factor/basis/csv/csv.factor

122 lines
3.4 KiB
Factor
Raw Normal View History

2008-04-17 16:29:04 -04:00
! Copyright (C) 2007, 2008 Phil Dawes
! See http://factorcode.org/license.txt for BSD license.
2012-07-27 16:22:32 -04:00
USING: combinators fry io io.files io.streams.string kernel
2013-03-15 00:05:41 -04:00
make math memoize namespaces sbufs sequences sequences.private
2012-07-27 16:22:32 -04:00
unicode.categories ;
2008-04-17 16:29:04 -04:00
IN: csv
SYMBOL: delimiter
2008-04-17 16:29:04 -04:00
CHAR: , delimiter set-global
<PRIVATE
2013-03-15 00:05:41 -04:00
MEMO: field-delimiters ( delimiter -- field-end quoted-field )
[ "\n" swap suffix ] [ "\"\n" swap suffix ] bi ; inline
2012-07-27 16:22:32 -04:00
DEFER: quoted-field
2013-03-15 00:05:41 -04:00
: maybe-escaped-quote ( delimeter stream quoted? -- delimiter stream endchar/f )
2over stream-read1 swap over =
2012-07-27 16:22:32 -04:00
[ nip ] [
{
{ CHAR: " [ [ CHAR: " , ] when quoted-field ] }
{ CHAR: \n [ ] } ! Error: newline inside string?
[ [ , drop f maybe-escaped-quote ] when* ]
} case
2013-03-15 00:05:41 -04:00
] if ; inline recursive
2012-07-18 15:24:24 -04:00
2013-03-15 00:05:41 -04:00
: quoted-field ( delimiter stream -- delimiter stream endchar/f )
"\"" over stream-read-until drop % t maybe-escaped-quote ;
2008-04-17 16:29:04 -04:00
2012-07-27 16:22:32 -04:00
: ?trim ( string -- string' )
dup length [ drop "" ] [
over first-unsafe blank?
[ drop t ] [ 1 - over nth-unsafe blank? ] if
[ [ blank? ] trim ] when
] if-zero ; inline
2013-03-15 00:05:41 -04:00
: field ( delimiter stream field-end quoted-field -- delimiter sep/f field )
pick stream-read-until dup CHAR: " = [
2012-07-27 16:22:32 -04:00
drop
2013-03-15 00:05:41 -04:00
[ drop [ quoted-field nip ] "" make ]
2012-07-27 16:22:32 -04:00
[
2013-03-15 00:05:41 -04:00
swap rot stream-read-until
2012-07-27 16:22:32 -04:00
[ "\"" glue ] dip swap ?trim
]
if-empty
2013-03-15 00:05:41 -04:00
] [ [ 2drop ] 2dip swap ?trim ] if ;
2012-07-27 16:22:32 -04:00
: (stream-read-row) ( delimiter stream field-end quoted-field -- delimiter sep/f fields )
[ dup [ 2dup = ] ] 3dip '[ drop _ _ _ field ] produce ;
2008-04-17 16:29:04 -04:00
: (stream-read-csv) ( stream -- )
delimiter get
[ dup [ empty? ] all? [ drop ] [ , ] if ]
rot pick field-delimiters
'[ _ _ _ (stream-read-row) ] do while drop ;
PRIVATE>
: stream-read-row ( stream -- row )
delimiter get swap over field-delimiters
(stream-read-row) 2nip ; inline
: read-row ( -- row )
input-stream get stream-read-row ; inline
2008-04-17 16:29:04 -04:00
: stream-read-csv ( stream -- rows )
[ (stream-read-csv) ] { } make
dup last { "" } = [ but-last ] when ; inline
: read-csv ( -- rows )
input-stream get stream-read-csv ; inline
2010-02-25 19:30:40 -05:00
: string>csv ( string -- csv )
[ read-csv ] with-string-reader ;
2010-02-25 19:30:40 -05:00
: file>csv ( path encoding -- csv )
[ read-csv ] with-file-reader ;
: with-delimiter ( ch quot -- )
[ delimiter ] dip with-variable ; inline
<PRIVATE
2008-07-25 17:02:07 -04:00
2012-07-27 16:47:15 -04:00
: needs-escaping? ( cell delimiter -- ? )
'[ dup "\n\"" member? [ drop t ] [ _ = ] if ] any? ; inline
2008-07-25 17:02:07 -04:00
2013-03-15 00:05:41 -04:00
: escape-quotes ( cell stream -- )
CHAR: " over stream-write1 swap [
[ over stream-write1 ]
[ dup CHAR: " = [ over stream-write1 ] [ drop ] if ] bi
] each CHAR: " swap stream-write1 ; inline
2013-03-15 00:05:41 -04:00
: escape-if-required ( cell delimiter stream -- )
[ dupd needs-escaping? ] dip
[ escape-quotes ] [ stream-write ] bi-curry if ; inline
: (stream-write-row) ( row delimiter stream -- )
2013-03-15 00:05:41 -04:00
[ '[ _ _ stream-write1 ] ] 2keep
'[ _ _ escape-if-required ] interleave nl ; inline
2012-07-27 16:22:32 -04:00
PRIVATE>
: stream-write-row ( row stream -- )
delimiter get swap (stream-write-row) ; inline
2010-02-25 19:30:40 -05:00
: write-row ( row -- )
output-stream get stream-write-row ; inline
: stream-write-csv ( rows stream -- )
delimiter get swap '[ _ _ (stream-write-row) ] each ;
2010-02-25 19:30:40 -05:00
: write-csv ( rows -- )
output-stream get stream-write-csv ;
2010-02-25 19:30:40 -05:00
: csv>string ( csv -- string )
[ write-csv ] with-string-writer ;
2013-03-15 00:05:41 -04:00
: csv>file ( rows path encoding -- )
[ write-csv ] with-file-writer ;