1 /*
2 * Copyright (C) 2007, 2014 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 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 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 <wtf/unicode/UTF8.h>
29
30 #include <wtf/ASCIICType.h>
31 #include <wtf/text/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 convertLatin1ToUTF8(
129 const LChar** sourceStart, const LChar* sourceEnd,
130 char** targetStart, char* targetEnd)
131 {
132 ConversionResult result = conversionOK;
133 const LChar* 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 LChar* oldSource = source; // In case we have to back up because of target overflow.
141 ch = static_cast<unsigned short>(*source++);
142
143 // Figure out how many bytes the result will require
144 if (ch < (UChar32)0x80)
145 bytesToWrite = 1;
146 else
147 bytesToWrite = 2;
148
149 target += bytesToWrite;
150 if (target > targetEnd) {
151 source = oldSource; // Back up source pointer!
152 target -= bytesToWrite;
153 result = targetExhausted;
154 break;
155 }
156 switch (bytesToWrite) { // note: everything falls through.
157 case 2:
158 *--target = (char)((ch | byteMark) & byteMask);
159 ch >>= 6;
160 FALLTHROUGH;
161 case 1:
162 *--target = (char)(ch | firstByteMark[bytesToWrite]);
163 }
164 target += bytesToWrite;
165 }
166 *sourceStart = source;
167 *targetStart = target;
168 return result;
169 }
170
171 ConversionResult convertUTF16ToUTF8(
172 const UChar** sourceStart, const UChar* sourceEnd,
173 char** targetStart, char* targetEnd, bool strict)
174 {
175 ConversionResult result = conversionOK;
176 const UChar* source = *sourceStart;
177 char* target = *targetStart;
178 while (source < sourceEnd) {
179 UChar32 ch;
180 unsigned short bytesToWrite = 0;
181 const UChar32 byteMask = 0xBF;
182 const UChar32 byteMark = 0x80;
183 const UChar* oldSource = source; // In case we have to back up because of target overflow.
184 ch = static_cast<unsigned short>(*source++);
185 // If we have a surrogate pair, convert to UChar32 first.
186 if (ch >= 0xD800 && ch <= 0xDBFF) {
187 // If the 16 bits following the high surrogate are in the source buffer...
188 if (source < sourceEnd) {
189 UChar32 ch2 = static_cast<unsigned short>(*source);
190 // If it's a low surrogate, convert to UChar32.
191 if (ch2 >= 0xDC00 && ch2 <= 0xDFFF) {
192 ch = ((ch - 0xD800) << 10) + (ch2 - 0xDC00) + 0x0010000;
193 ++source;
194 } else if (strict) { // it's an unpaired high surrogate
195 --source; // return to the illegal value itself
196 result = sourceIllegal;
197 break;
198 }
199 } else { // We don't have the 16 bits following the high surrogate.
200 --source; // return to the high surrogate
201 result = sourceExhausted;
202 break;
203 }
204 } else if (strict) {
205 // UTF-16 surrogate values are illegal in UTF-32
206 if (ch >= 0xDC00 && ch <= 0xDFFF) {
207 --source; // return to the illegal value itself
208 result = sourceIllegal;
209 break;
210 }
211 }
212 // Figure out how many bytes the result will require
213 if (ch < (UChar32)0x80) {
214 bytesToWrite = 1;
215 } else if (ch < (UChar32)0x800) {
216 bytesToWrite = 2;
217 } else if (ch < (UChar32)0x10000) {
218 bytesToWrite = 3;
219 } else if (ch < (UChar32)0x110000) {
220 bytesToWrite = 4;
221 } else {
222 bytesToWrite = 3;
223 ch = replacementCharacter;
224 }
225
226 target += bytesToWrite;
227 if (target > targetEnd) {
228 source = oldSource; // Back up source pointer!
229 target -= bytesToWrite;
230 result = targetExhausted;
231 break;
232 }
233 switch (bytesToWrite) { // note: everything falls through.
234 case 4: *--target = (char)((ch | byteMark) & byteMask); ch >>= 6; FALLTHROUGH;
235 case 3: *--target = (char)((ch | byteMark) & byteMask); ch >>= 6; FALLTHROUGH;
236 case 2: *--target = (char)((ch | byteMark) & byteMask); ch >>= 6; FALLTHROUGH;
237 case 1: *--target = (char)(ch | firstByteMark[bytesToWrite]);
238 }
239 target += bytesToWrite;
240 }
241 *sourceStart = source;
242 *targetStart = target;
243 return result;
244 }
245
246 // This must be called with the length pre-determined by the first byte.
