| 1 | // © 2016 and later: Unicode, Inc. and others. | 
|---|
| 2 | // License & terms of use: http://www.unicode.org/copyright.html | 
|---|
| 3 | /* | 
|---|
| 4 | ******************************************************************************* | 
|---|
| 5 | *   Copyright (C) 1996-2016, International Business Machines | 
|---|
| 6 | *   Corporation and others.  All Rights Reserved. | 
|---|
| 7 | ******************************************************************************* | 
|---|
| 8 | *   file name:  ucol_res.cpp | 
|---|
| 9 | *   encoding:   UTF-8 | 
|---|
| 10 | *   tab size:   8 (not used) | 
|---|
| 11 | *   indentation:4 | 
|---|
| 12 | * | 
|---|
| 13 | * Description: | 
|---|
| 14 | * This file contains dependencies that the collation run-time doesn't normally | 
|---|
| 15 | * need. This mainly contains resource bundle usage and collation meta information | 
|---|
| 16 | * | 
|---|
| 17 | * Modification history | 
|---|
| 18 | * Date        Name      Comments | 
|---|
| 19 | * 1996-1999   various members of ICU team maintained C API for collation framework | 
|---|
| 20 | * 02/16/2001  synwee    Added internal method getPrevSpecialCE | 
|---|
| 21 | * 03/01/2001  synwee    Added maxexpansion functionality. | 
|---|
| 22 | * 03/16/2001  weiv      Collation framework is rewritten in C and made UCA compliant | 
|---|
| 23 | * 12/08/2004  grhoten   Split part of ucol.cpp into ucol_res.cpp | 
|---|
| 24 | * 2012-2014   markus    Rewritten in C++ again. | 
|---|
| 25 | */ | 
|---|
| 26 |  | 
|---|
| 27 | #include "unicode/utypes.h" | 
|---|
| 28 |  | 
|---|
| 29 | #if !UCONFIG_NO_COLLATION | 
|---|
| 30 |  | 
|---|
| 31 | #include "unicode/coll.h" | 
|---|
| 32 | #include "unicode/localpointer.h" | 
|---|
| 33 | #include "unicode/locid.h" | 
|---|
| 34 | #include "unicode/tblcoll.h" | 
|---|
| 35 | #include "unicode/ucol.h" | 
|---|
| 36 | #include "unicode/uloc.h" | 
|---|
| 37 | #include "unicode/unistr.h" | 
|---|
| 38 | #include "unicode/ures.h" | 
|---|
| 39 | #include "charstr.h" | 
|---|
| 40 | #include "cmemory.h" | 
|---|
| 41 | #include "cstring.h" | 
|---|
| 42 | #include "collationdatareader.h" | 
|---|
| 43 | #include "collationroot.h" | 
|---|
| 44 | #include "collationtailoring.h" | 
|---|
| 45 | #include "resource.h" | 
|---|
| 46 | #include "putilimp.h" | 
|---|
| 47 | #include "uassert.h" | 
|---|
| 48 | #include "ucln_in.h" | 
|---|
| 49 | #include "ucol_imp.h" | 
|---|
| 50 | #include "uenumimp.h" | 
|---|
| 51 | #include "ulist.h" | 
|---|
| 52 | #include "umutex.h" | 
|---|
| 53 | #include "unifiedcache.h" | 
|---|
| 54 | #include "uresimp.h" | 
|---|
| 55 | #include "ustrenum.h" | 
|---|
| 56 | #include "utracimp.h" | 
|---|
| 57 |  | 
|---|
| 58 | U_NAMESPACE_BEGIN | 
|---|
| 59 |  | 
|---|
| 60 | namespace { | 
|---|
| 61 |  | 
|---|
| 62 | static const UChar *rootRules = NULL; | 
|---|
| 63 | static int32_t rootRulesLength = 0; | 
|---|
| 64 | static UResourceBundle *rootBundle = NULL; | 
|---|
| 65 | static UInitOnce gInitOnceUcolRes = U_INITONCE_INITIALIZER; | 
|---|
| 66 |  | 
|---|
| 67 | }  // namespace | 
|---|
| 68 |  | 
|---|
| 69 | U_CDECL_BEGIN | 
|---|
| 70 |  | 
|---|
| 71 | static UBool U_CALLCONV | 
|---|
| 72 | ucol_res_cleanup() { | 
|---|
| 73 | rootRules = NULL; | 
|---|
| 74 | rootRulesLength = 0; | 
|---|
| 75 | ures_close(rootBundle); | 
|---|
| 76 | rootBundle = NULL; | 
|---|
| 77 | gInitOnceUcolRes.reset(); | 
|---|
| 78 | return TRUE; | 
|---|
| 79 | } | 
|---|
| 80 |  | 
|---|
| 81 | void U_CALLCONV | 
|---|
| 82 | CollationLoader::loadRootRules(UErrorCode &errorCode) { | 
|---|
| 83 | if(U_FAILURE(errorCode)) { return; } | 
|---|
| 84 | rootBundle = ures_open(U_ICUDATA_COLL, kRootLocaleName, &errorCode); | 
|---|
| 85 | if(U_FAILURE(errorCode)) { return; } | 
|---|
| 86 | rootRules = ures_getStringByKey(rootBundle, "UCARules", &rootRulesLength, &errorCode); | 
|---|
| 87 | if(U_FAILURE(errorCode)) { | 
|---|
| 88 | ures_close(rootBundle); | 
|---|
| 89 | rootBundle = NULL; | 
|---|
| 90 | return; | 
|---|
| 91 | } | 
|---|
| 92 | ucln_i18n_registerCleanup(UCLN_I18N_UCOL_RES, ucol_res_cleanup); | 
|---|
| 93 | } | 
|---|
| 94 |  | 
|---|
| 95 | U_CDECL_END | 
