1 /*
2 *******************************************************************************
3 *
4 * Copyright (C) 2001-2007, International Business Machines
5 * Corporation and others. All Rights Reserved.
6 *
7 *******************************************************************************
8 * file name: ustrcase.c
9 * encoding: US-ASCII
10 * tab size: 8 (not used)
11 * indentation:4
12 *
13 * created on: 2002feb20
14 * created by: Markus W. Scherer
15 *
16 * Implementation file for string casing C API functions.
17 * Uses functions from uchar.c for basic functionality that requires access
18 * to the Unicode Character Database (uprops.dat).
19 */
20
21 #include "unicode/utypes.h"
22 #include "unicode/uloc.h"
23 #include "unicode/ustring.h"
24 #include "unicode/ucasemap.h"
25 #include "unicode/ubrk.h"
26 #include "cmemory.h"
27 #include "ucase.h"
28 #include "unormimp.h"
29 #include "ustr_imp.h"
30
31 /* string casing ------------------------------------------------------------ */
32
33 /* append a full case mapping result, see UCASE_MAX_STRING_LENGTH */
34 static U_INLINE int32_t
appendResult(UChar * dest,int32_t destIndex,int32_t destCapacity,int32_t result,const UChar * s)35 appendResult(UChar *dest, int32_t destIndex, int32_t destCapacity,
36 int32_t result, const UChar *s) {
37 UChar32 c;
38 int32_t length;
39
40 /* decode the result */
41 if(result<0) {
42 /* (not) original code point */
43 c=~result;
44 length=-1;
45 } else if(result<=UCASE_MAX_STRING_LENGTH) {
46 c=U_SENTINEL;
47 length=result;
48 } else {
49 c=result;
50 length=-1;
51 }
52
53 if(destIndex<destCapacity) {
54 /* append the result */
55 if(length<0) {
56 /* code point */
57 UBool isError=FALSE;
58 U16_APPEND(dest, destIndex, destCapacity, c, isError);
59 if(isError) {
60 /* overflow, nothing written */
61 destIndex+=U16_LENGTH(c);
62 }
63 } else {
64 /* string */
65 if((destIndex+length)<=destCapacity) {
66 while(length>0) {
67 dest[destIndex++]=*s++;
68 --length;
69 }
70 } else {
71 /* overflow */
72 destIndex+=length;
73 }
74 }
75 } else {
76 /* preflight */
77 if(length<0) {
78 destIndex+=U16_LENGTH(c);
79 } else {
80 destIndex+=length;
81 }
82 }
83 return destIndex;
84 }
85
86 static UChar32 U_CALLCONV
utf16_caseContextIterator(void * context,int8_t dir)87 utf16_caseContextIterator(void *context, int8_t dir) {
88 UCaseContext *csc=(UCaseContext *)context;
89 UChar32 c;
90
91 if(dir<0) {
92 /* reset for backward iteration */
93 csc->index=csc->cpStart;
94 csc->dir=dir;
95 } else if(dir>0) {
96 /* reset for forward iteration */
97 csc->index=csc->cpLimit;
98 csc->dir=dir;
99 } else {
100 /* continue current iteration direction */
101 dir=csc->dir;
102 }
103
104 if(dir<0) {
105 if(csc->start<csc->index) {
106 U16_PREV((const UChar *)csc->p, csc->start, csc->index, c);
107 return c;
108 }
109 } else {
110 if(csc->index<csc->limit) {
111 U16_NEXT((const UChar *)csc->p, csc->index, csc->limit, c);
112 return c;
113 }
114 }
115 return U_SENTINEL;
116 }
117
118 /*
119 * Case-maps [srcStart..srcLimit[ but takes
120 * context [0..srcLength[ into account.
