line |
stmt |
bran |
cond |
sub |
pod |
time |
code |
1
|
|
|
|
|
|
|
package NewsExtractor::SiteSpecificExtractor::news_tnn_tw; |
2
|
1
|
|
|
1
|
|
8
|
use utf8; |
|
1
|
|
|
|
|
2
|
|
|
1
|
|
|
|
|
8
|
|
3
|
1
|
|
|
1
|
|
37
|
use Moo; |
|
1
|
|
|
|
|
3
|
|
|
1
|
|
|
|
|
7
|
|
4
|
|
|
|
|
|
|
extends 'NewsExtractor::GenericExtractor'; |
5
|
|
|
|
|
|
|
|
6
|
1
|
|
|
1
|
|
348
|
use Importer 'NewsExtractor::TextUtil' => 'reformat_dateline'; |
|
1
|
|
|
|
|
3
|
|
|
1
|
|
|
|
|
9
|
|
7
|
|
|
|
|
|
|
|
8
|
|
|
|
|
|
|
sub dateline { |
9
|
0
|
|
|
0
|
0
|
|
my ($self) = @_; |
10
|
0
|
|
|
|
|
|
my $el = $self->dom->at("span.f12_15a_g2"); |
11
|
0
|
|
|
|
|
|
return reformat_dateline( $el->text(), '+08:00' ); |
12
|
|
|
|
|
|
|
} |
13
|
|
|
|
|
|
|
|
14
|
|
|
|
|
|
|
sub journalist { |
15
|
0
|
|
|
0
|
0
|
|
my ($self) = @_; |
16
|
0
|
|
|
|
|
|
my $txt = $self->content_text; |
17
|
|
|
|
|
|
|
|
18
|
0
|
|
|
|
|
|
my ($x) = $txt =~ m{[〔﹝【]記者 \s* ([ \p{Letter}]+?) \s* (?:[//]\p{Letter}+)?報導[﹞〕】]}x; |
19
|
0
|
|
|
|
|
|
return $x; |
20
|
|
|
|
|
|
|
} |
21
|
|
|
|
|
|
|
|
22
|
|
|
|
|
|
|
1; |