Thu, 22 Jan 2015 13:21:57 +0100
Incorporate requested changes from Mozilla in review:
https://bugzilla.mozilla.org/show_bug.cgi?id=1123480#c6
michael@0 | 1 | /* |
michael@0 | 2 | ********************************************************************** |
michael@0 | 3 | * Copyright (C) 1997-2013, International Business Machines |
michael@0 | 4 | * Corporation and others. All Rights Reserved. |
michael@0 | 5 | ********************************************************************** |
michael@0 | 6 | * |
michael@0 | 7 | * File resbund.cpp |
michael@0 | 8 | * |
michael@0 | 9 | * Modification History: |
michael@0 | 10 | * |
michael@0 | 11 | * Date Name Description |
michael@0 | 12 | * 02/05/97 aliu Fixed bug in chopLocale. Added scanForLocaleInFile |
michael@0 | 13 | * based on code taken from scanForLocale. Added |
michael@0 | 14 | * constructor which attempts to read resource bundle |
michael@0 | 15 | * from a specific file, without searching other files. |
michael@0 | 16 | * 02/11/97 aliu Added UErrorCode return values to constructors. Fixed |
michael@0 | 17 | * infinite loops in scanForFile and scanForLocale. |
michael@0 | 18 | * Modified getRawResourceData to not delete storage in |
michael@0 | 19 | * localeData and resourceData which it doesn't own. |
michael@0 | 20 | * Added Mac compatibility #ifdefs for tellp() and |
michael@0 | 21 | * ios::nocreate. |
michael@0 | 22 | * 03/04/97 aliu Modified to use ExpandingDataSink objects instead of |
michael@0 | 23 | * the highly inefficient ostrstream objects. |
michael@0 | 24 | * 03/13/97 aliu Rewrote to load in entire resource bundle and store |
michael@0 | 25 | * it as a Hashtable of ResourceBundleData objects. |
michael@0 | 26 | * Added state table to govern parsing of files. |
michael@0 | 27 | * Modified to load locale index out of new file distinct |
michael@0 | 28 | * from default.txt. |
michael@0 | 29 | * 03/25/97 aliu Modified to support 2-d arrays, needed for timezone data. |
michael@0 | 30 | * Added support for custom file suffixes. Again, needed |
michael@0 | 31 | * to support timezone data. Improved error handling to |
michael@0 | 32 | * detect duplicate tags and subtags. |
michael@0 | 33 | * 04/07/97 aliu Fixed bug in getHashtableForLocale(). Fixed handling |
michael@0 | 34 | * of failing UErrorCode values on entry to API methods. |
michael@0 | 35 | * Fixed bugs in getArrayItem() for negative indices. |
michael@0 | 36 | * 04/29/97 aliu Update to use new Hashtable deletion protocol. |
michael@0 | 37 | * 05/06/97 aliu Flattened kTransitionTable for HP compiler. |
michael@0 | 38 | * Fixed usage of CharString. |
michael@0 | 39 | * 06/11/99 stephen Removed parsing of .txt files. |
michael@0 | 40 | * Reworked to use new binary format. |
michael@0 | 41 | * Cleaned up. |
michael@0 | 42 | * 06/14/99 stephen Removed methods taking a filename suffix. |
michael@0 | 43 | * 06/22/99 stephen Added missing T_FileStream_close in parse() |
michael@0 | 44 | * 11/09/99 weiv Added getLocale(), rewritten constructForLocale() |
michael@0 | 45 | * March 2000 weiv complete overhaul. |
michael@0 | 46 | ****************************************************************************** |
michael@0 | 47 | */ |
michael@0 | 48 | |
michael@0 | 49 | #include "unicode/utypes.h" |
michael@0 | 50 | #include "unicode/resbund.h" |
michael@0 | 51 | |
michael@0 | 52 | #include "mutex.h" |
michael@0 | 53 | #include "uassert.h" |
michael@0 | 54 | #include "umutex.h" |
michael@0 | 55 | |
michael@0 | 56 | #include "uresimp.h" |
michael@0 | 57 | |
michael@0 | 58 | U_NAMESPACE_BEGIN |
michael@0 | 59 | |
