xref: /illumos-gate/usr/src/cmd/mandoc/man_validate.c (revision f985abb4a2473d3c04b086f7c9fab177e368ffef)
1 /*	$Id: man_validate.c,v 1.86 2013/10/17 20:54:58 schwarze Exp $ */
2 /*
3  * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
4  * Copyright (c) 2010, 2012, 2013 Ingo Schwarze <schwarze@openbsd.org>
5  *
6  * Permission to use, copy, modify, and distribute this software for any
7  * purpose with or without fee is hereby granted, provided that the above
8  * copyright notice and this permission notice appear in all copies.
9  *
10  * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
11  * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
12  * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
13  * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
14  * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
15  * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
16  * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
17  */
18 #ifdef HAVE_CONFIG_H
19 #include "config.h"
20 #endif
21 
22 #include <sys/types.h>
23 
24 #include <assert.h>
25 #include <ctype.h>
26 #include <errno.h>
27 #include <limits.h>
28 #include <stdarg.h>
29 #include <stdlib.h>
30 #include <string.h>
31 #include <time.h>
32 
33 #include "man.h"
34 #include "mandoc.h"
35 #include "libman.h"
36 #include "libmandoc.h"
37 
38 #define	CHKARGS	  struct man *man, struct man_node *n
39 
40 typedef	int	(*v_check)(CHKARGS);
41 
42 struct	man_valid {
43 	v_check	 *pres;
44 	v_check	 *posts;
45 };
46 
47 static	int	  check_eq0(CHKARGS);
48 static	int	  check_eq2(CHKARGS);
49 static	int	  check_le1(CHKARGS);
50 static	int	  check_ge2(CHKARGS);
51 static	int	  check_le5(CHKARGS);
52 static	int	  check_head1(CHKARGS);
53 static	int	  check_par(CHKARGS);
54 static	int	  check_part(CHKARGS);
55 static	int	  check_root(CHKARGS);
56 static	void	  check_text(CHKARGS);
57 
58 static	int	  post_AT(CHKARGS);
59 static	int	  post_IP(CHKARGS);
60 static	int	  post_vs(CHKARGS);
61 static	int	  post_fi(CHKARGS);
62 static	int	  post_ft(CHKARGS);
63 static	int	  post_nf(CHKARGS);
64 static	int	  post_sec(CHKARGS);
65 static	int	  post_TH(CHKARGS);
66 static	int	  post_UC(CHKARGS);
67 static	int	  pre_sec(CHKARGS);
68 
69 static	v_check	  posts_at[] = { post_AT, NULL };
70 static	v_check	  posts_br[] = { post_vs, check_eq0, NULL };
71 static	v_check	  posts_eq0[] = { check_eq0, NULL };
72 static	v_check	  posts_eq2[] = { check_eq2, NULL };
73 static	v_check	  posts_fi[] = { check_eq0, post_fi, NULL };
74 static	v_check	  posts_ft[] = { post_ft, NULL };
75 static	v_check	  posts_ip[] = { post_IP, NULL };
76 static	v_check	  posts_le1[] = { check_le1, NULL };
77 static	v_check	  posts_nf[] = { check_eq0, post_nf, NULL };
78 static	v_check	  posts_par[] = { check_par, NULL };
79 static	v_check	  posts_part[] = { check_part, NULL };
80 static	v_check	  posts_sec[] = { post_sec, NULL };
81 static	v_check	  posts_sp[] = { post_vs, check_le1, NULL };
82 static	v_check	  posts_th[] = { check_ge2, check_le5, post_TH, NULL };
83 static	v_check	  posts_uc[] = { post_UC, NULL };
84 static	v_check	  posts_ur[] = { check_head1, check_part, NULL };
