1*c1c95addSBrooks Davis /* $Id: man_validate.c,v 1.159 2023/10/24 20:53:12 schwarze Exp $ */
261d06d6bSBaptiste Daroussin /*
3*c1c95addSBrooks Davis * Copyright (c) 2010, 2012-2020, 2023 Ingo Schwarze <schwarze@openbsd.org>
461d06d6bSBaptiste Daroussin * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
561d06d6bSBaptiste Daroussin *
661d06d6bSBaptiste Daroussin * Permission to use, copy, modify, and distribute this software for any
761d06d6bSBaptiste Daroussin * purpose with or without fee is hereby granted, provided that the above
861d06d6bSBaptiste Daroussin * copyright notice and this permission notice appear in all copies.
961d06d6bSBaptiste Daroussin *
1061d06d6bSBaptiste Daroussin * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHORS DISCLAIM ALL WARRANTIES
1161d06d6bSBaptiste Daroussin * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
1261d06d6bSBaptiste Daroussin * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHORS BE LIABLE FOR
1361d06d6bSBaptiste Daroussin * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
1461d06d6bSBaptiste Daroussin * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
1561d06d6bSBaptiste Daroussin * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
1661d06d6bSBaptiste Daroussin * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
176d38604fSBaptiste Daroussin *
186d38604fSBaptiste Daroussin * Validation module for man(7) syntax trees used by mandoc(1).
1961d06d6bSBaptiste Daroussin */
2061d06d6bSBaptiste Daroussin #include "config.h"
2161d06d6bSBaptiste Daroussin
2261d06d6bSBaptiste Daroussin #include <sys/types.h>
2361d06d6bSBaptiste Daroussin
2461d06d6bSBaptiste Daroussin #include <assert.h>
2561d06d6bSBaptiste Daroussin #include <ctype.h>
2661d06d6bSBaptiste Daroussin #include <errno.h>
2761d06d6bSBaptiste Daroussin #include <limits.h>
2861d06d6bSBaptiste Daroussin #include <stdarg.h>
297295610fSBaptiste Daroussin #include <stdio.h>
3061d06d6bSBaptiste Daroussin #include <stdlib.h>
3161d06d6bSBaptiste Daroussin #include <string.h>
3261d06d6bSBaptiste Daroussin #include <time.h>
3361d06d6bSBaptiste Daroussin
3461d06d6bSBaptiste Daroussin #include "mandoc_aux.h"
3561d06d6bSBaptiste Daroussin #include "mandoc.h"
36*c1c95addSBrooks Davis #include "mandoc_xr.h"
3761d06d6bSBaptiste Daroussin #include "roff.h"
3861d06d6bSBaptiste Daroussin #include "man.h"
3961d06d6bSBaptiste Daroussin #include "libmandoc.h"
4061d06d6bSBaptiste Daroussin #include "roff_int.h"
4161d06d6bSBaptiste Daroussin #include "libman.h"
426d38604fSBaptiste Daroussin #include "tag.h"
4361d06d6bSBaptiste Daroussin
4461d06d6bSBaptiste Daroussin #define CHKARGS struct roff_man *man, struct roff_node *n
4561d06d6bSBaptiste Daroussin
4661d06d6bSBaptiste Daroussin typedef void (*v_check)(CHKARGS);
4761d06d6bSBaptiste Daroussin
4861d06d6bSBaptiste Daroussin static void check_par(CHKARGS);
4961d06d6bSBaptiste Daroussin static void check_part(CHKARGS);
5061d06d6bSBaptiste Daroussin static void check_root(CHKARGS);
516d38604fSBaptiste Daroussin static void check_tag(struct roff_node *, struct roff_node *);
5261d06d6bSBaptiste Daroussin static void check_text(CHKARGS);
5361d06d6bSBaptiste Daroussin
5461d06d6bSBaptiste Daroussin static void post_AT(CHKARGS);
557295610fSBaptiste Daroussin static void post_EE(CHKARGS);
567295610fSBaptiste Daroussin static void post_EX(CHKARGS);
5761d06d6bSBaptiste Daroussin static void post_IP(CHKARGS);
58*c1c95addSBrooks Davis static void post_MR(CHKARGS);
5961d06d6bSBaptiste Daroussin static void post_OP(CHKARGS);
607295610fSBaptiste Daroussin static void post_SH(CHKARGS);
6161d06d6bSBaptiste Daroussin static void post_TH(CHKARGS);
626d38604fSBaptiste Daroussin static void post_TP(CHKARGS);
6361d06d6bSBaptiste Daroussin static void post_UC(CHKARGS);
6461d06d6bSBaptiste Daroussin static void post_UR(CHKARGS);
6561d06d6bSBaptiste Daroussin static void post_in(CHKARGS);
6661d06d6bSBaptiste Daroussin
677295610fSBaptiste Daroussin static const v_check man_valids[MAN_MAX - MAN_TH] = {
6861d06d6bSBaptiste Daroussin post_TH, /* TH */
697295610fSBaptiste Daroussin post_SH, /* SH */
707295610fSBaptiste Daroussin post_SH, /* SS */
716d38604fSBaptiste Daroussin post_TP, /* TP */
726d38604fSBaptiste Daroussin post_TP, /* TQ */
73*c1c95addSBrooks Davis check_par, /* LP */
7461d06d6bSBaptiste Daroussin check_par, /* PP */
75*c1c95addSBrooks Davis check_par, /* P */
7661d06d6bSBaptiste Daroussin post_IP, /* IP */
7761d06d6bSBaptiste Daroussin NULL, /* HP */
7861d06d6bSBaptiste Daroussin NULL, /* SM */
7961d06d6bSBaptiste Daroussin NULL, /* SB */
8061d06d6bSBaptiste Daroussin NULL, /* BI */
8161d06d6bSBaptiste Daroussin NULL, /* IB */
