]> andersk Git - libyaml.git/blame - src/scanner.c
andersk / libyaml.git / blame
? search:
summary | shortlog | log | commit | commitdiff | tree
blob | blame (incremental) | history | HEAD
Scalar writers are completed.
[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 # ']'
41 * BLOCK-SEQUENCE-START # '{'
42 * BLOCK-SEQUENCE-END # '}'
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 *
73 * The correspoding sequence of tokens:
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
eb9cceb5
KS		 618yaml_parser_roll_indent(yaml_parser_t *parser, int column,
619 int number, yaml_token_type_t type, yaml_mark_t mark);
03be97ab
KS		 620
621static int
f2b59d4d 622yaml_parser_unroll_indent(yaml_parser_t *parser, int 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
KS		 764 /* Fetch the next token from the queue. */
765
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,
1081 "could not found expected ':'");
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
KS		 1105 int required = (!parser->flow_level
1106 && parser->indent == parser->mark.column);
eb9cceb5
KS		 1107
1108 /*
1109 * A simple key is required only when it is the first token in the current
1110 * line. Therefore it is always allowed. But we add a check anyway.
1111 */
1112
1113 assert(parser->simple_key_allowed || !required); /* Impossible. */
1114
1115 /*
1116 * If the current position may start a simple key, save it.
1117 */
1118
1119 if (parser->simple_key_allowed)
1120 {
625fcfe9
KS		 1121 yaml_simple_key_t simple_key = { 1, required,
1122 parser->tokens_parsed + parser->tokens.tail - parser->tokens.head,
1123 parser->mark };
eb9cceb5
KS		 1124
1125 if (!yaml_parser_remove_simple_key(parser)) return 0;
1126
625fcfe9 1127 *(parser->simple_keys.top-1) = simple_key;
eb9cceb5
KS		 1128 }
1129
1130 return 1;
1131}
1132
1133/*
1134 * Remove a potential simple key at the current flow level.
1135 */
1136
1137static int
1138yaml_parser_remove_simple_key(yaml_parser_t *parser)
1139{
625fcfe9 1140 yaml_simple_key_t *simple_key = parser->simple_keys.top-1;
eb9cceb5 1141
625fcfe9 1142 if (simple_key->possible)
eb9cceb5
KS		 1143 {
1144 /* If the key is required, it is an error. */
1145
1146 if (simple_key->required) {
1147 return yaml_parser_set_scanner_error(parser,
1148 "while scanning a simple key", simple_key->mark,
1149 "could not found expected ':'");
1150 }
625fcfe9 1151 }
eb9cceb5 1152
625fcfe9 1153 /* Remove the key from the stack. */
eb9cceb5 1154
625fcfe9 1155 simple_key->possible = 0;
eb9cceb5
KS		 1156
1157 return 1;
1158}
1159
1160/*
1161 * Increase the flow level and resize the simple key list if needed.
1162 */
1163
1164static int
1165yaml_parser_increase_flow_level(yaml_parser_t *parser)
1166{
625fcfe9 1167 yaml_simple_key_t empty_simple_key = { 0, 0, 0, { 0, 0, 0 } };
eb9cceb5 1168
625fcfe9 1169 /* Reset the simple key on the next level. */
eb9cceb5 1170
625fcfe9
KS		 1171 if (!PUSH(parser, parser->simple_keys, empty_simple_key))
1172 return 0;
1173
1174 /* Increase the flow level. */
eb9cceb5 1175
625fcfe9 1176 parser->flow_level++;
eb9cceb5
KS		 1177
1178 return 1;
1179}
1180
1181/*
1182 * Decrease the flow level.
1183 */
1184
1185static int
1186yaml_parser_decrease_flow_level(yaml_parser_t *parser)
1187{
625fcfe9
KS		 1188 if (parser->flow_level) {
1189 parser->flow_level --;
1190 POP(parser, parser->simple_keys);
eb9cceb5
KS		 1191 }
1192
eb9cceb5
KS		 1193 return 1;
1194}
1195
1196/*
1197 * Push the current indentation level to the stack and set the new level
1198 * the current column is greater than the indentation level. In this case,
1199 * append or insert the specified token into the token queue.
1200 *
1201 */
1202
1203static int
1204yaml_parser_roll_indent(yaml_parser_t *parser, int column,
1205 int number, yaml_token_type_t type, yaml_mark_t mark)
1206{
625fcfe9 1207 yaml_token_t token;
eb9cceb5
KS		 1208
1209 /* In the flow context, do nothing. */
1210
1211 if (parser->flow_level)
1212 return 1;
1213
1214 if (parser->indent < column)
1215 {
eb9cceb5
KS		 1216 /*
1217 * Push the current indentation level to the stack and set the new
1218 * indentation level.
1219 */
1220
625fcfe9
KS		 1221 if (!PUSH(parser, parser->indents, parser->indent))
1222 return 0;
1223
eb9cceb5
KS		 1224 parser->indent = column;
1225
625fcfe9 1226 /* Create a token and insert it into the queue. */
eb9cceb5 1227
625fcfe9 1228 TOKEN_INIT(token, type, mark, mark);
eb9cceb5 1229
625fcfe9
KS		 1230 if (number == -1) {
1231 if (!ENQUEUE(parser, parser->tokens, token))
1232 return 0;
1233 }
1234 else {
1235 if (!QUEUE_INSERT(parser,
1236 parser->tokens, number - parser->tokens_parsed, token))
1237 return 0;
eb9cceb5
KS		 1238 }
1239 }
1240
1241 return 1;
1242}
1243
1244/*
1245 * Pop indentation levels from the indents stack until the current level
1246 * becomes less or equal to the column. For each intendation level, append
1247 * the BLOCK-END token.
1248 */
1249
1250
1251static int
1252yaml_parser_unroll_indent(yaml_parser_t *parser, int column)
1253{
625fcfe9 1254 yaml_token_t token;
eb9cceb5
KS		 1255
1256 /* In the flow context, do nothing. */
1257
1258 if (parser->flow_level)
1259 return 1;
1260
1261 /* Loop through the intendation levels in the stack. */
1262
1263 while (parser->indent > column)
1264 {
625fcfe9 1265 /* Create a token and append it to the queue. */
eb9cceb5 1266
625fcfe9 1267 TOKEN_INIT(token, YAML_BLOCK_END_TOKEN, parser->mark, parser->mark);
eb9cceb5 1268
625fcfe9 1269 if (!ENQUEUE(parser, parser->tokens, token))
eb9cceb5 1270 return 0;
eb9cceb5
KS		 1271
1272 /* Pop the indentation level. */
1273
625fcfe9 1274 parser->indent = POP(parser, parser->indents);
eb9cceb5
KS		 1275 }
1276
1277 return 1;
1278}
1279
1280/*
1281 * Initialize the scanner and produce the STREAM-START token.
1282 */
1283
1284static int
1285yaml_parser_fetch_stream_start(yaml_parser_t *parser)
1286{
625fcfe9
KS		 1287 yaml_simple_key_t simple_key = { 0, 0, 0, { 0, 0, 0 } };
1288 yaml_token_t token;
eb9cceb5
KS		 1289
1290 /* Set the initial indentation. */
1291
1292 parser->indent = -1;
1293
625fcfe9
KS		 1294 /* Initialize the simple key stack. */
1295
1296 if (!PUSH(parser, parser->simple_keys, simple_key))
1297 return 0;
1298
eb9cceb5
KS		 1299 /* A simple key is allowed at the beginning of the stream. */
1300
1301 parser->simple_key_allowed = 1;
1302
1303 /* We have started. */
1304
1305 parser->stream_start_produced = 1;
1306
625fcfe9 1307 /* Create the STREAM-START token and append it to the queue. */
eb9cceb5 1308
625fcfe9
KS		 1309 STREAM_START_TOKEN_INIT(token, parser->encoding,
1310 parser->mark, parser->mark);
eb9cceb5 1311
625fcfe9 1312 if (!ENQUEUE(parser, parser->tokens, token))
eb9cceb5 1313 return 0;
eb9cceb5
KS		 1314
1315 return 1;
1316}
1317
1318/*
1319 * Produce the STREAM-END token and shut down the scanner.
1320 */
1321
1322static int
1323yaml_parser_fetch_stream_end(yaml_parser_t *parser)
1324{
625fcfe9 1325 yaml_token_t token;
eb9cceb5
KS		 1326
1327 /* Reset the indentation level. */
1328
1329 if (!yaml_parser_unroll_indent(parser, -1))
1330 return 0;
1331
7e32c194 1332 /* Reset simple keys. */
eb9cceb5 1333
7e32c194
KS		 1334 if (!yaml_parser_remove_simple_key(parser))
1335 return 0;
1336
1337 parser->simple_key_allowed = 0;
eb9cceb5 1338
625fcfe9 1339 /* Create the STREAM-END token and append it to the queue. */
eb9cceb5 1340
625fcfe9 1341 STREAM_END_TOKEN_INIT(token, parser->mark, parser->mark);
eb9cceb5 1342
625fcfe9 1343 if (!ENQUEUE(parser, parser->tokens, token))
eb9cceb5 1344 return 0;
eb9cceb5
KS		 1345
1346 return 1;
1347}
1348
1349/*
625fcfe9 1350 * Produce a VERSION-DIRECTIVE or TAG-DIRECTIVE token.
eb9cceb5
KS		 1351 */
1352
1353static int
1354yaml_parser_fetch_directive(yaml_parser_t *parser)
1355{
625fcfe9 1356 yaml_token_t token;
eb9cceb5
KS		 1357
1358 /* Reset the indentation level. */
1359
1360 if (!yaml_parser_unroll_indent(parser, -1))
1361 return 0;
1362
1363 /* Reset simple keys. */
1364
1365 if (!yaml_parser_remove_simple_key(parser))
1366 return 0;
1367
1368 parser->simple_key_allowed = 0;
1369
1370 /* Create the YAML-DIRECTIVE or TAG-DIRECTIVE token. */
1371
625fcfe9
KS		 1372 if (!yaml_parser_scan_directive(parser, &token))
1373 return 0;
eb9cceb5
KS		 1374
1375 /* Append the token to the queue. */
1376
625fcfe9
KS		 1377 if (!ENQUEUE(parser, parser->tokens, token)) {
1378 yaml_token_delete(&token);
eb9cceb5
KS		 1379 return 0;
1380 }
1381
1382 return 1;
1383}
1384
1385/*
1386 * Produce the DOCUMENT-START or DOCUMENT-END token.
1387 */
1388
1389static int
1390yaml_parser_fetch_document_indicator(yaml_parser_t *parser,
1391 yaml_token_type_t type)
1392{
1393 yaml_mark_t start_mark, end_mark;
625fcfe9 1394 yaml_token_t token;
eb9cceb5
KS		 1395
1396 /* Reset the indentation level. */
1397
1398 if (!yaml_parser_unroll_indent(parser, -1))
1399 return 0;
1400
1401 /* Reset simple keys. */
1402
1403 if (!yaml_parser_remove_simple_key(parser))
1404 return 0;
1405
1406 parser->simple_key_allowed = 0;
1407
1408 /* Consume the token. */
1409
625fcfe9 1410 start_mark = parser->mark;
eb9cceb5 1411
625fcfe9
KS		 1412 SKIP(parser);
1413 SKIP(parser);
1414 SKIP(parser);
eb9cceb5 1415
625fcfe9 1416 end_mark = parser->mark;
eb9cceb5
KS		 1417
1418 /* Create the DOCUMENT-START or DOCUMENT-END token. */
1419
625fcfe9 1420 TOKEN_INIT(token, type, start_mark, end_mark);
eb9cceb5
KS		 1421
1422 /* Append the token to the queue. */
1423
625fcfe9 1424 if (!ENQUEUE(parser, parser->tokens, token))
eb9cceb5 1425 return 0;
eb9cceb5
KS		 1426
1427 return 1;
1428}
1429
1430/*
1431 * Produce the FLOW-SEQUENCE-START or FLOW-MAPPING-START token.
