--- gforth/utf-8.fs 2007/06/30 23:00:14 1.27 +++ gforth/utf-8.fs 2007/09/15 20:11:44 1.29 @@ -94,15 +94,10 @@ Defer check-xy ' noop IS check-xy \ utf-8 stuff for xchars -: u8string+ ( xcaddr u -- xcaddr u' ) - over + u8>> over - ; -: u8string- ( xcaddr u -- xcaddr u' ) - over + u8<< over - ; - -: +u8string ( xc-addr1 u1 -- xc-addr2 u2 ) +: +u8/string ( xc-addr1 u1 -- xc-addr2 u2 ) over dup u8>> swap - /string ; -: -u8string ( xc-addr1 u1 -- xc-addr2 u2 ) - over dup u8<< swap - /string ; +: u8\string- ( xcaddr u -- xcaddr u' ) + over + u8<< over - ; : u8@ ( c-addr -- u ) u8@+ nip ; @@ -115,7 +110,7 @@ Defer check-xy ' noop IS check-xy >r u8!+ r> r> swap - true then ; -: u8addrlen ( u8-addr -- u ) +: u8addrlen ( u8-addr u -- u ) drop \ length of UTF-8 char starting at u8-addr (accesses only u8-addr) c@ dup $80 u< if drop 1 exit endif @@ -129,7 +124,7 @@ Defer check-xy ' noop IS check-xy : -u8trailing-garbage ( addr u1 -- addr u2 ) 2dup + dup u8<< ( addr u1 end1 end2 ) - 2dup dup u8addrlen + = if \ last character ok + 2dup dup over over - u8addrlen + = if \ last character ok 2drop else nip nip over - @@ -295,10 +290,12 @@ here wc-table - Constant #wc-table ['] u8>> is xchar+ ['] u8<< is xchar- [ [IFDEF] xstring+ ] - ['] u8string+ is xstring+ - ['] u8string- is xstring- - ['] +u8string is +xstring - ['] -u8string is -xstring + ['] u8\string- is xstring- + ['] +u8/string is +xstring +[ [THEN] ] +[ [IFDEF] x/string+ ] + ['] u8\string- is x\string- + ['] +u8/string is +x/string [ [THEN] ] ['] u8@ is xc@ ['] u8!+? is xc!+? @@ -307,6 +304,9 @@ here wc-table - Constant #wc-table [ [IFDEF] x-width ] ['] u8width is x-width [ [THEN] ] +[ [IFDEF] x-size ] + ['] u8addrlen is x-size +[ [THEN] ] ['] -u8trailing-garbage is -trailing-garbage ;