1 | /* |
2 | Copyright (C) 1999 Lars Knoll (knoll@mpi-hd.mpg.de) |
3 | Copyright (C) 2003-2017 Apple Inc. All rights reserved. |
4 | Copyright (C) 2005, 2006, 2007 Alexey Proskuryakov (ap@nypop.com) |
5 | |
6 | This library is free software; you can redistribute it and/or |
7 | modify it under the terms of the GNU Library General Public |
8 | License as published by the Free Software Foundation; either |
9 | version 2 of the License, or (at your option) any later version. |
10 | |
11 | This library is distributed in the hope that it will be useful, |
12 | but WITHOUT ANY WARRANTY; without even the implied warranty of |
13 | MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU |
14 | Library General Public License for more details. |
15 | |
16 | You should have received a copy of the GNU Library General Public License |
17 | along with this library; see the file COPYING.LIB. If not, write to |
18 | the Free Software Foundation, Inc., 51 Franklin Street, Fifth Floor, |
19 | Boston, MA 02110-1301, USA. |
20 | */ |
21 | |
22 | |
23 | #include "config.h" |
24 | #include "TextResourceDecoder.h" |
25 | |
26 | #include "HTMLMetaCharsetParser.h" |
27 | #include "HTMLNames.h" |
28 | #include "MIMETypeRegistry.h" |
29 | #include "TextCodec.h" |
30 | #include "TextEncoding.h" |
31 | #include "TextEncodingDetector.h" |
32 | #include "TextEncodingRegistry.h" |
33 | #include <wtf/ASCIICType.h> |
34 | |
35 | |
36 | namespace WebCore { |
37 | |
38 | using namespace HTMLNames; |
39 | |
40 | static inline bool bytesEqual(const char* p, char b0, char b1) |
41 | { |
42 | return p[0] == b0 && p[1] == b1; |
43 | } |
44 | |
45 | static inline bool bytesEqual(const char* p, char b0, char b1, char b2, char b3, char b4) |
46 | { |
47 | return p[0] == b0 && p[1] == b1 && p[2] == b2 && p[3] == b3 && p[4] == b4; |
48 | } |
49 | |
50 | static inline bool bytesEqual(const char* p, char b0, char b1, char b2, char b3, char b4, char b5) |
51 | { |
52 | return p[0] == b0 && p[1] == b1 && p[2] == b2 && p[3] == b3 && p[4] == b4 && p[5] == b5; |
53 | } |
54 | |
55 | static inline bool bytesEqual(const char* p, char b0, char b1, char b2, char b3, char b4, char b5, char b6, char b7) |
56 | { |
57 | return p[0] == b0 && p[1] == b1 && p[2] == b2 && p[3] == b3 && p[4] == b4 && p[5] == b5 && p[6] == b6 && p[7] == b7; |
58 | } |
59 | |
60 | static inline bool bytesEqual(const char* p, char b0, char b1, char b2, char b3, char b4, char b5, char b6, char b7, char b8, char b9) |
61 | { |
62 | return p[0] == b0 && p[1] == b1 && p[2] == b2 && p[3] == b3 && p[4] == b4 && p[5] == b5 && p[6] == b6 && p[7] == b7 && p[8] == b8 && p[9] == b9; |
63 | } |
64 | |
65 | // You might think we should put these find functions elsewhere, perhaps with the |
66 | // similar functions that operate on UChar, but arguably only the decoder has |
67 | // a reason to process strings of char rather than UChar. |
68 | |
69 | static int find(const char* subject, size_t subjectLength, const char* target) |
70 | { |
71 | size_t targetLength = strlen(target); |
72 | if (targetLength > subjectLength) |
73 | return -1; |
74 | for (size_t i = 0; i <= subjectLength - targetLength; ++i) { |
75 | bool match = true; |