8261d06d6bSBaptiste Daroussin NULL, /* BR */
8361d06d6bSBaptiste Daroussin NULL, /* RB */
8461d06d6bSBaptiste Daroussin NULL, /* R */
8561d06d6bSBaptiste Daroussin NULL, /* B */
8661d06d6bSBaptiste Daroussin NULL, /* I */
8761d06d6bSBaptiste Daroussin NULL, /* IR */
8861d06d6bSBaptiste Daroussin NULL, /* RI */
8961d06d6bSBaptiste Daroussin NULL, /* RE */
9061d06d6bSBaptiste Daroussin check_part, /* RS */
9161d06d6bSBaptiste Daroussin NULL, /* DT */
9261d06d6bSBaptiste Daroussin post_UC, /* UC */
9361d06d6bSBaptiste Daroussin NULL, /* PD */
9461d06d6bSBaptiste Daroussin post_AT, /* AT */
9561d06d6bSBaptiste Daroussin post_in, /* in */
967295610fSBaptiste Daroussin NULL, /* SY */
977295610fSBaptiste Daroussin NULL, /* YS */
9861d06d6bSBaptiste Daroussin post_OP, /* OP */
997295610fSBaptiste Daroussin post_EX, /* EX */
1007295610fSBaptiste Daroussin post_EE, /* EE */
10161d06d6bSBaptiste Daroussin post_UR, /* UR */
10261d06d6bSBaptiste Daroussin NULL, /* UE */
10361d06d6bSBaptiste Daroussin post_UR, /* MT */
10461d06d6bSBaptiste Daroussin NULL, /* ME */
105*c1c95addSBrooks Davis post_MR, /* MR */
10661d06d6bSBaptiste Daroussin };
10761d06d6bSBaptiste Daroussin
10861d06d6bSBaptiste Daroussin
1097295610fSBaptiste Daroussin /* Validate the subtree rooted at man->last. */
11061d06d6bSBaptiste Daroussin void
man_validate(struct roff_man * man)1117295610fSBaptiste Daroussin man_validate(struct roff_man *man)
11261d06d6bSBaptiste Daroussin {
11361d06d6bSBaptiste Daroussin struct roff_node *n;
11461d06d6bSBaptiste Daroussin const v_check *cp;
11561d06d6bSBaptiste Daroussin
1167295610fSBaptiste Daroussin /*
1177295610fSBaptiste Daroussin * Iterate over all children, recursing into each one
1187295610fSBaptiste Daroussin * in turn, depth-first.
1197295610fSBaptiste Daroussin */
1207295610fSBaptiste Daroussin
121*c1c95addSBrooks Davis n = man->last;
12261d06d6bSBaptiste Daroussin man->last = man->last->child;
12361d06d6bSBaptiste Daroussin while (man->last != NULL) {
1247295610fSBaptiste Daroussin man_validate(man);
12561d06d6bSBaptiste Daroussin if (man->last == n)
12661d06d6bSBaptiste Daroussin man->last = man->last->child;
12761d06d6bSBaptiste Daroussin else
12861d06d6bSBaptiste Daroussin man->last = man->last->next;
12961d06d6bSBaptiste Daroussin }
13061d06d6bSBaptiste Daroussin
1317295610fSBaptiste Daroussin /* Finally validate the macro itself. */
1327295610fSBaptiste Daroussin
13361d06d6bSBaptiste Daroussin man->last = n;
13461d06d6bSBaptiste Daroussin man->next = ROFF_NEXT_SIBLING;
13561d06d6bSBaptiste Daroussin switch (n->type) {
13661d06d6bSBaptiste Daroussin case ROFFT_TEXT:
13761d06d6bSBaptiste Daroussin check_text(man, n);
13861d06d6bSBaptiste Daroussin break;
13961d06d6bSBaptiste Daroussin case ROFFT_ROOT:
14061d06d6bSBaptiste Daroussin check_root(man, n);
14161d06d6bSBaptiste Daroussin break;
14261d06d6bSBaptiste Daroussin case ROFFT_COMMENT:
14361d06d6bSBaptiste Daroussin case ROFFT_EQN:
14461d06d6bSBaptiste Daroussin case ROFFT_TBL:
14561d06d6bSBaptiste Daroussin break;
14661d06d6bSBaptiste Daroussin default:
14761d06d6bSBaptiste Daroussin if (n->tok < ROFF_MAX) {
14861d06d6bSBaptiste Daroussin roff_validate(man);
14961d06d6bSBaptiste Daroussin break;
15061d06d6bSBaptiste Daroussin }
15161d06d6bSBaptiste Daroussin assert(n->tok >= MAN_TH && n->tok < MAN_MAX);
1527295610fSBaptiste Daroussin cp = man_valids + (n->tok - MAN_TH);
15361d06d6bSBaptiste Daroussin if (*cp)
15461d06d6bSBaptiste Daroussin (*cp)(man, n);
15561d06d6bSBaptiste Daroussin if (man->last == n)
1567295610fSBaptiste Daroussin n->flags |= NODE_VALID;
15761d06d6bSBaptiste Daroussin break;
15861d06d6bSBaptiste Daroussin }
15961d06d6bSBaptiste Daroussin }
16061d06d6bSBaptiste Daroussin
16161d06d6bSBaptiste Daroussin static void
check_root(CHKARGS)16261d06d6bSBaptiste Daroussin check_root(CHKARGS)
16361d06d6bSBaptiste Daroussin {
16461d06d6bSBaptiste Daroussin assert((man->flags & (MAN_BLINE | MAN_ELINE)) == 0);
16561d06d6bSBaptiste Daroussin
16661d06d6bSBaptiste Daroussin if (n->last == NULL || n->last->type == ROFFT_COMMENT)
1677295610fSBaptiste Daroussin mandoc_msg(MANDOCERR_DOC_EMPTY, n->line, n->pos, NULL);
16861d06d6bSBaptiste Daroussin else
16961d06d6bSBaptiste Daroussin man->meta.hasbody = 1;
17061d06d6bSBaptiste Daroussin
17161d06d6bSBaptiste Daroussin if (NULL == man->meta.title) {
1727295610fSBaptiste Daroussin mandoc_msg(MANDOCERR_TH_NOTITLE, n->line, n->pos, NULL);
17361d06d6bSBaptiste Daroussin
17461d06d6bSBaptiste Daroussin /*
17561d06d6bSBaptiste Daroussin * If a title hasn't been set, do so now (by
17661d06d6bSBaptiste Daroussin * implication, date and section also aren't set).
