===================================================================
RCS file: /cvs/mandoc/mdoc_validate.c,v
retrieving revision 1.122
retrieving revision 1.130
diff -u -p -r1.122 -r1.130
--- mandoc/mdoc_validate.c	2010/11/29 12:22:28	1.122
+++ mandoc/mdoc_validate.c	2010/11/30 10:32:05	1.130
@@ -1,4 +1,4 @@
-/*	$Id: mdoc_validate.c,v 1.122 2010/11/29 12:22:28 kristaps Exp $ */
+/*	$Id: mdoc_validate.c,v 1.130 2010/11/30 10:32:05 kristaps Exp $ */
 /*
  * Copyright (c) 2008, 2009, 2010 Kristaps Dzonsons <kristaps@bsd.lv>
  *
@@ -85,6 +85,9 @@ static	int	 post_at(POST_ARGS);
 static	int	 post_bf(POST_ARGS);
 static	int	 post_bl(POST_ARGS);
 static	int	 post_bl_head(POST_ARGS);
+static	int	 post_defaults(POST_ARGS);
+static	int	 post_literal(POST_ARGS);
+static	int	 post_eoln(POST_ARGS);
 static	int	 post_dt(POST_ARGS);
 static	int	 post_it(POST_ARGS);
 static	int	 post_lb(POST_ARGS);
@@ -95,7 +98,6 @@ static	int	 post_sh(POST_ARGS);
 static	int	 post_sh_body(POST_ARGS);
 static	int	 post_sh_head(POST_ARGS);
 static	int	 post_st(POST_ARGS);
-static	int	 post_eoln(POST_ARGS);
 static	int	 post_vt(POST_ARGS);
 static	int	 pre_an(PRE_ARGS);
 static	int	 pre_bd(PRE_ARGS);
@@ -104,19 +106,23 @@ static	int	 pre_dd(PRE_ARGS);
 static	int	 pre_display(PRE_ARGS);
 static	int	 pre_dt(PRE_ARGS);
 static	int	 pre_it(PRE_ARGS);
+static	int	 pre_literal(PRE_ARGS);
 static	int	 pre_os(PRE_ARGS);
-static	int	 pre_pp(PRE_ARGS);
+static	int	 pre_par(PRE_ARGS);
 static	int	 pre_rv(PRE_ARGS);
 static	int	 pre_sh(PRE_ARGS);
 static	int	 pre_ss(PRE_ARGS);
 
