1 /*
2 *******************************************************************************
3 * Copyright (C) 2007-2015, International Business Machines Corporation and
4 * others. All Rights Reserved.
5 *******************************************************************************
6 *
7 * File plurrule.cpp
8 */
9
10 #include <math.h>
11 #include <stdio.h>
12
13 #include "unicode/utypes.h"
14 #include "unicode/localpointer.h"
15 #include "unicode/plurrule.h"
16 #include "unicode/upluralrules.h"
17 #include "unicode/ures.h"
18 #include "charstr.h"
19 #include "cmemory.h"
20 #include "cstring.h"
21 #include "digitlst.h"
22 #include "hash.h"
23 #include "locutil.h"
24 #include "mutex.h"
25 #include "patternprops.h"
26 #include "plurrule_impl.h"
27 #include "putilimp.h"
28 #include "ucln_in.h"
29 #include "ustrfmt.h"
30 #include "uassert.h"
31 #include "uvectr32.h"
32 #include "sharedpluralrules.h"
33 #include "unifiedcache.h"
34 #include "digitinterval.h"
35 #include "visibledigits.h"
36
37
38 #if !UCONFIG_NO_FORMATTING
39
40 U_NAMESPACE_BEGIN
41
42 #define ARRAY_SIZE(array) (int32_t)(sizeof array / sizeof array[0])
43
44 static const UChar PLURAL_KEYWORD_OTHER[]={LOW_O,LOW_T,LOW_H,LOW_E,LOW_R,0};
45 static const UChar PLURAL_DEFAULT_RULE[]={LOW_O,LOW_T,LOW_H,LOW_E,LOW_R,COLON,SPACE,LOW_N,0};
46 static const UChar PK_IN[]={LOW_I,LOW_N,0};
47 static const UChar PK_NOT[]={LOW_N,LOW_O,LOW_T,0};
48 static const UChar PK_IS[]={LOW_I,LOW_S,0};
49 static const UChar PK_MOD[]={LOW_M,LOW_O,LOW_D,0};
50 static const UChar PK_AND[]={LOW_A,LOW_N,LOW_D,0};
51 static const UChar PK_OR[]={LOW_O,LOW_R,0};
52 static const UChar PK_VAR_N[]={LOW_N,0};
53 static const UChar PK_VAR_I[]={LOW_I,0};
54 static const UChar PK_VAR_F[]={LOW_F,0};
55 static const UChar PK_VAR_T[]={LOW_T,0};
56 static const UChar PK_VAR_V[]={LOW_V,0};
57 static const UChar PK_WITHIN[]={LOW_W,LOW_I,LOW_T,LOW_H,LOW_I,LOW_N,0};
58 static const UChar PK_DECIMAL[]={LOW_D,LOW_E,LOW_C,LOW_I,LOW_M,LOW_A,LOW_L,0};
59 static const UChar PK_INTEGER[]={LOW_I,LOW_N,LOW_T,LOW_E,LOW_G,LOW_E,LOW_R,0};
60
61 UOBJECT_DEFINE_RTTI_IMPLEMENTATION(PluralRules)
UOBJECT_DEFINE_RTTI_IMPLEMENTATION(PluralKeywordEnumeration)62 UOBJECT_DEFINE_RTTI_IMPLEMENTATION(PluralKeywordEnumeration)
63
64 PluralRules::PluralRules(UErrorCode& /*status*/)
65 : UObject(),
66 mRules(NULL)
67 {
68 }
69
PluralRules(const PluralRules & other)70 PluralRules::PluralRules(const PluralRules& other)
71 : UObject(other),
72 mRules(NULL)
73 {
74 *this=other;
75 }
76
~PluralRules()77 PluralRules::~PluralRules() {
78 delete mRules;
79 }
80
~SharedPluralRules()81 SharedPluralRules::~SharedPluralRules() {
82 delete ptr;
83 }
84
85 PluralRules*
clone() const86 PluralRules::clone() const {
87 return new PluralRules(*this);
88 }
89
90 PluralRules&
operator =(const PluralRules & other)91 PluralRules::operator=(const PluralRules& other) {
92 if (this != &other) {
93 delete mRules;
94 if (other.mRules==NULL) {
95 mRules = NULL;
96 }
97 else {
98 mRules = new RuleChain(*other.mRules);
99 }
100 }
101
102 return *this;
103 }
104
getAvailableLocales(UErrorCode & status)105 StringEnumeration* PluralRules::getAvailableLocales(UErrorCode &status) {
106 StringEnumeration *result = new PluralAvailableLocalesEnumeration(status);
107 if (result == NULL && U_SUCCESS(status)) {
108 status = U_MEMORY_ALLOCATION_ERROR;
109 }
110 if (U_FAILURE(status)) {
111 delete result;
112 result = NULL;
113 }
114 return result;
115 }
116
117
118 PluralRules* U_EXPORT2
createRules(const UnicodeString & description,UErrorCode & status)119 PluralRules::createRules(const UnicodeString& description, UErrorCode& status) {
120 if (U_FAILURE(status)) {
121 return NULL;
122 }
123
124 PluralRuleParser parser;
125 PluralRules *newRules = new PluralRules(status);
126 if (U_SUCCESS(status) && newRules == NULL) {
127 status = U_MEMORY_ALLOCATION_ERROR;
128 }
129 parser.parse(description, newRules, status);
130 if (U_FAILURE(status)) {
131 delete newRules;
132 newRules = NULL;
133 }
134 return newRules;
135 }
136
137
138 PluralRules* U_EXPORT2
createDefaultRules(UErrorCode & status)139 PluralRules::createDefaultRules(UErrorCode& status) {
140 return createRules(UnicodeString(TRUE, PLURAL_DEFAULT_RULE, -1), status);
141 }
142
143 /******************************************************************************/
144 /* Create PluralRules cache */
145
146 template<> U_I18N_API
createObject(const void *,UErrorCode & status) const147 const SharedPluralRules *LocaleCacheKey<SharedPluralRules>::createObject(
148 const void * /*unused*/, UErrorCode &status) const {
149 const char *localeId = fLoc.getName();
150 PluralRules *pr = PluralRules::internalForLocale(
151 localeId, UPLURAL_TYPE_CARDINAL, status);
152 if (U_FAILURE(status)) {
153 return NULL;
154 }
155 SharedPluralRules *result = new SharedPluralRules(pr);
156 if (result == NULL) {
157 status = U_MEMORY_ALLOCATION_ERROR;
158 delete pr;
159 return NULL;
160 }
161 result->addRef();
162 return result;
163 }
164
165 /* end plural rules cache */
166 /******************************************************************************/
167
168 const SharedPluralRules* U_EXPORT2
createSharedInstance(const Locale & locale,UPluralType type,UErrorCode & status)169 PluralRules::createSharedInstance(
170 const Locale& locale, UPluralType type, UErrorCode& status) {
171 if (U_FAILURE(status)) {
172 return NULL;
173 }
174 if (type != UPLURAL_TYPE_CARDINAL) {
175 status = U_UNSUPPORTED_ERROR;
176 return NULL;
177 }
178 const SharedPluralRules *result = NULL;
179 UnifiedCache::getByLocale(locale, result, status);
180 return result;
181 }
182
183 PluralRules* U_EXPORT2
forLocale(const Locale & locale,UErrorCode & status)184 PluralRules::forLocale(const Locale& locale, UErrorCode& status) {
185 return forLocale(locale, UPLURAL_TYPE_CARDINAL, status);
186 }
187
188 PluralRules* U_EXPORT2
forLocale(const Locale & locale,UPluralType type,UErrorCode & status)189 PluralRules::forLocale(const Locale& locale, UPluralType type, UErrorCode& status) {
190 if (type != UPLURAL_TYPE_CARDINAL) {
191 return internalForLocale(locale, type, status);
192 }
193 const SharedPluralRules *shared = createSharedInstance(
194 locale, type, status);
195 if (U_FAILURE(status)) {
196 return NULL;
197 }
198 PluralRules *result = (*shared)->clone();
199 shared->removeRef();
200 if (result == NULL) {
201 status = U_MEMORY_ALLOCATION_ERROR;
202 }
203 return result;
204 }
205
206 PluralRules* U_EXPORT2
internalForLocale(const Locale & locale,UPluralType type,UErrorCode & status)207 PluralRules::internalForLocale(const Locale& locale, UPluralType type, UErrorCode& status) {
208 if (U_FAILURE(status)) {
209 return NULL;
210 }
211 if (type >= UPLURAL_TYPE_COUNT) {
212 status = U_ILLEGAL_ARGUMENT_ERROR;
213 return NULL;
214 }
215 PluralRules *newObj = new PluralRules(status);
216 if (newObj==NULL || U_FAILURE(status)) {
217 delete newObj;
218 return NULL;
219 }
220 UnicodeString locRule = newObj->getRuleFromResource(locale, type, status);
221 // TODO: which errors, if any, should be returned?
