icuSources/i18n/region.cpp

   1 // © 2016 and later: Unicode, Inc. and others.
   2 // License & terms of use: http://www.unicode.org/copyright.html
   3 /*
   4 *******************************************************************************
   5 * Copyright (C) 2014-2016, International Business Machines Corporation and
   6 * others. All Rights Reserved.
   7 *******************************************************************************
   8 *
   9 *
  10 * File REGION.CPP
  11 *
  12 * Modification History:*
  13 *   Date        Name        Description
  14 * 01/15/13      Emmons      Original Port from ICU4J
  15 ********************************************************************************
  16 */
  17
  18 /**
  19  * \file
  20  * \brief C++ API: Region classes (territory containment)
  21  */
  22
  23 #include "unicode/region.h"
  24 #include "unicode/utypes.h"
  25 #include "unicode/uobject.h"
  26 #include "unicode/unistr.h"
  27 #include "unicode/ures.h"
  28 #include "unicode/decimfmt.h"
  29 #include "ucln_in.h"
  30 #include "cstring.h"
  31 #include "mutex.h"
  32 #include "uhash.h"
  33 #include "umutex.h"
  34 #include "uresimp.h"
  35 #include "region_impl.h"
  36
  37 #if !UCONFIG_NO_FORMATTING
  38
  39
  40 U_CDECL_BEGIN
  41
  42 static void U_CALLCONV
  43 deleteRegion(void *obj) {
  44     delete (icu::Region *)obj;
  45 }
  46
  47 /**
  48  * Cleanup callback func
  49  */
  50 static UBool U_CALLCONV region_cleanup(void)
  51 {
  52     icu::Region::cleanupRegionData();
  53
  54     return TRUE;
  55 }
  56
  57 U_CDECL_END
  58
  59 U_NAMESPACE_BEGIN
  60
  61 static UInitOnce gRegionDataInitOnce = U_INITONCE_INITIALIZER;
  62 static UVector* availableRegions[URGN_LIMIT];
  63
  64 static UHashtable *regionAliases = NULL;
  65 static UHashtable *regionIDMap = NULL;
  66 static UHashtable *numericCodeMap = NULL;
  67 static UVector *allRegions = NULL;
  68
  69 static const UChar UNKNOWN_REGION_ID [] = { 0x5A, 0x5A, 0 };  /* "ZZ" */
  70 static const UChar OUTLYING_OCEANIA_REGION_ID [] = { 0x51, 0x4F, 0 };  /* "QO" */
  71 static const UChar WORLD_ID [] = { 0x30, 0x30, 0x31, 0 };  /* "001" */
  72 static const UChar RANGE_MARKER = 0x7E; /* '~' */
  73
  74 UOBJECT_DEFINE_RTTI_IMPLEMENTATION(RegionNameEnumeration)
  75
  76 /*
  77  * Initializes the region data from the ICU resource bundles.  The region data
  78  * contains the basic relationships such as which regions are known, what the numeric
  79  * codes are, any known aliases, and the territory containment data.
  80  *
  81  * If the region data has already loaded, then this method simply returns without doing
  82  * anything meaningful.
  83  */
  84 void U_CALLCONV Region::loadRegionData(UErrorCode &status) {
  85
  86     // Construct service objs first
  87     LocalUHashtablePointer newRegionIDMap(uhash_open(uhash_hashUnicodeString, uhash_compareUnicodeString, NULL, &status));
  88     LocalUHashtablePointer newNumericCodeMap(uhash_open(uhash_hashLong,uhash_compareLong,NULL,&status));
  89     LocalUHashtablePointer newRegionAliases(uhash_open(uhash_hashUnicodeString,uhash_compareUnicodeString,NULL,&status));
  90     LocalPointer<DecimalFormat> df(new DecimalFormat(status), status);
  91
  92     LocalPointer<UVector> continents(new UVector(uprv_deleteUObject, uhash_compareUnicodeString, status), status);
  93     LocalPointer<UVector> groupings(new UVector(uprv_deleteUObject, uhash_compareUnicodeString, status), status);
  94     allRegions = new UVector(uprv_deleteUObject, uhash_compareUnicodeString, status);
  95
  96     LocalUResourceBundlePointer metadata(ures_openDirect(NULL,"metadata",&status));
  97     LocalUResourceBundlePointer metadataAlias(ures_getByKey(metadata.getAlias(),"alias",NULL,&status));
  98     LocalUResourceBundlePointer territoryAlias(ures_getByKey(metadataAlias.getAlias(),"territory",NULL,&status));
  99
 100     LocalUResourceBundlePointer supplementalData(ures_openDirect(NULL,"supplementalData",&status));
 101     LocalUResourceBundlePointer codeMappings(ures_getByKey(supplementalData.getAlias(),"codeMappings",NULL,&status));
 102
 103     LocalUResourceBundlePointer idValidity(ures_getByKey(supplementalData.getAlias(),"idValidity",NULL,&status));
 104     LocalUResourceBundlePointer regionList(ures_getByKey(idValidity.getAlias(),"region",NULL,&status));
 105     LocalUResourceBundlePointer regionRegular(ures_getByKey(regionList.getAlias(),"regular",NULL,&status));
 106     LocalUResourceBundlePointer regionMacro(ures_getByKey(regionList.getAlias(),"macroregion",NULL,&status));