121 */
122 static int32_t
_caseMap(const UCaseMap * csm,UCaseMapFull * map,UChar * dest,int32_t destCapacity,const UChar * src,UCaseContext * csc,int32_t srcStart,int32_t srcLimit,UErrorCode * pErrorCode)123 _caseMap(const UCaseMap *csm, UCaseMapFull *map,
124 UChar *dest, int32_t destCapacity,
125 const UChar *src, UCaseContext *csc,
126 int32_t srcStart, int32_t srcLimit,
127 UErrorCode *pErrorCode) {
128 const UChar *s;
129 UChar32 c, c2;
130 int32_t srcIndex, destIndex;
131 int32_t locCache;
132
133 locCache=csm->locCache;
134
135 /* case mapping loop */
136 srcIndex=srcStart;
137 destIndex=0;
138 while(srcIndex<srcLimit) {
139 csc->cpStart=srcIndex;
140 U16_NEXT(src, srcIndex, srcLimit, c);
141 csc->cpLimit=srcIndex;
142 c=map(csm->csp, c, utf16_caseContextIterator, csc, &s, csm->locale, &locCache);
143 if((destIndex<destCapacity) && (c<0 ? (c2=~c)<=0xffff : UCASE_MAX_STRING_LENGTH<c && (c2=c)<=0xffff)) {
144 /* fast path version of appendResult() for BMP results */
145 dest[destIndex++]=(UChar)c2;
146 } else {
147 destIndex=appendResult(dest, destIndex, destCapacity, c, s);
148 }
149 }
150
151 if(destIndex>destCapacity) {
152 *pErrorCode=U_BUFFER_OVERFLOW_ERROR;
153 }
154 return destIndex;
155 }
156
157 static void
setTempCaseMapLocale(UCaseMap * csm,const char * locale,UErrorCode * pErrorCode)158 setTempCaseMapLocale(UCaseMap *csm, const char *locale, UErrorCode *pErrorCode) {
159 /*
160 * We could call ucasemap_setLocale(), but here we really only care about
161 * the initial language subtag, we need not return the real string via
162 * ucasemap_getLocale(), and we don't care about only getting "x" from
163 * "x-some-thing" etc.
164 *
165 * We ignore locales with a longer-than-3 initial subtag.
166 *
167 * We also do not fill in the locCache because it is rarely used,
168 * and not worth setting unless we reuse it for many case mapping operations.
169 * (That's why UCaseMap was created.)
170 */
171 int i;
172 char c;
173
174 /* the internal functions require locale!=NULL */
175 if(locale==NULL) {
176 locale=uloc_getDefault();
177 }
178 for(i=0; i<4 && (c=locale[i])!=0 && c!='-' && c!='_'; ++i) {
179 csm->locale[i]=c;
180 }
181 if(i<=3) {
182 csm->locale[i]=0; /* Up to 3 non-separator characters. */
183 } else {
184 csm->locale[0]=0; /* Longer-than-3 initial subtag: Ignore. */
185 }
186 }
187
188 /*
189 * Set parameters on an empty UCaseMap, for UCaseMap-less API functions.
190 * Do this fast because it is called with every function call.
191 */
192 static U_INLINE void
setTempCaseMap(UCaseMap * csm,const char * locale,UErrorCode * pErrorCode)193 setTempCaseMap(UCaseMap *csm, const char *locale, UErrorCode *pErrorCode) {
194 if(csm->csp==NULL) {
195 csm->csp=ucase_getSingleton(pErrorCode);
196 if(U_FAILURE(*pErrorCode)) {
197 return;
198 }
199 }
200 if(locale!=NULL && locale[0]==0) {
201 csm->locale[0]=0;
202 } else {
203 setTempCaseMapLocale(csm, locale, pErrorCode);
204 }
205 }
206
207 #if !UCONFIG_NO_BREAK_ITERATION
208
209 /*
210 * Internal titlecasing function.
211 */
212 static int32_t
_toTitle(UCaseMap * csm,UChar * dest,int32_t destCapacity,const UChar * src,UCaseContext * csc,int32_t srcLength,UErrorCode * pErrorCode)213 _toTitle(UCaseMap *csm,
214 UChar *dest, int32_t destCapacity,
215 const UChar *src, UCaseContext *csc,
216 int32_t srcLength,
217 UErrorCode *pErrorCode) {
218 const UChar *s;
219 UChar32 c;
220 int32_t prev, titleStart, titleLimit, index, destIndex, length;
221 UBool isFirstIndex;
222
223 if(csm->iter!=NULL) {
224 ubrk_setText(csm->iter, src, srcLength, pErrorCode);
225 } else {
226 csm->iter=ubrk_open(UBRK_WORD, csm->locale,
227 src, srcLength,
228 pErrorCode);
229 }
230 if(U_FAILURE(*pErrorCode)) {
231 return 0;
232 }
233
234 /* set up local variables */
235 destIndex=0;
236 prev=0;
237 isFirstIndex=TRUE;
238
239 /* titlecasing loop */
240 while(prev<srcLength) {
241 /* find next index where to titlecase */
242 if(isFirstIndex) {
243 isFirstIndex=FALSE;
244 index=ubrk_first(csm->iter);
245 } else {
246 index=ubrk_next(csm->iter);
247 }
248 if(index==UBRK_DONE || index>srcLength) {
249 index=srcLength;
250 }
251
252 /*
253 * Unicode 4 & 5 section 3.13 Default Case Operations:
254 *
255 * R3 toTitlecase(X): Find the word boundaries based on Unicode Standard Annex
256 * #29, "Text Boundaries." Between each pair of word boundaries, find the first
257 * cased character F. If F exists, map F to default_title(F); then map each
258 * subsequent character C to default_lower(C).
