Wed, 14 May 2003 17:01:10 -0700

Committed by: keithp

Update of /local/src/CVS/jove
In directory home.keithp.com:/tmp/cvs-serv8425

Modified Files:
	abbrev.c ask.c buf.c c.c case.c ctype.h delete.c disp.c 
	error.c fp.c funcdefs.c insert.c io.c io.h iproc.c jove.c 
	jove.h keymaps.txt misc.c move.c paragraph.c portsrv.c proc.c 
	re1.c recover.c scandir.c screen.c temp.h term.c tune.h util.c 
	vars.c 
Log Message:
Switch to UTF-8 for internal storage and file format.  Permit Latin-1 output

Index: abbrev.c
===================================================================
RCS file: /local/src/CVS/jove/abbrev.c,v
retrieving revision 1.2
retrieving revision 1.3
diff -u -d -r1.2 -r1.3

--- abbrev.c	11 Jul 2001 06:42:26 -0000	1.2
+++ abbrev.c	15 May 2003 00:01:06 -0000	1.3
@@ -96,16 +96,16 @@
 {
 	Bufpos	point;
 	char	wordbuf[1024];
-	register char	*wp = wordbuf,
-			*cp;
-	register int	c;
+	char	*wp = wordbuf,
+		*cp;
+	int	c;
 	int	UC_count = 0;
 	struct abbrev	*ap;
 
 	DOTsave(&point);
     WITH_TABLE(curbuf->b_major)
 	b_word(1);
-	while (curchar < point.p_char && ismword(c = linebuf[curchar])) {
+	while (curchar < point.p_char && ismword(cur_char())) {
 		if (AutoCaseAbbrev) {
 			if (isupper(c)) {
 				UC_count++;
@@ -114,8 +114,8 @@
 		}
 		if (wp >= wordbuf + (sizeof (wordbuf) - 1))
 			break;
-		*wp++ = c;
-		curchar++;
+		wp = unparse_ucs4 (c, wp);
+		right_char ();
 	}
 	*wp = '\0';
     END_TABLE();
@@ -127,14 +127,14 @@
 	}
 	del_char(BACKWARD, (wp - wordbuf));
 
-	for (cp = ap->a_phrase; c = *cp; ) {
+	for (cp = ap->a_phrase; c = parse_ucs4(cp); ) {
 		if (AutoCaseAbbrev) {
 			insert_c(islower(c) && UC_count &&
 			       (cp == ap->a_phrase || (UC_count > 1 && (cp[-1] == ' '))) ?
 				toupper(c) : c, 1);
 		} else
 			insert_c(c, 1);
-		cp += 1;
+		cp = buf_next (cp);
 	}
 
 	if (ap->a_cmdhook != 0)
@@ -283,4 +283,4 @@
 	ap->a_cmdhook = hook;
 }
 
-#endif ABBREV
+#endif /* ABBREV */

Index: ask.c
===================================================================
RCS file: /local/src/CVS/jove/ask.c,v
retrieving revision 1.2
retrieving revision 1.3
diff -u -d -r1.2 -r1.3
--- ask.c	11 Jul 2001 06:42:26 -0000	1.2
+++ ask.c	15 May 2003 00:01:06 -0000	1.3
@@ -401,8 +401,8 @@
 		makedirty(curline);
 	}
 	is_ntdir = ((numfound == 1) &&
-		    (curchar > 0) &&
-		    (linebuf[curchar - 1] != '/') &&
+		    (!bolp()) &&
+		    (prev_char () != '/') &&
 		    (isdir(linebuf)));
 	if (the_same && !is_ntdir) {
 		add_mess((n == 1) ? " [Unique]" : " [Ambiguous]");

Index: buf.c
===================================================================
RCS file: /local/src/CVS/jove/buf.c,v
retrieving revision 1.2
retrieving revision 1.3
diff -u -d -r1.2 -r1.3
--- buf.c	11 Jul 2001 06:42:26 -0000	1.2
+++ buf.c	15 May 2003 00:01:06 -0000	1.3
@@ -740,6 +740,103 @@
 	return b;
 }
 
+ucs4 parse_ucs4 (char *buf)
+{
+	ucs4    result = (*buf++ & 0xff);
+
+	if ((result & 0xc0) == 0xc0)
+	{
+		ucs4	mask = 0x20;
+		int	extra = 1;
+
+		while ((result & mask) != 0)
+		{
+			extra++;
+			mask >>= 1;
+		}
+		result &= (mask - 1);
+		while (extra-- > 0)
+		{
+			ucs4 c = *buf++;
+
+			if (!(c & 0x80))
+				break;
+			result = (result << 6) | (c & 0x3f);
+		}
+	}
+	return result;
+}
+
+char *unparse_ucs4 (ucs4 c, char *buf)
+{
+	char	d;
+	int	bits;
+
+	     if (c <       0x80) { d = c;                         bits= -6; }
+	else if (c <      0x800) { d= ((c >>  6) & 0x1F) | 0xC0;  bits=  0; }
+	else if (c <    0x10000) { d= ((c >> 12) & 0x0F) | 0xE0;  bits=  6; }
+	else if (c <   0x200000) { d= ((c >> 18) & 0x07) | 0xF0;  bits= 12; }
+	else if (c <  0x4000000) { d= ((c >> 24) & 0x03) | 0xF8;  bits= 18; }
+	else if (c < 0x80000000) { d= ((c >> 30) & 0x01) | 0xFC;  bits= 24; }
+	else                     { d = c;                         bits= -6; }
+
+	*buf++ = d;
+
+	for ( ; bits >= 0; bits-= 6)
+		*buf++ = ((c >> bits) & 0x3F) | 0x80;
+	return buf;
+}
+
+ucs4 next_ucs4(char **bufp)
+{
+    char    *buf = *bufp;
+    ucs4    result = *buf++ & 0xff;
+
+    if ((result & 0xc0) == 0xc0)
+    {
+	int	extra = 1;
+	ucs4	mask = 0x20;
+
+	while ((result & mask))
+	{
+	    extra++;
+	    mask >>= 1;
+	}
+	result &= (mask - 1);
+	while (extra--)
+	{
+	    ucs4	c = *buf & 0xff;
+	    if ((c & 0xc0) != 0x80)
+		break;
+	    result = (result << 6) | (c & 0x3f);
+	    buf++;
+	}
+    }
+    *bufp = buf;
+    return result;
+}
+
+char *buf_next (char *buf)
+{
+	if ((*buf++ & 0xc0) == 0xc0)
+		while ((*buf & 0xc0) == 0x80)
+			buf++;
+	return buf;
+}
+
+char *buf_prev (char *buf, char *first)
+{
+	if ((*--buf & 0xc0) == 0x80)
+	{
+		char	*prev = buf;
+		while (prev > first && (*prev & 0xc0) == 0x80)
+			prev--;
+		if ((*prev & 0xc0) == 0xc0)
+			buf = prev;
+	}
+	return buf;
+}
+
 /* set alternate buffer */
 
 SetABuf(b)

Index: c.c
===================================================================
RCS file: /local/src/CVS/jove/c.c,v
retrieving revision 1.3
retrieving revision 1.4
diff -u -d -r1.3 -r1.4
--- c.c	26 Jul 2002 19:58:40 -0000	1.3
+++ c.c	15 May 2003 00:01:06 -0000	1.4
@@ -88,13 +88,13 @@
 	char	re_buf[V_STRING_LEN],
 		*re_alts[NALTS];
 	int	count = 0;
-	register char	*lp,
-			c;
-	char	p_match,
-		re_str[V_STRING_LEN],
-		*cp,
-		quote_c = 0;
-	register int	c_char;
+	char	*lp;
+	ucs4	c;
+	ucs4	p_match;
+	char	re_str[V_STRING_LEN],
+		*cp;
+	ucs4	quote_c = 0;
+	int	c_char;
 	Line	*comment_line = 0;
 	int	comment_cchar;
 	int	in_comment = COMMENT_NONE,
@@ -125,11 +125,11 @@
 		curchar = sp->p_char;	/* here's where I cheat */
 		c_char = curchar;
 		if (dir == FORWARD)
-			c_char--;
+			c_char = buf_prev (lp + c_char, lp) - lp;
 
 		if (backslashed(lp, c_char))
 			continue;
-		c = lp[c_char];
+		c = parse_ucs4 (lp + c_char);
 		/* check if this is a comment (if we're not inside quotes) */
 		if (quote_c == 0) {
 		    if (in_comment == COMMENT_NONE)
@@ -138,32 +138,15 @@
 			{
 			    if (c == '/')
 			    {
-				if (lp[c_char+1] == '*')
+				if (parse_ucs4 (buf_next (lp+c_char)) == '*')
 				    in_comment = COMMENT_C;
-#if 0
-				else if (lp[c_char+1] == '/')
-				{
-				    in_comment = COMMENT_CPP;
-				    comment_line = curline;
-				    comment_cchar = c_char;
-				}
-#endif
 			    }
 			} else {
 			    int	i;
 
