xref: /openbsd-src/usr.bin/mandoc/man_validate.c (revision 43003dfe3ad45d1698bed8a37f2b0f5b14f20d4f)
1 /*	$Id: man_validate.c,v 1.9 2009/09/18 22:46:14 schwarze Exp $ */
2 /*
3  * Copyright (c) 2008, 2009 Kristaps Dzonsons <kristaps@kth.se>
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 <ctype.h>
21 #include <errno.h>
22 #include <limits.h>
23 #include <stdarg.h>
24 #include <stdlib.h>
25 
26 #include "libman.h"
27 #include "libmandoc.h"
28 
29 #define	CHKARGS	  struct man *m, const struct man_node *n
30 
31 typedef	int	(*v_check)(CHKARGS);
32 
33 struct	man_valid {
34 	v_check	 *pres;
35 	v_check	 *posts;
36 };
37 
38 static	int	  check_bline(CHKARGS);
39 static	int	  check_eq0(CHKARGS);
40 static	int	  check_eq1(CHKARGS);
41 static	int	  check_ge2(CHKARGS);
42 static	int	  check_le5(CHKARGS);
43 static	int	  check_par(CHKARGS);
44 static	int	  check_part(CHKARGS);
45 static	int	  check_root(CHKARGS);
46 static	int	  check_sec(CHKARGS);
47 static	int	  check_sp(CHKARGS);
48 static	int	  check_text(CHKARGS);
49 
50 static	v_check	  posts_eq0[] = { check_eq0, NULL };
51 static	v_check	  posts_ge2_le5[] = { check_ge2, check_le5, NULL };
52 static	v_check	  posts_par[] = { check_par, NULL };
53 static	v_check	  posts_part[] = { check_part, NULL };
54 static	v_check	  posts_sec[] = { check_sec, NULL };
55 static	v_check	  posts_sp[] = { check_sp, NULL };
56 static	v_check	  pres_bline[] = { check_bline, NULL };
57 
58 static	const struct man_valid man_valids[MAN_MAX] = {
59 	{ pres_bline, posts_eq0 }, /* br */
60 	{ pres_bline, posts_ge2_le5 }, /* TH */
61 	{ pres_bline, posts_sec }, /* SH */
62 	{ pres_bline, posts_sec }, /* SS */
63 	{ pres_bline, posts_par }, /* TP */
64 	{ pres_bline, posts_par }, /* LP */
65 	{ pres_bline, posts_par }, /* PP */
66 	{ pres_bline, posts_par }, /* P */
67 	{ pres_bline, posts_par }, /* IP */
68 	{ pres_bline, posts_par }, /* HP */
69 	{ NULL, NULL }, /* SM */
70 	{ NULL, NULL }, /* SB */
71 	{ NULL, NULL }, /* BI */
72 	{ NULL, NULL }, /* IB */
73 	{ NULL, NULL }, /* BR */
74 	{ NULL, NULL }, /* RB */
75 	{ NULL, NULL }, /* R */
76 	{ NULL, NULL }, /* B */
77 	{ NULL, NULL }, /* I */
78 	{ NULL, NULL }, /* IR */
79 	{ NULL, NULL }, /* RI */
80 	{ pres_bline, posts_eq0 }, /* na */
81 	{ NULL, NULL }, /* i */
82 	{ pres_bline, posts_sp }, /* sp */
83 	{ pres_bline, posts_eq0 }, /* nf */
84 	{ pres_bline, posts_eq0 }, /* fi */
85 	{ NULL, NULL }, /* r */
86 	{ NULL, NULL }, /* RE */
87 	{ NULL, posts_part }, /* RS */
88 	{ NULL, NULL }, /* DT */
89 	{ NULL, NULL }, /* UC */
90 };
91 
92 
93 int
94 man_valid_pre(struct man *m, const struct man_node *n)
95 {
96 	v_check		*cp;
97 
98 	if (MAN_TEXT == n->type)
99 		return(1);
100 	if (MAN_ROOT == n->type)
101 		return(1);
102 
103 	if (NULL == (cp = man_valids[n->tok].pres))
104 		return(1);
105 	for ( ; *cp; cp++)
106 		if ( ! (*cp)(m, n))
107 			return(0);
108 	return(1);
109 }
110 
111 
112 int
113 man_valid_post(struct man *m)
114 {
115 	v_check		*cp;
116 
117 	if (MAN_VALID & m->last->flags)
118 		return(1);
119 	m->last->flags |= MAN_VALID;
120 
121 	switch (m->last->type) {
122 	case (MAN_TEXT):
123 		return(check_text(m, m->last));
124 	case (MAN_ROOT):
125 		return(check_root(m, m->last));
126 	default:
127 		break;
128 	}
129 
130 	if (NULL == (cp = man_valids[m->last->tok].posts))
131 		return(1);
132 	for ( ; *cp; cp++)
133 		if ( ! (*cp)(m, m->last))
134 			return(0);
135 
136 	return(1);
137 }
138 
139 
140 static int
141 check_root(CHKARGS)
142 {
143 
144 	if (MAN_BLINE & m->flags)
