File:  [gforth] / gforth / utf-8.fs
Revision 1.32: download - view: text, annotated - select for diffs
Wed Oct 17 16:05:22 2007 UTC (16 years, 5 months ago) by pazsan
Branches: MAIN
CVS tags: HEAD
Better xhold (no buffer)
Fixed +x/string deferred words

    1: \ UTF-8 handling                                       12dec04py
    2: 
    3: \ Copyright (C) 2004,2005,2006 Free Software Foundation, Inc.
    4: 
    5: \ This file is part of Gforth.
    6: 
    7: \ Gforth is free software; you can redistribute it and/or
    8: \ modify it under the terms of the GNU General Public License
    9: \ as published by the Free Software Foundation; either version 2
   10: \ of the License, or (at your option) any later version.
   11: 
   12: \ This program is distributed in the hope that it will be useful,
   13: \ but WITHOUT ANY WARRANTY; without even the implied warranty of
   14: \ MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
   15: \ GNU General Public License for more details.
   16: 
   17: \ You should have received a copy of the GNU General Public License
   18: \ along with this program; if not, write to the Free Software
   19: \ Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111, USA.
   20: 
   21: \ short: u8 means utf-8 encoded address
   22: 
   23: s" malformed UTF-8 character" exception Constant UTF-8-err
   24: 
   25: $80 Value max-single-byte
   26: 
   27: : u8len ( u8 -- n )
   28:     dup      max-single-byte u< IF  drop 1  EXIT  THEN \ special case ASCII
   29:     $800  2 >r
   30:     BEGIN  2dup u>=  WHILE  5 lshift r> 1+ >r  dup 0= UNTIL  THEN
   31:     2drop r> ;
   32: 
   33: : u8@+ ( u8addr -- u8addr' u )
   34:     count  dup max-single-byte u< ?EXIT  \ special case ASCII
   35:     dup $C2 u< IF  UTF-8-err throw  THEN  \ malformed character
   36:     $7F and  $40 >r
   37:     BEGIN  dup r@ and  WHILE  r@ xor
   38: 	    6 lshift r> 5 lshift >r >r count
   39: 	    dup $C0 and $80 <> IF   UTF-8-err throw  THEN
   40: 	    $3F and r> or
   41:     REPEAT  rdrop ;
   42: 
   43: : u8!+ ( u u8addr -- u8addr' )
   44:     over max-single-byte u< IF  tuck c! 1+  EXIT  THEN \ special case ASCII
   45:     >r 0 swap  $3F
   46:     BEGIN  2dup u>  WHILE
   47: 	    2/ >r  dup $3F and $80 or swap 6 rshift r>
   48:     REPEAT  $7F xor 2* or  r>
   49:     BEGIN   over $80 u>= WHILE  tuck c! 1+  REPEAT  nip ;
   50: 
   51: \ plug-in so that char and '<char> work for UTF-8
   52: 
   53: [ifundef] char@ \ !! bootstrapping help
   54:     Defer char@ ( addr u -- char addr' u' )
   55:     :noname  over c@ -rot 1 /string ; IS char@
   56: [then]
   57: 
   58: :noname  ( addr u -- char addr' u' )
   59:     \ !! the if here seems to work around some breakage, but not
   60:     \ entirely; e.g., try 'รง' with LANG=C.
