1 /* $Id: mdoc_validate.c,v 1.389 2021/07/18 11:41:23 schwarze Exp $ */
3 * Copyright (c) 2010-2020 Ingo Schwarze <schwarze@openbsd.org>
4 * Copyright (c) 2008-2012 Kristaps Dzonsons <kristaps@bsd.lv>
5 * Copyright (c) 2010 Joerg Sonnenberger <joerg@netbsd.org>
7 * Permission to use, copy, modify, and distribute this software for any
8 * purpose with or without fee is hereby granted, provided that the above
9 * copyright notice and this permission notice appear in all copies.
11 * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHORS DISCLAIM ALL WARRANTIES
12 * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
13 * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHORS BE LIABLE FOR
14 * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
15 * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
16 * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
17 * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
19 * Validation module for mdoc(7) syntax trees used by mandoc(1).
23 #include <sys/types.h>
25 #include <sys/utsname.h>
36 #include "mandoc_aux.h"
38 #include "mandoc_xr.h"
41 #include "libmandoc.h"
46 /* FIXME: .Bl -diag can't have non-text children in HEAD. */
48 #define POST_ARGS struct roff_man *mdoc
56 typedef void (*v_post)(POST_ARGS);
58 static int build_list(struct roff_man *, int);
59 static void check_argv(struct roff_man *,
60 struct roff_node *, struct mdoc_argv *);
61 static void check_args(struct roff_man *, struct roff_node *);
62 static void check_text(struct roff_man *, int, int, char *);
63 static void check_text_em(struct roff_man *, int, int, char *);
64 static void check_toptext(struct roff_man *, int, int, const char *);
65 static int child_an(const struct roff_node *);
66 static size_t macro2len(enum roff_tok);
67 static void rewrite_macro2len(struct roff_man *, char **);
68 static int similar(const char *, const char *);
70 static void post_abort(POST_ARGS) __attribute__((__noreturn__));
71 static void post_an(POST_ARGS);
72 static void post_an_norm(POST_ARGS);
73 static void post_at(POST_ARGS);
74 static void post_bd(POST_ARGS);
75 static void post_bf(POST_ARGS);
76 static void post_bk(POST_ARGS);
77 static void post_bl(POST_ARGS);
78 static void post_bl_block(POST_ARGS);
79 static void post_bl_head(POST_ARGS);
80 static void post_bl_norm(POST_ARGS);
81 static void post_bx(POST_ARGS);
82 static void post_defaults(POST_ARGS);
83 static void post_display(POST_ARGS);
84 static void post_dd(POST_ARGS);
85 static void post_delim(POST_ARGS);
86 static void post_delim_nb(POST_ARGS);
87 static void post_dt(POST_ARGS);
88 static void post_em(POST_ARGS);
89 static void post_en(POST_ARGS);
90 static void post_er(POST_ARGS);
91 static void post_es(POST_ARGS);
92 static void post_eoln(POST_ARGS);
93 static void post_ex(POST_ARGS);
94 static void post_fa(POST_ARGS);
95 static void post_fl(POST_ARGS);
96 static void post_fn(POST_ARGS);
97 static void post_fname(POST_ARGS);
98 static void post_fo(POST_ARGS);
99 static void post_hyph(POST_ARGS);
100 static void post_it(POST_ARGS);
101 static void post_lb(POST_ARGS);
102 static void post_nd(POST_ARGS);
103 static void post_nm(POST_ARGS);
104 static void post_ns(POST_ARGS);
105 static void post_obsolete(POST_ARGS);
106 static void post_os(POST_ARGS);
107 static void post_par(POST_ARGS);
108 static void post_prevpar(POST_ARGS);
109 static void post_root(POST_ARGS);
110 static void post_rs(POST_ARGS);
111 static void post_rv(POST_ARGS);
112 static void post_section(POST_ARGS);
113 static void post_sh(POST_ARGS);
114 static void post_sh_head(POST_ARGS);
115 static void post_sh_name(POST_ARGS);
116 static void post_sh_see_also(POST_ARGS);
117 static void post_sh_authors(POST_ARGS);
118 static void post_sm(POST_ARGS);
119 static void post_st(POST_ARGS);
120 static void post_std(POST_ARGS);
121 static void post_sx(POST_ARGS);
122 static void post_tag(POST_ARGS);
123 static void post_tg(POST_ARGS);
124 static void post_useless(POST_ARGS);
125 static void post_xr(POST_ARGS);
126 static void post_xx(POST_ARGS);
128 static const v_post mdoc_valids[MDOC_MAX - MDOC_Dd] = {
133 post_section, /* Ss */
135 post_display, /* D1 */
136 post_display, /* Dl */
137 post_display, /* Bd */
142 post_delim_nb, /* Ad */
145 post_defaults, /* Ar */
156 post_delim_nb, /* Ft */
158 post_delim_nb, /* In */
162 post_delim_nb, /* Op */
164 post_defaults, /* Pa */
168 post_delim_nb, /* Vt */
171 post_hyph, /* %B */ /* FIXME: can be used outside Rs/Re. */
179 post_hyph, /* %T */ /* FIXME: can be used outside Rs/Re. */
183 post_delim_nb, /* Aq */
191 post_obsolete, /* Db */
208 post_delim_nb, /* Pq */
210 post_delim_nb, /* Ql */
212 post_delim_nb, /* Qq */
217 post_delim_nb, /* Sq */
221 post_useless, /* Tn */
232 post_obsolete, /* Hf */
233 post_obsolete, /* Fr */
237 post_delim_nb, /* Lk */
238 post_defaults, /* Mt */
239 post_delim_nb, /* Brq */
252 #define RSORD_MAX 14 /* Number of `Rs' blocks. */
254 static const enum roff_tok rsord[RSORD_MAX] = {
271 static const char * const secnames[SEC__MAX] = {
278 "IMPLEMENTATION NOTES",
293 "SECURITY CONSIDERATIONS",
297 static int fn_prio = TAG_STRONG;
300 /* Validate the subtree rooted at mdoc->last. */
302 mdoc_validate(struct roff_man *mdoc)
304 struct roff_node *n, *np;
308 * Translate obsolete macros to modern macros first
309 * such that later code does not need to look
310 * for the obsolete versions.
327 * Iterate over all children, recursing into each one
328 * in turn, depth-first.
331 mdoc->last = mdoc->last->child;
332 while (mdoc->last != NULL) {
335 mdoc->last = mdoc->last->child;
337 mdoc->last = mdoc->last->next;
340 /* Finally validate the macro itself. */
343 mdoc->next = ROFF_NEXT_SIBLING;
347 if (n->sec != SEC_SYNOPSIS ||
348 (np->tok != MDOC_Cd && np->tok != MDOC_Fd))
349 check_text(mdoc, n->line, n->pos, n->string);
350 if ((n->flags & NODE_NOFILL) == 0 &&
351 (np->tok != MDOC_It || np->type != ROFFT_HEAD ||
352 np->parent->parent->norm->Bl.type != LIST_diag))
353 check_text_em(mdoc, n->line, n->pos, n->string);
354 if (np->tok == MDOC_It || (np->type == ROFFT_BODY &&
355 (np->tok == MDOC_Sh || np->tok == MDOC_Ss)))
356 check_toptext(mdoc, n->line, n->pos, n->string);
366 check_args(mdoc, mdoc->last);
369 * Closing delimiters are not special at the
370 * beginning of a block, opening delimiters
371 * are not special at the end.
374 if (n->child != NULL)
375 n->child->flags &= ~NODE_DELIMC;
377 n->last->flags &= ~NODE_DELIMO;
379 /* Call the macro's postprocessor. */
381 if (n->tok < ROFF_MAX) {
386 assert(n->tok >= MDOC_Dd && n->tok < MDOC_MAX);
387 p = mdoc_valids + (n->tok - MDOC_Dd);
397 check_args(struct roff_man *mdoc, struct roff_node *n)
404 assert(n->args->argc);
405 for (i = 0; i < (int)n->args->argc; i++)
406 check_argv(mdoc, n, &n->args->argv[i]);
410 check_argv(struct roff_man *mdoc, struct roff_node *n, struct mdoc_argv *v)
414 for (i = 0; i < (int)v->sz; i++)
415 check_text(mdoc, v->line, v->pos, v->value[i]);
419 check_text(struct roff_man *mdoc, int ln, int pos, char *p)
423 if (mdoc->last->flags & NODE_NOFILL)
426 for (cp = p; NULL != (p = strchr(p, '\t')); p++)
427 mandoc_msg(MANDOCERR_FI_TAB, ln, pos + (int)(p - cp), NULL);
431 check_text_em(struct roff_man *mdoc, int ln, int pos, char *p)
433 const struct roff_node *np, *nn;
436 np = mdoc->last->prev;
437 nn = mdoc->last->next;
439 /* Look for em-dashes wrongly encoded as "--". */
441 for (cp = p; *cp != '\0'; cp++) {
442 if (cp[0] != '-' || cp[1] != '-')
446 /* Skip input sequences of more than two '-'. */
454 /* Skip "--" directly attached to something else. */
456 if ((cp - p > 1 && cp[-2] != ' ') ||
457 (cp[1] != '\0' && cp[1] != ' '))
460 /* Require a letter right before or right afterwards. */
463 isalpha((unsigned char)cp[-3]) :
465 np->type == ROFFT_TEXT &&
466 *np->string != '\0' &&
467 isalpha((unsigned char)np->string[
468 strlen(np->string) - 1])) ||
469 (cp[1] != '\0' && cp[2] != '\0' ?
