xref: /aosp_15_r20/external/cronet/third_party/icu/source/i18n/zonemeta.cpp (revision 6777b5387eb2ff775bb5750e3f5d96f37fb7352b)
1 // © 2016 and later: Unicode, Inc. and others.
2 // License & terms of use: http://www.unicode.org/copyright.html
3 /*
4 *******************************************************************************
5 * Copyright (C) 2007-2014, 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 "zonemeta.h"
15 
16 #include "unicode/timezone.h"
17 #include "unicode/ustring.h"
18 #include "unicode/putil.h"
19 #include "unicode/simpletz.h"
20 #include "unicode/strenum.h"
21 #include "umutex.h"
22 #include "uvector.h"
23 #include "cmemory.h"
24 #include "gregoimp.h"
25 #include "cstring.h"
26 #include "ucln_in.h"
27 #include "uassert.h"
28 #include "uresimp.h"
29 #include "uhash.h"
30 #include "olsontz.h"
31 #include "uinvchar.h"
32 
33 static icu::UMutex gZoneMetaLock;
34 
35 // CLDR Canonical ID mapping table
36 static UHashtable *gCanonicalIDCache = nullptr;
37 static icu::UInitOnce gCanonicalIDCacheInitOnce {};
38 
39 // Metazone mapping table
40 static UHashtable *gOlsonToMeta = nullptr;
41 static icu::UInitOnce gOlsonToMetaInitOnce {};
42 
43 // Available metazone IDs vector and table
44 static icu::UVector *gMetaZoneIDs = nullptr;
45 static UHashtable *gMetaZoneIDTable = nullptr;
46 static icu::UInitOnce gMetaZoneIDsInitOnce {};
47 
48 // Country info vectors
49 static icu::UVector *gSingleZoneCountries = nullptr;
50 static icu::UVector *gMultiZonesCountries = nullptr;
51 static icu::UInitOnce gCountryInfoVectorsInitOnce {};
52 
53 U_CDECL_BEGIN
54 
55 /**
56  * Cleanup callback func
57  */
zoneMeta_cleanup()58 static UBool U_CALLCONV zoneMeta_cleanup()
59 {
60     if (gCanonicalIDCache != nullptr) {
61         uhash_close(gCanonicalIDCache);
62         gCanonicalIDCache = nullptr;
63     }
64     gCanonicalIDCacheInitOnce.reset();
65 
66     if (gOlsonToMeta != nullptr) {
67         uhash_close(gOlsonToMeta);
68         gOlsonToMeta = nullptr;
69     }
70     gOlsonToMetaInitOnce.reset();
71 
72     if (gMetaZoneIDTable != nullptr) {
73         uhash_close(gMetaZoneIDTable);
74         gMetaZoneIDTable = nullptr;
75     }
76     // delete after closing gMetaZoneIDTable, because it holds
77     // value objects held by the hashtable
78     delete gMetaZoneIDs;
79     gMetaZoneIDs = nullptr;
80     gMetaZoneIDsInitOnce.reset();
81 
82     delete gSingleZoneCountries;
83     gSingleZoneCountries = nullptr;
84     delete gMultiZonesCountries;
85     gMultiZonesCountries = nullptr;
86     gCountryInfoVectorsInitOnce.reset();
87 
88     return true;
89 }
90 
91 /**
92  * Deleter for char16_t* string
93  */
94 static void U_CALLCONV
deleteUCharString(void * obj)95 deleteUCharString(void *obj) {
96     char16_t *entry = (char16_t*)obj;
97     uprv_free(entry);
98 }
99 
100 /**
101  * Deleter for OlsonToMetaMappingEntry
102  */
103 static void U_CALLCONV
deleteOlsonToMetaMappingEntry(void * obj)104 deleteOlsonToMetaMappingEntry(void *obj) {
105     icu::OlsonToMetaMappingEntry *entry = (icu::OlsonToMetaMappingEntry*)obj;
106     delete entry;
107 }
108 
109 U_CDECL_END
110 
111 U_NAMESPACE_BEGIN
112 
113 #define ZID_KEY_MAX 128
114 
115 static const char gMetaZones[]          = "metaZones";
116 static const char gMetazoneInfo[]       = "metazoneInfo";
117 static const char gMapTimezonesTag[]    = "mapTimezones";
118 
119 static const char gKeyTypeData[]        = "keyTypeData";
120 static const char gTypeAliasTag[]       = "typeAlias";
121 static const char gTypeMapTag[]         = "typeMap";
122 static const char gTimezoneTag[]        = "timezone";
123 static const char gIanaMapTag[]         = "ianaMap";
124 
125 static const char gPrimaryZonesTag[]    = "primaryZones";
126 
127 static const char gWorldTag[]           = "001";
128 
129 static const char16_t gWorld[] = {0x30, 0x30, 0x31, 0x00}; // "001"
130 
131 static const char16_t gDefaultFrom[] = {0x31, 0x39, 0x37, 0x30, 0x2D, 0x30, 0x31, 0x2D, 0x30, 0x31,
132                                      0x20, 0x30, 0x30, 0x3A, 0x30, 0x30, 0x00}; // "1970-01-01 00:00"
133 static const char16_t gDefaultTo[]   = {0x39, 0x39, 0x39, 0x39, 0x2D, 0x31, 0x32, 0x2D, 0x33, 0x31,
134                                      0x20, 0x32, 0x33, 0x3A, 0x35, 0x39, 0x00}; // "9999-12-31 23:59"
135 
136 static const char16_t gCustomTzPrefix[]    = {0x47, 0x4D, 0x54, 0};    // "GMT"
137 
138 #define ASCII_DIGIT(c) (((c)>=0x30 && (c)<=0x39) ? (c)-0x30 : -1)
139 
140 /*
141  * Convert a date string used by metazone mappings to UDate.
