xref: /openbsd-src/usr.bin/mandoc/man.c (revision 5b859c19fe53bbea08f5c342e0a4470e99f883e1) 
1 /*	$OpenBSD: man.c,v 1.93 2014/11/28 06:26:46 schwarze Exp $ */
2 /*
3  * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
4  * Copyright (c) 2013, 2014 Ingo Schwarze <schwarze@openbsd.org>
5  * Copyright (c) 2011 Joerg Sonnenberger <joerg@netbsd.org>
6  *
7  * Permission to use, copy, modify, and distribute this software for any
8  * purpose with or without fee is hereby granted, provided that the above
9  * copyright notice and this permission notice appear in all copies.
10  *
11  * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
12  * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
13  * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
14  * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
15  * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
16  * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
17  * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
18  */
19 #include <sys/types.h>
20 
21 #include <assert.h>
22 #include <ctype.h>
23 #include <stdarg.h>
24 #include <stdlib.h>
25 #include <stdio.h>
26 #include <string.h>
27 
28 #include "man.h"
29 #include "mandoc.h"
30 #include "mandoc_aux.h"
31 #include "libman.h"
32 #include "libmandoc.h"
33 
34 const	char *const __man_macronames[MAN_MAX] = {
35 	"br",		"TH",		"SH",		"SS",
36 	"TP",		"LP",		"PP",		"P",
37 	"IP",		"HP",		"SM",		"SB",
38 	"BI",		"IB",		"BR",		"RB",
39 	"R",		"B",		"I",		"IR",
40 	"RI",		"na",		"sp",		"nf",
41 	"fi",		"RE",		"RS",		"DT",
42 	"UC",		"PD",		"AT",		"in",
43 	"ft",		"OP",		"EX",		"EE",
44 	"UR",		"UE",		"ll"
45 	};
46 
47 const	char * const *man_macronames = __man_macronames;
48 
49 static	struct man_node	*man_node_alloc(struct man *, int, int,
50 				enum man_type, enum mant);
51 static	void		 man_node_append(struct man *, struct man_node *);
52 static	void		 man_node_free(struct man_node *);
53 static	void		 man_node_unlink(struct man *,
54 				struct man_node *);
55 static	int		 man_ptext(struct man *, int, char *, int);
56 static	int		 man_pmacro(struct man *, int, char *, int);
57 static	void		 man_free1(struct man *);
58 static	void		 man_alloc1(struct man *);
59 static	void		 man_descope(struct man *, int, int);
60 
61 
62 const struct man_node *
63 man_node(const struct man *man)
64 {
65 
66 	return(man->first);
67 }
68 
69 const struct man_meta *
70 man_meta(const struct man *man)
71 {
72 
73 	return(&man->meta);
74 }
75 
76 void
77 man_reset(struct man *man)
78 {
79 
80 	man_free1(man);
81 	man_alloc1(man);
82 }
83 
84 void
85 man_free(struct man *man)
86 {
87 
88 	man_free1(man);
89 	free(man);
90 }
91 
92 struct man *
93 man_alloc(struct roff *roff, struct mparse *parse, int quick)
94 {
95 	struct man	*p;
96 
97 	p = mandoc_calloc(1, sizeof(struct man));
98 
99 	man_hash_init();
100 	p->parse = parse;
101 	p->quick = quick;
102 	p->roff = roff;
103 
104 	man_alloc1(p);
105 	return(p);
106 }
107 
108 int
109 man_endparse(struct man *man)
110 {
111 
112 	man_macroend(man);
113 	return(1);
114 }
115 
116 int
117 man_parseln(struct man *man, int ln, char *buf, int offs)
118 {
119 
120 	if (man->last->type != MAN_EQN || ln > man->last->line)
121 		man->flags |= MAN_NEWLINE;
122 
123 	return (roff_getcontrol(man->roff, buf, &offs) ?