-#if 0
-			    for (i = 0; i < c_char; i++)
-				if (lp[i] == '/' && lp[i+1] == '/')
-				{
-				    in_comment = COMMENT_CPP;
-				    comment_line = curline;
-				    comment_cchar = i;
-				    break;
-				}
-#endif
 			    if (in_comment == COMMENT_NONE)
-				if (c == '/' && c_char != 0 && lp[c_char-1] == '*')
+				if (c == '/' && c_char != 0 && 
+				    parse_ucs4 (buf_prev(lp+c_char, lp)) == '*')
 				    in_comment = COMMENT_C;
 			}
 		    }
@@ -188,13 +171,13 @@
 			{
 			    if (dir == FORWARD)
 			    {
-				if (c_char != 0 && lp[c_char - 1] == '*')
+				if (c_char != 0 && parse_ucs4 (buf_prev(lp+c_char, lp)) == '*')
 				    in_comment = COMMENT_NONE;
 			    }
 			    else
 			    {
-				if (lp[c_char+1] == '*')
-				    in_comment = COMMENT_NONE;
+				if (parse_ucs4 (buf_next (lp+c_char)) == '*')
+					in_comment = COMMENT_NONE;
 			    }
 			}
 		    }
@@ -251,12 +234,12 @@
 do_expr(dir, skip_words)
 register int	dir;
 {
-	register char	c,
+	register ucs4	c,
 			syntax = (dir == FORWARD) ? _Op : _Cl;
 
 	if (dir == BACKWARD)
 		b_char(1);
-	c = linebuf[curchar];
+	c = cur_char ();
 	for (;;) {
 		if (!skip_words && ismword(c)) {
 		    WITH_TABLE(curbuf->b_major)
@@ -270,7 +253,7 @@
 		f_char(dir);
 		if (eobp() || bobp())
 			return;
-		c = linebuf[curchar];
+		c = cur_char ();
 	}
 }
 
@@ -364,7 +347,7 @@
 FindMatch(int dir)
 {
 	register Bufpos	*bp;
-	register char	c = linebuf[curchar];
+	register ucs4	c = cur_char ();
 
 	if ((index(p_types, c) == 0) ||
 	    (backslashed(linebuf, curchar)))
@@ -415,7 +398,7 @@
 	}
 	while (num > 0) {
 		if (curchar - offset >= 0 && 
-		    iswhite (linebuf[curchar-offset]) &&
+		    iswhite (parse_ucs4 (linebuf+(curchar-offset))) &&
 		    ((((curpos = calc_pos (linebuf, curchar)) + extrapos)
 		      % CIndIncrmt) == 0) || dir != BACKWARD) 
 		{
@@ -432,9 +415,12 @@
 			
 			tempchar = curchar;
 			while (COMPAREPOS && 
-			       iswhite ((c = linebuf[tempchar-offset])))
+			       iswhite ((c = parse_ucs4 (linebuf+(tempchar-offset)))))
 			{
-				tempchar += step;
+				if (step > 0)
+					tempchar = buf_next(linebuf+tempchar)-linebuf;
+				else
+					tempchar = buf_prev(linebuf+tempchar,linebuf)-linebuf;
 				curpos += step;
 				if (c == '\t')
 					curpos = calc_pos (linebuf, tempchar);
@@ -467,9 +453,10 @@
 		lremove(before.p_line, after.p_line);
 	}
 	if (extrapos) {
+		int	oldpos = curchar;
 		insert_c(' ', abs (extrapos));
 		if (extrapos < 0)
-			curchar += extrapos;
+			curchar = oldpos;
 	}
 	return YES;
 }
@@ -484,11 +471,11 @@
 	/* be fancy - make it line up with the start of this
 	   statement if it crosses lines */
 	Eol ();
-	while (linebuf[curchar] != ')')
+	while (cur_char () != ')')
 	{
-		if (curchar <= 0)
+		if (bolp() <= 0)
 			return NO;
-		curchar--;
+		left_char ();
 	}
 	bp = m_paren (')', BACKWARD, NO, YES);
 	if (bp && bp->p_line != curline) {
@@ -535,14 +522,14 @@
     int		indent;
     int		incrmt = 0;
     int		fromIndent;
-    int		c;
+    ucs4	c;
     int		mp_save;
     
     DOTsave (&save);
     ToIndent ();
     fromIndent = save.p_char - curchar;
     
-    c = linebuf[curchar];
+    c = cur_char ();
     
     bp = 0;
     if (c == '#')
@@ -872,27 +859,28 @@
 	LineInsert(1);
 	DelWtSpace();
 	Bol();
-	for (cp = open_c; *cp; cp++) {
-		if (*cp == '\r') {
+	for (cp = open_c; *cp; cp = buf_next (cp)) {
+		ucs4	c = parse_ucs4(cp);
+		if (c == '\r') {
 			if (!eolp())
 				LineInsert(1);
 			else
 				line_move(FORWARD, 1, NO);
-		} else if (*cp == ' ' || *cp == '\t') {
-			if (linebuf[curchar] != *cp)
-				insert_c(*cp, 1);
+		} else if (c == ' ' || c == '\t') {
+			if (cur_char() != *cp)
+				insert_c(c, 1);
 		} else
 			/* Since we matched the open comment string on this
 			   line, we don't need to worry about crossing line
 			   boundaries. */
-			curchar++;
+			right_char ();
 	}
 	savedot = MakeMark(curline, curchar, M_FLOATER);
 
 	/* We need to strip the line header pattern of leading white space
 	   since we need to match the line after all of its leading
 	   whitespace is gone. */
-	for (cp = l_header; *cp && (isspace(*cp)); cp++)
+	for (cp = l_header; *cp && (isspace(parse_ucs4 (cp))); cp = buf_next(cp))
 		;
 	header_len = strlen(cp);
 	trailer_len = strlen(l_trailer);
@@ -968,5 +956,5 @@
 	del_char(FORWARD, 1);
 }
 
-#endif CMT_FMT
+#endif /* CMT_FMT */
 

Index: case.c
===================================================================
RCS file: /local/src/CVS/jove/case.c,v
retrieving revision 1.2
retrieving revision 1.3
diff -u -d -r1.2 -r1.3
--- case.c	11 Jul 2001 06:42:26 -0000	1.2
+++ case.c	15 May 2003 00:01:06 -0000	1.3
@@ -34,12 +34,7 @@
 			modify();
 			makedirty(curline);
 		}
-		if (eolp()) {
-			if (curline->l_next == 0)
-				break;
-			SetLine(curline->l_next);
-		} else
-			curchar++;
+		f_char (1);
 	}
 	if (restore)
 		SetDot(&b);
@@ -68,13 +63,13 @@
 			modify();
 			makedirty(curline);
 		}
-		curchar++;
-		while (!eolp() && isword(linebuf[curchar])) {
+		right_char ();
+		while (!eolp() && isword(cur_char ())) {
 			if (lower(&linebuf[curchar])) {
 				modify();
 				makedirty(curline);
 			}
-			curchar++;
+			right_char ();
 		}
 	}
 	if (restore)

Index: ctype.h
===================================================================
RCS file: /local/src/CVS/jove/ctype.h,v
retrieving revision 1.1.1.1
retrieving revision 1.2
diff -u -d -r1.1.1.1 -r1.2
--- ctype.h	10 Mar 2001 12:23:12 -0000	1.1.1.1
+++ ctype.h	15 May 2003 00:01:06 -0000	1.2
@@ -17,21 +17,22 @@
 #define _Cl	0200
 
 extern int	SyntaxTable;
+#define islatin1(c)	((unsigned) (c) < 0x100)
 #define iswhite(c)	(isspace(c))
-#define isword(c)	((CharTable[SyntaxTable])[((int) c) & 0377]&(_W))
-#define	isalpha(c)	((CharTable[SyntaxTable])[((int) c) & 0377]&(_U|_L))
-#define	isupper(c)	((CharTable[SyntaxTable])[((int) c) & 0377]&_U)
-#define	islower(c)	((CharTable[SyntaxTable])[((int) c) & 0377]&_L)
-#define	isdigit(c)	((CharTable[SyntaxTable])[((int) c) & 0377]&_N)
+#define isword(c)	(islatin1(c) && (CharTable[SyntaxTable])[((int) c) & 0377]&(_W))
+#define	isalpha(c)	(islatin1(c) && (CharTable[SyntaxTable])[((int) c) & 0377]&(_U|_L))
+#define	isupper(c)	(islatin1(c) && (CharTable[SyntaxTable])[((int) c) & 0377]&_U)
+#define	islower(c)	(islatin1(c) && (CharTable[SyntaxTable])[((int) c) & 0377]&_L)
+#define	isdigit(c)	(islatin1(c) && (CharTable[SyntaxTable])[((int) c) & 0377]&_N)
 #define	isspace(c)	(c == ' ' || c == '\t')
-#define ispunct(c)	((CharTable[SyntaxTable])[((int) c) & 0377]&_P)
+#define ispunct(c)	(islatin1(c) && (CharTable[SyntaxTable])[((int) c) & 0377]&_P)
 #define toupper(c)	((c)&~040)
 #define tolower(c)	((c)|040)
 #define toascii(c)	((c)&0377)
-#define isctrl(c)	((CharTable[0][((int) c) & 0377])&_C)
-#define isopenp(c)	((CharTable[0][((int) c) & 0377])&_Op)
-#define isclosep(c)	((CharTable[0][((int) c) & 0377])&_Cl)
-#define has_syntax(c,s)	((CharTable[SyntaxTable][((int) c) & 0377])&s)
+#define isctrl(c)	(islatin1(c) && (CharTable[0][((int) c) & 0377])&_C)
+#define isopenp(c)	(islatin1(c) && (CharTable[0][((int) c) & 0377])&_Op)
+#define isclosep(c)	(islatin1(c) && (CharTable[0][((int) c) & 0377])&_Cl)
+#define has_syntax(c,s)	(islatin1(c) && (CharTable[SyntaxTable][((int) c) & 0377])&s)
 #define WITH_TABLE(x) \
 { \
 	int	push = SyntaxTable; \