17761d06d6bSBaptiste Daroussin */
17861d06d6bSBaptiste Daroussin
17961d06d6bSBaptiste Daroussin man->meta.title = mandoc_strdup("");
18061d06d6bSBaptiste Daroussin man->meta.msec = mandoc_strdup("");
1816d38604fSBaptiste Daroussin man->meta.date = mandoc_normdate(NULL, NULL);
18261d06d6bSBaptiste Daroussin }
18361d06d6bSBaptiste Daroussin
18461d06d6bSBaptiste Daroussin if (man->meta.os_e &&
18561d06d6bSBaptiste Daroussin (man->meta.rcsids & (1 << man->meta.os_e)) == 0)
1867295610fSBaptiste Daroussin mandoc_msg(MANDOCERR_RCS_MISSING, 0, 0,
18761d06d6bSBaptiste Daroussin man->meta.os_e == MANDOC_OS_OPENBSD ?
18861d06d6bSBaptiste Daroussin "(OpenBSD)" : "(NetBSD)");
18961d06d6bSBaptiste Daroussin }
19061d06d6bSBaptiste Daroussin
1916d38604fSBaptiste Daroussin /*
1926d38604fSBaptiste Daroussin * Skip leading whitespace, dashes, backslashes, and font escapes,
1936d38604fSBaptiste Daroussin * then create a tag if the first following byte is a letter.
1946d38604fSBaptiste Daroussin * Priority is high unless whitespace is present.
1956d38604fSBaptiste Daroussin */
1966d38604fSBaptiste Daroussin static void
check_tag(struct roff_node * n,struct roff_node * nt)1976d38604fSBaptiste Daroussin check_tag(struct roff_node *n, struct roff_node *nt)
1986d38604fSBaptiste Daroussin {
1996d38604fSBaptiste Daroussin const char *cp, *arg;
2006d38604fSBaptiste Daroussin int prio, sz;
2016d38604fSBaptiste Daroussin
2026d38604fSBaptiste Daroussin if (nt == NULL || nt->type != ROFFT_TEXT)
2036d38604fSBaptiste Daroussin return;
2046d38604fSBaptiste Daroussin
2056d38604fSBaptiste Daroussin cp = nt->string;
2066d38604fSBaptiste Daroussin prio = TAG_STRONG;
2076d38604fSBaptiste Daroussin for (;;) {
2086d38604fSBaptiste Daroussin switch (*cp) {
2096d38604fSBaptiste Daroussin case ' ':
2106d38604fSBaptiste Daroussin case '\t':
2116d38604fSBaptiste Daroussin prio = TAG_WEAK;
2126d38604fSBaptiste Daroussin /* FALLTHROUGH */
2136d38604fSBaptiste Daroussin case '-':
2146d38604fSBaptiste Daroussin cp++;
2156d38604fSBaptiste Daroussin break;
2166d38604fSBaptiste Daroussin case '\\':
2176d38604fSBaptiste Daroussin cp++;
2186d38604fSBaptiste Daroussin switch (mandoc_escape(&cp, &arg, &sz)) {
2196d38604fSBaptiste Daroussin case ESCAPE_FONT:
2206d38604fSBaptiste Daroussin case ESCAPE_FONTBOLD:
2216d38604fSBaptiste Daroussin case ESCAPE_FONTITALIC:
2226d38604fSBaptiste Daroussin case ESCAPE_FONTBI:
2236d38604fSBaptiste Daroussin case ESCAPE_FONTROMAN:
2246d38604fSBaptiste Daroussin case ESCAPE_FONTCR:
2256d38604fSBaptiste Daroussin case ESCAPE_FONTCB:
2266d38604fSBaptiste Daroussin case ESCAPE_FONTCI:
2276d38604fSBaptiste Daroussin case ESCAPE_FONTPREV:
2286d38604fSBaptiste Daroussin case ESCAPE_IGNORE:
2296d38604fSBaptiste Daroussin break;
2306d38604fSBaptiste Daroussin case ESCAPE_SPECIAL:
2316d38604fSBaptiste Daroussin if (sz != 1)
2326d38604fSBaptiste Daroussin return;
2336d38604fSBaptiste Daroussin switch (*arg) {
2346d38604fSBaptiste Daroussin case '-':
2356d38604fSBaptiste Daroussin case 'e':
2366d38604fSBaptiste Daroussin break;
2376d38604fSBaptiste Daroussin default:
2386d38604fSBaptiste Daroussin return;
2396d38604fSBaptiste Daroussin }
2406d38604fSBaptiste Daroussin break;
2416d38604fSBaptiste Daroussin default:
2426d38604fSBaptiste Daroussin return;
2436d38604fSBaptiste Daroussin }
2446d38604fSBaptiste Daroussin break;
2456d38604fSBaptiste Daroussin default:
2466d38604fSBaptiste Daroussin if (isalpha((unsigned char)*cp))
2476d38604fSBaptiste Daroussin tag_put(cp, prio, n);
2486d38604fSBaptiste Daroussin return;
2496d38604fSBaptiste Daroussin }
2506d38604fSBaptiste Daroussin }
2516d38604fSBaptiste Daroussin }
2526d38604fSBaptiste Daroussin
2537295610fSBaptiste Daroussin static void
check_text(CHKARGS)25461d06d6bSBaptiste Daroussin check_text(CHKARGS)
25561d06d6bSBaptiste Daroussin {
25661d06d6bSBaptiste Daroussin char *cp, *p;
25761d06d6bSBaptiste Daroussin
2587295610fSBaptiste Daroussin if (n->flags & NODE_NOFILL)
25961d06d6bSBaptiste Daroussin return;
26061d06d6bSBaptiste Daroussin
26161d06d6bSBaptiste Daroussin cp = n->string;
26261d06d6bSBaptiste Daroussin for (p = cp; NULL != (p = strchr(p, '\t')); p++)
