• Home
  • Line#
  • Scopes#
  • Navigate#
  • Raw
  • Download
1 // © 2016 and later: Unicode, Inc. and others.
2 // License & terms of use: http://www.unicode.org/copyright.html
3 /*
4  **********************************************************************
5  *   Copyright (C) 1997-2016, International Business Machines
6  *   Corporation and others.  All Rights Reserved.
7  **********************************************************************
8 *
9 * File locid.cpp
10 *
11 * Created by: Richard Gillam
12 *
13 * Modification History:
14 *
15 *   Date        Name        Description
16 *   02/11/97    aliu        Changed gLocPath to fgDataDirectory and added
17 *                           methods to get and set it.
18 *   04/02/97    aliu        Made operator!= inline; fixed return value
19 *                           of getName().
20 *   04/15/97    aliu        Cleanup for AIX/Win32.
21 *   04/24/97    aliu        Numerous changes per code review.
22 *   08/18/98    stephen     Changed getDisplayName()
23 *                           Added SIMPLIFIED_CHINESE, TRADITIONAL_CHINESE
24 *                           Added getISOCountries(), getISOLanguages(),
25 *                           getLanguagesForCountry()
26 *   03/16/99    bertrand    rehaul.
27 *   07/21/99    stephen     Added U_CFUNC setDefault
28 *   11/09/99    weiv        Added const char * getName() const;
29 *   04/12/00    srl         removing unicodestring api's and cached hash code
30 *   08/10/01    grhoten     Change the static Locales to accessor functions
31 ******************************************************************************
32 */
33 
34 #include <utility>
35 
36 #include "unicode/bytestream.h"
37 #include "unicode/locid.h"
38 #include "unicode/strenum.h"
39 #include "unicode/stringpiece.h"
40 #include "unicode/uloc.h"
41 #include "putilimp.h"
42 #include "mutex.h"
43 #include "umutex.h"
44 #include "uassert.h"
45 #include "cmemory.h"
46 #include "cstring.h"
47 #include "uassert.h"
48 #include "uhash.h"
49 #include "ulocimp.h"
50 #include "ucln_cmn.h"
51 #include "ustr_imp.h"
52 #include "charstr.h"
53 #include "bytesinkutil.h"
54 
55 U_CDECL_BEGIN
56 static UBool U_CALLCONV locale_cleanup(void);
57 U_CDECL_END
58 
59 U_NAMESPACE_BEGIN
60 
61 static Locale   *gLocaleCache = NULL;
62 static UInitOnce gLocaleCacheInitOnce = U_INITONCE_INITIALIZER;
63 
64 // gDefaultLocaleMutex protects all access to gDefaultLocalesHashT and gDefaultLocale.
65 static UMutex gDefaultLocaleMutex = U_MUTEX_INITIALIZER;
66 static UHashtable *gDefaultLocalesHashT = NULL;
67 static Locale *gDefaultLocale = NULL;
68 
69 /**
70  * \def ULOC_STRING_LIMIT
71  * strings beyond this value crash in CharString
72  */
73 #define ULOC_STRING_LIMIT 357913941
74 
75 U_NAMESPACE_END
76 
77 typedef enum ELocalePos {
78     eENGLISH,
79     eFRENCH,
80     eGERMAN,
81     eITALIAN,
82     eJAPANESE,
83     eKOREAN,
84     eCHINESE,
85 
86     eFRANCE,
87     eGERMANY,
88     eITALY,
89     eJAPAN,
90     eKOREA,
91     eCHINA,      /* Alias for PRC */
92     eTAIWAN,
93     eUK,
94     eUS,
95     eCANADA,
96     eCANADA_FRENCH,
97     eROOT,
98 
99 
100     //eDEFAULT,
101     eMAX_LOCALES
102 } ELocalePos;
103 
104 U_CFUNC int32_t locale_getKeywords(const char *localeID,
105             char prev,
106             char *keywords, int32_t keywordCapacity,
107             char *values, int32_t valuesCapacity, int32_t *valLen,
108             UBool valuesToo,
109             UErrorCode *status);
110 
111 U_CDECL_BEGIN
112 //
113 // Deleter function for Locales owned by the default Locale hash table/
114 //
115 static void U_CALLCONV
deleteLocale(void * obj)116 deleteLocale(void *obj) {
117     delete (icu::Locale *) obj;
118 }
119 
locale_cleanup(void)120 static UBool U_CALLCONV locale_cleanup(void)
121 {
122     U_NAMESPACE_USE
123 
124     delete [] gLocaleCache;
125     gLocaleCache = NULL;
126     gLocaleCacheInitOnce.reset();
127 
128     if (gDefaultLocalesHashT) {
129         uhash_close(gDefaultLocalesHashT);   // Automatically deletes all elements, using deleter func.
130         gDefaultLocalesHashT = NULL;
131     }
132     gDefaultLocale = NULL;
133     return TRUE;
134 }
135 
136 
locale_init(UErrorCode & status)137 static void U_CALLCONV locale_init(UErrorCode &status) {
138     U_NAMESPACE_USE
139 
140     U_ASSERT(gLocaleCache == NULL);
141     gLocaleCache = new Locale[(int)eMAX_LOCALES];
142     if (gLocaleCache == NULL) {
143         status = U_MEMORY_ALLOCATION_ERROR;
144         return;
145     }
146     ucln_common_registerCleanup(UCLN_COMMON_LOCALE, locale_cleanup);
147     gLocaleCache[eROOT]          = Locale("");
148     gLocaleCache[eENGLISH]       = Locale("en");
149     gLocaleCache[eFRENCH]        = Locale("fr");
150     gLocaleCache[eGERMAN]        = Locale("de");
151     gLocaleCache[eITALIAN]       = Locale("it");
152     gLocaleCache[eJAPANESE]      = Locale("ja");
153     gLocaleCache[eKOREAN]        = Locale("ko");
154     gLocaleCache[eCHINESE]       = Locale("zh");
155     gLocaleCache[eFRANCE]        = Locale("fr", "FR");
156     gLocaleCache[eGERMANY]       = Locale("de", "DE");
157     gLocaleCache[eITALY]         = Locale("it", "IT");
158     gLocaleCache[eJAPAN]         = Locale("ja", "JP");
159     gLocaleCache[eKOREA]         = Locale("ko", "KR");
160     gLocaleCache[eCHINA]         = Locale("zh", "CN");
161     gLocaleCache[eTAIWAN]        = Locale("zh", "TW");
162     gLocaleCache[eUK]            = Locale("en", "GB");
163     gLocaleCache[eUS]            = Locale("en", "US");
164     gLocaleCache[eCANADA]        = Locale("en", "CA");
165     gLocaleCache[eCANADA_FRENCH] = Locale("fr", "CA");
166 }
167 
168 U_CDECL_END
169 
170 U_NAMESPACE_BEGIN
171 
locale_set_default_internal(const char * id,UErrorCode & status)172 Locale *locale_set_default_internal(const char *id, UErrorCode& status) {
173     // Synchronize this entire function.
