line |
stmt |
bran |
cond |
sub |
pod |
time |
code |
1
|
|
|
|
|
|
|
package WWW::Crawler::Mojo::Job; |
2
|
11
|
|
|
11
|
|
68776
|
use strict; |
|
11
|
|
|
|
|
28
|
|
|
11
|
|
|
|
|
272
|
|
3
|
11
|
|
|
11
|
|
48
|
use warnings; |
|
11
|
|
|
|
|
18
|
|
|
11
|
|
|
|
|
234
|
|
4
|
11
|
|
|
11
|
|
46
|
use utf8; |
|
11
|
|
|
|
|
15
|
|
|
11
|
|
|
|
|
51
|
|
5
|
11
|
|
|
11
|
|
648
|
use Mojo::Base -base; |
|
11
|
|
|
|
|
158696
|
|
|
11
|
|
|
|
|
64
|
|
6
|
11
|
|
|
11
|
|
1398
|
use Mojo::Util qw(md5_sum); |
|
11
|
|
|
|
|
21
|
|
|
11
|
|
|
|
|
5074
|
|
7
|
|
|
|
|
|
|
|
8
|
|
|
|
|
|
|
has 'closed'; |
9
|
|
|
|
|
|
|
has 'context'; |
10
|
|
|
|
|
|
|
has depth => 0; |
11
|
|
|
|
|
|
|
has 'literal_uri'; |
12
|
|
|
|
|
|
|
has 'method'; |
13
|
|
|
|
|
|
|
has 'referrer'; |
14
|
|
|
|
|
|
|
has redirect_history => sub { [] }; |
15
|
|
|
|
|
|
|
has 'tx_params'; |
16
|
|
|
|
|
|
|
has 'url'; |
17
|
|
|
|
|
|
|
|
18
|
|
|
|
|
|
|
sub upgrade { |
19
|
60
|
|
|
60
|
1
|
379
|
my ($class, $job) = @_; |
20
|
|
|
|
|
|
|
|
21
|
60
|
100
|
100
|
|
|
218
|
if (!ref $job || ref $job ne __PACKAGE__) { |
22
|
8
|
100
|
|
|
|
31
|
my $url = !ref $job ? Mojo::URL->new($job) : $job; |
23
|
8
|
|
|
|
|
462
|
$job = $class->new(url => $url); |
24
|
|
|
|
|
|
|
} |
25
|
|
|
|
|
|
|
|
26
|
60
|
|
|
|
|
193
|
return $job; |
27
|
|
|
|
|
|
|
} |
28
|
|
|
|
|
|
|
|
29
|
|
|
|
|
|
|
sub clone { |
30
|
1
|
|
|
1
|
1
|
732
|
my $self = shift; |
31
|
1
|
|
|
|
|
6
|
return __PACKAGE__->new(%$self); |
32
|
|
|
|
|
|
|
} |
33
|
|
|
|
|
|
|
|
34
|
|
|
|
|
|
|
sub close { |
35
|
14
|
|
|
14
|
1
|
1185
|
my $self = shift; |
36
|
14
|
|
|
|
|
45
|
$self->{closed} = 1; |
37
|
14
|
|
|
|
|
410
|
$self->{referrer} = undef; |
38
|
|
|
|
|
|
|
} |
39
|
|
|
|
|
|
|
|
40
|
|
|
|
|
|
|
sub child { |
41
|
77
|
|
|
77
|
1
|
2267
|
my $self = shift; |
42
|
77
|
|
|
|
|
184
|
return __PACKAGE__->new(@_, referrer => $self, depth => $self->depth + 1); |
43
|
|
|
|
|
|
|
} |
44
|
|
|
|
|
|
|
|
45
|
|
|
|
|
|
|
sub digest { |
46
|
64
|
|
|
64
|
1
|
118
|
my $self = shift; |
47
|
64
|
|
100
|
|
|
128
|
my $md5_seed = $self->url->to_string . ($self->method || ''); |
48
|
64
|
100
|
|
|
|
13427
|
$md5_seed .= $self->tx_params->to_string if ($self->tx_params); |
49
|
64
|
|
|
|
|
881
|
return md5_sum($md5_seed); |
50
|
|
|
|
|
|
|
} |
51
|
|
|
|
|
|
|
|
52
|
|
|
|
|
|
|
sub redirect { |
53
|
14
|
|
|
14
|
1
|
761
|
my ($self, $last, @history) = @_; |
54
|
14
|
|
|
|
|
51
|
$self->url($last); |
55
|
14
|
|
|
|
|
109
|
$self->redirect_history(\@history); |
56
|
|
|