2637295610fSBaptiste Daroussin mandoc_msg(MANDOCERR_FI_TAB,
2647295610fSBaptiste Daroussin n->line, n->pos + (int)(p - cp), NULL);
2657295610fSBaptiste Daroussin }
2667295610fSBaptiste Daroussin
2677295610fSBaptiste Daroussin static void
post_EE(CHKARGS)2687295610fSBaptiste Daroussin post_EE(CHKARGS)
2697295610fSBaptiste Daroussin {
2707295610fSBaptiste Daroussin if ((n->flags & NODE_NOFILL) == 0)
2717295610fSBaptiste Daroussin mandoc_msg(MANDOCERR_FI_SKIP, n->line, n->pos, "EE");
2727295610fSBaptiste Daroussin }
2737295610fSBaptiste Daroussin
2747295610fSBaptiste Daroussin static void
post_EX(CHKARGS)2757295610fSBaptiste Daroussin post_EX(CHKARGS)
2767295610fSBaptiste Daroussin {
2777295610fSBaptiste Daroussin if (n->flags & NODE_NOFILL)
2787295610fSBaptiste Daroussin mandoc_msg(MANDOCERR_NF_SKIP, n->line, n->pos, "EX");
27961d06d6bSBaptiste Daroussin }
28061d06d6bSBaptiste Daroussin
28161d06d6bSBaptiste Daroussin static void
post_OP(CHKARGS)28261d06d6bSBaptiste Daroussin post_OP(CHKARGS)
28361d06d6bSBaptiste Daroussin {
28461d06d6bSBaptiste Daroussin
28561d06d6bSBaptiste Daroussin if (n->child == NULL)
2867295610fSBaptiste Daroussin mandoc_msg(MANDOCERR_OP_EMPTY, n->line, n->pos, "OP");
28761d06d6bSBaptiste Daroussin else if (n->child->next != NULL && n->child->next->next != NULL) {
28861d06d6bSBaptiste Daroussin n = n->child->next->next;
2897295610fSBaptiste Daroussin mandoc_msg(MANDOCERR_ARG_EXCESS,
29061d06d6bSBaptiste Daroussin n->line, n->pos, "OP ... %s", n->string);
29161d06d6bSBaptiste Daroussin }
29261d06d6bSBaptiste Daroussin }
29361d06d6bSBaptiste Daroussin
29461d06d6bSBaptiste Daroussin static void
post_SH(CHKARGS)2957295610fSBaptiste Daroussin post_SH(CHKARGS)
2967295610fSBaptiste Daroussin {
2977295610fSBaptiste Daroussin struct roff_node *nc;
2986d38604fSBaptiste Daroussin char *cp, *tag;
2997295610fSBaptiste Daroussin
3006d38604fSBaptiste Daroussin nc = n->child;
3016d38604fSBaptiste Daroussin switch (n->type) {
3026d38604fSBaptiste Daroussin case ROFFT_HEAD:
3036d38604fSBaptiste Daroussin tag = NULL;
3046d38604fSBaptiste Daroussin deroff(&tag, n);
3056d38604fSBaptiste Daroussin if (tag != NULL) {
3066d38604fSBaptiste Daroussin for (cp = tag; *cp != '\0'; cp++)
3076d38604fSBaptiste Daroussin if (*cp == ' ')
3086d38604fSBaptiste Daroussin *cp = '_';
3096d38604fSBaptiste Daroussin if (nc != NULL && nc->type == ROFFT_TEXT &&
3106d38604fSBaptiste Daroussin strcmp(nc->string, tag) == 0)
3116d38604fSBaptiste Daroussin tag_put(NULL, TAG_STRONG, n);
3126d38604fSBaptiste Daroussin else
3136d38604fSBaptiste Daroussin tag_put(tag, TAG_FALLBACK, n);
3146d38604fSBaptiste Daroussin free(tag);
3156d38604fSBaptiste Daroussin }
3167295610fSBaptiste Daroussin return;
3176d38604fSBaptiste Daroussin case ROFFT_BODY:
3186d38604fSBaptiste Daroussin if (nc != NULL)
3196d38604fSBaptiste Daroussin break;
3206d38604fSBaptiste Daroussin return;
3216d38604fSBaptiste Daroussin default:
3226d38604fSBaptiste Daroussin return;
3236d38604fSBaptiste Daroussin }
3247295610fSBaptiste Daroussin
325*c1c95addSBrooks Davis if ((nc->tok == MAN_LP || nc->tok == MAN_PP || nc->tok == MAN_P) &&
326*c1c95addSBrooks Davis nc->body->child != NULL) {
3277295610fSBaptiste Daroussin while (nc->body->last != NULL) {
3287295610fSBaptiste Daroussin man->next = ROFF_NEXT_CHILD;
3297295610fSBaptiste Daroussin roff_node_relink(man, nc->body->last);
3307295610fSBaptiste Daroussin man->last = n;
3317295610fSBaptiste Daroussin }
3327295610fSBaptiste Daroussin }
3337295610fSBaptiste Daroussin
334*c1c95addSBrooks Davis if (nc->tok == MAN_LP || nc->tok == MAN_PP || nc->tok == MAN_P ||
335*c1c95addSBrooks Davis nc->tok == ROFF_sp || nc->tok == ROFF_br) {
3367295610fSBaptiste Daroussin mandoc_msg(MANDOCERR_PAR_SKIP, nc->line, nc->pos,
3377295610fSBaptiste Daroussin "%s after %s", roff_name[nc->tok], roff_name[n->tok]);
3387295610fSBaptiste Daroussin roff_node_delete(man, nc);
3397295610fSBaptiste Daroussin }
3407295610fSBaptiste Daroussin
3417295610fSBaptiste Daroussin /*
3427295610fSBaptiste Daroussin * Trailing PP is empty, so it is deleted by check_par().