|---|
| 96 |  | 
|---|
| 97 | void | 
|---|
| 98 | CollationLoader::appendRootRules(UnicodeString &s) { | 
|---|
| 99 | UErrorCode errorCode = U_ZERO_ERROR; | 
|---|
| 100 | umtx_initOnce(gInitOnceUcolRes, CollationLoader::loadRootRules, errorCode); | 
|---|
| 101 | if(U_SUCCESS(errorCode)) { | 
|---|
| 102 | s.append(rootRules, rootRulesLength); | 
|---|
| 103 | } | 
|---|
| 104 | } | 
|---|
| 105 |  | 
|---|
| 106 | void | 
|---|
| 107 | CollationLoader::loadRules(const char *localeID, const char *collationType, | 
|---|
| 108 | UnicodeString &rules, UErrorCode &errorCode) { | 
|---|
| 109 | if(U_FAILURE(errorCode)) { return; } | 
|---|
| 110 | U_ASSERT(collationType != NULL && *collationType != 0); | 
|---|
| 111 | // Copy the type for lowercasing. | 
|---|
| 112 | char type[16]; | 
|---|
| 113 | int32_t typeLength = static_cast<int32_t>(uprv_strlen(collationType)); | 
|---|
| 114 | if(typeLength >= UPRV_LENGTHOF(type)) { | 
|---|
| 115 | errorCode = U_ILLEGAL_ARGUMENT_ERROR; | 
|---|
| 116 | return; | 
|---|
| 117 | } | 
|---|
| 118 | uprv_memcpy(type, collationType, typeLength + 1); | 
|---|
| 119 | T_CString_toLowerCase(type); | 
|---|
| 120 |  | 
|---|
| 121 | LocalUResourceBundlePointer bundle(ures_open(U_ICUDATA_COLL, localeID, &errorCode)); | 
|---|
| 122 | LocalUResourceBundlePointer collations( | 
|---|
| 123 | ures_getByKey(bundle.getAlias(), "collations", NULL, &errorCode)); | 
|---|
| 124 | LocalUResourceBundlePointer data( | 
|---|
| 125 | ures_getByKeyWithFallback(collations.getAlias(), type, NULL, &errorCode)); | 
|---|
| 126 | int32_t length; | 
|---|
| 127 | const UChar *s =  ures_getStringByKey(data.getAlias(), "Sequence", &length, &errorCode); | 
|---|
| 128 | if(U_FAILURE(errorCode)) { return; } | 
|---|
| 129 |  | 
|---|
| 130 | // No string pointer aliasing so that we need not hold onto the resource bundle. | 
|---|
| 131 | rules.setTo(s, length); | 
|---|
| 132 | if(rules.isBogus()) { | 
|---|
| 133 | errorCode = U_MEMORY_ALLOCATION_ERROR; | 
|---|
| 134 | } | 
|---|
| 135 | } | 
|---|
| 136 |  | 
|---|
| 137 | template<> U_I18N_API | 
|---|
| 138 | const CollationCacheEntry * | 
|---|
| 139 | LocaleCacheKey<CollationCacheEntry>::createObject(const void *creationContext, | 
|---|
| 140 | UErrorCode &errorCode) const { | 
|---|
| 141 | CollationLoader *loader = | 
|---|
| 142 | reinterpret_cast<CollationLoader *>( | 
|---|
| 143 | const_cast<void *>(creationContext)); | 
|---|
| 144 | return loader->createCacheEntry(errorCode); | 
|---|
| 145 | } | 
|---|
| 146 |  | 
|---|
| 147 | const CollationCacheEntry * | 
|---|
| 148 | CollationLoader::loadTailoring(const Locale &locale, UErrorCode &errorCode) { | 
|---|
| 149 | const CollationCacheEntry *rootEntry = CollationRoot::getRootCacheEntry(errorCode); | 
|---|
| 150 | if(U_FAILURE(errorCode)) { return NULL; } | 
|---|
| 151 | const char *name = locale.getName(); | 
|---|
| 152 | if(*name == 0 || uprv_strcmp(name, "root") == 0) { | 
|---|
| 153 |  | 
|---|
| 154 | // Have to add a ref. | 
|---|
| 155 | rootEntry->addRef(); | 
|---|
| 156 | return rootEntry; | 
|---|
| 157 | } | 
|---|
| 158 |  | 
|---|
| 159 | // Clear warning codes before loading where they get cached. | 
|---|
| 160 | errorCode = U_ZERO_ERROR; | 
|---|
| 161 | CollationLoader loader(rootEntry, locale, errorCode); | 
|---|
| 162 |  | 
|---|
| 163 | // getCacheEntry adds a ref for us. | 
|---|
| 164 | return loader.getCacheEntry(errorCode); | 
|---|
| 165 | } | 
|---|
| 166 |  | 
|---|
| 167 | CollationLoader::CollationLoader(const CollationCacheEntry *re, const Locale &requested, | 
|---|
| 168 | UErrorCode &errorCode) | 
|---|
| 169 | : cache(UnifiedCache::getInstance(errorCode)), rootEntry(re), | 
|---|
| 170 | validLocale(re->validLocale), locale(requested), | 
|---|
| 171 | typesTried(0), typeFallback(FALSE), | 
|---|
| 172 | bundle(NULL), collations(NULL), data(NULL) { | 
|---|
| 173 | type[0] = 0; | 
|---|
| 174 | defaultType[0] = 0; | 
|---|
| 175 | if(U_FAILURE(errorCode)) { return; } | 
|---|
| 176 |  | 
|---|
| 177 | // Canonicalize the locale ID: Ignore all irrelevant keywords. | 
|---|
| 178 | const char *baseName = locale.getBaseName(); | 
|---|