85 static	v_check	  pres_sec[] = { pre_sec, NULL };
86 
87 static	const struct man_valid man_valids[MAN_MAX] = {
88 	{ NULL, posts_br }, /* br */
89 	{ NULL, posts_th }, /* TH */
90 	{ pres_sec, posts_sec }, /* SH */
91 	{ pres_sec, posts_sec }, /* SS */
92 	{ NULL, NULL }, /* TP */
93 	{ NULL, posts_par }, /* LP */
94 	{ NULL, posts_par }, /* PP */
95 	{ NULL, posts_par }, /* P */
96 	{ NULL, posts_ip }, /* IP */
97 	{ NULL, NULL }, /* HP */
98 	{ NULL, NULL }, /* SM */
99 	{ NULL, NULL }, /* SB */
100 	{ NULL, NULL }, /* BI */
101 	{ NULL, NULL }, /* IB */
102 	{ NULL, NULL }, /* BR */
103 	{ NULL, NULL }, /* RB */
104 	{ NULL, NULL }, /* R */
105 	{ NULL, NULL }, /* B */
106 	{ NULL, NULL }, /* I */
107 	{ NULL, NULL }, /* IR */
108 	{ NULL, NULL }, /* RI */
109 	{ NULL, posts_eq0 }, /* na */
110 	{ NULL, posts_sp }, /* sp */
111 	{ NULL, posts_nf }, /* nf */
112 	{ NULL, posts_fi }, /* fi */
113 	{ NULL, NULL }, /* RE */
114 	{ NULL, posts_part }, /* RS */
115 	{ NULL, NULL }, /* DT */
116 	{ NULL, posts_uc }, /* UC */
117 	{ NULL, posts_le1 }, /* PD */
118 	{ NULL, posts_at }, /* AT */
119 	{ NULL, NULL }, /* in */
120 	{ NULL, posts_ft }, /* ft */
121 	{ NULL, posts_eq2 }, /* OP */
122 	{ NULL, posts_nf }, /* EX */
123 	{ NULL, posts_fi }, /* EE */
124 	{ NULL, posts_ur }, /* UR */
125 	{ NULL, NULL }, /* UE */
126 };
127 
128 
129 int
130 man_valid_pre(struct man *man, struct man_node *n)
131 {
132 	v_check		*cp;
133 
134 	switch (n->type) {
135 	case (MAN_TEXT):
136 		/* FALLTHROUGH */
137 	case (MAN_ROOT):
138 		/* FALLTHROUGH */
139 	case (MAN_EQN):
140 		/* FALLTHROUGH */
141 	case (MAN_TBL):
142 		return(1);
143 	default:
144 		break;
145 	}
146 
147 	if (NULL == (cp = man_valids[n->tok].pres))
148 		return(1);
149 	for ( ; *cp; cp++)
150 		if ( ! (*cp)(man, n))
151 			return(0);
152 	return(1);
153 }
154 
155 
156 int
157 man_valid_post(struct man *man)
158 {
159 	v_check		*cp;
160 
161 	if (MAN_VALID & man->last->flags)
162 		return(1);
163 	man->last->flags |= MAN_VALID;
164 
165 	switch (man->last->type) {
166 	case (MAN_TEXT):
167 		check_text(man, man->last);
168 		return(1);
169 	case (MAN_ROOT):
170 		return(check_root(man, man->last));
171 	case (MAN_EQN):
172 		/* FALLTHROUGH */
173 	case (MAN_TBL):
174 		return(1);
175 	default:
176 		break;
177 	}
178 
179 	if (NULL == (cp = man_valids[man->last->tok].posts))
180 		return(1);
181 	for ( ; *cp; cp++)
182 		if ( ! (*cp)(man, man->last))
183 			return(0);
184 
185 	return(1);
186 }
187 
188 
189 static int
190 check_root(CHKARGS)
191 {
192 
193 	if (MAN_BLINE & man->flags)
194 		man_nmsg(man, n, MANDOCERR_SCOPEEXIT);
195 	else if (MAN_ELINE & man->flags)
196 		man_nmsg(man, n, MANDOCERR_SCOPEEXIT);
197 
198 	man->flags &= ~MAN_BLINE;
199 	man->flags &= ~MAN_ELINE;
200 
201 	if (NULL == man->first->child) {
202 		man_nmsg(man, n, MANDOCERR_NODOCBODY);
203 		return(0);
204 	} else if (NULL == man->meta.title) {
205 		man_nmsg(man, n, MANDOCERR_NOTITLE);
206 
207 		/*
208 		 * If a title hasn't been set, do so now (by
209 		 * implication, date and section also aren't set).
