CFUniChar.c

   1 /*
   2  * Copyright (c) 2014 Apple Inc. All rights reserved.
   3  *
   4  * @APPLE_LICENSE_HEADER_START@
   5  *
   6  * This file contains Original Code and/or Modifications of Original Code
   7  * as defined in and that are subject to the Apple Public Source License
   8  * Version 2.0 (the 'License'). You may not use this file except in
   9  * compliance with the License. Please obtain a copy of the License at
  10  * http://www.opensource.apple.com/apsl/ and read it before using this
  11  * file.
  12  *
  13  * The Original Code and all software distributed under the License are
  14  * distributed on an 'AS IS' basis, WITHOUT WARRANTY OF ANY KIND, EITHER
  15  * EXPRESS OR IMPLIED, AND APPLE HEREBY DISCLAIMS ALL SUCH WARRANTIES,
  16  * INCLUDING WITHOUT LIMITATION, ANY WARRANTIES OF MERCHANTABILITY,
  17  * FITNESS FOR A PARTICULAR PURPOSE, QUIET ENJOYMENT OR NON-INFRINGEMENT.
  18  * Please see the License for the specific language governing rights and
  19  * limitations under the License.
  20  *
  21  * @APPLE_LICENSE_HEADER_END@
  22  */
  23
  24 /*      CFUniChar.c
  25         Copyright (c) 2001-2014, Apple Inc. All rights reserved.
  26         Responsibility: Aki Inoue
  27 */
  28
  29 #include <CoreFoundation/CFByteOrder.h>
  30 #include "CFInternal.h"
  31 #include "CFUniChar.h"
  32 #include "CFStringEncodingConverterExt.h"
  33 #include "CFUnicodeDecomposition.h"
  34 #include "CFUniCharPriv.h"
  35 #if DEPLOYMENT_TARGET_MACOSX || DEPLOYMENT_TARGET_EMBEDDED || DEPLOYMENT_TARGET_EMBEDDED_MINI || DEPLOYMENT_TARGET_LINUX || DEPLOYMENT_TARGET_FREEBSD
  36 #include <fcntl.h>
  37 #include <sys/types.h>
  38 #include <sys/stat.h>
  39 #include <sys/param.h>
  40 #include <sys/mman.h>
  41 #include <unistd.h>
  42 #include <stdlib.h>
  43 #endif
  44 #if DEPLOYMENT_TARGET_MACOSX || DEPLOYMENT_TARGET_EMBEDDED
  45 #include <mach/mach.h>
  46 #endif
  47
  48 #if DEPLOYMENT_TARGET_WINDOWS
  49 extern void _CFGetFrameworkPath(wchar_t *path, int maxLength);
  50 #endif
  51
  52 #if DEPLOYMENT_TARGET_MACOSX || DEPLOYMENT_TARGET_EMBEDDED
  53 #define __kCFCharacterSetDir "/System/Library/CoreServices"
  54 #elif DEPLOYMENT_TARGET_LINUX || DEPLOYMENT_TARGET_FREEBSD || DEPLOYMENT_TARGET_EMBEDDED_MINI
  55 #define __kCFCharacterSetDir "/usr/local/share/CoreFoundation"
  56 #elif DEPLOYMENT_TARGET_WINDOWS
  57 #define __kCFCharacterSetDir "\\Windows\\CoreFoundation"
  58 #endif
  59
  60 #if DEPLOYMENT_TARGET_MACOSX || DEPLOYMENT_TARGET_EMBEDDED
  61 #define USE_MACHO_SEGMENT 1
  62 #endif
  63
  64 enum {
  65     kCFUniCharLastExternalSet = kCFUniCharNewlineCharacterSet,
  66     kCFUniCharFirstInternalSet = kCFUniCharCompatibilityDecomposableCharacterSet,
  67     kCFUniCharLastInternalSet = kCFUniCharGraphemeExtendCharacterSet,
  68     kCFUniCharFirstBitmapSet = kCFUniCharDecimalDigitCharacterSet
  69 };
  70
  71 CF_INLINE uint32_t __CFUniCharMapExternalSetToInternalIndex(uint32_t cset) { return ((kCFUniCharFirstInternalSet <= cset) ? ((cset - kCFUniCharFirstInternalSet) + kCFUniCharLastExternalSet) : cset) - kCFUniCharFirstBitmapSet; }
  72 CF_INLINE uint32_t __CFUniCharMapCompatibilitySetID(uint32_t cset) { return ((cset == kCFUniCharControlCharacterSet) ? kCFUniCharControlAndFormatterCharacterSet : (((cset > kCFUniCharLastExternalSet) && (cset < kCFUniCharFirstInternalSet)) ? ((cset - kCFUniCharLastExternalSet) + kCFUniCharFirstInternalSet) : cset)); }
  73
  74 #if (DEPLOYMENT_TARGET_MACOSX || DEPLOYMENT_TARGET_EMBEDDED) && USE_MACHO_SEGMENT
  75 #include <mach-o/getsect.h>
  76 #include <mach-o/dyld.h>
  77 #include <mach-o/ldsyms.h>
  78
  79 extern const void* unicode_csbitmaps_section_start      __asm("section$start$__UNICODE$__csbitmaps");
  80 extern const void* unicode_csbitmaps_section_end        __asm("section$end$__UNICODE$__csbitmaps");
  81 extern const void* unicode_properties_section_start     __asm("section$start$__UNICODE$__properties");
  82 extern const void* unicode_properties_section_end       __asm("section$end$__UNICODE$__properties");
  83 extern const void* unicode_data_section_start           __asm("section$start$__UNICODE$__data");
  84 extern const void* unicode_data_section_end             __asm("section$end$__UNICODE$__data");
  85
  86 static const void *__CFGetSectDataPtr(const char *segname, const char *sectname, uint64_t *sizep) {
  87     // special case three common sections to have fast access
  88     if ( strcmp(segname, "__UNICODE") == 0 ) {
  89         if ( strcmp(sectname, "__csbitmaps") == 0)  {
  90             if (sizep) *sizep = &unicode_csbitmaps_section_end - &unicode_csbitmaps_section_start;
  91             return &unicode_csbitmaps_section_start;
  92         }
  93         else if ( strcmp(sectname, "__properties") == 0 ) {
  94             if (sizep) *sizep = &unicode_properties_section_end - &unicode_properties_section_start;
  95             return &unicode_properties_section_start;
  96         }
  97         else if ( strcmp(sectname, "__data") == 0 ) {
  98             if (sizep) *sizep = &unicode_data_section_end - &unicode_data_section_start;
  99             return &unicode_data_section_start;
 100         }
 101     }
 102
 103     uint32_t idx, cnt = _dyld_image_count();
 104     for (idx = 0; idx < cnt; idx++) {
 105        void *mh = (void *)_dyld_get_image_header(idx);
 106        if (mh != &_mh_dylib_header) continue;
 107 #if __LP64__
 108        const struct section_64 *sect = getsectbynamefromheader_64((struct mach_header_64 *)mh, segname, sectname);
 109 #else
 110        const struct section *sect = getsectbynamefromheader((struct mach_header *)mh, segname, sectname);
 111 #endif
 112        if (!sect) break;
 113        if (sizep) *sizep = (uint64_t)sect->size;
 114        return (char *)sect->addr + _dyld_get_image_vmaddr_slide(idx);
 115     }
 116     if (sizep) *sizep = 0ULL;
 117     return NULL;
 118 }
 119 #endif
 120
 121 #if !USE_MACHO_SEGMENT
 122
 123 // Memory map the file
 124
 125 #if DEPLOYMENT_TARGET_MACOSX || DEPLOYMENT_TARGET_EMBEDDED || DEPLOYMENT_TARGET_EMBEDDED_MINI || DEPLOYMENT_TARGET_LINUX
 126 CF_INLINE void __CFUniCharCharacterSetPath(char *cpath) {
 127 #elif DEPLOYMENT_TARGET_WINDOWS
 128 CF_INLINE void __CFUniCharCharacterSetPath(wchar_t *wpath) {
 129 #else
 130 #error Unknown or unspecified DEPLOYMENT_TARGET
 131 #endif
 132 #if DEPLOYMENT_TARGET_MACOSX || DEPLOYMENT_TARGET_EMBEDDED
 133     strlcpy(cpath, __kCFCharacterSetDir, MAXPATHLEN);
 134 #elif DEPLOYMENT_TARGET_LINUX
 135     strlcpy(cpath, __kCFCharacterSetDir, MAXPATHLEN);
 136 #elif DEPLOYMENT_TARGET_WINDOWS
 137     wchar_t frameworkPath[MAXPATHLEN];
 138     _CFGetFrameworkPath(frameworkPath, MAXPATHLEN);
 139     wcsncpy(wpath, frameworkPath, MAXPATHLEN);
 140     wcsncat(wpath, L"\\CoreFoundation.resources\\", MAXPATHLEN - wcslen(wpath));
 141 #else
 142     strlcpy(cpath, __kCFCharacterSetDir, MAXPATHLEN);
 143     strlcat(cpath, "/CharacterSets/", MAXPATHLEN);
 144 #endif
 145 }
 146
 147 #if DEPLOYMENT_TARGET_WINDOWS
 148 #define MAX_BITMAP_STATE 512
 149 //
 150 //  If a string is placed into this array, then it has been previously
 151 //  determined that the bitmap-file cannot be found.  Thus, we make
 152 //  the assumption it won't be there in future calls and we avoid
 153 //  hitting the disk un-necessarily.  This assumption isn't 100%
 154 //  correct, as bitmap-files can be added.  We would have to re-start
 155 //  the application in order to pick-up the new bitmap info.
 156 //
 157 //  We should probably re-visit this.