1432 */
1433
1434static int
1435yaml_parser_fetch_flow_collection_start(yaml_parser_t *parser,
1436 yaml_token_type_t type)
1437{
1438 yaml_mark_t start_mark, end_mark;
625fcfe9 1439 yaml_token_t token;
eb9cceb5
KS		 1440
1441 /* The indicators '[' and '{' may start a simple key. */
1442
1443 if (!yaml_parser_save_simple_key(parser))
1444 return 0;
1445
1446 /* Increase the flow level. */
1447
1448 if (!yaml_parser_increase_flow_level(parser))
1449 return 0;
1450
1451 /* A simple key may follow the indicators '[' and '{'. */
1452
1453 parser->simple_key_allowed = 1;
1454
1455 /* Consume the token. */
1456
625fcfe9
KS		 1457 start_mark = parser->mark;
1458 SKIP(parser);
1459 end_mark = parser->mark;
eb9cceb5
KS		 1460
1461 /* Create the FLOW-SEQUENCE-START of FLOW-MAPPING-START token. */
1462
625fcfe9 1463 TOKEN_INIT(token, type, start_mark, end_mark);
eb9cceb5
KS		 1464
1465 /* Append the token to the queue. */
1466
625fcfe9 1467 if (!ENQUEUE(parser, parser->tokens, token))
eb9cceb5 1468 return 0;
eb9cceb5
KS		 1469
1470 return 1;
1471}
1472
1473/*
1474 * Produce the FLOW-SEQUENCE-END or FLOW-MAPPING-END token.
1475 */
1476
1477static int
1478yaml_parser_fetch_flow_collection_end(yaml_parser_t *parser,
1479 yaml_token_type_t type)
1480{
1481 yaml_mark_t start_mark, end_mark;
625fcfe9 1482 yaml_token_t token;
eb9cceb5
KS		 1483
1484 /* Reset any potential simple key on the current flow level. */
1485
1486 if (!yaml_parser_remove_simple_key(parser))
1487 return 0;
1488
1489 /* Decrease the flow level. */
1490
1491 if (!yaml_parser_decrease_flow_level(parser))
1492 return 0;
1493
1494 /* No simple keys after the indicators ']' and '}'. */
1495
1496 parser->simple_key_allowed = 0;
1497
1498 /* Consume the token. */
1499
625fcfe9
KS		 1500 start_mark = parser->mark;
1501 SKIP(parser);
1502 end_mark = parser->mark;
eb9cceb5
KS		 1503
1504 /* Create the FLOW-SEQUENCE-END of FLOW-MAPPING-END token. */
1505
625fcfe9 1506 TOKEN_INIT(token, type, start_mark, end_mark);
eb9cceb5
KS		 1507
1508 /* Append the token to the queue. */
1509
625fcfe9 1510 if (!ENQUEUE(parser, parser->tokens, token))
eb9cceb5 1511 return 0;
eb9cceb5
KS		 1512
1513 return 1;
1514}
1515
1516/*
1517 * Produce the FLOW-ENTRY token.
1518 */
1519
1520static int
1521yaml_parser_fetch_flow_entry(yaml_parser_t *parser)
1522{
1523 yaml_mark_t start_mark, end_mark;
625fcfe9 1524 yaml_token_t token;
eb9cceb5
KS		 1525
1526 /* Reset any potential simple keys on the current flow level. */
1527
1528 if (!yaml_parser_remove_simple_key(parser))
1529 return 0;
1530
1531 /* Simple keys are allowed after ','. */
1532
1533 parser->simple_key_allowed = 1;
1534
1535 /* Consume the token. */
1536
625fcfe9
KS		 1537 start_mark = parser->mark;
1538 SKIP(parser);
1539 end_mark = parser->mark;
eb9cceb5 1540
625fcfe9 1541 /* Create the FLOW-ENTRY token and append it to the queue. */
eb9cceb5 1542
625fcfe9 1543 TOKEN_INIT(token, YAML_FLOW_ENTRY_TOKEN, start_mark, end_mark);
eb9cceb5 1544
625fcfe9 1545 if (!ENQUEUE(parser, parser->tokens, token))
eb9cceb5 1546 return 0;
eb9cceb5
KS		 1547
1548 return 1;
1549}
1550
1551/*
1552 * Produce the BLOCK-ENTRY token.
1553 */
1554
1555static int
1556yaml_parser_fetch_block_entry(yaml_parser_t *parser)
1557{
1558 yaml_mark_t start_mark, end_mark;
625fcfe9 1559 yaml_token_t token;
eb9cceb5
KS		 1560
1561 /* Check if the scanner is in the block context. */
1562
1563 if (!parser->flow_level)
1564 {
1565 /* Check if we are allowed to start a new entry. */
1566
1567 if (!parser->simple_key_allowed) {
625fcfe9 1568 return yaml_parser_set_scanner_error(parser, NULL, parser->mark,
eb9cceb5
KS		 1569 "block sequence entries are not allowed in this context");
1570 }
1571
1572 /* Add the BLOCK-SEQUENCE-START token if needed. */
1573
625fcfe9
KS		 1574 if (!yaml_parser_roll_indent(parser, parser->mark.column, -1,
1575 YAML_BLOCK_SEQUENCE_START_TOKEN, parser->mark))
eb9cceb5
KS		 1576 return 0;
1577 }
1578 else
1579 {
1580 /*
1581 * It is an error for the '-' indicator to occur in the flow context,
1582 * but we let the Parser detect and report about it because the Parser
1583 * is able to point to the context.
1584 */
1585 }
1586
1587 /* Reset any potential simple keys on the current flow level. */
1588
1589 if (!yaml_parser_remove_simple_key(parser))
1590 return 0;
1591
1592 /* Simple keys are allowed after '-'. */
1593
1594 parser->simple_key_allowed = 1;
1595
1596 /* Consume the token. */
1597
625fcfe9
KS		 1598 start_mark = parser->mark;
1599 SKIP(parser);
1600 end_mark = parser->mark;
eb9cceb5 1601
625fcfe9 1602 /* Create the BLOCK-ENTRY token and append it to the queue. */
eb9cceb5 1603
625fcfe9 1604 TOKEN_INIT(token, YAML_BLOCK_ENTRY_TOKEN, start_mark, end_mark);
eb9cceb5 1605
625fcfe9 1606 if (!ENQUEUE(parser, parser->tokens, token))
eb9cceb5 1607 return 0;
eb9cceb5
KS		 1608
1609 return 1;
1610}
1611
1612/*
1613 * Produce the KEY token.
1614 */
1615
1616static int
1617yaml_parser_fetch_key(yaml_parser_t *parser)
1618{
1619 yaml_mark_t start_mark, end_mark;
625fcfe9 1620 yaml_token_t token;
eb9cceb5
KS		 1621
1622 /* In the block context, additional checks are required. */
1623
1624 if (!parser->flow_level)
1625 {
1626 /* Check if we are allowed to start a new key (not nessesary simple). */
1627
1628 if (!parser->simple_key_allowed) {
625fcfe9 1629 return yaml_parser_set_scanner_error(parser, NULL, parser->mark,
eb9cceb5
KS		 1630 "mapping keys are not allowed in this context");
1631 }
1632
1633 /* Add the BLOCK-MAPPING-START token if needed. */
1634
625fcfe9
KS		 1635 if (!yaml_parser_roll_indent(parser, parser->mark.column, -1,
1636 YAML_BLOCK_MAPPING_START_TOKEN, parser->mark))
eb9cceb5
KS		 1637 return 0;
1638 }
1639
1640 /* Reset any potential simple keys on the current flow level. */
1641
1642 if (!yaml_parser_remove_simple_key(parser))
1643 return 0;
1644
1645 /* Simple keys are allowed after '?' in the block context. */
1646
1647 parser->simple_key_allowed = (!parser->flow_level);
1648
1649 /* Consume the token. */
1650
625fcfe9
KS		 1651 start_mark = parser->mark;
1652 SKIP(parser);
1653 end_mark = parser->mark;
eb9cceb5 1654
625fcfe9 1655 /* Create the KEY token and append it to the queue. */
eb9cceb5 1656
625fcfe9 1657 TOKEN_INIT(token, YAML_KEY_TOKEN, start_mark, end_mark);
eb9cceb5 1658
625fcfe9 1659 if (!ENQUEUE(parser, parser->tokens, token))
eb9cceb5 1660 return 0;
eb9cceb5
KS		 1661
1662 return 1;
1663}
1664
1665/*
1666 * Produce the VALUE token.
1667 */
1668
1669static int
1670yaml_parser_fetch_value(yaml_parser_t *parser)
1671{
1672 yaml_mark_t start_mark, end_mark;
625fcfe9
KS		 1673 yaml_token_t token;
1674 yaml_simple_key_t *simple_key = parser->simple_keys.top-1;
eb9cceb5
KS		 1675
1676 /* Have we found a simple key? */
1677
625fcfe9 1678 if (simple_key->possible)
eb9cceb5 1679 {
eb9cceb5 1680
625fcfe9 1681 /* Create the KEY token and insert it into the queue. */
eb9cceb5 1682
625fcfe9 1683 TOKEN_INIT(token, YAML_KEY_TOKEN, simple_key->mark, simple_key->mark);
eb9cceb5 1684
625fcfe9
KS		 1685 if (!QUEUE_INSERT(parser, parser->tokens,
1686 simple_key->token_number - parser->tokens_parsed, token))
eb9cceb5 1687 return 0;
eb9cceb5
KS		 1688
1689 /* In the block context, we may need to add the BLOCK-MAPPING-START token. */
1690
625fcfe9 1691 if (!yaml_parser_roll_indent(parser, simple_key->mark.column,
eb9cceb5
KS		 1692 simple_key->token_number,
1693 YAML_BLOCK_MAPPING_START_TOKEN, simple_key->mark))
1694 return 0;
1695
625fcfe9 1696 /* Remove the simple key. */
eb9cceb5 1697
625fcfe9 1698 simple_key->possible = 0;
eb9cceb5
KS		 1699
1700 /* A simple key cannot follow another simple key. */
1701
1702 parser->simple_key_allowed = 0;
1703 }
1704 else
1705 {
1706 /* The ':' indicator follows a complex key. */
1707
1708 /* In the block context, extra checks are required. */
1709
1710 if (!parser->flow_level)
1711 {
1712 /* Check if we are allowed to start a complex value. */
1713
1714 if (!parser->simple_key_allowed) {
625fcfe9 1715 return yaml_parser_set_scanner_error(parser, NULL, parser->mark,
eb9cceb5
KS		 1716 "mapping values are not allowed in this context");
1717 }
1718
1719 /* Add the BLOCK-MAPPING-START token if needed. */
1720
625fcfe9
KS		 1721 if (!yaml_parser_roll_indent(parser, parser->mark.column, -1,
1722 YAML_BLOCK_MAPPING_START_TOKEN, parser->mark))
eb9cceb5
KS		 1723 return 0;
1724 }
1725
eb9cceb5
KS		 1726 /* Simple keys after ':' are allowed in the block context. */
1727
1728 parser->simple_key_allowed = (!parser->flow_level);
1729 }
1730
1731 /* Consume the token. */
1732
625fcfe9
KS		 1733 start_mark = parser->mark;
1734 SKIP(parser);
1735 end_mark = parser->mark;
eb9cceb5 1736
625fcfe9 1737 /* Create the VALUE token and append it to the queue. */
eb9cceb5 1738
625fcfe9 1739 TOKEN_INIT(token, YAML_VALUE_TOKEN, start_mark, end_mark);
eb9cceb5 1740
625fcfe9 1741 if (!ENQUEUE(parser, parser->tokens, token))
eb9cceb5 1742 return 0;
eb9cceb5
KS		 1743
1744 return 1;
1745}
1746
1747/*
1748 * Produce the ALIAS or ANCHOR token.
