Diff for /gforth/utf-8.fs between versions 1.1 and 1.22

version 1.1, 2004/12/12 20:14:07 version 1.22, 2006/02/20 08:22:28
Line 1 Line 1
 \ UTF-8 handling                                       12dec04py  \ UTF-8 handling                                       12dec04py
   
 \ Copyright (C) 2004 Free Software Foundation, Inc.  \ Copyright (C) 2004,2005 Free Software Foundation, Inc.
   
 \ This file is part of Gforth.  \ This file is part of Gforth.
   
Line 20 Line 20
   
 \ short: u8 means utf-8 encoded address  \ short: u8 means utf-8 encoded address
   
 : u8len ( u -- n )  s" malformed UTF-8 character" exception Constant UTF-8-err
     dup      $80 u< IF  drop 1  EXIT  THEN \ special case ASCII  
   $80 Value max-single-byte
   
   : u8len ( u8 -- n )
       dup      max-single-byte u< IF  drop 1  EXIT  THEN \ special case ASCII
     $800  2 >r      $800  2 >r
     BEGIN  2dup u>=  WHILE  5 lshift r> 1+ >r  REPEAT      BEGIN  2dup u>=  WHILE  5 lshift r> 1+ >r  dup 0= UNTIL  THEN
     2drop r> ;      2drop r> ;
   
 : u8@+ ( u8addr -- u8addr' u )  : u8@+ ( u8addr -- u8addr' u )
     count  dup $80 and 0= ?EXIT  \ special case ASCII      count  dup max-single-byte u< ?EXIT  \ special case ASCII
       dup $C2 u< IF  UTF-8-err throw  THEN  \ malformed character
     $7F and  $40 >r      $7F and  $40 >r
     BEGIN  dup r@ and  WHILE  r@ xor      BEGIN  dup r@ and  WHILE  r@ xor
             6 lshift r> 5 lshift >r >r count              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              $3F and r> or
     REPEAT  rdrop ;      REPEAT  rdrop ;
   
 : u8!+ ( u u8addr -- u8addr' )  : u8!+ ( u u8addr -- u8addr' )
     over $80 < IF  tuck c! 1+  EXIT  THEN \ special case ASCII      over max-single-byte u< IF  tuck c! 1+  EXIT  THEN \ special case ASCII
     >r 0 swap  $3F      >r 0 swap  $3F
     BEGIN  2dup u>  WHILE      BEGIN  2dup u>  WHILE
             2/ >r  dup $3F and $80 or swap 6 rshift r>              2/ >r  dup $3F and $80 or swap 6 rshift r>
     REPEAT  $7F xor 2* or  r>      REPEAT  $7F xor 2* or  r>
     BEGIN   over $80 u>= WHILE  tuck c! 1+  REPEAT  nip ;      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  \ scan to next/previous character
   
 : u8>> ( u8addr -- u8addr' )  \ alternative names: u8char+ u8char-
     BEGIN  count $C0 and $80 <>  UNTIL ;  
   : u8>> ( u8addr -- u8addr' )  u8@+ drop ;
 : u8<< ( u8addr -- u8addr' )  : u8<< ( u8addr -- u8addr' )
     BEGIN  1- dup c@ $C0 and $80 <>  UNTIL ;      BEGIN  1- dup c@ $C0 and max-single-byte <>  UNTIL ;
   
 \ utf key and emit  \ utf key and emit
   
 : u8key ( -- u )  : u8key ( -- u )
     defers key dup $80 and 0= ?EXIT  \ special case ASCII      defers key dup max-single-byte u< ?EXIT  \ special case ASCII
       dup $C2 u< IF  UTF-8-err throw  THEN  \ malformed character
     $7F and  $40 >r      $7F and  $40 >r
     BEGIN  dup r@ and  WHILE  r@ xor      BEGIN  dup r@ and  WHILE  r@ xor
             6 lshift r> 5 lshift >r >r defers key              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              $3F and r> or
     REPEAT  rdrop ;      REPEAT  rdrop ;
   
 : u8emit ( u -- )  : u8emit ( u -- )
     dup $80 < IF  defers emit  EXIT  THEN \ special case ASCII      dup max-single-byte u< IF  defers emit  EXIT  THEN \ special case ASCII
     0 swap  $3F      0 swap  $3F
     BEGIN  2dup u>  WHILE      BEGIN  2dup u>  WHILE
             2/ >r  dup $3F and $80 or swap 6 rshift r>              2/ >r  dup $3F and $80 or swap 6 rshift r>
     REPEAT  $7F xor 2* or      REPEAT  $7F xor 2* or
     BEGIN   dup $80 u>= WHILE  defers emit  REPEAT  drop ;      BEGIN   dup $80 u>= WHILE  defers emit  REPEAT  drop ;
   
