIX_MAP(rm), /* rm Raeto-Romance/Romansh 0x17 */
michael@0 834 ILCID_POSIX_MAP(ro), /* ro Romanian 0x18 */
michael@0 835 ILCID_POSIX_MAP(root), /* root 0x00 */
michael@0 836 ILCID_POSIX_MAP(ru), /* ru Russian 0x19 */
michael@0 837 ILCID_POSIX_MAP(rw), /* rw Kinyarwanda 0x87 */
michael@0 838 ILCID_POSIX_MAP(sa), /* sa Sanskrit 0x4f */
michael@0 839 ILCID_POSIX_MAP(sah), /* sah Yakut 0x85 */
michael@0 840 ILCID_POSIX_MAP(sd), /* sd Sindhi 0x59 */
michael@0 841 ILCID_POSIX_MAP(se), /* se Sami 0x3b */
michael@0 842 /* ILCID_POSIX_MAP(sh), // sh Serbo-Croatian 0x1a */
michael@0 843 ILCID_POSIX_MAP(si), /* si Sinhalese 0x5b */
michael@0 844 ILCID_POSIX_MAP(sk), /* sk Slovak 0x1b */
michael@0 845 ILCID_POSIX_MAP(sl), /* sl Slovenian 0x24 */
michael@0 846 ILCID_POSIX_MAP(so), /* so Somali 0x77 */
michael@0 847 ILCID_POSIX_MAP(sq), /* sq Albanian 0x1c */
michael@0 848 /* ILCID_POSIX_MAP(sr), // sr Serbian 0x1a */
michael@0 849 ILCID_POSIX_MAP(st), /* st Sutu 0x30 */
michael@0 850 ILCID_POSIX_MAP(sv), /* sv Swedish 0x1d */
michael@0 851 ILCID_POSIX_MAP(sw), /* sw Swahili 0x41 */
michael@0 852 ILCID_POSIX_MAP(syr), /* syr Syriac 0x5A */
michael@0 853 ILCID_POSIX_MAP(ta), /* ta Tamil 0x49 */
michael@0 854 ILCID_POSIX_MAP(te), /* te Telugu 0x4a */
michael@0 855 ILCID_POSIX_MAP(tg), /* tg Tajik 0x28 */
michael@0 856 ILCID_POSIX_MAP(th), /* th Thai 0x1e */
michael@0 857 ILCID_POSIX_MAP(ti), /* ti Tigrigna 0x73 */
michael@0 858 ILCID_POSIX_MAP(tk), /* tk Turkmen 0x42 */
michael@0 859 ILCID_POSIX_MAP(tn), /* tn Tswana 0x32 */
michael@0 860 ILCID_POSIX_MAP(tr), /* tr Turkish 0x1f */
michael@0 861 ILCID_POSIX_MAP(ts), /* ts Tsonga 0x31 */
michael@0 862 ILCID_POSIX_MAP(tt), /* tt Tatar 0x44 */
michael@0 863 ILCID_POSIX_MAP(tzm), /* tzm Tamazight 0x5f */
michael@0 864 ILCID_POSIX_MAP(ug), /* ug Uighur 0x80 */
michael@0 865 ILCID_POSIX_MAP(uk), /* uk Ukrainian 0x22 */
michael@0 866 ILCID_POSIX_MAP(ur), /* ur Urdu 0x20 */
michael@0 867 ILCID_POSIX_MAP(uz), /* uz Uzbek 0x43 */
michael@0 868 ILCID_POSIX_MAP(ve), /* ve Venda 0x33 */
michael@0 869 ILCID_POSIX_MAP(vi), /* vi Vietnamese 0x2a */
michael@0 870 ILCID_POSIX_MAP(wen), /* wen Sorbian 0x2e */
michael@0 871 ILCID_POSIX_MAP(wo), /* wo Wolof 0x88 */
michael@0 872 ILCID_POSIX_MAP(xh), /* xh Xhosa 0x34 */
michael@0 873 ILCID_POSIX_MAP(yi), /* yi Yiddish 0x3d */
michael@0 874 ILCID_POSIX_MAP(yo), /* yo Yoruba 0x6a */
michael@0 875 ILCID_POSIX_MAP(zh), /* zh Chinese 0x04 */
michael@0 876 ILCID_POSIX_MAP(zu), /* zu Zulu 0x35 */
michael@0 877 };
michael@0 878
michael@0 879 static const uint32_t gLocaleCount = sizeof(gPosixIDmap)/sizeof(ILcidPosixMap);
michael@0 880
michael@0 881 /**
michael@0 882 * Do not call this function. It is called by hostID.