259 *
260 * In this implementation, segment [prev..index[ into 3 parts:
261 * a) uncased characters (copy as-is) [prev..titleStart[
262 * b) first case letter (titlecase) [titleStart..titleLimit[
263 * c) subsequent characters (lowercase) [titleLimit..index[
264 */
265 if(prev<index) {
266 /* find and copy uncased characters [prev..titleStart[ */
267 titleStart=titleLimit=prev;
268 U16_NEXT(src, titleLimit, index, c);
269 if((csm->options&U_TITLECASE_NO_BREAK_ADJUSTMENT)==0 && UCASE_NONE==ucase_getType(csm->csp, c)) {
270 /* Adjust the titlecasing index (titleStart) to the next cased character. */
271 for(;;) {
272 titleStart=titleLimit;
273 if(titleLimit==index) {
274 /*
275 * only uncased characters in [prev..index[
276 * stop with titleStart==titleLimit==index
277 */
278 break;
279 }
280 U16_NEXT(src, titleLimit, index, c);
281 if(UCASE_NONE!=ucase_getType(csm->csp, c)) {
282 break; /* cased letter at [titleStart..titleLimit[ */
283 }
284 }
285 length=titleStart-prev;
286 if(length>0) {
287 if((destIndex+length)<=destCapacity) {
288 uprv_memcpy(dest+destIndex, src+prev, length*U_SIZEOF_UCHAR);
289 }
290 destIndex+=length;
291 }
292 }
293
294 if(titleStart<titleLimit) {
295 /* titlecase c which is from [titleStart..titleLimit[ */
296 csc->cpStart=titleStart;
297 csc->cpLimit=titleLimit;
298 c=ucase_toFullTitle(csm->csp, c, utf16_caseContextIterator, csc, &s, csm->locale, &csm->locCache);
299 destIndex=appendResult(dest, destIndex, destCapacity, c, s);
300
301 /* lowercase [titleLimit..index[ */
302 if(titleLimit<index) {
303 if((csm->options&U_TITLECASE_NO_LOWERCASE)==0) {
304 /* Normal operation: Lowercase the rest of the word. */
305 destIndex+=
306 _caseMap(
307 csm, ucase_toFullLower,
308 dest+destIndex, destCapacity-destIndex,
309 src, csc,
310 titleLimit, index,
311 pErrorCode);
312 } else {
313 /* Optionally just copy the rest of the word unchanged. */
314 length=index-titleLimit;
315 if((destIndex+length)<=destCapacity) {
316 uprv_memcpy(dest+destIndex, src+titleLimit, length*U_SIZEOF_UCHAR);
317 }
318 destIndex+=length;
319 }
320 }
321 }
322 }
323
324 prev=index;
325 }
326
327 if(destIndex>destCapacity) {
328 *pErrorCode=U_BUFFER_OVERFLOW_ERROR;
329 }
330 return destIndex;
331 }
332
333 #endif
334
335 /* functions available in the common library (for unistr_case.cpp) */
336
337 U_CFUNC int32_t
ustr_toLower(const UCaseProps * csp,UChar * dest,int32_t destCapacity,const UChar * src,int32_t srcLength,const char * locale,UErrorCode * pErrorCode)338 ustr_toLower(const UCaseProps *csp,
339 UChar *dest, int32_t destCapacity,
340 const UChar *src, int32_t srcLength,
341 const char *locale,
342 UErrorCode *pErrorCode) {
343 UCaseMap csm={ NULL };
344 UCaseContext csc={ NULL };
345
346 csm.csp=csp;
347 setTempCaseMap(&csm, locale, pErrorCode);
348 csc.p=(void *)src;
349 csc.limit=srcLength;
350
351 return _caseMap(&csm, ucase_toFullLower,
352 dest, destCapacity,
353 src, &csc, 0, srcLength,
354 pErrorCode);
355 }
356
357 U_CFUNC int32_t
ustr_toUpper(const UCaseProps * csp,UChar * dest,int32_t destCapacity,const UChar * src,int32_t srcLength,const char * locale,UErrorCode * pErrorCode)358 ustr_toUpper(const UCaseProps *csp,
359 UChar *dest, int32_t destCapacity,
360 const UChar *src, int32_t srcLength,
361 const char *locale,
362 UErrorCode *pErrorCode) {
363 UCaseMap csm={ NULL };
364 UCaseContext csc={ NULL };
365
366 csm.csp=csp;
367 setTempCaseMap(&csm, locale, pErrorCode);
368 csc.p=(void *)src;
369 csc.limit=srcLength;
370
371 return _caseMap(&csm, ucase_toFullUpper,
372 dest, destCapacity,
373 src, &csc, 0, srcLength,
374 pErrorCode);
375 }
376
377 #if !UCONFIG_NO_BREAK_ITERATION
378
379 U_CFUNC int32_t
