1 // © 2016 and later: Unicode, Inc. and others.
2 // License & terms of use: http://www.unicode.org/copyright.html
3 /*
4 ******************************************************************************
5 * Copyright (C) 1996-2014, International Business Machines Corporation and
6 * others. All Rights Reserved.
7 ******************************************************************************
8 */
9
10 /**
11 * File coll.cpp
12 *
13 * Created by: Helena Shih
14 *
15 * Modification History:
16 *
17 * Date Name Description
18 * 2/5/97 aliu Modified createDefault to load collation data from
19 * binary files when possible. Added related methods
20 * createCollationFromFile, chopLocale, createPathName.
21 * 2/11/97 aliu Added methods addToCache, findInCache, which implement
22 * a Collation cache. Modified createDefault to look in
23 * cache first, and also to store newly created Collation
24 * objects in the cache. Modified to not use gLocPath.
25 * 2/12/97 aliu Modified to create objects from RuleBasedCollator cache.
26 * Moved cache out of Collation class.
27 * 2/13/97 aliu Moved several methods out of this class and into
28 * RuleBasedCollator, with modifications. Modified
29 * createDefault() to call new RuleBasedCollator(Locale&)
30 * constructor. General clean up and documentation.
31 * 2/20/97 helena Added clone, operator==, operator!=, operator=, and copy
32 * constructor.
33 * 05/06/97 helena Added memory allocation error detection.
34 * 05/08/97 helena Added createInstance().
35 * 6/20/97 helena Java class name change.
36 * 04/23/99 stephen Removed EDecompositionMode, merged with
37 * Normalizer::EMode
38 * 11/23/9 srl Inlining of some critical functions
39 * 01/29/01 synwee Modified into a C++ wrapper calling C APIs (ucol.h)
40 * 2012-2014 markus Rewritten in C++ again.
41 */
42
43 #include "utypeinfo.h" // for 'typeid' to work
44
45 #include "unicode/utypes.h"
46
47 #if !UCONFIG_NO_COLLATION
48
49 #include "unicode/coll.h"
50 #include "unicode/tblcoll.h"
51 #include "collationdata.h"
52 #include "collationroot.h"
53 #include "collationtailoring.h"
54 #include "ucol_imp.h"
55 #include "cstring.h"
56 #include "cmemory.h"
57 #include "umutex.h"
58 #include "servloc.h"
59 #include "uassert.h"
60 #include "ustrenum.h"
61 #include "uresimp.h"
62 #include "ucln_in.h"
63
64 static icu::Locale* availableLocaleList = nullptr;
65 static int32_t availableLocaleListCount;
66 #if !UCONFIG_NO_SERVICE
67 static icu::ICULocaleService* gService = nullptr;
68 static icu::UInitOnce gServiceInitOnce {};
69 #endif
70 static icu::UInitOnce gAvailableLocaleListInitOnce {};
71
72 /**
73 * Release all static memory held by collator.
74 */
75 U_CDECL_BEGIN
collator_cleanup()76 static UBool U_CALLCONV collator_cleanup() {
77 #if !UCONFIG_NO_SERVICE
78 if (gService) {
79 delete gService;
80 gService = nullptr;
81 }
82 gServiceInitOnce.reset();
83 #endif
84 if (availableLocaleList) {
85 delete []availableLocaleList;
86 availableLocaleList = nullptr;
87 }
88 availableLocaleListCount = 0;
89 gAvailableLocaleListInitOnce.reset();
90 return true;
91 }
92
93 U_CDECL_END
94
95 U_NAMESPACE_BEGIN
96
97 #if !UCONFIG_NO_SERVICE
98
99 // ------------------------------------------
100 //
101 // Registration
102 //
103
104 //-------------------------------------------
105
~CollatorFactory()106 CollatorFactory::~CollatorFactory() {}
107
108 //-------------------------------------------
109
110 UBool
visible() const111 CollatorFactory::visible() const {
112 return true;
113 }
114
115 //-------------------------------------------
116
117 UnicodeString&
getDisplayName(const Locale & objectLocale,const Locale & displayLocale,UnicodeString & result)118 CollatorFactory::getDisplayName(const Locale& objectLocale,
119 const Locale& displayLocale,
120 UnicodeString& result)
121 {
122 return objectLocale.getDisplayName(displayLocale, result);
123 }
124
125 // -------------------------------------
126
127 class ICUCollatorFactory : public ICUResourceBundleFactory {
128 public:
ICUCollatorFactory()129 ICUCollatorFactory() : ICUResourceBundleFactory(UnicodeString(U_ICUDATA_COLL, -1, US_INV)) { }
130 virtual ~ICUCollatorFactory();
131 protected:
132 virtual UObject* create(const ICUServiceKey& key, const ICUService* service, UErrorCode& status) const override;
133 };
134
~ICUCollatorFactory()135 ICUCollatorFactory::~ICUCollatorFactory() {}
136
137 UObject*
create(const ICUServiceKey & key,const ICUService *,UErrorCode & status) const138 ICUCollatorFactory::create(const ICUServiceKey& key, const ICUService* /* service */, UErrorCode& status) const {
139 if (handlesKey(key, status)) {
140 const LocaleKey& lkey = static_cast<const LocaleKey&>(key);
141 Locale loc;
142 // make sure the requested locale is correct
143 // default LocaleFactory uses currentLocale since that's the one vetted by handlesKey
