1 /*
2 *******************************************************************************
3 *
4 *   Copyright (C) 2001-2015, International Business Machines
5 *   Corporation and others.  All Rights Reserved.
6 *
7 *******************************************************************************
8 *   file name:  ustrcase.cpp
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/brkiter.h"
23 #include "unicode/ustring.h"
24 #include "unicode/ucasemap.h"
25 #include "unicode/ubrk.h"
26 #include "unicode/utf.h"
27 #include "unicode/utf16.h"
28 #include "cmemory.h"
29 #include "ucase.h"
30 #include "ustr_imp.h"
31 #include "uassert.h"
32 
33 U_NAMESPACE_USE
34 
35 /* string casing ------------------------------------------------------------ */
36 
37 /* Appends a full case mapping result, see UCASE_MAX_STRING_LENGTH. */
38 static inline int32_t
appendResult(UChar * dest,int32_t destIndex,int32_t destCapacity,int32_t result,const UChar * s)39 appendResult(UChar *dest, int32_t destIndex, int32_t destCapacity,
40              int32_t result, const UChar *s) {
41     UChar32 c;
42     int32_t length;
43 
44     /* decode the result */
45     if(result<0) {
46         /* (not) original code point */
47         c=~result;
48         length=-1;
49     } else if(result<=UCASE_MAX_STRING_LENGTH) {
50         c=U_SENTINEL;
51         length=result;
52     } else {
53         c=result;
54         length=-1;
55     }
56 
57     if(destIndex<destCapacity) {
58         /* append the result */
59         if(length<0) {
60             /* code point */
61             UBool isError=FALSE;
62             U16_APPEND(dest, destIndex, destCapacity, c, isError);
63             if(isError) {
64                 /* overflow, nothing written */
65                 destIndex+=U16_LENGTH(c);
66             }
67         } else {
68             /* string */
69             if((destIndex+length)<=destCapacity) {
70                 while(length>0) {
71                     dest[destIndex++]=*s++;
72                     --length;
73                 }
74             } else {
75                 /* overflow */
76                 destIndex+=length;
77             }
78         }
79     } else {
80         /* preflight */
81         if(length<0) {
82             destIndex+=U16_LENGTH(c);
83         } else {
84             destIndex+=length;
85         }
86     }
87     return destIndex;
88 }
89 
90 static UChar32 U_CALLCONV
utf16_caseContextIterator(void * context,int8_t dir)91 utf16_caseContextIterator(void *context, int8_t dir) {
92     UCaseContext *csc=(UCaseContext *)context;
93     UChar32 c;
94 
95     if(dir<0) {
96         /* reset for backward iteration */
97         csc->index=csc->cpStart;
98         csc->dir=dir;
99     } else if(dir>0) {
100         /* reset for forward iteration */
101         csc->index=csc->cpLimit;
102         csc->dir=dir;
103     } else {
104         /* continue current iteration direction */
105         dir=csc->dir;
106     }
107 
108     if(dir<0) {
109         if(csc->start<csc->index) {
110             U16_PREV((const UChar *)csc->p, csc->start, csc->index, c);
111             return c;
112         }
113     } else {
114         if(csc->index<csc->limit) {
115             U16_NEXT((const UChar *)csc->p, csc->index, csc->limit, c);
116             return c;
117         }
118     }
119     return U_SENTINEL;
120 }
121 
122 /*
123  * Case-maps [srcStart..srcLimit[ but takes
124  * context [0..srcLength[ into account.