470 isalpha((unsigned char)cp[2]) :
472 nn->type == ROFFT_TEXT &&
473 isalpha((unsigned char)*nn->string))) {
474 mandoc_msg(MANDOCERR_DASHDASH,
475 ln, pos + (int)(cp - p) - 1, NULL);
482 check_toptext(struct roff_man *mdoc, int ln, int pos, const char *p)
484 const char *cp, *cpr;
489 if ((cp = strstr(p, "OpenBSD")) != NULL)
490 mandoc_msg(MANDOCERR_BX, ln, pos + (int)(cp - p), "Ox");
491 if ((cp = strstr(p, "NetBSD")) != NULL)
492 mandoc_msg(MANDOCERR_BX, ln, pos + (int)(cp - p), "Nx");
493 if ((cp = strstr(p, "FreeBSD")) != NULL)
494 mandoc_msg(MANDOCERR_BX, ln, pos + (int)(cp - p), "Fx");
495 if ((cp = strstr(p, "DragonFly")) != NULL)
496 mandoc_msg(MANDOCERR_BX, ln, pos + (int)(cp - p), "Dx");
499 while ((cp = strstr(cp + 1, "()")) != NULL) {
500 for (cpr = cp - 1; cpr >= p; cpr--)
501 if (*cpr != '_' && !isalnum((unsigned char)*cpr))
503 if ((cpr < p || *cpr == ' ') && cpr + 1 < cp) {
505 mandoc_msg(MANDOCERR_FUNC, ln, pos + (int)(cpr - p),
506 "%.*s()", (int)(cp - cpr), cpr);
512 post_abort(POST_ARGS)
518 post_delim(POST_ARGS)
520 const struct roff_node *nch;
525 tok = mdoc->last->tok;
526 nch = mdoc->last->last;
527 if (nch == NULL || nch->type != ROFFT_TEXT)
529 lc = strchr(nch->string, '\0') - 1;
530 if (lc < nch->string)
532 delim = mdoc_isdelim(lc);
533 if (delim == DELIM_NONE || delim == DELIM_OPEN)
535 if (*lc == ')' && (tok == MDOC_Nd || tok == MDOC_Sh ||
536 tok == MDOC_Ss || tok == MDOC_Fo))
539 mandoc_msg(MANDOCERR_DELIM, nch->line,
540 nch->pos + (int)(lc - nch->string), "%s%s %s", roff_name[tok],
541 nch == mdoc->last->child ? "" : " ...", nch->string);
545 post_delim_nb(POST_ARGS)
547 const struct roff_node *nch;
554 * Find candidates: at least two bytes,
555 * the last one a closing or middle delimiter.
558 tok = mdoc->last->tok;
559 nch = mdoc->last->last;
560 if (nch == NULL || nch->type != ROFFT_TEXT)
562 lc = strchr(nch->string, '\0') - 1;
563 if (lc <= nch->string)
565 delim = mdoc_isdelim(lc);
566 if (delim == DELIM_NONE || delim == DELIM_OPEN)
570 * Reduce false positives by allowing various cases.
573 /* Escaped delimiters. */
574 if (lc > nch->string + 1 && lc[-2] == '\\' &&
575 (lc[-1] == '&' || lc[-1] == 'e'))
578 /* Specific byte sequences. */
581 for (cp = lc; cp >= nch->string; cp--)
586 if (lc > nch->string + 1 && lc[-2] == '.' && lc[-1] == '.')
600 for (cp = lc; cp >= nch->string; cp--)
605 if (lc == nch->string + 1 && lc[-1] == '|')
611 /* Exactly two non-alphanumeric bytes. */
612 if (lc == nch->string + 1 && !isalnum((unsigned char)lc[-1]))
615 /* At least three alphabetic words with a sentence ending. */
616 if (strchr("!.:?", *lc) != NULL && (tok == MDOC_Em ||
617 tok == MDOC_Li || tok == MDOC_Pq || tok == MDOC_Sy)) {
619 for (cp = lc - 1; cp >= nch->string; cp--) {
622 if (cp > nch->string && cp[-1] == ',')
624 } else if (isalpha((unsigned int)*cp)) {
632 mandoc_msg(MANDOCERR_DELIM_NB, nch->line,
633 nch->pos + (int)(lc - nch->string), "%s%s %s", roff_name[tok],
634 nch == mdoc->last->child ? "" : " ...", nch->string);
638 post_bl_norm(POST_ARGS)
641 struct mdoc_argv *argv, *wa;
643 enum mdocargt mdoclt;
646 n = mdoc->last->parent;
647 n->norm->Bl.type = LIST__NONE;
650 * First figure out which kind of list to use: bind ourselves to
651 * the first mentioned list type and warn about any remaining
652 * ones. If we find no list type, we default to LIST_item.
655 wa = (n->args == NULL) ? NULL : n->args->argv;
656 mdoclt = MDOC_ARG_MAX;
657 for (i = 0; n->args && i < (int)n->args->argc; i++) {
658 argv = n->args->argv + i;
661 /* Set list types. */
695 /* Set list arguments. */
697 if (n->norm->Bl.comp)
698 mandoc_msg(MANDOCERR_ARG_REP,
699 argv->line, argv->pos, "Bl -compact");
700 n->norm->Bl.comp = 1;
705 mandoc_msg(MANDOCERR_ARG_EMPTY,
706 argv->line, argv->pos, "Bl -width");
707 n->norm->Bl.width = "0n";
710 if (NULL != n->norm->Bl.width)
711 mandoc_msg(MANDOCERR_ARG_REP,
712 argv->line, argv->pos,
713 "Bl -width %s", argv->value[0]);
714 rewrite_macro2len(mdoc, argv->value);
715 n->norm->Bl.width = argv->value[0];
719 mandoc_msg(MANDOCERR_ARG_EMPTY,
720 argv->line, argv->pos, "Bl -offset");
723 if (NULL != n->norm->Bl.offs)
724 mandoc_msg(MANDOCERR_ARG_REP,
725 argv->line, argv->pos,
726 "Bl -offset %s", argv->value[0]);
727 rewrite_macro2len(mdoc, argv->value);
728 n->norm->Bl.offs = argv->value[0];
733 if (LIST__NONE == lt)
737 /* Check: multiple list types. */
739 if (LIST__NONE != n->norm->Bl.type) {
740 mandoc_msg(MANDOCERR_BL_REP, n->line, n->pos,
741 "Bl -%s", mdoc_argnames[argv->arg]);
745 /* The list type should come first. */
747 if (n->norm->Bl.width ||
750 mandoc_msg(MANDOCERR_BL_LATETYPE,
751 n->line, n->pos, "Bl -%s",
752 mdoc_argnames[n->args->argv[0].arg]);
754 n->norm->Bl.type = lt;
755 if (LIST_column == lt) {
756 n->norm->Bl.ncols = argv->sz;
757 n->norm->Bl.cols = (void *)argv->value;
761 /* Allow lists to default to LIST_item. */
763 if (LIST__NONE == n->norm->Bl.type) {
764 mandoc_msg(MANDOCERR_BL_NOTYPE, n->line, n->pos, "Bl");
765 n->norm->Bl.type = LIST_item;
770 * Validate the width field. Some list types don't need width
771 * types and should be warned about them. Others should have it
772 * and must also be warned. Yet others have a default and need
776 switch (n->norm->Bl.type) {
778 if (n->norm->Bl.width == NULL)
779 mandoc_msg(MANDOCERR_BL_NOWIDTH,
780 n->line, n->pos, "Bl -tag");
787 if (n->norm->Bl.width != NULL)
788 mandoc_msg(MANDOCERR_BL_SKIPW, wa->line, wa->pos,
789 "Bl -%s", mdoc_argnames[mdoclt]);
790 n->norm->Bl.width = NULL;
795 if (n->norm->Bl.width == NULL)
796 n->norm->Bl.width = "2n";
799 if (n->norm->Bl.width == NULL)
800 n->norm->Bl.width = "3n";
811 struct mdoc_argv *argv;
816 for (i = 0; n->args && i < (int)n->args->argc; i++) {
817 argv = n->args->argv + i;
837 mandoc_msg(MANDOCERR_BD_FILE, n->line, n->pos, NULL);
841 mandoc_msg(MANDOCERR_ARG_EMPTY,
842 argv->line, argv->pos, "Bd -offset");
845 if (NULL != n->norm->Bd.offs)
846 mandoc_msg(MANDOCERR_ARG_REP,
847 argv->line, argv->pos,
848 "Bd -offset %s", argv->value[0]);
849 rewrite_macro2len(mdoc, argv->value);
850 n->norm->Bd.offs = argv->value[0];
853 if (n->norm->Bd.comp)
854 mandoc_msg(MANDOCERR_ARG_REP,
855 argv->line, argv->pos, "Bd -compact");
856 n->norm->Bd.comp = 1;
861 if (DISP__NONE == dt)
864 if (DISP__NONE == n->norm->Bd.type)
865 n->norm->Bd.type = dt;
867 mandoc_msg(MANDOCERR_BD_REP, n->line, n->pos,
868 "Bd -%s", mdoc_argnames[argv->arg]);
871 if (DISP__NONE == n->norm->Bd.type) {
872 mandoc_msg(MANDOCERR_BD_NOTYPE, n->line, n->pos, "Bd");
873 n->norm->Bd.type = DISP_ragged;
878 * Stand-alone line macros.