76 | for (size_t j = 0; j < targetLength; ++j) { |
77 | if (subject[i + j] != target[j]) { |
78 | match = false; |
79 | break; |
80 | } |
81 | } |
82 | if (match) |
83 | return i; |
84 | } |
85 | return -1; |
86 | } |
87 | |
88 | static TextEncoding findTextEncoding(const char* encodingName, int length) |
89 | { |
90 | Vector<char, 64> buffer(length + 1); |
91 | memcpy(buffer.data(), encodingName, length); |
92 | buffer[length] = '\0'; |
93 | return buffer.data(); |
94 | } |
95 | |
96 | class KanjiCode { |
97 | public: |
98 | enum Type { ASCII, JIS, EUC, SJIS, UTF16, UTF8 }; |
99 | static enum Type judge(const char* str, int length); |
100 | static const int ESC = 0x1b; |
101 | static const unsigned char sjisMap[256]; |
102 | static int ISkanji(int code) |
103 | { |
104 | if (code >= 0x100) |
105 | return 0; |
106 | return sjisMap[code & 0xff] & 1; |
107 | } |
108 | static int ISkana(int code) |
109 | { |
110 | if (code >= 0x100) |
111 | return 0; |
112 | return sjisMap[code & 0xff] & 2; |
113 | } |
114 | }; |
115 | |
116 | const unsigned char KanjiCode::sjisMap[256] = { |
117 | 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, |
118 | 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, |
119 | 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, |
120 | 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, |
121 | 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, |
122 | 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, |
123 | 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, |
124 | 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, |
125 | 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, |
126 | 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, |
127 | 0, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, |
128 | 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, |
129 | 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, |
130 | 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, |
131 | 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, |
132 | 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0 |
133 | }; |
134 | |
135 | /* |
136 | * EUC-JP is |
137 | * [0xa1 - 0xfe][0xa1 - 0xfe] |
138 | * 0x8e[0xa1 - 0xfe](SS2) |
139 | * 0x8f[0xa1 - 0xfe][0xa1 - 0xfe](SS3) |
140 | * |
141 | * Shift_Jis is |
142 | * [0x81 - 0x9f, 0xe0 - 0xef(0xfe?)][0x40 - 0x7e, 0x80 - 0xfc] |
143 | * |
144 | * Shift_Jis Hankaku Kana is |
145 | * [0xa1 - 0xdf] |
146 | */ |
147 | |
148 | /* |
149 | * KanjiCode::judge() is based on judge_jcode() from jvim |
150 | * http://hp.vector.co.jp/authors/VA003457/vim/ |
151 | * |
152 | * Special Thanks to Kenichi Tsuchida |
153 | */ |
154 | |
155 | enum KanjiCode::Type KanjiCode::judge(const char* str, int size) |
156 | { |
157 | enum Type code; |
158 | int i; |
159 | int bfr = false; /* Kana Moji */ |
160 | int bfk = 0; /* EUC Kana */ |
161 | int sjis = 0; |
162 | int euc = 0; |
163 | |
164 | const unsigned char* ptr = reinterpret_cast<const unsigned char*>(str); |
165 | |
166 | code = ASCII; |
167 | |
168 | i = 0; |
169 | while (i < size) { |
170 | if (ptr[i] == ESC && (size - i >= 3)) { |
171 | if (bytesEqual(str + i + 1, '$', 'B') |