michael@0 883 * The function is not private because this struct must stay as a C struct,
michael@0 884 * and this is an internal class.
michael@0 885 */
michael@0 886 static int32_t
michael@0 887 idCmp(const char* id1, const char* id2)
michael@0 888 {
michael@0 889 int32_t diffIdx = 0;
michael@0 890 while (*id1 == *id2 && *id1 != 0) {
michael@0 891 diffIdx++;
michael@0 892 id1++;
michael@0 893 id2++;
michael@0 894 }
michael@0 895 return diffIdx;
michael@0 896 }
michael@0 897
michael@0 898 /**
michael@0 899 * Searches for a Windows LCID
michael@0 900 *
michael@0 901 * @param posixid the Posix style locale id.
michael@0 902 * @param status gets set to U_ILLEGAL_ARGUMENT_ERROR when the Posix ID has
michael@0 903 * no equivalent Windows LCID.
michael@0 904 * @return the LCID
michael@0 905 */
michael@0 906 static uint32_t
michael@0 907 getHostID(const ILcidPosixMap *this_0, const char* posixID, UErrorCode* status)
michael@0 908 {
michael@0 909 int32_t bestIdx = 0;
michael@0 910 int32_t bestIdxDiff = 0;
michael@0 911 int32_t posixIDlen = (int32_t)uprv_strlen(posixID);
michael@0 912 uint32_t idx;
michael@0 913
michael@0 914 for (idx = 0; idx < this_0->numRegions; idx++ ) {
michael@0 915 int32_t sameChars = idCmp(posixID, this_0->regionMaps[idx].posixID);
michael@0 916 if (sameChars > bestIdxDiff && this_0->regionMaps[idx].posixID[sameChars] == 0) {
michael@0 917 if (posixIDlen == sameChars) {
michael@0 918 /* Exact match */
michael@0 919 return this_0->regionMaps[idx].hostID;
michael@0 920 }
michael@0 921 bestIdxDiff = sameChars;
michael@0 922 bestIdx = idx;
michael@0 923 }
michael@0 924 }
michael@0 925 /* We asked for something unusual, like en_ZZ, and we try to return the number for the same language. */
michael@0 926 /* We also have to make sure that sid and si and similar string subsets don't match. */
michael@0 927 if ((posixID[bestIdxDiff] == '_' || posixID[bestIdxDiff] == '@')
michael@0 928 && this_0->regionMaps[bestIdx].posixID[bestIdxDiff] == 0)
michael@0 929 {
michael@0 930 *status = U_USING_FALLBACK_WARNING;
michael@0 931 return this_0->regionMaps[bestIdx].hostID;
michael@0 932 }
michael@0 933
michael@0 934 /*no match found */
michael@0 935 *status = U_ILLEGAL_ARGUMENT_ERROR;
michael@0 936 return this_0->regionMaps->hostID;
michael@0 937 }
michael@0 938
michael@0 939 static const char*
michael@0 940 getPosixID(const ILcidPosixMap *this_0, uint32_t hostID)
michael@0 941 {
michael@0 942 uint32_t i;
michael@0 943 for (i = 0; i <= this_0->numRegions; i++)
michael@0 944 {
michael@0 945 if (this_0->regionMaps[i].hostID == hostID)
michael@0 946 {
michael@0 947 return this_0->regionMaps[i].posixID;
michael@0 948 }
michael@0 949 }
michael@0 950
michael@0 951 /* If you get here, then no matching region was found,
michael@0 952 so return the language id with the wild card region. */
michael@0 953 return this_0->regionMaps[0].posixID;
michael@0 954 }
michael@0 955
michael@0 956 /*
michael@0 957 //////////////////////////////////////
michael@0 958 //
michael@0 959 // LCID --> POSIX
michael@0 960 //
michael@0 961 /////////////////////////////////////
michael@0 962 */
michael@0 963 #ifdef USE_WINDOWS_LOCALE_API
michael@0 964 /*
michael@0 965 * Various language tags needs to be changed:
michael@0 966 * quz -> qu
michael@0 967 * prs -> fa
michael@0 968 */
michael@0 969 #define FIX_LANGUAGE_ID_TAG(buffer, len) \
michael@0 970 if (len >= 3) { \
michael@0 971 if (buffer[0] == 'q' && buffer[1] == 'u' && buffer[2] == 'z') {\
michael@0 972 buffer[2] = 0; \
michael@0 973 uprv_strcat(buffer, buffer+3); \
michael@0 974 } else if (buffer[0] == 'p' && buffer[1] == 'r' && buffer[2] == 's') {\