882 post_an_norm(POST_ARGS)
885 struct mdoc_argv *argv;
892 for (i = 1; i < n->args->argc; i++) {
893 argv = n->args->argv + i;
894 mandoc_msg(MANDOCERR_AN_REP, argv->line, argv->pos,
895 "An -%s", mdoc_argnames[argv->arg]);
898 argv = n->args->argv;
899 if (argv->arg == MDOC_Split)
900 n->norm->An.auth = AUTH_split;
901 else if (argv->arg == MDOC_Nosplit)
902 n->norm->An.auth = AUTH_nosplit;
914 if (n->child != NULL)
915 mandoc_msg(MANDOCERR_ARG_SKIP, n->line,
916 n->pos, "%s %s", roff_name[n->tok], n->child->string);
918 while (n->child != NULL)
919 roff_node_delete(mdoc, n->child);
921 roff_word_alloc(mdoc, n->line, n->pos, n->tok == MDOC_Bt ?
922 "is currently in beta test." : "currently under development.");
923 mdoc->last->flags |= NODE_EOS | NODE_NOSRC;
928 build_list(struct roff_man *mdoc, int tok)
933 n = mdoc->last->next;
934 for (ic = 1;; ic++) {
935 roff_elem_alloc(mdoc, n->line, n->pos, tok);
936 mdoc->last->flags |= NODE_NOSRC;
937 roff_node_relink(mdoc, n);
938 n = mdoc->last = mdoc->last->parent;
939 mdoc->next = ROFF_NEXT_SIBLING;
942 if (ic > 1 || n->next->next != NULL) {
943 roff_word_alloc(mdoc, n->line, n->pos, ",");
944 mdoc->last->flags |= NODE_DELIMC | NODE_NOSRC;
946 n = mdoc->last->next;
947 if (n->next == NULL) {
948 roff_word_alloc(mdoc, n->line, n->pos, "and");
949 mdoc->last->flags |= NODE_NOSRC;
963 mdoc->next = ROFF_NEXT_CHILD;
964 roff_word_alloc(mdoc, n->line, n->pos, "The");
965 mdoc->last->flags |= NODE_NOSRC;
967 if (mdoc->last->next != NULL)
968 ic = build_list(mdoc, MDOC_Nm);
969 else if (mdoc->meta.name != NULL) {
970 roff_elem_alloc(mdoc, n->line, n->pos, MDOC_Nm);
971 mdoc->last->flags |= NODE_NOSRC;
972 roff_word_alloc(mdoc, n->line, n->pos, mdoc->meta.name);
973 mdoc->last->flags |= NODE_NOSRC;
974 mdoc->last = mdoc->last->parent;
975 mdoc->next = ROFF_NEXT_SIBLING;
978 mandoc_msg(MANDOCERR_EX_NONAME, n->line, n->pos, "Ex");
982 roff_word_alloc(mdoc, n->line, n->pos,
983 ic > 1 ? "utilities exit\\~0" : "utility exits\\~0");
984 mdoc->last->flags |= NODE_NOSRC;
985 roff_word_alloc(mdoc, n->line, n->pos,
986 "on success, and\\~>0 if an error occurs.");
987 mdoc->last->flags |= NODE_EOS | NODE_NOSRC;
1000 assert(n->child->type == ROFFT_TEXT);
1001 mdoc->next = ROFF_NEXT_CHILD;
1003 if ((p = mdoc_a2lib(n->child->string)) != NULL) {
1004 n->child->flags |= NODE_NOPRT;
1005 roff_word_alloc(mdoc, n->line, n->pos, p);
1006 mdoc->last->flags = NODE_NOSRC;
1011 mandoc_msg(MANDOCERR_LB_BAD, n->child->line,
1012 n->child->pos, "Lb %s", n->child->string);
1014 roff_word_alloc(mdoc, n->line, n->pos, "library");
1015 mdoc->last->flags = NODE_NOSRC;
1016 roff_word_alloc(mdoc, n->line, n->pos, "\\(lq");
1017 mdoc->last->flags = NODE_DELIMO | NODE_NOSRC;
1018 mdoc->last = mdoc->last->next;
1019 roff_word_alloc(mdoc, n->line, n->pos, "\\(rq");
1020 mdoc->last->flags = NODE_DELIMC | NODE_NOSRC;
1027 struct roff_node *n;
1033 mdoc->next = ROFF_NEXT_CHILD;
1034 if (n->child != NULL) {
1035 roff_word_alloc(mdoc, n->line, n->pos, "The");
1036 mdoc->last->flags |= NODE_NOSRC;
1037 ic = build_list(mdoc, MDOC_Fn);
1038 roff_word_alloc(mdoc, n->line, n->pos,
1039 ic > 1 ? "functions return" : "function returns");
1040 mdoc->last->flags |= NODE_NOSRC;
1041 roff_word_alloc(mdoc, n->line, n->pos,
1042 "the value\\~0 if successful;");
1044 roff_word_alloc(mdoc, n->line, n->pos, "Upon successful "
1045 "completion, the value\\~0 is returned;");
1046 mdoc->last->flags |= NODE_NOSRC;
1048 roff_word_alloc(mdoc, n->line, n->pos, "otherwise "
1049 "the value\\~\\-1 is returned and the global variable");
1050 mdoc->last->flags |= NODE_NOSRC;
1051 roff_elem_alloc(mdoc, n->line, n->pos, MDOC_Va);
1052 mdoc->last->flags |= NODE_NOSRC;
1053 roff_word_alloc(mdoc, n->line, n->pos, "errno");
1054 mdoc->last->flags |= NODE_NOSRC;
1055 mdoc->last = mdoc->last->parent;
1056 mdoc->next = ROFF_NEXT_SIBLING;
1057 roff_word_alloc(mdoc, n->line, n->pos,
1058 "is set to indicate the error.");
1059 mdoc->last->flags |= NODE_EOS | NODE_NOSRC;
1066 struct roff_node *n;
1071 if (n->args && n->args->argc == 1)
1072 if (n->args->argv[0].arg == MDOC_Std)
1075 mandoc_msg(MANDOCERR_ARG_STD, n->line, n->pos,
1076 "%s", roff_name[n->tok]);
1082 struct roff_node *n, *nch;
1087 assert(nch->type == ROFFT_TEXT);
1089 if ((p = mdoc_a2st(nch->string)) == NULL) {
1090 mandoc_msg(MANDOCERR_ST_BAD,
1091 nch->line, nch->pos, "St %s", nch->string);
1092 roff_node_delete(mdoc, n);
1096 nch->flags |= NODE_NOPRT;
1097 mdoc->next = ROFF_NEXT_CHILD;
1098 roff_word_alloc(mdoc, nch->line, nch->pos, p);
1099 mdoc->last->flags |= NODE_NOSRC;
1106 struct roff_node *n; /* The .Tg node. */
1107 struct roff_node *nch; /* The first child of the .Tg node. */
1108 struct roff_node *nn; /* The next node after the .Tg node. */
1109 struct roff_node *np; /* The parent of the next node. */
1110 struct roff_node *nt; /* The TEXT node containing the tag. */
1111 size_t len; /* The number of bytes in the tag. */
1113 /* Find the next node. */
1115 for (nn = n; nn != NULL; nn = nn->parent) {
1116 if (nn->next != NULL) {
1123 nt = nch = n->child;
1124 if (nch == NULL && nn != NULL && nn->child != NULL &&
1125 nn->child->type == ROFFT_TEXT)
1128 /* Validate the tag. */
1129 if (nt == NULL || *nt->string == '\0')
1130 mandoc_msg(MANDOCERR_MACRO_EMPTY, n->line, n->pos, "Tg");
1132 roff_node_delete(mdoc, n);
1135 len = strcspn(nt->string, " \t\\");
1136 if (nt->string[len] != '\0')
1137 mandoc_msg(MANDOCERR_TG_SPC, nt->line,
1138 nt->pos + len, "Tg %s", nt->string);
1140 /* Keep only the first argument. */
1141 if (nch != NULL && nch->next != NULL) {
1142 mandoc_msg(MANDOCERR_ARG_EXCESS, nch->next->line,
1143 nch->next->pos, "Tg ... %s", nch->next->string);
1144 while (nch->next != NULL)
1145 roff_node_delete(mdoc, nch->next);
1148 /* Drop the macro if the first argument is invalid. */
1149 if (len == 0 || nt->string[len] != '\0') {
1150 roff_node_delete(mdoc, n);
1154 /* By default, tag the .Tg node itself. */
1155 if (nn == NULL || nn->flags & NODE_ID)
1158 /* Explicit tagging of specific macros. */
1163 nn = nn->head->child == NULL ? n : nn->head;
1167 while (np->tok != MDOC_Bl)
1169 switch (np->norm->Bl.type) {
1184 nn = nn->body->child == NULL ? n : nn->body;
1194 nn = nn->body->child == NULL ? n : nn->body;
1210 if (nn->child == NULL)
1217 tag_put(nt->string, TAG_MANUAL, nn);
1219 n->flags |= NODE_NOPRT;
1223 post_obsolete(POST_ARGS)
1225 struct roff_node *n;
1228 if (n->type == ROFFT_ELEM || n->type == ROFFT_BLOCK)
1229 mandoc_msg(MANDOCERR_MACRO_OBS, n->line, n->pos,
1230 "%s", roff_name[n->tok]);
1234 post_useless(POST_ARGS)
1236 struct roff_node *n;
1239 mandoc_msg(MANDOCERR_MACRO_USELESS, n->line, n->pos,
1240 "%s", roff_name[n->tok]);
1250 struct roff_node *np, *nch;
1253 * Unlike other data pointers, these are "housed" by the HEAD
1254 * element, which contains the goods.
