factor/basis/csv/csv.factor

128 lines
3.0 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
make math memoize namespaces sequences sequences.private
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
2008-12-15 23:49:35 -05:00
: delimiter> ( -- delimiter ) delimiter get ; inline
MEMO: (field-end) ( delimiter -- delimiter' )
2012-07-14 18:33:24 -04:00
"\n" swap suffix ; inline
MEMO: (quoted-field) ( delimiter -- delimiter' )
2012-07-14 18:33:24 -04:00
"\"\n" swap suffix ; inline
2012-07-27 16:22:32 -04:00
DEFER: quoted-field
2012-07-27 16:22:32 -04:00
: maybe-escaped-quote ( delimeter quoted? -- delimiter endchar )
read1 pick over =
[ nip ] [
{
{ CHAR: " [ [ CHAR: " , ] when quoted-field ] }
{ CHAR: \n [ ] } ! Error: newline inside string?
[ [ , drop f maybe-escaped-quote ] when* ]
} case
] if ;
2012-07-18 15:24:24 -04:00
2012-07-27 16:22:32 -04:00
: quoted-field ( delimiter -- delimiter endchar )
"\"" 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
: field ( delimiter -- delimiter sep string )
dup (quoted-field) read-until
dup CHAR: " = [
2012-07-27 16:22:32 -04:00
drop
[ [ quoted-field ] "" make ]
[
over (field-end) read-until
[ "\"" glue ] dip swap ?trim
]
if-empty
] [ swap ?trim ] if ;
: (row) ( delimiter -- delimiter sep )
f [ 2dup = ] [ drop field , ] do while ;
: row ( delimiter -- delimiter eof? array[string] )
[ (row) ] { } make ;
2008-04-17 16:29:04 -04:00
: (csv) ( -- )
2012-07-27 16:22:32 -04:00
delimiter>
[ dup [ empty? ] all? [ drop ] [ , ] if ]
2012-07-27 16:22:32 -04:00
[ row ] do while drop ;
PRIVATE>
2008-04-17 16:29:04 -04:00
: csv-row ( stream -- row )
2012-07-27 16:22:32 -04:00
[ delimiter> row 2nip ] with-input-stream ;
2008-04-17 16:29:04 -04:00
: csv ( stream -- rows )
2009-02-12 02:32:06 -05:00
[ [ (csv) ] { } make ] with-input-stream
dup last { "" } = [ but-last ] when ;
2010-02-25 19:30:40 -05:00
: string>csv ( string -- csv )
<string-reader> csv ;
: file>csv ( path encoding -- csv )
<file-reader> csv ;
: 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
: escape-quotes ( cell -- cell' )
[
[
[ , ]
[ dup CHAR: " = [ , ] [ drop ] if ] bi
] each
] "" make ; inline
2008-07-25 17:02:07 -04:00
: enclose-in-quotes ( cell -- cell' )
"\"" dup surround ; inline
2012-07-27 16:47:15 -04:00
: escape-if-required ( cell delimiter -- cell' )
dupd needs-escaping?
[ escape-quotes enclose-in-quotes ] when ; inline
2012-07-27 16:22:32 -04:00
: (write-row) ( row delimiter -- )
2012-07-27 16:47:15 -04:00
dup '[ _ write1 ] swap
'[ _ escape-if-required write ] interleave nl ; inline
2012-07-27 16:22:32 -04:00
PRIVATE>
2008-07-25 17:02:07 -04:00
: write-row ( row -- )
2012-07-27 16:22:32 -04:00
delimiter> (write-row) ; inline
2010-02-25 19:30:40 -05:00
<PRIVATE
: (write-csv) ( rows -- )
2012-07-27 16:22:32 -04:00
delimiter> '[ _ (write-row) ] each ;
2010-02-25 19:30:40 -05:00
PRIVATE>
2008-08-30 12:46:35 -04:00
: write-csv ( rows stream -- )
2010-02-25 19:30:40 -05:00
[ (write-csv) ] with-output-stream ;
2010-02-25 19:30:40 -05:00
: csv>string ( csv -- string )
[ (write-csv) ] with-string-writer ;
: csv>file ( rows path encoding -- ) <file-writer> write-csv ;