Loads of platforms appear to have old or broken Unicode character type

information and are missing widths for relatively common Unicode
characters (so mbtowc() works, but wcwidth() fails). So if wcwidth()
returns -1, assume a width of 1 instead of ignoring the character.

1 files changed, 8 insertions, 0 deletions

diff --git a/utf8.c b/utf8.c
index 56281aa2..22ab62c1 100644
--- a/utf8.c
+++ b/utf8.c
@@ -119,6 +119,14 @@ utf8_width(wchar_t wc)
 	width = wcwidth(wc);
 	if (width < 0 || width > 0xff) {
 		log_debug("Unicode %04x, wcwidth() %d", wc, width);
+
+		/*
+		 * Many platforms have no width for relatively common
+		 * characters (wcwidth() returns -1); assume width 1 in this
+		 * case and hope for the best.
+		 */
+		if (width < 0)
+			return (1);
 		return (-1);
 	}
 	return (width);