174     Mutex lock(&gDefaultLocaleMutex);
175 
176     UBool canonicalize = FALSE;
177 
178     // If given a NULL string for the locale id, grab the default
179     //   name from the system.
180     //   (Different from most other locale APIs, where a null name means use
181     //    the current ICU default locale.)
182     if (id == NULL) {
183         id = uprv_getDefaultLocaleID();   // This function not thread safe? TODO: verify.
184         canonicalize = TRUE; // always canonicalize host ID
185     }
186 
187     char localeNameBuf[512];
188 
189     if (canonicalize) {
190         uloc_canonicalize(id, localeNameBuf, sizeof(localeNameBuf)-1, &status);
191     } else {
192         uloc_getName(id, localeNameBuf, sizeof(localeNameBuf)-1, &status);
193     }
194     localeNameBuf[sizeof(localeNameBuf)-1] = 0;  // Force null termination in event of
195                                                  //   a long name filling the buffer.
196                                                  //   (long names are truncated.)
197                                                  //
198     if (U_FAILURE(status)) {
199         return gDefaultLocale;
200     }
201 
202     if (gDefaultLocalesHashT == NULL) {
203         gDefaultLocalesHashT = uhash_open(uhash_hashChars, uhash_compareChars, NULL, &status);
204         if (U_FAILURE(status)) {
205             return gDefaultLocale;
206         }
207         uhash_setValueDeleter(gDefaultLocalesHashT, deleteLocale);
208         ucln_common_registerCleanup(UCLN_COMMON_LOCALE, locale_cleanup);
209     }
210 
211     Locale *newDefault = (Locale *)uhash_get(gDefaultLocalesHashT, localeNameBuf);
212     if (newDefault == NULL) {
213         newDefault = new Locale(Locale::eBOGUS);
214         if (newDefault == NULL) {
215             status = U_MEMORY_ALLOCATION_ERROR;
216             return gDefaultLocale;
217         }
218         newDefault->init(localeNameBuf, FALSE);
219         uhash_put(gDefaultLocalesHashT, (char*) newDefault->getName(), newDefault, &status);
220         if (U_FAILURE(status)) {
221             return gDefaultLocale;
222         }
223     }
224     gDefaultLocale = newDefault;
225     return gDefaultLocale;
226 }
227 
228 U_NAMESPACE_END
229 
230 /* sfb 07/21/99 */
231 U_CFUNC void
locale_set_default(const char * id)232 locale_set_default(const char *id)
233 {
234     U_NAMESPACE_USE
235     UErrorCode status = U_ZERO_ERROR;
236     locale_set_default_internal(id, status);
237 }
238 /* end */
239 
240 U_CFUNC const char *
locale_get_default(void)241 locale_get_default(void)
242 {
243     U_NAMESPACE_USE
244     return Locale::getDefault().getName();
245 }
246 
247 
248 U_NAMESPACE_BEGIN
249 
UOBJECT_DEFINE_RTTI_IMPLEMENTATION(Locale)250 UOBJECT_DEFINE_RTTI_IMPLEMENTATION(Locale)
251 
252 /*Character separating the posix id fields*/
253 // '_'
254 // In the platform codepage.
255 #define SEP_CHAR '_'
256 
257 Locale::~Locale()
258 {
259     if (baseName != fullName) {
260         uprv_free(baseName);
261     }
262     baseName = NULL;
263     /*if fullName is on the heap, we free it*/
264     if (fullName != fullNameBuffer)
265     {
266         uprv_free(fullName);
267         fullName = NULL;
268     }
269 }
270 
Locale()271 Locale::Locale()
272     : UObject(), fullName(fullNameBuffer), baseName(NULL)
273 {
274     init(NULL, FALSE);
275 }
276 
277 /*
278  * Internal constructor to allow construction of a locale object with
279  *   NO side effects.   (Default constructor tries to get
280  *   the default locale.)
281  */
Locale(Locale::ELocaleType)282 Locale::Locale(Locale::ELocaleType)
283     : UObject(), fullName(fullNameBuffer), baseName(NULL)
284 {
285     setToBogus();
286 }
287 
288 
Locale(const char * newLanguage,const char * newCountry,const char * newVariant,const char * newKeywords)289 Locale::Locale( const   char * newLanguage,
290                 const   char * newCountry,
291                 const   char * newVariant,
292                 const   char * newKeywords)
293     : UObject(), fullName(fullNameBuffer), baseName(NULL)
294 {
295     if( (newLanguage==NULL) && (newCountry == NULL) && (newVariant == NULL) )
296     {
297         init(NULL, FALSE); /* shortcut */
298     }
299     else
300     {
301         UErrorCode status = U_ZERO_ERROR;
302         int32_t size = 0;
303         int32_t lsize = 0;
304         int32_t csize = 0;
305         int32_t vsize = 0;
306         int32_t ksize = 0;
307 
308         // Calculate the size of the resulting string.
309 
310         // Language
311         if ( newLanguage != NULL )
312         {
313             lsize = (int32_t)uprv_strlen(newLanguage);
314             if ( lsize < 0 || lsize > ULOC_STRING_LIMIT ) { // int32 wrap
315                 setToBogus();
316                 return;
317             }
318             size = lsize;
319         }
320 
321         CharString togo(newLanguage, lsize, status); // start with newLanguage
322 
323         // _Country
324         if ( newCountry != NULL )
325         {
326             csize = (int32_t)uprv_strlen(newCountry);
327             if ( csize < 0 || csize > ULOC_STRING_LIMIT ) { // int32 wrap
328                 setToBogus();
329                 return;
330             }
331             size += csize;
332         }
333 
334         // _Variant
335         if ( newVariant != NULL )
336         {
337             // remove leading _'s
338             while(newVariant[0] == SEP_CHAR)
339             {
340                 newVariant++;
341             }
342 
343             // remove trailing _'s
344             vsize = (int32_t)uprv_strlen(newVariant);
345             if ( vsize < 0 || vsize > ULOC_STRING_LIMIT ) { // int32 wrap
346                 setToBogus();
347                 return;
348             }
349             while( (vsize>1) && (newVariant[vsize-1] == SEP_CHAR) )
350             {
351                 vsize--;
352             }
353         }
354 
355         if( vsize > 0 )
356         {
357             size += vsize;
358         }
359 
360         // Separator rules:
361         if ( vsize > 0 )
362         {
363             size += 2;  // at least: __v
364         }
365         else if ( csize > 0 )
366         {
367             size += 1;  // at least: _v
368         }
369 
370         if ( newKeywords != NULL)
371         {
372             ksize = (int32_t)uprv_strlen(newKeywords);
373             if ( ksize < 0 || ksize > ULOC_STRING_LIMIT ) {
374               setToBogus();
375               return;
376             }
377             size += ksize + 1;
378         }
379 
380         //  NOW we have the full locale string..
381         // Now, copy it back.
