File Coverage

blib/lib/NewsExtractor/SiteSpecificExtractor/www_ntdtv_com.pm

Criterion	Covered	Total	%
statement	9	23	39.1
branch	0	2	0.0
condition			n/a
subroutine	3	5	60.0
pod	0	2	0.0
total	12	32	37.5

use utf8;
use Moo;
extends 'NewsExtractor::GenericExtractor';
use Importer 'NewsExtractor::TextUtil'  => qw( normalize_whitespace );

    my ($self) = @_;
    my $el = $self->dom->at("div[itemprop=articleBody]");
    $el->find("div.print_link, div.single_ad")->map("remove");
    my $last_child = $el->children->last;
    if ($last_child->all_text =~ /相關鏈接：/) {
        $last_child->remove;
    }
    my $txt = $el->all_text;
    $txt = normalize_whitespace($txt);
    return $txt;
}

    my ($self) = @_;
    my ($name) = $self->content_text =~ m{
                \n
                (
                        新唐人亞太電視\p{Letter}+?報導
                        | ( 新唐人記者[\p{Letter}、]+報導 )
                        | ( 新唐人\p{Letter}+記者站綜合報導 )
                        | （\s* 記者\p{Letter}+報導/責任編輯：\p{Letter}+ \s*）
                        | （\s* 責任編輯：\p{Letter}+ \s*）
                        | （轉自\p{Letter}+/責任編輯：\p{Letter}+ \s*）
                        | ( 採訪/\p{Letter}+ \s+ 編輯/\p{Letter}+ \s+ 後製/\p{Letter}+ )
                )
                \z}xs;

    $name =~ s/\A（//;
    $name =~ s/）\z//;

    return $name;
}

1;