|
|
|
|
} |
57
|
|
|
|
|
|
|
|
58
|
|
|
|
|
|
|
sub original_url { |
59
|
1
|
|
|
1
|
1
|
881
|
my $self = shift; |
60
|
1
|
|
|
|
|
10
|
my @histry = @{$self->redirect_history}; |
|
1
|
|
|
|
|
6
|
|
61
|
1
|
50
|
|
|
|
9
|
return $self->url unless (@histry); |
62
|
1
|
|
|
|
|
4
|
return $histry[$#histry]; |
63
|
|
|
|
|
|
|
} |
64
|
|
|
|
|
|
|
|
65
|
|
|
|
|
|
|
1; |
66
|
|
|
|
|
|
|
|
67
|
|
|
|
|
|
|
=head1 NAME |
68
|
|
|
|
|
|
|
|
69
|
|
|
|
|
|
|
WWW::Crawler::Mojo::Job - Single crawler job |
70
|
|
|
|
|
|
|
|
71
|
|
|
|
|
|
|
=head1 SYNOPSIS |
72
|
|
|
|
|
|
|
|
73
|
|
|
|
|
|
|
my $job1 = WWW::Crawler::Mojo::Job->new; |
74
|
|
|
|
|
|
|
$job1->url('http://example.com/'); |
75
|
|
|
|
|
|
|
my $job2 = $job1->child; |
76
|
|
|
|
|
|
|
|
77
|
|
|
|
|
|
|
=head1 DESCRIPTION |
78
|
|
|
|
|
|
|
|
79
|
|
|
|
|
|
|
This class represents a single crawler job. |
80
|
|
|
|
|
|
|
|
81
|
|
|
|
|
|
|
=head1 ATTRIBUTES |
82
|
|
|
|
|
|
|
|
83
|
|
|
|
|
|
|
=head2 context |
84
|
|
|
|
|
|
|
|
85
|
|
|
|
|
|
|
Either L or L instance that the job is referrered by. |
86
|
|
|
|
|
|
|
|
87
|
|
|
|
|
|
|
$job->context($dom); |
88
|
|
|
|
|
|
|
say $job->context; |
89
|
|
|
|
|
|
|
|
90
|
|
|
|
|
|
|
=head2 closed |
91
|
|
|
|
|
|
|
|
92
|
|
|
|
|
|
|
A flag indecates whether the job is closed or not. |
93
|
|
|
|
|
|
|
|
94
|
|
|
|
|
|
|
$job->closed(1); |
95
|
|
|
|
|
|
|
say $job->closed; |
96
|
|
|
|
|
|
|
|
97
|
|
|
|
|
|
|
=head2 depth |
98
|
|
|
|
|
|
|
|
99
|
|
|
|
|
|
|
The depth of the job in referrer series. |
100
|
|
|
|
|
|
|
|
101
|
|
|
|
|
|
|
my $job1 = WWW::Crawler::Mojo::Job->new; |
102
|
|
|
|
|
|
|
my $job2 = $job1->child; |
103
|
|
|
|
|
|
|
my $job3 = $job2->child; |
104
|
|
|
|
|
|
|
say $job1->depth; # 0 |
105
|
|
|
|
|
|
|
say $job2->depth; # 1 |
106
|
|
|
|
|
|
|
say $job3->depth; # 2 |
107
|
|
|
|
|
|
|
|
108
|
|
|
|
|
|
|
=head2 literal_uri |
109
|
|
|
|
|
|
|
|
110
|
|
|
|
|
|
|
A L instance of the literal URL that has appeared in the referrer |
111
|
|
|
|
|
|
|
document. |
112
|
|
|
|
|
|
|
|
113
|
|
|
|
|
|
|
$job1->literal_uri('./index.html'); |
114
|
|
|
|
|
|
|
say $job1->literal_uri; # './index.html' |
115
|
|
|
|
|
|
|
|
116
|
|
|
|
|
|
|
=head2 referrer |
117
|
|
|
|
|
|
|
|
118
|
|
|
|
|
|
|
A job instance that has referred the URL. |
119
|
|
|
|
|
|
|
|
120
|
|
|
|
|
|
|
$job1->referrer($job); |
121
|
|
|
|
|
|
|
my $job2 = $job1->referrer; |
122
|
|
|
|
|
|
|
|
123
|
|
|
|
|
|
|
=head2 redirect_history |
124
|
|
|
|
|
|
|
|
125
|
|
|
|
|
|
|
An array reference that contains URLs of redirect history. |
