--- lib/Grep/Source.pm	2007/02/24 12:32:31	100
+++ lib/Grep/Source.pm	2007/04/28 22:52:08	125
@@ -8,7 +8,7 @@
 use Carp qw/verbose/;
 use Module::Pluggable search_path => 'Grep::Source', sub_name => 'sources', require => 1;
 use base qw(Class::Accessor Jifty::Object);
-Grep::Source->mk_accessors( qw(feed uri q new_items collection) );
+Grep::Source->mk_accessors( qw(feed uri q new_items collection search_obj tree) );
 
 use HTML::TreeBuilder;
 use WWW::Mechanize;
@@ -108,11 +108,12 @@
 	my $class = $self->feed->source || 'Grep::Source::Feed';
 	$self->log->debug("using $class");
 
-	my $parent = $self;
-	$class->fetch( $parent );
-	undef $parent;
+	$self->search_obj( Grep::Search->new() );
+	$self->log->debug("created " . $self->search_obj);
 
-	Grep::Search->finish if $self->new_items;
+	$class->fetch( $self );
+
+	$self->search_obj->finish;
 
 	return $self->collection;
 }
@@ -131,11 +132,13 @@
 sub add_record {
 	my $self = shift;
 
+	$self->log->confess("no search_obj") unless ($self->search_obj);
+
 	my $i = Grep::Model::Item->new();
 
 	my $rec = {@_};
 
-	warn "resolving links";
+	$self->log->debug("resolving links using base ", $rec->{link});
 	my $resolver = HTML::ResolveLink->new( base => $rec->{link} );
 	$rec->{content} = $resolver->resolve( $rec->{content} );
 
@@ -149,7 +152,7 @@
 
 		# is new record?
 		if ( $msg !~ m/^Found/ ) {
-			Grep::Search->add( $i );
+			$self->search_obj->add( $i );
 			$self->new_items( ( $self->new_items || 0 ) + 1 );
 		}
 	} else {
@@ -192,12 +195,65 @@
 
 =cut
 
+sub element_by_triplet {
+	my $self = shift;
+
+	my $args = {@_};
+
+	my $tree = $args->{tree} || die "no tree";
+	my $message = $args->{message} || '';
+	my $fatal = $args->{fatal};
+	die "no templates" unless defined( $args->{templates} );
+	my @templates;
+warn dump( $args->{templates} );
+	if ( ref( $args->{templates} ) eq 'ARRAY' ) {
+		@templates = @{ $args->{templates} };
+	} else {
+		@templates = ( $args->{templates} );
+	}
+
+	push @templates, ( undef, undef ) if ( $#templates == 0 );
+
+	die "wrapper doesn't have 3 elements but ", $#templates unless (
+		( $#templates + 1 ) % 3 == 0
+	);
+
+	my ( $el, $attr, $value );
+
+	my @results;
+	my @tags;
+
+	while ( @templates ) {
+		( $el,$attr,$value ) = splice( @templates, 0, 3 );
+		my $tag = $attr ? "<$el $attr=\"$value\">" : "<$el>";
+		push @tags, $tag;
+		$self->log->debug("looking for $message $tag");
+		@results = $tree->look_down( '_tag', $el, sub {
+				return 1 unless ( $attr && $value );
+				( $_[0]->attr( $attr ) || '' ) eq $value;
+		});
+		last if @results;
+	}
+
+	if ( ! @results ) {
+		my $msg = "can't find $message ", join(" ", @tags);
+		die $msg if ( $fatal );
+		warn $msg;
+		return;
+	}
+
+	$self->log->debug("found ", $#results + 1, " results");
+
+	return @results if wantarray;
+	return shift @results;
+}
+
 sub scrape {
 	my $self = shift;
 
 	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);
@@ -219,38 +275,36 @@
 	$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";
 
-	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\">";
-
-	my $div = $tree->look_down( '_tag', $el, sub {
-			warn dump( $_[0]->attr( $attr ) ),$/;
-			( $_[0]->attr( $attr ) || '' ) eq $value;
-	});
-
-	if ( ! $div ) {
-		warn "can't find results wrapper <$el $attr=\"$value\">";
-		return;
-	}
+	my $div = $self->element_by_triplet(
+		tree => $tree,
+		templates => $args->{wrapper},
+		message => 'wrapper for all results',
+		fatal => 1
+	);
 
-	my $max = 5;
+	my $max = 15;
 	my $nr = 1;
 
 	my $base_uri = $uri;
 	$base_uri =~ s!\?.*$!!;
 
-	foreach my $dt ( $div->look_down( '_tag', $args->{results} ) ) {
+	my @r = $self->element_by_triplet(
+		tree => $tree,
+		templates => $args->{results},
+		message => 'result element',
+	);
+
+	foreach my $dt ( @r ) {
 		my $a = $dt->look_down( '_tag', 'a', sub { $_[0]->attr('href') } );
 		if ( $a ) {
 
@@ -259,18 +313,18 @@
 			$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 );
 
 				my $page_tree = HTML::TreeBuilder->new or die "can't create page tree";
 				$page_tree->parse( $mech->content ) or die "can't parse page at $page_uri";
-
-				( $el,$attr,$value ) = @{ $args->{scrape} };
-				$div = $page_tree->look_down( '_tag', $el, sub { ( $_[0]->attr( $attr ) || '' ) eq $value } );
-
-				die "can't find <$el $attr=\"$value\">" unless ($div);
+				$div = $self->element_by_triplet(
+					tree => $page_tree,
+					message => "result $nr",
+					templates => $args->{scrape}
+				);
 
 				$self->add_record(
 					in_feed => $feed,
@@ -282,7 +336,7 @@
 #					author => 
 #					issued => 
 #					modified => 
-				);
+				) if ( $div );
 
 				$mech->back;
 				$page_tree->delete;
@@ -290,6 +344,8 @@
 			} else {
 				warn "can't follow uri $page_uri: $!\n";
 			}
+		} else {
+			$self->log->debug("result $nr doesn't have link inside, ignoring...");
 		}
 
 		last if ($nr == $max);