142  * The format used by CLDR metazone mapping is "yyyy-MM-dd HH:mm".
143  */
144 static UDate
parseDate(const char16_t * text,UErrorCode & status)145 parseDate (const char16_t *text, UErrorCode &status) {
146     if (U_FAILURE(status)) {
147         return 0;
148     }
149     int32_t len = u_strlen(text);
150     if (len != 16 && len != 10) {
151         // It must be yyyy-MM-dd HH:mm (length 16) or yyyy-MM-dd (length 10)
152         status = U_INVALID_FORMAT_ERROR;
153         return 0;
154     }
155 
156     int32_t year = 0, month = 0, day = 0, hour = 0, min = 0, n;
157     int32_t idx;
158 
159     // "yyyy" (0 - 3)
160     for (idx = 0; idx <= 3 && U_SUCCESS(status); idx++) {
161         n = ASCII_DIGIT((int32_t)text[idx]);
162         if (n >= 0) {
163             year = 10*year + n;
164         } else {
165             status = U_INVALID_FORMAT_ERROR;
166         }
167     }
168     // "MM" (5 - 6)
169     for (idx = 5; idx <= 6 && U_SUCCESS(status); idx++) {
170         n = ASCII_DIGIT((int32_t)text[idx]);
171         if (n >= 0) {
172             month = 10*month + n;
173         } else {
174             status = U_INVALID_FORMAT_ERROR;
175         }
176     }
177     // "dd" (8 - 9)
178     for (idx = 8; idx <= 9 && U_SUCCESS(status); idx++) {
179         n = ASCII_DIGIT((int32_t)text[idx]);
180         if (n >= 0) {
181             day = 10*day + n;
182         } else {
183             status = U_INVALID_FORMAT_ERROR;
184         }
185     }
186     if (len == 16) {
187         // "HH" (11 - 12)
188         for (idx = 11; idx <= 12 && U_SUCCESS(status); idx++) {
189             n = ASCII_DIGIT((int32_t)text[idx]);
190             if (n >= 0) {
191                 hour = 10*hour + n;
192             } else {
193                 status = U_INVALID_FORMAT_ERROR;
194             }
195         }
196         // "mm" (14 - 15)
197         for (idx = 14; idx <= 15 && U_SUCCESS(status); idx++) {
198             n = ASCII_DIGIT((int32_t)text[idx]);
199             if (n >= 0) {
200                 min = 10*min + n;
201             } else {
202                 status = U_INVALID_FORMAT_ERROR;
203             }
204         }
205     }
206 
207     if (U_SUCCESS(status)) {
208         UDate date = Grego::fieldsToDay(year, month - 1, day) * U_MILLIS_PER_DAY
209             + hour * U_MILLIS_PER_HOUR + min * U_MILLIS_PER_MINUTE;
210         return date;
211     }
212     return 0;
213 }
214 
initCanonicalIDCache(UErrorCode & status)215 static void U_CALLCONV initCanonicalIDCache(UErrorCode &status) {
216     gCanonicalIDCache = uhash_open(uhash_hashUChars, uhash_compareUChars, nullptr, &status);
217     if (gCanonicalIDCache == nullptr) {
218         status = U_MEMORY_ALLOCATION_ERROR;
219     }
220     if (U_FAILURE(status)) {
221         gCanonicalIDCache = nullptr;
222     }
223     // No key/value deleters - keys/values are from a resource bundle
224     ucln_i18n_registerCleanup(UCLN_I18N_ZONEMETA, zoneMeta_cleanup);
225 }
226 
227 
228 const char16_t* U_EXPORT2
getCanonicalCLDRID(const UnicodeString & tzid,UErrorCode & status)229 ZoneMeta::getCanonicalCLDRID(const UnicodeString &tzid, UErrorCode& status) {
230     if (U_FAILURE(status)) {
231         return nullptr;
232     }
233 
234     if (tzid.isBogus() || tzid.length() > ZID_KEY_MAX) {
235         status = U_ILLEGAL_ARGUMENT_ERROR;
236         return nullptr;
237     }
238 
239     // Checking the cached results
240     umtx_initOnce(gCanonicalIDCacheInitOnce, &initCanonicalIDCache, status);
241     if (U_FAILURE(status)) {
242         return nullptr;
243     }
244 
245     const char16_t *canonicalID = nullptr;
246 
247     UErrorCode tmpStatus = U_ZERO_ERROR;
248     char16_t utzid[ZID_KEY_MAX + 1];
249     tzid.extract(utzid, ZID_KEY_MAX + 1, tmpStatus);
250     U_ASSERT(tmpStatus == U_ZERO_ERROR);    // we checked the length of tzid already
251 
252     if (!uprv_isInvariantUString(utzid, -1)) {
253         // All of known tz IDs are only containing ASCII invariant characters.