247 // If presented with a length > 4, this returns false. The Unicode
248 // definition of UTF-8 goes up to 4-byte sequences.
249 static bool isLegalUTF8(const unsigned char* source, int length)
250 {
251 unsigned char a;
252 const unsigned char* srcptr = source + length;
253 switch (length) {
254 default: return false;
255 // Everything else falls through when "true"...
256 case 4: if ((a = (*--srcptr)) < 0x80 || a > 0xBF) return false; FALLTHROUGH;
257 case 3: if ((a = (*--srcptr)) < 0x80 || a > 0xBF) return false; FALLTHROUGH;
258 case 2: if ((a = (*--srcptr)) > 0xBF) return false;
259
260 switch (*source) {
261 // no fall-through in this inner switch
262 case 0xE0: if (a < 0xA0) return false; break;
263 case 0xED: if (a > 0x9F) return false; break;
264 case 0xF0: if (a < 0x90) return false; break;
265 case 0xF4: if (a > 0x8F) return false; break;
266 default: if (a < 0x80) return false;
267 }
268 FALLTHROUGH;
269
270 case 1: if (*source >= 0x80 && *source < 0xC2) return false;
271 }
272 if (*source > 0xF4)
273 return false;
274 return true;
275 }
276
277 // Magic values subtracted from a buffer value during UTF8 conversion.
278 // This table contains as many values as there might be trailing bytes
279 // in a UTF-8 sequence.
280 static const UChar32 offsetsFromUTF8[6] = { 0x00000000UL, 0x00003080UL, 0x000E2080UL, 0x03C82080UL, static_cast<UChar32>(0xFA082080UL), static_cast<UChar32>(0x82082080UL) };
281
282 static inline UChar32 readUTF8Sequence(const char*& sequence, unsigned length)
283 {
284 UChar32 character = 0;
285
286 // The cases all fall through.
287 switch (length) {
288 case 6: character += static_cast<unsigned char>(*sequence++); character <<= 6; FALLTHROUGH;
289 case 5: character += static_cast<unsigned char>(*sequence++); character <<= 6; FALLTHROUGH;
290 case 4: character += static_cast<unsigned char>(*sequence++); character <<= 6; FALLTHROUGH;
291 case 3: character += static_cast<unsigned char>(*sequence++); character <<= 6; FALLTHROUGH;
292 case 2: character += static_cast<unsigned char>(*sequence++); character <<= 6; FALLTHROUGH;
293 case 1: character += static_cast<unsigned char>(*sequence++);
294 }
295
296 return character - offsetsFromUTF8[length - 1];
297 }
298
299 ConversionResult convertUTF8ToUTF16(
300 const char** sourceStart, const char* sourceEnd,
301 UChar** targetStart, UChar* targetEnd, bool* sourceAllASCII, bool strict)
302 {
303 ConversionResult result = conversionOK;
304 const char* source = *sourceStart;
305 UChar* target = *targetStart;
306 UChar orAllData = 0;
307 while (source < sourceEnd) {
308 int utf8SequenceLength = inlineUTF8SequenceLength(*source);
309 if (sourceEnd - source < utf8SequenceLength) {
310 result = sourceExhausted;
311 break;
312 }
313 // Do this check whether lenient or strict
314 if (!isLegalUTF8(reinterpret_cast<const unsigned char*>(source), utf8SequenceLength)) {
315 result = sourceIllegal;
316 break;
317 }
318
319 UChar32 character = readUTF8Sequence(source, utf8SequenceLength);
320
321 if (target >= targetEnd) {
322 source -= utf8SequenceLength; // Back up source pointer!
323 result = targetExhausted;
324 break;
325 }
326
327 if (U_IS_BMP(character)) {
328 // UTF-16 surrogate values are illegal in UTF-32
329 if (U_IS_SURROGATE(character)) {
330 if (strict) {
331 source -= utf8SequenceLength; // return to the illegal value itself
332 result = sourceIllegal;
333 break;
334 } else {
335 *target++ = replacementCharacter;
336 orAllData |= replacementCharacter;
337 }
338 } else {
339 *target++ = character; // normal case
340 orAllData |= character;
341 }
342 } else if (U_IS_SUPPLEMENTARY(character)) {
343 // target is a character in range 0xFFFF - 0x10FFFF
344 if (target + 1 >= targetEnd) {
345 source -= utf8SequenceLength; // Back up source pointer!
