Deprecated mdoc_msg (not being used anywhere).
[mandoc.git] / mdoc.c
1 /* $Id: mdoc.c,v 1.80 2009/06/15 10:36:01 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 int
246 mdoc_verr(struct mdoc *mdoc, int ln, int pos,
247 const char *fmt, ...)
248 {
249 char buf[256];
250 va_list ap;
251
252 if (NULL == mdoc->cb.mdoc_err)
253 return(0);
254
255 va_start(ap, fmt);
256 (void)vsnprintf(buf, sizeof(buf) - 1, fmt, ap);
257 va_end(ap);
258 return((*mdoc->cb.mdoc_err)(mdoc->data, ln, pos, buf));
259 }
260
261
262 int
263 mdoc_vwarn(struct mdoc *mdoc, int ln, int pos,
264 enum mdoc_warn type, const char *fmt, ...)
265 {
266 char buf[256];
267 va_list ap;
268
269 if (NULL == mdoc->cb.mdoc_warn)
270 return(0);
271
272 va_start(ap, fmt);
273 (void)vsnprintf(buf, sizeof(buf) - 1, fmt, ap);
274 va_end(ap);
275 return((*mdoc->cb.mdoc_warn)(mdoc->data, ln, pos, type, buf));
276 }
277
278
279 int
280 mdoc_nerr(struct mdoc *mdoc, const struct mdoc_node *node, const char *fmt, ...)
281 {
282 char buf[256];
283 va_list ap;
284
285 if (NULL == mdoc->cb.mdoc_err)
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_err)(mdoc->data, node->line, node->pos, buf));
292 }
293
294
295 int
296 mdoc_warn(struct mdoc *mdoc, enum mdoc_warn type, const char *fmt, ...)
297 {
298 char buf[256];
299 va_list ap;
300
301 if (NULL == mdoc->cb.mdoc_warn)
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_warn)(mdoc->data, mdoc->last->line,
308 mdoc->last->pos, type, buf));
309 }
310
311
312 int
313 mdoc_err(struct mdoc *mdoc, const char *fmt, ...)
314 {
315 char buf[256];
316 va_list ap;
317
318 if (NULL == mdoc->cb.mdoc_err)
319 return(0);
320
321 va_start(ap, fmt);
322 (void)vsnprintf(buf, sizeof(buf) - 1, fmt, ap);
323 va_end(ap);
324 return((*mdoc->cb.mdoc_err)(mdoc->data, mdoc->last->line,
325 mdoc->last->pos, buf));
326 }
327
328
329 int
330 mdoc_pwarn(struct mdoc *mdoc, int line, int pos, enum mdoc_warn type,
331 const char *fmt, ...)
332 {
333 char buf[256];
334 va_list ap;
335
336 if (NULL == mdoc->cb.mdoc_warn)
337 return(0);
338
339 va_start(ap, fmt);
340 (void)vsnprintf(buf, sizeof(buf) - 1, fmt, ap);
341 va_end(ap);
342 return((*mdoc->cb.mdoc_warn)(mdoc->data, line, pos, type, buf));
343 }
344
345 int
346 mdoc_perr(struct mdoc *mdoc, int line, int pos, const char *fmt, ...)