125  */
126 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)127 _caseMap(const UCaseMap *csm, UCaseMapFull *map,
128          UChar *dest, int32_t destCapacity,
129          const UChar *src, UCaseContext *csc,
130          int32_t srcStart, int32_t srcLimit,
131          UErrorCode *pErrorCode) {
132     const UChar *s;
133     UChar32 c, c2 = 0;
134     int32_t srcIndex, destIndex;
135     int32_t locCache;
136 
137     locCache=csm->locCache;
138 
139     /* case mapping loop */
140     srcIndex=srcStart;
141     destIndex=0;
142     while(srcIndex<srcLimit) {
143         csc->cpStart=srcIndex;
144         U16_NEXT(src, srcIndex, srcLimit, c);
145         csc->cpLimit=srcIndex;
146         c=map(csm->csp, c, utf16_caseContextIterator, csc, &s, csm->locale, &locCache);
147         if((destIndex<destCapacity) && (c<0 ? (c2=~c)<=0xffff : UCASE_MAX_STRING_LENGTH<c && (c2=c)<=0xffff)) {
148             /* fast path version of appendResult() for BMP results */
149             dest[destIndex++]=(UChar)c2;
150         } else {
151             destIndex=appendResult(dest, destIndex, destCapacity, c, s);
152         }
153     }
154 
155     if(destIndex>destCapacity) {
156         *pErrorCode=U_BUFFER_OVERFLOW_ERROR;
157     }
158     return destIndex;
159 }
160 
161 #if !UCONFIG_NO_BREAK_ITERATION
162 
163 U_CFUNC int32_t U_CALLCONV
ustrcase_internalToTitle(const UCaseMap * csm,UChar * dest,int32_t destCapacity,const UChar * src,int32_t srcLength,UErrorCode * pErrorCode)164 ustrcase_internalToTitle(const UCaseMap *csm,
165                          UChar *dest, int32_t destCapacity,
166                          const UChar *src, int32_t srcLength,
167                          UErrorCode *pErrorCode) {
168     const UChar *s;
169     UChar32 c;
170     int32_t prev, titleStart, titleLimit, idx, destIndex, length;
171     UBool isFirstIndex;
172 
173     if(U_FAILURE(*pErrorCode)) {
174         return 0;
175     }
176 
177     // Use the C++ abstract base class to minimize dependencies.
178     // TODO: Change UCaseMap.iter to store a BreakIterator directly.
179     BreakIterator *bi=reinterpret_cast<BreakIterator *>(csm->iter);
180 
181     /* set up local variables */
182     int32_t locCache=csm->locCache;
183     UCaseContext csc=UCASECONTEXT_INITIALIZER;
184     csc.p=(void *)src;
185     csc.limit=srcLength;
186     destIndex=0;
187     prev=0;
188     isFirstIndex=TRUE;
189 
190     /* titlecasing loop */
191     while(prev<srcLength) {
192         /* find next index where to titlecase */
193         if(isFirstIndex) {
194             isFirstIndex=FALSE;
195             idx=bi->first();
196         } else {
197             idx=bi->next();
198         }
199         if(idx==UBRK_DONE || idx>srcLength) {
200             idx=srcLength;
201         }
202 
203         /*
204          * Unicode 4 & 5 section 3.13 Default Case Operations:
205          *
206          * R3  toTitlecase(X): Find the word boundaries based on Unicode Standard Annex
207          * #29, "Text Boundaries." Between each pair of word boundaries, find the first
208          * cased character F. If F exists, map F to default_title(F); then map each
209          * subsequent character C to default_lower(C).
210          *
211          * In this implementation, segment [prev..index[ into 3 parts:
212          * a) uncased characters (copy as-is) [prev..titleStart[
213          * b) first case letter (titlecase)         [titleStart..titleLimit[
214          * c) subsequent characters (lowercase)                 [titleLimit..index[
215          */
216         if(prev<idx) {
217             /* find and copy uncased characters [prev..titleStart[ */
218             titleStart=titleLimit=prev;
219             U16_NEXT(src, titleLimit, idx, c);
220             if((csm->options&U_TITLECASE_NO_BREAK_ADJUSTMENT)==0 && UCASE_NONE==ucase_getType(csm->csp, c)) {
221                 /* Adjust the titlecasing index (titleStart) to the next cased character. */
222                 for(;;) {
223                     titleStart=titleLimit;
224                     if(titleLimit==idx) {
225                         /*
226                          * only uncased characters in [prev..index[
227                          * stop with titleStart==titleLimit==index
228                          */
229                         break;
230                     }
231                     U16_NEXT(src, titleLimit, idx, c);
232                     if(UCASE_NONE!=ucase_getType(csm->csp, c)) {
233                         break; /* cased letter at [titleStart..titleLimit[ */
