Tidy utf8.c a little: build table on first use, and make utf8_width take

a u_int rather than splitting and then combining again in utf8_split.

3 files changed, 33 insertions, 57 deletions

diff --git a/usr.bin/tmux/server.c b/usr.bin/tmux/server.c
index 256cbeb597d..8b2166668f3 100644
--- a/usr.bin/tmux/server.c
+++ b/usr.bin/tmux/server.c
@@ -1,4 +1,4 @@
-/* $OpenBSD: server.c,v 1.146 2015/10/31 13:12:03 nicm Exp $ */
+/* $OpenBSD: server.c,v 1.147 2015/11/12 12:19:57 nicm Exp $ */
 
 /*
  * Copyright (c) 2007 Nicholas Marriott <nicm@users.sourceforge.net>
@@ -186,7 +186,6 @@ server_start(struct event_base *base, int lockfd, char *lockfile)
 	TAILQ_INIT(&session_groups);
 	mode_key_init_trees();
 	key_bindings_init();
-	utf8_build();
 
 	start_time = time(NULL);
 
diff --git a/usr.bin/tmux/tmux.h b/usr.bin/tmux/tmux.h
index f44c98da3d8..f19ed3c3c5a 100644
--- a/usr.bin/tmux/tmux.h
+++ b/usr.bin/tmux/tmux.h
@@ -1,4 +1,4 @@
-/* $OpenBSD: tmux.h,v 1.573 2015/11/12 11:10:50 nicm Exp $ */
+/* $OpenBSD: tmux.h,v 1.574 2015/11/12 12:19:57 nicm Exp $ */
 
 /*
  * Copyright (c) 2007 Nicholas Marriott <nicm@users.sourceforge.net>
@@ -2181,7 +2181,7 @@ void		 session_group_synchronize1(struct session *, struct session *);
 void		 session_renumber_windows(struct session *);
 
 /* utf8.c */
-void		 utf8_build(void);
+u_int		 utf8_width(u_int);
 void		 utf8_set(struct utf8_data *, u_char);
 int		 utf8_open(struct utf8_data *, u_char);
 int		 utf8_append(struct utf8_data *, u_char);
diff --git a/usr.bin/tmux/utf8.c b/usr.bin/tmux/utf8.c
index 3b85264937c..f3d210ebefd 100644
--- a/usr.bin/tmux/utf8.c
+++ b/usr.bin/tmux/utf8.c
@@ -1,4 +1,4 @@
-/* $OpenBSD: utf8.c,v 1.16 2015/11/12 11:10:50 nicm Exp $ */
+/* $OpenBSD: utf8.c,v 1.17 2015/11/12 12:19:57 nicm Exp $ */
 
 /*
  * Copyright (c) 2008 Nicholas Marriott <nicm@users.sourceforge.net>
@@ -35,7 +35,7 @@ struct utf8_width_entry {
 };
 
 /* Sorted, then repeatedly split in the middle to balance the tree. */
-struct utf8_width_entry utf8_width_table[] = {
+static struct utf8_width_entry utf8_width_table[] = {
 	{ 0x00b41, 0x00b44, 0, NULL, NULL },
 	{ 0x008e4, 0x00902, 0, NULL, NULL },
 	{ 0x006d6, 0x006dd, 0, NULL, NULL },
@@ -344,12 +344,9 @@ struct utf8_width_entry utf8_width_table[] = {
 	{ 0xe0100, 0xe01ef, 0, NULL, NULL },
 	{ 0x100000, 0x10fffd, 0, NULL, NULL },
 };
+static struct utf8_width_entry	*utf8_width_root = NULL;
 
-struct utf8_width_entry	*utf8_width_root = NULL;
-
-int	utf8_overlap(struct utf8_width_entry *, struct utf8_width_entry *);
-u_int	utf8_combine(const struct utf8_data *);
-u_int	utf8_width(const struct utf8_data *);
+static void	utf8_build(void);
 
 /* Set a single character. */
 void
@@ -405,40 +402,20 @@ utf8_append(struct utf8_data *utf8data, u_char ch)
 	if (utf8data->have != utf8data->size)
 		return (1);
 
-	utf8data->width = utf8_width(utf8data);
-	return (0);
-}
-
-/* Check if two width tree entries overlap. */
-int
-utf8_overlap(struct utf8_width_entry *item1, struct utf8_width_entry *item2)
-{
-	if (item1->first >= item2->first && item1->first <= item2->last)
-		return (1);
-	if (item1->last >= item2->first && item1->last <= item2->last)
-		return (1);
-	if (item2->first >= item1->first && item2->first <= item1->last)
-		return (1);
-	if (item2->last >= item1->first && item2->last <= item1->last)
-		return (1);
+	utf8data->width = utf8_width(utf8_combine(utf8data));
 	return (0);
 }
 
 /* Build UTF-8 width tree. */
-void
+static void
 utf8_build(void)
 {
 	struct utf8_width_entry	**ptr, *item, *node;
-	u_int			  i, j;
+	u_int			  i;
 
 	for (i = 0; i < nitems(utf8_width_table); i++) {
 		item = &utf8_width_table[i];
 
-		for (j = 0; j < nitems(utf8_width_table); j++) {
-			if (i != j && utf8_overlap(item, &utf8_width_table[j]))
-				log_fatalx("utf8 overlap: %u %u", i, j);
-		}
-
 		ptr = &utf8_width_root;
 		while (*ptr != NULL) {
 			node = *ptr;
@@ -451,6 +428,27 @@ utf8_build(void)
 	}
 }
 
+/* Lookup width of UTF-8 data in tree. */
+u_int
+utf8_width(u_int uc)
+{
+	struct utf8_width_entry	*item;
+
+	if (utf8_width_root == NULL)
+		utf8_build();
+
+	item = utf8_width_root;
+	while (item != NULL) {
+		if (uc < item->first)
+			item = item->left;
+		else if (uc > item->last)
+			item = item->right;
+		else
+			return (item->width);
+	}
+	return (1);
+}
+
 /* Combine UTF-8 into 32-bit Unicode. */
 u_int
 utf8_combine(const struct utf8_data *utf8data)
@@ -481,7 +479,7 @@ utf8_combine(const struct utf8_data *utf8data)
 	return (value);
 }
 
-/* Split a UTF-8 character. */
+/* Split 32-bit Unicode into UTF-8. */
 int
 utf8_split(u_int uc, struct utf8_data *utf8data)
 {
@@ -505,7 +503,7 @@ utf8_split(u_int uc, struct utf8_data *utf8data)
 		utf8data->data[3] = 0x80 | (uc & 0x3f);
 	} else
 		return (-1);
-	utf8data->width = utf8_width(utf8data);
+	utf8data->width = utf8_width(uc);
 	return (0);
 }
 
@@ -522,27 +520,6 @@ utf8_split2(u_int uc, u_char *ptr)
 	return (1);
 }
 
-/* Lookup width of UTF-8 data in tree. */
-u_int
-utf8_width(const struct utf8_data *utf8data)
-{
-	struct utf8_width_entry	*item;
-	u_int			 value;
-
-	value = utf8_combine(utf8data);
-
-	item = utf8_width_root;
-	while (item != NULL) {
-		if (value < item->first)
-			item = item->left;
-		else if (value > item->last)
-			item = item->right;
-		else
-			return (item->width);
-	}
-	return (1);
-}
-
 /*
  * Encode len characters from src into dst, which is guaranteed to have four
  * bytes available for each character from src (for \abc or UTF-8) plus space