]> andersk Git - libyaml.git/blame - src/scanner.c
andersk / libyaml.git / blame
? search:
summary | shortlog | log | commit | commitdiff | tree
blob | blame (incremental) | history | HEAD
Allow colons in plain scalars inside flow collections
[libyaml.git] / src / scanner.c
CommitLineData
03be97ab
KS		 1
2/*
3 * Introduction
4 * ************
5 *
6 * The following notes assume that you are familiar with the YAML specification
7 * (http://yaml.org/spec/cvs/current.html). We mostly follow it, although in
8 * some cases we are less restrictive that it requires.
9 *
10 * The process of transforming a YAML stream into a sequence of events is
11 * divided on two steps: Scanning and Parsing.
12 *
13 * The Scanner transforms the input stream into a sequence of tokens, while the
14 * parser transform the sequence of tokens produced by the Scanner into a
15 * sequence of parsing events.
16 *
17 * The Scanner is rather clever and complicated. The Parser, on the contrary,
18 * is a straightforward implementation of a recursive-descendant parser (or,
19 * LL(1) parser, as it is usually called).
20 *
21 * Actually there are two issues of Scanning that might be called "clever", the
22 * rest is quite straightforward. The issues are "block collection start" and
23 * "simple keys". Both issues are explained below in details.
24 *
25 * Here the Scanning step is explained and implemented. We start with the list
26 * of all the tokens produced by the Scanner together with short descriptions.
27 *
28 * Now, tokens:
29 *
30 * STREAM-START(encoding) # The stream start.
31 * STREAM-END # The stream end.
32 * VERSION-DIRECTIVE(major,minor) # The '%YAML' directive.
33 * TAG-DIRECTIVE(handle,prefix) # The '%TAG' directive.
34 * DOCUMENT-START # '---'
35 * DOCUMENT-END # '...'
36 * BLOCK-SEQUENCE-START # Indentation increase denoting a block
37 * BLOCK-MAPPING-START # sequence or a block mapping.
38 * BLOCK-END # Indentation decrease.
39 * FLOW-SEQUENCE-START # '['
40 * FLOW-SEQUENCE-END # ']'
2e849b36
VS		 41 * FLOW-MAPPING-START # '{'
42 * FLOW-MAPPING-END # '}'
03be97ab
KS		 43 * BLOCK-ENTRY # '-'
44 * FLOW-ENTRY # ','
45 * KEY # '?' or nothing (simple keys).
46 * VALUE # ':'
47 * ALIAS(anchor) # '*anchor'
48 * ANCHOR(anchor) # '&anchor'
49 * TAG(handle,suffix) # '!handle!suffix'
50 * SCALAR(value,style) # A scalar.
51 *
52 * The following two tokens are "virtual" tokens denoting the beginning and the
53 * end of the stream:
54 *
55 * STREAM-START(encoding)
56 * STREAM-END
57 *
58 * We pass the information about the input stream encoding with the
59 * STREAM-START token.
60 *
61 * The next two tokens are responsible for tags:
62 *
63 * VERSION-DIRECTIVE(major,minor)
64 * TAG-DIRECTIVE(handle,prefix)
65 *
66 * Example:
67 *
68 * %YAML 1.1
69 * %TAG ! !foo
70 * %TAG !yaml! tag:yaml.org,2002:
71 * ---
72 *
bdf4d192 73 * The corresponding sequence of tokens:
03be97ab
KS		 74 *
75 * STREAM-START(utf-8)
76 * VERSION-DIRECTIVE(1,1)
77 * TAG-DIRECTIVE("!","!foo")
78 * TAG-DIRECTIVE("!yaml","tag:yaml.org,2002:")
79 * DOCUMENT-START
80 * STREAM-END
81 *
82 * Note that the VERSION-DIRECTIVE and TAG-DIRECTIVE tokens occupy a whole
83 * line.
84 *
85 * The document start and end indicators are represented by:
86 *
87 * DOCUMENT-START
88 * DOCUMENT-END
89 *
90 * Note that if a YAML stream contains an implicit document (without '---'
91 * and '...' indicators), no DOCUMENT-START and DOCUMENT-END tokens will be
92 * produced.
93 *
94 * In the following examples, we present whole documents together with the
95 * produced tokens.
96 *
97 * 1. An implicit document:
98 *
99 * 'a scalar'
100 *
101 * Tokens:
102 *
103 * STREAM-START(utf-8)
104 * SCALAR("a scalar",single-quoted)
105 * STREAM-END
106 *
107 * 2. An explicit document:
108 *
109 * ---
110 * 'a scalar'
111 * ...
112 *
113 * Tokens:
114 *
115 * STREAM-START(utf-8)
116 * DOCUMENT-START
117 * SCALAR("a scalar",single-quoted)
118 * DOCUMENT-END
119 * STREAM-END
120 *
121 * 3. Several documents in a stream:
122 *
123 * 'a scalar'
124 * ---
125 * 'another scalar'
126 * ---
127 * 'yet another scalar'
128 *
129 * Tokens:
130 *
131 * STREAM-START(utf-8)
132 * SCALAR("a scalar",single-quoted)
133 * DOCUMENT-START
134 * SCALAR("another scalar",single-quoted)
135 * DOCUMENT-START
136 * SCALAR("yet another scalar",single-quoted)
137 * STREAM-END
138 *
139 * We have already introduced the SCALAR token above. The following tokens are
140 * used to describe aliases, anchors, tag, and scalars:
141 *
142 * ALIAS(anchor)
143 * ANCHOR(anchor)
144 * TAG(handle,suffix)
145 * SCALAR(value,style)
146 *
147 * The following series of examples illustrate the usage of these tokens:
148 *
149 * 1. A recursive sequence:
150 *
151 * &A [ *A ]
152 *
153 * Tokens:
154 *
155 * STREAM-START(utf-8)
156 * ANCHOR("A")
157 * FLOW-SEQUENCE-START
158 * ALIAS("A")
159 * FLOW-SEQUENCE-END
160 * STREAM-END
161 *
162 * 2. A tagged scalar:
163 *
164 * !!float "3.14" # A good approximation.
165 *
166 * Tokens:
167 *
168 * STREAM-START(utf-8)
169 * TAG("!!","float")
170 * SCALAR("3.14",double-quoted)
171 * STREAM-END
172 *
173 * 3. Various scalar styles:
174 *
175 * --- # Implicit empty plain scalars do not produce tokens.
176 * --- a plain scalar
177 * --- 'a single-quoted scalar'
178 * --- "a double-quoted scalar"
179 * --- |-
180 * a literal scalar
181 * --- >-
182 * a folded
183 * scalar
184 *
185 * Tokens:
186 *
187 * STREAM-START(utf-8)
188 * DOCUMENT-START
189 * DOCUMENT-START
190 * SCALAR("a plain scalar",plain)
191 * DOCUMENT-START
192 * SCALAR("a single-quoted scalar",single-quoted)
193 * DOCUMENT-START
194 * SCALAR("a double-quoted scalar",double-quoted)
195 * DOCUMENT-START
196 * SCALAR("a literal scalar",literal)
197 * DOCUMENT-START
198 * SCALAR("a folded scalar",folded)
199 * STREAM-END
200 *
201 * Now it's time to review collection-related tokens. We will start with
202 * flow collections:
203 *
204 * FLOW-SEQUENCE-START
205 * FLOW-SEQUENCE-END
206 * FLOW-MAPPING-START
207 * FLOW-MAPPING-END
208 * FLOW-ENTRY
209 * KEY
210 * VALUE
211 *
212 * The tokens FLOW-SEQUENCE-START, FLOW-SEQUENCE-END, FLOW-MAPPING-START, and
213 * FLOW-MAPPING-END represent the indicators '[', ']', '{', and '}'
214 * correspondingly. FLOW-ENTRY represent the ',' indicator. Finally the
215 * indicators '?' and ':', which are used for denoting mapping keys and values,
216 * are represented by the KEY and VALUE tokens.
217 *
218 * The following examples show flow collections:
219 *
220 * 1. A flow sequence:
221 *
222 * [item 1, item 2, item 3]
223 *
224 * Tokens:
225 *
226 * STREAM-START(utf-8)
227 * FLOW-SEQUENCE-START
228 * SCALAR("item 1",plain)
229 * FLOW-ENTRY
230 * SCALAR("item 2",plain)
231 * FLOW-ENTRY
232 * SCALAR("item 3",plain)
233 * FLOW-SEQUENCE-END
234 * STREAM-END
235 *
236 * 2. A flow mapping:
237 *
238 * {
239 * a simple key: a value, # Note that the KEY token is produced.
240 * ? a complex key: another value,
241 * }
242 *
243 * Tokens:
244 *
245 * STREAM-START(utf-8)
246 * FLOW-MAPPING-START
247 * KEY
248 * SCALAR("a simple key",plain)
249 * VALUE
250 * SCALAR("a value",plain)
251 * FLOW-ENTRY
252 * KEY
253 * SCALAR("a complex key",plain)
254 * VALUE
255 * SCALAR("another value",plain)
256 * FLOW-ENTRY
257 * FLOW-MAPPING-END
258 * STREAM-END
259 *
260 * A simple key is a key which is not denoted by the '?' indicator. Note that
261 * the Scanner still produce the KEY token whenever it encounters a simple key.
262 *
263 * For scanning block collections, the following tokens are used (note that we
264 * repeat KEY and VALUE here):
265 *
266 * BLOCK-SEQUENCE-START
267 * BLOCK-MAPPING-START
268 * BLOCK-END
269 * BLOCK-ENTRY
270 * KEY
271 * VALUE
272 *
273 * The tokens BLOCK-SEQUENCE-START and BLOCK-MAPPING-START denote indentation
274 * increase that precedes a block collection (cf. the INDENT token in Python).
275 * The token BLOCK-END denote indentation decrease that ends a block collection
276 * (cf. the DEDENT token in Python). However YAML has some syntax pecularities
277 * that makes detections of these tokens more complex.
278 *
279 * The tokens BLOCK-ENTRY, KEY, and VALUE are used to represent the indicators
280 * '-', '?', and ':' correspondingly.
281 *
282 * The following examples show how the tokens BLOCK-SEQUENCE-START,
283 * BLOCK-MAPPING-START, and BLOCK-END are emitted by the Scanner:
284 *
285 * 1. Block sequences:
286 *
287 * - item 1
288 * - item 2
289 * -
290 * - item 3.1
291 * - item 3.2
292 * -
293 * key 1: value 1
294 * key 2: value 2
295 *
296 * Tokens:
297 *
298 * STREAM-START(utf-8)
299 * BLOCK-SEQUENCE-START
300 * BLOCK-ENTRY
301 * SCALAR("item 1",plain)
302 * BLOCK-ENTRY
303 * SCALAR("item 2",plain)
304 * BLOCK-ENTRY
305 * BLOCK-SEQUENCE-START
306 * BLOCK-ENTRY
307 * SCALAR("item 3.1",plain)
308 * BLOCK-ENTRY
309 * SCALAR("item 3.2",plain)
310 * BLOCK-END
311 * BLOCK-ENTRY
312 * BLOCK-MAPPING-START
313 * KEY
314 * SCALAR("key 1",plain)
315 * VALUE
316 * SCALAR("value 1",plain)
317 * KEY
318 * SCALAR("key 2",plain)
319 * VALUE
320 * SCALAR("value 2",plain)
321 * BLOCK-END
322 * BLOCK-END
323 * STREAM-END
324 *
325 * 2. Block mappings:
326 *
327 * a simple key: a value # The KEY token is produced here.
328 * ? a complex key
329 * : another value
330 * a mapping:
331 * key 1: value 1
332 * key 2: value 2
333 * a sequence:
334 * - item 1
335 * - item 2
336 *
337 * Tokens:
338 *
339 * STREAM-START(utf-8)
340 * BLOCK-MAPPING-START
341 * KEY
342 * SCALAR("a simple key",plain)
343 * VALUE
344 * SCALAR("a value",plain)
345 * KEY
346 * SCALAR("a complex key",plain)
347 * VALUE
348 * SCALAR("another value",plain)
349 * KEY
350 * SCALAR("a mapping",plain)
351 * BLOCK-MAPPING-START
352 * KEY
353 * SCALAR("key 1",plain)
354 * VALUE
355 * SCALAR("value 1",plain)
356 * KEY
357 * SCALAR("key 2",plain)
358 * VALUE
359 * SCALAR("value 2",plain)
360 * BLOCK-END
361 * KEY
362 * SCALAR("a sequence",plain)
363 * VALUE
364 * BLOCK-SEQUENCE-START
365 * BLOCK-ENTRY
366 * SCALAR("item 1",plain)
367 * BLOCK-ENTRY
368 * SCALAR("item 2",plain)
369 * BLOCK-END
370 * BLOCK-END
371 * STREAM-END
372 *
373 * YAML does not always require to start a new block collection from a new
374 * line. If the current line contains only '-', '?', and ':' indicators, a new
375 * block collection may start at the current line. The following examples
376 * illustrate this case:
377 *
378 * 1. Collections in a sequence:
379 *
380 * - - item 1
381 * - item 2
382 * - key 1: value 1
383 * key 2: value 2
384 * - ? complex key
385 * : complex value
386 *
387 * Tokens:
388 *
389 * STREAM-START(utf-8)
390 * BLOCK-SEQUENCE-START
391 * BLOCK-ENTRY
392 * BLOCK-SEQUENCE-START
393 * BLOCK-ENTRY
394 * SCALAR("item 1",plain)
395 * BLOCK-ENTRY
396 * SCALAR("item 2",plain)
397 * BLOCK-END
398 * BLOCK-ENTRY
399 * BLOCK-MAPPING-START
400 * KEY
401 * SCALAR("key 1",plain)
402 * VALUE
403 * SCALAR("value 1",plain)
404 * KEY
405 * SCALAR("key 2",plain)
406 * VALUE
407 * SCALAR("value 2",plain)
408 * BLOCK-END
409 * BLOCK-ENTRY
410 * BLOCK-MAPPING-START
411 * KEY
412 * SCALAR("complex key")
413 * VALUE
414 * SCALAR("complex value")
415 * BLOCK-END
416 * BLOCK-END
417 * STREAM-END
418 *
419 * 2. Collections in a mapping:
420 *
421 * ? a sequence
422 * : - item 1
423 * - item 2
424 * ? a mapping
425 * : key 1: value 1
426 * key 2: value 2
427 *
428 * Tokens:
429 *
430 * STREAM-START(utf-8)
431 * BLOCK-MAPPING-START
432 * KEY
433 * SCALAR("a sequence",plain)
434 * VALUE
435 * BLOCK-SEQUENCE-START
436 * BLOCK-ENTRY
437 * SCALAR("item 1",plain)
438 * BLOCK-ENTRY
439 * SCALAR("item 2",plain)
440 * BLOCK-END
441 * KEY
442 * SCALAR("a mapping",plain)
443 * VALUE
444 * BLOCK-MAPPING-START
445 * KEY
446 * SCALAR("key 1",plain)
447 * VALUE
448 * SCALAR("value 1",plain)
449 * KEY
450 * SCALAR("key 2",plain)
451 * VALUE
452 * SCALAR("value 2",plain)
453 * BLOCK-END
454 * BLOCK-END
455 * STREAM-END
456 *
457 * YAML also permits non-indented sequences if they are included into a block
458 * mapping. In this case, the token BLOCK-SEQUENCE-START is not produced:
459 *
460 * key:
461 * - item 1 # BLOCK-SEQUENCE-START is NOT produced here.
462 * - item 2
463 *
464 * Tokens:
465 *
466 * STREAM-START(utf-8)
467 * BLOCK-MAPPING-START
468 * KEY
469 * SCALAR("key",plain)
470 * VALUE
471 * BLOCK-ENTRY
472 * SCALAR("item 1",plain)
473 * BLOCK-ENTRY
474 * SCALAR("item 2",plain)
475 * BLOCK-END
476 */
477
625fcfe9 478#include "yaml_private.h"
03be97ab 479
f2b59d4d
KS		 480/*
481 * Ensure that the buffer contains the required number of characters.
482 * Return 1 on success, 0 on failure (reader error or memory error).
483 */
484
625fcfe9
KS		 485#define CACHE(parser,length) \
486 (parser->unread >= (length) \
487 ? 1 \
f2b59d4d
KS		 488 : yaml_parser_update_buffer(parser, (length)))
489
eb9cceb5
KS		 490/*
491 * Advance the buffer pointer.
492 */
493
625fcfe9
KS		 494#define SKIP(parser) \
495 (parser->mark.index ++, \
496 parser->mark.column ++, \
497 parser->unread --, \
e35af832 498 parser->buffer.pointer += WIDTH(parser->buffer))
e71095e3 499
625fcfe9 500#define SKIP_LINE(parser) \
e35af832 501 (IS_CRLF(parser->buffer) ? \
625fcfe9
KS		 502 (parser->mark.index += 2, \
503 parser->mark.column = 0, \
504 parser->mark.line ++, \
505 parser->unread -= 2, \
506 parser->buffer.pointer += 2) : \
e35af832 507 IS_BREAK(parser->buffer) ? \
625fcfe9
KS		 508 (parser->mark.index ++, \
509 parser->mark.column = 0, \
510 parser->mark.line ++, \
511 parser->unread --, \
e35af832 512 parser->buffer.pointer += WIDTH(parser->buffer)) : 0)
e71095e3
KS		 513
514/*
515 * Copy a character to a string buffer and advance pointers.
