2008-09-25 03:06:49 -04:00
|
|
|
! Copyright (C) 2008 Doug Coleman.
|
|
|
|
|
! See http://factorcode.org/license.txt for BSD license.
|
2008-02-11 17:27:40 -05:00
|
|
|
USING: assocs html.parser kernel math sequences strings ascii
|
2008-09-10 23:11:40 -04:00
|
|
|
arrays generalizations shuffle unicode.case namespaces make
|
2008-09-25 03:02:47 -04:00
|
|
|
splitting http accessors io combinators http.client urls
|
2008-10-01 18:53:46 -04:00
|
|
|
urls.encoding fry ;
|
2007-12-04 15:14:33 -05:00
|
|
|
IN: html.parser.analyzer
|
2007-09-20 18:09:08 -04:00
|
|
|
|
2008-05-24 00:34:10 -04:00
|
|
|
TUPLE: link attributes clickable ;
|
|
|
|
|
|
2008-09-29 21:21:12 -04:00
|
|
|
: scrape-html ( url -- headers vector )
|
|
|
|
|
http-get parse-html ;
|
2008-05-19 19:58:56 -04:00
|
|
|
|
2008-09-25 03:02:47 -04:00
|
|
|
: find-all ( seq quot -- alist )
|
|
|
|
|
[ <enum> >alist ] [ '[ second @ ] ] bi* filter ; inline
|
2008-03-06 18:09:48 -05:00
|
|
|
|
2008-09-25 03:02:47 -04:00
|
|
|
: find-nth ( seq quot n -- i elt )
|
|
|
|
|
[ <enum> >alist ] 2dip -rot
|
|
|
|
|
'[ _ [ second @ ] find-from rot drop swap 1+ ]
|
|
|
|
|
[ f 0 ] 2dip times drop first2 ; inline
|
2008-03-06 18:09:48 -05:00
|
|
|
|
2008-09-25 03:02:47 -04:00
|
|
|
: find-first-name ( str vector -- i/f tag/f )
|
|
|
|
|
[ >lower ] dip [ name>> = ] with find ; inline
|
|
|
|
|
|
|
|
|
|
: find-matching-close ( str vector -- i/f tag/f )
|
|
|
|
|
[ >lower ] dip
|
|
|
|
|
[ [ name>> = ] [ closing?>> ] bi and ] with find ; inline
|
|
|
|
|
|
|
|
|
|
: find-between* ( i/f tag/f vector -- vector )
|
|
|
|
|
pick integer? [
|
|
|
|
|
rot tail-slice
|
|
|
|
|
>r name>> r>
|
|
|
|
|
[ find-matching-close drop dup [ 1+ ] when ] keep
|
|
|
|
|
swap [ head ] [ first ] if*
|
2008-03-06 18:09:48 -05:00
|
|
|
] [
|
2008-09-25 03:02:47 -04:00
|
|
|
3drop V{ } clone
|
2008-05-24 00:34:10 -04:00
|
|
|
] if ; inline
|
2008-09-25 03:02:47 -04:00
|
|
|
|
|
|
|
|
: find-between ( i/f tag/f vector -- vector )
|
|
|
|
|
find-between* dup length 3 >= [
|
|
|
|
|
[ rest-slice but-last-slice ] keep like
|
|
|
|
|
] when ; inline
|
2008-03-06 18:09:48 -05:00
|
|
|
|
2008-09-25 03:02:47 -04:00
|
|
|
: find-between-first ( string vector -- vector' )
|
|
|
|
|
[ find-first-name ] keep find-between ; inline
|
|
|
|
|
|
|
|
|
|
: find-between-all ( vector quot -- seq )
|
|
|
|
|
[ [ [ closing?>> not ] bi and ] curry find-all ] curry
|
|
|
|
|
[ [ >r first2 r> find-between* ] curry map ] bi ; inline
|
2008-03-06 18:09:48 -05:00
|
|
