1 // © 2016 and later: Unicode, Inc. and others.
2 // License & terms of use: http://www.unicode.org/copyright.html
3 /*
4 *******************************************************************************
5 *
6 * Copyright (C) 2001-2015, International Business Machines
7 * Corporation and others. All Rights Reserved.
8 *
9 *******************************************************************************
10 * file name: ustrcase.cpp
11 * encoding: UTF-8
12 * tab size: 8 (not used)
13 * indentation:4
14 *
15 * created on: 2002feb20
16 * created by: Markus W. Scherer
17 *
18 * Implementation file for string casing C API functions.
19 * Uses functions from uchar.c for basic functionality that requires access
20 * to the Unicode Character Database (uprops.dat).
21 */
22
23 #include "unicode/utypes.h"
24 #include "unicode/brkiter.h"
25 #include "unicode/casemap.h"
26 #include "unicode/edits.h"
27 #include "unicode/stringoptions.h"
28 #include "unicode/ustring.h"
29 #include "unicode/ucasemap.h"
30 #include "unicode/ubrk.h"
31 #include "unicode/utf.h"
32 #include "unicode/utf16.h"
33 #include "cmemory.h"
34 #include "ucase.h"
35 #include "ucasemap_imp.h"
36 #include "ustr_imp.h"
37 #include "uassert.h"
38
39 /**
40 * Code point for COMBINING ACUTE ACCENT
41 * @internal
42 */
43 #define ACUTE u'\u0301'
44
45 U_NAMESPACE_BEGIN
46
47 namespace {
48
checkOverflowAndEditsError(int32_t destIndex,int32_t destCapacity,Edits * edits,UErrorCode & errorCode)49 int32_t checkOverflowAndEditsError(int32_t destIndex, int32_t destCapacity,
50 Edits *edits, UErrorCode &errorCode) {
51 if (U_SUCCESS(errorCode)) {
52 if (destIndex > destCapacity) {
53 errorCode = U_BUFFER_OVERFLOW_ERROR;
54 } else if (edits != nullptr) {
55 edits->copyErrorTo(errorCode);
56 }
57 }
58 return destIndex;
59 }
60
61 /* Appends a full case mapping result, see UCASE_MAX_STRING_LENGTH. */
62 inline int32_t
appendResult(char16_t * dest,int32_t destIndex,int32_t destCapacity,int32_t result,const char16_t * s,int32_t cpLength,uint32_t options,icu::Edits * edits)63 appendResult(char16_t *dest, int32_t destIndex, int32_t destCapacity,
64 int32_t result, const char16_t *s,
65 int32_t cpLength, uint32_t options, icu::Edits *edits) {
66 UChar32 c;
67 int32_t length;
68
69 /* decode the result */
70 if(result<0) {
71 /* (not) original code point */
72 if(edits!=nullptr) {
73 edits->addUnchanged(cpLength);
74 }
75 if(options & U_OMIT_UNCHANGED_TEXT) {
76 return destIndex;
77 }
78 c=~result;
79 if(destIndex<destCapacity && c<=0xffff) { // BMP slightly-fastpath
80 dest[destIndex++]=(char16_t)c;
81 return destIndex;
82 }
83 length=cpLength;
84 } else {
85 if(result<=UCASE_MAX_STRING_LENGTH) {
86 c=U_SENTINEL;
87 length=result;
88 } else if(destIndex<destCapacity && result<=0xffff) { // BMP slightly-fastpath
89 dest[destIndex++]=(char16_t)result;
90 if(edits!=nullptr) {
91 edits->addReplace(cpLength, 1);
92 }
93 return destIndex;
94 } else {
95 c=result;
96 length=U16_LENGTH(c);
97 }
98 if(edits!=nullptr) {
99 edits->addReplace(cpLength, length);
100 }
101 }
102 if(length>(INT32_MAX-destIndex)) {
103 return -1; // integer overflow
104 }
105
106 if(destIndex<destCapacity) {
107 /* append the result */
108 if(c>=0) {
109 /* code point */
110 UBool isError=false;
111 U16_APPEND(dest, destIndex, destCapacity, c, isError);
112 if(isError) {
113 /* overflow, nothing written */
114 destIndex+=length;
115 }
116 } else {
117 /* string */
118 if((destIndex+length)<=destCapacity) {
119 while(length>0) {
120 dest[destIndex++]=*s++;
121 --length;
122 }
123 } else {
124 /* overflow */
125 destIndex+=length;
126 }
127 }
128 } else {
129 /* preflight */
130 destIndex+=length;
131 }
132 return destIndex;
133 }
134
135 inline int32_t
appendUChar(char16_t * dest,int32_t destIndex,int32_t destCapacity,char16_t c)136 appendUChar(char16_t *dest, int32_t destIndex, int32_t destCapacity, char16_t c) {
137 if(destIndex<destCapacity) {
138 dest[destIndex]=c;
139 } else if(destIndex==INT32_MAX) {
140 return -1; // integer overflow
141 }
142 return destIndex+1;
143 }
144
145 int32_t
appendNonEmptyUnchanged(char16_t * dest,int32_t destIndex,int32_t destCapacity,const char16_t * s,int32_t length,uint32_t options,icu::Edits * edits)146 appendNonEmptyUnchanged(char16_t *dest, int32_t destIndex, int32_t destCapacity,
147 const char16_t *s, int32_t length, uint32_t options, icu::Edits *edits) {
148 if(edits!=nullptr) {
149 edits->addUnchanged(length);
150 }
151 if(options & U_OMIT_UNCHANGED_TEXT) {
152 return destIndex;
153 }
154 if(length>(INT32_MAX-destIndex)) {
155 return -1; // integer overflow
156 }
157 if((destIndex+length)<=destCapacity) {
158 u_memcpy(dest+destIndex, s, length);
159 }
160 return destIndex + length;
161 }
162
163 inline int32_t
appendUnchanged(char16_t * dest,int32_t destIndex,int32_t destCapacity,const char16_t * s,int32_t length,uint32_t options,icu::Edits * edits)164 appendUnchanged(char16_t *dest, int32_t destIndex, int32_t destCapacity,
165 const char16_t *s, int32_t length, uint32_t options, icu::Edits *edits) {
166 if (length <= 0) {
167 return destIndex;
168 }
169 return appendNonEmptyUnchanged(dest, destIndex, destCapacity, s, length, options, edits);
170 }
171
172 UChar32 U_CALLCONV
utf16_caseContextIterator(void * context,int8_t dir)173 utf16_caseContextIterator(void *context, int8_t dir) {
174 UCaseContext *csc=(UCaseContext *)context;
175 UChar32 c;
176
177 if(dir<0) {
178 /* reset for backward iteration */
179 csc->index=csc->cpStart;
180 csc->dir=dir;
181 } else if(dir>0) {
182 /* reset for forward iteration */
183 csc->index=csc->cpLimit;
184 csc->dir=dir;
185 } else {
186 /* continue current iteration direction */
187 dir=csc->dir;
188 }
189
190 if(dir<0) {
191 if(csc->start<csc->index) {
192 U16_PREV((const char16_t *)csc->p, csc->start, csc->index, c);
193 return c;
194 }
195 } else {
196 if(csc->index<csc->limit) {
197 U16_NEXT((const char16_t *)csc->p, csc->index, csc->limit, c);
198 return c;
199 }
200 }
201 return U_SENTINEL;
202 }
203
204 /**
205 * caseLocale >= 0: Lowercases [srcStart..srcLimit[ but takes context [0..srcLength[ into account.
206 * caseLocale < 0: Case-folds [srcStart..srcLimit[.
207 */
toLower(int32_t caseLocale,uint32_t options,char16_t * dest,int32_t destCapacity,const char16_t * src,UCaseContext * csc,int32_t srcStart,int32_t srcLimit,icu::Edits * edits,UErrorCode & errorCode)208 int32_t toLower(int32_t caseLocale, uint32_t options,
209 char16_t *dest, int32_t destCapacity,
210 const char16_t *src, UCaseContext *csc, int32_t srcStart, int32_t srcLimit,
211 icu::Edits *edits, UErrorCode &errorCode) {
212 const int8_t *latinToLower;
213 if (caseLocale == UCASE_LOC_ROOT ||
214 (caseLocale >= 0 ?
