xref: /dragonfly/contrib/mdocml/man_validate.c (revision 52f9f0d9)
1 /*	$Id: man_validate.c,v 1.69 2011/04/13 09:57:08 kristaps Exp $ */
2 /*
3  * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
4  * Copyright (c) 2010 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 *m, 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_bline(CHKARGS);
48 static	int	  check_eq0(CHKARGS);
49 static	int	  check_ft(CHKARGS);
50 static	int	  check_le1(CHKARGS);
51 static	int	  check_ge2(CHKARGS);
52 static	int	  check_le5(CHKARGS);
53 static	int	  check_par(CHKARGS);
54 static	int	  check_part(CHKARGS);
55 static	int	  check_root(CHKARGS);
56 static	int	  check_sec(CHKARGS);
57 static	void	  check_text(CHKARGS);
58 
59 static	int	  post_AT(CHKARGS);
60 static	int	  post_fi(CHKARGS);
61 static	int	  post_nf(CHKARGS);
62 static	int	  post_TH(CHKARGS);
63 static	int	  post_UC(CHKARGS);
64 
65 static	v_check	  posts_at[] = { post_AT, NULL };
66 static	v_check	  posts_eq0[] = { check_eq0, NULL };
67 static	v_check	  posts_fi[] = { check_eq0, post_fi, NULL };
68 static	v_check	  posts_le1[] = { check_le1, NULL };
69 static	v_check	  posts_ft[] = { check_ft, NULL };
70 static	v_check	  posts_nf[] = { check_eq0, post_nf, NULL };
71 static	v_check	  posts_par[] = { check_par, NULL };
72 static	v_check	  posts_part[] = { check_part, NULL };
73 static	v_check	  posts_sec[] = { check_sec, NULL };
74 static	v_check	  posts_th[] = { check_ge2, check_le5, post_TH, NULL };
75 static	v_check	  posts_uc[] = { post_UC, NULL };
76 static	v_check	  pres_bline[] = { check_bline, NULL };
77 
78 
79 static	const struct man_valid man_valids[MAN_MAX] = {
80 	{ NULL, posts_eq0 }, /* br */
81 	{ pres_bline, posts_th }, /* TH */
82 	{ pres_bline, posts_sec }, /* SH */
83 	{ pres_bline, posts_sec }, /* SS */
84 	{ pres_bline, NULL }, /* TP */
85 	{ pres_bline, posts_par }, /* LP */
86 	{ pres_bline, posts_par }, /* PP */
87 	{ pres_bline, posts_par }, /* P */
88 	{ pres_bline, NULL }, /* IP */
89 	{ pres_bline, NULL }, /* HP */
90 	{ NULL, NULL }, /* SM */
91 	{ NULL, NULL }, /* SB */
92 	{ NULL, NULL }, /* BI */
93 	{ NULL, NULL }, /* IB */
94 	{ NULL, NULL }, /* BR */
95 	{ NULL, NULL }, /* RB */
96 	{ NULL, NULL }, /* R */
97 	{ NULL, NULL }, /* B */
98 	{ NULL, NULL }, /* I */
99 	{ NULL, NULL }, /* IR */
100 	{ NULL, NULL }, /* RI */
101 	{ NULL, posts_eq0 }, /* na */ /* FIXME: should warn only. */
102 	{ NULL, posts_le1 }, /* sp */ /* FIXME: should warn only. */
103 	{ pres_bline, posts_nf }, /* nf */
104 	{ pres_bline, posts_fi }, /* fi */
105 	{ NULL, NULL }, /* RE */
106 	{ NULL, posts_part }, /* RS */
107 	{ NULL, NULL }, /* DT */
108 	{ NULL, posts_uc }, /* UC */
109 	{ NULL, NULL }, /* PD */
110 	{ NULL, posts_at }, /* AT */
111 	{ NULL, NULL }, /* in */
112 	{ NULL, posts_ft }, /* ft */
113 };
114 
115 
116 int
117 man_valid_pre(struct man *m, struct man_node *n)
118 {
119 	v_check		*cp;
120 
121 	switch (n->type) {
122 	case (MAN_TEXT):
123 		/* FALLTHROUGH */
124 	case (MAN_ROOT):
125 		/* FALLTHROUGH */
126 	case (MAN_EQN):
127 		/* FALLTHROUGH */
128 	case (MAN_TBL):
129 		return(1);
130 	default:
131 		break;
132 	}
133 
134 	if (NULL == (cp = man_valids[n->tok].pres))
135 		return(1);
136 	for ( ; *cp; cp++)
137 		if ( ! (*cp)(m, n))
