| 1 | // © 2016 and later: Unicode, Inc. and others. |
| 2 | // License & terms of use: http://www.unicode.org/copyright.html |
| 3 | /* |
| 4 | ******************************************************************************* |
| 5 | * Copyright (C) 2011-2016, International Business Machines Corporation and |
| 6 | * others. All Rights Reserved. |
| 7 | ******************************************************************************* |
| 8 | */ |
| 9 | |
| 10 | #include "unicode/utypes.h" |
| 11 | |
| 12 | #if !UCONFIG_NO_FORMATTING |
| 13 | |
| 14 | #include "tzgnames.h" |
| 15 | |
| 16 | #include "unicode/basictz.h" |
| 17 | #include "unicode/locdspnm.h" |
| 18 | #include "unicode/rbtz.h" |
| 19 | #include "unicode/simpleformatter.h" |
| 20 | #include "unicode/simpletz.h" |
| 21 | #include "unicode/strenum.h" |
| 22 | #include "unicode/vtzone.h" |
| 23 | |
| 24 | #include "bytesinkutil.h" |
| 25 | #include "charstr.h" |
| 26 | #include "cmemory.h" |
| 27 | #include "cstring.h" |
| 28 | #include "mutex.h" |
| 29 | #include "uhash.h" |
| 30 | #include "uassert.h" |
| 31 | #include "umutex.h" |
| 32 | #include "ulocimp.h" |
| 33 | #include "uresimp.h" |
| 34 | #include "ureslocs.h" |
| 35 | #include "zonemeta.h" |
| 36 | #include "tznames_impl.h" |
| 37 | #include "olsontz.h" |
| 38 | #include "ucln_in.h" |
| 39 | |
| 40 | U_NAMESPACE_BEGIN |
| 41 | |
| 42 | #define ZID_KEY_MAX 128 |
| 43 | |
| 44 | static const char gZoneStrings[] = "zoneStrings" ; |
| 45 | |
| 46 | static const char gRegionFormatTag[] = "regionFormat" ; |
| 47 | static const char gFallbackFormatTag[] = "fallbackFormat" ; |
| 48 | |
| 49 | static const UChar gEmpty[] = {0x00}; |
| 50 | |
| 51 | static const UChar gDefRegionPattern[] = {0x7B, 0x30, 0x7D, 0x00}; // "{0}" |
| 52 | static const UChar gDefFallbackPattern[] = {0x7B, 0x31, 0x7D, 0x20, 0x28, 0x7B, 0x30, 0x7D, 0x29, 0x00}; // "{1} ({0})" |
| 53 | |
| 54 | static const double kDstCheckRange = (double)184*U_MILLIS_PER_DAY; |
| 55 | |
| 56 | |
| 57 | |
| 58 | U_CDECL_BEGIN |
| 59 | |
| 60 | typedef struct PartialLocationKey { |
| 61 | const UChar* tzID; |
| 62 | const UChar* mzID; |
| 63 | UBool isLong; |
| 64 | } PartialLocationKey; |
| 65 | |
| 66 | /** |
| 67 | * Hash function for partial location name hash key |
| 68 | */ |
| 69 | static int32_t U_CALLCONV |
| 70 | hashPartialLocationKey(const UHashTok key) { |
| 71 | // <tzID>&<mzID>#[L|S] |
| 72 | PartialLocationKey *p = (PartialLocationKey *)key.pointer; |
| 73 | UnicodeString str(p->tzID); |
| 74 | str.append((UChar)0x26) |
| 75 | .append(p->mzID, -1) |
| 76 | .append((UChar)0x23) |
| 77 | .append((UChar)(p->isLong ? 0x4C : 0x53)); |
| 78 | return str.hashCode(); |
| 79 | } |
| 80 | |
| 81 | /** |
| 82 | * Comparer for partial location name hash key |
| 83 | */ |
| 84 | static UBool U_CALLCONV |
| 85 | comparePartialLocationKey(const UHashTok key1, const UHashTok key2) { |
| 86 | PartialLocationKey *p1 = (PartialLocationKey *)key1.pointer; |
| 87 | PartialLocationKey *p2 = (PartialLocationKey *)key2.pointer; |
| 88 | |
| 89 | if (p1 == p2) { |
| 90 | return TRUE; |
| 91 | } |
| 92 | if (p1 == NULL || p2 == NULL) { |
| 93 | return FALSE; |
| 94 | } |
| 95 | // We just check identity of tzID/mzID |
| 96 | return (p1->tzID == p2->tzID && p1->mzID == p2->mzID && p1->isLong == p2->isLong); |
| 97 | } |
| 98 | |
| 99 | /** |
| 100 | * Deleter for GNameInfo |
| 101 | */ |
| 102 | static void U_CALLCONV |
| 103 | deleteGNameInfo(void *obj) { |
| 104 | uprv_free(obj); |
| 105 | } |
| 106 | |
| 107 | /** |
| 108 | * GNameInfo stores zone name information in the local trie |
| 109 | */ |
| 110 | typedef struct GNameInfo { |
| 111 | UTimeZoneGenericNameType type; |
| 112 | const UChar* tzID; |
| 113 | } ZNameInfo; |
| 114 | |
| 115 | /** |
| 116 | * GMatchInfo stores zone name match information used by find method |
| 117 | */ |
| 118 | typedef struct GMatchInfo { |
| 119 | const GNameInfo* gnameInfo; |
| 120 | int32_t matchLength; |
| 121 | UTimeZoneFormatTimeType timeType; |
| 122 | } ZMatchInfo; |
| 123 | |
| 124 | U_CDECL_END |
| 125 | |
| 126 | // --------------------------------------------------- |
| 127 | // The class stores time zone generic name match information |
| 128 | // --------------------------------------------------- |
| 129 | class TimeZoneGenericNameMatchInfo : public UMemory { |
| 130 | public: |
| 131 | TimeZoneGenericNameMatchInfo(UVector* matches); |
| 132 | ~TimeZoneGenericNameMatchInfo(); |
| 133 | |
| 134 | int32_t size() const; |
| 135 | UTimeZoneGenericNameType getGenericNameType(int32_t index) const; |
| 136 | int32_t getMatchLength(int32_t index) const; |
| 137 | UnicodeString& getTimeZoneID(int32_t index, UnicodeString& tzID) const; |
| 138 | |
| 139 | private: |
| 140 | UVector* fMatches; // vector of MatchEntry |
| 141 | }; |
| 142 | |
| 143 | TimeZoneGenericNameMatchInfo::TimeZoneGenericNameMatchInfo(UVector* matches) |
| 144 | : fMatches(matches) { |
| 145 | } |
| 146 | |
| 147 | TimeZoneGenericNameMatchInfo::~TimeZoneGenericNameMatchInfo() { |
| 148 | if (fMatches != NULL) { |
| 149 | delete fMatches; |
| 150 | } |
| 151 | } |
| 152 | |
| 153 | int32_t |
| 154 | TimeZoneGenericNameMatchInfo::size() const { |
| 155 | if (fMatches == NULL) { |
| 156 | return 0; |
| 157 | } |
| 158 | return fMatches->size(); |
| 159 | } |
| 160 | |
| 161 | UTimeZoneGenericNameType |
| 162 | TimeZoneGenericNameMatchInfo::getGenericNameType(int32_t index) const { |
| 163 | GMatchInfo *minfo = (GMatchInfo *)fMatches->elementAt(index); |
| 164 | if (minfo != NULL) { |
| 165 | return static_cast<UTimeZoneGenericNameType>(minfo->gnameInfo->type); |
| 166 | } |
| 167 | return UTZGNM_UNKNOWN; |
| 168 | } |
| 169 | |
| 170 | int32_t |
| 171 | TimeZoneGenericNameMatchInfo::getMatchLength(int32_t index) const { |
| 172 | ZMatchInfo *minfo = (ZMatchInfo *)fMatches->elementAt(index); |
| 173 | if (minfo != NULL) { |
| 174 | return minfo->matchLength; |
| 175 | } |
| 176 | return -1; |
| 177 | } |
| 178 | |
| 179 | UnicodeString& |
| 180 | TimeZoneGenericNameMatchInfo::getTimeZoneID(int32_t index, UnicodeString& tzID) const { |