215 !(caseLocale == UCASE_LOC_TURKISH || caseLocale == UCASE_LOC_LITHUANIAN) :
216 (options & _FOLD_CASE_OPTIONS_MASK) == U_FOLD_CASE_DEFAULT)) {
217 latinToLower = LatinCase::TO_LOWER_NORMAL;
218 } else {
219 latinToLower = LatinCase::TO_LOWER_TR_LT;
220 }
221 const UTrie2 *trie = ucase_getTrie();
222 int32_t destIndex = 0;
223 int32_t prev = srcStart;
224 int32_t srcIndex = srcStart;
225 for (;;) {
226 // fast path for simple cases
227 char16_t lead = 0;
228 while (srcIndex < srcLimit) {
229 lead = src[srcIndex];
230 int32_t delta;
231 if (lead < LatinCase::LONG_S) {
232 int8_t d = latinToLower[lead];
233 if (d == LatinCase::EXC) { break; }
234 ++srcIndex;
235 if (d == 0) { continue; }
236 delta = d;
237 } else if (lead >= 0xd800) {
238 break; // surrogate or higher
239 } else {
240 uint16_t props = UTRIE2_GET16_FROM_U16_SINGLE_LEAD(trie, lead);
241 if (UCASE_HAS_EXCEPTION(props)) { break; }
242 ++srcIndex;
243 if (!UCASE_IS_UPPER_OR_TITLE(props) || (delta = UCASE_GET_DELTA(props)) == 0) {
244 continue;
245 }
246 }
247 lead += static_cast<char16_t>(delta);
248 destIndex = appendUnchanged(dest, destIndex, destCapacity,
249 src + prev, srcIndex - 1 - prev, options, edits);
250 if (destIndex >= 0) {
251 destIndex = appendUChar(dest, destIndex, destCapacity, lead);
252 if (edits != nullptr) {
253 edits->addReplace(1, 1);
254 }
255 }
256 if (destIndex < 0) {
257 errorCode = U_INDEX_OUTOFBOUNDS_ERROR;
258 return 0;
259 }
260 prev = srcIndex;
261 }
262 if (srcIndex >= srcLimit) {
263 break;
264 }
265 // slow path
266 int32_t cpStart = srcIndex++;
267 char16_t trail;
268 UChar32 c;
269 if (U16_IS_LEAD(lead) && srcIndex < srcLimit && U16_IS_TRAIL(trail = src[srcIndex])) {
270 c = U16_GET_SUPPLEMENTARY(lead, trail);
271 ++srcIndex;
272 } else {
273 c = lead;
274 }
275 const char16_t *s;
276 if (caseLocale >= 0) {
277 csc->cpStart = cpStart;
278 csc->cpLimit = srcIndex;
279 c = ucase_toFullLower(c, utf16_caseContextIterator, csc, &s, caseLocale);
280 } else {
281 c = ucase_toFullFolding(c, &s, options);
282 }
283 if (c >= 0) {
284 destIndex = appendUnchanged(dest, destIndex, destCapacity,
285 src + prev, cpStart - prev, options, edits);
286 if (destIndex >= 0) {
287 destIndex = appendResult(dest, destIndex, destCapacity, c, s,
288 srcIndex - cpStart, options, edits);
289 }
290 if (destIndex < 0) {
291 errorCode = U_INDEX_OUTOFBOUNDS_ERROR;
292 return 0;
293 }
294 prev = srcIndex;
295 }
296 }
297 destIndex = appendUnchanged(dest, destIndex, destCapacity,
298 src + prev, srcIndex - prev, options, edits);
299 if (destIndex < 0) {
300 errorCode = U_INDEX_OUTOFBOUNDS_ERROR;
301 return 0;
302 }
303 return destIndex;
304 }
305
toUpper(int32_t caseLocale,uint32_t options,char16_t * dest,int32_t destCapacity,const char16_t * src,UCaseContext * csc,int32_t srcLength,icu::Edits * edits,UErrorCode & errorCode)306 int32_t toUpper(int32_t caseLocale, uint32_t options,
307 char16_t *dest, int32_t destCapacity,
308 const char16_t *src, UCaseContext *csc, int32_t srcLength,
309 icu::Edits *edits, UErrorCode &errorCode) {
310 const int8_t *latinToUpper;
311 if (caseLocale == UCASE_LOC_TURKISH) {
312 latinToUpper = LatinCase::TO_UPPER_TR;
313 } else {
314 latinToUpper = LatinCase::TO_UPPER_NORMAL;
315 }
316 const UTrie2 *trie = ucase_getTrie();
317 int32_t destIndex = 0;
318 int32_t prev = 0;
319 int32_t srcIndex = 0;
320 for (;;) {
321 // fast path for simple cases
322 char16_t lead = 0;
323 while (srcIndex < srcLength) {
324 lead = src[srcIndex];
325 int32_t delta;
326 if (lead < LatinCase::LONG_S) {
327 int8_t d = latinToUpper[lead];
328 if (d == LatinCase::EXC) { break; }
329 ++srcIndex;
330 if (d == 0) { continue; }
331 delta = d;
332 } else if (lead >= 0xd800) {
333 break; // surrogate or higher
334 } else {
335 uint16_t props = UTRIE2_GET16_FROM_U16_SINGLE_LEAD(trie, lead);
336 if (UCASE_HAS_EXCEPTION(props)) { break; }
337 ++srcIndex;
338 if (UCASE_GET_TYPE(props) != UCASE_LOWER || (delta = UCASE_GET_DELTA(props)) == 0) {
339 continue;
340 }
341 }
342 lead += static_cast<char16_t>(delta);
343 destIndex = appendUnchanged(dest, destIndex, destCapacity,
344 src + prev, srcIndex - 1 - prev, options, edits);
345 if (destIndex >= 0) {
346 destIndex = appendUChar(dest, destIndex, destCapacity, lead);
347 if (edits != nullptr) {
348 edits->addReplace(1, 1);
349 }
350 }
351 if (destIndex < 0) {
352 errorCode = U_INDEX_OUTOFBOUNDS_ERROR;
353 return 0;
354 }
355 prev = srcIndex;
356 }
357 if (srcIndex >= srcLength) {
358 break;
359 }
360 // slow path
361 int32_t cpStart;
362 csc->cpStart = cpStart = srcIndex++;
363 char16_t trail;
364 UChar32 c;
365 if (U16_IS_LEAD(lead) && srcIndex < srcLength && U16_IS_TRAIL(trail = src[srcIndex])) {
366 c = U16_GET_SUPPLEMENTARY(lead, trail);
367 ++srcIndex;
368 } else {
369 c = lead;
370 }
371 csc->cpLimit = srcIndex;
372 const char16_t *s;
373 c = ucase_toFullUpper(c, utf16_caseContextIterator, csc, &s, caseLocale);
374 if (c >= 0) {
375 destIndex = appendUnchanged(dest, destIndex, destCapacity,
376 src + prev, cpStart - prev, options, edits);
377 if (destIndex >= 0) {
378 destIndex = appendResult(dest, destIndex, destCapacity, c, s,
379 srcIndex - cpStart, options, edits);
380 }
381 if (destIndex < 0) {
382 errorCode = U_INDEX_OUTOFBOUNDS_ERROR;
383 return 0;
384 }
385 prev = srcIndex;
386 }
387 }
388 destIndex = appendUnchanged(dest, destIndex, destCapacity,
389 src + prev, srcIndex - prev, options, edits);
390 if (destIndex < 0) {
391 errorCode = U_INDEX_OUTOFBOUNDS_ERROR;
392 return 0;
393 }
394 return destIndex;
395 }
396
397 } // namespace
398
399 U_NAMESPACE_END
400
401 U_NAMESPACE_USE
402
403 #if !UCONFIG_NO_BREAK_ITERATION
404
405 namespace {
406
407 /**
408 * Input: c is a letter I with or without acute accent.
409 * start is the index in src after c, and is less than segmentLimit.
410 * If a plain i/I is followed by a plain j/J,
411 * or an i/I with acute (precomposed or decomposed) is followed by a j/J with acute,
412 * then we output accordingly.
413 *
414 * @return the src index after the titlecased sequence, or the start index if no Dutch IJ
415 */
maybeTitleDutchIJ(const char16_t * src,UChar32 c,int32_t start,int32_t segmentLimit,char16_t * dest,int32_t & destIndex,int32_t destCapacity,uint32_t options,icu::Edits * edits)416 int32_t maybeTitleDutchIJ(const char16_t *src, UChar32 c, int32_t start, int32_t segmentLimit,
417 char16_t *dest, int32_t &destIndex, int32_t destCapacity, uint32_t options,
418 icu::Edits *edits) {
419 U_ASSERT(start < segmentLimit);
420
421 int32_t index = start;
422 bool withAcute = false;
423
424 // If the conditions are met, then the following variables tell us what to output.
425 int32_t unchanged1 = 0; // code units before the j, or the whole sequence (0..3)
426 bool doTitleJ = false; // true if the j needs to be titlecased
427 int32_t unchanged2 = 0; // after the j (0 or 1)
428
429 // next character after the first letter
430 char16_t c2 = src[index++];
431
432 // Is the first letter an i/I with accent?
433 if (c == u'I') {
434 if (c2 == ACUTE) {
435 withAcute = true;
436 unchanged1 = 1;
437 if (index == segmentLimit) { return start; }
438 c2 = src[index++];
439 }
440 } else { // Í
441 withAcute = true;
442 }
443
444 // Is the next character a j/J?
445 if (c2 == u'j') {
446 doTitleJ = true;
447 } else if (c2 == u'J') {
448 ++unchanged1;
449 } else {
450 return start;
451 }
452
453 // A plain i/I must be followed by a plain j/J.
454 // An i/I with acute must be followed by a j/J with acute.
455 if (withAcute) {
456 if (index == segmentLimit || src[index++] != ACUTE) { return start; }
457 if (doTitleJ) {
458 unchanged2 = 1;
459 } else {
460 ++unchanged1;
461 }
462 }
463
464 // There must not be another combining mark.
465 if (index < segmentLimit) {
466 int32_t cp;
467 int32_t i = index;
468 U16_NEXT(src, i, segmentLimit, cp);
469 uint32_t typeMask = U_GET_GC_MASK(cp);
470 if ((typeMask & U_GC_M_MASK) != 0) {
471 return start;
472 }
473 }
474
475 // Output the rest of the Dutch IJ.
476 destIndex = appendUnchanged(dest, destIndex, destCapacity, src + start, unchanged1, options, edits);
477 start += unchanged1;
478 if (doTitleJ) {
479 destIndex = appendUChar(dest, destIndex, destCapacity, u'J');
480 if (edits != nullptr) {
481 edits->addReplace(1, 1);
482 }
483 ++start;
484 }
485 destIndex = appendUnchanged(dest, destIndex, destCapacity, src + start, unchanged2, options, edits);
486
487 U_ASSERT(start + unchanged2 == index);
488 return index;
489 }
490
491 } // namespace
492
493 U_CFUNC int32_t U_CALLCONV
ustrcase_internalToTitle(int32_t caseLocale,uint32_t options,BreakIterator * iter,char16_t * dest,int32_t destCapacity,const char16_t * src,int32_t srcLength,icu::Edits * edits,UErrorCode & errorCode)494 ustrcase_internalToTitle(int32_t caseLocale, uint32_t options, BreakIterator *iter,
495 char16_t *dest, int32_t destCapacity,
496 const char16_t *src, int32_t srcLength,
497 icu::Edits *edits,
498 UErrorCode &errorCode) {
499 if (!ustrcase_checkTitleAdjustmentOptions(options, errorCode)) {
500 return 0;
501 }
502
503 /* set up local variables */
504 UCaseContext csc=UCASECONTEXT_INITIALIZER;
505 csc.p=(void *)src;
506 csc.limit=srcLength;
507 int32_t destIndex=0;
508 int32_t prev=0;
509 bool isFirstIndex=true;
510
511 /* titlecasing loop */
512 while(prev<srcLength) {
513 /* find next index where to titlecase */
514 int32_t index;
515 if(isFirstIndex) {
516 isFirstIndex=false;
517 index=iter->first();
518 } else {
519 index=iter->next();
520 }
521 if(index==UBRK_DONE || index>srcLength) {
522 index=srcLength;
523 }
524
525 /*
526 * Segment [prev..index[ into 3 parts:
527 * a) skipped characters (copy as-is) [prev..titleStart[
528 * b) first letter (titlecase) [titleStart..titleLimit[
529 * c) subsequent characters (lowercase) [titleLimit..index[
530 */
531 if(prev<index) {
532 // Find and copy skipped characters [prev..titleStart[
533 int32_t titleStart=prev;
534 int32_t titleLimit=prev;
535 UChar32 c;
536 U16_NEXT(src, titleLimit, index, c);
537 if ((options&U_TITLECASE_NO_BREAK_ADJUSTMENT)==0) {
538 // Adjust the titlecasing index to the next cased character,
539 // or to the next letter/number/symbol/private use.
