1/* $Id: man_validate.c,v 1.156 2021/08/10 12:55:03 schwarze Exp $ */
2/*
3 * Copyright (c) 2010, 2012-2020 Ingo Schwarze <schwarze@openbsd.org>
4 * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
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 AUTHORS DISCLAIM ALL WARRANTIES
11 * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
12 * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHORS 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 * Validation module for man(7) syntax trees used by mandoc(1).
19 */
20#include "config.h"
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 <stdio.h>
30#include <stdlib.h>
31#include <string.h>
32#include <time.h>
33
34#include "mandoc_aux.h"
35#include "mandoc.h"
36#include "roff.h"
37#include "man.h"
38#include "libmandoc.h"
39#include "roff_int.h"
40#include "libman.h"
41#include "tag.h"
42
43#define	CHKARGS	  struct roff_man *man, struct roff_node *n
44
45typedef	void	(*v_check)(CHKARGS);
46
47static	void	  check_abort(CHKARGS) __attribute__((__noreturn__));
48static	void	  check_par(CHKARGS);
49static	void	  check_part(CHKARGS);
50static	void	  check_root(CHKARGS);
51static	void	  check_tag(struct roff_node *, struct roff_node *);
52static	void	  check_text(CHKARGS);
53
54static	void	  post_AT(CHKARGS);
55static	void	  post_EE(CHKARGS);
56static	void	  post_EX(CHKARGS);
57static	void	  post_IP(CHKARGS);
58static	void	  post_OP(CHKARGS);
59static	void	  post_SH(CHKARGS);
60static	void	  post_TH(CHKARGS);
61static	void	  post_TP(CHKARGS);
62static	void	  post_UC(CHKARGS);
63static	void	  post_UR(CHKARGS);
64static	void	  post_in(CHKARGS);
65
66static	const v_check man_valids[MAN_MAX - MAN_TH] = {
67	post_TH,    /* TH */
68	post_SH,    /* SH */
69	post_SH,    /* SS */
70	post_TP,    /* TP */
71	post_TP,    /* TQ */
72	check_abort,/* LP */
73	check_par,  /* PP */
74	check_abort,/* P */
75	post_IP,    /* IP */
76	NULL,       /* HP */
77	NULL,       /* SM */
78	NULL,       /* SB */
79	NULL,       /* BI */
80	NULL,       /* IB */
81	NULL,       /* BR */
82	NULL,       /* RB */
83	NULL,       /* R */
84	NULL,       /* B */
85	NULL,       /* I */
86	NULL,       /* IR */
87	NULL,       /* RI */
88	NULL,       /* RE */
89	check_part, /* RS */
90	NULL,       /* DT */
91	post_UC,    /* UC */
92	NULL,       /* PD */
93	post_AT,    /* AT */
94	post_in,    /* in */
95	NULL,       /* SY */
96	NULL,       /* YS */
97	post_OP,    /* OP */
98	post_EX,    /* EX */
99	post_EE,    /* EE */
100	post_UR,    /* UR */
101	NULL,       /* UE */
102	post_UR,    /* MT */
103	NULL,       /* ME */
104};
105
106
107/* Validate the subtree rooted at man->last. */
108void
109man_validate(struct roff_man *man)
110{
111	struct roff_node *n;
112	const v_check	 *cp;
113
114	/*
115	 * Translate obsolete macros such that later code
116	 * does not need to look for them.
117	 */
118
119	n = man->last;
120	switch (n->tok) {
121	case MAN_LP:
122	case MAN_P:
123		n->tok = MAN_PP;
124		break;
125	default:
126		break;
127	}
128
129	/*
130	 * Iterate over all children, recursing into each one
131	 * in turn, depth-first.
