1/*
2 * Copyright (C) 2004, 2006, 2008, 2011 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 "config.h"
27#include "TextCodecUTF8.h"
28
29#include "TextCodecASCIIFastPath.h"
30#include <wtf/text/CString.h>
31#include <wtf/text/StringBuffer.h>
32#include <wtf/unicode/CharacterNames.h>
33
34using namespace WTF::Unicode;
35using namespace std;
36
37namespace WebCore {
38
39const int nonCharacter = -1;
40
41PassOwnPtr<TextCodec> TextCodecUTF8::create(const TextEncoding&, const void*)
42{
43    return adoptPtr(new TextCodecUTF8);
44}
45
46void TextCodecUTF8::registerEncodingNames(EncodingNameRegistrar registrar)
47{
48    registrar("UTF-8", "UTF-8");
49
50    // Additional aliases that originally were present in the encoding
51    // table in WebKit on Macintosh, and subsequently added by
52    // TextCodecICU. Perhaps we can prove some are not used on the web
53    // and remove them.
54    registrar("unicode11utf8", "UTF-8");
55    registrar("unicode20utf8", "UTF-8");
56    registrar("utf8", "UTF-8");
57    registrar("x-unicode20utf8", "UTF-8");
58}
59
60void TextCodecUTF8::registerCodecs(TextCodecRegistrar registrar)
61{
62    registrar("UTF-8", create, 0);
63}
64
65static inline int nonASCIISequenceLength(uint8_t firstByte)
66{
67    static const uint8_t lengths[256] = {
68        0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
69        0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
70        0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
71        0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
72        0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
73        0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
74        0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
75        0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
76        2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2,
77        2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2,
78        2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2,
79        2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2,
80        2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2,
81        2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2,
82        3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3,
83        4, 4, 4, 4, 4, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0
84    };
85    return lengths[firstByte];
86}
87
88static inline int decodeNonASCIISequence(const uint8_t* sequence, unsigned length)
89{
90    ASSERT(!isASCII(sequence[0]));
91    if (length == 2) {
92        ASSERT(sequence[0] <= 0xDF);
93        if (sequence[0] < 0xC2)
94            return nonCharacter;
95        if (sequence[1] < 0x80 || sequence[1] > 0xBF)
96            return nonCharacter;
97        return ((sequence[0] << 6) + sequence[1]) - 0x00003080;
98    }
99    if (length == 3) {
100        ASSERT(sequence[0] >= 0xE0 && sequence[0] <= 0xEF);
101        switch (sequence[0]) {
102        case 0xE0:
103            if (sequence[1] < 0xA0 || sequence[1] > 0xBF)
104                return nonCharacter;
105            break;
106        case 0xED:
107            if (sequence[1] < 0x80 || sequence[1] > 0x9F)
108                return nonCharacter;
109            break;
110        default:
111            if (sequence[1] < 0x80 || sequence[1] > 0xBF)
112                return nonCharacter;
113        }
114        if (sequence[2] < 0x80 || sequence[2] > 0xBF)
115            return nonCharacter;
116        return ((sequence[0] << 12) + (sequence[1] << 6) + sequence[2]) - 0x000E2080;
117    }
118    ASSERT(length == 4);
119    ASSERT(sequence[0] >= 0xF0 && sequence[0] <= 0xF4);
120    switch (sequence[0]) {
121    case 0xF0:
122        if (sequence[1] < 0x90 || sequence[1] > 0xBF)
123            return nonCharacter;
124        break;
125    case 0xF4:
126        if (sequence[1] < 0x80 || sequence[1] > 0x8F)
127            return nonCharacter;
128        break;
129    default:
130        if (sequence[1] < 0x80 || sequence[1] > 0xBF)
131            return nonCharacter;
132    }
133    if (sequence[2] < 0x80 || sequence[2] > 0xBF)
134        return nonCharacter;
135    if (sequence[3] < 0x80 || sequence[3] > 0xBF)
136        return nonCharacter;
137    return ((sequence[0] << 18) + (sequence[1] << 12) + (sequence[2] << 6) + sequence[3]) - 0x03C82080;
138}
139
140static inline UChar* appendCharacter(UChar* destination, int character)
141{
142    ASSERT(character != nonCharacter);
143    ASSERT(!U_IS_SURROGATE(character));
144    if (U_IS_BMP(character))
145        *destination++ = character;
146    else {
147        *destination++ = U16_LEAD(character);
148        *destination++ = U16_TRAIL(character);
149    }
150    return destination;
151}
152
153void TextCodecUTF8::consumePartialSequenceByte()
154{
155    --m_partialSequenceSize;
156    memmove(m_partialSequence, m_partialSequence + 1, m_partialSequenceSize);
157}
158
159void TextCodecUTF8::handleError(UChar*& destination, bool stopOnError, bool& sawError)
160{
161    sawError = true;
162    if (stopOnError)
163        return;
164    // Each error generates a replacement character and consumes one byte.