 158 //
 159 static wchar_t *mappedBitmapState[MAX_BITMAP_STATE];
 160 static int __nNumStateEntries = -1;
 161 CRITICAL_SECTION __bitmapStateLock = {0};
 162
 163 bool __GetBitmapStateForName(const wchar_t *bitmapName) {
 164     if (NULL == __bitmapStateLock.DebugInfo)
 165         InitializeCriticalSection(&__bitmapStateLock);
 166     EnterCriticalSection(&__bitmapStateLock);
 167     if (__nNumStateEntries >= 0) {
 168         for (int i = 0; i < __nNumStateEntries; i++) {
 169             if (wcscmp(mappedBitmapState[i], bitmapName) == 0) {
 170                 LeaveCriticalSection(&__bitmapStateLock);
 171                 return true;
 172             }
 173         }
 174     }
 175     LeaveCriticalSection(&__bitmapStateLock);
 176     return false;
 177 }
 178 void __AddBitmapStateForName(const wchar_t *bitmapName) {
 179     if (NULL == __bitmapStateLock.DebugInfo)
 180         InitializeCriticalSection(&__bitmapStateLock);
 181     EnterCriticalSection(&__bitmapStateLock);
 182     __nNumStateEntries++;
 183     mappedBitmapState[__nNumStateEntries] = (wchar_t *)malloc((lstrlenW(bitmapName)+1) * sizeof(wchar_t));
 184     lstrcpyW(mappedBitmapState[__nNumStateEntries], bitmapName);
 185     LeaveCriticalSection(&__bitmapStateLock);
 186 }
 187 #endif
 188
 189 #if DEPLOYMENT_TARGET_MACOSX || DEPLOYMENT_TARGET_EMBEDDED || DEPLOYMENT_TARGET_EMBEDDED_MINI || DEPLOYMENT_TARGET_LINUX
 190 static bool __CFUniCharLoadBytesFromFile(const char *fileName, const void **bytes, int64_t *fileSize) {
 191 #elif DEPLOYMENT_TARGET_WINDOWS
 192 static bool __CFUniCharLoadBytesFromFile(const wchar_t *fileName, const void **bytes, int64_t *fileSize) {
 193 #else
 194 #error Unknown or unspecified DEPLOYMENT_TARGET
 195 #endif
 196 #if DEPLOYMENT_TARGET_WINDOWS
 197     HANDLE bitmapFileHandle = NULL;
 198     HANDLE mappingHandle = NULL;
 199
 200     if (__GetBitmapStateForName(fileName)) {
 201         // The fileName has been tried in the past, so just return false
 202         // and move on.
 203         *bytes = NULL;
 204         return false;
 205     }
 206     mappingHandle = OpenFileMappingW(FILE_MAP_READ, TRUE, fileName);
 207     if (NULL == mappingHandle) {
 208         if ((bitmapFileHandle = CreateFileW(fileName, GENERIC_READ, FILE_SHARE_READ, NULL, OPEN_EXISTING, FILE_ATTRIBUTE_NORMAL, NULL)) == INVALID_HANDLE_VALUE) {
 209             // We tried to get the bitmap file for mapping, but it's not there.  Add to list of non-existant bitmap-files so
 210             // we don't have to try this again in the future.
 211             __AddBitmapStateForName(fileName);
 212             return false;
 213         }
 214         mappingHandle = CreateFileMapping(bitmapFileHandle, NULL, PAGE_READONLY, 0, 0, NULL);
 215         CloseHandle(bitmapFileHandle);
 216         if (!mappingHandle) return false;
 217     }
 218
 219     *bytes = MapViewOfFileEx(mappingHandle, FILE_MAP_READ, 0, 0, 0, 0);
 220
 221     if (NULL != fileSize) {
 222         MEMORY_BASIC_INFORMATION memoryInfo;
 223
 224         if (0 == VirtualQueryEx(mappingHandle, *bytes, &memoryInfo, sizeof(memoryInfo))) {
 225             *fileSize = 0; // This indicates no checking. Is it right ?
 226         } else {
 227             *fileSize = memoryInfo.RegionSize;
 228         }
 229     }
 230
 231     CloseHandle(mappingHandle);
 232
 233     return (*bytes ? true : false);
 234 #else
 235     struct stat statBuf;
 236     int fd = -1;
 237
 238     if ((fd = open(fileName, O_RDONLY, 0)) < 0) {
 239         return false;
 240     }
 241     if (fstat(fd, &statBuf) < 0 || (*bytes = mmap(0, statBuf.st_size, PROT_READ, MAP_PRIVATE, fd, 0)) == (void *)-1) {
 242         close(fd);
 243         return false;
 244     }
 245     close(fd);
 246
 247     if (NULL != fileSize) *fileSize = statBuf.st_size;
 248
 249     return true;
 250 #endif
 251 }
 252
 253 #endif // USE_MACHO_SEGMENT
 254
 255 #if DEPLOYMENT_TARGET_MACOSX || DEPLOYMENT_TARGET_EMBEDDED || DEPLOYMENT_TARGET_EMBEDDED_MINI || DEPLOYMENT_TARGET_LINUX
 256 static bool __CFUniCharLoadFile(const char *bitmapName, const void **bytes, int64_t *fileSize) {
 257 #elif DEPLOYMENT_TARGET_WINDOWS
 258 static bool __CFUniCharLoadFile(const wchar_t *bitmapName, const void **bytes, int64_t *fileSize) {
 259 #else
 260 #error Unknown or unspecified DEPLOYMENT_TARGET
 261 #endif
 262 #if USE_MACHO_SEGMENT
 263         *bytes = __CFGetSectDataPtr("__UNICODE", bitmapName, NULL);
 264
 265     if (NULL != fileSize) *fileSize = 0;
 266
 267     return *bytes ? true : false;
 268 #else
 269 #if DEPLOYMENT_TARGET_MACOSX || DEPLOYMENT_TARGET_EMBEDDED || DEPLOYMENT_TARGET_EMBEDDED_MINI || DEPLOYMENT_TARGET_LINUX
 270     char cpath[MAXPATHLEN];
 271     __CFUniCharCharacterSetPath(cpath);
 272     strlcat(cpath, bitmapName, MAXPATHLEN);
 273     Boolean needToFree = false;
 274     const char *possiblyFrameworkRootedCPath = CFPathRelativeToAppleFrameworksRoot(cpath, &needToFree);
 275     bool result = __CFUniCharLoadBytesFromFile(possiblyFrameworkRootedCPath, bytes, fileSize);
 276     if (needToFree) free((void *)possiblyFrameworkRootedCPath);
 277     return result;
 278 #elif DEPLOYMENT_TARGET_WINDOWS
 279     wchar_t wpath[MAXPATHLEN];
 280     __CFUniCharCharacterSetPath(wpath);
 281     wcsncat(wpath, bitmapName, MAXPATHLEN);
 282     return __CFUniCharLoadBytesFromFile(wpath, bytes, fileSize);
 283 #else
 284 #error Unknown or unspecified DEPLOYMENT_TARGET
 285 #endif
 286 #endif
 287 }
 288
 289 // Bitmap functions
 290 /*
 291  Currently unused but left in for symmetry/informative purposes
 292  CF_INLINE bool isControl(UTF32Char theChar, uint16_t charset, const void *data) { // ISO Control
 293     return (((theChar <= 0x001F) || (theChar >= 0x007F && theChar <= 0x009F)) ? true : false);
 294 }*/
 295
 296 CF_INLINE bool isWhitespace(UTF32Char theChar, uint16_t charset, const void *data) { // Space
 297     return (((theChar == 0x0020) || (theChar == 0x0009) || (theChar == 0x00A0) || (theChar == 0x1680) || (theChar >= 0x2000 && theChar <= 0x200B) || (theChar == 0x202F) || (theChar == 0x205F) || (theChar == 0x3000)) ? true : false);
 298 }
 299
 300 CF_INLINE bool isNewline(UTF32Char theChar, uint16_t charset, const void *data) { // White space
 301     return (((theChar >= 0x000A && theChar <= 0x000D) || (theChar == 0x0085) || (theChar == 0x2028) || (theChar == 0x2029)) ? true : false);
 302 }
 303
 304 CF_INLINE bool isWhitespaceAndNewline(UTF32Char theChar, uint16_t charset, const void *data) { // White space
 305     return ((isWhitespace(theChar, charset, data) || isNewline(theChar, charset, data)) ? true : false);
 306 }
 307
 308 #if USE_MACHO_SEGMENT
 309 CF_INLINE bool __CFSimpleFileSizeVerification(const void *bytes, int64_t fileSize) { return true; }
 310 #elif 1
 311 // <rdar://problem/8961744> __CFSimpleFileSizeVerification is broken
 312 static bool __CFSimpleFileSizeVerification(const void *bytes, int64_t fileSize) { return true; }
 313 #else
 314 static bool __CFSimpleFileSizeVerification(const void *bytes, int64_t fileSize) {
 315     bool result = true;
 316
 317     if (fileSize > 0) {
 318         if ((sizeof(uint32_t) * 2) > fileSize) {
 319             result = false;
 320         } else {
 321             uint32_t headerSize = CFSwapInt32BigToHost(*((uint32_t *)((char *)bytes + 4)));
 322
 323             if ((headerSize < (sizeof(uint32_t) * 4)) || (headerSize > fileSize)) {
 324                 result = false;
 325             } else {
 326                 const uint32_t *lastElement = (uint32_t *)(((uint8_t *)bytes) + headerSize) - 2;
 327
 328                 if ((headerSize + CFSwapInt32BigToHost(lastElement[0]) + CFSwapInt32BigToHost(lastElement[1])) > headerSize) result = false;
 329             }
 330         }
 331     }
 332
 333     if (!result) CFLog(kCFLogLevelCritical, CFSTR("File size verification for Unicode database file failed."));
 334
 335     return result;
 336 }
 337 #endif // USE_MACHO_SEGMENT
 338
 339 typedef struct {
 340     uint32_t _numPlanes;
 341     const uint8_t **_planes;
 342 } __CFUniCharBitmapData;
 343
 344 static char __CFUniCharUnicodeVersionString[8] = {0, 0, 0, 0, 0, 0, 0, 0};
 345
 346 static uint32_t __CFUniCharNumberOfBitmaps = 0;
 347 static __CFUniCharBitmapData *__CFUniCharBitmapDataArray = NULL;
 348
 349 static CFLock_t __CFUniCharBitmapLock = CFLockInit;
 350
 351 #if DEPLOYMENT_TARGET_MACOSX || DEPLOYMENT_TARGET_EMBEDDED || DEPLOYMENT_TARGET_EMBEDDED_MINI || DEPLOYMENT_TARGET_LINUX
 352 #if !defined(CF_UNICHAR_BITMAP_FILE)
 353 #if USE_MACHO_SEGMENT
 354 #define CF_UNICHAR_BITMAP_FILE "__csbitmaps"
 355 #else
 356 #define CF_UNICHAR_BITMAP_FILE "/CFCharacterSetBitmaps.bitmap"
 357 #endif
 358 #endif
 359 #elif DEPLOYMENT_TARGET_WINDOWS
 360 #if !defined(CF_UNICHAR_BITMAP_FILE)