 107     LocalUResourceBundlePointer regionUnknown(ures_getByKey(regionList.getAlias(),"unknown",NULL,&status));
 108
 109     LocalUResourceBundlePointer territoryContainment(ures_getByKey(supplementalData.getAlias(),"territoryContainment",NULL,&status));
 110     LocalUResourceBundlePointer worldContainment(ures_getByKey(territoryContainment.getAlias(),"001",NULL,&status));
 111     LocalUResourceBundlePointer groupingContainment(ures_getByKey(territoryContainment.getAlias(),"grouping",NULL,&status));
 112
 113     if (U_FAILURE(status)) {
 114         return;
 115     }
 116
 117     // now, initialize
 118     df->setParseIntegerOnly(TRUE);
 119     uhash_setValueDeleter(newRegionIDMap.getAlias(), deleteRegion);  // regionIDMap owns objs
 120     uhash_setKeyDeleter(newRegionAliases.getAlias(), uprv_deleteUObject); // regionAliases owns the string keys
 121
 122
 123     while ( ures_hasNext(regionRegular.getAlias()) ) {
 124         UnicodeString regionName = ures_getNextUnicodeString(regionRegular.getAlias(),NULL,&status);
 125         int32_t rangeMarkerLocation = regionName.indexOf(RANGE_MARKER);
 126         UChar buf[6];
 127         regionName.extract(buf,6,status);
 128         if ( rangeMarkerLocation > 0 ) {
 129             UChar endRange = regionName.charAt(rangeMarkerLocation+1);
 130             buf[rangeMarkerLocation] = 0;
 131             while ( buf[rangeMarkerLocation-1] <= endRange ) {
 132                 LocalPointer<UnicodeString> newRegion(new UnicodeString(buf), status);
 133                 allRegions->addElement(newRegion.orphan(),status);
 134                 buf[rangeMarkerLocation-1]++;
 135             }
 136         } else {
 137             LocalPointer<UnicodeString> newRegion(new UnicodeString(regionName), status);
 138             allRegions->addElement(newRegion.orphan(),status);
 139         }
 140     }
 141
 142     while ( ures_hasNext(regionMacro.getAlias()) ) {
 143         UnicodeString regionName = ures_getNextUnicodeString(regionMacro.getAlias(),NULL,&status);
 144         int32_t rangeMarkerLocation = regionName.indexOf(RANGE_MARKER);
 145         UChar buf[6];
 146         regionName.extract(buf,6,status);
 147         if ( rangeMarkerLocation > 0 ) {
 148             UChar endRange = regionName.charAt(rangeMarkerLocation+1);
 149             buf[rangeMarkerLocation] = 0;
 150             while ( buf[rangeMarkerLocation-1] <= endRange ) {
 151                 LocalPointer<UnicodeString> newRegion(new UnicodeString(buf), status);
 152                 allRegions->addElement(newRegion.orphan(),status);
 153                 buf[rangeMarkerLocation-1]++;
 154             }
 155         } else {
 156             LocalPointer<UnicodeString> newRegion(new UnicodeString(regionName), status);
 157             allRegions->addElement(newRegion.orphan(),status);
 158         }
 159     }
 160
 161     while ( ures_hasNext(regionUnknown.getAlias()) ) {
 162         LocalPointer<UnicodeString> regionName (new UnicodeString(ures_getNextUnicodeString(regionUnknown.getAlias(),NULL,&status),status));
 163         allRegions->addElement(regionName.orphan(),status);
 164     }
 165
 166     while ( ures_hasNext(worldContainment.getAlias()) ) {
 167         UnicodeString *continentName = new UnicodeString(ures_getNextUnicodeString(worldContainment.getAlias(),NULL,&status));
 168         continents->addElement(continentName,status);
 169     }
 170
 171     UResourceBundle *groupingBundle = nullptr;
 172     while ( ures_hasNext(groupingContainment.getAlias()) ) {
 173         groupingBundle = ures_getNextResource(groupingContainment.getAlias(), groupingBundle, &status);
 174         if (U_FAILURE(status)) {
 175             break;
 176         }
 177         UnicodeString *groupingName = new UnicodeString(ures_getKey(groupingBundle), -1, US_INV);
 178         if (groupingName) {
 179             groupings->addElement(groupingName,status);
 180         }
 181     }
 182     ures_close(groupingBundle);
 183
 184     for ( int32_t i = 0 ; i < allRegions->size() ; i++ ) {
 185         LocalPointer<Region> r(new Region(), status);
 186         if ( U_FAILURE(status) ) {
 187            return;
 188         }
 189         UnicodeString *regionName = (UnicodeString *)allRegions->elementAt(i);
 190         r->idStr = *regionName;
 191
 192         r->idStr.extract(0,r->idStr.length(),r->id,sizeof(r->id),US_INV);
 193         r->fType = URGN_TERRITORY; // Only temporary - figure out the real type later once the aliases are known.