144 // but for ICU rb resources we use the actual one since it will fallback again
145 lkey.canonicalLocale(loc);
146
147 return Collator::makeInstance(loc, status);
148 }
149 return nullptr;
150 }
151
152 // -------------------------------------
153
154 class ICUCollatorService : public ICULocaleService {
155 public:
ICUCollatorService()156 ICUCollatorService()
157 : ICULocaleService(UNICODE_STRING_SIMPLE("Collator"))
158 {
159 UErrorCode status = U_ZERO_ERROR;
160 registerFactory(new ICUCollatorFactory(), status);
161 }
162
163 virtual ~ICUCollatorService();
164
cloneInstance(UObject * instance) const165 virtual UObject* cloneInstance(UObject* instance) const override {
166 return ((Collator*)instance)->clone();
167 }
168
handleDefault(const ICUServiceKey & key,UnicodeString * actualID,UErrorCode & status) const169 virtual UObject* handleDefault(const ICUServiceKey& key, UnicodeString* actualID, UErrorCode& status) const override {
170 const LocaleKey* lkey = dynamic_cast<const LocaleKey*>(&key);
171 U_ASSERT(lkey != nullptr);
172 if (actualID) {
173 // Ugly Hack Alert! We return an empty actualID to signal
174 // to callers that this is a default object, not a "real"
175 // service-created object. (TODO remove in 3.0) [aliu]
176 actualID->truncate(0);
177 }
178 Locale loc("");
179 lkey->canonicalLocale(loc);
180 return Collator::makeInstance(loc, status);
181 }
182
getKey(ICUServiceKey & key,UnicodeString * actualReturn,UErrorCode & status) const183 virtual UObject* getKey(ICUServiceKey& key, UnicodeString* actualReturn, UErrorCode& status) const override {
184 UnicodeString ar;
185 if (actualReturn == nullptr) {
186 actualReturn = &ar;
187 }
188 return (Collator*)ICULocaleService::getKey(key, actualReturn, status);
189 }
190
isDefault() const191 virtual UBool isDefault() const override {
192 return countFactories() == 1;
193 }
194 };
195
~ICUCollatorService()196 ICUCollatorService::~ICUCollatorService() {}
197
198 // -------------------------------------
199
initService()200 static void U_CALLCONV initService() {
201 gService = new ICUCollatorService();
202 ucln_i18n_registerCleanup(UCLN_I18N_COLLATOR, collator_cleanup);
203 }
204
205
206 static ICULocaleService*
getService()207 getService()
208 {
209 umtx_initOnce(gServiceInitOnce, &initService);
210 return gService;
211 }
212
213 // -------------------------------------
214
215 static inline UBool
hasService()216 hasService()
217 {
218 UBool retVal = !gServiceInitOnce.isReset() && (getService() != nullptr);
219 return retVal;
220 }
221
222 #endif /* UCONFIG_NO_SERVICE */
223
224 static void U_CALLCONV
initAvailableLocaleList(UErrorCode & status)225 initAvailableLocaleList(UErrorCode &status) {
226 U_ASSERT(availableLocaleListCount == 0);
227 U_ASSERT(availableLocaleList == nullptr);
228 // for now, there is a hardcoded list, so just walk through that list and set it up.
229 UResourceBundle *index = nullptr;
230 StackUResourceBundle installed;
231 int32_t i = 0;
232
233 index = ures_openDirect(U_ICUDATA_COLL, "res_index", &status);
234 ures_getByKey(index, "InstalledLocales", installed.getAlias(), &status);
235
236 if(U_SUCCESS(status)) {
237 availableLocaleListCount = ures_getSize(installed.getAlias());
238 availableLocaleList = new Locale[availableLocaleListCount];
239
240 if (availableLocaleList != nullptr) {
241 ures_resetIterator(installed.getAlias());
242 while(ures_hasNext(installed.getAlias())) {
243 const char *tempKey = nullptr;
244 ures_getNextString(installed.getAlias(), nullptr, &tempKey, &status);
245 availableLocaleList[i++] = Locale(tempKey);
246 }
247 }
248 U_ASSERT(availableLocaleListCount == i);
249 }
250 ures_close(index);
251 ucln_i18n_registerCleanup(UCLN_I18N_COLLATOR, collator_cleanup);
252 }
253
isAvailableLocaleListInitialized(UErrorCode & status)254 static UBool isAvailableLocaleListInitialized(UErrorCode &status) {
255 umtx_initOnce(gAvailableLocaleListInitOnce, &initAvailableLocaleList, status);
256 return U_SUCCESS(status);
257 }
258
259
260 // Collator public methods -----------------------------------------------
261
262 namespace {
263
264 static const struct {
265 const char *name;
266 UColAttribute attr;
267 } collAttributes[] = {
268 { "colStrength", UCOL_STRENGTH },
269 { "colBackwards", UCOL_FRENCH_COLLATION },
270 { "colCaseLevel", UCOL_CASE_LEVEL },
271 { "colCaseFirst", UCOL_CASE_FIRST },
272 { "colAlternate", UCOL_ALTERNATE_HANDLING },
273 { "colNormalization", UCOL_NORMALIZATION_MODE },
274 { "colNumeric", UCOL_NUMERIC_COLLATION }
275 };
276
277 static const struct {
278 const char *name;
279 UColAttributeValue value;
280 } collAttributeValues[] = {
281 { "primary", UCOL_PRIMARY },
282 { "secondary", UCOL_SECONDARY },
283 { "tertiary", UCOL_TERTIARY },
284 { "quaternary", UCOL_QUATERNARY },
285 // Note: Not supporting typo "quarternary" because it was never supported in locale IDs.
