]>
Commit | Line | Data |
---|---|---|
1 | ||
2 | #include "yaml_private.h" | |
3 | ||
4 | /* | |
5 | * Declarations. | |
6 | */ | |
7 | ||
8 | static int | |
9 | yaml_parser_set_reader_error(yaml_parser_t *parser, const char *problem, | |
10 | size_t offset, int value); | |
11 | ||
12 | static int | |
13 | yaml_parser_update_raw_buffer(yaml_parser_t *parser); | |
14 | ||
15 | static int | |
16 | yaml_parser_determine_encoding(yaml_parser_t *parser); | |
17 | ||
18 | YAML_DECLARE(int) | |
19 | yaml_parser_update_buffer(yaml_parser_t *parser, size_t length); | |
20 | ||
21 | /* | |
22 | * Set the reader error and return 0. | |
23 | */ | |
24 | ||
25 | static int | |
26 | yaml_parser_set_reader_error(yaml_parser_t *parser, const char *problem, | |
27 | size_t offset, int value) | |
28 | { | |
29 | parser->error = YAML_READER_ERROR; | |
30 | parser->problem = problem; | |
31 | parser->problem_offset = offset; | |
32 | parser->problem_value = value; | |
33 | ||
34 | return 0; | |
35 | } | |
36 | ||
37 | /* | |
38 | * Byte order marks. | |
39 | */ | |
40 | ||
41 | #define BOM_UTF8 "\xef\xbb\xbf" | |
42 | #define BOM_UTF16LE "\xff\xfe" | |
43 | #define BOM_UTF16BE "\xfe\xff" | |
44 | ||
45 | /* | |
46 | * Determine the input stream encoding by checking the BOM symbol. If no BOM is | |
47 | * found, the UTF-8 encoding is assumed. Return 1 on success, 0 on failure. | |
48 | */ | |
49 | ||
50 | static int | |
51 | yaml_parser_determine_encoding(yaml_parser_t *parser) | |
52 | { | |
53 | /* Ensure that we had enough bytes in the raw buffer. */ | |
54 | ||
55 | while (!parser->eof | |
56 | && parser->raw_buffer.last - parser->raw_buffer.pointer < 3) { | |
57 | if (!yaml_parser_update_raw_buffer(parser)) { | |
58 | return 0; | |
59 | } | |
60 | } | |
61 | ||
62 | /* Determine the encoding. */ | |
63 | ||
64 | if (parser->raw_buffer.last - parser->raw_buffer.pointer >= 2 | |
65 | && !memcmp(parser->raw_buffer.pointer, BOM_UTF16LE, 2)) { | |
66 | parser->encoding = YAML_UTF16LE_ENCODING; | |
67 | parser->raw_buffer.pointer += 2; | |
68 | parser->offset += 2; | |
69 | } | |
70 | else if (parser->raw_buffer.last - parser->raw_buffer.pointer >= 2 | |
71 | && !memcmp(parser->raw_buffer.pointer, BOM_UTF16BE, 2)) { | |
72 | parser->encoding = YAML_UTF16BE_ENCODING; | |
73 | parser->raw_buffer.pointer += 2; | |
74 | parser->offset += 2; | |
75 | } | |
76 | else if (parser->raw_buffer.last - parser->raw_buffer.pointer >= 3 | |
77 | && !memcmp(parser->raw_buffer.pointer, BOM_UTF8, 3)) { | |
78 | parser->encoding = YAML_UTF8_ENCODING; | |
79 | parser->raw_buffer.pointer += 3; | |
80 | parser->offset += 3; | |
81 | } | |
82 | else { | |
83 | parser->encoding = YAML_UTF8_ENCODING; | |
84 | } | |
85 | ||
86 | return 1; | |
87 | } | |
88 | ||
89 | /* | |
90 | * Update the raw buffer. | |
91 | */ | |
92 | ||
93 | static int | |
94 | yaml_parser_update_raw_buffer(yaml_parser_t *parser) | |
95 | { | |
96 | size_t size_read = 0; | |
97 | ||
98 | /* Return if the raw buffer is full. */ | |
99 | ||
100 | if (parser->raw_buffer.start == parser->raw_buffer.pointer | |