 static	v_post	 posts_an[] = { post_an, NULL };
-static	v_post	 posts_at[] = { post_at, NULL };
-static	v_post	 posts_bd_bk[] = { hwarn_eq0, bwarn_ge1, NULL };
+static	v_post	 posts_at[] = { post_at, post_defaults, NULL };
+static	v_post	 posts_bd[] = { post_literal, hwarn_eq0, bwarn_ge1, NULL };
 static	v_post	 posts_bf[] = { hwarn_le1, post_bf, NULL };
+static	v_post	 posts_bk[] = { hwarn_eq0, bwarn_ge1, NULL };
 static	v_post	 posts_bl[] = { bwarn_ge1, post_bl, NULL };
 static	v_post	 posts_bool[] = { eerr_eq1, ebool, NULL };
 static	v_post	 posts_eoln[] = { post_eoln, NULL };
+static	v_post	 posts_defaults[] = { post_defaults, NULL };
+static	v_post	 posts_dl[] = { post_literal, bwarn_ge1, herr_eq0, NULL };
 static	v_post	 posts_dt[] = { post_dt, NULL };
 static	v_post	 posts_fo[] = { hwarn_eq1, bwarn_ge1, NULL };
 static	v_post	 posts_it[] = { post_it, NULL };
@@ -135,9 +141,10 @@ static	v_post	 posts_vt[] = { post_vt, NULL };
 static	v_post	 posts_wline[] = { bwarn_ge1, herr_eq0, NULL };
 static	v_post	 posts_wtext[] = { ewarn_ge1, NULL };
 static	v_pre	 pres_an[] = { pre_an, NULL };
-static	v_pre	 pres_bd[] = { pre_display, pre_bd, pre_pp, NULL };
-static	v_pre	 pres_bl[] = { pre_bl, pre_pp, NULL };
+static	v_pre	 pres_bd[] = { pre_display, pre_bd, pre_literal, pre_par, NULL };
+static	v_pre	 pres_bl[] = { pre_bl, pre_par, NULL };
 static	v_pre	 pres_d1[] = { pre_display, NULL };
+static	v_pre	 pres_dl[] = { pre_literal, pre_display, NULL };
 static	v_pre	 pres_dd[] = { pre_dd, NULL };
 static	v_pre	 pres_dt[] = { pre_dt, NULL };
 static	v_pre	 pres_er[] = { NULL, NULL };
@@ -145,7 +152,7 @@ static	v_pre	 pres_ex[] = { NULL, NULL };
 static	v_pre	 pres_fd[] = { NULL, NULL };
 static	v_pre	 pres_it[] = { pre_it, NULL };
 static	v_pre	 pres_os[] = { pre_os, NULL };
-static	v_pre	 pres_pp[] = { pre_pp, NULL };
+static	v_pre	 pres_pp[] = { pre_par, NULL };
 static	v_pre	 pres_rv[] = { pre_rv, NULL };
 static	v_pre	 pres_sh[] = { pre_sh, NULL };
 static	v_pre	 pres_ss[] = { pre_ss, NULL };
@@ -159,15 +166,15 @@ const	struct valids mdoc_valids[MDOC_MAX] = {
 	{ pres_ss, posts_ss },			/* Ss */ 
 	{ pres_pp, posts_notext },		/* Pp */ 
 	{ pres_d1, posts_wline },		/* D1 */
-	{ pres_d1, posts_wline },		/* Dl */
-	{ pres_bd, posts_bd_bk },		/* Bd */
+	{ pres_dl, posts_dl },			/* Dl */
+	{ pres_bd, posts_bd },			/* Bd */
 	{ NULL, NULL },				/* Ed */
 	{ pres_bl, posts_bl },			/* Bl */ 
 	{ NULL, NULL },				/* El */
 	{ pres_it, posts_it },			/* It */
 	{ NULL, posts_text },			/* Ad */ 
 	{ pres_an, posts_an },			/* An */ 
-	{ NULL, NULL },				/* Ar */
+	{ NULL, posts_defaults },		/* Ar */
 	{ NULL, posts_text },			/* Cd */ 
 	{ NULL, NULL },				/* Cm */
 	{ NULL, NULL },				/* Dv */ 
@@ -181,12 +188,12 @@ const	struct valids mdoc_valids[MDOC_MAX] = {
 	{ NULL, posts_wtext },			/* Ft */ 
 	{ NULL, posts_text },			/* Ic */ 
 	{ NULL, posts_text1 },			/* In */ 
-	{ NULL, NULL },				/* Li */
+	{ NULL, posts_defaults },		/* Li */
 	{ NULL, posts_nd },			/* Nd */
 	{ NULL, posts_nm },			/* Nm */
 	{ NULL, posts_wline },			/* Op */
 	{ NULL, NULL },				/* Ot */
-	{ NULL, NULL },				/* Pa */
+	{ NULL, posts_defaults },		/* Pa */
 	{ pres_rv, NULL },			/* Rv */
 	{ NULL, posts_st },			/* St */ 
 	{ NULL, NULL },				/* Va */
@@ -251,7 +258,7 @@ const	struct valids mdoc_valids[MDOC_MAX] = {
 	{ NULL, NULL },				/* Fc */ 
 	{ NULL, NULL },				/* Oo */
 	{ NULL, NULL },				/* Oc */
-	{ NULL, posts_bd_bk },			/* Bk */
+	{ NULL, posts_bk },			/* Bk */
 	{ NULL, NULL },				/* Ek */
 	{ NULL, posts_eoln },			/* Bt */
 	{ NULL, NULL },				/* Hf */
@@ -260,7 +267,7 @@ const	struct valids mdoc_valids[MDOC_MAX] = {
 	{ NULL, posts_lb },			/* Lb */
 	{ NULL, posts_notext },			/* Lp */ 
 	{ NULL, posts_text },			/* Lk */ 
-	{ NULL, posts_text },			/* Mt */ 
+	{ NULL, posts_defaults },		/* Mt */ 
 	{ NULL, posts_wline },			/* Brq */ 
 	{ NULL, NULL },				/* Bro */ 
 	{ NULL, NULL },				/* Brc */ 
@@ -580,6 +587,7 @@ pre_display(PRE_ARGS)
 		if (MDOC_BLOCK == node->type)
 			if (MDOC_Bd == node->tok)
 				break;
+
 	if (NULL == node)
 		return(1);
 
