xref: /openbsd-src/usr.bin/mandoc/man.c (revision 7d464165e831b6257b4befbd30d2fbb433d300de) 
1 /*	$Id: man.c,v 1.58 2011/04/21 22:59:54 schwarze Exp $ */
2 /*
3  * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
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 <sys/types.h>
18 
19 #include <assert.h>
20 #include <stdarg.h>
21 #include <stdlib.h>
22 #include <stdio.h>
23 #include <string.h>
24 
25 #include "mandoc.h"
26 #include "libman.h"
27 #include "libmandoc.h"
28 
29 const	char *const __man_macronames[MAN_MAX] = {
30 	"br",		"TH",		"SH",		"SS",
31 	"TP", 		"LP",		"PP",		"P",
32 	"IP",		"HP",		"SM",		"SB",
33 	"BI",		"IB",		"BR",		"RB",
34 	"R",		"B",		"I",		"IR",
35 	"RI",		"na",		"sp",		"nf",
36 	"fi",		"RE",		"RS",		"DT",
37 	"UC",		"PD",		"AT",		"in",
38 	"ft"
39 	};
40 
41 const	char * const *man_macronames = __man_macronames;
42 
43 static	struct man_node	*man_node_alloc(struct man *, int, int,
44 				enum man_type, enum mant);
45 static	int		 man_node_append(struct man *,
46 				struct man_node *);
47 static	void		 man_node_free(struct man_node *);
48 static	void		 man_node_unlink(struct man *,
49 				struct man_node *);
50 static	int		 man_ptext(struct man *, int, char *, int);
51 static	int		 man_pmacro(struct man *, int, char *, int);
52 static	void		 man_free1(struct man *);
53 static	void		 man_alloc1(struct man *);
54 static	int		 man_descope(struct man *, int, int);
55 
56 
57 const struct man_node *
58 man_node(const struct man *m)
59 {
60 
61 	assert( ! (MAN_HALT & m->flags));
62 	return(m->first);
63 }
64 
65 
66 const struct man_meta *
67 man_meta(const struct man *m)
68 {
69 
70 	assert( ! (MAN_HALT & m->flags));
71 	return(&m->meta);
72 }
73 
74 
75 void
76 man_reset(struct man *man)
77 {
78 
79 	man_free1(man);
80 	man_alloc1(man);
81 }
82 
83 
84 void
85 man_free(struct man *man)
86 {
87 
88 	man_free1(man);
89 	free(man);
90 }
91 
92 
93 struct man *
94 man_alloc(struct regset *regs, void *data, mandocmsg msg)
95 {
96 	struct man	*p;
97 
98 	p = mandoc_calloc(1, sizeof(struct man));
99 
100 	man_hash_init();
101 	p->data = data;
102 	p->msg = msg;
103 	p->regs = regs;
104 
105 	man_alloc1(p);
106 	return(p);
107 }
108 
109 
110 int
111 man_endparse(struct man *m)
112 {
113 
114 	assert( ! (MAN_HALT & m->flags));
115 	if (man_macroend(m))
116 		return(1);
117 	m->flags |= MAN_HALT;
118 	return(0);
119 }
120 
121 
122 int
123 man_parseln(struct man *m, int ln, char *buf, int offs)
124 {
125 
126 	m->flags |= MAN_NEWLINE;
127 
128 	assert( ! (MAN_HALT & m->flags));
129 	return(('.' == buf[offs] || '\'' == buf[offs]) ?
