wtf/unicode/UTF8.cpp

   1 /*
   2  * Copyright (C) 2007 Apple Inc.  All rights reserved.
   3  *
   4  * Redistribution and use in source and binary forms, with or without
   5  * modification, are permitted provided that the following conditions
   6  * are met:
   7  * 1. Redistributions of source code must retain the above copyright
   8  *    notice, this list of conditions and the following disclaimer.
   9  * 2. Redistributions in binary form must reproduce the above copyright
  10  *    notice, this list of conditions and the following disclaimer in the
  11  *    documentation and/or other materials provided with the distribution.
  12  *
  13  * THIS SOFTWARE IS PROVIDED BY APPLE COMPUTER, INC. ``AS IS'' AND ANY
  14  * EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
  15  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
  16  * PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL APPLE COMPUTER, INC. OR
  17  * CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
  18  * EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
  19  * PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
  20  * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY
  21  * OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
  22  * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
  23  * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
  24  */
  25
  26 #include "UTF8.h"
  27
  28 namespace WTF {
  29 namespace Unicode {
  30
  31 inline int inlineUTF8SequenceLengthNonASCII(char b0)
  32 {
  33     if ((b0 & 0xC0) != 0xC0)
  34         return 0;
  35     if ((b0 & 0xE0) == 0xC0)
  36         return 2;
  37     if ((b0 & 0xF0) == 0xE0)
  38         return 3;
  39     if ((b0 & 0xF8) == 0xF0)
  40         return 4;
  41     return 0;
  42 }
  43
  44 inline int inlineUTF8SequenceLength(char b0)
  45 {
  46     return (b0 & 0x80) == 0 ? 1 : inlineUTF8SequenceLengthNonASCII(b0);
  47 }
  48
  49 int UTF8SequenceLength(char b0)
  50 {
  51     return (b0 & 0x80) == 0 ? 1 : inlineUTF8SequenceLengthNonASCII(b0);
  52 }
  53
  54 int decodeUTF8Sequence(const char* sequence)
  55 {
  56     // Handle 0-byte sequences (never valid).
  57     const unsigned char b0 = sequence[0];
  58     const int length = inlineUTF8SequenceLength(b0);
  59     if (length == 0)
  60         return -1;
  61
  62     // Handle 1-byte sequences (plain ASCII).
  63     const unsigned char b1 = sequence[1];
  64     if (length == 1) {
  65         if (b1)
  66             return -1;
  67         return b0;
  68     }
  69
  70     // Handle 2-byte sequences.
  71     if ((b1 & 0xC0) != 0x80)
  72         return -1;
  73     const unsigned char b2 = sequence[2];
  74     if (length == 2) {
  75         if (b2)
  76             return -1;
  77         const int c = ((b0 & 0x1F) << 6) | (b1 & 0x3F);
  78         if (c < 0x80)
  79             return -1;
  80         return c;
  81     }
  82
  83     // Handle 3-byte sequences.
  84     if ((b2 & 0xC0) != 0x80)
  85         return -1;
  86     const unsigned char b3 = sequence[3];
  87     if (length == 3) {
  88         if (b3)
  89             return -1;
  90         const int c = ((b0 & 0xF) << 12) | ((b1 & 0x3F) << 6) | (b2 & 0x3F);
  91         if (c < 0x800)
  92             return -1;
  93         // UTF-16 surrogates should never appear in UTF-8 data.
  94         if (c >= 0xD800 && c <= 0xDFFF)
  95             return -1;
  96         return c;
  97     }
  98
  99     // Handle 4-byte sequences.
 100     if ((b3 & 0xC0) != 0x80)
 101         return -1;
 102     const unsigned char b4 = sequence[4];
 103     if (length == 4) {
 104         if (b4)
 105             return -1;
 106         const int c = ((b0 & 0x7) << 18) | ((b1 & 0x3F) << 12) | ((b2 & 0x3F) << 6) | (b3 & 0x3F);
 107         if (c < 0x10000 || c > 0x10FFFF)
 108             return -1;
 109         return c;
 110     }
 111
 112     return -1;
 113 }
 114
 115 // Once the bits are split out into bytes of UTF-8, this is a mask OR-ed
 116 // into the first byte, depending on how many bytes follow.  There are
 117 // as many entries in this table as there are UTF-8 sequence types.