516 */
517
625fcfe9
KS		 518#define READ(parser,string) \
519 (STRING_EXTEND(parser,string) ? \
e35af832 520 (COPY(string,parser->buffer), \
625fcfe9
KS		 521 parser->mark.index ++, \
522 parser->mark.column ++, \
523 parser->unread --, \
524 1) : 0)
92d41fe1
KS		 525
526/*
527 * Copy a line break character to a string buffer and advance pointers.
528 */
529
625fcfe9
KS		 530#define READ_LINE(parser,string) \
531 (STRING_EXTEND(parser,string) ? \
e35af832
KS		 532 (((CHECK_AT(parser->buffer,'\r',0) \
533 && CHECK_AT(parser->buffer,'\n',1)) ? /* CR LF -> LF */ \
92d41fe1 534 (*((string).pointer++) = (yaml_char_t) '\n', \
625fcfe9
KS		 535 parser->buffer.pointer += 2, \
536 parser->mark.index += 2, \
537 parser->mark.column = 0, \
538 parser->mark.line ++, \
92d41fe1 539 parser->unread -= 2) : \
e35af832
KS		 540 (CHECK_AT(parser->buffer,'\r',0) \
541 || CHECK_AT(parser->buffer,'\n',0)) ? /* CR|LF -> LF */ \
92d41fe1 542 (*((string).pointer++) = (yaml_char_t) '\n', \
625fcfe9
KS		 543 parser->buffer.pointer ++, \
544 parser->mark.index ++, \
545 parser->mark.column = 0, \
546 parser->mark.line ++, \
92d41fe1 547 parser->unread --) : \
e35af832
KS		 548 (CHECK_AT(parser->buffer,'\xC2',0) \
549 && CHECK_AT(parser->buffer,'\x85',1)) ? /* NEL -> LF */ \
92d41fe1 550 (*((string).pointer++) = (yaml_char_t) '\n', \
625fcfe9
KS		 551 parser->buffer.pointer += 2, \
552 parser->mark.index ++, \
553 parser->mark.column = 0, \
554 parser->mark.line ++, \
92d41fe1 555 parser->unread --) : \
e35af832
KS		 556 (CHECK_AT(parser->buffer,'\xE2',0) && \
557 CHECK_AT(parser->buffer,'\x80',1) && \
558 (CHECK_AT(parser->buffer,'\xA8',2) || \
559 CHECK_AT(parser->buffer,'\xA9',2))) ? /* LS|PS -> LS|PS */ \
625fcfe9
KS		 560 (*((string).pointer++) = *(parser->buffer.pointer++), \
561 *((string).pointer++) = *(parser->buffer.pointer++), \
562 *((string).pointer++) = *(parser->buffer.pointer++), \
563 parser->mark.index ++, \
564 parser->mark.column = 0, \
565 parser->mark.line ++, \
566 parser->unread --) : 0), \
567 1) : 0)
92d41fe1 568
625fcfe9
KS		 569/*
570 * Public API declarations.
571 */
e71095e3 572
625fcfe9
KS		 573YAML_DECLARE(int)
574yaml_parser_scan(yaml_parser_t *parser, yaml_token_t *token);
92d41fe1 575
625fcfe9
KS		 576/*
577 * Error handling.
578 */
92d41fe1 579
e71095e3 580static int
625fcfe9
KS		 581yaml_parser_set_scanner_error(yaml_parser_t *parser, const char *context,
582 yaml_mark_t context_mark, const char *problem);
e71095e3 583
03be97ab
KS		 584/*
585 * High-level token API.
586 */
587
625fcfe9 588YAML_DECLARE(int)
03be97ab
KS		 589yaml_parser_fetch_more_tokens(yaml_parser_t *parser);
590
591static int
592yaml_parser_fetch_next_token(yaml_parser_t *parser);
593
594/*
595 * Potential simple keys.
596 */
597
598static int
599yaml_parser_stale_simple_keys(yaml_parser_t *parser);
600
601static int
602yaml_parser_save_simple_key(yaml_parser_t *parser);
603
604static int
605yaml_parser_remove_simple_key(yaml_parser_t *parser);
606
eb9cceb5
KS		 607static int
608yaml_parser_increase_flow_level(yaml_parser_t *parser);
609
610static int
611yaml_parser_decrease_flow_level(yaml_parser_t *parser);
612
03be97ab
KS		 613/*
614 * Indentation treatment.
615 */
616
617static int
c201bf64
KS		 618yaml_parser_roll_indent(yaml_parser_t *parser, ptrdiff_t column,
619 ptrdiff_t number, yaml_token_type_t type, yaml_mark_t mark);
03be97ab
KS		 620
621static int
c201bf64 622yaml_parser_unroll_indent(yaml_parser_t *parser, ptrdiff_t column);
03be97ab
KS		 623
624/*
625 * Token fetchers.
626 */
627
628static int
629yaml_parser_fetch_stream_start(yaml_parser_t *parser);
630
631static int
632yaml_parser_fetch_stream_end(yaml_parser_t *parser);
633
634static int
635yaml_parser_fetch_directive(yaml_parser_t *parser);
636
03be97ab
KS		 637static int
638yaml_parser_fetch_document_indicator(yaml_parser_t *parser,
639 yaml_token_type_t type);
640
03be97ab
KS		 641static int
642yaml_parser_fetch_flow_collection_start(yaml_parser_t *parser,
643 yaml_token_type_t type);
644
03be97ab
KS		 645static int
646yaml_parser_fetch_flow_collection_end(yaml_parser_t *parser,
647 yaml_token_type_t type);
648
649static int
650yaml_parser_fetch_flow_entry(yaml_parser_t *parser);
651
652static int
653yaml_parser_fetch_block_entry(yaml_parser_t *parser);
654
655static int
656yaml_parser_fetch_key(yaml_parser_t *parser);
657
658static int
659yaml_parser_fetch_value(yaml_parser_t *parser);
660
661static int
eb9cceb5 662yaml_parser_fetch_anchor(yaml_parser_t *parser, yaml_token_type_t type);
03be97ab
KS		 663
664static int
665yaml_parser_fetch_tag(yaml_parser_t *parser);
666
03be97ab
KS		 667static int
668yaml_parser_fetch_block_scalar(yaml_parser_t *parser, int literal);
669
03be97ab
KS		 670static int
671yaml_parser_fetch_flow_scalar(yaml_parser_t *parser, int single);
672
673static int
674yaml_parser_fetch_plain_scalar(yaml_parser_t *parser);
675
676/*
677 * Token scanners.
678 */
679
680static int
681yaml_parser_scan_to_next_token(yaml_parser_t *parser);
682
625fcfe9
KS		 683static int
684yaml_parser_scan_directive(yaml_parser_t *parser, yaml_token_t *token);
03be97ab
KS		 685
686static int
687yaml_parser_scan_directive_name(yaml_parser_t *parser,
688 yaml_mark_t start_mark, yaml_char_t **name);
689
690static int
e71095e3 691yaml_parser_scan_version_directive_value(yaml_parser_t *parser,
03be97ab
KS		 692 yaml_mark_t start_mark, int *major, int *minor);
693
694static int
e71095e3 695yaml_parser_scan_version_directive_number(yaml_parser_t *parser,
03be97ab
KS		 696 yaml_mark_t start_mark, int *number);
697
698static int
699yaml_parser_scan_tag_directive_value(yaml_parser_t *parser,
e71095e3 700 yaml_mark_t mark, yaml_char_t **handle, yaml_char_t **prefix);
03be97ab 701
625fcfe9
KS		 702static int
703yaml_parser_scan_anchor(yaml_parser_t *parser, yaml_token_t *token,
03be97ab
KS		 704 yaml_token_type_t type);
705
625fcfe9
KS		 706static int
707yaml_parser_scan_tag(yaml_parser_t *parser, yaml_token_t *token);
03be97ab
KS		 708
709static int
710yaml_parser_scan_tag_handle(yaml_parser_t *parser, int directive,
711 yaml_mark_t start_mark, yaml_char_t **handle);
712
713static int
714yaml_parser_scan_tag_uri(yaml_parser_t *parser, int directive,
e71095e3
KS		 715 yaml_char_t *head, yaml_mark_t start_mark, yaml_char_t **uri);
716
717static int
718yaml_parser_scan_uri_escapes(yaml_parser_t *parser, int directive,
719 yaml_mark_t start_mark, yaml_string_t *string);
03be97ab 720
625fcfe9
KS		 721static int
722yaml_parser_scan_block_scalar(yaml_parser_t *parser, yaml_token_t *token,
723 int literal);
03be97ab 724
92d41fe1
KS		 725static int
726yaml_parser_scan_block_scalar_breaks(yaml_parser_t *parser,
727 int *indent, yaml_string_t *breaks,
728 yaml_mark_t start_mark, yaml_mark_t *end_mark);
729
625fcfe9
KS		 730static int
731yaml_parser_scan_flow_scalar(yaml_parser_t *parser, yaml_token_t *token,
732 int single);
03be97ab 733
625fcfe9
KS		 734static int
735yaml_parser_scan_plain_scalar(yaml_parser_t *parser, yaml_token_t *token);
03be97ab 736
f2b59d4d 737/*
625fcfe9 738 * Get the next token.
f2b59d4d
KS		 739 */
740
625fcfe9
KS		 741YAML_DECLARE(int)
742yaml_parser_scan(yaml_parser_t *parser, yaml_token_t *token)
f2b59d4d 743{
f2b59d4d 744 assert(parser); /* Non-NULL parser object is expected. */
625fcfe9 745 assert(token); /* Non-NULL token object is expected. */
f2b59d4d 746
5a00d8fe
KS		 747 /* Erase the token object. */
748
749 memset(token, 0, sizeof(yaml_token_t));
750
625fcfe9 751 /* No tokens after STREAM-END or error. */
f2b59d4d 752
625fcfe9 753 if (parser->stream_end_produced || parser->error) {
625fcfe9 754 return 1;
7e32c194
KS		 755 }
756
f2b59d4d
KS		 757 /* Ensure that the tokens queue contains enough tokens. */
758
625fcfe9
KS		 759 if (!parser->token_available) {
760 if (!yaml_parser_fetch_more_tokens(parser))
761 return 0;
92d41fe1
KS		 762 }
763
625fcfe9 764 /* Fetch the next token from the queue. */
986dbde7 765
625fcfe9
KS		 766 *token = DEQUEUE(parser, parser->tokens);
767 parser->token_available = 0;
768 parser->tokens_parsed ++;
e71095e3 769
625fcfe9
KS		 770 if (token->type == YAML_STREAM_END_TOKEN) {
771 parser->stream_end_produced = 1;
e71095e3
KS		 772 }
773
e71095e3
KS		 774 return 1;
775}
776
f2b59d4d
KS		 777/*
778 * Set the scanner error and return 0.
779 */
780
781static int
782yaml_parser_set_scanner_error(yaml_parser_t *parser, const char *context,
783 yaml_mark_t context_mark, const char *problem)
784{
785 parser->error = YAML_SCANNER_ERROR;
786 parser->context = context;
787 parser->context_mark = context_mark;
788 parser->problem = problem;
625fcfe9 789 parser->problem_mark = parser->mark;
7e32c194
KS		 790
791 return 0;
f2b59d4d
KS		 792}
793
f2b59d4d
KS		 794/*
795 * Ensure that the tokens queue contains at least one token which can be
796 * returned to the Parser.
797 */
798
625fcfe9 799YAML_DECLARE(int)
f2b59d4d
KS		 800yaml_parser_fetch_more_tokens(yaml_parser_t *parser)
801{
802 int need_more_tokens;
f2b59d4d
KS		 803
804 /* While we need more tokens to fetch, do it. */
805
806 while (1)
807 {
808 /*
809 * Check if we really need to fetch more tokens.
810 */
811
812 need_more_tokens = 0;
813
625fcfe9 814 if (parser->tokens.head == parser->tokens.tail)
f2b59d4d
KS		 815 {
816 /* Queue is empty. */
817
818 need_more_tokens = 1;
819 }
820 else
821 {
625fcfe9
KS		 822 yaml_simple_key_t *simple_key;
823
f2b59d4d
KS		 824 /* Check if any potential simple key may occupy the head position. */
825
7e32c194
KS		 826 if (!yaml_parser_stale_simple_keys(parser))
827 return 0;
828
625fcfe9
KS		 829 for (simple_key = parser->simple_keys.start;
830 simple_key != parser->simple_keys.top; simple_key++) {
831 if (simple_key->possible
832 && simple_key->token_number == parser->tokens_parsed) {
f2b59d4d
KS		 833 need_more_tokens = 1;
834 break;
835 }
836 }
837 }
838
839 /* We are finished. */
840
841 if (!need_more_tokens)
842 break;
843
844 /* Fetch the next token. */
845
846 if (!yaml_parser_fetch_next_token(parser))
847 return 0;
848 }
849
625fcfe9
KS		 850 parser->token_available = 1;
851
f2b59d4d
KS		 852 return 1;
853}
854
855/*
856 * The dispatcher for token fetchers.
857 */
858
859static int
860yaml_parser_fetch_next_token(yaml_parser_t *parser)
861{
862 /* Ensure that the buffer is initialized. */
863
625fcfe9 864 if (!CACHE(parser, 1))
f2b59d4d
KS		 865 return 0;
866
867 /* Check if we just started scanning. Fetch STREAM-START then. */
868
869 if (!parser->stream_start_produced)
870 return yaml_parser_fetch_stream_start(parser);
871
872 /* Eat whitespaces and comments until we reach the next token. */
873
874 if (!yaml_parser_scan_to_next_token(parser))
875 return 0;
876
7e32c194
KS		 877 /* Remove obsolete potential simple keys. */
878
879 if (!yaml_parser_stale_simple_keys(parser))
880 return 0;
881
f2b59d4d
KS		 882 /* Check the indentation level against the current column. */
883
625fcfe9 884 if (!yaml_parser_unroll_indent(parser, parser->mark.column))
f2b59d4d
KS		 885 return 0;
886
887 /*
888 * Ensure that the buffer contains at least 4 characters. 4 is the length
889 * of the longest indicators ('--- ' and '... ').
890 */
891
625fcfe9 892 if (!CACHE(parser, 4))
f2b59d4d
KS		 893 return 0;
894
895 /* Is it the end of the stream? */
896
e35af832 897 if (IS_Z(parser->buffer))
f2b59d4d
KS		 898 return yaml_parser_fetch_stream_end(parser);
899
900 /* Is it a directive? */
901
e35af832 902 if (parser->mark.column == 0 && CHECK(parser->buffer, '%'))
f2b59d4d
KS		 903 return yaml_parser_fetch_directive(parser);
904
905 /* Is it the document start indicator? */
906
625fcfe9 907 if (parser->mark.column == 0
e35af832
KS		 908 && CHECK_AT(parser->buffer, '-', 0)
909 && CHECK_AT(parser->buffer, '-', 1)
910 && CHECK_AT(parser->buffer, '-', 2)
911 && IS_BLANKZ_AT(parser->buffer, 3))
eb9cceb5
KS		 912 return yaml_parser_fetch_document_indicator(parser,
913 YAML_DOCUMENT_START_TOKEN);
f2b59d4d
KS		 914
915 /* Is it the document end indicator? */
916
625fcfe9 917 if (parser->mark.column == 0
e35af832
KS		 918 && CHECK_AT(parser->buffer, '.', 0)
919 && CHECK_AT(parser->buffer, '.', 1)
920 && CHECK_AT(parser->buffer, '.', 2)
921 && IS_BLANKZ_AT(parser->buffer, 3))
eb9cceb5
KS		 922 return yaml_parser_fetch_document_indicator(parser,
923 YAML_DOCUMENT_END_TOKEN);
f2b59d4d
KS		 924
925 /* Is it the flow sequence start indicator? */
926
e35af832 927 if (CHECK(parser->buffer, '['))
eb9cceb5
KS		 928 return yaml_parser_fetch_flow_collection_start(parser,
929 YAML_FLOW_SEQUENCE_START_TOKEN);
f2b59d4d
KS		 930
931 /* Is it the flow mapping start indicator? */
932
e35af832 933 if (CHECK(parser->buffer, '{'))
eb9cceb5
KS		 934 return yaml_parser_fetch_flow_collection_start(parser,
935 YAML_FLOW_MAPPING_START_TOKEN);
f2b59d4d
KS		 936
937 /* Is it the flow sequence end indicator? */
938
e35af832 939 if (CHECK(parser->buffer, ']'))
eb9cceb5
KS		 940 return yaml_parser_fetch_flow_collection_end(parser,
941 YAML_FLOW_SEQUENCE_END_TOKEN);
f2b59d4d
KS		 942
943 /* Is it the flow mapping end indicator? */
944
e35af832 945 if (CHECK(parser->buffer, '}'))
eb9cceb5
KS		 946 return yaml_parser_fetch_flow_collection_end(parser,
947 YAML_FLOW_MAPPING_END_TOKEN);
f2b59d4d
KS		 948
949 /* Is it the flow entry indicator? */
950
e35af832 951 if (CHECK(parser->buffer, ','))
f2b59d4d
KS		 952 return yaml_parser_fetch_flow_entry(parser);
953
954 /* Is it the block entry indicator? */
955
e35af832 956 if (CHECK(parser->buffer, '-') && IS_BLANKZ_AT(parser->buffer, 1))
f2b59d4d
KS		 957 return yaml_parser_fetch_block_entry(parser);
958
959 /* Is it the key indicator? */
960
e35af832
KS		 961 if (CHECK(parser->buffer, '?')
