line |
stmt |
bran |
cond |
sub |
pod |
time |
code |
1
|
|
|
|
|
|
|
use utf8; |
2
|
1
|
|
|
1
|
|
6
|
use Moo; |
|
1
|
|
|
|
|
2
|
|
|
1
|
|
|
|
|
5
|
|
3
|
1
|
|
|
1
|
|
25
|
extends 'NewsExtractor::GenericExtractor'; |
|
1
|
|
|
|
|
3
|
|
|
1
|
|
|
|
|
4
|
|
4
|
|
|
|
|
|
|
|
5
|
|
|
|
|
|
|
around headline => sub { |
6
|
|
|
|
|
|
|
my $orig = shift; |
7
|
|
|
|
|
|
|
my $self = $_[0]; |
8
|
|
|
|
|
|
|
my $headline = $orig->(@_); |
9
|
|
|
|
|
|
|
my $journalist = $self->journalist(); |
10
|
|
|
|
|
|
|
$headline =~ s/\s\|\s${journalist}$//; |
11
|
|
|
|
|
|
|
return $headline; |
12
|
|
|
|
|
|
|
}; |
13
|
|
|
|
|
|
|
|
14
|
|
|
|
|
|
|
my ($self) = @_; |
15
|
|
|
|
|
|
|
my $el = $self->dom->at('.article-info'); |
16
|
0
|
|
|
0
|
0
|
|
my ($n) = $el->all_text =~ m{^(?:Author|撰文): (?:(?:特約)?記者)?(.+?) \|}; |
17
|
0
|
|
|
|
|
|
return $n; |
18
|
0
|
|
|
|
|
|
} |
19
|
0
|
|
|
|
|
|
|
20
|
|
|
|
|
|
|
my ($self) = @_; |
21
|
|
|
|
|
|
|
my $el = $self->dom->at('.article-info'); |
22
|
|
|
|
|
|
|
my ($dd,$mm,$yy) = $el->all_text =~ m{(?:發佈日期|Publish Date):\s([0-9]{2})\.([0-9]{2})\.([0-9]{2})\b}; |
23
|
0
|
|
|
0
|
0
|
|
return undef unless ($dd && $mm && $yy); |
24
|
0
|
|
|
|
|
|
return "20${yy}-${mm}-${dd}T23:59:59+08:00"; |
25
|
0
|
|
|
|
|
|
} |
26
|
0
|
0
|
0
|
|
|
|
|
|
|
|
0
|
|
|
|
|
27
|
0
|
|
|
|
|
|
|
28
|
|
|
|
|
|
|
1; |