 194
 195         Formattable result;
 196         UErrorCode ps = U_ZERO_ERROR;
 197         df->parse(r->idStr,result,ps);
 198         if ( U_SUCCESS(ps) ) {
 199             r->code = result.getLong(); // Convert string to number
 200             uhash_iput(newNumericCodeMap.getAlias(),r->code,(void *)(r.getAlias()),&status);
 201             r->fType = URGN_SUBCONTINENT;
 202         } else {
 203             r->code = -1;
 204         }
 205         void* idStrAlias = (void*)&(r->idStr); // about to orphan 'r'. Save this off.
 206         uhash_put(newRegionIDMap.getAlias(),idStrAlias,(void *)(r.orphan()),&status); // regionIDMap takes ownership
 207     }
 208
 209     // Process the territory aliases
 210     while ( ures_hasNext(territoryAlias.getAlias()) ) {
 211         LocalUResourceBundlePointer res(ures_getNextResource(territoryAlias.getAlias(),NULL,&status));
 212         const char *aliasFrom = ures_getKey(res.getAlias());
 213         LocalPointer<UnicodeString> aliasFromStr(new UnicodeString(aliasFrom, -1, US_INV), status);
 214         UnicodeString aliasTo = ures_getUnicodeStringByKey(res.getAlias(),"replacement",&status);
 215         res.adoptInstead(NULL);
 216
 217         const Region *aliasToRegion = (Region *) uhash_get(newRegionIDMap.getAlias(),&aliasTo);
 218         Region *aliasFromRegion = (Region *)uhash_get(newRegionIDMap.getAlias(),aliasFromStr.getAlias());
 219
 220         if ( aliasToRegion != NULL && aliasFromRegion == NULL ) { // This is just an alias from some string to a region
 221             uhash_put(newRegionAliases.getAlias(),(void *)aliasFromStr.orphan(), (void *)aliasToRegion,&status);
 222         } else {
 223             if ( aliasFromRegion == NULL ) { // Deprecated region code not in the master codes list - so need to create a deprecated region for it.
 224                 LocalPointer<Region> newRgn(new Region, status);
 225                 if ( U_SUCCESS(status) ) {
 226                     aliasFromRegion = newRgn.orphan();
 227                 } else {
 228                     return; // error out
 229                 }
 230                 aliasFromRegion->idStr.setTo(*aliasFromStr);
 231                 aliasFromRegion->idStr.extract(0,aliasFromRegion->idStr.length(),aliasFromRegion->id,sizeof(aliasFromRegion->id),US_INV);
 232                 uhash_put(newRegionIDMap.getAlias(),(void *)&(aliasFromRegion->idStr),(void *)aliasFromRegion,&status);
 233                 Formattable result;
 234                 UErrorCode ps = U_ZERO_ERROR;
 235                 df->parse(aliasFromRegion->idStr,result,ps);
 236                 if ( U_SUCCESS(ps) ) {
 237                     aliasFromRegion->code = result.getLong(); // Convert string to number
 238                     uhash_iput(newNumericCodeMap.getAlias(),aliasFromRegion->code,(void *)aliasFromRegion,&status);
 239                 } else {
 240                     aliasFromRegion->code = -1;
 241                 }
 242                 aliasFromRegion->fType = URGN_DEPRECATED;
 243             } else {
 244                 aliasFromRegion->fType = URGN_DEPRECATED;
 245             }
 246
 247             {
 248                 LocalPointer<UVector> newPreferredValues(new UVector(uprv_deleteUObject, uhash_compareUnicodeString, status), status);
 249                 aliasFromRegion->preferredValues = newPreferredValues.orphan();
 250             }
 251             if( U_FAILURE(status)) {
 252                 return;
 253             }
 254             UnicodeString currentRegion;
 255             //currentRegion.remove();   TODO: was already 0 length?