130 			man_pmacro(m, ln, buf, offs) :
131 			man_ptext(m, ln, buf, offs));
132 }
133 
134 
135 static void
136 man_free1(struct man *man)
137 {
138 
139 	if (man->first)
140 		man_node_delete(man, man->first);
141 	if (man->meta.title)
142 		free(man->meta.title);
143 	if (man->meta.source)
144 		free(man->meta.source);
145 	if (man->meta.date)
146 		free(man->meta.date);
147 	if (man->meta.vol)
148 		free(man->meta.vol);
149 	if (man->meta.msec)
150 		free(man->meta.msec);
151 }
152 
153 
154 static void
155 man_alloc1(struct man *m)
156 {
157 
158 	memset(&m->meta, 0, sizeof(struct man_meta));
159 	m->flags = 0;
160 	m->last = mandoc_calloc(1, sizeof(struct man_node));
161 	m->first = m->last;
162 	m->last->type = MAN_ROOT;
163 	m->last->tok = MAN_MAX;
164 	m->next = MAN_NEXT_CHILD;
165 }
166 
167 
168 static int
169 man_node_append(struct man *man, struct man_node *p)
170 {
171 
172 	assert(man->last);
173 	assert(man->first);
174 	assert(MAN_ROOT != p->type);
175 
176 	switch (man->next) {
177 	case (MAN_NEXT_SIBLING):
178 		man->last->next = p;
179 		p->prev = man->last;
180 		p->parent = man->last->parent;
181 		break;
182 	case (MAN_NEXT_CHILD):
183 		man->last->child = p;
184 		p->parent = man->last;
185 		break;
186 	default:
187 		abort();
188 		/* NOTREACHED */
189 	}
190 
191 	assert(p->parent);
192 	p->parent->nchild++;
193 
194 	if ( ! man_valid_pre(man, p))
195 		return(0);
196 
197 	switch (p->type) {
198 	case (MAN_HEAD):
199 		assert(MAN_BLOCK == p->parent->type);
200 		p->parent->head = p;
201 		break;
202 	case (MAN_BODY):
203 		assert(MAN_BLOCK == p->parent->type);
204 		p->parent->body = p;
205 		break;
206 	default:
207 		break;
208 	}
209 
210 	man->last = p;
211 
212 	switch (p->type) {
213 	case (MAN_TBL):
214 		/* FALLTHROUGH */
215 	case (MAN_TEXT):
216 		if ( ! man_valid_post(man))
217 			return(0);
218 		break;
219 	default:
220 		break;
221 	}
222 
223 	return(1);
224 }
225 
226 
227 static struct man_node *
228 man_node_alloc(struct man *m, int line, int pos,
229 		enum man_type type, enum mant tok)
230 {
231 	struct man_node *p;
232 
233 	p = mandoc_calloc(1, sizeof(struct man_node));
234 	p->line = line;
235 	p->pos = pos;
236 	p->type = type;
237 	p->tok = tok;
238 
239 	if (MAN_NEWLINE & m->flags)
240 		p->flags |= MAN_LINE;
241 	m->flags &= ~MAN_NEWLINE;
242 	return(p);
243 }
244 
245 
246 int
247 man_elem_alloc(struct man *m, int line, int pos, enum mant tok)
248 {
249 	struct man_node *p;
250 
251 	p = man_node_alloc(m, line, pos, MAN_ELEM, tok);
252 	if ( ! man_node_append(m, p))
253 		return(0);
254 	m->next = MAN_NEXT_CHILD;
255 	return(1);
256 }
257 
258 
259 int
260 man_head_alloc(struct man *m, int line, int pos, enum mant tok)
261 {
262 	struct man_node *p;
263 
264 	p = man_node_alloc(m, line, pos, MAN_HEAD, tok);
265 	if ( ! man_node_append(m, p))
266 		return(0);
267 	m->next = MAN_NEXT_CHILD;
268 	return(1);
269 }
270 
271 
272 int
273 man_body_alloc(struct man *m, int line, int pos, enum mant tok)
274 {
275 	struct man_node *p;
276 
277 	p = man_node_alloc(m, line, pos, MAN_BODY, tok);
278 	if ( ! man_node_append(m, p))
279 		return(0);
280 	m->next = MAN_NEXT_CHILD;
281 	return(1);
282 }
283 
284 
285 int
286 man_block_alloc(struct man *m, int line, int pos, enum mant tok)
287 {
288 	struct man_node *p;
289 
290 	p = man_node_alloc(m, line, pos, MAN_BLOCK, tok);
291 	if ( ! man_node_append(m, p))
292 		return(0);
293 	m->next = MAN_NEXT_CHILD;
294 	return(1);
295 }
296 
297 int
298 man_word_alloc(struct man *m, int line, int pos, const char *word)
299 {
300 	struct man_node	*n;
301 	size_t		 sv, len;
302 
303 	len = strlen(word);
304 
305 	n = man_node_alloc(m, line, pos, MAN_TEXT, MAN_MAX);
306 	n->string = mandoc_malloc(len + 1);
307 	sv = strlcpy(n->string, word, len + 1);
308 
309 	/* Prohibit truncation. */
310 	assert(sv < len + 1);
311 
312 	if ( ! man_node_append(m, n))
313 		return(0);
314 
315 	m->next = MAN_NEXT_SIBLING;
316 	return(1);
317 }
318 
319 
320 /*
321  * Free all of the resources held by a node.  This does NOT unlink a
322  * node from its context; for that, see man_node_unlink().