|
|
|
|
|
|
|
2007-11-29 12:18:46 -05:00
|
|
|
: remove-blank-text ( vector -- vector' )
|
2007-09-20 18:09:08 -04:00
|
|
|
[
|
2008-09-25 03:02:47 -04:00
|
|
|
dup name>> text =
|
|
|
|
|
[ text>> [ blank? ] all? not ] [ drop t ] if
|
2008-04-26 00:17:08 -04:00
|
|
|
] filter ;
|
2007-09-20 18:09:08 -04:00
|
|
|
|
2007-11-29 12:18:46 -05:00
|
|
|
: trim-text ( vector -- vector' )
|
|
|
|
|
[
|
2008-09-25 03:02:47 -04:00
|
|
|
dup name>> text =
|
|
|
|
|
[ [ [ blank? ] trim ] change-text ] when
|
2007-11-29 12:18:46 -05:00
|
|
|
] map ;
|
|
|
|
|
|
2007-09-20 18:09:08 -04:00
|
|
|
: find-by-id ( id vector -- vector )
|
2008-05-19 19:58:56 -04:00
|
|
|
[ attributes>> "id" swap at = ] with filter ;
|
2007-09-20 18:09:08 -04:00
|
|
|
|
|
|
|
|
: find-by-class ( id vector -- vector )
|
2008-05-19 19:58:56 -04:00
|
|
|
[ attributes>> "class" swap at = ] with filter ;
|
2007-09-20 18:09:08 -04:00
|
|
|
|
|
|
|
|
: find-by-name ( str vector -- vector )
|
2008-09-25 03:02:47 -04:00
|
|
|
[ >lower ] dip [ name>> = ] with filter ;
|
2007-09-20 18:09:08 -04:00
|
|
|
|
|
|
|
|
: find-by-attribute-key ( key vector -- vector )
|
2008-09-25 03:02:47 -04:00
|
|
|
[ >lower ] dip
|
2008-05-19 19:58:56 -04:00
|
|
|
[ attributes>> at ] with filter
|
2008-05-14 00:36:55 -04:00
|
|
|
sift ;
|
2007-09-20 18:09:08 -04:00
|
|
|
|
|
|
|
|
: find-by-attribute-key-value ( value key vector -- vector )
|
2008-09-25 03:02:47 -04:00
|
|
|
[ >lower ] dip
|
2008-05-19 19:58:56 -04:00
|
|
|
[ attributes>> at over = ] with filter nip
|
2008-05-14 00:36:55 -04:00
|
|
|
sift ;
|
2007-09-20 18:09:08 -04:00
|
|
|
|
|
|
|
|
: find-first-attribute-key-value ( value key vector -- i/f tag/f )
|
2008-09-25 03:02:47 -04:00
|
|
|
[ >lower ] dip
|
2008-05-19 19:58:56 -04:00
|
|
|
[ attributes>> at over = ] with find rot drop ;
|
2007-09-20 18:09:08 -04:00
|
|
|
|
2008-03-06 18:09:48 -05:00
|
|
|
: tag-link ( tag -- link/f )
|
2008-05-19 19:58:56 -04:00
|
|
|
attributes>> [ "href" swap at ] [ f ] if* ;
|
2007-09-20 18:09:08 -04:00
|
|
|
|
2008-05-19 19:58:56 -04:00
|
|
|
: find-links ( vector -- vector' )
|
|
|
|
|
[ [ name>> "a" = ] [ attributes>> "href" swap at ] bi and ]
|
|
|
|
|
find-between-all ;
|
2007-09-20 18:09:08 -04:00
|
|
|
|
2008-05-24 00:34:10 -04:00
|
|
|
: <link> ( vector -- link )
|
|
|
|
|
[ first attributes>> ]
|
|
|
|
|
[ [ name>> { text "img" } member? ] filter ] bi
|
|
|
|
|
link boa ;
|
|
|
|
|
|
2008-05-19 19:58:56 -04:00
|
|
|
: link. ( vector -- )
|
2008-05-24 00:34:10 -04:00
|
|