ustr_toTitle(const UCaseProps * csp,UChar * dest,int32_t destCapacity,const UChar * src,int32_t srcLength,UBreakIterator * titleIter,const char * locale,uint32_t options,UErrorCode * pErrorCode)380 ustr_toTitle(const UCaseProps *csp,
381 UChar *dest, int32_t destCapacity,
382 const UChar *src, int32_t srcLength,
383 UBreakIterator *titleIter,
384 const char *locale, uint32_t options,
385 UErrorCode *pErrorCode) {
386 UCaseMap csm={ NULL };
387 UCaseContext csc={ NULL };
388 int32_t length;
389
390 csm.csp=csp;
391 csm.iter=titleIter;
392 csm.options=options;
393 setTempCaseMap(&csm, locale, pErrorCode);
394 csc.p=(void *)src;
395 csc.limit=srcLength;
396
397 length=_toTitle(&csm,
398 dest, destCapacity,
399 src, &csc, srcLength,
400 pErrorCode);
401 if(titleIter==NULL && csm.iter!=NULL) {
402 ubrk_close(csm.iter);
403 }
404 return length;
405 }
406
407 #endif
408
409 U_CFUNC int32_t
ustr_foldCase(const UCaseProps * csp,UChar * dest,int32_t destCapacity,const UChar * src,int32_t srcLength,uint32_t options,UErrorCode * pErrorCode)410 ustr_foldCase(const UCaseProps *csp,
411 UChar *dest, int32_t destCapacity,
412 const UChar *src, int32_t srcLength,
413 uint32_t options,
414 UErrorCode *pErrorCode) {
415 int32_t srcIndex, destIndex;
416
417 const UChar *s;
418 UChar32 c, c2;
419
420 /* case mapping loop */
421 srcIndex=destIndex=0;
422 while(srcIndex<srcLength) {
423 U16_NEXT(src, srcIndex, srcLength, c);
424 c=ucase_toFullFolding(csp, c, &s, options);
425 if((destIndex<destCapacity) && (c<0 ? (c2=~c)<=0xffff : UCASE_MAX_STRING_LENGTH<c && (c2=c)<=0xffff)) {
426 /* fast path version of appendResult() for BMP results */
427 dest[destIndex++]=(UChar)c2;
428 } else {
429 destIndex=appendResult(dest, destIndex, destCapacity, c, s);
430 }
431 }
432
433 if(destIndex>destCapacity) {
434 *pErrorCode=U_BUFFER_OVERFLOW_ERROR;
435 }
436 return destIndex;
437 }
438
439 /*
440 * Implement argument checking and buffer handling
441 * for string case mapping as a common function.
442 */
443
444 /* common internal function for public API functions */
445
446 static int32_t
caseMap(const UCaseMap * csm,UChar * dest,int32_t destCapacity,const UChar * src,int32_t srcLength,int32_t toWhichCase,UErrorCode * pErrorCode)447 caseMap(const UCaseMap *csm,
448 UChar *dest, int32_t destCapacity,
449 const UChar *src, int32_t srcLength,
450 int32_t toWhichCase,
451 UErrorCode *pErrorCode) {
452 UChar buffer[300];
453 UChar *temp;
454
455 int32_t destLength;
456
457 /* check argument values */
458 if(pErrorCode==NULL || U_FAILURE(*pErrorCode)) {
459 return 0;
460 }
461 if( destCapacity<0 ||
462 (dest==NULL && destCapacity>0) ||
463 src==NULL ||
464 srcLength<-1
465 ) {
466 *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR;
467 return 0;
468 }
469
470 /* get the string length */
471 if(srcLength==-1) {
472 srcLength=u_strlen(src);
473 }
474
475 /* check for overlapping source and destination */
476 if( dest!=NULL &&
477 ((src>=dest && src<(dest+destCapacity)) ||
478 (dest>=src && dest<(src+srcLength)))
479 ) {
480 /* overlap: provide a temporary destination buffer and later copy the result */
481 if(destCapacity<=(sizeof(buffer)/U_SIZEOF_UCHAR)) {
482 /* the stack buffer is large enough */
483 temp=buffer;
484 } else {
485 /* allocate a buffer */
486 temp=(UChar *)uprv_malloc(destCapacity*U_SIZEOF_UCHAR);
487 if(temp==NULL) {
488 *pErrorCode=U_MEMORY_ALLOCATION_ERROR;
489 return 0;
490 }
491 }
492 } else {
493 temp=dest;
494 }
495
496 destLength=0;
497
498 if(toWhichCase==FOLD_CASE) {
499 destLength=ustr_foldCase(csm->csp, temp, destCapacity, src, srcLength,
500 csm->options, pErrorCode);
501 } else {
502 UCaseContext csc={ NULL };
503
504 csc.p=(void *)src;
505 csc.limit=srcLength;
506
507 if(toWhichCase==TO_LOWER) {
508 destLength=_caseMap(csm, ucase_toFullLower,
509 temp, destCapacity,
510 src, &csc,