| 181 | GMatchInfo *minfo = (GMatchInfo *)fMatches->elementAt(index); |
| 182 | if (minfo != NULL && minfo->gnameInfo->tzID != NULL) { |
| 183 | tzID.setTo(TRUE, minfo->gnameInfo->tzID, -1); |
| 184 | } else { |
| 185 | tzID.setToBogus(); |
| 186 | } |
| 187 | return tzID; |
| 188 | } |
| 189 | |
| 190 | // --------------------------------------------------- |
| 191 | // GNameSearchHandler |
| 192 | // --------------------------------------------------- |
| 193 | class GNameSearchHandler : public TextTrieMapSearchResultHandler { |
| 194 | public: |
| 195 | GNameSearchHandler(uint32_t types); |
| 196 | virtual ~GNameSearchHandler(); |
| 197 | |
| 198 | UBool handleMatch(int32_t matchLength, const CharacterNode *node, UErrorCode &status); |
| 199 | UVector* getMatches(int32_t& maxMatchLen); |
| 200 | |
| 201 | private: |
| 202 | uint32_t fTypes; |
| 203 | UVector* fResults; |
| 204 | int32_t fMaxMatchLen; |
| 205 | }; |
| 206 | |
| 207 | GNameSearchHandler::GNameSearchHandler(uint32_t types) |
| 208 | : fTypes(types), fResults(NULL), fMaxMatchLen(0) { |
| 209 | } |
| 210 | |
| 211 | GNameSearchHandler::~GNameSearchHandler() { |
| 212 | if (fResults != NULL) { |
| 213 | delete fResults; |
| 214 | } |
| 215 | } |
| 216 | |
| 217 | UBool |
| 218 | GNameSearchHandler::handleMatch(int32_t matchLength, const CharacterNode *node, UErrorCode &status) { |
| 219 | if (U_FAILURE(status)) { |
| 220 | return FALSE; |
| 221 | } |
| 222 | if (node->hasValues()) { |
| 223 | int32_t valuesCount = node->countValues(); |
| 224 | for (int32_t i = 0; i < valuesCount; i++) { |
| 225 | GNameInfo *nameinfo = (ZNameInfo *)node->getValue(i); |
| 226 | if (nameinfo == NULL) { |
| 227 | break; |
| 228 | } |
| 229 | if ((nameinfo->type & fTypes) != 0) { |
| 230 | // matches a requested type |
| 231 | if (fResults == NULL) { |
| 232 | fResults = new UVector(uprv_free, NULL, status); |
| 233 | if (fResults == NULL) { |
| 234 | status = U_MEMORY_ALLOCATION_ERROR; |
| 235 | } |
| 236 | } |
| 237 | if (U_SUCCESS(status)) { |
| 238 | U_ASSERT(fResults != NULL); |
| 239 | GMatchInfo *gmatch = (GMatchInfo *)uprv_malloc(sizeof(GMatchInfo)); |
| 240 | if (gmatch == NULL) { |
| 241 | status = U_MEMORY_ALLOCATION_ERROR; |
| 242 | } else { |
| 243 | // add the match to the vector |
| 244 | gmatch->gnameInfo = nameinfo; |
| 245 | gmatch->matchLength = matchLength; |
| 246 | gmatch->timeType = UTZFMT_TIME_TYPE_UNKNOWN; |
| 247 | fResults->addElement(gmatch, status); |
| 248 | if (U_FAILURE(status)) { |
| 249 | uprv_free(gmatch); |
| 250 | } else { |
| 251 | if (matchLength > fMaxMatchLen) { |
| 252 | fMaxMatchLen = matchLength; |
| 253 | } |
| 254 | } |
| 255 | } |
| 256 | } |
| 257 | } |
| 258 | } |
| 259 | } |
| 260 | return TRUE; |
| 261 | } |
| 262 | |
| 263 | UVector* |
| 264 | GNameSearchHandler::getMatches(int32_t& maxMatchLen) { |
| 265 | // give the ownership to the caller |
| 266 | UVector *results = fResults; |
| 267 | maxMatchLen = fMaxMatchLen; |
| 268 | |
| 269 | // reset |
| 270 | fResults = NULL; |
| 271 | fMaxMatchLen = 0; |
| 272 | return results; |
| 273 | } |
| 274 | |
| 275 | static UMutex gLock; |
| 276 | |
| 277 | class TZGNCore : public UMemory { |
| 278 | public: |
| 279 | TZGNCore(const Locale& locale, UErrorCode& status); |
| 280 | virtual ~TZGNCore(); |
| 281 | |
| 282 | UnicodeString& getDisplayName(const TimeZone& tz, UTimeZoneGenericNameType type, |
| 283 | UDate date, UnicodeString& name) const; |
| 284 | |
| 285 | UnicodeString& getGenericLocationName(const UnicodeString& tzCanonicalID, UnicodeString& name) const; |
| 286 | |
| 287 | int32_t findBestMatch(const UnicodeString& text, int32_t start, uint32_t types, |
| 288 | UnicodeString& tzID, UTimeZoneFormatTimeType& timeType, UErrorCode& status) const; |
| 289 | |
| 290 | private: |
| 291 | Locale fLocale; |
| 292 | const TimeZoneNames* fTimeZoneNames; |
| 293 | UHashtable* fLocationNamesMap; |
| 294 | UHashtable* fPartialLocationNamesMap; |
| 295 | |
| 296 | SimpleFormatter fRegionFormat; |
| 297 | SimpleFormatter fFallbackFormat; |
| 298 | |
| 299 | LocaleDisplayNames* fLocaleDisplayNames; |
| 300 | ZNStringPool fStringPool; |
| 301 | |
| 302 | TextTrieMap fGNamesTrie; |
| 303 | UBool fGNamesTrieFullyLoaded; |
| 304 | |
| 305 | char fTargetRegion[ULOC_COUNTRY_CAPACITY]; |
| 306 | |
| 307 | void initialize(const Locale& locale, UErrorCode& status); |
| 308 | void cleanup(); |
| 309 | |
| 310 | void loadStrings(const UnicodeString& tzCanonicalID); |
| 311 | |
| 312 | const UChar* getGenericLocationName(const UnicodeString& tzCanonicalID); |
| 313 | |
| 314 | UnicodeString& formatGenericNonLocationName(const TimeZone& tz, UTimeZoneGenericNameType type, |
| 315 | UDate date, UnicodeString& name) const; |
| 316 | |
| 317 | UnicodeString& getPartialLocationName(const UnicodeString& tzCanonicalID, |
| 318 | const UnicodeString& mzID, UBool isLong, const UnicodeString& mzDisplayName, |
| 319 | UnicodeString& name) const; |
| 320 | |
| 321 | const UChar* getPartialLocationName(const UnicodeString& tzCanonicalID, |
| 322 | const UnicodeString& mzID, UBool isLong, const UnicodeString& mzDisplayName); |
| 323 | |
| 324 | TimeZoneGenericNameMatchInfo* findLocal(const UnicodeString& text, int32_t start, uint32_t types, UErrorCode& status) const; |
| 325 | |
| 326 | TimeZoneNames::MatchInfoCollection* findTimeZoneNames(const UnicodeString& text, int32_t start, uint32_t types, UErrorCode& status) const; |
| 327 | }; |
| 328 | |
| 329 | |
| 330 | // --------------------------------------------------- |
| 331 | // TZGNCore - core implmentation of TimeZoneGenericNames |
| 332 | // |
| 333 | // TimeZoneGenericNames is parallel to TimeZoneNames, |
| 334 | // but handles run-time generated time zone names. |
| 335 | // This is the main part of this module. |
| 336 | // --------------------------------------------------- |
| 337 | TZGNCore::TZGNCore(const Locale& locale, UErrorCode& status) |
| 338 | : fLocale(locale), |
| 339 | fTimeZoneNames(NULL), |
| 340 | fLocationNamesMap(NULL), |
| 341 | fPartialLocationNamesMap(NULL), |
| 342 | fLocaleDisplayNames(NULL), |
| 343 | fStringPool(status), |