962 && (parser->flow_level || IS_BLANKZ_AT(parser->buffer, 1)))
f2b59d4d
KS		 963 return yaml_parser_fetch_key(parser);
964
965 /* Is it the value indicator? */
966
e35af832
KS		 967 if (CHECK(parser->buffer, ':')
968 && (parser->flow_level || IS_BLANKZ_AT(parser->buffer, 1)))
f2b59d4d
KS		 969 return yaml_parser_fetch_value(parser);
970
971 /* Is it an alias? */
972
e35af832 973 if (CHECK(parser->buffer, '*'))
eb9cceb5 974 return yaml_parser_fetch_anchor(parser, YAML_ALIAS_TOKEN);
f2b59d4d
KS		 975
976 /* Is it an anchor? */
977
e35af832 978 if (CHECK(parser->buffer, '&'))
eb9cceb5 979 return yaml_parser_fetch_anchor(parser, YAML_ANCHOR_TOKEN);
f2b59d4d
KS		 980
981 /* Is it a tag? */
982
e35af832 983 if (CHECK(parser->buffer, '!'))
f2b59d4d
KS		 984 return yaml_parser_fetch_tag(parser);
985
986 /* Is it a literal scalar? */
987
e35af832 988 if (CHECK(parser->buffer, '|') && !parser->flow_level)
f2b59d4d
KS		 989 return yaml_parser_fetch_block_scalar(parser, 1);
990
991 /* Is it a folded scalar? */
992
e35af832 993 if (CHECK(parser->buffer, '>') && !parser->flow_level)
f2b59d4d
KS		 994 return yaml_parser_fetch_block_scalar(parser, 0);
995
996 /* Is it a single-quoted scalar? */
997
e35af832 998 if (CHECK(parser->buffer, '\''))
f2b59d4d
KS		 999 return yaml_parser_fetch_flow_scalar(parser, 1);
1000
1001 /* Is it a double-quoted scalar? */
1002
e35af832 1003 if (CHECK(parser->buffer, '"'))
f2b59d4d
KS		 1004 return yaml_parser_fetch_flow_scalar(parser, 0);
1005
1006 /*
1007 * Is it a plain scalar?
1008 *
1009 * A plain scalar may start with any non-blank characters except
1010 *
1011 * '-', '?', ':', ',', '[', ']', '{', '}',
1012 * '#', '&', '*', '!', '|', '>', '\'', '\"',
1013 * '%', '@', '`'.
1014 *
7e32c194
KS		 1015 * In the block context (and, for the '-' indicator, in the flow context
1016 * too), it may also start with the characters
f2b59d4d
KS		 1017 *
1018 * '-', '?', ':'
1019 *
1020 * if it is followed by a non-space character.
1021 *
1022 * The last rule is more restrictive than the specification requires.
1023 */
1024
e35af832
KS		 1025 if (!(IS_BLANKZ(parser->buffer) || CHECK(parser->buffer, '-')
1026 || CHECK(parser->buffer, '?') || CHECK(parser->buffer, ':')
1027 || CHECK(parser->buffer, ',') || CHECK(parser->buffer, '[')
1028 || CHECK(parser->buffer, ']') || CHECK(parser->buffer, '{')
1029 || CHECK(parser->buffer, '}') || CHECK(parser->buffer, '#')
1030 || CHECK(parser->buffer, '&') || CHECK(parser->buffer, '*')
1031 || CHECK(parser->buffer, '!') || CHECK(parser->buffer, '|')
1032 || CHECK(parser->buffer, '>') || CHECK(parser->buffer, '\'')
1033 || CHECK(parser->buffer, '"') || CHECK(parser->buffer, '%')
1034 || CHECK(parser->buffer, '@') || CHECK(parser->buffer, '`')) ||
1035 (CHECK(parser->buffer, '-') && !IS_BLANK_AT(parser->buffer, 1)) ||
f2b59d4d 1036 (!parser->flow_level &&
e35af832
KS		 1037 (CHECK(parser->buffer, '?') || CHECK(parser->buffer, ':'))
1038 && !IS_BLANKZ_AT(parser->buffer, 1)))
f2b59d4d
KS		 1039 return yaml_parser_fetch_plain_scalar(parser);
1040
1041 /*
1042 * If we don't determine the token type so far, it is an error.
1043 */
1044
625fcfe9
KS		 1045 return yaml_parser_set_scanner_error(parser,
1046 "while scanning for the next token", parser->mark,
1047 "found character that cannot start any token");
f2b59d4d
KS		 1048}
1049
eb9cceb5
KS		 1050/*
1051 * Check the list of potential simple keys and remove the positions that
1052 * cannot contain simple keys anymore.
1053 */
1054
1055static int
1056yaml_parser_stale_simple_keys(yaml_parser_t *parser)
1057{
625fcfe9 1058 yaml_simple_key_t *simple_key;
eb9cceb5
KS		 1059
1060 /* Check for a potential simple key for each flow level. */
1061
625fcfe9
KS		 1062 for (simple_key = parser->simple_keys.start;
1063 simple_key != parser->simple_keys.top; simple_key ++)
eb9cceb5 1064 {
eb9cceb5
KS		 1065 /*
1066 * The specification requires that a simple key
1067 *
1068 * - is limited to a single line,
1069 * - is shorter than 1024 characters.
1070 */
1071
625fcfe9
KS		 1072 if (simple_key->possible
1073 && (simple_key->mark.line < parser->mark.line
1074 || simple_key->mark.index+1024 < parser->mark.index)) {
eb9cceb5
KS		 1075
1076 /* Check if the potential simple key to be removed is required. */
1077
1078 if (simple_key->required) {
1079 return yaml_parser_set_scanner_error(parser,
1080 "while scanning a simple key", simple_key->mark,
6be8109b 1081 "could not find expected ':'");
eb9cceb5
KS		 1082 }
1083
625fcfe9 1084 simple_key->possible = 0;
eb9cceb5
KS		 1085 }
1086 }
1087
1088 return 1;
1089}
1090
1091/*
1092 * Check if a simple key may start at the current position and add it if
1093 * needed.
1094 */
1095
1096static int
1097yaml_parser_save_simple_key(yaml_parser_t *parser)
1098{
1099 /*
1100 * A simple key is required at the current position if the scanner is in
1101 * the block context and the current column coincides with the indentation
1102 * level.
1103 */
1104
625fcfe9 1105 int required = (!parser->flow_level
c201bf64 1106 && parser->indent == (ptrdiff_t)parser->mark.column);
eb9cceb5 1107
eb9cceb5
KS		 1108 /*
1109 * If the current position may start a simple key, save it.
1110 */
1111
1112 if (parser->simple_key_allowed)
1113 {
252c575a
KS		 1114 yaml_simple_key_t simple_key;
1115 simple_key.possible = 1;
1116 simple_key.required = required;
986dbde7 1117 simple_key.token_number =
3b160b60 1118 parser->tokens_parsed + (parser->tokens.tail - parser->tokens.head);
0174ed6e 1119 simple_key.mark = parser->mark;
eb9cceb5
KS		 1120
1121 if (!yaml_parser_remove_simple_key(parser)) return 0;
1122
625fcfe9 1123 *(parser->simple_keys.top-1) = simple_key;
eb9cceb5
KS		 1124 }
1125
1126 return 1;
1127}
1128
1129/*
1130 * Remove a potential simple key at the current flow level.
1131 */
1132
1133static int
1134yaml_parser_remove_simple_key(yaml_parser_t *parser)
1135{
625fcfe9 1136 yaml_simple_key_t *simple_key = parser->simple_keys.top-1;
eb9cceb5 1137
625fcfe9 1138 if (simple_key->possible)
eb9cceb5
KS		 1139 {
1140 /* If the key is required, it is an error. */
1141
1142 if (simple_key->required) {
1143 return yaml_parser_set_scanner_error(parser,
1144 "while scanning a simple key", simple_key->mark,
6be8109b 1145 "could not find expected ':'");
eb9cceb5 1146 }
625fcfe9 1147 }
eb9cceb5 1148
625fcfe9 1149 /* Remove the key from the stack. */
eb9cceb5 1150
625fcfe9 1151 simple_key->possible = 0;
eb9cceb5
KS		 1152
1153 return 1;
1154}
1155
1156/*
1157 * Increase the flow level and resize the simple key list if needed.
1158 */
1159
1160static int
1161yaml_parser_increase_flow_level(yaml_parser_t *parser)
1162{
625fcfe9 1163 yaml_simple_key_t empty_simple_key = { 0, 0, 0, { 0, 0, 0 } };
eb9cceb5 1164
625fcfe9 1165 /* Reset the simple key on the next level. */
eb9cceb5 1166
625fcfe9
KS		 1167 if (!PUSH(parser, parser->simple_keys, empty_simple_key))
1168 return 0;
1169
1170 /* Increase the flow level. */
eb9cceb5 1171
1ef11717
KS		 1172 if (parser->flow_level == INT_MAX) {
1173 parser->error = YAML_MEMORY_ERROR;
c201bf64 1174 return 0;
1ef11717 1175 }
c201bf64 1176
625fcfe9 1177 parser->flow_level++;
eb9cceb5
KS		 1178
1179 return 1;
1180}
1181
1182/*
1183 * Decrease the flow level.
1184 */
1185
1186static int
1187yaml_parser_decrease_flow_level(yaml_parser_t *parser)
1188{
625fcfe9
KS		 1189 if (parser->flow_level) {
1190 parser->flow_level --;
fc2dd942 1191 (void)POP(parser, parser->simple_keys);
eb9cceb5
KS		 1192 }
1193
eb9cceb5
KS		 1194 return 1;
1195}
1196
1197/*
1198 * Push the current indentation level to the stack and set the new level
1199 * the current column is greater than the indentation level. In this case,
1200 * append or insert the specified token into the token queue.
986dbde7 1201 *
eb9cceb5
KS		 1202 */
1203
1204static int
c201bf64
KS		 1205yaml_parser_roll_indent(yaml_parser_t *parser, ptrdiff_t column,
1206 ptrdiff_t number, yaml_token_type_t type, yaml_mark_t mark)
eb9cceb5 1207{
625fcfe9 1208 yaml_token_t token;
eb9cceb5
KS		 1209
1210 /* In the flow context, do nothing. */
1211
1212 if (parser->flow_level)
1213 return 1;
1214
1215 if (parser->indent < column)
1216 {
eb9cceb5
KS		 1217 /*
1218 * Push the current indentation level to the stack and set the new
1219 * indentation level.
1220 */
1221
625fcfe9
KS		 1222 if (!PUSH(parser, parser->indents, parser->indent))
1223 return 0;
1224
1ef11717
KS		 1225 if (column > INT_MAX) {
1226 parser->error = YAML_MEMORY_ERROR;
c201bf64 1227 return 0;
1ef11717 1228 }
c201bf64 1229
eb9cceb5
KS		 1230 parser->indent = column;
1231
625fcfe9 1232 /* Create a token and insert it into the queue. */
eb9cceb5 1233
625fcfe9 1234 TOKEN_INIT(token, type, mark, mark);
eb9cceb5 1235
625fcfe9
KS		 1236 if (number == -1) {
1237 if (!ENQUEUE(parser, parser->tokens, token))
1238 return 0;
1239 }
1240 else {
1241 if (!QUEUE_INSERT(parser,
1242 parser->tokens, number - parser->tokens_parsed, token))
1243 return 0;
eb9cceb5
KS		 1244 }
1245 }
1246
1247 return 1;
1248}
1249
1250/*
1251 * Pop indentation levels from the indents stack until the current level
bdf4d192 1252 * becomes less or equal to the column. For each indentation level, append
eb9cceb5
KS		 1253 * the BLOCK-END token.
1254 */
1255
1256
1257static int
c201bf64 1258yaml_parser_unroll_indent(yaml_parser_t *parser, ptrdiff_t column)
eb9cceb5 1259{
625fcfe9 1260 yaml_token_t token;
eb9cceb5
KS		 1261
1262 /* In the flow context, do nothing. */
1263
1264 if (parser->flow_level)
1265 return 1;
1266
bdf4d192 1267 /* Loop through the indentation levels in the stack. */
eb9cceb5
KS		 1268
1269 while (parser->indent > column)
1270 {
625fcfe9 1271 /* Create a token and append it to the queue. */
eb9cceb5 1272
625fcfe9 1273 TOKEN_INIT(token, YAML_BLOCK_END_TOKEN, parser->mark, parser->mark);
eb9cceb5 1274
625fcfe9 1275 if (!ENQUEUE(parser, parser->tokens, token))
eb9cceb5 1276 return 0;
eb9cceb5
KS		 1277
1278 /* Pop the indentation level. */
1279
625fcfe9 1280 parser->indent = POP(parser, parser->indents);
eb9cceb5
KS		 1281 }
1282
1283 return 1;
1284}
1285
1286/*
1287 * Initialize the scanner and produce the STREAM-START token.
1288 */
1289
1290static int
1291yaml_parser_fetch_stream_start(yaml_parser_t *parser)
1292{
625fcfe9
KS		 1293 yaml_simple_key_t simple_key = { 0, 0, 0, { 0, 0, 0 } };
1294 yaml_token_t token;
eb9cceb5
KS		 1295
1296 /* Set the initial indentation. */
1297
1298 parser->indent = -1;
1299
625fcfe9
KS		 1300 /* Initialize the simple key stack. */
1301
1302 if (!PUSH(parser, parser->simple_keys, simple_key))
1303 return 0;
1304
eb9cceb5
KS		 1305 /* A simple key is allowed at the beginning of the stream. */
1306
1307 parser->simple_key_allowed = 1;
1308
1309 /* We have started. */
1310
1311 parser->stream_start_produced = 1;
1312
625fcfe9 1313 /* Create the STREAM-START token and append it to the queue. */
eb9cceb5 1314
625fcfe9
KS		 1315 STREAM_START_TOKEN_INIT(token, parser->encoding,
1316 parser->mark, parser->mark);
eb9cceb5 1317
625fcfe9 1318 if (!ENQUEUE(parser, parser->tokens, token))
eb9cceb5 1319 return 0;
eb9cceb5
KS		 1320
1321 return 1;
1322}
1323
1324/*
1325 * Produce the STREAM-END token and shut down the scanner.
1326 */
1327
1328static int
1329yaml_parser_fetch_stream_end(yaml_parser_t *parser)
1330{
625fcfe9 1331 yaml_token_t token;
eb9cceb5 1332
c83b67a6
KS		 1333 /* Force new line. */
1334
1335 if (parser->mark.column != 0) {
1336 parser->mark.column = 0;
1337 parser->mark.line ++;
1338 }
1339
eb9cceb5
KS		 1340 /* Reset the indentation level. */
1341
1342 if (!yaml_parser_unroll_indent(parser, -1))
1343 return 0;
1344
7e32c194 1345 /* Reset simple keys. */
eb9cceb5 1346
7e32c194
KS		 1347 if (!yaml_parser_remove_simple_key(parser))
1348 return 0;
1349
1350 parser->simple_key_allowed = 0;
eb9cceb5 1351
625fcfe9 1352 /* Create the STREAM-END token and append it to the queue. */
eb9cceb5 1353
625fcfe9 1354 STREAM_END_TOKEN_INIT(token, parser->mark, parser->mark);
eb9cceb5 1355
625fcfe9 1356 if (!ENQUEUE(parser, parser->tokens, token))
eb9cceb5 1357 return 0;
eb9cceb5
KS		 1358
1359 return 1;
1360}
1361
1362/*
625fcfe9 1363 * Produce a VERSION-DIRECTIVE or TAG-DIRECTIVE token.
eb9cceb5
KS		 1364 */
1365
1366static int
1367yaml_parser_fetch_directive(yaml_parser_t *parser)
1368{
625fcfe9 1369 yaml_token_t token;
eb9cceb5
KS		 1370
1371 /* Reset the indentation level. */
1372
1373 if (!yaml_parser_unroll_indent(parser, -1))
1374 return 0;
1375
1376 /* Reset simple keys. */
1377
1378 if (!yaml_parser_remove_simple_key(parser))
1379 return 0;
1380
1381 parser->simple_key_allowed = 0;
1382
1383 /* Create the YAML-DIRECTIVE or TAG-DIRECTIVE token. */
1384
625fcfe9
KS		 1385 if (!yaml_parser_scan_directive(parser, &token))
1386 return 0;
eb9cceb5
KS		 1387
1388 /* Append the token to the queue. */
1389
625fcfe9
KS		 1390 if (!ENQUEUE(parser, parser->tokens, token)) {
1391 yaml_token_delete(&token);
eb9cceb5
KS		 1392 return 0;
1393 }
1394
1395 return 1;
1396}
1397
1398/*
1399 * Produce the DOCUMENT-START or DOCUMENT-END token.