382 
383         // newLanguage is already copied
384 
385         if ( ( vsize != 0 ) || (csize != 0) )  // at least:  __v
386         {                                      //            ^
387             togo.append(SEP_CHAR, status);
388         }
389 
390         if ( csize != 0 )
391         {
392             togo.append(newCountry, status);
393         }
394 
395         if ( vsize != 0)
396         {
397             togo.append(SEP_CHAR, status)
398                 .append(newVariant, vsize, status);
399         }
400 
401         if ( ksize != 0)
402         {
403             if (uprv_strchr(newKeywords, '=')) {
404                 togo.append('@', status); /* keyword parsing */
405             }
406             else {
407                 togo.append('_', status); /* Variant parsing with a script */
408                 if ( vsize == 0) {
409                     togo.append('_', status); /* No country found */
410                 }
411             }
412             togo.append(newKeywords, status);
413         }
414 
415         if (U_FAILURE(status)) {
416             // Something went wrong with appending, etc.
417             setToBogus();
418             return;
419         }
420         // Parse it, because for example 'language' might really be a complete
421         // string.
422         init(togo.data(), FALSE);
423     }
424 }
425 
Locale(const Locale & other)426 Locale::Locale(const Locale &other)
427     : UObject(other), fullName(fullNameBuffer), baseName(NULL)
428 {
429     *this = other;
430 }
431 
Locale(Locale && other)432 Locale::Locale(Locale&& other) U_NOEXCEPT
433     : UObject(other), fullName(fullNameBuffer), baseName(fullName) {
434   *this = std::move(other);
435 }
436 
operator =(const Locale & other)437 Locale& Locale::operator=(const Locale& other) {
438     if (this == &other) {
439         return *this;
440     }
441 
442     setToBogus();
443 
444     if (other.fullName == other.fullNameBuffer) {
445         uprv_strcpy(fullNameBuffer, other.fullNameBuffer);
446     } else if (other.fullName == nullptr) {
447         fullName = nullptr;
448     } else {
449         fullName = uprv_strdup(other.fullName);
450         if (fullName == nullptr) return *this;
451     }
452 
453     if (other.baseName == other.fullName) {
454         baseName = fullName;
455     } else if (other.baseName != nullptr) {
456         baseName = uprv_strdup(other.baseName);
457         if (baseName == nullptr) return *this;
458     }
459 
460     uprv_strcpy(language, other.language);
461     uprv_strcpy(script, other.script);
462     uprv_strcpy(country, other.country);
463 
464     variantBegin = other.variantBegin;
465     fIsBogus = other.fIsBogus;
466 
467     return *this;
468 }
469 
operator =(Locale && other)470 Locale& Locale::operator=(Locale&& other) U_NOEXCEPT {
471     if (baseName != fullName) uprv_free(baseName);
472     if (fullName != fullNameBuffer) uprv_free(fullName);
473 
474     if (other.fullName == other.fullNameBuffer) {
475         uprv_strcpy(fullNameBuffer, other.fullNameBuffer);
476         fullName = fullNameBuffer;
477     } else {
478         fullName = other.fullName;
479     }
480 
481     if (other.baseName == other.fullName) {
482         baseName = fullName;
483     } else {
484         baseName = other.baseName;
485     }
486 
487     uprv_strcpy(language, other.language);
488     uprv_strcpy(script, other.script);
489     uprv_strcpy(country, other.country);
490 
491     variantBegin = other.variantBegin;
492     fIsBogus = other.fIsBogus;
493 
494     other.baseName = other.fullName = other.fullNameBuffer;
495 
496     return *this;
497 }
498 
499 Locale *
clone() const500 Locale::clone() const {
501     return new Locale(*this);
502 }
503 
504 UBool
operator ==(const Locale & other) const505 Locale::operator==( const   Locale& other) const
506 {
507     return (uprv_strcmp(other.fullName, fullName) == 0);
508 }
509 
510 #define ISASCIIALPHA(c) (((c) >= 'a' && (c) <= 'z') || ((c) >= 'A' && (c) <= 'Z'))
511 
512 /*This function initializes a Locale from a C locale ID*/
init(const char * localeID,UBool canonicalize)513 Locale& Locale::init(const char* localeID, UBool canonicalize)
514 {
515     fIsBogus = FALSE;
516     /* Free our current storage */
517     if (baseName != fullName) {
518         uprv_free(baseName);
519     }
520     baseName = NULL;
521     if(fullName != fullNameBuffer) {
522         uprv_free(fullName);
523         fullName = fullNameBuffer;
524     }
525 
526     // not a loop:
527     // just an easy way to have a common error-exit
528     // without goto and without another function
529     do {
530         char *separator;
531         char *field[5] = {0};
532         int32_t fieldLen[5] = {0};
533         int32_t fieldIdx;
534         int32_t variantField;
535         int32_t length;
536         UErrorCode err;
537 
538         if(localeID == NULL) {
539             // not an error, just set the default locale
540             return *this = getDefault();
541         }
542 
543         /* preset all fields to empty */
544         language[0] = script[0] = country[0] = 0;
545 
546         // "canonicalize" the locale ID to ICU/Java format
547         err = U_ZERO_ERROR;
548         length = canonicalize ?
549             uloc_canonicalize(localeID, fullName, sizeof(fullNameBuffer), &err) :
550             uloc_getName(localeID, fullName, sizeof(fullNameBuffer), &err);
551 
552         if(err == U_BUFFER_OVERFLOW_ERROR || length >= (int32_t)sizeof(fullNameBuffer)) {
553             /*Go to heap for the fullName if necessary*/
554             fullName = (char *)uprv_malloc(sizeof(char)*(length + 1));
555             if(fullName == 0) {
556                 fullName = fullNameBuffer;
557                 break; // error: out of memory
558             }
559             err = U_ZERO_ERROR;
560             length = canonicalize ?