| 344 | fGNamesTrie(TRUE, deleteGNameInfo), |
| 345 | fGNamesTrieFullyLoaded(FALSE) { |
| 346 | initialize(locale, status); |
| 347 | } |
| 348 | |
| 349 | TZGNCore::~TZGNCore() { |
| 350 | cleanup(); |
| 351 | } |
| 352 | |
| 353 | void |
| 354 | TZGNCore::initialize(const Locale& locale, UErrorCode& status) { |
| 355 | if (U_FAILURE(status)) { |
| 356 | return; |
| 357 | } |
| 358 | |
| 359 | // TimeZoneNames |
| 360 | fTimeZoneNames = TimeZoneNames::createInstance(locale, status); |
| 361 | if (U_FAILURE(status)) { |
| 362 | return; |
| 363 | } |
| 364 | |
| 365 | // Initialize format patterns |
| 366 | UnicodeString rpat(TRUE, gDefRegionPattern, -1); |
| 367 | UnicodeString fpat(TRUE, gDefFallbackPattern, -1); |
| 368 | |
| 369 | UErrorCode tmpsts = U_ZERO_ERROR; // OK with fallback warning.. |
| 370 | UResourceBundle *zoneStrings = ures_open(U_ICUDATA_ZONE, locale.getName(), &tmpsts); |
| 371 | zoneStrings = ures_getByKeyWithFallback(zoneStrings, gZoneStrings, zoneStrings, &tmpsts); |
| 372 | |
| 373 | if (U_SUCCESS(tmpsts)) { |
| 374 | const UChar *regionPattern = ures_getStringByKeyWithFallback(zoneStrings, gRegionFormatTag, NULL, &tmpsts); |
| 375 | if (U_SUCCESS(tmpsts) && u_strlen(regionPattern) > 0) { |
| 376 | rpat.setTo(regionPattern, -1); |
| 377 | } |
| 378 | tmpsts = U_ZERO_ERROR; |
| 379 | const UChar *fallbackPattern = ures_getStringByKeyWithFallback(zoneStrings, gFallbackFormatTag, NULL, &tmpsts); |
| 380 | if (U_SUCCESS(tmpsts) && u_strlen(fallbackPattern) > 0) { |
| 381 | fpat.setTo(fallbackPattern, -1); |
| 382 | } |
| 383 | } |
| 384 | ures_close(zoneStrings); |
| 385 | |
| 386 | fRegionFormat.applyPatternMinMaxArguments(rpat, 1, 1, status); |
| 387 | fFallbackFormat.applyPatternMinMaxArguments(fpat, 2, 2, status); |
| 388 | if (U_FAILURE(status)) { |
| 389 | cleanup(); |
| 390 | return; |
| 391 | } |
| 392 | |
| 393 | // locale display names |
| 394 | fLocaleDisplayNames = LocaleDisplayNames::createInstance(locale); |
| 395 | |
| 396 | // hash table for names - no key/value deleters |
| 397 | fLocationNamesMap = uhash_open(uhash_hashUChars, uhash_compareUChars, NULL, &status); |
| 398 | if (U_FAILURE(status)) { |
| 399 | cleanup(); |
| 400 | return; |
| 401 | } |
| 402 | |
| 403 | fPartialLocationNamesMap = uhash_open(hashPartialLocationKey, comparePartialLocationKey, NULL, &status); |
| 404 | if (U_FAILURE(status)) { |
| 405 | cleanup(); |
| 406 | return; |
| 407 | } |
| 408 | uhash_setKeyDeleter(fPartialLocationNamesMap, uprv_free); |
| 409 | // no value deleter |
| 410 | |
| 411 | // target region |
| 412 | const char* region = fLocale.getCountry(); |
| 413 | int32_t regionLen = static_cast<int32_t>(uprv_strlen(region)); |
| 414 | if (regionLen == 0) { |
| 415 | CharString loc; |
| 416 | { |
| 417 | CharStringByteSink sink(&loc); |
| 418 | ulocimp_addLikelySubtags(fLocale.getName(), sink, &status); |
| 419 | } |
| 420 | |
| 421 | regionLen = uloc_getCountry(loc.data(), fTargetRegion, sizeof(fTargetRegion), &status); |
| 422 | if (U_SUCCESS(status)) { |
| 423 | fTargetRegion[regionLen] = 0; |
| 424 | } else { |
| 425 | cleanup(); |
| 426 | return; |
| 427 | } |
| 428 | } else if (regionLen < (int32_t)sizeof(fTargetRegion)) { |
| 429 | uprv_strcpy(fTargetRegion, region); |
| 430 | } else { |
| 431 | fTargetRegion[0] = 0; |
| 432 | } |
| 433 | |
| 434 | // preload generic names for the default zone |
| 435 | TimeZone *tz = TimeZone::createDefault(); |
| 436 | const UChar *tzID = ZoneMeta::getCanonicalCLDRID(*tz); |
| 437 | if (tzID != NULL) { |
| 438 | loadStrings(UnicodeString(TRUE, tzID, -1)); |
| 439 | } |
| 440 | delete tz; |
| 441 | } |
| 442 | |
| 443 | void |
| 444 | TZGNCore::cleanup() { |
| 445 | if (fLocaleDisplayNames != NULL) { |
| 446 | delete fLocaleDisplayNames; |
| 447 | } |
| 448 | if (fTimeZoneNames != NULL) { |
| 449 | delete fTimeZoneNames; |
| 450 | } |
| 451 | |
| 452 | uhash_close(fLocationNamesMap); |
| 453 | uhash_close(fPartialLocationNamesMap); |
| 454 | } |
| 455 | |
| 456 | |
| 457 | UnicodeString& |
| 458 | TZGNCore::getDisplayName(const TimeZone& tz, UTimeZoneGenericNameType type, UDate date, UnicodeString& name) const { |
| 459 | name.setToBogus(); |
| 460 | switch (type) { |
| 461 | case UTZGNM_LOCATION: |
| 462 | { |
| 463 | const UChar* tzCanonicalID = ZoneMeta::getCanonicalCLDRID(tz); |
| 464 | if (tzCanonicalID != NULL) { |
| 465 | getGenericLocationName(UnicodeString(TRUE, tzCanonicalID, -1), name); |
| 466 | } |
| 467 | } |
| 468 | break; |
| 469 | case UTZGNM_LONG: |
| 470 | case UTZGNM_SHORT: |
| 471 | formatGenericNonLocationName(tz, type, date, name); |
| 472 | if (name.isEmpty()) { |
| 473 | const UChar* tzCanonicalID = ZoneMeta::getCanonicalCLDRID(tz); |
| 474 | if (tzCanonicalID != NULL) { |
| 475 | getGenericLocationName(UnicodeString(TRUE, tzCanonicalID, -1), name); |
| 476 | } |
| 477 | } |
| 478 | break; |
| 479 | default: |
| 480 | break; |
| 481 | } |
| 482 | return name; |
| 483 | } |
| 484 | |
| 485 | UnicodeString& |
| 486 | TZGNCore::getGenericLocationName(const UnicodeString& tzCanonicalID, UnicodeString& name) const { |
| 487 | if (tzCanonicalID.isEmpty()) { |
| 488 | name.setToBogus(); |
| 489 | return name; |
| 490 | } |
| 491 | |
| 492 | const UChar *locname = NULL; |
| 493 | TZGNCore *nonConstThis = const_cast<TZGNCore *>(this); |
| 494 | umtx_lock(&gLock); |
| 495 | { |
| 496 | locname = nonConstThis->getGenericLocationName(tzCanonicalID); |
| 497 | } |
| 498 | umtx_unlock(&gLock); |
| 499 | |
| 500 | if (locname == NULL) { |
| 501 | name.setToBogus(); |
| 502 | } else { |
| 503 | name.setTo(locname, u_strlen(locname)); |
| 504 | } |
| 505 | |
| 506 | return name; |
| 507 | } |
| 508 | |
| 509 | /* |
| 510 | * This method updates the cache and must be called with a lock |
| 511 | */ |
| 512 | const UChar* |
| 513 | TZGNCore::getGenericLocationName(const UnicodeString& tzCanonicalID) { |
| 514 | U_ASSERT(!tzCanonicalID.isEmpty()); |
| 515 | if (tzCanonicalID.length() > ZID_KEY_MAX) { |
| 516 | return NULL; |
| 517 | } |
| 518 | |
| 519 | UErrorCode status = U_ZERO_ERROR; |
| 520 | UChar tzIDKey[ZID_KEY_MAX + 1]; |