101 | && parser->raw_buffer.last == parser->raw_buffer.end) | |
102 | return 1; | |
103 | ||
104 | /* Return on EOF. */ | |
105 | ||
106 | if (parser->eof) return 1; | |
107 | ||
108 | /* Move the remaining bytes in the raw buffer to the beginning. */ | |
109 | ||
110 | if (parser->raw_buffer.start < parser->raw_buffer.pointer | |
111 | && parser->raw_buffer.pointer < parser->raw_buffer.last) { | |
112 | memmove(parser->raw_buffer.start, parser->raw_buffer.pointer, | |
113 | parser->raw_buffer.last - parser->raw_buffer.pointer); | |
114 | } | |
115 | parser->raw_buffer.last -= | |
116 | parser->raw_buffer.pointer - parser->raw_buffer.start; | |
117 | parser->raw_buffer.pointer = parser->raw_buffer.start; | |
118 | ||
119 | /* Call the read handler to fill the buffer. */ | |
120 | ||
121 | if (!parser->read_handler(parser->read_handler_data, parser->raw_buffer.last, | |
122 | parser->raw_buffer.end - parser->raw_buffer.last, &size_read)) { | |
123 | return yaml_parser_set_reader_error(parser, "Input error", | |
124 | parser->offset, -1); | |
125 | } | |
126 | parser->raw_buffer.last += size_read; | |
127 | if (!size_read) { | |
128 | parser->eof = 1; | |
129 | } | |
130 | ||
131 | return 1; | |
132 | } | |
133 | ||
134 | /* | |
135 | * Ensure that the buffer contains at least `length` characters. | |
136 | * Return 1 on success, 0 on failure. | |
137 | * | |
138 | * The length is supposed to be significantly less that the buffer size. | |
139 | */ | |
140 | ||
141 | YAML_DECLARE(int) | |
142 | yaml_parser_update_buffer(yaml_parser_t *parser, size_t length) | |
143 | { | |
144 | assert(parser->read_handler); /* Read handler must be set. */ | |
145 | ||
146 | /* If the EOF flag is set and the raw buffer is empty, do nothing. */ | |
147 | ||
148 | if (parser->eof && parser->raw_buffer.pointer == parser->raw_buffer.last) | |
149 | return 1; | |
150 | ||
151 | /* Return if the buffer contains enough characters. */ | |
152 | ||
153 | if (parser->unread >= length) | |
154 | return 1; | |
155 | ||
156 | /* Determine the input encoding if it is not known yet. */ | |
157 | ||
158 | if (!parser->encoding) { | |
159 | if (!yaml_parser_determine_encoding(parser)) | |
160 | return 0; | |
161 | } | |
162 | ||
163 | /* Move the unread characters to the beginning of the buffer. */ | |
164 | ||
165 | if (parser->buffer.start < parser->buffer.pointer | |
166 | && parser->buffer.pointer < parser->buffer.last) { | |
167 | size_t size = parser->buffer.last - parser->buffer.pointer; | |
168 | memmove(parser->buffer.start, parser->buffer.pointer, size); | |
169 | parser->buffer.pointer = parser->buffer.start; | |
170 | parser->buffer.last = parser->buffer.start + size; | |
171 | } | |
172 | else if (parser->buffer.pointer == parser->buffer.last) { | |
173 | parser->buffer.pointer = parser->buffer.start; | |
174 | parser->buffer.last = parser->buffer.start; | |
175 | } | |
176 | ||
177 | /* Fill the buffer until it has enough characters. */ | |
178 | ||
179 | while (parser->unread < length) | |
180 | { | |
181 | /* Fill the raw buffer. */ | |
182 | ||
183 | if (!yaml_parser_update_raw_buffer(parser)) return 0; | |
