===================================================================
RCS file: /cvs/mandoc/mdoc_validate.c,v
retrieving revision 1.7
retrieving revision 1.17
diff -u -p -r1.7 -r1.17
--- mandoc/mdoc_validate.c	2009/06/10 20:18:43	1.7
+++ mandoc/mdoc_validate.c	2009/06/27 09:03:03	1.17
@@ -1,4 +1,4 @@
-/*	$Id: mdoc_validate.c,v 1.7 2009/06/10 20:18:43 kristaps Exp $ */
+/*	$Id: mdoc_validate.c,v 1.17 2009/06/27 09:03:03 kristaps Exp $ */
 /*
  * Copyright (c) 2008, 2009 Kristaps Dzonsons <kristaps@kth.se>
  *
@@ -42,15 +42,20 @@ enum	merr {
 	ELISTTYPE,
 	EDISPTYPE,
 	EMULTIDISP,
+	ESECNAME,
 	EMULTILIST,
 	EARGREP,
 	EBOOL,
+	ECOLMIS,
 	ENESTDISP
 };
 
 enum	mwarn {
 	WPRINT,
+	WNOWIDTH,
+	WMISSWIDTH,
 	WESCAPE,
+	WDEPCOL,
 	WWRONGMSEC,
 	WSECOOO,
 	WSECREP,
@@ -93,7 +98,6 @@ static	int	err_child_gt(struct mdoc *, const char *, i
 static	int	warn_child_gt(struct mdoc *, const char *, int);
 static	int	err_child_eq(struct mdoc *, const char *, int);
 static	int	warn_child_eq(struct mdoc *, const char *, int);
-static	int	count_child(struct mdoc *);
 static	int	warn_print(struct mdoc *, int, int);
 static	int	warn_count(struct mdoc *, const char *, 
 			int, const char *, int);
@@ -112,7 +116,6 @@ static	int	pre_fd(PRE_ARGS);
 static	int	pre_it(PRE_ARGS);
 static	int	pre_lb(PRE_ARGS);
 static	int	pre_os(PRE_ARGS);
-static	int	pre_prologue(PRE_ARGS);
 static	int	pre_rv(PRE_ARGS);
 static	int	pre_sh(PRE_ARGS);
 static	int	pre_ss(PRE_ARGS);
@@ -134,6 +137,7 @@ static	int	post_args(POST_ARGS);
 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_it(POST_ARGS);
 static	int	post_nm(POST_ARGS);
 static	int	post_root(POST_ARGS);
@@ -151,15 +155,15 @@ static	v_pre	pres_an[] = { pre_an, NULL };
 static	v_pre	pres_bd[] = { pre_display, pre_bd, NULL };
 static	v_pre	pres_bl[] = { pre_bl, NULL };
 static	v_pre	pres_cd[] = { pre_cd, NULL };
-static	v_pre	pres_dd[] = { pre_prologue, pre_dd, NULL };
+static	v_pre	pres_dd[] = { pre_dd, NULL };
 static	v_pre	pres_d1[] = { pre_display, NULL };
-static	v_pre	pres_dt[] = { pre_prologue, pre_dt, NULL };
+static	v_pre	pres_dt[] = { pre_dt, NULL };
 static	v_pre	pres_er[] = { pre_er, NULL };
 static	v_pre	pres_ex[] = { pre_ex, NULL };
 static	v_pre	pres_fd[] = { pre_fd, NULL };
 static	v_pre	pres_it[] = { pre_it, NULL };
 static	v_pre	pres_lb[] = { pre_lb, NULL };
-static	v_pre	pres_os[] = { pre_prologue, pre_os, NULL };
+static	v_pre	pres_os[] = { pre_os, 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 };
@@ -170,7 +174,7 @@ static	v_post	posts_wtext[] = { ewarn_ge1, NULL };
 static	v_post	posts_notext[] = { eerr_eq0, NULL };
 static	v_post	posts_wline[] = { bwarn_ge1, herr_eq0, NULL };
 static	v_post	posts_sh[] = { herr_ge1, bwarn_ge1, post_sh, NULL };
-static	v_post	posts_bl[] = { herr_eq0, bwarn_ge1, post_bl, NULL };
+static	v_post	posts_bl[] = { bwarn_ge1, post_bl, NULL };
 static	v_post	posts_it[] = { post_it, NULL };
 static	v_post	posts_in[] = { ewarn_eq1, NULL };
 static	v_post	posts_ss[] = { herr_ge1, NULL };
@@ -188,7 +192,7 @@ static	v_post	posts_bf[] = { hwarn_le1, post_bf, NULL 
 static	v_post	posts_fo[] = { hwarn_eq1, bwarn_ge1, NULL };
 