210 		 */
211 
212 	        man->meta.title = mandoc_strdup("unknown");
213 		man->meta.msec = mandoc_strdup("1");
214 		man->meta.date = mandoc_normdate
215 			(man->parse, NULL, n->line, n->pos);
216 	}
217 
218 	return(1);
219 }
220 
221 static void
222 check_text(CHKARGS)
223 {
224 	char		*cp, *p;
225 
226 	if (MAN_LITERAL & man->flags)
227 		return;
228 
229 	cp = n->string;
230 	for (p = cp; NULL != (p = strchr(p, '\t')); p++)
231 		man_pmsg(man, n->line, (int)(p - cp), MANDOCERR_BADTAB);
232 }
233 
234 #define	INEQ_DEFINE(x, ineq, name) \
235 static int \
236 check_##name(CHKARGS) \
237 { \
238 	if (n->nchild ineq (x)) \
239 		return(1); \
240 	mandoc_vmsg(MANDOCERR_ARGCOUNT, man->parse, n->line, n->pos, \
241 			"line arguments %s %d (have %d)", \
242 			#ineq, (x), n->nchild); \
243 	return(1); \
244 }
245 
246 INEQ_DEFINE(0, ==, eq0)
247 INEQ_DEFINE(2, ==, eq2)
248 INEQ_DEFINE(1, <=, le1)
249 INEQ_DEFINE(2, >=, ge2)
250 INEQ_DEFINE(5, <=, le5)
251 
252 static int
253 check_head1(CHKARGS)
254 {
255 
256 	if (MAN_HEAD == n->type && 1 != n->nchild)
257 		mandoc_vmsg(MANDOCERR_ARGCOUNT, man->parse, n->line,
258 		    n->pos, "line arguments eq 1 (have %d)", n->nchild);
259 
260 	return(1);
261 }
262 
263 static int
264 post_ft(CHKARGS)
265 {
266 	char	*cp;
267 	int	 ok;
268 
269 	if (0 == n->nchild)
270 		return(1);
271 
272 	ok = 0;
273 	cp = n->child->string;
274 	switch (*cp) {
275 	case ('1'):
276 		/* FALLTHROUGH */
277 	case ('2'):
278 		/* FALLTHROUGH */
279 	case ('3'):
280 		/* FALLTHROUGH */
281 	case ('4'):
282 		/* FALLTHROUGH */
283 	case ('I'):
284 		/* FALLTHROUGH */
285 	case ('P'):
286 		/* FALLTHROUGH */
287 	case ('R'):
288 		if ('\0' == cp[1])
289 			ok = 1;
290 		break;
291 	case ('B'):
292 		if ('\0' == cp[1] || ('I' == cp[1] && '\0' == cp[2]))
293 			ok = 1;
294 		break;
295 	case ('C'):
296 		if ('W' == cp[1] && '\0' == cp[2])
297 			ok = 1;
298 		break;
299 	default:
300 		break;
301 	}
302 
303 	if (0 == ok) {
304 		mandoc_vmsg
305 			(MANDOCERR_BADFONT, man->parse,
306 			 n->line, n->pos, "%s", cp);
307 		*cp = '\0';
308 	}
309 
310 	if (1 < n->nchild)
311 		mandoc_vmsg
312 			(MANDOCERR_ARGCOUNT, man->parse, n->line,
313 			 n->pos, "want one child (have %d)",
314 			 n->nchild);
315 
316 	return(1);
317 }
318 
319 static int
320 pre_sec(CHKARGS)
321 {
322 
323 	if (MAN_BLOCK == n->type)
324 		man->flags &= ~MAN_LITERAL;
325 	return(1);
326 }
327 
328 static int
329 post_sec(CHKARGS)
330 {
331 
332 	if ( ! (MAN_HEAD == n->type && 0 == n->nchild))
333 		return(1);
334 
335 	man_nmsg(man, n, MANDOCERR_SYNTARGCOUNT);
336 	return(0);
337 }
338 
339 static int
340 check_part(CHKARGS)
341 {
342 
343 	if (MAN_BODY == n->type && 0 == n->nchild)
344 		mandoc_msg(MANDOCERR_ARGCWARN, man->parse, n->line,
