1*f6697133Sschwarze /* $OpenBSD: man_validate.c,v 1.129 2023/10/24 20:30:49 schwarze Exp $ */
2f73abda9Skristaps /*
3*f6697133Sschwarze * Copyright (c) 2010, 2012-2020, 2023 Ingo Schwarze <schwarze@openbsd.org>
40ac7e6ecSschwarze * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
5f73abda9Skristaps *
6f73abda9Skristaps * Permission to use, copy, modify, and distribute this software for any
7a6464425Sschwarze * purpose with or without fee is hereby granted, provided that the above
8a6464425Sschwarze * copyright notice and this permission notice appear in all copies.
9f73abda9Skristaps *
10d1982c71Sschwarze * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHORS DISCLAIM ALL WARRANTIES
11a6464425Sschwarze * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
12d1982c71Sschwarze * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHORS BE LIABLE FOR
13a6464425Sschwarze * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
14a6464425Sschwarze * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
15a6464425Sschwarze * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
16a6464425Sschwarze * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
170ac7e6ecSschwarze *
180ac7e6ecSschwarze * Validation module for man(7) syntax trees used by mandoc(1).
19f73abda9Skristaps */
20f73abda9Skristaps #include <sys/types.h>
21f73abda9Skristaps
22f73abda9Skristaps #include <assert.h>
23f73abda9Skristaps #include <ctype.h>
248dc37b1cSschwarze #include <errno.h>
258dc37b1cSschwarze #include <limits.h>
26f73abda9Skristaps #include <stdarg.h>
27e501e731Sschwarze #include <stdio.h>
28f73abda9Skristaps #include <stdlib.h>
29ddce0b0cSschwarze #include <string.h>
30b58a7f4fSschwarze #include <time.h>
31f73abda9Skristaps
324f4f7972Sschwarze #include "mandoc_aux.h"
33d1982c71Sschwarze #include "mandoc.h"
34*f6697133Sschwarze #include "mandoc_xr.h"
35d1982c71Sschwarze #include "roff.h"
36d1982c71Sschwarze #include "man.h"
37f6854d5cSschwarze #include "libmandoc.h"
38fa2127f9Sschwarze #include "roff_int.h"
39d1982c71Sschwarze #include "libman.h"
406e2a0df9Sschwarze #include "tag.h"
41f73abda9Skristaps
42ede1b9d0Sschwarze #define CHKARGS struct roff_man *man, struct roff_node *n
43f73abda9Skristaps
44eae2af84Sschwarze typedef void (*v_check)(CHKARGS);
45f73abda9Skristaps
46eae2af84Sschwarze static void check_par(CHKARGS);
47eae2af84Sschwarze static void check_part(CHKARGS);
48eae2af84Sschwarze static void check_root(CHKARGS);
490ac7e6ecSschwarze static void check_tag(struct roff_node *, struct roff_node *);
50eae2af84Sschwarze static void check_text(CHKARGS);
51f73abda9Skristaps
52eae2af84Sschwarze static void post_AT(CHKARGS);
537faedc4aSschwarze static void post_EE(CHKARGS);
547faedc4aSschwarze static void post_EX(CHKARGS);
55eae2af84Sschwarze static void post_IP(CHKARGS);
56*f6697133Sschwarze static void post_MR(CHKARGS);
5739839348Sschwarze static void post_OP(CHKARGS);
588251afdeSschwarze static void post_SH(CHKARGS);
59eae2af84Sschwarze static void post_TH(CHKARGS);
600ac7e6ecSschwarze static void post_TP(CHKARGS);
61eae2af84Sschwarze static void post_UC(CHKARGS);
62eae2af84Sschwarze static void post_UR(CHKARGS);
63318aa4c1Sschwarze static void post_in(CHKARGS);
648dc07b1eSschwarze
65b3f54129Sschwarze static const v_check man_valids[MAN_MAX - MAN_TH] = {
66492e55cdSschwarze post_TH, /* TH */
678251afdeSschwarze post_SH, /* SH */
688251afdeSschwarze post_SH, /* SS */
690ac7e6ecSschwarze post_TP, /* TP */
700ac7e6ecSschwarze post_TP, /* TQ */
713f3c303aSschwarze check_par, /* LP */
72492e55cdSschwarze check_par, /* PP */
733f3c303aSschwarze check_par, /* P */
74492e55cdSschwarze post_IP, /* IP */
75492e55cdSschwarze NULL, /* HP */
76492e55cdSschwarze NULL, /* SM */
77492e55cdSschwarze NULL, /* SB */