511 0, srcLength,
512 pErrorCode);
513 } else if(toWhichCase==TO_UPPER) {
514 destLength=_caseMap(csm, ucase_toFullUpper,
515 temp, destCapacity,
516 src, &csc,
517 0, srcLength,
518 pErrorCode);
519 } else /* if(toWhichCase==TO_TITLE) */ {
520 #if UCONFIG_NO_BREAK_ITERATION
521 *pErrorCode=U_UNSUPPORTED_ERROR;
522 #else
523 /* UCaseMap is actually non-const in toTitle() APIs. */
524 destLength=_toTitle((UCaseMap *)csm, temp, destCapacity,
525 src, &csc, srcLength,
526 pErrorCode);
527 #endif
528 }
529 }
530 if(temp!=dest) {
531 /* copy the result string to the destination buffer */
532 if(destLength>0) {
533 int32_t copyLength= destLength<=destCapacity ? destLength : destCapacity;
534 if(copyLength>0) {
535 uprv_memmove(dest, temp, copyLength*U_SIZEOF_UCHAR);
536 }
537 }
538 if(temp!=buffer) {
539 uprv_free(temp);
540 }
541 }
542
543 return u_terminateUChars(dest, destCapacity, destLength, pErrorCode);
544 }
545
546 /* public API functions */
547
548 U_CAPI int32_t U_EXPORT2
u_strToLower(UChar * dest,int32_t destCapacity,const UChar * src,int32_t srcLength,const char * locale,UErrorCode * pErrorCode)549 u_strToLower(UChar *dest, int32_t destCapacity,
550 const UChar *src, int32_t srcLength,
551 const char *locale,
552 UErrorCode *pErrorCode) {
553 UCaseMap csm={ NULL };
554 setTempCaseMap(&csm, locale, pErrorCode);
555 return caseMap(&csm,
556 dest, destCapacity,
557 src, srcLength,
558 TO_LOWER, pErrorCode);
559 }
560
561 U_CAPI int32_t U_EXPORT2
u_strToUpper(UChar * dest,int32_t destCapacity,const UChar * src,int32_t srcLength,const char * locale,UErrorCode * pErrorCode)562 u_strToUpper(UChar *dest, int32_t destCapacity,
563 const UChar *src, int32_t srcLength,
564 const char *locale,
565 UErrorCode *pErrorCode) {
566 UCaseMap csm={ NULL };
567 setTempCaseMap(&csm, locale, pErrorCode);
568 return caseMap(&csm,
569 dest, destCapacity,
570 src, srcLength,
571 TO_UPPER, pErrorCode);
572 }
573
574 #if !UCONFIG_NO_BREAK_ITERATION
575
576 U_CAPI int32_t U_EXPORT2
u_strToTitle(UChar * dest,int32_t destCapacity,const UChar * src,int32_t srcLength,UBreakIterator * titleIter,const char * locale,UErrorCode * pErrorCode)577 u_strToTitle(UChar *dest, int32_t destCapacity,
578 const UChar *src, int32_t srcLength,
579 UBreakIterator *titleIter,
580 const char *locale,
581 UErrorCode *pErrorCode) {
582 UCaseMap csm={ NULL };
583 int32_t length;
584
585 csm.iter=titleIter;
586 setTempCaseMap(&csm, locale, pErrorCode);
587 length=caseMap(&csm,
588 dest, destCapacity,
589 src, srcLength,
590 TO_TITLE, pErrorCode);
591 if(titleIter==NULL && csm.iter!=NULL) {
592 ubrk_close(csm.iter);
593 }
594 return length;
595 }
596
597 U_CAPI int32_t U_EXPORT2
ucasemap_toTitle(UCaseMap * csm,UChar * dest,int32_t destCapacity,const UChar * src,int32_t srcLength,UErrorCode * pErrorCode)598 ucasemap_toTitle(UCaseMap *csm,
599 UChar *dest, int32_t destCapacity,
600 const UChar *src, int32_t srcLength,
601 UErrorCode *pErrorCode) {
602 return caseMap(csm,
603 dest, destCapacity,
604 src, srcLength,
605 TO_TITLE, pErrorCode);
606 }
607
608 #endif
609
610 U_CAPI int32_t U_EXPORT2
u_strFoldCase(UChar * dest,int32_t destCapacity,const UChar * src,int32_t srcLength,uint32_t options,UErrorCode * pErrorCode)611 u_strFoldCase(UChar *dest, int32_t destCapacity,
612 const UChar *src, int32_t srcLength,
613 uint32_t options,
614 UErrorCode *pErrorCode) {
615 UCaseMap csm={ NULL };
616 csm.csp=ucase_getSingleton(pErrorCode);
617 csm.options=options;
618 return caseMap(&csm,
619 dest, destCapacity,
620 src, srcLength,
621 FOLD_CASE, pErrorCode);
622 }
623
624 /* case-insensitive string comparisons -------------------------------------- */
625
626 /*
627 * This function is a copy of unorm_cmpEquivFold() minus the parts for
628 * canonical equivalence.
