icuSources/common/ustrcase.cpp

   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
  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
  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
 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
 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
 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
 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
 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
 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
 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
 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  */
 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
 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
 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
 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
 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
 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
 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 }