factor/basis/csv/csv.factor

90 lines
2.4 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.
! Simple CSV Parser
! Phil Dawes phil@phildawes.net
2008-09-10 23:11:40 -04:00
USING: kernel sequences io namespaces make
combinators 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
2008-12-15 23:49:35 -05:00
: delimiter> ( -- delimiter ) delimiter get ; inline
DEFER: quoted-field ( -- endchar )
2008-07-25 17:02:07 -04:00
! trims whitespace from either end of string
: trim-whitespace ( str -- str )
[ blank? ] trim ; inline
: skip-to-field-end ( -- endchar )
"\n" delimiter> suffix read-until nip ; inline
2008-04-17 16:29:04 -04:00
: not-quoted-field ( -- endchar )
"\"\n" delimiter> suffix read-until ! "
2008-04-17 16:29:04 -04:00
dup
{ { CHAR: " [ drop drop quoted-field ] } ! "
{ delimiter> [ swap trim-whitespace % ] }
{ CHAR: \n [ swap trim-whitespace % ] }
{ f [ swap trim-whitespace % ] } ! eof
2008-04-17 16:29:04 -04:00
} case ;
: maybe-escaped-quote ( -- endchar )
read1 dup
{ { CHAR: " [ , quoted-field ] } ! " is an escaped quote
{ delimiter> [ ] } ! end of quoted field
{ CHAR: \n [ ] }
[ 2drop skip-to-field-end ] ! end of quoted field + padding
2008-04-17 16:29:04 -04:00
} case ;
: quoted-field ( -- endchar )
"\"" read-until ! "
drop % maybe-escaped-quote ;
: field ( -- sep string )
[ not-quoted-field ] "" make ; ! trim-whitespace
2008-04-17 16:29:04 -04:00
: (row) ( -- sep )
field ,
dup delimiter get = [ drop (row) ] when ;
2008-04-17 16:29:04 -04:00
: row ( -- eof? array[string] )
[ (row) ] { } make ;
: append-if-row-not-empty ( row -- )
dup { "" } = [ drop ] [ , ] if ;
: (csv) ( -- )
row append-if-row-not-empty
[ (csv) ] when ;
2008-04-17 16:29:04 -04:00
: csv-row ( stream -- row )
[ row nip ] with-input-stream ;
2008-04-17 16:29:04 -04:00
: csv ( stream -- rows )
[ [ (csv) ] { } make ] with-input-stream ;
: with-delimiter ( char quot -- )
delimiter swap with-variable ; inline
2008-07-25 17:02:07 -04:00
: needs-escaping? ( cell -- ? )
[ [ "\n\"" member? ] [ delimiter get = ] bi or ] contains? ; inline ! "
2008-07-25 17:02:07 -04:00
: escape-quotes ( cell -- cell' )
[ [ dup , CHAR: " = [ CHAR: " , ] when ] each ] "" make ; inline
: enclose-in-quotes ( cell -- cell' )
CHAR: " [ prefix ] [ suffix ] bi ; inline ! "
: escape-if-required ( cell -- cell' )
dup needs-escaping? [ escape-quotes enclose-in-quotes ] when ; inline
: write-row ( row -- )
[ delimiter get write1 ] [ escape-if-required write ] interleave nl ; inline
2008-07-25 17:02:07 -04:00
2008-08-30 12:46:35 -04:00
: write-csv ( rows stream -- )
2008-07-25 17:02:07 -04:00
[ [ write-row ] each ] with-output-stream ;