factor/basis/csv/csv.factor

125 lines
2.8 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.
USING: kernel sequences io namespaces make combinators
2010-02-25 19:30:40 -05:00
unicode.categories io.files combinators.short-circuit
io.streams.string fry memoize ;
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
2012-07-18 15:24:24 -04:00
: field-end ( -- str sep )
delimiter> (field-end) read-until ; inline
DEFER: quoted-field
MEMO: (quoted-field) ( delimiter -- delimiter' )
2012-07-14 18:33:24 -04:00
"\"\n" swap suffix ; inline
2012-07-18 15:24:24 -04:00
: maybe-escaped-quote ( quoted? -- endchar )
read1 dup {
2012-07-18 15:24:24 -04:00
{ CHAR: " [ over [ , ] [ drop ] if quoted-field ] }
{ delimiter> [ ] }
2012-07-18 15:24:24 -04:00
{ CHAR: \n [ ] } ! Error: newline inside string?
[ [ , f maybe-escaped-quote ] when ]
} case nip ;
2008-04-17 16:29:04 -04:00
: quoted-field ( -- endchar )
"\"" read-until
2012-07-18 15:24:24 -04:00
drop % t maybe-escaped-quote ;
: ?trim ( string -- string' )
dup { [ first blank? ] [ last blank? ] } 1||
[ [ blank? ] trim ] when ;
2008-04-17 16:29:04 -04:00
: field ( -- sep string )
delimiter> (quoted-field) read-until
dup CHAR: " = [
2012-07-18 15:24:24 -04:00
over empty?
[ 2drop [ quoted-field ] "" make ]
[ drop field-end [ "\"" glue ] dip swap ?trim ]
if
] [
2012-07-18 15:24:24 -04:00
swap [ "" ] [ ?trim ] if-empty
] if ;
2008-04-17 16:29:04 -04:00
: (row) ( -- sep )
f delimiter> '[ dup _ = ]
[ drop field , ] do while ;
2008-04-17 16:29:04 -04:00
: row ( -- eof? array[string] )
[ (row) ] { } make ;
2008-04-17 16:29:04 -04:00
: (csv) ( -- )
[ dup [ empty? ] all? [ drop ] [ , ] if ]
[ row ] do while ;
PRIVATE>
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 )
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
: 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
2008-07-25 17:02:07 -04:00
: escape-if-required ( cell -- cell' )
dup needs-escaping?
[ escape-quotes enclose-in-quotes ] when ; inline
PRIVATE>
2008-07-25 17:02:07 -04:00
: write-row ( row -- )
delimiter> '[ _ write1 ]
[ escape-if-required write ] interleave nl ; inline
2010-02-25 19:30:40 -05:00
<PRIVATE
: (write-csv) ( rows -- )
[ 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 ;