1400 */
1401
1402static int
1403yaml_parser_fetch_document_indicator(yaml_parser_t *parser,
1404 yaml_token_type_t type)
1405{
1406 yaml_mark_t start_mark, end_mark;
625fcfe9 1407 yaml_token_t token;
eb9cceb5
KS		 1408
1409 /* Reset the indentation level. */
1410
1411 if (!yaml_parser_unroll_indent(parser, -1))
1412 return 0;
1413
1414 /* Reset simple keys. */
1415
1416 if (!yaml_parser_remove_simple_key(parser))
1417 return 0;
1418
1419 parser->simple_key_allowed = 0;
1420
1421 /* Consume the token. */
1422
625fcfe9 1423 start_mark = parser->mark;
eb9cceb5 1424
625fcfe9
KS		 1425 SKIP(parser);
1426 SKIP(parser);
1427 SKIP(parser);
eb9cceb5 1428
625fcfe9 1429 end_mark = parser->mark;
eb9cceb5
KS		 1430
1431 /* Create the DOCUMENT-START or DOCUMENT-END token. */
1432
625fcfe9 1433 TOKEN_INIT(token, type, start_mark, end_mark);
eb9cceb5
KS		 1434
1435 /* Append the token to the queue. */
1436
625fcfe9 1437 if (!ENQUEUE(parser, parser->tokens, token))
eb9cceb5 1438 return 0;
eb9cceb5
KS		 1439
1440 return 1;
1441}
1442
1443/*
1444 * Produce the FLOW-SEQUENCE-START or FLOW-MAPPING-START token.
1445 */
1446
1447static int
1448yaml_parser_fetch_flow_collection_start(yaml_parser_t *parser,
1449 yaml_token_type_t type)
1450{
1451 yaml_mark_t start_mark, end_mark;
625fcfe9 1452 yaml_token_t token;
eb9cceb5
KS		 1453
1454 /* The indicators '[' and '{' may start a simple key. */
1455
1456 if (!yaml_parser_save_simple_key(parser))
1457 return 0;
1458
1459 /* Increase the flow level. */
1460
1461 if (!yaml_parser_increase_flow_level(parser))
1462 return 0;
1463
1464 /* A simple key may follow the indicators '[' and '{'. */
1465
1466 parser->simple_key_allowed = 1;
1467
1468 /* Consume the token. */
1469
625fcfe9
KS		 1470 start_mark = parser->mark;
1471 SKIP(parser);
1472 end_mark = parser->mark;
eb9cceb5
KS		 1473
1474 /* Create the FLOW-SEQUENCE-START of FLOW-MAPPING-START token. */
1475
625fcfe9 1476 TOKEN_INIT(token, type, start_mark, end_mark);
eb9cceb5
KS		 1477
1478 /* Append the token to the queue. */
1479
625fcfe9 1480 if (!ENQUEUE(parser, parser->tokens, token))
eb9cceb5 1481 return 0;
eb9cceb5
KS		 1482
1483 return 1;
1484}
1485
1486/*
1487 * Produce the FLOW-SEQUENCE-END or FLOW-MAPPING-END token.
1488 */
1489
1490static int
1491yaml_parser_fetch_flow_collection_end(yaml_parser_t *parser,
1492 yaml_token_type_t type)
1493{
1494 yaml_mark_t start_mark, end_mark;
625fcfe9 1495 yaml_token_t token;
eb9cceb5
KS		 1496
1497 /* Reset any potential simple key on the current flow level. */
1498
1499 if (!yaml_parser_remove_simple_key(parser))
1500 return 0;
1501
1502 /* Decrease the flow level. */
1503
1504 if (!yaml_parser_decrease_flow_level(parser))
1505 return 0;
1506
1507 /* No simple keys after the indicators ']' and '}'. */
1508
1509 parser->simple_key_allowed = 0;
1510
1511 /* Consume the token. */
1512
625fcfe9
KS		 1513 start_mark = parser->mark;
1514 SKIP(parser);
1515 end_mark = parser->mark;
eb9cceb5
KS		 1516
1517 /* Create the FLOW-SEQUENCE-END of FLOW-MAPPING-END token. */
1518
625fcfe9 1519 TOKEN_INIT(token, type, start_mark, end_mark);
eb9cceb5
KS		 1520
1521 /* Append the token to the queue. */
1522
625fcfe9 1523 if (!ENQUEUE(parser, parser->tokens, token))
eb9cceb5 1524 return 0;
eb9cceb5
KS		 1525
1526 return 1;
1527}
1528
1529/*
1530 * Produce the FLOW-ENTRY token.
1531 */
1532
1533static int
1534yaml_parser_fetch_flow_entry(yaml_parser_t *parser)
1535{
1536 yaml_mark_t start_mark, end_mark;
625fcfe9 1537 yaml_token_t token;
eb9cceb5
KS		 1538
1539 /* Reset any potential simple keys on the current flow level. */
1540
1541 if (!yaml_parser_remove_simple_key(parser))
1542 return 0;
1543
1544 /* Simple keys are allowed after ','. */
1545
1546 parser->simple_key_allowed = 1;
1547
1548 /* Consume the token. */
1549
625fcfe9
KS		 1550 start_mark = parser->mark;
1551 SKIP(parser);
1552 end_mark = parser->mark;
eb9cceb5 1553
625fcfe9 1554 /* Create the FLOW-ENTRY token and append it to the queue. */
eb9cceb5 1555
625fcfe9 1556 TOKEN_INIT(token, YAML_FLOW_ENTRY_TOKEN, start_mark, end_mark);
eb9cceb5 1557
625fcfe9 1558 if (!ENQUEUE(parser, parser->tokens, token))
eb9cceb5 1559 return 0;
eb9cceb5
KS		 1560
1561 return 1;
1562}
1563
1564/*
1565 * Produce the BLOCK-ENTRY token.
1566 */
1567
1568static int
1569yaml_parser_fetch_block_entry(yaml_parser_t *parser)
1570{
1571 yaml_mark_t start_mark, end_mark;
625fcfe9 1572 yaml_token_t token;
eb9cceb5
KS		 1573
1574 /* Check if the scanner is in the block context. */
1575
1576 if (!parser->flow_level)
1577 {
1578 /* Check if we are allowed to start a new entry. */
1579
1580 if (!parser->simple_key_allowed) {
625fcfe9 1581 return yaml_parser_set_scanner_error(parser, NULL, parser->mark,
eb9cceb5
KS		 1582 "block sequence entries are not allowed in this context");
1583 }
1584
1585 /* Add the BLOCK-SEQUENCE-START token if needed. */
1586
625fcfe9
KS		 1587 if (!yaml_parser_roll_indent(parser, parser->mark.column, -1,
1588 YAML_BLOCK_SEQUENCE_START_TOKEN, parser->mark))
eb9cceb5
KS		 1589 return 0;
1590 }
1591 else
1592 {
1593 /*
1594 * It is an error for the '-' indicator to occur in the flow context,
1595 * but we let the Parser detect and report about it because the Parser
1596 * is able to point to the context.
1597 */
1598 }
1599
1600 /* Reset any potential simple keys on the current flow level. */
1601
1602 if (!yaml_parser_remove_simple_key(parser))
1603 return 0;
1604
1605 /* Simple keys are allowed after '-'. */
1606
1607 parser->simple_key_allowed = 1;
1608
1609 /* Consume the token. */
1610
625fcfe9
KS		 1611 start_mark = parser->mark;
1612 SKIP(parser);
1613 end_mark = parser->mark;
eb9cceb5 1614
625fcfe9 1615 /* Create the BLOCK-ENTRY token and append it to the queue. */
eb9cceb5 1616
625fcfe9 1617 TOKEN_INIT(token, YAML_BLOCK_ENTRY_TOKEN, start_mark, end_mark);
eb9cceb5 1618
625fcfe9 1619 if (!ENQUEUE(parser, parser->tokens, token))
eb9cceb5 1620 return 0;
eb9cceb5
KS		 1621
1622 return 1;
1623}
1624
1625/*
1626 * Produce the KEY token.
1627 */
1628
1629static int
1630yaml_parser_fetch_key(yaml_parser_t *parser)
1631{
1632 yaml_mark_t start_mark, end_mark;
625fcfe9 1633 yaml_token_t token;
eb9cceb5
KS		 1634
1635 /* In the block context, additional checks are required. */
1636
1637 if (!parser->flow_level)
1638 {
46574312 1639 /* Check if we are allowed to start a new key (not necessary simple). */
eb9cceb5
KS		 1640
1641 if (!parser->simple_key_allowed) {
625fcfe9 1642 return yaml_parser_set_scanner_error(parser, NULL, parser->mark,
eb9cceb5
KS		 1643 "mapping keys are not allowed in this context");
1644 }
1645
1646 /* Add the BLOCK-MAPPING-START token if needed. */
1647
625fcfe9
KS		 1648 if (!yaml_parser_roll_indent(parser, parser->mark.column, -1,
1649 YAML_BLOCK_MAPPING_START_TOKEN, parser->mark))
eb9cceb5
KS		 1650 return 0;
1651 }
1652
1653 /* Reset any potential simple keys on the current flow level. */
1654
1655 if (!yaml_parser_remove_simple_key(parser))
1656 return 0;
1657
1658 /* Simple keys are allowed after '?' in the block context. */
1659
1660 parser->simple_key_allowed = (!parser->flow_level);
1661
1662 /* Consume the token. */
1663
625fcfe9
KS		 1664 start_mark = parser->mark;
1665 SKIP(parser);
1666 end_mark = parser->mark;
eb9cceb5 1667
625fcfe9 1668 /* Create the KEY token and append it to the queue. */
eb9cceb5 1669
625fcfe9 1670 TOKEN_INIT(token, YAML_KEY_TOKEN, start_mark, end_mark);
eb9cceb5 1671
625fcfe9 1672 if (!ENQUEUE(parser, parser->tokens, token))
eb9cceb5 1673 return 0;
eb9cceb5
KS		 1674
1675 return 1;
1676}
1677
1678/*
1679 * Produce the VALUE token.
1680 */
1681
1682static int
1683yaml_parser_fetch_value(yaml_parser_t *parser)
1684{
1685 yaml_mark_t start_mark, end_mark;
625fcfe9
KS		 1686 yaml_token_t token;
1687 yaml_simple_key_t *simple_key = parser->simple_keys.top-1;
eb9cceb5
KS		 1688
1689 /* Have we found a simple key? */
1690
625fcfe9 1691 if (simple_key->possible)
eb9cceb5 1692 {
eb9cceb5 1693
625fcfe9 1694 /* Create the KEY token and insert it into the queue. */
eb9cceb5 1695
625fcfe9 1696 TOKEN_INIT(token, YAML_KEY_TOKEN, simple_key->mark, simple_key->mark);
eb9cceb5 1697
625fcfe9
KS		 1698 if (!QUEUE_INSERT(parser, parser->tokens,
1699 simple_key->token_number - parser->tokens_parsed, token))
eb9cceb5 1700 return 0;
eb9cceb5
KS		 1701
1702 /* In the block context, we may need to add the BLOCK-MAPPING-START token. */
1703
625fcfe9 1704 if (!yaml_parser_roll_indent(parser, simple_key->mark.column,
eb9cceb5
KS		 1705 simple_key->token_number,
1706 YAML_BLOCK_MAPPING_START_TOKEN, simple_key->mark))
1707 return 0;
1708
625fcfe9 1709 /* Remove the simple key. */
eb9cceb5 1710
625fcfe9 1711 simple_key->possible = 0;
eb9cceb5
KS		 1712
1713 /* A simple key cannot follow another simple key. */
1714
1715 parser->simple_key_allowed = 0;
1716 }
1717 else
1718 {
1719 /* The ':' indicator follows a complex key. */
1720
1721 /* In the block context, extra checks are required. */
1722
1723 if (!parser->flow_level)
1724 {
1725 /* Check if we are allowed to start a complex value. */
1726
1727 if (!parser->simple_key_allowed) {
625fcfe9 1728 return yaml_parser_set_scanner_error(parser, NULL, parser->mark,
eb9cceb5
KS		 1729 "mapping values are not allowed in this context");
1730 }
1731
1732 /* Add the BLOCK-MAPPING-START token if needed. */
1733
625fcfe9
KS		 1734 if (!yaml_parser_roll_indent(parser, parser->mark.column, -1,
1735 YAML_BLOCK_MAPPING_START_TOKEN, parser->mark))
eb9cceb5
KS		 1736 return 0;
1737 }
1738
eb9cceb5
KS		 1739 /* Simple keys after ':' are allowed in the block context. */
1740
1741 parser->simple_key_allowed = (!parser->flow_level);
1742 }
1743
1744 /* Consume the token. */
1745
625fcfe9
KS		 1746 start_mark = parser->mark;
1747 SKIP(parser);
1748 end_mark = parser->mark;
eb9cceb5 1749
625fcfe9 1750 /* Create the VALUE token and append it to the queue. */
eb9cceb5 1751
625fcfe9 1752 TOKEN_INIT(token, YAML_VALUE_TOKEN, start_mark, end_mark);
eb9cceb5 1753
625fcfe9 1754 if (!ENQUEUE(parser, parser->tokens, token))
eb9cceb5 1755 return 0;
eb9cceb5
KS		 1756
1757 return 1;
1758}
1759
1760/*
1761 * Produce the ALIAS or ANCHOR token.
1762 */
1763
1764static int
1765yaml_parser_fetch_anchor(yaml_parser_t *parser, yaml_token_type_t type)
1766{
625fcfe9 1767 yaml_token_t token;
eb9cceb5
KS		 1768
1769 /* An anchor or an alias could be a simple key. */
1770
1771 if (!yaml_parser_save_simple_key(parser))
1772 return 0;
1773
1774 /* A simple key cannot follow an anchor or an alias. */
1775
1776 parser->simple_key_allowed = 0;
1777
625fcfe9 1778 /* Create the ALIAS or ANCHOR token and append it to the queue. */
eb9cceb5 1779
625fcfe9
KS		 1780 if (!yaml_parser_scan_anchor(parser, &token, type))
1781 return 0;
eb9cceb5 1782
625fcfe9
KS		 1783 if (!ENQUEUE(parser, parser->tokens, token)) {
1784 yaml_token_delete(&token);
eb9cceb5
KS		 1785 return 0;
1786 }
eb9cceb5
KS		 1787 return 1;
1788}
1789
1790/*
1791 * Produce the TAG token.
1792 */
1793
1794static int
1795yaml_parser_fetch_tag(yaml_parser_t *parser)
1796{
625fcfe9 1797 yaml_token_t token;
eb9cceb5
KS		 1798
1799 /* A tag could be a simple key. */
1800
1801 if (!yaml_parser_save_simple_key(parser))
1802 return 0;
1803
1804 /* A simple key cannot follow a tag. */
1805
1806 parser->simple_key_allowed = 0;
1807
625fcfe9 1808 /* Create the TAG token and append it to the queue. */
eb9cceb5 1809
625fcfe9
KS		 1810 if (!yaml_parser_scan_tag(parser, &token))
1811 return 0;
eb9cceb5 1812
625fcfe9
KS		 1813 if (!ENQUEUE(parser, parser->tokens, token)) {
1814 yaml_token_delete(&token);
eb9cceb5
KS		 1815 return 0;
1816 }
1817
1818 return 1;
1819}
1820
1821/*
1822 * Produce the SCALAR(...,literal) or SCALAR(...,folded) tokens.
1823 */
1824
1825static int
1826yaml_parser_fetch_block_scalar(yaml_parser_t *parser, int literal)
1827{
625fcfe9 1828 yaml_token_t token;
eb9cceb5
KS		 1829
1830 /* Remove any potential simple keys. */
1831
1832 if (!yaml_parser_remove_simple_key(parser))
1833 return 0;
1834
1835 /* A simple key may follow a block scalar. */
1836
1837 parser->simple_key_allowed = 1;
1838
625fcfe9 1839 /* Create the SCALAR token and append it to the queue. */
eb9cceb5 1840
625fcfe9
KS		 1841 if (!yaml_parser_scan_block_scalar(parser, &token, literal))
1842 return 0;
eb9cceb5 1843
625fcfe9
KS		 1844 if (!ENQUEUE(parser, parser->tokens, token)) {
1845 yaml_token_delete(&token);
eb9cceb5
KS		 1846 return 0;
1847 }
1848
1849 return 1;
1850}
1851
1852/*
1853 * Produce the SCALAR(...,single-quoted) or SCALAR(...,double-quoted) tokens.