561                 uloc_canonicalize(localeID, fullName, length+1, &err) :
562                 uloc_getName(localeID, fullName, length+1, &err);
563         }
564         if(U_FAILURE(err) || err == U_STRING_NOT_TERMINATED_WARNING) {
565             /* should never occur */
566             break;
567         }
568 
569         variantBegin = length;
570 
571         /* after uloc_getName/canonicalize() we know that only '_' are separators */
572         separator = field[0] = fullName;
573         fieldIdx = 1;
574         while ((separator = uprv_strchr(field[fieldIdx-1], SEP_CHAR)) != 0 && fieldIdx < UPRV_LENGTHOF(field)-1) {
575             field[fieldIdx] = separator + 1;
576             fieldLen[fieldIdx-1] = (int32_t)(separator - field[fieldIdx-1]);
577             fieldIdx++;
578         }
579         // variant may contain @foo or .foo POSIX cruft; remove it
580         separator = uprv_strchr(field[fieldIdx-1], '@');
581         char* sep2 = uprv_strchr(field[fieldIdx-1], '.');
582         if (separator!=NULL || sep2!=NULL) {
583             if (separator==NULL || (sep2!=NULL && separator > sep2)) {
584                 separator = sep2;
585             }
586             fieldLen[fieldIdx-1] = (int32_t)(separator - field[fieldIdx-1]);
587         } else {
588             fieldLen[fieldIdx-1] = length - (int32_t)(field[fieldIdx-1] - fullName);
589         }
590 
591         if (fieldLen[0] >= (int32_t)(sizeof(language)))
592         {
593             break; // error: the language field is too long
594         }
595 
596         variantField = 1; /* Usually the 2nd one, except when a script or country is also used. */
597         if (fieldLen[0] > 0) {
598             /* We have a language */
599             uprv_memcpy(language, fullName, fieldLen[0]);
600             language[fieldLen[0]] = 0;
601         }
602         if (fieldLen[1] == 4 && ISASCIIALPHA(field[1][0]) &&
603                 ISASCIIALPHA(field[1][1]) && ISASCIIALPHA(field[1][2]) &&
604                 ISASCIIALPHA(field[1][3])) {
605             /* We have at least a script */
606             uprv_memcpy(script, field[1], fieldLen[1]);
607             script[fieldLen[1]] = 0;
608             variantField++;
609         }
610 
611         if (fieldLen[variantField] == 2 || fieldLen[variantField] == 3) {
612             /* We have a country */
613             uprv_memcpy(country, field[variantField], fieldLen[variantField]);
614             country[fieldLen[variantField]] = 0;
615             variantField++;
616         } else if (fieldLen[variantField] == 0) {
617             variantField++; /* script or country empty but variant in next field (i.e. en__POSIX) */
618         }
619 
620         if (fieldLen[variantField] > 0) {
621             /* We have a variant */
622             variantBegin = (int32_t)(field[variantField] - fullName);
623         }
624 
625         err = U_ZERO_ERROR;
626         initBaseName(err);
627         if (U_FAILURE(err)) {
628             break;
629         }
630 
631         // successful end of init()
632         return *this;
633     } while(0); /*loop doesn't iterate*/
634 
635     // when an error occurs, then set this object to "bogus" (there is no UErrorCode here)
636     setToBogus();
637 
638     return *this;
639 }
640 
641 /*
642  * Set up the base name.
643  * If there are no key words, it's exactly the full name.
644  * If key words exist, it's the full name truncated at the '@' character.
645  * Need to set up both at init() and after setting a keyword.
646  */
647 void
initBaseName(UErrorCode & status)648 Locale::initBaseName(UErrorCode &status) {
649     if (U_FAILURE(status)) {
650         return;
651     }
652     U_ASSERT(baseName==NULL || baseName==fullName);
653     const char *atPtr = uprv_strchr(fullName, '@');
654     const char *eqPtr = uprv_strchr(fullName, '=');
655     if (atPtr && eqPtr && atPtr < eqPtr) {
656         // Key words exist.
657         int32_t baseNameLength = (int32_t)(atPtr - fullName);
658         baseName = (char *)uprv_malloc(baseNameLength + 1);
659         if (baseName == NULL) {
660             status = U_MEMORY_ALLOCATION_ERROR;
661             return;
662         }
663         uprv_strncpy(baseName, fullName, baseNameLength);
664         baseName[baseNameLength] = 0;
665 
666         // The original computation of variantBegin leaves it equal to the length
667         // of fullName if there is no variant.  It should instead be
668         // the length of the baseName.
669         if (variantBegin > baseNameLength) {
670             variantBegin = baseNameLength;
671         }
672     } else {
673         baseName = fullName;
674     }
675 }
676 
677 
678 int32_t
hashCode() const679 Locale::hashCode() const
680 {
681     return ustr_hashCharsN(fullName, static_cast<int32_t>(uprv_strlen(fullName)));
682 }
683 
684 void
setToBogus()685 Locale::setToBogus() {
686     /* Free our current storage */
687     if(baseName != fullName) {
688         uprv_free(baseName);
689     }
690     baseName = NULL;
691     if(fullName != fullNameBuffer) {
692         uprv_free(fullName);
693         fullName = fullNameBuffer;
694     }
695     *fullNameBuffer = 0;
696     *language = 0;
697     *script = 0;
698     *country = 0;
699     fIsBogus = TRUE;
700     variantBegin = 0;
701 }
702 
703 const Locale& U_EXPORT2
getDefault()704 Locale::getDefault()
705 {
706     {
707         Mutex lock(&gDefaultLocaleMutex);
708         if (gDefaultLocale != NULL) {
709             return *gDefaultLocale;
710         }
711     }
712     UErrorCode status = U_ZERO_ERROR;
713     return *locale_set_default_internal(NULL, status);
714 }
715 
716 
717 
718 void U_EXPORT2
setDefault(const Locale & newLocale,UErrorCode & status)719 Locale::setDefault( const   Locale&     newLocale,
720                             UErrorCode&  status)
721 {
722     if (U_FAILURE(status)) {
723         return;
724     }
725 
726     /* Set the default from the full name string of the supplied locale.
727      * This is a convenient way to access the default locale caching mechanisms.
728      */
729     const char *localeID = newLocale.getName();
730     locale_set_default_internal(localeID, status);
731 }
732 
733 void
addLikelySubtags(UErrorCode & status)734 Locale::addLikelySubtags(UErrorCode& status) {
735     if (U_FAILURE(status)) {
736         return;
737     }
738 
739     // The maximized locale ID string is often longer, but there is no good
740     // heuristic to estimate just how much longer. Leave that to CharString.
741     CharString maximizedLocaleID;
742     int32_t maximizedLocaleIDCapacity = static_cast<int32_t>(uprv_strlen(fullName));
743 
744     char* buffer;
745     int32_t reslen;
746 
747     for (;;) {
748         buffer = maximizedLocaleID.getAppendBuffer(
749                 /*minCapacity=*/maximizedLocaleIDCapacity,
750                 /*desiredCapacityHint=*/maximizedLocaleIDCapacity,
751                 maximizedLocaleIDCapacity,
752                 status);
753 
754         if (U_FAILURE(status)) {
755             return;
756         }
757 
758         reslen = uloc_addLikelySubtags(
759                 fullName,
760                 buffer,
761                 maximizedLocaleIDCapacity,
762                 &status);
763 
764         if (status != U_BUFFER_OVERFLOW_ERROR) {
765             break;
766         }
767 
768         maximizedLocaleIDCapacity = reslen;
769         status = U_ZERO_ERROR;
770     }
771 
772     if (U_FAILURE(status)) {
773         return;
774     }
775 
776     maximizedLocaleID.append(buffer, reslen, status);
777     if (status == U_STRING_NOT_TERMINATED_WARNING) {
778         status = U_ZERO_ERROR;  // Terminators provided by CharString.
