line |
stmt |
bran |
cond |
sub |
pod |
time |
code |
1
|
|
|
|
|
|
|
package WWW::Leech::Parser; |
2
|
1
|
|
|
1
|
|
13184
|
use strict; |
|
1
|
|
|
|
|
2
|
|
|
1
|
|
|
|
|
38
|
|
3
|
1
|
|
|
1
|
|
396
|
use HTML::TreeBuilder::XPath; |
|
1
|
|
|
|
|
46925
|
|
|
1
|
|
|
|
|
6
|
|
4
|
1
|
|
|
1
|
|
435
|
use Encode qw|is_utf8 encode decode|; |
|
1
|
|
|
|
|
7241
|
|
|
1
|
|
|
|
|
53
|
|
5
|
1
|
|
|
1
|
|
413
|
use utf8; |
|
1
|
|
|
|
|
12
|
|
|
1
|
|
|
|
|
5
|
|
6
|
|
|
|
|
|
|
|
7
|
|
|
|
|
|
|
|
8
|
|
|
|
|
|
|
BEGIN { |
9
|
1
|
|
|
1
|
|
27
|
use Exporter (); |
|
1
|
|
|
|
|
2
|
|
|
1
|
|
|
|
|
17
|
|
10
|
1
|
|
|
1
|
|
4
|
use vars qw($VERSION @ISA @EXPORT @EXPORT_OK %EXPORT_TAGS); |
|
1
|
|
|
|
|
2
|
|
|
1
|
|
|
|
|
73
|
|
11
|
1
|
|
|
1
|
|
3
|
$VERSION = '0.01'; |
12
|
1
|
|
|
|
|
8
|
@ISA = qw(Exporter); |
13
|
1
|
|
|
|
|
2
|
@EXPORT = qw(); |
14
|
1
|
|
|
|
|
1
|
@EXPORT_OK = qw(); |
15
|
1
|
|
|
|
|
545
|
%EXPORT_TAGS = (); |
16
|
|
|
|
|
|
|
} |
17
|
|
|
|
|
|
|
|
18
|
|
|
|
|
|
|
sub new{ |
19
|
|
|
|
|
|
|
|
20
|
0
|
|
|
0
|
1
|
|
my $class = shift; |
21
|
0
|
|
|
|
|
|
my $params = shift; |
22
|
|
|
|
|
|
|
|
23
|
|
|
|
|
|
|
|
24
|
0
|
|
|
|
|
|
my $fields = {}; |
25
|
|
|
|
|
|
|
|
26
|
0
|
0
|
|
|
|
|
if(ref($params->{'fields'}) eq 'ARRAY'){ |
27
|
0
|
|
|
|
|
|
foreach( @{$params->{'fields'}} ){ |
|
0
|
|
|
|
|
|
|
28
|
0
|
|
|
|
|
|
$fields->{$_->{'name'}} = $_; |
29
|
|
|
|
|
|
|
} |
30
|
0
|
|
|
|
|
|
$params->{'fields'} = $fields; |
31
|
|
|
|
|
|
|
} |
32
|
|
|
|
|
|
|
|
33
|
0
|
|
|
|
|
|
return bless $params, __PACKAGE__; |
34
|
|
|
|
|
|
|
|
35
|
|
|
|
|
|
|
} |
36
|
|
|
|
|
|
|
|
37
|
|
|
|
|
|
|
sub parseList{ |
38
|
0
|
|
|
0
|
1
|
|
my $this = shift; |
39
|
0
|
|
|
|
|
|
my $str = shift; |
40
|
|
|
|
|
|
|
|
41
|
0
|
|
|
|
|
|
my $tree= HTML::TreeBuilder::XPath->new; |
42
|
0
|
|
|
|
|
|
$tree->parse_content($str); |
43
|
|
|
|
|
|
|
|
44
|
0
|
|
|
|
|
|
my $links = []; |
45
|
0
|
|
|
|
|
|
foreach(@{$tree->findnodes($this->{'item:link'})}){ |
|
0
|
|
|
|
|
|
|
46
|
0
|
|
|
|
|
|
push(@$links, $_->attr('href')); |
47
|
|
|
|
|
|
|
} |
48
|
|
|
|
|
|
|
|
49
|
0
|
|
|
|
|
|
my $np = $tree->findnodes($this->{'nextpage:link'})->[0]; |
50
|
|
|
|
|
|
|
|
51
|
0
|
0
|
|
|
|
|
if($np){ |
52
|
0
|
|
|
|
|
|
$np = $np->attr('href'); |
53
|
|
|
|
|
|
|
} |
54
|
|
|
|
|
|
|
|
55
|
|
|
|
|
|
|
return { |
56
|
0
|
|
|
|
|
|
'links' => $links, |
57
|
|
|
|
|
|
|
'next_page' => $np |
58
|
|
|
|
|
|
|
} |
59
|
|
|
|
|
|
|
} |
60
|
|
|
|
|
|
|
|
61
|
|
|
|
|
|
|
sub parse{ |
62
|
0
|
|
|
0
|
0
|
|
my $this = shift; |
63
|
0
|
|
|
|
|
|
my $str = shift; |
64
|
|
|
|
|
|
|
|
65
|
0
|
0
|
|
|
|
|
if(!is_utf8($str)){ |
66
|
0
|
|
|
|
|
|
$str = decode('UTF-8',$str); |
67
|
|
|
|
|
|
|
} |
68
|
|
|
|
|
|
|
|
69
|
0
|
|
|
|
|
|