124 	    man_pmacro(man, ln, buf, offs) :
125 	    man_ptext(man, ln, buf, offs));
126 }
127 
128 static void
129 man_free1(struct man *man)
130 {
131 
132 	if (man->first)
133 		man_node_delete(man, man->first);
134 	free(man->meta.title);
135 	free(man->meta.source);
136 	free(man->meta.date);
137 	free(man->meta.vol);
138 	free(man->meta.msec);
139 }
140 
141 static void
142 man_alloc1(struct man *man)
143 {
144 
145 	memset(&man->meta, 0, sizeof(struct man_meta));
146 	man->flags = 0;
147 	man->last = mandoc_calloc(1, sizeof(struct man_node));
148 	man->first = man->last;
149 	man->last->type = MAN_ROOT;
150 	man->last->tok = MAN_MAX;
151 	man->next = MAN_NEXT_CHILD;
152 }
153 
154 
155 static void
156 man_node_append(struct man *man, struct man_node *p)
157 {
158 
159 	assert(man->last);
160 	assert(man->first);
161 	assert(p->type != MAN_ROOT);
162 
163 	switch (man->next) {
164 	case MAN_NEXT_SIBLING:
165 		man->last->next = p;
166 		p->prev = man->last;
167 		p->parent = man->last->parent;
168 		break;
169 	case MAN_NEXT_CHILD:
170 		man->last->child = p;
171 		p->parent = man->last;
172 		break;
173 	default:
174 		abort();
175 		/* NOTREACHED */
176 	}
177 
178 	assert(p->parent);
179 	p->parent->nchild++;
180 
181 	switch (p->type) {
182 	case MAN_BLOCK:
183 		if (p->tok == MAN_SH || p->tok == MAN_SS)
184 			man->flags &= ~MAN_LITERAL;
185 		break;
186 	case MAN_HEAD:
187 		assert(p->parent->type == MAN_BLOCK);
188 		p->parent->head = p;
189 		break;
190 	case MAN_BODY:
191 		assert(p->parent->type == MAN_BLOCK);
192 		p->parent->body = p;
193 		break;
194 	default:
195 		break;
196 	}
197 
198 	man->last = p;
199 
200 	switch (p->type) {
201 	case MAN_TBL:
202 		/* FALLTHROUGH */
203 	case MAN_TEXT:
204 		man_valid_post(man);
205 		break;
206 	default:
207 		break;
208 	}
209 }
210 
211 static struct man_node *
212 man_node_alloc(struct man *man, int line, int pos,
213 		enum man_type type, enum mant tok)
214 {
215 	struct man_node *p;
216 
217 	p = mandoc_calloc(1, sizeof(struct man_node));
218 	p->line = line;
219 	p->pos = pos;
220 	p->type = type;
221 	p->tok = tok;
222 
223 	if (man->flags & MAN_NEWLINE)
224 		p->flags |= MAN_LINE;
225 	man->flags &= ~MAN_NEWLINE;
226 	return(p);
227 }
228 
229 void
230 man_elem_alloc(struct man *man, int line, int pos, enum mant tok)
231 {
232 	struct man_node *p;
233 
234 	p = man_node_alloc(man, line, pos, MAN_ELEM, tok);
235 	man_node_append(man, p);
236 	man->next = MAN_NEXT_CHILD;
237 }
238 
239 void
240 man_head_alloc(struct man *man, int line, int pos, enum mant tok)
241 {
242 	struct man_node *p;
243 
244 	p = man_node_alloc(man, line, pos, MAN_HEAD, tok);
245 	man_node_append(man, p);
246 	man->next = MAN_NEXT_CHILD;
247 }
248 
249 void
250 man_body_alloc(struct man *man, int line, int pos, enum mant tok)
251 {
252 	struct man_node *p;
253 
254 	p = man_node_alloc(man, line, pos, MAN_BODY, tok);
255 	man_node_append(man, p);
256 	man->next = MAN_NEXT_CHILD;
257 }
258 
259 void
260 man_block_alloc(struct man *man, int line, int pos, enum mant tok)
261 {
262 	struct man_node *p;
263 
264 	p = man_node_alloc(man, line, pos, MAN_BLOCK, tok);
265 	man_node_append(man, p);
266 	man->next = MAN_NEXT_CHILD;
267 }
268 
269 void
270 man_word_alloc(struct man *man, int line, int pos, const char *word)
271 {
272 	struct man_node	*n;
273 
274 	n = man_node_alloc(man, line, pos, MAN_TEXT, MAN_MAX);
275 	n->string = roff_strdup(man->roff, word);
276 	man_node_append(man, n);
277 	man->next = MAN_NEXT_SIBLING;
278 }
279 
280 void
281 man_word_append(struct man *man, const char *word)
282 {
283 	struct man_node	*n;
284 	char		*addstr, *newstr;
285 
286 	n = man->last;
287 	addstr = roff_strdup(man->roff, word);
288 	mandoc_asprintf(&newstr, "%s %s", n->string, addstr);
289 	free(addstr);
290 	free(n->string);
291 	n->string = newstr;
292 	man->next = MAN_NEXT_SIBLING;
293 }
294 
295 /*
296  * Free all of the resources held by a node.  This does NOT unlink a
297  * node from its context; for that, see man_node_unlink().