286 { "identical", UCOL_IDENTICAL },
287 { "no", UCOL_OFF },
288 { "yes", UCOL_ON },
289 { "shifted", UCOL_SHIFTED },
290 { "non-ignorable", UCOL_NON_IGNORABLE },
291 { "lower", UCOL_LOWER_FIRST },
292 { "upper", UCOL_UPPER_FIRST }
293 };
294
295 static const char *collReorderCodes[UCOL_REORDER_CODE_LIMIT - UCOL_REORDER_CODE_FIRST] = {
296 "space", "punct", "symbol", "currency", "digit"
297 };
298
getReorderCode(const char * s)299 int32_t getReorderCode(const char *s) {
300 for (int32_t i = 0; i < UPRV_LENGTHOF(collReorderCodes); ++i) {
301 if (uprv_stricmp(s, collReorderCodes[i]) == 0) {
302 return UCOL_REORDER_CODE_FIRST + i;
303 }
304 }
305 // Not supporting "others" = UCOL_REORDER_CODE_OTHERS
306 // as a synonym for Zzzz = USCRIPT_UNKNOWN for now:
307 // Avoid introducing synonyms/aliases.
308 return -1;
309 }
310
311 /**
312 * Sets collation attributes according to locale keywords. See
313 * http://www.unicode.org/reports/tr35/tr35-collation.html#Collation_Settings
314 *
315 * Using "alias" keywords and values where defined:
316 * http://www.unicode.org/reports/tr35/tr35.html#Old_Locale_Extension_Syntax
317 * http://unicode.org/repos/cldr/trunk/common/bcp47/collation.xml
318 */
setAttributesFromKeywords(const Locale & loc,Collator & coll,UErrorCode & errorCode)319 void setAttributesFromKeywords(const Locale &loc, Collator &coll, UErrorCode &errorCode) {
320 if (U_FAILURE(errorCode)) {
321 return;
322 }
323 if (uprv_strcmp(loc.getName(), loc.getBaseName()) == 0) {
324 // No keywords.
325 return;
326 }
327 char value[1024]; // The reordering value could be long.
328 // Check for collation keywords that were already deprecated
329 // before any were supported in createInstance() (except for "collation").
330 int32_t length = loc.getKeywordValue("colHiraganaQuaternary", value, UPRV_LENGTHOF(value), errorCode);
331 if (U_FAILURE(errorCode)) {
332 errorCode = U_ILLEGAL_ARGUMENT_ERROR;
333 return;
334 }
335 if (length != 0) {
336 errorCode = U_UNSUPPORTED_ERROR;
337 return;
338 }
339 length = loc.getKeywordValue("variableTop", value, UPRV_LENGTHOF(value), errorCode);
340 if (U_FAILURE(errorCode)) {
341 errorCode = U_ILLEGAL_ARGUMENT_ERROR;
342 return;
343 }
344 if (length != 0) {
345 errorCode = U_UNSUPPORTED_ERROR;
346 return;
347 }
348 // Parse known collation keywords, ignore others.
349 if (errorCode == U_STRING_NOT_TERMINATED_WARNING) {
350 errorCode = U_ZERO_ERROR;
351 }
352 for (int32_t i = 0; i < UPRV_LENGTHOF(collAttributes); ++i) {
353 length = loc.getKeywordValue(collAttributes[i].name, value, UPRV_LENGTHOF(value), errorCode);
354 if (U_FAILURE(errorCode) || errorCode == U_STRING_NOT_TERMINATED_WARNING) {
355 errorCode = U_ILLEGAL_ARGUMENT_ERROR;
356 return;
357 }
358 if (length == 0) { continue; }
359 for (int32_t j = 0;; ++j) {
360 if (j == UPRV_LENGTHOF(collAttributeValues)) {
361 errorCode = U_ILLEGAL_ARGUMENT_ERROR;
362 return;
363 }
364 if (uprv_stricmp(value, collAttributeValues[j].name) == 0) {
365 coll.setAttribute(collAttributes[i].attr, collAttributeValues[j].value, errorCode);
366 break;
367 }
368 }
369 }
370 length = loc.getKeywordValue("colReorder", value, UPRV_LENGTHOF(value), errorCode);
371 if (U_FAILURE(errorCode) || errorCode == U_STRING_NOT_TERMINATED_WARNING) {
372 errorCode = U_ILLEGAL_ARGUMENT_ERROR;
373 return;
374 }
375 if (length != 0) {
376 int32_t codes[USCRIPT_CODE_LIMIT + (UCOL_REORDER_CODE_LIMIT - UCOL_REORDER_CODE_FIRST)];
377 int32_t codesLength = 0;
378 char *scriptName = value;
379 for (;;) {
380 if (codesLength == UPRV_LENGTHOF(codes)) {
381 errorCode = U_ILLEGAL_ARGUMENT_ERROR;
382 return;
383 }
384 char *limit = scriptName;
385 char c;
386 while ((c = *limit) != 0 && c != '-') { ++limit; }
387 *limit = 0;
388 int32_t code;
389 if ((limit - scriptName) == 4) {
390 // Strict parsing, accept only 4-letter script codes, not long names.