| 521 | int32_t tzIDKeyLen = tzCanonicalID.extract(tzIDKey, ZID_KEY_MAX + 1, status); |
| 522 | U_ASSERT(status == U_ZERO_ERROR); // already checked length above |
| 523 | tzIDKey[tzIDKeyLen] = 0; |
| 524 | |
| 525 | const UChar *locname = (const UChar *)uhash_get(fLocationNamesMap, tzIDKey); |
| 526 | |
| 527 | if (locname != NULL) { |
| 528 | // gEmpty indicate the name is not available |
| 529 | if (locname == gEmpty) { |
| 530 | return NULL; |
| 531 | } |
| 532 | return locname; |
| 533 | } |
| 534 | |
| 535 | // Construct location name |
| 536 | UnicodeString name; |
| 537 | UnicodeString usCountryCode; |
| 538 | UBool isPrimary = FALSE; |
| 539 | |
| 540 | ZoneMeta::getCanonicalCountry(tzCanonicalID, usCountryCode, &isPrimary); |
| 541 | |
| 542 | if (!usCountryCode.isEmpty()) { |
| 543 | if (isPrimary) { |
| 544 | // If this is the primary zone in the country, use the country name. |
| 545 | char countryCode[ULOC_COUNTRY_CAPACITY]; |
| 546 | U_ASSERT(usCountryCode.length() < ULOC_COUNTRY_CAPACITY); |
| 547 | int32_t ccLen = usCountryCode.extract(0, usCountryCode.length(), countryCode, sizeof(countryCode), US_INV); |
| 548 | countryCode[ccLen] = 0; |
| 549 | |
| 550 | UnicodeString country; |
| 551 | fLocaleDisplayNames->regionDisplayName(countryCode, country); |
| 552 | fRegionFormat.format(country, name, status); |
| 553 | } else { |
| 554 | // If this is not the primary zone in the country, |
| 555 | // use the exemplar city name. |
| 556 | |
| 557 | // getExemplarLocationName should retur non-empty string |
| 558 | // if the time zone is associated with a region |
| 559 | |
| 560 | UnicodeString city; |
| 561 | fTimeZoneNames->getExemplarLocationName(tzCanonicalID, city); |
| 562 | fRegionFormat.format(city, name, status); |
| 563 | } |
| 564 | if (U_FAILURE(status)) { |
| 565 | return NULL; |
| 566 | } |
| 567 | } |
| 568 | |
| 569 | locname = name.isEmpty() ? NULL : fStringPool.get(name, status); |
| 570 | if (U_SUCCESS(status)) { |
| 571 | // Cache the result |
| 572 | const UChar* cacheID = ZoneMeta::findTimeZoneID(tzCanonicalID); |
| 573 | U_ASSERT(cacheID != NULL); |
| 574 | if (locname == NULL) { |
| 575 | // gEmpty to indicate - no location name available |
| 576 | uhash_put(fLocationNamesMap, (void *)cacheID, (void *)gEmpty, &status); |
| 577 | } else { |
| 578 | uhash_put(fLocationNamesMap, (void *)cacheID, (void *)locname, &status); |
| 579 | if (U_FAILURE(status)) { |
| 580 | locname = NULL; |
| 581 | } else { |
| 582 | // put the name info into the trie |
| 583 | GNameInfo *nameinfo = (ZNameInfo *)uprv_malloc(sizeof(GNameInfo)); |
| 584 | if (nameinfo != NULL) { |
| 585 | nameinfo->type = UTZGNM_LOCATION; |
| 586 | nameinfo->tzID = cacheID; |
| 587 | fGNamesTrie.put(locname, nameinfo, status); |
| 588 | } |
| 589 | } |
| 590 | } |
| 591 | } |
| 592 | |
| 593 | return locname; |
| 594 | } |
| 595 | |
| 596 | UnicodeString& |
| 597 | TZGNCore::formatGenericNonLocationName(const TimeZone& tz, UTimeZoneGenericNameType type, UDate date, UnicodeString& name) const { |
| 598 | U_ASSERT(type == UTZGNM_LONG || type == UTZGNM_SHORT); |
| 599 | name.setToBogus(); |
| 600 | |
| 601 | const UChar* uID = ZoneMeta::getCanonicalCLDRID(tz); |
| 602 | if (uID == NULL) { |
| 603 | return name; |
| 604 | } |
| 605 | |
| 606 | UnicodeString tzID(TRUE, uID, -1); |
| 607 | |
| 608 | // Try to get a name from time zone first |
| 609 | UTimeZoneNameType nameType = (type == UTZGNM_LONG) ? UTZNM_LONG_GENERIC : UTZNM_SHORT_GENERIC; |
| 610 | fTimeZoneNames->getTimeZoneDisplayName(tzID, nameType, name); |
| 611 | |
| 612 | if (!name.isEmpty()) { |
| 613 | return name; |
| 614 | } |
| 615 | |
| 616 | // Try meta zone |
| 617 | UChar mzIDBuf[32]; |
| 618 | UnicodeString mzID(mzIDBuf, 0, UPRV_LENGTHOF(mzIDBuf)); |
| 619 | fTimeZoneNames->getMetaZoneID(tzID, date, mzID); |
| 620 | if (!mzID.isEmpty()) { |
| 621 | UErrorCode status = U_ZERO_ERROR; |
| 622 | UBool useStandard = FALSE; |
| 623 | int32_t raw, sav; |
| 624 | UChar tmpNameBuf[ZONE_NAME_U16_MAX]; |
| 625 | |
| 626 | tz.getOffset(date, FALSE, raw, sav, status); |
| 627 | if (U_FAILURE(status)) { |
| 628 | return name; |
| 629 | } |
| 630 | |
| 631 | if (sav == 0) { |
| 632 | useStandard = TRUE; |
| 633 | |
| 634 | TimeZone *tmptz = tz.clone(); |
| 635 | // Check if the zone actually uses daylight saving time around the time |
| 636 | BasicTimeZone *btz = NULL; |
| 637 | if (dynamic_cast<OlsonTimeZone *>(tmptz) != NULL |
| 638 | || dynamic_cast<SimpleTimeZone *>(tmptz) != NULL |
| 639 | || dynamic_cast<RuleBasedTimeZone *>(tmptz) != NULL |
| 640 | || dynamic_cast<VTimeZone *>(tmptz) != NULL) { |
| 641 | btz = (BasicTimeZone*)tmptz; |
| 642 | } |
| 643 | |
| 644 | if (btz != NULL) { |
| 645 | TimeZoneTransition before; |
| 646 | UBool beforTrs = btz->getPreviousTransition(date, TRUE, before); |
| 647 | if (beforTrs |
| 648 | && (date - before.getTime() < kDstCheckRange) |
| 649 | && before.getFrom()->getDSTSavings() != 0) { |
| 650 | useStandard = FALSE; |
| 651 | } else { |
| 652 | TimeZoneTransition after; |
| 653 | UBool afterTrs = btz->getNextTransition(date, FALSE, after); |
| 654 | if (afterTrs |
| 655 | && (after.getTime() - date < kDstCheckRange) |
| 656 | && after.getTo()->getDSTSavings() != 0) { |
| 657 | useStandard = FALSE; |
| 658 | } |
| 659 | } |
| 660 | } else { |
| 661 | // If not BasicTimeZone... only if the instance is not an ICU's implementation. |
| 662 | // We may get a wrong answer in edge case, but it should practically work OK. |
| 663 | tmptz->getOffset(date - kDstCheckRange, FALSE, raw, sav, status); |
| 664 | if (sav != 0) { |
| 665 | useStandard = FALSE; |
| 666 | } else { |
| 667 | tmptz->getOffset(date + kDstCheckRange, FALSE, raw, sav, status); |
| 668 | if (sav != 0){ |
| 669 | useStandard = FALSE; |
| 670 | } |
| 671 | } |
| 672 | if (U_FAILURE(status)) { |
| 673 | delete tmptz; |
| 674 | return name; |
| 675 | } |
| 676 | } |
| 677 | delete tmptz; |
| 678 | } |
| 679 | if (useStandard) { |
| 680 | UTimeZoneNameType stdNameType = (nameType == UTZNM_LONG_GENERIC) |
| 681 | ? UTZNM_LONG_STANDARD : UTZNM_SHORT_STANDARD; |