234                     }
235                 }
236                 length=titleStart-prev;
237                 if(length>0) {
238                     if((destIndex+length)<=destCapacity) {
239                         uprv_memcpy(dest+destIndex, src+prev, length*U_SIZEOF_UCHAR);
240                     }
241                     destIndex+=length;
242                 }
243             }
244 
245             if(titleStart<titleLimit) {
246                 /* titlecase c which is from [titleStart..titleLimit[ */
247                 csc.cpStart=titleStart;
248                 csc.cpLimit=titleLimit;
249                 c=ucase_toFullTitle(csm->csp, c, utf16_caseContextIterator, &csc, &s, csm->locale, &locCache);
250                 destIndex=appendResult(dest, destIndex, destCapacity, c, s);
251 
252                 /* Special case Dutch IJ titlecasing */
253                 if ( titleStart+1 < idx &&
254                      ucase_getCaseLocale(csm->locale,&locCache) == UCASE_LOC_DUTCH &&
255                      ( src[titleStart] == (UChar32) 0x0049 || src[titleStart] == (UChar32) 0x0069 ) &&
256                      ( src[titleStart+1] == (UChar32) 0x004A || src[titleStart+1] == (UChar32) 0x006A )) {
257                             c=(UChar32) 0x004A;
258                             destIndex=appendResult(dest, destIndex, destCapacity, c, s);
259                             titleLimit++;
260                 }
261 
262                 /* lowercase [titleLimit..index[ */
263                 if(titleLimit<idx) {
264                     if((csm->options&U_TITLECASE_NO_LOWERCASE)==0) {
265                         /* Normal operation: Lowercase the rest of the word. */
266                         destIndex+=
267                             _caseMap(
268                                 csm, ucase_toFullLower,
269                                 dest+destIndex, destCapacity-destIndex,
270                                 src, &csc,
271                                 titleLimit, idx,
272                                 pErrorCode);
273                     } else {
274                         /* Optionally just copy the rest of the word unchanged. */
275                         length=idx-titleLimit;
276                         if((destIndex+length)<=destCapacity) {
277                             uprv_memcpy(dest+destIndex, src+titleLimit, length*U_SIZEOF_UCHAR);
278                         }
279                         destIndex+=length;
280                     }
281                 }
282             }
283         }
284 
285         prev=idx;
286     }
287 
288     if(destIndex>destCapacity) {
289         *pErrorCode=U_BUFFER_OVERFLOW_ERROR;
290     }
291     return destIndex;
292 }
293 
294 #endif  // !UCONFIG_NO_BREAK_ITERATION
295 
296 /* functions available in the common library (for unistr_case.cpp) */
297 
298 U_CFUNC int32_t U_CALLCONV
ustrcase_internalToLower(const UCaseMap * csm,UChar * dest,int32_t destCapacity,const UChar * src,int32_t srcLength,UErrorCode * pErrorCode)299 ustrcase_internalToLower(const UCaseMap *csm,
300                          UChar *dest, int32_t destCapacity,
301                          const UChar *src, int32_t srcLength,
302                          UErrorCode *pErrorCode) {
303     UCaseContext csc=UCASECONTEXT_INITIALIZER;
304     csc.p=(void *)src;
305     csc.limit=srcLength;
306     return _caseMap(
307         csm, ucase_toFullLower,
308         dest, destCapacity,
309         src, &csc, 0, srcLength,
310         pErrorCode);
311 }
312 
313 U_CFUNC int32_t U_CALLCONV
ustrcase_internalToUpper(const UCaseMap * csm,UChar * dest,int32_t destCapacity,const UChar * src,int32_t srcLength,UErrorCode * pErrorCode)314 ustrcase_internalToUpper(const UCaseMap *csm,
315                          UChar *dest, int32_t destCapacity,
316                          const UChar *src, int32_t srcLength,
317                          UErrorCode *pErrorCode) {
318     UCaseContext csc=UCASECONTEXT_INITIALIZER;
319     csc.p=(void *)src;
320     csc.limit=srcLength;
321     return _caseMap(
322         csm, ucase_toFullUpper,
323         dest, destCapacity,
324         src, &csc, 0, srcLength,
325         pErrorCode);
326 }
327 
328 static int32_t
ustr_foldCase(const UCaseProps * csp,UChar * dest,int32_t destCapacity,const UChar * src,int32_t srcLength,uint32_t options,UErrorCode * pErrorCode)329 ustr_foldCase(const UCaseProps *csp,
330               UChar *dest, int32_t destCapacity,
331               const UChar *src, int32_t srcLength,
332               uint32_t options,