298  */
299 static void
300 man_node_free(struct man_node *p)
301 {
302 
303 	free(p->string);
304 	free(p);
305 }
306 
307 void
308 man_node_delete(struct man *man, struct man_node *p)
309 {
310 
311 	while (p->child)
312 		man_node_delete(man, p->child);
313 
314 	man_node_unlink(man, p);
315 	man_node_free(p);
316 }
317 
318 void
319 man_addeqn(struct man *man, const struct eqn *ep)
320 {
321 	struct man_node	*n;
322 
323 	n = man_node_alloc(man, ep->ln, ep->pos, MAN_EQN, MAN_MAX);
324 	n->eqn = ep;
325 	if (ep->ln > man->last->line)
326 		n->flags |= MAN_LINE;
327 	man_node_append(man, n);
328 	man->next = MAN_NEXT_SIBLING;
329 	man_descope(man, ep->ln, ep->pos);
330 }
331 
332 void
333 man_addspan(struct man *man, const struct tbl_span *sp)
334 {
335 	struct man_node	*n;
336 
337 	n = man_node_alloc(man, sp->line, 0, MAN_TBL, MAN_MAX);
338 	n->span = sp;
339 	man_node_append(man, n);
340 	man->next = MAN_NEXT_SIBLING;
341 	man_descope(man, sp->line, 0);
342 }
343 
344 static void
345 man_descope(struct man *man, int line, int offs)
346 {
347 	/*
348 	 * Co-ordinate what happens with having a next-line scope open:
349 	 * first close out the element scope (if applicable), then close
350 	 * out the block scope (also if applicable).
351 	 */
352 
353 	if (man->flags & MAN_ELINE) {
354 		man->flags &= ~MAN_ELINE;
355 		man_unscope(man, man->last->parent);
356 	}
357 	if ( ! (man->flags & MAN_BLINE))
358 		return;
359 	man->flags &= ~MAN_BLINE;
360 	man_unscope(man, man->last->parent);
361 	man_body_alloc(man, line, offs, man->last->tok);
362 }
363 
364 static int
365 man_ptext(struct man *man, int line, char *buf, int offs)
366 {
367 	int		 i;
368 
369 	/* Literal free-form text whitespace is preserved. */
370 
371 	if (man->flags & MAN_LITERAL) {
372 		man_word_alloc(man, line, offs, buf + offs);
373 		man_descope(man, line, offs);
374 		return(1);
375 	}
376 
377 	for (i = offs; buf[i] == ' '; i++)
378 		/* Skip leading whitespace. */ ;
379 
380 	/*
381 	 * Blank lines are ignored right after headings
382 	 * but add a single vertical space elsewhere.
383 	 */
384 
385 	if (buf[i] == '\0') {
386 		/* Allocate a blank entry. */
387 		if (man->last->tok != MAN_SH &&
388 		    man->last->tok != MAN_SS) {
389 			man_elem_alloc(man, line, offs, MAN_sp);
390 			man->next = MAN_NEXT_SIBLING;
391 		}
392 		return(1);
393 	}
394 
395 	/*
396 	 * Warn if the last un-escaped character is whitespace. Then
397 	 * strip away the remaining spaces (tabs stay!).
398 	 */
399 
400 	i = (int)strlen(buf);
401 	assert(i);
402 
403 	if (' ' == buf[i - 1] || '\t' == buf[i - 1]) {
404 		if (i > 1 && '\\' != buf[i - 2])
405 			mandoc_msg(MANDOCERR_SPACE_EOL, man->parse,
406 			    line, i - 1, NULL);
407 
408 		for (--i; i && ' ' == buf[i]; i--)
409 			/* Spin back to non-space. */ ;
410 
411 		/* Jump ahead of escaped whitespace. */
412 		i += '\\' == buf[i] ? 2 : 1;
413 
414 		buf[i] = '\0';
415 	}
416 	man_word_alloc(man, line, offs, buf + offs);
417 
418 	/*
419 	 * End-of-sentence check.  If the last character is an unescaped
420 	 * EOS character, then flag the node as being the end of a
421 	 * sentence.  The front-end will know how to interpret this.