540 // Stop with titleStart<titleLimit<=index
541 // if there is a character to be titlecased,
542 // or else stop with titleStart==titleLimit==index.
543 bool toCased = (options&U_TITLECASE_ADJUST_TO_CASED) != 0;
544 while (toCased ? UCASE_NONE==ucase_getType(c) : !ustrcase_isLNS(c)) {
545 titleStart=titleLimit;
546 if(titleLimit==index) {
547 break;
548 }
549 U16_NEXT(src, titleLimit, index, c);
550 }
551 if (prev < titleStart) {
552 destIndex=appendUnchanged(dest, destIndex, destCapacity,
553 src+prev, titleStart-prev, options, edits);
554 if(destIndex<0) {
555 errorCode=U_INDEX_OUTOFBOUNDS_ERROR;
556 return 0;
557 }
558 }
559 }
560
561 if(titleStart<titleLimit) {
562 /* titlecase c which is from [titleStart..titleLimit[ */
563 csc.cpStart=titleStart;
564 csc.cpLimit=titleLimit;
565 const char16_t *s;
566 c=ucase_toFullTitle(c, utf16_caseContextIterator, &csc, &s, caseLocale);
567 destIndex=appendResult(dest, destIndex, destCapacity, c, s,
568 titleLimit-titleStart, options, edits);
569 if(destIndex<0) {
570 errorCode=U_INDEX_OUTOFBOUNDS_ERROR;
571 return 0;
572 }
573
574 /* Special case Dutch IJ titlecasing */
575 if (titleStart+1 < index &&
576 caseLocale == UCASE_LOC_DUTCH) {
577 if (c < 0) {
578 c = ~c;
579 }
580
581 if (c == u'I' || c == u'Í') {
582 titleLimit = maybeTitleDutchIJ(src, c, titleStart + 1, index,
583 dest, destIndex, destCapacity, options,
584 edits);
585 }
586 }
587
588 /* lowercase [titleLimit..index[ */
589 if(titleLimit<index) {
590 if((options&U_TITLECASE_NO_LOWERCASE)==0) {
591 /* Normal operation: Lowercase the rest of the word. */
592 destIndex+=
593 toLower(
594 caseLocale, options,
595 (dest==nullptr) ? nullptr: dest+destIndex, destCapacity-destIndex,
596 src, &csc, titleLimit, index,
597 edits, errorCode);
598 if(errorCode==U_BUFFER_OVERFLOW_ERROR) {
599 errorCode=U_ZERO_ERROR;
600 }
601 if(U_FAILURE(errorCode)) {
602 return destIndex;
603 }
604 } else {
605 /* Optionally just copy the rest of the word unchanged. */
606 destIndex=appendUnchanged(dest, destIndex, destCapacity,
607 src+titleLimit, index-titleLimit, options, edits);
608 if(destIndex<0) {
609 errorCode=U_INDEX_OUTOFBOUNDS_ERROR;
610 return 0;
611 }
612 }
613 }
614 }
615 }
616
617 prev=index;
618 }
619
620 return checkOverflowAndEditsError(destIndex, destCapacity, edits, errorCode);
621 }
622
623 #endif // !UCONFIG_NO_BREAK_ITERATION
624
625 U_NAMESPACE_BEGIN
626 namespace GreekUpper {
627
628 // Data generated by prototype code, see
629 // https://icu.unicode.org/design/case/greek-upper
630 // TODO: Move this data into ucase.icu.
631 static const uint16_t data0370[] = {
632 // U+0370..03FF
633 0x0370,
634 0x0370,
635 0x0372,
636 0x0372,
637 0,
638 0,
639 0x0376,
640 0x0376,
641 0,
642 0,
643 0x037A,
644 0x03FD,
645 0x03FE,
646 0x03FF,
647 0,
648 0x037F,
649 0,
650 0,
651 0,
652 0,
653 0,
654 0,
655 0x0391 | HAS_VOWEL | HAS_ACCENT,
656 0,
657 0x0395 | HAS_VOWEL | HAS_ACCENT,
658 0x0397 | HAS_VOWEL | HAS_ACCENT,
659 0x0399 | HAS_VOWEL | HAS_ACCENT,
660 0,
661 0x039F | HAS_VOWEL | HAS_ACCENT,
662 0,
663 0x03A5 | HAS_VOWEL | HAS_ACCENT,
664 0x03A9 | HAS_VOWEL | HAS_ACCENT,
665 0x0399 | HAS_VOWEL | HAS_ACCENT | HAS_DIALYTIKA,
666 0x0391 | HAS_VOWEL,
667 0x0392,
668 0x0393,
669 0x0394,
670 0x0395 | HAS_VOWEL,
671 0x0396,
672 0x0397 | HAS_VOWEL,
673 0x0398,
674 0x0399 | HAS_VOWEL,
675 0x039A,
676 0x039B,
677 0x039C,
678 0x039D,
679 0x039E,
680 0x039F | HAS_VOWEL,
681 0x03A0,
682 0x03A1,
683 0,
684 0x03A3,
685 0x03A4,
686 0x03A5 | HAS_VOWEL,
687 0x03A6,
688 0x03A7,
689 0x03A8,
690 0x03A9 | HAS_VOWEL,
691 0x0399 | HAS_VOWEL | HAS_DIALYTIKA,
692 0x03A5 | HAS_VOWEL | HAS_DIALYTIKA,
693 0x0391 | HAS_VOWEL | HAS_ACCENT,
694 0x0395 | HAS_VOWEL | HAS_ACCENT,
695 0x0397 | HAS_VOWEL | HAS_ACCENT,
696 0x0399 | HAS_VOWEL | HAS_ACCENT,
697 0x03A5 | HAS_VOWEL | HAS_ACCENT | HAS_DIALYTIKA,
698 0x0391 | HAS_VOWEL,
699 0x0392,
700 0x0393,
701 0x0394,
702 0x0395 | HAS_VOWEL,
703 0x0396,
704 0x0397 | HAS_VOWEL,
705 0x0398,
706 0x0399 | HAS_VOWEL,
707 0x039A,
708 0x039B,
709 0x039C,
710 0x039D,
711 0x039E,
712 0x039F | HAS_VOWEL,
713 0x03A0,
714 0x03A1,
715 0x03A3,
716 0x03A3,
717 0x03A4,
718 0x03A5 | HAS_VOWEL,
719 0x03A6,
720 0x03A7,
721 0x03A8,
722 0x03A9 | HAS_VOWEL,
723 0x0399 | HAS_VOWEL | HAS_DIALYTIKA,
724 0x03A5 | HAS_VOWEL | HAS_DIALYTIKA,
725 0x039F | HAS_VOWEL | HAS_ACCENT,
726 0x03A5 | HAS_VOWEL | HAS_ACCENT,
727 0x03A9 | HAS_VOWEL | HAS_ACCENT,
728 0x03CF,
729 0x0392,
730 0x0398,
731 0x03D2,
732 0x03D2 | HAS_ACCENT,
733 0x03D2 | HAS_DIALYTIKA,
734 0x03A6,
735 0x03A0,
736 0x03CF,
737 0x03D8,
738 0x03D8,
739 0x03DA,
740 0x03DA,
741 0x03DC,
742 0x03DC,
743 0x03DE,
744 0x03DE,
745 0x03E0,
746 0x03E0,
747 0,
748 0,
749 0,
750 0,
751 0,
752 0,
753 0,
754 0,
755 0,
756 0,
757 0,
758 0,
759 0,
760 0,
761 0x039A,
762 0x03A1,
763 0x03F9,
764 0x037F,
765 0x03F4,
766 0x0395 | HAS_VOWEL,
767 0,
768 0x03F7,
769 0x03F7,
770 0x03F9,
771 0x03FA,
772 0x03FA,
773 0x03FC,
774 0x03FD,
775 0x03FE,
776 0x03FF,
777 };
778
779 static const uint16_t data1F00[] = {
780 // U+1F00..1FFF
781 0x0391 | HAS_VOWEL,
782 0x0391 | HAS_VOWEL,
783 0x0391 | HAS_VOWEL | HAS_ACCENT,
784 0x0391 | HAS_VOWEL | HAS_ACCENT,
785 0x0391 | HAS_VOWEL | HAS_ACCENT,
786 0x0391 | HAS_VOWEL | HAS_ACCENT,
787 0x0391 | HAS_VOWEL | HAS_ACCENT,
788 0x0391 | HAS_VOWEL | HAS_ACCENT,
789 0x0391 | HAS_VOWEL,
790 0x0391 | HAS_VOWEL,
791 0x0391 | HAS_VOWEL | HAS_ACCENT,
792 0x0391 | HAS_VOWEL | HAS_ACCENT,
793 0x0391 | HAS_VOWEL | HAS_ACCENT,
794 0x0391 | HAS_VOWEL | HAS_ACCENT,
795 0x0391 | HAS_VOWEL | HAS_ACCENT,
796 0x0391 | HAS_VOWEL | HAS_ACCENT,
797 0x0395 | HAS_VOWEL,
798 0x0395 | HAS_VOWEL,
799 0x0395 | HAS_VOWEL | HAS_ACCENT,
800 0x0395 | HAS_VOWEL | HAS_ACCENT,
801 0x0395 | HAS_VOWEL | HAS_ACCENT,
802 0x0395 | HAS_VOWEL | HAS_ACCENT,
803 0,
804 0,
805 0x0395 | HAS_VOWEL,
806 0x0395 | HAS_VOWEL,
807 0x0395 | HAS_VOWEL | HAS_ACCENT,
808 0x0395 | HAS_VOWEL | HAS_ACCENT,
809 0x0395 | HAS_VOWEL | HAS_ACCENT,
810 0x0395 | HAS_VOWEL | HAS_ACCENT,
811 0,
812 0,
813 0x0397 | HAS_VOWEL,
814 0x0397 | HAS_VOWEL,
815 0x0397 | HAS_VOWEL | HAS_ACCENT,
816 0x0397 | HAS_VOWEL | HAS_ACCENT,
817 0x0397 | HAS_VOWEL | HAS_ACCENT,
818 0x0397 | HAS_VOWEL | HAS_ACCENT,