347 {
348 char buf[256];
349 va_list ap;
350
351 if (NULL == mdoc->cb.mdoc_err)
352 return(0);
353
354 va_start(ap, fmt);
355 (void)vsnprintf(buf, sizeof(buf) - 1, fmt, ap);
356 va_end(ap);
357 return((*mdoc->cb.mdoc_err)(mdoc->data, line, pos, buf));
358 }
359
360
361 int
362 mdoc_macro(struct mdoc *m, int tok,
363 int ln, int pp, int *pos, char *buf)
364 {
365
366 /* FIXME - these should happen during validation. */
367
368 if (MDOC_PROLOGUE & mdoc_macros[tok].flags &&
369 SEC_PROLOGUE != m->lastnamed)
370 return(perr(m, ln, pp, EPROLBODY));
371
372 if ( ! (MDOC_PROLOGUE & mdoc_macros[tok].flags) &&
373 SEC_PROLOGUE == m->lastnamed)
374 return(perr(m, ln, pp, EBODYPROL));
375
376 if (1 != pp && ! (MDOC_CALLABLE & mdoc_macros[tok].flags))
377 return(perr(m, ln, pp, ENOCALL));
378
379 return((*mdoc_macros[tok].fp)(m, tok, ln, pp, pos, buf));
380 }
381
382
383 static int
384 perr(struct mdoc *m, int line, int pos, enum merr type)
385 {
386 char *p;
387
388 p = NULL;
389 switch (type) {
390 case (ENOCALL):
391 p = "not callable";
392 break;
393 case (EPROLBODY):
394 p = "macro disallowed in document body";
395 break;
396 case (EBODYPROL):
397 p = "macro disallowed in document prologue";
398 break;
399 case (EMALLOC):
400 p = "memory exhausted";
401 break;
402 case (ETEXTPROL):
403 p = "text disallowed in document prologue";
404 break;
405 case (ENOBLANK):
406 p = "blank lines disallowed in non-literal contexts";
407 break;
408 case (ESPACE):
409 p = "whitespace disallowed after delimiter";
410 break;
411 }
412 assert(p);
413 return(mdoc_perr(m, line, pos, p));
414 }
415
416
417 static int
418 node_append(struct mdoc *mdoc, struct mdoc_node *p)
419 {
420
421 assert(mdoc->last);
422 assert(mdoc->first);
423 assert(MDOC_ROOT != p->type);
424
425 switch (mdoc->next) {
426 case (MDOC_NEXT_SIBLING):
427 mdoc->last->next = p;
428 p->prev = mdoc->last;
429 p->parent = mdoc->last->parent;
430 break;
431 case (MDOC_NEXT_CHILD):
432 mdoc->last->child = p;
433 p->parent = mdoc->last;
434 break;
435 default:
436 abort();
437 /* NOTREACHED */
438 }
439
440 if ( ! mdoc_valid_pre(mdoc, p))
441 return(0);
442 if ( ! mdoc_action_pre(mdoc, p))
443 return(0);
444
445 switch (p->type) {
446 case (MDOC_HEAD):
447 assert(MDOC_BLOCK == p->parent->type);
448 p->parent->head = p;
449 break;
450 case (MDOC_TAIL):
451 assert(MDOC_BLOCK == p->parent->type);
452 p->parent->tail = p;
453 break;
454 case (MDOC_BODY):
455 assert(MDOC_BLOCK == p->parent->type);
456 p->parent->body = p;
457 break;
458 default:
459 break;
460 }
461
462 mdoc->last = p;
463
464 switch (p->type) {
465 case (MDOC_TEXT):
466 if ( ! mdoc_valid_post(mdoc))
467 return(0);
468 if ( ! mdoc_action_post(mdoc))
469 return(0);
470 break;
471 default:
472 break;
473 }
474
475 return(1);
476 }
477
478
479 static struct mdoc_node *
480 node_alloc(struct mdoc *mdoc, int line,
481 int pos, int tok, enum mdoc_type type)
482 {
483 struct mdoc_node *p;
484
485 if (NULL == (p = calloc(1, sizeof(struct mdoc_node)))) {
486 (void)verr(mdoc, EMALLOC);
487 return(NULL);
488 }
489
490 p->sec = mdoc->lastsec;
491 p->line = line;
492 p->pos = pos;
493 p->tok = tok;
494 if (MDOC_TEXT != (p->type = type))
495 assert(p->tok >= 0);
496
497 return(p);
498 }
499
500
501 int
502 mdoc_tail_alloc(struct mdoc *mdoc, int line, int pos, int tok)
503 {
504 struct mdoc_node *p;
505
506 p = node_alloc(mdoc, line, pos, tok, MDOC_TAIL);
507 if (NULL == p)
508 return(0);
509 return(node_append(mdoc, p));
510 }
511
512
513 int
514 mdoc_head_alloc(struct mdoc *mdoc, int line, int pos, int tok)
515 {
516 struct mdoc_node *p;
517
518 assert(mdoc->first);
519 assert(mdoc->last);
520
521 p = node_alloc(mdoc, line, pos, tok, MDOC_HEAD);
522 if (NULL == p)
523 return(0);
524 return(node_append(mdoc, p));
525 }
526
527