 \ input editor  \ utf-8 stuff for xchars
   
   : +u8/string ( c-addr1 u1 -- c-addr2 u2 )
       over dup u8>> swap - /string ;
   
   : -u8/string ( c-addr1 u1 -- c-addr2 u2 )
       over dup u8<< swap - /string ;
   
   : u8@ ( c-addr -- u )
       u8@+ nip ;
   
   : u8!+? ( xc xc-addr1 u1 -- xc-addr2 u2 f )
       >r over u8len r@ over u< if ( xc xc-addr1 len r: u1 )
           \ not enough space
           drop nip r> false
       else
           >r u8!+ r> r> swap - true
       then ;
   
   : u8addrlen ( u8-addr -- u )
       \ length of UTF-8 char starting at u8-addr (accesses only u8-addr)
       c@
       dup $80 u< if drop 1 exit endif
       dup $c0 u< if UTF-8-err throw endif
       dup $e0 u< if drop 2 exit endif
       dup $f0 u< if drop 3 exit endif
       dup $f8 u< if drop 4 exit endif
       dup $fc u< if drop 5 exit endif
       dup $fe u< if drop 6 exit endif
       UTF-8-err throw ;
   
   : -u8trailing-garbage ( addr u1 -- addr u2 )
       2dup + dup u8<< ( addr u1 end1 end2 )
       2dup dup u8addrlen + = if \ last character ok
           2drop
       else
           nip nip over -
       then ;
   
   [IFUNDEF] wcwidth
       : wcwidth abort ;
   [THEN]
       
   : u8width ( xcaddr u -- n )
       0 rot rot over + swap ?DO
           I xc@+ swap >r wcwidth +
       r> I - +LOOP ;
   
   : set-encoding-utf-8 ( -- )
       ['] u8emit is xemit
       ['] u8key is xkey
       ['] u8>> is xchar+
       ['] u8<< is xchar-
       ['] +u8/string is +x/string
       ['] -u8/string is -x/string
       ['] u8@ is xc@
       ['] u8!+? is xc!+?
       ['] u8@+ is xc@+
       ['] u8len is xc-size
   [ [IFDEF] x-width ]
       ['] u8width is x-width
   [ [THEN] ]
       ['] -u8trailing-garbage is -trailing-garbage
   ;
   
   : utf-8-cold ( -- )
       s" LC_ALL" getenv 2dup d0= IF  2drop
           s" LC_CTYPE" getenv 2dup d0= IF  2drop
               s" LANG" getenv 2dup d0= IF  2drop
                   s" C"  THEN THEN THEN
       s" UTF-8" search nip nip
       IF  set-encoding-utf-8  ELSE  set-encoding-fixed-width  THEN ;
   
   ' utf-8-cold INIT8 chained
   
 : save-cursor ( -- )  27 emit '7 emit ;  utf-8-cold
 : restore-cursor ( -- )  27 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> u8len >r  rot r@ chars + -rot r> chars + .all .rest ;  
 : u8back  ( max span addr pos1 -- max span addr pos2 f )  
     dup  IF  over + u8<< over -  0 max  
     ELSE  #bell emit  THEN .rest 0 ;  
 : u8forw  ( max span addr pos1 -- max span addr pos2 f )  
     2 pick over <> IF  over + u8@+ u8emit over -  ELSE  #bell emit  THEN 0 ;  
 : (u8del)  ( max span addr pos1 -- max span addr pos2 )  
     over + dup u8<< tuck - >r over -  
     >string over r@ + -rot move  
     rot r> - -rot .all 2 spaces .rest ;  
 : ?u8del ( max span addr pos1 -- max span addr pos2 0 )  
   dup  IF  (u8del)  THEN  0 ;  
 : <u8del> ( max span addr pos1 -- max span addr pos2 0 )  
   2 pick over <>  
         IF  u8forw drop (u8del)  ELSE  #bell emit  THEN  0 ;  
 : u8eof  2 pick over or 0=  IF  bye  ELSE  <u8del>  THEN ;  
   
 ' u8forw  ctrl F bindkey  
 ' u8back  ctrl B bindkey  
 ' ?u8del  ctrl H bindkey  
 ' u8eof   ctrl D bindkey  
 ' <u8del> ctrl X bindkey  
 ' (u8ins) IS insert-char  
 ' noop IS everychar  
 ' save-cursor IS everyline  
 ' u8key IS key  
 ' u8emit IS emit  

Removed from v.1.1  
changed lines
  Added in v.1.22


FreeBSD-CVSweb <freebsd-cvsweb@FreeBSD.org>