819 0x0397 | HAS_VOWEL | HAS_ACCENT,
820 0x0397 | HAS_VOWEL | HAS_ACCENT,
821 0x0397 | HAS_VOWEL,
822 0x0397 | HAS_VOWEL,
823 0x0397 | HAS_VOWEL | HAS_ACCENT,
824 0x0397 | HAS_VOWEL | HAS_ACCENT,
825 0x0397 | HAS_VOWEL | HAS_ACCENT,
826 0x0397 | HAS_VOWEL | HAS_ACCENT,
827 0x0397 | HAS_VOWEL | HAS_ACCENT,
828 0x0397 | HAS_VOWEL | HAS_ACCENT,
829 0x0399 | HAS_VOWEL,
830 0x0399 | HAS_VOWEL,
831 0x0399 | HAS_VOWEL | HAS_ACCENT,
832 0x0399 | HAS_VOWEL | HAS_ACCENT,
833 0x0399 | HAS_VOWEL | HAS_ACCENT,
834 0x0399 | HAS_VOWEL | HAS_ACCENT,
835 0x0399 | HAS_VOWEL | HAS_ACCENT,
836 0x0399 | HAS_VOWEL | HAS_ACCENT,
837 0x0399 | HAS_VOWEL,
838 0x0399 | HAS_VOWEL,
839 0x0399 | HAS_VOWEL | HAS_ACCENT,
840 0x0399 | HAS_VOWEL | HAS_ACCENT,
841 0x0399 | HAS_VOWEL | HAS_ACCENT,
842 0x0399 | HAS_VOWEL | HAS_ACCENT,
843 0x0399 | HAS_VOWEL | HAS_ACCENT,
844 0x0399 | HAS_VOWEL | HAS_ACCENT,
845 0x039F | HAS_VOWEL,
846 0x039F | HAS_VOWEL,
847 0x039F | HAS_VOWEL | HAS_ACCENT,
848 0x039F | HAS_VOWEL | HAS_ACCENT,
849 0x039F | HAS_VOWEL | HAS_ACCENT,
850 0x039F | HAS_VOWEL | HAS_ACCENT,
851 0,
852 0,
853 0x039F | HAS_VOWEL,
854 0x039F | HAS_VOWEL,
855 0x039F | HAS_VOWEL | HAS_ACCENT,
856 0x039F | HAS_VOWEL | HAS_ACCENT,
857 0x039F | HAS_VOWEL | HAS_ACCENT,
858 0x039F | HAS_VOWEL | HAS_ACCENT,
859 0,
860 0,
861 0x03A5 | HAS_VOWEL,
862 0x03A5 | HAS_VOWEL,
863 0x03A5 | HAS_VOWEL | HAS_ACCENT,
864 0x03A5 | HAS_VOWEL | HAS_ACCENT,
865 0x03A5 | HAS_VOWEL | HAS_ACCENT,
866 0x03A5 | HAS_VOWEL | HAS_ACCENT,
867 0x03A5 | HAS_VOWEL | HAS_ACCENT,
868 0x03A5 | HAS_VOWEL | HAS_ACCENT,
869 0,
870 0x03A5 | HAS_VOWEL,
871 0,
872 0x03A5 | HAS_VOWEL | HAS_ACCENT,
873 0,
874 0x03A5 | HAS_VOWEL | HAS_ACCENT,
875 0,
876 0x03A5 | HAS_VOWEL | HAS_ACCENT,
877 0x03A9 | HAS_VOWEL,
878 0x03A9 | HAS_VOWEL,
879 0x03A9 | HAS_VOWEL | HAS_ACCENT,
880 0x03A9 | HAS_VOWEL | HAS_ACCENT,
881 0x03A9 | HAS_VOWEL | HAS_ACCENT,
882 0x03A9 | HAS_VOWEL | HAS_ACCENT,
883 0x03A9 | HAS_VOWEL | HAS_ACCENT,
884 0x03A9 | HAS_VOWEL | HAS_ACCENT,
885 0x03A9 | HAS_VOWEL,
886 0x03A9 | HAS_VOWEL,
887 0x03A9 | HAS_VOWEL | HAS_ACCENT,
888 0x03A9 | HAS_VOWEL | HAS_ACCENT,
889 0x03A9 | HAS_VOWEL | HAS_ACCENT,
890 0x03A9 | HAS_VOWEL | HAS_ACCENT,
891 0x03A9 | HAS_VOWEL | HAS_ACCENT,
892 0x03A9 | HAS_VOWEL | HAS_ACCENT,
893 0x0391 | HAS_VOWEL | HAS_ACCENT,
894 0x0391 | HAS_VOWEL | HAS_ACCENT,
895 0x0395 | HAS_VOWEL | HAS_ACCENT,
896 0x0395 | HAS_VOWEL | HAS_ACCENT,
897 0x0397 | HAS_VOWEL | HAS_ACCENT,
898 0x0397 | HAS_VOWEL | HAS_ACCENT,
899 0x0399 | HAS_VOWEL | HAS_ACCENT,
900 0x0399 | HAS_VOWEL | HAS_ACCENT,
901 0x039F | HAS_VOWEL | HAS_ACCENT,
902 0x039F | HAS_VOWEL | HAS_ACCENT,
903 0x03A5 | HAS_VOWEL | HAS_ACCENT,
904 0x03A5 | HAS_VOWEL | HAS_ACCENT,
905 0x03A9 | HAS_VOWEL | HAS_ACCENT,
906 0x03A9 | HAS_VOWEL | HAS_ACCENT,
907 0,
908 0,
909 0x0391 | HAS_VOWEL | HAS_YPOGEGRAMMENI,
910 0x0391 | HAS_VOWEL | HAS_YPOGEGRAMMENI,
911 0x0391 | HAS_VOWEL | HAS_YPOGEGRAMMENI | HAS_ACCENT,
912 0x0391 | HAS_VOWEL | HAS_YPOGEGRAMMENI | HAS_ACCENT,
913 0x0391 | HAS_VOWEL | HAS_YPOGEGRAMMENI | HAS_ACCENT,
914 0x0391 | HAS_VOWEL | HAS_YPOGEGRAMMENI | HAS_ACCENT,
915 0x0391 | HAS_VOWEL | HAS_YPOGEGRAMMENI | HAS_ACCENT,
916 0x0391 | HAS_VOWEL | HAS_YPOGEGRAMMENI | HAS_ACCENT,
917 0x0391 | HAS_VOWEL | HAS_YPOGEGRAMMENI,
918 0x0391 | HAS_VOWEL | HAS_YPOGEGRAMMENI,
919 0x0391 | HAS_VOWEL | HAS_YPOGEGRAMMENI | HAS_ACCENT,
920 0x0391 | HAS_VOWEL | HAS_YPOGEGRAMMENI | HAS_ACCENT,
921 0x0391 | HAS_VOWEL | HAS_YPOGEGRAMMENI | HAS_ACCENT,
922 0x0391 | HAS_VOWEL | HAS_YPOGEGRAMMENI | HAS_ACCENT,
923 0x0391 | HAS_VOWEL | HAS_YPOGEGRAMMENI | HAS_ACCENT,
924 0x0391 | HAS_VOWEL | HAS_YPOGEGRAMMENI | HAS_ACCENT,
925 0x0397 | HAS_VOWEL | HAS_YPOGEGRAMMENI,
926 0x0397 | HAS_VOWEL | HAS_YPOGEGRAMMENI,
927 0x0397 | HAS_VOWEL | HAS_YPOGEGRAMMENI | HAS_ACCENT,
928 0x0397 | HAS_VOWEL | HAS_YPOGEGRAMMENI | HAS_ACCENT,
929 0x0397 | HAS_VOWEL | HAS_YPOGEGRAMMENI | HAS_ACCENT,
930 0x0397 | HAS_VOWEL | HAS_YPOGEGRAMMENI | HAS_ACCENT,
931 0x0397 | HAS_VOWEL | HAS_YPOGEGRAMMENI | HAS_ACCENT,
932 0x0397 | HAS_VOWEL | HAS_YPOGEGRAMMENI | HAS_ACCENT,
933 0x0397 | HAS_VOWEL | HAS_YPOGEGRAMMENI,
934 0x0397 | HAS_VOWEL | HAS_YPOGEGRAMMENI,
935 0x0397 | HAS_VOWEL | HAS_YPOGEGRAMMENI | HAS_ACCENT,
936 0x0397 | HAS_VOWEL | HAS_YPOGEGRAMMENI | HAS_ACCENT,
937 0x0397 | HAS_VOWEL | HAS_YPOGEGRAMMENI | HAS_ACCENT,
938 0x0397 | HAS_VOWEL | HAS_YPOGEGRAMMENI | HAS_ACCENT,
939 0x0397 | HAS_VOWEL | HAS_YPOGEGRAMMENI | HAS_ACCENT,
940 0x0397 | HAS_VOWEL | HAS_YPOGEGRAMMENI | HAS_ACCENT,
941 0x03A9 | HAS_VOWEL | HAS_YPOGEGRAMMENI,
942 0x03A9 | HAS_VOWEL | HAS_YPOGEGRAMMENI,
943 0x03A9 | HAS_VOWEL | HAS_YPOGEGRAMMENI | HAS_ACCENT,
944 0x03A9 | HAS_VOWEL | HAS_YPOGEGRAMMENI | HAS_ACCENT,
945 0x03A9 | HAS_VOWEL | HAS_YPOGEGRAMMENI | HAS_ACCENT,
946 0x03A9 | HAS_VOWEL | HAS_YPOGEGRAMMENI | HAS_ACCENT,
947 0x03A9 | HAS_VOWEL | HAS_YPOGEGRAMMENI | HAS_ACCENT,
948 0x03A9 | HAS_VOWEL | HAS_YPOGEGRAMMENI | HAS_ACCENT,
949 0x03A9 | HAS_VOWEL | HAS_YPOGEGRAMMENI,
950 0x03A9 | HAS_VOWEL | HAS_YPOGEGRAMMENI,
951 0x03A9 | HAS_VOWEL | HAS_YPOGEGRAMMENI | HAS_ACCENT,
952 0x03A9 | HAS_VOWEL | HAS_YPOGEGRAMMENI | HAS_ACCENT,
953 0x03A9 | HAS_VOWEL | HAS_YPOGEGRAMMENI | HAS_ACCENT,
954 0x03A9 | HAS_VOWEL | HAS_YPOGEGRAMMENI | HAS_ACCENT,
955 0x03A9 | HAS_VOWEL | HAS_YPOGEGRAMMENI | HAS_ACCENT,
956 0x03A9 | HAS_VOWEL | HAS_YPOGEGRAMMENI | HAS_ACCENT,
957 0x0391 | HAS_VOWEL,
958 0x0391 | HAS_VOWEL,
959 0x0391 | HAS_VOWEL | HAS_YPOGEGRAMMENI | HAS_ACCENT,
960 0x0391 | HAS_VOWEL | HAS_YPOGEGRAMMENI,
961 0x0391 | HAS_VOWEL | HAS_YPOGEGRAMMENI | HAS_ACCENT,
962 0,
963 0x0391 | HAS_VOWEL | HAS_ACCENT,
964 0x0391 | HAS_VOWEL | HAS_YPOGEGRAMMENI | HAS_ACCENT,
965 0x0391 | HAS_VOWEL,
966 0x0391 | HAS_VOWEL,
967 0x0391 | HAS_VOWEL | HAS_ACCENT,
968 0x0391 | HAS_VOWEL | HAS_ACCENT,
969 0x0391 | HAS_VOWEL | HAS_YPOGEGRAMMENI,
970 0,
971 0x0399 | HAS_VOWEL,
972 0,
973 0,
974 0,
975 0x0397 | HAS_VOWEL | HAS_YPOGEGRAMMENI | HAS_ACCENT,
976 0x0397 | HAS_VOWEL | HAS_YPOGEGRAMMENI,
977 0x0397 | HAS_VOWEL | HAS_YPOGEGRAMMENI | HAS_ACCENT,
978 0,
979 0x0397 | HAS_VOWEL | HAS_ACCENT,
980 0x0397 | HAS_VOWEL | HAS_YPOGEGRAMMENI | HAS_ACCENT,
981 0x0395 | HAS_VOWEL | HAS_ACCENT,