254         status = U_ILLEGAL_ARGUMENT_ERROR;
255         return nullptr;
256     }
257 
258     // Check if it was already cached
259     umtx_lock(&gZoneMetaLock);
260     {
261         canonicalID = (const char16_t *)uhash_get(gCanonicalIDCache, utzid);
262     }
263     umtx_unlock(&gZoneMetaLock);
264 
265     if (canonicalID != nullptr) {
266         return canonicalID;
267     }
268 
269     // If not, resolve CLDR canonical ID with resource data
270     UBool isInputCanonical = false;
271     char id[ZID_KEY_MAX + 1];
272     tzid.extract(0, 0x7fffffff, id, UPRV_LENGTHOF(id), US_INV);
273 
274     // replace '/' with ':'
275     char *p = id;
276     while (*p++) {
277         if (*p == '/') {
278             *p = ':';
279         }
280     }
281 
282     UResourceBundle *top = ures_openDirect(nullptr, gKeyTypeData, &tmpStatus);
283     UResourceBundle *rb = ures_getByKey(top, gTypeMapTag, nullptr, &tmpStatus);
284     ures_getByKey(rb, gTimezoneTag, rb, &tmpStatus);
285     ures_getByKey(rb, id, rb, &tmpStatus);
286     if (U_SUCCESS(tmpStatus)) {
287         // type entry (canonical) found
288         // the input is the canonical ID. resolve to const char16_t*
289         canonicalID = TimeZone::findID(tzid);
290         isInputCanonical = true;
291     }
292 
293     if (canonicalID == nullptr) {
294         // If a map element not found, then look for an alias
295         tmpStatus = U_ZERO_ERROR;
296         ures_getByKey(top, gTypeAliasTag, rb, &tmpStatus);
297         ures_getByKey(rb, gTimezoneTag, rb, &tmpStatus);
298         const char16_t *canonical = ures_getStringByKey(rb,id,nullptr,&tmpStatus);
299         if (U_SUCCESS(tmpStatus)) {
300             // canonical map found
301             canonicalID = canonical;
302         }
303 
304         if (canonicalID == nullptr) {
305             // Dereference the input ID using the tz data
306             const char16_t *derefer = TimeZone::dereferOlsonLink(tzid);
307             if (derefer == nullptr) {
308                 status = U_ILLEGAL_ARGUMENT_ERROR;
309             } else {
310                 int32_t len = u_strlen(derefer);
311                 u_UCharsToChars(derefer,id,len);
312                 id[len] = (char) 0; // Make sure it is null terminated.
313 
314                 // replace '/' with ':'
315                 char *q = id;
316                 while (*q++) {
317                     if (*q == '/') {
318                         *q = ':';
319                     }
320                 }
321 
322                 // If a dereference turned something up then look for an alias.
323                 // rb still points to the alias table, so we don't have to go looking
324                 // for it.
325                 tmpStatus = U_ZERO_ERROR;
326                 canonical = ures_getStringByKey(rb,id,nullptr,&tmpStatus);
327                 if (U_SUCCESS(tmpStatus)) {
328                     // canonical map for the dereferenced ID found
329                     canonicalID = canonical;
330                 } else {
331                     canonicalID = derefer;
332                     isInputCanonical = true;
333                 }
334             }
335         }
336     }
337     ures_close(rb);
338     ures_close(top);
339 
340     if (U_SUCCESS(status)) {
341         U_ASSERT(canonicalID != nullptr);  // canocanilD must be non-nullptr here
342 
343         // Put the resolved canonical ID to the cache
344         umtx_lock(&gZoneMetaLock);
345         {
346             const char16_t* idInCache = (const char16_t *)uhash_get(gCanonicalIDCache, utzid);
347             if (idInCache == nullptr) {
348                 const char16_t* key = ZoneMeta::findTimeZoneID(tzid);
349                 U_ASSERT(key != nullptr);
350                 if (key != nullptr) {
351                     idInCache = (const char16_t *)uhash_put(gCanonicalIDCache, (void *)key, (void *)canonicalID, &status);
352                     U_ASSERT(idInCache == nullptr);
353                 }
354             }
355             if (U_SUCCESS(status) && isInputCanonical) {
356                 // Also put canonical ID itself into the cache if not exist
357                 const char16_t *canonicalInCache = (const char16_t*)uhash_get(gCanonicalIDCache, canonicalID);
358                 if (canonicalInCache == nullptr) {
359                     canonicalInCache = (const char16_t *)uhash_put(gCanonicalIDCache, (void *)canonicalID, (void *)canonicalID, &status);
360                     U_ASSERT(canonicalInCache == nullptr);
361                 }
362             }
363         }
364         umtx_unlock(&gZoneMetaLock);
365     }
366 
367     return canonicalID;
368 }
369 
370 UnicodeString& U_EXPORT2