| 179 | if(uprv_strcmp(locale.getName(), baseName) != 0) { | 
|---|
| 180 | locale = Locale(baseName); | 
|---|
| 181 |  | 
|---|
| 182 | // Fetch the collation type from the locale ID. | 
|---|
| 183 | int32_t typeLength = requested.getKeywordValue( "collation", | 
|---|
| 184 | type, UPRV_LENGTHOF(type) - 1, errorCode); | 
|---|
| 185 | if(U_FAILURE(errorCode)) { | 
|---|
| 186 | errorCode = U_ILLEGAL_ARGUMENT_ERROR; | 
|---|
| 187 | return; | 
|---|
| 188 | } | 
|---|
| 189 | type[typeLength] = 0;  // in case of U_NOT_TERMINATED_WARNING | 
|---|
| 190 | if(typeLength == 0) { | 
|---|
| 191 | // No collation type. | 
|---|
| 192 | } else if(uprv_stricmp(type, "default") == 0) { | 
|---|
| 193 | // Ignore "default" (case-insensitive). | 
|---|
| 194 | type[0] = 0; | 
|---|
| 195 | } else { | 
|---|
| 196 | // Copy the collation type. | 
|---|
| 197 | T_CString_toLowerCase(type); | 
|---|
| 198 | locale.setKeywordValue( "collation", type, errorCode); | 
|---|
| 199 | } | 
|---|
| 200 | } | 
|---|
| 201 | } | 
|---|
| 202 |  | 
|---|
| 203 | CollationLoader::~CollationLoader() { | 
|---|
| 204 | ures_close(data); | 
|---|
| 205 | ures_close(collations); | 
|---|
| 206 | ures_close(bundle); | 
|---|
| 207 | } | 
|---|
| 208 |  | 
|---|
| 209 | const CollationCacheEntry * | 
|---|
| 210 | CollationLoader::createCacheEntry(UErrorCode &errorCode) { | 
|---|
| 211 | // This is a linear lookup and fallback flow turned into a state machine. | 
|---|
| 212 | // Most local variables have been turned into instance fields. | 
|---|
| 213 | // In a cache miss, cache.get() calls CacheKey::createObject(), | 
|---|
| 214 | // which means that we progress via recursion. | 
|---|
| 215 | // loadFromCollations() will recurse to itself as well for collation type fallback. | 
|---|
| 216 | if(bundle == NULL) { | 
|---|
| 217 | return loadFromLocale(errorCode); | 
|---|
| 218 | } else if(collations == NULL) { | 
|---|
| 219 | return loadFromBundle(errorCode); | 
|---|
| 220 | } else if(data == NULL) { | 
|---|
| 221 | return loadFromCollations(errorCode); | 
|---|
| 222 | } else { | 
|---|
| 223 | return loadFromData(errorCode); | 
|---|
| 224 | } | 
|---|
| 225 | } | 
|---|
| 226 |  | 
|---|
| 227 | const CollationCacheEntry * | 
|---|
| 228 | CollationLoader::loadFromLocale(UErrorCode &errorCode) { | 
|---|
| 229 | if(U_FAILURE(errorCode)) { return NULL; } | 
|---|
| 230 | U_ASSERT(bundle == NULL); | 
|---|
| 231 | bundle = ures_openNoDefault(U_ICUDATA_COLL, locale.getBaseName(), &errorCode); | 
|---|
| 232 | if(errorCode == U_MISSING_RESOURCE_ERROR) { | 
|---|
| 233 | errorCode = U_USING_DEFAULT_WARNING; | 
|---|
| 234 |  | 
|---|
| 235 | // Have to add that ref that we promise. | 
|---|
| 236 | rootEntry->addRef(); | 
|---|
| 237 | return rootEntry; | 
|---|
| 238 | } | 
|---|
| 239 | Locale requestedLocale(locale); | 
|---|
| 240 | const char *vLocale = ures_getLocaleByType(bundle, ULOC_ACTUAL_LOCALE, &errorCode); | 
|---|
| 241 | if(U_FAILURE(errorCode)) { return NULL; } | 
|---|
| 242 | locale = validLocale = Locale(vLocale);  // no type until loadFromCollations() | 
|---|
| 243 | if(type[0] != 0) { | 
|---|
| 244 | locale.setKeywordValue( "collation", type, errorCode); | 
|---|
| 245 | } | 
|---|
| 246 | if(locale != requestedLocale) { | 
|---|
| 247 | return getCacheEntry(errorCode); | 
|---|
| 248 | } else { | 
|---|
| 249 | return loadFromBundle(errorCode); | 
|---|
| 250 | } | 
|---|
| 251 | } | 
|---|
| 252 |  | 
|---|
| 253 | const CollationCacheEntry * | 
|---|
| 254 | CollationLoader::loadFromBundle(UErrorCode &errorCode) { | 
|---|
| 255 | if(U_FAILURE(errorCode)) { return NULL; } | 
|---|
| 256 | U_ASSERT(collations == NULL); | 
|---|
| 257 | // There are zero or more tailorings in the collations table. | 
|---|
| 258 | collations = ures_getByKey(bundle, "collations", NULL, &errorCode); | 
|---|
| 259 | if(errorCode == U_MISSING_RESOURCE_ERROR) { | 
|---|
| 260 | errorCode = U_USING_DEFAULT_WARNING; | 
|---|
| 261 | // Return the root tailoring with the validLocale, without collation type. | 
|---|
| 262 | return makeCacheEntryFromRoot(validLocale, errorCode); | 
|---|
| 263 | } | 
|---|
| 264 | if(U_FAILURE(errorCode)) { return NULL; } | 