346 result = targetExhausted;
347 break;
348 }
349 *target++ = U16_LEAD(character);
350 *target++ = U16_TRAIL(character);
351 orAllData = 0xffff;
352 } else {
353 if (strict) {
354 source -= utf8SequenceLength; // return to the start
355 result = sourceIllegal;
356 break; // Bail out; shouldn't continue
357 } else {
358 *target++ = replacementCharacter;
359 orAllData |= replacementCharacter;
360 }
361 }
362 }
363 *sourceStart = source;
364 *targetStart = target;
365
366 if (sourceAllASCII)
367 *sourceAllASCII = !(orAllData & ~0x7f);
368
369 return result;
370 }
371
372 unsigned calculateStringHashAndLengthFromUTF8MaskingTop8Bits(const char* data, const char* dataEnd, unsigned& dataLength, unsigned& utf16Length)
373 {
374 if (!data)
375 return 0;
376
377 StringHasher stringHasher;
378 dataLength = 0;
379 utf16Length = 0;
380
381 while (data < dataEnd || (!dataEnd && *data)) {
382 if (isASCII(*data)) {
383 stringHasher.addCharacter(*data++);
384 dataLength++;
385 utf16Length++;
386 continue;
387 }
388
389 int utf8SequenceLength = inlineUTF8SequenceLengthNonASCII(*data);
390 dataLength += utf8SequenceLength;
391
392 if (!dataEnd) {
393 for (int i = 1; i < utf8SequenceLength; ++i) {
394 if (!data[i])
395 return 0;
396 }
397 } else if (dataEnd - data < utf8SequenceLength)
398 return 0;
399
400 if (!isLegalUTF8(reinterpret_cast<const unsigned char*>(data), utf8SequenceLength))
401 return 0;
402
403 UChar32 character = readUTF8Sequence(data, utf8SequenceLength);
404 ASSERT(!isASCII(character));
405
406 if (U_IS_BMP(character)) {
407 // UTF-16 surrogate values are illegal in UTF-32
408 if (U_IS_SURROGATE(character))
409 return 0;
410 stringHasher.addCharacter(static_cast<UChar>(character)); // normal case
411 utf16Length++;
412 } else if (U_IS_SUPPLEMENTARY(character)) {
413 stringHasher.addCharacters(static_cast<UChar>(U16_LEAD(character)),
414 static_cast<UChar>(U16_TRAIL(character)));
415 utf16Length += 2;
416 } else
417 return 0;
418 }
419
420 return stringHasher.hashWithTop8BitsMasked();
421 }
422
423 bool equalUTF16WithUTF8(const UChar* a, const char* b, const char* bEnd)
424 {
425 while (b < bEnd) {
426 if (isASCII(*a) || isASCII(*b)) {
427 if (*a++ != *b++)
428 return false;
429 continue;
430 }
431
432 int utf8SequenceLength = inlineUTF8SequenceLengthNonASCII(*b);
433
434 if (bEnd - b < utf8SequenceLength)
435 return false;
436
437 if (!isLegalUTF8(reinterpret_cast<const unsigned char*>(b), utf8SequenceLength))
438 return false;
439
440 UChar32 character = readUTF8Sequence(b, utf8SequenceLength);
441 ASSERT(!isASCII(character));
442
443 if (U_IS_BMP(character)) {
444 // UTF-16 surrogate values are illegal in UTF-32
445 if (U_IS_SURROGATE(character))
446 return false;
447 if (*a++ != character)
448 return false;
449 } else if (U_IS_SUPPLEMENTARY(character)) {
450 if (*a++ != U16_LEAD(character))
451 return false;
452 if (*a++ != U16_TRAIL(character))
453 return false;
454 } else
455 return false;
456 }
457
458 return true;
459 }
460
461 bool equalLatin1WithUTF8(const LChar* a, const char* b, const char* bEnd)
462 {
463 while (b < bEnd) {
464 if (isASCII(*a) || isASCII(*b)) {
465 if (*a++ != *b++)
466 return false;
467 continue;
468 }
469
470 if (b + 1 == bEnd)
471 return false;
472
473 if ((b[0] & 0xE0) != 0xC0 || (b[1] & 0xC0) != 0x80)
474 return false;
475
476 LChar character = ((b[0] & 0x1F) << 6) | (b[1] & 0x3F);
477
478 b += 2;
479
480 if (*a++ != character)
481 return false;
482 }
483
484 return true;
485 }
486
487 } // namespace Unicode
488 } // namespace WTF