982 0x0395 | HAS_VOWEL | HAS_ACCENT,
983 0x0397 | HAS_VOWEL | HAS_ACCENT,
984 0x0397 | HAS_VOWEL | HAS_ACCENT,
985 0x0397 | HAS_VOWEL | HAS_YPOGEGRAMMENI,
986 0,
987 0,
988 0,
989 0x0399 | HAS_VOWEL,
990 0x0399 | HAS_VOWEL,
991 0x0399 | HAS_VOWEL | HAS_ACCENT | HAS_DIALYTIKA,
992 0x0399 | HAS_VOWEL | HAS_ACCENT | HAS_DIALYTIKA,
993 0,
994 0,
995 0x0399 | HAS_VOWEL | HAS_ACCENT,
996 0x0399 | HAS_VOWEL | HAS_ACCENT | HAS_DIALYTIKA,
997 0x0399 | HAS_VOWEL,
998 0x0399 | HAS_VOWEL,
999 0x0399 | HAS_VOWEL | HAS_ACCENT,
1000 0x0399 | HAS_VOWEL | HAS_ACCENT,
1001 0,
1002 0,
1003 0,
1004 0,
1005 0x03A5 | HAS_VOWEL,
1006 0x03A5 | HAS_VOWEL,
1007 0x03A5 | HAS_VOWEL | HAS_ACCENT | HAS_DIALYTIKA,
1008 0x03A5 | HAS_VOWEL | HAS_ACCENT | HAS_DIALYTIKA,
1009 0x03A1,
1010 0x03A1,
1011 0x03A5 | HAS_VOWEL | HAS_ACCENT,
1012 0x03A5 | HAS_VOWEL | HAS_ACCENT | HAS_DIALYTIKA,
1013 0x03A5 | HAS_VOWEL,
1014 0x03A5 | HAS_VOWEL,
1015 0x03A5 | HAS_VOWEL | HAS_ACCENT,
1016 0x03A5 | HAS_VOWEL | HAS_ACCENT,
1017 0x03A1,
1018 0,
1019 0,
1020 0,
1021 0,
1022 0,
1023 0x03A9 | HAS_VOWEL | HAS_YPOGEGRAMMENI | HAS_ACCENT,
1024 0x03A9 | HAS_VOWEL | HAS_YPOGEGRAMMENI,
1025 0x03A9 | HAS_VOWEL | HAS_YPOGEGRAMMENI | HAS_ACCENT,
1026 0,
1027 0x03A9 | HAS_VOWEL | HAS_ACCENT,
1028 0x03A9 | HAS_VOWEL | HAS_YPOGEGRAMMENI | HAS_ACCENT,
1029 0x039F | HAS_VOWEL | HAS_ACCENT,
1030 0x039F | HAS_VOWEL | HAS_ACCENT,
1031 0x03A9 | HAS_VOWEL | HAS_ACCENT,
1032 0x03A9 | HAS_VOWEL | HAS_ACCENT,
1033 0x03A9 | HAS_VOWEL | HAS_YPOGEGRAMMENI,
1034 0,
1035 0,
1036 0,
1037 };
1038
1039 // U+2126 Ohm sign
1040 static const uint16_t data2126 = 0x03A9 | HAS_VOWEL;
1041
getLetterData(UChar32 c)1042 uint32_t getLetterData(UChar32 c) {
1043 if (c < 0x370 || 0x2126 < c || (0x3ff < c && c < 0x1f00)) {
1044 return 0;
1045 } else if (c <= 0x3ff) {
1046 return data0370[c - 0x370];
1047 } else if (c <= 0x1fff) {
1048 return data1F00[c - 0x1f00];
1049 } else if (c == 0x2126) {
1050 return data2126;
1051 } else {
1052 return 0;
1053 }
1054 }
1055
getDiacriticData(UChar32 c)1056 uint32_t getDiacriticData(UChar32 c) {
1057 switch (c) {
1058 case 0x0300: // varia
1059 case 0x0301: // tonos = oxia
1060 case 0x0342: // perispomeni
1061 case 0x0302: // circumflex can look like perispomeni
1062 case 0x0303: // tilde can look like perispomeni
1063 case 0x0311: // inverted breve can look like perispomeni
1064 return HAS_ACCENT;
1065 case 0x0308: // dialytika = diaeresis
1066 return HAS_COMBINING_DIALYTIKA;
1067 case 0x0344: // dialytika tonos
1068 return HAS_COMBINING_DIALYTIKA | HAS_ACCENT;
1069 case 0x0345: // ypogegrammeni = iota subscript
1070 return HAS_YPOGEGRAMMENI;
1071 case 0x0304: // macron
1072 case 0x0306: // breve
1073 case 0x0313: // comma above
1074 case 0x0314: // reversed comma above
1075 case 0x0343: // koronis
1076 return HAS_OTHER_GREEK_DIACRITIC;
1077 default:
1078 return 0;
1079 }
1080 }
1081
isFollowedByCasedLetter(const char16_t * s,int32_t i,int32_t length)1082 UBool isFollowedByCasedLetter(const char16_t *s, int32_t i, int32_t length) {
1083 while (i < length) {
1084 UChar32 c;
1085 U16_NEXT(s, i, length, c);
1086 int32_t type = ucase_getTypeOrIgnorable(c);
1087 if ((type & UCASE_IGNORABLE) != 0) {
1088 // Case-ignorable, continue with the loop.
1089 } else if (type != UCASE_NONE) {
1090 return true; // Followed by cased letter.
1091 } else {
1092 return false; // Uncased and not case-ignorable.
1093 }
1094 }
1095 return false; // Not followed by cased letter.
1096 }
1097
1098 /**
1099 * Greek string uppercasing with a state machine.
1100 * Probably simpler than a stateless function that has to figure out complex context-before
1101 * for each character.
1102 * TODO: Try to re-consolidate one way or another with the non-Greek function.
1103 */
toUpper(uint32_t options,char16_t * dest,int32_t destCapacity,const char16_t * src,int32_t srcLength,Edits * edits,UErrorCode & errorCode)1104 int32_t toUpper(uint32_t options,
1105 char16_t *dest, int32_t destCapacity,
1106 const char16_t *src, int32_t srcLength,
1107 Edits *edits,
1108 UErrorCode &errorCode) {
1109 int32_t destIndex=0;
1110 uint32_t state = 0;
1111 for (int32_t i = 0; i < srcLength;) {
1112 int32_t nextIndex = i;
1113 UChar32 c;
1114 U16_NEXT(src, nextIndex, srcLength, c);
1115 uint32_t nextState = 0;
1116 int32_t type = ucase_getTypeOrIgnorable(c);
1117 if ((type & UCASE_IGNORABLE) != 0) {
1118 // c is case-ignorable
1119 nextState |= (state & AFTER_CASED);
1120 } else if (type != UCASE_NONE) {
1121 // c is cased
1122 nextState |= AFTER_CASED;
1123 }
1124 uint32_t data = getLetterData(c);
1125 if (data > 0) {
1126 uint32_t upper = data & UPPER_MASK;
1127 // Add a dialytika to this iota or ypsilon vowel
1128 // if we removed a tonos from the previous vowel,
1129 // and that previous vowel did not also have (or gain) a dialytika.
1130 // Adding one only to the final vowel in a longer sequence
1131 // (which does not occur in normal writing) would require lookahead.
1132 // Set the same flag as for preserving an existing dialytika.
1133 if ((data & HAS_VOWEL) != 0 &&
1134 (state & (AFTER_VOWEL_WITH_PRECOMPOSED_ACCENT | AFTER_VOWEL_WITH_COMBINING_ACCENT)) !=
1135 0 &&
1136 (upper == 0x399 || upper == 0x3A5)) {
1137 data |= (state & AFTER_VOWEL_WITH_PRECOMPOSED_ACCENT) ? HAS_DIALYTIKA
1138 : HAS_COMBINING_DIALYTIKA;
1139 }
1140 int32_t numYpogegrammeni = 0; // Map each one to a trailing, spacing, capital iota.
1141 if ((data & HAS_YPOGEGRAMMENI) != 0) {
1142 numYpogegrammeni = 1;
1143 }
1144 const UBool hasPrecomposedAccent = (data & HAS_ACCENT) != 0;
1145 // Skip combining diacritics after this Greek letter.
1146 while (nextIndex < srcLength) {
1147 uint32_t diacriticData = getDiacriticData(src[nextIndex]);
1148 if (diacriticData != 0) {
1149 data |= diacriticData;
1150 if ((diacriticData & HAS_YPOGEGRAMMENI) != 0) {
1151 ++numYpogegrammeni;
1152 }
1153 ++nextIndex;
1154 } else {
1155 break; // not a Greek diacritic
1156 }
1157 }
1158 if ((data & HAS_VOWEL_AND_ACCENT_AND_DIALYTIKA) == HAS_VOWEL_AND_ACCENT) {
1159 nextState |= hasPrecomposedAccent ? AFTER_VOWEL_WITH_PRECOMPOSED_ACCENT
1160 : AFTER_VOWEL_WITH_COMBINING_ACCENT;
1161 }
1162 // Map according to Greek rules.