1258 if (np->type != ROFFT_HEAD)
1261 assert(np->parent->type == ROFFT_BLOCK);
1262 assert(np->parent->tok == MDOC_Bf);
1264 /* Check the number of arguments. */
1267 if (np->parent->args == NULL) {
1269 mandoc_msg(MANDOCERR_BF_NOFONT,
1270 np->line, np->pos, "Bf");
1276 mandoc_msg(MANDOCERR_ARG_EXCESS,
1277 nch->line, nch->pos, "Bf ... %s", nch->string);
1279 /* Extract argument into data. */
1281 if (np->parent->args != NULL) {
1282 switch (np->parent->args->argv[0].arg) {
1284 np->norm->Bf.font = FONT_Em;
1287 np->norm->Bf.font = FONT_Li;
1290 np->norm->Bf.font = FONT_Sy;
1298 /* Extract parameter into data. */
1300 if ( ! strcmp(np->child->string, "Em"))
1301 np->norm->Bf.font = FONT_Em;
1302 else if ( ! strcmp(np->child->string, "Li"))
1303 np->norm->Bf.font = FONT_Li;
1304 else if ( ! strcmp(np->child->string, "Sy"))
1305 np->norm->Bf.font = FONT_Sy;
1307 mandoc_msg(MANDOCERR_BF_BADFONT, np->child->line,
1308 np->child->pos, "Bf %s", np->child->string);
1312 post_fname(POST_ARGS)
1314 struct roff_node *n, *nch;
1322 if (cp[strlen(cp + 1)] == ')')
1326 pos = strcspn(cp, "()");
1327 if (cp[pos] == '\0') {
1328 if (n->sec == SEC_DESCRIPTION ||
1329 n->sec == SEC_CUSTOM)
1330 tag_put(NULL, fn_prio++, n);
1334 mandoc_msg(MANDOCERR_FN_PAREN, nch->line, nch->pos + pos, "%s", cp);
1347 const struct roff_node *n;
1351 if (n->type != ROFFT_HEAD)
1354 if (n->child == NULL) {
1355 mandoc_msg(MANDOCERR_FO_NOHEAD, n->line, n->pos, "Fo");
1358 if (n->child != n->last) {
1359 mandoc_msg(MANDOCERR_ARG_EXCESS,
1360 n->child->next->line, n->child->next->pos,
1361 "Fo ... %s", n->child->next->string);
1362 while (n->child != n->last)
1363 roff_node_delete(mdoc, n->last);
1373 const struct roff_node *n;
1376 for (n = mdoc->last->child; n != NULL; n = n->next) {
1377 for (cp = n->string; *cp != '\0'; cp++) {
1378 /* Ignore callbacks and alterations. */
1379 if (*cp == '(' || *cp == '{')
1383 mandoc_msg(MANDOCERR_FA_COMMA, n->line,
1384 n->pos + (int)(cp - n->string), "%s", n->string);
1388 post_delim_nb(mdoc);
1394 struct roff_node *n;
1398 if (n->sec == SEC_NAME && n->child != NULL &&
1399 n->child->type == ROFFT_TEXT && mdoc->meta.msec != NULL)
1400 mandoc_xr_add(mdoc->meta.msec, n->child->string, -1, -1);
1402 if (n->last != NULL && n->last->tok == MDOC_Pp)
1403 roff_node_relink(mdoc, n->last);
1405 if (mdoc->meta.name == NULL)
1406 deroff(&mdoc->meta.name, n);
1408 if (mdoc->meta.name == NULL ||
1409 (mdoc->lastsec == SEC_NAME && n->child == NULL))
1410 mandoc_msg(MANDOCERR_NM_NONAME, n->line, n->pos, "Nm");
1414 post_delim_nb(mdoc);
1423 if ((n->child != NULL && n->child->type == ROFFT_TEXT) ||
1424 mdoc->meta.name == NULL)
1427 mdoc->next = ROFF_NEXT_CHILD;
1428 roff_word_alloc(mdoc, n->line, n->pos, mdoc->meta.name);
1429 mdoc->last->flags |= NODE_NOSRC;
1436 struct roff_node *n;
1440 if (n->type != ROFFT_BODY)
1443 if (n->sec != SEC_NAME)
1444 mandoc_msg(MANDOCERR_ND_LATE, n->line, n->pos, "Nd");
1446 if (n->child == NULL)
1447 mandoc_msg(MANDOCERR_ND_EMPTY, n->line, n->pos, "Nd");
1455 post_display(POST_ARGS)
1457 struct roff_node *n, *np;
1462 if (n->end != ENDBODY_NOT) {
1463 if (n->tok == MDOC_Bd &&
1464 n->body->parent->args == NULL)
1465 roff_node_delete(mdoc, n);
1466 } else if (n->child == NULL)
1467 mandoc_msg(MANDOCERR_BLK_EMPTY, n->line, n->pos,
1468 "%s", roff_name[n->tok]);
1469 else if (n->tok == MDOC_D1)
1473 if (n->tok == MDOC_Bd) {
1474 if (n->args == NULL) {
1475 mandoc_msg(MANDOCERR_BD_NOARG,
1476 n->line, n->pos, "Bd");
1477 mdoc->next = ROFF_NEXT_SIBLING;
1478 while (n->body->child != NULL)
1479 roff_node_relink(mdoc,
1481 roff_node_delete(mdoc, n);
1487 for (np = n->parent; np != NULL; np = np->parent) {
1488 if (np->type == ROFFT_BLOCK && np->tok == MDOC_Bd) {
1489 mandoc_msg(MANDOCERR_BD_NEST, n->line,
1490 n->pos, "%s in Bd", roff_name[n->tok]);
1501 post_defaults(POST_ARGS)
1503 struct roff_node *n;
1506 if (n->child != NULL) {
1507 post_delim_nb(mdoc);
1510 mdoc->next = ROFF_NEXT_CHILD;
1513 roff_word_alloc(mdoc, n->line, n->pos, "file");
1514 mdoc->last->flags |= NODE_NOSRC;
1515 roff_word_alloc(mdoc, n->line, n->pos, "...");
1519 roff_word_alloc(mdoc, n->line, n->pos, "~");
1524 mdoc->last->flags |= NODE_NOSRC;
1531 struct roff_node *n, *nch;
1538 * If we have a child, look it up in the standard keys. If a
1539 * key exist, use that instead of the child; if it doesn't,
1540 * prefix "AT&T UNIX " to the existing data.
1544 if (nch != NULL && ((att = mdoc_a2att(nch->string)) == NULL))
1545 mandoc_msg(MANDOCERR_AT_BAD,
1546 nch->line, nch->pos, "At %s", nch->string);
1548 mdoc->next = ROFF_NEXT_CHILD;
1550 roff_word_alloc(mdoc, nch->line, nch->pos, att);
1551 nch->flags |= NODE_NOPRT;
1553 roff_word_alloc(mdoc, n->line, n->pos, "AT&T UNIX");
1554 mdoc->last->flags |= NODE_NOSRC;
1561 struct roff_node *np, *nch;
1567 if (np->norm->An.auth == AUTH__NONE) {
1569 mandoc_msg(MANDOCERR_MACRO_EMPTY,
1570 np->line, np->pos, "An");
1572 post_delim_nb(mdoc);
1573 } else if (nch != NULL)
1574 mandoc_msg(MANDOCERR_ARG_EXCESS,
1575 nch->line, nch->pos, "An ... %s", nch->string);
1582 tag_put(NULL, TAG_FALLBACK, mdoc->last);
1588 post_obsolete(mdoc);
1589 if (mdoc->last->type == ROFFT_BLOCK)
1590 mdoc->last->norm->Es = mdoc->last_es;
1596 struct roff_node *n;
1599 if (n->sec == SEC_ERRORS &&
1600 (n->parent->tok == MDOC_It ||
1601 (n->parent->tok == MDOC_Bq &&
1602 n->parent->parent->parent->tok == MDOC_It)))
1603 tag_put(NULL, TAG_STRONG, n);
1604 post_delim_nb(mdoc);
1610 struct roff_node *n;
1613 if ((n->prev == NULL ||
1614 (n->prev->type == ROFFT_TEXT &&
1615 strcmp(n->prev->string, "|") == 0)) &&
1616 (n->parent->tok == MDOC_It ||
1617 (n->parent->tok == MDOC_Xo &&
1618 n->parent->parent->prev == NULL &&
1619 n->parent->parent->parent->tok == MDOC_It)))
1620 tag_put(NULL, TAG_STRONG, n);
1621 post_delim_nb(mdoc);
1627 post_obsolete(mdoc);
1628 mdoc->last_es = mdoc->last;
1634 struct roff_node *n;
1638 * Transform ".Fl Fl long" to ".Fl \-long",
1639 * resulting for example in better HTML output.