| 682 | UnicodeString stdName(tmpNameBuf, 0, UPRV_LENGTHOF(tmpNameBuf)); |
| 683 | fTimeZoneNames->getDisplayName(tzID, stdNameType, date, stdName); |
| 684 | if (!stdName.isEmpty()) { |
| 685 | name.setTo(stdName); |
| 686 | |
| 687 | // TODO: revisit this issue later |
| 688 | // In CLDR, a same display name is used for both generic and standard |
| 689 | // for some meta zones in some locales. This looks like a data bugs. |
| 690 | // For now, we check if the standard name is different from its generic |
| 691 | // name below. |
| 692 | UChar genNameBuf[ZONE_NAME_U16_MAX]; |
| 693 | UnicodeString mzGenericName(genNameBuf, 0, UPRV_LENGTHOF(genNameBuf)); |
| 694 | fTimeZoneNames->getMetaZoneDisplayName(mzID, nameType, mzGenericName); |
| 695 | if (stdName.caseCompare(mzGenericName, 0) == 0) { |
| 696 | name.setToBogus(); |
| 697 | } |
| 698 | } |
| 699 | } |
| 700 | if (name.isEmpty()) { |
| 701 | // Get a name from meta zone |
| 702 | UnicodeString mzName(tmpNameBuf, 0, UPRV_LENGTHOF(tmpNameBuf)); |
| 703 | fTimeZoneNames->getMetaZoneDisplayName(mzID, nameType, mzName); |
| 704 | if (!mzName.isEmpty()) { |
| 705 | // Check if we need to use a partial location format. |
| 706 | // This check is done by comparing offset with the meta zone's |
| 707 | // golden zone at the given date. |
| 708 | UChar idBuf[32]; |
| 709 | UnicodeString goldenID(idBuf, 0, UPRV_LENGTHOF(idBuf)); |
| 710 | fTimeZoneNames->getReferenceZoneID(mzID, fTargetRegion, goldenID); |
| 711 | if (!goldenID.isEmpty() && goldenID != tzID) { |
| 712 | TimeZone *goldenZone = TimeZone::createTimeZone(goldenID); |
| 713 | int32_t raw1, sav1; |
| 714 | |
| 715 | // Check offset in the golden zone with wall time. |
| 716 | // With getOffset(date, false, offsets1), |
| 717 | // you may get incorrect results because of time overlap at DST->STD |
| 718 | // transition. |
| 719 | goldenZone->getOffset(date + raw + sav, TRUE, raw1, sav1, status); |
| 720 | delete goldenZone; |
| 721 | if (U_SUCCESS(status)) { |
| 722 | if (raw != raw1 || sav != sav1) { |
| 723 | // Now we need to use a partial location format |
| 724 | getPartialLocationName(tzID, mzID, (nameType == UTZNM_LONG_GENERIC), mzName, name); |
| 725 | } else { |
| 726 | name.setTo(mzName); |
| 727 | } |
| 728 | } |
| 729 | } else { |
| 730 | name.setTo(mzName); |
| 731 | } |
| 732 | } |
| 733 | } |
| 734 | } |
| 735 | return name; |
| 736 | } |
| 737 | |
| 738 | UnicodeString& |
| 739 | TZGNCore::getPartialLocationName(const UnicodeString& tzCanonicalID, |
| 740 | const UnicodeString& mzID, UBool isLong, const UnicodeString& mzDisplayName, |
| 741 | UnicodeString& name) const { |
| 742 | name.setToBogus(); |
| 743 | if (tzCanonicalID.isEmpty() || mzID.isEmpty() || mzDisplayName.isEmpty()) { |
| 744 | return name; |
| 745 | } |
| 746 | |
| 747 | const UChar *uplname = NULL; |
| 748 | TZGNCore *nonConstThis = const_cast<TZGNCore *>(this); |
| 749 | umtx_lock(&gLock); |
| 750 | { |
| 751 | uplname = nonConstThis->getPartialLocationName(tzCanonicalID, mzID, isLong, mzDisplayName); |
| 752 | } |
| 753 | umtx_unlock(&gLock); |
| 754 | |
| 755 | if (uplname == NULL) { |
| 756 | name.setToBogus(); |
| 757 | } else { |
| 758 | name.setTo(TRUE, uplname, -1); |
| 759 | } |
| 760 | return name; |
| 761 | } |
| 762 | |
| 763 | /* |
| 764 | * This method updates the cache and must be called with a lock |
| 765 | */ |
| 766 | const UChar* |
| 767 | TZGNCore::getPartialLocationName(const UnicodeString& tzCanonicalID, |
| 768 | const UnicodeString& mzID, UBool isLong, const UnicodeString& mzDisplayName) { |
| 769 | U_ASSERT(!tzCanonicalID.isEmpty()); |
| 770 | U_ASSERT(!mzID.isEmpty()); |
| 771 | U_ASSERT(!mzDisplayName.isEmpty()); |
| 772 | |
| 773 | PartialLocationKey key; |
| 774 | key.tzID = ZoneMeta::findTimeZoneID(tzCanonicalID); |
| 775 | key.mzID = ZoneMeta::findMetaZoneID(mzID); |
| 776 | key.isLong = isLong; |
| 777 | U_ASSERT(key.tzID != NULL && key.mzID != NULL); |
| 778 | |
| 779 | const UChar* uplname = (const UChar*)uhash_get(fPartialLocationNamesMap, (void *)&key); |
| 780 | if (uplname != NULL) { |
| 781 | return uplname; |
| 782 | } |
| 783 | |
| 784 | UnicodeString location; |
| 785 | UnicodeString usCountryCode; |
| 786 | ZoneMeta::getCanonicalCountry(tzCanonicalID, usCountryCode); |
| 787 | if (!usCountryCode.isEmpty()) { |
| 788 | char countryCode[ULOC_COUNTRY_CAPACITY]; |
| 789 | U_ASSERT(usCountryCode.length() < ULOC_COUNTRY_CAPACITY); |
| 790 | int32_t ccLen = usCountryCode.extract(0, usCountryCode.length(), countryCode, sizeof(countryCode), US_INV); |
| 791 | countryCode[ccLen] = 0; |
| 792 | |
| 793 | UnicodeString regionalGolden; |
| 794 | fTimeZoneNames->getReferenceZoneID(mzID, countryCode, regionalGolden); |
| 795 | if (tzCanonicalID == regionalGolden) { |
| 796 | // Use country name |
| 797 | fLocaleDisplayNames->regionDisplayName(countryCode, location); |
| 798 | } else { |
| 799 | // Otherwise, use exemplar city name |
| 800 | fTimeZoneNames->getExemplarLocationName(tzCanonicalID, location); |
| 801 | } |
| 802 | } else { |
| 803 | fTimeZoneNames->getExemplarLocationName(tzCanonicalID, location); |
| 804 | if (location.isEmpty()) { |
| 805 | // This could happen when the time zone is not associated with a country, |
| 806 | // and its ID is not hierarchical, for example, CST6CDT. |
| 807 | // We use the canonical ID itself as the location for this case. |
| 808 | location.setTo(tzCanonicalID); |
| 809 | } |
| 810 | } |
| 811 | |
| 812 | UErrorCode status = U_ZERO_ERROR; |
| 813 | UnicodeString name; |
| 814 | fFallbackFormat.format(location, mzDisplayName, name, status); |
| 815 | if (U_FAILURE(status)) { |
| 816 | return NULL; |
| 817 | } |
| 818 | |
| 819 | uplname = fStringPool.get(name, status); |
| 820 | if (U_SUCCESS(status)) { |
| 821 | // Add the name to cache |
| 822 | PartialLocationKey* cacheKey = (PartialLocationKey *)uprv_malloc(sizeof(PartialLocationKey)); |
| 823 | if (cacheKey != NULL) { |
| 824 | cacheKey->tzID = key.tzID; |
| 825 | cacheKey->mzID = key.mzID; |
| 826 | cacheKey->isLong = key.isLong; |
| 827 | uhash_put(fPartialLocationNamesMap, (void *)cacheKey, (void *)uplname, &status); |