1749 */
1750
1751static int
1752yaml_parser_fetch_anchor(yaml_parser_t *parser, yaml_token_type_t type)
1753{
625fcfe9 1754 yaml_token_t token;
eb9cceb5
KS		 1755
1756 /* An anchor or an alias could be a simple key. */
1757
1758 if (!yaml_parser_save_simple_key(parser))
1759 return 0;
1760
1761 /* A simple key cannot follow an anchor or an alias. */
1762
1763 parser->simple_key_allowed = 0;
1764
625fcfe9 1765 /* Create the ALIAS or ANCHOR token and append it to the queue. */
eb9cceb5 1766
625fcfe9
KS		 1767 if (!yaml_parser_scan_anchor(parser, &token, type))
1768 return 0;
eb9cceb5 1769
625fcfe9
KS		 1770 if (!ENQUEUE(parser, parser->tokens, token)) {
1771 yaml_token_delete(&token);
eb9cceb5
KS		 1772 return 0;
1773 }
eb9cceb5
KS		 1774 return 1;
1775}
1776
1777/*
1778 * Produce the TAG token.
1779 */
1780
1781static int
1782yaml_parser_fetch_tag(yaml_parser_t *parser)
1783{
625fcfe9 1784 yaml_token_t token;
eb9cceb5
KS		 1785
1786 /* A tag could be a simple key. */
1787
1788 if (!yaml_parser_save_simple_key(parser))
1789 return 0;
1790
1791 /* A simple key cannot follow a tag. */
1792
1793 parser->simple_key_allowed = 0;
1794
625fcfe9 1795 /* Create the TAG token and append it to the queue. */
eb9cceb5 1796
625fcfe9
KS		 1797 if (!yaml_parser_scan_tag(parser, &token))
1798 return 0;
eb9cceb5 1799
625fcfe9
KS		 1800 if (!ENQUEUE(parser, parser->tokens, token)) {
1801 yaml_token_delete(&token);
eb9cceb5
KS		 1802 return 0;
1803 }
1804
1805 return 1;
1806}
1807
1808/*
1809 * Produce the SCALAR(...,literal) or SCALAR(...,folded) tokens.
1810 */
1811
1812static int
1813yaml_parser_fetch_block_scalar(yaml_parser_t *parser, int literal)
1814{
625fcfe9 1815 yaml_token_t token;
eb9cceb5
KS		 1816
1817 /* Remove any potential simple keys. */
1818
1819 if (!yaml_parser_remove_simple_key(parser))
1820 return 0;
1821
1822 /* A simple key may follow a block scalar. */
1823
1824 parser->simple_key_allowed = 1;
1825
625fcfe9 1826 /* Create the SCALAR token and append it to the queue. */
eb9cceb5 1827
625fcfe9
KS		 1828 if (!yaml_parser_scan_block_scalar(parser, &token, literal))
1829 return 0;
eb9cceb5 1830
625fcfe9
KS		 1831 if (!ENQUEUE(parser, parser->tokens, token)) {
1832 yaml_token_delete(&token);
eb9cceb5
KS		 1833 return 0;
1834 }
1835
1836 return 1;
1837}
1838
1839/*
1840 * Produce the SCALAR(...,single-quoted) or SCALAR(...,double-quoted) tokens.
1841 */
1842
1843static int
1844yaml_parser_fetch_flow_scalar(yaml_parser_t *parser, int single)
1845{
625fcfe9 1846 yaml_token_t token;
eb9cceb5
KS		 1847
1848 /* A plain scalar could be a simple key. */
1849
1850 if (!yaml_parser_save_simple_key(parser))
1851 return 0;
1852
1853 /* A simple key cannot follow a flow scalar. */
1854
1855 parser->simple_key_allowed = 0;
1856
625fcfe9 1857 /* Create the SCALAR token and append it to the queue. */
eb9cceb5 1858
625fcfe9
KS		 1859 if (!yaml_parser_scan_flow_scalar(parser, &token, single))
1860 return 0;
eb9cceb5 1861
625fcfe9
KS		 1862 if (!ENQUEUE(parser, parser->tokens, token)) {
1863 yaml_token_delete(&token);
eb9cceb5
KS		 1864 return 0;
1865 }
1866
1867 return 1;
1868}
1869
1870/*
1871 * Produce the SCALAR(...,plain) token.
1872 */
1873
1874static int
1875yaml_parser_fetch_plain_scalar(yaml_parser_t *parser)
1876{
625fcfe9 1877 yaml_token_t token;
eb9cceb5
KS		 1878
1879 /* A plain scalar could be a simple key. */
1880
1881 if (!yaml_parser_save_simple_key(parser))
1882 return 0;
1883
1884 /* A simple key cannot follow a flow scalar. */
1885
1886 parser->simple_key_allowed = 0;
1887
625fcfe9 1888 /* Create the SCALAR token and append it to the queue. */
eb9cceb5 1889
625fcfe9
KS		 1890 if (!yaml_parser_scan_plain_scalar(parser, &token))
1891 return 0;
eb9cceb5 1892
625fcfe9
KS		 1893 if (!ENQUEUE(parser, parser->tokens, token)) {
1894 yaml_token_delete(&token);
eb9cceb5
KS		 1895 return 0;
1896 }
1897
1898 return 1;
1899}
1900
e71095e3
KS		 1901/*
1902 * Eat whitespaces and comments until the next token is found.
1903 */
1904
1905static int
1906yaml_parser_scan_to_next_token(yaml_parser_t *parser)
1907{
1908 /* Until the next token is not found. */
1909
1910 while (1)
1911 {
1912 /* Allow the BOM mark to start a line. */
1913
625fcfe9 1914 if (!CACHE(parser, 1)) return 0;
e71095e3 1915
e35af832 1916 if (parser->mark.column == 0 && IS_BOM(parser->buffer))
625fcfe9 1917 SKIP(parser);
e71095e3
KS		 1918
1919 /*
1920 * Eat whitespaces.
1921 *
1922 * Tabs are allowed:
1923 *
1924 * - in the flow context;
1925 * - in the block context, but not at the beginning of the line or
1926 * after '-', '?', or ':' (complex value).
1927 */
1928
625fcfe9 1929 if (!CACHE(parser, 1)) return 0;
e71095e3 1930
e35af832 1931 while (CHECK(parser->buffer,' ') ||
e71095e3 1932 ((parser->flow_level || !parser->simple_key_allowed) &&
e35af832 1933 CHECK(parser->buffer, '\t'))) {
625fcfe9
KS		 1934 SKIP(parser);
1935 if (!CACHE(parser, 1)) return 0;
e71095e3
KS		 1936 }
1937
1938 /* Eat a comment until a line break. */
1939
e35af832
KS		 1940 if (CHECK(parser->buffer, '#')) {
1941 while (!IS_BREAKZ(parser->buffer)) {
625fcfe9
KS		 1942 SKIP(parser);
1943 if (!CACHE(parser, 1)) return 0;
e71095e3
KS		 1944 }
1945 }
1946
1947 /* If it is a line break, eat it. */
1948
e35af832 1949 if (IS_BREAK(parser->buffer))
e71095e3 1950 {
625fcfe9
KS		 1951 if (!CACHE(parser, 2)) return 0;
1952 SKIP_LINE(parser);
e71095e3
KS		 1953
1954 /* In the block context, a new line may start a simple key. */
1955
1956 if (!parser->flow_level) {
1957 parser->simple_key_allowed = 1;
1958 }
1959 }
1960 else
1961 {
1962 /* We have found a token. */
1963
1964 break;
1965 }
1966 }
1967
1968 return 1;
1969}
1970
1971/*
1972 * Scan a YAML-DIRECTIVE or TAG-DIRECTIVE token.
1973 *
1974 * Scope:
1975 * %YAML 1.1 # a comment \n
1976 * ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
1977 * %TAG !yaml! tag:yaml.org,2002: \n
1978 * ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
1979 */
1980
625fcfe9
KS		 1981int
1982yaml_parser_scan_directive(yaml_parser_t *parser, yaml_token_t *token)
e71095e3
KS		 1983{
1984 yaml_mark_t start_mark, end_mark;
1985 yaml_char_t *name = NULL;
1986 int major, minor;
1987 yaml_char_t *handle = NULL, *prefix = NULL;
e71095e3
KS		 1988
1989 /* Eat '%'. */
1990
625fcfe9 1991 start_mark = parser->mark;
e71095e3 1992
625fcfe9 1993 SKIP(parser);
e71095e3
KS		 1994
1995 /* Scan the directive name. */
1996
1997 if (!yaml_parser_scan_directive_name(parser, start_mark, &name))
1998 goto error;
1999
2000 /* Is it a YAML directive? */
2001
2002 if (strcmp((char *)name, "YAML") == 0)
2003 {
2004 /* Scan the VERSION directive value. */
2005
2006 if (!yaml_parser_scan_version_directive_value(parser, start_mark,
2007 &major, &minor))
2008 goto error;
2009
625fcfe9 2010 end_mark = parser->mark;
e71095e3
KS		 2011
2012 /* Create a VERSION-DIRECTIVE token. */
2013
625fcfe9 2014 VERSION_DIRECTIVE_TOKEN_INIT(*token, major, minor,
e71095e3 2015 start_mark, end_mark);
e71095e3
KS		 2016 }
2017
2018 /* Is it a TAG directive? */
2019
2020 else if (strcmp((char *)name, "TAG") == 0)
2021 {
2022 /* Scan the TAG directive value. */
2023
2024 if (!yaml_parser_scan_tag_directive_value(parser, start_mark,
2025 &handle, &prefix))
2026 goto error;
2027
625fcfe9 2028 end_mark = parser->mark;
e71095e3
KS		 2029
2030 /* Create a TAG-DIRECTIVE token. */
2031
625fcfe9 2032 TAG_DIRECTIVE_TOKEN_INIT(*token, handle, prefix,
e71095e3 2033 start_mark, end_mark);
e71095e3
KS		 2034 }
2035
2036 /* Unknown directive. */
2037
2038 else
2039 {
92d41fe1 2040 yaml_parser_set_scanner_error(parser, "while scanning a directive",
e71095e3
KS		 2041 start_mark, "found uknown directive name");
2042 goto error;
2043 }
2044
2045 /* Eat the rest of the line including any comments. */
2046
625fcfe9
KS		 2047 if (!CACHE(parser, 1)) goto error;
2048
e35af832 2049 while (IS_BLANK(parser->buffer)) {
625fcfe9
KS		 2050 SKIP(parser);
2051 if (!CACHE(parser, 1)) goto error;
e71095e3
KS		 2052 }
2053
e35af832
KS		 2054 if (CHECK(parser->buffer, '#')) {
2055 while (!IS_BREAKZ(parser->buffer)) {
625fcfe9
KS		 2056 SKIP(parser);
2057 if (!CACHE(parser, 1)) goto error;
e71095e3
KS		 2058 }
2059 }
2060
2061 /* Check if we are at the end of the line. */
2062
e35af832 2063 if (!IS_BREAKZ(parser->buffer)) {
92d41fe1 2064 yaml_parser_set_scanner_error(parser, "while scanning a directive",
e71095e3
KS		 2065 start_mark, "did not found expected comment or line break");
2066 goto error;
2067 }
2068
2069 /* Eat a line break. */
2070
e35af832 2071 if (IS_BREAK(parser->buffer)) {
625fcfe9
KS		 2072 if (!CACHE(parser, 2)) goto error;
2073 SKIP_LINE(parser);
e71095e3
KS		 2074 }
2075
2076 yaml_free(name);
2077
625fcfe9 2078 return 1;
e71095e3
KS		 2079
2080error:
e71095e3
KS		 2081 yaml_free(prefix);
2082 yaml_free(handle);
2083 yaml_free(name);
625fcfe9 2084 return 0;
e71095e3
KS		 2085}
2086
2087/*
2088 * Scan the directive name.