 361 #define CF_UNICHAR_BITMAP_FILE L"CFCharacterSetBitmaps.bitmap"
 362 #endif
 363 #else
 364 #error Unknown or unspecified DEPLOYMENT_TARGET
 365 #endif
 366
 367 static bool __CFUniCharLoadBitmapData(void) {
 368     __CFUniCharBitmapData *array;
 369     uint32_t headerSize;
 370     uint32_t bitmapSize;
 371     int numPlanes;
 372     uint8_t currentPlane;
 373     const void *bytes;
 374     const void *bitmapBase;
 375     const void *bitmap;
 376     int idx, bitmapIndex;
 377     int64_t fileSize;
 378
 379     __CFLock(&__CFUniCharBitmapLock);
 380
 381     if (__CFUniCharBitmapDataArray || !__CFUniCharLoadFile(CF_UNICHAR_BITMAP_FILE, &bytes, &fileSize) || !__CFSimpleFileSizeVerification(bytes, fileSize)) {
 382         __CFUnlock(&__CFUniCharBitmapLock);
 383         return false;
 384     }
 385
 386     for (idx = 0;idx < 4 && ((const uint8_t *)bytes)[idx];idx++) {
 387         __CFUniCharUnicodeVersionString[idx * 2] = ((const uint8_t *)bytes)[idx];
 388         __CFUniCharUnicodeVersionString[idx * 2 + 1] = '.';
 389     }
 390     __CFUniCharUnicodeVersionString[(idx < 4 ? idx * 2 - 1 : 7)] = '\0';
 391
 392     headerSize = CFSwapInt32BigToHost(*((uint32_t *)((char *)bytes + 4)));
 393
 394     bitmapBase = (uint8_t *)bytes + headerSize;
 395     bytes = (uint8_t *)bytes + (sizeof(uint32_t) * 2);
 396     headerSize -= (sizeof(uint32_t) * 2);
 397
 398     __CFUniCharNumberOfBitmaps = headerSize / (sizeof(uint32_t) * 2);
 399
 400     array = (__CFUniCharBitmapData *)CFAllocatorAllocate(kCFAllocatorSystemDefault, sizeof(__CFUniCharBitmapData) * __CFUniCharNumberOfBitmaps, 0);
 401
 402     for (idx = 0;idx < (int)__CFUniCharNumberOfBitmaps;idx++) {
 403         bitmap = (uint8_t *)bitmapBase + CFSwapInt32BigToHost(*((uint32_t *)bytes)); bytes = (uint8_t *)bytes + sizeof(uint32_t);
 404         bitmapSize = CFSwapInt32BigToHost(*((uint32_t *)bytes)); bytes = (uint8_t *)bytes + sizeof(uint32_t);
 405
 406         numPlanes = bitmapSize / (8 * 1024);
 407         numPlanes = *(const uint8_t *)((char *)bitmap + (((numPlanes - 1) * ((8 * 1024) + 1)) - 1)) + 1;
 408         array[idx]._planes = (const uint8_t **)CFAllocatorAllocate(kCFAllocatorSystemDefault, sizeof(const void *) * numPlanes, 0);
 409         array[idx]._numPlanes = numPlanes;
 410
 411         currentPlane = 0;
 412         for (bitmapIndex = 0;bitmapIndex < numPlanes;bitmapIndex++) {
 413             if (bitmapIndex == currentPlane) {
 414                 array[idx]._planes[bitmapIndex] = (const uint8_t *)bitmap;
 415                 bitmap = (uint8_t *)bitmap + (8 * 1024);
 416 #if defined (__cplusplus)
 417                                 currentPlane = *(((const uint8_t*&)bitmap)++);
 418 #else
 419                                 currentPlane = *((const uint8_t *)bitmap++);
 420 #endif
 421
 422             } else {
 423                 array[idx]._planes[bitmapIndex] = NULL;
 424             }
 425         }
 426     }
 427
 428     __CFUniCharBitmapDataArray = array;
 429
 430     __CFUnlock(&__CFUniCharBitmapLock);
 431
 432     return true;
 433 }
 434
 435 CF_PRIVATE const char *__CFUniCharGetUnicodeVersionString(void) {
 436     if (NULL == __CFUniCharBitmapDataArray) __CFUniCharLoadBitmapData();
 437     return __CFUniCharUnicodeVersionString;
 438 }
 439
 440 bool CFUniCharIsMemberOf(UTF32Char theChar, uint32_t charset) {
 441     charset = __CFUniCharMapCompatibilitySetID(charset);
 442
 443     switch (charset) {
 444         case kCFUniCharWhitespaceCharacterSet:
 445             return isWhitespace(theChar, charset, NULL);
 446
 447         case kCFUniCharWhitespaceAndNewlineCharacterSet:
 448             return isWhitespaceAndNewline(theChar, charset, NULL);
 449
 450         case kCFUniCharNewlineCharacterSet:
 451             return isNewline(theChar, charset, NULL);
 452
 453         default: {
 454             uint32_t tableIndex = __CFUniCharMapExternalSetToInternalIndex(charset);
 455
 456             if (NULL == __CFUniCharBitmapDataArray) __CFUniCharLoadBitmapData();
 457
 458             if (tableIndex < __CFUniCharNumberOfBitmaps) {
 459                 __CFUniCharBitmapData *data = __CFUniCharBitmapDataArray + tableIndex;
 460                 uint8_t planeNo = (theChar >> 16) & 0xFF;
 461
 462                 // The bitmap data for kCFUniCharIllegalCharacterSet is actually LEGAL set less Plane 14 ~ 16
 463                 if (charset == kCFUniCharIllegalCharacterSet) {
 464                     if (planeNo == 0x0E) { // Plane 14
 465                         theChar &= 0xFF;
 466                         return (((theChar == 0x01) || ((theChar > 0x1F) && (theChar < 0x80))) ? false : true);
 467                     } else if (planeNo == 0x0F || planeNo == 0x10) { // Plane 15 & 16
 468                         return ((theChar & 0xFF) > 0xFFFD ? true : false);
 469                     } else {
 470                         return (planeNo < data->_numPlanes && data->_planes[planeNo] ? !CFUniCharIsMemberOfBitmap(theChar, data->_planes[planeNo]) : true);
 471                     }
 472                 } else if (charset == kCFUniCharControlAndFormatterCharacterSet) {
 473                     if (planeNo == 0x0E) { // Plane 14
 474                         theChar &= 0xFF;
 475                         return (((theChar == 0x01) || ((theChar > 0x1F) && (theChar < 0x80))) ? true : false);
 476                     } else {
 477                         return (planeNo < data->_numPlanes && data->_planes[planeNo] ? CFUniCharIsMemberOfBitmap(theChar, data->_planes[planeNo]) : false);
 478                     }
 479                 } else {
 480                     return (planeNo < data->_numPlanes && data->_planes[planeNo] ? CFUniCharIsMemberOfBitmap(theChar, data->_planes[planeNo]) : false);
 481                 }
 482             }
 483             return false;
 484         }
 485     }
 486 }
 487
 488 const uint8_t *CFUniCharGetBitmapPtrForPlane(uint32_t charset, uint32_t plane) {
 489     if (NULL == __CFUniCharBitmapDataArray) __CFUniCharLoadBitmapData();
 490
 491     charset = __CFUniCharMapCompatibilitySetID(charset);
 492
 493     if ((charset > kCFUniCharWhitespaceAndNewlineCharacterSet) && (charset != kCFUniCharIllegalCharacterSet) && (charset != kCFUniCharNewlineCharacterSet)) {
 494         uint32_t tableIndex = __CFUniCharMapExternalSetToInternalIndex(charset);
 495
 496         if (tableIndex < __CFUniCharNumberOfBitmaps) {
 497             __CFUniCharBitmapData *data = __CFUniCharBitmapDataArray + tableIndex;
 498
 499             return (plane < data->_numPlanes ? data->_planes[plane] : NULL);
 500         }
 501     }
 502     return NULL;
 503 }
 504
 505 CF_PRIVATE uint8_t CFUniCharGetBitmapForPlane(uint32_t charset, uint32_t plane, void *bitmap, bool isInverted) {
 506     const uint8_t *src = CFUniCharGetBitmapPtrForPlane(charset, plane);
 507     int numBytes = (8 * 1024);
 508
 509     if (src) {
 510         if (isInverted) {
 511 #if defined (__cplusplus)
 512                         while (numBytes-- > 0) *(((uint8_t *&)bitmap)++) = ~(*(src++));
 513 #else
 514                         while (numBytes-- > 0) *((uint8_t *)bitmap++) = ~(*(src++));
 515 #endif
 516         } else {
 517 #if defined (__cplusplus)
 518             while (numBytes-- > 0) *(((uint8_t *&)bitmap)++) = *(src++);
 519 #else
 520                         while (numBytes-- > 0) *((uint8_t *)bitmap++) = *(src++);
 521 #endif
 522         }
 523         return kCFUniCharBitmapFilled;
 524     } else if (charset == kCFUniCharIllegalCharacterSet) {
 525         __CFUniCharBitmapData *data = __CFUniCharBitmapDataArray + __CFUniCharMapExternalSetToInternalIndex(__CFUniCharMapCompatibilitySetID(charset));
 526
 527         if (plane < data->_numPlanes && (src = data->_planes[plane])) {
 528             if (isInverted) {
 529 #if defined (__cplusplus)
 530                                 while (numBytes-- > 0) *(((uint8_t *&)bitmap)++) = *(src++);
 531 #else
 532                                 while (numBytes-- > 0) *((uint8_t *)bitmap++) = *(src++);
 533 #endif
 534             } else {
 535 #if defined (__cplusplus)
 536                 while (numBytes-- > 0) *(((uint8_t *&)bitmap)++) = ~(*(src++));
 537 #else
 538                                 while (numBytes-- > 0) *((uint8_t *)bitmap++) = ~(*(src++));
 539 #endif
 540             }
 541             return kCFUniCharBitmapFilled;
 542         } else if (plane == 0x0E) { // Plane 14
 543             int idx;
 544             uint8_t asciiRange = (isInverted ? (uint8_t)0xFF : (uint8_t)0);
 545             uint8_t otherRange = (isInverted ? (uint8_t)0 : (uint8_t)0xFF);
 546
 547 #if defined (__cplusplus)
 548                         *(((uint8_t *&)bitmap)++) = 0x02; // UE0001 LANGUAGE TAG
 549 #else
 550                         *((uint8_t *)bitmap++) = 0x02; // UE0001 LANGUAGE TAG
