line |
stmt |
bran |
cond |
sub |
pod |
time |
code |
1
|
|
|
|
|
|
|
package NewsExtractor::SiteSpecificExtractor::www_taipeitimes_com; |
2
|
1
|
|
|
1
|
|
8
|
use utf8; |
|
1
|
|
|
|
|
3
|
|
|
1
|
|
|
|
|
6
|
|
3
|
1
|
|
|
1
|
|
35
|
use Moo; |
|
1
|
|
|
|
|
3
|
|
|
1
|
|
|
|
|
6
|
|
4
|
|
|
|
|
|
|
extends 'NewsExtractor::GenericExtractor'; |
5
|
|
|
|
|
|
|
|
6
|
1
|
|
|
1
|
|
330
|
use Importer 'NewsExtractor::TextUtil' => 'normalize_whitespace'; |
|
1
|
|
|
|
|
2
|
|
|
1
|
|
|
|
|
7
|
|
7
|
|
|
|
|
|
|
|
8
|
|
|
|
|
|
|
sub dateline { |
9
|
0
|
|
|
0
|
0
|
|
my ($self) = @_; |
10
|
0
|
|
|
|
|
|
my $el = $self->dom->at("meta[property='article:published_time']"); |
11
|
0
|
|
|
|
|
|
return "". $el->attr('content'); |
12
|
|
|
|
|
|
|
} |
13
|
|
|
|
|
|
|
|
14
|
|
|
|
|
|
|
sub journalist { |
15
|
0
|
|
|
0
|
0
|
|
my ($self) = @_; |
16
|
0
|
|
|
|
|
|
my $el = $self->dom->at(".name"); |
17
|
0
|
|
|
|
|
|
my $txt = $el->text; |
18
|
0
|
|
|
|
|
|
return normalize_whitespace($txt); |
19
|
|
|
|
|
|
|
} |
20
|
|
|
|
|
|
|
|
21
|
|
|
|
|
|
|
1; |