779     }
780 
781     if (U_FAILURE(status)) {
782         return;
783     }
784 
785     init(maximizedLocaleID.data(), /*canonicalize=*/FALSE);
786     if (isBogus()) {
787         status = U_ILLEGAL_ARGUMENT_ERROR;
788     }
789 }
790 
791 void
minimizeSubtags(UErrorCode & status)792 Locale::minimizeSubtags(UErrorCode& status) {
793     if (U_FAILURE(status)) {
794         return;
795     }
796 
797     // Except for a few edge cases (like the empty string, that is minimized to
798     // "en__POSIX"), minimized locale ID strings will be either the same length
799     // or shorter than their input.
800     CharString minimizedLocaleID;
801     int32_t minimizedLocaleIDCapacity = static_cast<int32_t>(uprv_strlen(fullName));
802 
803     char* buffer;
804     int32_t reslen;
805 
806     for (;;) {
807         buffer = minimizedLocaleID.getAppendBuffer(
808                 /*minCapacity=*/minimizedLocaleIDCapacity,
809                 /*desiredCapacityHint=*/minimizedLocaleIDCapacity,
810                 minimizedLocaleIDCapacity,
811                 status);
812 
813         if (U_FAILURE(status)) {
814             return;
815         }
816 
817         reslen = uloc_minimizeSubtags(
818                 fullName,
819                 buffer,
820                 minimizedLocaleIDCapacity,
821                 &status);
822 
823         if (status != U_BUFFER_OVERFLOW_ERROR) {
824             break;
825         }
826 
827         // Because of the internal minimal buffer size of CharString, I can't
828         // think of any input data for which this could possibly ever happen.
829         // Maybe it would be better replaced with an assertion instead?
830         minimizedLocaleIDCapacity = reslen;
831         status = U_ZERO_ERROR;
832     }
833 
834     if (U_FAILURE(status)) {
835         return;
836     }
837 
838     minimizedLocaleID.append(buffer, reslen, status);
839     if (status == U_STRING_NOT_TERMINATED_WARNING) {
840         status = U_ZERO_ERROR;  // Terminators provided by CharString.
841     }
842 
843     if (U_FAILURE(status)) {
844         return;
845     }
846 
847     init(minimizedLocaleID.data(), /*canonicalize=*/FALSE);
848     if (isBogus()) {
849         status = U_ILLEGAL_ARGUMENT_ERROR;
850     }
851 }
852 
853 Locale U_EXPORT2
forLanguageTag(StringPiece tag,UErrorCode & status)854 Locale::forLanguageTag(StringPiece tag, UErrorCode& status)
855 {
856     Locale result(Locale::eBOGUS);
857 
858     if (U_FAILURE(status)) {
859         return result;
860     }
861 
862     // If a BCP-47 language tag is passed as the language parameter to the
863     // normal Locale constructor, it will actually fall back to invoking
864     // uloc_forLanguageTag() to parse it if it somehow is able to detect that
865     // the string actually is BCP-47. This works well for things like strings
866     // using BCP-47 extensions, but it does not at all work for things like
867     // BCP-47 grandfathered tags (eg. "en-GB-oed") which are possible to also
868     // interpret as ICU locale IDs and because of that won't trigger the BCP-47
869     // parsing. Therefore the code here explicitly calls uloc_forLanguageTag()
870     // and then Locale::init(), instead of just calling the normal constructor.
871 
872     // All simple language tags will have the exact same length as ICU locale
873     // ID strings as they have as BCP-47 strings (like "en_US" for "en-US").
874     CharString localeID;
875     int32_t resultCapacity = tag.size();
876 
877     char* buffer;
878     int32_t parsedLength, reslen;
879 
880     for (;;) {
881         buffer = localeID.getAppendBuffer(
882                 /*minCapacity=*/resultCapacity,
883                 /*desiredCapacityHint=*/resultCapacity,
884                 resultCapacity,
885                 status);
886 
887         if (U_FAILURE(status)) {
888             return result;
889         }
890 
891         reslen = ulocimp_forLanguageTag(
892                 tag.data(),
893                 tag.length(),
894                 buffer,
895                 resultCapacity,
896                 &parsedLength,
897                 &status);
898 
899         if (status != U_BUFFER_OVERFLOW_ERROR) {
900             break;
901         }
902 
903         // For all BCP-47 language tags that use extensions, the corresponding
904         // ICU locale ID will be longer but uloc_forLanguageTag() does compute
905         // the exact length needed so this memory reallocation will be done at
906         // most once.
907         resultCapacity = reslen;
908         status = U_ZERO_ERROR;
909     }
910 
911     if (U_FAILURE(status)) {
912         return result;
913     }
914 
915     if (parsedLength != tag.size()) {
916         status = U_ILLEGAL_ARGUMENT_ERROR;
917         return result;
918     }
919 
920     localeID.append(buffer, reslen, status);
921     if (status == U_STRING_NOT_TERMINATED_WARNING) {
922         status = U_ZERO_ERROR;  // Terminators provided by CharString.
923     }
924 
925     if (U_FAILURE(status)) {
926         return result;
927     }
928 
929     result.init(localeID.data(), /*canonicalize=*/FALSE);
930     if (result.isBogus()) {
931         status = U_ILLEGAL_ARGUMENT_ERROR;
932     }
933     return result;
934 }
935 
936 void
toLanguageTag(ByteSink & sink,UErrorCode & status) const937 Locale::toLanguageTag(ByteSink& sink, UErrorCode& status) const
938 {
939     if (U_FAILURE(status)) {
940         return;
941     }
942 
943     if (fIsBogus) {
944         status = U_ILLEGAL_ARGUMENT_ERROR;
945         return;
946     }
947 
948     // All simple language tags will have the exact same length as BCP-47
949     // strings as they have as ICU locale IDs (like "en-US" for "en_US").
950     LocalMemory<char> scratch;
951     int32_t scratch_capacity = static_cast<int32_t>(uprv_strlen(fullName));
952 
953     if (scratch_capacity == 0) {
954         scratch_capacity = 3;  // "und"
955     }
956 
957     char* buffer;
958     int32_t result_capacity, reslen;
959 
960     for (;;) {
961         if (scratch.allocateInsteadAndReset(scratch_capacity) == nullptr) {
962             status = U_MEMORY_ALLOCATION_ERROR;
963             return;
964         }
965 
966         buffer = sink.GetAppendBuffer(
967                 /*min_capacity=*/scratch_capacity,
968                 /*desired_capacity_hint=*/scratch_capacity,
969                 scratch.getAlias(),
970                 scratch_capacity,
971                 &result_capacity);
972 
973         reslen = uloc_toLanguageTag(
974                 fullName,
975                 buffer,
976                 result_capacity,
977                 /*strict=*/FALSE,
978                 &status);
979 
980         if (status != U_BUFFER_OVERFLOW_ERROR) {
981             break;
982         }
983 
984         // For some very few edge cases a language tag will be longer as a
985         // BCP-47 string than it is as an ICU locale ID. Most notoriously "C"
986         // expands to the BCP-47 tag "en-US-u-va-posix", 16 times longer, and
987         // it'll take several calls to uloc_toLanguageTag() to figure that out.