michael@0 | 60 | /*----------------------------------------------------------------------------- |
michael@0 | 61 | * Implementation Notes |
michael@0 | 62 | * |
michael@0 | 63 | * Resource bundles are read in once, and thereafter cached. |
michael@0 | 64 | * ResourceBundle statically keeps track of which files have been |
michael@0 | 65 | * read, so we are guaranteed that each file is read at most once. |
michael@0 | 66 | * Resource bundles can be loaded from different data directories and |
michael@0 | 67 | * will be treated as distinct, even if they are for the same locale. |
michael@0 | 68 | * |
michael@0 | 69 | * Resource bundles are lightweight objects, which have pointers to |
michael@0 | 70 | * one or more shared Hashtable objects containing all the data. |
michael@0 | 71 | * Copying would be cheap, but there is no copy constructor, since |
michael@0 | 72 | * there wasn't one in the original API. |
michael@0 | 73 | * |
michael@0 | 74 | * The ResourceBundle parsing mechanism is implemented as a transition |
michael@0 | 75 | * network, for easy maintenance and modification. The network is |
michael@0 | 76 | * implemented as a matrix (instead of in code) to make this even |
michael@0 | 77 | * easier. The matrix contains Transition objects. Each Transition |
michael@0 | 78 | * object describes a destination node and an action to take before |
michael@0 | 79 | * moving to the destination node. The source node is encoded by the |
michael@0 | 80 | * index of the object in the array that contains it. The pieces |
michael@0 | 81 | * needed to understand the transition network are the enums for node |
michael@0 | 82 | * IDs and actions, the parse() method, which walks through the |
michael@0 | 83 | * network and implements the actions, and the network itself. The |
michael@0 | 84 | * network guarantees certain conditions, for example, that a new |
michael@0 | 85 | * resource will not be closed until one has been opened first; or |
michael@0 | 86 | * that data will not be stored into a TaggedList until a TaggedList |
michael@0 | 87 | * has been created. Nonetheless, the code in parse() does some |
michael@0 | 88 | * consistency checks as it runs the network, and fails with an |
michael@0 | 89 | * U_INTERNAL_PROGRAM_ERROR if one of these checks fails. If the input |
michael@0 | 90 | * data has a bad format, an U_INVALID_FORMAT_ERROR is returned. If you |
michael@0 | 91 | * see an U_INTERNAL_PROGRAM_ERROR the transition matrix has a bug in |
michael@0 | 92 | * it. |
michael@0 | 93 | * |
michael@0 | 94 | * Old functionality of multiple locales in a single file is still |
michael@0 | 95 | * supported. For this reason, LOCALE names override FILE names. If |
michael@0 | 96 | * data for en_US is located in the en.txt file, once it is loaded, |
michael@0 | 97 | * the code will not care where it came from (other than remembering |
michael@0 | 98 | * which directory it came from). However, if there is an en_US |
michael@0 | 99 | * resource in en_US.txt, that will take precedence. There is no |
michael@0 | 100 | * limit to the number or type of resources that can be stored in a |
michael@0 | 101 | * file, however, files are only searched in a specific way. If |
michael@0 | 102 | * en_US_CA is requested, then first en_US_CA.txt is searched, then |
michael@0 | 103 | * en_US.txt, then en.txt, then default.txt. So it only makes sense |
michael@0 | 104 | * to put certain locales in certain files. In this example, it would |
michael@0 | 105 | * be logical to put en_US_CA, en_US, and en into the en.txt file, |
