From 73641367db67f47c55d28d98b2d41467a8c87211 Mon Sep 17 00:00:00 2001
From: Leah Neukirchen <leah@vuxu.org>
Date: Wed, 22 Nov 2017 23:48:00 +0100
Subject: revamp utf-8 handling code

---
 safe_u8putstr.c | 73 +++++++++++++++++++++------------------------------------
 1 file changed, 27 insertions(+), 46 deletions(-)

(limited to 'safe_u8putstr.c')

diff --git a/safe_u8putstr.c b/safe_u8putstr.c
index e198b54..31cdb0b 100644
--- a/safe_u8putstr.c
+++ b/safe_u8putstr.c
@@ -1,76 +1,57 @@
 #include <stdint.h>
 #include <stdio.h>
 
+#include "u8decode.h"
+
 void
 safe_u8putstr(char *s0, size_t l, FILE *stream)
 {
 	// tty-safe output of s, with relaxed utf-8 semantics:
 	// - C0 and C1 are displayed as escape sequences
-	// - valid utf8 is printed as is
-	// - rest is assumed to be latin1, and translated into utf8
+	// - valid utf-8 is printed as is
+	// - rest is assumed to be latin-1, and translated into utf-8
 	// - translate CRLF to CR
 
 	unsigned char *s = (unsigned char *)s0;
 	unsigned char *e = s + l;
+	uint32_t c;
 
 	while (s < e) {
-		if ((*s & 0x80) == 0) {
-			if (*s < 32 &&
-			    *s != ' ' && *s != '\t' && *s != '\n' && *s != '\r') {
-				// C0
+		int l = u8decode((char *)s, &c);
+		if (l == -1) {
+			l = 1;
+			if (*s <= 0x9fu) {
+				// C1
 				fputc(0xe2, stream);
 				fputc(0x90, stream);
-				fputc(0x80+*s, stream);
-			} else if (*s == 127) {
-				// DEL
+				fputc(0x80+0x1b, stream);
+
 				fputc(0xe2, stream);
 				fputc(0x90, stream);
-				fputc(0xa1, stream);
-			} else if (*s == '\r') {
-				if (e - s > 1 && s[1] == '\n')
-					s++;
 				fputc(*s, stream);
 			} else {
-				// safe ASCII
-				fputc(*s, stream);
+				/* invalid utf-8, assume it was latin-1 */
+				fputc(0xc0 | (*s >> 6), stream);
+				fputc(0x80 | (*s & 0x3f), stream);
 			}
-		} else if ((*s & 0xc0) == 0x80) {
-			if (*s >= 0xa0)
-				goto latin1;
-
-			// C1
+		} else if (c < 32 && 
+		    *s != ' ' && *s != '\t' && *s != '\n' && *s != '\r') {
+			// C0
 			fputc(0xe2, stream);
 			fputc(0x90, stream);
-			fputc(0x80+0x1b, stream);
-
+			fputc(0x80+*s, stream);
+		} else if (c == 127) {
+			// DEL
 			fputc(0xe2, stream);
 			fputc(0x90, stream);
+			fputc(0xa1, stream);
+		} else if (c == '\r') {
+			if (e - s > 1 && s[1] == '\n')
+				s++;
 			fputc(*s, stream);
 		} else {
-			uint32_t f = 0;
-			if (e - s >= 4)
-				f = (s[0]<<24) | (s[1]<<16) | (s[2]<<8) | s[3];
-			else if (e - s == 3)
-				f = (s[0]<<24) | (s[1]<<16) | (s[2]<<8);
-			else if (e - s == 2)
-				f = (s[0]<<24) | (s[1]<<16);
-			else if (e - s == 1)
-				f = (s[0]<<24);
-
-			if      ((f & 0xe0c00000) == 0xc0800000) goto u2;
-			else if ((f & 0xf0c0c000) == 0xe0808000) goto u3;
-			else if ((f & 0xf8c0c0c0) == 0xf0808080) {
-				fputc(*s++, stream);
-u3:                             fputc(*s++, stream);
-u2:                             fputc(*s++, stream);
-				fputc(*s, stream);
-			} else {
-latin1:
-				/* invalid utf8, assume it was latin1 */
-				fputc(0xc0 | (*s >> 6), stream);
-				fputc(0x80 | (*s & 0x3f), stream);
-			}
+			fwrite(s, 1, l, stream);
 		}
-		s++;
+		s += l;
 	}
 }
-- 
cgit 1.4.1