 256             for (int32_t i = 0 ; i < aliasTo.length() ; i++ ) {
 257                 if ( aliasTo.charAt(i) != 0x0020 ) {
 258                     currentRegion.append(aliasTo.charAt(i));
 259                 }
 260                 if ( aliasTo.charAt(i) == 0x0020 || i+1 == aliasTo.length() ) {
 261                     Region *target = (Region *)uhash_get(newRegionIDMap.getAlias(),(void *)&currentRegion);
 262                     if (target) {
 263                         LocalPointer<UnicodeString> preferredValue(new UnicodeString(target->idStr), status);
 264                         aliasFromRegion->preferredValues->addElement((void *)preferredValue.orphan(),status);  // may add null if err
 265                     }
 266                     currentRegion.remove();
 267                 }
 268             }
 269         }
 270     }
 271
 272     // Process the code mappings - This will allow us to assign numeric codes to most of the territories.
 273     while ( ures_hasNext(codeMappings.getAlias()) ) {
 274         UResourceBundle *mapping = ures_getNextResource(codeMappings.getAlias(),NULL,&status);
 275         if ( ures_getType(mapping) == URES_ARRAY && ures_getSize(mapping) == 3) {
 276             UnicodeString codeMappingID = ures_getUnicodeStringByIndex(mapping,0,&status);
 277             UnicodeString codeMappingNumber = ures_getUnicodeStringByIndex(mapping,1,&status);
 278             UnicodeString codeMapping3Letter = ures_getUnicodeStringByIndex(mapping,2,&status);
 279
 280             Region *r = (Region *)uhash_get(newRegionIDMap.getAlias(),(void *)&codeMappingID);
 281             if ( r ) {
 282                 Formattable result;
 283                 UErrorCode ps = U_ZERO_ERROR;
 284                 df->parse(codeMappingNumber,result,ps);
 285                 if ( U_SUCCESS(ps) ) {
 286                     r->code = result.getLong(); // Convert string to number
 287                     uhash_iput(newNumericCodeMap.getAlias(),r->code,(void *)r,&status);
 288                 }
 289                 LocalPointer<UnicodeString> code3(new UnicodeString(codeMapping3Letter), status);
 290                 uhash_put(newRegionAliases.getAlias(),(void *)code3.orphan(), (void *)r,&status);
 291             }
 292         }
 293         ures_close(mapping);
 294     }
 295
 296     // Now fill in the special cases for WORLD, UNKNOWN, CONTINENTS, and GROUPINGS
 297     Region *r;
 298     UnicodeString WORLD_ID_STRING(WORLD_ID);
 299     r = (Region *) uhash_get(newRegionIDMap.getAlias(),(void *)&WORLD_ID_STRING);
 300     if ( r ) {
 301         r->fType = URGN_WORLD;
 302     }
 303
 304     UnicodeString UNKNOWN_REGION_ID_STRING(UNKNOWN_REGION_ID);
 305     r = (Region *) uhash_get(newRegionIDMap.getAlias(),(void *)&UNKNOWN_REGION_ID_STRING);
 306     if ( r ) {
 307         r->fType = URGN_UNKNOWN;
 308     }
 309
 310     for ( int32_t i = 0 ; i < continents->size() ; i++ ) {
 311         r = (Region *) uhash_get(newRegionIDMap.getAlias(),(void *)continents->elementAt(i));
 312         if ( r ) {
 313             r->fType = URGN_CONTINENT;
 314         }
 315     }
 316
 317     for ( int32_t i = 0 ; i < groupings->size() ; i++ ) {
 318         r = (Region *) uhash_get(newRegionIDMap.getAlias(),(void *)groupings->elementAt(i));
 319         if ( r ) {
 320             r->fType = URGN_GROUPING;
 321         }
 322     }
 323
 324     // Special case: The region code "QO" (Outlying Oceania) is a subcontinent code added by CLDR
 325     // even though it looks like a territory code.  Need to handle it here.
