uxd.c: add overlong sequence detection; ver.rst: regenerate

1 files changed, 39 insertions, 5 deletions

diff --git a/uxd.c b/uxd.c
index dbea5f0..38c5862 100644
--- a/uxd.c
+++ b/uxd.c
@@ -77,6 +77,7 @@ int cur_normal_hilite = 0;
 #define HL_NORM_INV 1
 #define HL_SPECIAL 2
 #define HL_BAD 3
+#define HL_OVERLONG 4
 
 /* terminal codes for mono highlighting. */
 #define MONO_NORMAL 0
@@ -417,8 +418,12 @@ void append_color(char *buf, int hl_type) {
 			fgcolor = special_color;
 			bgcolor = 0;
 			break;
-		default:
+		case HL_OVERLONG:
+			fgcolor = 0;
+			bgcolor = special_color;
+			break;
 		case HL_BAD:
+		default:
 			fgcolor = 0;
 			bgcolor = bad_color;
 			break;
@@ -447,6 +452,7 @@ void append_mono(char *buf, int hl_type) {
 			code = MONO_BOLD;
 			break;
 		default:
+		case HL_OVERLONG: /* maybe change this later */
 		case HL_BAD:
 			code = MONO_REVERSE;
 			break;
@@ -521,6 +527,26 @@ int is_bom(unsigned char *b) {
 	return (b[0] == 0xef && b[1] == 0xbb && b[2] == 0xbf);
 }
 
+/* Detect overlong encodings, without doing a full decode. */
+int is_overlong(int cont_count, unsigned char *b) {
+	/* 1 byte seqs are never overlong. */
+	if(!cont_count)
+		return 0;
+
+	/* 2 byte seqs, if the first byte is 0xc0 or 0xc1, it's overlong. */
+	if(cont_count == 1 && b[0] <= 0xc1)
+		return 1;
+
+	/* for 3 and 4 byte seqs, it's the 2nd byte that matters. */
+	if(cont_count == 2 && b[1] <= 0x9f)
+		return 1;
+
+	if(cont_count == 3 && b[1] <= 0x8f)
+		return 1;
+
+	return 0;
+}
+
 /* U+10FFFF is the last valid codepoint. It encodes to f4 8f bf bf.
    'count' is the count of continuation bytes only (so, 3 for a 4-byte
    sqeuence). */
@@ -560,7 +586,7 @@ int dump_utf8_char(void) {
 	unsigned char bytes[] = { 0, 0, 0, 0, 0 };
 	unsigned char *cont_bytes = bytes + 1;
 	char *printable;
-	int bad = 0, special = 0, hl_type;
+	int bad = 0, special = 0, overlong = 0, hl_type;
 	int c, cont_count, i;
 	static int byte0;
 
@@ -625,7 +651,10 @@ int dump_utf8_char(void) {
 	if(is_out_of_range(cont_count, bytes))
 		bad = 1;
 
-	if(bad) {
+	if(is_overlong(cont_count, bytes))
+		overlong = 1;
+
+	if(bad || overlong) {
 		bad_count++;
 	} else {
 		char_count++;
@@ -644,8 +673,13 @@ int dump_utf8_char(void) {
 		hl_type = HL_SPECIAL;
 		printable = PRINT_BOM;
 	} else {
-		hl_type = HL_NORMAL;
-		printable = (char *)bytes;
+		if(overlong) {
+			hl_type = HL_OVERLONG;
+			printable = PRINT_BAD;
+		} else {
+			hl_type = HL_NORMAL;
+			printable = (char *)bytes;
+		}
 	}
 
 	/* human-readable (right) column: */