]> git.cameronkatri.com Git - mandoc.git/blob - mdoc.c
Removed mdoc_nwarn (not being used anywhere).
[mandoc.git] / mdoc.c
1 /* $Id: mdoc.c,v 1.79 2009/06/15 10:02:53 kristaps Exp $ */
2 /*
3 * Copyright (c) 2008, 2009 Kristaps Dzonsons <kristaps@kth.se>
4 *
5 * Permission to use, copy, modify, and distribute this software for any
6 * purpose with or without fee is hereby granted, provided that the above
7 * copyright notice and this permission notice appear in all copies.
8 *
9 * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
10 * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
11 * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
12 * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
13 * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
14 * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
15 * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
16 */
17 #include <assert.h>
18 #include <ctype.h>
19 #include <stdarg.h>
20 #include <stdio.h>
21 #include <stdlib.h>
22 #include <string.h>
23
24 #include "libmdoc.h"
25
26 enum merr {
27 ENOCALL,
28 EBODYPROL,
29 EPROLBODY,
30 ESPACE,
31 ETEXTPROL,
32 ENOBLANK,
33 EMALLOC
34 };
35
36 const char *const __mdoc_macronames[MDOC_MAX] = {
37 "\\\"", "Dd", "Dt", "Os",
38 "Sh", "Ss", "Pp", "D1",
39 "Dl", "Bd", "Ed", "Bl",
40 "El", "It", "Ad", "An",
41 "Ar", "Cd", "Cm", "Dv",
42 "Er", "Ev", "Ex", "Fa",
43 "Fd", "Fl", "Fn", "Ft",
44 "Ic", "In", "Li", "Nd",
45 "Nm", "Op", "Ot", "Pa",
46 "Rv", "St", "Va", "Vt",
47 /* LINTED */
48 "Xr", "\%A", "\%B", "\%D",
49 /* LINTED */
50 "\%I", "\%J", "\%N", "\%O",
51 /* LINTED */
52 "\%P", "\%R", "\%T", "\%V",
53 "Ac", "Ao", "Aq", "At",
54 "Bc", "Bf", "Bo", "Bq",
55 "Bsx", "Bx", "Db", "Dc",
56 "Do", "Dq", "Ec", "Ef",
57 "Em", "Eo", "Fx", "Ms",
58 "No", "Ns", "Nx", "Ox",
59 "Pc", "Pf", "Po", "Pq",
60 "Qc", "Ql", "Qo", "Qq",
61 "Re", "Rs", "Sc", "So",
62 "Sq", "Sm", "Sx", "Sy",
63 "Tn", "Ux", "Xc", "Xo",
64 "Fo", "Fc", "Oo", "Oc",
65 "Bk", "Ek", "Bt", "Hf",
66 "Fr", "Ud", "Lb", "Ap",
67 "Lp", "Lk", "Mt", "Brq",
68 /* LINTED */
69 "Bro", "Brc", "\%C", "Es",
70 /* LINTED */
71 "En", "Dx", "\%Q"
72 };
73
74 const char *const __mdoc_argnames[MDOC_ARG_MAX] = {
75 "split", "nosplit", "ragged",
76 "unfilled", "literal", "file",
77 "offset", "bullet", "dash",
78 "hyphen", "item", "enum",
79 "tag", "diag", "hang",
80 "ohang", "inset", "column",
81 "width", "compact", "std",
82 "filled", "words", "emphasis",
83 "symbolic", "nested"
84 };
85
86 const char * const *mdoc_macronames = __mdoc_macronames;
87 const char * const *mdoc_argnames = __mdoc_argnames;
88
89 static void mdoc_free1(struct mdoc *);
90 static int mdoc_alloc1(struct mdoc *);
91 static struct mdoc_node *node_alloc(struct mdoc *, int, int,
92 int, enum mdoc_type);
93 static int node_append(struct mdoc *,
94 struct mdoc_node *);
95 static int parsetext(struct mdoc *, int, char *);
96 static int parsemacro(struct mdoc *, int, char *);
97 static int macrowarn(struct mdoc *, int, const char *);
98 static int perr(struct mdoc *, int, int, enum merr);
99
100 #define verr(m, t) perr((m), (m)->last->line, (m)->last->pos, (t))
101
102 /*
103 * Get the first (root) node of the parse tree.