my $tree= HTML::TreeBuilder::XPath->new; |
70
|
0
|
|
|
|
|
|
$tree->ignore_unknown(0); |
71
|
0
|
|
|
|
|
|
$tree->parse_content($str); |
72
|
|
|
|
|
|
|
|
73
|
0
|
|
|
|
|
|
my $item = {}; |
74
|
|
|
|
|
|
|
|
75
|
0
|
|
|
|
|
|
foreach my $field_name ( keys(%{$this->{'fields'}}) ){ |
|
0
|
|
|
|
|
|
|
76
|
|
|
|
|
|
|
|
77
|
0
|
|
|
|
|
|
my $field = $this->{'fields'}->{$field_name}; |
78
|
|
|
|
|
|
|
|
79
|
0
|
0
|
|
|
|
|
if(ref($field) ne 'HASH') { |
80
|
0
|
|
|
|
|
|
$field = { |
81
|
|
|
|
|
|
|
'xpath' => $field |
82
|
|
|
|
|
|
|
}; |
83
|
|
|
|
|
|
|
} |
84
|
|
|
|
|
|
|
|
85
|
0
|
|
|
|
|
|
$field->{'name'} = $field_name; |
86
|
|
|
|
|
|
|
|
87
|
0
|
0
|
|
|
|
|
if($field->{'xpath'}){ |
88
|
0
|
0
|
|
|
|
|
if(!is_utf8($field->{'xpath'})){ |
89
|
0
|
|
|
|
|
|
$field->{'xpath'} = decode('UTF-8',$field->{'xpath'}); |
90
|
|
|
|
|
|
|
} |
91
|
|
|
|
|
|
|
|
92
|
0
|
0
|
|
|
|
|
my $type = ($field->{'type'} ? $field->{'type'} : 'text'); |
93
|
0
|
0
|
|
|
|
|
my $wantarray = ( $field_name =~ /\[\]$/ ? 1 : undef ); |
94
|
|
|
|
|
|
|
|
95
|
0
|
|
|
|
|
|
my $value; |
96
|
|
|
|
|
|
|
|
97
|
0
|
0
|
|
|
|
|
if($wantarray){ |
98
|
0
|
|
|
|
|
|
$value = []; |
99
|
|
|
|
|
|
|
|
100
|
0
|
0
|
|
|
|
|
if($type eq 'html'){ |
101
|
|
|
|
|
|
|
|
102
|
0
|
|
|
|
|
|
foreach my $n ($tree->findnodes($field->{'xpath'})){ |
103
|
0
|
|
|
|
|
|
push(@$value,$n->as_XML); |
104
|
|
|
|
|
|
|
} |
105
|
|
|
|
|
|
|
|
106
|
|
|
|
|
|
|
} else { |
107
|
0
|
|
|
|
|
|
$value = [$tree->findvalues($field->{'xpath'})]; |
108
|
|
|
|
|
|
|
|
109
|
0
|
0
|
|
|
|
|
if($type eq 'unique'){ |
110
|
0
|
|
|
|
|
|
my %u; |
111
|
0
|
|
|
|
|
|
$u{$_} = 1 foreach(@$value); |
112
|
0
|
|
|
|
|
|
$value = [keys(%u)]; |
113
|
|
|
|
|
|
|
} |
114
|
|
|
|
|
|
|
} |
115
|
|
|
|
|
|
|
} else { |
116
|
0
|
0
|
|
|
|
|
if($type eq 'html'){ |
117
|
|
|
|
|
|
|
|
118
|
0
|
0
|
|
|
|
|
if([$tree->findnodes($field->{'xpath'})]->[0]){ |
119
|
0
|
|
|
|
|
|
$value = [$tree->findnodes($field->{'xpath'})]->[0]->as_XML; |
120
|
|
|
|
|
|
|
} |
121
|
|
|
|
|
|
|
|
122
|
|
|
|
|
|
|
} |
123
|
|
|
|
|
|
|
else { |
124
|
0
|
|
|
|
|
|
$value = $tree->findvalue($field->{'xpath'}); |
125
|
|
|
|
|
|
|
} |
126
|
|
|
|
|
|
|
|
127
|
0
|
0
|
|
|
|
|
if($type eq 'int'){ |
128
|
0
|
|
|
|
|
|
$value =~ s/[^\d]//g; |
129
|
|
|
|
|
|
|
} |
130
|
|
|
|
|
|
|
} |
131
|
|
|
|
|
|
|
|
132
|
0
|
0
|
|
|
|
|
if($field->{'filter'}){ |
133
|
0
|
|
|
|
|
|
$value = $field->{'filter'}->($value,$field); |
134
|
|
|
|
|
|
|
} |
135
|
|
|
|
|
|
|
|
136
|
0
|
|
|
|
|
|
$item->{$field_name} = $value; |
137
|
|
|
|
|
|
|
} |
138
|
|
|
|
|
|
|
} |
139
|
|
|
|
|
|
|
|
140
|
0
|
|
|
|
|
|
return $item; |
141
|
|
|
|
|
|
|
|
142
|
|
|
|
|
|
|
} |
143
|
|
|
|
|
|
|
|
144
|
|
|
|
|
|
|
1; |
145
|
|
|
|
|
|
|
|
146
|
|
|
|
|
|
|
__END__ |