 326
 327     UnicodeString OUTLYING_OCEANIA_REGION_ID_STRING(OUTLYING_OCEANIA_REGION_ID);
 328     r = (Region *) uhash_get(newRegionIDMap.getAlias(),(void *)&OUTLYING_OCEANIA_REGION_ID_STRING);
 329     if ( r ) {
 330         r->fType = URGN_SUBCONTINENT;
 331     }
 332
 333     // Load territory containment info from the supplemental data.
 334     while ( ures_hasNext(territoryContainment.getAlias()) ) {
 335         LocalUResourceBundlePointer mapping(ures_getNextResource(territoryContainment.getAlias(),NULL,&status));
 336         if( U_FAILURE(status) ) {
 337             return;  // error out
 338         }
 339         const char *parent = ures_getKey(mapping.getAlias());
 340         if (uprv_strcmp(parent, "containedGroupings") == 0 || uprv_strcmp(parent, "deprecated") == 0) {
 341             continue; // handle new pseudo-parent types added in ICU data per cldrbug 7808; for now just skip.
 342             // #11232 is to do something useful with these.
 343         }
 344         UnicodeString parentStr = UnicodeString(parent, -1 , US_INV);
 345         Region *parentRegion = (Region *) uhash_get(newRegionIDMap.getAlias(),(void *)&parentStr);
 346
 347         for ( int j = 0 ; j < ures_getSize(mapping.getAlias()); j++ ) {
 348             UnicodeString child = ures_getUnicodeStringByIndex(mapping.getAlias(),j,&status);
 349             Region *childRegion = (Region *) uhash_get(newRegionIDMap.getAlias(),(void *)&child);
 350             if ( parentRegion != NULL && childRegion != NULL ) {
 351
 352                 // Add the child region to the set of regions contained by the parent
 353                 if (parentRegion->containedRegions == NULL) {
 354                     parentRegion->containedRegions = new UVector(uprv_deleteUObject, uhash_compareUnicodeString, status);
 355                 }
 356
 357                 LocalPointer<UnicodeString> childStr(new UnicodeString(), status);
 358                 if( U_FAILURE(status) ) {
 359                     return;  // error out
 360                 }
 361                 childStr->fastCopyFrom(childRegion->idStr);
 362                 parentRegion->containedRegions->addElement((void *)childStr.orphan(),status);
 363
 364                 // Set the parent region to be the containing region of the child.
 365                 // Regions of type GROUPING can't be set as the parent, since another region
 366                 // such as a SUBCONTINENT, CONTINENT, or WORLD must always be the parent.
 367                 if ( parentRegion->fType != URGN_GROUPING) {
 368                     childRegion->containingRegion = parentRegion;
 369                 }
 370             }
 371         }
 372     }
 373
 374     // Create the availableRegions lists
 375     int32_t pos = UHASH_FIRST;
 376     while ( const UHashElement* element = uhash_nextElement(newRegionIDMap.getAlias(),&pos)) {
 377         Region *ar = (Region *)element->value.pointer;
 378         if ( availableRegions[ar->fType] == NULL ) {
 379             LocalPointer<UVector> newAr(new UVector(uprv_deleteUObject, uhash_compareUnicodeString, status), status);
 380             availableRegions[ar->fType] = newAr.orphan();
 381         }
 382         LocalPointer<UnicodeString> arString(new UnicodeString(ar->idStr), status);
 383         if( U_FAILURE(status) ) {
 384             return;  // error out
 385         }
 386         availableRegions[ar->fType]->addElement((void *)arString.orphan(),status);
 387     }
 388
 389     ucln_i18n_registerCleanup(UCLN_I18N_REGION, region_cleanup);
 390     // copy hashtables
 391     numericCodeMap = newNumericCodeMap.orphan();
 392     regionIDMap = newRegionIDMap.orphan();
 393     regionAliases = newRegionAliases.orphan();
 394 }
 395
 396 void Region::cleanupRegionData() {
 397     for (int32_t i = 0 ; i < URGN_LIMIT ; i++ ) {
 398         if ( availableRegions[i] ) {
 399             delete availableRegions[i];
 400         }
 401     }
 402
 403     if (regionAliases) {
 404         uhash_close(regionAliases);
 405     }
 406
 407     if (numericCodeMap) {
 408         uhash_close(numericCodeMap);
 409     }
 410
 411     if (regionIDMap) {
 412         uhash_close(regionIDMap);
 413     }
 414     if (allRegions) {
 415         allRegions->removeAllElements(); // Don't need the temporary list anymore.