 118 // (I.e., one byte sequence, two byte... etc.). Remember that sequencs
 119 // for *legal* UTF-8 will be 4 or fewer bytes total.
 120 static const unsigned char firstByteMark[7] = { 0x00, 0x00, 0xC0, 0xE0, 0xF0, 0xF8, 0xFC };
 121
 122 ConversionResult convertUTF16ToUTF8(
 123     const UChar** sourceStart, const UChar* sourceEnd,
 124     char** targetStart, char* targetEnd, bool strict)
 125 {
 126     ConversionResult result = conversionOK;
 127     const UChar* source = *sourceStart;
 128     char* target = *targetStart;
 129     while (source < sourceEnd) {
 130         UChar32 ch;
 131         unsigned short bytesToWrite = 0;
 132         const UChar32 byteMask = 0xBF;
 133         const UChar32 byteMark = 0x80;
 134         const UChar* oldSource = source; // In case we have to back up because of target overflow.
 135         ch = static_cast<unsigned short>(*source++);
 136         // If we have a surrogate pair, convert to UChar32 first.
 137         if (ch >= 0xD800 && ch <= 0xDBFF) {
 138             // If the 16 bits following the high surrogate are in the source buffer...
 139             if (source < sourceEnd) {
 140                 UChar32 ch2 = static_cast<unsigned short>(*source);
 141                 // If it's a low surrogate, convert to UChar32.
 142                 if (ch2 >= 0xDC00 && ch2 <= 0xDFFF) {
 143                     ch = ((ch - 0xD800) << 10) + (ch2 - 0xDC00) + 0x0010000;
 144                     ++source;
 145                 } else if (strict) { // it's an unpaired high surrogate
 146                     --source; // return to the illegal value itself
 147                     result = sourceIllegal;
 148                     break;
 149                 }
 150             } else { // We don't have the 16 bits following the high surrogate.
 151                 --source; // return to the high surrogate
 152                 result = sourceExhausted;
 153                 break;
 154             }
 155         } else if (strict) {
 156             // UTF-16 surrogate values are illegal in UTF-32
 157             if (ch >= 0xDC00 && ch <= 0xDFFF) {
 158                 --source; // return to the illegal value itself
 159                 result = sourceIllegal;
 160                 break;
 161             }
 162         }
 163         // Figure out how many bytes the result will require
 164         if (ch < (UChar32)0x80) {
 165             bytesToWrite = 1;
 166         } else if (ch < (UChar32)0x800) {
 167             bytesToWrite = 2;
 168         } else if (ch < (UChar32)0x10000) {
 169             bytesToWrite = 3;
 170         } else if (ch < (UChar32)0x110000) {
 171             bytesToWrite = 4;
 172         } else {
 173             bytesToWrite = 3;
 174             ch = 0xFFFD;
 175         }
 176
 177         target += bytesToWrite;
 178         if (target > targetEnd) {
 179             source = oldSource; // Back up source pointer!
 180             target -= bytesToWrite;
 181             result = targetExhausted;
 182             break;
 183         }
 184         switch (bytesToWrite) { // note: everything falls through.
 185             case 4: *--target = (char)((ch | byteMark) & byteMask); ch >>= 6;
 186             case 3: *--target = (char)((ch | byteMark) & byteMask); ch >>= 6;
 187             case 2: *--target = (char)((ch | byteMark) & byteMask); ch >>= 6;
 188             case 1: *--target =  (char)(ch | firstByteMark[bytesToWrite]);
 189         }
 190         target += bytesToWrite;
 191     }
 192     *sourceStart = source;
 193     *targetStart = target;
 194     return result;
 195 }
 196
 197 // This must be called with the length pre-determined by the first byte.
 198 // If presented with a length > 4, this returns false.  The Unicode
 199 // definition of UTF-8 goes up to 4-byte sequences.
 200 static bool isLegalUTF8(const unsigned char* source, int length)
 201 {
 202     unsigned char a;
 203     const unsigned char* srcptr = source + length;
 204     switch (length) {
 205         default: return false;
 206         // Everything else falls through when "true"...
