--- gforth/utf-8.fs 2004/12/12 20:14:07 1.1 +++ gforth/utf-8.fs 2004/12/28 21:09:47 1.7 @@ -20,7 +20,9 @@ \ short: u8 means utf-8 encoded address -: u8len ( u -- n ) +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 BEGIN 2dup u>= WHILE 5 lshift r> 1+ >r REPEAT @@ -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 ; @@ -57,7 +59,7 @@ $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 ; @@ -71,39 +73,90 @@ \ 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> ; +: ( max span addr pos1 u8char -- max span addr pos2 ) + >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 ) - >r >string over r@ u8len + swap move 2dup chars + r@ swap u8!+ drop - r> u8len >r rot r@ chars + -rot r> chars + .all .rest ; + .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 ; + dup IF over + u8<< over - 0 max .all .rest + 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 - rot r> - -rot .all 2 spaces .rest ; + rot r> - -rot ; : ?u8del ( max span addr pos1 -- max span addr pos2 0 ) - dup IF (u8del) THEN 0 ; + dup IF (u8del) .all 2 spaces .rest THEN 0 ; : ( max span addr pos1 -- max span addr pos2 0 ) 2 pick over <> - IF u8forw drop (u8del) ELSE #bell emit THEN 0 ; + IF u8forw drop (u8del) .all 2 spaces .rest + ELSE bell THEN 0 ; : u8eof 2 pick over or 0= IF bye ELSE THEN ; -' u8forw ctrl F bindkey -' u8back ctrl B bindkey -' ?u8del ctrl H bindkey -' u8eof ctrl D bindkey -' ctrl X bindkey -' (u8ins) IS insert-char -' noop IS everychar -' save-cursor IS everyline -' u8key IS key -' u8emit IS emit +: u8first-pos ( max span addr pos1 -- max span addr 0 0 ) + drop 0 .all .rest 0 ; +: u8end-pos ( max span addr pos1 -- max span addr span 0 ) + drop over .all 0 ; + + +: u8clear-line ( max span addr pos1 -- max addr ) + drop restore-cursor swap spaces restore-cursor ; +: u8clear-tib ( max span addr pos -- max 0 addr 0 false ) + u8clear-line 0 tuck dup ; + +: (u8enter) ( max span addr pos1 -- max span addr pos2 true ) + >r end^ 2@ hist-setpos + 2dup swap history write-line drop ( throw ) \ don't worry about errors + hist-pos 2dup backward^ 2! end^ 2! + r> .all space true ; + +: u8kill-expand ( max span addr pos1 -- max span addr pos2 ) + prefix-found cell+ @ ?dup IF >r + r@ - >string over r@ + -rot move + rot r@ - -rot .all r> spaces .rest THEN ; + +: insert ( string length buffer size -- ) + rot over min >r r@ - ( left over ) + over dup r@ + rot move r> move ; + +: 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 prefix-off + ELSE dup >r + 2>r >string r@ + 2r> 2swap insert + r@ + rot r> + -rot + THEN + prefix-found @ IF bl (u8ins) ELSE .all .rest THEN 0 ; + +: utf-8-io ( -- ) + ['] u8forw ctrl F bindkey + ['] u8back ctrl B bindkey + ['] ?u8del ctrl H bindkey + ['] u8eof ctrl D bindkey + ['] ctrl X bindkey + ['] u8clear-tib ctrl K bindkey + ['] u8first-pos ctrl A bindkey + ['] u8end-pos ctrl E bindkey + ['] (u8enter) #lf bindkey + ['] (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 ; +