1 ! Copyright (C) 2008 Daniel Ehrenberg
2 ! See http://factorcode.org/license.txt for BSD license.
3 USING: math.parser arrays io.encodings sequences kernel assocs
4 hashtables io.encodings.ascii generic parser classes.tuple words
5 words.symbol io io.files splitting namespaces math
6 compiler.units accessors classes.singleton classes.mixin
7 io.encodings.iana fry simple-flat-file ;
13 ! encoding-name iana-name file-name
14 { "latin1" "ISO_8859-1:1987" "8859-1" }
15 { "latin2" "ISO_8859-2:1987" "8859-2" }
16 { "latin3" "ISO_8859-3:1988" "8859-3" }
17 { "latin4" "ISO_8859-4:1988" "8859-4" }
18 { "latin/cyrillic" "ISO_8859-5:1988" "8859-5" }
19 { "latin/arabic" "ISO_8859-6:1987" "8859-6" }
20 { "latin/greek" "ISO_8859-7:1987" "8859-7" }
21 { "latin/hebrew" "ISO_8859-8:1988" "8859-8" }
22 { "latin5" "ISO_8859-9:1989" "8859-9" }
23 { "latin6" "ISO-8859-10" "8859-10" }
24 { "latin/thai" "TIS-620" "8859-11" }
25 { "latin7" "ISO-8859-13" "8859-13" }
26 { "latin8" "ISO-8859-14" "8859-14" }
27 { "latin9" "ISO-8859-15" "8859-15" }
28 { "latin10" "ISO-8859-16" "8859-16" }
29 { "koi8-r" "KOI8-R" "KOI8-R" }
30 { "windows-1252" "windows-1252" "CP1252" }
31 { "ebcdic" "IBM037" "CP037" }
32 { "mac-roman" "macintosh" "ROMAN" }
35 : encoding-file ( file-name -- stream )
36 "vocab:io/encodings/8-bit/" ".TXT" surround ;
38 SYMBOL: 8-bit-encodings
40 TUPLE: 8-bit biassoc ;
42 : encode-8-bit ( char stream assoc -- )
44 [ swap stream-write1 ] [ encode-error ] if* ; inline
46 M: 8-bit encode-char biassoc>> encode-8-bit ;
48 : decode-8-bit ( stream assoc -- char/f )
50 [ swap at [ replacement-char ] unless* ]
51 [ drop f ] if* ; inline
53 M: 8-bit decode-char biassoc>> decode-8-bit ;
57 M: 8-bit-encoding <encoder>
58 8-bit-encodings get-global at <encoder> ;
60 M: 8-bit-encoding <decoder>
61 8-bit-encodings get-global at <decoder> ;
63 : create-encoding ( name -- word )
64 "io.encodings.8-bit" create
65 [ define-singleton-class ]
66 [ 8-bit-encoding add-mixin-instance ]
75 [ dupd register-encoding ]
76 [ encoding-file flat-file>biassoc 8-bit boa ]
79 8-bit-encodings set-global
80 ] with-compilation-unit