@@ -1075,17 +1083,38 @@ post_bf(POST_ARGS)
 	return(1);
 }
 
-
 static int
 post_lb(POST_ARGS)
 {
+	const char	*p;
+	char		*buf;
+	size_t		 sz;
 
-	if (mdoc_a2lib(mdoc->last->child->string))
+	assert(mdoc->last->child);
+	assert(MDOC_TEXT == mdoc->last->child->type);
+
+	p = mdoc_a2lib(mdoc->last->child->string);
+
+	/* If lookup ok, replace with table value. */
+
+	if (p) {
+		free(mdoc->last->child->string);
+		mdoc->last->child->string = mandoc_strdup(p);
 		return(1);
-	return(mdoc_nmsg(mdoc, mdoc->last, MANDOCERR_BADLIB));
-}
+	}
 
+	/* If not, use "library ``xxxx''. */
 
+	sz = strlen(mdoc->last->child->string) +
+		2 + strlen("\\(lqlibrary\\(rq");
+	buf = mandoc_malloc(sz);
+	snprintf(buf, sz, "library \\(lq%s\\(rq", 
+			mdoc->last->child->string);
+	free(mdoc->last->child->string);
+	mdoc->last->child->string = buf;
+	return(1);
+}
+
 static int
 post_eoln(POST_ARGS)
 {
@@ -1125,28 +1154,146 @@ post_vt(POST_ARGS)
 static int
 post_nm(POST_ARGS)
 {
+	struct mdoc_node *nn;
+	char		  buf[BUFSIZ];
 
-	if (mdoc->last->child)
+	/* If no child specified, make sure we have the meta name. */
+
+	if (NULL == mdoc->last->child && NULL == mdoc->meta.name) {
+		mdoc_nmsg(mdoc, mdoc->last, MANDOCERR_NONAME);
 		return(1);
-	if (mdoc->meta.name)
+	} else if (mdoc->meta.name)
 		return(1);
-	return(mdoc_nmsg(mdoc, mdoc->last, MANDOCERR_NONAME));
+
+	/* If no meta name, set it from the child. */
+
+	buf[0] = '\0';
+
+	for (nn = mdoc->last->child; nn; nn = nn->next) {
+		/* XXX - copied from concat(). */
+		assert(MDOC_TEXT == nn->type);
+
+		if (strlcat(buf, nn->string, BUFSIZ) >= BUFSIZ) {
+			mdoc_nmsg(mdoc, nn, MANDOCERR_MEM);
+			return(0);
+		}
+
+		if (NULL == nn->next)
+			continue;
+
+		if (strlcat(buf, " ", BUFSIZ) >= BUFSIZ) {
+			mdoc_nmsg(mdoc, nn, MANDOCERR_MEM);
+			return(0);
+		}
+	}
+
+	mdoc->meta.name = mandoc_strdup(buf);
+	return(1);
 }
 
+static int
+post_literal(POST_ARGS)
+{
+	
+	/*
+	 * The `Dl' (note "el" not "one") and `Bd' macros unset the
+	 * MDOC_LITERAL flag as they leave.  Note that `Bd' only sets
+	 * this in literal mode, but it doesn't hurt to just switch it
+	 * off in general since displays can't be nested.
+	 */
 