1643 if (n->prev != NULL && n->prev->tok == MDOC_Fl &&
1644 n->prev->child == NULL && n->child != NULL &&
1645 (n->flags & NODE_LINE) == 0) {
1646 mandoc_asprintf(&cp, "\\-%s", n->child->string);
1647 free(n->child->string);
1648 n->child->string = cp;
1649 roff_node_delete(mdoc, n->prev);
1657 struct roff_node *n;
1661 post_delim_nb(mdoc);
1676 if (n->child == NULL)
1678 v = n->child->string;
1679 if ((v[0] != '0' && v[0] != '1') || v[1] != '.' ||
1680 v[2] < '0' || v[2] > '9' ||
1681 v[3] < 'a' || v[3] > 'z' || v[4] != '\0')
1683 n->child->flags |= NODE_NOPRT;
1684 mdoc->next = ROFF_NEXT_CHILD;
1685 roff_word_alloc(mdoc, n->child->line, n->child->pos, v);
1686 v = mdoc->last->string;
1687 v[3] = toupper((unsigned char)v[3]);
1688 mdoc->last->flags |= NODE_NOSRC;
1700 mdoc->next = ROFF_NEXT_CHILD;
1701 roff_word_alloc(mdoc, n->line, n->pos, os);
1702 mdoc->last->flags |= NODE_NOSRC;
1709 struct roff_node *nbl, *nit, *nch;
1716 if (nit->type != ROFFT_BLOCK)
1719 nbl = nit->parent->parent;
1720 lt = nbl->norm->Bl.type;
1728 if (nit->head->child == NULL)
1729 mandoc_msg(MANDOCERR_IT_NOHEAD,
1730 nit->line, nit->pos, "Bl -%s It",
1731 mdoc_argnames[nbl->args->argv[0].arg]);
1737 if (nit->body == NULL || nit->body->child == NULL)
1738 mandoc_msg(MANDOCERR_IT_NOBODY,
1739 nit->line, nit->pos, "Bl -%s It",
1740 mdoc_argnames[nbl->args->argv[0].arg]);
1743 if ((nch = nit->head->child) != NULL)
1744 mandoc_msg(MANDOCERR_ARG_SKIP,
1745 nit->line, nit->pos, "It %s",
1746 nch->type == ROFFT_TEXT ? nch->string :
1747 roff_name[nch->tok]);
1750 cols = (int)nbl->norm->Bl.ncols;
1752 assert(nit->head->child == NULL);
1754 if (nit->head->next->child == NULL &&
1755 nit->head->next->next == NULL) {
1756 mandoc_msg(MANDOCERR_MACRO_EMPTY,
1757 nit->line, nit->pos, "It");
1758 roff_node_delete(mdoc, nit);
1763 for (nch = nit->child; nch != NULL; nch = nch->next) {
1764 if (nch->type != ROFFT_BODY)
1766 if (i++ && nch->flags & NODE_LINE)
1767 mandoc_msg(MANDOCERR_TA_LINE,
1768 nch->line, nch->pos, "Ta");
1770 if (i < cols || i > cols + 1)
1771 mandoc_msg(MANDOCERR_BL_COL, nit->line, nit->pos,
1772 "%d columns, %d cells", cols, i);
1773 else if (nit->head->next->child != NULL &&
1774 nit->head->next->child->flags & NODE_LINE)
1775 mandoc_msg(MANDOCERR_IT_NOARG,
1776 nit->line, nit->pos, "Bl -column It");
1784 post_bl_block(POST_ARGS)
1786 struct roff_node *n, *ni, *nc;
1791 for (ni = n->body->child; ni != NULL; ni = ni->next) {
1792 if (ni->body == NULL)
1794 nc = ni->body->last;
1795 while (nc != NULL) {
1804 if (ni->next == NULL) {
1805 mandoc_msg(MANDOCERR_PAR_MOVE, nc->line,
1806 nc->pos, "%s", roff_name[nc->tok]);
1807 roff_node_relink(mdoc, nc);
1808 } else if (n->norm->Bl.comp == 0 &&
1809 n->norm->Bl.type != LIST_column) {
1810 mandoc_msg(MANDOCERR_PAR_SKIP,
1812 "%s before It", roff_name[nc->tok]);
1813 roff_node_delete(mdoc, nc);
1816 nc = ni->body->last;
1822 * If "in" begins with a dot, a word, and whitespace, return a dynamically
1823 * allocated copy of "in" that skips all of those. Otherwise, return NULL.
1825 * This is a partial workaround for the TODO list item beginning with:
1826 * - When the -width string contains macros, the macros must be rendered
1829 skip_leading_dot_word(const char *in)
1831 const char *iter = in;
1838 while (*iter != '\0' && !isspace(*iter))
1841 * If the dot was followed by space or NUL,
1842 * do not skip anything.
1848 while (isspace(*iter))
1851 * If the word was not followed by space,
1852 * do not skip anything.
1857 return strdup(iter);
1861 * If the argument of -offset or -width is a macro,
1862 * replace it with the associated default width.
1865 rewrite_macro2len(struct roff_man *mdoc, char **arg)
1874 else if ( ! strcmp(*arg, "Ds"))
1876 else if ((tok = roffhash_find(mdoc->mdocmac, *arg, 0)) != TOKEN_NONE)
1877 width = macro2len(tok);
1878 else if ((newarg = skip_leading_dot_word(*arg)) == NULL)
1885 mandoc_asprintf(arg, "%zun", width);
1889 post_bl_head(POST_ARGS)
1891 struct roff_node *nbl, *nh, *nch, *nnext;
1892 struct mdoc_argv *argv;
1898 if (nh->norm->Bl.type != LIST_column) {
1899 if ((nch = nh->child) == NULL)
1901 mandoc_msg(MANDOCERR_ARG_EXCESS,
1902 nch->line, nch->pos, "Bl ... %s", nch->string);
1903 while (nch != NULL) {
1904 roff_node_delete(mdoc, nch);
1911 * Append old-style lists, where the column width specifiers
1912 * trail as macro parameters, to the new-style ("normal-form")
1913 * lists where they're argument values following -column.
1916 if (nh->child == NULL)
1920 for (j = 0; j < (int)nbl->args->argc; j++)
1921 if (nbl->args->argv[j].arg == MDOC_Column)
1924 assert(j < (int)nbl->args->argc);
1927 * Accommodate for new-style groff column syntax. Shuffle the
1928 * child nodes, all of which must be TEXT, as arguments for the
1929 * column field. Then, delete the head children.
1932 argv = nbl->args->argv + j;
1934 for (nch = nh->child; nch != NULL; nch = nch->next)
1936 argv->value = mandoc_reallocarray(argv->value,
1937 argv->sz, sizeof(char *));
1939 nh->norm->Bl.ncols = argv->sz;
1940 nh->norm->Bl.cols = (void *)argv->value;
1942 for (nch = nh->child; nch != NULL; nch = nnext) {
1943 argv->value[i++] = nch->string;
1946 roff_node_delete(NULL, nch);
1954 struct roff_node *nbody; /* of the Bl */
1955 struct roff_node *nchild, *nnext; /* of the Bl body */
1956 const char *prev_Er;
1960 switch (nbody->type) {
1962 post_bl_block(mdoc);
1972 if (nbody->end != ENDBODY_NOT)
1976 * Up to the first item, move nodes before the list,
1977 * but leave transparent nodes where they are
1978 * if they precede an item.
1979 * The next non-transparent node is kept in nchild.
1980 * It only needs to be updated after a non-transparent
1981 * node was moved out, and at the very beginning
1982 * when no node at all was moved yet.
1985 nchild = mdoc->last;
1987 if (nchild == mdoc->last)
1988 nchild = roff_node_child(nbody);
1989 if (nchild == NULL) {
1991 mandoc_msg(MANDOCERR_BLK_EMPTY,
1992 nbody->line, nbody->pos, "Bl");
1995 if (nchild->tok == MDOC_It) {
1999 mandoc_msg(MANDOCERR_BL_MOVE, nbody->child->line,
2000 nbody->child->pos, "%s", roff_name[nbody->child->tok]);
2001 if (nbody->parent->prev == NULL) {
2002 mdoc->last = nbody->parent->parent;
2003 mdoc->next = ROFF_NEXT_CHILD;
2005 mdoc->last = nbody->parent->prev;
2006 mdoc->next = ROFF_NEXT_SIBLING;
2008 roff_node_relink(mdoc, nbody->child);
2012 * We have reached the first item,
2013 * so moving nodes out is no longer possible.
2014 * But in .Bl -column, the first rows may be implicit,
2015 * that is, they may not start with .It macros.
2016 * Such rows may be followed by nodes generated on the
2017 * roff level, for example .TS.
2018 * Wrap such roff nodes into an implicit row.