Index: delete.c
===================================================================
RCS file: /local/src/CVS/jove/delete.c,v
retrieving revision 1.1.1.1
retrieving revision 1.2
diff -u -d -r1.1.1.1 -r1.2
--- delete.c	10 Mar 2001 12:23:11 -0000	1.1.1.1
+++ delete.c	15 May 2003 00:01:06 -0000	1.2
@@ -107,9 +107,12 @@
 DelPChar()
 {
 	if (MinorMode(OverWrite)) {
-		int	count = min(arg_value(), curchar);
+		int	count = arg_value();
+		int	i;
 
-		b_char(count);
+		for (i = 0; i < count && !bolp(); i++)
+			b_char(1);
+		count = i;
 
 		/* overwrite with spaces */
 		set_arg_value(count);
@@ -238,13 +241,16 @@
 
 DelWtSpace()
 {
-	register char	*ep = &linebuf[curchar],
-			*sp = &linebuf[curchar];
+	char	*ep = &linebuf[curchar],
+		*sp = &linebuf[curchar], *pp;
+	ucs4	c;
+	
 
-	while (*ep == ' ' || *ep == '\t')
-		ep++;
-	while (sp > linebuf && *(sp - 1) == ' ' || *(sp - 1) == '\t')
-		sp--;
+	while ((c = parse_ucs4 (ep)) == ' ' || c == '\t')
+		ep = buf_next (ep);
+	while ((pp = buf_prev (sp, linebuf)) != sp &&
+	       (c = parse_ucs4 (pp)) == ' ' || c == '\t')
+		sp = pp;
 	if (sp != ep) {
 		curchar = sp - linebuf;
 		DFixMarks(curline, curchar, curline, curchar + (ep - sp));
@@ -265,7 +271,7 @@
 	register Mark	*dot;
 	int	all;
 
-	if (!blnkp(&linebuf[curchar]))
+	if (!blnkp(cur_char ()))
 		return;
 	dot = MakeMark(curline, curchar, M_FLOATER);
 	all = !blnkp(linebuf);

Index: disp.c
===================================================================
RCS file: /local/src/CVS/jove/disp.c,v
retrieving revision 1.2
retrieving revision 1.3
diff -u -d -r1.2 -r1.3
--- disp.c	11 Jul 2001 06:42:26 -0000	1.2
+++ disp.c	15 May 2003 00:01:06 -0000	1.3
@@ -152,14 +152,14 @@
 	return calc_pos(lcontents(line), c_char);
 }
 
-calc_pos(lp, c_char)
-register char	*lp;
-register int	c_char;
+int
+calc_pos(char *lp, int c_char)
 {
-	register int	pos = 0;
-	register int	c;
+	int	pos = 0;
+	ucs4	c;
+	char	*ep = lp + c_char;
 
-	while ((--c_char >= 0) && (c = *lp++) != 0) {
+	while (lp < ep && (c = next_ucs4 (&lp))) {
 		if (c == '\t')
 			pos += (tabstop - (pos % tabstop));
 		else if (isctrl(c))
@@ -515,7 +515,7 @@
 			else
 				PhysScreen[linenum].s_id = -1;
 		} else
-#endif ID_CHAR
+#endif /* ID_CHAR */
 		    if (BufSwrite(linenum))
 			do_cl_eol(linenum);
 		else
@@ -574,20 +574,18 @@
 }
 
 /* private*/
-DeTab(s_offset, buf, outbuf, limit, visspace)
-register char	*buf;
-char	*outbuf;
+DeTab(int s_offset, char *buf, ucs4 *outbuf, int limit, int visspace)
 {
-	register char	*phys_p = outbuf,
+	register ucs4	*phys_p = outbuf,
 			c;
 	register int	pos = 0;
-	char		*limitp = &outbuf[limit];
+	ucs4		*limitp = &outbuf[limit];
 
 #define OkayOut(ch)	if ((pos++ >= s_offset) && (phys_p < limitp))\
 				*phys_p++ = ch;\
 			else
 
-	while (c = *buf++) {
+	while (c = next_ucs4(&buf)) {
 		if (c == '\t') {
 			int	nchars = (tabstop - (pos % tabstop));
 
@@ -621,8 +619,7 @@
   	Returns Non-Zero if you are finished (no differences left). */
 
 /*private*/
-IDchar(new, lineno, col)
-register char	*new;
+IDchar(ucs4 *new, int lineno, int col)
 {
 	register int	i;
 	int	j,
@@ -810,7 +807,7 @@
 		CapCol = CO - 1;
 }
 
-#endif ID_CHAR
+#endif /* ID_CHAR */
 
 /* chkmail() returns nonzero if there is new mail since the
    last time we checked. */

Index: error.c
===================================================================
RCS file: /local/src/CVS/jove/error.c,v
retrieving revision 1.1.1.1
retrieving revision 1.2
diff -u -d -r1.1.1.1 -r1.2
--- error.c	10 Mar 2001 12:23:11 -0000	1.1.1.1
+++ error.c	15 May 2003 00:01:06 -0000	1.2
@@ -294,4 +294,4 @@
 	SetBuf(buftospel);
 }
 
-#endif SPELL
+#endif /* SPELL */

Index: fp.c
===================================================================
RCS file: /local/src/CVS/jove/fp.c,v
retrieving revision 1.3
retrieving revision 1.4
diff -u -d -r1.3 -r1.4
--- fp.c	20 Feb 2002 20:34:31 -0000	1.3
+++ fp.c	15 May 2003 00:01:06 -0000	1.4
@@ -31,7 +31,41 @@
 	termflush(EOF);
 }
 
-termflush(c)
+int Utf8Out = 0;
+
+void
+termoutchar (ucs4 c)
+{
+    char	d;
+    int	bits;
+
+    if (!Utf8Out)
+    {
+	bits = -6;
+	if (c < 0x100)
+	    d = c;
+	else
+	    d = '�';
+    }
+    else
+    {
+	     if (c <       0x80) { d = c;                         bits= -6; }
+	else if (c <      0x800) { d= ((c >>  6) & 0x1F) | 0xC0;  bits=  0; }
+	else if (c <    0x10000) { d= ((c >> 12) & 0x0F) | 0xE0;  bits=  6; }
+	else if (c <   0x200000) { d= ((c >> 18) & 0x07) | 0xF0;  bits= 12; }
+	else if (c <  0x4000000) { d= ((c >> 24) & 0x03) | 0xF8;  bits= 18; }
+	else if (c < 0x80000000) { d= ((c >> 30) & 0x01) | 0xFC;  bits= 24; }
+	else                     { d = c;                         bits= -6; }
+    }
+
+    termoutbyte (d);
+
+    for ( ; bits >= 0; bits-= 6)
+	termoutbyte (((c >> bits) & 0x3F) | 0x80);
+}
+
+void
+termflush(char c)
 {
 	register int	n;
 

Index: funcdefs.c
===================================================================
RCS file: /local/src/CVS/jove/funcdefs.c,v
retrieving revision 1.1.1.1
retrieving revision 1.2
diff -u -d -r1.1.1.1 -r1.2
--- funcdefs.c	10 Mar 2001 12:23:11 -0000	1.1.1.1
+++ funcdefs.c	15 May 2003 00:01:06 -0000	1.2
@@ -155,6 +155,7 @@
 	Tab(),
 	DescribeCompose(),
 	HandleCompose(),
+	HandleUnicode(),
 	DoParen(),
 	DoOpenBrace(),
 	ParseAll(),
@@ -283,11 +284,11 @@
 
 #	define WIRED_CMD(c)	c
 
-#else TXT_TO_C
+#else /* TXT_TO_C */
 
 #	define WIRED_CMD(c)	0
 
-#endif TXT_TO_C
+#endif /* TXT_TO_C */
 
 struct cmd	commands[] = {
 	FUNCTION, "Prefix-1", WIRED_CMD(EscPrefix),
@@ -400,7 +401,7 @@
 	FUNCTION, "exit-jove", WIRED_CMD(Leave),
 #ifdef CMT_FMT
  	FUNCMOD, "fill-comment", WIRED_CMD(Comment),
-#endif CMT_FMT
+#endif /* CMT_FMT */
 	FUNCMOD, "fill-paragraph", WIRED_CMD(Justify),
 	FUNCMOD, "fill-region", WIRED_CMD(RegJustify),
 	FUNCMOD, "filter-buffer", WIRED_CMD(FilterBuffer),
@@ -430,6 +431,7 @@
 	FUNCTION, "highlight-current-line", WIRED_CMD(WHighCur),
 	FUNCINS, "handle-compose", WIRED_CMD(HandleCompose),
 	FUNCINS, "handle-tab", WIRED_CMD(Tab),
+	FUNCINS, "handle-unicode", WIRED_CMD(HandleUnicode),
 	FUNCTION, "i-search-forward", WIRED_CMD(IncFSearch),
 	FUNCTION, "i-search-reverse", WIRED_CMD(IncRSearch),
 	FUNCMOD, "indent-line", WIRED_CMD(IndentLine),

