--- lib/Grep/Source.pm 2007/02/24 11:16:05 92 +++ lib/Grep/Source.pm 2007/03/14 18:46:37 109 @@ -7,13 +7,14 @@ use Carp qw/verbose/; use Module::Pluggable search_path => 'Grep::Source', sub_name => 'sources', require => 1; -use base qw(Class::Accessor); -Grep::Source->mk_accessors( qw(feed uri q new_items collection) ); +use base qw(Class::Accessor Jifty::Object); +Grep::Source->mk_accessors( qw(feed uri q new_items collection search) ); use HTML::TreeBuilder; use WWW::Mechanize; use XML::Feed; use URI; +use HTML::ResolveLink; use Data::Dump qw/dump/; @@ -100,16 +101,16 @@ $self->uri( $uri ); - Jifty->log->info( $message ); + $self->log->info( $message ); $self->collection( Grep::Model::ItemCollection->new() ); my $class = $self->feed->source || 'Grep::Source::Feed'; - Jifty->log->debug("using $class"); + $self->log->debug("using $class"); - $class->fetch( $self ); - - Grep::Search->finish if $self->new_items; + my $parent = $self; + $class->fetch( $parent ); + undef $parent; return $self->collection; } @@ -128,19 +129,27 @@ sub add_record { my $self = shift; + $self->search( Grep::Search->new() ) unless ($self->search); + my $i = Grep::Model::Item->new(); - my ($ok,$msg) = $i->load_or_create( @_ ); + my $rec = {@_}; + + $self->log->debug("resolving links using base ", $rec->{link}); + my $resolver = HTML::ResolveLink->new( base => $rec->{link} ); + $rec->{content} = $resolver->resolve( $rec->{content} ); + + my ($ok,$msg) = $i->load_or_create( %$rec ); $msg ||= ''; if ( $ok ) { - Jifty->log->debug("item ", $i->id, ": $msg"); + $self->log->debug("item ", $i->id, ": $msg"); $self->collection->add_record( $i ); # is new record? if ( $msg !~ m/^Found/ ) { - Grep::Search->add( $i ); + $search->add( $i ); $self->new_items( ( $self->new_items || 0 ) + 1 ); } } else { @@ -162,14 +171,14 @@ my $content = shift or die "no content?"; foreach my $s ( $self->sources ) { - Jifty->log->debug("testing source class $s"); + $self->log->debug("testing source class $s"); if ( $s->can('content_have') ) { my $regex = $s->content_have( $content ) or die "${s}->content_have didn't return anything"; die "${s}->content_have didn't return regex but ", dump( $regex ), " ref ", ref( $regex ) unless ( ref($regex) eq 'Regexp' ); if ( $content =~ $regex ) { - Jifty->log->debug("${s}->content_have succesful"); + $self->log->debug("${s}->content_have succesful"); return $s; } } @@ -188,7 +197,7 @@ my $args = {@_}; - warn "scrape got args ",dump($args); + $self->log->debug("scrape with args ",dump($args)); my ($feed,$uri,$q) = ($self->feed, $self->uri,$self->q); die "no uri" unless ($uri); @@ -210,12 +219,12 @@ $self->save( 'get.html', $mech->content ); if ( my $form = $args->{submit_form} ) { - warn "submit form on $uri with ", dump( $form ),"\n"; + $self->log->debug("submit form on $uri with ", dump( $form )); $mech->submit_form( %$form ) or die "can't submit form ", dump( $form ); $self->save( 'submit.html', $mech->content ); } - warn "parse result page\n"; + $self->log->debug("parse result page"); my $tree = HTML::TreeBuilder->new or die "can't create html tree"; $tree->parse( $mech->content ) or die "can't parse fetched content"; @@ -223,10 +232,9 @@ die "wrapper doesn't have 3 elements but ", $#{ $args->{wrapper} } unless ( $#{ $args->{wrapper} } == 2 ); my ( $el,$attr,$value ) = @{ $args->{wrapper} }; - warn "looking for <$el $attr=\"$value\">"; + $self->log->debug("looking for <$el $attr=\"$value\">"); my $div = $tree->look_down( '_tag', $el, sub { - warn dump( $_[0]->attr( $attr ) ),$/; ( $_[0]->attr( $attr ) || '' ) eq $value; }); @@ -235,7 +243,7 @@ return; } - my $max = 5; + my $max = 15; my $nr = 1; my $base_uri = $uri; @@ -250,7 +258,7 @@ $page_uri->query( undef ); $page_uri = $page_uri->canonical; - warn "fetching page: ",$a->as_text," from $page_uri\n"; + $self->log->debug("fetching page: ",$a->as_text," from $page_uri"); if ( $mech->follow_link( url => $a->attr('href') ) ) { $self->save( "page-${nr}.html", $mech->content ); @@ -302,11 +310,12 @@ sub save { my $self = shift; my ( $file, $content ) = @_; + return unless ( defined($file) && defined($content) ); if ( -w '/tmp/grep' ) { open(my $f, '>', "/tmp/grep/$file") or die "can't open $file: $!"; print $f $content or die "can't write to $file: $!"; close $f or die "can't close $file: $!"; - Jifty->log->debug("saved $file ",length($content)," bytes"); + $self->log->debug("saved $file ",length($content)," bytes"); } }