2089 *
2090 * Scope:
2091 * %YAML 1.1 # a comment \n
2092 * ^^^^
2093 * %TAG !yaml! tag:yaml.org,2002: \n
2094 * ^^^
2095 */
2096
2097static int
2098yaml_parser_scan_directive_name(yaml_parser_t *parser,
2099 yaml_mark_t start_mark, yaml_char_t **name)
2100{
625fcfe9 2101 yaml_string_t string = NULL_STRING;
e71095e3 2102
625fcfe9 2103 if (!STRING_INIT(parser, string, INITIAL_STRING_SIZE)) goto error;
e71095e3
KS		 2104
2105 /* Consume the directive name. */
2106
625fcfe9 2107 if (!CACHE(parser, 1)) goto error;
e71095e3 2108
e35af832 2109 while (IS_ALPHA(parser->buffer))
e71095e3 2110 {
625fcfe9
KS		 2111 if (!READ(parser, string)) goto error;
2112 if (!CACHE(parser, 1)) goto error;
e71095e3
KS		 2113 }
2114
2115 /* Check if the name is empty. */
2116
625fcfe9 2117 if (string.start == string.pointer) {
e71095e3
KS		 2118 yaml_parser_set_scanner_error(parser, "while scanning a directive",
2119 start_mark, "cannot found expected directive name");
2120 goto error;
2121 }
2122
2123 /* Check for an blank character after the name. */
2124
e35af832 2125 if (!IS_BLANKZ(parser->buffer)) {
e71095e3
KS		 2126 yaml_parser_set_scanner_error(parser, "while scanning a directive",
2127 start_mark, "found unexpected non-alphabetical character");
2128 goto error;
2129 }
2130
625fcfe9 2131 *name = string.start;
e71095e3
KS		 2132
2133 return 1;
2134
2135error:
625fcfe9 2136 STRING_DEL(parser, string);
e71095e3
KS		 2137 return 0;
2138}
2139
2140/*
2141 * Scan the value of VERSION-DIRECTIVE.
2142 *
2143 * Scope:
2144 * %YAML 1.1 # a comment \n
2145 * ^^^^^^
2146 */
2147
2148static int
2149yaml_parser_scan_version_directive_value(yaml_parser_t *parser,
2150 yaml_mark_t start_mark, int *major, int *minor)
2151{
2152 /* Eat whitespaces. */
2153
625fcfe9 2154 if (!CACHE(parser, 1)) return 0;
e71095e3 2155
e35af832 2156 while (IS_BLANK(parser->buffer)) {
625fcfe9
KS		 2157 SKIP(parser);
2158 if (!CACHE(parser, 1)) return 0;
e71095e3
KS		 2159 }
2160
2161 /* Consume the major version number. */
2162
2163 if (!yaml_parser_scan_version_directive_number(parser, start_mark, major))
2164 return 0;
2165
2166 /* Eat '.'. */
2167
e35af832 2168 if (!CHECK(parser->buffer, '.')) {
e71095e3
KS		 2169 return yaml_parser_set_scanner_error(parser, "while scanning a %YAML directive",
2170 start_mark, "did not find expected digit or '.' character");
2171 }
2172
625fcfe9 2173 SKIP(parser);
e71095e3
KS		 2174
2175 /* Consume the minor version number. */
2176
2177 if (!yaml_parser_scan_version_directive_number(parser, start_mark, minor))
2178 return 0;
ab01bac8
KS		 2179
2180 return 1;
e71095e3
KS		 2181}
2182
2183#define MAX_NUMBER_LENGTH 9
2184
2185/*
2186 * Scan the version number of VERSION-DIRECTIVE.
2187 *
2188 * Scope:
2189 * %YAML 1.1 # a comment \n
2190 * ^
2191 * %YAML 1.1 # a comment \n
2192 * ^
2193 */
2194
2195static int
2196yaml_parser_scan_version_directive_number(yaml_parser_t *parser,
2197 yaml_mark_t start_mark, int *number)
2198{
2199 int value = 0;
2200 size_t length = 0;
2201
2202 /* Repeat while the next character is digit. */
2203
625fcfe9 2204 if (!CACHE(parser, 1)) return 0;
e71095e3 2205
e35af832 2206 while (IS_DIGIT(parser->buffer))
e71095e3
KS		 2207 {
2208 /* Check if the number is too long. */
2209
2210 if (++length > MAX_NUMBER_LENGTH) {
2211 return yaml_parser_set_scanner_error(parser, "while scanning a %YAML directive",
2212 start_mark, "found extremely long version number");
2213 }
2214
e35af832 2215 value = value*10 + AS_DIGIT(parser->buffer);
e71095e3 2216
625fcfe9 2217 SKIP(parser);
e71095e3 2218
625fcfe9 2219 if (!CACHE(parser, 1)) return 0;
e71095e3
KS		 2220 }
2221
2222 /* Check if the number was present. */
2223
2224 if (!length) {
2225 return yaml_parser_set_scanner_error(parser, "while scanning a %YAML directive",
2226 start_mark, "did not find expected version number");
2227 }
2228
2229 *number = value;
2230
2231 return 1;
2232}
2233
2234/*
2235 * Scan the value of a TAG-DIRECTIVE token.
2236 *
2237 * Scope:
2238 * %TAG !yaml! tag:yaml.org,2002: \n
2239 * ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
2240 */
2241
2242static int
2243yaml_parser_scan_tag_directive_value(yaml_parser_t *parser,
2244 yaml_mark_t start_mark, yaml_char_t **handle, yaml_char_t **prefix)
2245{
2246 yaml_char_t *handle_value = NULL;
2247 yaml_char_t *prefix_value = NULL;
2248
2249 /* Eat whitespaces. */
2250
625fcfe9 2251 if (!CACHE(parser, 1)) goto error;
e71095e3 2252
e35af832 2253 while (IS_BLANK(parser->buffer)) {
625fcfe9
KS		 2254 SKIP(parser);
2255 if (!CACHE(parser, 1)) goto error;
e71095e3
KS		 2256 }
2257
2258 /* Scan a handle. */
2259
2260 if (!yaml_parser_scan_tag_handle(parser, 1, start_mark, &handle_value))
2261 goto error;
2262
2263 /* Expect a whitespace. */
2264
625fcfe9 2265 if (!CACHE(parser, 1)) goto error;
e71095e3 2266
e35af832 2267 if (!IS_BLANK(parser->buffer)) {
e71095e3
KS		 2268 yaml_parser_set_scanner_error(parser, "while scanning a %TAG directive",
2269 start_mark, "did not find expected whitespace");
2270 goto error;
2271 }
2272
2273 /* Eat whitespaces. */
2274
e35af832 2275 while (IS_BLANK(parser->buffer)) {
625fcfe9
KS		 2276 SKIP(parser);
2277 if (!CACHE(parser, 1)) goto error;
e71095e3
KS		 2278 }
2279
2280 /* Scan a prefix. */
2281
2282 if (!yaml_parser_scan_tag_uri(parser, 1, NULL, start_mark, &prefix_value))
2283 goto error;
2284
2285 /* Expect a whitespace or line break. */
2286
625fcfe9 2287 if (!CACHE(parser, 1)) goto error;
e71095e3 2288
e35af832 2289 if (!IS_BLANKZ(parser->buffer)) {
e71095e3
KS		 2290 yaml_parser_set_scanner_error(parser, "while scanning a %TAG directive",
2291 start_mark, "did not find expected whitespace or line break");
2292 goto error;
2293 }
2294
2295 *handle = handle_value;
2296 *prefix = prefix_value;
2297
2298 return 1;
2299
2300error:
2301 yaml_free(handle_value);
2302 yaml_free(prefix_value);
2303 return 0;
2304}
2305
625fcfe9
KS		 2306static int
2307yaml_parser_scan_anchor(yaml_parser_t *parser, yaml_token_t *token,
e71095e3
KS		 2308 yaml_token_type_t type)
2309{
2310 int length = 0;
2311 yaml_mark_t start_mark, end_mark;
625fcfe9 2312 yaml_string_t string = NULL_STRING;
e71095e3 2313
625fcfe9 2314 if (!STRING_INIT(parser, string, INITIAL_STRING_SIZE)) goto error;
e71095e3
KS		 2315
2316 /* Eat the indicator character. */
2317
625fcfe9 2318 start_mark = parser->mark;
e71095e3 2319
625fcfe9 2320 SKIP(parser);
e71095e3
KS		 2321
2322 /* Consume the value. */
2323
625fcfe9 2324 if (!CACHE(parser, 1)) goto error;
e71095e3 2325
e35af832 2326 while (IS_ALPHA(parser->buffer)) {
625fcfe9
KS		 2327 if (!READ(parser, string)) goto error;
2328 if (!CACHE(parser, 1)) goto error;
e71095e3
KS		 2329 length ++;
2330 }
2331
625fcfe9 2332 end_mark = parser->mark;
e71095e3
KS		 2333
2334 /*
2335 * Check if length of the anchor is greater than 0 and it is followed by
2336 * a whitespace character or one of the indicators:
2337 *
2338 * '?', ':', ',', ']', '}', '%', '@', '`'.
2339 */
2340
e35af832
KS		 2341 if (!length || !(IS_BLANKZ(parser->buffer) || CHECK(parser->buffer, '?')
2342 || CHECK(parser->buffer, ':') || CHECK(parser->buffer, ',')
2343 || CHECK(parser->buffer, ']') || CHECK(parser->buffer, '}')
2344 || CHECK(parser->buffer, '%') || CHECK(parser->buffer, '@')
2345 || CHECK(parser->buffer, '`'))) {
e71095e3
KS		 2346 yaml_parser_set_scanner_error(parser, type == YAML_ANCHOR_TOKEN ?
2347 "while scanning an anchor" : "while scanning an alias", start_mark,
2348 "did not find expected alphabetic or numeric character");
2349 goto error;
2350 }
2351
2352 /* Create a token. */
2353
625fcfe9
KS		 2354 if (type == YAML_ANCHOR_TOKEN) {
2355 ANCHOR_TOKEN_INIT(*token, string.start, start_mark, end_mark);
2356 }
2357 else {
2358 ALIAS_TOKEN_INIT(*token, string.start, start_mark, end_mark);
92d41fe1 2359 }
e71095e3 2360
625fcfe9 2361 return 1;
e71095e3
KS		 2362
2363error:
625fcfe9 2364 STRING_DEL(parser, string);
e71095e3
KS		 2365 return 0;
2366}
2367
2368/*
2369 * Scan a TAG token.