345 				n->pos, "want children (have none)");
346 
347 	return(1);
348 }
349 
350 
351 static int
352 check_par(CHKARGS)
353 {
354 
355 	switch (n->type) {
356 	case (MAN_BLOCK):
357 		if (0 == n->body->nchild)
358 			man_node_delete(man, n);
359 		break;
360 	case (MAN_BODY):
361 		if (0 == n->nchild)
362 			man_nmsg(man, n, MANDOCERR_IGNPAR);
363 		break;
364 	case (MAN_HEAD):
365 		if (n->nchild)
366 			man_nmsg(man, n, MANDOCERR_ARGSLOST);
367 		break;
368 	default:
369 		break;
370 	}
371 
372 	return(1);
373 }
374 
375 static int
376 post_IP(CHKARGS)
377 {
378 
379 	switch (n->type) {
380 	case (MAN_BLOCK):
381 		if (0 == n->head->nchild && 0 == n->body->nchild)
382 			man_node_delete(man, n);
383 		break;
384 	case (MAN_BODY):
385 		if (0 == n->parent->head->nchild && 0 == n->nchild)
386 			man_nmsg(man, n, MANDOCERR_IGNPAR);
387 		break;
388 	default:
389 		break;
390 	}
391 	return(1);
392 }
393 
394 static int
395 post_TH(CHKARGS)
396 {
397 	const char	*p;
398 	int		 line, pos;
399 
400 	free(man->meta.title);
401 	free(man->meta.vol);
402 	free(man->meta.source);
403 	free(man->meta.msec);
404 	free(man->meta.date);
405 
406 	line = n->line;
407 	pos = n->pos;
408 	man->meta.title = man->meta.vol = man->meta.date =
409 		man->meta.msec = man->meta.source = NULL;
410 
411 	/* ->TITLE<- MSEC DATE SOURCE VOL */
412 
413 	n = n->child;
414 	if (n && n->string) {
415 		for (p = n->string; '\0' != *p; p++) {
416 			/* Only warn about this once... */
417 			if (isalpha((unsigned char)*p) &&
418 					! isupper((unsigned char)*p)) {
419 				man_nmsg(man, n, MANDOCERR_UPPERCASE);
420 				break;
421 			}
422 		}
423 		man->meta.title = mandoc_strdup(n->string);
424 	} else
425 		man->meta.title = mandoc_strdup("");
426 
427 	/* TITLE ->MSEC<- DATE SOURCE VOL */
428 
429 	if (n)
430 		n = n->next;
431 	if (n && n->string)
432 		man->meta.msec = mandoc_strdup(n->string);
433 	else
434 		man->meta.msec = mandoc_strdup("");
435 
436 	/* TITLE MSEC ->DATE<- SOURCE VOL */
437 
438 	if (n)
439 		n = n->next;
440 	if (n && n->string && '\0' != n->string[0]) {
441 		pos = n->pos;
442 		man->meta.date = mandoc_normdate
443 		    (man->parse, n->string, line, pos);
444 	} else
445 		man->meta.date = mandoc_strdup("");
446 
447 	/* TITLE MSEC DATE ->SOURCE<- VOL */
448 
449 	if (n && (n = n->next))
450 		man->meta.source = mandoc_strdup(n->string);
451 
452 	/* TITLE MSEC DATE SOURCE ->VOL<- */
453 	/* If missing, use the default VOL name for MSEC. */
454 
455 	if (n && (n = n->next))
456 		man->meta.vol = mandoc_strdup(n->string);
457 	else if ('\0' != man->meta.msec[0] &&
458 	    (NULL != (p = mandoc_a2msec(man->meta.msec))))
459 		man->meta.vol = mandoc_strdup(p);
460 
461 	/*
462 	 * Remove the `TH' node after we've processed it for our
463 	 * meta-data.