|---|
| 265 |  | 
|---|
| 266 | // Fetch the default type from the data. | 
|---|
| 267 | { | 
|---|
| 268 | UErrorCode internalErrorCode = U_ZERO_ERROR; | 
|---|
| 269 | LocalUResourceBundlePointer def( | 
|---|
| 270 | ures_getByKeyWithFallback(collations, "default", NULL, &internalErrorCode)); | 
|---|
| 271 | int32_t length; | 
|---|
| 272 | const UChar *s = ures_getString(def.getAlias(), &length, &internalErrorCode); | 
|---|
| 273 | if(U_SUCCESS(internalErrorCode) && 0 < length && length < UPRV_LENGTHOF(defaultType)) { | 
|---|
| 274 | u_UCharsToChars(s, defaultType, length + 1); | 
|---|
| 275 | } else { | 
|---|
| 276 | uprv_strcpy(defaultType, "standard"); | 
|---|
| 277 | } | 
|---|
| 278 | } | 
|---|
| 279 |  | 
|---|
| 280 | // Record which collation types we have looked for already, | 
|---|
| 281 | // so that we do not deadlock in the cache. | 
|---|
| 282 | // | 
|---|
| 283 | // If there is no explicit type, then we look in the cache | 
|---|
| 284 | // for the entry with the default type. | 
|---|
| 285 | // If the explicit type is the default type, then we do not look in the cache | 
|---|
| 286 | // for the entry with an empty type. | 
|---|
| 287 | // Otherwise, two concurrent requests with opposite fallbacks would deadlock each other. | 
|---|
| 288 | // Also, it is easier to always enter the next method with a non-empty type. | 
|---|
| 289 | if(type[0] == 0) { | 
|---|
| 290 | uprv_strcpy(type, defaultType); | 
|---|
| 291 | typesTried |= TRIED_DEFAULT; | 
|---|
| 292 | if(uprv_strcmp(type, "search") == 0) { | 
|---|
| 293 | typesTried |= TRIED_SEARCH; | 
|---|
| 294 | } | 
|---|
| 295 | if(uprv_strcmp(type, "standard") == 0) { | 
|---|
| 296 | typesTried |= TRIED_STANDARD; | 
|---|
| 297 | } | 
|---|
| 298 | locale.setKeywordValue( "collation", type, errorCode); | 
|---|
| 299 | return getCacheEntry(errorCode); | 
|---|
| 300 | } else { | 
|---|
| 301 | if(uprv_strcmp(type, defaultType) == 0) { | 
|---|
| 302 | typesTried |= TRIED_DEFAULT; | 
|---|
| 303 | } | 
|---|
| 304 | if(uprv_strcmp(type, "search") == 0) { | 
|---|
| 305 | typesTried |= TRIED_SEARCH; | 
|---|
| 306 | } | 
|---|
| 307 | if(uprv_strcmp(type, "standard") == 0) { | 
|---|
| 308 | typesTried |= TRIED_STANDARD; | 
|---|
| 309 | } | 
|---|
| 310 | return loadFromCollations(errorCode); | 
|---|
| 311 | } | 
|---|
| 312 | } | 
|---|
| 313 |  | 
|---|
| 314 | const CollationCacheEntry * | 
|---|
| 315 | CollationLoader::loadFromCollations(UErrorCode &errorCode) { | 
|---|
| 316 | if(U_FAILURE(errorCode)) { return NULL; } | 
|---|
| 317 | U_ASSERT(data == NULL); | 
|---|
| 318 | // Load the collations/type tailoring, with type fallback. | 
|---|
| 319 | LocalUResourceBundlePointer localData( | 
|---|
| 320 | ures_getByKeyWithFallback(collations, type, NULL, &errorCode)); | 
|---|
| 321 | int32_t typeLength = static_cast<int32_t>(uprv_strlen(type)); | 
|---|
| 322 | if(errorCode == U_MISSING_RESOURCE_ERROR) { | 
|---|
| 323 | errorCode = U_USING_DEFAULT_WARNING; | 
|---|
| 324 | typeFallback = TRUE; | 
|---|
| 325 | if((typesTried & TRIED_SEARCH) == 0 && | 
|---|
| 326 | typeLength > 6 && uprv_strncmp(type, "search", 6) == 0) { | 
|---|
| 327 | // fall back from something like "searchjl" to "search" | 
|---|
| 328 | typesTried |= TRIED_SEARCH; | 
|---|
| 329 | type[6] = 0; | 
|---|
| 330 | } else if((typesTried & TRIED_DEFAULT) == 0) { | 
|---|
| 331 | // fall back to the default type | 
|---|
| 332 | typesTried |= TRIED_DEFAULT; | 
|---|
| 333 | uprv_strcpy(type, defaultType); | 
|---|
| 334 | } else if((typesTried & TRIED_STANDARD) == 0) { | 
|---|
| 335 | // fall back to the "standard" type | 
|---|
| 336 | typesTried |= TRIED_STANDARD; | 
|---|
| 337 | uprv_strcpy(type, "standard"); | 
|---|
| 338 | } else { | 
|---|
| 339 | // Return the root tailoring with the validLocale, without collation type. | 
|---|
| 340 | return makeCacheEntryFromRoot(validLocale, errorCode); | 
|---|
| 341 | } | 
|---|
| 342 | locale.setKeywordValue( "collation", type, errorCode); | 
|---|
| 343 | return getCacheEntry(errorCode); | 
|---|
| 344 | } | 
|---|
| 345 | if(U_FAILURE(errorCode)) { return NULL; } | 
|---|
| 346 |  | 