333               UErrorCode *pErrorCode) {
334     int32_t srcIndex, destIndex;
335 
336     const UChar *s;
337     UChar32 c, c2 = 0;
338 
339     /* case mapping loop */
340     srcIndex=destIndex=0;
341     while(srcIndex<srcLength) {
342         U16_NEXT(src, srcIndex, srcLength, c);
343         c=ucase_toFullFolding(csp, c, &s, options);
344         if((destIndex<destCapacity) && (c<0 ? (c2=~c)<=0xffff : UCASE_MAX_STRING_LENGTH<c && (c2=c)<=0xffff)) {
345             /* fast path version of appendResult() for BMP results */
346             dest[destIndex++]=(UChar)c2;
347         } else {
348             destIndex=appendResult(dest, destIndex, destCapacity, c, s);
349         }
350     }
351 
352     if(destIndex>destCapacity) {
353         *pErrorCode=U_BUFFER_OVERFLOW_ERROR;
354     }
355     return destIndex;
356 }
357 
358 U_CFUNC int32_t U_CALLCONV
ustrcase_internalFold(const UCaseMap * csm,UChar * dest,int32_t destCapacity,const UChar * src,int32_t srcLength,UErrorCode * pErrorCode)359 ustrcase_internalFold(const UCaseMap *csm,
360                       UChar *dest, int32_t destCapacity,
361                       const UChar *src, int32_t srcLength,
362                       UErrorCode *pErrorCode) {
363     return ustr_foldCase(csm->csp, dest, destCapacity, src, srcLength, csm->options, pErrorCode);
364 }
365 
366 U_CFUNC int32_t
ustrcase_map(const UCaseMap * csm,UChar * dest,int32_t destCapacity,const UChar * src,int32_t srcLength,UStringCaseMapper * stringCaseMapper,UErrorCode * pErrorCode)367 ustrcase_map(const UCaseMap *csm,
368              UChar *dest, int32_t destCapacity,
369              const UChar *src, int32_t srcLength,
370              UStringCaseMapper *stringCaseMapper,
371              UErrorCode *pErrorCode) {
372     UChar buffer[300];
373     UChar *temp;
374 
375     int32_t destLength;
376 
377     /* check argument values */
378     if(U_FAILURE(*pErrorCode)) {
379         return 0;
380     }
381     if( destCapacity<0 ||
382         (dest==NULL && destCapacity>0) ||
383         src==NULL ||
384         srcLength<-1
385     ) {
386         *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR;
387         return 0;
388     }
389 
390     /* get the string length */
391     if(srcLength==-1) {
392         srcLength=u_strlen(src);
393     }
394 
395     /* check for overlapping source and destination */
396     if( dest!=NULL &&
397         ((src>=dest && src<(dest+destCapacity)) ||
398          (dest>=src && dest<(src+srcLength)))
399     ) {
400         /* overlap: provide a temporary destination buffer and later copy the result */
401         if(destCapacity<=UPRV_LENGTHOF(buffer)) {
402             /* the stack buffer is large enough */
403             temp=buffer;
404         } else {
405             /* allocate a buffer */
406             temp=(UChar *)uprv_malloc(destCapacity*U_SIZEOF_UCHAR);
407             if(temp==NULL) {
408                 *pErrorCode=U_MEMORY_ALLOCATION_ERROR;
409                 return 0;
410             }
411         }
412     } else {
413         temp=dest;
414     }
415 
416     destLength=stringCaseMapper(csm, temp, destCapacity, src, srcLength, pErrorCode);
417     if(temp!=dest) {
418         /* copy the result string to the destination buffer */
419         if(destLength>0) {
420             int32_t copyLength= destLength<=destCapacity ? destLength : destCapacity;
421             if(copyLength>0) {
422                 uprv_memmove(dest, temp, copyLength*U_SIZEOF_UCHAR);
423             }
424         }
425         if(temp!=buffer) {
426             uprv_free(temp);
427         }
428     }
429 
430     return u_terminateUChars(dest, destCapacity, destLength, pErrorCode);
431 }
432 
433 /* public API functions */
434 
435 U_CAPI int32_t U_EXPORT2
u_strFoldCase(UChar * dest,int32_t destCapacity,const UChar * src,int32_t srcLength,uint32_t options,UErrorCode * pErrorCode)436 u_strFoldCase(UChar *dest, int32_t destCapacity,
437               const UChar *src, int32_t srcLength,
438               uint32_t options,
439               UErrorCode *pErrorCode) {
440     UCaseMap csm=UCASEMAP_INITIALIZER;
441     csm.csp=ucase_getSingleton();
442     csm.options=options;
443     return ustrcase_map(
444         &csm,
445         dest, destCapacity,
446         src, srcLength,
447         ustrcase_internalFold, pErrorCode);
448 }
449 
450 /* case-insensitive string comparisons -------------------------------------- */
451 
452 /*
453  * This function is a copy of unorm_cmpEquivFold() minus the parts for
454  * canonical equivalence.
