CFUniChar.c

   1 /*
   2  * Copyright (c) 2012 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) 2001-2011, 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_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
  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 static const void *__CFGetSectDataPtr(const char *segname, const char *sectname, uint64_t *sizep) {
  80     uint32_t idx, cnt = _dyld_image_count();
  81     for (idx = 0; idx < cnt; idx++) {
  82        void *mh = (void *)_dyld_get_image_header(idx);
  83        if (mh != &_mh_dylib_header) continue;
  84 #if __LP64__
  85        const struct section_64 *sect = getsectbynamefromheader_64((struct mach_header_64 *)mh, segname, sectname);
  86 #else
  87        const struct section *sect = getsectbynamefromheader((struct mach_header *)mh, segname, sectname);
  88 #endif
  89        if (!sect) break;
  90        if (sizep) *sizep = (uint64_t)sect->size;
  91        return (char *)sect->addr + _dyld_get_image_vmaddr_slide(idx);
  92     }
  93     if (sizep) *sizep = 0ULL;
  94     return NULL;
  95 }
  96 #endif
  97
  98 #if !USE_MACHO_SEGMENT
  99
 100 // Memory map the file
 101
 102 #if DEPLOYMENT_TARGET_MACOSX || DEPLOYMENT_TARGET_EMBEDDED || DEPLOYMENT_TARGET_LINUX
 103 CF_INLINE void __CFUniCharCharacterSetPath(char *cpath) {
 104 #elif DEPLOYMENT_TARGET_WINDOWS
 105 CF_INLINE void __CFUniCharCharacterSetPath(wchar_t *wpath) {
 106 #else
 107 #error Unknown or unspecified DEPLOYMENT_TARGET
 108 #endif
 109 #if DEPLOYMENT_TARGET_MACOSX || DEPLOYMENT_TARGET_EMBEDDED
 110     strlcpy(cpath, __kCFCharacterSetDir, MAXPATHLEN);
 111 #elif DEPLOYMENT_TARGET_LINUX
 112     strlcpy(cpath, __kCFCharacterSetDir, MAXPATHLEN);
 113 #elif DEPLOYMENT_TARGET_WINDOWS
 114     wchar_t frameworkPath[MAXPATHLEN];
 115     _CFGetFrameworkPath(frameworkPath, MAXPATHLEN);
 116     wcsncpy(wpath, frameworkPath, MAXPATHLEN);
 117     wcsncat(wpath, L"\\CoreFoundation.resources\\", MAXPATHLEN - wcslen(wpath));
 118 #else
 119     strlcpy(cpath, __kCFCharacterSetDir, MAXPATHLEN);
 120     strlcat(cpath, "/CharacterSets/", MAXPATHLEN);
 121 #endif
 122 }
 123
 124 #if DEPLOYMENT_TARGET_WINDOWS
 125 #define MAX_BITMAP_STATE 512
 126 //
 127 //  If a string is placed into this array, then it has been previously
 128 //  determined that the bitmap-file cannot be found.  Thus, we make
 129 //  the assumption it won't be there in future calls and we avoid
 130 //  hitting the disk un-necessarily.  This assumption isn't 100%
 131 //  correct, as bitmap-files can be added.  We would have to re-start
 132 //  the application in order to pick-up the new bitmap info.
 133 //
 134 //  We should probably re-visit this.
 135 //
 136 static wchar_t *mappedBitmapState[MAX_BITMAP_STATE];
 137 static int __nNumStateEntries = -1;
 138 CRITICAL_SECTION __bitmapStateLock = {0};
 139
 140 bool __GetBitmapStateForName(const wchar_t *bitmapName) {
 141     if (NULL == __bitmapStateLock.DebugInfo)
 142         InitializeCriticalSection(&__bitmapStateLock);
 143     EnterCriticalSection(&__bitmapStateLock);
 144     if (__nNumStateEntries >= 0) {
 145         for (int i = 0; i < __nNumStateEntries; i++) {
 146             if (wcscmp(mappedBitmapState[i], bitmapName) == 0) {
 147                 LeaveCriticalSection(&__bitmapStateLock);
 148                 return true;
 149             }
 150         }
 151     }
 152     LeaveCriticalSection(&__bitmapStateLock);
 153     return false;
 154 }
 155 void __AddBitmapStateForName(const wchar_t *bitmapName) {
 156     if (NULL == __bitmapStateLock.DebugInfo)
 157         InitializeCriticalSection(&__bitmapStateLock);
 158     EnterCriticalSection(&__bitmapStateLock);
 159     __nNumStateEntries++;
 160     mappedBitmapState[__nNumStateEntries] = (wchar_t *)malloc((lstrlenW(bitmapName)+1) * sizeof(wchar_t));
 161     lstrcpyW(mappedBitmapState[__nNumStateEntries], bitmapName);
 162     LeaveCriticalSection(&__bitmapStateLock);
 163 }
 164 #endif
 165
 166 #if DEPLOYMENT_TARGET_MACOSX || DEPLOYMENT_TARGET_EMBEDDED || DEPLOYMENT_TARGET_LINUX
 167 static bool __CFUniCharLoadBytesFromFile(const char *fileName, const void **bytes, int64_t *fileSize) {
 168 #elif DEPLOYMENT_TARGET_WINDOWS
 169 static bool __CFUniCharLoadBytesFromFile(const wchar_t *fileName, const void **bytes, int64_t *fileSize) {
 170 #else
 171 #error Unknown or unspecified DEPLOYMENT_TARGET
 172 #endif
 173 #if DEPLOYMENT_TARGET_WINDOWS
 174     HANDLE bitmapFileHandle = NULL;
 175     HANDLE mappingHandle = NULL;
 176
 177     if (__GetBitmapStateForName(fileName)) {
 178         // The fileName has been tried in the past, so just return false
 179         // and move on.
 180         *bytes = NULL;
 181         return false;
 182     }
 183     mappingHandle = OpenFileMappingW(FILE_MAP_READ, TRUE, fileName);
 184     if (NULL == mappingHandle) {
 185         if ((bitmapFileHandle = CreateFileW(fileName, GENERIC_READ, FILE_SHARE_READ, NULL, OPEN_EXISTING, FILE_ATTRIBUTE_NORMAL, NULL)) == INVALID_HANDLE_VALUE) {
 186             // We tried to get the bitmap file for mapping, but it's not there.  Add to list of non-existant bitmap-files so
 187             // we don't have to try this again in the future.
 188             __AddBitmapStateForName(fileName);
 189             return false;
 190         }
 191         mappingHandle = CreateFileMapping(bitmapFileHandle, NULL, PAGE_READONLY, 0, 0, NULL);
 192         CloseHandle(bitmapFileHandle);
 193         if (!mappingHandle) return false;
 194     }
 195
 196     *bytes = MapViewOfFileEx(mappingHandle, FILE_MAP_READ, 0, 0, 0, 0);
 197
 198     if (NULL != fileSize) {
 199         MEMORY_BASIC_INFORMATION memoryInfo;
 200
 201         if (0 == VirtualQueryEx(mappingHandle, *bytes, &memoryInfo, sizeof(memoryInfo))) {
 202             *fileSize = 0; // This indicates no checking. Is it right ?
