| line |
stmt |
bran |
cond |
sub |
pod |
time |
code |
|
1
|
|
|
|
|
|
|
package NewsExtractor::SiteSpecificExtractor::estate_ltn_com_tw; |
|
2
|
1
|
|
|
1
|
|
8
|
use utf8; |
|
|
1
|
|
|
|
|
2
|
|
|
|
1
|
|
|
|
|
8
|
|
|
3
|
1
|
|
|
1
|
|
42
|
use Moo; |
|
|
1
|
|
|
|
|
3
|
|
|
|
1
|
|
|
|
|
6
|
|
|
4
|
|
|
|
|
|
|
extends 'NewsExtractor::SiteSpecificExtractor'; |
|
5
|
|
|
|
|
|
|
|
|
6
|
1
|
|
|
1
|
|
364
|
use Importer 'NewsExtractor::TextUtil' => qw< normalize_whitespace html2text >; |
|
|
1
|
|
|
|
|
4
|
|
|
|
1
|
|
|
|
|
10
|
|
|
7
|
|
|
|
|
|
|
|
|
8
|
|
|
|
|
|
|
sub headline { |
|
9
|
0
|
|
|
0
|
0
|
|
my ($self) = @_; |
|
10
|
0
|
|
|
|
|
|
my $el = $self->dom->at('div.container.page-name > h1'); |
|
11
|
0
|
|
|
|
|
|
return $el->all_text; |
|
12
|
|
|
|
|
|
|
} |
|
13
|
|
|
|
|
|
|
|
|
14
|
|
|
|
|
|
|
sub dateline { |
|
15
|
0
|
|
|
0
|
0
|
|
my ($self) = @_; |
|
16
|
0
|
|
|
|
|
|
my $el = $self->dom->at('p.author .time'); |
|
17
|
0
|
|
|
|
|
|
my $txt = $el->all_text; |
|
18
|
0
|
|
|
|
|
|
return $txt; |
|
19
|
|
|
|
|
|
|
} |
|
20
|
|
|
|
|
|
|
|
|
21
|
|
|
|
|
|
|
sub journalist { |
|
22
|
0
|
|
|
0
|
0
|
|
my ($self) = @_; |
|
23
|
0
|
|
|
|
|
|
my $el = $self->dom->at('p.author'); |
|
24
|
0
|
|
|
|
|
|
$el->children("div.share")->map('remove'); |
|
25
|
0
|
|
|
|
|
|
return normalize_whitespace($el->text); |
|
26
|
|
|
|
|
|
|
} |
|
27
|
|
|
|
|
|
|
|
|
28
|
|
|
|
|
|
|
sub content_text { |
|
29
|
0
|
|
|
0
|
0
|
|
my ($self) = @_; |
|
30
|
0
|
|
|
|
|
|
my $el = $self->dom->at("div[itemprop=articleBody]"); |
|
31
|
0
|
|
|
|
|
|
$el->children("p.appE1121")->map('remove'); |
|
32
|
0
|
|
|
|
|
|
return html2text( $el->to_string ); |
|
33
|
|
|
|
|
|
|
} |
|
34
|
|
|
|
|
|
|
|
|
35
|
|
|
|
|
|
|
1; |