+	if (MDOC_BODY == mdoc->last->type)
+		mdoc->last->flags &= ~MDOC_LITERAL;
+
+	return(1);
+}
+
 static int
+post_defaults(POST_ARGS)
+{
+	struct mdoc_node *nn;
+
+	/*
+	 * The `Ar' defaults to "file ..." if no value is provided as an
+	 * argument; the `Mt' and `Pa' macros use "~"; the `Li' just
+	 * gets an empty string.
+	 */
+
+	if (mdoc->last->child)
+		return(1);
+	
+	nn = mdoc->last;
+	mdoc->next = MDOC_NEXT_CHILD;
+
+	switch (nn->tok) {
+	case (MDOC_Ar):
+		if ( ! mdoc_word_alloc(mdoc, nn->line, nn->pos, "file"))
+			return(0);
+		if ( ! mdoc_word_alloc(mdoc, nn->line, nn->pos, "..."))
+			return(0);
+		break;
+	case (MDOC_At):
+		if ( ! mdoc_word_alloc(mdoc, nn->line, nn->pos, "AT&T"))
+			return(0);
+		if ( ! mdoc_word_alloc(mdoc, nn->line, nn->pos, "UNIX"))
+			return(0);
+		break;
+	case (MDOC_Li):
+		if ( ! mdoc_word_alloc(mdoc, nn->line, nn->pos, ""))
+			return(0);
+		break;
+	case (MDOC_Pa):
+		/* FALLTHROUGH */
+	case (MDOC_Mt):
+		if ( ! mdoc_word_alloc(mdoc, nn->line, nn->pos, "~"))
+			return(0);
+		break;
+	default:
+		abort();
+		/* NOTREACHED */
+	} 
+
+	mdoc->last = nn;
+	return(1);
+}
+
+static int
 post_at(POST_ARGS)
 {
+	const char	 *p, *q;
+	char		 *buf;
+	size_t		  sz;
 
+	/*
+	 * If we have a child, look it up in the standard keys.  If a
+	 * key exist, use that instead of the child; if it doesn't,
+	 * prefix "AT&T UNIX " to the existing data.
+	 */
+	
 	if (NULL == mdoc->last->child)
 		return(1);
+
 	assert(MDOC_TEXT == mdoc->last->child->type);
-	if (mdoc_a2att(mdoc->last->child->string))
-		return(1);
-	return(mdoc_nmsg(mdoc, mdoc->last, MANDOCERR_BADATT));
-}
+	p = mdoc_a2att(mdoc->last->child->string);
 
+	if (p) {
+		free(mdoc->last->child->string);
+		mdoc->last->child->string = mandoc_strdup(p);
+	} else {
+		mdoc_nmsg(mdoc, mdoc->last, MANDOCERR_BADATT);
+		p = "AT&T UNIX ";
+		q = mdoc->last->child->string;
+		sz = strlen(p) + strlen(q) + 1;
+		buf = mandoc_malloc(sz);
+		strlcpy(buf, p, sz);
+		strlcat(buf, q, sz);
+		free(mdoc->last->child->string);
+		mdoc->last->child->string = buf;
+	}
 
+	return(1);
+}
+
 static int
 post_an(POST_ARGS)
 {
@@ -1255,24 +1402,69 @@ post_it(POST_ARGS)
 static int
 post_bl_head(POST_ARGS) 
 {
-	struct mdoc_node *n;
+	struct mdoc_node *np, *nn, *nnp;
+	int		  i, j;
 
-	assert(mdoc->last->parent);
-	n = mdoc->last->parent;
+	if (LIST_column != mdoc->last->data.Bl->type)
+		/* FIXME: this should be ERROR class... */
+		return(hwarn_eq0(mdoc));
 
-	if (LIST_column == n->data.Bl->type) {
-		if (n->data.Bl->ncols && mdoc->last->nchild) {
-			mdoc_nmsg(mdoc, n, MANDOCERR_COLUMNS);
-			return(0);
-		}
+	/*
+	 * Convert old-style lists, where the column width specifiers
+	 * trail as macro parameters, to the new-style ("normal-form")
+	 * lists where they're argument values following -column.
+	 */
+
+	/* First, disallow both types and allow normal-form. */
+
+	/* 
+	 * TODO: technically, we can accept both and just merge the two
+	 * lists, but I'll leave that for another day.
+	 */
+
+	if (mdoc->last->data.Bl->ncols && mdoc->last->nchild) {
+		mdoc_nmsg(mdoc, mdoc->last, MANDOCERR_COLUMNS);
+		return(0);
+	} else if (NULL == mdoc->last->child)
 		return(1);
+
+	np = mdoc->last->parent;
+	assert(np->args);
+
+	for (j = 0; j < (int)np->args->argc; j++) 
+		if (MDOC_Column == np->args->argv[j].arg)
+			break;
+
+	assert(j < (int)np->args->argc);
+	assert(0 == np->args->argv[j].sz);
+
+	/*
+	 * Accomodate for new-style groff column syntax.  Shuffle the
+	 * child nodes, all of which must be TEXT, as arguments for the
+	 * column field.  Then, delete the head children.
+	 */
+
+	np->args->argv[j].sz = (size_t)mdoc->last->nchild;
+	np->args->argv[j].value = mandoc_malloc
+		((size_t)mdoc->last->nchild * sizeof(char *));
+
+	mdoc->last->data.Bl->ncols = np->args->argv[j].sz;
+	mdoc->last->data.Bl->cols = (const char **)np->args->argv[j].value;
+
+	for (i = 0, nn = mdoc->last->child; nn; i++) {
+		np->args->argv[j].value[i] = nn->string;
+		nn->string = NULL;
+		nnp = nn;
+		nn = nn->next;
+		mdoc_node_delete(NULL, nnp);
 	}
 