455  * Keep the functions in sync, and see there for how this works.
456  * The duplication is for modularization:
457  * It makes caseless (but not canonical caseless) matches independent of
458  * the normalization code.
459  */
460 
461 /* stack element for previous-level source/decomposition pointers */
462 struct CmpEquivLevel {
463     const UChar *start, *s, *limit;
464 };
465 typedef struct CmpEquivLevel CmpEquivLevel;
466 
467 /**
468  * Internal implementation code comparing string with case fold.
469  * This function is called from u_strcmpFold() and u_caseInsensitivePrefixMatch().
470  *
471  * @param s1            input string 1
472  * @param length1       length of string 1, or -1 (NULL terminated)
473  * @param s2            input string 2
474  * @param length2       length of string 2, or -1 (NULL terminated)
475  * @param options       compare options
476  * @param matchLen1     (output) length of partial prefix match in s1
477  * @param matchLen2     (output) length of partial prefix match in s2
478  * @param pErrorCode    receives error status
479  * @return The result of comparison
480  */
_cmpFold(const UChar * s1,int32_t length1,const UChar * s2,int32_t length2,uint32_t options,int32_t * matchLen1,int32_t * matchLen2,UErrorCode * pErrorCode)481 static int32_t _cmpFold(
482             const UChar *s1, int32_t length1,
483             const UChar *s2, int32_t length2,
484             uint32_t options,
485             int32_t *matchLen1, int32_t *matchLen2,
486             UErrorCode *pErrorCode) {
487     int32_t cmpRes = 0;
488 
489     const UCaseProps *csp;
490 
491     /* current-level start/limit - s1/s2 as current */
492     const UChar *start1, *start2, *limit1, *limit2;
493 
494     /* points to the original start address */
495     const UChar *org1, *org2;
496 
497     /* points to the end of match + 1 */
498     const UChar *m1, *m2;
499 
500     /* case folding variables */
501     const UChar *p;
502     int32_t length;
503 
504     /* stacks of previous-level start/current/limit */
505     CmpEquivLevel stack1[2], stack2[2];
506 
507     /* case folding buffers, only use current-level start/limit */
508     UChar fold1[UCASE_MAX_STRING_LENGTH+1], fold2[UCASE_MAX_STRING_LENGTH+1];
509 
510     /* track which is the current level per string */
511     int32_t level1, level2;
512 
513     /* current code units, and code points for lookups */
514     UChar32 c1, c2, cp1, cp2;
515 
516     /* no argument error checking because this itself is not an API */
517 
518     /*
519      * assume that at least the option U_COMPARE_IGNORE_CASE is set
520      * otherwise this function would have to behave exactly as uprv_strCompare()
521      */
522     csp=ucase_getSingleton();
523     if(U_FAILURE(*pErrorCode)) {
524         return 0;
525     }
526 
527     /* initialize */
528     if(matchLen1) {
529         U_ASSERT(matchLen2 !=NULL);
530         *matchLen1=0;
531         *matchLen2=0;
532     }
533 
534     start1=m1=org1=s1;
535     if(length1==-1) {
536         limit1=NULL;
537     } else {
538         limit1=s1+length1;
539     }
540 
541     start2=m2=org2=s2;
542     if(length2==-1) {
543         limit2=NULL;
544     } else {
545         limit2=s2+length2;
546     }
547 
548     level1=level2=0;
549     c1=c2=-1;
550 
551     /* comparison loop */
552     for(;;) {
553         /*
554          * here a code unit value of -1 means "get another code unit"
555          * below it will mean "this source is finished"
556          */
557 
558         if(c1<0) {
559             /* get next code unit from string 1, post-increment */
560             for(;;) {
561                 if(s1==limit1 || ((c1=*s1)==0 && (limit1==NULL || (options&_STRNCMP_STYLE)))) {
562                     if(level1==0) {
563                         c1=-1;
564                         break;
565                     }
566                 } else {
567                     ++s1;
568                     break;
569                 }
570 
571                 /* reached end of level buffer, pop one level */
572                 do {
573                     --level1;
574                     start1=stack1[level1].start;    /*Not uninitialized*/
575                 } while(start1==NULL);
576                 s1=stack1[level1].s;                /*Not uninitialized*/
577                 limit1=stack1[level1].limit;        /*Not uninitialized*/
578             }
579         }
580 
581         if(c2<0) {
582             /* get next code unit from string 2, post-increment */
583             for(;;) {