145 		return(man_nwarn(m, n, WEXITSCOPE));
146 	if (MAN_ELINE & m->flags)
147 		return(man_nwarn(m, n, WEXITSCOPE));
148 
149 	m->flags &= ~MAN_BLINE;
150 	m->flags &= ~MAN_ELINE;
151 
152 	if (NULL == m->first->child)
153 		return(man_nerr(m, n, WNODATA));
154 	if (NULL == m->meta.title)
155 		return(man_nerr(m, n, WNOTITLE));
156 
157 	return(1);
158 }
159 
160 
161 static int
162 check_text(CHKARGS)
163 {
164 	const char	*p;
165 	int		 pos, c;
166 
167 	assert(n->string);
168 
169 	for (p = n->string, pos = n->pos + 1; *p; p++, pos++) {
170 		if ('\\' == *p) {
171 			c = mandoc_special(p);
172 			if (c) {
173 				p += c - 1;
174 				pos += c - 1;
175 				continue;
176 			}
177 			if ( ! (MAN_IGN_ESCAPE & m->pflags))
178 				return(man_perr(m, n->line, pos, WESCAPE));
179 			if ( ! man_pwarn(m, n->line, pos, WESCAPE))
180 				return(0);
181 			continue;
182 		}
183 
184 		if ('\t' == *p || isprint((u_char)*p))
185 			continue;
186 
187 		if (MAN_IGN_CHARS & m->pflags)
188 			return(man_pwarn(m, n->line, pos, WNPRINT));
189 		return(man_perr(m, n->line, pos, WNPRINT));
190 	}
191 
192 	return(1);
193 }
194 
195 
196 #define	INEQ_DEFINE(x, ineq, name) \
197 static int \
198 check_##name(CHKARGS) \
199 { \
200 	if (n->nchild ineq (x)) \
201 		return(1); \
202 	return(man_verr(m, n->line, n->pos, \
203 			"expected line arguments %s %d, have %d", \
204 			#ineq, (x), n->nchild)); \
205 }
206 
207 INEQ_DEFINE(0, ==, eq0)
208 INEQ_DEFINE(1, ==, eq1)
209 INEQ_DEFINE(2, >=, ge2)
210 INEQ_DEFINE(5, <=, le5)
211 
212 
213 static int
214 check_sp(CHKARGS)
215 {
216 	long		 lval;
217 	char		*ep, *buf;
218 
219 	if (NULL == n->child)
220 		return(1);
221 	else if ( ! check_eq1(m, n))
222 		return(0);
223 
224 	assert(MAN_TEXT == n->child->type);
225 	buf = n->child->string;
226 	assert(buf);
227 
228 	/* From OpenBSD's strtol(3). */
229 
230 	errno = 0;
231 	lval = strtol(buf, &ep, 10);
232 	if (buf[0] == '\0' || *ep != '\0')
233 		return(man_nerr(m, n->child, WNUMFMT));
234 
235 	if ((errno == ERANGE && (lval == LONG_MAX || lval == LONG_MIN)) ||
236 			(lval > INT_MAX || lval < 0))
237 		return(man_nerr(m, n->child, WNUMFMT));
238 
239 	return(1);
240 }
241 
242 
243 static int
244 check_sec(CHKARGS)
245 {
246 
247 	if (MAN_BODY == n->type && 0 == n->nchild)
248 		return(man_nwarn(m, n, WBODYARGS));
249 	if (MAN_HEAD == n->type && 0 == n->nchild)
250 		return(man_nerr(m, n, WHEADARGS));
251 	return(1);
252 }
253 
254 
255 static int
256 check_part(CHKARGS)
257 {
258 
259 	if (MAN_BODY == n->type && 0 == n->nchild)
260 		return(man_nwarn(m, n, WBODYARGS));
261 	return(1);
262 }
263 
264 
265 static int
266 check_par(CHKARGS)
267 {
268 
269 	if (MAN_BODY == n->type)
270 		switch (n->tok) {
271 		case (MAN_IP):
272 			/* FALLTHROUGH */
273 		case (MAN_HP):
274 			/* FALLTHROUGH */
275 		case (MAN_TP):
276 			/* Body-less lists are ok. */
277 			break;
278 		default:
279 			if (n->nchild)
280 				break;
281 			return(man_nwarn(m, n, WBODYARGS));
282 		}
283 	if (MAN_HEAD == n->type)
284 		switch (n->tok) {
285 		case (MAN_PP):
286 			/* FALLTHROUGH */
287 		case (MAN_P):
288 			/* FALLTHROUGH */
289 		case (MAN_LP):
290 			if (0 == n->nchild)
291 				break;
292 			return(man_nwarn(m, n, WNHEADARGS));
293 		default:
294 			if (n->nchild)
295 				break;
296 			return(man_nwarn(m, n, WHEADARGS));
297 		}
298 
299 	return(1);
300 }
301 
302 
303 static int
304 check_bline(CHKARGS)
305 {
306 
307 	assert( ! (MAN_ELINE & m->flags));
308 	if (MAN_BLINE & m->flags)
309 		return(man_nerr(m, n, WLNSCOPE));
310 	return(1);
311 }
312 
313