Index: insert.c
===================================================================
RCS file: /local/src/CVS/jove/insert.c,v
retrieving revision 1.2
retrieving revision 1.3
diff -u -d -r1.2 -r1.3
--- insert.c	11 Jul 2001 06:42:26 -0000	1.2
+++ insert.c	15 May 2003 00:01:06 -0000	1.3
@@ -1,5 +1,5 @@
 /************************************************************************
- * This program is Copyright (C) 1986 by Jonathan Payne.  JOVE is       *
+ * This program is Copyright © 1986 by Jonathan Payne.  JOVE is         *
  * provided to you without charge, and with no warranty.  You may give  *
  * away copies of JOVE, including sources, provided that this notice is *
  * included in all the files.                                           *
@@ -58,7 +58,7 @@
 	}
 
 	modify();
-	if (curchar != 0) {
+	if (!bolp()) {
 		strcpy(new_line, &linebuf[curchar]);
 		linebuf[curchar] = '\0';	/* Shorten this line */
 		SavLine(curline, linebuf);
@@ -77,15 +77,17 @@
 
 /* Replaces white space before dot to make dot reach goal */
  
-n_tab(goal)
-register int	goal;
+void
+n_tab(int goal)
 {
 	int	incrmt, dotcol;
-	register char	*sp, *ep;
+	char	*sp, *ep, *pp;
+	ucs4	c;
 	
 	sp = ep = &linebuf[curchar];
-	while (sp > linebuf && *(sp - 1) == ' ' || *(sp - 1) == '\t')
-		sp--;
+	while ((pp = buf_prev (sp, linebuf)) >= linebuf &&
+	       (c = parse_ucs4 (pp)) == ' ' || c == '\t')
+		sp = pp;
 	if (sp != ep) {
 		curchar = sp - linebuf;
 		DFixMarks(curline, curchar, curline, curchar + (ep - sp));
@@ -137,7 +139,7 @@
 {
 #ifdef ABBREV
 	if (MinorMode(Abbrev) && !ismword(LastKeyStruck) &&
-	    !bolp() && ismword(linebuf[curchar - 1]))
+	    !bolp() && ismword(prev_char ()))
 		AbbrevExpand();
 #endif
 	if (MinorMode(OverWrite)) {
@@ -148,7 +150,7 @@
 			int	pos = calc_pos(linebuf, curchar);
 
 			if (!eolp()) {
-				if (linebuf[curchar] == '\t') {
+				if (cur_char () == '\t') {
 					if ((pos + 1) == ((pos + tabstop) - (pos % tabstop)))
 						del_char(FORWARD, 1);
 				} else
@@ -168,16 +170,45 @@
 	insert_c(c, arg_value());
 }
 
+/* insert ucs4 character C at point */
+static void
+insert_ucs4 (ucs4 c)
+{
+	char	d;
+	int	bits;
+	int	s = curchar;
+
+	     if (c <       0x80) { d = c;                         bits= -6; }
+	else if (c <      0x800) { d= ((c >>  6) & 0x1F) | 0xC0;  bits=  0; }
+	else if (c <    0x10000) { d= ((c >> 12) & 0x0F) | 0xE0;  bits=  6; }
+	else if (c <   0x200000) { d= ((c >> 18) & 0x07) | 0xF0;  bits= 12; }
+	else if (c <  0x4000000) { d= ((c >> 24) & 0x03) | 0xF8;  bits= 18; }
+	else if (c < 0x80000000) { d= ((c >> 30) & 0x01) | 0xFC;  bits= 24; }
+	else d = c;
+
+	ins_c (d, linebuf, curchar, 1, LBSIZE);
+	curchar++;
+
+	for ( ; bits >= 0; bits-= 6)
+	{
+		ins_c ((((c >> bits) & 0x3F) | 0x80), linebuf, curchar, 1, LBSIZE);
+		curchar++;
+	}
+}
+
 /* insert character C N times at point */
-insert_c(c, n)
+void
+insert_c(ucs4 c, int n)
 {
+	int	s = curchar;
+	
 	if (n <= 0)
 		return;
 	modify();
 	makedirty(curline);
-	ins_c(c, linebuf, curchar, n, LBSIZE);
-	IFixMarks(curline, curchar, curline, curchar + n);
-	curchar += n;
+	while (n--)
+		insert_ucs4 (c);
+	IFixMarks(curline, s, curline, curchar);
 }	
 
 /* Tab in to the right place for C mode */
@@ -204,7 +235,7 @@
 		if (strlen (linebuf) == 0)
 		{
 			(void) c_indent();
-			if (curchar == 0)
+			if (bolp())
 				c_tab();
 		}
 		else
@@ -380,7 +411,7 @@
 		if ((curline != line1 || curchar >= char1) &&
 		    (curline != line2 || curchar < char2)) 
 		{
-			if (!linebuf[curchar])
+			if (!cur_char())
 				DelWtSpace();
 			else
 				IndentLine ();
@@ -429,24 +460,33 @@
 		Insert(c);
 }
 
+#define COMPOSE_SEP ':'
+
 char	ComposeSequences[V_STRING_LEN] = 
-"!=�|c=�#=�$=�y=�|=�ss=�\" =�cr=�_a=�<=�~ =�- =�rc=�__=�. =�+-=�2 =�3 =�' =�,u=�pp=�..=�, =�1 =�m=�>=�14=�12=�34=�?=�`A=�'A=�^A=�~A=�\"A=�.A=�AE=�,C=�`E=�'E=�^E=�\"E=�`I=�'I=�^I=�\"I=�-D=�~N=�`O=�'O=�^O=�~O=�\"O=�* =�/O=�`U=�'U=�^U=�\"U=�'Y=�TH=�SS=�`a=�'a=�^a=�~a=�\"a=�.a=�ae=�,c=�`e=�'e=�^e=�\"e=�`i=�'i=�^i=�\"i=�-d=�~n=�`o=�'o=�^o=�~o=�\"o=�/ =�/o=�`u=�'u=�^u=�\"u=�'y=�th=�\"y=\377";
+/* Latin 1 Supplement */
+"!:¡|c:¢#:£$:¤y:¥| :¦ss:§\" :¨cr:©_a:ª<<:«~:¬-:rc:®__:¯"
+". :°+-:±2:²3:³' :´,u:µpp:¶. :·, :¸1:¹m:º>>:»14:¼12:½34:¾?:¿"
+"`A:À'A:Á^A:Â~A:Ã\"A:Ä.A:ÅAE:ÆC,:Ç`E:È'E:É^E:Ê\"E:Ë`I:Ì'I:Í^I:Î\"I:Ï"
+"-D:Ð~N:Ñ`O:Ò'O:Ó^O:Ô~O:Õ\"O:Ö *:×/O:Ø`U:Ù'u:Ú^U:Û\"U:Ü'Y:ÝTH:ÞSS:ß"
+"`a:à'a:á^a:â~a:ã\"a:ä.a:åae:æ,c:ç`e:è'e:é^e:ê\"e:ë`i:ì'i:í^i:î\"i:ï"
+"et:ð~n:ñ`o:ò'o:ó^o:ô~o:õ\"o:ö /:÷/o:ø`u:ù'u:ú^u:û\"u:ü'y:ýth:þ\"y:ÿ"
+/* Currency Symbols */
+"CE:₠/C:₡CR:₢=F:₣=L:₤/m:₥=N:₦PT:₧RS:₨=W:₩NS:₪=d:₫=C:€-K:₭=T:₮DP=₯"
+/* General punctuation */
+"||:‖__:‗` :‘' :’, :‚``:“'':”,,:„+ :†++:‡o :•"
+;
 
 char *
-NextComposeSequence (compose)
-	char *compose;
+NextComposeSequence (char *compose)
 {
+	ucs4	c;
 	if (!compose)
 		return 0;
-	while (*compose && *compose != '=')
-		compose++;
-	if (!*compose)
-		return 0;
-	if (!compose[1])
-		return 0;
-	if (!compose[2])
-		return 0;
-	return compose + 2;
+	while ((c = next_ucs4 (&compose)) && c != ':')
+	    ;
+	if (!c)
+	    return 0;
+	return buf_next (compose);
 }
 
 char *
@@ -466,18 +506,20 @@
     int	    maxlen;
     char    *compose;
     char    *equal;
+    char    *semi;
     int	    nlinespercol;
     int	    nlines;
     int	    col;
     int	    which;
     char    seq[64];
+    char    val[64];
 
     maxlen = 0;
     for (compose = ComposeSequences; 
 	 compose; 
 	 compose = NextComposeSequence (compose))
     {
-	equal = index (compose, '=');
+	equal = index (compose, ':');
 	if (!equal)
 	    break;
 	ncompose++;
@@ -497,18 +539,49 @@
 	    compose = getCompose (which);
 	    if (!compose)
 		break;
-	    equal = index (compose, '=');
+	    equal = index (compose, ':');
 	    if (!equal)
 		break;
+	    semi = buf_next (equal+1);
 	    strncpy (seq, compose, equal - compose);
 	    seq[equal - compose] = '\0';
-	    Typeout ("%c = %-*s", equal[1], maxlen-4, seq);
+	    strncpy (val, equal + 1, semi - (equal + 1));
+	    val[semi - (equal + 1)] = '\0';
+	    Typeout ("%s = %-*s", val, maxlen-4, seq);
 	}
 	Typeout ((char *) 0);
     }
     TOstop ();
 }
 