391 code = u_getPropertyValueEnum(UCHAR_SCRIPT, scriptName);
392 } else {
393 code = getReorderCode(scriptName);
394 }
395 if (code < 0) {
396 errorCode = U_ILLEGAL_ARGUMENT_ERROR;
397 return;
398 }
399 codes[codesLength++] = code;
400 if (c == 0) { break; }
401 scriptName = limit + 1;
402 }
403 coll.setReorderCodes(codes, codesLength, errorCode);
404 }
405 length = loc.getKeywordValue("kv", value, UPRV_LENGTHOF(value), errorCode);
406 if (U_FAILURE(errorCode) || errorCode == U_STRING_NOT_TERMINATED_WARNING) {
407 errorCode = U_ILLEGAL_ARGUMENT_ERROR;
408 return;
409 }
410 if (length != 0) {
411 int32_t code = getReorderCode(value);
412 if (code < 0) {
413 errorCode = U_ILLEGAL_ARGUMENT_ERROR;
414 return;
415 }
416 coll.setMaxVariable((UColReorderCode)code, errorCode);
417 }
418 if (U_FAILURE(errorCode)) {
419 errorCode = U_ILLEGAL_ARGUMENT_ERROR;
420 }
421 }
422
423 } // namespace
424
createInstance(UErrorCode & success)425 Collator* U_EXPORT2 Collator::createInstance(UErrorCode& success)
426 {
427 return createInstance(Locale::getDefault(), success);
428 }
429
createInstance(const Locale & desiredLocale,UErrorCode & status)430 Collator* U_EXPORT2 Collator::createInstance(const Locale& desiredLocale,
431 UErrorCode& status)
432 {
433 if (U_FAILURE(status))
434 return 0;
435 if (desiredLocale.isBogus()) {
436 // Locale constructed from malformed locale ID or language tag.
437 status = U_ILLEGAL_ARGUMENT_ERROR;
438 return nullptr;
439 }
440
441 Collator* coll;
442 #if !UCONFIG_NO_SERVICE
443 if (hasService()) {
444 Locale actualLoc;
445 coll = (Collator*)gService->get(desiredLocale, &actualLoc, status);
446 } else
447 #endif
448 {
449 coll = makeInstance(desiredLocale, status);
450 // Either returns nullptr with U_FAILURE(status), or non-nullptr with U_SUCCESS(status)
451 }
452 // The use of *coll in setAttributesFromKeywords can cause the nullptr check to be
453 // optimized out of the delete even though setAttributesFromKeywords returns
454 // immediately if U_FAILURE(status), so we add a check here.
455 if (U_FAILURE(status)) {
456 return nullptr;
457 }
458 setAttributesFromKeywords(desiredLocale, *coll, status);
459 if (U_FAILURE(status)) {
460 delete coll;
461 return nullptr;
462 }
463 return coll;
464 }
465
466
makeInstance(const Locale & desiredLocale,UErrorCode & status)467 Collator* Collator::makeInstance(const Locale& desiredLocale, UErrorCode& status) {
468 const CollationCacheEntry *entry = CollationLoader::loadTailoring(desiredLocale, status);
469 if (U_SUCCESS(status)) {
470 Collator *result = new RuleBasedCollator(entry);
471 if (result != nullptr) {
472 // Both the unified cache's get() and the RBC constructor
473 // did addRef(). Undo one of them.
474 entry->removeRef();
475 return result;
476 }
477 status = U_MEMORY_ALLOCATION_ERROR;
478 }
479 if (entry != nullptr) {
480 // Undo the addRef() from the cache.get().
481 entry->removeRef();
482 }
483 return nullptr;
484 }
485
486 Collator *
safeClone() const487 Collator::safeClone() const {
488 return clone();
489 }
490
491 // implement deprecated, previously abstract method
compare(const UnicodeString & source,const UnicodeString & target) const492 Collator::EComparisonResult Collator::compare(const UnicodeString& source,
493 const UnicodeString& target) const
494 {
495 UErrorCode ec = U_ZERO_ERROR;
496 return (EComparisonResult)compare(source, target, ec);
497 }
498
499 // implement deprecated, previously abstract method
compare(const UnicodeString & source,const UnicodeString & target,int32_t length) const500 Collator::EComparisonResult Collator::compare(const UnicodeString& source,
501 const UnicodeString& target,
502 int32_t length) const
503 {
504 UErrorCode ec = U_ZERO_ERROR;
505 return (EComparisonResult)compare(source, target, length, ec);
506 }
507
508 // implement deprecated, previously abstract method
compare(const char16_t * source,int32_t sourceLength,const char16_t * target,int32_t targetLength) const509 Collator::EComparisonResult Collator::compare(const char16_t* source, int32_t sourceLength,
510 const char16_t* target, int32_t targetLength)
511 const
512 {
513 UErrorCode ec = U_ZERO_ERROR;
514 return (EComparisonResult)compare(source, sourceLength, target, targetLength, ec);
515 }
516
compare(UCharIterator &,UCharIterator &,UErrorCode & status) const517 UCollationResult Collator::compare(UCharIterator &/*sIter*/,
518 UCharIterator &/*tIter*/,
519 UErrorCode &status) const {
520 if(U_SUCCESS(status)) {
521 // Not implemented in the base class.