165    *destination++ = replacementCharacter;
166    consumePartialSequenceByte();
167}
168
169void TextCodecUTF8::handlePartialSequence(UChar*& destination, const uint8_t*& source, const uint8_t* end, bool flush, bool stopOnError, bool& sawError)
170{
171    ASSERT(m_partialSequenceSize);
172    do {
173        if (isASCII(m_partialSequence[0])) {
174            *destination++ = m_partialSequence[0];
175            consumePartialSequenceByte();
176            continue;
177        }
178        int count = nonASCIISequenceLength(m_partialSequence[0]);
179        if (!count) {
180            handleError(destination, stopOnError, sawError);
181            if (stopOnError)
182                return;
183            continue;
184        }
185        if (count > m_partialSequenceSize) {
186            if (count - m_partialSequenceSize > end - source) {
187                if (!flush) {
188                    // The new data is not enough to complete the sequence, so
189                    // add it to the existing partial sequence.
190                    memcpy(m_partialSequence + m_partialSequenceSize, source, end - source);
191                    m_partialSequenceSize += end - source;
192                    return;
193                }
194                // An incomplete partial sequence at the end is an error.
195                handleError(destination, stopOnError, sawError);
196                if (stopOnError)
197                    return;
198                continue;
199            }
200            memcpy(m_partialSequence + m_partialSequenceSize, source, count - m_partialSequenceSize);
201            source += count - m_partialSequenceSize;
202            m_partialSequenceSize = count;
203        }
204        int character = decodeNonASCIISequence(m_partialSequence, count);
205        if (character == nonCharacter) {
206            handleError(destination, stopOnError, sawError);
207            if (stopOnError)
208                return;
209            continue;
210        }
211        m_partialSequenceSize -= count;
212        destination = appendCharacter(destination, character);
213    } while (m_partialSequenceSize);
214}
215
216String TextCodecUTF8::decode(const char* bytes, size_t length, bool flush, bool stopOnError, bool& sawError)
217{
218    // Each input byte might turn into a character.
219    // That includes all bytes in the partial-sequence buffer because
220    // each byte in an invalid sequence will turn into a replacement character.
221    StringBuffer buffer(m_partialSequenceSize + length);
222
223    const uint8_t* source = reinterpret_cast<const uint8_t*>(bytes);
224    const uint8_t* end = source + length;
225    const uint8_t* alignedEnd = alignToMachineWord(end);
226    UChar* destination = buffer.characters();
227
228    do {
229        if (m_partialSequenceSize) {
230            // Explicitly copy destination and source pointers to avoid taking pointers to the
231            // local variables, which may harm code generation by disabling some optimizations
232            // in some compilers.
233            UChar* destinationForHandlePartialSequence = destination;
234            const uint8_t* sourceForHandlePartialSequence = source;
235            handlePartialSequence(destinationForHandlePartialSequence, sourceForHandlePartialSequence, end, flush, stopOnError, sawError);
236            destination = destinationForHandlePartialSequence;
237            source = sourceForHandlePartialSequence;
238            if (m_partialSequenceSize)
239                break;
240        }
241
242        while (source < end) {
243            if (isASCII(*source)) {
244                // Fast path for ASCII. Most UTF-8 text will be ASCII.
245                if (isAlignedToMachineWord(source)) {
246                    while (source < alignedEnd) {
247                        MachineWord chunk = *reinterpret_cast_ptr<const MachineWord*>(source);
248                        if (!isAllASCII(chunk))
249                            break;
250                        copyASCIIMachineWord(destination, source);
251                        source += sizeof(MachineWord);
252                        destination += sizeof(MachineWord);
253                    }
254                    if (source == end)
255                        break;
256                    if (!isASCII(*source))
257                        continue;
258                }
259                *destination++ = *source++;
260                continue;
261            }
262            int count = nonASCIISequenceLength(*source);
263            int character;
264            if (!count)
265                character = nonCharacter;
266            else {
267                if (count > end - source) {
268                    ASSERT(end - source < static_cast<ptrdiff_t>(sizeof(m_partialSequence)));
269                    ASSERT(!m_partialSequenceSize);
270                    m_partialSequenceSize = end - source;
271                    memcpy(m_partialSequence, source, m_partialSequenceSize);
272                    source = end;
273                    break;
274                }
275                character = decodeNonASCIISequence(source, count);
276            }
277            if (character == nonCharacter) {
278                sawError = true;
279                if (stopOnError)
280                    break;
281                // Each error generates a replacement character and consumes one byte.
282                *destination++ = replacementCharacter;
283                ++source;
284                continue;
285            }
286            source += count;
287            destination = appendCharacter(destination, character);
288        }
289    } while (flush && m_partialSequenceSize);
290
291    buffer.shrink(destination - buffer.characters());
292
293    return String::adopt(buffer);
294}
295
296CString TextCodecUTF8::encode(const UChar* characters, size_t length, UnencodableHandling)
297{
298    // The maximum number of UTF-8 bytes needed per UTF-16 code unit is 3.
299    // BMP characters take only one UTF-16 code unit and can take up to 3 bytes (3x).
300    // Non-BMP characters take two UTF-16 code units and can take up to 4 bytes (2x).
301    if (length > numeric_limits<size_t>::max() / 3)
302        CRASH();
303    Vector<uint8_t> bytes(length * 3);
304
305    size_t i = 0;
306    size_t bytesWritten = 0;
307    while (i < length) {
308        UChar32 character;
309        U16_NEXT(characters, i, length, character);
310        U8_APPEND_UNSAFE(bytes.data(), bytesWritten, character);
311    }
312
313    return CString(reinterpret_cast<char*>(bytes.data()), bytesWritten);
314}
315
316} // namespace WebCore
317