File Coverage

blib/lib/NewsExtractor/SiteSpecificExtractor/www_bcc_com_tw.pm

Criterion	Covered	Total	%
statement	9	29	31.0
branch	0	8	0.0
condition			n/a
subroutine	3	7	42.8
pod	0	4	0.0
total	12	48	25.0

use utf8;
use Moo;
extends 'NewsExtractor::SiteSpecificExtractor';

use Importer 'NewsExtractor::TextUtil' => 'html2text';

    my ($self) = @_;
    my $el = $self->dom->at('head > title');
    return $el->all_text;
}

    my ($self) = @_;
    my $el = $self->dom->at('div.tt27') or return undef;
    my $txt = $el->all_text;
    $txt =~ s/\s+報導\z//;
    return $txt;
}

    my ($self) = @_;
    my $content = $self->content_text or return undef;
    my ($o) = $content =~ m{。 [（\(] (\p{Letter}+?) 報導 [）\)] \n\n}x;
    unless ($o) {
        ($o) = $content =~ m{。 [（\(] 中廣記者 (\p{Letter}+?) [）\)] \z}x;
    }
    return $o;
}

    my ($self) = @_;
    my $el;

    for $el ($self->dom->find('script, div.ft')->each) {
        $el->remove();
    }

    $el = $self->dom->at('#some-class-name') or return undef;
    return html2text( $el->to_string );
}

1;