522 status = U_UNSUPPORTED_ERROR;
523 }
524 return UCOL_EQUAL;
525 }
526
compareUTF8(const StringPiece & source,const StringPiece & target,UErrorCode & status) const527 UCollationResult Collator::compareUTF8(const StringPiece &source,
528 const StringPiece &target,
529 UErrorCode &status) const {
530 if(U_FAILURE(status)) {
531 return UCOL_EQUAL;
532 }
533 UCharIterator sIter, tIter;
534 uiter_setUTF8(&sIter, source.data(), source.length());
535 uiter_setUTF8(&tIter, target.data(), target.length());
536 return compare(sIter, tIter, status);
537 }
538
equals(const UnicodeString & source,const UnicodeString & target) const539 UBool Collator::equals(const UnicodeString& source,
540 const UnicodeString& target) const
541 {
542 UErrorCode ec = U_ZERO_ERROR;
543 return (compare(source, target, ec) == UCOL_EQUAL);
544 }
545
greaterOrEqual(const UnicodeString & source,const UnicodeString & target) const546 UBool Collator::greaterOrEqual(const UnicodeString& source,
547 const UnicodeString& target) const
548 {
549 UErrorCode ec = U_ZERO_ERROR;
550 return (compare(source, target, ec) != UCOL_LESS);
551 }
552
greater(const UnicodeString & source,const UnicodeString & target) const553 UBool Collator::greater(const UnicodeString& source,
554 const UnicodeString& target) const
555 {
556 UErrorCode ec = U_ZERO_ERROR;
557 return (compare(source, target, ec) == UCOL_GREATER);
558 }
559
560 // this API ignores registered collators, since it returns an
561 // array of indefinite lifetime
getAvailableLocales(int32_t & count)562 const Locale* U_EXPORT2 Collator::getAvailableLocales(int32_t& count)
563 {
564 UErrorCode status = U_ZERO_ERROR;
565 Locale *result = nullptr;
566 count = 0;
567 if (isAvailableLocaleListInitialized(status))
568 {
569 result = availableLocaleList;
570 count = availableLocaleListCount;
571 }
572 return result;
573 }
574
getDisplayName(const Locale & objectLocale,const Locale & displayLocale,UnicodeString & name)575 UnicodeString& U_EXPORT2 Collator::getDisplayName(const Locale& objectLocale,
576 const Locale& displayLocale,
577 UnicodeString& name)
578 {
579 #if !UCONFIG_NO_SERVICE
580 if (hasService()) {
581 UnicodeString locNameStr;
582 LocaleUtility::initNameFromLocale(objectLocale, locNameStr);
583 return gService->getDisplayName(locNameStr, name, displayLocale);
584 }
585 #endif
586 return objectLocale.getDisplayName(displayLocale, name);
587 }
588
getDisplayName(const Locale & objectLocale,UnicodeString & name)589 UnicodeString& U_EXPORT2 Collator::getDisplayName(const Locale& objectLocale,
590 UnicodeString& name)
591 {
592 return getDisplayName(objectLocale, Locale::getDefault(), name);
593 }
594
595 /* This is useless information */
596 /*void Collator::getVersion(UVersionInfo versionInfo) const
597 {
598 if (versionInfo!=nullptr)
599 uprv_memcpy(versionInfo, fVersion, U_MAX_VERSION_LENGTH);
600 }
601 */
602
603 // UCollator protected constructor destructor ----------------------------
604
605 /**
606 * Default constructor.
607 * Constructor is different from the old default Collator constructor.
608 * The task for determining the default collation strength and normalization mode
609 * is left to the child class.
610 */
Collator()611 Collator::Collator()
612 : UObject()
613 {
614 }
615
616 /**
617 * Constructor.
618 * Empty constructor, does not handle the arguments.
619 * This constructor is done for backward compatibility with 1.7 and 1.8.
620 * The task for handling the argument collation strength and normalization
621 * mode is left to the child class.
622 * @param collationStrength collation strength
623 * @param decompositionMode
624 * @deprecated 2.4 use the default constructor instead
625 */
Collator(UCollationStrength,UNormalizationMode)626 Collator::Collator(UCollationStrength, UNormalizationMode )
627 : UObject()
628 {
629 }
630
~Collator()631 Collator::~Collator()
632 {
633 }
634
Collator(const Collator & other)635 Collator::Collator(const Collator &other)
636 : UObject(other)
637 {
638 }
639
operator ==(const Collator & other) const640 bool Collator::operator==(const Collator& other) const
641 {
642 // Subclasses: Call this method and then add more specific checks.