+#define ishex(c)	(('0' <= (c) && (c) <= '9') || \
+			 ('a' <= (c) && (c) <= 'f') || \
+			 ('A' <= (c) && (c) <= 'F'))
+			 
+HandleUnicode()
+{
+	int	len = 0;
+	ucs4	u = 0, v;
+	int	c, slow;
+	
+	for (;;) {
+		c = waitchar (&slow);
+		if (slow)
+			message (key_strokes);
+		if ('0' <= c && c <= '9')
+			v = c - '0';
+		else if ('a' <= c && c <= 'f')
+			v = c - 'a' + 10;
+		else if ('A' <= c && c <= 'F')
+			v = c - 'A' + 10;
+		else
+			break;
+		u = (u << 4) | v;
+	}
+	if (u)
+		insert_c (u, 1);
+}
+
 HandleCompose()
 {
 	char	sequence[V_STRING_LEN];
@@ -536,9 +609,9 @@
 		}
 		if (!compose)
 			complain("Invalid compose sequence");
-		if (compose[len] == '=')
+		if (compose[len] == ':')
 		{
-			Insert (compose[len+1]);
+			Insert (parse_ucs4(compose+len+1));
 			return;
 		}
 	}
@@ -631,7 +704,7 @@
 {
 #ifdef ABBREV
 	if (MinorMode(Abbrev) && !ismword(LastKeyStruck) &&
-	    !bolp() && ismword(linebuf[curchar - 1]))
+	    !bolp() && ismword(prev_char ()))
 		AbbrevExpand();
 #endif
 #ifdef LISP
@@ -653,7 +726,7 @@
 ins_str(str, ok_nl)
 register char	*str;
 {
-	register char	c;
+	register ucs4	c;
 	Bufpos	save;
 	int	llen;
 
@@ -661,7 +734,8 @@
 		return;		/* ain't nothing to insert! */
 	DOTsave(&save);
 	llen = strlen(linebuf);
-	while (c = *str++) {
+	while (c = parse_ucs4 (str)) {
+		str = buf_next (str);
 		if (c == '\n' || (ok_nl && llen >= LBSIZE - 2)) {
 			IFixMarks(save.p_line, save.p_char, curline, curchar);
 			modify();
@@ -671,7 +745,7 @@
 			llen = strlen(linebuf);
 		}
 		if (c != '\n') {
-			ins_c(c, linebuf, curchar++, 1, LBSIZE);
+			insert_ucs4 (c);
 			llen++;
 		}
 	}
@@ -952,7 +1026,7 @@
 	Bufpos	dot,
 		end;
 
-	if (!isopenp (linebuf[curchar]))
+	if (!isopenp (cur_char ()))
 		complain((char *) 0);
 	DOTsave(&dot);
 	FSexpr();
@@ -1038,7 +1112,7 @@
 
 	SetDot(bp);
 	f_char(1);
-	if (linebuf[curchar] != '(') {
+	if (cur_char () != '(') {
 		register Word	*wp;
 
 		if (specials == NIL)
@@ -1054,10 +1128,10 @@
 			END_TABLE();
 			if (LookingAt("[ \t]*;\\|[ \t]*$", linebuf, curchar))
 				curchar = c_char;
-			else while (linebuf[curchar] == ' ')
-				curchar++;
+			else while (cur_char () == ' ')
+				right_char ();
 		} else
-			curchar++;
+			right_char();
 	}
 	goal = calc_pos(linebuf, curchar);
 	SetDot(&savedot);
@@ -1067,4 +1141,4 @@
 
 	return bp;
 }
-#endif LISP
+#endif /* LISP */

Index: io.c
===================================================================
RCS file: /local/src/CVS/jove/io.c,v
retrieving revision 1.3
retrieving revision 1.4
diff -u -d -r1.3 -r1.4
--- io.c	11 Jul 2001 06:42:26 -0000	1.3
+++ io.c	15 May 2003 00:01:06 -0000	1.4
@@ -142,7 +142,7 @@
 	strcpy(end, linebuf + curchar);
 	xeof = jove_gets(fp, linebuf + curchar, LBSIZE - curchar);
 	len = strlen (linebuf);
-	if (linebuf[len-1] == '\r') {
+	if (parse_ucs4(buf_prev(linebuf+len, linebuf)) == '\r') {
 		if (!is_insert && AutoCarriageReturn)
 			curbuf->b_minor |= CarriageReturns;
 		if (MinorMode(CarriageReturns))
@@ -471,7 +471,7 @@
 	} while (sp != 0);
 }
 
-#endif CHDIR
+#endif /* CHDIR */
 
 # include	<pwd.h>
 

Index: io.h
===================================================================
RCS file: /local/src/CVS/jove/io.h,v
retrieving revision 1.1.1.1
retrieving revision 1.2
diff -u -d -r1.1.1.1 -r1.2
--- io.h	10 Mar 2001 12:23:12 -0000	1.1.1.1
+++ io.h	15 May 2003 00:01:06 -0000	1.2
@@ -15,7 +15,10 @@
 
 extern Terminal	terminal;
 
-#define termoutchar(c)	(--terminal.t_cnt >= 0 ? (*terminal.t_ptr++ = (c)) : termflush((c)))
+#define termoutbyte(c)	(--terminal.t_cnt >= 0 ? (*terminal.t_ptr++ = (c)) : termflush((c)))
+
+void	termoutchar (ucs4 c);
+void	termflush (char c);
 
 #define T_ERR		040
 

Index: iproc.c
===================================================================
RCS file: /local/src/CVS/jove/iproc.c,v
retrieving revision 1.2
retrieving revision 1.3
diff -u -d -r1.2 -r1.3
--- iproc.c	11 Jul 2001 06:42:26 -0000	1.2
+++ iproc.c	15 May 2003 00:01:06 -0000	1.3
@@ -524,13 +524,11 @@
 proc_kill(p, sig)
 register Process	*p;
 {
-#ifndef LINUX
-	extern char	*sys_errlist[];
-#endif
 	if (isdead(p))
 		return;
 	if (killpg(p->p_pid, sig) == -1)
-		s_mess("Cannot kill %s (%d): %s", p->p_name, p->p_pid, sys_errlist[errno]);
+		s_mess("Cannot kill %s (%d): %s", p->p_name, p->p_pid, 
+		       strerror (errno));
 }
 
 proc_eof (p)
@@ -946,8 +944,8 @@
 {
 	if (!eolp ())
 		complain ("[not at end of line]");
-	while (curchar && !iswhite (linebuf[curchar-1]))
-		--curchar;
+	while (!bolp() && !iswhite (cur_char ()))
+		left_char ();
 	f_complete (c, linebuf + curchar);
 	Eol ();
 }

Index: jove.c
===================================================================
RCS file: /local/src/CVS/jove/jove.c,v
retrieving revision 1.3
retrieving revision 1.4
diff -u -d -r1.3 -r1.4
--- jove.c	24 Aug 2001 09:02:38 -0000	1.3
+++ jove.c	15 May 2003 00:01:06 -0000	1.4
@@ -35,10 +35,10 @@
 #else
 #ifdef SYSV
 struct Termio	sg1, sg2;
-#else SYSV
+#else /* SYSV */
 struct sgttyb	sg1, sg2;
-#endif SYSV
-#endif BRLUNIX
+#endif /* SYSV */
+#endif /* BRLUNIX */
 
 #ifdef BIFF
 private struct stat	tt_stat;	/* for biff */
@@ -140,7 +140,7 @@
     if (fcntl(fd, F_SETFL, on ? blockf : nonblockf) == -1)
 	finish(SIGHUP);
 }
-#endif SYSV
+#endif /*  SYSV */
 
