factor/core/io/encodings/encodings.factor

135 lines
3.2 KiB
Factor
Raw Normal View History

2008-02-21 16:22:49 -05:00
! Copyright (C) 2008 Daniel Ehrenberg.
2007-09-20 18:09:08 -04:00
! See http://factorcode.org/license.txt for BSD license.
USING: math kernel sequences sbufs vectors namespaces growable
2008-05-15 00:23:12 -04:00
strings io classes continuations destructors combinators
io.styles io.streams.plain splitting byte-arrays
sequences.private accessors ;
2007-09-20 18:09:08 -04:00
IN: io.encodings
! The encoding descriptor protocol
2008-03-14 04:09:51 -04:00
GENERIC: decode-char ( stream encoding -- char/f )
2008-03-14 04:09:51 -04:00
GENERIC: encode-char ( char stream encoding -- )
2008-03-20 21:11:45 -04:00
GENERIC: <decoder> ( stream encoding -- newstream )
2008-03-14 04:09:51 -04:00
: replacement-char HEX: fffd ;
2008-02-01 18:45:35 -05:00
TUPLE: decoder stream code cr ;
2008-02-13 02:02:37 -05:00
2008-03-20 16:00:49 -04:00
ERROR: decode-error ;
2008-02-13 18:53:10 -05:00
GENERIC: <encoder> ( stream encoding -- newstream )
TUPLE: encoder stream code ;
2008-03-20 16:00:49 -04:00
ERROR: encode-error ;
! Decoding
<PRIVATE
M: object <decoder> f decoder boa ;
2008-02-11 00:14:42 -05:00
2008-03-14 04:09:51 -04:00
: >decoder< ( decoder -- stream encoding )
[ stream>> ] [ code>> ] bi ;
2008-02-11 00:14:42 -05:00
: cr+ t swap set-decoder-cr ; inline
2008-02-13 20:53:53 -05:00
: cr- f swap set-decoder-cr ; inline
2008-02-16 16:35:44 -05:00
: line-ends/eof ( stream str -- str ) f like swap cr- ; inline
2008-02-13 20:53:53 -05:00
2008-02-16 16:35:44 -05:00
: line-ends\r ( stream str -- str ) swap cr+ ; inline
2008-02-13 20:53:53 -05:00
2008-02-16 16:35:44 -05:00
: line-ends\n ( stream str -- str )
over decoder-cr over empty? and
2008-02-16 16:35:44 -05:00
[ drop dup cr- stream-readln ] [ swap cr- ] if ; inline
2008-02-13 20:53:53 -05:00
2008-02-16 16:35:44 -05:00
: handle-readln ( stream str ch -- str )
{
{ f [ line-ends/eof ] }
{ CHAR: \r [ line-ends\r ] }
{ CHAR: \n [ line-ends\n ] }
} case ;
2008-02-13 20:53:53 -05:00
2008-02-16 16:35:44 -05:00
: fix-read ( stream string -- string )
over decoder-cr [
2008-02-16 16:35:44 -05:00
over cr-
"\n" ?head [
over stream-read1 [ suffix ] when*
2008-03-14 04:09:51 -04:00
] when
] when nip ;
: read-loop ( n stream -- string )
2008-03-18 17:01:14 -04:00
SBUF" " clone [
2008-03-14 04:09:51 -04:00
[
2008-03-18 17:01:14 -04:00
>r nip stream-read1 dup
[ r> push f ] [ r> 2drop t ] if
] 2curry find-integer drop
] keep "" like f like ;
2008-02-13 20:53:53 -05:00
M: decoder stream-read
2008-03-14 04:09:51 -04:00
tuck read-loop fix-read ;
2008-02-16 16:35:44 -05:00
2008-03-18 17:01:14 -04:00
M: decoder stream-read-partial stream-read ;
2008-03-14 04:09:51 -04:00
: (read-until) ( buf quot -- string/f sep/f )
2008-03-18 17:01:14 -04:00
! quot: -- char stop?
2008-03-14 04:09:51 -04:00
dup call
[ >r drop "" like r> ]
[ pick push (read-until) ] if ; inline
M: decoder stream-read-until
2008-03-14 04:09:51 -04:00
SBUF" " clone -rot >decoder<
2008-03-18 17:01:14 -04:00
[ decode-char [ dup rot memq? ] [ drop f t ] if* ] 3curry
(read-until) ;
2008-02-13 20:53:53 -05:00
2008-02-16 16:35:44 -05:00
: fix-read1 ( stream char -- char )
over decoder-cr [
2008-02-16 16:35:44 -05:00
over cr-
dup CHAR: \n = [
2008-03-14 04:09:51 -04:00
drop dup stream-read1
] when
] when nip ;
2008-02-16 16:35:44 -05:00
M: decoder stream-read1
2008-03-14 04:09:51 -04:00
dup >decoder< decode-char fix-read1 ;
2008-02-16 16:35:44 -05:00
M: decoder stream-readln ( stream -- str )
2008-02-16 16:35:44 -05:00
"\r\n" over stream-read-until handle-readln ;
2008-03-14 04:09:51 -04:00
M: decoder dispose decoder-stream dispose ;
2008-02-16 16:35:44 -05:00
! Encoding
M: object <encoder> encoder boa ;
2008-03-14 04:09:51 -04:00
: >encoder< ( encoder -- stream encoding )
[ stream>> ] [ code>> ] bi ;
2008-02-16 16:35:44 -05:00
M: encoder stream-write1
2008-03-14 04:09:51 -04:00
>encoder< encode-char ;
2008-02-13 20:53:53 -05:00
M: encoder stream-write
2008-03-14 04:09:51 -04:00
>encoder< [ encode-char ] 2curry each ;
2008-02-16 16:35:44 -05:00
2008-03-14 04:09:51 -04:00
M: encoder dispose encoder-stream dispose ;
2008-02-13 20:53:53 -05:00
2008-03-18 17:01:14 -04:00
M: encoder stream-flush encoder-stream stream-flush ;
INSTANCE: encoder plain-writer
PRIVATE>
2008-02-15 20:44:35 -05:00
: re-encode ( stream encoding -- newstream )
2008-03-14 04:09:51 -04:00
over encoder? [ >r encoder-stream r> ] when <encoder> ;
2008-02-15 20:44:35 -05:00
: encode-output ( encoding -- )
output-stream [ swap re-encode ] change ;
: re-decode ( stream encoding -- newstream )
2008-03-14 04:09:51 -04:00
over decoder? [ >r decoder-stream r> ] when <decoder> ;
: decode-input ( encoding -- )
input-stream [ swap re-decode ] change ;