222 if (locRule.length() == 0) {
223 // Locales with no specific rules (all numbers have the "other" category
224 // will return a U_MISSING_RESOURCE_ERROR at this point. This is not
225 // an error.
226 locRule = UnicodeString(PLURAL_DEFAULT_RULE);
227 status = U_ZERO_ERROR;
228 }
229 PluralRuleParser parser;
230 parser.parse(locRule, newObj, status);
231 // TODO: should rule parse errors be returned, or
232 // should we silently use default rules?
233 // Original impl used default rules.
234 // Ask the question to ICU Core.
235
236 return newObj;
237 }
238
239 UnicodeString
select(int32_t number) const240 PluralRules::select(int32_t number) const {
241 return select(FixedDecimal(number));
242 }
243
244 UnicodeString
select(double number) const245 PluralRules::select(double number) const {
246 return select(FixedDecimal(number));
247 }
248
249 UnicodeString
select(const FixedDecimal & number) const250 PluralRules::select(const FixedDecimal &number) const {
251 if (mRules == NULL) {
252 return UnicodeString(TRUE, PLURAL_DEFAULT_RULE, -1);
253 }
254 else {
255 return mRules->select(number);
256 }
257 }
258
259 UnicodeString
select(const VisibleDigitsWithExponent & number) const260 PluralRules::select(const VisibleDigitsWithExponent &number) const {
261 if (number.getExponent() != NULL) {
262 return UnicodeString(TRUE, PLURAL_DEFAULT_RULE, -1);
263 }
264 return select(FixedDecimal(number.getMantissa()));
265 }
266
267
268
269 StringEnumeration*
getKeywords(UErrorCode & status) const270 PluralRules::getKeywords(UErrorCode& status) const {
271 if (U_FAILURE(status)) return NULL;
272 StringEnumeration* nameEnumerator = new PluralKeywordEnumeration(mRules, status);
273 if (U_FAILURE(status)) {
274 delete nameEnumerator;
275 return NULL;
276 }
277
278 return nameEnumerator;
279 }
280
281 double
getUniqueKeywordValue(const UnicodeString &)282 PluralRules::getUniqueKeywordValue(const UnicodeString& /* keyword */) {
283 // Not Implemented.
284 return UPLRULES_NO_UNIQUE_VALUE;
285 }
286
287 int32_t
getAllKeywordValues(const UnicodeString &,double *,int32_t,UErrorCode & error)288 PluralRules::getAllKeywordValues(const UnicodeString & /* keyword */, double * /* dest */,
289 int32_t /* destCapacity */, UErrorCode& error) {
290 error = U_UNSUPPORTED_ERROR;
291 return 0;
292 }
293
294
scaleForInt(double d)295 static double scaleForInt(double d) {
296 double scale = 1.0;
297 while (d != floor(d)) {
298 d = d * 10.0;
299 scale = scale * 10.0;
300 }
301 return scale;
302 }
303
304 static int32_t
getSamplesFromString(const UnicodeString & samples,double * dest,int32_t destCapacity,UErrorCode & status)305 getSamplesFromString(const UnicodeString &samples, double *dest,
306 int32_t destCapacity, UErrorCode& status) {
307 int32_t sampleCount = 0;
308 int32_t sampleStartIdx = 0;
309 int32_t sampleEndIdx = 0;
310
311 //std::string ss; // TODO: debugging.
312 // std::cout << "PluralRules::getSamples(), samples = \"" << samples.toUTF8String(ss) << "\"\n";
313 for (sampleCount = 0; sampleCount < destCapacity && sampleStartIdx < samples.length(); ) {
314 sampleEndIdx = samples.indexOf(COMMA, sampleStartIdx);
315 if (sampleEndIdx == -1) {
316 sampleEndIdx = samples.length();
317 }
318 const UnicodeString &sampleRange = samples.tempSubStringBetween(sampleStartIdx, sampleEndIdx);
319 // ss.erase();
320 // std::cout << "PluralRules::getSamples(), samplesRange = \"" << sampleRange.toUTF8String(ss) << "\"\n";
321 int32_t tildeIndex = sampleRange.indexOf(TILDE);
322 if (tildeIndex < 0) {
323 FixedDecimal fixed(sampleRange, status);
324 double sampleValue = fixed.source;
325 if (fixed.visibleDecimalDigitCount == 0 || sampleValue != floor(sampleValue)) {
326 dest[sampleCount++] = sampleValue;
327 }
328 } else {
329
330 FixedDecimal fixedLo(sampleRange.tempSubStringBetween(0, tildeIndex), status);
331 FixedDecimal fixedHi(sampleRange.tempSubStringBetween(tildeIndex+1), status);
332 double rangeLo = fixedLo.source;
333 double rangeHi = fixedHi.source;
334 if (U_FAILURE(status)) {
335 break;
336 }
337 if (rangeHi < rangeLo) {
338 status = U_INVALID_FORMAT_ERROR;
339 break;
340 }
341
342 // For ranges of samples with fraction decimal digits, scale the number up so that we
343 // are adding one in the units place. Avoids roundoffs from repetitive adds of tenths.
344
345 double scale = scaleForInt(rangeLo);
346 double t = scaleForInt(rangeHi);
347 if (t > scale) {
348 scale = t;
349 }
350 rangeLo *= scale;
351 rangeHi *= scale;
352 for (double n=rangeLo; n<=rangeHi; n+=1) {
353 // Hack Alert: don't return any decimal samples with integer values that
354 // originated from a format with trailing decimals.
355 // This API is returning doubles, which can't distinguish having displayed
356 // zeros to the right of the decimal.
357 // This results in test failures with values mapping back to a different keyword.