michael@0 | 106 | * since they would be found there if asked for. The extreme example |
michael@0 | 107 | * is to place all locale resources into default.txt, which should |
michael@0 | 108 | * also work. |
michael@0 | 109 | * |
michael@0 | 110 | * Inheritance is implemented. For example, xx_YY_zz inherits as |
michael@0 | 111 | * follows: xx_YY_zz, xx_YY, xx, default. Inheritance is implemented |
michael@0 | 112 | * as an array of hashtables. There will be from 1 to 4 hashtables in |
michael@0 | 113 | * the array. |
michael@0 | 114 | * |
michael@0 | 115 | * Fallback files are implemented. The fallback pattern is Language |
michael@0 | 116 | * Country Variant (LCV) -> LC -> L. Fallback is first done for the |
michael@0 | 117 | * requested locale. Then it is done for the default locale, as |
michael@0 | 118 | * returned by Locale::getDefault(). Then the special file |
michael@0 | 119 | * default.txt is searched for the default locale. The overall FILE |
michael@0 | 120 | * fallback path is LCV -> LC -> L -> dLCV -> dLC -> dL -> default. |
michael@0 | 121 | * |
michael@0 | 122 | * Note that although file name searching includes the default locale, |
michael@0 | 123 | * once a ResourceBundle object is constructed, the inheritance path |
michael@0 | 124 | * no longer includes the default locale. The path is LCV -> LC -> L |
michael@0 | 125 | * -> default. |
michael@0 | 126 | * |
michael@0 | 127 | * File parsing is lazy. Nothing is parsed unless it is called for by |
michael@0 | 128 | * someone. So when a ResourceBundle for xx_YY_zz is constructed, |
michael@0 | 129 | * only that locale is parsed (along with anything else in the same |
michael@0 | 130 | * file). Later, if the FooBar tag is asked for, and if it isn't |
michael@0 | 131 | * found in xx_YY_zz, then xx_YY.txt will be parsed and checked, and |
michael@0 | 132 | * so forth, until the chain is exhausted or the tag is found. |
michael@0 | 133 | * |
michael@0 | 134 | * Thread-safety is implemented around caches, both the cache that |
michael@0 | 135 | * stores all the resouce data, and the cache that stores flags |
michael@0 | 136 | * indicating whether or not a file has been visited. These caches |
michael@0 | 137 | * delete their storage at static cleanup time, when the process |
michael@0 | 138 | * quits. |
michael@0 | 139 | * |
michael@0 | 140 | * ResourceBundle supports TableCollation as a special case. This |
michael@0 | 141 | * involves having special ResourceBundle objects which DO own their |
michael@0 | 142 | * data, since we don't want large collation rule strings in the |
michael@0 | 143 | * ResourceBundle cache (these are already cached in the |
michael@0 | 144 | * TableCollation cache). TableCollation files (.ctx files) have the |
michael@0 | 145 | * same format as normal resource data files, with a different |
michael@0 | 146 | * interpretation, from the standpoint of ResourceBundle. .ctx files |
michael@0 | 147 | * are loaded into otherwise ordinary ResourceBundle objects. They |
michael@0 | 148 | * don't inherit (that's implemented by TableCollation) and they own |
michael@0 | 149 | * their data (as mentioned above). However, they still support |
michael@0 | 150 | * possible multiple locales in a single .ctx file. (This is in |
michael@0 | 151 | * practice a bad idea, since you only want the one locale you're |
michael@0 | 152 | * looking for, and only one tag will be present |
michael@0 | 153 | * ("CollationElements"), so you don't need an inheritance chain of |
michael@0 | 154 | * multiple locales.) Up to 4 locale resources will be loaded from a |