 203         } else {
 204             *fileSize = memoryInfo.RegionSize;
 205         }
 206     }
 207
 208     CloseHandle(mappingHandle);
 209
 210     return (*bytes ? true : false);
 211 #else
 212     struct stat statBuf;
 213     int fd = -1;
 214
 215     if ((fd = open(fileName, O_RDONLY, 0)) < 0) {
 216         return false;
 217     }
 218     if (fstat(fd, &statBuf) < 0 || (*bytes = mmap(0, statBuf.st_size, PROT_READ, MAP_PRIVATE, fd, 0)) == (void *)-1) {
 219         close(fd);
 220         return false;
 221     }
 222     close(fd);
 223
 224     if (NULL != fileSize) *fileSize = statBuf.st_size;
 225
 226     return true;
 227 #endif
 228 }
 229
 230 #endif // USE_MACHO_SEGMENT
 231
 232 #if DEPLOYMENT_TARGET_MACOSX || DEPLOYMENT_TARGET_EMBEDDED || DEPLOYMENT_TARGET_LINUX
 233 static bool __CFUniCharLoadFile(const char *bitmapName, const void **bytes, int64_t *fileSize) {
 234 #elif DEPLOYMENT_TARGET_WINDOWS
 235 static bool __CFUniCharLoadFile(const wchar_t *bitmapName, const void **bytes, int64_t *fileSize) {
 236 #else
 237 #error Unknown or unspecified DEPLOYMENT_TARGET
 238 #endif
 239 #if USE_MACHO_SEGMENT
 240         *bytes = __CFGetSectDataPtr("__UNICODE", bitmapName, NULL);
 241
 242     if (NULL != fileSize) *fileSize = 0;
 243
 244     return *bytes ? true : false;
 245 #else
 246 #if DEPLOYMENT_TARGET_MACOSX || DEPLOYMENT_TARGET_EMBEDDED || DEPLOYMENT_TARGET_LINUX
 247     char cpath[MAXPATHLEN];
 248     __CFUniCharCharacterSetPath(cpath);
 249     strlcat(cpath, bitmapName, MAXPATHLEN);
 250     return __CFUniCharLoadBytesFromFile(cpath, bytes, fileSize);
 251 #elif DEPLOYMENT_TARGET_WINDOWS
 252     wchar_t wpath[MAXPATHLEN];
 253     __CFUniCharCharacterSetPath(wpath);
 254     wcsncat(wpath, bitmapName, MAXPATHLEN);
 255     return __CFUniCharLoadBytesFromFile(wpath, bytes, fileSize);
 256 #else
 257 #error Unknown or unspecified DEPLOYMENT_TARGET
 258 #endif
 259 #endif
 260 }
 261
 262 // Bitmap functions
 263 CF_INLINE bool isControl(UTF32Char theChar, uint16_t charset, const void *data) { // ISO Control
 264     return (((theChar <= 0x001F) || (theChar >= 0x007F && theChar <= 0x009F)) ? true : false);
 265 }
 266
 267 CF_INLINE bool isWhitespace(UTF32Char theChar, uint16_t charset, const void *data) { // Space
 268     return (((theChar == 0x0020) || (theChar == 0x0009) || (theChar == 0x00A0) || (theChar == 0x1680) || (theChar >= 0x2000 && theChar <= 0x200B) || (theChar == 0x202F) || (theChar == 0x205F) || (theChar == 0x3000)) ? true : false);
 269 }
 270
 271 CF_INLINE bool isNewline(UTF32Char theChar, uint16_t charset, const void *data) { // White space
 272     return (((theChar >= 0x000A && theChar <= 0x000D) || (theChar == 0x0085) || (theChar == 0x2028) || (theChar == 0x2029)) ? true : false);
 273 }
 274
 275 CF_INLINE bool isWhitespaceAndNewline(UTF32Char theChar, uint16_t charset, const void *data) { // White space
 276     return ((isWhitespace(theChar, charset, data) || isNewline(theChar, charset, data)) ? true : false);
 277 }
 278
 279 #if USE_MACHO_SEGMENT
 280 CF_INLINE bool __CFSimpleFileSizeVerification(const void *bytes, int64_t fileSize) { return true; }
 281 #elif 1
 282 // <rdar://problem/8961744> __CFSimpleFileSizeVerification is broken
 283 static bool __CFSimpleFileSizeVerification(const void *bytes, int64_t fileSize) { return true; }
 284 #else
 285 static bool __CFSimpleFileSizeVerification(const void *bytes, int64_t fileSize) {
 286     bool result = true;
 287
 288     if (fileSize > 0) {
 289         if ((sizeof(uint32_t) * 2) > fileSize) {
 290             result = false;
 291         } else {
 292             uint32_t headerSize = CFSwapInt32BigToHost(*((uint32_t *)((char *)bytes + 4)));
 293
 294             if ((headerSize < (sizeof(uint32_t) * 4)) || (headerSize > fileSize)) {
 295                 result = false;
 296             } else {
 297                 const uint32_t *lastElement = (uint32_t *)(((uint8_t *)bytes) + headerSize) - 2;
 298
 299                 if ((headerSize + CFSwapInt32BigToHost(lastElement[0]) + CFSwapInt32BigToHost(lastElement[1])) > headerSize) result = false;
 300             }
 301         }
 302     }
 303
 304     if (!result) CFLog(kCFLogLevelCritical, CFSTR("File size verification for Unicode database file failed."));
 305
 306     return result;
 307 }
 308 #endif // USE_MACHO_SEGMENT
 309
 310 typedef struct {
 311     uint32_t _numPlanes;
 312     const uint8_t **_planes;
 313 } __CFUniCharBitmapData;
 314
 315 static char __CFUniCharUnicodeVersionString[8] = {0, 0, 0, 0, 0, 0, 0, 0};
 316
 317 static uint32_t __CFUniCharNumberOfBitmaps = 0;
 318 static __CFUniCharBitmapData *__CFUniCharBitmapDataArray = NULL;
 319
 320 static CFSpinLock_t __CFUniCharBitmapLock = CFSpinLockInit;
 321
 322 #if DEPLOYMENT_TARGET_MACOSX || DEPLOYMENT_TARGET_EMBEDDED || DEPLOYMENT_TARGET_LINUX
 323 #if !defined(CF_UNICHAR_BITMAP_FILE)
 324 #if USE_MACHO_SEGMENT
 325 #define CF_UNICHAR_BITMAP_FILE "__csbitmaps"
 326 #else
 327 #define CF_UNICHAR_BITMAP_FILE "/CFCharacterSetBitmaps.bitmap"
 328 #endif
 329 #endif
 330 #elif DEPLOYMENT_TARGET_WINDOWS
 331 #if !defined(CF_UNICHAR_BITMAP_FILE)
 332 #define CF_UNICHAR_BITMAP_FILE L"CFCharacterSetBitmaps.bitmap"
 333 #endif
 334 #else
 335 #error Unknown or unspecified DEPLOYMENT_TARGET
 336 #endif
 337
 338 static bool __CFUniCharLoadBitmapData(void) {
 339     __CFUniCharBitmapData *array;
 340     uint32_t headerSize;
 341     uint32_t bitmapSize;
 342     int numPlanes;
 343     uint8_t currentPlane;
 344     const void *bytes;
 345     const void *bitmapBase;
 346     const void *bitmap;
 347     int idx, bitmapIndex;
 348     int64_t fileSize;
 349
 350     __CFSpinLock(&__CFUniCharBitmapLock);
 351
 352     if (__CFUniCharBitmapDataArray || !__CFUniCharLoadFile(CF_UNICHAR_BITMAP_FILE, &bytes, &fileSize) || !__CFSimpleFileSizeVerification(bytes, fileSize)) {
 353         __CFSpinUnlock(&__CFUniCharBitmapLock);
 354         return false;
 355     }
 356
 357     for (idx = 0;idx < 4 && ((const uint8_t *)bytes)[idx];idx++) {
 358         __CFUniCharUnicodeVersionString[idx * 2] = ((const uint8_t *)bytes)[idx];
 359         __CFUniCharUnicodeVersionString[idx * 2 + 1] = '.';
 360     }
 361     __CFUniCharUnicodeVersionString[(idx < 4 ? idx * 2 - 1 : 7)] = '\0';
 362
 363     headerSize = CFSwapInt32BigToHost(*((uint32_t *)((char *)bytes + 4)));
 364
 365     bitmapBase = (uint8_t *)bytes + headerSize;
 366     bytes = (uint8_t *)bytes + (sizeof(uint32_t) * 2);
 367     headerSize -= (sizeof(uint32_t) * 2);
 368
 369     __CFUniCharNumberOfBitmaps = headerSize / (sizeof(uint32_t) * 2);
 370
 371     array = (__CFUniCharBitmapData *)CFAllocatorAllocate(kCFAllocatorSystemDefault, sizeof(__CFUniCharBitmapData) * __CFUniCharNumberOfBitmaps, 0);
 372
 373     for (idx = 0;idx < (int)__CFUniCharNumberOfBitmaps;idx++) {
 374         bitmap = (uint8_t *)bitmapBase + CFSwapInt32BigToHost(*((uint32_t *)bytes)); bytes = (uint8_t *)bytes + sizeof(uint32_t);
 375         bitmapSize = CFSwapInt32BigToHost(*((uint32_t *)bytes)); bytes = (uint8_t *)bytes + sizeof(uint32_t);
 376
 377         numPlanes = bitmapSize / (8 * 1024);
 378         numPlanes = *(const uint8_t *)((char *)bitmap + (((numPlanes - 1) * ((8 * 1024) + 1)) - 1)) + 1;
 379         array[idx]._planes = (const uint8_t **)CFAllocatorAllocate(kCFAllocatorSystemDefault, sizeof(const void *) * numPlanes, 0);
 380         array[idx]._numPlanes = numPlanes;
 381
 382         currentPlane = 0;
 383         for (bitmapIndex = 0;bitmapIndex < numPlanes;bitmapIndex++) {
 384             if (bitmapIndex == currentPlane) {
 385                 array[idx]._planes[bitmapIndex] = (const uint8_t *)bitmap;