358 double sampleValue = n/scale;
359 if (!(sampleValue == floor(sampleValue) && fixedLo.visibleDecimalDigitCount > 0)) {
360 dest[sampleCount++] = sampleValue;
361 }
362 if (sampleCount >= destCapacity) {
363 break;
364 }
365 }
366 }
367 sampleStartIdx = sampleEndIdx + 1;
368 }
369 return sampleCount;
370 }
371
372
373 int32_t
getSamples(const UnicodeString & keyword,double * dest,int32_t destCapacity,UErrorCode & status)374 PluralRules::getSamples(const UnicodeString &keyword, double *dest,
375 int32_t destCapacity, UErrorCode& status) {
376 RuleChain *rc = rulesForKeyword(keyword);
377 if (rc == NULL || destCapacity == 0 || U_FAILURE(status)) {
378 return 0;
379 }
380 int32_t numSamples = getSamplesFromString(rc->fIntegerSamples, dest, destCapacity, status);
381 if (numSamples == 0) {
382 numSamples = getSamplesFromString(rc->fDecimalSamples, dest, destCapacity, status);
383 }
384 return numSamples;
385 }
386
387
rulesForKeyword(const UnicodeString & keyword) const388 RuleChain *PluralRules::rulesForKeyword(const UnicodeString &keyword) const {
389 RuleChain *rc;
390 for (rc = mRules; rc != NULL; rc = rc->fNext) {
391 if (rc->fKeyword == keyword) {
392 break;
393 }
394 }
395 return rc;
396 }
397
398
399 UBool
isKeyword(const UnicodeString & keyword) const400 PluralRules::isKeyword(const UnicodeString& keyword) const {
401 if (0 == keyword.compare(PLURAL_KEYWORD_OTHER, 5)) {
402 return true;
403 }
404 return rulesForKeyword(keyword) != NULL;
405 }
406
407 UnicodeString
getKeywordOther() const408 PluralRules::getKeywordOther() const {
409 return UnicodeString(TRUE, PLURAL_KEYWORD_OTHER, 5);
410 }
411
412 UBool
operator ==(const PluralRules & other) const413 PluralRules::operator==(const PluralRules& other) const {
414 const UnicodeString *ptrKeyword;
415 UErrorCode status= U_ZERO_ERROR;
416
417 if ( this == &other ) {
418 return TRUE;
419 }
420 LocalPointer<StringEnumeration> myKeywordList(getKeywords(status));
421 LocalPointer<StringEnumeration> otherKeywordList(other.getKeywords(status));
422 if (U_FAILURE(status)) {
423 return FALSE;
424 }
425
426 if (myKeywordList->count(status)!=otherKeywordList->count(status)) {
427 return FALSE;
428 }
429 myKeywordList->reset(status);
430 while ((ptrKeyword=myKeywordList->snext(status))!=NULL) {
431 if (!other.isKeyword(*ptrKeyword)) {
432 return FALSE;
433 }
434 }
435 otherKeywordList->reset(status);
436 while ((ptrKeyword=otherKeywordList->snext(status))!=NULL) {
437 if (!this->isKeyword(*ptrKeyword)) {
438 return FALSE;
439 }
440 }
441 if (U_FAILURE(status)) {
442 return FALSE;
443 }
444
445 return TRUE;
446 }
447
448
449 void
parse(const UnicodeString & ruleData,PluralRules * prules,UErrorCode & status)450 PluralRuleParser::parse(const UnicodeString& ruleData, PluralRules *prules, UErrorCode &status)
451 {
452 if (U_FAILURE(status)) {
453 return;
454 }
455 U_ASSERT(ruleIndex == 0); // Parsers are good for a single use only!
456 ruleSrc = &ruleData;
457
458 while (ruleIndex< ruleSrc->length()) {
459 getNextToken(status);
460 if (U_FAILURE(status)) {
461 return;
462 }
463 checkSyntax(status);
464 if (U_FAILURE(status)) {
465 return;
466 }
467 switch (type) {
468 case tAnd:
469 U_ASSERT(curAndConstraint != NULL);
470 curAndConstraint = curAndConstraint->add();
471 break;
472 case tOr:
473 {
474 U_ASSERT(currentChain != NULL);
475 OrConstraint *orNode=currentChain->ruleHeader;
476 while (orNode->next != NULL) {
477 orNode = orNode->next;
478 }
479 orNode->next= new OrConstraint();
480 orNode=orNode->next;
481 orNode->next=NULL;
482 curAndConstraint = orNode->add();
483 }
484 break;
485 case tIs:
486 U_ASSERT(curAndConstraint != NULL);
487 U_ASSERT(curAndConstraint->value == -1);
488 U_ASSERT(curAndConstraint->rangeList == NULL);
489 break;
490 case tNot:
491 U_ASSERT(curAndConstraint != NULL);
492 curAndConstraint->negated=TRUE;
493 break;
494
495 case tNotEqual:
496 curAndConstraint->negated=TRUE;
497 case tIn:
498 case tWithin:
499 case tEqual:
500 U_ASSERT(curAndConstraint != NULL);
501 curAndConstraint->rangeList = new UVector32(status);
502 curAndConstraint->rangeList->addElement(-1, status); // range Low
503 curAndConstraint->rangeList->addElement(-1, status); // range Hi
504 rangeLowIdx = 0;
505 rangeHiIdx = 1;
506 curAndConstraint->value=PLURAL_RANGE_HIGH;
507 curAndConstraint->integerOnly = (type != tWithin);
508 break;
509 case tNumber:
510 U_ASSERT(curAndConstraint != NULL);
511 if ( (curAndConstraint->op==AndConstraint::MOD)&&
512 (curAndConstraint->opNum == -1 ) ) {
513 curAndConstraint->opNum=getNumberValue(token);
514 }
515 else {
516 if (curAndConstraint->rangeList == NULL) {
517 // this is for an 'is' rule
518 curAndConstraint->value = getNumberValue(token);
519 } else {
520 // this is for an 'in' or 'within' rule
521 if (curAndConstraint->rangeList->elementAti(rangeLowIdx) == -1) {
522 curAndConstraint->rangeList->setElementAt(getNumberValue(token), rangeLowIdx);
523 curAndConstraint->rangeList->setElementAt(getNumberValue(token), rangeHiIdx);
524 }
525 else {
526 curAndConstraint->rangeList->setElementAt(getNumberValue(token), rangeHiIdx);
527 if (curAndConstraint->rangeList->elementAti(rangeLowIdx) >
528 curAndConstraint->rangeList->elementAti(rangeHiIdx)) {
529 // Range Lower bound > Range Upper bound.
530 // U_UNEXPECTED_TOKEN seems a little funny, but it is consistently
531 // used for all plural rule parse errors.
532 status = U_UNEXPECTED_TOKEN;
533 break;
534 }
535 }
536 }
537 }
538 break;
539 case tComma:
540 // TODO: rule syntax checking is inadequate, can happen with badly formed rules.
541 // Catch cases like "n mod 10, is 1" here instead.
542 if (curAndConstraint == NULL || curAndConstraint->rangeList == NULL) {
543 status = U_UNEXPECTED_TOKEN;
544 break;
545 }
546 U_ASSERT(curAndConstraint->rangeList->size() >= 2);
547 rangeLowIdx = curAndConstraint->rangeList->size();
548 curAndConstraint->rangeList->addElement(-1, status); // range Low
549 rangeHiIdx = curAndConstraint->rangeList->size();
550 curAndConstraint->rangeList->addElement(-1, status); // range Hi
551 break;
552 case tMod:
553 U_ASSERT(curAndConstraint != NULL);
554 curAndConstraint->op=AndConstraint::MOD;
555 break;
556 case tVariableN:
557 case tVariableI:
558 case tVariableF:
559 case tVariableT:
560 case tVariableV:
561 U_ASSERT(curAndConstraint != NULL);
562 curAndConstraint->digitsType = type;
563 break;
564 case tKeyword:
565 {
566 RuleChain *newChain = new RuleChain;
567 if (newChain == NULL) {
568 status = U_MEMORY_ALLOCATION_ERROR;
569 break;
570 }
571 newChain->fKeyword = token;
572 if (prules->mRules == NULL) {
573 prules->mRules = newChain;
574 } else {
575 // The new rule chain goes at the end of the linked list of rule chains,
576 // unless there is an "other" keyword & chain. "other" must remain last.