2021 while (nchild != NULL) {
2022 if (nchild->tok == MDOC_It) {
2023 nchild = roff_node_next(nchild);
2026 nnext = nchild->next;
2027 mdoc->last = nchild->prev;
2028 mdoc->next = ROFF_NEXT_SIBLING;
2029 roff_block_alloc(mdoc, nchild->line, nchild->pos, MDOC_It);
2030 roff_head_alloc(mdoc, nchild->line, nchild->pos, MDOC_It);
2031 mdoc->next = ROFF_NEXT_SIBLING;
2032 roff_body_alloc(mdoc, nchild->line, nchild->pos, MDOC_It);
2033 while (nchild->tok != MDOC_It) {
2034 roff_node_relink(mdoc, nchild);
2038 nnext = nchild->next;
2039 mdoc->next = ROFF_NEXT_SIBLING;
2044 if (mdoc->meta.os_e != MANDOC_OS_NETBSD)
2048 for (nchild = nbody->child; nchild != NULL; nchild = nchild->next) {
2049 if (nchild->tok != MDOC_It)
2051 if ((nnext = nchild->head->child) == NULL)
2053 if (nnext->type == ROFFT_BLOCK)
2054 nnext = nnext->body->child;
2055 if (nnext == NULL || nnext->tok != MDOC_Er)
2057 nnext = nnext->child;
2058 if (prev_Er != NULL) {
2059 order = strcmp(prev_Er, nnext->string);
2061 mandoc_msg(MANDOCERR_ER_ORDER,
2062 nnext->line, nnext->pos,
2063 "Er %s %s (NetBSD)",
2064 prev_Er, nnext->string);
2065 else if (order == 0)
2066 mandoc_msg(MANDOCERR_ER_REP,
2067 nnext->line, nnext->pos,
2068 "Er %s (NetBSD)", prev_Er);
2070 prev_Er = nnext->string;
2077 struct roff_node *n;
2081 if (n->type == ROFFT_BLOCK && n->body->child == NULL) {
2082 mandoc_msg(MANDOCERR_BLK_EMPTY, n->line, n->pos, "Bk");
2083 roff_node_delete(mdoc, n);
2090 struct roff_node *nch;
2092 nch = mdoc->last->child;
2095 mdoc->flags ^= MDOC_SMOFF;
2099 assert(nch->type == ROFFT_TEXT);
2101 if ( ! strcmp(nch->string, "on")) {
2102 mdoc->flags &= ~MDOC_SMOFF;
2105 if ( ! strcmp(nch->string, "off")) {
2106 mdoc->flags |= MDOC_SMOFF;
2110 mandoc_msg(MANDOCERR_SM_BAD, nch->line, nch->pos,
2111 "%s %s", roff_name[mdoc->last->tok], nch->string);
2112 roff_node_relink(mdoc, nch);
2117 post_root(POST_ARGS)
2119 struct roff_node *n;
2121 /* Add missing prologue data. */
2123 if (mdoc->meta.date == NULL)
2124 mdoc->meta.date = mandoc_normdate(NULL, NULL);
2126 if (mdoc->meta.title == NULL) {
2127 mandoc_msg(MANDOCERR_DT_NOTITLE, 0, 0, "EOF");
2128 mdoc->meta.title = mandoc_strdup("UNTITLED");
2131 if (mdoc->meta.vol == NULL)
2132 mdoc->meta.vol = mandoc_strdup("LOCAL");
2134 if (mdoc->meta.os == NULL) {
2135 mandoc_msg(MANDOCERR_OS_MISSING, 0, 0, NULL);
2136 mdoc->meta.os = mandoc_strdup("");
2137 } else if (mdoc->meta.os_e &&
2138 (mdoc->meta.rcsids & (1 << mdoc->meta.os_e)) == 0)
2139 mandoc_msg(MANDOCERR_RCS_MISSING, 0, 0,
2140 mdoc->meta.os_e == MANDOC_OS_OPENBSD ?
2141 "(OpenBSD)" : "(NetBSD)");
2143 if (mdoc->meta.arch != NULL &&
2144 arch_valid(mdoc->meta.arch, mdoc->meta.os_e) == 0) {
2145 n = mdoc->meta.first->child;
2146 while (n->tok != MDOC_Dt ||
2148 n->child->next == NULL ||
2149 n->child->next->next == NULL)
2151 n = n->child->next->next;
2152 mandoc_msg(MANDOCERR_ARCH_BAD, n->line, n->pos,
2153 "Dt ... %s %s", mdoc->meta.arch,
2154 mdoc->meta.os_e == MANDOC_OS_OPENBSD ?
2155 "(OpenBSD)" : "(NetBSD)");
2158 /* Check that we begin with a proper `Sh'. */
2160 n = mdoc->meta.first->child;
2162 (n->type == ROFFT_COMMENT ||
2163 (n->tok >= MDOC_Dd &&
2164 mdoc_macro(n->tok)->flags & MDOC_PROLOGUE)))
2168 mandoc_msg(MANDOCERR_DOC_EMPTY, 0, 0, NULL);
2169 else if (n->tok != MDOC_Sh)
2170 mandoc_msg(MANDOCERR_SEC_BEFORE, n->line, n->pos,
2171 "%s", roff_name[n->tok]);
2177 struct roff_node *np, *nch, *next, *prev;
2182 if (np->type != ROFFT_BODY)
2185 if (np->child == NULL) {
2186 mandoc_msg(MANDOCERR_RS_EMPTY, np->line, np->pos, "Rs");
2191 * The full `Rs' block needs special handling to order the
2192 * sub-elements according to `rsord'. Pick through each element
2193 * and correctly order it. This is an insertion sort.
2197 for (nch = np->child->next; nch != NULL; nch = next) {
2198 /* Determine order number of this child. */
2199 for (i = 0; i < RSORD_MAX; i++)
2200 if (rsord[i] == nch->tok)
2203 if (i == RSORD_MAX) {
2204 mandoc_msg(MANDOCERR_RS_BAD, nch->line, nch->pos,
2205 "%s", roff_name[nch->tok]);
2207 } else if (nch->tok == MDOC__J || nch->tok == MDOC__B)
2208 np->norm->Rs.quote_T++;
2211 * Remove this child from the chain. This somewhat
2212 * repeats roff_node_unlink(), but since we're
2213 * just re-ordering, there's no need for the
2214 * full unlink process.
2217 if ((next = nch->next) != NULL)
2218 next->prev = nch->prev;
2220 if ((prev = nch->prev) != NULL)
2221 prev->next = nch->next;
2223 nch->prev = nch->next = NULL;
2226 * Scan back until we reach a node that's
2227 * to be ordered before this child.
2230 for ( ; prev ; prev = prev->prev) {
2231 /* Determine order of `prev'. */
2232 for (j = 0; j < RSORD_MAX; j++)
2233 if (rsord[j] == prev->tok)
2243 * Set this child back into its correct place
2244 * in front of the `prev' node.
2250 np->child->prev = nch;
2251 nch->next = np->child;
2255 prev->next->prev = nch;
2256 nch->next = prev->next;
2263 * For some arguments of some macros,
2264 * convert all breakable hyphens into ASCII_HYPH.
2267 post_hyph(POST_ARGS)
2269 struct roff_node *n, *nch;
2273 for (nch = n->child; nch != NULL; nch = nch->next) {
2274 if (nch->type != ROFFT_TEXT)
2279 while (*(++cp) != '\0')
2281 isalpha((unsigned char)cp[-1]) &&
2282 isalpha((unsigned char)cp[1])) {
2283 if (n->tag == NULL && n->flags & NODE_ID)
2284 n->tag = mandoc_strdup(nch->string);
2293 struct roff_node *n;
2296 if (n->flags & NODE_LINE ||
2297 (n->next != NULL && n->next->flags & NODE_DELIMC))
2298 mandoc_msg(MANDOCERR_NS_SKIP, n->line, n->pos, NULL);
2313 switch (mdoc->last->type) {
2318 switch (mdoc->lastsec) {
2323 post_sh_see_also(mdoc);
2326 post_sh_authors(mdoc);
2338 post_sh_name(POST_ARGS)
2340 struct roff_node *n;
2345 for (n = mdoc->last->child; n != NULL; n = n->next) {
2348 if (hasnm && n->child != NULL)
2349 mandoc_msg(MANDOCERR_NAMESEC_PUNCT,
2351 "Nm %s", n->child->string);
2356 if (n->next != NULL)
2357 mandoc_msg(MANDOCERR_NAMESEC_ND,
2358 n->line, n->pos, NULL);
2361 if (n->type == ROFFT_TEXT &&
2362 n->string[0] == ',' && n->string[1] == '\0' &&
2363 n->next != NULL && n->next->tok == MDOC_Nm) {
2369 mandoc_msg(MANDOCERR_NAMESEC_BAD,
2370 n->line, n->pos, "%s", roff_name[n->tok]);
2377 mandoc_msg(MANDOCERR_NAMESEC_NONM,
2378 mdoc->last->line, mdoc->last->pos, NULL);
2380 mandoc_msg(MANDOCERR_NAMESEC_NOND,
2381 mdoc->last->line, mdoc->last->pos, NULL);
2385 post_sh_see_also(POST_ARGS)
2387 const struct roff_node *n;
2388 const char *name, *sec;
2389 const char *lastname, *lastsec, *lastpunct;
2392 n = mdoc->last->child;
2393 lastname = lastsec = lastpunct = NULL;
2395 if (n->tok != MDOC_Xr ||
2397 n->child->next == NULL)
2400 /* Process one .Xr node. */
2402 name = n->child->string;
2403 sec = n->child->next->string;
2404 if (lastsec != NULL) {
2405 if (lastpunct[0] != ',' || lastpunct[1] != '\0')
2406 mandoc_msg(MANDOCERR_XR_PUNCT, n->line,
2407 n->pos, "%s before %s(%s)",
2408 lastpunct, name, sec);
2409 cmp = strcmp(lastsec, sec);
2411 mandoc_msg(MANDOCERR_XR_ORDER, n->line,
2412 n->pos, "%s(%s) after %s(%s)",
2413 name, sec, lastname, lastsec);
2414 else if (cmp == 0 &&
2415 strcasecmp(lastname, name) > 0)
2416 mandoc_msg(MANDOCERR_XR_ORDER, n->line,
2417 n->pos, "%s after %s", name, lastname);
2422 /* Process the following node. */
2427 if (n->tok == MDOC_Xr) {
2431 if (n->type != ROFFT_TEXT)
2433 for (name = n->string; *name != '\0'; name++)
2434 if (isalpha((const unsigned char)*name))
2436 lastpunct = n->string;
2437 if (n->next == NULL || n->next->tok == MDOC_Rs)
2438 mandoc_msg(MANDOCERR_XR_PUNCT, n->line,
2439 n->pos, "%s after %s(%s)",
2440 lastpunct, lastname, lastsec);
2446 child_an(const struct roff_node *n)
2449 for (n = n->child; n != NULL; n = n->next)
2450 if ((n->tok == MDOC_An && n->child != NULL) || child_an(n))
2456 post_sh_authors(POST_ARGS)
2459 if ( ! child_an(mdoc->last))
2460 mandoc_msg(MANDOCERR_AN_MISSING,
2461 mdoc->last->line, mdoc->last->pos, NULL);
2465 * Return an upper bound for the string distance (allowing
2466 * transpositions). Not a full Levenshtein implementation
2467 * because Levenshtein is quadratic in the string length
2468 * and this function is called for every standard name,
2469 * so the check for each custom name would be cubic.
