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 "wtf/unicode/UTF8.h"
29
30#include "wtf/ASCIICType.h"
31#include "wtf/StringHasher.h"
32#include "wtf/unicode/CharacterNames.h"
33
34namespace WTF {
35namespace Unicode {
36
37inline 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
50inline int inlineUTF8SequenceLength(char b0)
51{
52    return isASCII(b0) ? 1 : inlineUTF8SequenceLengthNonASCII(b0);
53}
54
55int UTF8SequenceLength(char b0)
56{
57    return isASCII(b0) ? 1 : inlineUTF8SequenceLengthNonASCII(b0);
58}
59
60int 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.
126static const unsigned char firstByteMark[7] = { 0x00, 0x00, 0xC0, 0xE0, 0xF0, 0xF8, 0xFC };
127
128ConversionResult 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        case 1:
161            *--target =  (char)(ch | firstByteMark[bytesToWrite]);
162        }
163        target += bytesToWrite;
164    }
165    *sourceStart = source;
166    *targetStart = target;
167    return result;
168}
169
170ConversionResult convertUTF16ToUTF8(
171    const UChar** sourceStart, const UChar* sourceEnd,
172    char** targetStart, char* targetEnd, bool strict)
173{
174    ConversionResult result = conversionOK;
175    const UChar* source = *sourceStart;
176    char* target = *targetStart;
177    while (source < sourceEnd) {
178        UChar32 ch;
179        unsigned short bytesToWrite = 0;
180        const UChar32 byteMask = 0xBF;
181        const UChar32 byteMark = 0x80;
182        const UChar* oldSource = source; // In case we have to back up because of target overflow.
183        ch = static_cast<unsigned short>(*source++);
184        // If we have a surrogate pair, convert to UChar32 first.
185        if (ch >= 0xD800 && ch <= 0xDBFF) {
186            // If the 16 bits following the high surrogate are in the source buffer...
187            if (source < sourceEnd) {
188                UChar32 ch2 = static_cast<unsigned short>(*source);
189                // If it's a low surrogate, convert to UChar32.
190                if (ch2 >= 0xDC00 && ch2 <= 0xDFFF) {
191                    ch = ((ch - 0xD800) << 10) + (ch2 - 0xDC00) + 0x0010000;
192                    ++source;
193                } else if (strict) { // it's an unpaired high surrogate
194                    --source; // return to the illegal value itself
195                    result = sourceIllegal;
196                    break;
197                }
198            } else { // We don't have the 16 bits following the high surrogate.
199                --source; // return to the high surrogate
200                result = sourceExhausted;
201                break;
202            }
203        } else if (strict) {
204            // UTF-16 surrogate values are illegal in UTF-32
205            if (ch >= 0xDC00 && ch <= 0xDFFF) {
206                --source; // return to the illegal value itself
207                result = sourceIllegal;
208                break;
209            }
210        }
211        // Figure out how many bytes the result will require
212        if (ch < (UChar32)0x80) {
213            bytesToWrite = 1;
214        } else if (ch < (UChar32)0x800) {
215            bytesToWrite = 2;
216        } else if (ch < (UChar32)0x10000) {
217            bytesToWrite = 3;
218        } else if (ch < (UChar32)0x110000) {
219            bytesToWrite = 4;
220        } else {
221            bytesToWrite = 3;
222            ch = replacementCharacter;
223        }
224
225        target += bytesToWrite;
226        if (target > targetEnd) {
227            source = oldSource; // Back up source pointer!
228            target -= bytesToWrite;
229            result = targetExhausted;
230            break;
231        }
232        switch (bytesToWrite) { // note: everything falls through.
233            case 4: *--target = (char)((ch | byteMark) & byteMask); ch >>= 6;
234            case 3: *--target = (char)((ch | byteMark) & byteMask); ch >>= 6;
235            case 2: *--target = (char)((ch | byteMark) & byteMask); ch >>= 6;
236            case 1: *--target =  (char)(ch | firstByteMark[bytesToWrite]);
237        }
238        target += bytesToWrite;
239    }
240    *sourceStart = source;
241    *targetStart = target;
242    return result;
243}
244
245// This must be called with the length pre-determined by the first byte.