629 * Keep the functions in sync, and see there for how this works.
630 * The duplication is for modularization:
631 * It makes caseless (but not canonical caseless) matches independent of
632 * the normalization code.
633 */
634
635 /* stack element for previous-level source/decomposition pointers */
636 struct CmpEquivLevel {
637 const UChar *start, *s, *limit;
638 };
639 typedef struct CmpEquivLevel CmpEquivLevel;
640
641 /* internal function */
642 U_CFUNC int32_t
u_strcmpFold(const UChar * s1,int32_t length1,const UChar * s2,int32_t length2,uint32_t options,UErrorCode * pErrorCode)643 u_strcmpFold(const UChar *s1, int32_t length1,
644 const UChar *s2, int32_t length2,
645 uint32_t options,
646 UErrorCode *pErrorCode) {
647 const UCaseProps *csp;
648
649 /* current-level start/limit - s1/s2 as current */
650 const UChar *start1, *start2, *limit1, *limit2;
651
652 /* case folding variables */
653 const UChar *p;
654 int32_t length;
655
656 /* stacks of previous-level start/current/limit */
657 CmpEquivLevel stack1[2], stack2[2];
658
659 /* case folding buffers, only use current-level start/limit */
660 UChar fold1[UCASE_MAX_STRING_LENGTH+1], fold2[UCASE_MAX_STRING_LENGTH+1];
661
662 /* track which is the current level per string */
663 int32_t level1, level2;
664
665 /* current code units, and code points for lookups */
666 UChar32 c1, c2, cp1, cp2;
667
668 /* no argument error checking because this itself is not an API */
669
670 /*
671 * assume that at least the option U_COMPARE_IGNORE_CASE is set
672 * otherwise this function would have to behave exactly as uprv_strCompare()
673 */
674 csp=ucase_getSingleton(pErrorCode);
675 if(U_FAILURE(*pErrorCode)) {
676 return 0;
677 }
678
679 /* initialize */
680 start1=s1;
681 if(length1==-1) {
682 limit1=NULL;
683 } else {
684 limit1=s1+length1;
685 }
686
687 start2=s2;
688 if(length2==-1) {
689 limit2=NULL;
690 } else {
691 limit2=s2+length2;
692 }
693
694 level1=level2=0;
695 c1=c2=-1;
696
697 /* comparison loop */
698 for(;;) {
699 /*
700 * here a code unit value of -1 means "get another code unit"
701 * below it will mean "this source is finished"
702 */
703
704 if(c1<0) {
705 /* get next code unit from string 1, post-increment */
706 for(;;) {
707 if(s1==limit1 || ((c1=*s1)==0 && (limit1==NULL || (options&_STRNCMP_STYLE)))) {
708 if(level1==0) {
709 c1=-1;
710 break;
711 }
712 } else {
713 ++s1;
714 break;
715 }
716
717 /* reached end of level buffer, pop one level */
718 do {
719 --level1;
720 start1=stack1[level1].start;
721 } while(start1==NULL);
722 s1=stack1[level1].s;
723 limit1=stack1[level1].limit;
724 }
725 }
726
727 if(c2<0) {
728 /* get next code unit from string 2, post-increment */
729 for(;;) {
730 if(s2==limit2 || ((c2=*s2)==0 && (limit2==NULL || (options&_STRNCMP_STYLE)))) {
731 if(level2==0) {
732 c2=-1;
733 break;
734 }
735 } else {
736 ++s2;
737 break;
738 }
739
740 /* reached end of level buffer, pop one level */
741 do {
742 --level2;
743 start2=stack2[level2].start;
744 } while(start2==NULL);
745 s2=stack2[level2].s;
746 limit2=stack2[level2].limit;
747 }
748 }
749
750 /*
751 * compare c1 and c2
752 * either variable c1, c2 is -1 only if the corresponding string is finished
753 */
754 if(c1==c2) {
755 if(c1<0) {
756 return 0; /* c1==c2==-1 indicating end of strings */
757 }
758 c1=c2=-1; /* make us fetch new code units */
759 continue;
760 } else if(c1<0) {
761 return -1; /* string 1 ends before string 2 */