|---|
| 347 | data = localData.orphan(); | 
|---|
| 348 | const char *actualLocale = ures_getLocaleByType(data, ULOC_ACTUAL_LOCALE, &errorCode); | 
|---|
| 349 | if(U_FAILURE(errorCode)) { return NULL; } | 
|---|
| 350 | const char *vLocale = validLocale.getBaseName(); | 
|---|
| 351 | UBool actualAndValidLocalesAreDifferent = Locale(actualLocale) != Locale(vLocale); | 
|---|
| 352 |  | 
|---|
| 353 | // Set the collation types on the informational locales, | 
|---|
| 354 | // except when they match the default types (for brevity and backwards compatibility). | 
|---|
| 355 | // For the valid locale, suppress the default type. | 
|---|
| 356 | if(uprv_strcmp(type, defaultType) != 0) { | 
|---|
| 357 | validLocale.setKeywordValue( "collation", type, errorCode); | 
|---|
| 358 | if(U_FAILURE(errorCode)) { return NULL; } | 
|---|
| 359 | } | 
|---|
| 360 |  | 
|---|
| 361 | // Is this the same as the root collator? If so, then use that instead. | 
|---|
| 362 | if((*actualLocale == 0 || uprv_strcmp(actualLocale, "root") == 0) && | 
|---|
| 363 | uprv_strcmp(type, "standard") == 0) { | 
|---|
| 364 | if(typeFallback) { | 
|---|
| 365 | errorCode = U_USING_DEFAULT_WARNING; | 
|---|
| 366 | } | 
|---|
| 367 | return makeCacheEntryFromRoot(validLocale, errorCode); | 
|---|
| 368 | } | 
|---|
| 369 |  | 
|---|
| 370 | locale = Locale(actualLocale); | 
|---|
| 371 | if(actualAndValidLocalesAreDifferent) { | 
|---|
| 372 | locale.setKeywordValue( "collation", type, errorCode); | 
|---|
| 373 | const CollationCacheEntry *entry = getCacheEntry(errorCode); | 
|---|
| 374 | return makeCacheEntry(validLocale, entry, errorCode); | 
|---|
| 375 | } else { | 
|---|
| 376 | return loadFromData(errorCode); | 
|---|
| 377 | } | 
|---|
| 378 | } | 
|---|
| 379 |  | 
|---|
| 380 | const CollationCacheEntry * | 
|---|
| 381 | CollationLoader::loadFromData(UErrorCode &errorCode) { | 
|---|
| 382 | if(U_FAILURE(errorCode)) { return NULL; } | 
|---|
| 383 | LocalPointer<CollationTailoring> t(new CollationTailoring(rootEntry->tailoring->settings)); | 
|---|
| 384 | if(t.isNull() || t->isBogus()) { | 
|---|
| 385 | errorCode = U_MEMORY_ALLOCATION_ERROR; | 
|---|
| 386 | return NULL; | 
|---|
| 387 | } | 
|---|
| 388 |  | 
|---|
| 389 | // deserialize | 
|---|
| 390 | LocalUResourceBundlePointer binary(ures_getByKey(data, "%%CollationBin", NULL, &errorCode)); | 
|---|
| 391 | // Note: U_MISSING_RESOURCE_ERROR --> The old code built from rules if available | 
|---|
| 392 | // but that created undesirable dependencies. | 
|---|
| 393 | int32_t length; | 
|---|
| 394 | const uint8_t *inBytes = ures_getBinary(binary.getAlias(), &length, &errorCode); | 
|---|
| 395 | CollationDataReader::read(rootEntry->tailoring, inBytes, length, *t, errorCode); | 
|---|
| 396 | // Note: U_COLLATOR_VERSION_MISMATCH --> The old code built from rules if available | 
|---|
| 397 | // but that created undesirable dependencies. | 
|---|
| 398 | if(U_FAILURE(errorCode)) { return NULL; } | 
|---|
| 399 |  | 
|---|
| 400 | // Try to fetch the optional rules string. | 
|---|
| 401 | { | 
|---|
| 402 | UErrorCode internalErrorCode = U_ZERO_ERROR; | 
|---|
| 403 | int32_t len; | 
|---|
| 404 | const UChar *s = ures_getStringByKey(data, "Sequence", &len, | 
|---|
| 405 | &internalErrorCode); | 
|---|
| 406 | if(U_SUCCESS(internalErrorCode)) { | 
|---|
| 407 | t->rules.setTo(TRUE, s, len); | 
|---|
| 408 | } | 
|---|
| 409 | } | 
|---|
| 410 |  | 
|---|
| 411 | const char *actualLocale = locale.getBaseName();  // without type | 
|---|
| 412 | const char *vLocale = validLocale.getBaseName(); | 
|---|
| 413 | UBool actualAndValidLocalesAreDifferent = Locale(actualLocale) != Locale(vLocale); | 
|---|
| 414 |  | 
|---|
| 415 | // For the actual locale, suppress the default type *according to the actual locale*. | 
|---|
| 416 | // For example, zh has default=pinyin and contains all of the Chinese tailorings. | 
|---|
| 417 | // zh_Hant has default=stroke but has no other data. | 
|---|
| 418 | // For the valid locale "zh_Hant" we need to suppress stroke. | 
|---|
| 419 | // For the actual locale "zh" we need to suppress pinyin instead. | 
|---|
| 420 | if(actualAndValidLocalesAreDifferent) { | 