104 */
105 const struct mdoc_node *
106 mdoc_node(const struct mdoc *m)
107 {
108
109 return(MDOC_HALT & m->flags ? NULL : m->first);
110 }
111
112
113 const struct mdoc_meta *
114 mdoc_meta(const struct mdoc *m)
115 {
116
117 return(MDOC_HALT & m->flags ? NULL : &m->meta);
118 }
119
120
121 static void
122 mdoc_free1(struct mdoc *mdoc)
123 {
124
125 if (mdoc->first)
126 mdoc_node_freelist(mdoc->first);
127 if (mdoc->meta.title)
128 free(mdoc->meta.title);
129 if (mdoc->meta.os)
130 free(mdoc->meta.os);
131 if (mdoc->meta.name)
132 free(mdoc->meta.name);
133 if (mdoc->meta.arch)
134 free(mdoc->meta.arch);
135 if (mdoc->meta.vol)
136 free(mdoc->meta.vol);
137 }
138
139
140 static int
141 mdoc_alloc1(struct mdoc *mdoc)
142 {
143
144 bzero(&mdoc->meta, sizeof(struct mdoc_meta));
145 mdoc->flags = 0;
146 mdoc->lastnamed = mdoc->lastsec = 0;
147 mdoc->last = calloc(1, sizeof(struct mdoc_node));
148 if (NULL == mdoc->last)
149 return(0);
150
151 mdoc->first = mdoc->last;
152 mdoc->last->type = MDOC_ROOT;
153 mdoc->next = MDOC_NEXT_CHILD;
154 return(1);
155 }
156
157
158 /*
159 * Free up all resources contributed by a parse: the node tree,
160 * meta-data and so on. Then reallocate the root node for another
161 * parse.
162 */
163 int
164 mdoc_reset(struct mdoc *mdoc)
165 {
166
167 mdoc_free1(mdoc);
168 return(mdoc_alloc1(mdoc));
169 }
170
171
172 /*
173 * Completely free up all resources.
174 */
175 void
176 mdoc_free(struct mdoc *mdoc)
177 {
178
179 mdoc_free1(mdoc);
180 if (mdoc->htab)
181 mdoc_hash_free(mdoc->htab);
182 free(mdoc);
183 }
184
185
186 struct mdoc *
187 mdoc_alloc(void *data, int pflags, const struct mdoc_cb *cb)
188 {
189 struct mdoc *p;
190
191 if (NULL == (p = calloc(1, sizeof(struct mdoc))))
192 return(NULL);
193 if (cb)
194 (void)memcpy(&p->cb, cb, sizeof(struct mdoc_cb));
195
196 p->data = data;
197 p->pflags = pflags;
198
199 if (NULL == (p->htab = mdoc_hash_alloc())) {
200 free(p);
201 return(NULL);
202 } else if (mdoc_alloc1(p))
203 return(p);
204
205 free(p);
206 return(NULL);
207 }
208
209
210 /*
211 * Climb back up the parse tree, validating open scopes. Mostly calls
212 * through to macro_end in macro.c.
213 */
214 int
215 mdoc_endparse(struct mdoc *m)
216 {
217
218 if (MDOC_HALT & m->flags)
219 return(0);
220 else if (mdoc_macroend(m))
221 return(1);
222 m->flags |= MDOC_HALT;
223 return(0);
224 }
225
226
227 /*
228 * Main parse routine. Parses a single line -- really just hands off to
229 * the macro or text parser.
