line |
stmt |
bran |
cond |
sub |
pod |
time |
code |
1
|
|
|
|
|
|
|
package NewsExtractor::SiteSpecificExtractor::www_digitimes_com_tw; |
2
|
1
|
|
|
1
|
|
149
|
use utf8; |
|
1
|
|
|
|
|
4
|
|
|
1
|
|
|
|
|
6
|
|
3
|
1
|
|
|
1
|
|
35
|
use Moo; |
|
1
|
|
|
|
|
4
|
|
|
1
|
|
|
|
|
6
|
|
4
|
|
|
|
|
|
|
extends 'NewsExtractor::GenericExtractor'; |
5
|
|
|
|
|
|
|
|
6
|
1
|
|
|
1
|
|
447
|
use Importer 'NewsExtractor::TextUtil' => qw(u); |
|
1
|
|
|
|
|
2
|
|
|
1
|
|
|
|
|
7
|
|
7
|
|
|
|
|
|
|
|
8
|
|
|
|
|
|
|
sub journalist { |
9
|
0
|
|
|
0
|
0
|
|
my ($self) = @_; |
10
|
0
|
0
|
|
|
|
|
my $el = $self->dom->at('li > time') or return; |
11
|
0
|
|
|
|
|
|
$el = $el->parent->previous->at('span > span'); |
12
|
0
|
|
|
|
|
|
return $el->all_text; |
13
|
|
|
|
|
|
|
} |
14
|
|
|
|
|
|
|
|
15
|
|
|
|
|
|
|
sub dateline { |
16
|
0
|
|
|
0
|
0
|
|
my ($self) = @_; |
17
|
0
|
0
|
|
|
|
|
my $el = $self->dom->at('li > time') or return; |
18
|
|
|
|
|
|
|
|
19
|
0
|
|
|
|
|
|
my @t = $el->all_text =~ m/([0-9]+)/g; |
20
|
0
|
|
|
|
|
|
return u( |
21
|
|
|
|
|
|
|
sprintf( |
22
|
|
|
|
|
|
|
'%04d-%02d-%02dT%02d:%02d:%02d+08:00', |
23
|
|
|
|
|
|
|
$t[0], $t[1], $t[2], 23, 59, 59 |
24
|
|
|
|
|
|
|
) |
25
|
|
|
|
|
|
|
); |
26
|
|
|
|
|
|
|
} |
27
|
|
|
|
|
|
|
|
28
|
|
|
|
|
|
|
1; |