2370 */
2371
625fcfe9
KS		 2372static int
2373yaml_parser_scan_tag(yaml_parser_t *parser, yaml_token_t *token)
e71095e3
KS		 2374{
2375 yaml_char_t *handle = NULL;
2376 yaml_char_t *suffix = NULL;
e71095e3
KS		 2377 yaml_mark_t start_mark, end_mark;
2378
625fcfe9 2379 start_mark = parser->mark;
e71095e3
KS		 2380
2381 /* Check if the tag is in the canonical form. */
2382
625fcfe9 2383 if (!CACHE(parser, 2)) goto error;
e71095e3 2384
e35af832 2385 if (CHECK_AT(parser->buffer, '<', 1))
e71095e3
KS		 2386 {
2387 /* Set the handle to '' */
2388
2389 handle = yaml_malloc(1);
2390 if (!handle) goto error;
2391 handle[0] = '\0';
2392
2393 /* Eat '!<' */
2394
625fcfe9
KS		 2395 SKIP(parser);
2396 SKIP(parser);
e71095e3
KS		 2397
2398 /* Consume the tag value. */
2399
2400 if (!yaml_parser_scan_tag_uri(parser, 0, NULL, start_mark, &suffix))
2401 goto error;
2402
2403 /* Check for '>' and eat it. */
2404
e35af832 2405 if (!CHECK(parser->buffer, '>')) {
e71095e3
KS		 2406 yaml_parser_set_scanner_error(parser, "while scanning a tag",
2407 start_mark, "did not find the expected '>'");
2408 goto error;
2409 }
2410
625fcfe9 2411 SKIP(parser);
e71095e3
KS		 2412 }
2413 else
2414 {
2415 /* The tag has either the '!suffix' or the '!handle!suffix' form. */
2416
2417 /* First, try to scan a handle. */
2418
2419 if (!yaml_parser_scan_tag_handle(parser, 0, start_mark, &handle))
2420 goto error;
2421
2422 /* Check if it is, indeed, handle. */
2423
2424 if (handle[0] == '!' && handle[1] != '\0' && handle[strlen((char *)handle)-1] == '!')
2425 {
2426 /* Scan the suffix now. */
2427
2428 if (!yaml_parser_scan_tag_uri(parser, 0, NULL, start_mark, &suffix))
2429 goto error;
2430 }
2431 else
2432 {
2433 /* It wasn't a handle after all. Scan the rest of the tag. */
2434
2435 if (!yaml_parser_scan_tag_uri(parser, 0, handle, start_mark, &suffix))
2436 goto error;
2437
2438 /* Set the handle to '!'. */
2439
2440 yaml_free(handle);
2441 handle = yaml_malloc(2);
2442 if (!handle) goto error;
2443 handle[0] = '!';
2444 handle[1] = '\0';
7e32c194
KS		 2445
2446 /*
625fcfe9
KS		 2447 * A special case: the '!' tag. Set the handle to '' and the
2448 * suffix to '!'.
7e32c194
KS		 2449 */
2450
2451 if (suffix[0] == '\0') {
2452 yaml_char_t *tmp = handle;
2453 handle = suffix;
2454 suffix = tmp;
2455 }
e71095e3
KS		 2456 }
2457 }
2458
2459 /* Check the character which ends the tag. */
2460
625fcfe9 2461 if (!CACHE(parser, 1)) goto error;
e71095e3 2462
e35af832 2463 if (!IS_BLANKZ(parser->buffer)) {
e71095e3
KS		 2464 yaml_parser_set_scanner_error(parser, "while scanning a tag",
2465 start_mark, "did not found expected whitespace or line break");
2466 goto error;
2467 }
2468
625fcfe9 2469 end_mark = parser->mark;
e71095e3
KS		 2470
2471 /* Create a token. */
2472
625fcfe9 2473 TAG_TOKEN_INIT(*token, handle, suffix, start_mark, end_mark);
e71095e3 2474
625fcfe9 2475 return 1;
e71095e3
KS		 2476
2477error:
2478 yaml_free(handle);
2479 yaml_free(suffix);
625fcfe9 2480 return 0;
e71095e3
KS		 2481}
2482
2483/*
2484 * Scan a tag handle.
2485 */
2486
2487static int
2488yaml_parser_scan_tag_handle(yaml_parser_t *parser, int directive,
2489 yaml_mark_t start_mark, yaml_char_t **handle)
2490{
625fcfe9 2491 yaml_string_t string = NULL_STRING;
e71095e3 2492
625fcfe9 2493 if (!STRING_INIT(parser, string, INITIAL_STRING_SIZE)) goto error;
e71095e3
KS		 2494
2495 /* Check the initial '!' character. */
2496
625fcfe9 2497 if (!CACHE(parser, 1)) goto error;
e71095e3 2498
e35af832 2499 if (!CHECK(parser->buffer, '!')) {
e71095e3
KS		 2500 yaml_parser_set_scanner_error(parser, directive ?
2501 "while scanning a tag directive" : "while scanning a tag",
2502 start_mark, "did not find expected '!'");
2503 goto error;
2504 }
2505
2506 /* Copy the '!' character. */
2507
625fcfe9 2508 if (!READ(parser, string)) goto error;
e71095e3
KS		 2509
2510 /* Copy all subsequent alphabetical and numerical characters. */
2511
625fcfe9 2512 if (!CACHE(parser, 1)) goto error;
e71095e3 2513
e35af832 2514 while (IS_ALPHA(parser->buffer))
e71095e3 2515 {
625fcfe9
KS		 2516 if (!READ(parser, string)) goto error;
2517 if (!CACHE(parser, 1)) goto error;
e71095e3
KS		 2518 }
2519
2520 /* Check if the trailing character is '!' and copy it. */
2521
e35af832 2522 if (CHECK(parser->buffer, '!'))
e71095e3 2523 {
625fcfe9 2524 if (!READ(parser, string)) goto error;
e71095e3
KS		 2525 }
2526 else
2527 {
2528 /*
7e32c194
KS		 2529 * It's either the '!' tag or not really a tag handle. If it's a %TAG
2530 * directive, it's an error. If it's a tag token, it must be a part of
2531 * URI.
e71095e3
KS		 2532 */
2533
625fcfe9 2534 if (directive && !(string.start[0] == '!' && string.start[1] == '\0')) {
7e32c194 2535 yaml_parser_set_scanner_error(parser, "while parsing a tag directive",
e71095e3
KS		 2536 start_mark, "did not find expected '!'");
2537 goto error;
2538 }
2539 }
2540
625fcfe9 2541 *handle = string.start;
e71095e3
KS		 2542
2543 return 1;
2544
2545error:
625fcfe9 2546 STRING_DEL(parser, string);
e71095e3
KS		 2547 return 0;
2548}
2549
2550/*
2551 * Scan a tag.
2552 */
2553
2554static int
2555yaml_parser_scan_tag_uri(yaml_parser_t *parser, int directive,
2556 yaml_char_t *head, yaml_mark_t start_mark, yaml_char_t **uri)
2557{
2558 size_t length = head ? strlen((char *)head) : 0;
625fcfe9 2559 yaml_string_t string = NULL_STRING;
e71095e3 2560
625fcfe9 2561 if (!STRING_INIT(parser, string, INITIAL_STRING_SIZE)) goto error;
e71095e3
KS		 2562
2563 /* Resize the string to include the head. */
2564
625fcfe9
KS		 2565 while (string.end - string.start <= length) {
2566 if (!yaml_string_extend(&string.start, &string.pointer, &string.end)) {
2567 parser->error = YAML_MEMORY_ERROR;
2568 goto error;
2569 }
e71095e3
KS		 2570 }
2571
7e32c194
KS		 2572 /*
2573 * Copy the head if needed.
2574 *
2575 * Note that we don't copy the leading '!' character.
2576 */
e71095e3 2577
7e32c194 2578 if (length > 1) {
625fcfe9 2579 memcpy(string.start, head+1, length-1);
7e32c194 2580 string.pointer += length-1;
e71095e3
KS		 2581 }
2582
2583 /* Scan the tag. */
2584
625fcfe9 2585 if (!CACHE(parser, 1)) goto error;
e71095e3
KS		 2586
2587 /*
2588 * The set of characters that may appear in URI is as follows:
2589 *
2590 * '0'-'9', 'A'-'Z', 'a'-'z', '_', '-', ';', '/', '?', ':', '@', '&',
2591 * '=', '+', '$', ',', '.', '!', '~', '*', '\'', '(', ')', '[', ']',
2592 * '%'.
2593 */
2594
e35af832
KS		 2595 while (IS_ALPHA(parser->buffer) || CHECK(parser->buffer, ';')
2596 || CHECK(parser->buffer, '/') || CHECK(parser->buffer, '?')
2597 || CHECK(parser->buffer, ':') || CHECK(parser->buffer, '@')
2598 || CHECK(parser->buffer, '&') || CHECK(parser->buffer, '=')
2599 || CHECK(parser->buffer, '+') || CHECK(parser->buffer, '$')
2600 || CHECK(parser->buffer, ',') || CHECK(parser->buffer, '.')
2601 || CHECK(parser->buffer, '!') || CHECK(parser->buffer, '~')
2602 || CHECK(parser->buffer, '*') || CHECK(parser->buffer, '\'')
2603 || CHECK(parser->buffer, '(') || CHECK(parser->buffer, ')')
2604 || CHECK(parser->buffer, '[') || CHECK(parser->buffer, ']')
2605 || CHECK(parser->buffer, '%'))
e71095e3 2606 {
e71095e3
KS		 2607 /* Check if it is a URI-escape sequence. */
2608
e35af832 2609 if (CHECK(parser->buffer, '%')) {
e71095e3
KS		 2610 if (!yaml_parser_scan_uri_escapes(parser,
2611 directive, start_mark, &string)) goto error;
2612 }
2613 else {
625fcfe9 2614 if (!READ(parser, string)) goto error;
e71095e3
KS		 2615 }
2616
2617 length ++;
625fcfe9 2618 if (!CACHE(parser, 1)) goto error;
e71095e3
KS		 2619 }
2620
2621 /* Check if the tag is non-empty. */
2622
2623 if (!length) {
625fcfe9
KS		 2624 if (!STRING_EXTEND(parser, string))
2625 goto error;
2626
e71095e3
KS		 2627 yaml_parser_set_scanner_error(parser, directive ?
2628 "while parsing a %TAG directive" : "while parsing a tag",
2629 start_mark, "did not find expected tag URI");
2630 goto error;
2631 }
2632
625fcfe9 2633 *uri = string.start;
e71095e3
KS		 2634
2635 return 1;
2636
2637error:
625fcfe9 2638 STRING_DEL(parser, string);
e71095e3
KS		 2639 return 0;
2640}
2641
2642/*
2643 * Decode an URI-escape sequence corresponding to a single UTF-8 character.
2644 */
2645
2646static int
2647yaml_parser_scan_uri_escapes(yaml_parser_t *parser, int directive,
2648 yaml_mark_t start_mark, yaml_string_t *string)
2649{
2650 int width = 0;
2651
2652 /* Decode the required number of characters. */
2653
2654 do {
2655
2656 unsigned char octet = 0;
2657
2658 /* Check for a URI-escaped octet. */
2659
625fcfe9 2660 if (!CACHE(parser, 3)) return 0;
e71095e3 2661
e35af832
KS		 2662 if (!(CHECK(parser->buffer, '%')
2663 && IS_HEX_AT(parser->buffer, 1)
2664 && IS_HEX_AT(parser->buffer, 2))) {
e71095e3
KS		 2665 return yaml_parser_set_scanner_error(parser, directive ?
2666 "while parsing a %TAG directive" : "while parsing a tag",
2667 start_mark, "did not find URI escaped octet");
2668 }
2669
2670 /* Get the octet. */
2671
e35af832 2672 octet = (AS_HEX_AT(parser->buffer, 1) << 4) + AS_HEX_AT(parser->buffer, 2);
e71095e3
KS		 2673
2674 /* If it is the leading octet, determine the length of the UTF-8 sequence. */
2675
2676 if (!width)
2677 {
2678 width = (octet & 0x80) == 0x00 ? 1 :
2679 (octet & 0xE0) == 0xC0 ? 2 :
2680 (octet & 0xF0) == 0xE0 ? 3 :
2681 (octet & 0xF8) == 0xF0 ? 4 : 0;
2682 if (!width) {
2683 return yaml_parser_set_scanner_error(parser, directive ?
2684 "while parsing a %TAG directive" : "while parsing a tag",
2685 start_mark, "found an incorrect leading UTF-8 octet");
2686 }
2687 }
2688 else
2689 {
2690 /* Check if the trailing octet is correct. */
2691
2692 if ((octet & 0xC0) != 0x80) {
2693 return yaml_parser_set_scanner_error(parser, directive ?
