-! Copyright (C) 2006, 2008 Daniel Ehrenberg.
+! Copyright (C) 2006, 2009 Daniel Ehrenberg.
! See http://factorcode.org/license.txt for BSD license.
-USING: math kernel sequences sbufs vectors namespaces io.binary
-io.encodings combinators splitting io byte-arrays inspector ;
+USING: accessors alien.accessors byte-arrays io io.binary
+io.encodings kernel math math.private sequences
+sequences.private strings strings.private ;
IN: io.encodings.utf16
-TUPLE: utf16be ;
+SINGLETON: utf16be
-TUPLE: utf16le ;
+SINGLETON: utf16le
-TUPLE: utf16 ;
+SINGLETON: utf16
+
+ERROR: missing-bom ;
<PRIVATE
: quad-be ( stream byte -- stream char )
double-be over stream-read1 [
- dup -2 shift BIN: 110111 number= [
- >r 2 shift r> BIN: 11 bitand bitor
- over stream-read1 swap append-nums HEX: 10000 +
+ dup -2 shift 0b110111 number= [
+ [ 2 shift ] dip 0b11 bitand bitor
+ over stream-read1 swap append-nums 0x10000 +
] [ 2drop dup stream-read1 drop replacement-char ] if
] when* ;
dup stream-read1 drop replacement-char ;
: begin-utf16be ( stream byte -- stream char )
- dup -3 shift BIN: 11011 number= [
- dup BIN: 00000100 bitand zero?
- [ BIN: 11 bitand quad-be ]
+ dup -3 shift 0b11011 number= [
+ dup 0b00000100 bitand zero?
+ [ 0b11 bitand quad-be ]
[ drop ignore ] if
] [ double-be ] if ;
-
+
M: utf16be decode-char
drop dup stream-read1 dup [ begin-utf16be ] when nip ;
: quad-le ( stream ch -- stream char )
over stream-read1 swap 10 shift bitor
- over stream-read1 dup -2 shift BIN: 110111 = [
- BIN: 11 bitand append-nums HEX: 10000 +
+ over stream-read1 dup -2 shift 0b110111 = [
+ 0b11 bitand append-nums 0x10000 +
] [ 2drop replacement-char ] if ;
: double-le ( stream byte1 byte2 -- stream char )
- dup -3 shift BIN: 11011 = [
- dup BIN: 100 bitand 0 number=
- [ BIN: 11 bitand 8 shift bitor quad-le ]
+ dup -3 shift 0b11011 = [
+ dup 0b100 bitand 0 number=
+ [ 0b11 bitand 8 shift bitor quad-le ]
[ 2drop replacement-char ] if
] [ append-nums ] if ;
: encode-first ( char -- byte1 byte2 )
-10 shift
- dup -8 shift BIN: 11011000 bitor
- swap HEX: FF bitand ;
+ [ -8 shift 0b11011000 bitor ] [ 0xFF bitand ] bi ; inline
: encode-second ( char -- byte3 byte4 )
- BIN: 1111111111 bitand
- dup -8 shift BIN: 11011100 bitor
- swap BIN: 11111111 bitand ;
+ 0b1111111111 bitand
+ [ -8 shift 0b11011100 bitor ] [ 0b11111111 bitand ] bi ; inline
-: stream-write2 ( stream char1 char2 -- )
- rot [ stream-write1 ] curry bi@ ;
+: stream-write2 ( char1 char2 stream -- )
+ [ B{ } 2sequence ] dip stream-write ; inline
+ ! [ stream-write1 ] curry bi@ ; inline
-: char>utf16be ( stream char -- )
- dup HEX: FFFF > [
- HEX: 10000 -
- 2dup encode-first stream-write2
- encode-second stream-write2
- ] [ h>b/b swap stream-write2 ] if ;
+: char>utf16be ( char stream -- )
+ over 0xFFFF > [
+ [ 0x10000 - ] dip