 const	struct valids mdoc_valids[MDOC_MAX] = {
-	{ NULL, NULL }, 			/* \" */
+	{ NULL, NULL },				/* Ap */
 	{ pres_dd, posts_text },		/* Dd */
 	{ pres_dt, NULL },			/* Dt */
 	{ pres_os, NULL },			/* Os */
@@ -295,9 +299,8 @@ const	struct valids mdoc_valids[MDOC_MAX] = {
 	{ NULL, NULL },				/* Fr */
 	{ NULL, posts_notext },			/* Ud */
 	{ pres_lb, posts_lb },			/* Lb */
-	{ NULL, NULL },				/* Ap */
 	{ NULL, posts_pp },			/* Lp */ 
-	{ NULL, posts_text },			/* Lk */ 
+	{ NULL, NULL },				/* Lk */ 
 	{ NULL, posts_text },			/* Mt */ 
 	{ NULL, posts_wline },			/* Brq */ 
 	{ NULL, NULL },				/* Bro */ 
@@ -410,6 +413,9 @@ perr(struct mdoc *m, int line, int pos, enum merr type
 	case (EDISPTYPE):
 		p = "missing display type";
 		break;
+	case (ESECNAME):
+		p = "the NAME section must come first";
+		break;
 	case (ELINE):
 		p = "expected line arguments";
 		break;
@@ -419,6 +425,9 @@ perr(struct mdoc *m, int line, int pos, enum merr type
 	case (ENODATA):
 		p = "document has no data";
 		break;
+	case (ECOLMIS):
+		p = "column syntax style mismatch";
+		break;
 	case (EATT):
 		p = "expected valid AT&T symbol";
 		break;
@@ -460,6 +469,16 @@ pwarn(struct mdoc *m, int line, int pos, enum mwarn ty
 		p = "prologue macros out-of-order";
 		c = WARN_COMPAT;
 		break;
+	case (WDEPCOL):
+		p = "deprecated column argument syntax";
+		c = WARN_COMPAT;
+		break;
+	case (WNOWIDTH):
+		p = "superfluous width argument";
+		break;
+	case (WMISSWIDTH):
+		p = "missing width argument";
+		break;
 	case (WPRINT):
 		p = "invalid character";
 		break;
@@ -529,19 +548,6 @@ err_count(struct mdoc *m, const char *k,
 }
 
 
-static inline int
-count_child(struct mdoc *mdoc)
-{
-	int		  i;
-	struct mdoc_node *n;
-
-	for (i = 0, n = mdoc->last->child; n; n = n->next, i++)
-		/* Do nothing */ ;
-
-	return(i);
-}
-
-
 /*
  * Build these up with macros because they're basically the same check
  * for different inequalities.  Yes, this could be done with functions,
@@ -552,10 +558,9 @@ count_child(struct mdoc *mdoc)
 static int 							\
 lvl##_child_##name(struct mdoc *mdoc, const char *p, int sz) 	\
 { 								\
-	int i; 							\
-	if ((i = count_child(mdoc)) ineq sz) 			\
+	if (mdoc->last->nchild ineq sz)				\
 		return(1); 					\
-	return(lvl##_count(mdoc, #ineq, sz, p, i)); 		\
+	return(lvl##_count(mdoc, #ineq, sz, p, mdoc->last->nchild)); \
 }
 
 #define CHECK_BODY_DEFN(name, lvl, func, num) 			\
@@ -705,8 +710,6 @@ check_text(struct mdoc *mdoc, int line, int pos, const
 {
 	size_t		 c;
 