528 int
529 mdoc_body_alloc(struct mdoc *mdoc, int line, int pos, int tok)
530 {
531 struct mdoc_node *p;
532
533 p = node_alloc(mdoc, line, pos, tok, MDOC_BODY);
534 if (NULL == p)
535 return(0);
536 return(node_append(mdoc, p));
537 }
538
539
540 int
541 mdoc_block_alloc(struct mdoc *mdoc, int line, int pos,
542 int tok, struct mdoc_arg *args)
543 {
544 struct mdoc_node *p;
545
546 p = node_alloc(mdoc, line, pos, tok, MDOC_BLOCK);
547 if (NULL == p)
548 return(0);
549 p->args = args;
550 if (p->args)
551 (args->refcnt)++;
552 return(node_append(mdoc, p));
553 }
554
555
556 int
557 mdoc_elem_alloc(struct mdoc *mdoc, int line, int pos,
558 int tok, struct mdoc_arg *args)
559 {
560 struct mdoc_node *p;
561
562 p = node_alloc(mdoc, line, pos, tok, MDOC_ELEM);
563 if (NULL == p)
564 return(0);
565 p->args = args;
566 if (p->args)
567 (args->refcnt)++;
568 return(node_append(mdoc, p));
569 }
570
571
572 int
573 mdoc_word_alloc(struct mdoc *mdoc,
574 int line, int pos, const char *word)
575 {
576 struct mdoc_node *p;
577
578 p = node_alloc(mdoc, line, pos, -1, MDOC_TEXT);
579 if (NULL == p)
580 return(0);
581 if (NULL == (p->string = strdup(word))) {
582 (void)verr(mdoc, EMALLOC);
583 return(0);
584 }
585 return(node_append(mdoc, p));
586 }
587
588
589 void
590 mdoc_node_free(struct mdoc_node *p)
591 {
592
593 if (p->string)
594 free(p->string);
595 if (p->args)
596 mdoc_argv_free(p->args);
597 free(p);
598 }
599
600
601 void
602 mdoc_node_freelist(struct mdoc_node *p)
603 {
604
605 if (p->child)
606 mdoc_node_freelist(p->child);
607 if (p->next)
608 mdoc_node_freelist(p->next);
609
610 mdoc_node_free(p);
611 }
612
613
614 /*
615 * Parse free-form text, that is, a line that does not begin with the
616 * control character.
617 */
618 static int
619 parsetext(struct mdoc *m, int line, char *buf)
620 {
621
622 if (SEC_PROLOGUE == m->lastnamed)
623 return(perr(m, line, 0, ETEXTPROL));
624
625 if (0 == buf[0] && ! (MDOC_LITERAL & m->flags))
626 return(perr(m, line, 0, ENOBLANK));
627
628 if ( ! mdoc_word_alloc(m, line, 0, buf))
629 return(0);
630
631 m->next = MDOC_NEXT_SIBLING;
632 return(1);
633 }
634
635
636 static int
637 macrowarn(struct mdoc *m, int ln, const char *buf)
638 {
639 if ( ! (MDOC_IGN_MACRO & m->pflags))
640 return(mdoc_perr(m, ln, 1,
641 "unknown macro: %s%s",
642 buf, strlen(buf) > 3 ? "..." : ""));
643 return(mdoc_pwarn(m, ln, 1, WARN_SYNTAX,
644 "unknown macro: %s%s",
645 buf, strlen(buf) > 3 ? "..." : ""));
646 }
647
648
649
650 /*
651 * Parse a macro line, that is, a line beginning with the control
652 * character.
653 */
654 int
655 parsemacro(struct mdoc *m, int ln, char *buf)
656 {
657 int i, c;
658 char mac[5];
659
660 /* Comments and empties are quickly ignored. */
661
662 if (0 == buf[1])
663 return(1);
664
665 if (' ' == buf[1]) {
666 i = 2;
667 while (buf[i] && ' ' == buf[i])
668 i++;
669 if (0 == buf[i])
670 return(1);
671 return(perr(m, ln, 1, ESPACE));
672 }
673
674 if (buf[1] && '\\' == buf[1])
675 if (buf[2] && '\"' == buf[2])
676 return(1);
677
678 /* Copy the first word into a nil-terminated buffer. */
679
680 for (i = 1; i < 5; i++) {
681 if (0 == (mac[i - 1] = buf[i]))
682 break;
683 else if (' ' == buf[i])
684 break;
685 }
686
687 mac[i - 1] = 0;
688
689 if (i == 5 || i <= 2) {
690 if ( ! macrowarn(m, ln, mac))
691 goto err;
692 return(1);
693 }
694
695 if (MDOC_MAX == (c = mdoc_hash_find(m->htab, mac))) {
696 if ( ! macrowarn(m, ln, mac))
697 goto err;
698 return(1);
699 }
700
701 /* The macro is sane. Jump to the next word. */
702
703 while (buf[i] && ' ' == buf[i])
704 i++;
705
706 /* Begin recursive parse sequence. */
707
708 if ( ! mdoc_macro(m, c, ln, 1, &i, buf))
709 goto err;
710
711 return(1);
712
713 err: /* Error out. */
714
715 m->flags |= MDOC_HALT;
716 return(0);
717 }