323  */
324 static void
325 man_node_free(struct man_node *p)
326 {
327 
328 	if (p->string)
329 		free(p->string);
330 	free(p);
331 }
332 
333 
334 void
335 man_node_delete(struct man *m, struct man_node *p)
336 {
337 
338 	while (p->child)
339 		man_node_delete(m, p->child);
340 
341 	man_node_unlink(m, p);
342 	man_node_free(p);
343 }
344 
345 int
346 man_addeqn(struct man *m, const struct eqn *ep)
347 {
348 	struct man_node	*n;
349 
350 	assert( ! (MAN_HALT & m->flags));
351 
352 	n = man_node_alloc(m, ep->line, ep->pos, MAN_EQN, MAN_MAX);
353 	n->eqn = ep;
354 
355 	if ( ! man_node_append(m, n))
356 		return(0);
357 
358 	m->next = MAN_NEXT_SIBLING;
359 	return(man_descope(m, ep->line, ep->pos));
360 }
361 
362 int
363 man_addspan(struct man *m, const struct tbl_span *sp)
364 {
365 	struct man_node	*n;
366 
367 	assert( ! (MAN_HALT & m->flags));
368 
369 	n = man_node_alloc(m, sp->line, 0, MAN_TBL, MAN_MAX);
370 	n->span = sp;
371 
372 	if ( ! man_node_append(m, n))
373 		return(0);
374 
375 	m->next = MAN_NEXT_SIBLING;
376 	return(man_descope(m, sp->line, 0));
377 }
378 
379 static int
380 man_descope(struct man *m, int line, int offs)
381 {
382 	/*
383 	 * Co-ordinate what happens with having a next-line scope open:
384 	 * first close out the element scope (if applicable), then close
385 	 * out the block scope (also if applicable).
386 	 */
387 
388 	if (MAN_ELINE & m->flags) {
389 		m->flags &= ~MAN_ELINE;
390 		if ( ! man_unscope(m, m->last->parent, MANDOCERR_MAX))
391 			return(0);
392 	}
393 
394 	if ( ! (MAN_BLINE & m->flags))
395 		return(1);
396 	m->flags &= ~MAN_BLINE;
397 
398 	if ( ! man_unscope(m, m->last->parent, MANDOCERR_MAX))
399 		return(0);
400 	return(man_body_alloc(m, line, offs, m->last->tok));
401 }
402 
403 
404 static int
405 man_ptext(struct man *m, int line, char *buf, int offs)
406 {
407 	int		 i;
408 
409 	/* Ignore bogus comments. */
410 
411 	if ('\\' == buf[offs] &&
412 			'.' == buf[offs + 1] &&
413 			'"' == buf[offs + 2]) {
414 		man_pmsg(m, line, offs, MANDOCERR_BADCOMMENT);
415 		return(1);
416 	}
417 
418 	/* Literal free-form text whitespace is preserved. */
419 
420 	if (MAN_LITERAL & m->flags) {
421 		if ( ! man_word_alloc(m, line, offs, buf + offs))
422 			return(0);
423 		return(man_descope(m, line, offs));
424 	}
425 
426 	/* Pump blank lines directly into the backend. */
427 
428 	for (i = offs; ' ' == buf[i]; i++)
429 		/* Skip leading whitespace. */ ;
430 
431 	if ('\0' == buf[i]) {
432 		/* Allocate a blank entry. */
433 		if ( ! man_word_alloc(m, line, offs, ""))
434 			return(0);
435 		return(man_descope(m, line, offs));
436 	}
437 
438 	/*
439 	 * Warn if the last un-escaped character is whitespace. Then
440 	 * strip away the remaining spaces (tabs stay!).