584                 if(s2==limit2 || ((c2=*s2)==0 && (limit2==NULL || (options&_STRNCMP_STYLE)))) {
585                     if(level2==0) {
586                         c2=-1;
587                         break;
588                     }
589                 } else {
590                     ++s2;
591                     break;
592                 }
593 
594                 /* reached end of level buffer, pop one level */
595                 do {
596                     --level2;
597                     start2=stack2[level2].start;    /*Not uninitialized*/
598                 } while(start2==NULL);
599                 s2=stack2[level2].s;                /*Not uninitialized*/
600                 limit2=stack2[level2].limit;        /*Not uninitialized*/
601             }
602         }
603 
604         /*
605          * compare c1 and c2
606          * either variable c1, c2 is -1 only if the corresponding string is finished
607          */
608         if(c1==c2) {
609             const UChar *next1, *next2;
610 
611             if(c1<0) {
612                 cmpRes=0;   /* c1==c2==-1 indicating end of strings */
613                 break;
614             }
615 
616             /*
617              * Note: Move the match positions in both strings at the same time
618              *      only when corresponding code point(s) in the original strings
619              *      are fully consumed. For example, when comparing s1="Fust" and
620              *      s2="Fu\u00dfball", s2[2] is folded into "ss", and s1[2] matches
621              *      the first code point in the case-folded data. But the second "s"
622              *      has no matching code point in s1, so this implementation returns
623              *      2 as the prefix match length ("Fu").
624              */
625             next1=next2=NULL;
626             if(level1==0) {
627                 next1=s1;
628             } else if(s1==limit1) {
629                 /* Note: This implementation only use a single level of stack.
630                  *      If this code needs to be changed to use multiple levels
631                  *      of stacks, the code above should check if the current
632                  *      code is at the end of all stacks.
633                  */
634                 U_ASSERT(level1==1);
635 
636                 /* is s1 at the end of the current stack? */
637                 next1=stack1[0].s;
638             }
639 
640             if (next1!=NULL) {
641                 if(level2==0) {
642                     next2=s2;
643                 } else if(s2==limit2) {
644                     U_ASSERT(level2==1);
645 
646                     /* is s2 at the end of the current stack? */
647                     next2=stack2[0].s;
648                 }
649                 if(next2!=NULL) {
650                     m1=next1;
651                     m2=next2;
652                 }
653             }
654             c1=c2=-1;       /* make us fetch new code units */
655             continue;
656         } else if(c1<0) {
657             cmpRes=-1;      /* string 1 ends before string 2 */
658             break;
659         } else if(c2<0) {
660             cmpRes=1;       /* string 2 ends before string 1 */
661             break;
662         }
663         /* c1!=c2 && c1>=0 && c2>=0 */
664 
665         /* get complete code points for c1, c2 for lookups if either is a surrogate */
666         cp1=c1;
667         if(U_IS_SURROGATE(c1)) {
668             UChar c;
669 
670             if(U_IS_SURROGATE_LEAD(c1)) {
671                 if(s1!=limit1 && U16_IS_TRAIL(c=*s1)) {
672                     /* advance ++s1; only below if cp1 decomposes/case-folds */
673                     cp1=U16_GET_SUPPLEMENTARY(c1, c);
674                 }
675             } else /* isTrail(c1) */ {
676                 if(start1<=(s1-2) && U16_IS_LEAD(c=*(s1-2))) {
677                     cp1=U16_GET_SUPPLEMENTARY(c, c1);
678                 }
679             }
680         }
681 
682         cp2=c2;
683         if(U_IS_SURROGATE(c2)) {
684             UChar c;
685 
686             if(U_IS_SURROGATE_LEAD(c2)) {
687                 if(s2!=limit2 && U16_IS_TRAIL(c=*s2)) {
688                     /* advance ++s2; only below if cp2 decomposes/case-folds */
689                     cp2=U16_GET_SUPPLEMENTARY(c2, c);
690                 }
691             } else /* isTrail(c2) */ {
692                 if(start2<=(s2-2) && U16_IS_LEAD(c=*(s2-2))) {
693                     cp2=U16_GET_SUPPLEMENTARY(c, c2);
694                 }
695             }
696         }
697 