-	/* FIXME: should be ERROR class. */
-	return(hwarn_eq0(mdoc));
+	mdoc->last->nchild = 0;
+	mdoc->last->child = NULL;
+
+	return(1);
 }
 
-
 static int
 post_bl(POST_ARGS)
 {
@@ -1345,17 +1537,26 @@ post_root(POST_ARGS)
 	return(0);
 }
 
-
 static int
 post_st(POST_ARGS)
 {
+	const char	*p;
 
-	if (mdoc_a2st(mdoc->last->child->string))
-		return(1);
-	return(mdoc_nmsg(mdoc, mdoc->last, MANDOCERR_BADSTANDARD));
-}
+	assert(MDOC_TEXT == mdoc->last->child->type);
 
+	p = mdoc_a2st(mdoc->last->child->string);
 
+	if (p == NULL) {
+		mdoc_nmsg(mdoc, mdoc->last, MANDOCERR_BADSTANDARD);
+		mdoc_node_delete(mdoc, mdoc->last);
+	} else {
+		free(mdoc->last->child->string);
+		mdoc->last->child->string = mandoc_strdup(p);
+	}
+
+	return(1);
+}
+
 static int
 post_rs(POST_ARGS)
 {
@@ -1502,34 +1703,33 @@ post_sh_body(POST_ARGS)
 static int
 post_sh_head(POST_ARGS)
 {
-	char		        buf[BUFSIZ];
-	enum mdoc_sec	        sec;
-	const struct mdoc_node *n;
+	char		  buf[BUFSIZ];
+	enum mdoc_sec	  sec;
+	struct mdoc_node *n;
 
 	/*
 	 * Process a new section.  Sections are either "named" or
-	 * "custom"; custom sections are user-defined, while named ones
-	 * usually follow a conventional order and may only appear in
-	 * certain manual sections.
+	 * "custom".  Custom sections are user-defined, while named ones
+	 * follow a conventional order and may only appear in certain
+	 * manual sections.
 	 */
 
 	buf[0] = '\0';
 
-	/*
-	 * FIXME: yes, these can use a dynamic buffer, but I don't do so
-	 * in the interests of simplicity.
-	 */
+	/* FIXME: use dynamic buffer... */
 
 	for (n = mdoc->last->child; n; n = n->next) {
-		/* XXX - copied from compact(). */
+		/* XXX - copied from concat(). */
 		assert(MDOC_TEXT == n->type);
 
 		if (strlcat(buf, n->string, BUFSIZ) >= BUFSIZ) {
 			mdoc_nmsg(mdoc, n, MANDOCERR_MEM);
 			return(0);
 		}
+
 		if (NULL == n->next)
 			continue;
+
 		if (strlcat(buf, " ", BUFSIZ) >= BUFSIZ) {
 			mdoc_nmsg(mdoc, n, MANDOCERR_MEM);
 			return(0);
@@ -1538,41 +1738,60 @@ post_sh_head(POST_ARGS)
 
 	sec = mdoc_str2sec(buf);
 
-	/* 
-	 * Check: NAME should always be first, CUSTOM has no roles,
-	 * non-CUSTOM has a conventional order to be followed.
-	 */
+	/* The NAME should be first. */
 
 	if (SEC_NAME != sec && SEC_NONE == mdoc->lastnamed)
-		if ( ! mdoc_nmsg(mdoc, mdoc->last, MANDOCERR_NAMESECFIRST))
-			return(0);
+		mdoc_nmsg(mdoc, mdoc->last, MANDOCERR_NAMESECFIRST);
 