 416         delete allRegions;
 417         allRegions = NULL;
 418     }
 419
 420     regionAliases = numericCodeMap = regionIDMap = NULL;
 421
 422     gRegionDataInitOnce.reset();
 423 }
 424
 425 Region::Region ()
 426         : code(-1),
 427           fType(URGN_UNKNOWN),
 428           containingRegion(NULL),
 429           containedRegions(NULL),
 430           preferredValues(NULL) {
 431     id[0] = 0;
 432 }
 433
 434 Region::~Region () {
 435         if (containedRegions) {
 436             delete containedRegions;
 437         }
 438         if (preferredValues) {
 439             delete preferredValues;
 440         }
 441 }
 442
 443 /**
 444  * Returns true if the two regions are equal.
 445  * Per PMC, just use pointer compare, since we have at most one instance of each Region.
 446  */
 447 UBool
 448 Region::operator==(const Region &that) const {
 449     return (idStr == that.idStr);
 450 }
 451
 452 /**
 453  * Returns true if the two regions are NOT equal; that is, if operator ==() returns false.
 454  * Per PMC, just use pointer compare, since we have at most one instance of each Region.
 455  */
 456 UBool
 457 Region::operator!=(const Region &that) const {
 458         return (idStr != that.idStr);
 459 }
 460
 461 /**
 462  * Returns a pointer to a Region using the given region code.  The region code can be either 2-letter ISO code,
 463  * 3-letter ISO code,  UNM.49 numeric code, or other valid Unicode Region Code as defined by the LDML specification.
 464  * The identifier will be canonicalized internally using the supplemental metadata as defined in the CLDR.
 465  * If the region code is NULL or not recognized, the appropriate error code will be set ( U_ILLEGAL_ARGUMENT_ERROR )
 466  */
 467 const Region* U_EXPORT2
 468 Region::getInstance(const char *region_code, UErrorCode &status) {
 469
 470     umtx_initOnce(gRegionDataInitOnce, &loadRegionData, status);
 471     if (U_FAILURE(status)) {
 472         return NULL;
 473     }
 474
 475     if ( !region_code ) {
 476         status = U_ILLEGAL_ARGUMENT_ERROR;
 477         return NULL;
 478     }
 479
 480     UnicodeString regionCodeString = UnicodeString(region_code, -1, US_INV);
 481     Region *r = (Region *)uhash_get(regionIDMap,(void *)&regionCodeString);
 482
 483     if ( !r ) {
 484         r = (Region *)uhash_get(regionAliases,(void *)&regionCodeString);
 485     }
 486
 487     if ( !r ) { // Unknown region code
 488         status = U_ILLEGAL_ARGUMENT_ERROR;
 489         return NULL;
 490     }
 491
 492     if ( r->fType == URGN_DEPRECATED && r->preferredValues->size() == 1) {
 493         StringEnumeration *pv = r->getPreferredValues(status);
 494         pv->reset(status);
 495         const UnicodeString *ustr = pv->snext(status);
 496         r = (Region *)uhash_get(regionIDMap,(void *)ustr);
 497         delete pv;
 498     }
 499
 500     return r;
 501
 502 }
 503
 504 /**
 505  * Returns a pointer to a Region using the given numeric region code. If the numeric region code is not recognized,
 506  * the appropriate error code will be set ( U_ILLEGAL_ARGUMENT_ERROR ).
 507  */
 508 const Region* U_EXPORT2
 509 Region::getInstance (int32_t code, UErrorCode &status) {
 510
 511     umtx_initOnce(gRegionDataInitOnce, &loadRegionData, status);
 512     if (U_FAILURE(status)) {
 513         return NULL;
 514     }
 515
 516     Region *r = (Region *)uhash_iget(numericCodeMap,code);
 517
 518     if ( !r ) { // Just in case there's an alias that's numeric, try to find it.
 519         UnicodeString pat = UNICODE_STRING_SIMPLE("0");
 520         LocalPointer<DecimalFormat> df(new DecimalFormat(pat,status), status);
 521         if( U_FAILURE(status) ) {
 522             return NULL;
 523         }
 524         UnicodeString id;
 525         id.remove();
 526         FieldPosition posIter;
 527         df->format(code,id, posIter, status);
 528         r = (Region *)uhash_get(regionAliases,&id);
 529     }
 530
 531     if( U_FAILURE(status) ) {
 532         return NULL;
 533     }
 534
 535     if ( !r ) {
 536         status = U_ILLEGAL_ARGUMENT_ERROR;
 537         return NULL;
 538     }
 539
 540     if ( r->fType == URGN_DEPRECATED && r->preferredValues->size() == 1) {
 541         StringEnumeration *pv = r->getPreferredValues(status);
 542         pv->reset(status);
 543         const UnicodeString *ustr = pv->snext(status);
 544         r = (Region *)uhash_get(regionIDMap,(void *)ustr);
 545         delete pv;
 546     }
 547
 548     return r;
 549 }
 550
 551
 552 /**
 553  * Returns an enumeration over the IDs of all known regions that match the given type.
