line |
stmt |
bran |
cond |
sub |
pod |
time |
code |
1
|
|
|
|
|
|
|
package WWW::Ruten; |
2
|
|
|
|
|
|
|
{ |
3
|
|
|
|
|
|
|
$WWW::Ruten::VERSION = '0.03'; |
4
|
|
|
|
|
|
|
} |
5
|
|
|
|
|
|
|
# ABSTRACT: Scripting www.ruten.com.tw |
6
|
|
|
|
|
|
|
|
7
|
2
|
|
|
2
|
|
70840
|
use 5.008; |
|
2
|
|
|
|
|
9
|
|
|
2
|
|
|
|
|
68
|
|
8
|
2
|
|
|
2
|
|
10
|
use strict; |
|
2
|
|
|
|
|
4
|
|
|
2
|
|
|
|
|
59
|
|
9
|
2
|
|
|
2
|
|
10
|
use warnings; |
|
2
|
|
|
|
|
8
|
|
|
2
|
|
|
|
|
72
|
|
10
|
2
|
|
|
2
|
|
2584
|
use WWW::Mechanize 1.66; |
|
2
|
|
|
|
|
675051
|
|
|
2
|
|
|
|
|
129
|
|
11
|
2
|
|
|
2
|
|
27
|
use Encode; |
|
2
|
|
|
|
|
4
|
|
|
2
|
|
|
|
|
282
|
|
12
|
2
|
|
|
2
|
|
3905
|
use HTML::TreeBuilder::XPath; |
|
2
|
|
|
|
|
349029
|
|
|
2
|
|
|
|
|
31
|
|
13
|
2
|
|
|
2
|
|
3087
|
use HTML::Selector::XPath; |
|
2
|
|
|
|
|
12346
|
|
|
2
|
|
|
|
|
1121
|
|
14
|
|
|
|
|
|
|
|
15
|
|
|
|
|
|
|
sub new { |
16
|
1
|
|
|
1
|
1
|
14
|
my ($class) = @_; |
17
|
1
|
|
|
|
|
3
|
my $self = {}; |
18
|
1
|
|
|
|
|
3
|
bless $self, $class; |
19
|
1
|
|
|
|
|
4
|
return $self; |
20
|
|
|
|
|
|
|
} |
21
|
|
|
|
|
|
|
|
22
|
|
|
|
|
|
|
sub search { |
23
|
1
|
|
|
1
|
1
|
9
|
my ($self, $term, @params) = @_; |
24
|
1
|
|
|
|
|
10
|
$self->{search_params} = { |
25
|
|
|
|
|
|
|
term => $term |
26
|
|
|
|
|
|
|
}; |
27
|
|
|
|
|
|
|
|
28
|
1
|
|
|
|
|
4
|
return $self; |
29
|
|
|
|
|
|
|
} |
30
|
|
|
|
|
|
|
|
31
|
|
|
|
|
|
|
sub each { |
32
|
1
|
|
|
1
|
1
|
12
|
my ($self, $cb) = @_; |
33
|
1
|
|
|
|
|
7
|
$self->_do_search(); |
34
|
|
|
|
|
|
|
|
35
|
1
|
|
|
|
|
3
|
foreach my $result (@{ $self->{search_results} }) { |
|
1
|
|
|
|
|
4
|
|
36
|
0
|
|
|
|
|
0
|
$cb->($result); |
37
|
|
|
|
|
|
|
} |
38
|
1
|
|
|
|
|
3
|
return $self; |
39
|
|
|
|
|
|
|
} |
40
|
|
|
|
|
|
|
|
41
|
|
|
|
|
|
|
sub _do_search { |
42
|
1
|
|
|
1
|
|
3
|
my ($self) = @_; |
43
|
1
|
50
|
|
|
|
6
|
die unless defined $self->{search_params}{term}; |
44
|
|
|
|
|
|
|
|
45
|
1
|
|
33
|
|
|
14
|
$self->{mech} ||= WWW::Mechanize->new; |
46
|
|
|
|
|
|
|
|
47
|
1
|
|
|
|
|
33407
|
my $mech = $self->{mech}; |
48
|
1
|
|
|
|
|
9
|
$mech->get("http://www.ruten.com.tw"); |
49
|
1
|
|
|
|
|
1425516
|
$mech->submit_form( |
50
|
|
|
|
|
|
|
form_name => "srch", |
51
|
|
|
|
|
|
|
fields => { |
52
|
|
|
|
|
|
|
k => $self->{search_params}{term} |
53
|
|
|
|
|
|
|
} |
54
|
|
|
|
|
|
|
); |
55
|
|
|
|
|
|
|
|
56
|
1
|
|
|
|
|
1457522
|
my $content = $mech->content; |
57
|
|
|
|
|
|
|
|
58
|
1
|
|
|
|
|
42
|
my @results = (); |
59
|
1
|
|
|
|
|
21
|
my $html = HTML::TreeBuilder::XPath->new; |
60
|
1
|
|
|
|
|
492
|
$html->parse($content); |
61
|
|
|
|
|
|
|
|
62
|
1
|
|
|
|
|
55382
|
my $selector = HTML::Selector::XPath->new("ul.items h3.title a"); |
63
|
|
|
|
|
|
|
|
64
|
1
|
|
|
|
|
13
|
for my $node ($html->findnodes($selector->to_xpath)) { |
65
|
0
|
|
|
|
|
0
|
push @results, { |
66
|
|
|
|
|
|
|
url => $node->attr("href"), |
67
|
|
|
|
|
|
|
title => join("", $node->content_list) |
68
|
|
|
|
|
|
|
} |
69
|
|
|
|
|
|
|
} |
70
|
|
|
|
|
|
|
|
71
|
1
|
|
50
|
|
|
48120
|
$self->{search_results} ||=[]; |
72
|
1
|
|
|
|
|
2
|
push @{ $self->{search_results} }, @results; |
|
1
|
|
|
|
|
3
|
|
73
|
|
|
|
|
|
|
|
74
|
1
|
|
|
|
|
211
|
return $self; |
75
|
|
|
|
|
|
|
} |
76
|
|
|
|
|
|
|
|
77
|
|
|
|
|
|
|
1; |
78
|
|
|
|
|
|
|
|
79
|
|
|
|
|
|
|
|
80
|
|
|
|
|
|
|
__END__ |