78492e55cdSschwarze NULL, /* BI */
79492e55cdSschwarze NULL, /* IB */
80492e55cdSschwarze NULL, /* BR */
81492e55cdSschwarze NULL, /* RB */
82492e55cdSschwarze NULL, /* R */
83492e55cdSschwarze NULL, /* B */
84492e55cdSschwarze NULL, /* I */
85492e55cdSschwarze NULL, /* IR */
86492e55cdSschwarze NULL, /* RI */
87492e55cdSschwarze NULL, /* RE */
88492e55cdSschwarze check_part, /* RS */
89492e55cdSschwarze NULL, /* DT */
90492e55cdSschwarze post_UC, /* UC */
91e5d5fd76Sschwarze NULL, /* PD */
92492e55cdSschwarze post_AT, /* AT */
93318aa4c1Sschwarze post_in, /* in */
945e5a9c61Sschwarze NULL, /* SY */
955e5a9c61Sschwarze NULL, /* YS */
9639839348Sschwarze post_OP, /* OP */
977faedc4aSschwarze post_EX, /* EX */
987faedc4aSschwarze post_EE, /* EE */
99492e55cdSschwarze post_UR, /* UR */
100492e55cdSschwarze NULL, /* UE */
101df9a9479Sbentley post_UR, /* MT */
102df9a9479Sbentley NULL, /* ME */
103*f6697133Sschwarze post_MR, /* MR */
104f73abda9Skristaps };
105f73abda9Skristaps
106f73abda9Skristaps
1077c539ecbSschwarze /* Validate the subtree rooted at man->last. */
108eae2af84Sschwarze void
man_validate(struct roff_man * man)10983d65a5aSschwarze man_validate(struct roff_man *man)
110e35cb253Sschwarze {
1113a0d07afSschwarze struct roff_node *n;
11214a309e3Sschwarze const v_check *cp;
113e35cb253Sschwarze
1147c539ecbSschwarze /*
1157c539ecbSschwarze * Iterate over all children, recursing into each one
1167c539ecbSschwarze * in turn, depth-first.
1177c539ecbSschwarze */
1187c539ecbSschwarze
1193f3c303aSschwarze n = man->last;
120b3cd4c9aSschwarze man->last = man->last->child;
121b3cd4c9aSschwarze while (man->last != NULL) {
12283d65a5aSschwarze man_validate(man);
123b3cd4c9aSschwarze if (man->last == n)
124b3cd4c9aSschwarze man->last = man->last->child;
125b3cd4c9aSschwarze else
126b3cd4c9aSschwarze man->last = man->last->next;
127b3cd4c9aSschwarze }
128492e55cdSschwarze
1297c539ecbSschwarze /* Finally validate the macro itself. */
1307c539ecbSschwarze
131b3cd4c9aSschwarze man->last = n;
132b3cd4c9aSschwarze man->next = ROFF_NEXT_SIBLING;
1332791bd1cSschwarze switch (n->type) {
134d1982c71Sschwarze case ROFFT_TEXT:
135eae2af84Sschwarze check_text(man, n);
136eae2af84Sschwarze break;
137d1982c71Sschwarze case ROFFT_ROOT:
138eae2af84Sschwarze check_root(man, n);
139eae2af84Sschwarze break;
1404c293873Sschwarze case ROFFT_COMMENT:
141d1982c71Sschwarze case ROFFT_EQN:
142d1982c71Sschwarze case ROFFT_TBL:
143eae2af84Sschwarze break;
1442791bd1cSschwarze default:
14529478532Sschwarze if (n->tok < ROFF_MAX) {
146c4d3fa85Sschwarze roff_validate(man);
147c4d3fa85Sschwarze break;
14829478532Sschwarze }
14929478532Sschwarze assert(n->tok >= MAN_TH && n->tok < MAN_MAX);
150b3f54129Sschwarze cp = man_valids + (n->tok - MAN_TH);
151eae2af84Sschwarze if (*cp)
152eae2af84Sschwarze (*cp)(man, n);
153b3cd4c9aSschwarze if (man->last == n)
154542a495fSschwarze n->flags |= NODE_VALID;
155eae2af84Sschwarze break;
1562791bd1cSschwarze }
157f73abda9Skristaps }
158f73abda9Skristaps
159eae2af84Sschwarze static void
check_root(CHKARGS)160e35cb253Sschwarze check_root(CHKARGS)
161f73abda9Skristaps {
16247400284Sschwarze assert((man->flags & (MAN_BLINE | MAN_ELINE)) == 0);
163e35cb253Sschwarze
1644c293873Sschwarze if (n->last == NULL || n->last->type == ROFFT_COMMENT)
165a5a5f808Sschwarze mandoc_msg(MANDOCERR_DOC_EMPTY, n->line, n->pos, NULL);
16643edbcc8Sschwarze else
16743edbcc8Sschwarze man->meta.hasbody = 1;
16843edbcc8Sschwarze
16943edbcc8Sschwarze if (NULL == man->meta.title) {
170a5a5f808Sschwarze mandoc_msg(MANDOCERR_TH_NOTITLE, n->line, n->pos, NULL);
1715eb77048Sschwarze
172b16e7ddfSschwarze /*
173b16e7ddfSschwarze * If a title hasn't been set, do so now (by
174b16e7ddfSschwarze * implication, date and section also aren't set).