762 } else if(c2<0) {
763 return 1; /* string 2 ends before string 1 */
764 }
765 /* c1!=c2 && c1>=0 && c2>=0 */
766
767 /* get complete code points for c1, c2 for lookups if either is a surrogate */
768 cp1=c1;
769 if(U_IS_SURROGATE(c1)) {
770 UChar c;
771
772 if(U_IS_SURROGATE_LEAD(c1)) {
773 if(s1!=limit1 && U16_IS_TRAIL(c=*s1)) {
774 /* advance ++s1; only below if cp1 decomposes/case-folds */
775 cp1=U16_GET_SUPPLEMENTARY(c1, c);
776 }
777 } else /* isTrail(c1) */ {
778 if(start1<=(s1-2) && U16_IS_LEAD(c=*(s1-2))) {
779 cp1=U16_GET_SUPPLEMENTARY(c, c1);
780 }
781 }
782 }
783
784 cp2=c2;
785 if(U_IS_SURROGATE(c2)) {
786 UChar c;
787
788 if(U_IS_SURROGATE_LEAD(c2)) {
789 if(s2!=limit2 && U16_IS_TRAIL(c=*s2)) {
790 /* advance ++s2; only below if cp2 decomposes/case-folds */
791 cp2=U16_GET_SUPPLEMENTARY(c2, c);
792 }
793 } else /* isTrail(c2) */ {
794 if(start2<=(s2-2) && U16_IS_LEAD(c=*(s2-2))) {
795 cp2=U16_GET_SUPPLEMENTARY(c, c2);
796 }
797 }
798 }
799
800 /*
801 * go down one level for each string
802 * continue with the main loop as soon as there is a real change
803 */
804
805 if( level1==0 &&
806 (length=ucase_toFullFolding(csp, (UChar32)cp1, &p, options))>=0
807 ) {
808 /* cp1 case-folds to the code point "length" or to p[length] */
809 if(U_IS_SURROGATE(c1)) {
810 if(U_IS_SURROGATE_LEAD(c1)) {
811 /* advance beyond source surrogate pair if it case-folds */
812 ++s1;
813 } else /* isTrail(c1) */ {
814 /*
815 * we got a supplementary code point when hitting its trail surrogate,
816 * therefore the lead surrogate must have been the same as in the other string;
817 * compare this decomposition with the lead surrogate in the other string
818 * remember that this simulates bulk text replacement:
819 * the decomposition would replace the entire code point
820 */
821 --s2;
822 c2=*(s2-1);
823 }
824 }
825
826 /* push current level pointers */
827 stack1[0].start=start1;
828 stack1[0].s=s1;
829 stack1[0].limit=limit1;
830 ++level1;
831
832 /* copy the folding result to fold1[] */
833 if(length<=UCASE_MAX_STRING_LENGTH) {
834 u_memcpy(fold1, p, length);
835 } else {
836 int32_t i=0;
837 U16_APPEND_UNSAFE(fold1, i, length);
838 length=i;
839 }
840
841 /* set next level pointers to case folding */
842 start1=s1=fold1;
843 limit1=fold1+length;
844
845 /* get ready to read from decomposition, continue with loop */
846 c1=-1;
847 continue;
848 }
849
850 if( level2==0 &&
851 (length=ucase_toFullFolding(csp, (UChar32)cp2, &p, options))>=0
852 ) {
853 /* cp2 case-folds to the code point "length" or to p[length] */
854 if(U_IS_SURROGATE(c2)) {
855 if(U_IS_SURROGATE_LEAD(c2)) {
856 /* advance beyond source surrogate pair if it case-folds */
857 ++s2;
858 } else /* isTrail(c2) */ {
859 /*
860 * we got a supplementary code point when hitting its trail surrogate,
861 * therefore the lead surrogate must have been the same as in the other string;
862 * compare this decomposition with the lead surrogate in the other string
863 * remember that this simulates bulk text replacement:
864 * the decomposition would replace the entire code point
865 */
866 --s1;
867 c1=*(s1-1);
868 }
869 }
870
871 /* push current level pointers */
872 stack2[0].start=start2;
873 stack2[0].s=s2;
874 stack2[0].limit=limit2;
875 ++level2;
876
877 /* copy the folding result to fold2[] */
878 if(length<=UCASE_MAX_STRING_LENGTH) {
879 u_memcpy(fold2, p, length);
880 } else {
881 int32_t i=0;
882 U16_APPEND_UNSAFE(fold2, i, length);