getCanonicalCLDRID(const UnicodeString & tzid,UnicodeString & systemID,UErrorCode & status)371 ZoneMeta::getCanonicalCLDRID(const UnicodeString &tzid, UnicodeString &systemID, UErrorCode& status) {
372     const char16_t *canonicalID = getCanonicalCLDRID(tzid, status);
373     if (U_FAILURE(status) || canonicalID == nullptr) {
374         systemID.setToBogus();
375         return systemID;
376     }
377     systemID.setTo(true, canonicalID, -1);
378     return systemID;
379 }
380 
381 const char16_t* U_EXPORT2
getCanonicalCLDRID(const TimeZone & tz)382 ZoneMeta::getCanonicalCLDRID(const TimeZone& tz) {
383     if (dynamic_cast<const OlsonTimeZone *>(&tz) != nullptr) {
384         // short cut for OlsonTimeZone
385         const OlsonTimeZone *otz = (const OlsonTimeZone*)&tz;
386         return otz->getCanonicalID();
387     }
388     UErrorCode status = U_ZERO_ERROR;
389     UnicodeString tzID;
390     return getCanonicalCLDRID(tz.getID(tzID), status);
391 }
392 
393 UnicodeString& U_EXPORT2
getIanaID(const UnicodeString & tzid,UnicodeString & ianaID,UErrorCode & status)394 ZoneMeta::getIanaID(const UnicodeString& tzid, UnicodeString& ianaID, UErrorCode& status) {
395     // First, get CLDR canonical ID
396     const char16_t *canonicalID = getCanonicalCLDRID(tzid, status);
397     if (U_FAILURE(status) || canonicalID == nullptr) {
398         ianaID.setToBogus();
399         return ianaID;
400     }
401     // Find IANA mapping if any.
402     UErrorCode tmpStatus = U_ZERO_ERROR;
403     UnicodeString tmpKey(canonicalID);
404     tmpKey.findAndReplace(UnicodeString("/"), UnicodeString(":"));
405     char keyBuf[ZID_KEY_MAX + 1];
406     /* int32_t keyLen = */ tmpKey.extract(0, tmpKey.length(), keyBuf, sizeof(keyBuf), US_INV);
407 
408     StackUResourceBundle r;
409     ures_openDirectFillIn(r.getAlias(), nullptr, gKeyTypeData, &tmpStatus);
410     ures_getByKey(r.getAlias(), gIanaMapTag, r.getAlias(), &tmpStatus);
411     ures_getByKey(r.getAlias(), gTimezoneTag, r.getAlias(), &tmpStatus);
412     int32_t tmpLen = 0;
413     const char16_t* tmpIana = ures_getStringByKey(r.getAlias(), keyBuf, &tmpLen, &tmpStatus);
414     if (U_SUCCESS(tmpStatus)) {
415         ianaID.setTo(true, tmpIana, -1);
416     } else {
417         ianaID.setTo(true, canonicalID, -1);
418     }
419     return ianaID;
420 }
421 
countryInfoVectorsInit(UErrorCode & status)422 static void U_CALLCONV countryInfoVectorsInit(UErrorCode &status) {
423     // Create empty vectors
424     // No deleters for these UVectors, it's a reference to a resource bundle string.
425     gSingleZoneCountries = new UVector(nullptr, uhash_compareUChars, status);
426     if (gSingleZoneCountries == nullptr) {
427         status = U_MEMORY_ALLOCATION_ERROR;
428     }
429     gMultiZonesCountries = new UVector(nullptr, uhash_compareUChars, status);
430     if (gMultiZonesCountries == nullptr) {
431         status = U_MEMORY_ALLOCATION_ERROR;
432     }
433 
434     if (U_FAILURE(status)) {
435         delete gSingleZoneCountries;
436         delete gMultiZonesCountries;
437         gSingleZoneCountries = nullptr;
438         gMultiZonesCountries  = nullptr;
439     }
440     ucln_i18n_registerCleanup(UCLN_I18N_ZONEMETA, zoneMeta_cleanup);
441 }
442 
443 
444 UnicodeString& U_EXPORT2
getCanonicalCountry(const UnicodeString & tzid,UnicodeString & country,UBool * isPrimary)445 ZoneMeta::getCanonicalCountry(const UnicodeString &tzid, UnicodeString &country, UBool *isPrimary /* = nullptr */) {
446     if (isPrimary != nullptr) {
447         *isPrimary = false;
448     }
449 
450     const char16_t *region = TimeZone::getRegion(tzid);
451     if (region != nullptr && u_strcmp(gWorld, region) != 0) {
452         country.setTo(region, -1);
453     } else {
454         country.setToBogus();
455         return country;
456     }
457 
458     if (isPrimary != nullptr) {
459         char regionBuf[] = {0, 0, 0};
460 
461         // Checking the cached results
462         UErrorCode status = U_ZERO_ERROR;
463         umtx_initOnce(gCountryInfoVectorsInitOnce, &countryInfoVectorsInit, status);
464         if (U_FAILURE(status)) {
465             return country;
466         }
467 
468         // Check if it was already cached
469         UBool cached = false;
470         UBool singleZone = false;
471         umtx_lock(&gZoneMetaLock);
472         {
473             singleZone = cached = gSingleZoneCountries->contains((void*)region);
474             if (!cached) {
475                 cached = gMultiZonesCountries->contains((void*)region);
476             }
477         }
478         umtx_unlock(&gZoneMetaLock);
479 
480         if (!cached) {
481             // We need to go through all zones associated with the region.