3437295610fSBaptiste Daroussin * Trailing sp is significant.
3447295610fSBaptiste Daroussin */
3457295610fSBaptiste Daroussin
3467295610fSBaptiste Daroussin if ((nc = n->last) != NULL && nc->tok == ROFF_br) {
3477295610fSBaptiste Daroussin mandoc_msg(MANDOCERR_PAR_SKIP,
3487295610fSBaptiste Daroussin nc->line, nc->pos, "%s at the end of %s",
3497295610fSBaptiste Daroussin roff_name[nc->tok], roff_name[n->tok]);
3507295610fSBaptiste Daroussin roff_node_delete(man, nc);
3517295610fSBaptiste Daroussin }
3527295610fSBaptiste Daroussin }
3537295610fSBaptiste Daroussin
3547295610fSBaptiste Daroussin static void
post_UR(CHKARGS)35561d06d6bSBaptiste Daroussin post_UR(CHKARGS)
35661d06d6bSBaptiste Daroussin {
35761d06d6bSBaptiste Daroussin if (n->type == ROFFT_HEAD && n->child == NULL)
3587295610fSBaptiste Daroussin mandoc_msg(MANDOCERR_UR_NOHEAD, n->line, n->pos,
3597295610fSBaptiste Daroussin "%s", roff_name[n->tok]);
36061d06d6bSBaptiste Daroussin }
36161d06d6bSBaptiste Daroussin
36261d06d6bSBaptiste Daroussin static void
check_part(CHKARGS)36361d06d6bSBaptiste Daroussin check_part(CHKARGS)
36461d06d6bSBaptiste Daroussin {
36561d06d6bSBaptiste Daroussin if (n->type == ROFFT_BODY && n->child == NULL)
3667295610fSBaptiste Daroussin mandoc_msg(MANDOCERR_BLK_EMPTY, n->line, n->pos,
3677295610fSBaptiste Daroussin "%s", roff_name[n->tok]);
36861d06d6bSBaptiste Daroussin }
36961d06d6bSBaptiste Daroussin
37061d06d6bSBaptiste Daroussin static void
check_par(CHKARGS)37161d06d6bSBaptiste Daroussin check_par(CHKARGS)
37261d06d6bSBaptiste Daroussin {
37361d06d6bSBaptiste Daroussin
37461d06d6bSBaptiste Daroussin switch (n->type) {
37561d06d6bSBaptiste Daroussin case ROFFT_BLOCK:
37661d06d6bSBaptiste Daroussin if (n->body->child == NULL)
37761d06d6bSBaptiste Daroussin roff_node_delete(man, n);
37861d06d6bSBaptiste Daroussin break;
37961d06d6bSBaptiste Daroussin case ROFFT_BODY:
3807295610fSBaptiste Daroussin if (n->child != NULL &&
3817295610fSBaptiste Daroussin (n->child->tok == ROFF_sp || n->child->tok == ROFF_br)) {
3827295610fSBaptiste Daroussin mandoc_msg(MANDOCERR_PAR_SKIP,
3837295610fSBaptiste Daroussin n->child->line, n->child->pos,
3847295610fSBaptiste Daroussin "%s after %s", roff_name[n->child->tok],
3857295610fSBaptiste Daroussin roff_name[n->tok]);
3867295610fSBaptiste Daroussin roff_node_delete(man, n->child);
3877295610fSBaptiste Daroussin }
38861d06d6bSBaptiste Daroussin if (n->child == NULL)
3897295610fSBaptiste Daroussin mandoc_msg(MANDOCERR_PAR_SKIP, n->line, n->pos,
39061d06d6bSBaptiste Daroussin "%s empty", roff_name[n->tok]);
39161d06d6bSBaptiste Daroussin break;
39261d06d6bSBaptiste Daroussin case ROFFT_HEAD:
39361d06d6bSBaptiste Daroussin if (n->child != NULL)
3947295610fSBaptiste Daroussin mandoc_msg(MANDOCERR_ARG_SKIP,
3957295610fSBaptiste Daroussin n->line, n->pos, "%s %s%s",
39661d06d6bSBaptiste Daroussin roff_name[n->tok], n->child->string,
39761d06d6bSBaptiste Daroussin n->child->next != NULL ? " ..." : "");
39861d06d6bSBaptiste Daroussin break;
39961d06d6bSBaptiste Daroussin default:
40061d06d6bSBaptiste Daroussin break;
40161d06d6bSBaptiste Daroussin }
40261d06d6bSBaptiste Daroussin }
40361d06d6bSBaptiste Daroussin
40461d06d6bSBaptiste Daroussin static void
post_IP(CHKARGS)40561d06d6bSBaptiste Daroussin post_IP(CHKARGS)
40661d06d6bSBaptiste Daroussin {
40761d06d6bSBaptiste Daroussin switch (n->type) {
40861d06d6bSBaptiste Daroussin case ROFFT_BLOCK:
40961d06d6bSBaptiste Daroussin if (n->head->child == NULL && n->body->child == NULL)
41061d06d6bSBaptiste Daroussin roff_node_delete(man, n);
41161d06d6bSBaptiste Daroussin break;
4126d38604fSBaptiste Daroussin case ROFFT_HEAD:
4136d38604fSBaptiste Daroussin check_tag(n, n->child);
4146d38604fSBaptiste Daroussin break;
41561d06d6bSBaptiste Daroussin case ROFFT_BODY:
41661d06d6bSBaptiste Daroussin if (n->parent->head->child == NULL && n->child == NULL)
4177295610fSBaptiste Daroussin mandoc_msg(MANDOCERR_PAR_SKIP, n->line, n->pos,
41861d06d6bSBaptiste Daroussin "%s empty", roff_name[n->tok]);
41961d06d6bSBaptiste Daroussin break;
42061d06d6bSBaptiste Daroussin default:
42161d06d6bSBaptiste Daroussin break;
42261d06d6bSBaptiste Daroussin }
42361d06d6bSBaptiste Daroussin }
42461d06d6bSBaptiste Daroussin
4256d38604fSBaptiste Daroussin /*
4266d38604fSBaptiste Daroussin * The first next-line element in the head is the tag.