172 | || bytesEqual(str + i + 1, '(', 'B') |
173 | || bytesEqual(str + i + 1, '$', '@') |
174 | || bytesEqual(str + i + 1, '(', 'J')) { |
175 | code = JIS; |
176 | goto breakBreak; |
177 | } |
178 | if (bytesEqual(str + i + 1, '(', 'I') || bytesEqual(str + i + 1, ')', 'I')) { |
179 | code = JIS; |
180 | i += 3; |
181 | } else { |
182 | i++; |
183 | } |
184 | bfr = false; |
185 | bfk = 0; |
186 | } else { |
187 | if (ptr[i] < 0x20) { |
188 | bfr = false; |
189 | bfk = 0; |
190 | /* ?? check kudokuten ?? && ?? hiragana ?? */ |
191 | if ((i >= 2) && (ptr[i - 2] == 0x81) |
192 | && (0x41 <= ptr[i - 1] && ptr[i - 1] <= 0x49)) { |
193 | code = SJIS; |
194 | sjis += 100; /* kudokuten */ |
195 | } else if ((i >= 2) && (ptr[i - 2] == 0xa1) |
196 | && (0xa2 <= ptr[i - 1] && ptr[i - 1] <= 0xaa)) { |
197 | code = EUC; |
198 | euc += 100; /* kudokuten */ |
199 | } else if ((i >= 2) && (ptr[i - 2] == 0x82) && (0xa0 <= ptr[i - 1])) { |
200 | sjis += 40; /* hiragana */ |
201 | } else if ((i >= 2) && (ptr[i - 2] == 0xa4) && (0xa0 <= ptr[i - 1])) { |
202 | euc += 40; /* hiragana */ |
203 | } |
204 | } else { |
205 | /* ?? check hiragana or katana ?? */ |
206 | if ((size - i > 1) && (ptr[i] == 0x82) && (0xa0 <= ptr[i + 1])) { |
207 | sjis++; /* hiragana */ |
208 | } else if ((size - i > 1) && (ptr[i] == 0x83) |
209 | && (0x40 <= ptr[i + 1] && ptr[i + 1] <= 0x9f)) { |
210 | sjis++; /* katakana */ |
211 | } else if ((size - i > 1) && (ptr[i] == 0xa4) && (0xa0 <= ptr[i + 1])) { |
212 | euc++; /* hiragana */ |
213 | } else if ((size - i > 1) && (ptr[i] == 0xa5) && (0xa0 <= ptr[i + 1])) { |
214 | euc++; /* katakana */ |
215 | } |
216 | if (bfr) { |
217 | if ((i >= 1) && (0x40 <= ptr[i] && ptr[i] <= 0xa0) && ISkanji(ptr[i - 1])) { |
218 | code = SJIS; |
219 | goto breakBreak; |
220 | } else if ((i >= 1) && (0x81 <= ptr[i - 1] && ptr[i - 1] <= 0x9f) && ((0x40 <= ptr[i] && ptr[i] < 0x7e) || (0x7e < ptr[i] && ptr[i] <= 0xfc))) { |
221 | code = SJIS; |
222 | goto breakBreak; |
223 | } else if ((i >= 1) && (0xfd <= ptr[i] && ptr[i] <= 0xfe) && (0xa1 <= ptr[i - 1] && ptr[i - 1] <= 0xfe)) { |
224 | code = EUC; |
225 | goto breakBreak; |
226 | } else if ((i >= 1) && (0xfd <= ptr[i - 1] && ptr[i - 1] <= 0xfe) && (0xa1 <= ptr[i] && ptr[i] <= 0xfe)) { |
227 | code = EUC; |
228 | goto breakBreak; |
229 | } else if ((i >= 1) && (ptr[i] < 0xa0 || 0xdf < ptr[i]) && (0x8e == ptr[i - 1])) { |
230 | code = SJIS; |
231 | goto breakBreak; |
232 | } else if (ptr[i] <= 0x7f) { |
233 | code = SJIS; |
234 | goto breakBreak; |
235 | } else { |
236 | if (0xa1 <= ptr[i] && ptr[i] <= 0xa6) { |
237 | euc++; /* sjis hankaku kana kigo */ |
238 | } else if (0xa1 <= ptr[i] && ptr[i] <= 0xdf) { |
239 | ; /* sjis hankaku kana */ |
240 | } else if (0xa1 <= ptr[i] && ptr[i] <= 0xfe) { |
241 | euc++; |
242 | } else if (0x8e == ptr[i]) { |
243 | euc++; |
244 | } else if (0x20 <= ptr[i] && ptr[i] <= 0x7f) { |
245 | sjis++; |
246 | } |
247 | bfr = false; |
248 | bfk = 0; |
249 | } |
250 | } else if (0x8e == ptr[i]) { |
251 | if (size - i <= 1) { |
252 | ; |
253 | } else if (0xa1 <= ptr[i + 1] && ptr[i + 1] <= 0xdf) { |
254 | /* EUC KANA or SJIS KANJI */ |
255 | if (bfk == 1) { |
256 | euc += 100; |