698         /*
699          * go down one level for each string
700          * continue with the main loop as soon as there is a real change
701          */
702 
703         if( level1==0 &&
704             (length=ucase_toFullFolding(csp, (UChar32)cp1, &p, options))>=0
705         ) {
706             /* cp1 case-folds to the code point "length" or to p[length] */
707             if(U_IS_SURROGATE(c1)) {
708                 if(U_IS_SURROGATE_LEAD(c1)) {
709                     /* advance beyond source surrogate pair if it case-folds */
710                     ++s1;
711                 } else /* isTrail(c1) */ {
712                     /*
713                      * we got a supplementary code point when hitting its trail surrogate,
714                      * therefore the lead surrogate must have been the same as in the other string;
715                      * compare this decomposition with the lead surrogate in the other string
716                      * remember that this simulates bulk text replacement:
717                      * the decomposition would replace the entire code point
718                      */
719                     --s2;
720                     --m2;
721                     c2=*(s2-1);
722                 }
723             }
724 
725             /* push current level pointers */
726             stack1[0].start=start1;
727             stack1[0].s=s1;
728             stack1[0].limit=limit1;
729             ++level1;
730 
731             /* copy the folding result to fold1[] */
732             if(length<=UCASE_MAX_STRING_LENGTH) {
733                 u_memcpy(fold1, p, length);
734             } else {
735                 int32_t i=0;
736                 U16_APPEND_UNSAFE(fold1, i, length);
737                 length=i;
738             }
739 
740             /* set next level pointers to case folding */
741             start1=s1=fold1;
742             limit1=fold1+length;
743 
744             /* get ready to read from decomposition, continue with loop */
745             c1=-1;
746             continue;
747         }
748 
749         if( level2==0 &&
750             (length=ucase_toFullFolding(csp, (UChar32)cp2, &p, options))>=0
751         ) {
752             /* cp2 case-folds to the code point "length" or to p[length] */
753             if(U_IS_SURROGATE(c2)) {
754                 if(U_IS_SURROGATE_LEAD(c2)) {
755                     /* advance beyond source surrogate pair if it case-folds */
756                     ++s2;
757                 } else /* isTrail(c2) */ {
758                     /*
759                      * we got a supplementary code point when hitting its trail surrogate,
760                      * therefore the lead surrogate must have been the same as in the other string;
761                      * compare this decomposition with the lead surrogate in the other string
762                      * remember that this simulates bulk text replacement:
763                      * the decomposition would replace the entire code point
764                      */
765                     --s1;
766                     --m2;
767                     c1=*(s1-1);
768                 }
769             }
770 
771             /* push current level pointers */
772             stack2[0].start=start2;
773             stack2[0].s=s2;
774             stack2[0].limit=limit2;
775             ++level2;
776 
777             /* copy the folding result to fold2[] */
778             if(length<=UCASE_MAX_STRING_LENGTH) {
779                 u_memcpy(fold2, p, length);
780             } else {
781                 int32_t i=0;
782                 U16_APPEND_UNSAFE(fold2, i, length);
783                 length=i;
784             }
785 
786             /* set next level pointers to case folding */
787             start2=s2=fold2;
788             limit2=fold2+length;
789 
790             /* get ready to read from decomposition, continue with loop */
791             c2=-1;
792             continue;
793         }
794 
795         /*
796          * no decomposition/case folding, max level for both sides:
797          * return difference result
798          *
799          * code point order comparison must not just return cp1-cp2
800          * because when single surrogates are present then the surrogate pairs
801          * that formed cp1 and cp2 may be from different string indexes
802          *
803          * example: { d800 d800 dc01 } vs. { d800 dc00 }, compare at second code units
804          * c1=d800 cp1=10001 c2=dc00 cp2=10000
805          * cp1-cp2>0 but c1-c2<0 and in fact in UTF-32 it is { d800 10001 } < { 10000 }