175b16e7ddfSschwarze */
1765eb77048Sschwarze
1773fdead0cSschwarze man->meta.title = mandoc_strdup("");
1783fdead0cSschwarze man->meta.msec = mandoc_strdup("");
179ea5923abSschwarze man->meta.date = mandoc_normdate(NULL, NULL);
180ca68802eSschwarze }
181172864f7Sschwarze
182172864f7Sschwarze if (man->meta.os_e &&
183172864f7Sschwarze (man->meta.rcsids & (1 << man->meta.os_e)) == 0)
184a5a5f808Sschwarze mandoc_msg(MANDOCERR_RCS_MISSING, 0, 0,
185f3476b07Sschwarze man->meta.os_e == MANDOC_OS_OPENBSD ?
186f3476b07Sschwarze "(OpenBSD)" : "(NetBSD)");
1870adfad74Sschwarze }
1880adfad74Sschwarze
1890ac7e6ecSschwarze /*
1900ac7e6ecSschwarze * Skip leading whitespace, dashes, backslashes, and font escapes,
1910ac7e6ecSschwarze * then create a tag if the first following byte is a letter.
1920ac7e6ecSschwarze * Priority is high unless whitespace is present.
1930ac7e6ecSschwarze */
1940ac7e6ecSschwarze static void
check_tag(struct roff_node * n,struct roff_node * nt)1950ac7e6ecSschwarze check_tag(struct roff_node *n, struct roff_node *nt)
1960ac7e6ecSschwarze {
1970ac7e6ecSschwarze const char *cp, *arg;
1980ac7e6ecSschwarze int prio, sz;
1990ac7e6ecSschwarze
2000ac7e6ecSschwarze if (nt == NULL || nt->type != ROFFT_TEXT)
2010ac7e6ecSschwarze return;
2020ac7e6ecSschwarze
2030ac7e6ecSschwarze cp = nt->string;
2040ac7e6ecSschwarze prio = TAG_STRONG;
2050ac7e6ecSschwarze for (;;) {
2060ac7e6ecSschwarze switch (*cp) {
2070ac7e6ecSschwarze case ' ':
2080ac7e6ecSschwarze case '\t':
2090ac7e6ecSschwarze prio = TAG_WEAK;
2100ac7e6ecSschwarze /* FALLTHROUGH */
2110ac7e6ecSschwarze case '-':
2120ac7e6ecSschwarze cp++;
2130ac7e6ecSschwarze break;
2140ac7e6ecSschwarze case '\\':
2150ac7e6ecSschwarze cp++;
2160ac7e6ecSschwarze switch (mandoc_escape(&cp, &arg, &sz)) {
2170ac7e6ecSschwarze case ESCAPE_FONT:
2180ac7e6ecSschwarze case ESCAPE_FONTBOLD:
2190ac7e6ecSschwarze case ESCAPE_FONTITALIC:
2200ac7e6ecSschwarze case ESCAPE_FONTBI:
2210ac7e6ecSschwarze case ESCAPE_FONTROMAN:
2227d063611Sschwarze case ESCAPE_FONTCR:
2237d063611Sschwarze case ESCAPE_FONTCB:
2247d063611Sschwarze case ESCAPE_FONTCI:
2250ac7e6ecSschwarze case ESCAPE_FONTPREV:
2260ac7e6ecSschwarze case ESCAPE_IGNORE:
2270ac7e6ecSschwarze break;
2280ac7e6ecSschwarze case ESCAPE_SPECIAL:
2290ac7e6ecSschwarze if (sz != 1)
2300ac7e6ecSschwarze return;
2310ac7e6ecSschwarze switch (*arg) {
2320ac7e6ecSschwarze case '-':
2330ac7e6ecSschwarze case 'e':
2340ac7e6ecSschwarze break;
2350ac7e6ecSschwarze default:
2360ac7e6ecSschwarze return;
2370ac7e6ecSschwarze }
2380ac7e6ecSschwarze break;
2390ac7e6ecSschwarze default:
2400ac7e6ecSschwarze return;
2410ac7e6ecSschwarze }
2420ac7e6ecSschwarze break;
2430ac7e6ecSschwarze default:
2440ac7e6ecSschwarze if (isalpha((unsigned char)*cp))
2450ac7e6ecSschwarze tag_put(cp, prio, n);
2460ac7e6ecSschwarze return;
2470ac7e6ecSschwarze }
2480ac7e6ecSschwarze }
2490ac7e6ecSschwarze }
2500ac7e6ecSschwarze
2517c539ecbSschwarze static void
check_text(CHKARGS)252e35cb253Sschwarze check_text(CHKARGS)
2530adfad74Sschwarze {
25404e980cbSschwarze char *cp, *p;
2550adfad74Sschwarze
2567faedc4aSschwarze if (n->flags & NODE_NOFILL)
257eae2af84Sschwarze return;
2581cdbf331Sschwarze
2591cdbf331Sschwarze cp = n->string;
2601cdbf331Sschwarze for (p = cp; NULL != (p = strchr(p, '\t')); p++)
261a5a5f808Sschwarze mandoc_msg(MANDOCERR_FI_TAB,