   61:     dup 1 u<= IF defers char@ EXIT THEN
   62:     over + >r u8@+ swap r> over - ; IS char@
   63: 
   64: \ scan to next/previous character
   65: 
   66: \ alternative names: u8char+ u8char-
   67: 
   68: : u8>> ( u8addr -- u8addr' )  u8@+ drop ;
   69: : u8<< ( u8addr -- u8addr' )
   70:     BEGIN  1- dup c@ $C0 and max-single-byte <>  UNTIL ;
   71: 
   72: \ utf key and emit
   73: 
   74: Defer check-xy  ' noop IS check-xy
   75: 
   76: : u8key ( -- u )
   77:     defers key dup max-single-byte u< ?EXIT  \ special case ASCII
   78:     dup $FF = ?EXIT  \ special resize character
   79:     dup $C2 u< IF  UTF-8-err throw  THEN  \ malformed character
   80:     $7F and  $40 >r
   81:     BEGIN  dup r@ and  WHILE  r@ xor
   82: 	    6 lshift r> 5 lshift >r >r defers key
   83: 	    dup $C0 and $80 <> IF  UTF-8-err throw  THEN
   84: 	    $3F and r> or
   85:     REPEAT  rdrop ;
   86: 
   87: : u8emit ( u -- )
   88:     dup max-single-byte u< IF  defers emit  EXIT  THEN \ special case ASCII
   89:     0 swap  $3F
   90:     BEGIN  2dup u>  WHILE
   91: 	    2/ >r  dup $3F and $80 or swap 6 rshift r>
   92:     REPEAT  $7F xor 2* or
   93:     BEGIN   dup $80 u>= WHILE  defers emit  REPEAT  drop ;
   94: 
   95: \ utf-8 stuff for xchars
   96: 
   97: : +u8/string ( xc-addr1 u1 -- xc-addr2 u2 )
   98:     over dup u8>> swap - /string ;
   99: : u8\string- ( xcaddr u -- xcaddr u' )
  100:     over + u8<< over - ;
  101: 
  102: : u8@ ( c-addr -- u )
  103:     u8@+ nip ;
  104: 
  105: : u8!+? ( xc xc-addr1 u1 -- xc-addr2 u2 f )
  106:     >r over u8len r@ over u< if ( xc xc-addr1 len r: u1 )
  107: 	\ not enough space
  108: 	drop nip r> false
  109:     else
  110: 	>r u8!+ r> r> swap - true
  111:     then ;
  112: 
  113: : u8addrlen ( u8-addr u -- u )  drop
  114:     \ length of UTF-8 char starting at u8-addr (accesses only u8-addr)
  115:     c@
  116:     dup $80 u< if drop 1 exit endif
  117:     dup $c0 u< if UTF-8-err throw endif
  118:     dup $e0 u< if drop 2 exit endif
  119:     dup $f0 u< if drop 3 exit endif
  120:     dup $f8 u< if drop 4 exit endif
  121:     dup $fc u< if drop 5 exit endif
  122:     dup $fe u< if drop 6 exit endif
  123:     UTF-8-err throw ;
  124: 
  125: : -u8trailing-garbage ( addr u1 -- addr u2 )
  126:     2dup + dup u8<< ( addr u1 end1 end2 )
  127:     2dup dup over over - u8addrlen + = if \ last character ok
  128: 	2drop
  129:     else
  130: 	nip nip over -
  131:     then ;
  132: 
  133: [IFUNDEF] wcwidth
  134: : wc,3 ( n low high -- )  1+ , , , ;
  135: 
  136: Create wc-table \ derived from wcwidth source code, for UCS32
  137: 0 0300 0357 wc,3
  138: 0 035D 036F wc,3
  139: 0 0483 0486 wc,3