-	/* FIXME: indicate deprecated escapes \*(xx and \*x. */
-
 	for ( ; *p; p++) {
 		if ('\t' == *p) {
 			if ( ! (MDOC_LITERAL & mdoc->flags))
@@ -776,7 +779,7 @@ pre_display(PRE_ARGS)
 static int
 pre_bl(PRE_ARGS)
 {
-	int		 i, type, width, offset;
+	int		 pos, type, width, offset;
 
 	if (MDOC_BLOCK != n->type)
 		return(1);
@@ -788,8 +791,8 @@ pre_bl(PRE_ARGS)
 	type = offset = width = -1;
 
 	/* LINTED */
-	for (i = 0; i < (int)n->args->argc; i++)
-		switch (n->args->argv[i].arg) {
+	for (pos = 0; pos < (int)n->args->argc; pos++)
+		switch (n->args->argv[pos].arg) {
 		case (MDOC_Bullet):
 			/* FALLTHROUGH */
 		case (MDOC_Dash):
@@ -811,23 +814,20 @@ pre_bl(PRE_ARGS)
 		case (MDOC_Inset):
 			/* FALLTHROUGH */
 		case (MDOC_Column):
-			if (-1 == type) {
-				type = n->args->argv[i].arg;
-				break;
-			}
-			return(nerr(mdoc, n, EMULTILIST));
+			if (-1 != type) 
+				return(nerr(mdoc, n, EMULTILIST));
+			type = n->args->argv[pos].arg;
+			break;
 		case (MDOC_Width):
-			if (-1 == width) {
-				width = n->args->argv[i].arg;
-				break;
-			}
-			return(nerr(mdoc, n, EARGREP));
+			if (-1 != width)
+				return(nerr(mdoc, n, EARGREP));
+			width = n->args->argv[pos].arg;
+			break;
 		case (MDOC_Offset):
-			if (-1 == offset) {
-				offset = n->args->argv[i].arg;
-				break;
-			}
-			return(nerr(mdoc, n, EARGREP));
+			if (-1 != offset)
+				return(nerr(mdoc, n, EARGREP));
+			offset = n->args->argv[pos].arg;
+			break;
 		default:
 			break;
 		}
@@ -835,7 +835,17 @@ pre_bl(PRE_ARGS)
 	if (-1 == type)
 		return(nerr(mdoc, n, ELISTTYPE));
 
+	/* 
+	 * Validate the width field.  Some list types don't need width
+	 * types and should be warned about them.  Others should have it
+	 * and must also be warned.
+	 */
+
 	switch (type) {
+	case (MDOC_Tag):
+		if (-1 == width && ! nwarn(mdoc, n, WMISSWIDTH))
+			return(0);
+		break;
 	case (MDOC_Column):
 		/* FALLTHROUGH */
 	case (MDOC_Diag):
@@ -843,17 +853,9 @@ pre_bl(PRE_ARGS)
 	case (MDOC_Inset):
 		/* FALLTHROUGH */
 	case (MDOC_Item):
-		if (-1 == width)
-			break;
-		return(mdoc_nwarn(mdoc, n, WARN_SYNTAX,
-				"superfluous %s argument",
-				mdoc_argnames[MDOC_Width]));
-	case (MDOC_Tag):
-		if (-1 != width)
-			break;
-		return(mdoc_nwarn(mdoc, n, WARN_SYNTAX, 
-				"suggest %s argument", 
-				mdoc_argnames[MDOC_Width]));
+		if (-1 != width && ! nwarn(mdoc, n, WNOWIDTH))
+			return(0);
+		break;
 	default:
 		break;
 	}
@@ -985,14 +987,6 @@ pre_cd(PRE_ARGS)
 
 
 static int
-pre_prologue(PRE_ARGS)
-{
-
-	return(check_sec(mdoc, n, SEC_PROLOGUE, SEC_CUSTOM));
-}
-
-
-static int
 pre_dt(PRE_ARGS)
 {
 
@@ -1045,25 +1039,24 @@ post_bf(POST_ARGS)
 
 	head = mdoc->last->head;
 
-	if (NULL == mdoc->last->args) {
-		if (NULL == head->child || 
-				MDOC_TEXT != head->child->type)
-			return(mdoc_err(mdoc, "text argument expected"));
+	if (mdoc->last->args && head->child)
+		return(mdoc_err(mdoc, "one argument expected"));
+	else if (mdoc->last->args)
+		return(1);
 