262a5a5f808Sschwarze n->line, n->pos + (int)(p - cp), NULL);
263a5e11edeSschwarze }
264f73abda9Skristaps
26539839348Sschwarze static void
post_EE(CHKARGS)2667faedc4aSschwarze post_EE(CHKARGS)
2677faedc4aSschwarze {
2687faedc4aSschwarze if ((n->flags & NODE_NOFILL) == 0)
2697faedc4aSschwarze mandoc_msg(MANDOCERR_FI_SKIP, n->line, n->pos, "EE");
2707faedc4aSschwarze }
2717faedc4aSschwarze
2727faedc4aSschwarze static void
post_EX(CHKARGS)2737faedc4aSschwarze post_EX(CHKARGS)
2747faedc4aSschwarze {
2757faedc4aSschwarze if (n->flags & NODE_NOFILL)
2767faedc4aSschwarze mandoc_msg(MANDOCERR_NF_SKIP, n->line, n->pos, "EX");
2777faedc4aSschwarze }
2787faedc4aSschwarze
2797faedc4aSschwarze static void
post_OP(CHKARGS)28039839348Sschwarze post_OP(CHKARGS)
28139839348Sschwarze {
282f73abda9Skristaps
28330e5ee06Sschwarze if (n->child == NULL)
284a5a5f808Sschwarze mandoc_msg(MANDOCERR_OP_EMPTY, n->line, n->pos, "OP");
28530e5ee06Sschwarze else if (n->child->next != NULL && n->child->next->next != NULL) {
28639839348Sschwarze n = n->child->next->next;
287a5a5f808Sschwarze mandoc_msg(MANDOCERR_ARG_EXCESS,
28839839348Sschwarze n->line, n->pos, "OP ... %s", n->string);
28939839348Sschwarze }
29039839348Sschwarze }
291f73abda9Skristaps
292eae2af84Sschwarze static void
post_SH(CHKARGS)2938251afdeSschwarze post_SH(CHKARGS)
2948251afdeSschwarze {
2958251afdeSschwarze struct roff_node *nc;
296e9f36175Sschwarze char *cp, *tag;
2978251afdeSschwarze
298e9f36175Sschwarze nc = n->child;
299e9f36175Sschwarze switch (n->type) {
300e9f36175Sschwarze case ROFFT_HEAD:
301e9f36175Sschwarze tag = NULL;
302e9f36175Sschwarze deroff(&tag, n);
303e9f36175Sschwarze if (tag != NULL) {
304e9f36175Sschwarze for (cp = tag; *cp != '\0'; cp++)
305e9f36175Sschwarze if (*cp == ' ')
306e9f36175Sschwarze *cp = '_';
307e9f36175Sschwarze if (nc != NULL && nc->type == ROFFT_TEXT &&
308e9f36175Sschwarze strcmp(nc->string, tag) == 0)
309e79243feSschwarze tag_put(NULL, TAG_STRONG, n);
310e9f36175Sschwarze else
311e9f36175Sschwarze tag_put(tag, TAG_FALLBACK, n);
312e9f36175Sschwarze free(tag);
313e9f36175Sschwarze }
3148251afdeSschwarze return;
315e9f36175Sschwarze case ROFFT_BODY:
316e9f36175Sschwarze if (nc != NULL)
317e9f36175Sschwarze break;
318e9f36175Sschwarze return;
319e9f36175Sschwarze default:
320e9f36175Sschwarze return;
321e9f36175Sschwarze }
3228251afdeSschwarze
3233f3c303aSschwarze if ((nc->tok == MAN_LP || nc->tok == MAN_PP || nc->tok == MAN_P) &&
3243f3c303aSschwarze nc->body->child != NULL) {
3258251afdeSschwarze while (nc->body->last != NULL) {
3268251afdeSschwarze man->next = ROFF_NEXT_CHILD;
3278251afdeSschwarze roff_node_relink(man, nc->body->last);
3288251afdeSschwarze man->last = n;
3298251afdeSschwarze }
3308251afdeSschwarze }
3318251afdeSschwarze
3323f3c303aSschwarze if (nc->tok == MAN_LP || nc->tok == MAN_PP || nc->tok == MAN_P ||
3333f3c303aSschwarze nc->tok == ROFF_sp || nc->tok == ROFF_br) {
334a5a5f808Sschwarze mandoc_msg(MANDOCERR_PAR_SKIP, nc->line, nc->pos,
335a5a5f808Sschwarze "%s after %s", roff_name[nc->tok], roff_name[n->tok]);
3368251afdeSschwarze roff_node_delete(man, nc);
3378251afdeSschwarze }
3388251afdeSschwarze
3398251afdeSschwarze /*
3408251afdeSschwarze * Trailing PP is empty, so it is deleted by check_par().
3418251afdeSschwarze * Trailing sp is significant.