1854 */
1855
1856static int
1857yaml_parser_fetch_flow_scalar(yaml_parser_t *parser, int single)
1858{
625fcfe9 1859 yaml_token_t token;
eb9cceb5
KS		 1860
1861 /* A plain scalar could be a simple key. */
1862
1863 if (!yaml_parser_save_simple_key(parser))
1864 return 0;
1865
1866 /* A simple key cannot follow a flow scalar. */
1867
1868 parser->simple_key_allowed = 0;
1869
625fcfe9 1870 /* Create the SCALAR token and append it to the queue. */
eb9cceb5 1871
625fcfe9
KS		 1872 if (!yaml_parser_scan_flow_scalar(parser, &token, single))
1873 return 0;
eb9cceb5 1874
625fcfe9
KS		 1875 if (!ENQUEUE(parser, parser->tokens, token)) {
1876 yaml_token_delete(&token);
eb9cceb5
KS		 1877 return 0;
1878 }
1879
1880 return 1;
1881}
1882
1883/*
1884 * Produce the SCALAR(...,plain) token.
1885 */
1886
1887static int
1888yaml_parser_fetch_plain_scalar(yaml_parser_t *parser)
1889{
625fcfe9 1890 yaml_token_t token;
eb9cceb5
KS		 1891
1892 /* A plain scalar could be a simple key. */
1893
1894 if (!yaml_parser_save_simple_key(parser))
1895 return 0;
1896
1897 /* A simple key cannot follow a flow scalar. */
1898
1899 parser->simple_key_allowed = 0;
1900
625fcfe9 1901 /* Create the SCALAR token and append it to the queue. */
eb9cceb5 1902
625fcfe9
KS		 1903 if (!yaml_parser_scan_plain_scalar(parser, &token))
1904 return 0;
eb9cceb5 1905
625fcfe9
KS		 1906 if (!ENQUEUE(parser, parser->tokens, token)) {
1907 yaml_token_delete(&token);
eb9cceb5
KS		 1908 return 0;
1909 }
1910
1911 return 1;
1912}
1913
e71095e3
KS		 1914/*
1915 * Eat whitespaces and comments until the next token is found.
1916 */
1917
1918static int
1919yaml_parser_scan_to_next_token(yaml_parser_t *parser)
1920{
1921 /* Until the next token is not found. */
1922
1923 while (1)
1924 {
1925 /* Allow the BOM mark to start a line. */
1926
625fcfe9 1927 if (!CACHE(parser, 1)) return 0;
e71095e3 1928
e35af832 1929 if (parser->mark.column == 0 && IS_BOM(parser->buffer))
625fcfe9 1930 SKIP(parser);
e71095e3
KS		 1931
1932 /*
1933 * Eat whitespaces.
1934 *
1935 * Tabs are allowed:
1936 *
1937 * - in the flow context;
1938 * - in the block context, but not at the beginning of the line or
986dbde7 1939 * after '-', '?', or ':' (complex value).
e71095e3
KS		 1940 */
1941
625fcfe9 1942 if (!CACHE(parser, 1)) return 0;
e71095e3 1943
e35af832 1944 while (CHECK(parser->buffer,' ') ||
e71095e3 1945 ((parser->flow_level || !parser->simple_key_allowed) &&
e35af832 1946 CHECK(parser->buffer, '\t'))) {
625fcfe9
KS		 1947 SKIP(parser);
1948 if (!CACHE(parser, 1)) return 0;
e71095e3
KS		 1949 }
1950
1951 /* Eat a comment until a line break. */
1952
e35af832
KS		 1953 if (CHECK(parser->buffer, '#')) {
1954 while (!IS_BREAKZ(parser->buffer)) {
625fcfe9
KS		 1955 SKIP(parser);
1956 if (!CACHE(parser, 1)) return 0;
e71095e3
KS		 1957 }
1958 }
1959
1960 /* If it is a line break, eat it. */
1961
e35af832 1962 if (IS_BREAK(parser->buffer))
e71095e3 1963 {
625fcfe9
KS		 1964 if (!CACHE(parser, 2)) return 0;
1965 SKIP_LINE(parser);
e71095e3
KS		 1966
1967 /* In the block context, a new line may start a simple key. */
1968
1969 if (!parser->flow_level) {
1970 parser->simple_key_allowed = 1;
1971 }
1972 }
1973 else
1974 {
1975 /* We have found a token. */
1976
1977 break;
1978 }
1979 }
1980
1981 return 1;
1982}
1983
1984/*
1985 * Scan a YAML-DIRECTIVE or TAG-DIRECTIVE token.
1986 *
1987 * Scope:
1988 * %YAML 1.1 # a comment \n
1989 * ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
1990 * %TAG !yaml! tag:yaml.org,2002: \n
1991 * ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
1992 */
1993
625fcfe9
KS		 1994int
1995yaml_parser_scan_directive(yaml_parser_t *parser, yaml_token_t *token)
e71095e3
KS		 1996{
1997 yaml_mark_t start_mark, end_mark;
1998 yaml_char_t *name = NULL;
1999 int major, minor;
2000 yaml_char_t *handle = NULL, *prefix = NULL;
e71095e3
KS		 2001
2002 /* Eat '%'. */
2003
625fcfe9 2004 start_mark = parser->mark;
e71095e3 2005
625fcfe9 2006 SKIP(parser);
e71095e3
KS		 2007
2008 /* Scan the directive name. */
2009
2010 if (!yaml_parser_scan_directive_name(parser, start_mark, &name))
2011 goto error;
2012
2013 /* Is it a YAML directive? */
2014
2015 if (strcmp((char *)name, "YAML") == 0)
2016 {
2017 /* Scan the VERSION directive value. */
2018
2019 if (!yaml_parser_scan_version_directive_value(parser, start_mark,
2020 &major, &minor))
2021 goto error;
2022
625fcfe9 2023 end_mark = parser->mark;
e71095e3
KS		 2024
2025 /* Create a VERSION-DIRECTIVE token. */
2026
625fcfe9 2027 VERSION_DIRECTIVE_TOKEN_INIT(*token, major, minor,
e71095e3 2028 start_mark, end_mark);
e71095e3
KS		 2029 }
2030
2031 /* Is it a TAG directive? */
2032
2033 else if (strcmp((char *)name, "TAG") == 0)
2034 {
2035 /* Scan the TAG directive value. */
2036
2037 if (!yaml_parser_scan_tag_directive_value(parser, start_mark,
2038 &handle, &prefix))
2039 goto error;
2040
625fcfe9 2041 end_mark = parser->mark;
e71095e3
KS		 2042
2043 /* Create a TAG-DIRECTIVE token. */
2044
625fcfe9 2045 TAG_DIRECTIVE_TOKEN_INIT(*token, handle, prefix,
e71095e3 2046 start_mark, end_mark);
e71095e3
KS		 2047 }
2048
2049 /* Unknown directive. */
2050
2051 else
2052 {
92d41fe1 2053 yaml_parser_set_scanner_error(parser, "while scanning a directive",
bdf4d192 2054 start_mark, "found unknown directive name");
e71095e3
KS		 2055 goto error;
2056 }
2057
2058 /* Eat the rest of the line including any comments. */
2059
625fcfe9
KS		 2060 if (!CACHE(parser, 1)) goto error;
2061
e35af832 2062 while (IS_BLANK(parser->buffer)) {
625fcfe9
KS		 2063 SKIP(parser);
2064 if (!CACHE(parser, 1)) goto error;
e71095e3
KS		 2065 }
2066
e35af832
KS		 2067 if (CHECK(parser->buffer, '#')) {
2068 while (!IS_BREAKZ(parser->buffer)) {
625fcfe9
KS		 2069 SKIP(parser);
2070 if (!CACHE(parser, 1)) goto error;
e71095e3
KS		 2071 }
2072 }
2073
2074 /* Check if we are at the end of the line. */
2075
e35af832 2076 if (!IS_BREAKZ(parser->buffer)) {
92d41fe1 2077 yaml_parser_set_scanner_error(parser, "while scanning a directive",
6be8109b 2078 start_mark, "did not find expected comment or line break");
e71095e3
KS		 2079 goto error;
2080 }
2081
2082 /* Eat a line break. */
2083
e35af832 2084 if (IS_BREAK(parser->buffer)) {
625fcfe9
KS		 2085 if (!CACHE(parser, 2)) goto error;
2086 SKIP_LINE(parser);
e71095e3
KS		 2087 }
2088
2089 yaml_free(name);
2090
625fcfe9 2091 return 1;
e71095e3
KS		 2092
2093error:
e71095e3
KS		 2094 yaml_free(prefix);
2095 yaml_free(handle);
2096 yaml_free(name);
625fcfe9 2097 return 0;
e71095e3
KS		 2098}
2099
2100/*
2101 * Scan the directive name.
2102 *
2103 * Scope:
2104 * %YAML 1.1 # a comment \n
2105 * ^^^^
2106 * %TAG !yaml! tag:yaml.org,2002: \n
2107 * ^^^
2108 */
2109
2110static int
2111yaml_parser_scan_directive_name(yaml_parser_t *parser,
2112 yaml_mark_t start_mark, yaml_char_t **name)
2113{
625fcfe9 2114 yaml_string_t string = NULL_STRING;
e71095e3 2115
625fcfe9 2116 if (!STRING_INIT(parser, string, INITIAL_STRING_SIZE)) goto error;
e71095e3
KS		 2117
2118 /* Consume the directive name. */
2119
625fcfe9 2120 if (!CACHE(parser, 1)) goto error;
e71095e3 2121
e35af832 2122 while (IS_ALPHA(parser->buffer))
e71095e3 2123 {
625fcfe9
KS		 2124 if (!READ(parser, string)) goto error;
2125 if (!CACHE(parser, 1)) goto error;
e71095e3
KS		 2126 }
2127
2128 /* Check if the name is empty. */
2129
625fcfe9 2130 if (string.start == string.pointer) {
e71095e3 2131 yaml_parser_set_scanner_error(parser, "while scanning a directive",
6be8109b 2132 start_mark, "could not find expected directive name");
e71095e3
KS		 2133 goto error;
2134 }
2135
2136 /* Check for an blank character after the name. */
2137
e35af832 2138 if (!IS_BLANKZ(parser->buffer)) {
e71095e3
KS		 2139 yaml_parser_set_scanner_error(parser, "while scanning a directive",
2140 start_mark, "found unexpected non-alphabetical character");
2141 goto error;
2142 }
2143
625fcfe9 2144 *name = string.start;
e71095e3
KS		 2145
2146 return 1;
2147
2148error:
625fcfe9 2149 STRING_DEL(parser, string);
e71095e3
KS		 2150 return 0;
2151}
2152
2153/*
2154 * Scan the value of VERSION-DIRECTIVE.
2155 *
2156 * Scope:
2157 * %YAML 1.1 # a comment \n
2158 * ^^^^^^
2159 */
2160
2161static int
2162yaml_parser_scan_version_directive_value(yaml_parser_t *parser,
2163 yaml_mark_t start_mark, int *major, int *minor)
2164{
2165 /* Eat whitespaces. */
2166
625fcfe9 2167 if (!CACHE(parser, 1)) return 0;
e71095e3 2168
e35af832 2169 while (IS_BLANK(parser->buffer)) {
625fcfe9
KS		 2170 SKIP(parser);
2171 if (!CACHE(parser, 1)) return 0;
e71095e3
KS		 2172 }
2173
2174 /* Consume the major version number. */
2175
2176 if (!yaml_parser_scan_version_directive_number(parser, start_mark, major))
2177 return 0;
2178
2179 /* Eat '.'. */
2180
e35af832 2181 if (!CHECK(parser->buffer, '.')) {
e71095e3
KS		 2182 return yaml_parser_set_scanner_error(parser, "while scanning a %YAML directive",
2183 start_mark, "did not find expected digit or '.' character");
2184 }
2185
625fcfe9 2186 SKIP(parser);
e71095e3
KS		 2187
2188 /* Consume the minor version number. */
2189
2190 if (!yaml_parser_scan_version_directive_number(parser, start_mark, minor))
2191 return 0;
ab01bac8
KS		 2192
2193 return 1;
e71095e3
KS		 2194}
2195
2196#define MAX_NUMBER_LENGTH 9
2197
2198/*
2199 * Scan the version number of VERSION-DIRECTIVE.
2200 *
2201 * Scope:
2202 * %YAML 1.1 # a comment \n
2203 * ^
2204 * %YAML 1.1 # a comment \n
2205 * ^
2206 */
2207
2208static int
2209yaml_parser_scan_version_directive_number(yaml_parser_t *parser,
2210 yaml_mark_t start_mark, int *number)
2211{
2212 int value = 0;
2213 size_t length = 0;
2214
2215 /* Repeat while the next character is digit. */
2216
625fcfe9 2217 if (!CACHE(parser, 1)) return 0;
e71095e3 2218
e35af832 2219 while (IS_DIGIT(parser->buffer))
e71095e3
KS		 2220 {
2221 /* Check if the number is too long. */
2222
2223 if (++length > MAX_NUMBER_LENGTH) {
2224 return yaml_parser_set_scanner_error(parser, "while scanning a %YAML directive",
2225 start_mark, "found extremely long version number");
2226 }
2227
e35af832 2228 value = value*10 + AS_DIGIT(parser->buffer);
e71095e3 2229
625fcfe9 2230 SKIP(parser);
e71095e3 2231
625fcfe9 2232 if (!CACHE(parser, 1)) return 0;
e71095e3
KS		 2233 }
2234
2235 /* Check if the number was present. */
2236
2237 if (!length) {
2238 return yaml_parser_set_scanner_error(parser, "while scanning a %YAML directive",
2239 start_mark, "did not find expected version number");
2240 }
2241
2242 *number = value;
2243
2244 return 1;
2245}
2246
2247/*
2248 * Scan the value of a TAG-DIRECTIVE token.
2249 *
2250 * Scope:
2251 * %TAG !yaml! tag:yaml.org,2002: \n
2252 * ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
2253 */
2254
2255static int
2256yaml_parser_scan_tag_directive_value(yaml_parser_t *parser,
2257 yaml_mark_t start_mark, yaml_char_t **handle, yaml_char_t **prefix)
2258{
2259 yaml_char_t *handle_value = NULL;
2260 yaml_char_t *prefix_value = NULL;
2261
2262 /* Eat whitespaces. */
2263
625fcfe9 2264 if (!CACHE(parser, 1)) goto error;
e71095e3 2265
e35af832 2266 while (IS_BLANK(parser->buffer)) {
625fcfe9
KS		 2267 SKIP(parser);
2268 if (!CACHE(parser, 1)) goto error;
e71095e3
KS		 2269 }
2270
2271 /* Scan a handle. */
2272
2273 if (!yaml_parser_scan_tag_handle(parser, 1, start_mark, &handle_value))
2274 goto error;
2275
2276 /* Expect a whitespace. */
2277
625fcfe9 2278 if (!CACHE(parser, 1)) goto error;
e71095e3 2279
e35af832 2280 if (!IS_BLANK(parser->buffer)) {
e71095e3
KS		 2281 yaml_parser_set_scanner_error(parser, "while scanning a %TAG directive",
2282 start_mark, "did not find expected whitespace");
2283 goto error;
2284 }
2285
2286 /* Eat whitespaces. */
2287
e35af832 2288 while (IS_BLANK(parser->buffer)) {
625fcfe9
KS		 2289 SKIP(parser);
2290 if (!CACHE(parser, 1)) goto error;
e71095e3
KS		 2291 }
2292
2293 /* Scan a prefix. */
2294
2295 if (!yaml_parser_scan_tag_uri(parser, 1, NULL, start_mark, &prefix_value))
2296 goto error;
2297
2298 /* Expect a whitespace or line break. */
2299
625fcfe9 2300 if (!CACHE(parser, 1)) goto error;
e71095e3 2301
e35af832 2302 if (!IS_BLANKZ(parser->buffer)) {
e71095e3
KS		 2303 yaml_parser_set_scanner_error(parser, "while scanning a %TAG directive",
2304 start_mark, "did not find expected whitespace or line break");
2305 goto error;
2306 }
2307
2308 *handle = handle_value;
2309 *prefix = prefix_value;
2310
2311 return 1;
2312
2313error:
2314 yaml_free(handle_value);
2315 yaml_free(prefix_value);
2316 return 0;
2317}
2318
625fcfe9
KS		 2319static int
2320yaml_parser_scan_anchor(yaml_parser_t *parser, yaml_token_t *token,
e71095e3
KS		 2321 yaml_token_type_t type)
2322{
2323 int length = 0;
2324 yaml_mark_t start_mark, end_mark;
625fcfe9 2325 yaml_string_t string = NULL_STRING;
e71095e3 2326
625fcfe9 2327 if (!STRING_INIT(parser, string, INITIAL_STRING_SIZE)) goto error;
e71095e3
KS		 2328
2329 /* Eat the indicator character. */
2330
625fcfe9 2331 start_mark = parser->mark;
e71095e3 2332
625fcfe9 2333 SKIP(parser);
e71095e3
KS		 2334
2335 /* Consume the value. */
2336
625fcfe9 2337 if (!CACHE(parser, 1)) goto error;
e71095e3 2338
e35af832 2339 while (IS_ALPHA(parser->buffer)) {
625fcfe9
KS		 2340 if (!READ(parser, string)) goto error;
2341 if (!CACHE(parser, 1)) goto error;
e71095e3
KS		 2342 length ++;
2343 }
2344
625fcfe9 2345 end_mark = parser->mark;
e71095e3
KS		 2346
2347 /*
2348 * Check if length of the anchor is greater than 0 and it is followed by
2349 * a whitespace character or one of the indicators:
2350 *
2351 * '?', ':', ',', ']', '}', '%', '@', '`'.