422 	 */
423 
424 	assert(i);
425 	if (mandoc_eos(buf, (size_t)i))
426 		man->last->flags |= MAN_EOS;
427 
428 	man_descope(man, line, offs);
429 	return(1);
430 }
431 
432 static int
433 man_pmacro(struct man *man, int ln, char *buf, int offs)
434 {
435 	struct man_node	*n;
436 	const char	*cp;
437 	enum mant	 tok;
438 	int		 i, ppos;
439 	int		 bline;
440 	char		 mac[5];
441 
442 	ppos = offs;
443 
444 	/*
445 	 * Copy the first word into a nil-terminated buffer.
446 	 * Stop when a space, tab, escape, or eoln is encountered.
447 	 */
448 
449 	i = 0;
450 	while (i < 4 && strchr(" \t\\", buf[offs]) == NULL)
451 		mac[i++] = buf[offs++];
452 
453 	mac[i] = '\0';
454 
455 	tok = (i > 0 && i < 4) ? man_hash_find(mac) : MAN_MAX;
456 
457 	if (tok == MAN_MAX) {
458 		mandoc_msg(MANDOCERR_MACRO, man->parse,
459 		    ln, ppos, buf + ppos - 1);
460 		return(1);
461 	}
462 
463 	/* Skip a leading escape sequence or tab. */
464 
465 	switch (buf[offs]) {
466 	case '\\':
467 		cp = buf + offs + 1;
468 		mandoc_escape(&cp, NULL, NULL);
469 		offs = cp - buf;
470 		break;
471 	case '\t':
472 		offs++;
473 		break;
474 	default:
475 		break;
476 	}
477 
478 	/* Jump to the next non-whitespace word. */
479 
480 	while (buf[offs] && buf[offs] == ' ')
481 		offs++;
482 
483 	/*
484 	 * Trailing whitespace.  Note that tabs are allowed to be passed
485 	 * into the parser as "text", so we only warn about spaces here.
486 	 */
487 
488 	if (buf[offs] == '\0' && buf[offs - 1] == ' ')
489 		mandoc_msg(MANDOCERR_SPACE_EOL, man->parse,
490 		    ln, offs - 1, NULL);
491 
492 	/*
493 	 * Remove prior ELINE macro, as it's being clobbered by a new
494 	 * macro.  Note that NSCOPED macros do not close out ELINE
495 	 * macros---they don't print text---so we let those slip by.
496 	 */
497 
498 	if ( ! (man_macros[tok].flags & MAN_NSCOPED) &&
499 			man->flags & MAN_ELINE) {
500 		n = man->last;
501 		assert(MAN_TEXT != n->type);
502 
503 		/* Remove repeated NSCOPED macros causing ELINE. */
504 
505 		if (man_macros[n->tok].flags & MAN_NSCOPED)
506 			n = n->parent;
507 
508 		mandoc_vmsg(MANDOCERR_BLK_LINE, man->parse, n->line,
509 		    n->pos, "%s breaks %s", man_macronames[tok],
510 		    man_macronames[n->tok]);
511 
512 		man_node_delete(man, n);
513 		man->flags &= ~MAN_ELINE;
514 	}
515 
516 	/*
517 	 * Remove prior BLINE macro that is being clobbered.