577 RuleChain *insertAfter = prules->mRules;
578 while (insertAfter->fNext!=NULL &&
579 insertAfter->fNext->fKeyword.compare(PLURAL_KEYWORD_OTHER, 5) != 0 ){
580 insertAfter=insertAfter->fNext;
581 }
582 newChain->fNext = insertAfter->fNext;
583 insertAfter->fNext = newChain;
584 }
585 OrConstraint *orNode = new OrConstraint();
586 newChain->ruleHeader = orNode;
587 curAndConstraint = orNode->add();
588 currentChain = newChain;
589 }
590 break;
591
592 case tInteger:
593 for (;;) {
594 getNextToken(status);
595 if (U_FAILURE(status) || type == tSemiColon || type == tEOF || type == tAt) {
596 break;
597 }
598 if (type == tEllipsis) {
599 currentChain->fIntegerSamplesUnbounded = TRUE;
600 continue;
601 }
602 currentChain->fIntegerSamples.append(token);
603 }
604 break;
605
606 case tDecimal:
607 for (;;) {
608 getNextToken(status);
609 if (U_FAILURE(status) || type == tSemiColon || type == tEOF || type == tAt) {
610 break;
611 }
612 if (type == tEllipsis) {
613 currentChain->fDecimalSamplesUnbounded = TRUE;
614 continue;
615 }
616 currentChain->fDecimalSamples.append(token);
617 }
618 break;
619
620 default:
621 break;
622 }
623 prevType=type;
624 if (U_FAILURE(status)) {
625 break;
626 }
627 }
628 }
629
630 UnicodeString
getRuleFromResource(const Locale & locale,UPluralType type,UErrorCode & errCode)631 PluralRules::getRuleFromResource(const Locale& locale, UPluralType type, UErrorCode& errCode) {
632 UnicodeString emptyStr;
633
634 if (U_FAILURE(errCode)) {
635 return emptyStr;
636 }
637 LocalUResourceBundlePointer rb(ures_openDirect(NULL, "plurals", &errCode));
638 if(U_FAILURE(errCode)) {
639 return emptyStr;
640 }
641 const char *typeKey;
642 switch (type) {
643 case UPLURAL_TYPE_CARDINAL:
644 typeKey = "locales";
645 break;
646 case UPLURAL_TYPE_ORDINAL:
647 typeKey = "locales_ordinals";
648 break;
649 default:
650 // Must not occur: The caller should have checked for valid types.
651 errCode = U_ILLEGAL_ARGUMENT_ERROR;
652 return emptyStr;
653 }
654 LocalUResourceBundlePointer locRes(ures_getByKey(rb.getAlias(), typeKey, NULL, &errCode));
655 if(U_FAILURE(errCode)) {
656 return emptyStr;
657 }
658 int32_t resLen=0;
659 const char *curLocaleName=locale.getName();
660 const UChar* s = ures_getStringByKey(locRes.getAlias(), curLocaleName, &resLen, &errCode);
661
662 if (s == NULL) {
663 // Check parent locales.
664 UErrorCode status = U_ZERO_ERROR;
665 char parentLocaleName[ULOC_FULLNAME_CAPACITY];
666 const char *curLocaleName=locale.getName();
667 uprv_strcpy(parentLocaleName, curLocaleName);
668
669 while (uloc_getParent(parentLocaleName, parentLocaleName,
670 ULOC_FULLNAME_CAPACITY, &status) > 0) {
671 resLen=0;
672 s = ures_getStringByKey(locRes.getAlias(), parentLocaleName, &resLen, &status);
673 if (s != NULL) {
674 errCode = U_ZERO_ERROR;
675 break;
676 }
677 status = U_ZERO_ERROR;
678 }
679 }
680 if (s==NULL) {
681 return emptyStr;
682 }
683
684 char setKey[256];
685 u_UCharsToChars(s, setKey, resLen + 1);
686 // printf("\n PluralRule: %s\n", setKey);
687
688 LocalUResourceBundlePointer ruleRes(ures_getByKey(rb.getAlias(), "rules", NULL, &errCode));
689 if(U_FAILURE(errCode)) {
690 return emptyStr;
691 }
692 LocalUResourceBundlePointer setRes(ures_getByKey(ruleRes.getAlias(), setKey, NULL, &errCode));
693 if (U_FAILURE(errCode)) {
694 return emptyStr;
695 }
696
697 int32_t numberKeys = ures_getSize(setRes.getAlias());
698 UnicodeString result;
699 const char *key=NULL;
700 for(int32_t i=0; i<numberKeys; ++i) { // Keys are zero, one, few, ...
701 UnicodeString rules = ures_getNextUnicodeString(setRes.getAlias(), &key, &errCode);
702 UnicodeString uKey(key, -1, US_INV);
703 result.append(uKey);
704 result.append(COLON);
705 result.append(rules);
706 result.append(SEMI_COLON);
707 }
708 return result;
709 }
710
711
712 UnicodeString
getRules() const713 PluralRules::getRules() const {
714 UnicodeString rules;
715 if (mRules != NULL) {
716 mRules->dumpRules(rules);
717 }
718 return rules;
719 }
720
721
AndConstraint()722 AndConstraint::AndConstraint() {
723 op = AndConstraint::NONE;
724 opNum=-1;
725 value = -1;
726 rangeList = NULL;
727 negated = FALSE;
728 integerOnly = FALSE;
729 digitsType = none;
730 next=NULL;
731 }
732
733
AndConstraint(const AndConstraint & other)734 AndConstraint::AndConstraint(const AndConstraint& other) {
735 this->op = other.op;
736 this->opNum=other.opNum;
737 this->value=other.value;
738 this->rangeList=NULL;
739 if (other.rangeList != NULL) {
740 UErrorCode status = U_ZERO_ERROR;
741 this->rangeList = new UVector32(status);
742 this->rangeList->assign(*other.rangeList, status);
743 }
744 this->integerOnly=other.integerOnly;
745 this->negated=other.negated;
746 this->digitsType = other.digitsType;
747 if (other.next==NULL) {
748 this->next=NULL;
749 }
750 else {
751 this->next = new AndConstraint(*other.next);
752 }
753 }
754
~AndConstraint()755 AndConstraint::~AndConstraint() {
756 delete rangeList;
757 if (next!=NULL) {
758 delete next;
759 }
760 }
761
762
763 UBool
isFulfilled(const FixedDecimal & number)764 AndConstraint::isFulfilled(const FixedDecimal &number) {
765 UBool result = TRUE;
766 if (digitsType == none) {
767 // An empty AndConstraint, created by a rule with a keyword but no following expression.
768 return TRUE;
769 }
770 double n = number.get(digitsType); // pulls n | i | v | f value for the number.
771 // Will always be positive.