184 | ||
185 | /* Decode the raw buffer. */ | |
186 | ||
187 | while (parser->raw_buffer.pointer != parser->raw_buffer.last) | |
188 | { | |
189 | unsigned int value = 0, value2 = 0; | |
190 | int incomplete = 0; | |
191 | unsigned char octet; | |
192 | unsigned int width = 0; | |
193 | int low, high; | |
194 | size_t k; | |
195 | size_t raw_unread = parser->raw_buffer.last - parser->raw_buffer.pointer; | |
196 | ||
197 | /* Decode the next character. */ | |
198 | ||
199 | switch (parser->encoding) | |
200 | { | |
201 | case YAML_UTF8_ENCODING: | |
202 | ||
203 | /* | |
204 | * Decode a UTF-8 character. Check RFC 3629 | |
205 | * (http://www.ietf.org/rfc/rfc3629.txt) for more details. | |
206 | * | |
207 | * The following table (taken from the RFC) is used for | |
208 | * decoding. | |
209 | * | |
210 | * Char. number range | UTF-8 octet sequence | |
211 | * (hexadecimal) | (binary) | |
212 | * --------------------+------------------------------------ | |
213 | * 0000 0000-0000 007F | 0xxxxxxx | |
214 | * 0000 0080-0000 07FF | 110xxxxx 10xxxxxx | |
215 | * 0000 0800-0000 FFFF | 1110xxxx 10xxxxxx 10xxxxxx | |
216 | * 0001 0000-0010 FFFF | 11110xxx 10xxxxxx 10xxxxxx 10xxxxxx | |
217 | * | |
218 | * Additionally, the characters in the range 0xD800-0xDFFF | |
219 | * are prohibited as they are reserved for use with UTF-16 | |
220 | * surrogate pairs. | |
221 | */ | |
222 | ||
223 | /* Determine the length of the UTF-8 sequence. */ | |
224 | ||
225 | octet = parser->raw_buffer.pointer[0]; | |
226 | width = (octet & 0x80) == 0x00 ? 1 : | |
227 | (octet & 0xE0) == 0xC0 ? 2 : | |
228 | (octet & 0xF0) == 0xE0 ? 3 : | |
229 | (octet & 0xF8) == 0xF0 ? 4 : 0; | |
230 | ||
231 | /* Check if the leading octet is valid. */ | |
232 | ||
233 | if (!width) | |
234 | return yaml_parser_set_reader_error(parser, | |
235 | "Invalid leading UTF-8 octet", | |
236 | parser->offset, octet); | |
237 | ||
238 | /* Check if the raw buffer contains an incomplete character. */ | |
239 | ||
240 | if (width > raw_unread) { | |
241 | if (parser->eof) { | |
242 | return yaml_parser_set_reader_error(parser, | |
243 | "Incomplete UTF-8 octet sequence", | |
244 | parser->offset, -1); | |
245 | } | |
246 | incomplete = 1; | |
247 | break; | |
248 | } | |
249 | ||
250 | /* Decode the leading octet. */ | |
251 | ||
252 | value = (octet & 0x80) == 0x00 ? octet & 0x7F : | |
253 | (octet & 0xE0) == 0xC0 ? octet & 0x1F : | |
254 | (octet & 0xF0) == 0xE0 ? octet & 0x0F : | |
255 | (octet & 0xF8) == 0xF0 ? octet & 0x07 : 0; | |
256 | ||
257 | /* Check and decode the trailing octets. */ | |
258 | ||
259 | for (k = 1; k < width; k ++) | |
260 | { | |
261 | octet = parser->raw_buffer.pointer[k]; | |
262 | ||
263 | /* Check if the octet is valid. */ | |
264 | ||
265 | if ((octet & 0xC0) != 0x80) | |
266 | return yaml_parser_set_reader_error(parser, | |
267 | "Invalid trailing UTF-8 octet", | |
268 | parser->offset+k, octet); | |
269 | ||
270 | /* Decode the octet. */ | |
271 | ||
272 | value = (value << 6) + (octet & 0x3F); | |
273 | } | |
274 | ||
275 | /* Check the length of the sequence against the value. */ | |