 551 #endif
 552             for (idx = 1;idx < numBytes;idx++) {
 553 #if defined (__cplusplus)
 554                                 *(((uint8_t *&)bitmap)++) = ((idx >= (0x20 / 8) && (idx < (0x80 / 8))) ? asciiRange : otherRange);
 555 #else
 556                                 *((uint8_t *)bitmap++) = ((idx >= (0x20 / 8) && (idx < (0x80 / 8))) ? asciiRange : otherRange);
 557 #endif
 558             }
 559             return kCFUniCharBitmapFilled;
 560         } else if (plane == 0x0F || plane == 0x10) { // Plane 15 & 16
 561             uint32_t value = (isInverted ? ~0 : 0);
 562             numBytes /= 4; // for 32bit
 563
 564             while (numBytes-- > 0) {
 565                 *((uint32_t *)bitmap) = value;
 566 #if defined (__cplusplus)
 567                                 bitmap = (uint8_t *)bitmap + sizeof(uint32_t);
 568 #else
 569                                 bitmap += sizeof(uint32_t);
 570 #endif
 571             }
 572             *(((uint8_t *)bitmap) - 5) = (isInverted ? 0x3F : 0xC0); // 0xFFFE & 0xFFFF
 573             return kCFUniCharBitmapFilled;
 574         }
 575         return (isInverted ? kCFUniCharBitmapEmpty : kCFUniCharBitmapAll);
 576     } else if ((charset < kCFUniCharDecimalDigitCharacterSet) || (charset == kCFUniCharNewlineCharacterSet)) {
 577         if (plane) return (isInverted ? kCFUniCharBitmapAll : kCFUniCharBitmapEmpty);
 578
 579         uint8_t *bitmapBase = (uint8_t *)bitmap;
 580         CFIndex idx;
 581         uint8_t nonFillValue = (isInverted ? (uint8_t)0xFF : (uint8_t)0);
 582
 583 #if defined (__cplusplus)
 584                     while (numBytes-- > 0) *(((uint8_t *&)bitmap)++) = nonFillValue;
 585 #else
 586                     while (numBytes-- > 0) *((uint8_t *)bitmap++) = nonFillValue;
 587 #endif
 588
 589         if ((charset == kCFUniCharWhitespaceAndNewlineCharacterSet) || (charset == kCFUniCharNewlineCharacterSet)) {
 590             const UniChar newlines[] = {0x000A, 0x000B, 0x000C, 0x000D, 0x0085, 0x2028, 0x2029};
 591
 592             for (idx = 0;idx < (int)(sizeof(newlines) / sizeof(*newlines)); idx++) {
 593                 if (isInverted) {
 594                     CFUniCharRemoveCharacterFromBitmap(newlines[idx], bitmapBase);
 595                 } else {
 596                     CFUniCharAddCharacterToBitmap(newlines[idx], bitmapBase);
 597                 }
 598             }
 599
 600             if (charset == kCFUniCharNewlineCharacterSet) return kCFUniCharBitmapFilled;
 601         }
 602
 603         if (isInverted) {
 604             CFUniCharRemoveCharacterFromBitmap(0x0009, bitmapBase);
 605             CFUniCharRemoveCharacterFromBitmap(0x0020, bitmapBase);
 606             CFUniCharRemoveCharacterFromBitmap(0x00A0, bitmapBase);
 607             CFUniCharRemoveCharacterFromBitmap(0x1680, bitmapBase);
 608             CFUniCharRemoveCharacterFromBitmap(0x202F, bitmapBase);
 609             CFUniCharRemoveCharacterFromBitmap(0x205F, bitmapBase);
 610             CFUniCharRemoveCharacterFromBitmap(0x3000, bitmapBase);
 611         } else {
 612             CFUniCharAddCharacterToBitmap(0x0009, bitmapBase);
 613             CFUniCharAddCharacterToBitmap(0x0020, bitmapBase);
 614             CFUniCharAddCharacterToBitmap(0x00A0, bitmapBase);
 615             CFUniCharAddCharacterToBitmap(0x1680, bitmapBase);
 616             CFUniCharAddCharacterToBitmap(0x202F, bitmapBase);
 617             CFUniCharAddCharacterToBitmap(0x205F, bitmapBase);
 618             CFUniCharAddCharacterToBitmap(0x3000, bitmapBase);
 619         }
 620
 621         for (idx = 0x2000;idx <= 0x200B;idx++) {
 622             if (isInverted) {
 623                 CFUniCharRemoveCharacterFromBitmap(idx, bitmapBase);
 624             } else {
 625                 CFUniCharAddCharacterToBitmap(idx, bitmapBase);
 626             }
 627         }
 628         return kCFUniCharBitmapFilled;
 629     }
 630     return (isInverted ? kCFUniCharBitmapAll : kCFUniCharBitmapEmpty);
 631 }
 632
 633 CF_PRIVATE uint32_t CFUniCharGetNumberOfPlanes(uint32_t charset) {
 634     if ((charset == kCFUniCharControlCharacterSet) || (charset == kCFUniCharControlAndFormatterCharacterSet)) {
 635         return 15; // 0 to 14
 636     } else if (charset < kCFUniCharDecimalDigitCharacterSet) {
 637         return 1;
 638     } else if (charset == kCFUniCharIllegalCharacterSet) {
 639         return 17;
 640     } else {
 641         uint32_t numPlanes;
 642
 643         if (NULL == __CFUniCharBitmapDataArray) __CFUniCharLoadBitmapData();
 644
 645         numPlanes = __CFUniCharBitmapDataArray[__CFUniCharMapExternalSetToInternalIndex(__CFUniCharMapCompatibilitySetID(charset))]._numPlanes;
 646
 647         return numPlanes;
 648     }
 649 }
 650
 651 // Mapping data loading
 652 static const void **__CFUniCharMappingTables = NULL;
 653
 654 static CFLock_t __CFUniCharMappingTableLock = CFLockInit;
 655
 656 #if DEPLOYMENT_TARGET_MACOSX || DEPLOYMENT_TARGET_EMBEDDED || DEPLOYMENT_TARGET_EMBEDDED_MINI || DEPLOYMENT_TARGET_LINUX
 657 #if __CF_BIG_ENDIAN__
 658 #if USE_MACHO_SEGMENT
 659 #define MAPPING_TABLE_FILE "__data"
 660 #else
 661 #define MAPPING_TABLE_FILE "/CFUnicodeData-B.mapping"
 662 #endif
 663 #else
 664 #if USE_MACHO_SEGMENT
 665 #define MAPPING_TABLE_FILE "__data"
 666 #else
 667 #define MAPPING_TABLE_FILE "/CFUnicodeData-L.mapping"
 668 #endif
 669 #endif
 670 #elif DEPLOYMENT_TARGET_WINDOWS
 671 #if __CF_BIG_ENDIAN__
 672 #if USE_MACHO_SEGMENT
 673 #define MAPPING_TABLE_FILE "__data"
 674 #else
 675 #define MAPPING_TABLE_FILE L"CFUnicodeData-B.mapping"
 676 #endif
 677 #else
 678 #if USE_MACHO_SEGMENT
 679 #define MAPPING_TABLE_FILE "__data"
 680 #else
 681 #define MAPPING_TABLE_FILE L"CFUnicodeData-L.mapping"
 682 #endif
 683 #endif
 684 #else
 685 #error Unknown or unspecified DEPLOYMENT_TARGET
 686 #endif
 687
 688 CF_PRIVATE const void *CFUniCharGetMappingData(uint32_t type) {
 689
 690     __CFLock(&__CFUniCharMappingTableLock);
 691
 692     if (NULL == __CFUniCharMappingTables) {
 693         const void *bytes;
 694         const void *bodyBase;
 695         int headerSize;
 696         int idx, count;
 697         int64_t fileSize;
 698
 699         if (!__CFUniCharLoadFile(MAPPING_TABLE_FILE, &bytes, &fileSize) || !__CFSimpleFileSizeVerification(bytes, fileSize)) {
 700             __CFUnlock(&__CFUniCharMappingTableLock);
 701             return NULL;
 702         }
 703
 704 #if defined (__cplusplus)
 705                 bytes = (uint8_t *)bytes + 4; // Skip Unicode version
 706                 headerSize = *((uint8_t *)bytes); bytes = (uint8_t *)bytes + sizeof(uint32_t);
 707 #else
 708                 bytes += 4; // Skip Unicode version
 709                 headerSize = *((uint32_t *)bytes); bytes += sizeof(uint32_t);
 710 #endif
 711         headerSize -= (sizeof(uint32_t) * 2);
 712         bodyBase = (char *)bytes + headerSize;
 713
 714         count = headerSize / sizeof(uint32_t);
 715
 716         __CFUniCharMappingTables = (const void **)CFAllocatorAllocate(kCFAllocatorSystemDefault, sizeof(const void *) * count, 0);
 717
 718         for (idx = 0;idx < count;idx++) {
 719 #if defined (__cplusplus)
 720                         __CFUniCharMappingTables[idx] = (char *)bodyBase + *((uint32_t *)bytes); bytes = (uint8_t *)bytes + sizeof(uint32_t);
 721 #else
 722                         __CFUniCharMappingTables[idx] = (char *)bodyBase + *((uint32_t *)bytes); bytes += sizeof(uint32_t);
 723 #endif
 724         }
 725     }
 726
 727     __CFUnlock(&__CFUniCharMappingTableLock);
 728
 729     return __CFUniCharMappingTables[type];
 730 }
 731
 732 // Case mapping functions
 733 #define DO_SPECIAL_CASE_MAPPING 1
 734
 735 static uint32_t *__CFUniCharCaseMappingTableCounts = NULL;
 736 static uint32_t **__CFUniCharCaseMappingTable = NULL;
 737 static const uint32_t **__CFUniCharCaseMappingExtraTable = NULL;
 738
 739 typedef struct {
 740     uint32_t _key;
 741     uint32_t _value;
 742 } __CFUniCharCaseMappings;
 743
 744 /* Binary searches CFStringEncodingUnicodeTo8BitCharMap */
 745 static uint32_t __CFUniCharGetMappedCase(const __CFUniCharCaseMappings *theTable, uint32_t numElem, UTF32Char character) {
 746     const __CFUniCharCaseMappings *p, *q, *divider;
 747
 748     if ((character < theTable[0]._key) || (character > theTable[numElem-1]._key)) {
 749         return 0;
 750     }
 751     p = theTable;
 752     q = p + (numElem-1);
 753     while (p <= q) {
 754         divider = p + ((q - p) >> 1);   /* divide by 2 */
 755         if (character < divider->_key) { q = divider - 1; }
 756         else if (character > divider->_key) { p = divider + 1; }
 757         else { return divider->_value; }
 758     }
 759     return 0;
 760 }
 761
 762 #define NUM_CASE_MAP_DATA (kCFUniCharCaseFold + 1)
