File Coverage

blib/lib/HTML/AutoPagerize.pm

Criterion	Covered	Total	%
statement	15	49	30.6
branch	0	14	0.0
condition			n/a
subroutine	5	11	45.4
pod	0	6	0.0
total	20	80	25.0

package HTML::AutoPagerize;

use strict;
use 5.8.1;
our $VERSION = '0.02';

use Carp;
use HTML::TreeBuilder::XPath;
use URI;

sub new {
    my $class = shift;
    bless { sites => [] }, $class;
}

sub sites {
    my $self = shift;
    $self->{sites} = shift if @_;
    $self->{sites};
}

sub sorted_sites {
    my $self = shift;
    return [ sort { length $b->{url} <=> length $a->{url} } @{ $self->sites } ];
}

sub add_site {
    my($self, %site) = @_;

    for my $key (qw( url nextLink )) {
        unless (defined $site{$key}) {
            croak "key '$key' needed for SITEINFO";
        }
    }

    $site{url} = qr/$site{url}/; # compile the regexp
    push @{$self->{sites}}, \%site;
}

sub handle {
    my($self, $uri, $html) = @_;

    my $siteinfo = $self->site_info_for($uri) or return;

    my $tree = HTML::TreeBuilder::XPath->new;
    $tree->parse($html);

    my $res;

    my $next_link = $siteinfo->{nextLink};
    if (my $nodes = $tree->findnodes($next_link)) {
        $res->{next_link} = URI->new_abs($nodes->shift->attr('href'), $uri);
    }

    if (my $page_element = $siteinfo->{pageElement}) {
        if (my $nodes = $tree->findnodes($page_element)) {
            $res->{page_element} = $nodes;
        }
    }

    return $res;
}

sub site_info_for {
    my($self, $uri) = @_;

    for my $site (@{ $self->sorted_sites }) {
        if ($uri =~ $site->{url}) {
            return $site;
        }
    }

    return;
}

1;
__END__