-		p = head->child->string;
-		if (0 == strcmp(p, "Em"))
-			return(1);
-		else if (0 == strcmp(p, "Li"))
-			return(1);
-		else if (0 == strcmp(p, "Sm"))
-			return(1);
-		return(mdoc_nerr(mdoc, head->child, "invalid font"));
-	}
+	if (NULL == head->child || MDOC_TEXT != head->child->type)
+		return(mdoc_err(mdoc, "text argument expected"));
 
-	if (head->child)
-		return(mdoc_err(mdoc, "one argument expected"));
+	p = head->child->string;
 
-	return(1);
+	if (0 == strcmp(p, "Em"))
+		return(1);
+	else if (0 == strcmp(p, "Li"))
+		return(1);
+	else if (0 == strcmp(p, "Sm"))
+		return(1);
+
+	return(mdoc_nerr(mdoc, head->child, "invalid font mode"));
 }
 
 
@@ -1229,10 +1222,35 @@ post_it(POST_ARGS)
 
 
 static int
+post_bl_head(POST_ARGS) 
+{
+	int			i;
+	const struct mdoc_node *n;
+
+	n = mdoc->last->parent;
+	assert(n->args);
+
+	for (i = 0; i < (int)n->args->argc; i++)
+		if (n->args->argv[i].arg == MDOC_Column)
+			break;
+
+	if (i == (int)n->args->argc)
+		return(1);
+
+	if (n->args->argv[i].sz && mdoc->last->child)
+		return(nerr(mdoc, n, ECOLMIS));
+
+	return(1);
+}
+
+
+static int
 post_bl(POST_ARGS)
 {
 	struct mdoc_node	*n;
 
+	if (MDOC_HEAD == mdoc->last->type) 
+		return(post_bl_head(mdoc));
 	if (MDOC_BODY != mdoc->last->type)
 		return(1);
 	if (NULL == mdoc->last->child)
@@ -1279,7 +1297,7 @@ post_root(POST_ARGS)
 
 	if (NULL == mdoc->first->child)
 		return(verr(mdoc, ENODATA));
-	if (SEC_PROLOGUE == mdoc->lastnamed)
+	if ( ! (MDOC_PBODY & mdoc->flags))
 		return(verr(mdoc, ENOPROLOGUE));
 
 	if (MDOC_BLOCK != mdoc->first->child->type)
@@ -1360,13 +1378,12 @@ post_sh_head(POST_ARGS)
 	 * certain manual sections.
 	 */
 
-	assert(MDOC_Sh == mdoc->last->tok);
-
-	/* This is just concat() inlined, which is irritating. */
-
 	buf[0] = 0;
+
 	for (n = mdoc->last->child; n; n = n->next) {
+		/* XXX - copied from compact(). */
 		assert(MDOC_TEXT == n->type);
+
 		if (strlcat(buf, n->string, 64) >= 64)
 			return(nerr(mdoc, n, ETOOLONG));
 		if (NULL == n->next)
@@ -1377,21 +1394,24 @@ post_sh_head(POST_ARGS)
 
 	sec = mdoc_atosec(buf);
 
-	/* The NAME section should always be first. */
+	/* 
+	 * Check: NAME should always be first, CUSTOM has no roles,
+	 * non-CUSTOM has a conventional order to be followed.
+	 */
 
-	if (SEC_BODY == mdoc->lastnamed && SEC_NAME != sec)
-		return(vwarn(mdoc, WSECOOO));
+	if (SEC_NAME != sec && SEC_NONE == mdoc->lastnamed)
+		return(verr(mdoc, ESECNAME));
 	if (SEC_CUSTOM == sec)
 		return(1);
-
-	/* Check for repeated or out-of-order sections. */
-
 	if (sec == mdoc->lastnamed)
 		return(vwarn(mdoc, WSECREP));
 	if (sec < mdoc->lastnamed)
 		return(vwarn(mdoc, WSECOOO));
 
-	/* Check particular section/manual section conventions. */
+	/* 
+	 * Check particular section/manual conventions.  LIBRARY can
+	 * only occur in msec 2, 3 (TODO: are there more of these?).
+	 */
 
 	switch (sec) {
 	case (SEC_LIBRARY):