 763
 764 static bool __CFUniCharLoadCaseMappingTable(void) {
 765     uint32_t *countArray;
 766     int idx;
 767
 768     if (NULL == __CFUniCharMappingTables) (void)CFUniCharGetMappingData(kCFUniCharToLowercase);
 769     if (NULL == __CFUniCharMappingTables) return false;
 770
 771     __CFLock(&__CFUniCharMappingTableLock);
 772
 773     if (__CFUniCharCaseMappingTableCounts) {
 774         __CFUnlock(&__CFUniCharMappingTableLock);
 775         return true;
 776     }
 777
 778     countArray = (uint32_t *)CFAllocatorAllocate(kCFAllocatorSystemDefault, sizeof(uint32_t) * NUM_CASE_MAP_DATA + sizeof(uint32_t *) * NUM_CASE_MAP_DATA * 2, 0);
 779     __CFUniCharCaseMappingTable = (uint32_t **)((char *)countArray + sizeof(uint32_t) * NUM_CASE_MAP_DATA);
 780     __CFUniCharCaseMappingExtraTable = (const uint32_t **)__CFUniCharCaseMappingTable + NUM_CASE_MAP_DATA;
 781
 782     for (idx = 0;idx < NUM_CASE_MAP_DATA;idx++) {
 783         countArray[idx] = *((uint32_t *)__CFUniCharMappingTables[idx]) / (sizeof(uint32_t) * 2);
 784         __CFUniCharCaseMappingTable[idx] = ((uint32_t *)__CFUniCharMappingTables[idx]) + 1;
 785         __CFUniCharCaseMappingExtraTable[idx] = (const uint32_t *)((char *)__CFUniCharCaseMappingTable[idx] + *((uint32_t *)__CFUniCharMappingTables[idx]));
 786     }
 787
 788     __CFUniCharCaseMappingTableCounts = countArray;
 789
 790     __CFUnlock(&__CFUniCharMappingTableLock);
 791     return true;
 792 }
 793
 794 #if __CF_BIG_ENDIAN__
 795 #define TURKISH_LANG_CODE       (0x7472) // tr
 796 #define LITHUANIAN_LANG_CODE    (0x6C74) // lt
 797 #define AZERI_LANG_CODE         (0x617A) // az
 798 #define DUTCH_LANG_CODE         (0x6E6C) // nl
 799 #define GREEK_LANG_CODE         (0x656C) // el
 800 #else
 801 #define TURKISH_LANG_CODE       (0x7274) // tr
 802 #define LITHUANIAN_LANG_CODE    (0x746C) // lt
 803 #define AZERI_LANG_CODE         (0x7A61) // az
 804 #define DUTCH_LANG_CODE         (0x6C6E) // nl
 805 #define GREEK_LANG_CODE         (0x6C65) // el
 806 #endif
 807
 808 CFIndex CFUniCharMapCaseTo(UTF32Char theChar, UTF16Char *convertedChar, CFIndex maxLength, uint32_t ctype, uint32_t flags, const uint8_t *langCode) {
 809     __CFUniCharBitmapData *data;
 810     uint8_t planeNo = (theChar >> 16) & 0xFF;
 811
 812 caseFoldRetry:
 813
 814 #if DO_SPECIAL_CASE_MAPPING
 815     if (flags & kCFUniCharCaseMapFinalSigma) {
 816         if (theChar == 0x03A3) { // Final sigma
 817             *convertedChar = (ctype == kCFUniCharToLowercase ? 0x03C2 : 0x03A3);
 818             return 1;
 819         }
 820     }
 821
 822     if (langCode) {
 823         if (flags & kCFUniCharCaseMapGreekTonos) { // localized Greek uppercasing
 824             if (theChar == 0x0301) { // GREEK TONOS
 825                 return 0;
 826             } else if (theChar == 0x0344) {// COMBINING GREEK DIALYTIKA TONOS
 827                 *convertedChar = 0x0308; // COMBINING GREEK DIALYTIKA
 828                 return 1;
 829             } else if (CFUniCharIsMemberOf(theChar, kCFUniCharDecomposableCharacterSet)) {
 830                 UTF32Char buffer[MAX_DECOMPOSED_LENGTH];
 831                 CFIndex length = CFUniCharDecomposeCharacter(theChar, buffer, MAX_DECOMPOSED_LENGTH);
 832
 833                 if (length > 1) {
 834                     UTF32Char *characters = buffer + 1;
 835                     UTF32Char *tail = buffer + length;
 836
 837                     while (characters < tail) {
 838                         if (*characters == 0x0301) break;
 839                         ++characters;
 840                     }
 841
 842                     if (characters < tail) { // found a tonos
 843                         CFIndex convertedLength = CFUniCharMapCaseTo(*buffer, convertedChar, maxLength, ctype, 0, langCode);
 844
 845                         if (convertedLength == 0) {
 846                             *convertedChar = (UTF16Char)*buffer;
 847                             convertedLength = 1;
 848                         }
 849
 850                         characters = buffer + 1;
 851
 852                         while (characters < tail) {
 853                             if (*characters != 0x0301) { // not tonos
 854                                 if (*characters < 0x10000) { // BMP
 855                                     convertedChar[convertedLength] = (UTF16Char)*characters;
 856                                     ++convertedLength;
 857                                 } else {
 858                                     UTF32Char character = *characters - 0x10000;
 859                                     convertedChar[convertedLength++] = (UTF16Char)((character >> 10) + 0xD800UL);
 860                                     convertedChar[convertedLength++] = (UTF16Char)((character & 0x3FF) + 0xDC00UL);
 861                                 }
 862                             }
 863                             ++characters;
 864                         }
 865
 866                         return convertedLength;
 867                     }
 868                 }
 869             }
 870         }
 871         switch (*(uint16_t *)langCode) {
 872             case LITHUANIAN_LANG_CODE:
 873                 if (theChar == 0x0307 && (flags & kCFUniCharCaseMapAfter_i)) {
 874                     return 0;
 875                 } else if (ctype == kCFUniCharToLowercase) {
 876                     if (flags & kCFUniCharCaseMapMoreAbove) {
 877                         switch (theChar) {
 878                             case 0x0049: // LATIN CAPITAL LETTER I
 879                                 *(convertedChar++) = 0x0069;
 880                                 *(convertedChar++) = 0x0307;
 881                                 return 2;
 882
 883                             case 0x004A: // LATIN CAPITAL LETTER J
 884                                 *(convertedChar++) = 0x006A;
 885                                 *(convertedChar++) = 0x0307;
 886                                 return 2;
 887
 888                             case 0x012E: // LATIN CAPITAL LETTER I WITH OGONEK
 889                                 *(convertedChar++) = 0x012F;
 890                                 *(convertedChar++) = 0x0307;
 891                                 return 2;
 892
 893                             default: break;
 894                         }
 895                     }
 896                     switch (theChar) {
 897                         case 0x00CC: // LATIN CAPITAL LETTER I WITH GRAVE
 898                             *(convertedChar++) = 0x0069;
 899                             *(convertedChar++) = 0x0307;
 900                             *(convertedChar++) = 0x0300;
 901                             return 3;
 902
 903                         case 0x00CD: // LATIN CAPITAL LETTER I WITH ACUTE
 904                             *(convertedChar++) = 0x0069;
 905                             *(convertedChar++) = 0x0307;
 906                             *(convertedChar++) = 0x0301;
 907                             return 3;
 908
 909                         case 0x0128: // LATIN CAPITAL LETTER I WITH TILDE
 910                             *(convertedChar++) = 0x0069;
 911                             *(convertedChar++) = 0x0307;
 912                             *(convertedChar++) = 0x0303;
 913                             return 3;
 914
 915                         default: break;
 916                     }
 917                 }
 918             break;
 919
 920             case TURKISH_LANG_CODE:
 921             case AZERI_LANG_CODE:
 922                 if ((theChar == 0x0049) || (theChar == 0x0131)) { // LATIN CAPITAL LETTER I & LATIN SMALL LETTER DOTLESS I
 923                     *convertedChar = (((ctype == kCFUniCharToLowercase) || (ctype == kCFUniCharCaseFold))  ? ((kCFUniCharCaseMapMoreAbove & flags) ? 0x0069 : 0x0131) : 0x0049);
 924                     return 1;
 925                 } else if ((theChar == 0x0069) || (theChar == 0x0130)) { // LATIN SMALL LETTER I & LATIN CAPITAL LETTER I WITH DOT ABOVE
 926                     *convertedChar = (((ctype == kCFUniCharToLowercase) || (ctype == kCFUniCharCaseFold)) ? 0x0069 : 0x0130);
 927                     return 1;
 928                 } else if (theChar == 0x0307 && (kCFUniCharCaseMapAfter_i & flags)) { // COMBINING DOT ABOVE AFTER_i
 929                     if (ctype == kCFUniCharToLowercase) {
 930                         return 0;
 931                     } else {
 932                         *convertedChar = 0x0307;
 933                         return 1;
 934                     }
 935                 }
 936                 break;
 937
 938             case DUTCH_LANG_CODE:
 939                 if ((theChar == 0x004A) || (theChar == 0x006A)) {