643 return typeid(*this) == typeid(other);
644 }
645
operator !=(const Collator & other) const646 bool Collator::operator!=(const Collator& other) const
647 {
648 return !operator==(other);
649 }
650
getBound(const uint8_t * source,int32_t sourceLength,UColBoundMode boundType,uint32_t noOfLevels,uint8_t * result,int32_t resultLength,UErrorCode & status)651 int32_t U_EXPORT2 Collator::getBound(const uint8_t *source,
652 int32_t sourceLength,
653 UColBoundMode boundType,
654 uint32_t noOfLevels,
655 uint8_t *result,
656 int32_t resultLength,
657 UErrorCode &status)
658 {
659 return ucol_getBound(source, sourceLength, boundType, noOfLevels, result, resultLength, &status);
660 }
661
662 void
setLocales(const Locale &,const Locale &,const Locale &)663 Collator::setLocales(const Locale& /* requestedLocale */, const Locale& /* validLocale */, const Locale& /*actualLocale*/) {
664 }
665
getTailoredSet(UErrorCode & status) const666 UnicodeSet *Collator::getTailoredSet(UErrorCode &status) const
667 {
668 if(U_FAILURE(status)) {
669 return nullptr;
670 }
671 // everything can be changed
672 return new UnicodeSet(0, 0x10FFFF);
673 }
674
675 // -------------------------------------
676
677 #if !UCONFIG_NO_SERVICE
678 URegistryKey U_EXPORT2
registerInstance(Collator * toAdopt,const Locale & locale,UErrorCode & status)679 Collator::registerInstance(Collator* toAdopt, const Locale& locale, UErrorCode& status)
680 {
681 if (U_SUCCESS(status)) {
682 // Set the collator locales while registering so that createInstance()
683 // need not guess whether the collator's locales are already set properly
684 // (as they are by the data loader).
685 toAdopt->setLocales(locale, locale, locale);
686 return getService()->registerInstance(toAdopt, locale, status);
687 }
688 return nullptr;
689 }
690
691 // -------------------------------------
692
693 class CFactory : public LocaleKeyFactory {
694 private:
695 CollatorFactory* _delegate;
696 Hashtable* _ids;
697
698 public:
CFactory(CollatorFactory * delegate,UErrorCode & status)699 CFactory(CollatorFactory* delegate, UErrorCode& status)
700 : LocaleKeyFactory(delegate->visible() ? VISIBLE : INVISIBLE)
701 , _delegate(delegate)
702 , _ids(nullptr)
703 {
704 if (U_SUCCESS(status)) {
705 int32_t count = 0;
706 _ids = new Hashtable(status);
707 if (_ids) {
708 const UnicodeString * idlist = _delegate->getSupportedIDs(count, status);
709 for (int i = 0; i < count; ++i) {
710 _ids->put(idlist[i], (void*)this, status);
711 if (U_FAILURE(status)) {
712 delete _ids;
713 _ids = nullptr;
714 return;
715 }
716 }
717 } else {
718 status = U_MEMORY_ALLOCATION_ERROR;
719 }
720 }
721 }
722
723 virtual ~CFactory();
724
725 virtual UObject* create(const ICUServiceKey& key, const ICUService* service, UErrorCode& status) const override;
726
727 protected:
getSupportedIDs(UErrorCode & status) const728 virtual const Hashtable* getSupportedIDs(UErrorCode& status) const override
729 {
730 if (U_SUCCESS(status)) {
731 return _ids;
732 }
733 return nullptr;
734 }
735
736 virtual UnicodeString&
737 getDisplayName(const UnicodeString& id, const Locale& locale, UnicodeString& result) const override;
738 };
739
~CFactory()740 CFactory::~CFactory()
741 {
742 delete _delegate;
743 delete _ids;
744 }
745
746 UObject*
create(const ICUServiceKey & key,const ICUService *,UErrorCode & status) const747 CFactory::create(const ICUServiceKey& key, const ICUService* /* service */, UErrorCode& status) const
748 {
749 if (handlesKey(key, status)) {
750 const LocaleKey* lkey = dynamic_cast<const LocaleKey*>(&key);
751 U_ASSERT(lkey != nullptr);
752 Locale validLoc;
753 lkey->currentLocale(validLoc);
754 return _delegate->createCollator(validLoc);
755 }
756 return nullptr;
757 }
758
759 UnicodeString&
getDisplayName(const UnicodeString & id,const Locale & locale,UnicodeString & result) const760 CFactory::getDisplayName(const UnicodeString& id, const Locale& locale, UnicodeString& result) const
761 {
762 if ((_coverage & 0x1) == 0) {
763 UErrorCode status = U_ZERO_ERROR;
764 const Hashtable* ids = getSupportedIDs(status);
765 if (ids && (ids->get(id) != nullptr)) {
766 Locale loc;
767 LocaleUtility::initLocaleFromName(id, loc);
768 return _delegate->getDisplayName(loc, locale, result);
769 }
770 }
771 result.setToBogus();
772 return result;
773 }
774
775 URegistryKey U_EXPORT2
registerFactory(CollatorFactory * toAdopt,UErrorCode & status)776 Collator::registerFactory(CollatorFactory* toAdopt, UErrorCode& status)
777 {
778 if (U_SUCCESS(status)) {
779 CFactory* f = new CFactory(toAdopt, status);
780 if (f) {
781 return getService()->registerFactory(f, status);
782 }
783 status = U_MEMORY_ALLOCATION_ERROR;
784 }
785 return nullptr;
786 }
787
788 // -------------------------------------
789
790 UBool U_EXPORT2
unregister(URegistryKey key,UErrorCode & status)791 Collator::unregister(URegistryKey key, UErrorCode& status)
792 {
793 if (U_SUCCESS(status)) {
794 if (hasService()) {
795 return gService->unregister(key, status);
796 }
797 status = U_ILLEGAL_ARGUMENT_ERROR;
798 }
799 return false;
800 }
801 #endif /* UCONFIG_NO_SERVICE */
802
803 class CollationLocaleListEnumeration : public StringEnumeration {
804 private:
805 int32_t index;
806 public:
807 static UClassID U_EXPORT2 getStaticClassID();
808 virtual UClassID getDynamicClassID() const override;
809 public:
CollationLocaleListEnumeration()810 CollationLocaleListEnumeration()
811 : index(0)
812 {
813 // The global variables should already be initialized.