| 828 | if (U_FAILURE(status)) { |
| 829 | uprv_free(cacheKey); |
| 830 | } else { |
| 831 | // put the name to the local trie as well |
| 832 | GNameInfo *nameinfo = (ZNameInfo *)uprv_malloc(sizeof(GNameInfo)); |
| 833 | if (nameinfo != NULL) { |
| 834 | nameinfo->type = isLong ? UTZGNM_LONG : UTZGNM_SHORT; |
| 835 | nameinfo->tzID = key.tzID; |
| 836 | fGNamesTrie.put(uplname, nameinfo, status); |
| 837 | } |
| 838 | } |
| 839 | } |
| 840 | } |
| 841 | return uplname; |
| 842 | } |
| 843 | |
| 844 | /* |
| 845 | * This method updates the cache and must be called with a lock, |
| 846 | * except initializer. |
| 847 | */ |
| 848 | void |
| 849 | TZGNCore::loadStrings(const UnicodeString& tzCanonicalID) { |
| 850 | // load the generic location name |
| 851 | getGenericLocationName(tzCanonicalID); |
| 852 | |
| 853 | // partial location names |
| 854 | UErrorCode status = U_ZERO_ERROR; |
| 855 | |
| 856 | const UnicodeString *mzID; |
| 857 | UnicodeString goldenID; |
| 858 | UnicodeString mzGenName; |
| 859 | UTimeZoneNameType genNonLocTypes[] = { |
| 860 | UTZNM_LONG_GENERIC, UTZNM_SHORT_GENERIC, |
| 861 | UTZNM_UNKNOWN /*terminator*/ |
| 862 | }; |
| 863 | |
| 864 | StringEnumeration *mzIDs = fTimeZoneNames->getAvailableMetaZoneIDs(tzCanonicalID, status); |
| 865 | while ((mzID = mzIDs->snext(status)) != NULL) { |
| 866 | if (U_FAILURE(status)) { |
| 867 | break; |
| 868 | } |
| 869 | // if this time zone is not the golden zone of the meta zone, |
| 870 | // partial location name (such as "PT (Los Angeles)") might be |
| 871 | // available. |
| 872 | fTimeZoneNames->getReferenceZoneID(*mzID, fTargetRegion, goldenID); |
| 873 | if (tzCanonicalID != goldenID) { |
| 874 | for (int32_t i = 0; genNonLocTypes[i] != UTZNM_UNKNOWN; i++) { |
| 875 | fTimeZoneNames->getMetaZoneDisplayName(*mzID, genNonLocTypes[i], mzGenName); |
| 876 | if (!mzGenName.isEmpty()) { |
| 877 | // getPartialLocationName formats a name and put it into the trie |
| 878 | getPartialLocationName(tzCanonicalID, *mzID, |
| 879 | (genNonLocTypes[i] == UTZNM_LONG_GENERIC), mzGenName); |
| 880 | } |
| 881 | } |
| 882 | } |
| 883 | } |
| 884 | if (mzIDs != NULL) { |
| 885 | delete mzIDs; |
| 886 | } |
| 887 | } |
| 888 | |
| 889 | int32_t |
| 890 | TZGNCore::findBestMatch(const UnicodeString& text, int32_t start, uint32_t types, |
| 891 | UnicodeString& tzID, UTimeZoneFormatTimeType& timeType, UErrorCode& status) const { |
| 892 | timeType = UTZFMT_TIME_TYPE_UNKNOWN; |
| 893 | tzID.setToBogus(); |
| 894 | |
| 895 | if (U_FAILURE(status)) { |
| 896 | return 0; |
| 897 | } |
| 898 | |
| 899 | // Find matches in the TimeZoneNames first |
| 900 | TimeZoneNames::MatchInfoCollection *tznamesMatches = findTimeZoneNames(text, start, types, status); |
| 901 | if (U_FAILURE(status)) { |
| 902 | return 0; |
| 903 | } |
| 904 | |
| 905 | int32_t bestMatchLen = 0; |
| 906 | UTimeZoneFormatTimeType bestMatchTimeType = UTZFMT_TIME_TYPE_UNKNOWN; |
| 907 | UnicodeString bestMatchTzID; |
| 908 | // UBool isLongStandard = FALSE; // workaround - see the comments below |
| 909 | UBool isStandard = FALSE; // TODO: Temporary hack (on hack) for short standard name/location name conflict (found in zh_Hant), should be removed after CLDR 21m1 integration |
| 910 | |
| 911 | if (tznamesMatches != NULL) { |
| 912 | UnicodeString mzID; |
| 913 | for (int32_t i = 0; i < tznamesMatches->size(); i++) { |
| 914 | int32_t len = tznamesMatches->getMatchLengthAt(i); |
| 915 | if (len > bestMatchLen) { |
| 916 | bestMatchLen = len; |
| 917 | if (!tznamesMatches->getTimeZoneIDAt(i, bestMatchTzID)) { |
| 918 | // name for a meta zone |
| 919 | if (tznamesMatches->getMetaZoneIDAt(i, mzID)) { |
| 920 | fTimeZoneNames->getReferenceZoneID(mzID, fTargetRegion, bestMatchTzID); |
| 921 | } |
| 922 | } |
| 923 | UTimeZoneNameType nameType = tznamesMatches->getNameTypeAt(i); |
| 924 | if (U_FAILURE(status)) { |
| 925 | break; |
| 926 | } |
| 927 | switch (nameType) { |
| 928 | case UTZNM_LONG_STANDARD: |
| 929 | // isLongStandard = TRUE; |
| 930 | case UTZNM_SHORT_STANDARD: // this one is never used for generic, but just in case |
| 931 | isStandard = TRUE; // TODO: Remove this later, see the comments above. |
| 932 | bestMatchTimeType = UTZFMT_TIME_TYPE_STANDARD; |
| 933 | break; |
| 934 | case UTZNM_LONG_DAYLIGHT: |
| 935 | case UTZNM_SHORT_DAYLIGHT: // this one is never used for generic, but just in case |
| 936 | bestMatchTimeType = UTZFMT_TIME_TYPE_DAYLIGHT; |
| 937 | break; |
| 938 | default: |
| 939 | bestMatchTimeType = UTZFMT_TIME_TYPE_UNKNOWN; |
| 940 | } |
| 941 | } |
| 942 | } |
| 943 | delete tznamesMatches; |
| 944 | if (U_FAILURE(status)) { |
| 945 | return 0; |
| 946 | } |
| 947 | |
| 948 | if (bestMatchLen == (text.length() - start)) { |
| 949 | // Full match |
| 950 | |
| 951 | //tzID.setTo(bestMatchTzID); |
| 952 | //timeType = bestMatchTimeType; |
| 953 | //return bestMatchLen; |
| 954 | |
| 955 | // TODO Some time zone uses a same name for the long standard name |
| 956 | // and the location name. When the match is a long standard name, |
| 957 | // then we need to check if the name is same with the location name. |
| 958 | // This is probably a data error or a design bug. |
| 959 | /* |
| 960 | if (!isLongStandard) { |
| 961 | tzID.setTo(bestMatchTzID); |
| 962 | timeType = bestMatchTimeType; |
| 963 | return bestMatchLen; |
| 964 | } |
| 965 | */ |
| 966 | // TODO The deprecation of commonlyUsed flag introduced the name |
| 967 | // conflict not only for long standard names, but short standard names too. |
| 968 | // These short names (found in zh_Hant) should be gone once we clean |
| 969 | // up CLDR time zone display name data. Once the short name conflict |
| 970 | // problem (with location name) is resolved, we should change the condition |
| 971 | // below back to the original one above. -Yoshito (2011-09-14) |
| 972 | if (!isStandard) { |
| 973 | tzID.setTo(bestMatchTzID); |
| 974 | timeType = bestMatchTimeType; |
| 975 | return bestMatchLen; |
| 976 | } |
| 977 | } |
| 978 | } |
| 979 | |
| 980 | // Find matches in the local trie |