 940                     *convertedChar = (((ctype == kCFUniCharToUppercase) || (ctype == kCFUniCharToTitlecase) || (kCFUniCharCaseMapDutchDigraph & flags)) ? 0x004A  : 0x006A);
 941                     return 1;
 942                 }
 943                 break;
 944
 945             default: break;
 946         }
 947     }
 948 #endif // DO_SPECIAL_CASE_MAPPING
 949
 950     if (NULL == __CFUniCharBitmapDataArray) __CFUniCharLoadBitmapData();
 951
 952     data = __CFUniCharBitmapDataArray + __CFUniCharMapExternalSetToInternalIndex(__CFUniCharMapCompatibilitySetID(ctype + kCFUniCharHasNonSelfLowercaseCharacterSet));
 953
 954     if (planeNo < data->_numPlanes && data->_planes[planeNo] && CFUniCharIsMemberOfBitmap(theChar, data->_planes[planeNo]) && (__CFUniCharCaseMappingTableCounts || __CFUniCharLoadCaseMappingTable())) {
 955         uint32_t value = __CFUniCharGetMappedCase((const __CFUniCharCaseMappings *)__CFUniCharCaseMappingTable[ctype], __CFUniCharCaseMappingTableCounts[ctype], theChar);
 956
 957         if (!value && ctype == kCFUniCharToTitlecase) {
 958             value = __CFUniCharGetMappedCase((const __CFUniCharCaseMappings *)__CFUniCharCaseMappingTable[kCFUniCharToUppercase], __CFUniCharCaseMappingTableCounts[kCFUniCharToUppercase], theChar);
 959             if (value) ctype = kCFUniCharToUppercase;
 960         }
 961
 962         if (value) {
 963             CFIndex count = CFUniCharConvertFlagToCount(value);
 964
 965             if (count == 1) {
 966                 if (value & kCFUniCharNonBmpFlag) {
 967                     if (maxLength > 1) {
 968                         value = (value & 0xFFFFFF) - 0x10000;
 969                         *(convertedChar++) = (UTF16Char)(value >> 10) + 0xD800UL;
 970                         *(convertedChar++) = (UTF16Char)(value & 0x3FF) + 0xDC00UL;
 971                         return 2;
 972                     }
 973                 } else {
 974                     *convertedChar = (UTF16Char)value;
 975                     return 1;
 976                 }
 977             } else if (count < maxLength) {
 978                 const uint32_t *extraMapping = __CFUniCharCaseMappingExtraTable[ctype] + (value & 0xFFFFFF);
 979
 980                 if (value & kCFUniCharNonBmpFlag) {
 981                     CFIndex copiedLen = 0;
 982
 983                     while (count-- > 0) {
 984                         value = *(extraMapping++);
 985                         if (value > 0xFFFF) {
 986                             if (copiedLen + 2 >= maxLength) break;
 987                             value = (value & 0xFFFFFF) - 0x10000;
 988                             convertedChar[copiedLen++] = (UTF16Char)(value >> 10) + 0xD800UL;
 989                             convertedChar[copiedLen++] = (UTF16Char)(value & 0x3FF) + 0xDC00UL;
 990                         } else {
 991                             if (copiedLen + 1 >= maxLength) break;
 992                             convertedChar[copiedLen++] = value;
 993                         }
 994                     }
 995                     if (!count) return copiedLen;
 996                 } else {
 997                     CFIndex idx;
 998
 999                     for (idx = 0;idx < count;idx++) *(convertedChar++) = (UTF16Char)*(extraMapping++);
1000                     return count;
1001                 }
1002             }
1003         }
1004     } else if (ctype == kCFUniCharCaseFold) {
1005         ctype = kCFUniCharToLowercase;
1006         goto caseFoldRetry;
1007     }
1008
1009     if (theChar > 0xFFFF) { // non-BMP
1010         theChar = (theChar & 0xFFFFFF) - 0x10000;
1011         *(convertedChar++) = (UTF16Char)(theChar >> 10) + 0xD800UL;
1012         *(convertedChar++) = (UTF16Char)(theChar & 0x3FF) + 0xDC00UL;
1013         return 2;
1014     } else {
1015         *convertedChar = theChar;
1016         return 1;
1017     }
1018 }
1019
1020 CFIndex CFUniCharMapTo(UniChar theChar, UniChar *convertedChar, CFIndex maxLength, uint16_t ctype, uint32_t flags) {
1021     if (ctype == kCFUniCharCaseFold + 1) { // kCFUniCharDecompose
1022         if (CFUniCharIsDecomposableCharacter(theChar, false)) {
1023             UTF32Char buffer[MAX_DECOMPOSED_LENGTH];
1024             CFIndex usedLength = CFUniCharDecomposeCharacter(theChar, buffer, MAX_DECOMPOSED_LENGTH);
1025             CFIndex idx;
1026
1027             for (idx = 0;idx < usedLength;idx++) *(convertedChar++) = buffer[idx];
1028             return usedLength;
1029         } else {
1030             *convertedChar = theChar;
1031             return 1;
1032         }
1033     } else {
1034         return CFUniCharMapCaseTo(theChar, convertedChar, maxLength, ctype, flags, NULL);
1035     }
1036 }
1037
1038 CF_INLINE bool __CFUniCharIsMoreAbove(UTF16Char *buffer, CFIndex length) {
1039     UTF32Char currentChar;
1040     uint32_t property;
1041
1042     while (length-- > 0) {
1043         currentChar = *(buffer)++;
1044         if (CFUniCharIsSurrogateHighCharacter(currentChar) && (length > 0) && CFUniCharIsSurrogateLowCharacter(*(buffer + 1))) {
1045             currentChar = CFUniCharGetLongCharacterForSurrogatePair(currentChar, *(buffer++));
1046             --length;
1047         }
1048         if (!CFUniCharIsMemberOf(currentChar, kCFUniCharNonBaseCharacterSet)) break;
1049
1050         property = CFUniCharGetCombiningPropertyForCharacter(currentChar, (const uint8_t *)CFUniCharGetUnicodePropertyDataForPlane(kCFUniCharCombiningProperty, (currentChar >> 16) & 0xFF));
1051
1052         if (property == 230) return true; // Above priority
1053     }
1054     return false;
1055 }
1056
1057 CF_INLINE bool __CFUniCharIsAfter_i(UTF16Char *buffer, CFIndex length) {
1058     UTF32Char currentChar = 0;
1059     uint32_t property;
1060     UTF32Char decomposed[MAX_DECOMPOSED_LENGTH];
1061     CFIndex decompLength;
1062     CFIndex idx;
1063
1064     if (length < 1) return 0;
1065
1066     buffer += length;
1067     while (length-- > 1) {
1068         currentChar = *(--buffer);
1069         if (CFUniCharIsSurrogateLowCharacter(currentChar)) {
1070             if ((length > 1) && CFUniCharIsSurrogateHighCharacter(*(buffer - 1))) {
1071                 currentChar = CFUniCharGetLongCharacterForSurrogatePair(*(--buffer), currentChar);
1072                 --length;
1073             } else {
1074                 break;
1075             }
1076         }
1077         if (!CFUniCharIsMemberOf(currentChar, kCFUniCharNonBaseCharacterSet)) break;
1078
1079         property = CFUniCharGetCombiningPropertyForCharacter(currentChar, (const uint8_t *)CFUniCharGetUnicodePropertyDataForPlane(kCFUniCharCombiningProperty, (currentChar >> 16) & 0xFF));
1080
1081         if (property == 230) return false; // Above priority
1082     }
1083     if (length == 0) {
1084         currentChar = *(--buffer);
1085     } else if (CFUniCharIsSurrogateLowCharacter(currentChar) && CFUniCharIsSurrogateHighCharacter(*(--buffer))) {
1086         currentChar = CFUniCharGetLongCharacterForSurrogatePair(*buffer, currentChar);
1087     }
1088
1089     decompLength = CFUniCharDecomposeCharacter(currentChar, decomposed, MAX_DECOMPOSED_LENGTH);
1090     currentChar = *decomposed;
1091
1092
1093     for (idx = 1;idx < decompLength;idx++) {
1094         currentChar = decomposed[idx];
1095         property = CFUniCharGetCombiningPropertyForCharacter(currentChar, (const uint8_t *)CFUniCharGetUnicodePropertyDataForPlane(kCFUniCharCombiningProperty, (currentChar >> 16) & 0xFF));
1096
1097         if (property == 230) return false; // Above priority
1098     }
1099     return true;
1100 }
1101
1102 CF_PRIVATE uint32_t CFUniCharGetConditionalCaseMappingFlags(UTF32Char theChar, UTF16Char *buffer, CFIndex currentIndex, CFIndex length, uint32_t type, const uint8_t *langCode, uint32_t lastFlags) {
1103     if (theChar == 0x03A3) { // GREEK CAPITAL LETTER SIGMA
1104         if ((type == kCFUniCharToLowercase) && (currentIndex > 0)) {
1105             UTF16Char *start = buffer;
1106             UTF16Char *end = buffer + length;
1107             UTF32Char otherChar;
1108
1109             // First check if we're after a cased character
1110             buffer += (currentIndex - 1);
1111             while (start <= buffer) {
1112                 otherChar = *(buffer--);
1113                 if (CFUniCharIsSurrogateLowCharacter(otherChar) && (start <= buffer) && CFUniCharIsSurrogateHighCharacter(*buffer)) {
1114                     otherChar = CFUniCharGetLongCharacterForSurrogatePair(*(buffer--), otherChar);
1115                 }
1116                 if (!CFUniCharIsMemberOf(otherChar, kCFUniCharCaseIgnorableCharacterSet)) {
1117                     if (!CFUniCharIsMemberOf(otherChar, kCFUniCharUppercaseLetterCharacterSet) && !CFUniCharIsMemberOf(otherChar, kCFUniCharLowercaseLetterCharacterSet)) return 0; // Uppercase set contains titlecase