230 */
231 int
232 mdoc_parseln(struct mdoc *m, int ln, char *buf)
233 {
234
235 /* If in error-mode, then we parse no more. */
236
237 if (MDOC_HALT & m->flags)
238 return(0);
239
240 return('.' == *buf ? parsemacro(m, ln, buf) :
241 parsetext(m, ln, buf));
242 }
243
244
245 void
246 mdoc_vmsg(struct mdoc *mdoc, int ln, int pos, const char *fmt, ...)
247 {
248 char buf[256];
249 va_list ap;
250
251 if (NULL == mdoc->cb.mdoc_msg)
252 return;
253
254 va_start(ap, fmt);
255 (void)vsnprintf(buf, sizeof(buf) - 1, fmt, ap);
256 va_end(ap);
257 (*mdoc->cb.mdoc_msg)(mdoc->data, ln, pos, buf);
258 }
259
260
261 int
262 mdoc_verr(struct mdoc *mdoc, int ln, int pos,
263 const char *fmt, ...)
264 {
265 char buf[256];
266 va_list ap;
267
268 if (NULL == mdoc->cb.mdoc_err)
269 return(0);
270
271 va_start(ap, fmt);
272 (void)vsnprintf(buf, sizeof(buf) - 1, fmt, ap);
273 va_end(ap);
274 return((*mdoc->cb.mdoc_err)(mdoc->data, ln, pos, buf));
275 }
276
277
278 int
279 mdoc_vwarn(struct mdoc *mdoc, int ln, int pos,
280 enum mdoc_warn type, const char *fmt, ...)
281 {
282 char buf[256];
283 va_list ap;
284
285 if (NULL == mdoc->cb.mdoc_warn)
286 return(0);
287
288 va_start(ap, fmt);
289 (void)vsnprintf(buf, sizeof(buf) - 1, fmt, ap);
290 va_end(ap);
291 return((*mdoc->cb.mdoc_warn)(mdoc->data, ln, pos, type, buf));
292 }
293
294
295 int
296 mdoc_nerr(struct mdoc *mdoc, const struct mdoc_node *node, const char *fmt, ...)
297 {
298 char buf[256];
299 va_list ap;
300
301 if (NULL == mdoc->cb.mdoc_err)
302 return(0);
303
304 va_start(ap, fmt);
305 (void)vsnprintf(buf, sizeof(buf) - 1, fmt, ap);
306 va_end(ap);
307 return((*mdoc->cb.mdoc_err)(mdoc->data, node->line, node->pos, buf));
308 }
309
310
311 int
312 mdoc_warn(struct mdoc *mdoc, enum mdoc_warn type, const char *fmt, ...)
313 {
314 char buf[256];
315 va_list ap;
316
317 if (NULL == mdoc->cb.mdoc_warn)
318 return(0);
319
320 va_start(ap, fmt);
321 (void)vsnprintf(buf, sizeof(buf) - 1, fmt, ap);
322 va_end(ap);
323 return((*mdoc->cb.mdoc_warn)(mdoc->data, mdoc->last->line,
324 mdoc->last->pos, type, buf));
325 }
326
327
328 int
329 mdoc_err(struct mdoc *mdoc, const char *fmt, ...)
330 {
331 char buf[256];
332 va_list ap;
333
334 if (NULL == mdoc->cb.mdoc_err)
335 return(0);
336
337 va_start(ap, fmt);
338 (void)vsnprintf(buf, sizeof(buf) - 1, fmt, ap);
339 va_end(ap);
340 return((*mdoc->cb.mdoc_err)(mdoc->data, mdoc->last->line,
341 mdoc->last->pos, buf));
342 }
343
344
345 void
346 mdoc_msg(struct mdoc *mdoc, const char *fmt, ...)
347 {
348 char buf[256];
349 va_list ap;
350
351 if (NULL == mdoc->cb.mdoc_msg)
352 return;
353
354 va_start(ap, fmt);
355 (void)vsnprintf(buf, sizeof(buf) - 1, fmt, ap);
356 va_end(ap);
357 (*mdoc->cb.mdoc_msg)(mdoc->data, mdoc->last->line, mdoc->last->pos,
358 buf);
359 }
360
361
362 void
363 mdoc_pmsg(struct mdoc *mdoc, int line, int pos, const char *fmt, ...)