257 | } |
258 | bfk++; |
259 | i++; |
260 | } else { |
261 | /* SJIS only */ |
262 | code = SJIS; |
263 | goto breakBreak; |
264 | } |
265 | } else if (0x81 <= ptr[i] && ptr[i] <= 0x9f) { |
266 | /* SJIS only */ |
267 | code = SJIS; |
268 | if ((size - i >= 1) |
269 | && ((0x40 <= ptr[i + 1] && ptr[i + 1] <= 0x7e) |
270 | || (0x80 <= ptr[i + 1] && ptr[i + 1] <= 0xfc))) { |
271 | goto breakBreak; |
272 | } |
273 | } else if (0xfd <= ptr[i] && ptr[i] <= 0xfe) { |
274 | /* EUC only */ |
275 | code = EUC; |
276 | if ((size - i >= 1) |
277 | && (0xa1 <= ptr[i + 1] && ptr[i + 1] <= 0xfe)) { |
278 | goto breakBreak; |
279 | } |
280 | } else if (ptr[i] <= 0x7f) { |
281 | ; |
282 | } else { |
283 | bfr = true; |
284 | bfk = 0; |
285 | } |
286 | } |
287 | i++; |
288 | } |
289 | } |
290 | if (code == ASCII) { |
291 | if (sjis > euc) { |
292 | code = SJIS; |
293 | } else if (sjis < euc) { |
294 | code = EUC; |
295 | } |
296 | } |
297 | breakBreak: |
298 | return (code); |
299 | } |
300 | |
301 | TextResourceDecoder::ContentType TextResourceDecoder::determineContentType(const String& mimeType) |
302 | { |
303 | if (equalLettersIgnoringASCIICase(mimeType, "text/css" )) |
304 | return CSS; |
305 | if (equalLettersIgnoringASCIICase(mimeType, "text/html" )) |
306 | return HTML; |
307 | if (MIMETypeRegistry::isXMLMIMEType(mimeType)) |
308 | return XML; |
309 | return PlainText; |
310 | } |
311 | |
312 | const TextEncoding& TextResourceDecoder::defaultEncoding(ContentType contentType, const TextEncoding& specifiedDefaultEncoding) |
313 | { |
314 | // Despite 8.5 "Text/xml with Omitted Charset" of RFC 3023, we assume UTF-8 instead of US-ASCII |
315 | // for text/xml. This matches Firefox. |
316 | if (contentType == XML) |
317 | return UTF8Encoding(); |
318 | if (!specifiedDefaultEncoding.isValid()) |
319 | return Latin1Encoding(); |
320 | return specifiedDefaultEncoding; |
321 | } |
322 | |
323 | inline TextResourceDecoder::TextResourceDecoder(const String& mimeType, const TextEncoding& specifiedDefaultEncoding, bool usesEncodingDetector) |
324 | : m_contentType(determineContentType(mimeType)) |
325 | , m_encoding(defaultEncoding(m_contentType, specifiedDefaultEncoding)) |
326 | , m_usesEncodingDetector(usesEncodingDetector) |
327 | { |
328 | } |
329 | |
330 | Ref<TextResourceDecoder> TextResourceDecoder::create(const String& mimeType, const TextEncoding& defaultEncoding, bool usesEncodingDetector) |
331 | { |
332 | return adoptRef(*new TextResourceDecoder(mimeType, defaultEncoding, usesEncodingDetector)); |
333 | } |
334 | |
335 | TextResourceDecoder::~TextResourceDecoder() = default; |
336 | |
337 | void TextResourceDecoder::setEncoding(const TextEncoding& encoding, EncodingSource source) |
338 | { |
339 | // In case the encoding didn't exist, we keep the old one (helps some sites specifying invalid encodings). |
340 | if (!encoding.isValid()) |
341 | return; |
342 | |
343 | // When encoding comes from meta tag (i.e. it cannot be XML files sent via XHR), |
344 | // treat x-user-defined as windows-1252 (bug 18270) |
345 | if (source == EncodingFromMetaTag && equalLettersIgnoringASCIICase(encoding.name(), "x-user-defined" )) |