3428251afdeSschwarze */
3438251afdeSschwarze
3448251afdeSschwarze if ((nc = n->last) != NULL && nc->tok == ROFF_br) {
345a5a5f808Sschwarze mandoc_msg(MANDOCERR_PAR_SKIP,
3468251afdeSschwarze nc->line, nc->pos, "%s at the end of %s",
3478251afdeSschwarze roff_name[nc->tok], roff_name[n->tok]);
3488251afdeSschwarze roff_node_delete(man, nc);
3498251afdeSschwarze }
3508251afdeSschwarze }
3518251afdeSschwarze
3528251afdeSschwarze static void
post_UR(CHKARGS)353492e55cdSschwarze post_UR(CHKARGS)
3543aeff926Sschwarze {
355d1982c71Sschwarze if (n->type == ROFFT_HEAD && n->child == NULL)
356a5a5f808Sschwarze mandoc_msg(MANDOCERR_UR_NOHEAD, n->line, n->pos,
357a5a5f808Sschwarze "%s", roff_name[n->tok]);
3583aeff926Sschwarze }
3593aeff926Sschwarze
360eae2af84Sschwarze static void
check_part(CHKARGS)361fbaebbd2Sschwarze check_part(CHKARGS)
362fbaebbd2Sschwarze {
363d1982c71Sschwarze if (n->type == ROFFT_BODY && n->child == NULL)
364a5a5f808Sschwarze mandoc_msg(MANDOCERR_BLK_EMPTY, n->line, n->pos,
365a5a5f808Sschwarze "%s", roff_name[n->tok]);
366fbaebbd2Sschwarze }
367fbaebbd2Sschwarze
368eae2af84Sschwarze static void
check_par(CHKARGS)369b58a7f4fSschwarze check_par(CHKARGS)
370b58a7f4fSschwarze {
371b58a7f4fSschwarze
3725e94b789Sschwarze switch (n->type) {
373d1982c71Sschwarze case ROFFT_BLOCK:
37430e5ee06Sschwarze if (n->body->child == NULL)
375fa2127f9Sschwarze roff_node_delete(man, n);
376b58a7f4fSschwarze break;
377d1982c71Sschwarze case ROFFT_BODY:
3788251afdeSschwarze if (n->child != NULL &&
3798251afdeSschwarze (n->child->tok == ROFF_sp || n->child->tok == ROFF_br)) {
380a5a5f808Sschwarze mandoc_msg(MANDOCERR_PAR_SKIP,
381a5a5f808Sschwarze n->child->line, n->child->pos,
3828251afdeSschwarze "%s after %s", roff_name[n->child->tok],
3838251afdeSschwarze roff_name[n->tok]);
3848251afdeSschwarze roff_node_delete(man, n->child);
3858251afdeSschwarze }
38630e5ee06Sschwarze if (n->child == NULL)
387a5a5f808Sschwarze mandoc_msg(MANDOCERR_PAR_SKIP, n->line, n->pos,
38814a309e3Sschwarze "%s empty", roff_name[n->tok]);
389b58a7f4fSschwarze break;
390d1982c71Sschwarze case ROFFT_HEAD:
39130e5ee06Sschwarze if (n->child != NULL)
392a5a5f808Sschwarze mandoc_msg(MANDOCERR_ARG_SKIP,
393a5a5f808Sschwarze n->line, n->pos, "%s %s%s",
39414a309e3Sschwarze roff_name[n->tok], n->child->string,
39530e5ee06Sschwarze n->child->next != NULL ? " ..." : "");
396b58a7f4fSschwarze break;
397b58a7f4fSschwarze default:
398b58a7f4fSschwarze break;
399b58a7f4fSschwarze }
400b58a7f4fSschwarze }
401b58a7f4fSschwarze
402eae2af84Sschwarze static void
post_IP(CHKARGS)403fe61e474Sschwarze post_IP(CHKARGS)
404fe61e474Sschwarze {
405fe61e474Sschwarze switch (n->type) {
406d1982c71Sschwarze case ROFFT_BLOCK:
40730e5ee06Sschwarze if (n->head->child == NULL && n->body->child == NULL)
408fa2127f9Sschwarze roff_node_delete(man, n);
409fe61e474Sschwarze break;
4100ac7e6ecSschwarze case ROFFT_HEAD:
4110ac7e6ecSschwarze check_tag(n, n->child);
4120ac7e6ecSschwarze break;
413d1982c71Sschwarze case ROFFT_BODY:
41430e5ee06Sschwarze if (n->parent->head->child == NULL && n->child == NULL)
415a5a5f808Sschwarze mandoc_msg(MANDOCERR_PAR_SKIP, n->line, n->pos,
41614a309e3Sschwarze "%s empty", roff_name[n->tok]);
417fe61e474Sschwarze break;
418fe61e474Sschwarze default:
419fe61e474Sschwarze break;
420fe61e474Sschwarze }
421fe61e474Sschwarze }
422b58a7f4fSschwarze
4230ac7e6ecSschwarze /*
4240ac7e6ecSschwarze * The first next-line element in the head is the tag.
