File Coverage

blib/lib/WWW/Search/GalerieIlonkaCZ.pm

Criterion	Covered	Total	%
statement	21	51	41.1
branch	0	4	0.0
condition			n/a
subroutine	7	11	63.6
pod	2	2	100.0
total	30	68	44.1

package WWW::Search::GalerieIlonkaCZ;

# Pragmas.
use base qw(WWW::Search);
use strict;
use warnings;

# Modules.
use Encode qw(decode_utf8);
use LWP::UserAgent;
use Readonly;
use Web::Scraper;

# Constants.
Readonly::Scalar our $MAINTAINER => 'Michal Spacek ';
Readonly::Scalar my $GILONKA_CZ => 'http://www.galerie-ilonka.cz';
Readonly::Scalar my $GILONKA_CZ_ACTION1 => '/galerie-ilonka/0/0/3/42/9/0/?hledatjak=2';

# Version.
our $VERSION = 0.01;

# Setup.
sub native_setup_search {
	my ($self, $query) = @_;
	$self->{'_def'} = scraper {
		process '//div[@id="incenterpage2"]/div'.
			'/div[@class="productBody"]', 'books[]' => scraper {

			process '//div[@class="productTitle"]/div/a',
				'title' => 'TEXT';
# XXX Jak vybrat podtitul, kdyz tam je. Treba pri vyhledavani exlibris
#			process '//div[@class="productTitle"]/div',
#				'subtitle' => 'RAW';
			process '//div[@class="img_box"]/a', 'url' => '@href';
			process '//div[@class="productText"]',
				'description' => 'TEXT';
# XXX Cannot parse . Breaks everything after this.
			process '//div[@class="productPriceBox"]'.
				'/div[@class="productPrice"]'.
				'/span[@itemprop="price"]', 'price' => 'TEXT';
			process '//div[@class="img_box"]/a/img',
				'cover_url' => '@src';

			return;
		};
		return;
	};
	$self->{'_query'} = $query;
	return 1;
}

# Get data.
sub native_retrieve_some {
	my $self = shift;

	# Query.
	my $query = decode_utf8($self->{'_query'});

	# Get content.
	my $ua = LWP::UserAgent->new(
		'agent' => "WWW::Search::GalerieIlonkaCZ/$VERSION",
	);
	my $response = $ua->get($GILONKA_CZ.$GILONKA_CZ_ACTION1."&slovo=$query");

	# Process.
	if ($response->is_success) {
		my $content = $response->content;

		# Get books structure.
		my $books_hr = $self->{'_def'}->scrape($content);

		# Process each book.
		foreach my $book_hr (@{$books_hr->{'books'}}) {
			_fix_url($book_hr, 'url');
			_fix_url($book_hr, 'cover_url');
			push @{$self->{'cache'}}, $book_hr;
		}
	}

	return;
}

# Fix URL to absolute path.
sub _fix_url {
	my ($book_hr, $url) = @_;
	if (exists $book_hr->{$url}) {
		$book_hr->{$url} = $GILONKA_CZ.$book_hr->{$url};
	}
	return;
}

1;

__END__