 386                 bitmap = (uint8_t *)bitmap + (8 * 1024);
 387 #if defined (__cplusplus)
 388                                 currentPlane = *(((const uint8_t*&)bitmap)++);
 389 #else
 390                                 currentPlane = *((const uint8_t *)bitmap++);
 391 #endif
 392
 393             } else {
 394                 array[idx]._planes[bitmapIndex] = NULL;
 395             }
 396         }
 397     }
 398
 399     __CFUniCharBitmapDataArray = array;
 400
 401     __CFSpinUnlock(&__CFUniCharBitmapLock);
 402
 403     return true;
 404 }
 405
 406 __private_extern__ const char *__CFUniCharGetUnicodeVersionString(void) {
 407     if (NULL == __CFUniCharBitmapDataArray) __CFUniCharLoadBitmapData();
 408     return __CFUniCharUnicodeVersionString;
 409 }
 410
 411 bool CFUniCharIsMemberOf(UTF32Char theChar, uint32_t charset) {
 412     charset = __CFUniCharMapCompatibilitySetID(charset);
 413
 414     switch (charset) {
 415         case kCFUniCharWhitespaceCharacterSet:
 416             return isWhitespace(theChar, charset, NULL);
 417
 418         case kCFUniCharWhitespaceAndNewlineCharacterSet:
 419             return isWhitespaceAndNewline(theChar, charset, NULL);
 420
 421         case kCFUniCharNewlineCharacterSet:
 422             return isNewline(theChar, charset, NULL);
 423
 424         default: {
 425             uint32_t tableIndex = __CFUniCharMapExternalSetToInternalIndex(charset);
 426
 427             if (NULL == __CFUniCharBitmapDataArray) __CFUniCharLoadBitmapData();
 428
 429             if (tableIndex < __CFUniCharNumberOfBitmaps) {
 430                 __CFUniCharBitmapData *data = __CFUniCharBitmapDataArray + tableIndex;
 431                 uint8_t planeNo = (theChar >> 16) & 0xFF;
 432
 433                 // The bitmap data for kCFUniCharIllegalCharacterSet is actually LEGAL set less Plane 14 ~ 16
 434                 if (charset == kCFUniCharIllegalCharacterSet) {
 435                     if (planeNo == 0x0E) { // Plane 14
 436                         theChar &= 0xFF;
 437                         return (((theChar == 0x01) || ((theChar > 0x1F) && (theChar < 0x80))) ? false : true);
 438                     } else if (planeNo == 0x0F || planeNo == 0x10) { // Plane 15 & 16
 439                         return ((theChar & 0xFF) > 0xFFFD ? true : false);
 440                     } else {
 441                         return (planeNo < data->_numPlanes && data->_planes[planeNo] ? !CFUniCharIsMemberOfBitmap(theChar, data->_planes[planeNo]) : true);
 442                     }
 443                 } else if (charset == kCFUniCharControlAndFormatterCharacterSet) {
 444                     if (planeNo == 0x0E) { // Plane 14
 445                         theChar &= 0xFF;
 446                         return (((theChar == 0x01) || ((theChar > 0x1F) && (theChar < 0x80))) ? true : false);
 447                     } else {
 448                         return (planeNo < data->_numPlanes && data->_planes[planeNo] ? CFUniCharIsMemberOfBitmap(theChar, data->_planes[planeNo]) : false);
 449                     }
 450                 } else {
 451                     return (planeNo < data->_numPlanes && data->_planes[planeNo] ? CFUniCharIsMemberOfBitmap(theChar, data->_planes[planeNo]) : false);
 452                 }
 453             }
 454             return false;
 455         }
 456     }
 457 }
 458
 459 const uint8_t *CFUniCharGetBitmapPtrForPlane(uint32_t charset, uint32_t plane) {
 460     if (NULL == __CFUniCharBitmapDataArray) __CFUniCharLoadBitmapData();
 461
 462     charset = __CFUniCharMapCompatibilitySetID(charset);
 463
 464     if ((charset > kCFUniCharWhitespaceAndNewlineCharacterSet) && (charset != kCFUniCharIllegalCharacterSet) && (charset != kCFUniCharNewlineCharacterSet)) {
 465         uint32_t tableIndex = __CFUniCharMapExternalSetToInternalIndex(charset);
 466
 467         if (tableIndex < __CFUniCharNumberOfBitmaps) {
 468             __CFUniCharBitmapData *data = __CFUniCharBitmapDataArray + tableIndex;
 469
 470             return (plane < data->_numPlanes ? data->_planes[plane] : NULL);
 471         }
 472     }
 473     return NULL;
 474 }
 475
 476 __private_extern__ uint8_t CFUniCharGetBitmapForPlane(uint32_t charset, uint32_t plane, void *bitmap, bool isInverted) {
 477     const uint8_t *src = CFUniCharGetBitmapPtrForPlane(charset, plane);
 478     int numBytes = (8 * 1024);
 479
 480     if (src) {
 481         if (isInverted) {
 482 #if defined (__cplusplus)
 483                         while (numBytes-- > 0) *(((uint8_t *&)bitmap)++) = ~(*(src++));
 484 #else
 485                         while (numBytes-- > 0) *((uint8_t *)bitmap++) = ~(*(src++));
 486 #endif
 487         } else {
 488 #if defined (__cplusplus)
 489             while (numBytes-- > 0) *(((uint8_t *&)bitmap)++) = *(src++);
 490 #else
 491                         while (numBytes-- > 0) *((uint8_t *)bitmap++) = *(src++);
 492 #endif
 493         }
 494         return kCFUniCharBitmapFilled;
 495     } else if (charset == kCFUniCharIllegalCharacterSet) {
 496         __CFUniCharBitmapData *data = __CFUniCharBitmapDataArray + __CFUniCharMapExternalSetToInternalIndex(__CFUniCharMapCompatibilitySetID(charset));
 497
 498         if (plane < data->_numPlanes && (src = data->_planes[plane])) {
 499             if (isInverted) {
 500 #if defined (__cplusplus)
 501                                 while (numBytes-- > 0) *(((uint8_t *&)bitmap)++) = *(src++);
 502 #else
 503                                 while (numBytes-- > 0) *((uint8_t *)bitmap++) = *(src++);
 504 #endif
 505             } else {
 506 #if defined (__cplusplus)
 507                 while (numBytes-- > 0) *(((uint8_t *&)bitmap)++) = ~(*(src++));
 508 #else
 509                                 while (numBytes-- > 0) *((uint8_t *)bitmap++) = ~(*(src++));
 510 #endif
 511             }
 512             return kCFUniCharBitmapFilled;
 513         } else if (plane == 0x0E) { // Plane 14
 514             int idx;
 515             uint8_t asciiRange = (isInverted ? (uint8_t)0xFF : (uint8_t)0);
 516             uint8_t otherRange = (isInverted ? (uint8_t)0 : (uint8_t)0xFF);
 517
 518 #if defined (__cplusplus)
 519                         *(((uint8_t *&)bitmap)++) = 0x02; // UE0001 LANGUAGE TAG
 520 #else
 521                         *((uint8_t *)bitmap++) = 0x02; // UE0001 LANGUAGE TAG
 522 #endif
 523             for (idx = 1;idx < numBytes;idx++) {
 524 #if defined (__cplusplus)
 525                                 *(((uint8_t *&)bitmap)++) = ((idx >= (0x20 / 8) && (idx < (0x80 / 8))) ? asciiRange : otherRange);
 526 #else
 527                                 *((uint8_t *)bitmap++) = ((idx >= (0x20 / 8) && (idx < (0x80 / 8))) ? asciiRange : otherRange);
 528 #endif
 529             }
 530             return kCFUniCharBitmapFilled;
 531         } else if (plane == 0x0F || plane == 0x10) { // Plane 15 & 16
 532             uint32_t value = (isInverted ? ~0 : 0);
 533             numBytes /= 4; // for 32bit
 534
 535             while (numBytes-- > 0) {
 536                 *((uint32_t *)bitmap) = value;
 537 #if defined (__cplusplus)
 538                                 bitmap = (uint8_t *)bitmap + sizeof(uint32_t);
 539 #else
 540                                 bitmap += sizeof(uint32_t);
 541 #endif
 542             }
 543             *(((uint8_t *)bitmap) - 5) = (isInverted ? 0x3F : 0xC0); // 0xFFFE & 0xFFFF
 544             return kCFUniCharBitmapFilled;
 545         }
 546         return (isInverted ? kCFUniCharBitmapEmpty : kCFUniCharBitmapAll);
 547     } else if ((charset < kCFUniCharDecimalDigitCharacterSet) || (charset == kCFUniCharNewlineCharacterSet)) {