|---|
| 421 | // Opening a bundle for the actual locale should always succeed. | 
|---|
| 422 | LocalUResourceBundlePointer actualBundle( | 
|---|
| 423 | ures_open(U_ICUDATA_COLL, actualLocale, &errorCode)); | 
|---|
| 424 | if(U_FAILURE(errorCode)) { return NULL; } | 
|---|
| 425 | UErrorCode internalErrorCode = U_ZERO_ERROR; | 
|---|
| 426 | LocalUResourceBundlePointer def( | 
|---|
| 427 | ures_getByKeyWithFallback(actualBundle.getAlias(), "collations/default", NULL, | 
|---|
| 428 | &internalErrorCode)); | 
|---|
| 429 | int32_t len; | 
|---|
| 430 | const UChar *s = ures_getString(def.getAlias(), &len, &internalErrorCode); | 
|---|
| 431 | if(U_SUCCESS(internalErrorCode) && len < UPRV_LENGTHOF(defaultType)) { | 
|---|
| 432 | u_UCharsToChars(s, defaultType, len + 1); | 
|---|
| 433 | } else { | 
|---|
| 434 | uprv_strcpy(defaultType, "standard"); | 
|---|
| 435 | } | 
|---|
| 436 | } | 
|---|
| 437 | t->actualLocale = locale; | 
|---|
| 438 | if(uprv_strcmp(type, defaultType) != 0) { | 
|---|
| 439 | t->actualLocale.setKeywordValue( "collation", type, errorCode); | 
|---|
| 440 | } else if(uprv_strcmp(locale.getName(), locale.getBaseName()) != 0) { | 
|---|
| 441 | // Remove the collation keyword if it was set. | 
|---|
| 442 | t->actualLocale.setKeywordValue( "collation", NULL, errorCode); | 
|---|
| 443 | } | 
|---|
| 444 | if(U_FAILURE(errorCode)) { return NULL; } | 
|---|
| 445 |  | 
|---|
| 446 | if(typeFallback) { | 
|---|
| 447 | errorCode = U_USING_DEFAULT_WARNING; | 
|---|
| 448 | } | 
|---|
| 449 | t->bundle = bundle; | 
|---|
| 450 | bundle = NULL; | 
|---|
| 451 | const CollationCacheEntry *entry = new CollationCacheEntry(validLocale, t.getAlias()); | 
|---|
| 452 | if(entry == NULL) { | 
|---|
| 453 | errorCode = U_MEMORY_ALLOCATION_ERROR; | 
|---|
| 454 | return nullptr; | 
|---|
| 455 | } else { | 
|---|
| 456 | t.orphan(); | 
|---|
| 457 | } | 
|---|
| 458 | // Have to add that reference that we promise. | 
|---|
| 459 | entry->addRef(); | 
|---|
| 460 | return entry; | 
|---|
| 461 | } | 
|---|
| 462 |  | 
|---|
| 463 | const CollationCacheEntry * | 
|---|
| 464 | CollationLoader::getCacheEntry(UErrorCode &errorCode) { | 
|---|
| 465 | LocaleCacheKey<CollationCacheEntry> key(locale); | 
|---|
| 466 | const CollationCacheEntry *entry = NULL; | 
|---|
| 467 | cache->get(key, this, entry, errorCode); | 
|---|
| 468 | return entry; | 
|---|
| 469 | } | 
|---|
| 470 |  | 
|---|
| 471 | const CollationCacheEntry * | 
|---|
| 472 | CollationLoader::makeCacheEntryFromRoot( | 
|---|
| 473 | const Locale &/*loc*/, | 
|---|
| 474 | UErrorCode &errorCode) const { | 
|---|
| 475 | if (U_FAILURE(errorCode)) { | 
|---|
| 476 | return NULL; | 
|---|
| 477 | } | 
|---|
| 478 | rootEntry->addRef(); | 
|---|
| 479 | return makeCacheEntry(validLocale, rootEntry, errorCode); | 
|---|
| 480 | } | 
|---|
| 481 |  | 
|---|
| 482 | const CollationCacheEntry * | 
|---|
| 483 | CollationLoader::makeCacheEntry( | 
|---|
| 484 | const Locale &loc, | 
|---|
| 485 | const CollationCacheEntry *entryFromCache, | 
|---|
| 486 | UErrorCode &errorCode) { | 
|---|
| 487 | if(U_FAILURE(errorCode) || loc == entryFromCache->validLocale) { | 
|---|
| 488 | return entryFromCache; | 
|---|
| 489 | } | 
|---|
| 490 | CollationCacheEntry *entry = new CollationCacheEntry(loc, entryFromCache->tailoring); | 
|---|
| 491 | if(entry == NULL) { | 
|---|
| 492 | errorCode = U_MEMORY_ALLOCATION_ERROR; | 
|---|
| 493 | entryFromCache->removeRef(); | 
|---|
| 494 | return NULL; | 
|---|
| 495 | } | 
|---|
| 496 | entry->addRef(); | 
|---|
| 497 | entryFromCache->removeRef(); | 
|---|
| 498 | return entry; | 
|---|
| 499 | } | 
|---|
| 500 |  | 
|---|
| 501 | U_NAMESPACE_END | 
|---|
| 502 |  | 
|---|
| 503 | U_NAMESPACE_USE | 
|---|
| 504 |  | 
|---|
| 505 | U_CAPI UCollator* | 
|---|
| 506 | ucol_open(const char *loc, | 
|---|
| 507 | UErrorCode *status) | 
|---|
| 508 | { | 
|---|
| 509 | UTRACE_ENTRY_OC(UTRACE_UCOL_OPEN); | 
|---|
| 510 | UTRACE_DATA1(UTRACE_INFO, "locale = \"%s\"", loc); | 
|---|
| 511 | UCollator *result = NULL; | 
|---|
| 512 |  | 
|---|
| 513 | Collator *coll = Collator::createInstance(loc, *status); | 
