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)) {
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
yaml_parser_update_raw_buffer(yaml_parser_t * parser)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)
yaml_parser_update_buffer(yaml_parser_t * parser,size_t length)142 yaml_parser_update_buffer(yaml_parser_t *parser, size_t length)
143 {
144 int first = 1;
145
146 assert(parser->read_handler); /* Read handler must be set. */
147
148 /* If the EOF flag is set and the raw buffer is empty, do nothing. */
149
150 if (parser->eof && parser->raw_buffer.pointer == parser->raw_buffer.last)
151 return 1;
152
153 /* Return if the buffer contains enough characters. */
154
155 if (parser->unread >= length)
156 return 1;
157
158 /* Determine the input encoding if it is not known yet. */
159
160 if (!parser->encoding) {
161 if (!yaml_parser_determine_encoding(parser))
162 return 0;
163 }
164
165 /* Move the unread characters to the beginning of the buffer. */
166
167 if (parser->buffer.start < parser->buffer.pointer
168 && parser->buffer.pointer < parser->buffer.last) {
169 size_t size = parser->buffer.last - parser->buffer.pointer;
170 memmove(parser->buffer.start, parser->buffer.pointer, size);
171 parser->buffer.pointer = parser->buffer.start;
172 parser->buffer.last = parser->buffer.start + size;
173 }
174 else if (parser->buffer.pointer == parser->buffer.last) {
175 parser->buffer.pointer = parser->buffer.start;
176 parser->buffer.last = parser->buffer.start;
177 }
178
179 /* Fill the buffer until it has enough characters. */
180
181 while (parser->unread < length)
182 {
183 /* Fill the raw buffer if necessary. */
184
185 if (!first || parser->raw_buffer.pointer == parser->raw_buffer.last) {
186 if (!yaml_parser_update_raw_buffer(parser)) return 0;
187 }
188 first = 0;
189
190 /* Decode the raw buffer. */
191
192 while (parser->raw_buffer.pointer != parser->raw_buffer.last)
193 {
194 unsigned int value = 0, value2 = 0;
195 int incomplete = 0;
196 unsigned char octet;
197 unsigned int width = 0;
198 int low, high;
199 size_t k;
200 size_t raw_unread = parser->raw_buffer.last - parser->raw_buffer.pointer;
201
202 /* Decode the next character. */
203
204 switch (parser->encoding)
205 {
206 case YAML_UTF8_ENCODING:
207
208 /*
209 * Decode a UTF-8 character. Check RFC 3629
210 * (http://www.ietf.org/rfc/rfc3629.txt) for more details.
211 *
212 * The following table (taken from the RFC) is used for
213 * decoding.
214 *
215 * Char. number range | UTF-8 octet sequence
216 * (hexadecimal) | (binary)
217 * --------------------+------------------------------------
218 * 0000 0000-0000 007F | 0xxxxxxx
219 * 0000 0080-0000 07FF | 110xxxxx 10xxxxxx
220 * 0000 0800-0000 FFFF | 1110xxxx 10xxxxxx 10xxxxxx
221 * 0001 0000-0010 FFFF | 11110xxx 10xxxxxx 10xxxxxx 10xxxxxx
222 *
223 * Additionally, the characters in the range 0xD800-0xDFFF
224 * are prohibited as they are reserved for use with UTF-16
225 * surrogate pairs.