michael@0 975 buffer[0] = 'f'; buffer[1] = 'a'; buffer[2] = 0; \
michael@0 976 uprv_strcat(buffer, buffer+3); \
michael@0 977 } \
michael@0 978 }
michael@0 979
michael@0 980 #endif
michael@0 981 U_CAPI int32_t
michael@0 982 uprv_convertToPosix(uint32_t hostid, char *posixID, int32_t posixIDCapacity, UErrorCode* status)
michael@0 983 {
michael@0 984 uint16_t langID;
michael@0 985 uint32_t localeIndex;
michael@0 986 UBool bLookup = TRUE;
michael@0 987 const char *pPosixID = NULL;
michael@0 988
michael@0 989 #ifdef USE_WINDOWS_LOCALE_API
michael@0 990 int32_t tmpLen = 0;
michael@0 991 char locName[157]; /* ULOC_FULLNAME_CAPACITY */
michael@0 992
michael@0 993 tmpLen = GetLocaleInfoA(hostid, LOCALE_SNAME, (LPSTR)locName, sizeof(locName)/sizeof(locName[0]));
michael@0 994 if (tmpLen > 1) {
michael@0 995 /* Windows locale name may contain sorting variant, such as "es-ES_tradnl".
michael@0 996 In such case, we need special mapping data found in the hardcoded table
michael@0 997 in this source file. */
michael@0 998 char *p = uprv_strchr(locName, '_');
michael@0 999 if (p) {
michael@0 1000 /* Keep the base locale, without variant */
michael@0 1001 *p = 0;
michael@0 1002 tmpLen = uprv_strlen(locName);
michael@0 1003 } else {
michael@0 1004 /* No hardcoded table lookup necessary */
michael@0 1005 bLookup = FALSE;
michael@0 1006 }
michael@0 1007 /* Change the tag separator from '-' to '_' */
michael@0 1008 p = locName;
michael@0 1009 while (*p) {
michael@0 1010 if (*p == '-') {
michael@0 1011 *p = '_';
michael@0 1012 }
michael@0 1013 p++;
michael@0 1014 }
michael@0 1015 FIX_LANGUAGE_ID_TAG(locName, tmpLen);
michael@0 1016 pPosixID = locName;
michael@0 1017 }
michael@0 1018 #endif
michael@0 1019 if (bLookup) {
michael@0 1020 const char *pCandidate = NULL;
michael@0 1021 langID = LANGUAGE_LCID(hostid);
michael@0 1022
michael@0 1023 for (localeIndex = 0; localeIndex < gLocaleCount; localeIndex++) {
michael@0 1024 if (langID == gPosixIDmap[localeIndex].regionMaps->hostID) {
michael@0 1025 pCandidate = getPosixID(&gPosixIDmap[localeIndex], hostid);
michael@0 1026 break;
michael@0 1027 }
michael@0 1028 }
michael@0 1029
michael@0 1030 /* On Windows, when locale name has a variant, we still look up the hardcoded table.
michael@0 1031 If a match in the hardcoded table is longer than the Windows locale name without
michael@0 1032 variant, we use the one as the result */
michael@0 1033 if (pCandidate && (pPosixID == NULL || uprv_strlen(pCandidate) > uprv_strlen(pPosixID))) {
michael@0 1034 pPosixID = pCandidate;
michael@0 1035 }
michael@0 1036 }
michael@0 1037
michael@0 1038 if (pPosixID) {
michael@0 1039 int32_t resLen = uprv_strlen(pPosixID);
michael@0 1040 int32_t copyLen = resLen <= posixIDCapacity ? resLen : posixIDCapacity;
michael@0 1041 uprv_memcpy(posixID, pPosixID, copyLen);
michael@0 1042 if (resLen < posixIDCapacity) {
michael@0 1043 posixID[resLen] = 0;
michael@0 1044 if (*status == U_STRING_NOT_TERMINATED_WARNING) {
michael@0 1045 *status = U_ZERO_ERROR;
michael@0 1046 }
michael@0 1047 } else if (resLen == posixIDCapacity) {
michael@0 1048 *status = U_STRING_NOT_TERMINATED_WARNING;
michael@0 1049 } else {
michael@0 1050 *status = U_BUFFER_OVERFLOW_ERROR;
michael@0 1051 }
michael@0 1052 return resLen;
michael@0 1053 }
michael@0 1054
michael@0 1055 /* no match found */
michael@0 1056 *status = U_ILLEGAL_ARGUMENT_ERROR;
michael@0 1057 return -1;
michael@0 1058 }
michael@0 1059
michael@0 1060 /*
michael@0 1061 //////////////////////////////////////
michael@0 1062 //
michael@0 1063 // POSIX --> LCID
michael@0 1064 // This should only be called from uloc_getLCID.