806          *
807          * therefore, use same fix-up as in ustring.c/uprv_strCompare()
808          * except: uprv_strCompare() fetches c=*s while this functions fetches c=*s++
809          * so we have slightly different pointer/start/limit comparisons here
810          */
811 
812         if(c1>=0xd800 && c2>=0xd800 && (options&U_COMPARE_CODE_POINT_ORDER)) {
813             /* subtract 0x2800 from BMP code points to make them smaller than supplementary ones */
814             if(
815                 (c1<=0xdbff && s1!=limit1 && U16_IS_TRAIL(*s1)) ||
816                 (U16_IS_TRAIL(c1) && start1!=(s1-1) && U16_IS_LEAD(*(s1-2)))
817             ) {
818                 /* part of a surrogate pair, leave >=d800 */
819             } else {
820                 /* BMP code point - may be surrogate code point - make <d800 */
821                 c1-=0x2800;
822             }
823 
824             if(
825                 (c2<=0xdbff && s2!=limit2 && U16_IS_TRAIL(*s2)) ||
826                 (U16_IS_TRAIL(c2) && start2!=(s2-1) && U16_IS_LEAD(*(s2-2)))
827             ) {
828                 /* part of a surrogate pair, leave >=d800 */
829             } else {
830                 /* BMP code point - may be surrogate code point - make <d800 */
831                 c2-=0x2800;
832             }
833         }
834 
835         cmpRes=c1-c2;
836         break;
837     }
838 
839     if(matchLen1) {
840         *matchLen1=m1-org1;
841         *matchLen2=m2-org2;
842     }
843     return cmpRes;
844 }
845 
846 /* internal function */
847 U_CFUNC int32_t
u_strcmpFold(const UChar * s1,int32_t length1,const UChar * s2,int32_t length2,uint32_t options,UErrorCode * pErrorCode)848 u_strcmpFold(const UChar *s1, int32_t length1,
849              const UChar *s2, int32_t length2,
850              uint32_t options,
851              UErrorCode *pErrorCode) {
852     return _cmpFold(s1, length1, s2, length2, options, NULL, NULL, pErrorCode);
853 }
854 
855 /* public API functions */
856 
857 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)858 u_strCaseCompare(const UChar *s1, int32_t length1,
859                  const UChar *s2, int32_t length2,
860                  uint32_t options,
861                  UErrorCode *pErrorCode) {
862     /* argument checking */
863     if(pErrorCode==0 || U_FAILURE(*pErrorCode)) {
864         return 0;
865     }
866     if(s1==NULL || length1<-1 || s2==NULL || length2<-1) {
867         *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR;
868         return 0;
869     }
870     return u_strcmpFold(s1, length1, s2, length2,
871                         options|U_COMPARE_IGNORE_CASE,
872                         pErrorCode);
873 }
874 
875 U_CAPI int32_t U_EXPORT2
u_strcasecmp(const UChar * s1,const UChar * s2,uint32_t options)876 u_strcasecmp(const UChar *s1, const UChar *s2, uint32_t options) {
877     UErrorCode errorCode=U_ZERO_ERROR;
878     return u_strcmpFold(s1, -1, s2, -1,
879                         options|U_COMPARE_IGNORE_CASE,
880                         &errorCode);
881 }
882 
883 U_CAPI int32_t U_EXPORT2
u_memcasecmp(const UChar * s1,const UChar * s2,int32_t length,uint32_t options)884 u_memcasecmp(const UChar *s1, const UChar *s2, int32_t length, uint32_t options) {
885     UErrorCode errorCode=U_ZERO_ERROR;
886     return u_strcmpFold(s1, length, s2, length,
887                         options|U_COMPARE_IGNORE_CASE,
888                         &errorCode);
889 }
890 
891 U_CAPI int32_t U_EXPORT2
u_strncasecmp(const UChar * s1,const UChar * s2,int32_t n,uint32_t options)892 u_strncasecmp(const UChar *s1, const UChar *s2, int32_t n, uint32_t options) {
893     UErrorCode errorCode=U_ZERO_ERROR;
894     return u_strcmpFold(s1, n, s2, n,
895                         options|(U_COMPARE_IGNORE_CASE|_STRNCMP_STYLE),
896                         &errorCode);
897 }
898 
899 /* internal API - detect length of shared prefix */
900 U_CAPI void
u_caseInsensitivePrefixMatch(const UChar * s1,int32_t length1,const UChar * s2,int32_t length2,uint32_t options,int32_t * matchLen1,int32_t * matchLen2,UErrorCode * pErrorCode)901 u_caseInsensitivePrefixMatch(const UChar *s1, int32_t length1,
902                              const UChar *s2, int32_t length2,
903                              uint32_t options,
904                              int32_t *matchLen1, int32_t *matchLen2,
905                              UErrorCode *pErrorCode) {
906     _cmpFold(s1, length1, s2, length2, options,
907         matchLen1, matchLen2, pErrorCode);
908 }
909