--- gforth/utf-8.fs	2004/12/24 14:08:43	1.4
+++ gforth/utf-8.fs	2005/01/04 22:22:08	1.11
@@ -20,49 +20,66 @@
 
 \ short: u8 means utf-8 encoded address
 
+s" malformed UTF-8 character" exception Constant UTF-8-err
+
+$80 Value maxascii
+
 : u8len ( u8 -- n )
-    dup      $80 u< IF  drop 1  EXIT  THEN \ special case ASCII
+    dup      maxascii u< IF  drop 1  EXIT  THEN \ special case ASCII
     $800  2 >r
     BEGIN  2dup u>=  WHILE  5 lshift r> 1+ >r  REPEAT
     2drop r> ;
 
 : u8@+ ( u8addr -- u8addr' u )
-    count  dup $80 and 0= ?EXIT  \ special case ASCII
+    count  dup maxascii u< ?EXIT  \ special case ASCII
     $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 ;
 
 : u8!+ ( u u8addr -- u8addr' )
-    over $80 < IF  tuck c! 1+  EXIT  THEN \ special case ASCII
+    over maxascii u< IF  tuck c! 1+  EXIT  THEN \ special case ASCII
     >r 0 swap  $3F
     BEGIN  2dup u>  WHILE
 	    2/ >r  dup $3F and $80 or swap 6 rshift r>
     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
+
+[ifundef] char@ \ !! bootstrapping help
+    Defer char@ ( addr u -- char addr' u' )
+    :noname  over c@ -rot 1 /string ; IS char@
+[then]
+
+:noname  ( addr u -- char addr' u' )
+    \ !! the if here seems to work around some breakage, but not
+    \ entirely; e.g., try 'ç' with LANG=C.
+    dup 1 u<= IF defers char@ EXIT THEN
+    over + >r u8@+ swap r> over - ; IS char@
+
 \ scan to next/previous character
 
-: u8>> ( u8addr -- u8addr' )
-    BEGIN  count $C0 and $80 <>  UNTIL ;
+: u8>> ( u8addr -- u8addr' )  $80 >r
+    BEGIN  count r> and maxascii <> $C0 >r  UNTIL  rdrop ;
 : u8<< ( u8addr -- u8addr' )
-    BEGIN  1- dup c@ $C0 and $80 <>  UNTIL ;
+    BEGIN  1- dup c@ $C0 and maxascii <>  UNTIL ;
 
 \ utf key and emit
 
 : u8key ( -- u )
-    defers key dup $80 and 0= ?EXIT  \ special case ASCII
+    defers key dup maxascii u< ?EXIT  \ special case ASCII
     $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 ;
 
 : u8emit ( u -- )
-    dup $80 < IF  defers emit  EXIT  THEN \ special case ASCII
+    dup maxascii u< IF  defers emit  EXIT  THEN \ special case ASCII
     0 swap  $3F
     BEGIN  2dup u>  WHILE
 	    2/ >r  dup $3F and $80 or swap 6 rshift r>
@@ -71,23 +88,26 @@
 
 \ 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 +117,7 @@
 : <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 )
@@ -129,13 +149,13 @@
 : u8tab-expand ( max span addr pos1 -- max span addr pos2 0 )
     key? IF  #tab (u8ins) 0  EXIT  THEN
     u8kill-expand 2dup extract-word dup 0= IF  nip EXIT  THEN
-    search-prefix  tib-full?
-    IF    7 emit  2drop  0 0 prefix-found 2!
+    search-prefix tib-full?
+    IF    7 emit  2drop  prefix-off
     ELSE  dup >r
 	2>r >string r@ + 2r> 2swap insert
 	r@ + rot r> + -rot
     THEN
-    prefix-found @ IF  bl (u8ins)  THEN  0 ;
+    prefix-found @ IF  bl (u8ins)  ELSE  .all .rest  THEN  0 ;
 
 : utf-8-io ( -- )
     ['] u8forw       ctrl F bindkey
@@ -150,6 +170,7 @@
     ['] (u8enter)    #cr    bindkey
     ['] u8tab-expand #tab   bindkey
     ['] (u8ins)      IS insert-char
+    ['] kill-prefix  IS everychar
     ['] save-cursor  IS everyline
     ['] u8key        IS key
     ['] u8emit       IS emit ;