| line | stmt | bran | cond | sub | pod | time | code | 
| 1 |  |  |  |  |  |  | use utf8; | 
| 2 | 1 |  |  | 1 |  | 20 | use Moo; | 
|  | 1 |  |  |  |  | 2 |  | 
|  | 1 |  |  |  |  | 5 |  | 
| 3 | 1 |  |  | 1 |  | 27 | extends 'NewsExtractor::JSONLDExtractor'; | 
|  | 1 |  |  |  |  | 3 |  | 
|  | 1 |  |  |  |  | 4 |  | 
| 4 |  |  |  |  |  |  |  | 
| 5 |  |  |  |  |  |  | use Importer 'NewsExtractor::TextUtil'  => qw( html2text ); | 
| 6 | 1 |  |  | 1 |  | 249 |  | 
|  | 1 |  |  |  |  | 3 |  | 
|  | 1 |  |  |  |  | 6 |  | 
| 7 |  |  |  |  |  |  | my ($self) = @_; | 
| 8 |  |  |  |  |  |  | # my $text = $self->schema_ld->{articleBody} // $self->schema_ld->{description} // ''; | 
| 9 | 0 |  |  | 0 | 0 |  | my $body = $self->tx->result->dom->at("div.article-body"); | 
| 10 |  |  |  |  |  |  | $body->find(".article-hash-tag")->map('remove'); | 
| 11 | 0 |  |  |  |  |  | my $text = html2text( $body->content ); | 
| 12 | 0 |  |  |  |  |  | return $text; | 
| 13 | 0 |  |  |  |  |  | } | 
| 14 | 0 |  |  |  |  |  |  | 
| 15 |  |  |  |  |  |  | 1; |