Added -fno-ign-chars support to libman.

man_validate.c checks for non-tab/isprint words.
libman hashtable fixed (was ignoring .br).
Added ncount field to man_node, deprecating count() functions.
Documented use of tabs in man.7.

6 files changed, 66 insertions, 21 deletions

diff --git a/main.c b/main.c
index b76aa2ac..dd7df9fe 100644
--- a/main.c
+++ b/main.c
@@ -232,10 +232,12 @@ man_init(struct curparse *curp)
 
 	/* Defaults from mandoc.1. */
 
-	pflags = MAN_IGN_MACRO;
+	pflags = MAN_IGN_MACRO | MAN_IGN_CHARS;
 
 	if (curp->fflags & NO_IGN_MACRO)
 		pflags &= ~MAN_IGN_MACRO;
+	if (curp->fflags & NO_IGN_CHARS)
+		pflags &= ~MAN_IGN_CHARS;
 
 	if (NULL == (man = man_alloc(curp, pflags, &mancb)))
 		warnx("memory exhausted");
diff --git a/man.7 b/man.7
index 480004f4..01031d02 100644
--- a/man.7
+++ b/man.7
@@ -52,9 +52,11 @@ Other lines are interpreted within the current state.
 .\" SECTION
 .Sh INPUT ENCODING
 .Nm
-documents may contain only graphable 7-bit ASCII characters and the
+documents may contain only graphable 7-bit ASCII characters, the
 space character
-.Sq \  .
+.Sq \  ,
+and tabs
+.Sq \et . 
 All manuals must have
 .Ux
 .Sq \en
diff --git a/man.c b/man.c
index eba5fa94..ef28b99f 100644
--- a/man.c
+++ b/man.c
@@ -182,6 +182,8 @@ man_node_append(struct man *man, struct man_node *p)
 		abort();
 		/* NOTREACHED */
 	}
+	
+	p->parent->nchild++;
 
 	man->last = p;
 
diff --git a/man.h b/man.h
index 2b31637b..497b237c 100644
--- a/man.h
+++ b/man.h
@@ -63,6 +63,7 @@ struct	man_node {
 	struct man_node	*child;
 	struct man_node	*next;
 	struct man_node	*prev;
+	int		 nchild;
 	int		 line;
 	int		 pos;
 	int		 tok;
@@ -73,7 +74,8 @@ struct	man_node {
 	char		*string;
 };
 
-#define	MAN_IGN_MACRO	 (1 << 0) /* Ignore unknown macros. */
+#define	MAN_IGN_MACRO	 (1 << 0)
+#define	MAN_IGN_CHARS	 (1 << 1)
 
 extern	const char *const *man_macronames;
 
diff --git a/man_hash.c b/man_hash.c
index 68ce712a..7f1686fb 100644
--- a/man_hash.c
+++ b/man_hash.c
@@ -37,9 +37,13 @@ man_hash_alloc(void)
 	int		*htab;
 	int		 i, j, x;
 
-	htab = calloc(26 * 5, sizeof(int));
+	/* Initialised to -1. */
+
+	htab = malloc(26 * 5 * sizeof(int));
 	if (NULL == htab)
 		return(NULL);
+	for (i = 0; i < 26 * 5; i++)
+		htab[i] = -1;
 
 	for (i = 0; i < MAN_MAX; i++) {
 		x = man_macronames[i][0];
@@ -51,7 +55,7 @@ man_hash_alloc(void)
 		x *= 5;
 
 		for (j = 0; j < 5; j++)
-			if (0 == htab[x + j]) {
+			if (-1 == htab[x + j]) {
 				htab[x + j] = i;
 				break;
 			}
@@ -80,7 +84,7 @@ man_hash_find(const void *arg, const char *tmp)
 	x *= 5;
 
 	for (i = 0; i < 5; i++) {
-		if (0 == (tok = htab[x + i]))
+		if (-1 == (tok = htab[x + i]))
 			return(MAN_MAX);
 		if (0 == strcmp(tmp, man_macronames[tok]))
 			return(tok);
diff --git a/man_validate.c b/man_validate.c
index 703e8118..24081eb8 100644
--- a/man_validate.c
+++ b/man_validate.c
@@ -23,23 +23,26 @@
 
 #include "libman.h"
 
-/* FIXME: validate text. */
-
 #define	POSTARGS  struct man *m, const struct man_node *n
 
+enum	merr {
+	WPRINT
+};
+
 typedef	int	(*v_post)(POSTARGS);
 
 struct	man_valid {
 	v_post	 *posts;
 };
 
-static	int	  count(const struct man_node *);
 static	int	  check_eq0(POSTARGS);
 static	int	  check_ge1(POSTARGS);
 static	int	  check_ge2(POSTARGS);
 static	int	  check_le1(POSTARGS);
 static	int	  check_le2(POSTARGS);
 static	int	  check_le5(POSTARGS);
+static	int	  check_text(POSTARGS);
+static	int	  perr(struct man *, int, int, int, enum merr);
 
 static	v_post	  posts_le1[] = { check_le1, NULL };
 static	v_post	  posts_le2[] = { check_le2, NULL };
@@ -85,7 +88,7 @@ man_valid_post(struct man *m)
 
 	switch (m->last->type) {
 	case (MAN_TEXT): 
-		/* FALLTHROUGH */
+		return(check_text(m, m->last));
 	case (MAN_ROOT):
 		return(1);
 	default:
@@ -102,14 +105,45 @@ man_valid_post(struct man *m)
 }
 
 
-static inline int
-count(const struct man_node *n)
-{ 
-	int		 i;
+static int
+perr(struct man *m, int line, int pos, 
+		int iserr, enum merr type)
+{
+	const char	 *p;
+	
+	p = NULL;
+	switch (type) {
+	case (WPRINT):
+		p = "invalid character";
+		break;
+	}
+	assert(p);
+
+	if (iserr)
+		return(man_verr(m, line, pos, p));
+
+	return(man_vwarn(m, line, pos, p));
+}
 
-	for (i = 0; n; n = n->next, i++) 
-		/* Loop. */ ;
-	return(i);
+
+static int
+check_text(POSTARGS) 
+{
+	const char	*p;
+	int		 pos;
+
+	assert(n->string);
+
+	for (p = n->string, pos = n->pos + 1; *p; p++, pos++) {
+		if ('\t' == *p || isprint((u_char)*p))
+			continue;
+
+		if (MAN_IGN_CHARS & m->pflags)
+			return(perr(m, n->line, pos, 0, WPRINT));
+		return(perr(m, n->line, pos, 1, WPRINT));
+	}
+
+	return(1);
 }
 
 
@@ -117,12 +151,11 @@ count(const struct man_node *n)
 static int \
 check_##name(POSTARGS) \
 { \
-	int		 c; \
-	if ((c = count(n->child)) ineq (x)) \
+	if (n->nchild ineq (x)) \
 		return(1); \
 	return(man_verr(m, n->line, n->pos, \
 			"expected line arguments %s %d, have %d", \
-			#ineq, (x), c)); \
+			#ineq, (x), n->nchild)); \
 }
 
 INEQ_DEFINE(0, ==, eq0)