  140: 0 0488 0489 wc,3
  141: 0 0591 05A1 wc,3
  142: 0 05A3 05B9 wc,3
  143: 0 05BB 05BD wc,3
  144: 0 05BF 05BF wc,3
  145: 0 05C1 05C2 wc,3
  146: 0 05C4 05C4 wc,3
  147: 0 0600 0603 wc,3
  148: 0 0610 0615 wc,3
  149: 0 064B 0658 wc,3
  150: 0 0670 0670 wc,3
  151: 0 06D6 06E4 wc,3
  152: 0 06E7 06E8 wc,3
  153: 0 06EA 06ED wc,3
  154: 0 070F 070F wc,3
  155: 0 0711 0711 wc,3
  156: 0 0730 074A wc,3
  157: 0 07A6 07B0 wc,3
  158: 0 0901 0902 wc,3
  159: 0 093C 093C wc,3
  160: 0 0941 0948 wc,3
  161: 0 094D 094D wc,3
  162: 0 0951 0954 wc,3
  163: 0 0962 0963 wc,3
  164: 0 0981 0981 wc,3
  165: 0 09BC 09BC wc,3
  166: 0 09C1 09C4 wc,3
  167: 0 09CD 09CD wc,3
  168: 0 09E2 09E3 wc,3
  169: 0 0A01 0A02 wc,3
  170: 0 0A3C 0A3C wc,3
  171: 0 0A41 0A42 wc,3
  172: 0 0A47 0A48 wc,3
  173: 0 0A4B 0A4D wc,3
  174: 0 0A70 0A71 wc,3
  175: 0 0A81 0A82 wc,3
  176: 0 0ABC 0ABC wc,3
  177: 0 0AC1 0AC5 wc,3
  178: 0 0AC7 0AC8 wc,3
  179: 0 0ACD 0ACD wc,3
  180: 0 0AE2 0AE3 wc,3
  181: 0 0B01 0B01 wc,3
  182: 0 0B3C 0B3C wc,3
  183: 0 0B3F 0B3F wc,3
  184: 0 0B41 0B43 wc,3
  185: 0 0B4D 0B4D wc,3
  186: 0 0B56 0B56 wc,3
  187: 0 0B82 0B82 wc,3
  188: 0 0BC0 0BC0 wc,3
  189: 0 0BCD 0BCD wc,3
  190: 0 0C3E 0C40 wc,3
  191: 0 0C46 0C48 wc,3
  192: 0 0C4A 0C4D wc,3
  193: 0 0C55 0C56 wc,3
  194: 0 0CBC 0CBC wc,3
  195: 0 0CBF 0CBF wc,3
  196: 0 0CC6 0CC6 wc,3
  197: 0 0CCC 0CCD wc,3
  198: 0 0D41 0D43 wc,3
  199: 0 0D4D 0D4D wc,3
  200: 0 0DCA 0DCA wc,3
  201: 0 0DD2 0DD4 wc,3
  202: 0 0DD6 0DD6 wc,3
  203: 0 0E31 0E31 wc,3
  204: 0 0E34 0E3A wc,3
  205: 0 0E47 0E4E wc,3
  206: 0 0EB1 0EB1 wc,3
  207: 0 0EB4 0EB9 wc,3
  208: 0 0EBB 0EBC wc,3
  209: 0 0EC8 0ECD wc,3
  210: 0 0F18 0F19 wc,3
  211: 0 0F35 0F35 wc,3
  212: 0 0F37 0F37 wc,3
  213: 0 0F39 0F39 wc,3
  214: 0 0F71 0F7E wc,3
  215: 0 0F80 0F84 wc,3
  216: 0 0F86 0F87 wc,3
  217: 0 0F90 0F97 wc,3
  218: 0 0F99 0FBC wc,3
  219: 0 0FC6 0FC6 wc,3
  220: 0 102D 1030 wc,3
  221: 0 1032 1032 wc,3
  222: 0 1036 1037 wc,3
  223: 0 1039 1039 wc,3
  224: 0 1058 1059 wc,3
  225: 1 0000 1100 wc,3
  226: 2 1100 115f wc,3
  227: 0 1160 11FF wc,3
  228: 0 1712 1714 wc,3
  229: 0 1732 1734 wc,3
  230: 0 1752 1753 wc,3
  231: 0 1772 1773 wc,3
  232: 0 17B4 17B5 wc,3
  233: 0 17B7 17BD wc,3
  234: 0 17C6 17C6 wc,3
  235: 0 17C9 17D3 wc,3
  236: 0 17DD 17DD wc,3
  237: 0 180B 180D wc,3
  238: 0 18A9 18A9 wc,3
  239: 0 1920 1922 wc,3
  240: 0 1927 1928 wc,3
  241: 0 1932 1932 wc,3
  242: 0 1939 193B wc,3
  243: 0 200B 200F wc,3
  244: 0 202A 202E wc,3
  245: 0 2060 2063 wc,3
  246: 0 206A 206F wc,3
  247: 0 20D0 20EA wc,3
  248: 2 2329 232A wc,3
  249: 0 302A 302F wc,3
  250: 2 2E80 303E wc,3
  251: 0 3099 309A wc,3