2470 * The following crude heuristics is linear, resulting
2471 * in quadratic behaviour for checking one custom name,
2472 * which does not cause measurable slowdown.
2475 similar(const char *s1, const char *s2)
2477 const int maxdist = 3;
2480 while (s1[0] != '\0' && s2[0] != '\0') {
2481 if (s1[0] == s2[0]) {
2486 if (++dist > maxdist)
2488 if (s1[1] == s2[1]) { /* replacement */
2491 } else if (s1[0] == s2[1] && s1[1] == s2[0]) {
2492 s1 += 2; /* transposition */
2494 } else if (s1[0] == s2[1]) /* insertion */
2496 else if (s1[1] == s2[0]) /* deletion */
2501 dist += strlen(s1) + strlen(s2);
2502 return dist > maxdist ? INT_MAX : dist;
2506 post_sh_head(POST_ARGS)
2508 struct roff_node *nch;
2509 const char *goodsec;
2510 const char *const *testsec;
2515 * Process a new section. Sections are either "named" or
2516 * "custom". Custom sections are user-defined, while named ones
2517 * follow a conventional order and may only appear in certain
2521 sec = mdoc->last->sec;
2523 /* The NAME should be first. */
2525 if (sec != SEC_NAME && mdoc->lastnamed == SEC_NONE)
2526 mandoc_msg(MANDOCERR_NAMESEC_FIRST,
2527 mdoc->last->line, mdoc->last->pos, "Sh %s",
2528 sec != SEC_CUSTOM ? secnames[sec] :
2529 (nch = mdoc->last->child) == NULL ? "" :
2530 nch->type == ROFFT_TEXT ? nch->string :
2531 roff_name[nch->tok]);
2533 /* The SYNOPSIS gets special attention in other areas. */
2535 if (sec == SEC_SYNOPSIS) {
2536 roff_setreg(mdoc->roff, "nS", 1, '=');
2537 mdoc->flags |= MDOC_SYNOPSIS;
2539 roff_setreg(mdoc->roff, "nS", 0, '=');
2540 mdoc->flags &= ~MDOC_SYNOPSIS;
2542 if (sec == SEC_DESCRIPTION)
2543 fn_prio = TAG_STRONG;
2545 /* Mark our last section. */
2547 mdoc->lastsec = sec;
2549 /* We don't care about custom sections after this. */
2551 if (sec == SEC_CUSTOM) {
2552 if ((nch = mdoc->last->child) == NULL ||
2553 nch->type != ROFFT_TEXT || nch->next != NULL)
2557 for (testsec = secnames + 1; *testsec != NULL; testsec++) {
2558 dist = similar(nch->string, *testsec);
2559 if (dist < mindist) {
2564 if (goodsec != NULL)
2565 mandoc_msg(MANDOCERR_SEC_TYPO, nch->line, nch->pos,
2566 "Sh %s instead of %s", nch->string, goodsec);
2571 * Check whether our non-custom section is being repeated or is
2575 if (sec == mdoc->lastnamed)
2576 mandoc_msg(MANDOCERR_SEC_REP, mdoc->last->line,
2577 mdoc->last->pos, "Sh %s", secnames[sec]);
2579 if (sec < mdoc->lastnamed)
2580 mandoc_msg(MANDOCERR_SEC_ORDER, mdoc->last->line,
2581 mdoc->last->pos, "Sh %s", secnames[sec]);
2583 /* Mark the last named section. */
2585 mdoc->lastnamed = sec;
2587 /* Check particular section/manual conventions. */
2589 if (mdoc->meta.msec == NULL)
2595 if (*mdoc->meta.msec == '4')
2597 goodsec = "2, 3, 4, 9";
2599 case SEC_RETURN_VALUES:
2601 if (*mdoc->meta.msec == '2')
2603 if (*mdoc->meta.msec == '3')
2605 if (NULL == goodsec)
2606 goodsec = "2, 3, 9";
2609 if (*mdoc->meta.msec == '9')
2611 if (NULL == goodsec)
2613 mandoc_msg(MANDOCERR_SEC_MSEC,
2614 mdoc->last->line, mdoc->last->pos,
2615 "Sh %s for %s only", secnames[sec], goodsec);
2625 struct roff_node *n, *nch;
2629 if (nch->next == NULL) {
2630 mandoc_msg(MANDOCERR_XR_NOSEC,
2631 n->line, n->pos, "Xr %s", nch->string);
2633 assert(nch->next == n->last);
2634 if(mandoc_xr_add(nch->next->string, nch->string,
2635 nch->line, nch->pos))
2636 mandoc_msg(MANDOCERR_XR_SELF,
2637 nch->line, nch->pos, "Xr %s %s",
2638 nch->string, nch->next->string);
2640 post_delim_nb(mdoc);
2644 post_section(POST_ARGS)
2646 struct roff_node *n, *nch;
2658 for (cp = tag; *cp != '\0'; cp++)
2661 if ((nch = n->child) != NULL &&
2662 nch->type == ROFFT_TEXT &&
2663 strcmp(nch->string, tag) == 0)
2664 tag_put(NULL, TAG_STRONG, n);
2666 tag_put(tag, TAG_FALLBACK, n);
2677 if ((nch = n->child) != NULL &&
2678 (nch->tok == MDOC_Pp || nch->tok == ROFF_br ||
2679 nch->tok == ROFF_sp)) {
2680 mandoc_msg(MANDOCERR_PAR_SKIP, nch->line, nch->pos,
2681 "%s after %s", roff_name[nch->tok],
2683 roff_node_delete(mdoc, nch);
2685 if ((nch = n->last) != NULL &&
2686 (nch->tok == MDOC_Pp || nch->tok == ROFF_br)) {
2687 mandoc_msg(MANDOCERR_PAR_SKIP, nch->line, nch->pos,
2688 "%s at the end of %s", roff_name[nch->tok],
2690 roff_node_delete(mdoc, nch);
2695 post_prevpar(POST_ARGS)
2697 struct roff_node *n, *np;
2700 if (n->type != ROFFT_ELEM && n->type != ROFFT_BLOCK)
2702 if ((np = roff_node_prev(n)) == NULL)
2706 * Don't allow `Pp' prior to a paragraph-type
2707 * block: `Pp' or non-compact `Bd' or `Bl'.