464 	 */
465 	man_node_delete(man, man->last);
466 	return(1);
467 }
468 
469 static int
470 post_nf(CHKARGS)
471 {
472 
473 	if (MAN_LITERAL & man->flags)
474 		man_nmsg(man, n, MANDOCERR_SCOPEREP);
475 
476 	man->flags |= MAN_LITERAL;
477 	return(1);
478 }
479 
480 static int
481 post_fi(CHKARGS)
482 {
483 
484 	if ( ! (MAN_LITERAL & man->flags))
485 		man_nmsg(man, n, MANDOCERR_WNOSCOPE);
486 
487 	man->flags &= ~MAN_LITERAL;
488 	return(1);
489 }
490 
491 static int
492 post_UC(CHKARGS)
493 {
494 	static const char * const bsd_versions[] = {
495 	    "3rd Berkeley Distribution",
496 	    "4th Berkeley Distribution",
497 	    "4.2 Berkeley Distribution",
498 	    "4.3 Berkeley Distribution",
499 	    "4.4 Berkeley Distribution",
500 	};
501 
502 	const char	*p, *s;
503 
504 	n = n->child;
505 
506 	if (NULL == n || MAN_TEXT != n->type)
507 		p = bsd_versions[0];
508 	else {
509 		s = n->string;
510 		if (0 == strcmp(s, "3"))
511 			p = bsd_versions[0];
512 		else if (0 == strcmp(s, "4"))
513 			p = bsd_versions[1];
514 		else if (0 == strcmp(s, "5"))
515 			p = bsd_versions[2];
516 		else if (0 == strcmp(s, "6"))
517 			p = bsd_versions[3];
518 		else if (0 == strcmp(s, "7"))
519 			p = bsd_versions[4];
520 		else
521 			p = bsd_versions[0];
522 	}
523 
524 	free(man->meta.source);
525 	man->meta.source = mandoc_strdup(p);
526 	return(1);
527 }
528 
529 static int
530 post_AT(CHKARGS)
531 {
532 	static const char * const unix_versions[] = {
533 	    "7th Edition",
534 	    "System III",
535 	    "System V",
536 	    "System V Release 2",
537 	};
538 
539 	const char	*p, *s;
540 	struct man_node	*nn;
541 
542 	n = n->child;
543 
544 	if (NULL == n || MAN_TEXT != n->type)
545 		p = unix_versions[0];
546 	else {
547 		s = n->string;
548 		if (0 == strcmp(s, "3"))
549 			p = unix_versions[0];
550 		else if (0 == strcmp(s, "4"))
551 			p = unix_versions[1];
552 		else if (0 == strcmp(s, "5")) {
553 			nn = n->next;
554 			if (nn && MAN_TEXT == nn->type && nn->string[0])
555 				p = unix_versions[3];
556 			else
557 				p = unix_versions[2];
558 		} else
559 			p = unix_versions[0];
560 	}
561 
562 	free(man->meta.source);
563 	man->meta.source = mandoc_strdup(p);
564 	return(1);
565 }
566 
567 static int
568 post_vs(CHKARGS)
569 {
570 
571 	if (NULL != n->prev)
572 		return(1);
573 
574 	switch (n->parent->tok) {
575 	case (MAN_SH):
576 		/* FALLTHROUGH */
577 	case (MAN_SS):
578 		man_nmsg(man, n, MANDOCERR_IGNPAR);
579 		/* FALLTHROUGH */
580 	case (MAN_MAX):
581 		/*
582 		 * Don't warn about this because it occurs in pod2man
583 		 * and would cause considerable (unfixable) warnage.
584 		 */
585 		man_node_delete(man, n);
586 		break;
587 	default:
588 		break;
589 	}
590 
591 	return(1);
592 }
593