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
yaml_parser_set_reader_error(yaml_parser_t * parser,const char * problem,size_t offset,int value)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
yaml_parser_determine_encoding(yaml_parser_t * parser)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)) return 0;
58 }
59
60 /* Determine the encoding. */
61
62 if (parser->raw_buffer.last - parser->raw_buffer.pointer >= 2
63 && !memcmp(parser->raw_buffer.pointer, BOM_UTF16LE, 2)) {
64 parser->encoding = YAML_UTF16LE_ENCODING;
65 parser->raw_buffer.pointer += 2;
66 parser->offset += 2;
67 }
68 else if (parser->raw_buffer.last - parser->raw_buffer.pointer >= 2
69 && !memcmp(parser->raw_buffer.pointer, BOM_UTF16BE, 2)) {
70 parser->encoding = YAML_UTF16BE_ENCODING;
71 parser->raw_buffer.pointer += 2;
72 parser->offset += 2;
73 }
74 else if (parser->raw_buffer.last - parser->raw_buffer.pointer >= 3
75 && !memcmp(parser->raw_buffer.pointer, BOM_UTF8, 3)) {
76 parser->encoding = YAML_UTF8_ENCODING;
77 parser->raw_buffer.pointer += 3;
78 parser->offset += 3;
79 }
80 else {
81 parser->encoding = YAML_UTF8_ENCODING;
82 }
83
84 return 1;
85 }
86
87 /*
88 * Update the raw buffer.
89 */
90
91 static int
yaml_parser_update_raw_buffer(yaml_parser_t * parser)92 yaml_parser_update_raw_buffer(yaml_parser_t *parser)
93 {
94 size_t size_read = 0;
95
96 /* Return if the raw buffer is full. */
97
98 if (parser->raw_buffer.start == parser->raw_buffer.pointer
99 && parser->raw_buffer.last == parser->raw_buffer.end)
100 return 1;
101
102 /* Return on EOF. */
103
104 if (parser->eof) return 1;
105
106 /* Move the remaining bytes in the raw buffer to the beginning. */
107
108 if (parser->raw_buffer.start < parser->raw_buffer.pointer
109 && parser->raw_buffer.pointer < parser->raw_buffer.last) {
110 memmove(parser->raw_buffer.start, parser->raw_buffer.pointer,
111 parser->raw_buffer.last - parser->raw_buffer.pointer);
112 }
113 parser->raw_buffer.last -=
114 parser->raw_buffer.pointer - parser->raw_buffer.start;
115 parser->raw_buffer.pointer = parser->raw_buffer.start;
116
117 /* Call the read handler to fill the buffer. */
118
119 if (!parser->read_handler(parser->read_handler_data, parser->raw_buffer.last,
120 parser->raw_buffer.end - parser->raw_buffer.last, &size_read)) {
121 return yaml_parser_set_reader_error(parser, "input error",
122 parser->offset, -1);
123 }
124 parser->raw_buffer.last += size_read;
125 if (!size_read) parser->eof = 1;
126
127 return 1;
128 }
129
130 /*
131 * Ensure that the buffer contains at least `length` characters.
132 * Return 1 on success, 0 on failure.
133 *
134 * The length is supposed to be significantly less that the buffer size.
135 */
136
137 YAML_DECLARE(int)
yaml_parser_update_buffer(yaml_parser_t * parser,size_t length)138 yaml_parser_update_buffer(yaml_parser_t *parser, size_t length)
139 {
140 int first = 1;
141
142 assert(parser->read_handler); /* Read handler must be set. */
143
144 /* If the EOF flag is set and the raw buffer is empty, do nothing. */
145
146 if (parser->eof && parser->raw_buffer.pointer == parser->raw_buffer.last)
147 return 1;
148
149 /* Return if the buffer contains enough characters. */
150
151 if (parser->unread >= length)
152 return 1;
153
154 /* Determine the input encoding if it is not known yet. */
155
156 if (!parser->encoding) {
157 if (!yaml_parser_determine_encoding(parser))
158 return 0;
159 }
160
161 /* Move the unread characters to the beginning of the buffer. */
162
163 if (parser->buffer.start < parser->buffer.pointer
164 && parser->buffer.pointer < parser->buffer.last) {
165 size_t size = parser->buffer.last - parser->buffer.pointer;
166 memmove(parser->buffer.start, parser->buffer.pointer, size);
167 parser->buffer.pointer = parser->buffer.start;
168 parser->buffer.last = parser->buffer.start + size;
169 }
170 else if (parser->buffer.pointer == parser->buffer.last) {
171 parser->buffer.pointer = parser->buffer.start;
172 parser->buffer.last = parser->buffer.start;
173 }
174
175 /* Fill the buffer until it has enough characters. */
176
177 while (parser->unread < length)
178 {
179 /* Fill the raw buffer if necessary. */
180
181 if (!first || parser->raw_buffer.pointer == parser->raw_buffer.last) {
182 if (!yaml_parser_update_raw_buffer(parser)) return 0;
183 }
184 first = 0;
185
186 /* Decode the raw buffer. */
187
188 while (parser->raw_buffer.pointer != parser->raw_buffer.last)
189 {
190 unsigned int value = 0, value2 = 0;
191 int incomplete = 0;
192 unsigned char octet;
193 unsigned int width = 0;
194 int low, high;
195 size_t k;
196 size_t raw_unread = parser->raw_buffer.last - parser->raw_buffer.pointer;
197
198 /* Decode the next character. */
199
200 switch (parser->encoding)
201 {
202 case YAML_UTF8_ENCODING:
203
204 /*
205 * Decode a UTF-8 character. Check RFC 3629
206 * (http://www.ietf.org/rfc/rfc3629.txt) for more details.