michael@0 1065 // The locale ID must be in canonical form.
michael@0 1066 // langID is separate so that this file doesn't depend on the uloc_* API.
michael@0 1067 //
michael@0 1068 /////////////////////////////////////
michael@0 1069 */
michael@0 1070
michael@0 1071 U_CAPI uint32_t
michael@0 1072 uprv_convertToLCID(const char *langID, const char* posixID, UErrorCode* status)
michael@0 1073 {
michael@0 1074
michael@0 1075 uint32_t low = 0;
michael@0 1076 uint32_t high = gLocaleCount;
michael@0 1077 uint32_t mid;
michael@0 1078 uint32_t oldmid = 0;
michael@0 1079 int32_t compVal;
michael@0 1080
michael@0 1081 uint32_t value = 0;
michael@0 1082 uint32_t fallbackValue = (uint32_t)-1;
michael@0 1083 UErrorCode myStatus;
michael@0 1084 uint32_t idx;
michael@0 1085
michael@0 1086 /* Check for incomplete id. */
michael@0 1087 if (!langID || !posixID || uprv_strlen(langID) < 2 || uprv_strlen(posixID) < 2) {
michael@0 1088 return 0;
michael@0 1089 }
michael@0 1090
michael@0 1091 /*Binary search for the map entry for normal cases */
michael@0 1092
michael@0 1093 while (high > low) /*binary search*/{
michael@0 1094
michael@0 1095 mid = (high+low) >> 1; /*Finds median*/
michael@0 1096
michael@0 1097 if (mid == oldmid)
michael@0 1098 break;
michael@0 1099
michael@0 1100 compVal = uprv_strcmp(langID, gPosixIDmap[mid].regionMaps->posixID);
michael@0 1101 if (compVal < 0){
michael@0 1102 high = mid;
michael@0 1103 }
michael@0 1104 else if (compVal > 0){
michael@0 1105 low = mid;
michael@0 1106 }
michael@0 1107 else /*we found it*/{
michael@0 1108 return getHostID(&gPosixIDmap[mid], posixID, status);
michael@0 1109 }
michael@0 1110 oldmid = mid;
michael@0 1111 }
michael@0 1112
michael@0 1113 /*
michael@0 1114 * Sometimes we can't do a binary search on posixID because some LCIDs
michael@0 1115 * go to different locales. We hit one of those special cases.
michael@0 1116 */
michael@0 1117 for (idx = 0; idx < gLocaleCount; idx++ ) {
michael@0 1118 myStatus = U_ZERO_ERROR;
michael@0 1119 value = getHostID(&gPosixIDmap[idx], posixID, &myStatus);
michael@0 1120 if (myStatus == U_ZERO_ERROR) {
michael@0 1121 return value;
michael@0 1122 }
michael@0 1123 else if (myStatus == U_USING_FALLBACK_WARNING) {
michael@0 1124 fallbackValue = value;
michael@0 1125 }
michael@0 1126 }
michael@0 1127
michael@0 1128 if (fallbackValue != (uint32_t)-1) {
michael@0 1129 *status = U_USING_FALLBACK_WARNING;
michael@0 1130 return fallbackValue;
michael@0 1131 }
michael@0 1132
michael@0 1133 /* no match found */
michael@0 1134 *status = U_ILLEGAL_ARGUMENT_ERROR;
michael@0 1135 return 0; /* return international (root) */
michael@0 1136 }
michael@0 1137

Mercurial Repository: The Tor Browser

Europalab SCM Repository Server

mercurial