138 			return(0);
139 	return(1);
140 }
141 
142 
143 int
144 man_valid_post(struct man *m)
145 {
146 	v_check		*cp;
147 
148 	if (MAN_VALID & m->last->flags)
149 		return(1);
150 	m->last->flags |= MAN_VALID;
151 
152 	switch (m->last->type) {
153 	case (MAN_TEXT):
154 		check_text(m, m->last);
155 		return(1);
156 	case (MAN_ROOT):
157 		return(check_root(m, m->last));
158 	case (MAN_EQN):
159 		/* FALLTHROUGH */
160 	case (MAN_TBL):
161 		return(1);
162 	default:
163 		break;
164 	}
165 
166 	if (NULL == (cp = man_valids[m->last->tok].posts))
167 		return(1);
168 	for ( ; *cp; cp++)
169 		if ( ! (*cp)(m, m->last))
170 			return(0);
171 
172 	return(1);
173 }
174 
175 
176 static int
177 check_root(CHKARGS)
178 {
179 
180 	if (MAN_BLINE & m->flags)
181 		man_nmsg(m, n, MANDOCERR_SCOPEEXIT);
182 	else if (MAN_ELINE & m->flags)
183 		man_nmsg(m, n, MANDOCERR_SCOPEEXIT);
184 
185 	m->flags &= ~MAN_BLINE;
186 	m->flags &= ~MAN_ELINE;
187 
188 	if (NULL == m->first->child) {
189 		man_nmsg(m, n, MANDOCERR_NODOCBODY);
190 		return(0);
191 	} else if (NULL == m->meta.title) {
192 		man_nmsg(m, n, MANDOCERR_NOTITLE);
193 
194 		/*
195 		 * If a title hasn't been set, do so now (by
196 		 * implication, date and section also aren't set).
197 		 */
198 
199 	        m->meta.title = mandoc_strdup("unknown");
200 		m->meta.msec = mandoc_strdup("1");
201 		m->meta.date = mandoc_normdate
202 			(m->parse, NULL, n->line, n->pos);
203 	}
204 
205 	return(1);
206 }
207 
208 static void
209 check_text(CHKARGS)
210 {
211 	char		*p, *pp, *cpp;
212 	int		 pos;
213 	size_t		 sz;
214 
215 	p = n->string;
216 	pos = n->pos + 1;
217 
218 	while ('\0' != *p) {
219 		sz = strcspn(p, "\t\\");
220 
221 		p += (int)sz;
222 		pos += (int)sz;
223 
224 		if ('\t' == *p) {
225 			if ( ! (MAN_LITERAL & m->flags))
226 				man_pmsg(m, n->line, pos, MANDOCERR_BADTAB);
227 			p++;
228 			pos++;
229 			continue;
230 		} else if ('\0' == *p)
231 			break;
232 
233 		pos++;
234 		pp = ++p;
235 
236 		if (ESCAPE_ERROR == mandoc_escape
237 				((const char **)&pp, NULL, NULL)) {
238 			man_pmsg(m, n->line, pos, MANDOCERR_BADESCAPE);
239 			break;
240 		}
241 
242 		cpp = p;
243 		while (NULL != (cpp = memchr(cpp, ASCII_HYPH, pp - cpp)))
244 			*cpp = '-';
245 
246 		pos += pp - p;
247 		p = pp;
248 	}
249 }
250 
251 #define	INEQ_DEFINE(x, ineq, name) \
252 static int \
253 check_##name(CHKARGS) \
254 { \
255 	if (n->nchild ineq (x)) \
256 		return(1); \
257 	mandoc_vmsg(MANDOCERR_ARGCOUNT, m->parse, n->line, n->pos, \
258 			"line arguments %s %d (have %d)", \
259 			#ineq, (x), n->nchild); \
260 	return(1); \
261 }
262 
263 INEQ_DEFINE(0, ==, eq0)
264 INEQ_DEFINE(1, <=, le1)
265 INEQ_DEFINE(2, >=, ge2)
266 INEQ_DEFINE(5, <=, le5)
267 
268 static int
269 check_ft(CHKARGS)
270 {
271 	char	*cp;
272 	int	 ok;
273 
274 	if (0 == n->nchild)
275 		return(1);
276 
277 	ok = 0;
278 	cp = n->child->string;
279 	switch (*cp) {
280 	case ('1'):
281 		/* FALLTHROUGH */
282 	case ('2'):
283 		/* FALLTHROUGH */
284 	case ('3'):
285 		/* FALLTHROUGH */
286 	case ('4'):
287 		/* FALLTHROUGH */
288 	case ('I'):
289 		/* FALLTHROUGH */
290 	case ('P'):
291 		/* FALLTHROUGH */
292 	case ('R'):
293 		if ('\0' == cp[1])
294 			ok = 1;
295 		break;
296 	case ('B'):
