line |
stmt |
bran |
cond |
sub |
pod |
time |
code |
1
|
|
|
|
|
|
|
package NewsExtractor::SiteSpecificExtractor::newtalk_tw; |
2
|
1
|
|
|
1
|
|
8
|
use utf8; |
|
1
|
|
|
|
|
3
|
|
|
1
|
|
|
|
|
6
|
|
3
|
1
|
|
|
1
|
|
33
|
use Moo; |
|
1
|
|
|
|
|
3
|
|
|
1
|
|
|
|
|
6
|
|
4
|
|
|
|
|
|
|
extends 'NewsExtractor::GenericExtractor'; |
5
|
|
|
|
|
|
|
|
6
|
1
|
|
|
1
|
|
336
|
use Importer 'NewsExtractor::TextUtil' => 'u'; |
|
1
|
|
|
|
|
3
|
|
|
1
|
|
|
|
|
5
|
|
7
|
|
|
|
|
|
|
|
8
|
|
|
|
|
|
|
sub dateline { |
9
|
0
|
|
|
0
|
0
|
|
my ($self) = @_; |
10
|
0
|
|
|
|
|
|
my ($el, $dateline); |
11
|
0
|
0
|
|
|
|
|
if ($el = $self->dom->at('meta[property="article:published_time"]')) { |
|
|
0
|
|
|
|
|
|
12
|
0
|
|
|
|
|
|
$dateline = $el->attr('content'); |
13
|
|
|
|
|
|
|
} elsif ($el = $self->dom->at('div.contentBox div.content_date')) { |
14
|
0
|
|
|
|
|
|
my @t = $el->all_text =~ m/([0-9]+)/g; |
15
|
0
|
|
|
|
|
|
$dateline = u(sprintf('%04d-%02d-%02dT%02d:%02d:%02d+08:00', |
16
|
|
|
|
|
|
|
$t[0], $t[1], $t[2], $t[3], $t[4], 59)); |
17
|
|
|
|
|
|
|
} |
18
|
0
|
|
|
|
|
|
return $dateline; |
19
|
|
|
|
|
|
|
} |
20
|
|
|
|
|
|
|
|
21
|
|
|
|
|
|
|
1; |