2694 "while parsing a %TAG directive" : "while parsing a tag",
2695 start_mark, "found an incorrect trailing UTF-8 octet");
2696 }
2697 }
2698
2699 /* Copy the octet and move the pointers. */
2700
2701 *(string->pointer++) = octet;
625fcfe9
KS		 2702 SKIP(parser);
2703 SKIP(parser);
2704 SKIP(parser);
e71095e3
KS		 2705
2706 } while (--width);
2707
2708 return 1;
2709}
2710
92d41fe1
KS		 2711/*
2712 * Scan a block scalar.
2713 */
2714
625fcfe9
KS		 2715static int
2716yaml_parser_scan_block_scalar(yaml_parser_t *parser, yaml_token_t *token,
2717 int literal)
92d41fe1
KS		 2718{
2719 yaml_mark_t start_mark;
2720 yaml_mark_t end_mark;
625fcfe9
KS		 2721 yaml_string_t string = NULL_STRING;
2722 yaml_string_t leading_break = NULL_STRING;
2723 yaml_string_t trailing_breaks = NULL_STRING;
92d41fe1
KS		 2724 int chomping = 0;
2725 int increment = 0;
2726 int indent = 0;
2727 int leading_blank = 0;
2728 int trailing_blank = 0;
2729
625fcfe9
KS		 2730 if (!STRING_INIT(parser, string, INITIAL_STRING_SIZE)) goto error;
2731 if (!STRING_INIT(parser, leading_break, INITIAL_STRING_SIZE)) goto error;
2732 if (!STRING_INIT(parser, trailing_breaks, INITIAL_STRING_SIZE)) goto error;
92d41fe1
KS		 2733
2734 /* Eat the indicator '|' or '>'. */
2735
625fcfe9 2736 start_mark = parser->mark;
92d41fe1 2737
625fcfe9 2738 SKIP(parser);
92d41fe1
KS		 2739
2740 /* Scan the additional block scalar indicators. */
2741
625fcfe9 2742 if (!CACHE(parser, 1)) goto error;
92d41fe1
KS		 2743
2744 /* Check for a chomping indicator. */
2745
e35af832 2746 if (CHECK(parser->buffer, '+') || CHECK(parser->buffer, '-'))
92d41fe1
KS		 2747 {
2748 /* Set the chomping method and eat the indicator. */
2749
e35af832 2750 chomping = CHECK(parser->buffer, '+') ? +1 : -1;
92d41fe1 2751
625fcfe9 2752 SKIP(parser);
92d41fe1
KS		 2753
2754 /* Check for an indentation indicator. */
2755
625fcfe9 2756 if (!CACHE(parser, 1)) goto error;
92d41fe1 2757
e35af832 2758 if (IS_DIGIT(parser->buffer))
92d41fe1
KS		 2759 {
2760 /* Check that the intendation is greater than 0. */
2761
e35af832 2762 if (CHECK(parser->buffer, '0')) {
92d41fe1
KS		 2763 yaml_parser_set_scanner_error(parser, "while scanning a block scalar",
2764 start_mark, "found an intendation indicator equal to 0");
2765 goto error;
2766 }
2767
2768 /* Get the intendation level and eat the indicator. */
2769
e35af832 2770 increment = AS_DIGIT(parser->buffer);
92d41fe1 2771
625fcfe9 2772 SKIP(parser);
92d41fe1
KS		 2773 }
2774 }
2775
2776 /* Do the same as above, but in the opposite order. */
2777
e35af832 2778 else if (IS_DIGIT(parser->buffer))
92d41fe1 2779 {
e35af832 2780 if (CHECK(parser->buffer, '0')) {
92d41fe1
KS		 2781 yaml_parser_set_scanner_error(parser, "while scanning a block scalar",
2782 start_mark, "found an intendation indicator equal to 0");
2783 goto error;
2784 }
2785
e35af832 2786 increment = AS_DIGIT(parser->buffer);
92d41fe1 2787
625fcfe9 2788 SKIP(parser);
92d41fe1 2789
625fcfe9 2790 if (!CACHE(parser, 1)) goto error;
92d41fe1 2791
e35af832
KS		 2792 if (CHECK(parser->buffer, '+') || CHECK(parser->buffer, '-')) {
2793 chomping = CHECK(parser->buffer, '+') ? +1 : -1;
625fcfe9
KS		 2794
2795 SKIP(parser);
92d41fe1
KS		 2796 }
2797 }
2798
2799 /* Eat whitespaces and comments to the end of the line. */
2800
625fcfe9 2801 if (!CACHE(parser, 1)) goto error;
92d41fe1 2802
e35af832 2803 while (IS_BLANK(parser->buffer)) {
625fcfe9
KS		 2804 SKIP(parser);
2805 if (!CACHE(parser, 1)) goto error;
92d41fe1
KS		 2806 }
2807
e35af832
KS		 2808 if (CHECK(parser->buffer, '#')) {
2809 while (!IS_BREAKZ(parser->buffer)) {
625fcfe9
KS		 2810 SKIP(parser);
2811 if (!CACHE(parser, 1)) goto error;
92d41fe1
KS		 2812 }
2813 }
2814
2815 /* Check if we are at the end of the line. */
2816
e35af832 2817 if (!IS_BREAKZ(parser->buffer)) {
92d41fe1
KS		 2818 yaml_parser_set_scanner_error(parser, "while scanning a block scalar",
2819 start_mark, "did not found expected comment or line break");
2820 goto error;
2821 }
2822
2823 /* Eat a line break. */
2824
e35af832 2825 if (IS_BREAK(parser->buffer)) {
625fcfe9
KS		 2826 if (!CACHE(parser, 2)) goto error;
2827 SKIP_LINE(parser);
92d41fe1
KS		 2828 }
2829
625fcfe9 2830 end_mark = parser->mark;
92d41fe1
KS		 2831
2832 /* Set the intendation level if it was specified. */
2833
2834 if (increment) {
2835 indent = parser->indent >= 0 ? parser->indent+increment : increment;
2836 }
2837
2838 /* Scan the leading line breaks and determine the indentation level if needed. */
2839
21fbedd4 2840 if (!yaml_parser_scan_block_scalar_breaks(parser, &indent, &trailing_breaks,
92d41fe1
KS		 2841 start_mark, &end_mark)) goto error;
2842
2843 /* Scan the block scalar content. */
2844
625fcfe9 2845 if (!CACHE(parser, 1)) goto error;
92d41fe1 2846
e35af832 2847 while (parser->mark.column == indent && !IS_Z(parser->buffer))
92d41fe1
KS		 2848 {
2849 /*
2850 * We are at the beginning of a non-empty line.
2851 */
2852
2853 /* Is it a trailing whitespace? */
2854
e35af832 2855 trailing_blank = IS_BLANK(parser->buffer);
92d41fe1
KS		 2856
2857 /* Check if we need to fold the leading line break. */
2858
625fcfe9 2859 if (!literal && (*leading_break.start == '\n')
92d41fe1
KS		 2860 && !leading_blank && !trailing_blank)
2861 {
2862 /* Do we need to join the lines by space? */
2863
625fcfe9
KS		 2864 if (*trailing_breaks.start == '\0') {
2865 if (!STRING_EXTEND(parser, string)) goto error;
92d41fe1
KS		 2866 *(string.pointer ++) = ' ';
2867 }
2868
625fcfe9 2869 CLEAR(parser, leading_break);
92d41fe1
KS		 2870 }
2871 else {
21fbedd4 2872 if (!JOIN(parser, string, leading_break)) goto error;
625fcfe9 2873 CLEAR(parser, leading_break);
92d41fe1
KS		 2874 }
2875
2876 /* Append the remaining line breaks. */
2877
21fbedd4 2878 if (!JOIN(parser, string, trailing_breaks)) goto error;
625fcfe9 2879 CLEAR(parser, trailing_breaks);
92d41fe1
KS		 2880
2881 /* Is it a leading whitespace? */
2882
e35af832 2883 leading_blank = IS_BLANK(parser->buffer);
92d41fe1
KS		 2884
2885 /* Consume the current line. */
2886
e35af832 2887 while (!IS_BREAKZ(parser->buffer)) {
625fcfe9
KS		 2888 if (!READ(parser, string)) goto error;
2889 if (!CACHE(parser, 1)) goto error;
92d41fe1
KS		 2890 }
2891
2892 /* Consume the line break. */
2893
625fcfe9 2894 if (!CACHE(parser, 2)) goto error;
92d41fe1 2895
625fcfe9 2896 if (!READ_LINE(parser, leading_break)) goto error;
92d41fe1
KS		 2897
2898 /* Eat the following intendation spaces and line breaks. */
2899
2900 if (!yaml_parser_scan_block_scalar_breaks(parser,
21fbedd4 2901 &indent, &trailing_breaks, start_mark, &end_mark)) goto error;
92d41fe1
KS		 2902 }
2903
2904 /* Chomp the tail. */
2905
2906 if (chomping != -1) {
21fbedd4 2907 if (!JOIN(parser, string, leading_break)) goto error;
92d41fe1
KS		 2908 }
2909 if (chomping == 1) {
21fbedd4 2910 if (!JOIN(parser, string, trailing_breaks)) goto error;
92d41fe1
KS		 2911 }
2912
2913 /* Create a token. */
2914
625fcfe9 2915 SCALAR_TOKEN_INIT(*token, string.start, string.pointer-string.start,
92d41fe1
KS		 2916 literal ? YAML_LITERAL_SCALAR_STYLE : YAML_FOLDED_SCALAR_STYLE,
2917 start_mark, end_mark);
92d41fe1 2918
625fcfe9
KS		 2919 STRING_DEL(parser, leading_break);
2920 STRING_DEL(parser, trailing_breaks);
92d41fe1 2921
625fcfe9 2922 return 1;
92d41fe1
KS		 2923
2924error:
625fcfe9
KS		 2925 STRING_DEL(parser, string);
2926 STRING_DEL(parser, leading_break);
2927 STRING_DEL(parser, trailing_breaks);
92d41fe1 2928
625fcfe9 2929 return 0;
92d41fe1
KS		 2930}
2931
2932/*
2933 * Scan intendation spaces and line breaks for a block scalar. Determine the
2934 * intendation level if needed.
2935 */
2936
2937static int
2938yaml_parser_scan_block_scalar_breaks(yaml_parser_t *parser,
2939 int *indent, yaml_string_t *breaks,
2940 yaml_mark_t start_mark, yaml_mark_t *end_mark)
2941{
2942 int max_indent = 0;
2943
625fcfe9 2944 *end_mark = parser->mark;
92d41fe1
KS		 2945
2946 /* Eat the intendation spaces and line breaks. */
2947
2948 while (1)
2949 {
2950 /* Eat the intendation spaces. */
2951
625fcfe9 2952 if (!CACHE(parser, 1)) return 0;
92d41fe1 2953
e35af832
KS		 2954 while ((!*indent || parser->mark.column < *indent)
2955 && IS_SPACE(parser->buffer)) {
625fcfe9
KS		 2956 SKIP(parser);
2957 if (!CACHE(parser, 1)) return 0;
92d41fe1
KS		 2958 }
2959
625fcfe9
KS		 2960 if (parser->mark.column > max_indent)
2961 max_indent = parser->mark.column;
92d41fe1
KS		 2962
2963 /* Check for a tab character messing the intendation. */
2964
e35af832
KS		 2965 if ((!*indent || parser->mark.column < *indent)
2966 && IS_TAB(parser->buffer)) {
92d41fe1
KS		 2967 return yaml_parser_set_scanner_error(parser, "while scanning a block scalar",
2968 start_mark, "found a tab character where an intendation space is expected");
2969 }
2970
2971 /* Have we found a non-empty line? */
2972
e35af832 2973 if (!IS_BREAK(parser->buffer)) break;
92d41fe1
KS		 2974
2975 /* Consume the line break. */
2976
625fcfe9
KS		 2977 if (!CACHE(parser, 2)) return 0;
2978 if (!READ_LINE(parser, *breaks)) return 0;
2979 *end_mark = parser->mark;
92d41fe1
KS		 2980 }
2981
2982 /* Determine the indentation level if needed. */
2983
2984 if (!*indent) {
2985 *indent = max_indent;
2986 if (*indent < parser->indent + 1)
2987 *indent = parser->indent + 1;
2988 if (*indent < 1)
2989 *indent = 1;
2990 }
2991
2992 return 1;
2993}
2994
21fbedd4
KS		 2995/*
2996 * Scan a quoted scalar.