2352 */
2353
e35af832
KS		 2354 if (!length || !(IS_BLANKZ(parser->buffer) || CHECK(parser->buffer, '?')
2355 || CHECK(parser->buffer, ':') || CHECK(parser->buffer, ',')
2356 || CHECK(parser->buffer, ']') || CHECK(parser->buffer, '}')
2357 || CHECK(parser->buffer, '%') || CHECK(parser->buffer, '@')
2358 || CHECK(parser->buffer, '`'))) {
e71095e3
KS		 2359 yaml_parser_set_scanner_error(parser, type == YAML_ANCHOR_TOKEN ?
2360 "while scanning an anchor" : "while scanning an alias", start_mark,
2361 "did not find expected alphabetic or numeric character");
2362 goto error;
2363 }
2364
2365 /* Create a token. */
2366
625fcfe9
KS		 2367 if (type == YAML_ANCHOR_TOKEN) {
2368 ANCHOR_TOKEN_INIT(*token, string.start, start_mark, end_mark);
2369 }
2370 else {
2371 ALIAS_TOKEN_INIT(*token, string.start, start_mark, end_mark);
92d41fe1 2372 }
e71095e3 2373
625fcfe9 2374 return 1;
e71095e3
KS		 2375
2376error:
625fcfe9 2377 STRING_DEL(parser, string);
e71095e3
KS		 2378 return 0;
2379}
2380
2381/*
2382 * Scan a TAG token.
2383 */
2384
625fcfe9
KS		 2385static int
2386yaml_parser_scan_tag(yaml_parser_t *parser, yaml_token_t *token)
e71095e3
KS		 2387{
2388 yaml_char_t *handle = NULL;
2389 yaml_char_t *suffix = NULL;
e71095e3
KS		 2390 yaml_mark_t start_mark, end_mark;
2391
625fcfe9 2392 start_mark = parser->mark;
e71095e3
KS		 2393
2394 /* Check if the tag is in the canonical form. */
2395
625fcfe9 2396 if (!CACHE(parser, 2)) goto error;
e71095e3 2397
e35af832 2398 if (CHECK_AT(parser->buffer, '<', 1))
e71095e3
KS		 2399 {
2400 /* Set the handle to '' */
2401
fc2dd942 2402 handle = YAML_MALLOC(1);
e71095e3
KS		 2403 if (!handle) goto error;
2404 handle[0] = '\0';
2405
2406 /* Eat '!<' */
2407
625fcfe9
KS		 2408 SKIP(parser);
2409 SKIP(parser);
e71095e3
KS		 2410
2411 /* Consume the tag value. */
2412
2413 if (!yaml_parser_scan_tag_uri(parser, 0, NULL, start_mark, &suffix))
2414 goto error;
2415
2416 /* Check for '>' and eat it. */
2417
e35af832 2418 if (!CHECK(parser->buffer, '>')) {
e71095e3
KS		 2419 yaml_parser_set_scanner_error(parser, "while scanning a tag",
2420 start_mark, "did not find the expected '>'");
2421 goto error;
2422 }
2423
625fcfe9 2424 SKIP(parser);
e71095e3
KS		 2425 }
2426 else
2427 {
2428 /* The tag has either the '!suffix' or the '!handle!suffix' form. */
2429
2430 /* First, try to scan a handle. */
2431
2432 if (!yaml_parser_scan_tag_handle(parser, 0, start_mark, &handle))
2433 goto error;
2434
2435 /* Check if it is, indeed, handle. */
2436
2437 if (handle[0] == '!' && handle[1] != '\0' && handle[strlen((char *)handle)-1] == '!')
2438 {
2439 /* Scan the suffix now. */
2440
2441 if (!yaml_parser_scan_tag_uri(parser, 0, NULL, start_mark, &suffix))
2442 goto error;
2443 }
2444 else
2445 {
2446 /* It wasn't a handle after all. Scan the rest of the tag. */
2447
2448 if (!yaml_parser_scan_tag_uri(parser, 0, handle, start_mark, &suffix))
2449 goto error;
2450
2451 /* Set the handle to '!'. */
2452
2453 yaml_free(handle);
fc2dd942 2454 handle = YAML_MALLOC(2);
e71095e3
KS		 2455 if (!handle) goto error;
2456 handle[0] = '!';
2457 handle[1] = '\0';
7e32c194
KS		 2458
2459 /*
625fcfe9
KS		 2460 * A special case: the '!' tag. Set the handle to '' and the
2461 * suffix to '!'.
7e32c194
KS		 2462 */
2463
2464 if (suffix[0] == '\0') {
2465 yaml_char_t *tmp = handle;
2466 handle = suffix;
2467 suffix = tmp;
2468 }
e71095e3
KS		 2469 }
2470 }
2471
2472 /* Check the character which ends the tag. */
2473
625fcfe9 2474 if (!CACHE(parser, 1)) goto error;
e71095e3 2475
e35af832 2476 if (!IS_BLANKZ(parser->buffer)) {
e71095e3 2477 yaml_parser_set_scanner_error(parser, "while scanning a tag",
6be8109b 2478 start_mark, "did not find expected whitespace or line break");
e71095e3
KS		 2479 goto error;
2480 }
2481
625fcfe9 2482 end_mark = parser->mark;
e71095e3
KS		 2483
2484 /* Create a token. */
2485
625fcfe9 2486 TAG_TOKEN_INIT(*token, handle, suffix, start_mark, end_mark);
e71095e3 2487
625fcfe9 2488 return 1;
e71095e3
KS		 2489
2490error:
2491 yaml_free(handle);
2492 yaml_free(suffix);
625fcfe9 2493 return 0;
e71095e3
KS		 2494}
2495
2496/*
2497 * Scan a tag handle.
2498 */
2499
2500static int
2501yaml_parser_scan_tag_handle(yaml_parser_t *parser, int directive,
2502 yaml_mark_t start_mark, yaml_char_t **handle)
2503{
625fcfe9 2504 yaml_string_t string = NULL_STRING;
e71095e3 2505
625fcfe9 2506 if (!STRING_INIT(parser, string, INITIAL_STRING_SIZE)) goto error;
e71095e3
KS		 2507
2508 /* Check the initial '!' character. */
2509
625fcfe9 2510 if (!CACHE(parser, 1)) goto error;
e71095e3 2511
e35af832 2512 if (!CHECK(parser->buffer, '!')) {
e71095e3
KS		 2513 yaml_parser_set_scanner_error(parser, directive ?
2514 "while scanning a tag directive" : "while scanning a tag",
2515 start_mark, "did not find expected '!'");
2516 goto error;
2517 }
2518
2519 /* Copy the '!' character. */
2520
625fcfe9 2521 if (!READ(parser, string)) goto error;
e71095e3
KS		 2522
2523 /* Copy all subsequent alphabetical and numerical characters. */
2524
625fcfe9 2525 if (!CACHE(parser, 1)) goto error;
e71095e3 2526
e35af832 2527 while (IS_ALPHA(parser->buffer))
e71095e3 2528 {
625fcfe9
KS		 2529 if (!READ(parser, string)) goto error;
2530 if (!CACHE(parser, 1)) goto error;
e71095e3
KS		 2531 }
2532
2533 /* Check if the trailing character is '!' and copy it. */
2534
e35af832 2535 if (CHECK(parser->buffer, '!'))
e71095e3 2536 {
625fcfe9 2537 if (!READ(parser, string)) goto error;
e71095e3
KS		 2538 }
2539 else
2540 {
2541 /*
7e32c194
KS		 2542 * It's either the '!' tag or not really a tag handle. If it's a %TAG
2543 * directive, it's an error. If it's a tag token, it must be a part of
2544 * URI.
e71095e3
KS		 2545 */
2546
625fcfe9 2547 if (directive && !(string.start[0] == '!' && string.start[1] == '\0')) {
7e32c194 2548 yaml_parser_set_scanner_error(parser, "while parsing a tag directive",
e71095e3
KS		 2549 start_mark, "did not find expected '!'");
2550 goto error;
2551 }
2552 }
2553
625fcfe9 2554 *handle = string.start;
e71095e3
KS		 2555
2556 return 1;
2557
2558error:
625fcfe9 2559 STRING_DEL(parser, string);
e71095e3
KS		 2560 return 0;
2561}
2562
2563/*
2564 * Scan a tag.
2565 */
2566
2567static int
2568yaml_parser_scan_tag_uri(yaml_parser_t *parser, int directive,
2569 yaml_char_t *head, yaml_mark_t start_mark, yaml_char_t **uri)
2570{
2571 size_t length = head ? strlen((char *)head) : 0;
625fcfe9 2572 yaml_string_t string = NULL_STRING;
e71095e3 2573
625fcfe9 2574 if (!STRING_INIT(parser, string, INITIAL_STRING_SIZE)) goto error;
e71095e3
KS		 2575
2576 /* Resize the string to include the head. */
2577
f56726b9 2578 while ((size_t)(string.end - string.start) <= length) {
625fcfe9
KS		 2579 if (!yaml_string_extend(&string.start, &string.pointer, &string.end)) {
2580 parser->error = YAML_MEMORY_ERROR;
2581 goto error;
2582 }
e71095e3
KS		 2583 }
2584
7e32c194
KS		 2585 /*
2586 * Copy the head if needed.
2587 *
2588 * Note that we don't copy the leading '!' character.
2589 */
e71095e3 2590
7e32c194 2591 if (length > 1) {
625fcfe9 2592 memcpy(string.start, head+1, length-1);
7e32c194 2593 string.pointer += length-1;
e71095e3
KS		 2594 }
2595
2596 /* Scan the tag. */
2597
625fcfe9 2598 if (!CACHE(parser, 1)) goto error;
e71095e3
KS		 2599
2600 /*
2601 * The set of characters that may appear in URI is as follows:
2602 *
2603 * '0'-'9', 'A'-'Z', 'a'-'z', '_', '-', ';', '/', '?', ':', '@', '&',
2604 * '=', '+', '$', ',', '.', '!', '~', '*', '\'', '(', ')', '[', ']',
2605 * '%'.
2606 */
2607
e35af832
KS		 2608 while (IS_ALPHA(parser->buffer) || CHECK(parser->buffer, ';')
2609 || CHECK(parser->buffer, '/') || CHECK(parser->buffer, '?')
2610 || CHECK(parser->buffer, ':') || CHECK(parser->buffer, '@')
2611 || CHECK(parser->buffer, '&') || CHECK(parser->buffer, '=')
2612 || CHECK(parser->buffer, '+') || CHECK(parser->buffer, '$')
2613 || CHECK(parser->buffer, ',') || CHECK(parser->buffer, '.')
2614 || CHECK(parser->buffer, '!') || CHECK(parser->buffer, '~')
2615 || CHECK(parser->buffer, '*') || CHECK(parser->buffer, '\'')
2616 || CHECK(parser->buffer, '(') || CHECK(parser->buffer, ')')
2617 || CHECK(parser->buffer, '[') || CHECK(parser->buffer, ']')
2618 || CHECK(parser->buffer, '%'))
e71095e3 2619 {
e71095e3
KS		 2620 /* Check if it is a URI-escape sequence. */
2621
e35af832 2622 if (CHECK(parser->buffer, '%')) {
d1003a9d
KS		 2623 if (!STRING_EXTEND(parser, string))
2624 goto error;
2625
e71095e3
KS		 2626 if (!yaml_parser_scan_uri_escapes(parser,
2627 directive, start_mark, &string)) goto error;
2628 }
2629 else {
625fcfe9 2630 if (!READ(parser, string)) goto error;
e71095e3
KS		 2631 }
2632
2633 length ++;
625fcfe9 2634 if (!CACHE(parser, 1)) goto error;
e71095e3
KS		 2635 }
2636
2637 /* Check if the tag is non-empty. */
2638
2639 if (!length) {
625fcfe9
KS		 2640 if (!STRING_EXTEND(parser, string))
2641 goto error;
2642
e71095e3
KS		 2643 yaml_parser_set_scanner_error(parser, directive ?
2644 "while parsing a %TAG directive" : "while parsing a tag",
2645 start_mark, "did not find expected tag URI");
2646 goto error;
2647 }
2648
625fcfe9 2649 *uri = string.start;
e71095e3
KS		 2650
2651 return 1;
2652
2653error:
625fcfe9 2654 STRING_DEL(parser, string);
e71095e3
KS		 2655 return 0;
2656}
2657
2658/*
2659 * Decode an URI-escape sequence corresponding to a single UTF-8 character.
2660 */
2661
2662static int
2663yaml_parser_scan_uri_escapes(yaml_parser_t *parser, int directive,
2664 yaml_mark_t start_mark, yaml_string_t *string)
2665{
2666 int width = 0;
2667
2668 /* Decode the required number of characters. */
2669
2670 do {
2671
2672 unsigned char octet = 0;
2673
2674 /* Check for a URI-escaped octet. */
2675
625fcfe9 2676 if (!CACHE(parser, 3)) return 0;
e71095e3 2677
e35af832
KS		 2678 if (!(CHECK(parser->buffer, '%')
2679 && IS_HEX_AT(parser->buffer, 1)
2680 && IS_HEX_AT(parser->buffer, 2))) {
e71095e3
KS		 2681 return yaml_parser_set_scanner_error(parser, directive ?
2682 "while parsing a %TAG directive" : "while parsing a tag",
2683 start_mark, "did not find URI escaped octet");
2684 }
2685
2686 /* Get the octet. */
2687
e35af832 2688 octet = (AS_HEX_AT(parser->buffer, 1) << 4) + AS_HEX_AT(parser->buffer, 2);
e71095e3
KS		 2689
2690 /* If it is the leading octet, determine the length of the UTF-8 sequence. */
2691
2692 if (!width)
2693 {
2694 width = (octet & 0x80) == 0x00 ? 1 :
2695 (octet & 0xE0) == 0xC0 ? 2 :
2696 (octet & 0xF0) == 0xE0 ? 3 :
2697 (octet & 0xF8) == 0xF0 ? 4 : 0;
2698 if (!width) {
2699 return yaml_parser_set_scanner_error(parser, directive ?
2700 "while parsing a %TAG directive" : "while parsing a tag",
2701 start_mark, "found an incorrect leading UTF-8 octet");
2702 }
2703 }
2704 else
2705 {
2706 /* Check if the trailing octet is correct. */
2707
2708 if ((octet & 0xC0) != 0x80) {
2709 return yaml_parser_set_scanner_error(parser, directive ?
2710 "while parsing a %TAG directive" : "while parsing a tag",
2711 start_mark, "found an incorrect trailing UTF-8 octet");
2712 }
2713 }
2714
2715 /* Copy the octet and move the pointers. */
2716
2717 *(string->pointer++) = octet;
625fcfe9
KS		 2718 SKIP(parser);
2719 SKIP(parser);
2720 SKIP(parser);
e71095e3
KS		 2721
2722 } while (--width);
2723
2724 return 1;
2725}
2726
92d41fe1
KS		 2727/*
2728 * Scan a block scalar.
