File Coverage

lib/WWW/Mixi/Scraper/Plugin/ListDiary.pm

Criterion	Covered	Total	%
statement	9	34	26.4
branch	0	4	0.0
condition	0	2	0.0
subroutine	3	7	42.8
pod	1	1	100.0
total	13	48	27.0

package WWW::Mixi::Scraper::Plugin::ListDiary;

use strict;
use warnings;
use WWW::Mixi::Scraper::Plugin;

validator {qw(
  id     is_number
  page   is_number
  year   is_number
  month  is_number
)};

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

  my %scraper;
  $scraper{meta} = scraper {
    process 'a',
      text => 'TEXT',
      href => '@href';
    result qw( text href );
  };

  $scraper{diaries} = scraper {
    process 'div.listDiaryTitle>dl>dd',
      time => 'TEXT';
    process 'div.listDiaryTitle>dl>dt>a',
      link   => '@href',
      subject => 'TEXT';
    process 'p',
      description => 'TEXT';
    process 'div.diaryPhoto>a>img',
      'images[]' => '@src';
    process 'div.diaryEditMenu>ul>li',
      'meta[]' => $scraper{meta};
    result qw( time link subject description images meta );
  };

  $scraper{list} = scraper {
    process 'div.listDiaryBlock,div.listDiaryBlockLast',
      'diaries[]' => $scraper{diaries};
    result qw( diaries );
  };

  my $stash = $self->post_process($scraper{list}->scrape(\$html));

  foreach my $diary ( @{ $stash } ) {
    my $meta = delete $diary->{meta};
    foreach my $item ( @{ $meta || [] } ) {
      if ( ($item->{href} || '') =~ /#(?:write|comment)$/ ) {
        my ($count) = $item->{text} =~ /(\d+)/;
        $diary->{count} = $count;
      }
    }
  }

  return $stash;
}

1;

__END__