Re: Bug#317549: tin: Linebreaks between words when lines are longer than window width

On Sat, Jul 09, 2005 at 08:04:35PM +0200, Urs Janßen wrote:
> > There is a small issue related to this though. tin generates a warning 
> > when a line in a composed message exceeds 78 characters. It's good 
> > feature but actually it counts bytes not characters. I use UTF-8 locale 
> > and messages often contain multibyte characters. Even when I save my 
> > message with no longer lines than 72 characters I get this warning from 
> > lines which contain enough multibyte characters to exceed the limit of 78 
> > bytes per line.
> 
> yes, this is a bug in post.c:check_article_to_be_posted() ~ line 1118
> the code is much older than tins multibyte support and when we added
> the multibyte stuff we failed to update that bits of code as well...

something like the following (untested) should fix this:

--- post.c	2005-07-10 02:07:33.541466569 +0200
+++ post.c	2005-07-10 02:20:51.870511937 +0200
@@ -1114,15 +1114,33 @@
 		}
 #endif /* CHARSET_CONVERSION */
 
-		/* FIXME: this code is wrong for a multibyte environment */
-		col = 0;
-		for (cp = line; *cp; cp++) {
-			if (!contains_8bit && !isascii(*cp))
-				contains_8bit = TRUE;
-			if (*cp == '\t')
-				col += 8 - (col % 8);
-			else
+		{
+#if defined(MULTIBYTE_ABLE) && !defined(NO_LOCALE)
+			int num_bytes;
+			wchar_t wc;
+#endif /* MULTIBYTE_ABLE && !NO_LOCALE */
+
+			col = 0;
+			for (cp = line; *cp; ) {
+				if (*cp == '\t') {
+					col += 8 - (col % 8);
+					cp++;
+				} else {
+#if defined(MULTIBYTE_ABLE) && !defined(NO_LOCALE)
+					if ((num_bytes = mbtowc(&wc, cp, MB_CUR_MAX)) != -1) {
+						cp += num_bytes;
+						if (!contains_8bit && num_bytes > 1)
+							contains_8bit = TRUE;
+					} else
+						cp++;
+#else
+					if (!contains_8bit && !isascii(*cp))
+						contains_8bit = TRUE;
+					cp++;
+#endif
 				col++;
+				}
+			}
 		}
 		if (col > MAX_COL && !got_long_line) {
 			setup_check_article_screen(&init);