 548         if (plane) return (isInverted ? kCFUniCharBitmapAll : kCFUniCharBitmapEmpty);
 549
 550         uint8_t *bitmapBase = (uint8_t *)bitmap;
 551         CFIndex idx;
 552         uint8_t nonFillValue = (isInverted ? (uint8_t)0xFF : (uint8_t)0);
 553
 554 #if defined (__cplusplus)
 555                     while (numBytes-- > 0) *(((uint8_t *&)bitmap)++) = nonFillValue;
 556 #else
 557                     while (numBytes-- > 0) *((uint8_t *)bitmap++) = nonFillValue;
 558 #endif
 559
 560         if ((charset == kCFUniCharWhitespaceAndNewlineCharacterSet) || (charset == kCFUniCharNewlineCharacterSet)) {
 561             const UniChar newlines[] = {0x000A, 0x000B, 0x000C, 0x000D, 0x0085, 0x2028, 0x2029};
 562
 563             for (idx = 0;idx < (int)(sizeof(newlines) / sizeof(*newlines)); idx++) {
 564                 if (isInverted) {
 565                     CFUniCharRemoveCharacterFromBitmap(newlines[idx], bitmapBase);
 566                 } else {
 567                     CFUniCharAddCharacterToBitmap(newlines[idx], bitmapBase);
 568                 }
 569             }
 570
 571             if (charset == kCFUniCharNewlineCharacterSet) return kCFUniCharBitmapFilled;
 572         }
 573
 574         if (isInverted) {
 575             CFUniCharRemoveCharacterFromBitmap(0x0009, bitmapBase);
 576             CFUniCharRemoveCharacterFromBitmap(0x0020, bitmapBase);
 577             CFUniCharRemoveCharacterFromBitmap(0x00A0, bitmapBase);
 578             CFUniCharRemoveCharacterFromBitmap(0x1680, bitmapBase);
 579             CFUniCharRemoveCharacterFromBitmap(0x202F, bitmapBase);
 580             CFUniCharRemoveCharacterFromBitmap(0x205F, bitmapBase);
 581             CFUniCharRemoveCharacterFromBitmap(0x3000, bitmapBase);
 582         } else {
 583             CFUniCharAddCharacterToBitmap(0x0009, bitmapBase);
 584             CFUniCharAddCharacterToBitmap(0x0020, bitmapBase);
 585             CFUniCharAddCharacterToBitmap(0x00A0, bitmapBase);
 586             CFUniCharAddCharacterToBitmap(0x1680, bitmapBase);
 587             CFUniCharAddCharacterToBitmap(0x202F, bitmapBase);
 588             CFUniCharAddCharacterToBitmap(0x205F, bitmapBase);
 589             CFUniCharAddCharacterToBitmap(0x3000, bitmapBase);
 590         }
 591
 592         for (idx = 0x2000;idx <= 0x200B;idx++) {
 593             if (isInverted) {
 594                 CFUniCharRemoveCharacterFromBitmap(idx, bitmapBase);
 595             } else {
 596                 CFUniCharAddCharacterToBitmap(idx, bitmapBase);
 597             }
 598         }
 599         return kCFUniCharBitmapFilled;
 600     }
 601     return (isInverted ? kCFUniCharBitmapAll : kCFUniCharBitmapEmpty);
 602 }
 603
 604 __private_extern__ uint32_t CFUniCharGetNumberOfPlanes(uint32_t charset) {
 605     if ((charset == kCFUniCharControlCharacterSet) || (charset == kCFUniCharControlAndFormatterCharacterSet)) {
 606         return 15; // 0 to 14
 607     } else if (charset < kCFUniCharDecimalDigitCharacterSet) {
 608         return 1;
 609     } else if (charset == kCFUniCharIllegalCharacterSet) {
 610         return 17;
 611     } else {
 612         uint32_t numPlanes;
 613
 614         if (NULL == __CFUniCharBitmapDataArray) __CFUniCharLoadBitmapData();
 615
 616         numPlanes = __CFUniCharBitmapDataArray[__CFUniCharMapExternalSetToInternalIndex(__CFUniCharMapCompatibilitySetID(charset))]._numPlanes;
 617
 618         return numPlanes;
 619     }
 620 }
 621
 622 // Mapping data loading
 623 static const void **__CFUniCharMappingTables = NULL;
 624
 625 static CFSpinLock_t __CFUniCharMappingTableLock = CFSpinLockInit;
 626
 627 #if DEPLOYMENT_TARGET_MACOSX || DEPLOYMENT_TARGET_EMBEDDED || DEPLOYMENT_TARGET_LINUX
 628 #if __CF_BIG_ENDIAN__
 629 #if USE_MACHO_SEGMENT
 630 #define MAPPING_TABLE_FILE "__data"
 631 #else
 632 #define MAPPING_TABLE_FILE "/CFUnicodeData-B.mapping"
 633 #endif
 634 #else
 635 #if USE_MACHO_SEGMENT
 636 #define MAPPING_TABLE_FILE "__data"
 637 #else
 638 #define MAPPING_TABLE_FILE "/CFUnicodeData-L.mapping"
 639 #endif
 640 #endif
 641 #elif DEPLOYMENT_TARGET_WINDOWS
 642 #if __CF_BIG_ENDIAN__
 643 #if USE_MACHO_SEGMENT
 644 #define MAPPING_TABLE_FILE "__data"
 645 #else
 646 #define MAPPING_TABLE_FILE L"CFUnicodeData-B.mapping"
 647 #endif
 648 #else
 649 #if USE_MACHO_SEGMENT
 650 #define MAPPING_TABLE_FILE "__data"
 651 #else
 652 #define MAPPING_TABLE_FILE L"CFUnicodeData-L.mapping"
 653 #endif
 654 #endif
 655 #else
 656 #error Unknown or unspecified DEPLOYMENT_TARGET
 657 #endif
 658
 659 __private_extern__ const void *CFUniCharGetMappingData(uint32_t type) {
 660
 661     __CFSpinLock(&__CFUniCharMappingTableLock);
 662
 663     if (NULL == __CFUniCharMappingTables) {
 664         const void *bytes;
 665         const void *bodyBase;
 666         int headerSize;
 667         int idx, count;
 668         int64_t fileSize;
 669
 670         if (!__CFUniCharLoadFile(MAPPING_TABLE_FILE, &bytes, &fileSize) || !__CFSimpleFileSizeVerification(bytes, fileSize)) {
 671             __CFSpinUnlock(&__CFUniCharMappingTableLock);
 672             return NULL;
 673         }
 674
 675 #if defined (__cplusplus)
 676                 bytes = (uint8_t *)bytes + 4; // Skip Unicode version
 677                 headerSize = *((uint8_t *)bytes); bytes = (uint8_t *)bytes + sizeof(uint32_t);
 678 #else
 679                 bytes += 4; // Skip Unicode version
 680                 headerSize = *((uint32_t *)bytes); bytes += sizeof(uint32_t);
 681 #endif
 682         headerSize -= (sizeof(uint32_t) * 2);
 683         bodyBase = (char *)bytes + headerSize;
 684
 685         count = headerSize / sizeof(uint32_t);
 686
 687         __CFUniCharMappingTables = (const void **)CFAllocatorAllocate(kCFAllocatorSystemDefault, sizeof(const void *) * count, 0);
 688
 689         for (idx = 0;idx < count;idx++) {
 690 #if defined (__cplusplus)
 691                         __CFUniCharMappingTables[idx] = (char *)bodyBase + *((uint32_t *)bytes); bytes = (uint8_t *)bytes + sizeof(uint32_t);
 692 #else
 693                         __CFUniCharMappingTables[idx] = (char *)bodyBase + *((uint32_t *)bytes); bytes += sizeof(uint32_t);
 694 #endif
 695         }
 696     }
 697
 698     __CFSpinUnlock(&__CFUniCharMappingTableLock);
 699
 700     return __CFUniCharMappingTables[type];
 701 }
 702
 703 // Case mapping functions
 704 #define DO_SPECIAL_CASE_MAPPING 1
 705
 706 static uint32_t *__CFUniCharCaseMappingTableCounts = NULL;
 707 static uint32_t **__CFUniCharCaseMappingTable = NULL;
 708 static const uint32_t **__CFUniCharCaseMappingExtraTable = NULL;
 709
 710 typedef struct {
 711     uint32_t _key;
 712     uint32_t _value;
 713 } __CFUniCharCaseMappings;
 714
 715 /* Binary searches CFStringEncodingUnicodeTo8BitCharMap */
 716 static uint32_t __CFUniCharGetMappedCase(const __CFUniCharCaseMappings *theTable, uint32_t numElem, UTF32Char character) {
 717     const __CFUniCharCaseMappings *p, *q, *divider;
 718
 719     if ((character < theTable[0]._key) || (character > theTable[numElem-1]._key)) {
 720         return 0;
 721     }
 722     p = theTable;
 723     q = p + (numElem-1);
 724     while (p <= q) {
 725         divider = p + ((q - p) >> 1);   /* divide by 2 */
 726         if (character < divider->_key) { q = divider - 1; }
 727         else if (character > divider->_key) { p = divider + 1; }
 728         else { return divider->_value; }
 729     }
 730     return 0;
 731 }
 732
 733 #define NUM_CASE_MAP_DATA (kCFUniCharCaseFold + 1)
 734
 735 static bool __CFUniCharLoadCaseMappingTable(void) {
 736     uint32_t *countArray;
 737     int idx;
 738
 739     if (NULL == __CFUniCharMappingTables) (void)CFUniCharGetMappingData(kCFUniCharToLowercase);