 207         case 4: if ((a = (*--srcptr)) < 0x80 || a > 0xBF) return false;
 208         case 3: if ((a = (*--srcptr)) < 0x80 || a > 0xBF) return false;
 209         case 2: if ((a = (*--srcptr)) > 0xBF) return false;
 210
 211         switch (*source) {
 212             // no fall-through in this inner switch
 213             case 0xE0: if (a < 0xA0) return false; break;
 214             case 0xED: if (a > 0x9F) return false; break;
 215             case 0xF0: if (a < 0x90) return false; break;
 216             case 0xF4: if (a > 0x8F) return false; break;
 217             default:   if (a < 0x80) return false;
 218         }
 219
 220         case 1: if (*source >= 0x80 && *source < 0xC2) return false;
 221     }
 222     if (*source > 0xF4)
 223         return false;
 224     return true;
 225 }
 226
 227 // Magic values subtracted from a buffer value during UTF8 conversion.
 228 // This table contains as many values as there might be trailing bytes
 229 // in a UTF-8 sequence.
 230 static const UChar32 offsetsFromUTF8[6] = { 0x00000000UL, 0x00003080UL, 0x000E2080UL,
 231             0x03C82080UL, 0xFA082080UL, 0x82082080UL };
 232
 233 ConversionResult convertUTF8ToUTF16(
 234     const char** sourceStart, const char* sourceEnd,
 235     UChar** targetStart, UChar* targetEnd, bool strict)
 236 {
 237     ConversionResult result = conversionOK;
 238     const char* source = *sourceStart;
 239     UChar* target = *targetStart;
 240     while (source < sourceEnd) {
 241         UChar32 ch = 0;
 242         int extraBytesToRead = UTF8SequenceLength(*source) - 1;
 243         if (source + extraBytesToRead >= sourceEnd) {
 244             result = sourceExhausted;
 245             break;
 246         }
 247         // Do this check whether lenient or strict
 248         if (!isLegalUTF8(reinterpret_cast<const unsigned char*>(source), extraBytesToRead + 1)) {
 249             result = sourceIllegal;
 250             break;
 251         }
 252         // The cases all fall through.
 253         switch (extraBytesToRead) {
 254             case 5: ch += static_cast<unsigned char>(*source++); ch <<= 6; // remember, illegal UTF-8
 255             case 4: ch += static_cast<unsigned char>(*source++); ch <<= 6; // remember, illegal UTF-8
 256             case 3: ch += static_cast<unsigned char>(*source++); ch <<= 6;
 257             case 2: ch += static_cast<unsigned char>(*source++); ch <<= 6;
 258             case 1: ch += static_cast<unsigned char>(*source++); ch <<= 6;
 259             case 0: ch += static_cast<unsigned char>(*source++);
 260         }
 261         ch -= offsetsFromUTF8[extraBytesToRead];
 262
 263         if (target >= targetEnd) {
 264             source -= (extraBytesToRead + 1); // Back up source pointer!
 265             result = targetExhausted; break;
 266         }
 267         if (ch <= 0xFFFF) {
 268             // UTF-16 surrogate values are illegal in UTF-32
 269             if (ch >= 0xD800 && ch <= 0xDFFF) {
 270                 if (strict) {
 271                     source -= (extraBytesToRead + 1); // return to the illegal value itself
 272                     result = sourceIllegal;
 273                     break;
 274                 } else
 275                     *target++ = 0xFFFD;
 276             } else
 277                 *target++ = (UChar)ch; // normal case
 278         } else if (ch > 0x10FFFF) {
 279             if (strict) {
 280                 result = sourceIllegal;
 281                 source -= (extraBytesToRead + 1); // return to the start
 282                 break; // Bail out; shouldn't continue
 283             } else
 284                 *target++ = 0xFFFD;
 285         } else {
 286             // target is a character in range 0xFFFF - 0x10FFFF
 287             if (target + 1 >= targetEnd) {
 288                 source -= (extraBytesToRead + 1); // Back up source pointer!
 289                 result = targetExhausted;
 290                 break;
 291             }
 292             ch -= 0x0010000UL;
 293             *target++ = (UChar)((ch >> 10) + 0xD800);
 294             *target++ = (UChar)((ch & 0x03FF) + 0xDC00);
 295         }
 296     }
 297     *sourceStart = source;
 298     *targetStart = target;
 299     return result;
 300 }
 301
 302 }
 303 }