|
[ attributes>> "href" swap at write nl ]
|
|
|
|
|
[ clickable>> [ bl bl text>> print ] each nl ] bi ;
|
2007-09-20 18:09:08 -04:00
|
|
|
|
2008-03-06 18:09:48 -05:00
|
|
|
: find-by-text ( seq quot -- tag )
|
2008-05-19 19:58:56 -04:00
|
|
|
[ dup name>> text = ] prepose find drop ;
|
2007-09-20 18:09:08 -04:00
|
|
|
|
2008-02-11 17:27:40 -05:00
|
|
|
: find-opening-tags-by-name ( name seq -- seq )
|
2008-09-25 03:02:47 -04:00
|
|
|
[ [ name>> = ] [ closing?>> not ] bi and ] with find-all ;
|
2007-09-20 18:09:08 -04:00
|
|
|
|
2008-02-11 17:27:40 -05:00
|
|
|
: href-contains? ( str tag -- ? )
|
2008-05-19 19:58:56 -04:00
|
|
|
attributes>> "href" swap at* [ subseq? ] [ 2drop f ] if ;
|
2007-09-20 18:09:08 -04:00
|
|
|
|
2008-08-14 00:09:43 -04:00
|
|
|
: find-hrefs ( vector -- vector' )
|
|
|
|
|
find-links
|
|
|
|
|
[ [
|
|
|
|
|
[ name>> "a" = ]
|
|
|
|
|
[ attributes>> "href" swap key? ] bi and ] filter
|
|
|
|
|
] map sift [ [ attributes>> "href" swap at ] map ] map concat ;
|
2008-02-11 18:19:27 -05:00
|
|
|
|
2008-05-19 19:58:56 -04:00
|
|
|
: find-forms ( vector -- vector' )
|
|
|
|
|
"form" over find-opening-tags-by-name
|
2008-05-24 00:34:10 -04:00
|
|
|
swap [ >r first2 r> find-between* ] curry map
|
2008-05-19 19:58:56 -04:00
|
|
|
[ [ name>> { "form" "input" } member? ] filter ] map ;
|
|
|
|
|
|
|
|
|
|
: find-html-objects ( string vector -- vector' )
|
2008-05-24 00:34:10 -04:00
|
|
|
[ find-opening-tags-by-name ] keep
|
2008-09-25 03:02:47 -04:00
|
|
|
[ [ first2 ] dip find-between* ] curry map ;
|
2008-05-19 19:58:56 -04:00
|
|
|
|
|
|
|
|
: form-action ( vector -- string )
|
|
|
|
|
[ name>> "form" = ] find nip
|
|
|
|
|
attributes>> "action" swap at ;
|
2007-09-20 18:09:08 -04:00
|
|
|
|
2008-05-19 19:58:56 -04:00
|
|
|
: hidden-form-values ( vector -- strings )
|
|
|
|
|
[ attributes>> "type" swap at "hidden" = ] filter ;
|
|
|
|
|
|
|
|
|
|
: input. ( tag -- )
|
|
|
|
|
dup name>> print
|
|
|
|
|
attributes>>
|
|
|
|
|
[ bl bl bl bl [ write "=" write ] [ write bl ] bi* nl ] assoc-each ;
|
|
|
|
|
|
|
|
|
|
: form. ( vector -- )
|
|
|
|
|
[ closing?>> not ] filter
|
|
|
|
|
[
|
|
|
|
|
{
|
|
|
|
|
{ [ dup name>> "form" = ]
|
2008-08-17 11:38:34 -04:00
|
|
|
[ "form action: " write attributes>> "action" swap at print ] }
|
2008-05-19 19:58:56 -04:00
|
|
|
{ [ dup name>> "input" = ] [ input. ] }
|
|
|
|
|
[ drop ]
|
|
|
|
|
} cond
|
|
|
|
|
] each ;
|
|
|
|
|
|
|
|
|
|
: query>assoc* ( str -- hash )
|
|
|
|
|
"?" split1 nip query>assoc ;
|