4250ac7e6ecSschwarze * If that's a font macro, use its first child instead.
4260ac7e6ecSschwarze */
4270ac7e6ecSschwarze static void
post_TP(CHKARGS)4280ac7e6ecSschwarze post_TP(CHKARGS)
4290ac7e6ecSschwarze {
4300ac7e6ecSschwarze struct roff_node *nt;
4310ac7e6ecSschwarze
4320ac7e6ecSschwarze if (n->type != ROFFT_HEAD || (nt = n->child) == NULL)
4330ac7e6ecSschwarze return;
4340ac7e6ecSschwarze
4350ac7e6ecSschwarze while ((nt->flags & NODE_LINE) == 0)
4360ac7e6ecSschwarze if ((nt = nt->next) == NULL)
4370ac7e6ecSschwarze return;
4380ac7e6ecSschwarze
4390ac7e6ecSschwarze switch (nt->tok) {
4400ac7e6ecSschwarze case MAN_B:
4410ac7e6ecSschwarze case MAN_BI:
4420ac7e6ecSschwarze case MAN_BR:
4430ac7e6ecSschwarze case MAN_I:
4440ac7e6ecSschwarze case MAN_IB:
4450ac7e6ecSschwarze case MAN_IR:
4460ac7e6ecSschwarze nt = nt->child;
4470ac7e6ecSschwarze break;
4480ac7e6ecSschwarze default:
4490ac7e6ecSschwarze break;
4500ac7e6ecSschwarze }
4510ac7e6ecSschwarze check_tag(n, nt);
4520ac7e6ecSschwarze }
4530ac7e6ecSschwarze
454eae2af84Sschwarze static void
post_TH(CHKARGS)4558dc07b1eSschwarze post_TH(CHKARGS)
4568dc07b1eSschwarze {
4573a0d07afSschwarze struct roff_node *nb;
458ea200224Sschwarze const char *p;
4598dc07b1eSschwarze
4607ead8a4eSschwarze free(man->meta.title);
4617ead8a4eSschwarze free(man->meta.vol);
4622a238f45Sschwarze free(man->meta.os);
4637ead8a4eSschwarze free(man->meta.msec);
4647ead8a4eSschwarze free(man->meta.date);
4658dc07b1eSschwarze
4667ead8a4eSschwarze man->meta.title = man->meta.vol = man->meta.date =
4672a238f45Sschwarze man->meta.msec = man->meta.os = NULL;
4688dc07b1eSschwarze
4691d5a8eebSschwarze nb = n;
4701d5a8eebSschwarze
4712a238f45Sschwarze /* ->TITLE<- MSEC DATE OS VOL */
4728dc07b1eSschwarze
4738dc07b1eSschwarze n = n->child;
4740c0f292bSschwarze if (n != NULL && n->string != NULL) {
4750c0f292bSschwarze for (p = n->string; *p != '\0'; p++) {
476ea200224Sschwarze /* Only warn about this once... */
47704e980cbSschwarze if (isalpha((unsigned char)*p) &&
47804e980cbSschwarze ! isupper((unsigned char)*p)) {
479a5a5f808Sschwarze mandoc_msg(MANDOCERR_TITLE_CASE, n->line,
480a5a5f808Sschwarze n->pos + (int)(p - n->string),
481bd594191Sschwarze "TH %s", n->string);
482ea200224Sschwarze break;
483ea200224Sschwarze }
484ea200224Sschwarze }
4857ead8a4eSschwarze man->meta.title = mandoc_strdup(n->string);
4863fdead0cSschwarze } else {
4877ead8a4eSschwarze man->meta.title = mandoc_strdup("");
488a5a5f808Sschwarze mandoc_msg(MANDOCERR_TH_NOTITLE, nb->line, nb->pos, "TH");
4893fdead0cSschwarze }
4908dc07b1eSschwarze
4912a238f45Sschwarze /* TITLE ->MSEC<- DATE OS VOL */
4928dc07b1eSschwarze
4930c0f292bSschwarze if (n != NULL)
4948dc07b1eSschwarze n = n->next;
495ee646987Sschwarze if (n != NULL && n->string != NULL) {
4967ead8a4eSschwarze man->meta.msec = mandoc_strdup(n->string);
497ee646987Sschwarze if (man->filesec != '\0' &&
498ee646987Sschwarze man->filesec != *n->string &&
499ee646987Sschwarze *n->string >= '1' && *n->string <= '9')
500ee646987Sschwarze mandoc_msg(MANDOCERR_MSEC_FILE, n->line, n->pos,
501ee646987Sschwarze "*.%c vs TH ... %c", man->filesec, *n->string);
502ee646987Sschwarze } else {
5037ead8a4eSschwarze man->meta.msec = mandoc_strdup("");
504a5a5f808Sschwarze mandoc_msg(MANDOCERR_MSEC_MISSING,
5053fdead0cSschwarze nb->line, nb->pos, "TH %s", man->meta.title);
5063fdead0cSschwarze }