772 // May be non-integer (n option only)
773 do {
774 if (integerOnly && n != uprv_floor(n)) {
775 result = FALSE;
776 break;
777 }
778
779 if (op == MOD) {
780 n = fmod(n, opNum);
781 }
782 if (rangeList == NULL) {
783 result = value == -1 || // empty rule
784 n == value; // 'is' rule
785 break;
786 }
787 result = FALSE; // 'in' or 'within' rule
788 for (int32_t r=0; r<rangeList->size(); r+=2) {
789 if (rangeList->elementAti(r) <= n && n <= rangeList->elementAti(r+1)) {
790 result = TRUE;
791 break;
792 }
793 }
794 } while (FALSE);
795
796 if (negated) {
797 result = !result;
798 }
799 return result;
800 }
801
802
803 AndConstraint*
add()804 AndConstraint::add()
805 {
806 this->next = new AndConstraint();
807 return this->next;
808 }
809
OrConstraint()810 OrConstraint::OrConstraint() {
811 childNode=NULL;
812 next=NULL;
813 }
814
OrConstraint(const OrConstraint & other)815 OrConstraint::OrConstraint(const OrConstraint& other) {
816 if ( other.childNode == NULL ) {
817 this->childNode = NULL;
818 }
819 else {
820 this->childNode = new AndConstraint(*(other.childNode));
821 }
822 if (other.next == NULL ) {
823 this->next = NULL;
824 }
825 else {
826 this->next = new OrConstraint(*(other.next));
827 }
828 }
829
~OrConstraint()830 OrConstraint::~OrConstraint() {
831 if (childNode!=NULL) {
832 delete childNode;
833 }
834 if (next!=NULL) {
835 delete next;
836 }
837 }
838
839 AndConstraint*
add()840 OrConstraint::add()
841 {
842 OrConstraint *curOrConstraint=this;
843 {
844 while (curOrConstraint->next!=NULL) {
845 curOrConstraint = curOrConstraint->next;
846 }
847 U_ASSERT(curOrConstraint->childNode == NULL);
848 curOrConstraint->childNode = new AndConstraint();
849 }
850 return curOrConstraint->childNode;
851 }
852
853 UBool
isFulfilled(const FixedDecimal & number)854 OrConstraint::isFulfilled(const FixedDecimal &number) {
855 OrConstraint* orRule=this;
856 UBool result=FALSE;
857
858 while (orRule!=NULL && !result) {
859 result=TRUE;
860 AndConstraint* andRule = orRule->childNode;
861 while (andRule!=NULL && result) {
862 result = andRule->isFulfilled(number);
863 andRule=andRule->next;
864 }
865 orRule = orRule->next;
866 }
867
868 return result;
869 }
870
871
RuleChain()872 RuleChain::RuleChain(): fKeyword(), fNext(NULL), ruleHeader(NULL), fDecimalSamples(), fIntegerSamples(),
873 fDecimalSamplesUnbounded(FALSE), fIntegerSamplesUnbounded(FALSE) {
874 }
875
RuleChain(const RuleChain & other)876 RuleChain::RuleChain(const RuleChain& other) :
877 fKeyword(other.fKeyword), fNext(NULL), ruleHeader(NULL), fDecimalSamples(other.fDecimalSamples),
878 fIntegerSamples(other.fIntegerSamples), fDecimalSamplesUnbounded(other.fDecimalSamplesUnbounded),
879 fIntegerSamplesUnbounded(other.fIntegerSamplesUnbounded) {
880 if (other.ruleHeader != NULL) {
881 this->ruleHeader = new OrConstraint(*(other.ruleHeader));
882 }
883 if (other.fNext != NULL ) {
884 this->fNext = new RuleChain(*other.fNext);
885 }
886 }
887
~RuleChain()888 RuleChain::~RuleChain() {
889 delete fNext;
890 delete ruleHeader;
891 }
892
893
894 UnicodeString
select(const FixedDecimal & number) const895 RuleChain::select(const FixedDecimal &number) const {
896 if (!number.isNanOrInfinity) {
897 for (const RuleChain *rules = this; rules != NULL; rules = rules->fNext) {
898 if (rules->ruleHeader->isFulfilled(number)) {
899 return rules->fKeyword;
900 }
901 }
902 }
903 return UnicodeString(TRUE, PLURAL_KEYWORD_OTHER, 5);
904 }
905
tokenString(tokenType tok)906 static UnicodeString tokenString(tokenType tok) {
907 UnicodeString s;
908 switch (tok) {
909 case tVariableN:
910 s.append(LOW_N); break;
911 case tVariableI:
912 s.append(LOW_I); break;
913 case tVariableF:
914 s.append(LOW_F); break;
915 case tVariableV:
916 s.append(LOW_V); break;
917 case tVariableT:
918 s.append(LOW_T); break;
919 default:
920 s.append(TILDE);
921 }
922 return s;
923 }
924
925 void
dumpRules(UnicodeString & result)926 RuleChain::dumpRules(UnicodeString& result) {
927 UChar digitString[16];
928
929 if ( ruleHeader != NULL ) {
930 result += fKeyword;
931 result += COLON;
932 result += SPACE;
933 OrConstraint* orRule=ruleHeader;
934 while ( orRule != NULL ) {
935 AndConstraint* andRule=orRule->childNode;
936 while ( andRule != NULL ) {
937 if ((andRule->op==AndConstraint::NONE) && (andRule->rangeList==NULL) && (andRule->value == -1)) {
938 // Empty Rules.
939 } else if ( (andRule->op==AndConstraint::NONE) && (andRule->rangeList==NULL) ) {
940 result += tokenString(andRule->digitsType);
941 result += UNICODE_STRING_SIMPLE(" is ");
942 if (andRule->negated) {
943 result += UNICODE_STRING_SIMPLE("not ");
944 }
945 uprv_itou(digitString,16, andRule->value,10,0);
946 result += UnicodeString(digitString);
947 }
948 else {
949 result += tokenString(andRule->digitsType);
950 result += SPACE;
951 if (andRule->op==AndConstraint::MOD) {
952 result += UNICODE_STRING_SIMPLE("mod ");
953 uprv_itou(digitString,16, andRule->opNum,10,0);
954 result += UnicodeString(digitString);
955 }
956 if (andRule->rangeList==NULL) {
957 if (andRule->negated) {
958 result += UNICODE_STRING_SIMPLE(" is not ");
959 uprv_itou(digitString,16, andRule->value,10,0);
960 result += UnicodeString(digitString);
961 }
962 else {
963 result += UNICODE_STRING_SIMPLE(" is ");
964 uprv_itou(digitString,16, andRule->value,10,0);
965 result += UnicodeString(digitString);
966 }
967 }
968 else {
969 if (andRule->negated) {
970 if ( andRule->integerOnly ) {
971 result += UNICODE_STRING_SIMPLE(" not in ");
972 }
973 else {
974 result += UNICODE_STRING_SIMPLE(" not within ");
975 }
976 }
977 else {
978 if ( andRule->integerOnly ) {
979 result += UNICODE_STRING_SIMPLE(" in ");
980 }
981 else {
982 result += UNICODE_STRING_SIMPLE(" within ");
983 }
984 }
985 for (int32_t r=0; r<andRule->rangeList->size(); r+=2) {
986 int32_t rangeLo = andRule->rangeList->elementAti(r);
987 int32_t rangeHi = andRule->rangeList->elementAti(r+1);
988 uprv_itou(digitString,16, rangeLo, 10, 0);
989 result += UnicodeString(digitString);
990 result += UNICODE_STRING_SIMPLE("..");
991 uprv_itou(digitString,16, rangeHi, 10,0);
992 result += UnicodeString(digitString);
993 if (r+2 < andRule->rangeList->size()) {
994 result += UNICODE_STRING_SIMPLE(", ");
995 }
996 }
997 }
998 }
999 if ( (andRule=andRule->next) != NULL) {
1000 result += UNICODE_STRING_SIMPLE(" and ");
1001 }
1002 }
1003 if ( (orRule = orRule->next) != NULL ) {
1004 result += UNICODE_STRING_SIMPLE(" or ");
1005 }
1006 }
1007 }
1008 if ( fNext != NULL ) {
1009 result += UNICODE_STRING_SIMPLE("; ");
1010 fNext->dumpRules(result);
1011 }
1012 }
1013
1014
1015 UErrorCode
getKeywords(int32_t capacityOfKeywords,UnicodeString * keywords,int32_t & arraySize) const1016 RuleChain::getKeywords(int32_t capacityOfKeywords, UnicodeString* keywords, int32_t& arraySize) const {
1017 if ( arraySize < capacityOfKeywords-1 ) {
1018 keywords[arraySize++]=fKeyword;
1019 }
1020 else {
1021 return U_BUFFER_OVERFLOW_ERROR;
1022 }
1023
1024 if ( fNext != NULL ) {
1025 return fNext->getKeywords(capacityOfKeywords, keywords, arraySize);
1026 }
1027 else {
1028 return U_ZERO_ERROR;
1029 }
1030 }
1031
1032 UBool
isKeyword(const UnicodeString & keywordParam) const1033 RuleChain::isKeyword(const UnicodeString& keywordParam) const {
1034 if ( fKeyword == keywordParam ) {
1035 return TRUE;
1036 }
1037
1038 if ( fNext != NULL ) {
1039 return fNext->isKeyword(keywordParam);
1040 }
1041 else {
1042 return FALSE;
1043 }
1044 }
1045
1046
PluralRuleParser()1047 PluralRuleParser::PluralRuleParser() :
1048 ruleIndex(0), token(), type(none), prevType(none),
1049 curAndConstraint(NULL), currentChain(NULL), rangeLowIdx(-1), rangeHiIdx(-1)
1050 {
1051 }
1052
~PluralRuleParser()1053 PluralRuleParser::~PluralRuleParser() {
1054 }
1055
1056
1057 int32_t
getNumberValue(const UnicodeString & token)1058 PluralRuleParser::getNumberValue(const UnicodeString& token) {
1059 int32_t i;
1060 char digits[128];
1061
1062 i = token.extract(0, token.length(), digits, ARRAY_SIZE(digits), US_INV);
1063 digits[i]='\0';
1064
1065 return((int32_t)atoi(digits));
1066 }
1067
1068
1069 void
checkSyntax(UErrorCode & status)1070 PluralRuleParser::checkSyntax(UErrorCode &status)
1071 {
1072 if (U_FAILURE(status)) {
1073 return;
1074 }
1075 if (!(prevType==none || prevType==tSemiColon)) {
1076 type = getKeyType(token, type); // Switch token type from tKeyword if we scanned a reserved word,
1077 // and we are not at the start of a rule, where a
1078 // keyword is expected.