1163 UBool addTonos = false;
1164 if (upper == 0x397 &&
1165 (data & HAS_ACCENT) != 0 &&
1166 numYpogegrammeni == 0 &&
1167 (state & AFTER_CASED) == 0 &&
1168 !isFollowedByCasedLetter(src, nextIndex, srcLength)) {
1169 // Keep disjunctive "or" with (only) a tonos.
1170 // We use the same "word boundary" conditions as for the Final_Sigma test.
1171 if (hasPrecomposedAccent) {
1172 upper = 0x389; // Preserve the precomposed form.
1173 } else {
1174 addTonos = true;
1175 }
1176 } else if ((data & HAS_DIALYTIKA) != 0) {
1177 // Preserve a vowel with dialytika in precomposed form if it exists.
1178 if (upper == 0x399) {
1179 upper = 0x3AA;
1180 data &= ~HAS_EITHER_DIALYTIKA;
1181 } else if (upper == 0x3A5) {
1182 upper = 0x3AB;
1183 data &= ~HAS_EITHER_DIALYTIKA;
1184 }
1185 }
1186
1187 UBool change;
1188 if (edits == nullptr && (options & U_OMIT_UNCHANGED_TEXT) == 0) {
1189 change = true; // common, simple usage
1190 } else {
1191 // Find out first whether we are changing the text.
1192 change = src[i] != upper || numYpogegrammeni > 0;
1193 int32_t i2 = i + 1;
1194 if ((data & HAS_EITHER_DIALYTIKA) != 0) {
1195 change |= i2 >= nextIndex || src[i2] != 0x308;
1196 ++i2;
1197 }
1198 if (addTonos) {
1199 change |= i2 >= nextIndex || src[i2] != 0x301;
1200 ++i2;
1201 }
1202 int32_t oldLength = nextIndex - i;
1203 int32_t newLength = (i2 - i) + numYpogegrammeni;
1204 change |= oldLength != newLength;
1205 if (change) {
1206 if (edits != nullptr) {
1207 edits->addReplace(oldLength, newLength);
1208 }
1209 } else {
1210 if (edits != nullptr) {
1211 edits->addUnchanged(oldLength);
1212 }
1213 // Write unchanged text?
1214 change = (options & U_OMIT_UNCHANGED_TEXT) == 0;
1215 }
1216 }
1217
1218 if (change) {
1219 destIndex=appendUChar(dest, destIndex, destCapacity, (char16_t)upper);
1220 if (destIndex >= 0 && (data & HAS_EITHER_DIALYTIKA) != 0) {
1221 destIndex=appendUChar(dest, destIndex, destCapacity, 0x308); // restore or add a dialytika
1222 }
1223 if (destIndex >= 0 && addTonos) {
1224 destIndex=appendUChar(dest, destIndex, destCapacity, 0x301);
1225 }
1226 while (destIndex >= 0 && numYpogegrammeni > 0) {
1227 destIndex=appendUChar(dest, destIndex, destCapacity, 0x399);
1228 --numYpogegrammeni;
1229 }
1230 if(destIndex<0) {
1231 errorCode=U_INDEX_OUTOFBOUNDS_ERROR;
1232 return 0;
1233 }
1234 }
1235 } else {
1236 const char16_t *s;
1237 c=ucase_toFullUpper(c, nullptr, nullptr, &s, UCASE_LOC_GREEK);
1238 destIndex = appendResult(dest, destIndex, destCapacity, c, s,
1239 nextIndex - i, options, edits);
1240 if (destIndex < 0) {
1241 errorCode = U_INDEX_OUTOFBOUNDS_ERROR;
1242 return 0;
1243 }
1244 }
1245 i = nextIndex;
1246 state = nextState;
1247 }
1248
1249 return destIndex;
1250 }
1251
1252 } // namespace GreekUpper
1253 U_NAMESPACE_END
1254
1255 /* functions available in the common library (for unistr_case.cpp) */
1256
1257 U_CFUNC int32_t U_CALLCONV
ustrcase_internalToLower(int32_t caseLocale,uint32_t options,UCASEMAP_BREAK_ITERATOR_UNUSED char16_t * dest,int32_t destCapacity,const char16_t * src,int32_t srcLength,icu::Edits * edits,UErrorCode & errorCode)1258 ustrcase_internalToLower(int32_t caseLocale, uint32_t options, UCASEMAP_BREAK_ITERATOR_UNUSED
1259 char16_t *dest, int32_t destCapacity,
1260 const char16_t *src, int32_t srcLength,
1261 icu::Edits *edits,
1262 UErrorCode &errorCode) {
1263 UCaseContext csc=UCASECONTEXT_INITIALIZER;
1264 csc.p=(void *)src;
1265 csc.limit=srcLength;
1266 int32_t destIndex = toLower(
1267 caseLocale, options,
1268 dest, destCapacity,
1269 src, &csc, 0, srcLength,
1270 edits, errorCode);
1271 return checkOverflowAndEditsError(destIndex, destCapacity, edits, errorCode);
1272 }
1273
1274 U_CFUNC int32_t U_CALLCONV
ustrcase_internalToUpper(int32_t caseLocale,uint32_t options,UCASEMAP_BREAK_ITERATOR_UNUSED char16_t * dest,int32_t destCapacity,const char16_t * src,int32_t srcLength,icu::Edits * edits,UErrorCode & errorCode)1275 ustrcase_internalToUpper(int32_t caseLocale, uint32_t options, UCASEMAP_BREAK_ITERATOR_UNUSED
1276 char16_t *dest, int32_t destCapacity,
1277 const char16_t *src, int32_t srcLength,
1278 icu::Edits *edits,
1279 UErrorCode &errorCode) {
1280 int32_t destIndex;
1281 if (caseLocale == UCASE_LOC_GREEK) {
1282 destIndex = GreekUpper::toUpper(options, dest, destCapacity,
1283 src, srcLength, edits, errorCode);
1284 } else {
1285 UCaseContext csc=UCASECONTEXT_INITIALIZER;
1286 csc.p=(void *)src;
1287 csc.limit=srcLength;
1288 destIndex = toUpper(
1289 caseLocale, options,
1290 dest, destCapacity,
1291 src, &csc, srcLength,
1292 edits, errorCode);
1293 }
1294 return checkOverflowAndEditsError(destIndex, destCapacity, edits, errorCode);
1295 }
1296
1297 U_CFUNC int32_t U_CALLCONV
ustrcase_internalFold(int32_t,uint32_t options,UCASEMAP_BREAK_ITERATOR_UNUSED char16_t * dest,int32_t destCapacity,const char16_t * src,int32_t srcLength,icu::Edits * edits,UErrorCode & errorCode)1298 ustrcase_internalFold(int32_t /* caseLocale */, uint32_t options, UCASEMAP_BREAK_ITERATOR_UNUSED
1299 char16_t *dest, int32_t destCapacity,
1300 const char16_t *src, int32_t srcLength,
1301 icu::Edits *edits,
1302 UErrorCode &errorCode) {
1303 int32_t destIndex = toLower(
1304 -1, options,
1305 dest, destCapacity,
1306 src, nullptr, 0, srcLength,
1307 edits, errorCode);
1308 return checkOverflowAndEditsError(destIndex, destCapacity, edits, errorCode);
1309 }
1310
1311 U_CFUNC int32_t
ustrcase_map(int32_t caseLocale,uint32_t options,UCASEMAP_BREAK_ITERATOR_PARAM char16_t * dest,int32_t destCapacity,const char16_t * src,int32_t srcLength,UStringCaseMapper * stringCaseMapper,icu::Edits * edits,UErrorCode & errorCode)1312 ustrcase_map(int32_t caseLocale, uint32_t options, UCASEMAP_BREAK_ITERATOR_PARAM
1313 char16_t *dest, int32_t destCapacity,
1314 const char16_t *src, int32_t srcLength,
1315 UStringCaseMapper *stringCaseMapper,
1316 icu::Edits *edits,
1317 UErrorCode &errorCode) {
1318 int32_t destLength;
1319
1320 /* check argument values */
1321 if(U_FAILURE(errorCode)) {
1322 return 0;
1323 }
1324 if( destCapacity<0 ||
1325 (dest==nullptr && destCapacity>0) ||
1326 src==nullptr ||
1327 srcLength<-1
1328 ) {
1329 errorCode=U_ILLEGAL_ARGUMENT_ERROR;
1330 return 0;
1331 }
1332
1333 /* get the string length */
1334 if(srcLength==-1) {
1335 srcLength=u_strlen(src);
1336 }
1337
1338 /* check for overlapping source and destination */
1339 if( dest!=nullptr &&
1340 ((src>=dest && src<(dest+destCapacity)) ||
1341 (dest>=src && dest<(src+srcLength)))
1342 ) {
1343 errorCode=U_ILLEGAL_ARGUMENT_ERROR;
1344 return 0;
1345 }
1346
1347 if (edits != nullptr && (options & U_EDITS_NO_RESET) == 0) {
1348 edits->reset();
1349 }
1350 destLength=stringCaseMapper(caseLocale, options, UCASEMAP_BREAK_ITERATOR
1351 dest, destCapacity, src, srcLength, edits, errorCode);
1352 return u_terminateUChars(dest, destCapacity, destLength, &errorCode);
1353 }
1354
1355 U_CFUNC int32_t
ustrcase_mapWithOverlap(int32_t caseLocale,uint32_t options,UCASEMAP_BREAK_ITERATOR_PARAM char16_t * dest,int32_t destCapacity,const char16_t * src,int32_t srcLength,UStringCaseMapper * stringCaseMapper,UErrorCode & errorCode)1356 ustrcase_mapWithOverlap(int32_t caseLocale, uint32_t options, UCASEMAP_BREAK_ITERATOR_PARAM
1357 char16_t *dest, int32_t destCapacity,
1358 const char16_t *src, int32_t srcLength,
1359 UStringCaseMapper *stringCaseMapper,
1360 UErrorCode &errorCode) {
1361 char16_t buffer[300];
1362 char16_t *temp;
1363
1364 int32_t destLength;
1365
1366 /* check argument values */
1367 if(U_FAILURE(errorCode)) {
1368 return 0;
1369 }
1370 if( destCapacity<0 ||
1371 (dest==nullptr && destCapacity>0) ||
1372 src==nullptr ||
1373 srcLength<-1
1374 ) {
1375 errorCode=U_ILLEGAL_ARGUMENT_ERROR;
1376 return 0;
1377 }
1378
1379 /* get the string length */
1380 if(srcLength==-1) {
1381 srcLength=u_strlen(src);
1382 }
1383
1384 /* check for overlapping source and destination */
1385 if( dest!=nullptr &&
1386 ((src>=dest && src<(dest+destCapacity)) ||
1387 (dest>=src && dest<(src+srcLength)))
1388 ) {
1389 /* overlap: provide a temporary destination buffer and later copy the result */
1390 if(destCapacity<=UPRV_LENGTHOF(buffer)) {
1391 /* the stack buffer is large enough */
1392 temp=buffer;
1393 } else {
1394 /* allocate a buffer */
1395 temp=(char16_t *)uprv_malloc(destCapacity*U_SIZEOF_UCHAR);
1396 if(temp==nullptr) {
1397 errorCode=U_MEMORY_ALLOCATION_ERROR;
1398 return 0;
1399 }
1400 }
1401 } else {
1402 temp=dest;
1403 }
1404
1405 destLength=stringCaseMapper(caseLocale, options, UCASEMAP_BREAK_ITERATOR
1406 temp, destCapacity, src, srcLength, nullptr, errorCode);
1407 if(temp!=dest) {
1408 /* copy the result string to the destination buffer */
1409 if (U_SUCCESS(errorCode) && 0 < destLength && destLength <= destCapacity) {
1410 u_memmove(dest, temp, destLength);
1411 }
1412 if(temp!=buffer) {
1413 uprv_free(temp);
1414 }
1415 }
1416
1417 return u_terminateUChars(dest, destCapacity, destLength, &errorCode);
1418 }
1419
1420 /* public API functions */
1421
1422 U_CAPI int32_t U_EXPORT2
u_strFoldCase(char16_t * dest,int32_t destCapacity,const char16_t * src,int32_t srcLength,uint32_t options,UErrorCode * pErrorCode)1423 u_strFoldCase(char16_t *dest, int32_t destCapacity,
1424 const char16_t *src, int32_t srcLength,
1425 uint32_t options,
1426 UErrorCode *pErrorCode) {
1427 return ustrcase_mapWithOverlap(
1428 UCASE_LOC_ROOT, options, UCASEMAP_BREAK_ITERATOR_NULL
1429 dest, destCapacity,
1430 src, srcLength,
1431 ustrcase_internalFold, *pErrorCode);
1432 }
1433
1434 U_NAMESPACE_BEGIN
1435
fold(uint32_t options,const char16_t * src,int32_t srcLength,char16_t * dest,int32_t destCapacity,Edits * edits,UErrorCode & errorCode)1436 int32_t CaseMap::fold(
1437 uint32_t options,
1438 const char16_t *src, int32_t srcLength,
1439 char16_t *dest, int32_t destCapacity, Edits *edits,
1440 UErrorCode &errorCode) {
1441 return ustrcase_map(
1442 UCASE_LOC_ROOT, options, UCASEMAP_BREAK_ITERATOR_NULL
1443 dest, destCapacity,
1444 src, srcLength,
1445 ustrcase_internalFold, edits, errorCode);
1446 }
1447
1448 U_NAMESPACE_END
1449
1450 /* case-insensitive string comparisons -------------------------------------- */
1451
1452 /*
1453 * This function is a copy of unorm_cmpEquivFold() minus the parts for
1454 * canonical equivalence.
