wtf/unicode/UTF8.cpp

   1 /*
   2  * Copyright (C) 2007 Apple Inc.  All rights reserved.
   3  * Copyright (C) 2010 Patrick Gansterer <paroga@paroga.com>
   4  *
   5  * Redistribution and use in source and binary forms, with or without
   6  * modification, are permitted provided that the following conditions
   7  * are met:
   8  * 1. Redistributions of source code must retain the above copyright
   9  *    notice, this list of conditions and the following disclaimer.
  10  * 2. Redistributions in binary form must reproduce the above copyright
  11  *    notice, this list of conditions and the following disclaimer in the
  12  *    documentation and/or other materials provided with the distribution.
  13  *
  14  * THIS SOFTWARE IS PROVIDED BY APPLE COMPUTER, INC. ``AS IS'' AND ANY
  15  * EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
  16  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
  17  * PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL APPLE COMPUTER, INC. OR
  18  * CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
  19  * EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
  20  * PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
  21  * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY
  22  * OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
  23  * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
  24  * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
  25  */
  26
  27 #include "config.h"
  28 #include "UTF8.h"
  29
  30 #include "ASCIICType.h"
  31 #include <wtf/StringHasher.h>
  32 #include <wtf/unicode/CharacterNames.h>
  33
  34 namespace WTF {
  35 namespace Unicode {
  36
  37 inline int inlineUTF8SequenceLengthNonASCII(char b0)
  38 {
  39     if ((b0 & 0xC0) != 0xC0)
  40         return 0;
  41     if ((b0 & 0xE0) == 0xC0)
  42         return 2;
  43     if ((b0 & 0xF0) == 0xE0)
  44         return 3;
  45     if ((b0 & 0xF8) == 0xF0)
  46         return 4;
  47     return 0;
  48 }
  49
  50 inline int inlineUTF8SequenceLength(char b0)
  51 {
  52     return isASCII(b0) ? 1 : inlineUTF8SequenceLengthNonASCII(b0);
  53 }
  54
  55 int UTF8SequenceLength(char b0)
  56 {
  57     return isASCII(b0) ? 1 : inlineUTF8SequenceLengthNonASCII(b0);
  58 }
  59
  60 int decodeUTF8Sequence(const char* sequence)
  61 {
  62     // Handle 0-byte sequences (never valid).
  63     const unsigned char b0 = sequence[0];
  64     const int length = inlineUTF8SequenceLength(b0);
  65     if (length == 0)
  66         return -1;
  67
  68     // Handle 1-byte sequences (plain ASCII).
  69     const unsigned char b1 = sequence[1];
  70     if (length == 1) {
  71         if (b1)
  72             return -1;
  73         return b0;
  74     }
  75
  76     // Handle 2-byte sequences.
  77     if ((b1 & 0xC0) != 0x80)
  78         return -1;
  79     const unsigned char b2 = sequence[2];
  80     if (length == 2) {
  81         if (b2)
  82             return -1;
  83         const int c = ((b0 & 0x1F) << 6) | (b1 & 0x3F);
  84         if (c < 0x80)
  85             return -1;
  86         return c;
  87     }
  88
  89     // Handle 3-byte sequences.
  90     if ((b2 & 0xC0) != 0x80)
  91         return -1;
  92     const unsigned char b3 = sequence[3];
  93     if (length == 3) {
  94         if (b3)
  95             return -1;
  96         const int c = ((b0 & 0xF) << 12) | ((b1 & 0x3F) << 6) | (b2 & 0x3F);
  97         if (c < 0x800)
  98             return -1;
  99         // UTF-16 surrogates should never appear in UTF-8 data.
 100         if (c >= 0xD800 && c <= 0xDFFF)
 101             return -1;
 102         return c;
 103     }
 104
 105     // Handle 4-byte sequences.