michael@0 | 155 | * .ctx file; everything after the first 4 is ignored (parsed and |
michael@0 | 156 | * deleted). (Normal .txt files have no limit.) Instead of being |
michael@0 | 157 | * loaded into the cache, and then looked up as needed, the locale |
michael@0 | 158 | * resources are read straight into the ResourceBundle object. |
michael@0 | 159 | * |
michael@0 | 160 | * The Index, which used to reside in default.txt, has been moved to a |
michael@0 | 161 | * new file, index.txt. This file contains a slightly modified format |
michael@0 | 162 | * with the addition of the "InstalledLocales" tag; it looks like: |
michael@0 | 163 | * |
michael@0 | 164 | * Index { |
michael@0 | 165 | * InstalledLocales { |
michael@0 | 166 | * ar |
michael@0 | 167 | * .. |
michael@0 | 168 | * zh_TW |
michael@0 | 169 | * } |
michael@0 | 170 | * } |
michael@0 | 171 | */ |
michael@0 | 172 | //----------------------------------------------------------------------------- |
michael@0 | 173 | |
michael@0 | 174 | UOBJECT_DEFINE_RTTI_IMPLEMENTATION(ResourceBundle) |
michael@0 | 175 | |
michael@0 | 176 | ResourceBundle::ResourceBundle(UErrorCode &err) |
michael@0 | 177 | :UObject(), fLocale(NULL) |
michael@0 | 178 | { |
michael@0 | 179 | fResource = ures_open(0, Locale::getDefault().getName(), &err); |
michael@0 | 180 | } |
michael@0 | 181 | |
michael@0 | 182 | ResourceBundle::ResourceBundle(const ResourceBundle &other) |
michael@0 | 183 | :UObject(other), fLocale(NULL) |
michael@0 | 184 | { |
michael@0 | 185 | UErrorCode status = U_ZERO_ERROR; |
michael@0 | 186 | |
michael@0 | 187 | if (other.fResource) { |
michael@0 | 188 | fResource = ures_copyResb(0, other.fResource, &status); |
michael@0 | 189 | } else { |
michael@0 | 190 | /* Copying a bad resource bundle */ |
michael@0 | 191 | fResource = NULL; |
michael@0 | 192 | } |
michael@0 | 193 | } |
michael@0 | 194 | |
michael@0 | 195 | ResourceBundle::ResourceBundle(UResourceBundle *res, UErrorCode& err) |
michael@0 | 196 | :UObject(), fLocale(NULL) |
michael@0 | 197 | { |
michael@0 | 198 | if (res) { |
michael@0 | 199 | fResource = ures_copyResb(0, res, &err); |
michael@0 | 200 | } else { |
michael@0 | 201 | /* Copying a bad resource bundle */ |
michael@0 | 202 | fResource = NULL; |
michael@0 | 203 | } |
michael@0 | 204 | } |
michael@0 | 205 | |
michael@0 | 206 | ResourceBundle::ResourceBundle(const char* path, const Locale& locale, UErrorCode& err) |
michael@0 | 207 | :UObject(), fLocale(NULL) |
michael@0 | 208 | { |
michael@0 | 209 | fResource = ures_open(path, locale.getName(), &err); |
michael@0 | 210 | } |
michael@0 | 211 | |
michael@0 | 212 | |
michael@0 | 213 | ResourceBundle& ResourceBundle::operator=(const ResourceBundle& other) |
michael@0 | 214 | { |
michael@0 | 215 | if(this == &other) { |
michael@0 | 216 | return *this; |
michael@0 | 217 | } |
michael@0 | 218 | if(fResource != 0) { |
michael@0 | 219 | ures_close(fResource); |
michael@0 | 220 | fResource = NULL; |
michael@0 | 221 | } |
michael@0 | 222 | if (fLocale != NULL) { |
michael@0 | 223 | delete fLocale; |
michael@0 | 224 | fLocale = NULL; |
michael@0 | 225 | } |
michael@0 | 226 | UErrorCode status = U_ZERO_ERROR; |
michael@0 | 227 | if (other.fResource) { |
michael@0 | 228 | fResource = ures_copyResb(0, other.fResource, &status); |
michael@0 | 229 | } else { |
michael@0 | 230 | /* Copying a bad resource bundle */ |
michael@0 | 231 | fResource = NULL; |
michael@0 | 232 | } |
michael@0 | 233 | return *this; |
michael@0 | 234 | } |
michael@0 | 235 | |
michael@0 | 236 | ResourceBundle::~ResourceBundle() |
michael@0 | 237 | { |
michael@0 | 238 | if(fResource != 0) { |