814 //isAvailableLocaleListInitialized(status);
815 }
816
817 virtual ~CollationLocaleListEnumeration();
818
clone() const819 virtual StringEnumeration * clone() const override
820 {
821 CollationLocaleListEnumeration *result = new CollationLocaleListEnumeration();
822 if (result) {
823 result->index = index;
824 }
825 return result;
826 }
827
count(UErrorCode &) const828 virtual int32_t count(UErrorCode &/*status*/) const override {
829 return availableLocaleListCount;
830 }
831
next(int32_t * resultLength,UErrorCode &)832 virtual const char* next(int32_t* resultLength, UErrorCode& /*status*/) override {
833 const char* result;
834 if(index < availableLocaleListCount) {
835 result = availableLocaleList[index++].getName();
836 if(resultLength != nullptr) {
837 *resultLength = (int32_t)uprv_strlen(result);
838 }
839 } else {
840 if(resultLength != nullptr) {
841 *resultLength = 0;
842 }
843 result = nullptr;
844 }
845 return result;
846 }
847
snext(UErrorCode & status)848 virtual const UnicodeString* snext(UErrorCode& status) override {
849 int32_t resultLength = 0;
850 const char *s = next(&resultLength, status);
851 return setChars(s, resultLength, status);
852 }
853
reset(UErrorCode &)854 virtual void reset(UErrorCode& /*status*/) override {
855 index = 0;
856 }
857 };
858
~CollationLocaleListEnumeration()859 CollationLocaleListEnumeration::~CollationLocaleListEnumeration() {}
860
UOBJECT_DEFINE_RTTI_IMPLEMENTATION(CollationLocaleListEnumeration)861 UOBJECT_DEFINE_RTTI_IMPLEMENTATION(CollationLocaleListEnumeration)
862
863
864 // -------------------------------------
865
866 StringEnumeration* U_EXPORT2
867 Collator::getAvailableLocales()
868 {
869 #if !UCONFIG_NO_SERVICE
870 if (hasService()) {
871 return getService()->getAvailableLocales();
872 }
873 #endif /* UCONFIG_NO_SERVICE */
874 UErrorCode status = U_ZERO_ERROR;
875 if (isAvailableLocaleListInitialized(status)) {
876 return new CollationLocaleListEnumeration();
877 }
878 return nullptr;
879 }
880
881 StringEnumeration* U_EXPORT2
getKeywords(UErrorCode & status)882 Collator::getKeywords(UErrorCode& status) {
883 return UStringEnumeration::fromUEnumeration(
884 ucol_getKeywords(&status), status);
885 }
886
887 StringEnumeration* U_EXPORT2
getKeywordValues(const char * keyword,UErrorCode & status)888 Collator::getKeywordValues(const char *keyword, UErrorCode& status) {
889 return UStringEnumeration::fromUEnumeration(
890 ucol_getKeywordValues(keyword, &status), status);
891 }
892
893 StringEnumeration* U_EXPORT2
getKeywordValuesForLocale(const char * key,const Locale & locale,UBool commonlyUsed,UErrorCode & status)894 Collator::getKeywordValuesForLocale(const char* key, const Locale& locale,
895 UBool commonlyUsed, UErrorCode& status) {
896 return UStringEnumeration::fromUEnumeration(
897 ucol_getKeywordValuesForLocale(
898 key, locale.getName(), commonlyUsed, &status),
899 status);
900 }
901
902 Locale U_EXPORT2
getFunctionalEquivalent(const char * keyword,const Locale & locale,UBool & isAvailable,UErrorCode & status)903 Collator::getFunctionalEquivalent(const char* keyword, const Locale& locale,
904 UBool& isAvailable, UErrorCode& status) {
905 // This is a wrapper over ucol_getFunctionalEquivalent
906 char loc[ULOC_FULLNAME_CAPACITY];
907 /*int32_t len =*/ ucol_getFunctionalEquivalent(loc, sizeof(loc),
908 keyword, locale.getName(), &isAvailable, &status);
909 if (U_FAILURE(status)) {
910 *loc = 0; // root
911 }
912 return Locale::createFromName(loc);
913 }
914
915 Collator::ECollationStrength
getStrength() const916 Collator::getStrength() const {
917 UErrorCode intStatus = U_ZERO_ERROR;
918 return (ECollationStrength)getAttribute(UCOL_STRENGTH, intStatus);