518 	 */
519 	if ((man->flags & MAN_BLINE) &&
520 	    (man_macros[tok].flags & MAN_BSCOPE)) {
521 		n = man->last;
522 
523 		/* Might be a text node like 8 in
524 		 * .TP 8
525 		 * .SH foo
526 		 */
527 		if (n->type == MAN_TEXT)
528 			n = n->parent;
529 
530 		/* Remove element that didn't end BLINE, if any. */
531 		if ( ! (man_macros[n->tok].flags & MAN_BSCOPE))
532 			n = n->parent;
533 
534 		assert(n->type == MAN_HEAD);
535 		n = n->parent;
536 		assert(n->type == MAN_BLOCK);
537 		assert(man_macros[n->tok].flags & MAN_SCOPED);
538 
539 		mandoc_vmsg(MANDOCERR_BLK_LINE, man->parse, n->line,
540 		    n->pos, "%s breaks %s", man_macronames[tok],
541 		    man_macronames[n->tok]);
542 
543 		man_node_delete(man, n);
544 		man->flags &= ~MAN_BLINE;
545 	}
546 
547 	/* Remember whether we are in next-line scope for a block head. */
548 
549 	bline = man->flags & MAN_BLINE;
550 
551 	/* Call to handler... */
552 
553 	assert(man_macros[tok].fp);
554 	(*man_macros[tok].fp)(man, tok, ln, ppos, &offs, buf);
555 
556 	/* In quick mode (for mandocdb), abort after the NAME section. */
557 
558 	if (man->quick && tok == MAN_SH) {
559 		n = man->last;
560 		if (n->type == MAN_BODY &&
561 		    strcmp(n->prev->child->string, "NAME"))
562 			return(2);
563 	}
564 
565 	/*
566 	 * If we are in a next-line scope for a block head,
567 	 * close it out now and switch to the body,
568 	 * unless the next-line scope is allowed to continue.
569 	 */
570 
571 	if ( ! bline || man->flags & MAN_ELINE ||
572 	    man_macros[tok].flags & MAN_NSCOPED)
573 		return(1);
574 
575 	assert(man->flags & MAN_BLINE);
576 	man->flags &= ~MAN_BLINE;
577 
578 	man_unscope(man, man->last->parent);
579 	man_body_alloc(man, ln, ppos, man->last->tok);
580 	return(1);
581 }
582 
583 /*
584  * Unlink a node from its context.  If "man" is provided, the last parse
585  * point will also be adjusted accordingly.
586  */
587 static void
588 man_node_unlink(struct man *man, struct man_node *n)
589 {
590 
591 	/* Adjust siblings. */
592 
593 	if (n->prev)
594 		n->prev->next = n->next;
595 	if (n->next)
596 		n->next->prev = n->prev;
597 
598 	/* Adjust parent. */
599 
600 	if (n->parent) {
601 		n->parent->nchild--;
602 		if (n->parent->child == n)
603 			n->parent->child = n->prev ? n->prev : n->next;
604 	}
605 
606 	/* Adjust parse point, if applicable. */
607 
608 	if (man && man->last == n) {
609 		/*XXX: this can occur when bailing from validation. */
610 		/*assert(NULL == n->next);*/
611 		if (n->prev) {
612 			man->last = n->prev;
613 			man->next = MAN_NEXT_SIBLING;
614 		} else {
615 			man->last = n->parent;
616 			man->next = MAN_NEXT_CHILD;
617 		}
618 	}
619 
620 	if (man && man->first == n)
621 		man->first = NULL;
622 }
623 
624 const struct mparse *
625 man_mparse(const struct man *man)
626 {
627 
628 	assert(man && man->parse);
629 	return(man->parse);
630 }
631 
632 void
633 man_deroff(char **dest, const struct man_node *n)
634 {
635 	char	*cp;
636 	size_t	 sz;
637 
638 	if (n->type != MAN_TEXT) {
639 		for (n = n->child; n; n = n->next)
640 			man_deroff(dest, n);
641 		return;
642 	}
643 
644 	/* Skip leading whitespace and escape sequences. */
645 
646 	cp = n->string;
647 	while ('\0' != *cp) {
648 		if ('\\' == *cp) {
649 			cp++;
650 			mandoc_escape((const char **)&cp, NULL, NULL);
651 		} else if (isspace((unsigned char)*cp))
652 			cp++;
653 		else
654 			break;
655 	}
656 
657 	/* Skip trailing whitespace. */
658 
659 	for (sz = strlen(cp); sz; sz--)
660 		if (0 == isspace((unsigned char)cp[sz-1]))
661 			break;
662 
663 	/* Skip empty strings. */
664 
665 	if (0 == sz)
666 		return;
667 
668 	if (NULL == *dest) {
669 		*dest = mandoc_strndup(cp, sz);
670 		return;
671 	}
672 
673 	mandoc_asprintf(&cp, "%s %*s", *dest, (int)sz, cp);
674 	free(*dest);
675 	*dest = cp;
676 }
677