+	/* The SYNOPSIS gets special attention in other areas. */
+
+	if (SEC_SYNOPSIS == sec)
+		mdoc->flags |= MDOC_SYNOPSIS;
+	else
+		mdoc->flags &= ~MDOC_SYNOPSIS;
+
+	/* Mark our last section. */
+
+	mdoc->lastsec = sec;
+
+	/* We don't care about custom sections after this. */
+
 	if (SEC_CUSTOM == sec)
 		return(1);
 
+	/*
+	 * Check whether our non-custom section is being repeated or is
+	 * out of order.
+	 */
+
 	if (sec == mdoc->lastnamed)
-		if ( ! mdoc_nmsg(mdoc, mdoc->last, MANDOCERR_SECREP))
-			return(0);
+		mdoc_nmsg(mdoc, mdoc->last, MANDOCERR_SECREP);
 
 	if (sec < mdoc->lastnamed)
-		if ( ! mdoc_nmsg(mdoc, mdoc->last, MANDOCERR_SECOOO))
-			return(0);
+		mdoc_nmsg(mdoc, mdoc->last, MANDOCERR_SECOOO);
 
-	/* 
-	 * Check particular section/manual conventions.  LIBRARY can
-	 * only occur in manual section 2, 3, and 9.
-	 */
+	/* Mark the last named section. */
 
+	mdoc->lastnamed = sec;
+
+	/* Check particular section/manual conventions. */
+
+	assert(mdoc->meta.msec);
+
 	switch (sec) {
+	case (SEC_RETURN_VALUES):
+		/* FALLTHROUGH */
+	case (SEC_ERRORS):
+		/* FALLTHROUGH */
 	case (SEC_LIBRARY):
-		assert(mdoc->meta.msec);
 		if (*mdoc->meta.msec == '2')
 			break;
 		if (*mdoc->meta.msec == '3')
 			break;
 		if (*mdoc->meta.msec == '9')
 			break;
-		return(mdoc_nmsg(mdoc, mdoc->last, MANDOCERR_SECMSEC));
+		mdoc_nmsg(mdoc, mdoc->last, MANDOCERR_SECMSEC);
+		break;
 	default:
 		break;
 	}
@@ -1580,15 +1799,17 @@ post_sh_head(POST_ARGS)
 	return(1);
 }
 
-
 static int
-pre_pp(PRE_ARGS)
+pre_par(PRE_ARGS)
 {
 
 	if (NULL == mdoc->last)
 		return(1);
 
-	/* Don't allow prior `Lp' or `Pp'. */
+	/* 
+	 * Don't allow prior `Lp' or `Pp' prior to a paragraph-type
+	 * block:  `Lp', `Pp', or non-compact `Bd' or `Bl'.
+	 */
 
 	if (MDOC_Pp != mdoc->last->tok && MDOC_Lp != mdoc->last->tok)
 		return(1);
@@ -1600,5 +1821,36 @@ pre_pp(PRE_ARGS)
 
 	mdoc_nmsg(mdoc, mdoc->last, MANDOCERR_IGNPAR);
 	mdoc_node_delete(mdoc, mdoc->last);
+	return(1);
+}
+
+static int
+pre_literal(PRE_ARGS)
+{
+
+	if (MDOC_BODY != n->type)
+		return(1);
+
+	/*
+	 * The `Dl' (note "el" not "one") and `Bd -literal' and `Bd
+	 * -unfilled' macros set MDOC_LITERAL on entrance to the body.
+	 */
+
+	switch (n->tok) {
+	case (MDOC_Dl):
+		mdoc->flags |= MDOC_LITERAL;
+		break;
+	case (MDOC_Bd):
+		assert(n->data.Bd);
+		if (DISP_literal == n->data.Bd->type)
+			mdoc->flags |= MDOC_LITERAL;
+		if (DISP_unfilled == n->data.Bd->type)
+			mdoc->flags |= MDOC_LITERAL;
+		break;
+	default:
+		abort();
+		/* NOTREACHED */
+	}
+	
 	return(1);
 }