File Coverage

blib/lib/NewsExtractor/JSONLDExtractor.pm

Criterion	Covered	Total	%
statement	15	32	46.8
branch	0	6	0.0
condition	0	5	0.0
subroutine	5	10	50.0
pod	0	4	0.0
total	20	57	35.0

use Moo;
extends 'NewsExtractor::TXExtractor';

use Mojo::Transaction::HTTP;
use Types::Standard qw( InstanceOf HashRef ArrayRef );
use Mojo::JSON qw(from_json);
use Importer 'NewsExtractor::TextUtil' => qw(u remove_control_characters);

has tx => (
    required => 1, is => 'ro',
    isa => InstanceOf['Mojo::Transaction::HTTP'] );

has schema_ld => (
    required => 0,
    is => 'lazy',
    isa => HashRef,
    builder => 1,
);

    my ($self) = @_;
    my $el = $self->dom->at('script[type="application/ld+json"]') or return {};
    my $x = from_json( $el->text );
    if (HashRef->check($x)) {
        return $x;
    }
    if (ArrayRef->check($x)) {
        return $x->[0];
    }
    return {};
}

    my ($self) = @_;
    return remove_control_characters(u($self->schema_ld->{author}{name}));
}

    my ($self) = @_;
    return remove_control_characters(u($self->schema_ld->{headline}));
}

    my ($self) = @_;
    return remove_control_characters(u($self->schema_ld->{datePublished}));
}

    my ($self) = @_;
    my $text = $self->schema_ld->{articleBody} // $self->schema_ld->{description} // '';
    return remove_control_characters(u($text));
}

1;