File Coverage

blib/lib/NewsExtractor/SiteSpecificExtractor/www_mdnkids_com.pm

Criterion	Covered	Total	%
statement	9	27	33.3
branch	0	8	0.0
condition			n/a
subroutine	3	5	60.0
pod	0	2	0.0
total	12	42	28.5

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

use Importer 'NewsExtractor::TextUtil' => qw(reformat_dateline normalize_whitespace);

    my ($self) = @_;
    my $el = $self->dom->at('h2 ~ div.col span');
    if ($el) {
        my ($ymd) = $el->all_text() =~ m<\(([0-9]{4}/[0-9]{1,2}/[0-9]{1,2})\)\z>;
        return reformat_dateline( $ymd, '+08:00' );
    }
    return undef;
}

    my ($self) = @_;
    my ($x, $el);
    if ($el = $self->dom->at('h2 ~ div.col span')) {
        $x = $el->all_text() =~ s<\([0-9]{4}/[0-9]{1,2}/[0-9]{1,2}\)\z><>r;
        $x = normalize_whitespace($x);
    }
    unless ($x) {
        my $t = $self->content_text;

        my @regexps = (
            qr{\A(\S+)／\S+報導\n},
            qr{\A\S*(報導／\S+\s+攝影／\S+)\n},
            qr{\A\S*(文／\S+\s+圖／\S+)\n},
            qr{\A\S*(\S{3})\n},
        );

        for my $re (@regexps) {
            ($x) = $t =~ /$re/;
            last if $x;
        }
    }

    return $x;
}

1;