line |
stmt |
bran |
cond |
sub |
pod |
time |
code |
1
|
|
|
|
|
|
|
package WWW::Leech::Walker; |
2
|
1
|
|
|
1
|
|
13124
|
use strict; |
|
1
|
|
|
|
|
2
|
|
|
1
|
|
|
|
|
24
|
|
3
|
1
|
|
|
1
|
|
393
|
use LWP; |
|
1
|
|
|
|
|
34671
|
|
|
1
|
|
|
|
|
32
|
|
4
|
1
|
|
|
1
|
|
374
|
use URI::URL; |
|
1
|
|
|
|
|
2589
|
|
|
1
|
|
|
|
|
44
|
|
5
|
1
|
|
|
1
|
|
375
|
use WWW::Leech::Parser; |
|
1
|
|
|
|
|
54483
|
|
|
1
|
|
|
|
|
45
|
|
6
|
|
|
|
|
|
|
|
7
|
|
|
|
|
|
|
BEGIN { |
8
|
1
|
|
|
1
|
|
7
|
use Exporter (); |
|
1
|
|
|
|
|
2
|
|
|
1
|
|
|
|
|
17
|
|
9
|
1
|
|
|
1
|
|
5
|
use vars qw($VERSION @ISA @EXPORT @EXPORT_OK %EXPORT_TAGS); |
|
1
|
|
|
|
|
2
|
|
|
1
|
|
|
|
|
66
|
|
10
|
1
|
|
|
1
|
|
3
|
$VERSION = '0.01'; |
11
|
1
|
|
|
|
|
8
|
@ISA = qw(Exporter); |
12
|
1
|
|
|
|
|
3
|
@EXPORT = qw(); |
13
|
1
|
|
|
|
|
5
|
@EXPORT_OK = qw(); |
14
|
1
|
|
|
|
|
453
|
%EXPORT_TAGS = (); |
15
|
|
|
|
|
|
|
} |
16
|
|
|
|
|
|
|
|
17
|
|
|
|
|
|
|
sub new{ |
18
|
0
|
|
|
0
|
1
|
|
my $class = shift; |
19
|
0
|
|
|
|
|
|
my $this = shift; |
20
|
|
|
|
|
|
|
|
21
|
0
|
|
|
|
|
|
$this->{'url'} = url $this->{'url'}; |
22
|
0
|
|
0
|
|
|
|
$this->{'state'} ||= {}; |
23
|
|
|
|
|
|
|
|
24
|
0
|
|
|
|
|
|
$this->{'_stop_flag'} = undef; |
25
|
|
|
|
|
|
|
|
26
|
0
|
|
|
|
|
|
$this = bless $this, __PACKAGE__; |
27
|
|
|
|
|
|
|
|
28
|
0
|
|
|
|
|
|
$this->{'parser'} = new WWW::Leech::Parser($this->{'parser'}); |
29
|
|
|
|
|
|
|
|
30
|
0
|
|
|
|
|
|
return $this; |
31
|
|
|
|
|
|
|
} |
32
|
|
|
|
|
|
|
|
33
|
|
|
|
|
|
|
sub leech{ |
34
|
0
|
|
|
0
|
1
|
|
my $this = shift; |
35
|
0
|
|
|
|
|
|
$this->{'_stop_flag'} = undef; |
36
|
|
|
|
|
|
|
|
37
|
0
|
|
|
|
|
|
my $current_url = $this->{'url'}; |
38
|
|
|
|
|
|
|
|
39
|
0
|
|
|
|
|
|
while(1){ |
40
|
|
|
|
|
|
|
|
41
|
0
|
|
|
|
|
|
$this->log("Getting links list: $current_url "); |
42
|
|
|
|
|
|
|
|
43
|
|
|
|
|
|
|
# get and parse link list |
44
|
0
|
|
|
|
|
|
my $data = $this->_get($current_url); |
45
|
|
|
|
|
|
|
|
46
|
0
|
|
|
|
|
|
my $parse_result = $this->{'parser'}->parseList($data); |
47
|
|
|
|
|
|
|
|
48
|
0
|
|
|
|
|
|
my $links = [map { url($_, $this->{'url'})->abs } @{$parse_result->{'links'}}]; |
|
0
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
49
|
|
|
|
|
|
|
|
50
|
0
|
|
|
|
|
|
$this->log("Links list length: ".scalar(@$links)); |
51
|
|
|
|
|
|
|
|
52
|
0
|
0
|
|
|
|
|
if(scalar(@$links) == 0){ |
53
|
0
|
|
|
|
|
|
$this->log("No items links found, check your XPath: ".$this->{'parser'}->{'item:link'}); |
54
|
|
|
|
|
|
|
|
55
|
0
|
|
|
|
|
|
$this->stop(); |
56
|
0
|
|
|
|
|
|
return; |
57
|
|
|
|
|
|
|
} |
58
|
|
|
|
|
|
|
|
59
|
|
|
|
|
|
|
# filtering if required |
60
|
0
|
0
|
|
|
|
|
if( $this->{'filter'} ){ |
61
|
0
|
|
|
|
|
|
$this->log("Filtering links"); |
62
|
0
|
|
|
|
|
|