226 */
227
228 /* Determine the length of the UTF-8 sequence. */
229
230 octet = parser->raw_buffer.pointer[0];
231 width = (octet & 0x80) == 0x00 ? 1 :
232 (octet & 0xE0) == 0xC0 ? 2 :
233 (octet & 0xF0) == 0xE0 ? 3 :
234 (octet & 0xF8) == 0xF0 ? 4 : 0;
235
236 /* Check if the leading octet is valid. */
237
238 if (!width)
239 return yaml_parser_set_reader_error(parser,
240 "invalid leading UTF-8 octet",
241 parser->offset, octet);
242
243 /* Check if the raw buffer contains an incomplete character. */
244
245 if (width > raw_unread) {
246 if (parser->eof) {
247 return yaml_parser_set_reader_error(parser,
248 "incomplete UTF-8 octet sequence",
249 parser->offset, -1);
250 }
251 incomplete = 1;
252 break;
253 }
254
255 /* Decode the leading octet. */
256
257 value = (octet & 0x80) == 0x00 ? octet & 0x7F :
258 (octet & 0xE0) == 0xC0 ? octet & 0x1F :
259 (octet & 0xF0) == 0xE0 ? octet & 0x0F :
260 (octet & 0xF8) == 0xF0 ? octet & 0x07 : 0;
261
262 /* Check and decode the trailing octets. */
263
264 for (k = 1; k < width; k ++)
265 {
266 octet = parser->raw_buffer.pointer[k];
267
268 /* Check if the octet is valid. */
269
270 if ((octet & 0xC0) != 0x80)
271 return yaml_parser_set_reader_error(parser,
272 "invalid trailing UTF-8 octet",
273 parser->offset+k, octet);
274
275 /* Decode the octet. */
276
277 value = (value << 6) + (octet & 0x3F);
278 }
279
280 /* Check the length of the sequence against the value. */
281
282 if (!((width == 1) ||
283 (width == 2 && value >= 0x80) ||
284 (width == 3 && value >= 0x800) ||
285 (width == 4 && value >= 0x10000)))
286 return yaml_parser_set_reader_error(parser,
287 "invalid length of a UTF-8 sequence",
288 parser->offset, -1);
289
290 /* Check the range of the value. */
291
292 if ((value >= 0xD800 && value <= 0xDFFF) || value > 0x10FFFF)
293 return yaml_parser_set_reader_error(parser,
294 "invalid Unicode character",
295 parser->offset, value);
296
297 break;
298
299 case YAML_UTF16LE_ENCODING:
300 case YAML_UTF16BE_ENCODING:
301
302 low = (parser->encoding == YAML_UTF16LE_ENCODING ? 0 : 1);
303 high = (parser->encoding == YAML_UTF16LE_ENCODING ? 1 : 0);
304
305 /*
306 * The UTF-16 encoding is not as simple as one might
307 * naively think. Check RFC 2781
308 * (http://www.ietf.org/rfc/rfc2781.txt).
309 *
310 * Normally, two subsequent bytes describe a Unicode
311 * character. However a special technique (called a
312 * surrogate pair) is used for specifying character
313 * values larger than 0xFFFF.
314 *
315 * A surrogate pair consists of two pseudo-characters:
316 * high surrogate area (0xD800-0xDBFF)
317 * low surrogate area (0xDC00-0xDFFF)
318 *
319 * The following formulas are used for decoding
320 * and encoding characters using surrogate pairs:
321 *
322 * U = U' + 0x10000 (0x01 00 00 <= U <= 0x10 FF FF)
323 * U' = yyyyyyyyyyxxxxxxxxxx (0 <= U' <= 0x0F FF FF)
324 * W1 = 110110yyyyyyyyyy
325 * W2 = 110111xxxxxxxxxx
326 *
327 * where U is the character value, W1 is the high surrogate
328 * area, W2 is the low surrogate area.