1455 * Keep the functions in sync, and see there for how this works.
1456 * The duplication is for modularization:
1457 * It makes caseless (but not canonical caseless) matches independent of
1458 * the normalization code.
1459 */
1460
1461 /* stack element for previous-level source/decomposition pointers */
1462 struct CmpEquivLevel {
1463 const char16_t *start, *s, *limit;
1464 };
1465 typedef struct CmpEquivLevel CmpEquivLevel;
1466
1467 /**
1468 * Internal implementation code comparing string with case fold.
1469 * This function is called from u_strcmpFold() and u_caseInsensitivePrefixMatch().
1470 *
1471 * @param s1 input string 1
1472 * @param length1 length of string 1, or -1 (NUL terminated)
1473 * @param s2 input string 2
1474 * @param length2 length of string 2, or -1 (NUL terminated)
1475 * @param options compare options
1476 * @param matchLen1 (output) length of partial prefix match in s1
1477 * @param matchLen2 (output) length of partial prefix match in s2
1478 * @param pErrorCode receives error status
1479 * @return The result of comparison
1480 */
_cmpFold(const char16_t * s1,int32_t length1,const char16_t * s2,int32_t length2,uint32_t options,int32_t * matchLen1,int32_t * matchLen2,UErrorCode * pErrorCode)1481 static int32_t _cmpFold(
1482 const char16_t *s1, int32_t length1,
1483 const char16_t *s2, int32_t length2,
1484 uint32_t options,
1485 int32_t *matchLen1, int32_t *matchLen2,
1486 UErrorCode *pErrorCode) {
1487 int32_t cmpRes = 0;
1488
1489 /* current-level start/limit - s1/s2 as current */
1490 const char16_t *start1, *start2, *limit1, *limit2;
1491
1492 /* points to the original start address */
1493 const char16_t *org1, *org2;
1494
1495 /* points to the end of match + 1 */
1496 const char16_t *m1, *m2;
1497
1498 /* case folding variables */
1499 const char16_t *p;
1500 int32_t length;
1501
1502 /* stacks of previous-level start/current/limit */
1503 CmpEquivLevel stack1[2], stack2[2];
1504
1505 /* case folding buffers, only use current-level start/limit */
1506 char16_t fold1[UCASE_MAX_STRING_LENGTH+1], fold2[UCASE_MAX_STRING_LENGTH+1];
1507
1508 /* track which is the current level per string */
1509 int32_t level1, level2;
1510
1511 /* current code units, and code points for lookups */
1512 UChar32 c1, c2, cp1, cp2;
1513
1514 /* no argument error checking because this itself is not an API */
1515
1516 /*
1517 * assume that at least the option U_COMPARE_IGNORE_CASE is set
1518 * otherwise this function would have to behave exactly as uprv_strCompare()
1519 */
1520 if(U_FAILURE(*pErrorCode)) {
1521 return 0;
1522 }
1523
1524 /* initialize */
1525 if(matchLen1) {
1526 U_ASSERT(matchLen2 !=nullptr);
1527 *matchLen1=0;
1528 *matchLen2=0;
1529 }
1530
1531 start1=m1=org1=s1;
1532 if(length1==-1) {
1533 limit1=nullptr;
1534 } else {
1535 limit1=s1+length1;
1536 }
1537
1538 start2=m2=org2=s2;
1539 if(length2==-1) {
1540 limit2=nullptr;
1541 } else {
1542 limit2=s2+length2;
1543 }
1544
1545 level1=level2=0;
1546 c1=c2=-1;
1547
1548 /* comparison loop */
1549 for(;;) {
1550 /*
1551 * here a code unit value of -1 means "get another code unit"
1552 * below it will mean "this source is finished"
1553 */
1554
1555 if(c1<0) {
1556 /* get next code unit from string 1, post-increment */
1557 for(;;) {
1558 if(s1==limit1 || ((c1=*s1)==0 && (limit1==nullptr || (options&_STRNCMP_STYLE)))) {
1559 if(level1==0) {
1560 c1=-1;
1561 break;
1562 }
1563 } else {
1564 ++s1;
1565 break;
1566 }
1567
1568 /* reached end of level buffer, pop one level */
1569 do {
1570 --level1;
1571 start1=stack1[level1].start; /*Not uninitialized*/
1572 } while(start1==nullptr);
1573 s1=stack1[level1].s; /*Not uninitialized*/
1574 limit1=stack1[level1].limit; /*Not uninitialized*/
1575 }
1576 }
1577
1578 if(c2<0) {
1579 /* get next code unit from string 2, post-increment */
1580 for(;;) {
1581 if(s2==limit2 || ((c2=*s2)==0 && (limit2==nullptr || (options&_STRNCMP_STYLE)))) {
1582 if(level2==0) {
1583 c2=-1;
1584 break;
1585 }
1586 } else {
1587 ++s2;
1588 break;
1589 }
1590
1591 /* reached end of level buffer, pop one level */
1592 do {
1593 --level2;
1594 start2=stack2[level2].start; /*Not uninitialized*/
1595 } while(start2==nullptr);
1596 s2=stack2[level2].s; /*Not uninitialized*/
1597 limit2=stack2[level2].limit; /*Not uninitialized*/
1598 }
1599 }
1600
1601 /*
1602 * compare c1 and c2
1603 * either variable c1, c2 is -1 only if the corresponding string is finished
1604 */
1605 if(c1==c2) {
1606 const char16_t *next1, *next2;
1607
1608 if(c1<0) {
1609 cmpRes=0; /* c1==c2==-1 indicating end of strings */
1610 break;
1611 }
1612
1613 /*
1614 * Note: Move the match positions in both strings at the same time
1615 * only when corresponding code point(s) in the original strings
1616 * are fully consumed. For example, when comparing s1="Fust" and
1617 * s2="Fu\u00dfball", s2[2] is folded into "ss", and s1[2] matches
1618 * the first code point in the case-folded data. But the second "s"
1619 * has no matching code point in s1, so this implementation returns
1620 * 2 as the prefix match length ("Fu").
1621 */
1622 next1=next2=nullptr;
1623 if(level1==0) {
1624 next1=s1;
1625 } else if(s1==limit1) {
1626 /* Note: This implementation only use a single level of stack.
1627 * If this code needs to be changed to use multiple levels
1628 * of stacks, the code above should check if the current
1629 * code is at the end of all stacks.