$links = $this->{'filter'}->($links,$this); |
63
|
0
|
|
|
|
|
|
$this->log("Filtered links list length: ".scalar(@$links)); |
64
|
|
|
|
|
|
|
|
65
|
0
|
0
|
|
|
|
|
return if($this->{'_stop_flag'}); |
66
|
|
|
|
|
|
|
} |
67
|
|
|
|
|
|
|
|
68
|
|
|
|
|
|
|
# get and parse individual items |
69
|
0
|
|
|
|
|
|
foreach(@$links){ |
70
|
0
|
|
|
|
|
|
$this->log("\tGetting item: $_ "); |
71
|
|
|
|
|
|
|
|
72
|
0
|
|
|
|
|
|
my $item = $this->{'parser'}->parse($this->_get($_)); |
73
|
0
|
|
|
|
|
|
$item->{'_url'} = $_->as_string; |
74
|
|
|
|
|
|
|
|
75
|
0
|
|
|
|
|
|
$this->{'processor'}->($item,$this); |
76
|
|
|
|
|
|
|
|
77
|
0
|
0
|
|
|
|
|
return if($this->{'_stop_flag'}); |
78
|
|
|
|
|
|
|
} |
79
|
|
|
|
|
|
|
|
80
|
0
|
0
|
|
|
|
|
if($parse_result->{'next_page'}){ |
81
|
|
|
|
|
|
|
|
82
|
0
|
|
|
|
|
|
my $pnurl; |
83
|
|
|
|
|
|
|
|
84
|
0
|
0
|
|
|
|
|
if($this->{'next_page_link_post_process'}){ |
85
|
0
|
|
|
|
|
|
$pnurl = $this->{'next_page_link_post_process'}->($parse_result->{'next_page'}, $this, $current_url); |
86
|
|
|
|
|
|
|
} |
87
|
|
|
|
|
|
|
|
88
|
0
|
0
|
|
|
|
|
if($pnurl){ |
89
|
0
|
|
|
|
|
|
$current_url = url($pnurl)->abs; |
90
|
|
|
|
|
|
|
} else { |
91
|
0
|
|
|
|
|
|
$current_url = url($parse_result->{'next_page'}, $this->{'url'})->abs; |
92
|
|
|
|
|
|
|
} |
93
|
|
|
|
|
|
|
|
94
|
0
|
0
|
|
|
|
|
return if($this->{'_stop_flag'}); |
95
|
|
|
|
|
|
|
|
96
|
|
|
|
|
|
|
} else { |
97
|
0
|
|
|
|
|
|
$this->stop("No next page link found"); |
98
|
0
|
|
|
|
|
|
return; |
99
|
|
|
|
|
|
|
} |
100
|
|
|
|
|
|
|
} |
101
|
|
|
|
|
|
|
|
102
|
|
|
|
|
|
|
|
103
|
|
|
|
|
|
|
} |
104
|
|
|
|
|
|
|
|
105
|
|
|
|
|
|
|
sub log{ |
106
|
0
|
|
|
0
|
1
|
|
my $this = shift; |
107
|
0
|
|
|
|
|
|
my $message = shift; |
108
|
|
|
|
|
|
|
|
109
|
0
|
0
|
|
|
|
|
if($this->{'logger'}){ |
110
|
0
|
|
|
|
|
|
$this->{'logger'}->($message,$this); |
111
|
|
|
|
|
|
|
} |
112
|
|
|
|
|
|
|
} |
113
|
|
|
|
|
|
|
|
114
|
|
|
|
|
|
|
sub stop{ |
115
|
0
|
|
|
0
|
1
|
|
my $this = shift; |
116
|
0
|
|
|
|
|
|
my $reason = shift; |
117
|
|
|
|
|
|
|
|
118
|
0
|
|
|
|
|
|
$this->{'_stop_flag'} = 1; |
119
|
0
|
|
|
|
|
|
$this->log("\tWalker stopped."); |
120
|
0
|
0
|
|
|
|
|
if($reason){ |
121
|
0
|
|
|
|
|
|
$this->log("\tReason: $reason."); |
122
|
|
|
|
|
|
|
} |
123
|
|
|
|
|
|
|
} |
124
|
|
|
|
|
|
|
|
125
|
|
|
|
|
|
|
sub _get{ |
126
|
0
|
|
|
0
|
|
|
my $this = shift; |
127
|
0
|
|
|
|
|
|
my $url = shift; |
128
|
|
|
|
|
|
|
|
129
|
0
|
|
|
|
|
|
my $req = new HTTP::Request('GET',$url); |
130
|
0
|
|
|
|
|
|
my $res = $this->{'ua'}->request($req); |
131
|
|
|
|
|
|
|
|
132
|
0
|
|
|
|
|
|
return $res->decoded_content(); |
133
|
|
|
|
|
|
|
} |
134
|
|
|
|
|
|
|
|
135
|
|
|
|
|
|
|
|
136
|
|
|
|
|
|
|
1; |
137
|
|
|
|
|
|
|
|
138
|
|
|
|
|
|
|
|
139
|
|
|
|
|
|
|
__END__ |