5078dc07b1eSschwarze
5082a238f45Sschwarze /* TITLE MSEC ->DATE<- OS VOL */
5098dc07b1eSschwarze
5100c0f292bSschwarze if (n != NULL)
5118dc07b1eSschwarze n = n->next;
512ea5923abSschwarze if (man->quick && n != NULL)
5137ead8a4eSschwarze man->meta.date = mandoc_strdup("");
514ea5923abSschwarze else
515ea5923abSschwarze man->meta.date = mandoc_normdate(n, nb);
5168dc07b1eSschwarze
5172a238f45Sschwarze /* TITLE MSEC DATE ->OS<- VOL */
5188dc07b1eSschwarze
5198dc07b1eSschwarze if (n && (n = n->next))
5202a238f45Sschwarze man->meta.os = mandoc_strdup(n->string);
521f3476b07Sschwarze else if (man->os_s != NULL)
522f3476b07Sschwarze man->meta.os = mandoc_strdup(man->os_s);
523f3476b07Sschwarze if (man->meta.os_e == MANDOC_OS_OTHER && man->meta.os != NULL) {
524f3476b07Sschwarze if (strstr(man->meta.os, "OpenBSD") != NULL)
525f3476b07Sschwarze man->meta.os_e = MANDOC_OS_OPENBSD;
526f3476b07Sschwarze else if (strstr(man->meta.os, "NetBSD") != NULL)
527f3476b07Sschwarze man->meta.os_e = MANDOC_OS_NETBSD;
528f3476b07Sschwarze }
5298dc07b1eSschwarze
5302a238f45Sschwarze /* TITLE MSEC DATE OS ->VOL<- */
53188ec69e3Sschwarze /* If missing, use the default VOL name for MSEC. */
5328dc07b1eSschwarze
5338dc07b1eSschwarze if (n && (n = n->next))
5347ead8a4eSschwarze man->meta.vol = mandoc_strdup(n->string);
5357ead8a4eSschwarze else if ('\0' != man->meta.msec[0] &&
5367ead8a4eSschwarze (NULL != (p = mandoc_a2msec(man->meta.msec))))
5377ead8a4eSschwarze man->meta.vol = mandoc_strdup(p);
5388dc07b1eSschwarze
53939839348Sschwarze if (n != NULL && (n = n->next) != NULL)
540a5a5f808Sschwarze mandoc_msg(MANDOCERR_ARG_EXCESS,
54139839348Sschwarze n->line, n->pos, "TH ... %s", n->string);
54239839348Sschwarze
5438dc07b1eSschwarze /*
5448dc07b1eSschwarze * Remove the `TH' node after we've processed it for our
5458dc07b1eSschwarze * meta-data.
5468dc07b1eSschwarze */
547fa2127f9Sschwarze roff_node_delete(man, man->last);
5488dc07b1eSschwarze }
5498dc07b1eSschwarze
550eae2af84Sschwarze static void
post_MR(CHKARGS)551*f6697133Sschwarze post_MR(CHKARGS)
552*f6697133Sschwarze {
553*f6697133Sschwarze struct roff_node *nch;
554*f6697133Sschwarze
555*f6697133Sschwarze if ((nch = n->child) == NULL) {
556*f6697133Sschwarze mandoc_msg(MANDOCERR_NM_NONAME, n->line, n->pos, "MR");
557*f6697133Sschwarze return;
558*f6697133Sschwarze }
559*f6697133Sschwarze if (nch->next == NULL) {
560*f6697133Sschwarze mandoc_msg(MANDOCERR_XR_NOSEC,
561*f6697133Sschwarze n->line, n->pos, "MR %s", nch->string);
562*f6697133Sschwarze return;
563*f6697133Sschwarze }
564*f6697133Sschwarze if (mandoc_xr_add(nch->next->string, nch->string, nch->line, nch->pos))
565*f6697133Sschwarze mandoc_msg(MANDOCERR_XR_SELF, nch->line, nch->pos,
566*f6697133Sschwarze "MR %s %s", nch->string, nch->next->string);
567*f6697133Sschwarze if ((nch = nch->next->next) == NULL || nch->next == NULL)
568*f6697133Sschwarze return;
569*f6697133Sschwarze
570*f6697133Sschwarze mandoc_msg(MANDOCERR_ARG_EXCESS, nch->next->line, nch->next->pos,
571*f6697133Sschwarze "MR ... %s", nch->next->string);
572*f6697133Sschwarze while (nch->next != NULL)
573*f6697133Sschwarze roff_node_delete(man, nch->next);
574*f6697133Sschwarze }
575*f6697133Sschwarze
576*f6697133Sschwarze static void
post_UC(CHKARGS)5778dc07b1eSschwarze post_UC(CHKARGS)
5788dc07b1eSschwarze {
5798dc07b1eSschwarze static const char * const bsd_versions[] = {
