File Coverage

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

Criterion	Covered	Total	%
statement	9	40	22.5
branch			n/a
condition	0	2	0.0
subroutine	3	8	37.5
pod	1	1	100.0
total	13	51	25.4

package WWW::Mixi::Scraper::Plugin::ViewDiary;

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

validator {qw(
  id        is_number
  owner_id  is_number
)};

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

  my %scraper;
  $scraper{images} = scraper {
    process 'a',
      link => '@onClick';
    process 'a>img',
      thumb_link => '@src';
    result qw( link thumb_link );
  };

  $scraper{diary} = scraper {
    process 'div.viewDiaryBox>div.listDiaryTitle>dl>dd',
      time => 'TEXT';
    process 'div.viewDiaryBox>div.listDiaryTitle>dl>dt',
      subject => 'TEXT';
    process 'div.viewDiaryBox>div.listDiaryTitle>dl>dt>span',
      string => 'TEXT';
    process 'div#diary_body',
      description => $self->html_or_text;
    process 'div.diaryPhoto>table>tr>td',
      'images[]' => $scraper{images};
    process 'div.diaryPaging01>div.diaryPagingLeft>a',
      prev_link => '@href';
    result qw( time subject description images prev_link string );
  };

  my $stash = $scraper{diary}->scrape(\$html);
  $stash->{link} = delete $stash->{prev_link};
  $stash->{link} =~ s/neighbor_diary/view_diary/;
  $stash->{link} =~ s/&direction=prev.*//;
  $stash = $self->post_process($stash)->[0];

  my $string = delete $stash->{string} || '';
  $stash->{subject} =~ s/$string$//;

  $scraper{comments} = scraper {
    process 'dl.comment>dt>span.date',
      time => 'TEXT';
    process 'dl.comment>dt>a',
      link => '@href',
      name => 'TEXT';
    process 'dl.comment>dd',
      description => $self->html_or_text;
    result qw( time link name description );
  };

  $scraper{list} = scraper {
    process 'div.commentListArea>ul>li',
      'comments[]' => $scraper{comments};
    result qw( comments );
  };

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

  return $stash;
}

1;

__END__