| 981 | TimeZoneGenericNameMatchInfo *localMatches = findLocal(text, start, types, status); |
| 982 | if (U_FAILURE(status)) { |
| 983 | return 0; |
| 984 | } |
| 985 | if (localMatches != NULL) { |
| 986 | for (int32_t i = 0; i < localMatches->size(); i++) { |
| 987 | int32_t len = localMatches->getMatchLength(i); |
| 988 | |
| 989 | // TODO See the above TODO. We use len >= bestMatchLen |
| 990 | // because of the long standard/location name collision |
| 991 | // problem. If it is also a location name, carrying |
| 992 | // timeType = UTZFMT_TIME_TYPE_STANDARD will cause a |
| 993 | // problem in SimpleDateFormat |
| 994 | if (len >= bestMatchLen) { |
| 995 | bestMatchLen = localMatches->getMatchLength(i); |
| 996 | bestMatchTimeType = UTZFMT_TIME_TYPE_UNKNOWN; // because generic |
| 997 | localMatches->getTimeZoneID(i, bestMatchTzID); |
| 998 | } |
| 999 | } |
| 1000 | delete localMatches; |
| 1001 | } |
| 1002 | |
| 1003 | if (bestMatchLen > 0) { |
| 1004 | timeType = bestMatchTimeType; |
| 1005 | tzID.setTo(bestMatchTzID); |
| 1006 | } |
| 1007 | return bestMatchLen; |
| 1008 | } |
| 1009 | |
| 1010 | TimeZoneGenericNameMatchInfo* |
| 1011 | TZGNCore::findLocal(const UnicodeString& text, int32_t start, uint32_t types, UErrorCode& status) const { |
| 1012 | GNameSearchHandler handler(types); |
| 1013 | |
| 1014 | TZGNCore *nonConstThis = const_cast<TZGNCore *>(this); |
| 1015 | |
| 1016 | umtx_lock(&gLock); |
| 1017 | { |
| 1018 | fGNamesTrie.search(text, start, (TextTrieMapSearchResultHandler *)&handler, status); |
| 1019 | } |
| 1020 | umtx_unlock(&gLock); |
| 1021 | |
| 1022 | if (U_FAILURE(status)) { |
| 1023 | return NULL; |
| 1024 | } |
| 1025 | |
| 1026 | TimeZoneGenericNameMatchInfo *gmatchInfo = NULL; |
| 1027 | |
| 1028 | int32_t maxLen = 0; |
| 1029 | UVector *results = handler.getMatches(maxLen); |
| 1030 | if (results != NULL && ((maxLen == (text.length() - start)) || fGNamesTrieFullyLoaded)) { |
| 1031 | // perfect match |
| 1032 | gmatchInfo = new TimeZoneGenericNameMatchInfo(results); |
| 1033 | if (gmatchInfo == NULL) { |
| 1034 | status = U_MEMORY_ALLOCATION_ERROR; |
| 1035 | delete results; |
| 1036 | return NULL; |
| 1037 | } |
| 1038 | return gmatchInfo; |
| 1039 | } |
| 1040 | |
| 1041 | if (results != NULL) { |
| 1042 | delete results; |
| 1043 | } |
| 1044 | |
| 1045 | // All names are not yet loaded into the local trie. |
| 1046 | // Load all available names into the trie. This could be very heavy. |
| 1047 | umtx_lock(&gLock); |
| 1048 | { |
| 1049 | if (!fGNamesTrieFullyLoaded) { |
| 1050 | StringEnumeration *tzIDs = TimeZone::createTimeZoneIDEnumeration(UCAL_ZONE_TYPE_CANONICAL, NULL, NULL, status); |
| 1051 | if (U_SUCCESS(status)) { |
| 1052 | const UnicodeString *tzID; |
| 1053 | while ((tzID = tzIDs->snext(status)) != NULL) { |
| 1054 | if (U_FAILURE(status)) { |
| 1055 | break; |
| 1056 | } |
| 1057 | nonConstThis->loadStrings(*tzID); |
| 1058 | } |
| 1059 | } |
| 1060 | if (tzIDs != NULL) { |
| 1061 | delete tzIDs; |
| 1062 | } |
| 1063 | |
| 1064 | if (U_SUCCESS(status)) { |
| 1065 | nonConstThis->fGNamesTrieFullyLoaded = TRUE; |
| 1066 | } |
| 1067 | } |
| 1068 | } |
| 1069 | umtx_unlock(&gLock); |
| 1070 | |
| 1071 | if (U_FAILURE(status)) { |
| 1072 | return NULL; |
| 1073 | } |
| 1074 | |
| 1075 | umtx_lock(&gLock); |
| 1076 | { |
| 1077 | // now try it again |
| 1078 | fGNamesTrie.search(text, start, (TextTrieMapSearchResultHandler *)&handler, status); |
| 1079 | } |
| 1080 | umtx_unlock(&gLock); |
| 1081 | |
| 1082 | results = handler.getMatches(maxLen); |
| 1083 | if (results != NULL && maxLen > 0) { |
| 1084 | gmatchInfo = new TimeZoneGenericNameMatchInfo(results); |
| 1085 | if (gmatchInfo == NULL) { |
| 1086 | status = U_MEMORY_ALLOCATION_ERROR; |
| 1087 | delete results; |
| 1088 | return NULL; |
| 1089 | } |
| 1090 | } |
| 1091 | |
| 1092 | return gmatchInfo; |
| 1093 | } |
| 1094 | |
| 1095 | TimeZoneNames::MatchInfoCollection* |
| 1096 | TZGNCore::findTimeZoneNames(const UnicodeString& text, int32_t start, uint32_t types, UErrorCode& status) const { |
| 1097 | // Check if the target name typs is really in the TimeZoneNames |
| 1098 | uint32_t nameTypes = 0; |
| 1099 | if (types & UTZGNM_LONG) { |
| 1100 | nameTypes |= (UTZNM_LONG_GENERIC | UTZNM_LONG_STANDARD); |
| 1101 | } |
| 1102 | if (types & UTZGNM_SHORT) { |
| 1103 | nameTypes |= (UTZNM_SHORT_GENERIC | UTZNM_SHORT_STANDARD); |
| 1104 | } |
| 1105 | |
| 1106 | if (types) { |
| 1107 | // Find matches in the TimeZoneNames |
| 1108 | return fTimeZoneNames->find(text, start, nameTypes, status); |
| 1109 | } |
| 1110 | |
| 1111 | return NULL; |
| 1112 | } |
| 1113 | |
| 1114 | typedef struct TZGNCoreRef { |
| 1115 | TZGNCore* obj; |
| 1116 | int32_t refCount; |
| 1117 | double lastAccess; |
| 1118 | } TZGNCoreRef; |
| 1119 | |
| 1120 | // TZGNCore object cache handling |
| 1121 | static UMutex gTZGNLock; |
| 1122 | static UHashtable *gTZGNCoreCache = NULL; |
| 1123 | static UBool gTZGNCoreCacheInitialized = FALSE; |
| 1124 | |
| 1125 | // Access count - incremented every time up to SWEEP_INTERVAL, |
| 1126 | // then reset to 0 |
| 1127 | static int32_t gAccessCount = 0; |
| 1128 | |
| 1129 | // Interval for calling the cache sweep function - every 100 times |
| 1130 | #define SWEEP_INTERVAL 100 |
| 1131 | |
| 1132 | // Cache expiration in millisecond. When a cached entry is no |
| 1133 | // longer referenced and exceeding this threshold since last |
| 1134 | // access time, then the cache entry will be deleted by the sweep |
| 1135 | // function. For now, 3 minutes. |
| 1136 | #define CACHE_EXPIRATION 180000.0 |
| 1137 | |
| 1138 | U_CDECL_BEGIN |
| 1139 | /** |
| 1140 | * Cleanup callback func |
| 1141 | */ |
| 1142 | static UBool U_CALLCONV tzgnCore_cleanup(void) |
| 1143 | { |
| 1144 | if (gTZGNCoreCache != NULL) { |
| 1145 | uhash_close(gTZGNCoreCache); |
| 1146 | gTZGNCoreCache = NULL; |
| 1147 | } |
| 1148 | gTZGNCoreCacheInitialized = FALSE; |
| 1149 | return TRUE; |
| 1150 | } |
| 1151 | |
| 1152 | /** |
| 1153 | * Deleter for TZGNCoreRef |
| 1154 | */ |
| 1155 | static void U_CALLCONV |
