detect UTF-16 surrogates as bad, use red for overlong

1 files changed, 14 insertions, 3 deletions

diff --git a/uxd.c b/uxd.c
index e32356b..d141c2e 100644
--- a/uxd.c
+++ b/uxd.c
@@ -425,9 +425,11 @@ void append_color(char *buf, int hl_type) {
 			bgcolor = 0;
 			break;
 		case HL_OVERLONG:
+			/* don't use a separate color for this any more
 			fgcolor = 0;
 			bgcolor = special_color;
 			break;
+			*/
 		case HL_BAD:
 		default:
 			fgcolor = 0;
@@ -563,6 +565,12 @@ int is_out_of_range(int cont_count, unsigned char *b) {
 	return 1;
 }
 
+/* surrogates for UTF-16 are not valid Unicode (therefore not UTF-8) */
+int is_surrogate(int cont_count, unsigned char *b) {
+	if(cont_count != 2) return 0;
+	return b[0] == 0xed && b[1] > 0x9f;
+}
+
 int get_next_byte(void) {
 	int c;
 
@@ -655,9 +663,12 @@ int dump_utf8_char(void) {
 		}
 	}
 
-	/* don't check bad sequences for out-of-range */
-	if(!bad && is_out_of_range(cont_count, bytes))
-		bad = 1;
+	/* don't check bad sequences for out-of-range or surrogate */
+	if(!bad) {
+		if(is_out_of_range(cont_count, bytes) || is_surrogate(cont_count, bytes))
+			bad = 1;
+	}
+
 
 	if(is_overlong(cont_count, bytes))
 		overlong = 1;