346 | m_encoding = "windows-1252" ; |
347 | else if (source == EncodingFromMetaTag || source == EncodingFromXMLHeader || source == EncodingFromCSSCharset) |
348 | m_encoding = encoding.closestByteBasedEquivalent(); |
349 | else |
350 | m_encoding = encoding; |
351 | |
352 | m_codec = nullptr; |
353 | m_source = source; |
354 | } |
355 | |
356 | bool TextResourceDecoder::hasEqualEncodingForCharset(const String& charset) const |
357 | { |
358 | return defaultEncoding(m_contentType, charset) == m_encoding; |
359 | } |
360 | |
361 | // Returns the position of the encoding string. |
362 | static int findXMLEncoding(const char* str, int len, int& encodingLength) |
363 | { |
364 | int pos = find(str, len, "encoding" ); |
365 | if (pos == -1) |
366 | return -1; |
367 | pos += 8; |
368 | |
369 | // Skip spaces and stray control characters. |
370 | while (pos < len && str[pos] <= ' ') |
371 | ++pos; |
372 | |
373 | // Skip equals sign. |
374 | if (pos >= len || str[pos] != '=') |
375 | return -1; |
376 | ++pos; |
377 | |
378 | // Skip spaces and stray control characters. |
379 | while (pos < len && str[pos] <= ' ') |
380 | ++pos; |
381 | |
382 | // Skip quotation mark. |
383 | if (pos >= len) |
384 | return - 1; |
385 | char quoteMark = str[pos]; |
386 | if (quoteMark != '"' && quoteMark != '\'') |
387 | return -1; |
388 | ++pos; |
389 | |
390 | // Find the trailing quotation mark. |
391 | int end = pos; |
392 | while (end < len && str[end] != quoteMark) |
393 | ++end; |
394 | if (end >= len) |
395 | return -1; |
396 | |
397 | encodingLength = end - pos; |
398 | return pos; |
399 | } |
400 | |
401 | size_t TextResourceDecoder::checkForBOM(const char* data, size_t len) |
402 | { |
403 | // Check for UTF-16 or UTF-8 BOM mark at the beginning, which is a sure sign of a Unicode encoding. |
404 | // We let it override even a user-chosen encoding. |
405 | const size_t maximumBOMLength = 3; |
406 | |
407 | ASSERT(!m_checkedForBOM); |
408 | |
409 | size_t lengthOfBOM = 0; |
410 | |
411 | size_t bufferLength = m_buffer.size(); |
412 | |
413 | size_t buf1Len = bufferLength; |
414 | size_t buf2Len = len; |
415 | const unsigned char* buf1 = reinterpret_cast<const unsigned char*>(m_buffer.data()); |
416 | const unsigned char* buf2 = reinterpret_cast<const unsigned char*>(data); |
417 | unsigned char c1 = buf1Len ? (static_cast<void>(--buf1Len), *buf1++) : buf2Len ? (static_cast<void>(--buf2Len), *buf2++) : 0; |
418 | unsigned char c2 = buf1Len ? (static_cast<void>(--buf1Len), *buf1++) : buf2Len ? (static_cast<void>(--buf2Len), *buf2++) : 0; |
419 | unsigned char c3 = buf1Len ? (static_cast<void>(--buf1Len), *buf1++) : buf2Len ? (static_cast<void>(--buf2Len), *buf2++) : 0; |
420 | |
421 | // Check for the BOM. |
422 | if (c1 == 0xFF && c2 == 0xFE) { |
423 | ASSERT(UTF16LittleEndianEncoding().isValid()); |
424 | setEncoding(UTF16LittleEndianEncoding(), AutoDetectedEncoding); |
425 | lengthOfBOM = 2; |
426 | } else if (c1 == 0xFE && c2 == 0xFF) { |
427 | ASSERT(UTF16BigEndianEncoding().isValid()); |
428 | setEncoding(UTF16BigEndianEncoding(), AutoDetectedEncoding); |
429 | lengthOfBOM = 2; |
430 | } else if (c1 == 0xEF && c2 == 0xBB && c3 == 0xBF) { |