207 *
208 * The following table (taken from the RFC) is used for
209 * decoding.
210 *
211 * Char. number range | UTF-8 octet sequence
212 * (hexadecimal) | (binary)
213 * --------------------+------------------------------------
214 * 0000 0000-0000 007F | 0xxxxxxx
215 * 0000 0080-0000 07FF | 110xxxxx 10xxxxxx
216 * 0000 0800-0000 FFFF | 1110xxxx 10xxxxxx 10xxxxxx
217 * 0001 0000-0010 FFFF | 11110xxx 10xxxxxx 10xxxxxx 10xxxxxx
218 *
219 * Additionally, the characters in the range 0xD800-0xDFFF
220 * are prohibited as they are reserved for use with UTF-16
221 * surrogate pairs.
222 */
223
224 /* Determine the length of the UTF-8 sequence. */
225
226 octet = parser->raw_buffer.pointer[0];
227 width = (octet & 0x80) == 0x00 ? 1 :
228 (octet & 0xE0) == 0xC0 ? 2 :
229 (octet & 0xF0) == 0xE0 ? 3 :
230 (octet & 0xF8) == 0xF0 ? 4 : 0;
231
232 /* Check if the leading octet is valid. */
233
234 if (!width)
235 return yaml_parser_set_reader_error(parser,
236 "invalid leading UTF-8 octet",
237 parser->offset, octet);
238
239 /* Check if the raw buffer contains an incomplete character. */
240
241 if (width > raw_unread) {
242 if (parser->eof) {
243 return yaml_parser_set_reader_error(parser,
244 "incomplete UTF-8 octet sequence",
245 parser->offset, -1);
246 }
247 incomplete = 1;
248 break;
249 }
250
251 /* Decode the leading octet. */
252
253 value = (octet & 0x80) == 0x00 ? octet & 0x7F :
254 (octet & 0xE0) == 0xC0 ? octet & 0x1F :
255 (octet & 0xF0) == 0xE0 ? octet & 0x0F :
256 (octet & 0xF8) == 0xF0 ? octet & 0x07 : 0;
257
258 /* Check and decode the trailing octets. */
259
260 for (k = 1; k < width; k ++)
261 {
262 octet = parser->raw_buffer.pointer[k];
263
264 /* Check if the octet is valid. */
265
266 if ((octet & 0xC0) != 0x80)
267 return yaml_parser_set_reader_error(parser,
268 "invalid trailing UTF-8 octet",
269 parser->offset+k, octet);
270
271 /* Decode the octet. */
272
273 value = (value << 6) + (octet & 0x3F);
274 }
275
276 /* Check the length of the sequence against the value. */
277
278 if (!((width == 1) ||
279 (width == 2 && value >= 0x80) ||
280 (width == 3 && value >= 0x800) ||
281 (width == 4 && value >= 0x10000)))
282 return yaml_parser_set_reader_error(parser,
283 "invalid length of a UTF-8 sequence",
284 parser->offset, -1);
285
286 /* Check the range of the value. */
287
288 if ((value >= 0xD800 && value <= 0xDFFF) || value > 0x10FFFF)
289 return yaml_parser_set_reader_error(parser,
290 "invalid Unicode character",
291 parser->offset, value);
292
293 break;
294
295 case YAML_UTF16LE_ENCODING:
296 case YAML_UTF16BE_ENCODING:
297
298 low = (parser->encoding == YAML_UTF16LE_ENCODING ? 0 : 1);
299 high = (parser->encoding == YAML_UTF16LE_ENCODING ? 1 : 0);
300
301 /*
302 * The UTF-16 encoding is not as simple as one might
303 * naively think. Check RFC 2781
304 * (http://www.ietf.org/rfc/rfc2781.txt).
305 *
306 * Normally, two subsequent bytes describe a Unicode
307 * character. However a special technique (called a
308 * surrogate pair) is used for specifying character
309 * values larger than 0xFFFF.
310 *
311 * A surrogate pair consists of two pseudo-characters:
312 * high surrogate area (0xD800-0xDBFF)
313 * low surrogate area (0xDC00-0xDFFF)
314 *
315 * The following formulas are used for decoding
316 * and encoding characters using surrogate pairs:
317 *
318 * U = U' + 0x10000 (0x01 00 00 <= U <= 0x10 FF FF)
319 * U' = yyyyyyyyyyxxxxxxxxxx (0 <= U' <= 0x0F FF FF)
320 * W1 = 110110yyyyyyyyyy
321 * W2 = 110111xxxxxxxxxx
322 *
323 * where U is the character value, W1 is the high surrogate
324 * area, W2 is the low surrogate area.