364 {
365 char buf[256];
366 va_list ap;
367
368 if (NULL == mdoc->cb.mdoc_msg)
369 return;
370
371 va_start(ap, fmt);
372 (void)vsnprintf(buf, sizeof(buf) - 1, fmt, ap);
373 va_end(ap);
374 (*mdoc->cb.mdoc_msg)(mdoc->data, line, pos, buf);
375 }
376
377
378 int
379 mdoc_pwarn(struct mdoc *mdoc, int line, int pos, enum mdoc_warn type,
380 const char *fmt, ...)
381 {
382 char buf[256];
383 va_list ap;
384
385 if (NULL == mdoc->cb.mdoc_warn)
386 return(0);
387
388 va_start(ap, fmt);
389 (void)vsnprintf(buf, sizeof(buf) - 1, fmt, ap);
390 va_end(ap);
391 return((*mdoc->cb.mdoc_warn)(mdoc->data, line, pos, type, buf));
392 }
393
394 int
395 mdoc_perr(struct mdoc *mdoc, int line, int pos, const char *fmt, ...)
396 {
397 char buf[256];
398 va_list ap;
399
400 if (NULL == mdoc->cb.mdoc_err)
401 return(0);
402
403 va_start(ap, fmt);
404 (void)vsnprintf(buf, sizeof(buf) - 1, fmt, ap);
405 va_end(ap);
406 return((*mdoc->cb.mdoc_err)(mdoc->data, line, pos, buf));
407 }
408
409
410 int
411 mdoc_macro(struct mdoc *m, int tok,
412 int ln, int pp, int *pos, char *buf)
413 {
414
415 /* FIXME - these should happen during validation. */
416
417 if (MDOC_PROLOGUE & mdoc_macros[tok].flags &&
418 SEC_PROLOGUE != m->lastnamed)
419 return(perr(m, ln, pp, EPROLBODY));
420
421 if ( ! (MDOC_PROLOGUE & mdoc_macros[tok].flags) &&
422 SEC_PROLOGUE == m->lastnamed)
423 return(perr(m, ln, pp, EBODYPROL));
424
425 if (1 != pp && ! (MDOC_CALLABLE & mdoc_macros[tok].flags))
426 return(perr(m, ln, pp, ENOCALL));
427
428 return((*mdoc_macros[tok].fp)(m, tok, ln, pp, pos, buf));
429 }
430
431
432 static int
433 perr(struct mdoc *m, int line, int pos, enum merr type)
434 {
435 char *p;
436
437 p = NULL;
438 switch (type) {
439 case (ENOCALL):
440 p = "not callable";
441 break;
442 case (EPROLBODY):
443 p = "macro disallowed in document body";
444 break;
445 case (EBODYPROL):
446 p = "macro disallowed in document prologue";
447 break;
448 case (EMALLOC):
449 p = "memory exhausted";
450 break;
451 case (ETEXTPROL):
452 p = "text disallowed in document prologue";
453 break;
454 case (ENOBLANK):
455 p = "blank lines disallowed in non-literal contexts";
456 break;
457 case (ESPACE):
458 p = "whitespace disallowed after delimiter";
459 break;
460 }
461 assert(p);
462 return(mdoc_perr(m, line, pos, p));
463 }
464
465
466 static int
467 node_append(struct mdoc *mdoc, struct mdoc_node *p)
468 {
469
470 assert(mdoc->last);
471 assert(mdoc->first);
472 assert(MDOC_ROOT != p->type);
473
474 switch (mdoc->next) {
475 case (MDOC_NEXT_SIBLING):
476 mdoc->last->next = p;
477 p->prev = mdoc->last;
478 p->parent = mdoc->last->parent;
479 break;
480 case (MDOC_NEXT_CHILD):
481 mdoc->last->child = p;
482 p->parent = mdoc->last;
483 break;
484 default:
485 abort();
486 /* NOTREACHED */
487 }
488
489 if ( ! mdoc_valid_pre(mdoc, p))
490 return(0);
491 if ( ! mdoc_action_pre(mdoc, p))
492 return(0);
493
494 switch (p->type) {
495 case (MDOC_HEAD):
496 assert(MDOC_BLOCK == p->parent->type);
497 p->parent->head = p;
498 break;
499 case (MDOC_TAIL):
500 assert(MDOC_BLOCK == p->parent->type);
501 p->parent->tail = p;
502 break;
