From cf823e2fadb80924656ac177ef37d117bd4a38be Mon Sep 17 00:00:00 2001
From: Renaud Fivet <renaud.fivet@gmail.com>
Date: Wed, 24 Jun 2020 16:38:03 +0800
Subject: [PATCH] Clean up handling of unicode character width (non printable
 are displayed as \u) and insure modeline displays filename including double
 and zero width characters.

---
 display.c | 30 ++++++++++++++++++------------
 display.h |  1 -
 main.c    |  4 ++--
 utf8.c    | 12 ++++--------
 utf8.h    |  2 +-
 5 files changed, 25 insertions(+), 24 deletions(-)

diff --git a/display.c b/display.c
index 2beac7f..3cd42d5 100644
--- a/display.c
+++ b/display.c
@@ -176,8 +176,7 @@ void vttidy(void)
  * screen. There is no checking for nonsense values; this might be a good
  * idea during the early stages.
  */
-void vtmove(int row, int col)
-{
+static void vtmove( int row, int col) {
 	vtrow = row;
 	vtcol = col;
 }
@@ -220,6 +219,9 @@ static void vtputc( unicode_t c) {
 		sane_vtputc( '\\') ;
 		sane_vtputc( hex[ c >> 4]) ;
 		sane_vtputc( hex[ c & 15]) ;
+	} else if( utf8_width( c) < 0) {
+		sane_vtputc( '\\') ;		/* show as non printable */
+		sane_vtputc( 'u') ;
 	} else
 		sane_vtputc( c) ;
 }
@@ -232,7 +234,7 @@ static int vtputs( const char *s) {
 		
 		s += utf8_to_unicode( s, 0, 4, &c) ;
 		vtputc( c) ;
-		n += utf8_width( c) ;
+		n += utf8_width( c) ;	/* To Do: only works if all printable */
 	}
 
 	return n ;
@@ -569,8 +571,10 @@ void updpos(void)
 				curcol += 2 ;	/* displayed as ^c */
 		else if( c >= 0x80 && c <= 0xA0)
 				curcol += 3 ;	/* displayed as \xx */
-		else
-			curcol += utf8_width( c) ;
+		else {
+			int width = utf8_width( c) ;
+			curcol += (width < 0) ? 2 : width ;	/* non printable are displayed as \u */
+		}
 	}
 
 	/* if extended, flag so and update the virtual line image */
@@ -977,9 +981,10 @@ static int updateline(int row, struct video *vp1, struct video *vp2)
 		   the virtual screen array                             */
 		while( ttcol < term.t_ncol) {
 		/* TODO: handle double width unicode char at last screen col */
-			TTputc(*cp1);
-			ttcol += utf8_width( *cp1) ;
-			*cp2++ = *cp1++;
+			unicode_t c = *cp1++ ;
+			TTputc( c) ;
+			ttcol += utf8_width( c) ;
+			*cp2++ = c ;
 		}
 
 		TTrev( FALSE) ;		/* turn rev video off */
@@ -1045,9 +1050,10 @@ static int updateline(int row, struct video *vp1, struct video *vp2)
 #endif
 
 	while (cp1 != cp5) {	/* Ordinary. */
-		TTputc(*cp1);
-		ttcol += utf8_width( *cp1) ;
-		*cp2++ = *cp1++;
+		unicode_t c = *cp1++ ;
+		TTputc( c) ;
+		ttcol += utf8_width( c) ;
+		*cp2++ = c ;
 	}
 
 	if (cp5 != cp3) {	/* Erase. */
@@ -1153,7 +1159,7 @@ static void modeline(struct window *wp)
 		char *msg = NULL;
 		char tline[ 6] ;	/* buffer for part of mode line */
 
-		vtcol = n - 7;	/* strlen(" top ") plus a couple */
+		vtcol -= 7 ;	/* strlen(" top ") plus a couple */
 		while (rows--) {
 			lp = lforw(lp);
 			if (lp == wp->w_bufp->b_linep) {
diff --git a/display.h b/display.h
index 50957ca..0f94d1a 100644
--- a/display.h
+++ b/display.h
@@ -16,7 +16,6 @@ extern int gbcolor ;		/* global backgrnd color (black) */
 void vtinit( void) ;
 void vtfree( void) ;
 void vttidy( void) ;
-void vtmove( int row, int col) ;
 int upscreen( int f, int n) ;
 int update( int force) ;
 void updpos( void) ;
diff --git a/main.c b/main.c
index f0f0c2f..50c1757 100644
--- a/main.c
+++ b/main.c
@@ -141,7 +141,7 @@ int main(int argc, char **argv)
 	int errflag;		/* C error processing? */
 	bname_t bname ;	/* buffer name of file to read */
 
-	setlocale( LC_CTYPE, "en_GB.UTF-8") ; /* wide character support (UTF-32) */
+	setlocale( LC_CTYPE, "") ; /* expects $LANG like en_GB.UTF-8 */
 
 #if	PKCODE & BSD
 	sleep(1); /* Time for window manager. */
@@ -226,7 +226,7 @@ int main(int argc, char **argv)
 				} else if( argv[ carg + 1]) {	/* -X filename */
 					if( startup( &argv[ carg + 1][ 0]) == TRUE)
 						startflag = TRUE ;	/* don't execute emacs.rc */
-					
+
 					carg += 1 ;
 				}
 
diff --git a/utf8.c b/utf8.c
index 9448356..81a8513 100644
--- a/utf8.c
+++ b/utf8.c
@@ -10,16 +10,12 @@
 /*
  * Display width of UTF-8 character
  */
-unsigned utf8_width( unicode_t c) {
+int utf8_width( unicode_t c) {
 #if CYGWIN
-	assert( sizeof( wchar_t) == 2) ;	/* wcwidth only handles UTF-16 */
-	return (c < 0x10000) ? (unsigned) wcwidth( (wchar_t) c) : 2 ;
-#elif BSD
-//	assert( sizeof( wchar_t) == 4) ;	/* wcwidth should handle UTF-32 */
-	int ret = wcwidth( (wchar_t) c) ;
-	return (ret < 0) ? 1 : (unsigned) ret ;
+	assert( sizeof( wchar_t) == 2) ;	/* wcwidth only supports UTF-16 */
+	return (c < 0x10000) ? wcwidth( (wchar_t) c) : -1 ;
 #else
-	return (unsigned) wcwidth( (wchar_t) c) ;
+	return wcwidth( (wchar_t) c) ;
 #endif
 }
 
diff --git a/utf8.h b/utf8.h
index ae61faf..b3b9505 100644
--- a/utf8.h
+++ b/utf8.h
@@ -3,7 +3,7 @@
 
 typedef unsigned int unicode_t ;
 
-unsigned utf8_width( unicode_t c) ;
+int utf8_width( unicode_t c) ;
 unsigned utf8_to_unicode( const char *line, unsigned index, unsigned len,
                                                             unicode_t *res) ;
 unsigned utf8_revdelta( unsigned char *buf, unsigned pos) ;