1630 */
1631 U_ASSERT(level1==1);
1632
1633 /* is s1 at the end of the current stack? */
1634 next1=stack1[0].s;
1635 }
1636
1637 if (next1!=nullptr) {
1638 if(level2==0) {
1639 next2=s2;
1640 } else if(s2==limit2) {
1641 U_ASSERT(level2==1);
1642
1643 /* is s2 at the end of the current stack? */
1644 next2=stack2[0].s;
1645 }
1646 if(next2!=nullptr) {
1647 m1=next1;
1648 m2=next2;
1649 }
1650 }
1651 c1=c2=-1; /* make us fetch new code units */
1652 continue;
1653 } else if(c1<0) {
1654 cmpRes=-1; /* string 1 ends before string 2 */
1655 break;
1656 } else if(c2<0) {
1657 cmpRes=1; /* string 2 ends before string 1 */
1658 break;
1659 }
1660 /* c1!=c2 && c1>=0 && c2>=0 */
1661
1662 /* get complete code points for c1, c2 for lookups if either is a surrogate */
1663 cp1=c1;
1664 if(U_IS_SURROGATE(c1)) {
1665 char16_t c;
1666
1667 if(U_IS_SURROGATE_LEAD(c1)) {
1668 if(s1!=limit1 && U16_IS_TRAIL(c=*s1)) {
1669 /* advance ++s1; only below if cp1 decomposes/case-folds */
1670 cp1=U16_GET_SUPPLEMENTARY(c1, c);
1671 }
1672 } else /* isTrail(c1) */ {
1673 if(start1<=(s1-2) && U16_IS_LEAD(c=*(s1-2))) {
1674 cp1=U16_GET_SUPPLEMENTARY(c, c1);
1675 }
1676 }
1677 }
1678
1679 cp2=c2;
1680 if(U_IS_SURROGATE(c2)) {
1681 char16_t c;
1682
1683 if(U_IS_SURROGATE_LEAD(c2)) {
1684 if(s2!=limit2 && U16_IS_TRAIL(c=*s2)) {
1685 /* advance ++s2; only below if cp2 decomposes/case-folds */
1686 cp2=U16_GET_SUPPLEMENTARY(c2, c);
1687 }
1688 } else /* isTrail(c2) */ {
1689 if(start2<=(s2-2) && U16_IS_LEAD(c=*(s2-2))) {
1690 cp2=U16_GET_SUPPLEMENTARY(c, c2);
1691 }
1692 }
1693 }
1694
1695 /*
1696 * go down one level for each string
1697 * continue with the main loop as soon as there is a real change
1698 */
1699
1700 if( level1==0 &&
1701 (length=ucase_toFullFolding((UChar32)cp1, &p, options))>=0
1702 ) {
1703 /* cp1 case-folds to the code point "length" or to p[length] */
1704 if(U_IS_SURROGATE(c1)) {
1705 if(U_IS_SURROGATE_LEAD(c1)) {
1706 /* advance beyond source surrogate pair if it case-folds */
1707 ++s1;
1708 } else /* isTrail(c1) */ {
1709 /*
1710 * we got a supplementary code point when hitting its trail surrogate,
1711 * therefore the lead surrogate must have been the same as in the other string;
1712 * compare this decomposition with the lead surrogate in the other string
1713 * remember that this simulates bulk text replacement:
1714 * the decomposition would replace the entire code point
1715 */
1716 --s2;
1717 --m2;
1718 c2=*(s2-1);
1719 }
1720 }
1721
1722 /* push current level pointers */
1723 stack1[0].start=start1;
1724 stack1[0].s=s1;
1725 stack1[0].limit=limit1;
1726 ++level1;
1727
1728 /* copy the folding result to fold1[] */
1729 if(length<=UCASE_MAX_STRING_LENGTH) {
1730 u_memcpy(fold1, p, length);
1731 } else {
1732 int32_t i=0;
1733 U16_APPEND_UNSAFE(fold1, i, length);
1734 length=i;
1735 }
1736
1737 /* set next level pointers to case folding */
1738 start1=s1=fold1;
1739 limit1=fold1+length;
1740
1741 /* get ready to read from decomposition, continue with loop */
1742 c1=-1;
1743 continue;
1744 }
1745
1746 if( level2==0 &&
1747 (length=ucase_toFullFolding((UChar32)cp2, &p, options))>=0
1748 ) {
1749 /* cp2 case-folds to the code point "length" or to p[length] */
1750 if(U_IS_SURROGATE(c2)) {
1751 if(U_IS_SURROGATE_LEAD(c2)) {
1752 /* advance beyond source surrogate pair if it case-folds */
1753 ++s2;
1754 } else /* isTrail(c2) */ {
1755 /*
1756 * we got a supplementary code point when hitting its trail surrogate,
1757 * therefore the lead surrogate must have been the same as in the other string;
1758 * compare this decomposition with the lead surrogate in the other string
1759 * remember that this simulates bulk text replacement:
1760 * the decomposition would replace the entire code point
1761 */
1762 --s1;
1763 --m2;
1764 c1=*(s1-1);
1765 }
1766 }
1767
1768 /* push current level pointers */
1769 stack2[0].start=start2;
1770 stack2[0].s=s2;
1771 stack2[0].limit=limit2;
1772 ++level2;
1773
1774 /* copy the folding result to fold2[] */
1775 if(length<=UCASE_MAX_STRING_LENGTH) {
1776 u_memcpy(fold2, p, length);
1777 } else {
1778 int32_t i=0;
1779 U16_APPEND_UNSAFE(fold2, i, length);
1780 length=i;
1781 }
1782
1783 /* set next level pointers to case folding */
1784 start2=s2=fold2;
1785 limit2=fold2+length;
1786
1787 /* get ready to read from decomposition, continue with loop */
1788 c2=-1;
1789 continue;
1790 }
1791
1792 /*
1793 * no decomposition/case folding, max level for both sides:
1794 * return difference result
1795 *
1796 * code point order comparison must not just return cp1-cp2
1797 * because when single surrogates are present then the surrogate pairs
1798 * that formed cp1 and cp2 may be from different string indexes
1799 *
1800 * example: { d800 d800 dc01 } vs. { d800 dc00 }, compare at second code units
1801 * c1=d800 cp1=10001 c2=dc00 cp2=10000
1802 * cp1-cp2>0 but c1-c2<0 and in fact in UTF-32 it is { d800 10001 } < { 10000 }
1803 *
1804 * therefore, use same fix-up as in ustring.c/uprv_strCompare()
1805 * except: uprv_strCompare() fetches c=*s while this functions fetches c=*s++
1806 * so we have slightly different pointer/start/limit comparisons here
1807 */
1808
1809 if(c1>=0xd800 && c2>=0xd800 && (options&U_COMPARE_CODE_POINT_ORDER)) {
1810 /* subtract 0x2800 from BMP code points to make them smaller than supplementary ones */
1811 if(
1812 (c1<=0xdbff && s1!=limit1 && U16_IS_TRAIL(*s1)) ||
1813 (U16_IS_TRAIL(c1) && start1!=(s1-1) && U16_IS_LEAD(*(s1-2)))
1814 ) {
1815 /* part of a surrogate pair, leave >=d800 */
1816 } else {
1817 /* BMP code point - may be surrogate code point - make <d800 */
1818 c1-=0x2800;
1819 }
1820
1821 if(
1822 (c2<=0xdbff && s2!=limit2 && U16_IS_TRAIL(*s2)) ||
1823 (U16_IS_TRAIL(c2) && start2!=(s2-1) && U16_IS_LEAD(*(s2-2)))
1824 ) {
1825 /* part of a surrogate pair, leave >=d800 */
1826 } else {
1827 /* BMP code point - may be surrogate code point - make <d800 */
1828 c2-=0x2800;
1829 }
1830 }
1831
1832 cmpRes=c1-c2;
1833 break;
1834 }
1835
1836 if(matchLen1) {
1837 *matchLen1=static_cast<int32_t>(m1-org1);
1838 *matchLen2=static_cast<int32_t>(m2-org2);
1839 }
1840 return cmpRes;
1841 }
1842
1843 /* internal function */
1844 U_CFUNC int32_t
u_strcmpFold(const char16_t * s1,int32_t length1,const char16_t * s2,int32_t length2,uint32_t options,UErrorCode * pErrorCode)1845 u_strcmpFold(const char16_t *s1, int32_t length1,
1846 const char16_t *s2, int32_t length2,
1847 uint32_t options,
1848 UErrorCode *pErrorCode) {
1849 return _cmpFold(s1, length1, s2, length2, options, nullptr, nullptr, pErrorCode);
1850 }
1851
1852 /* public API functions */
1853
1854 U_CAPI int32_t U_EXPORT2
u_strCaseCompare(const char16_t * s1,int32_t length1,const char16_t * s2,int32_t length2,uint32_t options,UErrorCode * pErrorCode)1855 u_strCaseCompare(const char16_t *s1, int32_t length1,
1856 const char16_t *s2, int32_t length2,
1857 uint32_t options,
1858 UErrorCode *pErrorCode) {
1859 /* argument checking */
1860 if(pErrorCode==0 || U_FAILURE(*pErrorCode)) {
1861 return 0;
1862 }
1863 if(s1==nullptr || length1<-1 || s2==nullptr || length2<-1) {
1864 *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR;
1865 return 0;
1866 }
1867 return u_strcmpFold(s1, length1, s2, length2,
1868 options|U_COMPARE_IGNORE_CASE,
1869 pErrorCode);
1870 }
1871
1872 U_CAPI int32_t U_EXPORT2
u_strcasecmp(const char16_t * s1,const char16_t * s2,uint32_t options)1873 u_strcasecmp(const char16_t *s1, const char16_t *s2, uint32_t options) {
1874 UErrorCode errorCode=U_ZERO_ERROR;
1875 return u_strcmpFold(s1, -1, s2, -1,
1876 options|U_COMPARE_IGNORE_CASE,
1877 &errorCode);
1878 }
1879
1880 U_CAPI int32_t U_EXPORT2
u_memcasecmp(const char16_t * s1,const char16_t * s2,int32_t length,uint32_t options)1881 u_memcasecmp(const char16_t *s1, const char16_t *s2, int32_t length, uint32_t options) {
1882 UErrorCode errorCode=U_ZERO_ERROR;
1883 return u_strcmpFold(s1, length, s2, length,
1884 options|U_COMPARE_IGNORE_CASE,
1885 &errorCode);
1886 }
1887
1888 U_CAPI int32_t U_EXPORT2
u_strncasecmp(const char16_t * s1,const char16_t * s2,int32_t n,uint32_t options)1889 u_strncasecmp(const char16_t *s1, const char16_t *s2, int32_t n, uint32_t options) {
1890 UErrorCode errorCode=U_ZERO_ERROR;
1891 return u_strcmpFold(s1, n, s2, n,
1892 options|(U_COMPARE_IGNORE_CASE|_STRNCMP_STYLE),
1893 &errorCode);
1894 }
1895
1896 /* internal API - detect length of shared prefix */
1897 U_CAPI void
u_caseInsensitivePrefixMatch(const char16_t * s1,int32_t length1,const char16_t * s2,int32_t length2,uint32_t options,int32_t * matchLen1,int32_t * matchLen2,UErrorCode * pErrorCode)1898 u_caseInsensitivePrefixMatch(const char16_t *s1, int32_t length1,
1899 const char16_t *s2, int32_t length2,
1900 uint32_t options,
1901 int32_t *matchLen1, int32_t *matchLen2,
1902 UErrorCode *pErrorCode) {
1903 _cmpFold(s1, length1, s2, length2, options,
1904 matchLen1, matchLen2, pErrorCode);
1905 }
1906