fix Makefile, dashes in hex dump, red for codepoints > U+10FFFF, fix spacing.

1 files changed, 25 insertions, 4 deletions

diff --git a/uxd.c b/uxd.c
index cff81cf..d71d6d5 100644
--- a/uxd.c
+++ b/uxd.c
@@ -135,6 +135,7 @@ void print_line(void) {
 
 	/* line up the rightmost field (human-readable) */
 	while(spacing--) printf("   ");
+	if(dump_column < (MAX_DUMP_COLS / 2)) putchar(' ');
 
 	printf(" %s\n", right_buf);
 
@@ -178,12 +179,21 @@ void append_left(unsigned char byte, int dash, int fgcolor, int bgcolor) {
 	append_color(left_buf, fgcolor, bgcolor);
 	sprintf(tmpbuf, "%02x", byte);
 	strcat(left_buf, tmpbuf);
-	if(dash) strcat(left_buf, "-");
-	append_color_off(left_buf);
-	if(!dash) strcat(left_buf, " ");
 
-	if(dump_column == 7) strcat(left_buf, " ");
 	dump_column++;
+
+	if(dash) {
+		strcat(left_buf, "-");
+		if(dump_column == (MAX_DUMP_COLS / 2))
+			strcat(left_buf, "-");
+		append_color_off(left_buf);
+	} else {
+		append_color_off(left_buf);
+		strcat(left_buf, " ");
+		if(dump_column == (MAX_DUMP_COLS / 2))
+			strcat(left_buf, " ");
+	}
+
 	if(dump_column == MAX_DUMP_COLS)
 		print_line();
 
@@ -208,6 +218,14 @@ int is_bom(unsigned char *b) {
 	return (b[0] == 0xef && b[1] == 0xbb && b[2] == 0xbf);
 }
 
+/* U+10FFFF is the last valid codepoint. It encodes to f4 8f bf bf. */
+int is_out_of_range(int count, unsigned char *b) {
+	if(count < 3) return 0;
+	if(b[0] < 0xf4) return 0;
+	if(b[1] < 0x90) return 0;
+	return 1;
+}
+
 /* return value: false = EOF, true = more data to read */
 int dump_utf8_char(void) {
 	unsigned char bytes[] = { 0, 0, 0, 0, 0 };
@@ -265,6 +283,9 @@ int dump_utf8_char(void) {
 		}
 	}
 
+	if(is_out_of_range(cont_count, bytes))
+		bad = 1;
+
 	if(bad) {
 		fg = BAD_FG;
 		bg = BAD_BG;