michael@0 | 239 | ures_close(fResource); |
michael@0 | 240 | } |
michael@0 | 241 | if(fLocale != NULL) { |
michael@0 | 242 | delete(fLocale); |
michael@0 | 243 | } |
michael@0 | 244 | } |
michael@0 | 245 | |
michael@0 | 246 | ResourceBundle * |
michael@0 | 247 | ResourceBundle::clone() const { |
michael@0 | 248 | return new ResourceBundle(*this); |
michael@0 | 249 | } |
michael@0 | 250 | |
michael@0 | 251 | UnicodeString ResourceBundle::getString(UErrorCode& status) const { |
michael@0 | 252 | int32_t len = 0; |
michael@0 | 253 | const UChar *r = ures_getString(fResource, &len, &status); |
michael@0 | 254 | return UnicodeString(TRUE, r, len); |
michael@0 | 255 | } |
michael@0 | 256 | |
michael@0 | 257 | const uint8_t *ResourceBundle::getBinary(int32_t& len, UErrorCode& status) const { |
michael@0 | 258 | return ures_getBinary(fResource, &len, &status); |
michael@0 | 259 | } |
michael@0 | 260 | |
michael@0 | 261 | const int32_t *ResourceBundle::getIntVector(int32_t& len, UErrorCode& status) const { |
michael@0 | 262 | return ures_getIntVector(fResource, &len, &status); |
michael@0 | 263 | } |
michael@0 | 264 | |
michael@0 | 265 | uint32_t ResourceBundle::getUInt(UErrorCode& status) const { |
michael@0 | 266 | return ures_getUInt(fResource, &status); |
michael@0 | 267 | } |
michael@0 | 268 | |
michael@0 | 269 | int32_t ResourceBundle::getInt(UErrorCode& status) const { |
michael@0 | 270 | return ures_getInt(fResource, &status); |
michael@0 | 271 | } |
michael@0 | 272 | |
michael@0 | 273 | const char *ResourceBundle::getName(void) const { |
michael@0 | 274 | return ures_getName(fResource); |
michael@0 | 275 | } |
michael@0 | 276 | |
michael@0 | 277 | const char *ResourceBundle::getKey(void) const { |
michael@0 | 278 | return ures_getKey(fResource); |
michael@0 | 279 | } |
michael@0 | 280 | |
michael@0 | 281 | UResType ResourceBundle::getType(void) const { |
michael@0 | 282 | return ures_getType(fResource); |
michael@0 | 283 | } |
michael@0 | 284 | |
michael@0 | 285 | int32_t ResourceBundle::getSize(void) const { |
michael@0 | 286 | return ures_getSize(fResource); |
michael@0 | 287 | } |
michael@0 | 288 | |
michael@0 | 289 | UBool ResourceBundle::hasNext(void) const { |
michael@0 | 290 | return ures_hasNext(fResource); |
michael@0 | 291 | } |
michael@0 | 292 | |
michael@0 | 293 | void ResourceBundle::resetIterator(void) { |
michael@0 | 294 | ures_resetIterator(fResource); |
michael@0 | 295 | } |
michael@0 | 296 | |
michael@0 | 297 | ResourceBundle ResourceBundle::getNext(UErrorCode& status) { |
michael@0 | 298 | UResourceBundle r; |
michael@0 | 299 | |
michael@0 | 300 | ures_initStackObject(&r); |
michael@0 | 301 | ures_getNextResource(fResource, &r, &status); |
michael@0 | 302 | ResourceBundle res(&r, status); |
michael@0 | 303 | if (U_SUCCESS(status)) { |
michael@0 | 304 | ures_close(&r); |
michael@0 | 305 | } |
michael@0 | 306 | return res; |
michael@0 | 307 | } |
michael@0 | 308 | |
michael@0 | 309 | UnicodeString ResourceBundle::getNextString(UErrorCode& status) { |
michael@0 | 310 | int32_t len = 0; |
michael@0 | 311 | const UChar* r = ures_getNextString(fResource, &len, 0, &status); |
michael@0 | 312 | return UnicodeString(TRUE, r, len); |
michael@0 | 313 | } |
michael@0 | 314 | |
michael@0 | 315 | UnicodeString ResourceBundle::getNextString(const char ** key, UErrorCode& status) { |
michael@0 | 316 | int32_t len = 0; |
michael@0 | 317 | const UChar* r = ures_getNextString(fResource, &len, key, &status); |
michael@0 | 318 | return UnicodeString(TRUE, r, len); |
michael@0 | 319 | } |
michael@0 | 320 | |