276 | ||
277 | if (!((width == 1) || | |
278 | (width == 2 && value >= 0x80) || | |
279 | (width == 3 && value >= 0x800) || | |
280 | (width == 4 && value >= 0x10000))) | |
281 | return yaml_parser_set_reader_error(parser, | |
282 | "Invalid length of a UTF-8 sequence", | |
283 | parser->offset, -1); | |
284 | ||
285 | /* Check the range of the value. */ | |
286 | ||
287 | if ((value >= 0xD800 && value <= 0xDFFF) || value > 0x10FFFF) | |
288 | return yaml_parser_set_reader_error(parser, | |
289 | "Invalid Unicode character", | |
290 | parser->offset, value); | |
291 | ||
292 | break; | |
293 | ||
294 | case YAML_UTF16LE_ENCODING: | |
295 | case YAML_UTF16BE_ENCODING: | |
296 | ||
297 | low = (parser->encoding == YAML_UTF16LE_ENCODING ? 0 : 1); | |
298 | high = (parser->encoding == YAML_UTF16LE_ENCODING ? 1 : 0); | |
299 | ||
300 | /* | |
301 | * The UTF-16 encoding is not as simple as one might | |
302 | * naively think. Check RFC 2781 | |
303 | * (http://www.ietf.org/rfc/rfc2781.txt). | |
304 | * | |
305 | * Normally, two subsequent bytes describe a Unicode | |
306 | * character. However a special technique (called a | |
307 | * surrogate pair) is used for specifying character | |
308 | * values larger than 0xFFFF. | |
309 | * | |
310 | * A surrogate pair consists of two pseudo-characters: | |
311 | * high surrogate area (0xD800-0xDBFF) | |
312 | * low surrogate area (0xDC00-0xDFFF) | |
313 | * | |
314 | * The following formulas are used for decoding | |
315 | * and encoding characters using surrogate pairs: | |
316 | * | |
317 | * U = U' + 0x10000 (0x01 00 00 <= U <= 0x10 FF FF) | |
318 | * U' = yyyyyyyyyyxxxxxxxxxx (0 <= U' <= 0x0F FF FF) | |
319 | * W1 = 110110yyyyyyyyyy | |
320 | * W2 = 110111xxxxxxxxxx | |
321 | * | |
322 | * where U is the character value, W1 is the high surrogate | |
323 | * area, W2 is the low surrogate area. | |
324 | */ | |
325 | ||
326 | /* Check for incomplete UTF-16 character. */ | |
327 | ||
328 | if (raw_unread < 2) { | |
329 | if (parser->eof) { | |
330 | return yaml_parser_set_reader_error(parser, | |
331 | "Incomplete UTF-16 character", | |
332 | parser->offset, -1); | |
333 | } | |
334 | incomplete = 1; | |
335 | break; | |
336 | } | |
337 | ||
338 | /* Get the character. */ | |
339 | ||
340 | value = parser->raw_buffer.pointer[low] | |
341 | + (parser->raw_buffer.pointer[high] << 8); | |
342 | ||
343 | /* Check for unexpected low surrogate area. */ | |
344 | ||
345 | if ((value & 0xFC00) == 0xDC00) | |
346 | return yaml_parser_set_reader_error(parser, | |
347 | "Unexpected low surrogate area", | |
348 | parser->offset, value); | |
349 | ||
350 | /* Check for a high surrogate area. */ | |
351 | ||
352 | if ((value & 0xFC00) == 0xD800) { | |
353 | ||
354 | width = 4; | |
355 | ||
356 | /* Check for incomplete surrogate pair. */ | |
357 | ||
358 | if (raw_unread < 4) { | |
359 | if (parser->eof) { | |
360 | return yaml_parser_set_reader_error(parser, | |
361 | "Incomplete UTF-16 surrogate pair", | |
362 | parser->offset, -1); | |
363 | } | |
364 | incomplete = 1; | |
365 | break; | |
366 | } | |
