WWW::Scraper::ISBN::TWCwbook_Driver - Search driver for TWCwbook's online catalog.

# ex:ts=8

package WWW::Scraper::ISBN::TWCwbook_Driver;

use strict;
use warnings;

use vars qw($VERSION @ISA);
$VERSION = '0.02';

#--------------------------------------------------------------------------

#--------------------------------------------------------------------------

###########################################################################
#Library Modules                                                          #
###########################################################################

use WWW::Scraper::ISBN::Driver;
use WWW::Mechanize;
use Template::Extract;
use Text::Iconv;

###########################################################################
#Constants                                                                #
###########################################################################

use constant	QUERY	=> "http://www.cwbook.com.tw/search/result1.jsp?field=2&keyWord=%s";

#--------------------------------------------------------------------------

###########################################################################
#Inheritence                                                              #
###########################################################################

@ISA = qw(WWW::Scraper::ISBN::Driver);

###########################################################################
#Interface Functions                                                      #
###########################################################################

sub search {
	my $self = shift;
	my $isbn = shift;
	$self->found(0);
	$self->book(undef);

	my $url = sprintf(QUERY, $isbn);
	my $mechanize = WWW::Mechanize->new();
	$mechanize->get($url);
	return undef unless($mechanize->success());

	my $template = <<END;
<input type="checkbox" name="productID" value="[% bookid %]">
END

	my $extract = Template::Extract->new;
	my $data = $extract->extract($template, $mechanize->content());

	return $self->handler("Could not extract data from TWCwbook result page.")
		unless(defined $data);

	$url = "http://www.cwbook.com.tw/common/book.jsp?productID=$data->{bookid}";
	$mechanize->get($url);
	return undef unless($mechanize->success());

	$template = <<END;
<div id="main" class="product">[% ... %]
<h2>[% title %]</h2>[% ... %]
<div class="block"><img src="[% image_link %]" ></div>[% ... %]
<li class="author">[% author %]</li>[% ... %]
<li class="publisher">[% publisher %]</li>[% ... %]
<li class="pubdate">[% pubdate %]</li>[% ... %]
<li class="price1">[% price_list %]</li>
END

	$data = $extract->extract($template, $mechanize->content());
	return $self->handler("Could not extract data from TWCwbook result page.")
		unless(defined $data);

	my $conv = Text::Iconv->new("utf-8", "big5");
	$data->{title} = $conv->convert($data->{title});
	$data->{title} =~ s/[ \r\t]//g;
	$data->{author} = $conv->convert($data->{author});
	$data->{author} =~ s/§@ªÌ¡G(.*)/$1/;
	$data->{publisher} = $conv->convert($data->{publisher});
	$data->{publisher} =~ s/¥Xª©ªÀ¡G(.*)/$1/;
	$data->{pubdate} =~ s/\D*(\d+\/\d+\/\d+).*/$1/;
	$data->{price_list} =~ s/\D+(\d+).*/$1/;

	my $bk = {
		'isbn'		=> $isbn,
		'title'		=> $data->{title},
		'author'	=> $data->{author},
		'book_link'	=> $url,
		'image_link'	=> "http://www.cwbook.com.tw".$data->{image_link},
		'pubdate'	=> $data->{pubdate},
		'publisher'	=> $data->{publisher},
		'price_list'	=> $data->{price_list},
	};

	$self->book($bk);
	$self->found(1);
	return $self->book;
}

1;
__END__