441 	 */
442 
443 	i = (int)strlen(buf);
444 	assert(i);
445 
446 	if (' ' == buf[i - 1] || '\t' == buf[i - 1]) {
447 		if (i > 1 && '\\' != buf[i - 2])
448 			man_pmsg(m, line, i - 1, MANDOCERR_EOLNSPACE);
449 
450 		for (--i; i && ' ' == buf[i]; i--)
451 			/* Spin back to non-space. */ ;
452 
453 		/* Jump ahead of escaped whitespace. */
454 		i += '\\' == buf[i] ? 2 : 1;
455 
456 		buf[i] = '\0';
457 	}
458 
459 	if ( ! man_word_alloc(m, line, offs, buf + offs))
460 		return(0);
461 
462 	/*
463 	 * End-of-sentence check.  If the last character is an unescaped
464 	 * EOS character, then flag the node as being the end of a
465 	 * sentence.  The front-end will know how to interpret this.
466 	 */
467 
468 	assert(i);
469 	if (mandoc_eos(buf, (size_t)i, 0))
470 		m->last->flags |= MAN_EOS;
471 
472 	return(man_descope(m, line, offs));
473 }
474 
475 
476 static int
477 man_pmacro(struct man *m, int ln, char *buf, int offs)
478 {
479 	int		 i, j, ppos;
480 	enum mant	 tok;
481 	char		 mac[5];
482 	struct man_node	*n;
483 
484 	/* Comments and empties are quickly ignored. */
485 
486 	offs++;
487 
488 	if ('\0' == buf[offs])
489 		return(1);
490 
491 	i = offs;
492 
493 	/*
494 	 * Skip whitespace between the control character and initial
495 	 * text.  "Whitespace" is both spaces and tabs.
496 	 */
497 
498 	if (' ' == buf[i] || '\t' == buf[i]) {
499 		i++;
500 		while (buf[i] && (' ' == buf[i] || '\t' == buf[i]))
501 			i++;
502 		if ('\0' == buf[i])
503 			goto out;
504 	}
505 
506 	ppos = i;
507 
508 	/*
509 	 * Copy the first word into a nil-terminated buffer.
510 	 * Stop copying when a tab, space, or eoln is encountered.
511 	 */
512 
513 	j = 0;
514 	while (j < 4 && '\0' != buf[i] && ' ' != buf[i] && '\t' != buf[i])
515 		mac[j++] = buf[i++];
516 	mac[j] = '\0';
517 
518 	tok = (j > 0 && j < 4) ? man_hash_find(mac) : MAN_MAX;
519 	if (MAN_MAX == tok) {
520 		man_vmsg(m, MANDOCERR_MACRO, ln, ppos, "%s", buf + ppos - 1);
521 		return(1);
522 	}
523 
524 	/* The macro is sane.  Jump to the next word. */
525 
526 	while (buf[i] && ' ' == buf[i])
527 		i++;
528 
529 	/*
530 	 * Trailing whitespace.  Note that tabs are allowed to be passed
531 	 * into the parser as "text", so we only warn about spaces here.
532 	 */
533 
534 	if ('\0' == buf[i] && ' ' == buf[i - 1])
535 		man_pmsg(m, ln, i - 1, MANDOCERR_EOLNSPACE);
536 
537 	/*
538 	 * Remove prior ELINE macro, as it's being clobbered by a new
539 	 * macro.  Note that NSCOPED macros do not close out ELINE
540 	 * macros---they don't print text---so we let those slip by.