132	 */
133
134	man->last = man->last->child;
135	while (man->last != NULL) {
136		man_validate(man);
137		if (man->last == n)
138			man->last = man->last->child;
139		else
140			man->last = man->last->next;
141	}
142
143	/* Finally validate the macro itself. */
144
145	man->last = n;
146	man->next = ROFF_NEXT_SIBLING;
147	switch (n->type) {
148	case ROFFT_TEXT:
149		check_text(man, n);
150		break;
151	case ROFFT_ROOT:
152		check_root(man, n);
153		break;
154	case ROFFT_COMMENT:
155	case ROFFT_EQN:
156	case ROFFT_TBL:
157		break;
158	default:
159		if (n->tok < ROFF_MAX) {
160			roff_validate(man);
161			break;
162		}
163		assert(n->tok >= MAN_TH && n->tok < MAN_MAX);
164		cp = man_valids + (n->tok - MAN_TH);
165		if (*cp)
166			(*cp)(man, n);
167		if (man->last == n)
168			n->flags |= NODE_VALID;
169		break;
170	}
171}
172
173static void
174check_root(CHKARGS)
175{
176	assert((man->flags & (MAN_BLINE | MAN_ELINE)) == 0);
177
178	if (n->last == NULL || n->last->type == ROFFT_COMMENT)
179		mandoc_msg(MANDOCERR_DOC_EMPTY, n->line, n->pos, NULL);
180	else
181		man->meta.hasbody = 1;
182
183	if (NULL == man->meta.title) {
184		mandoc_msg(MANDOCERR_TH_NOTITLE, n->line, n->pos, NULL);
185
186		/*
187		 * If a title hasn't been set, do so now (by
188		 * implication, date and section also aren't set).
189		 */
190
191		man->meta.title = mandoc_strdup("");
192		man->meta.msec = mandoc_strdup("");
193		man->meta.date = mandoc_normdate(NULL, NULL);
194	}
195
196	if (man->meta.os_e &&
197	    (man->meta.rcsids & (1 << man->meta.os_e)) == 0)
198		mandoc_msg(MANDOCERR_RCS_MISSING, 0, 0,
199		    man->meta.os_e == MANDOC_OS_OPENBSD ?
200		    "(OpenBSD)" : "(NetBSD)");
201}
202
203static void
204check_abort(CHKARGS)
205{
206	abort();
207}
208
209/*
210 * Skip leading whitespace, dashes, backslashes, and font escapes,
211 * then create a tag if the first following byte is a letter.
212 * Priority is high unless whitespace is present.
213 */
214static void
215check_tag(struct roff_node *n, struct roff_node *nt)
216{
217	const char	*cp, *arg;
218	int		 prio, sz;
219
220	if (nt == NULL || nt->type != ROFFT_TEXT)
221		return;
222
223	cp = nt->string;
224	prio = TAG_STRONG;
225	for (;;) {
226		switch (*cp) {
227		case ' ':
228		case '\t':
229			prio = TAG_WEAK;
230			/* FALLTHROUGH */
231		case '-':
232			cp++;
233			break;
234		case '\\':
235			cp++;
236			switch (mandoc_escape(&cp, &arg, &sz)) {
237			case ESCAPE_FONT:
238			case ESCAPE_FONTBOLD:
239			case ESCAPE_FONTITALIC:
240			case ESCAPE_FONTBI:
241			case ESCAPE_FONTROMAN:
242			case ESCAPE_FONTCR:
243			case ESCAPE_FONTCB:
244			case ESCAPE_FONTCI:
245			case ESCAPE_FONTPREV:
246			case ESCAPE_IGNORE:
247				break;
248			case ESCAPE_SPECIAL:
249				if (sz != 1)
250					return;
251				switch (*arg) {
252				case '-':
253				case 'e':
254					break;
255				default:
256					return;
257				}
258				break;
259			default:
260				return;
261			}
262			break;
263		default:
264			if (isalpha((unsigned char)*cp))
265				tag_put(cp, prio, n);
266			return;
267		}
268	}
269}
270
271static void
272check_text(CHKARGS)
273{
274	char		*cp, *p;
275
276	if (n->flags & NODE_NOFILL)
277		return;
278
279	cp = n->string;
280	for (p = cp; NULL != (p = strchr(p, '\t')); p++)
281		mandoc_msg(MANDOCERR_FI_TAB,
282		    n->line, n->pos + (int)(p - cp), NULL);
283}
284
285static void
286post_EE(CHKARGS)
287{
288	if ((n->flags & NODE_NOFILL) == 0)
289		mandoc_msg(MANDOCERR_FI_SKIP, n->line, n->pos, "EE");
290}
291
292static void
293post_EX(CHKARGS)
294{
295	if (n->flags & NODE_NOFILL)
296		mandoc_msg(MANDOCERR_NF_SKIP, n->line, n->pos, "EX");