 106     if ((b3 & 0xC0) != 0x80)
 107         return -1;
 108     const unsigned char b4 = sequence[4];
 109     if (length == 4) {
 110         if (b4)
 111             return -1;
 112         const int c = ((b0 & 0x7) << 18) | ((b1 & 0x3F) << 12) | ((b2 & 0x3F) << 6) | (b3 & 0x3F);
 113         if (c < 0x10000 || c > 0x10FFFF)
 114             return -1;
 115         return c;
 116     }
 117
 118     return -1;
 119 }
 120
 121 // Once the bits are split out into bytes of UTF-8, this is a mask OR-ed
 122 // into the first byte, depending on how many bytes follow.  There are
 123 // as many entries in this table as there are UTF-8 sequence types.
 124 // (I.e., one byte sequence, two byte... etc.). Remember that sequencs
 125 // for *legal* UTF-8 will be 4 or fewer bytes total.
 126 static const unsigned char firstByteMark[7] = { 0x00, 0x00, 0xC0, 0xE0, 0xF0, 0xF8, 0xFC };
 127
 128 ConversionResult convertUTF16ToUTF8(
 129     const UChar** sourceStart, const UChar* sourceEnd,
 130     char** targetStart, char* targetEnd, bool strict)
 131 {
 132     ConversionResult result = conversionOK;
 133     const UChar* source = *sourceStart;
 134     char* target = *targetStart;
 135     while (source < sourceEnd) {
 136         UChar32 ch;
 137         unsigned short bytesToWrite = 0;
 138         const UChar32 byteMask = 0xBF;
 139         const UChar32 byteMark = 0x80;
 140         const UChar* oldSource = source; // In case we have to back up because of target overflow.
 141         ch = static_cast<unsigned short>(*source++);
 142         // If we have a surrogate pair, convert to UChar32 first.
 143         if (ch >= 0xD800 && ch <= 0xDBFF) {
 144             // If the 16 bits following the high surrogate are in the source buffer...
 145             if (source < sourceEnd) {
 146                 UChar32 ch2 = static_cast<unsigned short>(*source);
 147                 // If it's a low surrogate, convert to UChar32.
 148                 if (ch2 >= 0xDC00 && ch2 <= 0xDFFF) {
 149                     ch = ((ch - 0xD800) << 10) + (ch2 - 0xDC00) + 0x0010000;
 150                     ++source;
 151                 } else if (strict) { // it's an unpaired high surrogate
 152                     --source; // return to the illegal value itself
 153                     result = sourceIllegal;
 154                     break;
 155                 }
 156             } else { // We don't have the 16 bits following the high surrogate.
 157                 --source; // return to the high surrogate
 158                 result = sourceExhausted;
 159                 break;
 160             }
 161         } else if (strict) {
 162             // UTF-16 surrogate values are illegal in UTF-32
 163             if (ch >= 0xDC00 && ch <= 0xDFFF) {
 164                 --source; // return to the illegal value itself
 165                 result = sourceIllegal;
 166                 break;
 167             }
 168         }
 169         // Figure out how many bytes the result will require
 170         if (ch < (UChar32)0x80) {
 171             bytesToWrite = 1;
 172         } else if (ch < (UChar32)0x800) {
 173             bytesToWrite = 2;
 174         } else if (ch < (UChar32)0x10000) {
 175             bytesToWrite = 3;
 176         } else if (ch < (UChar32)0x110000) {
 177             bytesToWrite = 4;
 178         } else {
 179             bytesToWrite = 3;
 180             ch = replacementCharacter;
 181         }
 182
 183         target += bytesToWrite;
 184         if (target > targetEnd) {
 185             source = oldSource; // Back up source pointer!
 186             target -= bytesToWrite;
 187             result = targetExhausted;
 188             break;
 189         }
 190         switch (bytesToWrite) { // note: everything falls through.
 191             case 4: *--target = (char)((ch | byteMark) & byteMask); ch >>= 6;
 192             case 3: *--target = (char)((ch | byteMark) & byteMask); ch >>= 6;
 193             case 2: *--target = (char)((ch | byteMark) & byteMask); ch >>= 6;
 194             case 1: *--target =  (char)(ch | firstByteMark[bytesToWrite]);
 195         }
 196         target += bytesToWrite;
 197     }
 198     *sourceStart = source;
 199     *targetStart = target;
 200     return result;
 201 }
 202
 203 // This must be called with the length pre-determined by the first byte.