1118                     break;
1119                 }
1120             }
1121
1122             // Next check if we're before a cased character
1123             buffer = start + currentIndex + 1;
1124             while (buffer < end) {
1125                 otherChar = *(buffer++);
1126                 if (CFUniCharIsSurrogateHighCharacter(otherChar) && (buffer < end) && CFUniCharIsSurrogateLowCharacter(*buffer)) {
1127                     otherChar = CFUniCharGetLongCharacterForSurrogatePair(otherChar, *(buffer++));
1128                 }
1129                 if (!CFUniCharIsMemberOf(otherChar, kCFUniCharCaseIgnorableCharacterSet)) {
1130                     if (CFUniCharIsMemberOf(otherChar, kCFUniCharUppercaseLetterCharacterSet) || CFUniCharIsMemberOf(otherChar, kCFUniCharLowercaseLetterCharacterSet)) return 0; // Uppercase set contains titlecase
1131                     break;
1132                 }
1133             }
1134             return kCFUniCharCaseMapFinalSigma;
1135         }
1136     } else if (langCode) {
1137         if (*((const uint16_t *)langCode) == LITHUANIAN_LANG_CODE) {
1138             if ((theChar == 0x0307) && ((kCFUniCharCaseMapAfter_i|kCFUniCharCaseMapMoreAbove) & lastFlags) == (kCFUniCharCaseMapAfter_i|kCFUniCharCaseMapMoreAbove)) {
1139                 return (__CFUniCharIsAfter_i(buffer, currentIndex) ? kCFUniCharCaseMapAfter_i : 0);
1140             } else if (type == kCFUniCharToLowercase) {
1141                 if ((theChar == 0x0049) || (theChar == 0x004A) || (theChar == 0x012E)) {
1142                     ++currentIndex;
1143                     return (__CFUniCharIsMoreAbove(buffer + currentIndex, length - currentIndex) ? kCFUniCharCaseMapMoreAbove : 0);
1144                 }
1145             } else if ((theChar == 'i') || (theChar == 'j')) {
1146                 ++currentIndex;
1147                 return (__CFUniCharIsMoreAbove(buffer + currentIndex, length - currentIndex) ? (kCFUniCharCaseMapAfter_i|kCFUniCharCaseMapMoreAbove) : 0);
1148             }
1149         } else if ((*((const uint16_t *)langCode) == TURKISH_LANG_CODE) || (*((const uint16_t *)langCode) == AZERI_LANG_CODE)) {
1150             if (type == kCFUniCharToLowercase) {
1151                 if (theChar == 0x0307) {
1152                     return (kCFUniCharCaseMapMoreAbove & lastFlags ? kCFUniCharCaseMapAfter_i : 0);
1153                 } else if (theChar == 0x0049) {
1154                     return (((++currentIndex < length) && (buffer[currentIndex] == 0x0307)) ? kCFUniCharCaseMapMoreAbove : 0);
1155                 }
1156             }
1157         } else if (*((const uint16_t *)langCode) == DUTCH_LANG_CODE) {
1158             if (kCFUniCharCaseMapDutchDigraph & lastFlags) {
1159                 return (((theChar == 0x006A) || (theChar == 0x004A)) ? kCFUniCharCaseMapDutchDigraph : 0);
1160             } else {
1161                 if ((type == kCFUniCharToTitlecase) && ((theChar == 0x0069) || (theChar == 0x0049))) {
1162                     return (((++currentIndex < length) && ((buffer[currentIndex] == 0x006A) || (buffer[currentIndex] == 0x004A))) ? kCFUniCharCaseMapDutchDigraph : 0);
1163                 }
1164             }
1165         }
1166
1167         if (kCFUniCharCaseMapGreekTonos & lastFlags) { // still searching for tonos
1168             if (CFUniCharIsMemberOf(theChar, kCFUniCharNonBaseCharacterSet)) {
1169                 return kCFUniCharCaseMapGreekTonos;
1170             }
1171         }
1172         if (((theChar >= 0x0370) && (theChar < 0x0400)) || ((theChar >= 0x1F00) && (theChar < 0x2000))) { // Greek/Coptic & Greek extended ranges
1173             if ((type == kCFUniCharToUppercase) && (CFUniCharIsMemberOf(theChar, kCFUniCharLetterCharacterSet))) return kCFUniCharCaseMapGreekTonos;
1174         }
1175     }
1176     return 0;
1177 }
1178
1179 // Unicode property database
1180 static __CFUniCharBitmapData *__CFUniCharUnicodePropertyTable = NULL;
1181 static int __CFUniCharUnicodePropertyTableCount = 0;
1182
1183 static CFLock_t __CFUniCharPropTableLock = CFLockInit;
1184
1185 #if DEPLOYMENT_TARGET_MACOSX || DEPLOYMENT_TARGET_EMBEDDED || DEPLOYMENT_TARGET_EMBEDDED_MINI || DEPLOYMENT_TARGET_LINUX
1186 #if USE_MACHO_SEGMENT
1187 #define PROP_DB_FILE "__properties"
1188 #else
1189 #define PROP_DB_FILE "/CFUniCharPropertyDatabase.data"
1190 #endif
1191 #elif DEPLOYMENT_TARGET_WINDOWS
1192 #if USE_MACHO_SEGMENT
1193 #define PROP_DB_FILE "__properties"
1194 #else
1195 #define PROP_DB_FILE L"CFUniCharPropertyDatabase.data"
1196 #endif
1197 #else
1198 #error Unknown or unspecified DEPLOYMENT_TARGET
1199 #endif
1200
1201 const void *CFUniCharGetUnicodePropertyDataForPlane(uint32_t propertyType, uint32_t plane) {
1202
1203     __CFLock(&__CFUniCharPropTableLock);
1204
1205     if (NULL == __CFUniCharUnicodePropertyTable) {
1206         __CFUniCharBitmapData *table;
1207         const void *bytes;
1208         const void *bodyBase;
1209         const void *planeBase;
1210         int headerSize;
1211         int idx, count;
1212         int planeIndex, planeCount;
1213         int planeSize;
1214         int64_t fileSize;
1215
1216         if (!__CFUniCharLoadFile(PROP_DB_FILE, &bytes, &fileSize) || !__CFSimpleFileSizeVerification(bytes, fileSize)) {
1217             __CFUnlock(&__CFUniCharPropTableLock);
1218             return NULL;
1219         }
1220
1221 #if defined (__cplusplus)
1222                 bytes = (uint8_t*)bytes + 4; // Skip Unicode version
1223                 headerSize = CFSwapInt32BigToHost(*((uint32_t *)bytes)); bytes = (uint8_t *)bytes + sizeof(uint32_t);
1224 #else
1225                 bytes += 4; // Skip Unicode version
1226                 headerSize = CFSwapInt32BigToHost(*((uint32_t *)bytes)); bytes += sizeof(uint32_t);
1227 #endif
1228
1229         headerSize -= (sizeof(uint32_t) * 2);
1230         bodyBase = (char *)bytes + headerSize;
1231
1232         count = headerSize / sizeof(uint32_t);
1233         __CFUniCharUnicodePropertyTableCount = count;
1234
1235         table = (__CFUniCharBitmapData *)CFAllocatorAllocate(kCFAllocatorSystemDefault, sizeof(__CFUniCharBitmapData) * count, 0);
1236
1237         for (idx = 0;idx < count;idx++) {
1238             planeCount = *((const uint8_t *)bodyBase);
1239             planeBase = (char *)bodyBase + planeCount + (planeCount % 4 ? 4 - (planeCount % 4) : 0);
1240             table[idx]._planes = (const uint8_t **)CFAllocatorAllocate(kCFAllocatorSystemDefault, sizeof(const void *) * planeCount, 0);
1241
1242             for (planeIndex = 0;planeIndex < planeCount;planeIndex++) {
1243                 if ((planeSize = ((const uint8_t *)bodyBase)[planeIndex + 1])) {
1244                     table[idx]._planes[planeIndex] = (const uint8_t *)planeBase;
1245 #if defined (__cplusplus)
1246                                         planeBase = (char*)planeBase + (planeSize * 256);
1247 #else
1248                                         planeBase += (planeSize * 256);
1249 #endif
1250                 } else {
1251                     table[idx]._planes[planeIndex] = NULL;
1252                 }
1253             }
1254
1255             table[idx]._numPlanes = planeCount;
1256 #if defined (__cplusplus)
1257                         bodyBase = (const uint8_t *)bodyBase + (CFSwapInt32BigToHost(*(uint32_t *)bytes));
1258                         ((uint32_t *&)bytes) ++;
1259 #else
1260                         bodyBase += (CFSwapInt32BigToHost(*((uint32_t *)bytes++)));
1261 #endif
1262         }
1263
1264         __CFUniCharUnicodePropertyTable = table;
1265     }
1266
1267     __CFUnlock(&__CFUniCharPropTableLock);
1268
1269     return (plane < __CFUniCharUnicodePropertyTable[propertyType]._numPlanes ? __CFUniCharUnicodePropertyTable[propertyType]._planes[plane] : NULL);
1270 }
1271
1272 CF_PRIVATE uint32_t CFUniCharGetNumberOfPlanesForUnicodePropertyData(uint32_t propertyType) {
1273     (void)CFUniCharGetUnicodePropertyDataForPlane(propertyType, 0);
1274     return __CFUniCharUnicodePropertyTable[propertyType]._numPlanes;
1275 }
1276
1277 CF_PRIVATE uint32_t CFUniCharGetUnicodeProperty(UTF32Char character, uint32_t propertyType) {
1278     if (propertyType == kCFUniCharCombiningProperty) {
1279         return CFUniCharGetCombiningPropertyForCharacter(character, (const uint8_t *)CFUniCharGetUnicodePropertyDataForPlane(propertyType, (character >> 16) & 0xFF));
1280     } else if (propertyType == kCFUniCharBidiProperty) {
1281         return CFUniCharGetBidiPropertyForCharacter(character, (const uint8_t *)CFUniCharGetUnicodePropertyDataForPlane(propertyType, (character >> 16) & 0xFF));
1282     } else {
1283         return 0;
1284     }
1285 }
1286
1287
1288
1289 /*
1290     The UTF8 conversion in the following function is derived from ConvertUTF.c
1291 */
1292 /*
1293  * Copyright 2001 Unicode, Inc.