| 1156 | deleteTZGNCoreRef(void *obj) { |
| 1157 | icu::TZGNCoreRef *entry = (icu::TZGNCoreRef*)obj; |
| 1158 | delete (icu::TZGNCore*) entry->obj; |
| 1159 | uprv_free(entry); |
| 1160 | } |
| 1161 | U_CDECL_END |
| 1162 | |
| 1163 | /** |
| 1164 | * Function used for removing unreferrenced cache entries exceeding |
| 1165 | * the expiration time. This function must be called with in the mutex |
| 1166 | * block. |
| 1167 | */ |
| 1168 | static void sweepCache() { |
| 1169 | int32_t pos = UHASH_FIRST; |
| 1170 | const UHashElement* elem; |
| 1171 | double now = (double)uprv_getUTCtime(); |
| 1172 | |
| 1173 | while ((elem = uhash_nextElement(gTZGNCoreCache, &pos)) != NULL) { |
| 1174 | TZGNCoreRef *entry = (TZGNCoreRef *)elem->value.pointer; |
| 1175 | if (entry->refCount <= 0 && (now - entry->lastAccess) > CACHE_EXPIRATION) { |
| 1176 | // delete this entry |
| 1177 | uhash_removeElement(gTZGNCoreCache, elem); |
| 1178 | } |
| 1179 | } |
| 1180 | } |
| 1181 | |
| 1182 | TimeZoneGenericNames::TimeZoneGenericNames() |
| 1183 | : fRef(0) { |
| 1184 | } |
| 1185 | |
| 1186 | TimeZoneGenericNames::~TimeZoneGenericNames() { |
| 1187 | umtx_lock(&gTZGNLock); |
| 1188 | { |
| 1189 | U_ASSERT(fRef->refCount > 0); |
| 1190 | // Just decrement the reference count |
| 1191 | fRef->refCount--; |
| 1192 | } |
| 1193 | umtx_unlock(&gTZGNLock); |
| 1194 | } |
| 1195 | |
| 1196 | TimeZoneGenericNames* |
| 1197 | TimeZoneGenericNames::createInstance(const Locale& locale, UErrorCode& status) { |
| 1198 | if (U_FAILURE(status)) { |
| 1199 | return NULL; |
| 1200 | } |
| 1201 | TimeZoneGenericNames* instance = new TimeZoneGenericNames(); |
| 1202 | if (instance == NULL) { |
| 1203 | status = U_MEMORY_ALLOCATION_ERROR; |
| 1204 | return NULL; |
| 1205 | } |
| 1206 | |
| 1207 | TZGNCoreRef *cacheEntry = NULL; |
| 1208 | { |
| 1209 | Mutex lock(&gTZGNLock); |
| 1210 | |
| 1211 | if (!gTZGNCoreCacheInitialized) { |
| 1212 | // Create empty hashtable |
| 1213 | gTZGNCoreCache = uhash_open(uhash_hashChars, uhash_compareChars, NULL, &status); |
| 1214 | if (U_SUCCESS(status)) { |
| 1215 | uhash_setKeyDeleter(gTZGNCoreCache, uprv_free); |
| 1216 | uhash_setValueDeleter(gTZGNCoreCache, deleteTZGNCoreRef); |
| 1217 | gTZGNCoreCacheInitialized = TRUE; |
| 1218 | ucln_i18n_registerCleanup(UCLN_I18N_TIMEZONEGENERICNAMES, tzgnCore_cleanup); |
| 1219 | } |
| 1220 | } |
| 1221 | if (U_FAILURE(status)) { |
| 1222 | return NULL; |
| 1223 | } |
| 1224 | |
| 1225 | // Check the cache, if not available, create new one and cache |
| 1226 | const char *key = locale.getName(); |
| 1227 | cacheEntry = (TZGNCoreRef *)uhash_get(gTZGNCoreCache, key); |
| 1228 | if (cacheEntry == NULL) { |
| 1229 | TZGNCore *tzgnCore = NULL; |
| 1230 | char *newKey = NULL; |
| 1231 | |
| 1232 | tzgnCore = new TZGNCore(locale, status); |
| 1233 | if (tzgnCore == NULL) { |
| 1234 | status = U_MEMORY_ALLOCATION_ERROR; |
| 1235 | } |
| 1236 | if (U_SUCCESS(status)) { |
| 1237 | newKey = (char *)uprv_malloc(uprv_strlen(key) + 1); |
| 1238 | if (newKey == NULL) { |
| 1239 | status = U_MEMORY_ALLOCATION_ERROR; |
| 1240 | } else { |
| 1241 | uprv_strcpy(newKey, key); |
| 1242 | } |
| 1243 | } |
| 1244 | if (U_SUCCESS(status)) { |
| 1245 | cacheEntry = (TZGNCoreRef *)uprv_malloc(sizeof(TZGNCoreRef)); |
| 1246 | if (cacheEntry == NULL) { |
| 1247 | status = U_MEMORY_ALLOCATION_ERROR; |
| 1248 | } else { |
| 1249 | cacheEntry->obj = tzgnCore; |
| 1250 | cacheEntry->refCount = 1; |
| 1251 | cacheEntry->lastAccess = (double)uprv_getUTCtime(); |
| 1252 | |
| 1253 | uhash_put(gTZGNCoreCache, newKey, cacheEntry, &status); |
| 1254 | } |
| 1255 | } |
| 1256 | if (U_FAILURE(status)) { |
| 1257 | if (tzgnCore != NULL) { |
| 1258 | delete tzgnCore; |
| 1259 | } |
| 1260 | if (newKey != NULL) { |
| 1261 | uprv_free(newKey); |
| 1262 | } |
| 1263 | if (cacheEntry != NULL) { |
| 1264 | uprv_free(cacheEntry); |
| 1265 | } |
| 1266 | cacheEntry = NULL; |
| 1267 | } |
| 1268 | } else { |
| 1269 | // Update the reference count |
| 1270 | cacheEntry->refCount++; |
| 1271 | cacheEntry->lastAccess = (double)uprv_getUTCtime(); |
| 1272 | } |
| 1273 | gAccessCount++; |
| 1274 | if (gAccessCount >= SWEEP_INTERVAL) { |
| 1275 | // sweep |
| 1276 | sweepCache(); |
| 1277 | gAccessCount = 0; |
| 1278 | } |
| 1279 | } // End of mutex locked block |
| 1280 | |
| 1281 | if (cacheEntry == NULL) { |
| 1282 | delete instance; |
| 1283 | return NULL; |
| 1284 | } |
| 1285 | |
| 1286 | instance->fRef = cacheEntry; |
| 1287 | return instance; |
| 1288 | } |
| 1289 | |
| 1290 | UBool |
| 1291 | TimeZoneGenericNames::operator==(const TimeZoneGenericNames& other) const { |
| 1292 | // Just compare if the other object also use the same |
| 1293 | // ref entry |
| 1294 | return fRef == other.fRef; |
| 1295 | } |
| 1296 | |
| 1297 | TimeZoneGenericNames* |
| 1298 | TimeZoneGenericNames::clone() const { |
| 1299 | TimeZoneGenericNames* other = new TimeZoneGenericNames(); |
| 1300 | if (other) { |
| 1301 | umtx_lock(&gTZGNLock); |
| 1302 | { |
| 1303 | // Just increments the reference count |
| 1304 | fRef->refCount++; |
| 1305 | other->fRef = fRef; |
| 1306 | } |
| 1307 | umtx_unlock(&gTZGNLock); |
| 1308 | } |
| 1309 | return other; |
| 1310 | } |
| 1311 | |
| 1312 | UnicodeString& |
| 1313 | TimeZoneGenericNames::getDisplayName(const TimeZone& tz, UTimeZoneGenericNameType type, |
| 1314 | UDate date, UnicodeString& name) const { |
| 1315 | return fRef->obj->getDisplayName(tz, type, date, name); |
| 1316 | } |
| 1317 | |
| 1318 | UnicodeString& |
| 1319 | TimeZoneGenericNames::getGenericLocationName(const UnicodeString& tzCanonicalID, UnicodeString& name) const { |
| 1320 | return fRef->obj->getGenericLocationName(tzCanonicalID, name); |
| 1321 | } |
| 1322 | |
| 1323 | int32_t |
| 1324 | TimeZoneGenericNames::findBestMatch(const UnicodeString& text, int32_t start, uint32_t types, |
| 1325 | UnicodeString& tzID, UTimeZoneFormatTimeType& timeType, UErrorCode& status) const { |
| 1326 | return fRef->obj->findBestMatch(text, start, types, tzID, timeType, status); |
| 1327 | } |
| 1328 | |
| 1329 | U_NAMESPACE_END |
| 1330 | #endif |
| 1331 | |