1079 }
1080
1081 switch(prevType) {
1082 case none:
1083 case tSemiColon:
1084 if (type!=tKeyword && type != tEOF) {
1085 status = U_UNEXPECTED_TOKEN;
1086 }
1087 break;
1088 case tVariableN:
1089 case tVariableI:
1090 case tVariableF:
1091 case tVariableT:
1092 case tVariableV:
1093 if (type != tIs && type != tMod && type != tIn &&
1094 type != tNot && type != tWithin && type != tEqual && type != tNotEqual) {
1095 status = U_UNEXPECTED_TOKEN;
1096 }
1097 break;
1098 case tKeyword:
1099 if (type != tColon) {
1100 status = U_UNEXPECTED_TOKEN;
1101 }
1102 break;
1103 case tColon:
1104 if (!(type == tVariableN ||
1105 type == tVariableI ||
1106 type == tVariableF ||
1107 type == tVariableT ||
1108 type == tVariableV ||
1109 type == tAt)) {
1110 status = U_UNEXPECTED_TOKEN;
1111 }
1112 break;
1113 case tIs:
1114 if ( type != tNumber && type != tNot) {
1115 status = U_UNEXPECTED_TOKEN;
1116 }
1117 break;
1118 case tNot:
1119 if (type != tNumber && type != tIn && type != tWithin) {
1120 status = U_UNEXPECTED_TOKEN;
1121 }
1122 break;
1123 case tMod:
1124 case tDot2:
1125 case tIn:
1126 case tWithin:
1127 case tEqual:
1128 case tNotEqual:
1129 if (type != tNumber) {
1130 status = U_UNEXPECTED_TOKEN;
1131 }
1132 break;
1133 case tAnd:
1134 case tOr:
1135 if ( type != tVariableN &&
1136 type != tVariableI &&
1137 type != tVariableF &&
1138 type != tVariableT &&
1139 type != tVariableV) {
1140 status = U_UNEXPECTED_TOKEN;
1141 }
1142 break;
1143 case tComma:
1144 if (type != tNumber) {
1145 status = U_UNEXPECTED_TOKEN;
1146 }
1147 break;
1148 case tNumber:
1149 if (type != tDot2 && type != tSemiColon && type != tIs && type != tNot &&
1150 type != tIn && type != tEqual && type != tNotEqual && type != tWithin &&
1151 type != tAnd && type != tOr && type != tComma && type != tAt &&
1152 type != tEOF)
1153 {
1154 status = U_UNEXPECTED_TOKEN;
1155 }
1156 // TODO: a comma following a number that is not part of a range will be allowed.
1157 // It's not the only case of this sort of thing. Parser needs a re-write.
1158 break;
1159 case tAt:
1160 if (type != tDecimal && type != tInteger) {
1161 status = U_UNEXPECTED_TOKEN;
1162 }
1163 break;
1164 default:
1165 status = U_UNEXPECTED_TOKEN;
1166 break;
1167 }
1168 }
1169
1170
1171 /*
1172 * Scan the next token from the input rules.
1173 * rules and returned token type are in the parser state variables.
1174 */
1175 void
getNextToken(UErrorCode & status)1176 PluralRuleParser::getNextToken(UErrorCode &status)
1177 {
1178 if (U_FAILURE(status)) {
1179 return;
1180 }
1181
1182 UChar ch;
1183 while (ruleIndex < ruleSrc->length()) {
1184 ch = ruleSrc->charAt(ruleIndex);
1185 type = charType(ch);
1186 if (type != tSpace) {
1187 break;
1188 }
1189 ++(ruleIndex);
1190 }
1191 if (ruleIndex >= ruleSrc->length()) {
1192 type = tEOF;
1193 return;
1194 }
1195 int32_t curIndex= ruleIndex;
1196
1197 switch (type) {
1198 case tColon:
1199 case tSemiColon:
1200 case tComma:
1201 case tEllipsis:
1202 case tTilde: // scanned '~'
1203 case tAt: // scanned '@'
1204 case tEqual: // scanned '='
1205 case tMod: // scanned '%'
1206 // Single character tokens.
1207 ++curIndex;
1208 break;
1209
1210 case tNotEqual: // scanned '!'
1211 if (ruleSrc->charAt(curIndex+1) == EQUALS) {
1212 curIndex += 2;
1213 } else {
1214 type = none;
1215 curIndex += 1;
1216 }
1217 break;
1218
1219 case tKeyword:
1220 while (type == tKeyword && ++curIndex < ruleSrc->length()) {
1221 ch = ruleSrc->charAt(curIndex);
1222 type = charType(ch);
1223 }
1224 type = tKeyword;
1225 break;
1226
1227 case tNumber:
1228 while (type == tNumber && ++curIndex < ruleSrc->length()) {
1229 ch = ruleSrc->charAt(curIndex);
1230 type = charType(ch);
1231 }
1232 type = tNumber;
1233 break;
1234
1235 case tDot:
1236 // We could be looking at either ".." in a range, or "..." at the end of a sample.
1237 if (curIndex+1 >= ruleSrc->length() || ruleSrc->charAt(curIndex+1) != DOT) {
1238 ++curIndex;
1239 break; // Single dot
1240 }
1241 if (curIndex+2 >= ruleSrc->length() || ruleSrc->charAt(curIndex+2) != DOT) {
1242 curIndex += 2;
1243 type = tDot2;
1244 break; // double dot
1245 }
1246 type = tEllipsis;
1247 curIndex += 3;
1248 break; // triple dot
1249
1250 default:
1251 status = U_UNEXPECTED_TOKEN;
1252 ++curIndex;
1253 break;
1254 }
1255
1256 U_ASSERT(ruleIndex <= ruleSrc->length());
1257 U_ASSERT(curIndex <= ruleSrc->length());
1258 token=UnicodeString(*ruleSrc, ruleIndex, curIndex-ruleIndex);
1259 ruleIndex = curIndex;
1260 }
1261
1262 tokenType
charType(UChar ch)1263 PluralRuleParser::charType(UChar ch) {
1264 if ((ch>=U_ZERO) && (ch<=U_NINE)) {
1265 return tNumber;
1266 }
1267 if (ch>=LOW_A && ch<=LOW_Z) {
1268 return tKeyword;
1269 }
1270 switch (ch) {
1271 case COLON:
1272 return tColon;
1273 case SPACE:
1274 return tSpace;
1275 case SEMI_COLON:
1276 return tSemiColon;
1277 case DOT:
1278 return tDot;
1279 case COMMA:
1280 return tComma;
1281 case EXCLAMATION:
1282 return tNotEqual;
1283 case EQUALS:
1284 return tEqual;
1285 case PERCENT_SIGN:
1286 return tMod;
1287 case AT:
1288 return tAt;
1289 case ELLIPSIS:
1290 return tEllipsis;
1291 case TILDE:
1292 return tTilde;
1293 default :
1294 return none;
1295 }
1296 }
1297
1298
1299 // Set token type for reserved words in the Plural Rule syntax.