 740     if (NULL == __CFUniCharMappingTables) return false;
 741
 742     __CFSpinLock(&__CFUniCharMappingTableLock);
 743
 744     if (__CFUniCharCaseMappingTableCounts) {
 745         __CFSpinUnlock(&__CFUniCharMappingTableLock);
 746         return true;
 747     }
 748
 749     countArray = (uint32_t *)CFAllocatorAllocate(kCFAllocatorSystemDefault, sizeof(uint32_t) * NUM_CASE_MAP_DATA + sizeof(uint32_t *) * NUM_CASE_MAP_DATA * 2, 0);
 750     __CFUniCharCaseMappingTable = (uint32_t **)((char *)countArray + sizeof(uint32_t) * NUM_CASE_MAP_DATA);
 751     __CFUniCharCaseMappingExtraTable = (const uint32_t **)__CFUniCharCaseMappingTable + NUM_CASE_MAP_DATA;
 752
 753     for (idx = 0;idx < NUM_CASE_MAP_DATA;idx++) {
 754         countArray[idx] = *((uint32_t *)__CFUniCharMappingTables[idx]) / (sizeof(uint32_t) * 2);
 755         __CFUniCharCaseMappingTable[idx] = ((uint32_t *)__CFUniCharMappingTables[idx]) + 1;
 756         __CFUniCharCaseMappingExtraTable[idx] = (const uint32_t *)((char *)__CFUniCharCaseMappingTable[idx] + *((uint32_t *)__CFUniCharMappingTables[idx]));
 757     }
 758
 759     __CFUniCharCaseMappingTableCounts = countArray;
 760
 761     __CFSpinUnlock(&__CFUniCharMappingTableLock);
 762     return true;
 763 }
 764
 765 #if __CF_BIG_ENDIAN__
 766 #define TURKISH_LANG_CODE       (0x7472) // tr
 767 #define LITHUANIAN_LANG_CODE    (0x6C74) // lt
 768 #define AZERI_LANG_CODE         (0x617A) // az
 769 #define DUTCH_LANG_CODE         (0x6E6C) // nl
 770 #else
 771 #define TURKISH_LANG_CODE       (0x7274) // tr
 772 #define LITHUANIAN_LANG_CODE    (0x746C) // lt
 773 #define AZERI_LANG_CODE         (0x7A61) // az
 774 #define DUTCH_LANG_CODE         (0x6C6E) // nl
 775 #endif
 776
 777 CFIndex CFUniCharMapCaseTo(UTF32Char theChar, UTF16Char *convertedChar, CFIndex maxLength, uint32_t ctype, uint32_t flags, const uint8_t *langCode) {
 778     __CFUniCharBitmapData *data;
 779     uint8_t planeNo = (theChar >> 16) & 0xFF;
 780
 781 caseFoldRetry:
 782
 783 #if DO_SPECIAL_CASE_MAPPING
 784     if (flags & kCFUniCharCaseMapFinalSigma) {
 785         if (theChar == 0x03A3) { // Final sigma
 786             *convertedChar = (ctype == kCFUniCharToLowercase ? 0x03C2 : 0x03A3);
 787             return 1;
 788         }
 789     }
 790
 791     if (langCode) {
 792         switch (*(uint16_t *)langCode) {
 793             case LITHUANIAN_LANG_CODE:
 794                 if (theChar == 0x0307 && (flags & kCFUniCharCaseMapAfter_i)) {
 795                     return 0;
 796                 } else if (ctype == kCFUniCharToLowercase) {
 797                     if (flags & kCFUniCharCaseMapMoreAbove) {
 798                         switch (theChar) {
 799                             case 0x0049: // LATIN CAPITAL LETTER I
 800                                 *(convertedChar++) = 0x0069;
 801                                 *(convertedChar++) = 0x0307;
 802                                 return 2;
 803
 804                             case 0x004A: // LATIN CAPITAL LETTER J
 805                                 *(convertedChar++) = 0x006A;
 806                                 *(convertedChar++) = 0x0307;
 807                                 return 2;
 808
 809                             case 0x012E: // LATIN CAPITAL LETTER I WITH OGONEK
 810                                 *(convertedChar++) = 0x012F;
 811                                 *(convertedChar++) = 0x0307;
 812                                 return 2;
 813
 814                             default: break;
 815                         }
 816                     }
 817                     switch (theChar) {
 818                         case 0x00CC: // LATIN CAPITAL LETTER I WITH GRAVE
 819                             *(convertedChar++) = 0x0069;
 820                             *(convertedChar++) = 0x0307;
 821                             *(convertedChar++) = 0x0300;
 822                             return 3;
 823
 824                         case 0x00CD: // LATIN CAPITAL LETTER I WITH ACUTE
 825                             *(convertedChar++) = 0x0069;
 826                             *(convertedChar++) = 0x0307;
 827                             *(convertedChar++) = 0x0301;
 828                             return 3;
 829
 830                         case 0x0128: // LATIN CAPITAL LETTER I WITH TILDE
 831                             *(convertedChar++) = 0x0069;
 832                             *(convertedChar++) = 0x0307;
 833                             *(convertedChar++) = 0x0303;
 834                             return 3;
 835
 836                         default: break;
 837                     }
 838                 }
 839             break;
 840
 841             case TURKISH_LANG_CODE:
 842             case AZERI_LANG_CODE:
 843                 if ((theChar == 0x0049) || (theChar == 0x0131)) { // LATIN CAPITAL LETTER I & LATIN SMALL LETTER DOTLESS I
 844                     *convertedChar = (((ctype == kCFUniCharToLowercase) || (ctype == kCFUniCharCaseFold))  ? ((kCFUniCharCaseMapMoreAbove & flags) ? 0x0069 : 0x0131) : 0x0049);
 845                     return 1;
 846                 } else if ((theChar == 0x0069) || (theChar == 0x0130)) { // LATIN SMALL LETTER I & LATIN CAPITAL LETTER I WITH DOT ABOVE
 847                     *convertedChar = (((ctype == kCFUniCharToLowercase) || (ctype == kCFUniCharCaseFold)) ? 0x0069 : 0x0130);
 848                     return 1;
 849                 } else if (theChar == 0x0307 && (kCFUniCharCaseMapAfter_i & flags)) { // COMBINING DOT ABOVE AFTER_i
 850                     if (ctype == kCFUniCharToLowercase) {
 851                         return 0;
 852                     } else {
 853                         *convertedChar = 0x0307;
 854                         return 1;
 855                     }
 856                 }
 857                 break;
 858
 859             case DUTCH_LANG_CODE:
 860                 if ((theChar == 0x004A) || (theChar == 0x006A)) {
 861                     *convertedChar = (((ctype == kCFUniCharToUppercase) || (ctype == kCFUniCharToTitlecase) || (kCFUniCharCaseMapDutchDigraph & flags)) ? 0x004A  : 0x006A);
 862                     return 1;
 863                 }
 864                 break;
 865
 866             default: break;
 867         }
 868     }
 869 #endif // DO_SPECIAL_CASE_MAPPING
 870
 871     if (NULL == __CFUniCharBitmapDataArray) __CFUniCharLoadBitmapData();
 872
 873     data = __CFUniCharBitmapDataArray + __CFUniCharMapExternalSetToInternalIndex(__CFUniCharMapCompatibilitySetID(ctype + kCFUniCharHasNonSelfLowercaseCharacterSet));
 874
 875     if (planeNo < data->_numPlanes && data->_planes[planeNo] && CFUniCharIsMemberOfBitmap(theChar, data->_planes[planeNo]) && (__CFUniCharCaseMappingTableCounts || __CFUniCharLoadCaseMappingTable())) {
 876         uint32_t value = __CFUniCharGetMappedCase((const __CFUniCharCaseMappings *)__CFUniCharCaseMappingTable[ctype], __CFUniCharCaseMappingTableCounts[ctype], theChar);
 877
 878         if (!value && ctype == kCFUniCharToTitlecase) {
 879             value = __CFUniCharGetMappedCase((const __CFUniCharCaseMappings *)__CFUniCharCaseMappingTable[kCFUniCharToUppercase], __CFUniCharCaseMappingTableCounts[kCFUniCharToUppercase], theChar);
 880             if (value) ctype = kCFUniCharToUppercase;
 881         }
 882
 883         if (value) {
 884             CFIndex count = CFUniCharConvertFlagToCount(value);
 885
 886             if (count == 1) {
 887                 if (value & kCFUniCharNonBmpFlag) {
 888                     if (maxLength > 1) {
 889                         value = (value & 0xFFFFFF) - 0x10000;