|---|
| 514 | if(U_SUCCESS(*status)) { | 
|---|
| 515 | result = coll->toUCollator(); | 
|---|
| 516 | } | 
|---|
| 517 | UTRACE_EXIT_PTR_STATUS(result, *status); | 
|---|
| 518 | return result; | 
|---|
| 519 | } | 
|---|
| 520 |  | 
|---|
| 521 |  | 
|---|
| 522 | U_CAPI int32_t U_EXPORT2 | 
|---|
| 523 | ucol_getDisplayName(    const    char        *objLoc, | 
|---|
| 524 | const    char        *dispLoc, | 
|---|
| 525 | UChar             *result, | 
|---|
| 526 | int32_t         resultLength, | 
|---|
| 527 | UErrorCode        *status) | 
|---|
| 528 | { | 
|---|
| 529 | if(U_FAILURE(*status)) return -1; | 
|---|
| 530 | UnicodeString dst; | 
|---|
| 531 | if(!(result==NULL && resultLength==0)) { | 
|---|
| 532 | // NULL destination for pure preflighting: empty dummy string | 
|---|
| 533 | // otherwise, alias the destination buffer | 
|---|
| 534 | dst.setTo(result, 0, resultLength); | 
|---|
| 535 | } | 
|---|
| 536 | Collator::getDisplayName(Locale(objLoc), Locale(dispLoc), dst); | 
|---|
| 537 | return dst.extract(result, resultLength, *status); | 
|---|
| 538 | } | 
|---|
| 539 |  | 
|---|
| 540 | U_CAPI const char* U_EXPORT2 | 
|---|
| 541 | ucol_getAvailable(int32_t index) | 
|---|
| 542 | { | 
|---|
| 543 | int32_t count = 0; | 
|---|
| 544 | const Locale *loc = Collator::getAvailableLocales(count); | 
|---|
| 545 | if (loc != NULL && index < count) { | 
|---|
| 546 | return loc[index].getName(); | 
|---|
| 547 | } | 
|---|
| 548 | return NULL; | 
|---|
| 549 | } | 
|---|
| 550 |  | 
|---|
| 551 | U_CAPI int32_t U_EXPORT2 | 
|---|
| 552 | ucol_countAvailable() | 
|---|
| 553 | { | 
|---|
| 554 | int32_t count = 0; | 
|---|
| 555 | Collator::getAvailableLocales(count); | 
|---|
| 556 | return count; | 
|---|
| 557 | } | 
|---|
| 558 |  | 
|---|
| 559 | #if !UCONFIG_NO_SERVICE | 
|---|
| 560 | U_CAPI UEnumeration* U_EXPORT2 | 
|---|
| 561 | ucol_openAvailableLocales(UErrorCode *status) { | 
|---|
| 562 | // This is a wrapper over Collator::getAvailableLocales() | 
|---|
| 563 | if (U_FAILURE(*status)) { | 
|---|
| 564 | return NULL; | 
|---|
| 565 | } | 
|---|
| 566 | StringEnumeration *s = icu::Collator::getAvailableLocales(); | 
|---|
| 567 | if (s == NULL) { | 
|---|
| 568 | *status = U_MEMORY_ALLOCATION_ERROR; | 
|---|
| 569 | return NULL; | 
|---|
| 570 | } | 
|---|
| 571 | return uenum_openFromStringEnumeration(s, status); | 
|---|
| 572 | } | 
|---|
| 573 | #endif | 
|---|
| 574 |  | 
|---|
| 575 | // Note: KEYWORDS[0] != RESOURCE_NAME - alan | 
|---|
| 576 |  | 
|---|
| 577 | static const char RESOURCE_NAME[] = "collations"; | 
|---|
| 578 |  | 
|---|
| 579 | static const char* const KEYWORDS[] = { "collation"}; | 
|---|
| 580 |  | 
|---|
| 581 | #define KEYWORD_COUNT UPRV_LENGTHOF(KEYWORDS) | 
|---|
| 582 |  | 
|---|
| 583 | U_CAPI UEnumeration* U_EXPORT2 | 
|---|
| 584 | ucol_getKeywords(UErrorCode *status) { | 
|---|
| 585 | UEnumeration *result = NULL; | 
|---|
| 586 | if (U_SUCCESS(*status)) { | 
|---|
| 587 | return uenum_openCharStringsEnumeration(KEYWORDS, KEYWORD_COUNT, status); | 
|---|
| 588 | } | 
|---|
| 589 | return result; | 
|---|
| 590 | } | 
|---|
| 591 |  | 
|---|
| 592 | U_CAPI UEnumeration* U_EXPORT2 | 
|---|
| 593 | ucol_getKeywordValues(const char *keyword, UErrorCode *status) { | 
|---|
| 594 | if (U_FAILURE(*status)) { | 
|---|
| 595 | return NULL; | 
|---|
| 596 | } | 
|---|
| 597 | // hard-coded to accept exactly one collation keyword | 
|---|
| 598 | // modify if additional collation keyword is added later | 
|---|
| 599 | if (keyword==NULL || uprv_strcmp(keyword, KEYWORDS[0])!=0) | 
|---|
| 600 | { | 
|---|
| 601 | *status = U_ILLEGAL_ARGUMENT_ERROR; | 
|---|
| 602 | return NULL; | 
|---|
| 603 | } | 
|---|
| 604 | return ures_getKeywordValues(U_ICUDATA_COLL, RESOURCE_NAME, status); | 
|---|
| 605 | } | 
|---|
| 606 |  | 
|---|
| 607 | static const UEnumeration defaultKeywordValues = { | 
|---|
| 608 | NULL, | 
|---|
| 609 | NULL, | 
|---|
| 610 | ulist_close_keyword_values_iterator, | 
|---|
| 611 | ulist_count_keyword_values, | 
|---|
| 612 | uenum_unextDefault, | 
|---|
| 613 | ulist_next_keyword_value, | 
|---|
| 614 | ulist_reset_keyword_values_iterator | 
|---|