482             // This is relatively heavy operation.
483 
484             U_ASSERT(u_strlen(region) == 2);
485 
486             u_UCharsToChars(region, regionBuf, 2);
487 
488             StringEnumeration *ids = TimeZone::createTimeZoneIDEnumeration(UCAL_ZONE_TYPE_CANONICAL_LOCATION, regionBuf, nullptr, status);
489             int32_t idsLen = ids->count(status);
490             if (U_SUCCESS(status) && idsLen == 1) {
491                 // only the single zone is available for the region
492                 singleZone = true;
493             }
494             delete ids;
495 
496             // Cache the result
497             umtx_lock(&gZoneMetaLock);
498             {
499                 UErrorCode ec = U_ZERO_ERROR;
500                 if (singleZone) {
501                     if (!gSingleZoneCountries->contains((void*)region)) {
502                         gSingleZoneCountries->addElement((void*)region, ec);
503                     }
504                 } else {
505                     if (!gMultiZonesCountries->contains((void*)region)) {
506                         gMultiZonesCountries->addElement((void*)region, ec);
507                     }
508                 }
509             }
510             umtx_unlock(&gZoneMetaLock);
511         }
512 
513         if (singleZone) {
514             *isPrimary = true;
515         } else {
516             // Note: We may cache the primary zone map in future.
517 
518             // Even a country has multiple zones, one of them might be
519             // dominant and treated as a primary zone
520             int32_t idLen = 0;
521             if (regionBuf[0] == 0) {
522                 u_UCharsToChars(region, regionBuf, 2);
523             }
524 
525             UResourceBundle *rb = ures_openDirect(nullptr, gMetaZones, &status);
526             ures_getByKey(rb, gPrimaryZonesTag, rb, &status);
527             const char16_t *primaryZone = ures_getStringByKey(rb, regionBuf, &idLen, &status);
528             if (U_SUCCESS(status)) {
529                 if (tzid.compare(primaryZone, idLen) == 0) {
530                     *isPrimary = true;
531                 } else {
532                     // The given ID might not be a canonical ID
533                     UnicodeString canonicalID;
534                     TimeZone::getCanonicalID(tzid, canonicalID, status);
535                     if (U_SUCCESS(status) && canonicalID.compare(primaryZone, idLen) == 0) {
536                         *isPrimary = true;
537                     }
538                 }
539             }
540             ures_close(rb);
541         }
542     }
543 
544     return country;
545 }
546 
547 UnicodeString& U_EXPORT2
getMetazoneID(const UnicodeString & tzid,UDate date,UnicodeString & result)548 ZoneMeta::getMetazoneID(const UnicodeString &tzid, UDate date, UnicodeString &result) {
549     UBool isSet = false;
550     const UVector *mappings = getMetazoneMappings(tzid);
551     if (mappings != nullptr) {
552         for (int32_t i = 0; i < mappings->size(); i++) {
553             OlsonToMetaMappingEntry *mzm = (OlsonToMetaMappingEntry*)mappings->elementAt(i);
554             if (mzm->from <= date && mzm->to > date) {
555                 result.setTo(mzm->mzid, -1);
556                 isSet = true;
557                 break;
558             }
559         }
560     }
561     if (!isSet) {
562         result.setToBogus();
563     }
564     return result;
565 }
566 
olsonToMetaInit(UErrorCode & status)567 static void U_CALLCONV olsonToMetaInit(UErrorCode &status) {
568     U_ASSERT(gOlsonToMeta == nullptr);
569     ucln_i18n_registerCleanup(UCLN_I18N_ZONEMETA, zoneMeta_cleanup);
570     gOlsonToMeta = uhash_open(uhash_hashUChars, uhash_compareUChars, nullptr, &status);
571     if (U_FAILURE(status)) {
572         gOlsonToMeta = nullptr;
573     } else {
574         uhash_setKeyDeleter(gOlsonToMeta, deleteUCharString);
575         uhash_setValueDeleter(gOlsonToMeta, uprv_deleteUObject);
576     }
577 }
578 
579 
580 const UVector* U_EXPORT2
getMetazoneMappings(const UnicodeString & tzid)581 ZoneMeta::getMetazoneMappings(const UnicodeString &tzid) {