246// If presented with a length > 4, this returns false.  The Unicode
247// definition of UTF-8 goes up to 4-byte sequences.
248static bool isLegalUTF8(const unsigned char* source, int length)
249{
250    unsigned char a;
251    const unsigned char* srcptr = source + length;
252    switch (length) {
253        default: return false;
254        // Everything else falls through when "true"...
255        case 4: if ((a = (*--srcptr)) < 0x80 || a > 0xBF) return false;
256        case 3: if ((a = (*--srcptr)) < 0x80 || a > 0xBF) return false;
257        case 2: if ((a = (*--srcptr)) > 0xBF) return false;
258
259        switch (*source) {
260            // no fall-through in this inner switch
261            case 0xE0: if (a < 0xA0) return false; break;
262            case 0xED: if (a > 0x9F) return false; break;
263            case 0xF0: if (a < 0x90) return false; break;
264            case 0xF4: if (a > 0x8F) return false; break;
265            default:   if (a < 0x80) return false;
266        }
267
268        case 1: if (*source >= 0x80 && *source < 0xC2) return false;
269    }
270    if (*source > 0xF4)
271        return false;
272    return true;
273}
274
275// Magic values subtracted from a buffer value during UTF8 conversion.
276// This table contains as many values as there might be trailing bytes
277// in a UTF-8 sequence.
278static const UChar32 offsetsFromUTF8[6] = { 0x00000000UL, 0x00003080UL, 0x000E2080UL, 0x03C82080UL, static_cast<UChar32>(0xFA082080UL), static_cast<UChar32>(0x82082080UL) };
279
280static inline UChar32 readUTF8Sequence(const char*& sequence, unsigned length)
281{
282    UChar32 character = 0;
283
284    // The cases all fall through.
285    switch (length) {
286        case 6: character += static_cast<unsigned char>(*sequence++); character <<= 6;
287        case 5: character += static_cast<unsigned char>(*sequence++); character <<= 6;
288        case 4: character += static_cast<unsigned char>(*sequence++); character <<= 6;
289        case 3: character += static_cast<unsigned char>(*sequence++); character <<= 6;
290        case 2: character += static_cast<unsigned char>(*sequence++); character <<= 6;
291        case 1: character += static_cast<unsigned char>(*sequence++);
292    }
293
294    return character - offsetsFromUTF8[length - 1];
295}
296
297ConversionResult convertUTF8ToUTF16(
298    const char** sourceStart, const char* sourceEnd,
299    UChar** targetStart, UChar* targetEnd, bool* sourceAllASCII, bool strict)
300{
301    ConversionResult result = conversionOK;
302    const char* source = *sourceStart;
303    UChar* target = *targetStart;
304    UChar orAllData = 0;
305    while (source < sourceEnd) {
306        int utf8SequenceLength = inlineUTF8SequenceLength(*source);
307        if (sourceEnd - source < utf8SequenceLength)  {
308            result = sourceExhausted;
309            break;
310        }
311        // Do this check whether lenient or strict
312        if (!isLegalUTF8(reinterpret_cast<const unsigned char*>(source), utf8SequenceLength)) {
313            result = sourceIllegal;
314            break;
315        }
316
317        UChar32 character = readUTF8Sequence(source, utf8SequenceLength);
318
319        if (target >= targetEnd) {
320            source -= utf8SequenceLength; // Back up source pointer!
321            result = targetExhausted;
322            break;
323        }
324
325        if (U_IS_BMP(character)) {
326            // UTF-16 surrogate values are illegal in UTF-32
327            if (U_IS_SURROGATE(character)) {
328                if (strict) {
329                    source -= utf8SequenceLength; // return to the illegal value itself
330                    result = sourceIllegal;
331                    break;
332                } else {
333                    *target++ = replacementCharacter;
334                    orAllData |= replacementCharacter;
335                }
336            } else {
337                *target++ = character; // normal case
338                orAllData |= character;
339            }
340        } else if (U_IS_SUPPLEMENTARY(character)) {
341            // target is a character in range 0xFFFF - 0x10FFFF
342            if (target + 1 >= targetEnd) {
343                source -= utf8SequenceLength; // Back up source pointer!