4276d38604fSBaptiste Daroussin * If that's a font macro, use its first child instead.
4286d38604fSBaptiste Daroussin */
4296d38604fSBaptiste Daroussin static void
post_TP(CHKARGS)4306d38604fSBaptiste Daroussin post_TP(CHKARGS)
4316d38604fSBaptiste Daroussin {
4326d38604fSBaptiste Daroussin struct roff_node *nt;
4336d38604fSBaptiste Daroussin
4346d38604fSBaptiste Daroussin if (n->type != ROFFT_HEAD || (nt = n->child) == NULL)
4356d38604fSBaptiste Daroussin return;
4366d38604fSBaptiste Daroussin
4376d38604fSBaptiste Daroussin while ((nt->flags & NODE_LINE) == 0)
4386d38604fSBaptiste Daroussin if ((nt = nt->next) == NULL)
4396d38604fSBaptiste Daroussin return;
4406d38604fSBaptiste Daroussin
4416d38604fSBaptiste Daroussin switch (nt->tok) {
4426d38604fSBaptiste Daroussin case MAN_B:
4436d38604fSBaptiste Daroussin case MAN_BI:
4446d38604fSBaptiste Daroussin case MAN_BR:
4456d38604fSBaptiste Daroussin case MAN_I:
4466d38604fSBaptiste Daroussin case MAN_IB:
4476d38604fSBaptiste Daroussin case MAN_IR:
4486d38604fSBaptiste Daroussin nt = nt->child;
4496d38604fSBaptiste Daroussin break;
4506d38604fSBaptiste Daroussin default:
4516d38604fSBaptiste Daroussin break;
4526d38604fSBaptiste Daroussin }
4536d38604fSBaptiste Daroussin check_tag(n, nt);
4546d38604fSBaptiste Daroussin }
4556d38604fSBaptiste Daroussin
45661d06d6bSBaptiste Daroussin static void
post_TH(CHKARGS)45761d06d6bSBaptiste Daroussin post_TH(CHKARGS)
45861d06d6bSBaptiste Daroussin {
45961d06d6bSBaptiste Daroussin struct roff_node *nb;
46061d06d6bSBaptiste Daroussin const char *p;
46161d06d6bSBaptiste Daroussin
46261d06d6bSBaptiste Daroussin free(man->meta.title);
46361d06d6bSBaptiste Daroussin free(man->meta.vol);
46461d06d6bSBaptiste Daroussin free(man->meta.os);
46561d06d6bSBaptiste Daroussin free(man->meta.msec);
46661d06d6bSBaptiste Daroussin free(man->meta.date);
46761d06d6bSBaptiste Daroussin
46861d06d6bSBaptiste Daroussin man->meta.title = man->meta.vol = man->meta.date =
46961d06d6bSBaptiste Daroussin man->meta.msec = man->meta.os = NULL;
47061d06d6bSBaptiste Daroussin
47161d06d6bSBaptiste Daroussin nb = n;
47261d06d6bSBaptiste Daroussin
47361d06d6bSBaptiste Daroussin /* ->TITLE<- MSEC DATE OS VOL */
47461d06d6bSBaptiste Daroussin
47561d06d6bSBaptiste Daroussin n = n->child;
47645a5aec3SBaptiste Daroussin if (n != NULL && n->string != NULL) {
47745a5aec3SBaptiste Daroussin for (p = n->string; *p != '\0'; p++) {
47861d06d6bSBaptiste Daroussin /* Only warn about this once... */
47961d06d6bSBaptiste Daroussin if (isalpha((unsigned char)*p) &&
48061d06d6bSBaptiste Daroussin ! isupper((unsigned char)*p)) {
4817295610fSBaptiste Daroussin mandoc_msg(MANDOCERR_TITLE_CASE, n->line,
4827295610fSBaptiste Daroussin n->pos + (int)(p - n->string),
48361d06d6bSBaptiste Daroussin "TH %s", n->string);
48461d06d6bSBaptiste Daroussin break;
48561d06d6bSBaptiste Daroussin }
48661d06d6bSBaptiste Daroussin }
48761d06d6bSBaptiste Daroussin man->meta.title = mandoc_strdup(n->string);
48861d06d6bSBaptiste Daroussin } else {
48961d06d6bSBaptiste Daroussin man->meta.title = mandoc_strdup("");
4907295610fSBaptiste Daroussin mandoc_msg(MANDOCERR_TH_NOTITLE, nb->line, nb->pos, "TH");
49161d06d6bSBaptiste Daroussin }
49261d06d6bSBaptiste Daroussin
49361d06d6bSBaptiste Daroussin /* TITLE ->MSEC<- DATE OS VOL */
49461d06d6bSBaptiste Daroussin
49545a5aec3SBaptiste Daroussin if (n != NULL)
49661d06d6bSBaptiste Daroussin n = n->next;
4976d38604fSBaptiste Daroussin if (n != NULL && n->string != NULL) {
49861d06d6bSBaptiste Daroussin man->meta.msec = mandoc_strdup(n->string);
4996d38604fSBaptiste Daroussin if (man->filesec != '\0' &&
5006d38604fSBaptiste Daroussin man->filesec != *n->string &&
5016d38604fSBaptiste Daroussin *n->string >= '1' && *n->string <= '9')
5026d38604fSBaptiste Daroussin mandoc_msg(MANDOCERR_MSEC_FILE, n->line, n->pos,
5036d38604fSBaptiste Daroussin "*.%c vs TH ... %c", man->filesec, *n->string);
5046d38604fSBaptiste Daroussin } else {
50561d06d6bSBaptiste Daroussin man->meta.msec = mandoc_strdup("");
5067295610fSBaptiste Daroussin mandoc_msg(MANDOCERR_MSEC_MISSING,
50761d06d6bSBaptiste Daroussin nb->line, nb->pos, "TH %s", man->meta.title);