297 		if ('\0' == cp[1] || ('I' == cp[1] && '\0' == cp[2]))
298 			ok = 1;
299 		break;
300 	case ('C'):
301 		if ('W' == cp[1] && '\0' == cp[2])
302 			ok = 1;
303 		break;
304 	default:
305 		break;
306 	}
307 
308 	if (0 == ok) {
309 		mandoc_vmsg
310 			(MANDOCERR_BADFONT, m->parse,
311 			 n->line, n->pos, "%s", cp);
312 		*cp = '\0';
313 	}
314 
315 	if (1 < n->nchild)
316 		mandoc_vmsg
317 			(MANDOCERR_ARGCOUNT, m->parse, n->line,
318 			 n->pos, "want one child (have %d)",
319 			 n->nchild);
320 
321 	return(1);
322 }
323 
324 static int
325 check_sec(CHKARGS)
326 {
327 
328 	if ( ! (MAN_HEAD == n->type && 0 == n->nchild))
329 		return(1);
330 
331 	man_nmsg(m, n, MANDOCERR_SYNTARGCOUNT);
332 	return(0);
333 }
334 
335 
336 static int
337 check_part(CHKARGS)
338 {
339 
340 	if (MAN_BODY == n->type && 0 == n->nchild)
341 		mandoc_msg(MANDOCERR_ARGCWARN, m->parse, n->line,
342 				n->pos, "want children (have none)");
343 
344 	return(1);
345 }
346 
347 
348 static int
349 check_par(CHKARGS)
350 {
351 
352 	switch (n->type) {
353 	case (MAN_BLOCK):
354 		if (0 == n->body->nchild)
355 			man_node_delete(m, n);
356 		break;
357 	case (MAN_BODY):
358 		if (0 == n->nchild)
359 			man_nmsg(m, n, MANDOCERR_IGNPAR);
360 		break;
361 	case (MAN_HEAD):
362 		if (n->nchild)
363 			man_nmsg(m, n, MANDOCERR_ARGSLOST);
364 		break;
365 	default:
366 		break;
367 	}
368 
369 	return(1);
370 }
371 
372 
373 static int
374 check_bline(CHKARGS)
375 {
376 
377 	assert( ! (MAN_ELINE & m->flags));
378 	if (MAN_BLINE & m->flags) {
379 		man_nmsg(m, n, MANDOCERR_SYNTLINESCOPE);
380 		return(0);
381 	}
382 
383 	return(1);
384 }
385 
386 static int
387 post_TH(CHKARGS)
388 {
389 	const char	*p;
390 	int		 line, pos;
391 
392 	if (m->meta.title)
393 		free(m->meta.title);
394 	if (m->meta.vol)
395 		free(m->meta.vol);
396 	if (m->meta.source)
397 		free(m->meta.source);
398 	if (m->meta.msec)
399 		free(m->meta.msec);
400 	if (m->meta.date)
401 		free(m->meta.date);
402 
403 	line = n->line;
404 	pos = n->pos;
405 	m->meta.title = m->meta.vol = m->meta.date =
406 		m->meta.msec = m->meta.source = NULL;
407 
408 	/* ->TITLE<- MSEC DATE SOURCE VOL */
409 
410 	n = n->child;
411 	if (n && n->string) {
412 		for (p = n->string; '\0' != *p; p++) {
413 			/* Only warn about this once... */
414 			if (isalpha((u_char)*p) && ! isupper((u_char)*p)) {
415 				man_nmsg(m, n, MANDOCERR_UPPERCASE);
416 				break;
417 			}
418 		}
419 		m->meta.title = mandoc_strdup(n->string);
420 	} else
421 		m->meta.title = mandoc_strdup("");
422 
423 	/* TITLE ->MSEC<- DATE SOURCE VOL */
424 
425 	if (n)
426 		n = n->next;
427 	if (n && n->string)
428 		m->meta.msec = mandoc_strdup(n->string);
429 	else
430 		m->meta.msec = mandoc_strdup("");
431 
432 	/* TITLE MSEC ->DATE<- SOURCE VOL */
433 
434 	if (n)
435 		n = n->next;
436 	if (n)
437 		pos = n->pos;
438 	m->meta.date = mandoc_normdate
439 		(m->parse, n ? n->string : NULL, line, pos);
440 
441 	/* TITLE MSEC DATE ->SOURCE<- VOL */
442 
443 	if (n && (n = n->next))
444 		m->meta.source = mandoc_strdup(n->string);
445 
446 	/* TITLE MSEC DATE SOURCE ->VOL<- */
447 
448 	if (n && (n = n->next))
449 		m->meta.vol = mandoc_strdup(n->string);
450 
451 	/*
452 	 * Remove the `TH' node after we've processed it for our
453 	 * meta-data.