325 */
326
327 /* Check for incomplete UTF-16 character. */
328
329 if (raw_unread < 2) {
330 if (parser->eof) {
331 return yaml_parser_set_reader_error(parser,
332 "incomplete UTF-16 character",
333 parser->offset, -1);
334 }
335 incomplete = 1;
336 break;
337 }
338
339 /* Get the character. */
340
341 value = parser->raw_buffer.pointer[low]
342 + (parser->raw_buffer.pointer[high] << 8);
343
344 /* Check for unexpected low surrogate area. */
345
346 if ((value & 0xFC00) == 0xDC00)
347 return yaml_parser_set_reader_error(parser,
348 "unexpected low surrogate area",
349 parser->offset, value);
350
351 /* Check for a high surrogate area. */
352
353 if ((value & 0xFC00) == 0xD800) {
354
355 width = 4;
356
357 /* Check for incomplete surrogate pair. */
358
359 if (raw_unread < 4) {
360 if (parser->eof) {
361 return yaml_parser_set_reader_error(parser,
362 "incomplete UTF-16 surrogate pair",
363 parser->offset, -1);
364 }
365 incomplete = 1;
366 break;
367 }
368
369 /* Get the next character. */
370
371 value2 = parser->raw_buffer.pointer[low+2]
372 + (parser->raw_buffer.pointer[high+2] << 8);
373
374 /* Check for a low surrogate area. */
375
376 if ((value2 & 0xFC00) != 0xDC00)
377 return yaml_parser_set_reader_error(parser,
378 "expected low surrogate area",
379 parser->offset+2, value2);
380
381 /* Generate the value of the surrogate pair. */
382
383 value = 0x10000 + ((value & 0x3FF) << 10) + (value2 & 0x3FF);
384 }
385
386 else {
387 width = 2;
388 }
389
390 break;
391
392 default:
393 assert(1); /* Impossible. */
394 }
395
396 /* Check if the raw buffer contains enough bytes to form a character. */
397
398 if (incomplete) break;
399
400 /*
401 * Check if the character is in the allowed range:
402 * #x9 | #xA | #xD | [#x20-#x7E] (8 bit)
403 * | #x85 | [#xA0-#xD7FF] | [#xE000-#xFFFD] (16 bit)
404 * | [#x10000-#x10FFFF] (32 bit)
405 */
406
407 if (! (value == 0x09 || value == 0x0A || value == 0x0D
408 || (value >= 0x20 && value <= 0x7E)
409 || (value == 0x85) || (value >= 0xA0 && value <= 0xD7FF)
410 || (value >= 0xE000 && value <= 0xFFFD)
411 || (value >= 0x10000 && value <= 0x10FFFF)))
412 return yaml_parser_set_reader_error(parser,
413 "control characters are not allowed",
414 parser->offset, value);
415
416 /* Move the raw pointers. */
417
418 parser->raw_buffer.pointer += width;
419 parser->offset += width;
420
421 /* Finally put the character into the buffer. */
422
423 /* 0000 0000-0000 007F -> 0xxxxxxx */
424 if (value <= 0x7F) *(parser->buffer.last++) = value;
425 /* 0000 0080-0000 07FF -> 110xxxxx 10xxxxxx */
426 else if (value <= 0x7FF) {
427 *(parser->buffer.last++) = 0xC0 + (value >> 6);
428 *(parser->buffer.last++) = 0x80 + (value & 0x3F);
429 }
430 /* 0000 0800-0000 FFFF -> 1110xxxx 10xxxxxx 10xxxxxx */
431 else if (value <= 0xFFFF) {
432 *(parser->buffer.last++) = 0xE0 + (value >> 12);
433 *(parser->buffer.last++) = 0x80 + ((value >> 6) & 0x3F);
434 *(parser->buffer.last++) = 0x80 + (value & 0x3F);
435 }
436 /* 0001 0000-0010 FFFF -> 11110xxx 10xxxxxx 10xxxxxx 10xxxxxx */
437 else {
438 *(parser->buffer.last++) = 0xF0 + (value >> 18);
439 *(parser->buffer.last++) = 0x80 + ((value >> 12) & 0x3F);
440 *(parser->buffer.last++) = 0x80 + ((value >> 6) & 0x3F);
441 *(parser->buffer.last++) = 0x80 + (value & 0x3F);
442 }
443
444 parser->unread ++;
445 }
446
447 /* On EOF, put NUL into the buffer and return. */
448
449 if (parser->eof) {
450 *(parser->buffer.last++) = '\0';
451 parser->unread ++;
452 return 1;
453 }
454
455 }
456
457 if (parser->offset >= MAX_FILE_SIZE) {
458 return yaml_parser_set_reader_error(parser, "input is too long",
459 parser->offset, -1);
460 }
461
462 return 1;
463 }
464