503 case (MDOC_BODY):
504 assert(MDOC_BLOCK == p->parent->type);
505 p->parent->body = p;
506 break;
507 default:
508 break;
509 }
510
511 mdoc->last = p;
512
513 switch (p->type) {
514 case (MDOC_TEXT):
515 if ( ! mdoc_valid_post(mdoc))
516 return(0);
517 if ( ! mdoc_action_post(mdoc))
518 return(0);
519 break;
520 default:
521 break;
522 }
523
524 return(1);
525 }
526
527
528 static struct mdoc_node *
529 node_alloc(struct mdoc *mdoc, int line,
530 int pos, int tok, enum mdoc_type type)
531 {
532 struct mdoc_node *p;
533
534 if (NULL == (p = calloc(1, sizeof(struct mdoc_node)))) {
535 (void)verr(mdoc, EMALLOC);
536 return(NULL);
537 }
538
539 p->sec = mdoc->lastsec;
540 p->line = line;
541 p->pos = pos;
542 p->tok = tok;
543 if (MDOC_TEXT != (p->type = type))
544 assert(p->tok >= 0);
545
546 return(p);
547 }
548
549
550 int
551 mdoc_tail_alloc(struct mdoc *mdoc, int line, int pos, int tok)
552 {
553 struct mdoc_node *p;
554
555 p = node_alloc(mdoc, line, pos, tok, MDOC_TAIL);
556 if (NULL == p)
557 return(0);
558 return(node_append(mdoc, p));
559 }
560
561
562 int
563 mdoc_head_alloc(struct mdoc *mdoc, int line, int pos, int tok)
564 {
565 struct mdoc_node *p;
566
567 assert(mdoc->first);
568 assert(mdoc->last);
569
570 p = node_alloc(mdoc, line, pos, tok, MDOC_HEAD);
571 if (NULL == p)
572 return(0);
573 return(node_append(mdoc, p));
574 }
575
576
577 int
578 mdoc_body_alloc(struct mdoc *mdoc, int line, int pos, int tok)
579 {
580 struct mdoc_node *p;
581
582 p = node_alloc(mdoc, line, pos, tok, MDOC_BODY);
583 if (NULL == p)
584 return(0);
585 return(node_append(mdoc, p));
586 }
587
588
589 int
590 mdoc_block_alloc(struct mdoc *mdoc, int line, int pos,
591 int tok, struct mdoc_arg *args)
592 {
593 struct mdoc_node *p;
594
595 p = node_alloc(mdoc, line, pos, tok, MDOC_BLOCK);
596 if (NULL == p)
597 return(0);
598 p->args = args;
599 if (p->args)
600 (args->refcnt)++;
601 return(node_append(mdoc, p));
602 }
603
604
605 int
606 mdoc_elem_alloc(struct mdoc *mdoc, int line, int pos,
607 int tok, struct mdoc_arg *args)
608 {
609 struct mdoc_node *p;
610
611 p = node_alloc(mdoc, line, pos, tok, MDOC_ELEM);
612 if (NULL == p)
613 return(0);
614 p->args = args;
615 if (p->args)
616 (args->refcnt)++;
617 return(node_append(mdoc, p));
618 }
619
620
621 int
622 mdoc_word_alloc(struct mdoc *mdoc,
623 int line, int pos, const char *word)
624 {
625 struct mdoc_node *p;
626
627 p = node_alloc(mdoc, line, pos, -1, MDOC_TEXT);
628 if (NULL == p)
629 return(0);
630 if (NULL == (p->string = strdup(word))) {
631 (void)verr(mdoc, EMALLOC);
632 return(0);
633 }
634 return(node_append(mdoc, p));
635 }
636
637
638 void
639 mdoc_node_free(struct mdoc_node *p)
640 {
641
642 if (p->string)
643 free(p->string);
644 if (p->args)
645 mdoc_argv_free(p->args);
646 free(p);
647 }
648
649
650 void
651 mdoc_node_freelist(struct mdoc_node *p)
652 {
653
654 if (p->child)
655 mdoc_node_freelist(p->child);
656 if (p->next)
657 mdoc_node_freelist(p->next);
658
659 mdoc_node_free(p);
660 }
661
662
663 /*
664 * Parse free-form text, that is, a line that does not begin with the
665 * control character.