988         // https://unicode-org.atlassian.net/browse/ICU-20132
989         scratch_capacity = reslen;
990         status = U_ZERO_ERROR;
991     }
992 
993     if (U_FAILURE(status)) {
994         return;
995     }
996 
997     sink.Append(buffer, reslen);
998     if (status == U_STRING_NOT_TERMINATED_WARNING) {
999         status = U_ZERO_ERROR;  // Terminators not used.
1000     }
1001 }
1002 
1003 Locale U_EXPORT2
createFromName(const char * name)1004 Locale::createFromName (const char *name)
1005 {
1006     if (name) {
1007         Locale l("");
1008         l.init(name, FALSE);
1009         return l;
1010     }
1011     else {
1012         return getDefault();
1013     }
1014 }
1015 
1016 Locale U_EXPORT2
createCanonical(const char * name)1017 Locale::createCanonical(const char* name) {
1018     Locale loc("");
1019     loc.init(name, TRUE);
1020     return loc;
1021 }
1022 
1023 const char *
getISO3Language() const1024 Locale::getISO3Language() const
1025 {
1026     return uloc_getISO3Language(fullName);
1027 }
1028 
1029 
1030 const char *
getISO3Country() const1031 Locale::getISO3Country() const
1032 {
1033     return uloc_getISO3Country(fullName);
1034 }
1035 
1036 /**
1037  * Return the LCID value as specified in the "LocaleID" resource for this
1038  * locale.  The LocaleID must be expressed as a hexadecimal number, from
1039  * one to four digits.  If the LocaleID resource is not present, or is
1040  * in an incorrect format, 0 is returned.  The LocaleID is for use in
1041  * Windows (it is an LCID), but is available on all platforms.
1042  */
1043 uint32_t
getLCID() const1044 Locale::getLCID() const
1045 {
1046     return uloc_getLCID(fullName);
1047 }
1048 
getISOCountries()1049 const char* const* U_EXPORT2 Locale::getISOCountries()
1050 {
1051     return uloc_getISOCountries();
1052 }
1053 
getISOLanguages()1054 const char* const* U_EXPORT2 Locale::getISOLanguages()
1055 {
1056     return uloc_getISOLanguages();
1057 }
1058 
1059 // Set the locale's data based on a posix id.
setFromPOSIXID(const char * posixID)1060 void Locale::setFromPOSIXID(const char *posixID)
1061 {
1062     init(posixID, TRUE);
1063 }
1064 
1065 const Locale & U_EXPORT2
getRoot(void)1066 Locale::getRoot(void)
1067 {
1068     return getLocale(eROOT);
1069 }
1070 
1071 const Locale & U_EXPORT2
getEnglish(void)1072 Locale::getEnglish(void)
1073 {
1074     return getLocale(eENGLISH);
1075 }
1076 
1077 const Locale & U_EXPORT2
getFrench(void)1078 Locale::getFrench(void)
1079 {
1080     return getLocale(eFRENCH);
1081 }
1082 
1083 const Locale & U_EXPORT2
getGerman(void)1084 Locale::getGerman(void)
1085 {
1086     return getLocale(eGERMAN);
1087 }
1088 
1089 const Locale & U_EXPORT2
getItalian(void)1090 Locale::getItalian(void)
1091 {
1092     return getLocale(eITALIAN);
1093 }
1094 
1095 const Locale & U_EXPORT2
getJapanese(void)1096 Locale::getJapanese(void)
1097 {
1098     return getLocale(eJAPANESE);
1099 }
1100 
1101 const Locale & U_EXPORT2
getKorean(void)1102 Locale::getKorean(void)
1103 {
1104     return getLocale(eKOREAN);
1105 }
1106 
1107 const Locale & U_EXPORT2
getChinese(void)1108 Locale::getChinese(void)
1109 {
1110     return getLocale(eCHINESE);
1111 }
1112 
1113 const Locale & U_EXPORT2
getSimplifiedChinese(void)1114 Locale::getSimplifiedChinese(void)
1115 {
1116     return getLocale(eCHINA);
1117 }
1118 
1119 const Locale & U_EXPORT2
getTraditionalChinese(void)1120 Locale::getTraditionalChinese(void)
1121 {
1122     return getLocale(eTAIWAN);
1123 }
1124 
1125 
1126 const Locale & U_EXPORT2
getFrance(void)1127 Locale::getFrance(void)
1128 {
1129     return getLocale(eFRANCE);
1130 }
1131 
1132 const Locale & U_EXPORT2
getGermany(void)1133 Locale::getGermany(void)
1134 {
1135     return getLocale(eGERMANY);
1136 }
1137 
1138 const Locale & U_EXPORT2
getItaly(void)1139 Locale::getItaly(void)
1140 {
1141     return getLocale(eITALY);
1142 }
1143 
1144 const Locale & U_EXPORT2
getJapan(void)1145 Locale::getJapan(void)
1146 {
1147     return getLocale(eJAPAN);
1148 }
1149 
1150 const Locale & U_EXPORT2
getKorea(void)1151 Locale::getKorea(void)
1152 {
1153     return getLocale(eKOREA);
1154 }
1155 
1156 const Locale & U_EXPORT2
getChina(void)1157 Locale::getChina(void)
1158 {
1159     return getLocale(eCHINA);
1160 }
1161 
1162 const Locale & U_EXPORT2
getPRC(void)1163 Locale::getPRC(void)
1164 {
1165     return getLocale(eCHINA);
1166 }
1167 
1168 const Locale & U_EXPORT2
getTaiwan(void)1169 Locale::getTaiwan(void)
1170 {
1171     return getLocale(eTAIWAN);
1172 }
1173 
1174 const Locale & U_EXPORT2
getUK(void)1175 Locale::getUK(void)
1176 {
1177     return getLocale(eUK);
1178 }
1179 
1180 const Locale & U_EXPORT2
getUS(void)1181 Locale::getUS(void)
1182 {
1183     return getLocale(eUS);
1184 }
1185 
1186 const Locale & U_EXPORT2
getCanada(void)1187 Locale::getCanada(void)
1188 {
1189     return getLocale(eCANADA);
1190 }
1191 
1192 const Locale & U_EXPORT2
getCanadaFrench(void)1193 Locale::getCanadaFrench(void)
1194 {
1195     return getLocale(eCANADA_FRENCH);
1196 }
1197 
1198 const Locale &
getLocale(int locid)1199 Locale::getLocale(int locid)
1200 {
1201     Locale *localeCache = getLocaleCache();
1202     U_ASSERT((locid < eMAX_LOCALES)&&(locid>=0));
1203     if (localeCache == NULL) {
1204         // Failure allocating the locale cache.
1205         //   The best we can do is return a NULL reference.
1206         locid = 0;
1207     }
1208     return localeCache[locid]; /*operating on NULL*/
1209 }
1210 
1211 /*
1212 This function is defined this way in order to get around static
1213 initialization and static destruction.
