factor/extra/html/parser/analyzer/analyzer.factor

92 lines
2.5 KiB
Factor
Raw Normal View History

2007-12-04 15:14:33 -05:00
USING: assocs html.parser kernel math sequences strings ;
IN: html.parser.analyzer
2007-09-20 18:09:08 -04:00
: remove-blank-text ( vector -- vector' )
2007-09-20 18:09:08 -04:00
[
dup tag-name text = [
tag-text [ blank? ] all? not
2007-09-20 18:09:08 -04:00
] [
drop t
] if
] subset ;
: trim-text ( vector -- vector' )
[
dup tag-name text = [
[ tag-text [ blank? ] trim ] keep
[ set-tag-text ] keep
] when
] map ;
2007-09-20 18:09:08 -04:00
: find-by-id ( id vector -- vector )
2008-01-09 17:36:30 -05:00
[ tag-attributes "id" swap at = ] with subset ;
2007-09-20 18:09:08 -04:00
: find-by-class ( id vector -- vector )
2008-01-09 17:36:30 -05:00
[ tag-attributes "class" swap at = ] with subset ;
2007-09-20 18:09:08 -04:00
: find-by-name ( str vector -- vector )
>r >lower r>
2008-01-09 17:36:30 -05:00
[ tag-name = ] with subset ;
2007-09-20 18:09:08 -04:00
: find-first-name ( str vector -- i/f tag/f )
>r >lower r>
2008-01-09 17:36:30 -05:00
[ tag-name = ] with find ;
2007-09-20 18:09:08 -04:00
: find-matching-close ( str vector -- i/f tag/f )
>r >lower r>
2008-01-09 17:36:30 -05:00
[ [ tag-name = ] keep tag-closing? and ] with find ;
2007-09-20 18:09:08 -04:00
: find-by-attribute-key ( key vector -- vector )
>r >lower r>
2008-01-09 17:36:30 -05:00
[ tag-attributes at ] with subset
2007-09-20 18:09:08 -04:00
[ ] subset ;
: find-by-attribute-key-value ( value key vector -- vector )
>r >lower r>
2008-01-09 17:36:30 -05:00
[ tag-attributes at over = ] with subset nip
2007-09-20 18:09:08 -04:00
[ ] subset ;
: find-first-attribute-key-value ( value key vector -- i/f tag/f )
>r >lower r>
2008-01-09 17:36:30 -05:00
[ tag-attributes at over = ] with find rot drop ;
2007-09-20 18:09:08 -04:00
: find-between ( i/f tag/f vector -- vector )
pick integer? [
rot 1+ tail-slice
>r tag-name r>
[ find-matching-close drop ] keep swap head
] [
3drop V{ } clone
] if ;
: find-links ( vector -- vector )
[ tag-name "a" = ] subset
[ tag-attributes "href" swap at ] map
[ ] subset ;
! : find-last-tag ( name vector -- index tag )
! [
! dup tag-matched? [ 2drop f ] [ tag-name = ] if
2008-01-09 17:36:30 -05:00
! ] with find-last ;
2007-09-20 18:09:08 -04:00
! : find-last-tag* ( name n vector -- tag )
! 0 -rot <slice> find-last-tag ;
! : find-matching-tag ( tag -- tag )
! dup tag-closing? [
! find-last-tag
! ] [
! ] if ;
! clear "/Users/erg/web/fark.html" file-contents parse-html find-links [ "go.pl" swap start ] subset [ "=" split peek ] map
2007-09-20 18:09:08 -04:00
! clear "http://fark.com" http-get parse-html find-links [ "go.pl" swap start ] subset [ "=" split peek ] map
! clear "/Users/erg/web/hostels.html" file-contents parse-html "Currency" "name" pick find-first-attribute-key-value
2007-09-20 18:09:08 -04:00
! clear "/Users/erg/web/hostels.html" file-contents parse-html
2007-12-04 15:14:33 -05:00
! "Currency" "name" pick find-first-attribute-key-value
2007-09-20 18:09:08 -04:00
! pick find-between remove-blank-text