1300
1301 tokenType
getKeyType(const UnicodeString & token,tokenType keyType)1302 PluralRuleParser::getKeyType(const UnicodeString &token, tokenType keyType)
1303 {
1304 if (keyType != tKeyword) {
1305 return keyType;
1306 }
1307
1308 if (0 == token.compare(PK_VAR_N, 1)) {
1309 keyType = tVariableN;
1310 } else if (0 == token.compare(PK_VAR_I, 1)) {
1311 keyType = tVariableI;
1312 } else if (0 == token.compare(PK_VAR_F, 1)) {
1313 keyType = tVariableF;
1314 } else if (0 == token.compare(PK_VAR_T, 1)) {
1315 keyType = tVariableT;
1316 } else if (0 == token.compare(PK_VAR_V, 1)) {
1317 keyType = tVariableV;
1318 } else if (0 == token.compare(PK_IS, 2)) {
1319 keyType = tIs;
1320 } else if (0 == token.compare(PK_AND, 3)) {
1321 keyType = tAnd;
1322 } else if (0 == token.compare(PK_IN, 2)) {
1323 keyType = tIn;
1324 } else if (0 == token.compare(PK_WITHIN, 6)) {
1325 keyType = tWithin;
1326 } else if (0 == token.compare(PK_NOT, 3)) {
1327 keyType = tNot;
1328 } else if (0 == token.compare(PK_MOD, 3)) {
1329 keyType = tMod;
1330 } else if (0 == token.compare(PK_OR, 2)) {
1331 keyType = tOr;
1332 } else if (0 == token.compare(PK_DECIMAL, 7)) {
1333 keyType = tDecimal;
1334 } else if (0 == token.compare(PK_INTEGER, 7)) {
1335 keyType = tInteger;
1336 }
1337 return keyType;
1338 }
1339
1340
PluralKeywordEnumeration(RuleChain * header,UErrorCode & status)1341 PluralKeywordEnumeration::PluralKeywordEnumeration(RuleChain *header, UErrorCode& status)
1342 : pos(0), fKeywordNames(status) {
1343 if (U_FAILURE(status)) {
1344 return;
1345 }
1346 fKeywordNames.setDeleter(uprv_deleteUObject);
1347 UBool addKeywordOther=TRUE;
1348 RuleChain *node=header;
1349 while(node!=NULL) {
1350 fKeywordNames.addElement(new UnicodeString(node->fKeyword), status);
1351 if (U_FAILURE(status)) {
1352 return;
1353 }
1354 if (0 == node->fKeyword.compare(PLURAL_KEYWORD_OTHER, 5)) {
1355 addKeywordOther= FALSE;
1356 }
1357 node=node->fNext;
1358 }
1359
1360 if (addKeywordOther) {
1361 fKeywordNames.addElement(new UnicodeString(PLURAL_KEYWORD_OTHER), status);
1362 }
1363 }
1364
1365 const UnicodeString*
snext(UErrorCode & status)1366 PluralKeywordEnumeration::snext(UErrorCode& status) {
1367 if (U_SUCCESS(status) && pos < fKeywordNames.size()) {
1368 return (const UnicodeString*)fKeywordNames.elementAt(pos++);
1369 }
1370 return NULL;
1371 }
1372
1373 void
reset(UErrorCode &)1374 PluralKeywordEnumeration::reset(UErrorCode& /*status*/) {
1375 pos=0;
1376 }
1377
1378 int32_t
count(UErrorCode &) const1379 PluralKeywordEnumeration::count(UErrorCode& /*status*/) const {
1380 return fKeywordNames.size();
1381 }
1382
~PluralKeywordEnumeration()1383 PluralKeywordEnumeration::~PluralKeywordEnumeration() {
1384 }
1385
FixedDecimal(const VisibleDigits & digits)1386 FixedDecimal::FixedDecimal(const VisibleDigits &digits) {
1387 digits.getFixedDecimal(
1388 source, intValue, decimalDigits,
1389 decimalDigitsWithoutTrailingZeros,
1390 visibleDecimalDigitCount, hasIntegerValue);
1391 isNegative = digits.isNegative();
1392 isNanOrInfinity = digits.isNaNOrInfinity();
1393 }
1394
FixedDecimal(double n,int32_t v,int64_t f)1395 FixedDecimal::FixedDecimal(double n, int32_t v, int64_t f) {
1396 init(n, v, f);
1397 // check values. TODO make into unit test.
1398 //
1399 // long visiblePower = (int) Math.pow(10, v);
1400 // if (decimalDigits > visiblePower) {
1401 // throw new IllegalArgumentException();
1402 // }
1403 // double fraction = intValue + (decimalDigits / (double) visiblePower);
1404 // if (fraction != source) {
1405 // double diff = Math.abs(fraction - source)/(Math.abs(fraction) + Math.abs(source));
1406 // if (diff > 0.00000001d) {
1407 // throw new IllegalArgumentException();
1408 // }
1409 // }
1410 }
1411
FixedDecimal(double n,int32_t v)1412 FixedDecimal::FixedDecimal(double n, int32_t v) {
1413 // Ugly, but for samples we don't care.
1414 init(n, v, getFractionalDigits(n, v));
1415 }
1416
FixedDecimal(double n)1417 FixedDecimal::FixedDecimal(double n) {
1418 init(n);
1419 }
1420
FixedDecimal()1421 FixedDecimal::FixedDecimal() {
1422 init(0, 0, 0);
1423 }
1424
1425
1426 // Create a FixedDecimal from a UnicodeString containing a number.
1427 // Inefficient, but only used for samples, so simplicity trumps efficiency.
1428
FixedDecimal(const UnicodeString & num,UErrorCode & status)1429 FixedDecimal::FixedDecimal(const UnicodeString &num, UErrorCode &status) {
1430 CharString cs;
1431 cs.appendInvariantChars(num, status);
1432 DigitList dl;
1433 dl.set(cs.toStringPiece(), status);
1434 if (U_FAILURE(status)) {
1435 init(0, 0, 0);
1436 return;
1437 }
1438 int32_t decimalPoint = num.indexOf(DOT);
1439 double n = dl.getDouble();
1440 if (decimalPoint == -1) {
1441 init(n, 0, 0);
1442 } else {
1443 int32_t v = num.length() - decimalPoint - 1;
1444 init(n, v, getFractionalDigits(n, v));
1445 }
1446 }
1447
1448
FixedDecimal(const FixedDecimal & other)1449 FixedDecimal::FixedDecimal(const FixedDecimal &other) {
1450 source = other.source;
1451 visibleDecimalDigitCount = other.visibleDecimalDigitCount;
1452 decimalDigits = other.decimalDigits;
1453 decimalDigitsWithoutTrailingZeros = other.decimalDigitsWithoutTrailingZeros;
1454 intValue = other.intValue;
1455 hasIntegerValue = other.hasIntegerValue;
1456 isNegative = other.isNegative;
1457 isNanOrInfinity = other.isNanOrInfinity;
1458 }
1459
1460
init(double n)1461 void FixedDecimal::init(double n) {
1462 int32_t numFractionDigits = decimals(n);
1463 init(n, numFractionDigits, getFractionalDigits(n, numFractionDigits));
1464 }
1465
1466
init(double n,int32_t v,int64_t f)1467 void FixedDecimal::init(double n, int32_t v, int64_t f) {
1468 isNegative = n < 0.0;
1469 source = fabs(n);
1470 isNanOrInfinity = uprv_isNaN(source) || uprv_isPositiveInfinity(source);
1471 if (isNanOrInfinity) {
1472 v = 0;
1473 f = 0;
1474 intValue = 0;
1475 hasIntegerValue = FALSE;
1476 } else {
1477 intValue = (int64_t)source;
1478 hasIntegerValue = (source == intValue);
1479 }
1480
1481 visibleDecimalDigitCount = v;
1482 decimalDigits = f;
1483 if (f == 0) {
1484 decimalDigitsWithoutTrailingZeros = 0;
1485 } else {
1486 int64_t fdwtz = f;
1487 while ((fdwtz%10) == 0) {
1488 fdwtz /= 10;
1489 }
1490 decimalDigitsWithoutTrailingZeros = fdwtz;
1491 }
1492 }
1493
1494
1495 // Fast path only exact initialization. Return true if successful.