 Peekc()
 {
@@ -178,7 +178,7 @@
 #else
 		while (nchars < 0 && errno == EINTR);
 		if (nchars <= 0)
-#endif SYSV
+#endif /*  SYSV */
 			finish(SIGHUP);
 		bp = smbuf;
 		InputPending = nchars > 1;
@@ -186,7 +186,7 @@
 	nchars--;
 	return *bp++ & 0377;
 }
-#else PIPEPROCS
+#else /*  PIPEPROCS */
 
 terminchar()
 {
@@ -270,8 +270,8 @@
 	nchars--;
 	return *bp++ & 0377;
 }
-#endif PIPEPROCS
-#else IPROCS
+#endif /*  PIPEPROCS */
+#else /*  IPROCS */
 terminchar()
 {
 	extern int	errno;
@@ -290,7 +290,7 @@
 	nchars--;
 	return *bp++ & 0377;
 }
-#endif IPROCS
+#endif /*  IPROCS */
 
 int	InputPending = 0;
 
@@ -477,7 +477,7 @@
 			*lip = win.ws_row;
 	}
 	}
-#else TIOCGWINSZ
+#else /*  TIOCGWINSZ */
 #ifdef BTL_BLIT
 #include <sys/jioctl.h>
 	{
@@ -490,8 +490,8 @@
 			*lip = jwin.bytesy;
 	}
 	}
-#endif BTL_BLIT
-#endif TIOCGWINSZ
+#endif /*  BTL_BLIT */
+#endif /*  TIOCGWINSZ */
 	/*
 	 * bounds check for absurd values - this will break when terminals get
 	 * more than 1000 characters across or down but I don't think that's
@@ -518,7 +518,7 @@
 		win.ws_row = li;
 		ioctl (0, TIOCSWINSZ, &win);
 	}
-#else TIOCGWINSZ
+#else /*  TIOCGWINSZ */
 #ifdef BTL_BLIT
 #include <sys/jioctl.h>
 	{
@@ -533,8 +533,8 @@
 				*lip = jwin.bytesy;
 		}
 	}
-#endif BTL_BLIT
-#endif TIOCGWINSZ
+#endif /*  BTL_BLIT */
+#endif /*  TIOCGWINSZ */
 }
 
 #ifdef BIFF
@@ -580,7 +580,7 @@
 	(void) ioctl(0, GetAttr, (char *) &sg1);
 #else
 	(void) gtty(0, &sg1);
-#endif SYSV
+#endif /*  SYSV */
 	sg2 = sg1;
 
 #ifdef SYSV
@@ -613,7 +613,7 @@
 	sg2.sg_xflags &= ~((sg2.sg_xflags&DC3DC1 ? 0 : STALL) | PAGE);
 #else
 	sg2.sg_flags &= ~(ECHO | CRMOD);
-#endif BRLUNIX
+#endif /*  BRLUNIX */
 
 #ifdef EUNICE
 	sg2.sg_flags |= RAW;	/* Eunice needs RAW mode last I heard. */
@@ -626,8 +626,8 @@
 #   endif
 #else
 	sg2.sg_flags |= (MetaKey ? RAW : CBREAK);
-#endif PURDUE_EE
-#endif EUNICE
+#endif /*  PURDUE_EE */
+#endif /*  EUNICE */
 #ifdef TIOCSLTC
 	(void) ioctl(0, TIOCGLTC, (struct sgttyb *) &ls1);
 	ls2 = ls1;
@@ -646,8 +646,8 @@
 		tc2.t_stopc = (char) -1;
 		tc2.t_startc = (char) -1;
 	}
-#endif TIOCGETC
-#endif SYSV
+#endif /*  TIOCGETC */
+#endif /*  SYSV */
 }
 
 tty_reset()
@@ -679,16 +679,16 @@
 		perror ("TIOCSETN");
 	}
 	}
-#endif BRLUNIX
-#endif SYSV
+#endif /*  BRLUNIX */
+#endif /*  SYSV */
 
 #ifndef SYSV
 #ifdef TIOCSETC
 	(void) ioctl(0, TIOCSETC, n == 0 ? (struct sgttyb *) &tc1 : (struct sgttyb *) &tc2);
-#endif TIOCSETC
+#endif /*  TIOCSETC */
 #ifdef TIOCSLTC
 	(void) ioctl(0, TIOCSLTC, n == 0 ? (struct sgttyb *) &ls1 : (struct sgttyb *) &ls2);
-#endif TIOCSLTC
+#endif /*  TIOCSLTC */
 #endif
 	done_ttinit = 1;
 #ifdef BIFF
@@ -1471,7 +1471,7 @@
 	sprintf(Mailbox, "/usr/mail/%s", getenv("LOGNAME"));
 #else
 	sprintf(Mailbox, "/usr/spool/mail/%s", getenv("USER"));
-#endif SYSV
+#endif /*  SYSV */
 	(void) joverc(Joverc, 0);
 	if (!scanvec(argv, "-j")) {
 		char	tmpbuf[100];

Index: jove.h
===================================================================
RCS file: /local/src/CVS/jove/jove.h,v
retrieving revision 1.4
retrieving revision 1.5
diff -u -d -r1.4 -r1.5
--- jove.h	20 Feb 2002 20:34:31 -0000	1.4
+++ jove.h	15 May 2003 00:01:06 -0000	1.5
@@ -134,6 +134,19 @@
 #define CharUpcase(c)	(CaseEquiv[c])
 #define WinText(w)	(((w)->w_flags & W_NUMLINES ? 8 : 0) + ((w)->w_flags & W_HIGHCUR ? 1 : 0))
 
+typedef unsigned int	ucs4;
+
+ucs4 parse_ucs4 (char *buf);
+ucs4 next_ucs4 (char **bufp);
+char *unparse_ucs4 (ucs4 c, char *buf);
+char *buf_next (char *buf);
+char *buf_prev (char *buf, char *first);
+#define char_at(i)	parse_ucs4(linebuf + (i))
+#define cur_char()	char_at(curchar)
+#define prev_char()	parse_ucs4(buf_prev (linebuf + curchar, linebuf))
+#define right_char()	(curchar = buf_next(linebuf+curchar) - linebuf)
+#define left_char()	(curchar = buf_prev(linebuf+curchar,linebuf)-linebuf)
+    
 extern int	OkayAbort,	/* okay to abort redisplay */
 		BufSize;
 
@@ -283,7 +296,7 @@
 # define PROC_IO	0	/* interactive process using pty */
 # define PROC_IN	1	/* non-interactive process using pipe */
 
-#endif IPROCS
+#endif /* IPROCS */
 
 struct window {
 	Window	*w_prev,	/* circular list */
@@ -459,6 +472,7 @@
 extern int
 	OKXonXoff,		/* disable start/stop characters */
 	MetaKey,		/* this terminal has a meta key */
+        Utf8Out,		/* this terminal accepts UTF-8 output */
 	VisBell,		/* use visible bell (if possible) */
 	WrapScan,		/* make searches wrap */
 	phystab,		/* terminal's tabstop settings */ 
@@ -570,15 +584,12 @@
 #define MESG_SIZE SPRINTF_LEN
 extern char	mesgbuf[MESG_SIZE];
 
-#ifdef SEVENBITCHARS
-typedef char screenchar;
-#define INVERSE_BIT 0x80
-#define CHAR_MASK   0x7F
-#else
-typedef short screenchar;
-#define INVERSE_BIT 0x100
-#define CHAR_MASK   0xFF
-#endif
+
+ucs4	next_ucs4(char **bufp);
+    
+typedef ucs4	    screenchar;
+#define INVERSE_BIT 0x80000000
+#define CHAR_MASK   0x7FFFFFFF
 
 struct screenline {
 	screenchar	*s_line,
@@ -685,6 +696,8 @@
 	*mak_buf(),
 	*buf_exists(),
 	*file_exists();
+
+extern void insert_c (ucs4 c, int n);
 
 struct cmd *
 	FindCmd();

Index: keymaps.txt
===================================================================
RCS file: /local/src/CVS/jove/keymaps.txt,v
retrieving revision 1.1.1.1
retrieving revision 1.2
diff -u -d -r1.1.1.1 -r1.2
--- keymaps.txt	10 Mar 2001 12:23:12 -0000	1.1.1.1
+++ keymaps.txt	15 May 2003 00:01:06 -0000	1.2
@@ -42,7 +42,7 @@
 	"pause-jove",			/* ^Z */
 	"Prefix-1",			/* ^[ */
 	"i-search-forward",		/* ^\ */
-	"unbound",			/* ^] */
+	"handle-unicode",		/* ^] */
 	"quoted-insert",		/* ^^ */
 	"unbound",			/* ^_ */
 	"self-insert",			/*    */

Index: misc.c
===================================================================
RCS file: /local/src/CVS/jove/misc.c,v
retrieving revision 1.2
retrieving revision 1.3
diff -u -d -r1.2 -r1.3
--- misc.c	11 Jul 2001 06:42:26 -0000	1.2
+++ misc.c	15 May 2003 00:01:06 -0000	1.3
@@ -60,13 +60,13 @@
 
 TransChar()
 {
-	char	before;
+	ucs4	before;
 
-	if (curchar == 0 || (eolp() && curchar == 1))
+	if (bolp() || (eolp() && curchar == 1))
 		complain((char *) 0);	/* BEEP */
 	if (eolp())
 		b_char(1);
-	before = linebuf[curchar - 1];
+	before = prev_char ();
 	del_char(BACKWARD, 1);
 	f_char(1);
 	insert_c(before, 1);
@@ -283,10 +283,10 @@
 