2997 */
2998
625fcfe9
KS		 2999static int
3000yaml_parser_scan_flow_scalar(yaml_parser_t *parser, yaml_token_t *token,
3001 int single)
21fbedd4
KS		 3002{
3003 yaml_mark_t start_mark;
3004 yaml_mark_t end_mark;
625fcfe9
KS		 3005 yaml_string_t string = NULL_STRING;
3006 yaml_string_t leading_break = NULL_STRING;
3007 yaml_string_t trailing_breaks = NULL_STRING;
3008 yaml_string_t whitespaces = NULL_STRING;
21fbedd4
KS		 3009 int leading_blanks;
3010
625fcfe9
KS		 3011 if (!STRING_INIT(parser, string, INITIAL_STRING_SIZE)) goto error;
3012 if (!STRING_INIT(parser, leading_break, INITIAL_STRING_SIZE)) goto error;
3013 if (!STRING_INIT(parser, trailing_breaks, INITIAL_STRING_SIZE)) goto error;
3014 if (!STRING_INIT(parser, whitespaces, INITIAL_STRING_SIZE)) goto error;
21fbedd4
KS		 3015
3016 /* Eat the left quote. */
3017
625fcfe9 3018 start_mark = parser->mark;
21fbedd4 3019
625fcfe9 3020 SKIP(parser);
21fbedd4
KS		 3021
3022 /* Consume the content of the quoted scalar. */
3023
3024 while (1)
3025 {
3026 /* Check that there are no document indicators at the beginning of the line. */
3027
625fcfe9 3028 if (!CACHE(parser, 4)) goto error;
21fbedd4 3029
625fcfe9 3030 if (parser->mark.column == 0 &&
e35af832
KS		 3031 ((CHECK_AT(parser->buffer, '-', 0) &&
3032 CHECK_AT(parser->buffer, '-', 1) &&
3033 CHECK_AT(parser->buffer, '-', 2)) ||
3034 (CHECK_AT(parser->buffer, '.', 0) &&
3035 CHECK_AT(parser->buffer, '.', 1) &&
3036 CHECK_AT(parser->buffer, '.', 2))) &&
3037 IS_BLANKZ_AT(parser->buffer, 3))
21fbedd4
KS		 3038 {
3039 yaml_parser_set_scanner_error(parser, "while scanning a quoted scalar",
3040 start_mark, "found unexpected document indicator");
3041 goto error;
3042 }
3043
3044 /* Check for EOF. */
3045
e35af832 3046 if (IS_Z(parser->buffer)) {
21fbedd4
KS		 3047 yaml_parser_set_scanner_error(parser, "while scanning a quoted scalar",
3048 start_mark, "found unexpected end of stream");
3049 goto error;
3050 }
3051
3052 /* Consume non-blank characters. */
3053
625fcfe9 3054 if (!CACHE(parser, 2)) goto error;
21fbedd4
KS		 3055
3056 leading_blanks = 0;
3057
e35af832 3058 while (!IS_BLANKZ(parser->buffer))
21fbedd4
KS		 3059 {
3060 /* Check for an escaped single quote. */
3061
e35af832
KS		 3062 if (single && CHECK_AT(parser->buffer, '\'', 0)
3063 && CHECK_AT(parser->buffer, '\'', 1))
21fbedd4 3064 {
625fcfe9 3065 if (!STRING_EXTEND(parser, string)) goto error;
21fbedd4 3066 *(string.pointer++) = '\'';
625fcfe9
KS		 3067 SKIP(parser);
3068 SKIP(parser);
21fbedd4
KS		 3069 }
3070
3071 /* Check for the right quote. */
3072
e35af832 3073 else if (CHECK(parser->buffer, single ? '\'' : '"'))
21fbedd4
KS		 3074 {
3075 break;
3076 }
3077
3078 /* Check for an escaped line break. */
3079
e35af832
KS		 3080 else if (!single && CHECK(parser->buffer, '\\')
3081 && IS_BREAK_AT(parser->buffer, 1))
21fbedd4 3082 {
625fcfe9
KS		 3083 if (!CACHE(parser, 3)) goto error;
3084 SKIP(parser);
3085 SKIP_LINE(parser);
21fbedd4
KS		 3086 leading_blanks = 1;
3087 break;
3088 }
3089
3090 /* Check for an escape sequence. */
3091
e35af832 3092 else if (!single && CHECK(parser->buffer, '\\'))
21fbedd4
KS		 3093 {
3094 int code_length = 0;
3095
625fcfe9
KS		 3096 if (!STRING_EXTEND(parser, string)) goto error;
3097
21fbedd4
KS		 3098 /* Check the escape character. */
3099
625fcfe9 3100 switch (parser->buffer.pointer[1])
21fbedd4
KS		 3101 {
3102 case '0':
3103 *(string.pointer++) = '\0';
3104 break;
3105
3106 case 'a':
3107 *(string.pointer++) = '\x07';
3108 break;
3109
3110 case 'b':
3111 *(string.pointer++) = '\x08';
3112 break;
3113
3114 case 't':
3115 case '\t':
3116 *(string.pointer++) = '\x09';
3117 break;
3118
3119 case 'n':
3120 *(string.pointer++) = '\x0A';
3121 break;
3122
3123 case 'v':
3124 *(string.pointer++) = '\x0B';
3125 break;
3126
3127 case 'f':
3128 *(string.pointer++) = '\x0C';
3129 break;
3130
3131 case 'r':
3132 *(string.pointer++) = '\x0D';
3133 break;
3134
3135 case 'e':
3136 *(string.pointer++) = '\x1B';
3137 break;
3138
3139 case ' ':
3140 *(string.pointer++) = '\x20';
3141 break;
3142
3143 case '"':
3144 *(string.pointer++) = '"';
3145 break;
3146
3147 case '\'':
3148 *(string.pointer++) = '\'';
3149 break;
3150
7e32c194
KS		 3151 case '\\':
3152 *(string.pointer++) = '\\';
3153 break;
3154
21fbedd4
KS		 3155 case 'N': /* NEL (#x85) */
3156 *(string.pointer++) = '\xC2';
3157 *(string.pointer++) = '\x85';
3158 break;
3159
3160 case '_': /* #xA0 */
3161 *(string.pointer++) = '\xC2';
3162 *(string.pointer++) = '\xA0';
3163 break;
3164
3165 case 'L': /* LS (#x2028) */
3166 *(string.pointer++) = '\xE2';
3167 *(string.pointer++) = '\x80';
3168 *(string.pointer++) = '\xA8';
3169 break;
3170
3171 case 'P': /* PS (#x2029) */
3172 *(string.pointer++) = '\xE2';
3173 *(string.pointer++) = '\x80';
7e32c194 3174 *(string.pointer++) = '\xA9';
21fbedd4
KS		 3175 break;
3176
3177 case 'x':
3178 code_length = 2;
3179 break;
3180
3181 case 'u':
3182 code_length = 4;
3183 break;
3184
3185 case 'U':
3186 code_length = 8;
3187 break;
3188
3189 default:
3190 yaml_parser_set_scanner_error(parser, "while parsing a quoted scalar",
3191 start_mark, "found unknown escape character");
3192 goto error;
3193 }
3194
625fcfe9
KS		 3195 SKIP(parser);
3196 SKIP(parser);
21fbedd4
KS		 3197
3198 /* Consume an arbitrary escape code. */
3199
3200 if (code_length)
3201 {
3202 unsigned int value = 0;
3203 int k;
3204
3205 /* Scan the character value. */
3206
625fcfe9 3207 if (!CACHE(parser, code_length)) goto error;
21fbedd4
KS		 3208
3209 for (k = 0; k < code_length; k ++) {
e35af832 3210 if (!IS_HEX_AT(parser->buffer, k)) {
21fbedd4
KS		 3211 yaml_parser_set_scanner_error(parser, "while parsing a quoted scalar",
3212 start_mark, "did not find expected hexdecimal number");
3213 goto error;
3214 }
e35af832 3215 value = (value << 4) + AS_HEX_AT(parser->buffer, k);
21fbedd4
KS		 3216 }
3217
3218 /* Check the value and write the character. */
3219
3220 if ((value >= 0xD800 && value <= 0xDFFF) || value > 0x10FFFF) {
3221 yaml_parser_set_scanner_error(parser, "while parsing a quoted scalar",
3222 start_mark, "found invalid Unicode character escape code");
3223 goto error;
3224 }
3225
3226 if (value <= 0x7F) {
3227 *(string.pointer++) = value;
3228 }
3229 else if (value <= 0x7FF) {
3230 *(string.pointer++) = 0xC0 + (value >> 6);
3231 *(string.pointer++) = 0x80 + (value & 0x3F);
3232 }
3233 else if (value <= 0xFFFF) {
3234 *(string.pointer++) = 0xE0 + (value >> 12);
3235 *(string.pointer++) = 0x80 + ((value >> 6) & 0x3F);
3236 *(string.pointer++) = 0x80 + (value & 0x3F);
3237 }
3238 else {
3239 *(string.pointer++) = 0xF0 + (value >> 18);
3240 *(string.pointer++) = 0x80 + ((value >> 12) & 0x3F);
3241 *(string.pointer++) = 0x80 + ((value >> 6) & 0x3F);
3242 *(string.pointer++) = 0x80 + (value & 0x3F);
3243 }
3244
3245 /* Advance the pointer. */
3246
3247 for (k = 0; k < code_length; k ++) {
625fcfe9 3248 SKIP(parser);
21fbedd4
KS		 3249 }
3250 }
3251 }
3252
3253 else
3254 {
3255 /* It is a non-escaped non-blank character. */
3256
625fcfe9 3257 if (!READ(parser, string)) goto error;
21fbedd4
KS		 3258 }
3259
625fcfe9 3260 if (!CACHE(parser, 2)) goto error;
21fbedd4
KS		 3261 }
3262
3263 /* Check if we are at the end of the scalar. */
3264
e35af832 3265 if (CHECK(parser->buffer, single ? '\'' : '"'))
21fbedd4
KS		 3266 break;
3267
3268 /* Consume blank characters. */
3269
625fcfe9 3270 if (!CACHE(parser, 1)) goto error;
21fbedd4 3271
e35af832 3272 while (IS_BLANK(parser->buffer) || IS_BREAK(parser->buffer))
21fbedd4 3273 {
e35af832 3274 if (IS_BLANK(parser->buffer))
21fbedd4
KS		 3275 {
3276 /* Consume a space or a tab character. */
3277
3278 if (!leading_blanks) {
625fcfe9 3279 if (!READ(parser, whitespaces)) goto error;
21fbedd4 3280 }
7e32c194 3281 else {
625fcfe9 3282 SKIP(parser);
7e32c194 3283 }
21fbedd4
KS		 3284 }
3285 else
3286 {
625fcfe9 3287 if (!CACHE(parser, 2)) goto error;
21fbedd4
KS		 3288
3289 /* Check if it is a first line break. */
3290
3291 if (!leading_blanks)
3292 {
625fcfe9
KS		 3293 CLEAR(parser, whitespaces);
3294 if (!READ_LINE(parser, leading_break)) goto error;
21fbedd4
KS		 3295 leading_blanks = 1;
3296 }
3297 else
3298 {
625fcfe9 3299 if (!READ_LINE(parser, trailing_breaks)) goto error;
21fbedd4
KS		 3300 }
3301 }
625fcfe9 3302 if (!CACHE(parser, 1)) goto error;
21fbedd4
KS		 3303 }
3304
3305 /* Join the whitespaces or fold line breaks. */
3306
21fbedd4
KS		 3307 if (leading_blanks)
3308 {
3309 /* Do we need to fold line breaks? */
3310
625fcfe9
KS		 3311 if (leading_break.start[0] == '\n') {
3312 if (trailing_breaks.start[0] == '\0') {
3313 if (!STRING_EXTEND(parser, string)) goto error;
21fbedd4
KS		 3314 *(string.pointer++) = ' ';
3315 }
3316 else {
3317 if (!JOIN(parser, string, trailing_breaks)) goto error;
625fcfe9 3318 CLEAR(parser, trailing_breaks);
21fbedd4 3319 }
625fcfe9 3320 CLEAR(parser, leading_break);
21fbedd4
KS		 3321 }
3322 else {
3323 if (!JOIN(parser, string, leading_break)) goto error;
3324 if (!JOIN(parser, string, trailing_breaks)) goto error;
625fcfe9
KS		 3325 CLEAR(parser, leading_break);
3326 CLEAR(parser, trailing_breaks);
21fbedd4
KS		 3327 }
3328 }
3329 else
3330 {
3331 if (!JOIN(parser, string, whitespaces)) goto error;
625fcfe9 3332 CLEAR(parser, whitespaces);
21fbedd4
KS		 3333 }
3334 }
3335
3336 /* Eat the right quote. */
3337
625fcfe9 3338 SKIP(parser);
21fbedd4 3339
625fcfe9 3340 end_mark = parser->mark;
21fbedd4
KS		 3341
3342 /* Create a token. */
3343
625fcfe9 3344 SCALAR_TOKEN_INIT(*token, string.start, string.pointer-string.start,
21fbedd4
KS		 3345 single ? YAML_SINGLE_QUOTED_SCALAR_STYLE : YAML_DOUBLE_QUOTED_SCALAR_STYLE,
3346 start_mark, end_mark);
21fbedd4 3347
625fcfe9
KS		 3348 STRING_DEL(parser, leading_break);
3349 STRING_DEL(parser, trailing_breaks);
3350 STRING_DEL(parser, whitespaces);
21fbedd4 3351
625fcfe9 3352 return 1;
21fbedd4
KS		 3353
3354error:
625fcfe9
KS		 3355 STRING_DEL(parser, string);
3356 STRING_DEL(parser, leading_break);
3357 STRING_DEL(parser, trailing_breaks);
3358 STRING_DEL(parser, whitespaces);
21fbedd4 3359
625fcfe9 3360 return 0;
21fbedd4
KS		 3361}
3362
3363/*
3364 * Scan a plain scalar.