2729 */
2730
625fcfe9
KS		 2731static int
2732yaml_parser_scan_block_scalar(yaml_parser_t *parser, yaml_token_t *token,
2733 int literal)
92d41fe1
KS		 2734{
2735 yaml_mark_t start_mark;
2736 yaml_mark_t end_mark;
625fcfe9
KS		 2737 yaml_string_t string = NULL_STRING;
2738 yaml_string_t leading_break = NULL_STRING;
2739 yaml_string_t trailing_breaks = NULL_STRING;
92d41fe1
KS		 2740 int chomping = 0;
2741 int increment = 0;
2742 int indent = 0;
2743 int leading_blank = 0;
2744 int trailing_blank = 0;
2745
625fcfe9
KS		 2746 if (!STRING_INIT(parser, string, INITIAL_STRING_SIZE)) goto error;
2747 if (!STRING_INIT(parser, leading_break, INITIAL_STRING_SIZE)) goto error;
2748 if (!STRING_INIT(parser, trailing_breaks, INITIAL_STRING_SIZE)) goto error;
92d41fe1
KS		 2749
2750 /* Eat the indicator '|' or '>'. */
2751
625fcfe9 2752 start_mark = parser->mark;
92d41fe1 2753
625fcfe9 2754 SKIP(parser);
92d41fe1
KS		 2755
2756 /* Scan the additional block scalar indicators. */
2757
625fcfe9 2758 if (!CACHE(parser, 1)) goto error;
92d41fe1
KS		 2759
2760 /* Check for a chomping indicator. */
2761
e35af832 2762 if (CHECK(parser->buffer, '+') || CHECK(parser->buffer, '-'))
92d41fe1
KS		 2763 {
2764 /* Set the chomping method and eat the indicator. */
2765
e35af832 2766 chomping = CHECK(parser->buffer, '+') ? +1 : -1;
92d41fe1 2767
625fcfe9 2768 SKIP(parser);
92d41fe1
KS		 2769
2770 /* Check for an indentation indicator. */
2771
625fcfe9 2772 if (!CACHE(parser, 1)) goto error;
92d41fe1 2773
e35af832 2774 if (IS_DIGIT(parser->buffer))
92d41fe1 2775 {
bdf4d192 2776 /* Check that the indentation is greater than 0. */
92d41fe1 2777
e35af832 2778 if (CHECK(parser->buffer, '0')) {
92d41fe1 2779 yaml_parser_set_scanner_error(parser, "while scanning a block scalar",
bdf4d192 2780 start_mark, "found an indentation indicator equal to 0");
92d41fe1
KS		 2781 goto error;
2782 }
2783
bdf4d192 2784 /* Get the indentation level and eat the indicator. */
92d41fe1 2785
e35af832 2786 increment = AS_DIGIT(parser->buffer);
92d41fe1 2787
625fcfe9 2788 SKIP(parser);
92d41fe1
KS		 2789 }
2790 }
2791
2792 /* Do the same as above, but in the opposite order. */
2793
e35af832 2794 else if (IS_DIGIT(parser->buffer))
92d41fe1 2795 {
e35af832 2796 if (CHECK(parser->buffer, '0')) {
92d41fe1 2797 yaml_parser_set_scanner_error(parser, "while scanning a block scalar",
bdf4d192 2798 start_mark, "found an indentation indicator equal to 0");
92d41fe1
KS		 2799 goto error;
2800 }
2801
e35af832 2802 increment = AS_DIGIT(parser->buffer);
92d41fe1 2803
625fcfe9 2804 SKIP(parser);
92d41fe1 2805
625fcfe9 2806 if (!CACHE(parser, 1)) goto error;
92d41fe1 2807
e35af832
KS		 2808 if (CHECK(parser->buffer, '+') || CHECK(parser->buffer, '-')) {
2809 chomping = CHECK(parser->buffer, '+') ? +1 : -1;
625fcfe9
KS		 2810
2811 SKIP(parser);
92d41fe1
KS		 2812 }
2813 }
2814
2815 /* Eat whitespaces and comments to the end of the line. */
2816
625fcfe9 2817 if (!CACHE(parser, 1)) goto error;
92d41fe1 2818
e35af832 2819 while (IS_BLANK(parser->buffer)) {
625fcfe9
KS		 2820 SKIP(parser);
2821 if (!CACHE(parser, 1)) goto error;
92d41fe1
KS		 2822 }
2823
e35af832
KS		 2824 if (CHECK(parser->buffer, '#')) {
2825 while (!IS_BREAKZ(parser->buffer)) {
625fcfe9
KS		 2826 SKIP(parser);
2827 if (!CACHE(parser, 1)) goto error;
92d41fe1
KS		 2828 }
2829 }
2830
2831 /* Check if we are at the end of the line. */
2832
e35af832 2833 if (!IS_BREAKZ(parser->buffer)) {
92d41fe1 2834 yaml_parser_set_scanner_error(parser, "while scanning a block scalar",
6be8109b 2835 start_mark, "did not find expected comment or line break");
92d41fe1
KS		 2836 goto error;
2837 }
2838
2839 /* Eat a line break. */
2840
e35af832 2841 if (IS_BREAK(parser->buffer)) {
625fcfe9
KS		 2842 if (!CACHE(parser, 2)) goto error;
2843 SKIP_LINE(parser);
92d41fe1
KS		 2844 }
2845
625fcfe9 2846 end_mark = parser->mark;
92d41fe1 2847
bdf4d192 2848 /* Set the indentation level if it was specified. */
92d41fe1
KS		 2849
2850 if (increment) {
2851 indent = parser->indent >= 0 ? parser->indent+increment : increment;
2852 }
2853
2854 /* Scan the leading line breaks and determine the indentation level if needed. */
2855
21fbedd4 2856 if (!yaml_parser_scan_block_scalar_breaks(parser, &indent, &trailing_breaks,
92d41fe1
KS		 2857 start_mark, &end_mark)) goto error;
2858
2859 /* Scan the block scalar content. */
2860
625fcfe9 2861 if (!CACHE(parser, 1)) goto error;
92d41fe1 2862
0174ed6e 2863 while ((int)parser->mark.column == indent && !IS_Z(parser->buffer))
92d41fe1
KS		 2864 {
2865 /*
2866 * We are at the beginning of a non-empty line.
2867 */
2868
2869 /* Is it a trailing whitespace? */
2870
e35af832 2871 trailing_blank = IS_BLANK(parser->buffer);
92d41fe1
KS		 2872
2873 /* Check if we need to fold the leading line break. */
2874
625fcfe9 2875 if (!literal && (*leading_break.start == '\n')
92d41fe1
KS		 2876 && !leading_blank && !trailing_blank)
2877 {
2878 /* Do we need to join the lines by space? */
2879
625fcfe9
KS		 2880 if (*trailing_breaks.start == '\0') {
2881 if (!STRING_EXTEND(parser, string)) goto error;
92d41fe1
KS		 2882 *(string.pointer ++) = ' ';
2883 }
2884
625fcfe9 2885 CLEAR(parser, leading_break);
92d41fe1
KS		 2886 }
2887 else {
21fbedd4 2888 if (!JOIN(parser, string, leading_break)) goto error;
625fcfe9 2889 CLEAR(parser, leading_break);
92d41fe1
KS		 2890 }
2891
2892 /* Append the remaining line breaks. */
2893
21fbedd4 2894 if (!JOIN(parser, string, trailing_breaks)) goto error;
625fcfe9 2895 CLEAR(parser, trailing_breaks);
92d41fe1
KS		 2896
2897 /* Is it a leading whitespace? */
2898
e35af832 2899 leading_blank = IS_BLANK(parser->buffer);
92d41fe1
KS		 2900
2901 /* Consume the current line. */
2902
e35af832 2903 while (!IS_BREAKZ(parser->buffer)) {
625fcfe9
KS		 2904 if (!READ(parser, string)) goto error;
2905 if (!CACHE(parser, 1)) goto error;
92d41fe1
KS		 2906 }
2907
2908 /* Consume the line break. */
2909
625fcfe9 2910 if (!CACHE(parser, 2)) goto error;
92d41fe1 2911
625fcfe9 2912 if (!READ_LINE(parser, leading_break)) goto error;
92d41fe1 2913
bdf4d192 2914 /* Eat the following indentation spaces and line breaks. */
92d41fe1
KS		 2915
2916 if (!yaml_parser_scan_block_scalar_breaks(parser,
21fbedd4 2917 &indent, &trailing_breaks, start_mark, &end_mark)) goto error;
92d41fe1
KS		 2918 }
2919
2920 /* Chomp the tail. */
2921
2922 if (chomping != -1) {
21fbedd4 2923 if (!JOIN(parser, string, leading_break)) goto error;
92d41fe1
KS		 2924 }
2925 if (chomping == 1) {
21fbedd4 2926 if (!JOIN(parser, string, trailing_breaks)) goto error;
92d41fe1
KS		 2927 }
2928
2929 /* Create a token. */
2930
625fcfe9 2931 SCALAR_TOKEN_INIT(*token, string.start, string.pointer-string.start,
92d41fe1
KS		 2932 literal ? YAML_LITERAL_SCALAR_STYLE : YAML_FOLDED_SCALAR_STYLE,
2933 start_mark, end_mark);
92d41fe1 2934
625fcfe9
KS		 2935 STRING_DEL(parser, leading_break);
2936 STRING_DEL(parser, trailing_breaks);
92d41fe1 2937
625fcfe9 2938 return 1;
92d41fe1
KS		 2939
2940error:
625fcfe9
KS		 2941 STRING_DEL(parser, string);
2942 STRING_DEL(parser, leading_break);
2943 STRING_DEL(parser, trailing_breaks);
92d41fe1 2944
625fcfe9 2945 return 0;
92d41fe1
KS		 2946}
2947
2948/*
bdf4d192
SH		 2949 * Scan indentation spaces and line breaks for a block scalar. Determine the
2950 * indentation level if needed.
92d41fe1
KS		 2951 */
2952
2953static int
2954yaml_parser_scan_block_scalar_breaks(yaml_parser_t *parser,
2955 int *indent, yaml_string_t *breaks,
2956 yaml_mark_t start_mark, yaml_mark_t *end_mark)
2957{
2958 int max_indent = 0;
2959
625fcfe9 2960 *end_mark = parser->mark;
92d41fe1 2961
bdf4d192 2962 /* Eat the indentation spaces and line breaks. */
92d41fe1
KS		 2963
2964 while (1)
2965 {
bdf4d192 2966 /* Eat the indentation spaces. */
92d41fe1 2967
625fcfe9 2968 if (!CACHE(parser, 1)) return 0;
92d41fe1 2969
0174ed6e 2970 while ((!*indent || (int)parser->mark.column < *indent)
e35af832 2971 && IS_SPACE(parser->buffer)) {
625fcfe9
KS		 2972 SKIP(parser);
2973 if (!CACHE(parser, 1)) return 0;
92d41fe1
KS		 2974 }
2975
0174ed6e
KS		 2976 if ((int)parser->mark.column > max_indent)
2977 max_indent = (int)parser->mark.column;
92d41fe1 2978
bdf4d192 2979 /* Check for a tab character messing the indentation. */
92d41fe1 2980
0174ed6e 2981 if ((!*indent || (int)parser->mark.column < *indent)
e35af832 2982 && IS_TAB(parser->buffer)) {
92d41fe1 2983 return yaml_parser_set_scanner_error(parser, "while scanning a block scalar",
bdf4d192 2984 start_mark, "found a tab character where an indentation space is expected");
92d41fe1
KS		 2985 }
2986
2987 /* Have we found a non-empty line? */
2988
e35af832 2989 if (!IS_BREAK(parser->buffer)) break;
92d41fe1
KS		 2990
2991 /* Consume the line break. */
2992
625fcfe9
KS		 2993 if (!CACHE(parser, 2)) return 0;
2994 if (!READ_LINE(parser, *breaks)) return 0;
2995 *end_mark = parser->mark;
92d41fe1
KS		 2996 }
2997
2998 /* Determine the indentation level if needed. */
2999
3000 if (!*indent) {
3001 *indent = max_indent;
3002 if (*indent < parser->indent + 1)
3003 *indent = parser->indent + 1;
3004 if (*indent < 1)
3005 *indent = 1;
3006 }
3007
986dbde7 3008 return 1;
92d41fe1
KS		 3009}
3010
21fbedd4
KS		 3011/*
3012 * Scan a quoted scalar.
3013 */
3014
625fcfe9
KS		 3015static int
3016yaml_parser_scan_flow_scalar(yaml_parser_t *parser, yaml_token_t *token,
3017 int single)
21fbedd4
KS		 3018{
3019 yaml_mark_t start_mark;
3020 yaml_mark_t end_mark;
625fcfe9
KS		 3021 yaml_string_t string = NULL_STRING;
3022 yaml_string_t leading_break = NULL_STRING;
3023 yaml_string_t trailing_breaks = NULL_STRING;
3024 yaml_string_t whitespaces = NULL_STRING;
21fbedd4
KS		 3025 int leading_blanks;
3026
625fcfe9
KS		 3027 if (!STRING_INIT(parser, string, INITIAL_STRING_SIZE)) goto error;
3028 if (!STRING_INIT(parser, leading_break, INITIAL_STRING_SIZE)) goto error;
3029 if (!STRING_INIT(parser, trailing_breaks, INITIAL_STRING_SIZE)) goto error;
3030 if (!STRING_INIT(parser, whitespaces, INITIAL_STRING_SIZE)) goto error;
21fbedd4
KS		 3031
3032 /* Eat the left quote. */
3033
625fcfe9 3034 start_mark = parser->mark;
21fbedd4 3035
625fcfe9 3036 SKIP(parser);
21fbedd4
KS		 3037
3038 /* Consume the content of the quoted scalar. */
3039
3040 while (1)
3041 {
3042 /* Check that there are no document indicators at the beginning of the line. */
3043
625fcfe9 3044 if (!CACHE(parser, 4)) goto error;
21fbedd4 3045
625fcfe9 3046 if (parser->mark.column == 0 &&
e35af832
KS		 3047 ((CHECK_AT(parser->buffer, '-', 0) &&
3048 CHECK_AT(parser->buffer, '-', 1) &&
3049 CHECK_AT(parser->buffer, '-', 2)) ||
3050 (CHECK_AT(parser->buffer, '.', 0) &&
3051 CHECK_AT(parser->buffer, '.', 1) &&
3052 CHECK_AT(parser->buffer, '.', 2))) &&
3053 IS_BLANKZ_AT(parser->buffer, 3))
21fbedd4
KS		 3054 {
3055 yaml_parser_set_scanner_error(parser, "while scanning a quoted scalar",
3056 start_mark, "found unexpected document indicator");
3057 goto error;
3058 }
3059
3060 /* Check for EOF. */
3061
e35af832 3062 if (IS_Z(parser->buffer)) {
21fbedd4
KS		 3063 yaml_parser_set_scanner_error(parser, "while scanning a quoted scalar",
3064 start_mark, "found unexpected end of stream");
3065 goto error;
3066 }
3067
3068 /* Consume non-blank characters. */
3069
625fcfe9 3070 if (!CACHE(parser, 2)) goto error;
21fbedd4
KS		 3071
3072 leading_blanks = 0;
3073
e35af832 3074 while (!IS_BLANKZ(parser->buffer))
21fbedd4
KS		 3075 {
3076 /* Check for an escaped single quote. */
3077
e35af832
KS		 3078 if (single && CHECK_AT(parser->buffer, '\'', 0)
3079 && CHECK_AT(parser->buffer, '\'', 1))
21fbedd4 3080 {
625fcfe9 3081 if (!STRING_EXTEND(parser, string)) goto error;
21fbedd4 3082 *(string.pointer++) = '\'';
625fcfe9
KS		 3083 SKIP(parser);
3084 SKIP(parser);
21fbedd4
KS		 3085 }
3086
3087 /* Check for the right quote. */
3088
e35af832 3089 else if (CHECK(parser->buffer, single ? '\'' : '"'))
21fbedd4
KS		 3090 {
3091 break;
3092 }
3093
3094 /* Check for an escaped line break. */
3095
e35af832
KS		 3096 else if (!single && CHECK(parser->buffer, '\\')
3097 && IS_BREAK_AT(parser->buffer, 1))
21fbedd4 3098 {
625fcfe9
KS		 3099 if (!CACHE(parser, 3)) goto error;
3100 SKIP(parser);
3101 SKIP_LINE(parser);
21fbedd4
KS		 3102 leading_blanks = 1;
3103 break;
3104 }
3105
3106 /* Check for an escape sequence. */
3107
e35af832 3108 else if (!single && CHECK(parser->buffer, '\\'))
21fbedd4 3109 {
0174ed6e 3110 size_t code_length = 0;
21fbedd4 3111
625fcfe9
KS		 3112 if (!STRING_EXTEND(parser, string)) goto error;
3113
21fbedd4
KS		 3114 /* Check the escape character. */
3115
625fcfe9 3116 switch (parser->buffer.pointer[1])
21fbedd4
KS		 3117 {
3118 case '0':
3119 *(string.pointer++) = '\0';
3120 break;
3121
3122 case 'a':
3123 *(string.pointer++) = '\x07';
3124 break;
3125
3126 case 'b':
3127 *(string.pointer++) = '\x08';
3128 break;
3129
3130 case 't':
3131 case '\t':
3132 *(string.pointer++) = '\x09';
3133 break;
3134
3135 case 'n':
3136 *(string.pointer++) = '\x0A';
3137 break;
3138
3139 case 'v':
3140 *(string.pointer++) = '\x0B';
3141 break;
3142
3143 case 'f':
3144 *(string.pointer++) = '\x0C';
3145 break;
3146
3147 case 'r':
3148 *(string.pointer++) = '\x0D';
3149 break;
3150
3151 case 'e':
3152 *(string.pointer++) = '\x1B';