2710 if (np->tok != MDOC_Pp && np->tok != ROFF_br)
2712 if (n->tok == MDOC_Bl && n->norm->Bl.comp)
2714 if (n->tok == MDOC_Bd && n->norm->Bd.comp)
2716 if (n->tok == MDOC_It && n->parent->norm->Bl.comp)
2719 mandoc_msg(MANDOCERR_PAR_SKIP, np->line, np->pos,
2720 "%s before %s", roff_name[np->tok], roff_name[n->tok]);
2721 roff_node_delete(mdoc, np);
2727 struct roff_node *np;
2729 fn_prio = TAG_STRONG;
2733 if (np->child != NULL)
2734 mandoc_msg(MANDOCERR_ARG_SKIP, np->line, np->pos,
2735 "%s %s", roff_name[np->tok], np->child->string);
2741 struct roff_node *n;
2744 n->flags |= NODE_NOPRT;
2746 if (mdoc->meta.date != NULL) {
2747 mandoc_msg(MANDOCERR_PROLOG_REP, n->line, n->pos, "Dd");
2748 free(mdoc->meta.date);
2749 } else if (mdoc->flags & MDOC_PBODY)
2750 mandoc_msg(MANDOCERR_PROLOG_LATE, n->line, n->pos, "Dd");
2751 else if (mdoc->meta.title != NULL)
2752 mandoc_msg(MANDOCERR_PROLOG_ORDER,
2753 n->line, n->pos, "Dd after Dt");
2754 else if (mdoc->meta.os != NULL)
2755 mandoc_msg(MANDOCERR_PROLOG_ORDER,
2756 n->line, n->pos, "Dd after Os");
2758 if (mdoc->quick && n != NULL)
2759 mdoc->meta.date = mandoc_strdup("");
2761 mdoc->meta.date = mandoc_normdate(n->child, n);
2767 struct roff_node *nn, *n;
2772 n->flags |= NODE_NOPRT;
2774 if (mdoc->flags & MDOC_PBODY) {
2775 mandoc_msg(MANDOCERR_DT_LATE, n->line, n->pos, "Dt");
2779 if (mdoc->meta.title != NULL)
2780 mandoc_msg(MANDOCERR_PROLOG_REP, n->line, n->pos, "Dt");
2781 else if (mdoc->meta.os != NULL)
2782 mandoc_msg(MANDOCERR_PROLOG_ORDER,
2783 n->line, n->pos, "Dt after Os");
2785 free(mdoc->meta.title);
2786 free(mdoc->meta.msec);
2787 free(mdoc->meta.vol);
2788 free(mdoc->meta.arch);
2790 mdoc->meta.title = NULL;
2791 mdoc->meta.msec = NULL;
2792 mdoc->meta.vol = NULL;
2793 mdoc->meta.arch = NULL;
2795 /* Mandatory first argument: title. */
2798 if (nn == NULL || *nn->string == '\0') {
2799 mandoc_msg(MANDOCERR_DT_NOTITLE, n->line, n->pos, "Dt");
2800 mdoc->meta.title = mandoc_strdup("UNTITLED");
2802 mdoc->meta.title = mandoc_strdup(nn->string);
2804 /* Check that all characters are uppercase. */
2806 for (p = nn->string; *p != '\0'; p++)
2807 if (islower((unsigned char)*p)) {
2808 mandoc_msg(MANDOCERR_TITLE_CASE, nn->line,
2809 nn->pos + (int)(p - nn->string),
2810 "Dt %s", nn->string);
2815 /* Mandatory second argument: section. */
2821 mandoc_msg(MANDOCERR_MSEC_MISSING, n->line, n->pos,
2822 "Dt %s", mdoc->meta.title);
2823 mdoc->meta.vol = mandoc_strdup("LOCAL");
2824 return; /* msec and arch remain NULL. */
2827 mdoc->meta.msec = mandoc_strdup(nn->string);
2829 /* Infer volume title from section number. */
2831 cp = mandoc_a2msec(nn->string);
2833 mandoc_msg(MANDOCERR_MSEC_BAD,
2834 nn->line, nn->pos, "Dt ... %s", nn->string);
2835 mdoc->meta.vol = mandoc_strdup(nn->string);
2837 mdoc->meta.vol = mandoc_strdup(cp);
2838 if (mdoc->filesec != '\0' &&
2839 mdoc->filesec != *nn->string &&
2840 *nn->string >= '1' && *nn->string <= '9')
2841 mandoc_msg(MANDOCERR_MSEC_FILE, nn->line, nn->pos,
2842 "*.%c vs Dt ... %c", mdoc->filesec, *nn->string);
2845 /* Optional third argument: architecture. */
2847 if ((nn = nn->next) == NULL)
2850 for (p = nn->string; *p != '\0'; p++)
2851 *p = tolower((unsigned char)*p);
2852 mdoc->meta.arch = mandoc_strdup(nn->string);
2854 /* Ignore fourth and later arguments. */
2856 if ((nn = nn->next) != NULL)
2857 mandoc_msg(MANDOCERR_ARG_EXCESS,
2858 nn->line, nn->pos, "Dt ... %s", nn->string);
2864 struct roff_node *n, *nch;
2867 post_delim_nb(mdoc);
2873 macro = !strcmp(nch->string, "Open") ? "Ox" :
2874 !strcmp(nch->string, "Net") ? "Nx" :
2875 !strcmp(nch->string, "Free") ? "Fx" :
2876 !strcmp(nch->string, "DragonFly") ? "Dx" : NULL;
2878 mandoc_msg(MANDOCERR_BX,
2879 n->line, n->pos, "%s", macro);
2882 mdoc->next = ROFF_NEXT_SIBLING;
2883 roff_elem_alloc(mdoc, n->line, n->pos, MDOC_Ns);
2884 mdoc->last->flags |= NODE_NOSRC;
2885 mdoc->next = ROFF_NEXT_SIBLING;
2887 mdoc->next = ROFF_NEXT_CHILD;
2888 roff_word_alloc(mdoc, n->line, n->pos, "BSD");
2889 mdoc->last->flags |= NODE_NOSRC;
2896 roff_elem_alloc(mdoc, n->line, n->pos, MDOC_Ns);
2897 mdoc->last->flags |= NODE_NOSRC;
2898 mdoc->next = ROFF_NEXT_SIBLING;
2899 roff_word_alloc(mdoc, n->line, n->pos, "-");
2900 mdoc->last->flags |= NODE_NOSRC;
2901 roff_elem_alloc(mdoc, n->line, n->pos, MDOC_Ns);
2902 mdoc->last->flags |= NODE_NOSRC;
2906 * Make `Bx's second argument always start with an uppercase
2907 * letter. Groff checks if it's an "accepted" term, but we just
2908 * uppercase blindly.
2911 *nch->string = (char)toupper((unsigned char)*nch->string);
2918 struct utsname utsname;
2919 static char *defbuf;
2921 struct roff_node *n;
2924 n->flags |= NODE_NOPRT;
2926 if (mdoc->meta.os != NULL)
2927 mandoc_msg(MANDOCERR_PROLOG_REP, n->line, n->pos, "Os");
2928 else if (mdoc->flags & MDOC_PBODY)
2929 mandoc_msg(MANDOCERR_PROLOG_LATE, n->line, n->pos, "Os");
2934 * Set the operating system by way of the `Os' macro.
2935 * The order of precedence is:
2936 * 1. the argument of the `Os' macro, unless empty
2937 * 2. the -Ios=foo command line argument, if provided
2938 * 3. -DOSNAME="\"foo\"", if provided during compilation
2939 * 4. "sysname release" from uname(3)
2942 free(mdoc->meta.os);
2943 mdoc->meta.os = NULL;
2944 deroff(&mdoc->meta.os, n);
2948 if (mdoc->os_s != NULL) {
2949 mdoc->meta.os = mandoc_strdup(mdoc->os_s);
2954 mdoc->meta.os = mandoc_strdup(OSNAME);
2956 if (defbuf == NULL) {
2957 if (uname(&utsname) == -1) {
2958 mandoc_msg(MANDOCERR_OS_UNAME, n->line, n->pos, "Os");
2959 defbuf = mandoc_strdup("UNKNOWN");
2961 mandoc_asprintf(&defbuf, "%s %s",
2962 utsname.sysname, utsname.release);
2964 mdoc->meta.os = mandoc_strdup(defbuf);
2968 if (mdoc->meta.os_e == MANDOC_OS_OTHER) {
2969 if (strstr(mdoc->meta.os, "OpenBSD") != NULL)
2970 mdoc->meta.os_e = MANDOC_OS_OPENBSD;
2971 else if (strstr(mdoc->meta.os, "NetBSD") != NULL)
2972 mdoc->meta.os_e = MANDOC_OS_NETBSD;
2976 * This is the earliest point where we can check
2977 * Mdocdate conventions because we don't know
2978 * the operating system earlier.
2981 if (n->child != NULL)
2982 mandoc_msg(MANDOCERR_OS_ARG, n->child->line, n->child->pos,
2983 "Os %s (%s)", n->child->string,
2984 mdoc->meta.os_e == MANDOC_OS_OPENBSD ?
2985 "OpenBSD" : "NetBSD");
2987 while (n->tok != MDOC_Dd)
2988 if ((n = n->prev) == NULL)
2990 if ((n = n->child) == NULL)
2992 if (strncmp(n->string, "$" "Mdocdate", 9)) {
2993 if (mdoc->meta.os_e == MANDOC_OS_OPENBSD)
2994 mandoc_msg(MANDOCERR_MDOCDATE_MISSING, n->line,
2995 n->pos, "Dd %s (OpenBSD)", n->string);
2997 if (mdoc->meta.os_e == MANDOC_OS_NETBSD)
2998 mandoc_msg(MANDOCERR_MDOCDATE, n->line,
2999 n->pos, "Dd %s (NetBSD)", n->string);
3004 mdoc_a2sec(const char *p)
3008 for (i = 0; i < (int)SEC__MAX; i++)
3009 if (secnames[i] && 0 == strcmp(p, secnames[i]))
3010 return (enum roff_sec)i;
3016 macro2len(enum roff_tok macro)