50861d06d6bSBaptiste Daroussin }
50961d06d6bSBaptiste Daroussin
51061d06d6bSBaptiste Daroussin /* TITLE MSEC ->DATE<- OS VOL */
51161d06d6bSBaptiste Daroussin
51245a5aec3SBaptiste Daroussin if (n != NULL)
51361d06d6bSBaptiste Daroussin n = n->next;
5146d38604fSBaptiste Daroussin if (man->quick && n != NULL)
51561d06d6bSBaptiste Daroussin man->meta.date = mandoc_strdup("");
5166d38604fSBaptiste Daroussin else
5176d38604fSBaptiste Daroussin man->meta.date = mandoc_normdate(n, nb);
51861d06d6bSBaptiste Daroussin
51961d06d6bSBaptiste Daroussin /* TITLE MSEC DATE ->OS<- VOL */
52061d06d6bSBaptiste Daroussin
52161d06d6bSBaptiste Daroussin if (n && (n = n->next))
52261d06d6bSBaptiste Daroussin man->meta.os = mandoc_strdup(n->string);
52361d06d6bSBaptiste Daroussin else if (man->os_s != NULL)
52461d06d6bSBaptiste Daroussin man->meta.os = mandoc_strdup(man->os_s);
52561d06d6bSBaptiste Daroussin if (man->meta.os_e == MANDOC_OS_OTHER && man->meta.os != NULL) {
52661d06d6bSBaptiste Daroussin if (strstr(man->meta.os, "OpenBSD") != NULL)
52761d06d6bSBaptiste Daroussin man->meta.os_e = MANDOC_OS_OPENBSD;
52861d06d6bSBaptiste Daroussin else if (strstr(man->meta.os, "NetBSD") != NULL)
52961d06d6bSBaptiste Daroussin man->meta.os_e = MANDOC_OS_NETBSD;
53061d06d6bSBaptiste Daroussin }
53161d06d6bSBaptiste Daroussin
53261d06d6bSBaptiste Daroussin /* TITLE MSEC DATE OS ->VOL<- */
53361d06d6bSBaptiste Daroussin /* If missing, use the default VOL name for MSEC. */
53461d06d6bSBaptiste Daroussin
53561d06d6bSBaptiste Daroussin if (n && (n = n->next))
53661d06d6bSBaptiste Daroussin man->meta.vol = mandoc_strdup(n->string);
53761d06d6bSBaptiste Daroussin else if ('\0' != man->meta.msec[0] &&
53861d06d6bSBaptiste Daroussin (NULL != (p = mandoc_a2msec(man->meta.msec))))
53961d06d6bSBaptiste Daroussin man->meta.vol = mandoc_strdup(p);
54061d06d6bSBaptiste Daroussin
54161d06d6bSBaptiste Daroussin if (n != NULL && (n = n->next) != NULL)
5427295610fSBaptiste Daroussin mandoc_msg(MANDOCERR_ARG_EXCESS,
54361d06d6bSBaptiste Daroussin n->line, n->pos, "TH ... %s", n->string);
54461d06d6bSBaptiste Daroussin
54561d06d6bSBaptiste Daroussin /*
54661d06d6bSBaptiste Daroussin * Remove the `TH' node after we've processed it for our
54761d06d6bSBaptiste Daroussin * meta-data.
54861d06d6bSBaptiste Daroussin */
54961d06d6bSBaptiste Daroussin roff_node_delete(man, man->last);
55061d06d6bSBaptiste Daroussin }
55161d06d6bSBaptiste Daroussin
55261d06d6bSBaptiste Daroussin static void
post_MR(CHKARGS)553*c1c95addSBrooks Davis post_MR(CHKARGS)
554*c1c95addSBrooks Davis {
555*c1c95addSBrooks Davis struct roff_node *nch;
556*c1c95addSBrooks Davis
557*c1c95addSBrooks Davis if ((nch = n->child) == NULL) {
558*c1c95addSBrooks Davis mandoc_msg(MANDOCERR_NM_NONAME, n->line, n->pos, "MR");
559*c1c95addSBrooks Davis return;
560*c1c95addSBrooks Davis }
561*c1c95addSBrooks Davis if (nch->next == NULL) {
562*c1c95addSBrooks Davis mandoc_msg(MANDOCERR_XR_NOSEC,
563*c1c95addSBrooks Davis n->line, n->pos, "MR %s", nch->string);
564*c1c95addSBrooks Davis return;
565*c1c95addSBrooks Davis }
566*c1c95addSBrooks Davis if (mandoc_xr_add(nch->next->string, nch->string, nch->line, nch->pos))
567*c1c95addSBrooks Davis mandoc_msg(MANDOCERR_XR_SELF, nch->line, nch->pos,
568*c1c95addSBrooks Davis "MR %s %s", nch->string, nch->next->string);
569*c1c95addSBrooks Davis if ((nch = nch->next->next) == NULL || nch->next == NULL)
570*c1c95addSBrooks Davis return;
571*c1c95addSBrooks Davis
572*c1c95addSBrooks Davis mandoc_msg(MANDOCERR_ARG_EXCESS, nch->next->line, nch->next->pos,
573*c1c95addSBrooks Davis "MR ... %s", nch->next->string);
574*c1c95addSBrooks Davis while (nch->next != NULL)
575*c1c95addSBrooks Davis roff_node_delete(man, nch->next);
576*c1c95addSBrooks Davis }
577*c1c95addSBrooks Davis
578*c1c95addSBrooks Davis static void
post_UC(CHKARGS)57961d06d6bSBaptiste Daroussin post_UC(CHKARGS)
58061d06d6bSBaptiste Daroussin {
58161d06d6bSBaptiste Daroussin static const char * const bsd_versions[] = {
58261d06d6bSBaptiste Daroussin "3rd Berkeley Distribution",
58361d06d6bSBaptiste Daroussin "4th Berkeley Distribution",