 890                         *(convertedChar++) = (UTF16Char)(value >> 10) + 0xD800UL;
 891                         *(convertedChar++) = (UTF16Char)(value & 0x3FF) + 0xDC00UL;
 892                         return 2;
 893                     }
 894                 } else {
 895                     *convertedChar = (UTF16Char)value;
 896                     return 1;
 897                 }
 898             } else if (count < maxLength) {
 899                 const uint32_t *extraMapping = __CFUniCharCaseMappingExtraTable[ctype] + (value & 0xFFFFFF);
 900
 901                 if (value & kCFUniCharNonBmpFlag) {
 902                     CFIndex copiedLen = 0;
 903
 904                     while (count-- > 0) {
 905                         value = *(extraMapping++);
 906                         if (value > 0xFFFF) {
 907                             if (copiedLen + 2 >= maxLength) break;
 908                             value = (value & 0xFFFFFF) - 0x10000;
 909                             convertedChar[copiedLen++] = (UTF16Char)(value >> 10) + 0xD800UL;
 910                             convertedChar[copiedLen++] = (UTF16Char)(value & 0x3FF) + 0xDC00UL;
 911                         } else {
 912                             if (copiedLen + 1 >= maxLength) break;
 913                             convertedChar[copiedLen++] = value;
 914                         }
 915                     }
 916                     if (!count) return copiedLen;
 917                 } else {
 918                     CFIndex idx;
 919
 920                     for (idx = 0;idx < count;idx++) *(convertedChar++) = (UTF16Char)*(extraMapping++);
 921                     return count;
 922                 }
 923             }
 924         }
 925     } else if (ctype == kCFUniCharCaseFold) {
 926         ctype = kCFUniCharToLowercase;
 927         goto caseFoldRetry;
 928     }
 929
 930     if (theChar > 0xFFFF) { // non-BMP
 931         theChar = (theChar & 0xFFFFFF) - 0x10000;
 932         *(convertedChar++) = (UTF16Char)(theChar >> 10) + 0xD800UL;
 933         *(convertedChar++) = (UTF16Char)(theChar & 0x3FF) + 0xDC00UL;
 934         return 2;
 935     } else {
 936         *convertedChar = theChar;
 937         return 1;
 938     }
 939 }
 940
 941 CFIndex CFUniCharMapTo(UniChar theChar, UniChar *convertedChar, CFIndex maxLength, uint16_t ctype, uint32_t flags) {
 942     if (ctype == kCFUniCharCaseFold + 1) { // kCFUniCharDecompose
 943         if (CFUniCharIsDecomposableCharacter(theChar, false)) {
 944             UTF32Char buffer[MAX_DECOMPOSED_LENGTH];
 945             CFIndex usedLength = CFUniCharDecomposeCharacter(theChar, buffer, MAX_DECOMPOSED_LENGTH);
 946             CFIndex idx;
 947
 948             for (idx = 0;idx < usedLength;idx++) *(convertedChar++) = buffer[idx];
 949             return usedLength;
 950         } else {
 951             *convertedChar = theChar;
 952             return 1;
 953         }
 954     } else {
 955         return CFUniCharMapCaseTo(theChar, convertedChar, maxLength, ctype, flags, NULL);
 956     }
 957 }
 958
 959 CF_INLINE bool __CFUniCharIsMoreAbove(UTF16Char *buffer, CFIndex length) {
 960     UTF32Char currentChar;
 961     uint32_t property;
 962
 963     while (length-- > 0) {
 964         currentChar = *(buffer)++;
 965         if (CFUniCharIsSurrogateHighCharacter(currentChar) && (length > 0) && CFUniCharIsSurrogateLowCharacter(*(buffer + 1))) {
 966             currentChar = CFUniCharGetLongCharacterForSurrogatePair(currentChar, *(buffer++));
 967             --length;
 968         }
 969         if (!CFUniCharIsMemberOf(currentChar, kCFUniCharNonBaseCharacterSet)) break;
 970
 971         property = CFUniCharGetCombiningPropertyForCharacter(currentChar, (const uint8_t *)CFUniCharGetUnicodePropertyDataForPlane(kCFUniCharCombiningProperty, (currentChar >> 16) & 0xFF));
 972
 973         if (property == 230) return true; // Above priority
 974     }
 975     return false;
 976 }
 977
 978 CF_INLINE bool __CFUniCharIsAfter_i(UTF16Char *buffer, CFIndex length) {
 979     UTF32Char currentChar = 0;
 980     uint32_t property;
 981     UTF32Char decomposed[MAX_DECOMPOSED_LENGTH];
 982     CFIndex decompLength;
 983     CFIndex idx;
 984
 985     if (length < 1) return 0;
 986
 987     buffer += length;
 988     while (length-- > 1) {
 989         currentChar = *(--buffer);
 990         if (CFUniCharIsSurrogateLowCharacter(currentChar)) {
 991             if ((length > 1) && CFUniCharIsSurrogateHighCharacter(*(buffer - 1))) {
 992                 currentChar = CFUniCharGetLongCharacterForSurrogatePair(*(--buffer), currentChar);
 993                 --length;
 994             } else {
 995                 break;
 996             }
 997         }
 998         if (!CFUniCharIsMemberOf(currentChar, kCFUniCharNonBaseCharacterSet)) break;
 999
1000         property = CFUniCharGetCombiningPropertyForCharacter(currentChar, (const uint8_t *)CFUniCharGetUnicodePropertyDataForPlane(kCFUniCharCombiningProperty, (currentChar >> 16) & 0xFF));
1001
1002         if (property == 230) return false; // Above priority
1003     }
1004     if (length == 0) {
1005         currentChar = *(--buffer);
1006     } else if (CFUniCharIsSurrogateLowCharacter(currentChar) && CFUniCharIsSurrogateHighCharacter(*(--buffer))) {
1007         currentChar = CFUniCharGetLongCharacterForSurrogatePair(*buffer, currentChar);
1008     }
1009
1010     decompLength = CFUniCharDecomposeCharacter(currentChar, decomposed, MAX_DECOMPOSED_LENGTH);
1011     currentChar = *decomposed;
1012
1013
1014     for (idx = 1;idx < decompLength;idx++) {
1015         currentChar = decomposed[idx];
1016         property = CFUniCharGetCombiningPropertyForCharacter(currentChar, (const uint8_t *)CFUniCharGetUnicodePropertyDataForPlane(kCFUniCharCombiningProperty, (currentChar >> 16) & 0xFF));
1017
1018         if (property == 230) return false; // Above priority
1019     }
1020     return true;
1021 }
1022
1023 __private_extern__ uint32_t CFUniCharGetConditionalCaseMappingFlags(UTF32Char theChar, UTF16Char *buffer, CFIndex currentIndex, CFIndex length, uint32_t type, const uint8_t *langCode, uint32_t lastFlags) {
1024     if (theChar == 0x03A3) { // GREEK CAPITAL LETTER SIGMA
1025         if ((type == kCFUniCharToLowercase) && (currentIndex > 0)) {
1026             UTF16Char *start = buffer;
1027             UTF16Char *end = buffer + length;
1028             UTF32Char otherChar;
1029
1030             // First check if we're after a cased character
1031             buffer += (currentIndex - 1);
1032             while (start <= buffer) {
1033                 otherChar = *(buffer--);
1034                 if (CFUniCharIsSurrogateLowCharacter(otherChar) && (start <= buffer) && CFUniCharIsSurrogateHighCharacter(*buffer)) {
1035                     otherChar = CFUniCharGetLongCharacterForSurrogatePair(*(buffer--), otherChar);
1036                 }
1037                 if (!CFUniCharIsMemberOf(otherChar, kCFUniCharCaseIgnorableCharacterSet)) {
1038                     if (!CFUniCharIsMemberOf(otherChar, kCFUniCharUppercaseLetterCharacterSet) && !CFUniCharIsMemberOf(otherChar, kCFUniCharLowercaseLetterCharacterSet)) return 0; // Uppercase set contains titlecase
1039                     break;
1040                 }
1041             }
1042
1043             // Next check if we're before a cased character
1044             buffer = start + currentIndex + 1;
1045             while (buffer < end) {
1046                 otherChar = *(buffer++);
1047                 if (CFUniCharIsSurrogateHighCharacter(otherChar) && (buffer < end) && CFUniCharIsSurrogateLowCharacter(*buffer)) {
1048                     otherChar = CFUniCharGetLongCharacterForSurrogatePair(otherChar, *(buffer++));