 554  */
 555 StringEnumeration* U_EXPORT2
 556 Region::getAvailable(URegionType type, UErrorCode &status) {
 557     umtx_initOnce(gRegionDataInitOnce, &loadRegionData, status); // returns immediately if U_FAILURE(status)
 558     if (U_FAILURE(status)) {
 559         return NULL;
 560     }
 561     return new RegionNameEnumeration(availableRegions[type],status);
 562 }
 563
 564 /**
 565  * Returns a pointer to the region that contains this region.  Returns NULL if this region is code "001" (World)
 566  * or "ZZ" (Unknown region). For example, calling this method with region "IT" (Italy) returns the
 567  * region "039" (Southern Europe).
 568  */
 569 const Region*
 570 Region::getContainingRegion() const {
 571     UErrorCode status = U_ZERO_ERROR;
 572     umtx_initOnce(gRegionDataInitOnce, &loadRegionData, status);
 573     return containingRegion;
 574 }
 575
 576 /**
 577  * Return a pointer to the region that geographically contains this region and matches the given type,
 578  * moving multiple steps up the containment chain if necessary.  Returns NULL if no containing region can be found
 579  * that matches the given type. Note: The URegionTypes = "URGN_GROUPING", "URGN_DEPRECATED", or "URGN_UNKNOWN"
 580  * are not appropriate for use in this API. NULL will be returned in this case. For example, calling this method
 581  * with region "IT" (Italy) for type "URGN_CONTINENT" returns the region "150" ( Europe ).
 582  */
 583 const Region*
 584 Region::getContainingRegion(URegionType type) const {
 585     UErrorCode status = U_ZERO_ERROR;
 586     umtx_initOnce(gRegionDataInitOnce, &loadRegionData, status);
 587     if ( containingRegion == NULL ) {
 588         return NULL;
 589     }
 590
 591     return ( containingRegion->fType == type)? containingRegion: containingRegion->getContainingRegion(type);
 592 }
 593
 594 /**
 595  * Return an enumeration over the IDs of all the regions that are immediate children of this region in the
 596  * region hierarchy. These returned regions could be either macro regions, territories, or a mixture of the two,
 597  * depending on the containment data as defined in CLDR.  This API may return NULL if this region doesn't have
 598  * any sub-regions. For example, calling this method with region "150" (Europe) returns an enumeration containing
 599  * the various sub regions of Europe - "039" (Southern Europe) - "151" (Eastern Europe) - "154" (Northern Europe)
 600  * and "155" (Western Europe).
 601  */
 602 StringEnumeration*
 603 Region::getContainedRegions(UErrorCode &status) const {
 604     umtx_initOnce(gRegionDataInitOnce, &loadRegionData, status); // returns immediately if U_FAILURE(status)
 605     if (U_FAILURE(status)) {
 606         return NULL;
 607     }
 608     return new RegionNameEnumeration(containedRegions,status);
 609 }
 610
 611 /**
 612  * Returns an enumeration over the IDs of all the regions that are children of this region anywhere in the region
 613  * hierarchy and match the given type.  This API may return an empty enumeration if this region doesn't have any
 614  * sub-regions that match the given type. For example, calling this method with region "150" (Europe) and type
 615  * "URGN_TERRITORY" returns a set containing all the territories in Europe ( "FR" (France) - "IT" (Italy) - "DE" (Germany) etc. )
 616  */
 617 StringEnumeration*
 618 Region::getContainedRegions( URegionType type, UErrorCode &status ) const {
 619     umtx_initOnce(gRegionDataInitOnce, &loadRegionData, status); // returns immediately if U_FAILURE(status)
 620     if (U_FAILURE(status)) {
 621         return NULL;
 622     }
 623
 624     UVector *result = new UVector(NULL, uhash_compareChars, status);
 625
 626     StringEnumeration *cr = getContainedRegions(status);
 627
 628     for ( int32_t i = 0 ; i < cr->count(status) ; i++ ) {
 629         const char *regionId = cr->next(NULL,status);
 630         const Region *r = Region::getInstance(regionId,status);
 631         if ( r->getType() == type) {
 632             result->addElement((void *)&r->idStr,status);
 633         } else {
 634             StringEnumeration *children = r->getContainedRegions(type, status);
 635             for ( int32_t j = 0 ; j < children->count(status) ; j++ ) {
 636                 const char *id2 = children->next(NULL,status);
 637                 const Region *r2 = Region::getInstance(id2,status);
 638                 result->addElement((void *)&r2->idStr,status);
 639             }
 640             delete children;
 641         }
 642     }
 643     delete cr;
 644     StringEnumeration* resultEnumeration = new RegionNameEnumeration(result,status);
 645     delete result;
 646     return resultEnumeration;
 647 }
 648
 649 /**
 650  * Returns true if this region contains the supplied other region anywhere in the region hierarchy.