 ToIndent()
 {
-	register char	*cp,
-			c;
+	char	*cp;
+	ucs4	c;
 
-	for (cp = linebuf; c = *cp; cp++)
+	for (cp = linebuf; c = parse_ucs4(cp); cp = buf_next (cp))
 		if (c != ' ' && c != '\t')
 			break;
 	curchar = cp - linebuf;
@@ -294,10 +294,10 @@
 
 AtIndent ()
 {
-	register char	*cp,
-			c;
+	char	*cp;
+	ucs4	c;
 
-	for (cp = linebuf; c = *cp; cp++)
+	for (cp = linebuf; c = parse_ucs4(cp); cp = buf_next (cp))
 		if (c != ' ' && c != '\t')
 			break;
 	return curchar == cp - linebuf;
@@ -405,7 +405,7 @@
 		complain(unsupported);
 	}
 }
-#endif ANSICODES
+#endif /* ANSICODES */
 
 NotModified()
 {

Index: move.c
===================================================================
RCS file: /local/src/CVS/jove/move.c,v
retrieving revision 1.1.1.1
retrieving revision 1.2
diff -u -d -r1.1.1.1 -r1.2
--- move.c	10 Mar 2001 12:23:12 -0000	1.1.1.1
+++ move.c	15 May 2003 00:01:06 -0000	1.2
@@ -24,7 +24,7 @@
 				break;
 			SetLine(curline->l_next);
 		} else
-			curchar++;
+			right_char();
 	}
 }
 
@@ -43,7 +43,7 @@
 			SetLine(curline->l_prev);
 			Eol();
 		} else
-			--curchar;
+			left_char ();
 	}
 }
 
@@ -416,9 +416,9 @@
 	} else {
 		extern int	REbom;
 
-		curchar = REbom + 1;	/* Just after the [?.!] */
+		curchar = buf_next (linebuf + REbom) - linebuf;	/* Just after the [?.!] */
 		if (LookingAt("[\")]  *\\|[\")]$", linebuf, curchar))
-			curchar++;
+			right_char ();
 		else if (!eolp() && !LookingAt("  *", linebuf, curchar))
 			to_sent(dir);
 	}
@@ -461,7 +461,7 @@
 f_word(num)
 register int	num;
 {
-	register char	c;
+	ucs4	c;
 
 	if (num < 0) {
 		b_word(-num);
@@ -469,8 +469,8 @@
 	}
 	while (--num >= 0) {
 		to_word(FORWARD);
-		while ((c = linebuf[curchar]) != 0 && isword(c))
-			curchar++;
+		while ((c = cur_char()) != 0 && isword(c))
+			right_char ();
 		if (eobp())
 			break;
 	}
@@ -480,16 +480,14 @@
 b_word(num)
 register int	num;
 {
-	register char	c;
-
 	if (num < 0) {
 		f_word(-num);
 		return;
 	}
 	while (--num >= 0) {
 		to_word(BACKWARD);
-		while (!bolp() && (c = linebuf[curchar - 1], isword(c)))
-			--curchar;
+		while (!bolp() && isword(prev_char ()))
+			left_char ();
 		if (bobp())
 			break;
 	}

Index: paragraph.c
===================================================================
RCS file: /local/src/CVS/jove/paragraph.c,v
retrieving revision 1.3
retrieving revision 1.4
diff -u -d -r1.3 -r1.4
--- paragraph.c	20 Feb 2002 20:34:31 -0000	1.3
+++ paragraph.c	15 May 2003 00:01:06 -0000	1.4
@@ -387,32 +387,33 @@
 
 do_space()
 {
-	int	c1 = curchar,
-		c2 = c1,
-		diff,
+	int	diff,
 		nspace;
-	char	ch;
+	ucs4	ch;
+	char	*sp = linebuf + curchar, *pp;
+	char	*ep = linebuf + curchar;
 
-	while (c1 > 0 && ((ch = linebuf[c1 - 1]) == ' ' || ch == '\t'))
-		c1--;
-	while ((ch = linebuf[c2]) == ' ' || ch == '\t')
-		c2++;
-	diff = (c2 - c1);
-	curchar = c2;
+	while ((pp = buf_prev (sp, linebuf)) != sp && 
+	       ((ch = parse_ucs4 (pp)) == ' ' || ch == '\t'))
+		sp = pp;
+	while ((ch = parse_ucs4 (ep)) == ' ' || ch == '\t')
+		ep = buf_next (ep);
+	diff = (ep - sp);
+	curchar = ep - linebuf;
 
 	if (diff == 0)
 		return;
-	if (c1 > 0) {
-		int	topunct = c1 - 1;
+	if (sp != linebuf) {
+		pp = buf_prev (sp, linebuf);
 
 		nspace = 1;
 		if (diff >= 2) {
-			while (index("\")]", linebuf[topunct])) {
-				if (topunct == 0)
+			while (index("\")]", parse_ucs4 (pp))) {
+				if (pp == linebuf)
 					break;
-				topunct--;
+				pp = buf_prev (pp, linebuf);
 			}
-			if (index("?!.:", linebuf[topunct]))
+			if (index("?!.:", parse_ucs4 (pp)))
 				nspace = 2;
 		}
 	} else
@@ -454,7 +455,8 @@
 				del_char(FORWARD, 1);	/* Delete line separator. */
 				ins_str("  ", NO);
 			} else {
-				cp = StrIndex(1, linebuf, curchar + 1, ' ');
+				int	n = buf_next (linebuf+curchar)-linebuf;
+				cp = StrIndex(1, linebuf, n, ' ');
 				if (cp == 0)
 					Eol();
 				else

Index: portsrv.c
===================================================================
RCS file: /local/src/CVS/jove/portsrv.c,v
retrieving revision 1.1.1.1
retrieving revision 1.2
diff -u -d -r1.1.1.1 -r1.2
--- portsrv.c	10 Mar 2001 12:23:12 -0000	1.1.1.1
+++ portsrv.c	15 May 2003 00:01:06 -0000	1.2
@@ -151,7 +151,7 @@
 	}
 }
 
-#else PIPEPROCS
+#else /* PIPEPROCS */
 main()
 {
 }

Index: proc.c
===================================================================
RCS file: /local/src/CVS/jove/proc.c,v
retrieving revision 1.1.1.1
retrieving revision 1.2
diff -u -d -r1.1.1.1 -r1.2
--- proc.c	10 Mar 2001 12:23:12 -0000	1.1.1.1
+++ proc.c	15 May 2003 00:01:06 -0000	1.2
@@ -87,7 +87,7 @@
 		else
 			kill_off(rpid, w);
 	}
-#endif IPROCS
+#endif /* IPROCS */
 }
 
 private int	holdCounts[NSIG];

Index: re1.c
===================================================================
RCS file: /local/src/CVS/jove/re1.c,v
retrieving revision 1.2
retrieving revision 1.3
diff -u -d -r1.2 -r1.3
--- re1.c	11 Jul 2001 06:42:26 -0000	1.2
+++ re1.c	15 May 2003 00:01:06 -0000	1.3
@@ -51,8 +51,9 @@
 				case BS:
 				case RUBOUT:
 				case 'N':
-					if (linebuf[offset++] == '\0')
+					if (linebuf[offset] == '\0')
 						goto nxtline;
+					offset = buf_next(linebuf+offset)-linebuf;
 					continue;
 
 				case CTL('W'):
@@ -817,12 +818,12 @@
 	char	*tagfname;
 
 	set_max_tag ();
-	if (!ismword(linebuf[curchar]))
+	if (!ismword(cur_char ()))
 		complain("[Not a tag!]");
-	while (c1 > 0 && ismword(linebuf[c1 - 1]))
-		c1--;
-	while (ismword(linebuf[c2]))
-		c2++;
+	while (c1 > 0 && ismword(parse_ucs4 (buf_prev (linebuf + c1, linebuf))))
+		c1 = buf_prev (linebuf+c1, linebuf) - linebuf;
+	while (ismword(parse_ucs4 (linebuf + c2)))
+		c2 = buf_next (linebuf+c2) - linebuf;
 
 	null_ncpy(tagname, linebuf + c1, c2 - c1);
 	find_tag(tagname);
@@ -866,8 +867,8 @@
 		return 0;
 	DOTsave(&buf);
 	if (dir == FORWARD) {
-		if (cmp_char(linebuf[curchar], c)) {
-			buf.p_char = curchar + 1;
+		if (cmp_char(cur_char (), c)) {
+			buf.p_char = buf_next (linebuf+curchar) - linebuf;
 			return &buf;
 		}
 	} else {

Index: recover.c
===================================================================
RCS file: /local/src/CVS/jove/recover.c,v
retrieving revision 1.3
retrieving revision 1.4
diff -u -d -r1.3 -r1.4
--- recover.c	11 Jul 2001 06:42:26 -0000	1.3
+++ recover.c	15 May 2003 00:01:07 -0000	1.4
@@ -550,7 +550,7 @@
 #ifdef KILL0
 	if (kill(Header.Pid, 0) == 0)
 		return 0;
-#endif KILL0
+#endif /* KILL0 */
 
 	if (Header.Nbuffers == 0) {
 		printf("There are no modified buffers in %s; should I delete the tmp file?", pntrfile);