297}
298
299static void
300post_OP(CHKARGS)
301{
302
303	if (n->child == NULL)
304		mandoc_msg(MANDOCERR_OP_EMPTY, n->line, n->pos, "OP");
305	else if (n->child->next != NULL && n->child->next->next != NULL) {
306		n = n->child->next->next;
307		mandoc_msg(MANDOCERR_ARG_EXCESS,
308		    n->line, n->pos, "OP ... %s", n->string);
309	}
310}
311
312static void
313post_SH(CHKARGS)
314{
315	struct roff_node	*nc;
316	char			*cp, *tag;
317
318	nc = n->child;
319	switch (n->type) {
320	case ROFFT_HEAD:
321		tag = NULL;
322		deroff(&tag, n);
323		if (tag != NULL) {
324			for (cp = tag; *cp != '\0'; cp++)
325				if (*cp == ' ')
326					*cp = '_';
327			if (nc != NULL && nc->type == ROFFT_TEXT &&
328			    strcmp(nc->string, tag) == 0)
329				tag_put(NULL, TAG_STRONG, n);
330			else
331				tag_put(tag, TAG_FALLBACK, n);
332			free(tag);
333		}
334		return;
335	case ROFFT_BODY:
336		if (nc != NULL)
337			break;
338		return;
339	default:
340		return;
341	}
342
343	if (nc->tok == MAN_PP && nc->body->child != NULL) {
344		while (nc->body->last != NULL) {
345			man->next = ROFF_NEXT_CHILD;
346			roff_node_relink(man, nc->body->last);
347			man->last = n;
348		}
349	}
350
351	if (nc->tok == MAN_PP || nc->tok == ROFF_sp || nc->tok == ROFF_br) {
352		mandoc_msg(MANDOCERR_PAR_SKIP, nc->line, nc->pos,
353		    "%s after %s", roff_name[nc->tok], roff_name[n->tok]);
354		roff_node_delete(man, nc);
355	}
356
357	/*
358	 * Trailing PP is empty, so it is deleted by check_par().
359	 * Trailing sp is significant.
360	 */
361
362	if ((nc = n->last) != NULL && nc->tok == ROFF_br) {
363		mandoc_msg(MANDOCERR_PAR_SKIP,
364		    nc->line, nc->pos, "%s at the end of %s",
365		    roff_name[nc->tok], roff_name[n->tok]);
366		roff_node_delete(man, nc);
367	}
368}
369
370static void
371post_UR(CHKARGS)
372{
373	if (n->type == ROFFT_HEAD && n->child == NULL)
374		mandoc_msg(MANDOCERR_UR_NOHEAD, n->line, n->pos,
375		    "%s", roff_name[n->tok]);
376	check_part(man, n);
377}
378
379static void
380check_part(CHKARGS)
381{
382
383	if (n->type == ROFFT_BODY && n->child == NULL)
384		mandoc_msg(MANDOCERR_BLK_EMPTY, n->line, n->pos,
385		    "%s", roff_name[n->tok]);
386}
387
388static void
389check_par(CHKARGS)
390{
391
392	switch (n->type) {
393	case ROFFT_BLOCK:
394		if (n->body->child == NULL)
395			roff_node_delete(man, n);
396		break;
397	case ROFFT_BODY:
398		if (n->child != NULL &&
399		    (n->child->tok == ROFF_sp || n->child->tok == ROFF_br)) {
400			mandoc_msg(MANDOCERR_PAR_SKIP,
401			    n->child->line, n->child->pos,
402			    "%s after %s", roff_name[n->child->tok],
403			    roff_name[n->tok]);
404			roff_node_delete(man, n->child);
405		}
406		if (n->child == NULL)
407			mandoc_msg(MANDOCERR_PAR_SKIP, n->line, n->pos,
408			    "%s empty", roff_name[n->tok]);
409		break;
410	case ROFFT_HEAD:
411		if (n->child != NULL)
412			mandoc_msg(MANDOCERR_ARG_SKIP,
413			    n->line, n->pos, "%s %s%s",
414			    roff_name[n->tok], n->child->string,
415			    n->child->next != NULL ? " ..." : "");
416		break;
417	default:
418		break;
419	}
420}
421
422static void
423post_IP(CHKARGS)
424{
425	switch (n->type) {
426	case ROFFT_BLOCK:
427		if (n->head->child == NULL && n->body->child == NULL)
428			roff_node_delete(man, n);
429		break;
430	case ROFFT_HEAD:
431		check_tag(n, n->child);
432		break;
433	case ROFFT_BODY:
434		if (n->parent->head->child == NULL && n->child == NULL)
435			mandoc_msg(MANDOCERR_PAR_SKIP, n->line, n->pos,
436			    "%s empty", roff_name[n->tok]);
437		break;
438	default:
439		break;
440	}
441}
442
443/*
444 * The first next-line element in the head is the tag.