582     UErrorCode status = U_ZERO_ERROR;
583     char16_t tzidUChars[ZID_KEY_MAX + 1];
584     tzid.extract(tzidUChars, ZID_KEY_MAX + 1, status);
585     if (U_FAILURE(status) || status == U_STRING_NOT_TERMINATED_WARNING) {
586         return nullptr;
587     }
588 
589     umtx_initOnce(gOlsonToMetaInitOnce, &olsonToMetaInit, status);
590     if (U_FAILURE(status)) {
591         return nullptr;
592     }
593 
594     // get the mapping from cache
595     const UVector *result = nullptr;
596 
597     umtx_lock(&gZoneMetaLock);
598     {
599         result = (UVector*) uhash_get(gOlsonToMeta, tzidUChars);
600     }
601     umtx_unlock(&gZoneMetaLock);
602 
603     if (result != nullptr) {
604         return result;
605     }
606 
607     // miss the cache - create new one
608     UVector *tmpResult = createMetazoneMappings(tzid);
609     if (tmpResult == nullptr) {
610         // not available
611         return nullptr;
612     }
613 
614     // put the new one into the cache
615     umtx_lock(&gZoneMetaLock);
616     {
617         // make sure it's already created
618         result = (UVector*) uhash_get(gOlsonToMeta, tzidUChars);
619         if (result == nullptr) {
620             // add the one just created
621             int32_t tzidLen = tzid.length() + 1;
622             char16_t *key = (char16_t*)uprv_malloc(tzidLen * sizeof(char16_t));
623             if (key == nullptr) {
624                 // memory allocation error..  just return nullptr
625                 result = nullptr;
626                 delete tmpResult;
627             } else {
628                 tzid.extract(key, tzidLen, status);
629                 uhash_put(gOlsonToMeta, key, tmpResult, &status);
630                 if (U_FAILURE(status)) {
631                     // delete the mapping
632                     result = nullptr;
633                     delete tmpResult;
634                 } else {
635                     result = tmpResult;
636                 }
637             }
638         } else {
639             // another thread already put the one
640             delete tmpResult;
641         }
642     }
643     umtx_unlock(&gZoneMetaLock);
644 
645     return result;
646 }
647 
648 UVector*
createMetazoneMappings(const UnicodeString & tzid)649 ZoneMeta::createMetazoneMappings(const UnicodeString &tzid) {
650     LocalPointer <UVector> mzMappings;
651     UErrorCode status = U_ZERO_ERROR;
652 
653     UnicodeString canonicalID;
654     UResourceBundle *rb = ures_openDirect(nullptr, gMetaZones, &status);
655     ures_getByKey(rb, gMetazoneInfo, rb, &status);
656     getCanonicalCLDRID(tzid, canonicalID, status);
657 
658     if (U_SUCCESS(status)) {
659         char tzKey[ZID_KEY_MAX + 1];
660         int32_t tzKeyLen = canonicalID.extract(0, canonicalID.length(), tzKey, sizeof(tzKey), US_INV);
661         tzKey[tzKeyLen] = 0;
662 
663         // tzid keys are using ':' as separators
664         char *p = tzKey;
665         while (*p) {
666             if (*p == '/') {
667                 *p = ':';
668             }
669             p++;
670         }
671 
672         ures_getByKey(rb, tzKey, rb, &status);
673 
674         if (U_SUCCESS(status)) {
675             UResourceBundle *mz = nullptr;
676             while (ures_hasNext(rb)) {
677                 mz = ures_getNextResource(rb, mz, &status);
678 
679                 const char16_t *mz_name = ures_getStringByIndex(mz, 0, nullptr, &status);
680                 const char16_t *mz_from = gDefaultFrom;
681                 const char16_t *mz_to = gDefaultTo;
682 
683                 if (ures_getSize(mz) == 3) {
684                     mz_from = ures_getStringByIndex(mz, 1, nullptr, &status);
685                     mz_to   = ures_getStringByIndex(mz, 2, nullptr, &status);
686                 }
687 
688                 if(U_FAILURE(status)){
689                     status = U_ZERO_ERROR;
690                     continue;
691                 }
692                 // We do not want to use SimpleDateformat to parse boundary dates,
693                 // because this code could be triggered by the initialization code
694                 // used by SimpleDateFormat.