1496 // Note: Do not multiply by 10 each time through loop, rounding cruft can build
1497 // up that makes the check for an integer result fail.
1498 // A single multiply of the original number works more reliably.
1499 static int32_t p10[] = {1, 10, 100, 1000, 10000};
quickInit(double n)1500 UBool FixedDecimal::quickInit(double n) {
1501 UBool success = FALSE;
1502 n = fabs(n);
1503 int32_t numFractionDigits;
1504 for (numFractionDigits = 0; numFractionDigits <= 3; numFractionDigits++) {
1505 double scaledN = n * p10[numFractionDigits];
1506 if (scaledN == floor(scaledN)) {
1507 success = TRUE;
1508 break;
1509 }
1510 }
1511 if (success) {
1512 init(n, numFractionDigits, getFractionalDigits(n, numFractionDigits));
1513 }
1514 return success;
1515 }
1516
1517
1518
decimals(double n)1519 int32_t FixedDecimal::decimals(double n) {
1520 // Count the number of decimal digits in the fraction part of the number, excluding trailing zeros.
1521 // fastpath the common cases, integers or fractions with 3 or fewer digits
1522 n = fabs(n);
1523 for (int ndigits=0; ndigits<=3; ndigits++) {
1524 double scaledN = n * p10[ndigits];
1525 if (scaledN == floor(scaledN)) {
1526 return ndigits;
1527 }
1528 }
1529
1530 // Slow path, convert with sprintf, parse converted output.
1531 char buf[30] = {0};
1532 sprintf(buf, "%1.15e", n);
1533 // formatted number looks like this: 1.234567890123457e-01
1534 int exponent = atoi(buf+18);
1535 int numFractionDigits = 15;
1536 for (int i=16; ; --i) {
1537 if (buf[i] != '0') {
1538 break;
1539 }
1540 --numFractionDigits;
1541 }
1542 numFractionDigits -= exponent; // Fraction part of fixed point representation.
1543 return numFractionDigits;
1544 }
1545
1546
1547 // Get the fraction digits of a double, represented as an integer.
1548 // v is the number of visible fraction digits in the displayed form of the number.
1549 // Example: n = 1001.234, v = 6, result = 234000
1550 // TODO: need to think through how this is used in the plural rule context.
1551 // This function can easily encounter integer overflow,
1552 // and can easily return noise digits when the precision of a double is exceeded.
1553
getFractionalDigits(double n,int32_t v)1554 int64_t FixedDecimal::getFractionalDigits(double n, int32_t v) {
1555 if (v == 0 || n == floor(n) || uprv_isNaN(n) || uprv_isPositiveInfinity(n)) {
1556 return 0;
1557 }
1558 n = fabs(n);
1559 double fract = n - floor(n);
1560 switch (v) {
1561 case 1: return (int64_t)(fract*10.0 + 0.5);
1562 case 2: return (int64_t)(fract*100.0 + 0.5);
1563 case 3: return (int64_t)(fract*1000.0 + 0.5);
1564 default:
1565 double scaled = floor(fract * pow(10.0, (double)v) + 0.5);
1566 if (scaled > U_INT64_MAX) {
1567 return U_INT64_MAX;
1568 } else {
1569 return (int64_t)scaled;
1570 }
1571 }
1572 }
1573
1574
adjustForMinFractionDigits(int32_t minFractionDigits)1575 void FixedDecimal::adjustForMinFractionDigits(int32_t minFractionDigits) {
1576 int32_t numTrailingFractionZeros = minFractionDigits - visibleDecimalDigitCount;
1577 if (numTrailingFractionZeros > 0) {
1578 for (int32_t i=0; i<numTrailingFractionZeros; i++) {
1579 // Do not let the decimalDigits value overflow if there are many trailing zeros.
1580 // Limit the value to 18 digits, the most that a 64 bit int can fully represent.
1581 if (decimalDigits >= 100000000000000000LL) {
1582 break;
1583 }
1584 decimalDigits *= 10;
1585 }
1586 visibleDecimalDigitCount += numTrailingFractionZeros;
1587 }
1588 }
1589
1590
get(tokenType operand) const1591 double FixedDecimal::get(tokenType operand) const {
1592 switch(operand) {
1593 case tVariableN: return source;
1594 case tVariableI: return (double)intValue;
1595 case tVariableF: return (double)decimalDigits;
1596 case tVariableT: return (double)decimalDigitsWithoutTrailingZeros;
1597 case tVariableV: return visibleDecimalDigitCount;
1598 default:
1599 U_ASSERT(FALSE); // unexpected.
1600 return source;
1601 }
1602 }
1603
getVisibleFractionDigitCount() const1604 int32_t FixedDecimal::getVisibleFractionDigitCount() const {
1605 return visibleDecimalDigitCount;
1606 }
1607
1608
1609
PluralAvailableLocalesEnumeration(UErrorCode & status)1610 PluralAvailableLocalesEnumeration::PluralAvailableLocalesEnumeration(UErrorCode &status) {
1611 fLocales = NULL;
1612 fRes = NULL;
1613 fOpenStatus = status;
1614 if (U_FAILURE(status)) {
1615 return;
1616 }
1617 fOpenStatus = U_ZERO_ERROR;
1618 LocalUResourceBundlePointer rb(ures_openDirect(NULL, "plurals", &fOpenStatus));
1619 fLocales = ures_getByKey(rb.getAlias(), "locales", NULL, &fOpenStatus);
1620 }
1621
~PluralAvailableLocalesEnumeration()1622 PluralAvailableLocalesEnumeration::~PluralAvailableLocalesEnumeration() {
1623 ures_close(fLocales);
1624 ures_close(fRes);
1625 fLocales = NULL;
1626 fRes = NULL;
1627 }
1628
next(int32_t * resultLength,UErrorCode & status)1629 const char *PluralAvailableLocalesEnumeration::next(int32_t *resultLength, UErrorCode &status) {
1630 if (U_FAILURE(status)) {
1631 return NULL;
1632 }
1633 if (U_FAILURE(fOpenStatus)) {
1634 status = fOpenStatus;
1635 return NULL;
1636 }
1637 fRes = ures_getNextResource(fLocales, fRes, &status);
1638 if (fRes == NULL || U_FAILURE(status)) {
1639 if (status == U_INDEX_OUTOFBOUNDS_ERROR) {
1640 status = U_ZERO_ERROR;
1641 }
1642 return NULL;
1643 }
1644 const char *result = ures_getKey(fRes);
1645 if (resultLength != NULL) {
1646 *resultLength = uprv_strlen(result);
1647 }
1648 return result;
1649 }
1650
1651
reset(UErrorCode & status)1652 void PluralAvailableLocalesEnumeration::reset(UErrorCode &status) {
1653 if (U_FAILURE(status)) {
1654 return;
1655 }
1656 if (U_FAILURE(fOpenStatus)) {
1657 status = fOpenStatus;
1658 return;
1659 }
1660 ures_resetIterator(fLocales);
1661 }
1662
count(UErrorCode & status) const1663 int32_t PluralAvailableLocalesEnumeration::count(UErrorCode &status) const {
1664 if (U_FAILURE(status)) {
1665 return 0;
1666 }
1667 if (U_FAILURE(fOpenStatus)) {
1668 status = fOpenStatus;
1669 return 0;
1670 }
1671 return ures_getSize(fLocales);
1672 }
1673
1674 U_NAMESPACE_END
1675
1676
1677 #endif /* #if !UCONFIG_NO_FORMATTING */
1678
1679 //eof
1680