 204 // If presented with a length > 4, this returns false.  The Unicode
 205 // definition of UTF-8 goes up to 4-byte sequences.
 206 static bool isLegalUTF8(const unsigned char* source, int length)
 207 {
 208     unsigned char a;
 209     const unsigned char* srcptr = source + length;
 210     switch (length) {
 211         default: return false;
 212         // Everything else falls through when "true"...
 213         case 4: if ((a = (*--srcptr)) < 0x80 || a > 0xBF) return false;
 214         case 3: if ((a = (*--srcptr)) < 0x80 || a > 0xBF) return false;
 215         case 2: if ((a = (*--srcptr)) > 0xBF) return false;
 216
 217         switch (*source) {
 218             // no fall-through in this inner switch
 219             case 0xE0: if (a < 0xA0) return false; break;
 220             case 0xED: if (a > 0x9F) return false; break;
 221             case 0xF0: if (a < 0x90) return false; break;
 222             case 0xF4: if (a > 0x8F) return false; break;
 223             default:   if (a < 0x80) return false;
 224         }
 225
 226         case 1: if (*source >= 0x80 && *source < 0xC2) return false;
 227     }
 228     if (*source > 0xF4)
 229         return false;
 230     return true;
 231 }
 232
 233 // Magic values subtracted from a buffer value during UTF8 conversion.
 234 // This table contains as many values as there might be trailing bytes
 235 // in a UTF-8 sequence.
 236 static const UChar32 offsetsFromUTF8[6] = { 0x00000000UL, 0x00003080UL, 0x000E2080UL,
 237             0x03C82080UL, 0xFA082080UL, 0x82082080UL };
 238
 239 static inline UChar32 readUTF8Sequence(const char*& sequence, unsigned length)
 240 {
 241     UChar32 character = 0;
 242
 243     // The cases all fall through.
 244     switch (length) {
 245         case 6: character += static_cast<unsigned char>(*sequence++); character <<= 6;
 246         case 5: character += static_cast<unsigned char>(*sequence++); character <<= 6;
 247         case 4: character += static_cast<unsigned char>(*sequence++); character <<= 6;
 248         case 3: character += static_cast<unsigned char>(*sequence++); character <<= 6;
 249         case 2: character += static_cast<unsigned char>(*sequence++); character <<= 6;
 250         case 1: character += static_cast<unsigned char>(*sequence++);
 251     }
 252
 253     return character - offsetsFromUTF8[length - 1];
 254 }
 255
 256 ConversionResult convertUTF8ToUTF16(
 257     const char** sourceStart, const char* sourceEnd,
 258     UChar** targetStart, UChar* targetEnd, bool strict)
 259 {
 260     ConversionResult result = conversionOK;
 261     const char* source = *sourceStart;
 262     UChar* target = *targetStart;
 263     while (source < sourceEnd) {
 264         int utf8SequenceLength = inlineUTF8SequenceLength(*source);
 265         if (sourceEnd - source < utf8SequenceLength)  {
 266             result = sourceExhausted;
 267             break;
 268         }
 269         // Do this check whether lenient or strict
 270         if (!isLegalUTF8(reinterpret_cast<const unsigned char*>(source), utf8SequenceLength)) {
 271             result = sourceIllegal;
 272             break;
 273         }
 274
 275         UChar32 character = readUTF8Sequence(source, utf8SequenceLength);
 276
 277         if (target >= targetEnd) {
 278             source -= utf8SequenceLength; // Back up source pointer!
 279             result = targetExhausted;
 280             break;
 281         }
 282
 283         if (U_IS_BMP(character)) {
 284             // UTF-16 surrogate values are illegal in UTF-32
 285             if (U_IS_SURROGATE(character)) {
 286                 if (strict) {
 287                     source -= utf8SequenceLength; // return to the illegal value itself
 288                     result = sourceIllegal;
 289                     break;
 290                 } else
 291                     *target++ = replacementCharacter;
 292             } else
 293                 *target++ = character; // normal case
 294         } else if (U_IS_SUPPLEMENTARY(character)) {
 295             // target is a character in range 0xFFFF - 0x10FFFF
 296             if (target + 1 >= targetEnd) {
 297                 source -= utf8SequenceLength; // Back up source pointer!