695                 UDate from = parseDate(mz_from, status);
696                 UDate to = parseDate(mz_to, status);
697                 if (U_FAILURE(status)) {
698                     status = U_ZERO_ERROR;
699                     continue;
700                 }
701 
702                 LocalPointer<OlsonToMetaMappingEntry> entry(new OlsonToMetaMappingEntry, status);
703                 if (U_FAILURE(status)) {
704                     break;
705                 }
706                 entry->mzid = mz_name;
707                 entry->from = from;
708                 entry->to = to;
709 
710                 if (mzMappings.isNull()) {
711                     mzMappings.adoptInsteadAndCheckErrorCode(
712                         new UVector(deleteOlsonToMetaMappingEntry, nullptr, status), status);
713                     if (U_FAILURE(status)) {
714                         break;
715                     }
716                 }
717 
718                 mzMappings->adoptElement(entry.orphan(), status);
719                 if (U_FAILURE(status)) {
720                     break;
721                 }
722             }
723             ures_close(mz);
724         }
725     }
726     ures_close(rb);
727     return U_SUCCESS(status) ? mzMappings.orphan() : nullptr;
728 }
729 
730 UnicodeString& U_EXPORT2
getZoneIdByMetazone(const UnicodeString & mzid,const UnicodeString & region,UnicodeString & result)731 ZoneMeta::getZoneIdByMetazone(const UnicodeString &mzid, const UnicodeString &region, UnicodeString &result) {
732     UErrorCode status = U_ZERO_ERROR;
733     const char16_t *tzid = nullptr;
734     int32_t tzidLen = 0;
735     char keyBuf[ZID_KEY_MAX + 1];
736     int32_t keyLen = 0;
737 
738     if (mzid.isBogus() || mzid.length() > ZID_KEY_MAX) {
739         result.setToBogus();
740         return result;
741     }
742 
743     keyLen = mzid.extract(0, mzid.length(), keyBuf, ZID_KEY_MAX + 1, US_INV);
744     keyBuf[keyLen] = 0;
745 
746     UResourceBundle *rb = ures_openDirect(nullptr, gMetaZones, &status);
747     ures_getByKey(rb, gMapTimezonesTag, rb, &status);
748     ures_getByKey(rb, keyBuf, rb, &status);
749 
750     if (U_SUCCESS(status)) {
751         // check region mapping
752         if (region.length() == 2 || region.length() == 3) {
753             keyLen = region.extract(0, region.length(), keyBuf, ZID_KEY_MAX + 1, US_INV);
754             keyBuf[keyLen] = 0;
755             tzid = ures_getStringByKey(rb, keyBuf, &tzidLen, &status);
756             if (status == U_MISSING_RESOURCE_ERROR) {
757                 status = U_ZERO_ERROR;
758             }
759         }
760         if (U_SUCCESS(status) && tzid == nullptr) {
761             // try "001"
762             tzid = ures_getStringByKey(rb, gWorldTag, &tzidLen, &status);
763         }
764     }
765     ures_close(rb);
766 
767     if (tzid == nullptr) {
768         result.setToBogus();
769     } else {
770         result.setTo(tzid, tzidLen);
771     }
772 
773     return result;
774 }
775 
initAvailableMetaZoneIDs()776 static void U_CALLCONV initAvailableMetaZoneIDs () {
777     U_ASSERT(gMetaZoneIDs == nullptr);
778     U_ASSERT(gMetaZoneIDTable == nullptr);
779     ucln_i18n_registerCleanup(UCLN_I18N_ZONEMETA, zoneMeta_cleanup);
780 
781     UErrorCode status = U_ZERO_ERROR;
782     gMetaZoneIDTable = uhash_open(uhash_hashUnicodeString, uhash_compareUnicodeString, nullptr, &status);
783     if (U_FAILURE(status) || gMetaZoneIDTable == nullptr) {
784         gMetaZoneIDTable = nullptr;
785         return;
786     }
787     uhash_setKeyDeleter(gMetaZoneIDTable, uprv_deleteUObject);
788     // No valueDeleter, because the vector maintain the value objects
789     gMetaZoneIDs = new UVector(nullptr, uhash_compareUChars, status);
790     if (U_FAILURE(status) || gMetaZoneIDs == nullptr) {
791         delete gMetaZoneIDs;
792         gMetaZoneIDs = nullptr;
793         uhash_close(gMetaZoneIDTable);
794         gMetaZoneIDTable = nullptr;
795         return;
796     }
797     gMetaZoneIDs->setDeleter(uprv_free);
798 
799     UResourceBundle *rb = ures_openDirect(nullptr, gMetaZones, &status);
800     UResourceBundle *bundle = ures_getByKey(rb, gMapTimezonesTag, nullptr, &status);
801     StackUResourceBundle res;
802     while (U_SUCCESS(status) && ures_hasNext(bundle)) {
803         ures_getNextResource(bundle, res.getAlias(), &status);
804         if (U_FAILURE(status)) {
805             break;
806         }
807         const char *mzID = ures_getKey(res.getAlias());
808         int32_t len = static_cast<int32_t>(uprv_strlen(mzID));
809         LocalMemory<char16_t> uMzID((char16_t*)uprv_malloc(sizeof(char16_t) * (len + 1)));
810         if (uMzID.isNull()) {
811             status = U_MEMORY_ALLOCATION_ERROR;
812             break;
813         }
814         u_charsToUChars(mzID, uMzID.getAlias(), len);
815         uMzID[len] = 0;
816         LocalPointer<UnicodeString> usMzID(new UnicodeString(uMzID.getAlias()), status);
817         if (U_FAILURE(status)) {
818             break;
819         }
820         if (uhash_get(gMetaZoneIDTable, usMzID.getAlias()) == nullptr) {
821             // Note: gMetaZoneIDTable adopts its keys, but not its values.
822             //       gMetaZoneIDs adopts its values.