+ [ [ encode-first ] dip stream-write2 ]
+ [ [ encode-second ] dip stream-write2 ] 2bi
+ ] [ [ h>b/b swap ] dip stream-write2 ] if ; inline
M: utf16be encode-char ( char stream encoding -- )
- drop swap char>utf16be ;
+ drop char>utf16be ;
: char>utf16le ( char stream -- )
- dup HEX: FFFF > [
- HEX: 10000 -
- 2dup encode-first swap stream-write2
- encode-second swap stream-write2
- ] [ h>b/b stream-write2 ] if ;
+ over 0xFFFF > [
+ [ 0x10000 - ] dip
+ [ [ encode-first swap ] dip stream-write2 ]
+ [ [ encode-second swap ] dip stream-write2 ] 2bi
+ ] [ [ h>b/b ] dip stream-write2 ] if ; inline
M: utf16le encode-char ( char stream encoding -- )
- drop swap char>utf16le ;
+ drop char>utf16le ;
-! UTF-16
+: ascii-char>utf16-byte-array ( off n byte-array string -- )
+ overd string-nth-fast -rot
+ [ 2 fixnum*fast rot fixnum+fast ] dip
+ set-nth-unsafe ; inline
+
+: ascii-string>utf16-byte-array ( off string -- byte-array )
+ [ length >fixnum [ <iota> ] [ 2 fixnum*fast <byte-array> ] bi ] keep
+ [ [ ascii-char>utf16-byte-array ] 2curry with each ] keepd ; inline
+
+: ascii-string>utf16le ( string stream -- )
+ [ 0 swap ascii-string>utf16-byte-array ] dip stream-write ; inline
+: ascii-string>utf16be ( string stream -- )
+ [ 1 swap ascii-string>utf16-byte-array ] dip stream-write ; inline
+
+GENERIC#: encode-string-utf16le 1 ( string stream -- )
+
+M: object encode-string-utf16le
+ [ char>utf16le ] curry each ; inline
+
+M: string encode-string-utf16le
+ over aux>>
+ [ call-next-method ]
+ [ ascii-string>utf16le ] if ; inline
-: bom-le B{ HEX: ff HEX: fe } ; inline
+M: utf16le encode-string drop encode-string-utf16le ;
-: bom-be B{ HEX: fe HEX: ff } ; inline
+GENERIC#: encode-string-utf16be 1 ( string stream -- )
-: start-utf16le? ( seq1 -- seq2 ? ) bom-le ?head ;
+M: object encode-string-utf16be
+ [ char>utf16be ] curry each ; inline
-: start-utf16be? ( seq1 -- seq2 ? ) bom-be ?head ;
+M: string encode-string-utf16be
+ over aux>>
+ [ call-next-method ]
+ [ ascii-string>utf16be ] if ; inline
-TUPLE: missing-bom ;
-M: missing-bom summary drop "The BOM for a UTF-16 stream was missing" ;
+M: utf16be encode-string drop encode-string-utf16be ;
+
+M: utf16le guess-encoded-length drop 2 * ; inline
+M: utf16le guess-decoded-length drop 2 /i ; inline
+
+M: utf16be guess-encoded-length drop 2 * ; inline
+M: utf16be guess-decoded-length drop 2 /i ; inline
+
+! UTF-16
+
+CONSTANT: bom-le B{ 0xff 0xfe }
+
+CONSTANT: bom-be B{ 0xfe 0xff }
: bom>le/be ( bom -- le/be )
dup bom-le sequence= [ drop utf16le ] [
M: utf16 <encoder> ( stream utf16 -- encoder )
drop bom-le over stream-write utf16le <encoder> ;
+: le? ( -- ? ) B{ 1 0 0 0 } 0 alien-unsigned-4 1 = ; foldable
+
PRIVATE>
+
+: utf16n ( -- value ) le? utf16le utf16be ? ;
projects / factor.git / blobdiff