329 */
330
331 /* Check for incomplete UTF-16 character. */
332
333 if (raw_unread < 2) {
334 if (parser->eof) {
335 return yaml_parser_set_reader_error(parser,
336 "incomplete UTF-16 character",
337 parser->offset, -1);
338 }
339 incomplete = 1;
340 break;
341 }
342
343 /* Get the character. */
344
345 value = parser->raw_buffer.pointer[low]
346 + (parser->raw_buffer.pointer[high] << 8);
347
348 /* Check for unexpected low surrogate area. */
349
350 if ((value & 0xFC00) == 0xDC00)
351 return yaml_parser_set_reader_error(parser,
352 "unexpected low surrogate area",
353 parser->offset, value);
354
355 /* Check for a high surrogate area. */
356
357 if ((value & 0xFC00) == 0xD800) {
358
359 width = 4;
360
361 /* Check for incomplete surrogate pair. */
362
363 if (raw_unread < 4) {
364 if (parser->eof) {
365 return yaml_parser_set_reader_error(parser,
366 "incomplete UTF-16 surrogate pair",
367 parser->offset, -1);
368 }
369 incomplete = 1;
370 break;
371 }
372
373 /* Get the next character. */
374
375 value2 = parser->raw_buffer.pointer[low+2]
376 + (parser->raw_buffer.pointer[high+2] << 8);
377
378 /* Check for a low surrogate area. */
379
380 if ((value2 & 0xFC00) != 0xDC00)
381 return yaml_parser_set_reader_error(parser,
382 "expected low surrogate area",
383 parser->offset+2, value2);
384
385 /* Generate the value of the surrogate pair. */
386
387 value = 0x10000 + ((value & 0x3FF) << 10) + (value2 & 0x3FF);
388 }
389
390 else {
391 width = 2;
392 }
393
394 break;
395
396 default:
397 assert(1); /* Impossible. */
398 }
399
400 /* Check if the raw buffer contains enough bytes to form a character. */
401
402 if (incomplete) break;
403
404 /*
405 * Check if the character is in the allowed range:
406 * #x9 | #xA | #xD | [#x20-#x7E] (8 bit)
407 * | #x85 | [#xA0-#xD7FF] | [#xE000-#xFFFD] (16 bit)
408 * | [#x10000-#x10FFFF] (32 bit)
409 */
410
411 if (! (value == 0x09 || value == 0x0A || value == 0x0D
412 || (value >= 0x20 && value <= 0x7E)
413 || (value == 0x85) || (value >= 0xA0 && value <= 0xD7FF)
414 || (value >= 0xE000 && value <= 0xFFFD)
415 || (value >= 0x10000 && value <= 0x10FFFF)))
416 return yaml_parser_set_reader_error(parser,
417 "control characters are not allowed",
418 parser->offset, value);
419
420 /* Move the raw pointers. */
421
422 parser->raw_buffer.pointer += width;
423 parser->offset += width;
424
425 /* Finally put the character into the buffer. */
426
427 /* 0000 0000-0000 007F -> 0xxxxxxx */
428 if (value <= 0x7F) {
429 *(parser->buffer.last++) = value;
430 }
431 /* 0000 0080-0000 07FF -> 110xxxxx 10xxxxxx */
432 else if (value <= 0x7FF) {
433 *(parser->buffer.last++) = 0xC0 + (value >> 6);
434 *(parser->buffer.last++) = 0x80 + (value & 0x3F);
435 }
436 /* 0000 0800-0000 FFFF -> 1110xxxx 10xxxxxx 10xxxxxx */
437 else if (value <= 0xFFFF) {
438 *(parser->buffer.last++) = 0xE0 + (value >> 12);
439 *(parser->buffer.last++) = 0x80 + ((value >> 6) & 0x3F);
440 *(parser->buffer.last++) = 0x80 + (value & 0x3F);
441 }
442 /* 0001 0000-0010 FFFF -> 11110xxx 10xxxxxx 10xxxxxx 10xxxxxx */
443 else {
444 *(parser->buffer.last++) = 0xF0 + (value >> 18);
445 *(parser->buffer.last++) = 0x80 + ((value >> 12) & 0x3F);
446 *(parser->buffer.last++) = 0x80 + ((value >> 6) & 0x3F);
447 *(parser->buffer.last++) = 0x80 + (value & 0x3F);
448 }
449
450 parser->unread ++;
451 }
452
453 /* On EOF, put NUL into the buffer and return. */
454
455 if (parser->eof) {
456 *(parser->buffer.last++) = '\0';
457 parser->unread ++;
458 return 1;
459 }
460
461 }
462
463 if (parser->offset >= PTRDIFF_MAX)
464 return yaml_parser_set_reader_error(parser, "input is too long",
465 PTRDIFF_MAX, -1);
466
467 return 1;
468 }
469
470