 298                 result = targetExhausted;
 299                 break;
 300             }
 301             *target++ = U16_LEAD(character);
 302             *target++ = U16_TRAIL(character);
 303         } else {
 304             if (strict) {
 305                 source -= utf8SequenceLength; // return to the start
 306                 result = sourceIllegal;
 307                 break; // Bail out; shouldn't continue
 308             } else
 309                 *target++ = replacementCharacter;
 310         }
 311     }
 312     *sourceStart = source;
 313     *targetStart = target;
 314     return result;
 315 }
 316
 317 unsigned calculateStringHashAndLengthFromUTF8(const char* data, const char* dataEnd, unsigned& dataLength, unsigned& utf16Length)
 318 {
 319     if (!data)
 320         return 0;
 321
 322     StringHasher stringHasher;
 323     dataLength = 0;
 324     utf16Length = 0;
 325
 326     while (data < dataEnd || (!dataEnd && *data)) {
 327         if (isASCII(*data)) {
 328             stringHasher.addCharacter(*data++);
 329             dataLength++;
 330             utf16Length++;
 331             continue;
 332         }
 333
 334         int utf8SequenceLength = inlineUTF8SequenceLengthNonASCII(*data);
 335         dataLength += utf8SequenceLength;
 336
 337         if (!dataEnd) {
 338             for (int i = 1; i < utf8SequenceLength; ++i) {
 339                 if (!data[i])
 340                     return 0;
 341             }
 342         } else if (dataEnd - data < utf8SequenceLength)
 343             return 0;
 344
 345         if (!isLegalUTF8(reinterpret_cast<const unsigned char*>(data), utf8SequenceLength))
 346             return 0;
 347
 348         UChar32 character = readUTF8Sequence(data, utf8SequenceLength);
 349         ASSERT(!isASCII(character));
 350
 351         if (U_IS_BMP(character)) {
 352             // UTF-16 surrogate values are illegal in UTF-32
 353             if (U_IS_SURROGATE(character))
 354                 return 0;
 355             stringHasher.addCharacter(static_cast<UChar>(character)); // normal case
 356             utf16Length++;
 357         } else if (U_IS_SUPPLEMENTARY(character)) {
 358             stringHasher.addCharacters(static_cast<UChar>(U16_LEAD(character)),
 359                                        static_cast<UChar>(U16_TRAIL(character)));
 360             utf16Length += 2;
 361         } else
 362             return 0;
 363     }
 364
 365     return stringHasher.hash();
 366 }
 367
 368 bool equalUTF16WithUTF8(const UChar* a, const UChar* aEnd, const char* b, const char* bEnd)
 369 {
 370     while (b < bEnd) {
 371         if (isASCII(*b)) {
 372             if (*a++ != *b++)
 373                 return false;
 374             continue;
 375         }
 376
 377         int utf8SequenceLength = inlineUTF8SequenceLengthNonASCII(*b);
 378
 379         if (bEnd - b < utf8SequenceLength)
 380             return false;
 381
 382         if (!isLegalUTF8(reinterpret_cast<const unsigned char*>(b), utf8SequenceLength))
 383             return 0;
 384
 385         UChar32 character = readUTF8Sequence(b, utf8SequenceLength);
 386         ASSERT(!isASCII(character));
 387
 388         if (U_IS_BMP(character)) {
 389             // UTF-16 surrogate values are illegal in UTF-32
 390             if (U_IS_SURROGATE(character))
 391                 return false;
 392             if (*a++ != character)
 393                 return false;
 394         } else if (U_IS_SUPPLEMENTARY(character)) {
 395             if (*a++ != U16_LEAD(character))
 396                 return false;
 397             if (*a++ != U16_TRAIL(character))
 398                 return false;
 399         } else
 400             return false;
 401     }
 402
 403     return a == aEnd;
 404 }
 405
 406 } // namespace Unicode
 407 } // namespace WTF