344                result = targetExhausted;
345                break;
346            }
347            *target++ = U16_LEAD(character);
348            *target++ = U16_TRAIL(character);
349            orAllData = 0xffff;
350        } else {
351            if (strict) {
352                source -= utf8SequenceLength; // return to the start
353                result = sourceIllegal;
354                break; // Bail out; shouldn't continue
355            } else {
356                *target++ = replacementCharacter;
357                orAllData |= replacementCharacter;
358            }
359        }
360    }
361    *sourceStart = source;
362    *targetStart = target;
363
364    if (sourceAllASCII)
365        *sourceAllASCII = !(orAllData & ~0x7f);
366
367    return result;
368}
369
370unsigned calculateStringHashAndLengthFromUTF8MaskingTop8Bits(const char* data, const char* dataEnd, unsigned& dataLength, unsigned& utf16Length)
371{
372    if (!data)
373        return 0;
374
375    StringHasher stringHasher;
376    dataLength = 0;
377    utf16Length = 0;
378
379    while (data < dataEnd || (!dataEnd && *data)) {
380        if (isASCII(*data)) {
381            stringHasher.addCharacter(*data++);
382            dataLength++;
383            utf16Length++;
384            continue;
385        }
386
387        int utf8SequenceLength = inlineUTF8SequenceLengthNonASCII(*data);
388        dataLength += utf8SequenceLength;
389
390        if (!dataEnd) {
391            for (int i = 1; i < utf8SequenceLength; ++i) {
392                if (!data[i])
393                    return 0;
394            }
395        } else if (dataEnd - data < utf8SequenceLength)
396            return 0;
397
398        if (!isLegalUTF8(reinterpret_cast<const unsigned char*>(data), utf8SequenceLength))
399            return 0;
400
401        UChar32 character = readUTF8Sequence(data, utf8SequenceLength);
402        ASSERT(!isASCII(character));
403
404        if (U_IS_BMP(character)) {
405            // UTF-16 surrogate values are illegal in UTF-32
406            if (U_IS_SURROGATE(character))
407                return 0;
408            stringHasher.addCharacter(static_cast<UChar>(character)); // normal case
409            utf16Length++;
410        } else if (U_IS_SUPPLEMENTARY(character)) {
411            stringHasher.addCharacters(static_cast<UChar>(U16_LEAD(character)),
412                                       static_cast<UChar>(U16_TRAIL(character)));
413            utf16Length += 2;
414        } else
415            return 0;
416    }
417
418    return stringHasher.hashWithTop8BitsMasked();
419}
420
421template<typename CharType>
422ALWAYS_INLINE bool equalWithUTF8Internal(const CharType* a, const CharType* aEnd, const char* b, const char* bEnd)
423{
424    while (b < bEnd) {
425        if (isASCII(*b)) {
426            if (*a++ != *b++)
427                return false;
428            continue;
429        }
430
431        int utf8SequenceLength = inlineUTF8SequenceLengthNonASCII(*b);
432
433        if (bEnd - b < utf8SequenceLength)
434            return false;
435
436        if (!isLegalUTF8(reinterpret_cast<const unsigned char*>(b), utf8SequenceLength))
437            return 0;
438
439        UChar32 character = readUTF8Sequence(b, utf8SequenceLength);
440        ASSERT(!isASCII(character));
441
442        if (U_IS_BMP(character)) {
443            // UTF-16 surrogate values are illegal in UTF-32
444            if (U_IS_SURROGATE(character))
445                return false;
446            if (*a++ != character)
447                return false;
448        } else if (U_IS_SUPPLEMENTARY(character)) {
449            if (*a++ != U16_LEAD(character))
450                return false;
451            if (*a++ != U16_TRAIL(character))
452                return false;
453        } else
454            return false;
455    }
456
457    return a == aEnd;
458}
459
460bool equalUTF16WithUTF8(const UChar* a, const UChar* aEnd, const char* b, const char* bEnd)
461{
462    return equalWithUTF8Internal(a, aEnd, b, bEnd);
463}
464
465bool equalLatin1WithUTF8(const LChar* a, const LChar* aEnd, const char* b, const char* bEnd)
466{
467    return equalWithUTF8Internal(a, aEnd, b, bEnd);
468}
469
470} // namespace Unicode
471} // namespace WTF
472