431 | ASSERT(UTF8Encoding().isValid()); |
432 | setEncoding(UTF8Encoding(), AutoDetectedEncoding); |
433 | lengthOfBOM = 3; |
434 | } |
435 | |
436 | if (lengthOfBOM || bufferLength + len >= maximumBOMLength) |
437 | m_checkedForBOM = true; |
438 | |
439 | ASSERT(lengthOfBOM <= maximumBOMLength); |
440 | return lengthOfBOM; |
441 | } |
442 | |
443 | bool TextResourceDecoder::checkForCSSCharset(const char* data, size_t len, bool& movedDataToBuffer) |
444 | { |
445 | if (m_source != DefaultEncoding && m_source != EncodingFromParentFrame) { |
446 | m_checkedForCSSCharset = true; |
447 | return true; |
448 | } |
449 | |
450 | size_t oldSize = m_buffer.size(); |
451 | m_buffer.grow(oldSize + len); |
452 | memcpy(m_buffer.data() + oldSize, data, len); |
453 | |
454 | movedDataToBuffer = true; |
455 | |
456 | if (m_buffer.size() <= 13) // strlen('@charset "x";') == 13 |
457 | return false; |
458 | |
459 | const char* dataStart = m_buffer.data(); |
460 | const char* dataEnd = dataStart + m_buffer.size(); |
461 | |
462 | if (bytesEqual(dataStart, '@', 'c', 'h', 'a', 'r', 's', 'e', 't', ' ', '"')) { |
463 | dataStart += 10; |
464 | const char* pos = dataStart; |
465 | |
466 | while (pos < dataEnd && *pos != '"') |
467 | ++pos; |
468 | if (pos == dataEnd) |
469 | return false; |
470 | |
471 | int encodingNameLength = pos - dataStart; |
472 | |
473 | ++pos; |
474 | if (pos == dataEnd) |
475 | return false; |
476 | |
477 | if (*pos == ';') |
478 | setEncoding(findTextEncoding(dataStart, encodingNameLength), EncodingFromCSSCharset); |
479 | } |
480 | |
481 | m_checkedForCSSCharset = true; |
482 | return true; |
483 | } |
484 | |
485 | bool TextResourceDecoder::checkForHeadCharset(const char* data, size_t len, bool& movedDataToBuffer) |
486 | { |
487 | if (m_source != DefaultEncoding && m_source != EncodingFromParentFrame) { |
488 | m_checkedForHeadCharset = true; |
489 | return true; |
490 | } |
491 | |
492 | // This is not completely efficient, since the function might go |
493 | // through the HTML head several times. |
494 | |
495 | size_t oldSize = m_buffer.size(); |
496 | m_buffer.grow(oldSize + len); |
497 | memcpy(m_buffer.data() + oldSize, data, len); |
498 | |
499 | movedDataToBuffer = true; |
500 | |
501 | // Continue with checking for an HTML meta tag if we were already doing so. |
502 | if (m_charsetParser) |
503 | return checkForMetaCharset(data, len); |
504 | |
505 | const char* ptr = m_buffer.data(); |
506 | const char* pEnd = ptr + m_buffer.size(); |
507 | |
508 | // Is there enough data available to check for XML declaration? |
509 | if (m_buffer.size() < 8) |
510 | return false; |
511 | |
512 | // Handle XML declaration, which can have encoding in it. This encoding is honored even for HTML documents. |
513 | // It is an error for an XML declaration not to be at the start of an XML document, and it is ignored in HTML documents in such case. |
514 | if (bytesEqual(ptr, '<', '?', 'x', 'm', 'l')) { |
515 | const char* xmlDeclarationEnd = ptr; |
516 | while (xmlDeclarationEnd != pEnd && *xmlDeclarationEnd != '>') |
517 | ++xmlDeclarationEnd; |
518 | if (xmlDeclarationEnd == pEnd) |
519 | return false; |