883 length=i;
884 }
885
886 /* set next level pointers to case folding */
887 start2=s2=fold2;
888 limit2=fold2+length;
889
890 /* get ready to read from decomposition, continue with loop */
891 c2=-1;
892 continue;
893 }
894
895 /*
896 * no decomposition/case folding, max level for both sides:
897 * return difference result
898 *
899 * code point order comparison must not just return cp1-cp2
900 * because when single surrogates are present then the surrogate pairs
901 * that formed cp1 and cp2 may be from different string indexes
902 *
903 * example: { d800 d800 dc01 } vs. { d800 dc00 }, compare at second code units
904 * c1=d800 cp1=10001 c2=dc00 cp2=10000
905 * cp1-cp2>0 but c1-c2<0 and in fact in UTF-32 it is { d800 10001 } < { 10000 }
906 *
907 * therefore, use same fix-up as in ustring.c/uprv_strCompare()
908 * except: uprv_strCompare() fetches c=*s while this functions fetches c=*s++
909 * so we have slightly different pointer/start/limit comparisons here
910 */
911
912 if(c1>=0xd800 && c2>=0xd800 && (options&U_COMPARE_CODE_POINT_ORDER)) {
913 /* subtract 0x2800 from BMP code points to make them smaller than supplementary ones */
914 if(
915 (c1<=0xdbff && s1!=limit1 && U16_IS_TRAIL(*s1)) ||
916 (U16_IS_TRAIL(c1) && start1!=(s1-1) && U16_IS_LEAD(*(s1-2)))
917 ) {
918 /* part of a surrogate pair, leave >=d800 */
919 } else {
920 /* BMP code point - may be surrogate code point - make <d800 */
921 c1-=0x2800;
922 }
923
924 if(
925 (c2<=0xdbff && s2!=limit2 && U16_IS_TRAIL(*s2)) ||
926 (U16_IS_TRAIL(c2) && start2!=(s2-1) && U16_IS_LEAD(*(s2-2)))
927 ) {
928 /* part of a surrogate pair, leave >=d800 */
929 } else {
930 /* BMP code point - may be surrogate code point - make <d800 */
931 c2-=0x2800;
932 }
933 }
934
935 return c1-c2;
936 }
937 }
938
939 /* public API functions */
940
941 U_CAPI int32_t U_EXPORT2
u_strCaseCompare(const UChar * s1,int32_t length1,const UChar * s2,int32_t length2,uint32_t options,UErrorCode * pErrorCode)942 u_strCaseCompare(const UChar *s1, int32_t length1,
943 const UChar *s2, int32_t length2,
944 uint32_t options,
945 UErrorCode *pErrorCode) {
946 /* argument checking */
947 if(pErrorCode==0 || U_FAILURE(*pErrorCode)) {
948 return 0;
949 }
950 if(s1==NULL || length1<-1 || s2==NULL || length2<-1) {
951 *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR;
952 return 0;
953 }
954 return u_strcmpFold(s1, length1, s2, length2,
955 options|U_COMPARE_IGNORE_CASE,
956 pErrorCode);
957 }
958
959 U_CAPI int32_t U_EXPORT2
u_strcasecmp(const UChar * s1,const UChar * s2,uint32_t options)960 u_strcasecmp(const UChar *s1, const UChar *s2, uint32_t options) {
961 UErrorCode errorCode=U_ZERO_ERROR;
962 return u_strcmpFold(s1, -1, s2, -1,
963 options|U_COMPARE_IGNORE_CASE,
964 &errorCode);
965 }
966
967 U_CAPI int32_t U_EXPORT2
u_memcasecmp(const UChar * s1,const UChar * s2,int32_t length,uint32_t options)968 u_memcasecmp(const UChar *s1, const UChar *s2, int32_t length, uint32_t options) {
969 UErrorCode errorCode=U_ZERO_ERROR;
970 return u_strcmpFold(s1, length, s2, length,
971 options|U_COMPARE_IGNORE_CASE,
972 &errorCode);
973 }
974
975 U_CAPI int32_t U_EXPORT2
u_strncasecmp(const UChar * s1,const UChar * s2,int32_t n,uint32_t options)976 u_strncasecmp(const UChar *s1, const UChar *s2, int32_t n, uint32_t options) {
977 UErrorCode errorCode=U_ZERO_ERROR;
978 return u_strcmpFold(s1, n, s2, n,
979 options|(U_COMPARE_IGNORE_CASE|_STRNCMP_STYLE),
980 &errorCode);
981 }
982