823             uhash_put(gMetaZoneIDTable, usMzID.orphan(), uMzID.getAlias(), &status);
824             gMetaZoneIDs->adoptElement(uMzID.orphan(), status);
825         }
826     }
827     ures_close(bundle);
828     ures_close(rb);
829 
830     if (U_FAILURE(status)) {
831         uhash_close(gMetaZoneIDTable);
832         delete gMetaZoneIDs;
833         gMetaZoneIDTable = nullptr;
834         gMetaZoneIDs = nullptr;
835     }
836 }
837 
838 const UVector*
getAvailableMetazoneIDs()839 ZoneMeta::getAvailableMetazoneIDs() {
840     umtx_initOnce(gMetaZoneIDsInitOnce, &initAvailableMetaZoneIDs);
841     return gMetaZoneIDs;
842 }
843 
844 const char16_t*
findMetaZoneID(const UnicodeString & mzid)845 ZoneMeta::findMetaZoneID(const UnicodeString& mzid) {
846     umtx_initOnce(gMetaZoneIDsInitOnce, &initAvailableMetaZoneIDs);
847     if (gMetaZoneIDTable == nullptr) {
848         return nullptr;
849     }
850     return (const char16_t*)uhash_get(gMetaZoneIDTable, &mzid);
851 }
852 
853 const char16_t*
findTimeZoneID(const UnicodeString & tzid)854 ZoneMeta::findTimeZoneID(const UnicodeString& tzid) {
855     return TimeZone::findID(tzid);
856 }
857 
858 
859 TimeZone*
createCustomTimeZone(int32_t offset)860 ZoneMeta::createCustomTimeZone(int32_t offset) {
861     UBool negative = false;
862     int32_t tmp = offset;
863     if (offset < 0) {
864         negative = true;
865         tmp = -offset;
866     }
867     uint8_t hour, min, sec;
868 
869     tmp /= 1000;
870     sec = static_cast<uint8_t>(tmp % 60);
871     tmp /= 60;
872     min = static_cast<uint8_t>(tmp % 60);
873     hour = static_cast<uint8_t>(tmp / 60);
874 
875     UnicodeString zid;
876     formatCustomID(hour, min, sec, negative, zid);
877     return new SimpleTimeZone(offset, zid);
878 }
879 
880 UnicodeString&
formatCustomID(uint8_t hour,uint8_t min,uint8_t sec,UBool negative,UnicodeString & id)881 ZoneMeta::formatCustomID(uint8_t hour, uint8_t min, uint8_t sec, UBool negative, UnicodeString& id) {
882     // Create normalized time zone ID - GMT[+|-]HH:mm[:ss]
883     id.setTo(gCustomTzPrefix, -1);
884     if (hour != 0 || min != 0) {
885         if (negative) {
886           id.append((char16_t)0x2D);    // '-'
887         } else {
888           id.append((char16_t)0x2B);    // '+'
889         }
890         // Always use US-ASCII digits
891         id.append((char16_t)(0x30 + (hour%100)/10));
892         id.append((char16_t)(0x30 + (hour%10)));
893         id.append((char16_t)0x3A);    // ':'
894         id.append((char16_t)(0x30 + (min%100)/10));
895         id.append((char16_t)(0x30 + (min%10)));
896         if (sec != 0) {
897           id.append((char16_t)0x3A);    // ':'
898           id.append((char16_t)(0x30 + (sec%100)/10));
899           id.append((char16_t)(0x30 + (sec%10)));
900         }
901     }
902     return id;
903 }
904 
905 const char16_t*
getShortID(const TimeZone & tz)906 ZoneMeta::getShortID(const TimeZone& tz) {
907     const char16_t* canonicalID = nullptr;
908     if (dynamic_cast<const OlsonTimeZone *>(&tz) != nullptr) {
909         // short cut for OlsonTimeZone
910         const OlsonTimeZone *otz = (const OlsonTimeZone*)&tz;
911         canonicalID = otz->getCanonicalID();
912     }
913     if (canonicalID == nullptr) {
914         return nullptr;
915     }
916     return getShortIDFromCanonical(canonicalID);
917 }
918 
919 const char16_t*
getShortID(const UnicodeString & id)920 ZoneMeta::getShortID(const UnicodeString& id) {
921     UErrorCode status = U_ZERO_ERROR;
922     const char16_t* canonicalID = ZoneMeta::getCanonicalCLDRID(id, status);
923     if (U_FAILURE(status) || canonicalID == nullptr) {
924         return nullptr;
925     }
926     return ZoneMeta::getShortIDFromCanonical(canonicalID);
927 }
928 
929 const char16_t*
getShortIDFromCanonical(const char16_t * canonicalID)930 ZoneMeta::getShortIDFromCanonical(const char16_t* canonicalID) {
931     const char16_t* shortID = nullptr;
932     int32_t len = u_strlen(canonicalID);
933     char tzidKey[ZID_KEY_MAX + 1];
934 
935     u_UCharsToChars(canonicalID, tzidKey, len);
936     tzidKey[len] = (char) 0; // Make sure it is null terminated.
937 
938     // replace '/' with ':'
939     char *p = tzidKey;
940     while (*p++) {
941         if (*p == '/') {
942             *p = ':';
943         }
944     }
945 
946     UErrorCode status = U_ZERO_ERROR;
947     UResourceBundle *rb = ures_openDirect(nullptr, gKeyTypeData, &status);
948     ures_getByKey(rb, gTypeMapTag, rb, &status);
949     ures_getByKey(rb, gTimezoneTag, rb, &status);
950     shortID = ures_getStringByKey(rb, tzidKey, nullptr, &status);
951     ures_close(rb);
952 
953     return shortID;
954 }
955 
956 U_NAMESPACE_END
957 
958 #endif /* #if !UCONFIG_NO_FORMATTING */
959