3365 */
3366
625fcfe9
KS		 3367static int
3368yaml_parser_scan_plain_scalar(yaml_parser_t *parser, yaml_token_t *token)
21fbedd4
KS		 3369{
3370 yaml_mark_t start_mark;
3371 yaml_mark_t end_mark;
625fcfe9
KS		 3372 yaml_string_t string = NULL_STRING;
3373 yaml_string_t leading_break = NULL_STRING;
3374 yaml_string_t trailing_breaks = NULL_STRING;
3375 yaml_string_t whitespaces = NULL_STRING;
21fbedd4
KS		 3376 int leading_blanks = 0;
3377 int indent = parser->indent+1;
3378
625fcfe9
KS		 3379 if (!STRING_INIT(parser, string, INITIAL_STRING_SIZE)) goto error;
3380 if (!STRING_INIT(parser, leading_break, INITIAL_STRING_SIZE)) goto error;
3381 if (!STRING_INIT(parser, trailing_breaks, INITIAL_STRING_SIZE)) goto error;
3382 if (!STRING_INIT(parser, whitespaces, INITIAL_STRING_SIZE)) goto error;
21fbedd4 3383
54815ffd 3384 start_mark = end_mark = parser->mark;
21fbedd4
KS		 3385
3386 /* Consume the content of the plain scalar. */
3387
3388 while (1)
3389 {
3390 /* Check for a document indicator. */
3391
625fcfe9 3392 if (!CACHE(parser, 4)) goto error;
21fbedd4 3393
625fcfe9 3394 if (parser->mark.column == 0 &&
e35af832
KS		 3395 ((CHECK_AT(parser->buffer, '-', 0) &&
3396 CHECK_AT(parser->buffer, '-', 1) &&
3397 CHECK_AT(parser->buffer, '-', 2)) ||
3398 (CHECK_AT(parser->buffer, '.', 0) &&
3399 CHECK_AT(parser->buffer, '.', 1) &&
3400 CHECK_AT(parser->buffer, '.', 2))) &&
3401 IS_BLANKZ_AT(parser->buffer, 3)) break;
21fbedd4
KS		 3402
3403 /* Check for a comment. */
3404
e35af832 3405 if (CHECK(parser->buffer, '#'))
21fbedd4
KS		 3406 break;
3407
3408 /* Consume non-blank characters. */
3409
e35af832 3410 while (!IS_BLANKZ(parser->buffer))
21fbedd4 3411 {
7e32c194 3412 /* Check for 'x:x' in the flow context. TODO: Fix the test "spec-08-13". */
21fbedd4 3413
e35af832
KS		 3414 if (parser->flow_level
3415 && CHECK(parser->buffer, ':')
3416 && !IS_BLANKZ_AT(parser->buffer, 1)) {
21fbedd4
KS		 3417 yaml_parser_set_scanner_error(parser, "while scanning a plain scalar",
3418 start_mark, "found unexpected ':'");
3419 goto error;
3420 }
3421
3422 /* Check for indicators that may end a plain scalar. */
3423
e35af832
KS		 3424 if ((CHECK(parser->buffer, ':') && IS_BLANKZ_AT(parser->buffer, 1))
3425 || (parser->flow_level &&
3426 (CHECK(parser->buffer, ',') || CHECK(parser->buffer, ':')
3427 || CHECK(parser->buffer, '?') || CHECK(parser->buffer, '[')
3428 || CHECK(parser->buffer, ']') || CHECK(parser->buffer, '{')
3429 || CHECK(parser->buffer, '}'))))
21fbedd4
KS		 3430 break;
3431
3432 /* Check if we need to join whitespaces and breaks. */
3433
625fcfe9 3434 if (leading_blanks || whitespaces.start != whitespaces.pointer)
21fbedd4 3435 {
21fbedd4
KS		 3436 if (leading_blanks)
3437 {
3438 /* Do we need to fold line breaks? */
3439
625fcfe9
KS		 3440 if (leading_break.start[0] == '\n') {
3441 if (trailing_breaks.start[0] == '\0') {
3442 if (!STRING_EXTEND(parser, string)) goto error;
21fbedd4
KS		 3443 *(string.pointer++) = ' ';
3444 }
3445 else {
3446 if (!JOIN(parser, string, trailing_breaks)) goto error;
625fcfe9 3447 CLEAR(parser, trailing_breaks);
21fbedd4 3448 }
625fcfe9 3449 CLEAR(parser, leading_break);
21fbedd4
KS		 3450 }
3451 else {
3452 if (!JOIN(parser, string, leading_break)) goto error;
3453 if (!JOIN(parser, string, trailing_breaks)) goto error;
625fcfe9
KS		 3454 CLEAR(parser, leading_break);
3455 CLEAR(parser, trailing_breaks);
21fbedd4
KS		 3456 }
3457
3458 leading_blanks = 0;
3459 }
3460 else
3461 {
3462 if (!JOIN(parser, string, whitespaces)) goto error;
625fcfe9 3463 CLEAR(parser, whitespaces);
21fbedd4
KS		 3464 }
3465 }
3466
3467 /* Copy the character. */
3468
625fcfe9 3469 if (!READ(parser, string)) goto error;
21fbedd4 3470
625fcfe9 3471 end_mark = parser->mark;
21fbedd4 3472
625fcfe9 3473 if (!CACHE(parser, 2)) goto error;
21fbedd4
KS		 3474 }
3475
3476 /* Is it the end? */
3477
e35af832 3478 if (!(IS_BLANK(parser->buffer) || IS_BREAK(parser->buffer)))
21fbedd4
KS		 3479 break;
3480
3481 /* Consume blank characters. */
3482
625fcfe9 3483 if (!CACHE(parser, 1)) goto error;
21fbedd4 3484
e35af832 3485 while (IS_BLANK(parser->buffer) || IS_BREAK(parser->buffer))
21fbedd4 3486 {
e35af832 3487 if (IS_BLANK(parser->buffer))
21fbedd4
KS		 3488 {
3489 /* Check for tab character that abuse intendation. */
3490
e35af832
KS		 3491 if (leading_blanks && parser->mark.column < indent
3492 && IS_TAB(parser->buffer)) {
21fbedd4
KS		 3493 yaml_parser_set_scanner_error(parser, "while scanning a plain scalar",
3494 start_mark, "found a tab character that violate intendation");
7e32c194 3495 goto error;
21fbedd4
KS		 3496 }
3497
3498 /* Consume a space or a tab character. */
3499
3500 if (!leading_blanks) {
625fcfe9 3501 if (!READ(parser, whitespaces)) goto error;
21fbedd4 3502 }
7e32c194 3503 else {
625fcfe9 3504 SKIP(parser);
7e32c194 3505 }
21fbedd4
KS		 3506 }
3507 else
3508 {
625fcfe9 3509 if (!CACHE(parser, 2)) goto error;
21fbedd4
KS		 3510
3511 /* Check if it is a first line break. */
3512
3513 if (!leading_blanks)
3514 {
625fcfe9
KS		 3515 CLEAR(parser, whitespaces);
3516 if (!READ_LINE(parser, leading_break)) goto error;
21fbedd4
KS		 3517 leading_blanks = 1;
3518 }
3519 else
3520 {
625fcfe9 3521 if (!READ_LINE(parser, trailing_breaks)) goto error;
21fbedd4
KS		 3522 }
3523 }
625fcfe9 3524 if (!CACHE(parser, 1)) goto error;
21fbedd4
KS		 3525 }
3526
3527 /* Check intendation level. */
3528
625fcfe9 3529 if (!parser->flow_level && parser->mark.column < indent)
21fbedd4
KS		 3530 break;
3531 }
3532
3533 /* Create a token. */
3534
625fcfe9 3535 SCALAR_TOKEN_INIT(*token, string.start, string.pointer-string.start,
21fbedd4 3536 YAML_PLAIN_SCALAR_STYLE, start_mark, end_mark);
21fbedd4
KS		 3537
3538 /* Note that we change the 'simple_key_allowed' flag. */
3539
3540 if (leading_blanks) {
3541 parser->simple_key_allowed = 1;
3542 }
3543
625fcfe9
KS		 3544 STRING_DEL(parser, leading_break);
3545 STRING_DEL(parser, trailing_breaks);
3546 STRING_DEL(parser, whitespaces);
21fbedd4 3547
625fcfe9 3548 return 1;
21fbedd4
KS		 3549
3550error:
625fcfe9
KS		 3551 STRING_DEL(parser, string);
3552 STRING_DEL(parser, leading_break);
3553 STRING_DEL(parser, trailing_breaks);
3554 STRING_DEL(parser, whitespaces);
21fbedd4 3555
625fcfe9 3556 return 0;
21fbedd4
KS		 3557}
3558
Fast YAML 1.1 parser and emitter library
This page took 0.62772 seconds and 5 git commands to generate.