-! Copyright (C) 2007, 2008 Phil Dawes
+! Copyright (C) 2007, 2008 Phil Dawes, 2013 John Benediktsson
! See http://factorcode.org/license.txt for BSD license.
-USING: combinators fry io io.files io.streams.string kernel
-make math memoize namespaces sbufs sequences sequences.private
-unicode.categories ;
+USING: combinators io io.files io.streams.string kernel make
+math namespaces sequences sequences.private unicode ;
IN: csv
SYMBOL: delimiter
<PRIVATE
-: delimiter> ( -- delimiter ) delimiter get ; inline
+MEMO: field-delimiters ( delimiter -- field-seps quote-seps )
+ [ "\r\n" swap prefix ] [ "\r\"\n" swap prefix ] bi ; inline
-MEMO: field-delimiters ( delimiter -- field-end quoted-field )
- [ "\n" swap suffix ] [ "\"\n" swap suffix ] bi ; inline
+DEFER: quoted-field,
-DEFER: quoted-field
-
-: maybe-escaped-quote ( delimeter stream quoted? -- delimiter stream endchar/f )
- 2over stream-read1 swap over =
+: maybe-escaped-quote ( delimeter stream quoted? -- delimiter stream sep/f )
+ 2over stream-read1 tuck =
[ nip ] [
{
- { CHAR: " [ [ CHAR: " , ] when quoted-field ] }
- { CHAR: \n [ ] } ! Error: newline inside string?
+ { CHAR: \" [ [ CHAR: \" , ] when quoted-field, ] }
+ { CHAR: \n [ ] } ! Error: cr inside string?
+ { CHAR: \r [ ] } ! Error: lf inside string?
[ [ , drop f maybe-escaped-quote ] when* ]
} case
] if ; inline recursive
-: quoted-field ( delimiter stream -- delimiter stream endchar/f )
+: quoted-field, ( delimiter stream -- delimiter stream sep/f )
"\"" over stream-read-until drop % t maybe-escaped-quote ;
+: quoted-field ( delimiter stream -- sep/f field )
+ [ quoted-field, 2nip ] "" make ;
+
: ?trim ( string -- string' )
dup length [ drop "" ] [
- over first-unsafe blank?
- [ drop t ] [ 1 - over nth-unsafe blank? ] if
- [ [ blank? ] trim ] when
+ over first-unsafe unicode:blank?
+ [ drop t ] [ 1 - over nth-unsafe unicode:blank? ] if
+ [ [ unicode:blank? ] trim ] when
] if-zero ; inline
-: field ( delimiter stream field-end quoted-field -- delimiter sep/f field )
- pick stream-read-until dup CHAR: " = [
- drop
- [ drop [ quoted-field nip ] "" make ]
- [
- swap rot stream-read-until
- [ "\"" glue ] dip swap ?trim
- ]
- if-empty
- ] [ [ 2drop ] 2dip swap ?trim ] if ;
-
-: (row) ( delimiter stream field-end quoted-field -- delimiter sep/f fields )
- [ dup [ 2dup = ] ] 3dip
- [ [ drop ] 3dip field ] 3curry produce ;
-
-: row ( delimiter -- delimiter sep/f fields )
- input-stream get over field-delimiters (row) ;
-
-: (csv) ( -- )
- delimiter>
+: continue-field ( delimiter stream field-seps seq -- sep/f field )
+ swap rot stream-read-until [ "\"" glue ] dip
+ swap ?trim nipd ; inline
+
+: field ( delimiter stream field-seps quote-seps -- sep/f field )
+ pick stream-read-until dup CHAR: \" = [
+ drop [ drop quoted-field ] [ continue-field ] if-empty
+ ] [ 3nipd swap ?trim ] if ;
+
+: (stream-read-row) ( delimiter stream field-end quoted-field -- sep/f fields )
+ [ [ dup '[ dup _ = ] ] keep ] 3dip
+ '[ drop _ _ _ _ field ] produce ; inline
+
+: (stream-read-csv) ( stream -- )
[ dup [ empty? ] all? [ drop ] [ , ] if ]
- input-stream get pick field-delimiters
- [ (row) ] 3curry do while drop ;
+ delimiter get rot over field-delimiters
+ '[ _ _ _ _ (stream-read-row) ] do while ;
PRIVATE>
-: read-row ( stream -- row )
- [ delimiter> row 2nip ] with-input-stream ;
+: stream-read-row ( stream -- row )
+ delimiter get tuck field-delimiters
+ (stream-read-row) nip ; inline
+
+: read-row ( -- row )
+ input-stream get stream-read-row ; inline
-: read-csv ( stream -- rows )
- [ [ (csv) ] { } make ] with-input-stream
- dup last { "" } = [ but-last ] when ;
+: 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
: string>csv ( string -- csv )
- <string-reader> read-csv ;
+ [ read-csv ] with-string-reader ;
: file>csv ( path encoding -- csv )
- <file-reader> read-csv ;
+ [ read-csv ] with-file-reader ;
: with-delimiter ( ch quot -- )
[ delimiter ] dip with-variable ; inline
<PRIVATE
: needs-escaping? ( cell delimiter -- ? )
- '[ dup "\n\"" member? [ drop t ] [ _ = ] if ] any? ; inline
+ '[ dup "\n\"\r" member? [ drop t ] [ _ = ] if ] any? ; inline
: escape-quotes ( cell stream -- )
- CHAR: " over stream-write1 swap [
+ CHAR: \" over stream-write1 swap [
[ over stream-write1 ]
- [ dup CHAR: " = [ over stream-write1 ] [ drop ] if ] bi
- ] each CHAR: " swap stream-write1 ; inline
+ [ dup CHAR: \" = [ over stream-write1 ] [ drop ] if ] bi
+ ] each CHAR: \" swap stream-write1 ;
: escape-if-required ( cell delimiter stream -- )
[ dupd needs-escaping? ] dip
[ escape-quotes ] [ stream-write ] bi-curry if ; inline
-: (write-row) ( row delimiter stream -- )
+: (stream-write-row) ( row delimiter stream -- )
[ '[ _ _ stream-write1 ] ] 2keep
'[ _ _ escape-if-required ] interleave nl ; inline
PRIVATE>
-: write-row ( row -- )
- delimiter> output-stream get (write-row) ; inline
-
-<PRIVATE
+: stream-write-row ( row stream -- )
+ delimiter get swap (stream-write-row) ; inline
-: (write-csv) ( rows -- )
- delimiter> output-stream get '[ _ _ (write-row) ] each ;
+: write-row ( row -- )
+ output-stream get stream-write-row ; inline
-PRIVATE>
+: stream-write-csv ( rows stream -- )
+ delimiter get swap '[ _ _ (stream-write-row) ] each ;
-: write-csv ( rows stream -- )
- [ (write-csv) ] with-output-stream ;
+: write-csv ( rows -- )
+ output-stream get stream-write-csv ;
: csv>string ( csv -- string )
- [ (write-csv) ] with-string-writer ;
+ [ write-csv ] with-string-writer ;
: csv>file ( rows path encoding -- )
- <file-writer> write-csv ;
+ [ write-csv ] with-file-writer ;
projects / factor.git / blobdiff