1214  */
1215 Locale *
getLocaleCache(void)1216 Locale::getLocaleCache(void)
1217 {
1218     UErrorCode status = U_ZERO_ERROR;
1219     umtx_initOnce(gLocaleCacheInitOnce, locale_init, status);
1220     return gLocaleCache;
1221 }
1222 
1223 class KeywordEnumeration : public StringEnumeration {
1224 private:
1225     char *keywords;
1226     char *current;
1227     int32_t length;
1228     UnicodeString currUSKey;
1229     static const char fgClassID;/* Warning this is used beyond the typical RTTI usage. */
1230 
1231 public:
getStaticClassID(void)1232     static UClassID U_EXPORT2 getStaticClassID(void) { return (UClassID)&fgClassID; }
getDynamicClassID(void) const1233     virtual UClassID getDynamicClassID(void) const { return getStaticClassID(); }
1234 public:
KeywordEnumeration(const char * keys,int32_t keywordLen,int32_t currentIndex,UErrorCode & status)1235     KeywordEnumeration(const char *keys, int32_t keywordLen, int32_t currentIndex, UErrorCode &status)
1236         : keywords((char *)&fgClassID), current((char *)&fgClassID), length(0) {
1237         if(U_SUCCESS(status) && keywordLen != 0) {
1238             if(keys == NULL || keywordLen < 0) {
1239                 status = U_ILLEGAL_ARGUMENT_ERROR;
1240             } else {
1241                 keywords = (char *)uprv_malloc(keywordLen+1);
1242                 if (keywords == NULL) {
1243                     status = U_MEMORY_ALLOCATION_ERROR;
1244                 }
1245                 else {
1246                     uprv_memcpy(keywords, keys, keywordLen);
1247                     keywords[keywordLen] = 0;
1248                     current = keywords + currentIndex;
1249                     length = keywordLen;
1250                 }
1251             }
1252         }
1253     }
1254 
1255     virtual ~KeywordEnumeration();
1256 
clone() const1257     virtual StringEnumeration * clone() const
1258     {
1259         UErrorCode status = U_ZERO_ERROR;
1260         return new KeywordEnumeration(keywords, length, (int32_t)(current - keywords), status);
1261     }
1262 
count(UErrorCode &) const1263     virtual int32_t count(UErrorCode &/*status*/) const {
1264         char *kw = keywords;
1265         int32_t result = 0;
1266         while(*kw) {
1267             result++;
1268             kw += uprv_strlen(kw)+1;
1269         }
1270         return result;
1271     }
1272 
next(int32_t * resultLength,UErrorCode & status)1273     virtual const char* next(int32_t* resultLength, UErrorCode& status) {
1274         const char* result;
1275         int32_t len;
1276         if(U_SUCCESS(status) && *current != 0) {
1277             result = current;
1278             len = (int32_t)uprv_strlen(current);
1279             current += len+1;
1280             if(resultLength != NULL) {
1281                 *resultLength = len;
1282             }
1283         } else {
1284             if(resultLength != NULL) {
1285                 *resultLength = 0;
1286             }
1287             result = NULL;
1288         }
1289         return result;
1290     }
1291 
snext(UErrorCode & status)1292     virtual const UnicodeString* snext(UErrorCode& status) {
1293         int32_t resultLength = 0;
1294         const char *s = next(&resultLength, status);
1295         return setChars(s, resultLength, status);
1296     }
1297 
reset(UErrorCode &)1298     virtual void reset(UErrorCode& /*status*/) {
1299         current = keywords;
1300     }
1301 };
1302 
1303 const char KeywordEnumeration::fgClassID = '\0';
1304 
~KeywordEnumeration()1305 KeywordEnumeration::~KeywordEnumeration() {
1306     uprv_free(keywords);
1307 }
1308 
1309 // A wrapper around KeywordEnumeration that calls uloc_toUnicodeLocaleKey() in
1310 // the next() method for each keyword before returning it.
1311 class UnicodeKeywordEnumeration : public KeywordEnumeration {
1312 public:
1313     using KeywordEnumeration::KeywordEnumeration;
1314     virtual ~UnicodeKeywordEnumeration();
1315 
next(int32_t * resultLength,UErrorCode & status)1316     virtual const char* next(int32_t* resultLength, UErrorCode& status) {
1317         const char* legacy_key = KeywordEnumeration::next(nullptr, status);
1318         if (U_SUCCESS(status) && legacy_key != nullptr) {
1319             const char* key = uloc_toUnicodeLocaleKey(legacy_key);
1320             if (key == nullptr) {
1321                 status = U_ILLEGAL_ARGUMENT_ERROR;
1322             } else {
1323                 if (resultLength != nullptr) {
1324                     *resultLength = static_cast<int32_t>(uprv_strlen(key));
1325                 }
1326                 return key;
1327             }
1328         }
1329         if (resultLength != nullptr) *resultLength = 0;
1330         return nullptr;
1331     }
1332 };
1333 
1334 // Out-of-line virtual destructor to serve as the "key function".
1335 UnicodeKeywordEnumeration::~UnicodeKeywordEnumeration() = default;
1336 
1337 StringEnumeration *
createKeywords(UErrorCode & status) const1338 Locale::createKeywords(UErrorCode &status) const
1339 {
1340     char keywords[256];
1341     int32_t keywordCapacity = sizeof keywords;
1342     StringEnumeration *result = NULL;
1343 
1344     if (U_FAILURE(status)) {
1345         return result;
1346     }
1347 
1348     const char* variantStart = uprv_strchr(fullName, '@');
1349     const char* assignment = uprv_strchr(fullName, '=');
1350     if(variantStart) {
1351         if(assignment > variantStart) {
1352             int32_t keyLen = locale_getKeywords(variantStart+1, '@', keywords, keywordCapacity, NULL, 0, NULL, FALSE, &status);
1353             if(U_SUCCESS(status) && keyLen) {
1354                 result = new KeywordEnumeration(keywords, keyLen, 0, status);
1355                 if (!result) {
1356                     status = U_MEMORY_ALLOCATION_ERROR;
1357                 }
1358             }
1359         } else {
1360             status = U_INVALID_FORMAT_ERROR;
1361         }
1362     }
1363     return result;
1364 }
1365 
1366 StringEnumeration *
createUnicodeKeywords(UErrorCode & status) const1367 Locale::createUnicodeKeywords(UErrorCode &status) const
1368 {
1369     char keywords[256];
1370     int32_t keywordCapacity = sizeof keywords;
1371     StringEnumeration *result = NULL;
1372 
1373     if (U_FAILURE(status)) {
1374         return result;
1375     }
1376 
1377     const char* variantStart = uprv_strchr(fullName, '@');
1378     const char* assignment = uprv_strchr(fullName, '=');
1379     if(variantStart) {
1380         if(assignment > variantStart) {
1381             int32_t keyLen = locale_getKeywords(variantStart+1, '@', keywords, keywordCapacity, NULL, 0, NULL, FALSE, &status);
1382             if(U_SUCCESS(status) && keyLen) {
1383                 result = new UnicodeKeywordEnumeration(keywords, keyLen, 0, status);
1384                 if (!result) {
1385                     status = U_MEMORY_ALLOCATION_ERROR;
1386                 }
1387             }
1388         } else {
1389             status = U_INVALID_FORMAT_ERROR;
1390         }
1391     }
1392     return result;
1393 }
1394 
1395 int32_t
getKeywordValue(const char * keywordName,char * buffer,int32_t bufLen,UErrorCode & status) const1396 Locale::getKeywordValue(const char* keywordName, char *buffer, int32_t bufLen, UErrorCode &status) const
1397 {
1398     return uloc_getKeywordValue(fullName, keywordName, buffer, bufLen, &status);
1399 }
1400 
1401 void
getKeywordValue(StringPiece keywordName,ByteSink & sink,UErrorCode & status) const1402 Locale::getKeywordValue(StringPiece keywordName, ByteSink& sink, UErrorCode& status) const {
1403     if (U_FAILURE(status)) {
1404         return;
1405     }
1406 
1407     if (fIsBogus) {
1408         status = U_ILLEGAL_ARGUMENT_ERROR;
1409         return;
1410     }
1411 
1412     // TODO: Remove the need for a const char* to a NUL terminated buffer.