454 	 */
455 	man_node_delete(m, m->last);
456 	return(1);
457 }
458 
459 static int
460 post_nf(CHKARGS)
461 {
462 
463 	if (MAN_LITERAL & m->flags)
464 		man_nmsg(m, n, MANDOCERR_SCOPEREP);
465 
466 	m->flags |= MAN_LITERAL;
467 	return(1);
468 }
469 
470 static int
471 post_fi(CHKARGS)
472 {
473 
474 	if ( ! (MAN_LITERAL & m->flags))
475 		man_nmsg(m, n, MANDOCERR_WNOSCOPE);
476 
477 	m->flags &= ~MAN_LITERAL;
478 	return(1);
479 }
480 
481 static int
482 post_UC(CHKARGS)
483 {
484 	static const char * const bsd_versions[] = {
485 	    "3rd Berkeley Distribution",
486 	    "4th Berkeley Distribution",
487 	    "4.2 Berkeley Distribution",
488 	    "4.3 Berkeley Distribution",
489 	    "4.4 Berkeley Distribution",
490 	};
491 
492 	const char	*p, *s;
493 
494 	n = n->child;
495 	n = m->last->child;
496 
497 	if (NULL == n || MAN_TEXT != n->type)
498 		p = bsd_versions[0];
499 	else {
500 		s = n->string;
501 		if (0 == strcmp(s, "3"))
502 			p = bsd_versions[0];
503 		else if (0 == strcmp(s, "4"))
504 			p = bsd_versions[1];
505 		else if (0 == strcmp(s, "5"))
506 			p = bsd_versions[2];
507 		else if (0 == strcmp(s, "6"))
508 			p = bsd_versions[3];
509 		else if (0 == strcmp(s, "7"))
510 			p = bsd_versions[4];
511 		else
512 			p = bsd_versions[0];
513 	}
514 
515 	if (m->meta.source)
516 		free(m->meta.source);
517 
518 	m->meta.source = mandoc_strdup(p);
519 	return(1);
520 }
521 
522 static int
523 post_AT(CHKARGS)
524 {
525 	static const char * const unix_versions[] = {
526 	    "7th Edition",
527 	    "System III",
528 	    "System V",
529 	    "System V Release 2",
530 	};
531 
532 	const char	*p, *s;
533 	struct man_node	*nn;
534 
535 	n = n->child;
536 
537 	if (NULL == n || MAN_TEXT != n->type)
538 		p = unix_versions[0];
539 	else {
540 		s = n->string;
541 		if (0 == strcmp(s, "3"))
542 			p = unix_versions[0];
543 		else if (0 == strcmp(s, "4"))
544 			p = unix_versions[1];
545 		else if (0 == strcmp(s, "5")) {
546 			nn = n->next;
547 			if (nn && MAN_TEXT == nn->type && nn->string[0])
548 				p = unix_versions[3];
549 			else
550 				p = unix_versions[2];
551 		} else
552 			p = unix_versions[0];
553 	}
554 
555 	if (m->meta.source)
556 		free(m->meta.source);
557 
558 	m->meta.source = mandoc_strdup(p);
559 	return(1);
560 }
561