michael@0 | 321 | ResourceBundle ResourceBundle::get(int32_t indexR, UErrorCode& status) const { |
michael@0 | 322 | UResourceBundle r; |
michael@0 | 323 | |
michael@0 | 324 | ures_initStackObject(&r); |
michael@0 | 325 | ures_getByIndex(fResource, indexR, &r, &status); |
michael@0 | 326 | ResourceBundle res(&r, status); |
michael@0 | 327 | if (U_SUCCESS(status)) { |
michael@0 | 328 | ures_close(&r); |
michael@0 | 329 | } |
michael@0 | 330 | return res; |
michael@0 | 331 | } |
michael@0 | 332 | |
michael@0 | 333 | UnicodeString ResourceBundle::getStringEx(int32_t indexS, UErrorCode& status) const { |
michael@0 | 334 | int32_t len = 0; |
michael@0 | 335 | const UChar* r = ures_getStringByIndex(fResource, indexS, &len, &status); |
michael@0 | 336 | return UnicodeString(TRUE, r, len); |
michael@0 | 337 | } |
michael@0 | 338 | |
michael@0 | 339 | ResourceBundle ResourceBundle::get(const char* key, UErrorCode& status) const { |
michael@0 | 340 | UResourceBundle r; |
michael@0 | 341 | |
michael@0 | 342 | ures_initStackObject(&r); |
michael@0 | 343 | ures_getByKey(fResource, key, &r, &status); |
michael@0 | 344 | ResourceBundle res(&r, status); |
michael@0 | 345 | if (U_SUCCESS(status)) { |
michael@0 | 346 | ures_close(&r); |
michael@0 | 347 | } |
michael@0 | 348 | return res; |
michael@0 | 349 | } |
michael@0 | 350 | |
michael@0 | 351 | ResourceBundle ResourceBundle::getWithFallback(const char* key, UErrorCode& status){ |
michael@0 | 352 | UResourceBundle r; |
michael@0 | 353 | ures_initStackObject(&r); |
michael@0 | 354 | ures_getByKeyWithFallback(fResource, key, &r, &status); |
michael@0 | 355 | ResourceBundle res(&r, status); |
michael@0 | 356 | if(U_SUCCESS(status)){ |
michael@0 | 357 | ures_close(&r); |
michael@0 | 358 | } |
michael@0 | 359 | return res; |
michael@0 | 360 | } |
michael@0 | 361 | UnicodeString ResourceBundle::getStringEx(const char* key, UErrorCode& status) const { |
michael@0 | 362 | int32_t len = 0; |
michael@0 | 363 | const UChar* r = ures_getStringByKey(fResource, key, &len, &status); |
michael@0 | 364 | return UnicodeString(TRUE, r, len); |
michael@0 | 365 | } |
michael@0 | 366 | |
michael@0 | 367 | const char* |
michael@0 | 368 | ResourceBundle::getVersionNumber() const |
michael@0 | 369 | { |
michael@0 | 370 | return ures_getVersionNumberInternal(fResource); |
michael@0 | 371 | } |
michael@0 | 372 | |
michael@0 | 373 | void ResourceBundle::getVersion(UVersionInfo versionInfo) const { |
michael@0 | 374 | ures_getVersion(fResource, versionInfo); |
michael@0 | 375 | } |
michael@0 | 376 | |
michael@0 | 377 | static UMutex gLocaleLock = U_MUTEX_INITIALIZER; |
michael@0 | 378 | const Locale &ResourceBundle::getLocale(void) const { |
michael@0 | 379 | Mutex lock(&gLocaleLock); |
michael@0 | 380 | if (fLocale != NULL) { |
michael@0 | 381 | return *fLocale; |
michael@0 | 382 | } |
michael@0 | 383 | UErrorCode status = U_ZERO_ERROR; |
michael@0 | 384 | const char *localeName = ures_getLocaleInternal(fResource, &status); |
michael@0 | 385 | ResourceBundle *ncThis = const_cast<ResourceBundle *>(this); |
michael@0 | 386 | ncThis->fLocale = new Locale(localeName); |
michael@0 | 387 | return ncThis->fLocale != NULL ? *ncThis->fLocale : Locale::getDefault(); |
michael@0 | 388 | } |
michael@0 | 389 | |
michael@0 | 390 | const Locale ResourceBundle::getLocale(ULocDataLocaleType type, UErrorCode &status) const |
michael@0 | 391 | { |
michael@0 | 392 | return ures_getLocaleByType(fResource, type, &status); |
michael@0 | 393 | } |
michael@0 | 394 | |
michael@0 | 395 | U_NAMESPACE_END |
michael@0 | 396 | //eof |