1413     const CharString keywordName_nul(keywordName, status);
1414     if (U_FAILURE(status)) {
1415         return;
1416     }
1417 
1418     LocalMemory<char> scratch;
1419     int32_t scratch_capacity = 16;  // Arbitrarily chosen default size.
1420 
1421     char* buffer;
1422     int32_t result_capacity, reslen;
1423 
1424     for (;;) {
1425         if (scratch.allocateInsteadAndReset(scratch_capacity) == nullptr) {
1426             status = U_MEMORY_ALLOCATION_ERROR;
1427             return;
1428         }
1429 
1430         buffer = sink.GetAppendBuffer(
1431                 /*min_capacity=*/scratch_capacity,
1432                 /*desired_capacity_hint=*/scratch_capacity,
1433                 scratch.getAlias(),
1434                 scratch_capacity,
1435                 &result_capacity);
1436 
1437         reslen = uloc_getKeywordValue(
1438                 fullName,
1439                 keywordName_nul.data(),
1440                 buffer,
1441                 result_capacity,
1442                 &status);
1443 
1444         if (status != U_BUFFER_OVERFLOW_ERROR) {
1445             break;
1446         }
1447 
1448         scratch_capacity = reslen;
1449         status = U_ZERO_ERROR;
1450     }
1451 
1452     if (U_FAILURE(status)) {
1453         return;
1454     }
1455 
1456     sink.Append(buffer, reslen);
1457     if (status == U_STRING_NOT_TERMINATED_WARNING) {
1458         status = U_ZERO_ERROR;  // Terminators not used.
1459     }
1460 }
1461 
1462 void
getUnicodeKeywordValue(StringPiece keywordName,ByteSink & sink,UErrorCode & status) const1463 Locale::getUnicodeKeywordValue(StringPiece keywordName,
1464                                ByteSink& sink,
1465                                UErrorCode& status) const {
1466     // TODO: Remove the need for a const char* to a NUL terminated buffer.
1467     const CharString keywordName_nul(keywordName, status);
1468     if (U_FAILURE(status)) {
1469         return;
1470     }
1471 
1472     const char* legacy_key = uloc_toLegacyKey(keywordName_nul.data());
1473 
1474     if (legacy_key == nullptr) {
1475         status = U_ILLEGAL_ARGUMENT_ERROR;
1476         return;
1477     }
1478 
1479     CharString legacy_value;
1480     {
1481         CharStringByteSink sink(&legacy_value);
1482         getKeywordValue(legacy_key, sink, status);
1483     }
1484 
1485     if (U_FAILURE(status)) {
1486         return;
1487     }
1488 
1489     const char* unicode_value = uloc_toUnicodeLocaleType(
1490             keywordName_nul.data(), legacy_value.data());
1491 
1492     if (unicode_value == nullptr) {
1493         status = U_ILLEGAL_ARGUMENT_ERROR;
1494         return;
1495     }
1496 
1497     sink.Append(unicode_value, static_cast<int32_t>(uprv_strlen(unicode_value)));
1498 }
1499 
1500 void
setKeywordValue(const char * keywordName,const char * keywordValue,UErrorCode & status)1501 Locale::setKeywordValue(const char* keywordName, const char* keywordValue, UErrorCode &status)
1502 {
1503     uloc_setKeywordValue(keywordName, keywordValue, fullName, ULOC_FULLNAME_CAPACITY, &status);
1504     if (U_SUCCESS(status) && baseName == fullName) {
1505         // May have added the first keyword, meaning that the fullName is no longer also the baseName.
1506         initBaseName(status);
1507     }
1508 }
1509 
1510 void
setKeywordValue(StringPiece keywordName,StringPiece keywordValue,UErrorCode & status)1511 Locale::setKeywordValue(StringPiece keywordName,
1512                         StringPiece keywordValue,
1513                         UErrorCode& status) {
1514     // TODO: Remove the need for a const char* to a NUL terminated buffer.
1515     const CharString keywordName_nul(keywordName, status);
1516     const CharString keywordValue_nul(keywordValue, status);
1517     setKeywordValue(keywordName_nul.data(), keywordValue_nul.data(), status);
1518 }
1519 
1520 void
setUnicodeKeywordValue(StringPiece keywordName,StringPiece keywordValue,UErrorCode & status)1521 Locale::setUnicodeKeywordValue(StringPiece keywordName,
1522                                StringPiece keywordValue,
1523                                UErrorCode& status) {
1524     // TODO: Remove the need for a const char* to a NUL terminated buffer.
1525     const CharString keywordName_nul(keywordName, status);
1526     const CharString keywordValue_nul(keywordValue, status);
1527 
1528     if (U_FAILURE(status)) {
1529         return;
1530     }
1531 
1532     const char* legacy_key = uloc_toLegacyKey(keywordName_nul.data());
1533 
1534     if (legacy_key == nullptr) {
1535         status = U_ILLEGAL_ARGUMENT_ERROR;
1536         return;
1537     }
1538 
1539     const char* legacy_value =
1540         uloc_toLegacyType(keywordName_nul.data(), keywordValue_nul.data());
1541 
1542     if (legacy_value == nullptr) {
1543         status = U_ILLEGAL_ARGUMENT_ERROR;
1544         return;
1545     }
1546 
1547     setKeywordValue(legacy_key, legacy_value, status);
1548 }
1549 
1550 const char *
getBaseName() const1551 Locale::getBaseName() const {
1552     return baseName;
1553 }
1554 
1555 //eof
1556 U_NAMESPACE_END
1557