Index: scandir.c
===================================================================
RCS file: /local/src/CVS/jove/scandir.c,v
retrieving revision 1.1.1.1
retrieving revision 1.2
diff -u -d -r1.1.1.1 -r1.2
--- scandir.c	10 Mar 2001 12:23:12 -0000	1.1.1.1
+++ scandir.c	15 May 2003 00:01:07 -0000	1.2
@@ -71,8 +71,8 @@
 	return &dir;
 }
 
-#endif hpux
-#endif BSD4_2
+#endif /* hpux */
+#endif /* BSD4_2 */
 
 /* Scandir returns the number of entries or -1 if the directory cannoot
    be opened or malloc fails. */

Index: screen.c
===================================================================
RCS file: /local/src/CVS/jove/screen.c,v
retrieving revision 1.2
retrieving revision 1.3
diff -u -d -r1.2 -r1.3
--- screen.c	11 Jul 2001 06:42:26 -0000	1.2
+++ screen.c	15 May 2003 00:01:07 -0000	1.3
@@ -146,8 +146,7 @@
 
 /* Output one character (if necessary) at the current position */
 
-dosputc(c)
-register screenchar	c;
+dosputc(screenchar c)
 {
 	if (*cursor != c) {
 #ifdef ID_CHAR
@@ -193,23 +192,21 @@
    reach the end of the screen.  Aborts if there is a character
    waiting.  */
 
-swrite(line, inversep, abortable)
-register char	*line;
-register int	abortable;
+swrite(char *line, int inversep, int abortable)
 {
-	register int	c;
+	ucs4	c;
 	int	col = i_col,
 		aborted = 0;
-	register int	n = cursend - cursor;
-	int	or_byte = inversep ? INVERSE_BIT : 0,
-		thebyte;
+	int	n = cursend - cursor;
+	ucs4	or_byte = inversep ? INVERSE_BIT : 0;
+	ucs4	thebyte;
 	int	togo = 0;
 
 	if (n <= 0)
 		return 1;
 
 	OkayAbort = 0;
-	while (c = *line++) {
+	while (c = next_ucs4 (&line)) {
 		if (abortable && OkayAbort) {
 			OkayAbort = NO;
 			if (InputPending = charp()) {
@@ -261,7 +258,7 @@
 	register int	n = cursend - cursor,
 			col = 0,
 			c;
-	register char	*bp;
+	char	*bp;
 	int	StartCol = DesiredScreen[linenum].s_offset,
 		visspace = DesiredScreen[linenum].s_window->w_flags & W_VISSPACE,
 		aborted = 0;
@@ -274,7 +271,7 @@
 			break;
 		}
 
-		c = *bp++ & CHAR_MASK;
+		c = next_ucs4 (&bp);
 		if (c == '\t')
 			col += (tabstop - (col % tabstop));
 		else if (isctrl(c))
@@ -284,7 +281,7 @@
 	}
 
 	OkayAbort = 0;
-	while (c = (*bp++ & CHAR_MASK)) {
+	while (c = next_ucs4(&bp)) {
 		if (OkayAbort) {
 			OkayAbort = NO;
 			if (InputPending = charp()) {

Index: temp.h
===================================================================
RCS file: /local/src/CVS/jove/temp.h,v
retrieving revision 1.1.1.1
retrieving revision 1.2
diff -u -d -r1.1.1.1 -r1.2
--- temp.h	10 Mar 2001 12:23:12 -0000	1.1.1.1
+++ temp.h	15 May 2003 00:01:07 -0000	1.2
@@ -42,7 +42,7 @@
 #else
 #   define CH_BITS		0
 #   define MAX_BLOCKS		(1 << 20)	/* basically unlimited */
-#endif SMALL
+#endif /* SMALL */
 
 #if LBSIZE == 512
 #    define BNO_SHIFT		(9 - CH_BITS)

Index: term.c
===================================================================
RCS file: /local/src/CVS/jove/term.c,v
retrieving revision 1.3
retrieving revision 1.4
diff -u -d -r1.3 -r1.4
--- term.c	11 Jul 2001 06:42:26 -0000	1.3
+++ term.c	15 May 2003 00:01:07 -0000	1.4
@@ -78,7 +78,7 @@
 char	PC;
 #else
 extern char	PC;
-#endif SYSV SYSVR2
+#endif /* SYSV SYSVR2 */
 
 static char	tspace[256];
 
@@ -245,7 +245,9 @@
 	/*
 	 * initialize for insert/delete char
 	 */
+#ifdef ID_CHAR
 	disp_opt_init ();
+#endif
 	return 1;
 }
 

Index: tune.h
===================================================================
RCS file: /local/src/CVS/jove/tune.h,v
retrieving revision 1.3
retrieving revision 1.4
diff -u -d -r1.3 -r1.4
--- tune.h	24 Aug 2001 09:02:38 -0000	1.3
+++ tune.h	15 May 2003 00:01:07 -0000	1.4
@@ -93,7 +93,7 @@
 #   endif
 #
 #   define SUBPROCS	/* only on UNIX systems (NOT INCORPORATED YET) */
-#endif MSDOS
+#endif /* MSDOS */
 
 #ifndef NO_C
 #ifdef SMALL
@@ -112,7 +112,7 @@
 #       define CHDIR		/* cd command and absolute pathnames */
 #	define	KILL0	/* kill(pid, 0) returns 0 if proc exists */
 #       define SPELL		/* spell words and buffer commands */
-#       define ID_CHAR		/* include code to IDchar */
+/* #       define ID_CHAR		/* include code to IDchar */
 #       define WIRED_TERMS	/* include code for wired terminals */
 #       define ANSICODES	/* extra commands that process ANSI codes */
 #	ifdef BSD4_2
@@ -121,7 +121,7 @@
 #   endif
 #   define LISP			/* include the code for Lisp Mode */
 #   define CMT_FMT		/* include the comment formatting routines */
-#endif SMALL
+#endif /* SMALL */
 
 #if !sun
 /* #   define MY_MALLOC	/* use more memory efficient malloc (not on suns) */
@@ -172,4 +172,4 @@
 	TmpFilePath[V_STRING_LEN],/* directory/device to store tmp files */
 	Shell[V_STRING_LEN],	/* shell to use */
 	ShFlags[V_STRING_LEN];
-#endif NOEXTERNS
+#endif /* NOEXTERNS */

Index: util.c
===================================================================
RCS file: /local/src/CVS/jove/util.c,v
retrieving revision 1.2
retrieving revision 1.3
diff -u -d -r1.2 -r1.3
--- util.c	11 Jul 2001 06:42:26 -0000	1.2
+++ util.c	15 May 2003 00:01:07 -0000	1.3
@@ -328,11 +328,11 @@
 to_word(dir)
 register int	dir;
 {
-	register char	c;
+	ucs4	c;
 
 	if (dir == FORWARD) {
-		while ((c = linebuf[curchar]) != 0 && !isword(c))
-			curchar++;
+		while ((c = cur_char ()) != 0 && !isword(c))
+			right_char ();
 		if (eolp()) {
 			if (curline->l_next == 0)
 				return;
@@ -341,8 +341,8 @@
 			return;
 		}
 	} else {
-		while (!bolp() && (c = linebuf[curchar - 1], !isword(c)))
-			--curchar;
+		while (!bolp() && !isword(prev_char()))
+			left_char ();
 		if (bolp()) {
 			if (curline->l_prev == 0)
 				return;
@@ -754,7 +754,7 @@
 		return 4.0;
 }
 
-# else  LOAD_AV_PROG
+# else /*  LOAD_AV_PROG */
 #  ifdef PROC_FILESYS
 #define LOAD_FILE "/proc/loadavg"
 
@@ -854,10 +854,10 @@
 }
 
 #  endif BSD4_2
-# endif PROC_FILESYS
-# endif LOAD_AV_PROG
-#endif LAVD
-#endif LOAD_AV
+# endif /* PROC_FILESYS */
+# endif /* LOAD_AV_PROG */
+#endif /* LAVD */
+#endif /* LOAD_AV */
 
 /* get the time buf, designated by *timep, from FROM to TO. */
 char *

Index: vars.c
===================================================================
RCS file: /local/src/CVS/jove/vars.c,v
retrieving revision 1.1.1.1
retrieving revision 1.2
diff -u -d -r1.1.1.1 -r1.2
--- vars.c	10 Mar 2001 12:23:12 -0000	1.1.1.1
+++ vars.c	15 May 2003 00:01:07 -0000	1.2
@@ -85,6 +85,7 @@
 #ifdef ID_CHAR
 	VARIABLE, "use-i/d-char", &UseIC, V_BOOL,
 #endif
+	VARIABLE, "utf8-output", &Utf8Out, V_BOOL|V_CLRSCREEN,
 	VARIABLE, "visible-bell", &VisBell, V_BOOL,
 	VARIABLE, "wrap-search", &WrapScan, V_BOOL,
 	VARIABLE, "write-files-on-make", &WtOnMk, V_BOOL,