919 }
920
921 void
setStrength(ECollationStrength newStrength)922 Collator::setStrength(ECollationStrength newStrength) {
923 UErrorCode intStatus = U_ZERO_ERROR;
924 setAttribute(UCOL_STRENGTH, (UColAttributeValue)newStrength, intStatus);
925 }
926
927 Collator &
setMaxVariable(UColReorderCode,UErrorCode & errorCode)928 Collator::setMaxVariable(UColReorderCode /*group*/, UErrorCode &errorCode) {
929 if (U_SUCCESS(errorCode)) {
930 errorCode = U_UNSUPPORTED_ERROR;
931 }
932 return *this;
933 }
934
935 UColReorderCode
getMaxVariable() const936 Collator::getMaxVariable() const {
937 return UCOL_REORDER_CODE_PUNCTUATION;
938 }
939
940 int32_t
getReorderCodes(int32_t *,int32_t,UErrorCode & status) const941 Collator::getReorderCodes(int32_t* /* dest*/,
942 int32_t /* destCapacity*/,
943 UErrorCode& status) const
944 {
945 if (U_SUCCESS(status)) {
946 status = U_UNSUPPORTED_ERROR;
947 }
948 return 0;
949 }
950
951 void
setReorderCodes(const int32_t *,int32_t,UErrorCode & status)952 Collator::setReorderCodes(const int32_t* /* reorderCodes */,
953 int32_t /* reorderCodesLength */,
954 UErrorCode& status)
955 {
956 if (U_SUCCESS(status)) {
957 status = U_UNSUPPORTED_ERROR;
958 }
959 }
960
961 int32_t
getEquivalentReorderCodes(int32_t reorderCode,int32_t * dest,int32_t capacity,UErrorCode & errorCode)962 Collator::getEquivalentReorderCodes(int32_t reorderCode,
963 int32_t *dest, int32_t capacity,
964 UErrorCode &errorCode) {
965 if(U_FAILURE(errorCode)) { return 0; }
966 if(capacity < 0 || (dest == nullptr && capacity > 0)) {
967 errorCode = U_ILLEGAL_ARGUMENT_ERROR;
968 return 0;
969 }
970 const CollationData *baseData = CollationRoot::getData(errorCode);
971 if(U_FAILURE(errorCode)) { return 0; }
972 return baseData->getEquivalentScripts(reorderCode, dest, capacity, errorCode);
973 }
974
975 int32_t
internalGetShortDefinitionString(const char *,char *,int32_t,UErrorCode & status) const976 Collator::internalGetShortDefinitionString(const char * /*locale*/,
977 char * /*buffer*/,
978 int32_t /*capacity*/,
979 UErrorCode &status) const {
980 if(U_SUCCESS(status)) {
981 status = U_UNSUPPORTED_ERROR; /* Shouldn't happen, internal function */
982 }
983 return 0;
984 }
985
986 UCollationResult
internalCompareUTF8(const char * left,int32_t leftLength,const char * right,int32_t rightLength,UErrorCode & errorCode) const987 Collator::internalCompareUTF8(const char *left, int32_t leftLength,
988 const char *right, int32_t rightLength,
989 UErrorCode &errorCode) const {
990 if(U_FAILURE(errorCode)) { return UCOL_EQUAL; }
991 if((left == nullptr && leftLength != 0) || (right == nullptr && rightLength != 0)) {
992 errorCode = U_ILLEGAL_ARGUMENT_ERROR;
993 return UCOL_EQUAL;
994 }
995 return compareUTF8(
996 StringPiece(left, (leftLength < 0) ? static_cast<int32_t>(uprv_strlen(left)) : leftLength),
997 StringPiece(right, (rightLength < 0) ? static_cast<int32_t>(uprv_strlen(right)) : rightLength),
998 errorCode);
999 }
1000
1001 int32_t
internalNextSortKeyPart(UCharIterator *,uint32_t[2],uint8_t *,int32_t,UErrorCode & errorCode) const1002 Collator::internalNextSortKeyPart(UCharIterator * /*iter*/, uint32_t /*state*/[2],
1003 uint8_t * /*dest*/, int32_t /*count*/, UErrorCode &errorCode) const {
1004 if (U_SUCCESS(errorCode)) {
1005 errorCode = U_UNSUPPORTED_ERROR;
1006 }
1007 return 0;
1008 }
1009
1010 // UCollator private data members ----------------------------------------
1011
1012 /* This is useless information */
1013 /*const UVersionInfo Collator::fVersion = {1, 1, 0, 0};*/
1014
1015 // -------------------------------------
1016
1017 U_NAMESPACE_END
1018
1019 #endif /* #if !UCONFIG_NO_COLLATION */
1020
1021 /* eof */
1022