1049                 }
1050                 if (!CFUniCharIsMemberOf(otherChar, kCFUniCharCaseIgnorableCharacterSet)) {
1051                     if (CFUniCharIsMemberOf(otherChar, kCFUniCharUppercaseLetterCharacterSet) || CFUniCharIsMemberOf(otherChar, kCFUniCharLowercaseLetterCharacterSet)) return 0; // Uppercase set contains titlecase
1052                     break;
1053                 }
1054             }
1055             return kCFUniCharCaseMapFinalSigma;
1056         }
1057     } else if (langCode) {
1058         if (*((const uint16_t *)langCode) == LITHUANIAN_LANG_CODE) {
1059             if ((theChar == 0x0307) && ((kCFUniCharCaseMapAfter_i|kCFUniCharCaseMapMoreAbove) & lastFlags) == (kCFUniCharCaseMapAfter_i|kCFUniCharCaseMapMoreAbove)) {
1060                 return (__CFUniCharIsAfter_i(buffer, currentIndex) ? kCFUniCharCaseMapAfter_i : 0);
1061             } else if (type == kCFUniCharToLowercase) {
1062                 if ((theChar == 0x0049) || (theChar == 0x004A) || (theChar == 0x012E)) {
1063                     return (__CFUniCharIsMoreAbove(buffer + (++currentIndex), length - currentIndex) ? kCFUniCharCaseMapMoreAbove : 0);
1064                 }
1065             } else if ((theChar == 'i') || (theChar == 'j')) {
1066                 return (__CFUniCharIsMoreAbove(buffer + (++currentIndex), length - currentIndex) ? (kCFUniCharCaseMapAfter_i|kCFUniCharCaseMapMoreAbove) : 0);
1067             }
1068         } else if ((*((const uint16_t *)langCode) == TURKISH_LANG_CODE) || (*((const uint16_t *)langCode) == AZERI_LANG_CODE)) {
1069             if (type == kCFUniCharToLowercase) {
1070                 if (theChar == 0x0307) {
1071                     return (kCFUniCharCaseMapMoreAbove & lastFlags ? kCFUniCharCaseMapAfter_i : 0);
1072                 } else if (theChar == 0x0049) {
1073                     return (((++currentIndex < length) && (buffer[currentIndex] == 0x0307)) ? kCFUniCharCaseMapMoreAbove : 0);
1074                 }
1075             }
1076         } else if (*((const uint16_t *)langCode) == DUTCH_LANG_CODE) {
1077             if (kCFUniCharCaseMapDutchDigraph & lastFlags) {
1078                 return (((theChar == 0x006A) || (theChar == 0x004A)) ? kCFUniCharCaseMapDutchDigraph : 0);
1079             } else {
1080                 if ((type == kCFUniCharToTitlecase) && ((theChar == 0x0069) || (theChar == 0x0049))) {
1081                     return (((++currentIndex < length) && ((buffer[currentIndex] == 0x006A) || (buffer[currentIndex] == 0x004A))) ? kCFUniCharCaseMapDutchDigraph : 0);
1082                 }
1083             }
1084         }
1085     }
1086     return 0;
1087 }
1088
1089 // Unicode property database
1090 static __CFUniCharBitmapData *__CFUniCharUnicodePropertyTable = NULL;
1091 static int __CFUniCharUnicodePropertyTableCount = 0;
1092
1093 static CFSpinLock_t __CFUniCharPropTableLock = CFSpinLockInit;
1094
1095 #if DEPLOYMENT_TARGET_MACOSX || DEPLOYMENT_TARGET_EMBEDDED || DEPLOYMENT_TARGET_LINUX
1096 #if USE_MACHO_SEGMENT
1097 #define PROP_DB_FILE "__properties"
1098 #else
1099 #define PROP_DB_FILE "/CFUniCharPropertyDatabase.data"
1100 #endif
1101 #elif DEPLOYMENT_TARGET_WINDOWS
1102 #if USE_MACHO_SEGMENT
1103 #define PROP_DB_FILE "__properties"
1104 #else
1105 #define PROP_DB_FILE L"CFUniCharPropertyDatabase.data"
1106 #endif
1107 #else
1108 #error Unknown or unspecified DEPLOYMENT_TARGET
1109 #endif
1110
1111 const void *CFUniCharGetUnicodePropertyDataForPlane(uint32_t propertyType, uint32_t plane) {
1112
1113     __CFSpinLock(&__CFUniCharPropTableLock);
1114
1115     if (NULL == __CFUniCharUnicodePropertyTable) {
1116         __CFUniCharBitmapData *table;
1117         const void *bytes;
1118         const void *bodyBase;
1119         const void *planeBase;
1120         int headerSize;
1121         int idx, count;
1122         int planeIndex, planeCount;
1123         int planeSize;
1124         int64_t fileSize;
1125
1126         if (!__CFUniCharLoadFile(PROP_DB_FILE, &bytes, &fileSize) || !__CFSimpleFileSizeVerification(bytes, fileSize)) {
1127             __CFSpinUnlock(&__CFUniCharPropTableLock);
1128             return NULL;
1129         }
1130
1131 #if defined (__cplusplus)
1132                 bytes = (uint8_t*)bytes + 4; // Skip Unicode version
1133                 headerSize = CFSwapInt32BigToHost(*((uint32_t *)bytes)); bytes = (uint8_t *)bytes + sizeof(uint32_t);
1134 #else
1135                 bytes += 4; // Skip Unicode version
1136                 headerSize = CFSwapInt32BigToHost(*((uint32_t *)bytes)); bytes += sizeof(uint32_t);
1137 #endif
1138
1139         headerSize -= (sizeof(uint32_t) * 2);
1140         bodyBase = (char *)bytes + headerSize;
1141
1142         count = headerSize / sizeof(uint32_t);
1143         __CFUniCharUnicodePropertyTableCount = count;
1144
1145         table = (__CFUniCharBitmapData *)CFAllocatorAllocate(kCFAllocatorSystemDefault, sizeof(__CFUniCharBitmapData) * count, 0);
1146
1147         for (idx = 0;idx < count;idx++) {
1148             planeCount = *((const uint8_t *)bodyBase);
1149             planeBase = (char *)bodyBase + planeCount + (planeCount % 4 ? 4 - (planeCount % 4) : 0);
1150             table[idx]._planes = (const uint8_t **)CFAllocatorAllocate(kCFAllocatorSystemDefault, sizeof(const void *) * planeCount, 0);
1151
1152             for (planeIndex = 0;planeIndex < planeCount;planeIndex++) {
1153                 if ((planeSize = ((const uint8_t *)bodyBase)[planeIndex + 1])) {
1154                     table[idx]._planes[planeIndex] = (const uint8_t *)planeBase;
1155 #if defined (__cplusplus)
1156                                         planeBase = (char*)planeBase + (planeSize * 256);
1157 #else
1158                                         planeBase += (planeSize * 256);
1159 #endif
1160                 } else {
1161                     table[idx]._planes[planeIndex] = NULL;
1162                 }
1163             }
1164
1165             table[idx]._numPlanes = planeCount;
1166 #if defined (__cplusplus)
1167                         bodyBase = (const uint8_t *)bodyBase + (CFSwapInt32BigToHost(*(uint32_t *)bytes));
1168                         ((uint32_t *&)bytes) ++;
1169 #else
1170                         bodyBase += (CFSwapInt32BigToHost(*((uint32_t *)bytes++)));
1171 #endif
1172         }
1173
1174         __CFUniCharUnicodePropertyTable = table;
1175     }
1176
1177     __CFSpinUnlock(&__CFUniCharPropTableLock);
1178
1179     return (plane < __CFUniCharUnicodePropertyTable[propertyType]._numPlanes ? __CFUniCharUnicodePropertyTable[propertyType]._planes[plane] : NULL);
1180 }
1181
1182 __private_extern__ uint32_t CFUniCharGetNumberOfPlanesForUnicodePropertyData(uint32_t propertyType) {
1183     (void)CFUniCharGetUnicodePropertyDataForPlane(propertyType, 0);
1184     return __CFUniCharUnicodePropertyTable[propertyType]._numPlanes;
1185 }
1186
1187 __private_extern__ uint32_t CFUniCharGetUnicodeProperty(UTF32Char character, uint32_t propertyType) {
1188     if (propertyType == kCFUniCharCombiningProperty) {
1189         return CFUniCharGetCombiningPropertyForCharacter(character, (const uint8_t *)CFUniCharGetUnicodePropertyDataForPlane(propertyType, (character >> 16) & 0xFF));
1190     } else if (propertyType == kCFUniCharBidiProperty) {
1191         return CFUniCharGetBidiPropertyForCharacter(character, (const uint8_t *)CFUniCharGetUnicodePropertyDataForPlane(propertyType, (character >> 16) & 0xFF));
1192     } else {
1193         return 0;
1194     }
1195 }
1196
1197
1198
1199 /*
1200     The UTF8 conversion in the following function is derived from ConvertUTF.c
1201 */
1202 /*
1203  * Copyright 2001 Unicode, Inc.