520 | // No need for +1, because we have an extra "?" to lose at the end of XML declaration. |
521 | int len = 0; |
522 | int pos = findXMLEncoding(ptr, xmlDeclarationEnd - ptr, len); |
523 | if (pos != -1) |
524 | setEncoding(findTextEncoding(ptr + pos, len), EncodingFromXMLHeader); |
525 | // continue looking for a charset - it may be specified in an HTTP-Equiv meta |
526 | } else if (bytesEqual(ptr, '<', 0, '?', 0, 'x', 0)) { |
527 | setEncoding(UTF16LittleEndianEncoding(), AutoDetectedEncoding); |
528 | return true; |
529 | } else if (bytesEqual(ptr, 0, '<', 0, '?', 0, 'x')) { |
530 | setEncoding(UTF16BigEndianEncoding(), AutoDetectedEncoding); |
531 | return true; |
532 | } |
533 | |
534 | // The HTTP-EQUIV meta has no effect on XHTML. |
535 | if (m_contentType == XML) |
536 | return true; |
537 | |
538 | m_charsetParser = std::make_unique<HTMLMetaCharsetParser>(); |
539 | return checkForMetaCharset(data, len); |
540 | } |
541 | |
542 | bool TextResourceDecoder::checkForMetaCharset(const char* data, size_t length) |
543 | { |
544 | if (!m_charsetParser->checkForMetaCharset(data, length)) |
545 | return false; |
546 | |
547 | setEncoding(m_charsetParser->encoding(), EncodingFromMetaTag); |
548 | m_charsetParser = nullptr; |
549 | m_checkedForHeadCharset = true; |
550 | return true; |
551 | } |
552 | |
553 | void TextResourceDecoder::detectJapaneseEncoding(const char* data, size_t len) |
554 | { |
555 | switch (KanjiCode::judge(data, len)) { |
556 | case KanjiCode::JIS: |
557 | setEncoding("ISO-2022-JP" , AutoDetectedEncoding); |
558 | break; |
559 | case KanjiCode::EUC: |
560 | setEncoding("EUC-JP" , AutoDetectedEncoding); |
561 | break; |
562 | case KanjiCode::SJIS: |
563 | setEncoding("Shift_JIS" , AutoDetectedEncoding); |
564 | break; |
565 | case KanjiCode::ASCII: |
566 | case KanjiCode::UTF16: |
567 | case KanjiCode::UTF8: |
568 | break; |
569 | } |
570 | } |
571 | |
572 | // We use the encoding detector in two cases: |
573 | // 1. Encoding detector is turned ON and no other encoding source is |
574 | // available (that is, it's DefaultEncoding). |
575 | // 2. Encoding detector is turned ON and the encoding is set to |
576 | // the encoding of the parent frame, which is also auto-detected. |
577 | // Note that condition #2 is NOT satisfied unless parent-child frame |
578 | // relationship is compliant to the same-origin policy. If they're from |
579 | // different domains, |m_source| would not be set to EncodingFromParentFrame |
580 | // in the first place. |
581 | bool TextResourceDecoder::shouldAutoDetect() const |
582 | { |
583 | return m_usesEncodingDetector |
584 | && (m_source == DefaultEncoding || (m_source == EncodingFromParentFrame && m_parentFrameAutoDetectedEncoding)); |
585 | } |
586 | |
587 | String TextResourceDecoder::decode(const char* data, size_t length) |
588 | { |
589 | size_t lengthOfBOM = 0; |
590 | if (!m_checkedForBOM) |
591 | lengthOfBOM = checkForBOM(data, length); |
592 | |
593 | bool movedDataToBuffer = false; |
594 | |
595 | if (m_contentType == CSS && !m_checkedForCSSCharset) |
596 | if (!checkForCSSCharset(data, length, movedDataToBuffer)) |
597 | return emptyString(); |
598 | |