666 */
667 static int
668 parsetext(struct mdoc *m, int line, char *buf)
669 {
670
671 if (SEC_PROLOGUE == m->lastnamed)
672 return(perr(m, line, 0, ETEXTPROL));
673
674 if (0 == buf[0] && ! (MDOC_LITERAL & m->flags))
675 return(perr(m, line, 0, ENOBLANK));
676
677 if ( ! mdoc_word_alloc(m, line, 0, buf))
678 return(0);
679
680 m->next = MDOC_NEXT_SIBLING;
681 return(1);
682 }
683
684
685 static int
686 macrowarn(struct mdoc *m, int ln, const char *buf)
687 {
688 if ( ! (MDOC_IGN_MACRO & m->pflags))
689 return(mdoc_perr(m, ln, 1,
690 "unknown macro: %s%s",
691 buf, strlen(buf) > 3 ? "..." : ""));
692 return(mdoc_pwarn(m, ln, 1, WARN_SYNTAX,
693 "unknown macro: %s%s",
694 buf, strlen(buf) > 3 ? "..." : ""));
695 }
696
697
698
699 /*
700 * Parse a macro line, that is, a line beginning with the control
701 * character.
702 */
703 int
704 parsemacro(struct mdoc *m, int ln, char *buf)
705 {
706 int i, c;
707 char mac[5];
708
709 /* Comments and empties are quickly ignored. */
710
711 if (0 == buf[1])
712 return(1);
713
714 if (' ' == buf[1]) {
715 i = 2;
716 while (buf[i] && ' ' == buf[i])
717 i++;
718 if (0 == buf[i])
719 return(1);
720 return(perr(m, ln, 1, ESPACE));
721 }
722
723 if (buf[1] && '\\' == buf[1])
724 if (buf[2] && '\"' == buf[2])
725 return(1);
726
727 /* Copy the first word into a nil-terminated buffer. */
728
729 for (i = 1; i < 5; i++) {
730 if (0 == (mac[i - 1] = buf[i]))
731 break;
732 else if (' ' == buf[i])
733 break;
734 }
735
736 mac[i - 1] = 0;
737
738 if (i == 5 || i <= 2) {
739 if ( ! macrowarn(m, ln, mac))
740 goto err;
741 return(1);
742 }
743
744 if (MDOC_MAX == (c = mdoc_hash_find(m->htab, mac))) {
745 if ( ! macrowarn(m, ln, mac))
746 goto err;
747 return(1);
748 }
749
750 /* The macro is sane. Jump to the next word. */
751
752 while (buf[i] && ' ' == buf[i])
753 i++;
754
755 /* Begin recursive parse sequence. */
756
757 if ( ! mdoc_macro(m, c, ln, 1, &i, buf))
758 goto err;
759
760 return(1);
761
762 err: /* Error out. */
763
764 m->flags |= MDOC_HALT;
765 return(0);
766 }