445 * If that's a font macro, use its first child instead.
446 */
447static void
448post_TP(CHKARGS)
449{
450	struct roff_node *nt;
451
452	if (n->type != ROFFT_HEAD || (nt = n->child) == NULL)
453		return;
454
455	while ((nt->flags & NODE_LINE) == 0)
456		if ((nt = nt->next) == NULL)
457			return;
458
459	switch (nt->tok) {
460	case MAN_B:
461	case MAN_BI:
462	case MAN_BR:
463	case MAN_I:
464	case MAN_IB:
465	case MAN_IR:
466		nt = nt->child;
467		break;
468	default:
469		break;
470	}
471	check_tag(n, nt);
472}
473
474static void
475post_TH(CHKARGS)
476{
477	struct roff_node *nb;
478	const char	*p;
479
480	free(man->meta.title);
481	free(man->meta.vol);
482	free(man->meta.os);
483	free(man->meta.msec);
484	free(man->meta.date);
485
486	man->meta.title = man->meta.vol = man->meta.date =
487	    man->meta.msec = man->meta.os = NULL;
488
489	nb = n;
490
491	/* ->TITLE<- MSEC DATE OS VOL */
492
493	n = n->child;
494	if (n != NULL && n->string != NULL) {
495		for (p = n->string; *p != '\0'; p++) {
496			/* Only warn about this once... */
497			if (isalpha((unsigned char)*p) &&
498			    ! isupper((unsigned char)*p)) {
499				mandoc_msg(MANDOCERR_TITLE_CASE, n->line,
500				    n->pos + (int)(p - n->string),
501				    "TH %s", n->string);
502				break;
503			}
504		}
505		man->meta.title = mandoc_strdup(n->string);
506	} else {
507		man->meta.title = mandoc_strdup("");
508		mandoc_msg(MANDOCERR_TH_NOTITLE, nb->line, nb->pos, "TH");
509	}
510
511	/* TITLE ->MSEC<- DATE OS VOL */
512
513	if (n != NULL)
514		n = n->next;
515	if (n != NULL && n->string != NULL) {
516		man->meta.msec = mandoc_strdup(n->string);
517		if (man->filesec != '\0' &&
518		    man->filesec != *n->string &&
519		    *n->string >= '1' && *n->string <= '9')
520			mandoc_msg(MANDOCERR_MSEC_FILE, n->line, n->pos,
521			    "*.%c vs TH ... %c", man->filesec, *n->string);
522	} else {
523		man->meta.msec = mandoc_strdup("");
524		mandoc_msg(MANDOCERR_MSEC_MISSING,
525		    nb->line, nb->pos, "TH %s", man->meta.title);
526	}
527
528	/* TITLE MSEC ->DATE<- OS VOL */
529
530	if (n != NULL)
531		n = n->next;
532	if (man->quick && n != NULL)
533		man->meta.date = mandoc_strdup("");
534	else
535		man->meta.date = mandoc_normdate(n, nb);
536
537	/* TITLE MSEC DATE ->OS<- VOL */
538
539	if (n && (n = n->next))
540		man->meta.os = mandoc_strdup(n->string);
541	else if (man->os_s != NULL)
542		man->meta.os = mandoc_strdup(man->os_s);
543	if (man->meta.os_e == MANDOC_OS_OTHER && man->meta.os != NULL) {
544		if (strstr(man->meta.os, "OpenBSD") != NULL)
545			man->meta.os_e = MANDOC_OS_OPENBSD;
546		else if (strstr(man->meta.os, "NetBSD") != NULL)
547			man->meta.os_e = MANDOC_OS_NETBSD;