  252: 2 3040 A4CF wc,3
  253: 2 AC00 D7A3 wc,3
  254: 2 F900 FAFF wc,3
  255: 0 FB1E FB1E wc,3
  256: 0 FE00 FE0F wc,3
  257: 0 FE20 FE23 wc,3
  258: 2 FE30 FE6F wc,3
  259: 0 FEFF FEFF wc,3
  260: 2 FF00 FF60 wc,3
  261: 2 FFE0 FFE6 wc,3
  262: 0 FFF9 FFFB wc,3
  263: 0 1D167 1D169 wc,3
  264: 0 1D173 1D182 wc,3
  265: 0 1D185 1D18B wc,3
  266: 0 1D1AA 1D1AD wc,3
  267: 2 20000 2FFFD wc,3
  268: 2 30000 3FFFD wc,3
  269: 0 E0001 E0001 wc,3
  270: 0 E0020 E007F wc,3
  271: 0 E0100 E01EF wc,3
  272: here wc-table - Constant #wc-table
  273: 
  274: \ inefficient table walk:
  275: 
  276: : wcwidth ( xc -- n )
  277:     wc-table #wc-table over + swap ?DO
  278: 	dup I 2@ within IF  I 2 cells + @  UNLOOP EXIT  THEN
  279:     3 cells +LOOP  1 ;
  280: [THEN]
  281:     
  282: : u8width ( xcaddr u -- n )
  283:     0 rot rot over + swap ?DO
  284:         I xc@+ swap >r wcwidth +
  285:     r> I - +LOOP ;
  286: 
  287: : set-encoding-utf-8 ( -- )
  288:     ['] u8emit is xemit
  289:     ['] u8key is xkey
  290:     ['] u8>> is xchar+
  291:     ['] u8<< is xchar-
  292: [ [IFDEF] xstring+ ]
  293:     ['] u8\string- is xstring-
  294:     ['] +u8/string is +xstring
  295: [ [THEN] ]
  296: [ [IFDEF] +x/string ]
  297:     ['] u8\string- is x\string-
  298:     ['] +u8/string is +x/string
  299: [ [THEN] ]
  300:     ['] u8@ is xc@
  301:     ['] u8!+? is xc!+?
  302:     ['] u8@+ is xc@+
  303:     ['] u8len is xc-size
  304: [ [IFDEF] x-width ]
  305:     ['] u8width is x-width
  306: [ [THEN] ]
  307: [ [IFDEF] x-size ]
  308:     ['] u8addrlen is x-size
  309: [ [THEN] ]
  310:     ['] -u8trailing-garbage is -trailing-garbage
  311: ;
  312: 
  313: : utf-8-cold ( -- )
  314:     s" LC_ALL" getenv 2dup d0= IF  2drop
  315: 	s" LC_CTYPE" getenv 2dup d0= IF  2drop
  316: 	    s" LANG" getenv 2dup d0= IF  2drop
  317: 		s" C"  THEN THEN THEN
  318:     s" UTF-8" search nip nip
  319:     IF  set-encoding-utf-8  ELSE  set-encoding-fixed-width  THEN ;
  320: 
  321: environment-wordlist set-current
  322: : xchar-encoding ( -- addr u ) \ xchar-ext
  323:     \G Returns a printable ASCII string that reperesents the encoding,
  324:     \G and use the preferred MIME name (if any) or the name in
  325:     \G @url{http://www.iana.org/assignments/character-sets} like
  326:     \G ``ISO-LATIN-1'' or ``UTF-8'', with the exception of ``ASCII'', where
  327:     \G we prefer the alias ``ASCII''.
  328:     max-single-byte $80 = IF s" UTF-8" ELSE s" ISO-LATIN-1" THEN ;
  329: forth definitions
  330: 
  331: :noname ( -- )
  332:     defers 'cold
  333:     utf-8-cold
  334: ; is 'cold
  335: 
  336: utf-8-cold

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