--- gforth/utf-8.fs	2004/12/27 20:05:53	1.5
+++ gforth/utf-8.fs	2004/12/29 14:15:12	1.8
@@ -20,6 +20,8 @@
 
 \ short: u8 means utf-8 encoded address
 
+s" malformed UTF-8 character" exception Constant UTF-8-err
+
 : u8len ( u8 -- n )
     dup      $80 u< IF  drop 1  EXIT  THEN \ special case ASCII
     $800  2 >r
@@ -31,7 +33,7 @@
     $7F and  $40 >r
     BEGIN  dup r@ and  WHILE  r@ xor
 	    6 lshift r> 5 lshift >r >r count
-\	    dup $C0 and $80 <> abort" malformed character"
+	    dup $C0 and $80 <> IF   UTF-8-err throw  THEN
 	    $3F and r> or
     REPEAT  rdrop ;
 
@@ -43,6 +45,12 @@
     REPEAT  $7F xor 2* or  r>
     BEGIN   over $80 u>= WHILE  tuck c! 1+  REPEAT  nip ;
 
+\ plug-in so that char and '<char> work for UTF-8
+
+:noname  ( addr u -- char addr' u' )
+    dup 1 u<= IF  defers char@  EXIT  THEN
+    over + >r u8@+ swap r> over - ; IS char@
+
 \ scan to next/previous character
 
 : u8>> ( u8addr -- u8addr' )
@@ -57,7 +65,7 @@
     $7F and  $40 >r
     BEGIN  dup r@ and  WHILE  r@ xor
 	    6 lshift r> 5 lshift >r >r defers key
-\	    dup $C0 and $80 <> abort" malformed character"
+	    dup $C0 and $80 <> IF  UTF-8-err throw  THEN
 	    $3F and r> or
     REPEAT  rdrop ;
 
@@ -71,23 +79,26 @@
 
 \ input editor
 
-: save-cursor ( -- )  27 emit '7 emit ;
-: restore-cursor ( -- )  27 emit '8 emit ;
+[IFUNDEF] #esc  27 Constant #esc  [THEN]
+
+: save-cursor ( -- )  #esc emit '7 emit ;
+: restore-cursor ( -- )  #esc emit '8 emit ;
 : .rest ( addr pos1 -- addr pos1 )
     restore-cursor 2dup type ;
 : .all ( span addr pos1 -- span addr pos1 )
     restore-cursor >r 2dup swap type r> ;
 
 : <u8ins>  ( max span addr pos1 u8char -- max span addr pos2 )
-    >r >string over r@ u8len + swap move 2dup chars + r@ swap u8!+ drop
+    >r  2over r@ u8len + u< IF  rdrop bell  EXIT  THEN
+    >string over r@ u8len + swap move 2dup chars + r@ swap u8!+ drop
     r> u8len >r  rot r@ chars + -rot r> chars + ;
 : (u8ins)  ( max span addr pos1 u8char -- max span addr pos2 )
     <u8ins> .all .rest ;
 : u8back  ( max span addr pos1 -- max span addr pos2 f )
     dup  IF  over + u8<< over -  0 max .all .rest
-    ELSE  #bell emit  THEN 0 ;
+    ELSE  bell  THEN 0 ;
 : u8forw  ( max span addr pos1 -- max span addr pos2 f )
-    2 pick over <> IF  over + u8@+ u8emit over -  ELSE  #bell emit  THEN 0 ;
+    2 pick over <> IF  over + u8@+ u8emit over -  ELSE  bell  THEN 0 ;
 : (u8del)  ( max span addr pos1 -- max span addr pos2 )
     over + dup u8<< tuck - >r over -
     >string over r@ + -rot move
@@ -97,7 +108,7 @@
 : <u8del> ( max span addr pos1 -- max span addr pos2 0 )
   2 pick over <>
     IF  u8forw drop (u8del) .all 2 spaces .rest
-    ELSE  #bell emit  THEN  0 ;
+    ELSE  bell  THEN  0 ;
 : u8eof  2 pick over or 0=  IF  bye  ELSE  <u8del>  THEN ;
 
 : u8first-pos  ( max span addr pos1 -- max span addr 0 0 )