599 | if ((m_contentType == HTML || m_contentType == XML) && !m_checkedForHeadCharset) // HTML and XML |
600 | if (!checkForHeadCharset(data, length, movedDataToBuffer)) |
601 | return emptyString(); |
602 | |
603 | // FIXME: It is wrong to change the encoding downstream after we have already done some decoding. |
604 | if (shouldAutoDetect()) { |
605 | if (m_encoding.isJapanese()) |
606 | detectJapaneseEncoding(data, length); // FIXME: We should use detectTextEncoding() for all languages. |
607 | else { |
608 | TextEncoding detectedEncoding; |
609 | if (detectTextEncoding(data, length, m_parentFrameAutoDetectedEncoding, &detectedEncoding)) |
610 | setEncoding(detectedEncoding, AutoDetectedEncoding); |
611 | } |
612 | } |
613 | |
614 | ASSERT(m_encoding.isValid()); |
615 | |
616 | if (!m_codec) |
617 | m_codec = newTextCodec(m_encoding); |
618 | |
619 | if (m_buffer.isEmpty()) |
620 | return m_codec->decode(data + lengthOfBOM, length - lengthOfBOM, false, m_contentType == XML, m_sawError); |
621 | |
622 | if (!movedDataToBuffer) { |
623 | size_t oldSize = m_buffer.size(); |
624 | m_buffer.grow(oldSize + length); |
625 | memcpy(m_buffer.data() + oldSize, data, length); |
626 | } |
627 | |
628 | String result = m_codec->decode(m_buffer.data() + lengthOfBOM, m_buffer.size() - lengthOfBOM, false, m_contentType == XML && !m_useLenientXMLDecoding, m_sawError); |
629 | m_buffer.clear(); |
630 | return result; |
631 | } |
632 | |
633 | String TextResourceDecoder::flush() |
634 | { |
635 | // If we can not identify the encoding even after a document is completely |
636 | // loaded, we need to detect the encoding if other conditions for |
637 | // autodetection is satisfied. |
638 | if (m_buffer.size() && shouldAutoDetect() |
639 | && ((!m_checkedForHeadCharset && (m_contentType == HTML || m_contentType == XML)) || (!m_checkedForCSSCharset && (m_contentType == CSS)))) { |
640 | TextEncoding detectedEncoding; |
641 | if (detectTextEncoding(m_buffer.data(), m_buffer.size(), m_parentFrameAutoDetectedEncoding, &detectedEncoding)) |
642 | setEncoding(detectedEncoding, AutoDetectedEncoding); |
643 | } |
644 | |
645 | if (!m_codec) |
646 | m_codec = newTextCodec(m_encoding); |
647 | |
648 | String result = m_codec->decode(m_buffer.data(), m_buffer.size(), true, m_contentType == XML && !m_useLenientXMLDecoding, m_sawError); |
649 | m_buffer.clear(); |
650 | m_codec = nullptr; |
651 | m_checkedForBOM = false; // Skip BOM again when re-decoding. |
652 | return result; |
653 | } |
654 | |
655 | String TextResourceDecoder::decodeAndFlush(const char* data, size_t length) |
656 | { |
657 | String decoded = decode(data, length); |
658 | return decoded + flush(); |
659 | } |
660 | |
661 | const TextEncoding* TextResourceDecoder::encodingForURLParsing() |
662 | { |
663 | // For UTF-{7,16,32}, we want to use UTF-8 for the query part as |
664 | // we do when submitting a form. A form with GET method |
665 | // has its contents added to a URL as query params and it makes sense |
666 | // to be consistent. |
667 | auto& encoding = m_encoding.encodingForFormSubmissionOrURLParsing(); |
668 | if (encoding == UTF8Encoding()) |
669 | return nullptr; |
670 | return &encoding; |
671 | } |
672 | |
673 | } |
674 | |