541 	 */
542 
543 	if ( ! (MAN_NSCOPED & man_macros[tok].flags) &&
544 			m->flags & MAN_ELINE) {
545 		n = m->last;
546 		assert(MAN_TEXT != n->type);
547 
548 		/* Remove repeated NSCOPED macros causing ELINE. */
549 
550 		if (MAN_NSCOPED & man_macros[n->tok].flags)
551 			n = n->parent;
552 
553 		man_vmsg(m, MANDOCERR_LINESCOPE, n->line, n->pos,
554 				"%s", man_macronames[n->tok]);
555 
556 		man_node_delete(m, n);
557 		m->flags &= ~MAN_ELINE;
558 	}
559 
560 	/*
561 	 * Save the fact that we're in the next-line for a block.  In
562 	 * this way, embedded roff instructions can "remember" state
563 	 * when they exit.
564 	 */
565 
566 	if (MAN_BLINE & m->flags)
567 		m->flags |= MAN_BPLINE;
568 
569 	/* Call to handler... */
570 
571 	assert(man_macros[tok].fp);
572 	if ( ! (*man_macros[tok].fp)(m, tok, ln, ppos, &i, buf))
573 		goto err;
574 
575 out:
576 	/*
577 	 * We weren't in a block-line scope when entering the
578 	 * above-parsed macro, so return.
579 	 */
580 
581 	if ( ! (MAN_BPLINE & m->flags)) {
582 		m->flags &= ~MAN_ILINE;
583 		return(1);
584 	}
585 	m->flags &= ~MAN_BPLINE;
586 
587 	/*
588 	 * If we're in a block scope, then allow this macro to slip by
589 	 * without closing scope around it.
590 	 */
591 
592 	if (MAN_ILINE & m->flags) {
593 		m->flags &= ~MAN_ILINE;
594 		return(1);
595 	}
596 
597 	/*
598 	 * If we've opened a new next-line element scope, then return
599 	 * now, as the next line will close out the block scope.
600 	 */
601 
602 	if (MAN_ELINE & m->flags)
603 		return(1);
604 
605 	/* Close out the block scope opened in the prior line.  */
606 
607 	assert(MAN_BLINE & m->flags);
608 	m->flags &= ~MAN_BLINE;
609 
610 	if ( ! man_unscope(m, m->last->parent, MANDOCERR_MAX))
611 		return(0);
612 	return(man_body_alloc(m, ln, offs, m->last->tok));
613 
614 err:	/* Error out. */
615 
616 	m->flags |= MAN_HALT;
617 	return(0);
618 }
619 
620 
621 void
622 man_vmsg(struct man *man, enum mandocerr t,
623 		int ln, int pos, const char *fmt, ...)
624 {
625 	char		 buf[256];
626 	va_list		 ap;
627 
628 	va_start(ap, fmt);
629 	vsnprintf(buf, sizeof(buf) - 1, fmt, ap);
630 	va_end(ap);
631 	(*man->msg)(t, man->data, ln, pos, buf);
632 }
633 
634 
635 /*
636  * Unlink a node from its context.  If "m" is provided, the last parse
637  * point will also be adjusted accordingly.
638  */
639 static void
640 man_node_unlink(struct man *m, struct man_node *n)
641 {
642 
643 	/* Adjust siblings. */
644 
645 	if (n->prev)
646 		n->prev->next = n->next;
647 	if (n->next)
648 		n->next->prev = n->prev;
649 
650 	/* Adjust parent. */
651 
652 	if (n->parent) {
653 		n->parent->nchild--;
654 		if (n->parent->child == n)
655 			n->parent->child = n->prev ? n->prev : n->next;
656 	}
657 
658 	/* Adjust parse point, if applicable. */
659 
660 	if (m && m->last == n) {
661 		/*XXX: this can occur when bailing from validation. */
662 		/*assert(NULL == n->next);*/
663 		if (n->prev) {
664 			m->last = n->prev;
665 			m->next = MAN_NEXT_SIBLING;
666 		} else {
667 			m->last = n->parent;
668 			m->next = MAN_NEXT_CHILD;
669 		}
670 	}
671 
672 	if (m && m->first == n)
673 		m->first = NULL;
674 }
675