1294  *
1295  * Disclaimer
1296  *
1297  * This source code is provided as is by Unicode, Inc. No claims are
1298  * made as to fitness for any particular purpose. No warranties of any
1299  * kind are expressed or implied. The recipient agrees to determine
1300  * applicability of information provided. If this file has been
1301  * purchased on magnetic or optical media from Unicode, Inc., the
1302  * sole remedy for any claim will be exchange of defective media
1303  * within 90 days of receipt.
1304  *
1305  * Limitations on Rights to Redistribute This Code
1306  *
1307  * Unicode, Inc. hereby grants the right to freely use the information
1308  * supplied in this file in the creation of products supporting the
1309  * Unicode Standard, and to make copies of this file in any form
1310  * for internal or external distribution as long as this notice
1311  * remains attached.
1312  */
1313 #define UNI_REPLACEMENT_CHAR (0x0000FFFDUL)
1314
1315 bool CFUniCharFillDestinationBuffer(const UTF32Char *src, CFIndex srcLength, void **dst, CFIndex dstLength, CFIndex *filledLength, uint32_t dstFormat) {
1316     UTF32Char currentChar;
1317     CFIndex usedLength = *filledLength;
1318
1319     if (dstFormat == kCFUniCharUTF16Format) {
1320         UTF16Char *dstBuffer = (UTF16Char *)*dst;
1321
1322         while (srcLength-- > 0) {
1323             currentChar = *(src++);
1324
1325             if (currentChar > 0xFFFF) { // Non-BMP
1326                 usedLength += 2;
1327                 if (dstLength) {
1328                     if (usedLength > dstLength) return false;
1329                     currentChar -= 0x10000;
1330                     *(dstBuffer++) = (UTF16Char)((currentChar >> 10) + 0xD800UL);
1331                     *(dstBuffer++) = (UTF16Char)((currentChar & 0x3FF) + 0xDC00UL);
1332                 }
1333             } else {
1334                 ++usedLength;
1335                 if (dstLength) {
1336                     if (usedLength > dstLength) return false;
1337                     *(dstBuffer++) = (UTF16Char)currentChar;
1338                 }
1339             }
1340         }
1341
1342         *dst = dstBuffer;
1343     } else if (dstFormat == kCFUniCharUTF8Format) {
1344         uint8_t *dstBuffer = (uint8_t *)*dst;
1345         uint16_t bytesToWrite = 0;
1346         const UTF32Char byteMask = 0xBF;
1347         const UTF32Char byteMark = 0x80;
1348         static const uint8_t firstByteMark[7] = { 0x00, 0x00, 0xC0, 0xE0, 0xF0, 0xF8, 0xFC };
1349
1350         while (srcLength-- > 0) {
1351             currentChar = *(src++);
1352
1353             /* Figure out how many bytes the result will require */
1354             if (currentChar < (UTF32Char)0x80) {
1355                 bytesToWrite = 1;
1356             } else if (currentChar < (UTF32Char)0x800) {
1357                 bytesToWrite = 2;
1358             } else if (currentChar < (UTF32Char)0x10000) {
1359                 bytesToWrite = 3;
1360             } else if (currentChar < (UTF32Char)0x200000) {
1361                 bytesToWrite = 4;
1362             } else {
1363                 bytesToWrite = 2;
1364                 currentChar = UNI_REPLACEMENT_CHAR;
1365             }
1366
1367             usedLength += bytesToWrite;
1368
1369             if (dstLength) {
1370                 if (usedLength > dstLength) return false;
1371
1372                 dstBuffer += bytesToWrite;
1373                 switch (bytesToWrite) { /* note: everything falls through. */
1374                     case 4:     *--dstBuffer = (currentChar | byteMark) & byteMask; currentChar >>= 6;
1375                     case 3:     *--dstBuffer = (currentChar | byteMark) & byteMask; currentChar >>= 6;
1376                     case 2:     *--dstBuffer = (currentChar | byteMark) & byteMask; currentChar >>= 6;
1377                     case 1:     *--dstBuffer =  currentChar | firstByteMark[bytesToWrite];
1378                 }
1379                 dstBuffer += bytesToWrite;
1380             }
1381         }
1382
1383         *dst = dstBuffer;
1384     } else {
1385         UTF32Char *dstBuffer = (UTF32Char *)*dst;
1386
1387         while (srcLength-- > 0) {
1388             currentChar = *(src++);
1389
1390             ++usedLength;
1391             if (dstLength) {
1392                 if (usedLength > dstLength) return false;
1393                 *(dstBuffer++) = currentChar;
1394             }
1395         }
1396
1397         *dst = dstBuffer;
1398     }
1399
1400     *filledLength = usedLength;
1401
1402     return true;
1403 }
1404
1405 #if DEPLOYMENT_TARGET_WINDOWS
1406 void __CFUniCharCleanup(void)
1407 {
1408     int idx;
1409
1410     // cleanup memory allocated by __CFUniCharLoadBitmapData()
1411     __CFLock(&__CFUniCharBitmapLock);
1412
1413     if (__CFUniCharBitmapDataArray != NULL) {
1414         for (idx = 0; idx < (int)__CFUniCharNumberOfBitmaps; idx++) {
1415             CFAllocatorDeallocate(kCFAllocatorSystemDefault, __CFUniCharBitmapDataArray[idx]._planes);
1416             __CFUniCharBitmapDataArray[idx]._planes = NULL;
1417         }
1418
1419         CFAllocatorDeallocate(kCFAllocatorSystemDefault, __CFUniCharBitmapDataArray);
1420         __CFUniCharBitmapDataArray = NULL;
1421         __CFUniCharNumberOfBitmaps = 0;
1422     }
1423
1424     __CFUnlock(&__CFUniCharBitmapLock);
1425
1426     // cleanup memory allocated by CFUniCharGetMappingData()
1427     __CFLock(&__CFUniCharMappingTableLock);
1428
1429     if (__CFUniCharMappingTables != NULL) {
1430         CFAllocatorDeallocate(kCFAllocatorSystemDefault, __CFUniCharMappingTables);
1431         __CFUniCharMappingTables = NULL;
1432     }
1433
1434     // cleanup memory allocated by __CFUniCharLoadCaseMappingTable()
1435     if (__CFUniCharCaseMappingTableCounts != NULL) {
1436         CFAllocatorDeallocate(kCFAllocatorSystemDefault, __CFUniCharCaseMappingTableCounts);
1437         __CFUniCharCaseMappingTableCounts = NULL;
1438
1439         __CFUniCharCaseMappingTable = NULL;
1440         __CFUniCharCaseMappingExtraTable = NULL;
1441     }
1442
1443     __CFUnlock(&__CFUniCharMappingTableLock);
1444
1445     // cleanup memory allocated by CFUniCharGetUnicodePropertyDataForPlane()
1446     __CFLock(&__CFUniCharPropTableLock);
1447
1448     if (__CFUniCharUnicodePropertyTable != NULL) {
1449         for (idx = 0; idx < __CFUniCharUnicodePropertyTableCount; idx++) {
1450             CFAllocatorDeallocate(kCFAllocatorSystemDefault, __CFUniCharUnicodePropertyTable[idx]._planes);
1451             __CFUniCharUnicodePropertyTable[idx]._planes = NULL;
1452         }
1453
1454         CFAllocatorDeallocate(kCFAllocatorSystemDefault, __CFUniCharUnicodePropertyTable);
1455         __CFUniCharUnicodePropertyTable = NULL;
1456         __CFUniCharUnicodePropertyTableCount = 0;
1457     }
1458
1459     __CFUnlock(&__CFUniCharPropTableLock);
1460 }
1461 #endif
1462
1463 #undef USE_MACHO_SEGMENT
1464