58461d06d6bSBaptiste Daroussin "4.2 Berkeley Distribution",
58561d06d6bSBaptiste Daroussin "4.3 Berkeley Distribution",
58661d06d6bSBaptiste Daroussin "4.4 Berkeley Distribution",
58761d06d6bSBaptiste Daroussin };
58861d06d6bSBaptiste Daroussin
58961d06d6bSBaptiste Daroussin const char *p, *s;
59061d06d6bSBaptiste Daroussin
59161d06d6bSBaptiste Daroussin n = n->child;
59261d06d6bSBaptiste Daroussin
59361d06d6bSBaptiste Daroussin if (n == NULL || n->type != ROFFT_TEXT)
59461d06d6bSBaptiste Daroussin p = bsd_versions[0];
59561d06d6bSBaptiste Daroussin else {
59661d06d6bSBaptiste Daroussin s = n->string;
59761d06d6bSBaptiste Daroussin if (0 == strcmp(s, "3"))
59861d06d6bSBaptiste Daroussin p = bsd_versions[0];
59961d06d6bSBaptiste Daroussin else if (0 == strcmp(s, "4"))
60061d06d6bSBaptiste Daroussin p = bsd_versions[1];
60161d06d6bSBaptiste Daroussin else if (0 == strcmp(s, "5"))
60261d06d6bSBaptiste Daroussin p = bsd_versions[2];
60361d06d6bSBaptiste Daroussin else if (0 == strcmp(s, "6"))
60461d06d6bSBaptiste Daroussin p = bsd_versions[3];
60561d06d6bSBaptiste Daroussin else if (0 == strcmp(s, "7"))
60661d06d6bSBaptiste Daroussin p = bsd_versions[4];
60761d06d6bSBaptiste Daroussin else
60861d06d6bSBaptiste Daroussin p = bsd_versions[0];
60961d06d6bSBaptiste Daroussin }
61061d06d6bSBaptiste Daroussin
61161d06d6bSBaptiste Daroussin free(man->meta.os);
61261d06d6bSBaptiste Daroussin man->meta.os = mandoc_strdup(p);
61361d06d6bSBaptiste Daroussin }
61461d06d6bSBaptiste Daroussin
61561d06d6bSBaptiste Daroussin static void
post_AT(CHKARGS)61661d06d6bSBaptiste Daroussin post_AT(CHKARGS)
61761d06d6bSBaptiste Daroussin {
61861d06d6bSBaptiste Daroussin static const char * const unix_versions[] = {
61961d06d6bSBaptiste Daroussin "7th Edition",
62061d06d6bSBaptiste Daroussin "System III",
62161d06d6bSBaptiste Daroussin "System V",
62261d06d6bSBaptiste Daroussin "System V Release 2",
62361d06d6bSBaptiste Daroussin };
62461d06d6bSBaptiste Daroussin
62561d06d6bSBaptiste Daroussin struct roff_node *nn;
62661d06d6bSBaptiste Daroussin const char *p, *s;
62761d06d6bSBaptiste Daroussin
62861d06d6bSBaptiste Daroussin n = n->child;
62961d06d6bSBaptiste Daroussin
63061d06d6bSBaptiste Daroussin if (n == NULL || n->type != ROFFT_TEXT)
63161d06d6bSBaptiste Daroussin p = unix_versions[0];
63261d06d6bSBaptiste Daroussin else {
63361d06d6bSBaptiste Daroussin s = n->string;
63461d06d6bSBaptiste Daroussin if (0 == strcmp(s, "3"))
63561d06d6bSBaptiste Daroussin p = unix_versions[0];
63661d06d6bSBaptiste Daroussin else if (0 == strcmp(s, "4"))
63761d06d6bSBaptiste Daroussin p = unix_versions[1];
63861d06d6bSBaptiste Daroussin else if (0 == strcmp(s, "5")) {
63961d06d6bSBaptiste Daroussin nn = n->next;
64061d06d6bSBaptiste Daroussin if (nn != NULL &&
64161d06d6bSBaptiste Daroussin nn->type == ROFFT_TEXT &&
64261d06d6bSBaptiste Daroussin nn->string[0] != '\0')
64361d06d6bSBaptiste Daroussin p = unix_versions[3];
64461d06d6bSBaptiste Daroussin else
64561d06d6bSBaptiste Daroussin p = unix_versions[2];
64661d06d6bSBaptiste Daroussin } else
64761d06d6bSBaptiste Daroussin p = unix_versions[0];
64861d06d6bSBaptiste Daroussin }
64961d06d6bSBaptiste Daroussin
65061d06d6bSBaptiste Daroussin free(man->meta.os);
65161d06d6bSBaptiste Daroussin man->meta.os = mandoc_strdup(p);
65261d06d6bSBaptiste Daroussin }
65361d06d6bSBaptiste Daroussin
65461d06d6bSBaptiste Daroussin static void
post_in(CHKARGS)65561d06d6bSBaptiste Daroussin post_in(CHKARGS)
65661d06d6bSBaptiste Daroussin {
65761d06d6bSBaptiste Daroussin char *s;
65861d06d6bSBaptiste Daroussin
65961d06d6bSBaptiste Daroussin if (n->parent->tok != MAN_TP ||
66061d06d6bSBaptiste Daroussin n->parent->type != ROFFT_HEAD ||
66161d06d6bSBaptiste Daroussin n->child == NULL ||
66261d06d6bSBaptiste Daroussin *n->child->string == '+' ||
66361d06d6bSBaptiste Daroussin *n->child->string == '-')
66461d06d6bSBaptiste Daroussin return;
66561d06d6bSBaptiste Daroussin mandoc_asprintf(&s, "+%s", n->child->string);
66661d06d6bSBaptiste Daroussin free(n->child->string);
66761d06d6bSBaptiste Daroussin n->child->string = s;
66861d06d6bSBaptiste Daroussin }
669