 651  */
 652 UBool
 653 Region::contains(const Region &other) const {
 654     UErrorCode status = U_ZERO_ERROR;
 655     umtx_initOnce(gRegionDataInitOnce, &loadRegionData, status);
 656
 657     if (!containedRegions) {
 658           return FALSE;
 659     }
 660     if (containedRegions->contains((void *)&other.idStr)) {
 661         return TRUE;
 662     } else {
 663         for ( int32_t i = 0 ; i < containedRegions->size() ; i++ ) {
 664             UnicodeString *crStr = (UnicodeString *)containedRegions->elementAt(i);
 665             Region *cr = (Region *) uhash_get(regionIDMap,(void *)crStr);
 666             if ( cr && cr->contains(other) ) {
 667                 return TRUE;
 668             }
 669         }
 670     }
 671
 672     return FALSE;
 673 }
 674
 675 /**
 676  * For deprecated regions, return an enumeration over the IDs of the regions that are the preferred replacement
 677  * regions for this region.  Returns NULL for a non-deprecated region.  For example, calling this method with region
 678  * "SU" (Soviet Union) would return a list of the regions containing "RU" (Russia), "AM" (Armenia), "AZ" (Azerbaijan), etc...
 679  */
 680 StringEnumeration*
 681 Region::getPreferredValues(UErrorCode &status) const {
 682     umtx_initOnce(gRegionDataInitOnce, &loadRegionData, status); // returns immediately if U_FAILURE(status)
 683     if (U_FAILURE(status) || fType != URGN_DEPRECATED) {
 684         return NULL;
 685     }
 686     return new RegionNameEnumeration(preferredValues,status);
 687 }
 688
 689
 690 /**
 691  * Return this region's canonical region code.
 692  */
 693 const char*
 694 Region::getRegionCode() const {
 695     return id;
 696 }
 697
 698 int32_t
 699 Region::getNumericCode() const {
 700     return code;
 701 }
 702
 703 /**
 704  * Returns the region type of this region.
 705  */
 706 URegionType
 707 Region::getType() const {
 708     return fType;
 709 }
 710
 711 RegionNameEnumeration::RegionNameEnumeration(UVector *fNameList, UErrorCode& status) {
 712     pos=0;
 713     if (fNameList && U_SUCCESS(status)) {
 714         fRegionNames = new UVector(uprv_deleteUObject, uhash_compareUnicodeString, fNameList->size(),status);
 715         for ( int32_t i = 0 ; i < fNameList->size() ; i++ ) {
 716             UnicodeString* this_region_name = (UnicodeString *)fNameList->elementAt(i);
 717             UnicodeString* new_region_name = new UnicodeString(*this_region_name);
 718             fRegionNames->addElement((void *)new_region_name,status);
 719         }
 720     }
 721     else {
 722         fRegionNames = NULL;
 723     }
 724 }
 725
 726 const UnicodeString*
 727 RegionNameEnumeration::snext(UErrorCode& status) {
 728   if (U_FAILURE(status) || (fRegionNames==NULL)) {
 729     return NULL;
 730   }
 731   const UnicodeString* nextStr = (const UnicodeString *)fRegionNames->elementAt(pos);
 732   if (nextStr!=NULL) {
 733     pos++;
 734   }
 735   return nextStr;
 736 }
 737
 738 void
 739 RegionNameEnumeration::reset(UErrorCode& /*status*/) {
 740     pos=0;
 741 }
 742
 743 int32_t
 744 RegionNameEnumeration::count(UErrorCode& /*status*/) const {
 745     return (fRegionNames==NULL) ? 0 : fRegionNames->size();
 746 }
 747
 748 RegionNameEnumeration::~RegionNameEnumeration() {
 749     delete fRegionNames;
 750 }
 751
 752 U_NAMESPACE_END
 753
 754 #endif /* #if !UCONFIG_NO_FORMATTING */
 755
 756 //eof