1 ! Copyright (C) 2006, 2009 Daniel Ehrenberg.
2 ! See http://factorcode.org/license.txt for BSD license.
3 USING: accessors byte-arrays combinators io io.binary
4 io.encodings kernel math math.private namespaces sbufs
5 sequences sequences.private splitting strings strings.private
21 : append-nums ( byte ch -- ch )
22 over [ 8 shift bitor ] [ 2drop replacement-char ] if ;
24 : double-be ( stream byte -- stream char )
25 over stream-read1 swap append-nums ;
27 : quad-be ( stream byte -- stream char )
28 double-be over stream-read1 [
29 dup -2 shift 0b110111 number= [
30 [ 2 shift ] dip 0b11 bitand bitor
31 over stream-read1 swap append-nums 0x10000 +
32 ] [ 2drop dup stream-read1 drop replacement-char ] if
35 : ignore ( stream -- stream char )
36 dup stream-read1 drop replacement-char ;
38 : begin-utf16be ( stream byte -- stream char )
39 dup -3 shift 0b11011 number= [
40 dup 0b00000100 bitand zero?
41 [ 0b11 bitand quad-be ]
45 M: utf16be decode-char
46 drop dup stream-read1 dup [ begin-utf16be ] when nip ;
50 : quad-le ( stream ch -- stream char )
51 over stream-read1 swap 10 shift bitor
52 over stream-read1 dup -2 shift 0b110111 = [
53 0b11 bitand append-nums 0x10000 +
54 ] [ 2drop replacement-char ] if ;
56 : double-le ( stream byte1 byte2 -- stream char )
57 dup -3 shift 0b11011 = [
58 dup 0b100 bitand 0 number=
59 [ 0b11 bitand 8 shift bitor quad-le ]
60 [ 2drop replacement-char ] if
61 ] [ append-nums ] if ;
63 : begin-utf16le ( stream byte -- stream char )
64 over stream-read1 [ double-le ] [ drop replacement-char ] if* ;
66 M: utf16le decode-char
67 drop dup stream-read1 dup [ begin-utf16le ] when nip ;
69 ! UTF-16LE/BE encoding
71 : encode-first ( char -- byte1 byte2 )
73 [ -8 shift 0b11011000 bitor ] [ 0xFF bitand ] bi ; inline
75 : encode-second ( char -- byte3 byte4 )
77 [ -8 shift 0b11011100 bitor ] [ 0b11111111 bitand ] bi ; inline
79 : stream-write2 ( char1 char2 stream -- )
80 [ B{ } 2sequence ] dip stream-write ; inline
81 ! [ stream-write1 ] curry bi@ ; inline
83 : char>utf16be ( char stream -- )
86 [ [ encode-first ] dip stream-write2 ]
87 [ [ encode-second ] dip stream-write2 ] 2bi
88 ] [ [ h>b/b swap ] dip stream-write2 ] if ; inline
90 M: utf16be encode-char
93 : char>utf16le ( char stream -- )
96 [ [ encode-first swap ] dip stream-write2 ]
97 [ [ encode-second swap ] dip stream-write2 ] 2bi
98 ] [ [ h>b/b ] dip stream-write2 ] if ; inline
100 M: utf16le encode-char
103 : ascii-char>utf16-byte-array ( off n byte-array string -- )
104 overd string-nth-fast -rot
105 [ 2 fixnum*fast rot fixnum+fast ] dip
106 set-nth-unsafe ; inline
108 : ascii-string>utf16-byte-array ( off string -- byte-array )
109 [ length >fixnum [ <iota> ] [ 2 fixnum*fast <byte-array> ] bi ] keep
110 [ [ ascii-char>utf16-byte-array ] 2curry with each ] keepd ; inline
112 : ascii-string>utf16le ( string stream -- )
113 [ 0 swap ascii-string>utf16-byte-array ] dip stream-write ; inline
114 : ascii-string>utf16be ( string stream -- )
115 [ 1 swap ascii-string>utf16-byte-array ] dip stream-write ; inline
117 GENERIC#: encode-string-utf16le 1 ( string stream -- )
119 M: object encode-string-utf16le
120 [ char>utf16le ] curry each ; inline
122 M: string encode-string-utf16le
125 [ ascii-string>utf16le ] if ; inline
127 M: utf16le encode-string drop encode-string-utf16le ;
129 GENERIC#: encode-string-utf16be 1 ( string stream -- )
131 M: object encode-string-utf16be
132 [ char>utf16be ] curry each ; inline
134 M: string encode-string-utf16be
137 [ ascii-string>utf16be ] if ; inline
139 M: utf16be encode-string drop encode-string-utf16be ;
141 M: utf16le guess-encoded-length drop 2 * ; inline
142 M: utf16le guess-decoded-length drop 2 /i ; inline
144 M: utf16be guess-encoded-length drop 2 * ; inline
145 M: utf16be guess-decoded-length drop 2 /i ; inline
149 CONSTANT: bom-le B{ 0xff 0xfe }
151 CONSTANT: bom-be B{ 0xfe 0xff }
153 : bom>le/be ( bom -- le/be )
154 dup bom-le sequence= [ drop utf16le ] [
155 bom-be sequence= [ utf16be ] [ missing-bom ] if
159 drop 2 over stream-read bom>le/be <decoder> ;
162 drop bom-le over stream-write utf16le <encoder> ;