1204  *
1205  * Disclaimer
1206  *
1207  * This source code is provided as is by Unicode, Inc. No claims are
1208  * made as to fitness for any particular purpose. No warranties of any
1209  * kind are expressed or implied. The recipient agrees to determine
1210  * applicability of information provided. If this file has been
1211  * purchased on magnetic or optical media from Unicode, Inc., the
1212  * sole remedy for any claim will be exchange of defective media
1213  * within 90 days of receipt.
1214  *
1215  * Limitations on Rights to Redistribute This Code
1216  *
1217  * Unicode, Inc. hereby grants the right to freely use the information
1218  * supplied in this file in the creation of products supporting the
1219  * Unicode Standard, and to make copies of this file in any form
1220  * for internal or external distribution as long as this notice
1221  * remains attached.
1222  */
1223 #define UNI_REPLACEMENT_CHAR (0x0000FFFDUL)
1224
1225 bool CFUniCharFillDestinationBuffer(const UTF32Char *src, CFIndex srcLength, void **dst, CFIndex dstLength, CFIndex *filledLength, uint32_t dstFormat) {
1226     UTF32Char currentChar;
1227     CFIndex usedLength = *filledLength;
1228
1229     if (dstFormat == kCFUniCharUTF16Format) {
1230         UTF16Char *dstBuffer = (UTF16Char *)*dst;
1231
1232         while (srcLength-- > 0) {
1233             currentChar = *(src++);
1234
1235             if (currentChar > 0xFFFF) { // Non-BMP
1236                 usedLength += 2;
1237                 if (dstLength) {
1238                     if (usedLength > dstLength) return false;
1239                     currentChar -= 0x10000;
1240                     *(dstBuffer++) = (UTF16Char)((currentChar >> 10) + 0xD800UL);
1241                     *(dstBuffer++) = (UTF16Char)((currentChar & 0x3FF) + 0xDC00UL);
1242                 }
1243             } else {
1244                 ++usedLength;
1245                 if (dstLength) {
1246                     if (usedLength > dstLength) return false;
1247                     *(dstBuffer++) = (UTF16Char)currentChar;
1248                 }
1249             }
1250         }
1251
1252         *dst = dstBuffer;
1253     } else if (dstFormat == kCFUniCharUTF8Format) {
1254         uint8_t *dstBuffer = (uint8_t *)*dst;
1255         uint16_t bytesToWrite = 0;
1256         const UTF32Char byteMask = 0xBF;
1257         const UTF32Char byteMark = 0x80;
1258         static const uint8_t firstByteMark[7] = { 0x00, 0x00, 0xC0, 0xE0, 0xF0, 0xF8, 0xFC };
1259
1260         while (srcLength-- > 0) {
1261             currentChar = *(src++);
1262
1263             /* Figure out how many bytes the result will require */
1264             if (currentChar < (UTF32Char)0x80) {
1265                 bytesToWrite = 1;
1266             } else if (currentChar < (UTF32Char)0x800) {
1267                 bytesToWrite = 2;
1268             } else if (currentChar < (UTF32Char)0x10000) {
1269                 bytesToWrite = 3;
1270             } else if (currentChar < (UTF32Char)0x200000) {
1271                 bytesToWrite = 4;
1272             } else {
1273                 bytesToWrite = 2;
1274                 currentChar = UNI_REPLACEMENT_CHAR;
1275             }
1276
1277             usedLength += bytesToWrite;
1278
1279             if (dstLength) {
1280                 if (usedLength > dstLength) return false;
1281
1282                 dstBuffer += bytesToWrite;
1283                 switch (bytesToWrite) { /* note: everything falls through. */
1284                     case 4:     *--dstBuffer = (currentChar | byteMark) & byteMask; currentChar >>= 6;
1285                     case 3:     *--dstBuffer = (currentChar | byteMark) & byteMask; currentChar >>= 6;
1286                     case 2:     *--dstBuffer = (currentChar | byteMark) & byteMask; currentChar >>= 6;
1287                     case 1:     *--dstBuffer =  currentChar | firstByteMark[bytesToWrite];
1288                 }
1289                 dstBuffer += bytesToWrite;
1290             }
1291         }
1292
1293         *dst = dstBuffer;
1294     } else {
1295         UTF32Char *dstBuffer = (UTF32Char *)*dst;
1296
1297         while (srcLength-- > 0) {
1298             currentChar = *(src++);
1299
1300             ++usedLength;
1301             if (dstLength) {
1302                 if (usedLength > dstLength) return false;
1303                 *(dstBuffer++) = currentChar;
1304             }
1305         }
1306
1307         *dst = dstBuffer;
1308     }
1309
1310     *filledLength = usedLength;
1311
1312     return true;
1313 }
1314
1315 #if DEPLOYMENT_TARGET_WINDOWS
1316 void __CFUniCharCleanup(void)
1317 {
1318     int idx;
1319
1320     // cleanup memory allocated by __CFUniCharLoadBitmapData()
1321     __CFSpinLock(&__CFUniCharBitmapLock);
1322
1323     if (__CFUniCharBitmapDataArray != NULL) {
1324         for (idx = 0; idx < (int)__CFUniCharNumberOfBitmaps; idx++) {
1325             CFAllocatorDeallocate(kCFAllocatorSystemDefault, __CFUniCharBitmapDataArray[idx]._planes);
1326             __CFUniCharBitmapDataArray[idx]._planes = NULL;
1327         }
1328
1329         CFAllocatorDeallocate(kCFAllocatorSystemDefault, __CFUniCharBitmapDataArray);
1330         __CFUniCharBitmapDataArray = NULL;
1331         __CFUniCharNumberOfBitmaps = 0;
1332     }
1333
1334     __CFSpinUnlock(&__CFUniCharBitmapLock);
1335
1336     // cleanup memory allocated by CFUniCharGetMappingData()
1337     __CFSpinLock(&__CFUniCharMappingTableLock);
1338
1339     if (__CFUniCharMappingTables != NULL) {
1340         CFAllocatorDeallocate(kCFAllocatorSystemDefault, __CFUniCharMappingTables);
1341         __CFUniCharMappingTables = NULL;
1342     }
1343
1344     // cleanup memory allocated by __CFUniCharLoadCaseMappingTable()
1345     if (__CFUniCharCaseMappingTableCounts != NULL) {
1346         CFAllocatorDeallocate(kCFAllocatorSystemDefault, __CFUniCharCaseMappingTableCounts);
1347         __CFUniCharCaseMappingTableCounts = NULL;
1348
1349         __CFUniCharCaseMappingTable = NULL;
1350         __CFUniCharCaseMappingExtraTable = NULL;
1351     }
1352
1353     __CFSpinUnlock(&__CFUniCharMappingTableLock);
1354
1355     // cleanup memory allocated by CFUniCharGetUnicodePropertyDataForPlane()
1356     __CFSpinLock(&__CFUniCharPropTableLock);
1357
1358     if (__CFUniCharUnicodePropertyTable != NULL) {
1359         for (idx = 0; idx < __CFUniCharUnicodePropertyTableCount; idx++) {
1360             CFAllocatorDeallocate(kCFAllocatorSystemDefault, __CFUniCharUnicodePropertyTable[idx]._planes);
1361             __CFUniCharUnicodePropertyTable[idx]._planes = NULL;
1362         }
1363
1364         CFAllocatorDeallocate(kCFAllocatorSystemDefault, __CFUniCharUnicodePropertyTable);
1365         __CFUniCharUnicodePropertyTable = NULL;
1366         __CFUniCharUnicodePropertyTableCount = 0;
1367     }
1368
1369     __CFSpinUnlock(&__CFUniCharPropTableLock);
1370 }
1371 #endif
1372
1373 #undef USE_MACHO_SEGMENT
1374