5808dc07b1eSschwarze "3rd Berkeley Distribution",
5818dc07b1eSschwarze "4th Berkeley Distribution",
5828dc07b1eSschwarze "4.2 Berkeley Distribution",
5838dc07b1eSschwarze "4.3 Berkeley Distribution",
5848dc07b1eSschwarze "4.4 Berkeley Distribution",
5858dc07b1eSschwarze };
5868dc07b1eSschwarze
5878dc07b1eSschwarze const char *p, *s;
5888dc07b1eSschwarze
5898dc07b1eSschwarze n = n->child;
5908dc07b1eSschwarze
591d1982c71Sschwarze if (n == NULL || n->type != ROFFT_TEXT)
5928dc07b1eSschwarze p = bsd_versions[0];
5938dc07b1eSschwarze else {
5948dc07b1eSschwarze s = n->string;
5958dc07b1eSschwarze if (0 == strcmp(s, "3"))
5968dc07b1eSschwarze p = bsd_versions[0];
5978dc07b1eSschwarze else if (0 == strcmp(s, "4"))
5988dc07b1eSschwarze p = bsd_versions[1];
5998dc07b1eSschwarze else if (0 == strcmp(s, "5"))
6008dc07b1eSschwarze p = bsd_versions[2];
6018dc07b1eSschwarze else if (0 == strcmp(s, "6"))
6028dc07b1eSschwarze p = bsd_versions[3];
6038dc07b1eSschwarze else if (0 == strcmp(s, "7"))
6048dc07b1eSschwarze p = bsd_versions[4];
6058dc07b1eSschwarze else
6068dc07b1eSschwarze p = bsd_versions[0];
6078dc07b1eSschwarze }
6088dc07b1eSschwarze
6092a238f45Sschwarze free(man->meta.os);
6102a238f45Sschwarze man->meta.os = mandoc_strdup(p);
6118dc07b1eSschwarze }
6128dc07b1eSschwarze
613eae2af84Sschwarze static void
post_AT(CHKARGS)6148dc07b1eSschwarze post_AT(CHKARGS)
6158dc07b1eSschwarze {
6168dc07b1eSschwarze static const char * const unix_versions[] = {
6178dc07b1eSschwarze "7th Edition",
6188dc07b1eSschwarze "System III",
6198dc07b1eSschwarze "System V",
6208dc07b1eSschwarze "System V Release 2",
6218dc07b1eSschwarze };
6228dc07b1eSschwarze
6233a0d07afSschwarze struct roff_node *nn;
6248dc07b1eSschwarze const char *p, *s;
6258dc07b1eSschwarze
6268dc07b1eSschwarze n = n->child;
6278dc07b1eSschwarze
628d1982c71Sschwarze if (n == NULL || n->type != ROFFT_TEXT)
6298dc07b1eSschwarze p = unix_versions[0];
6308dc07b1eSschwarze else {
6318dc07b1eSschwarze s = n->string;
6328dc07b1eSschwarze if (0 == strcmp(s, "3"))
6338dc07b1eSschwarze p = unix_versions[0];
6348dc07b1eSschwarze else if (0 == strcmp(s, "4"))
6358dc07b1eSschwarze p = unix_versions[1];
6368dc07b1eSschwarze else if (0 == strcmp(s, "5")) {
6378dc07b1eSschwarze nn = n->next;
638d1982c71Sschwarze if (nn != NULL &&
639d1982c71Sschwarze nn->type == ROFFT_TEXT &&
640d1982c71Sschwarze nn->string[0] != '\0')
6418dc07b1eSschwarze p = unix_versions[3];
6428dc07b1eSschwarze else
6438dc07b1eSschwarze p = unix_versions[2];
6448dc07b1eSschwarze } else
6458dc07b1eSschwarze p = unix_versions[0];
6468dc07b1eSschwarze }
6478dc07b1eSschwarze
6482a238f45Sschwarze free(man->meta.os);
6492a238f45Sschwarze man->meta.os = mandoc_strdup(p);
6508dc07b1eSschwarze }
65188d02d4fSschwarze
652eae2af84Sschwarze static void
post_in(CHKARGS)653318aa4c1Sschwarze post_in(CHKARGS)
654318aa4c1Sschwarze {
655318aa4c1Sschwarze char *s;
656318aa4c1Sschwarze
657318aa4c1Sschwarze if (n->parent->tok != MAN_TP ||
658318aa4c1Sschwarze n->parent->type != ROFFT_HEAD ||
659318aa4c1Sschwarze n->child == NULL ||
660318aa4c1Sschwarze *n->child->string == '+' ||
661318aa4c1Sschwarze *n->child->string == '-')
662318aa4c1Sschwarze return;
663318aa4c1Sschwarze mandoc_asprintf(&s, "+%s", n->child->string);
664318aa4c1Sschwarze free(n->child->string);
665318aa4c1Sschwarze n->child->string = s;
666318aa4c1Sschwarze }
667