| 615 | }; | 
|---|
| 616 |  | 
|---|
| 617 | namespace { | 
|---|
| 618 |  | 
|---|
| 619 | struct KeywordsSink : public ResourceSink { | 
|---|
| 620 | public: | 
|---|
| 621 | KeywordsSink(UErrorCode &errorCode) : | 
|---|
| 622 | values(ulist_createEmptyList(&errorCode)), hasDefault(FALSE) {} | 
|---|
| 623 | virtual ~KeywordsSink(); | 
|---|
| 624 |  | 
|---|
| 625 | virtual void put(const char *key, ResourceValue &value, UBool /*noFallback*/, | 
|---|
| 626 | UErrorCode &errorCode) { | 
|---|
| 627 | if (U_FAILURE(errorCode)) { return; } | 
|---|
| 628 | ResourceTable collations = value.getTable(errorCode); | 
|---|
| 629 | for (int32_t i = 0; collations.getKeyAndValue(i, key, value); ++i) { | 
|---|
| 630 | UResType type = value.getType(); | 
|---|
| 631 | if (type == URES_STRING) { | 
|---|
| 632 | if (!hasDefault && uprv_strcmp(key, "default") == 0) { | 
|---|
| 633 | CharString defcoll; | 
|---|
| 634 | defcoll.appendInvariantChars(value.getUnicodeString(errorCode), errorCode); | 
|---|
| 635 | if (U_SUCCESS(errorCode) && !defcoll.isEmpty()) { | 
|---|
| 636 | char *ownedDefault = uprv_strdup(defcoll.data()); | 
|---|
| 637 | if (ownedDefault == NULL) { | 
|---|
| 638 | errorCode = U_MEMORY_ALLOCATION_ERROR; | 
|---|
| 639 | return; | 
|---|
| 640 | } | 
|---|
| 641 | ulist_removeString(values, defcoll.data()); | 
|---|
| 642 | ulist_addItemBeginList(values, ownedDefault, TRUE, &errorCode); | 
|---|
| 643 | hasDefault = TRUE; | 
|---|
| 644 | } | 
|---|
| 645 | } | 
|---|
| 646 | } else if (type == URES_TABLE && uprv_strncmp(key, "private-", 8) != 0) { | 
|---|
| 647 | if (!ulist_containsString(values, key, (int32_t)uprv_strlen(key))) { | 
|---|
| 648 | ulist_addItemEndList(values, key, FALSE, &errorCode); | 
|---|
| 649 | } | 
|---|
| 650 | } | 
|---|
| 651 | if (U_FAILURE(errorCode)) { return; } | 
|---|
| 652 | } | 
|---|
| 653 | } | 
|---|
| 654 |  | 
|---|
| 655 | UList *values; | 
|---|
| 656 | UBool hasDefault; | 
|---|
| 657 | }; | 
|---|
| 658 |  | 
|---|
| 659 | KeywordsSink::~KeywordsSink() { | 
|---|
| 660 | ulist_deleteList(values); | 
|---|
| 661 | } | 
|---|
| 662 |  | 
|---|
| 663 | }  // namespace | 
|---|
| 664 |  | 
|---|
| 665 | U_CAPI UEnumeration* U_EXPORT2 | 
|---|
| 666 | ucol_getKeywordValuesForLocale(const char* /*key*/, const char* locale, | 
|---|
| 667 | UBool /*commonlyUsed*/, UErrorCode* status) { | 
|---|
| 668 | // Note: The parameter commonlyUsed is not used. | 
|---|
| 669 | // The switch is in the method signature for consistency | 
|---|
| 670 | // with other locale services. | 
|---|
| 671 |  | 
|---|
| 672 | // Read available collation values from collation bundles. | 
|---|
| 673 | LocalUResourceBundlePointer bundle(ures_open(U_ICUDATA_COLL, locale, status)); | 
|---|
| 674 | KeywordsSink sink(*status); | 
|---|
| 675 | ures_getAllItemsWithFallback(bundle.getAlias(), RESOURCE_NAME, sink, *status); | 
|---|
| 676 | if (U_FAILURE(*status)) { return NULL; } | 
|---|
| 677 |  | 
|---|
| 678 | UEnumeration *en = (UEnumeration *)uprv_malloc(sizeof(UEnumeration)); | 
|---|
| 679 | if (en == NULL) { | 
|---|
| 680 | *status = U_MEMORY_ALLOCATION_ERROR; | 
|---|
| 681 | return NULL; | 
|---|
| 682 | } | 
|---|
| 683 | memcpy(en, &defaultKeywordValues, sizeof(UEnumeration)); | 
|---|
| 684 | ulist_resetList(sink.values);  // Initialize the iterator. | 
|---|
| 685 | en->context = sink.values; | 
|---|
| 686 | sink.values = NULL;  // Avoid deletion in the sink destructor. | 
|---|
| 687 | return en; | 
|---|
| 688 | } | 
|---|
| 689 |  | 
|---|
| 690 | U_CAPI int32_t U_EXPORT2 | 
|---|
| 691 | ucol_getFunctionalEquivalent(char* result, int32_t resultCapacity, | 
|---|
| 692 | const char* keyword, const char* locale, | 
|---|
| 693 | UBool* isAvailable, UErrorCode* status) | 
|---|
| 694 | { | 
|---|
| 695 | // N.B.: Resource name is "collations" but keyword is "collation" | 
|---|
| 696 | return ures_getFunctionalEquivalent(result, resultCapacity, U_ICUDATA_COLL, | 
|---|
| 697 | "collations", keyword, locale, | 
|---|
| 698 | isAvailable, TRUE, status); | 
|---|
| 699 | } | 
|---|
| 700 |  | 
|---|
| 701 | #endif /* #if !UCONFIG_NO_COLLATION */ | 
|---|
| 702 |  | 
|---|