3153 break;
3154
3155 case ' ':
3156 *(string.pointer++) = '\x20';
3157 break;
3158
3159 case '"':
3160 *(string.pointer++) = '"';
3161 break;
3162
6db69c72
IN		 3163 case '/':
3164 *(string.pointer++) = '/';
3165 break;
3166
7e32c194
KS		 3167 case '\\':
3168 *(string.pointer++) = '\\';
3169 break;
3170
21fbedd4
KS		 3171 case 'N': /* NEL (#x85) */
3172 *(string.pointer++) = '\xC2';
3173 *(string.pointer++) = '\x85';
3174 break;
3175
3176 case '_': /* #xA0 */
3177 *(string.pointer++) = '\xC2';
3178 *(string.pointer++) = '\xA0';
3179 break;
3180
3181 case 'L': /* LS (#x2028) */
3182 *(string.pointer++) = '\xE2';
3183 *(string.pointer++) = '\x80';
3184 *(string.pointer++) = '\xA8';
3185 break;
3186
3187 case 'P': /* PS (#x2029) */
3188 *(string.pointer++) = '\xE2';
3189 *(string.pointer++) = '\x80';
7e32c194 3190 *(string.pointer++) = '\xA9';
21fbedd4
KS		 3191 break;
3192
3193 case 'x':
3194 code_length = 2;
3195 break;
3196
3197 case 'u':
3198 code_length = 4;
3199 break;
3200
3201 case 'U':
3202 code_length = 8;
3203 break;
3204
3205 default:
3206 yaml_parser_set_scanner_error(parser, "while parsing a quoted scalar",
3207 start_mark, "found unknown escape character");
3208 goto error;
3209 }
3210
625fcfe9
KS		 3211 SKIP(parser);
3212 SKIP(parser);
21fbedd4
KS		 3213
3214 /* Consume an arbitrary escape code. */
3215
3216 if (code_length)
3217 {
3218 unsigned int value = 0;
0174ed6e 3219 size_t k;
21fbedd4
KS		 3220
3221 /* Scan the character value. */
3222
625fcfe9 3223 if (!CACHE(parser, code_length)) goto error;
21fbedd4
KS		 3224
3225 for (k = 0; k < code_length; k ++) {
e35af832 3226 if (!IS_HEX_AT(parser->buffer, k)) {
21fbedd4
KS		 3227 yaml_parser_set_scanner_error(parser, "while parsing a quoted scalar",
3228 start_mark, "did not find expected hexdecimal number");
3229 goto error;
3230 }
e35af832 3231 value = (value << 4) + AS_HEX_AT(parser->buffer, k);
21fbedd4
KS		 3232 }
3233
3234 /* Check the value and write the character. */
3235
3236 if ((value >= 0xD800 && value <= 0xDFFF) || value > 0x10FFFF) {
3237 yaml_parser_set_scanner_error(parser, "while parsing a quoted scalar",
3238 start_mark, "found invalid Unicode character escape code");
3239 goto error;
3240 }
3241
3242 if (value <= 0x7F) {
3243 *(string.pointer++) = value;
3244 }
3245 else if (value <= 0x7FF) {
3246 *(string.pointer++) = 0xC0 + (value >> 6);
3247 *(string.pointer++) = 0x80 + (value & 0x3F);
3248 }
3249 else if (value <= 0xFFFF) {
3250 *(string.pointer++) = 0xE0 + (value >> 12);
3251 *(string.pointer++) = 0x80 + ((value >> 6) & 0x3F);
3252 *(string.pointer++) = 0x80 + (value & 0x3F);
3253 }
3254 else {
3255 *(string.pointer++) = 0xF0 + (value >> 18);
3256 *(string.pointer++) = 0x80 + ((value >> 12) & 0x3F);
3257 *(string.pointer++) = 0x80 + ((value >> 6) & 0x3F);
3258 *(string.pointer++) = 0x80 + (value & 0x3F);
3259 }
3260
3261 /* Advance the pointer. */
3262
3263 for (k = 0; k < code_length; k ++) {
625fcfe9 3264 SKIP(parser);
21fbedd4
KS		 3265 }
3266 }
3267 }
3268
3269 else
3270 {
3271 /* It is a non-escaped non-blank character. */
3272
625fcfe9 3273 if (!READ(parser, string)) goto error;
21fbedd4
KS		 3274 }
3275
625fcfe9 3276 if (!CACHE(parser, 2)) goto error;
21fbedd4
KS		 3277 }
3278
3279 /* Check if we are at the end of the scalar. */
3280
6bbc217f
IC		 3281 /* Fix for crash unitialized value crash
3282 * Credit for the bug and input is to OSS Fuzz
3283 * Credit for the fix to Alex Gaynor
3284 */
3285 if (!CACHE(parser, 1)) goto error;
e35af832 3286 if (CHECK(parser->buffer, single ? '\'' : '"'))
21fbedd4
KS		 3287 break;
3288
3289 /* Consume blank characters. */
3290
625fcfe9 3291 if (!CACHE(parser, 1)) goto error;
21fbedd4 3292
e35af832 3293 while (IS_BLANK(parser->buffer) || IS_BREAK(parser->buffer))
21fbedd4 3294 {
e35af832 3295 if (IS_BLANK(parser->buffer))
21fbedd4
KS		 3296 {
3297 /* Consume a space or a tab character. */
3298
3299 if (!leading_blanks) {
625fcfe9 3300 if (!READ(parser, whitespaces)) goto error;
21fbedd4 3301 }
7e32c194 3302 else {
625fcfe9 3303 SKIP(parser);
7e32c194 3304 }
21fbedd4
KS		 3305 }
3306 else
3307 {
625fcfe9 3308 if (!CACHE(parser, 2)) goto error;
21fbedd4
KS		 3309
3310 /* Check if it is a first line break. */
3311
3312 if (!leading_blanks)
3313 {
625fcfe9
KS		 3314 CLEAR(parser, whitespaces);
3315 if (!READ_LINE(parser, leading_break)) goto error;
21fbedd4
KS		 3316 leading_blanks = 1;
3317 }
3318 else
3319 {
625fcfe9 3320 if (!READ_LINE(parser, trailing_breaks)) goto error;
21fbedd4
KS		 3321 }
3322 }
625fcfe9 3323 if (!CACHE(parser, 1)) goto error;
21fbedd4
KS		 3324 }
3325
3326 /* Join the whitespaces or fold line breaks. */
3327
21fbedd4
KS		 3328 if (leading_blanks)
3329 {
3330 /* Do we need to fold line breaks? */
3331
625fcfe9
KS		 3332 if (leading_break.start[0] == '\n') {
3333 if (trailing_breaks.start[0] == '\0') {
3334 if (!STRING_EXTEND(parser, string)) goto error;
21fbedd4
KS		 3335 *(string.pointer++) = ' ';
3336 }
3337 else {
3338 if (!JOIN(parser, string, trailing_breaks)) goto error;
625fcfe9 3339 CLEAR(parser, trailing_breaks);
21fbedd4 3340 }
625fcfe9 3341 CLEAR(parser, leading_break);
21fbedd4
KS		 3342 }
3343 else {
3344 if (!JOIN(parser, string, leading_break)) goto error;
3345 if (!JOIN(parser, string, trailing_breaks)) goto error;
625fcfe9
KS		 3346 CLEAR(parser, leading_break);
3347 CLEAR(parser, trailing_breaks);
21fbedd4
KS		 3348 }
3349 }
3350 else
3351 {
3352 if (!JOIN(parser, string, whitespaces)) goto error;
625fcfe9 3353 CLEAR(parser, whitespaces);
21fbedd4
KS		 3354 }
3355 }
3356
3357 /* Eat the right quote. */
3358
625fcfe9 3359 SKIP(parser);
21fbedd4 3360
625fcfe9 3361 end_mark = parser->mark;
21fbedd4
KS		 3362
3363 /* Create a token. */
3364
625fcfe9 3365 SCALAR_TOKEN_INIT(*token, string.start, string.pointer-string.start,
21fbedd4
KS		 3366 single ? YAML_SINGLE_QUOTED_SCALAR_STYLE : YAML_DOUBLE_QUOTED_SCALAR_STYLE,
3367 start_mark, end_mark);
21fbedd4 3368
625fcfe9
KS		 3369 STRING_DEL(parser, leading_break);
3370 STRING_DEL(parser, trailing_breaks);
3371 STRING_DEL(parser, whitespaces);
21fbedd4 3372
625fcfe9 3373 return 1;
21fbedd4
KS		 3374
3375error:
625fcfe9
KS		 3376 STRING_DEL(parser, string);
3377 STRING_DEL(parser, leading_break);
3378 STRING_DEL(parser, trailing_breaks);
3379 STRING_DEL(parser, whitespaces);
21fbedd4 3380
625fcfe9 3381 return 0;
21fbedd4
KS		 3382}
3383
3384/*
3385 * Scan a plain scalar.
3386 */
3387
625fcfe9
KS		 3388static int
3389yaml_parser_scan_plain_scalar(yaml_parser_t *parser, yaml_token_t *token)
21fbedd4
KS		 3390{
3391 yaml_mark_t start_mark;
3392 yaml_mark_t end_mark;
625fcfe9
KS		 3393 yaml_string_t string = NULL_STRING;
3394 yaml_string_t leading_break = NULL_STRING;
3395 yaml_string_t trailing_breaks = NULL_STRING;
3396 yaml_string_t whitespaces = NULL_STRING;
21fbedd4
KS		 3397 int leading_blanks = 0;
3398 int indent = parser->indent+1;
3399
625fcfe9
KS		 3400 if (!STRING_INIT(parser, string, INITIAL_STRING_SIZE)) goto error;
3401 if (!STRING_INIT(parser, leading_break, INITIAL_STRING_SIZE)) goto error;
3402 if (!STRING_INIT(parser, trailing_breaks, INITIAL_STRING_SIZE)) goto error;
3403 if (!STRING_INIT(parser, whitespaces, INITIAL_STRING_SIZE)) goto error;
21fbedd4 3404
54815ffd 3405 start_mark = end_mark = parser->mark;
21fbedd4
KS		 3406
3407 /* Consume the content of the plain scalar. */
3408
3409 while (1)
3410 {
3411 /* Check for a document indicator. */
3412
625fcfe9 3413 if (!CACHE(parser, 4)) goto error;
21fbedd4 3414
625fcfe9 3415 if (parser->mark.column == 0 &&
e35af832
KS		 3416 ((CHECK_AT(parser->buffer, '-', 0) &&
3417 CHECK_AT(parser->buffer, '-', 1) &&
3418 CHECK_AT(parser->buffer, '-', 2)) ||
3419 (CHECK_AT(parser->buffer, '.', 0) &&
3420 CHECK_AT(parser->buffer, '.', 1) &&
3421 CHECK_AT(parser->buffer, '.', 2))) &&
3422 IS_BLANKZ_AT(parser->buffer, 3)) break;
21fbedd4
KS		 3423
3424 /* Check for a comment. */
3425
e35af832 3426 if (CHECK(parser->buffer, '#'))
21fbedd4
KS		 3427 break;
3428
3429 /* Consume non-blank characters. */
3430
e35af832 3431 while (!IS_BLANKZ(parser->buffer))
21fbedd4 3432 {
20496ee2
TM		 3433 /* Check for "x:" + one of ',?[]{}' in the flow context. TODO: Fix the test "spec-08-13".
3434 * This is not completely according to the spec
3435 * See http://yaml.org/spec/1.1/#id907281 9.1.3. Plain
3436 */
21fbedd4 3437
e35af832
KS		 3438 if (parser->flow_level
3439 && CHECK(parser->buffer, ':')
20496ee2
TM		 3440 && (
3441 CHECK_AT(parser->buffer, ',', 1)
3442 || CHECK_AT(parser->buffer, '?', 1)
3443 || CHECK_AT(parser->buffer, '[', 1)
3444 || CHECK_AT(parser->buffer, ']', 1)
3445 || CHECK_AT(parser->buffer, '{', 1)
3446 || CHECK_AT(parser->buffer, '}', 1)
3447 )
3448 ) {
21fbedd4
KS		 3449 yaml_parser_set_scanner_error(parser, "while scanning a plain scalar",
3450 start_mark, "found unexpected ':'");
3451 goto error;
3452 }
3453
3454 /* Check for indicators that may end a plain scalar. */
3455
e35af832
KS		 3456 if ((CHECK(parser->buffer, ':') && IS_BLANKZ_AT(parser->buffer, 1))
3457 || (parser->flow_level &&
20496ee2 3458 (CHECK(parser->buffer, ',')
e35af832
KS		 3459 || CHECK(parser->buffer, '?') || CHECK(parser->buffer, '[')
3460 || CHECK(parser->buffer, ']') || CHECK(parser->buffer, '{')
3461 || CHECK(parser->buffer, '}'))))
21fbedd4
KS		 3462 break;
3463
3464 /* Check if we need to join whitespaces and breaks. */
3465
625fcfe9 3466 if (leading_blanks || whitespaces.start != whitespaces.pointer)
21fbedd4 3467 {
21fbedd4
KS		 3468 if (leading_blanks)
3469 {
3470 /* Do we need to fold line breaks? */
3471
625fcfe9
KS		 3472 if (leading_break.start[0] == '\n') {
3473 if (trailing_breaks.start[0] == '\0') {
3474 if (!STRING_EXTEND(parser, string)) goto error;
21fbedd4
KS		 3475 *(string.pointer++) = ' ';
3476 }
3477 else {
3478 if (!JOIN(parser, string, trailing_breaks)) goto error;
625fcfe9 3479 CLEAR(parser, trailing_breaks);
21fbedd4 3480 }
625fcfe9 3481 CLEAR(parser, leading_break);
21fbedd4
KS		 3482 }
3483 else {
3484 if (!JOIN(parser, string, leading_break)) goto error;
3485 if (!JOIN(parser, string, trailing_breaks)) goto error;
625fcfe9
KS		 3486 CLEAR(parser, leading_break);
3487 CLEAR(parser, trailing_breaks);
21fbedd4
KS		 3488 }
3489
3490 leading_blanks = 0;
3491 }
3492 else
3493 {
3494 if (!JOIN(parser, string, whitespaces)) goto error;
625fcfe9 3495 CLEAR(parser, whitespaces);
21fbedd4
KS		 3496 }
3497 }
3498
3499 /* Copy the character. */
3500
625fcfe9 3501 if (!READ(parser, string)) goto error;
21fbedd4 3502
625fcfe9 3503 end_mark = parser->mark;
21fbedd4 3504
625fcfe9 3505 if (!CACHE(parser, 2)) goto error;
21fbedd4
KS		 3506 }
3507
3508 /* Is it the end? */
3509
e35af832 3510 if (!(IS_BLANK(parser->buffer) || IS_BREAK(parser->buffer)))
21fbedd4
KS		 3511 break;
3512
3513 /* Consume blank characters. */
3514
625fcfe9 3515 if (!CACHE(parser, 1)) goto error;
21fbedd4 3516
e35af832 3517 while (IS_BLANK(parser->buffer) || IS_BREAK(parser->buffer))
21fbedd4 3518 {
e35af832 3519 if (IS_BLANK(parser->buffer))
21fbedd4 3520 {
bdf4d192 3521 /* Check for tab character that abuse indentation. */
21fbedd4 3522
0174ed6e 3523 if (leading_blanks && (int)parser->mark.column < indent
e35af832 3524 && IS_TAB(parser->buffer)) {
21fbedd4 3525 yaml_parser_set_scanner_error(parser, "while scanning a plain scalar",
bdf4d192 3526 start_mark, "found a tab character that violate indentation");
7e32c194 3527 goto error;
21fbedd4
KS		 3528 }
3529
3530 /* Consume a space or a tab character. */
3531
3532 if (!leading_blanks) {
625fcfe9 3533 if (!READ(parser, whitespaces)) goto error;
21fbedd4 3534 }
7e32c194 3535 else {
625fcfe9 3536 SKIP(parser);
7e32c194 3537 }
21fbedd4
KS		 3538 }
3539 else
3540 {
625fcfe9 3541 if (!CACHE(parser, 2)) goto error;
21fbedd4
KS		 3542
3543 /* Check if it is a first line break. */
3544
3545 if (!leading_blanks)
3546 {
625fcfe9
KS		 3547 CLEAR(parser, whitespaces);
3548 if (!READ_LINE(parser, leading_break)) goto error;
21fbedd4
KS		 3549 leading_blanks = 1;
3550 }
3551 else
3552 {
625fcfe9 3553 if (!READ_LINE(parser, trailing_breaks)) goto error;
21fbedd4
KS		 3554 }
3555 }
625fcfe9 3556 if (!CACHE(parser, 1)) goto error;
21fbedd4
KS		 3557 }
3558
bdf4d192 3559 /* Check indentation level. */
21fbedd4 3560
0174ed6e 3561 if (!parser->flow_level && (int)parser->mark.column < indent)
21fbedd4
KS		 3562 break;
3563 }
3564
3565 /* Create a token. */
3566
625fcfe9 3567 SCALAR_TOKEN_INIT(*token, string.start, string.pointer-string.start,
21fbedd4 3568 YAML_PLAIN_SCALAR_STYLE, start_mark, end_mark);
21fbedd4
KS		 3569
3570 /* Note that we change the 'simple_key_allowed' flag. */
3571
3572 if (leading_blanks) {
3573 parser->simple_key_allowed = 1;
3574 }
3575
625fcfe9
KS		 3576 STRING_DEL(parser, leading_break);
3577 STRING_DEL(parser, trailing_breaks);
3578 STRING_DEL(parser, whitespaces);
21fbedd4 3579
625fcfe9 3580 return 1;
21fbedd4
KS		 3581
3582error:
625fcfe9
KS		 3583 STRING_DEL(parser, string);
3584 STRING_DEL(parser, leading_break);
3585 STRING_DEL(parser, trailing_breaks);
3586 STRING_DEL(parser, whitespaces);
21fbedd4 3587
625fcfe9 3588 return 0;
21fbedd4 3589}
Fast YAML 1.1 parser and emitter library
This page took 0.617703 seconds and 5 git commands to generate.