367 | ||
368 | /* Get the next character. */ | |
369 | ||
370 | value2 = parser->raw_buffer.pointer[low+2] | |
371 | + (parser->raw_buffer.pointer[high+2] << 8); | |
372 | ||
373 | /* Check for a low surrogate area. */ | |
374 | ||
375 | if ((value2 & 0xFC00) != 0xDC00) | |
376 | return yaml_parser_set_reader_error(parser, | |
377 | "Expected low surrogate area", | |
378 | parser->offset+2, value2); | |
379 | ||
380 | /* Generate the value of the surrogate pair. */ | |
381 | ||
382 | value = 0x10000 + ((value & 0x3FF) << 10) + (value2 & 0x3FF); | |
383 | } | |
384 | ||
385 | else { | |
386 | width = 2; | |
387 | } | |
388 | ||
389 | break; | |
390 | ||
391 | default: | |
392 | assert(1); /* Impossible. */ | |
393 | } | |
394 | ||
395 | /* Check if the raw buffer contains enough bytes to form a character. */ | |
396 | ||
397 | if (incomplete) break; | |
398 | ||
399 | /* | |
400 | * Check if the character is in the allowed range: | |
401 | * #x9 | #xA | #xD | [#x20-#x7E] (8 bit) | |
402 | * | #x85 | [#xA0-#xD7FF] | [#xE000-#xFFFD] (16 bit) | |
403 | * | [#x10000-#x10FFFF] (32 bit) | |
404 | */ | |
405 | ||
406 | if (! (value == 0x09 || value == 0x0A || value == 0x0D | |
407 | || (value >= 0x20 && value <= 0x7E) | |
408 | || (value == 0x85) || (value >= 0xA0 && value <= 0xD7FF) | |
409 | || (value >= 0xE000 && value <= 0xFFFD) | |
410 | || (value >= 0x10000 && value <= 0x10FFFF))) | |
411 | return yaml_parser_set_reader_error(parser, | |
412 | "Control characters are not allowed", | |
413 | parser->offset, value); | |
414 | ||
415 | /* Move the raw pointers. */ | |
416 | ||
417 | parser->raw_buffer.pointer += width; | |
418 | parser->offset += width; | |
419 | ||
420 | /* Finally put the character into the buffer. */ | |
421 | ||
422 | /* 0000 0000-0000 007F -> 0xxxxxxx */ | |
423 | if (value <= 0x7F) { | |
424 | *(parser->buffer.last++) = value; | |
425 | } | |
426 | /* 0000 0080-0000 07FF -> 110xxxxx 10xxxxxx */ | |
427 | else if (value <= 0x7FF) { | |
428 | *(parser->buffer.last++) = 0xC0 + (value >> 6); | |
429 | *(parser->buffer.last++) = 0x80 + (value & 0x3F); | |
430 | } | |
431 | /* 0000 0800-0000 FFFF -> 1110xxxx 10xxxxxx 10xxxxxx */ | |
432 | else if (value <= 0xFFFF) { | |
433 | *(parser->buffer.last++) = 0xE0 + (value >> 12); | |
434 | *(parser->buffer.last++) = 0x80 + ((value >> 6) & 0x3F); | |
435 | *(parser->buffer.last++) = 0x80 + (value & 0x3F); | |
436 | } | |
437 | /* 0001 0000-0010 FFFF -> 11110xxx 10xxxxxx 10xxxxxx 10xxxxxx */ | |
438 | else { | |
439 | *(parser->buffer.last++) = 0xF0 + (value >> 18); | |
440 | *(parser->buffer.last++) = 0x80 + ((value >> 12) & 0x3F); | |
441 | *(parser->buffer.last++) = 0x80 + ((value >> 6) & 0x3F); | |
442 | *(parser->buffer.last++) = 0x80 + (value & 0x3F); | |
443 | } | |
444 | ||
445 | parser->unread ++; | |
446 | } | |
447 | ||
448 | /* On EOF, put NUL into the buffer and return. */ | |
449 | ||
450 | if (parser->eof) { | |
451 | *(parser->buffer.last++) = '\0'; | |
452 | parser->unread ++; | |
453 | return 1; | |
454 | } | |
455 | ||
456 | } | |
457 | ||
458 | return 1; | |
459 | } | |
460 |