548	}
549
550	/* TITLE MSEC DATE OS ->VOL<- */
551	/* If missing, use the default VOL name for MSEC. */
552
553	if (n && (n = n->next))
554		man->meta.vol = mandoc_strdup(n->string);
555	else if ('\0' != man->meta.msec[0] &&
556	    (NULL != (p = mandoc_a2msec(man->meta.msec))))
557		man->meta.vol = mandoc_strdup(p);
558
559	if (n != NULL && (n = n->next) != NULL)
560		mandoc_msg(MANDOCERR_ARG_EXCESS,
561		    n->line, n->pos, "TH ... %s", n->string);
562
563	/*
564	 * Remove the `TH' node after we've processed it for our
565	 * meta-data.
566	 */
567	roff_node_delete(man, man->last);
568}
569
570static void
571post_UC(CHKARGS)
572{
573	static const char * const bsd_versions[] = {
574	    "3rd Berkeley Distribution",
575	    "4th Berkeley Distribution",
576	    "4.2 Berkeley Distribution",
577	    "4.3 Berkeley Distribution",
578	    "4.4 Berkeley Distribution",
579	};
580
581	const char	*p, *s;
582
583	n = n->child;
584
585	if (n == NULL || n->type != ROFFT_TEXT)
586		p = bsd_versions[0];
587	else {
588		s = n->string;
589		if (0 == strcmp(s, "3"))
590			p = bsd_versions[0];
591		else if (0 == strcmp(s, "4"))
592			p = bsd_versions[1];
593		else if (0 == strcmp(s, "5"))
594			p = bsd_versions[2];
595		else if (0 == strcmp(s, "6"))
596			p = bsd_versions[3];
597		else if (0 == strcmp(s, "7"))
598			p = bsd_versions[4];
599		else
600			p = bsd_versions[0];
601	}
602
603	free(man->meta.os);
604	man->meta.os = mandoc_strdup(p);
605}
606
607static void
608post_AT(CHKARGS)
609{
610	static const char * const unix_versions[] = {
611	    "7th Edition",
612	    "System III",
613	    "System V",
614	    "System V Release 2",
615	};
616
617	struct roff_node *nn;
618	const char	*p, *s;
619
620	n = n->child;
621
622	if (n == NULL || n->type != ROFFT_TEXT)
623		p = unix_versions[0];
624	else {
625		s = n->string;
626		if (0 == strcmp(s, "3"))
627			p = unix_versions[0];
628		else if (0 == strcmp(s, "4"))
629			p = unix_versions[1];
630		else if (0 == strcmp(s, "5")) {
631			nn = n->next;
632			if (nn != NULL &&
633			    nn->type == ROFFT_TEXT &&
634			    nn->string[0] != '\0')
635				p = unix_versions[3];
636			else
637				p = unix_versions[2];
638		} else
639			p = unix_versions[0];
640	}
641
642	free(man->meta.os);
643	man->meta.os = mandoc_strdup(p);
644}
645
646static void
647post_in(CHKARGS)
648{
649	char	*s;
650
651	if (n->parent->tok != MAN_TP ||
652	    n->parent->type != ROFFT_HEAD ||
653	    n->child == NULL ||
654	    *n->child->string == '+' ||
655	    *n->child->string == '-')
656		return;
657	mandoc_asprintf(&s, "+%s", n->child->string);
658	free(n->child->string);
659	n->child->string = s;
660}
661