126
|
|
|
|
|
|
|
|
127
|
|
|
|
|
|
|
$job1->redirect_history([$url1, $url2, $url3]); |
128
|
|
|
|
|
|
|
my $history = $job1->redirect_history; |
129
|
|
|
|
|
|
|
|
130
|
|
|
|
|
|
|
=head2 url |
131
|
|
|
|
|
|
|
|
132
|
|
|
|
|
|
|
A L instance of the resolved URL. |
133
|
|
|
|
|
|
|
|
134
|
|
|
|
|
|
|
$job1->url('http://example.com/'); |
135
|
|
|
|
|
|
|
say $job1->url; # 'http://example.com/' |
136
|
|
|
|
|
|
|
|
137
|
|
|
|
|
|
|
=head2 method |
138
|
|
|
|
|
|
|
|
139
|
|
|
|
|
|
|
HTTP request method such as GET or POST. |
140
|
|
|
|
|
|
|
|
141
|
|
|
|
|
|
|
$job1->method('GET'); |
142
|
|
|
|
|
|
|
say $job1->method; # GET |
143
|
|
|
|
|
|
|
|
144
|
|
|
|
|
|
|
=head2 tx_params |
145
|
|
|
|
|
|
|
|
146
|
|
|
|
|
|
|
A hash reference that contains params for L. |
147
|
|
|
|
|
|
|
|
148
|
|
|
|
|
|
|
$job1->tx_params({foo => 'bar'}); |
149
|
|
|
|
|
|
|
$params = $job1->tx_params; |
150
|
|
|
|
|
|
|
|
151
|
|
|
|
|
|
|
=head1 METHODS |
152
|
|
|
|
|
|
|
|
153
|
|
|
|
|
|
|
=head2 clone |
154
|
|
|
|
|
|
|
|
155
|
|
|
|
|
|
|
Clones the job. |
156
|
|
|
|
|
|
|
|
157
|
|
|
|
|
|
|
my $job2 = $job1->clone; |
158
|
|
|
|
|
|
|
|
159
|
|
|
|
|
|
|
=head2 close |
160
|
|
|
|
|
|
|
|
161
|
|
|
|
|
|
|
Closes the job and cuts the referrer series. |
162
|
|
|
|
|
|
|
|
163
|
|
|
|
|
|
|
$job->close; |
164
|
|
|
|
|
|
|
|
165
|
|
|
|
|
|
|
=head2 child |
166
|
|
|
|
|
|
|
|
167
|
|
|
|
|
|
|
Instantiates a child job by parent job. The parent URL is set to child referrer. |
168
|
|
|
|
|
|
|
|
169
|
|
|
|
|
|
|
my $job1 = WWW::Crawler::Mojo::Job->new(url => 'http://example.com/1'); |
170
|
|
|
|
|
|
|
my $job2 = $job1->child(url => 'http://example.com/2'); |
171
|
|
|
|
|
|
|
say $job2->referrer->url # 'http://example.com/1' |
172
|
|
|
|
|
|
|
|
173
|
|
|
|
|
|
|
=head2 digest |
174
|
|
|
|
|
|
|
|
175
|
|
|
|
|
|
|
Generates digest string with C, C, C attributes. |
176
|
|
|
|
|
|
|
|
177
|
|
|
|
|
|
|
say $job->digest; |
178
|
|
|
|
|
|
|
|
179
|
|
|
|
|
|
|
=head2 redirect |
180
|
|
|
|
|
|
|
|
181
|
|
|
|
|
|
|
Replaces the resolved URL and history at once. |
182
|
|
|
|
|
|
|
|
183
|
|
|
|
|
|
|
my $job = WWW::Crawler::Mojo::Job->new; |
184
|
|
|
|
|
|
|
$job->url($url1); |
185
|
|
|
|
|
|
|
$job->redirect($url2, $url3); |
186
|
|
|
|
|
|
|
say $job->url # $url2 |
187
|
|
|
|
|
|
|
say $job->redirect_history # [$url1, $url3] |
188
|
|
|
|
|
|
|
|
189
|
|
|
|
|
|
|
=head2 original_url |
190
|
|
|
|
|
|
|
|
191
|
|
|
|
|
|
|
Returns the original URL of redirected job. If redirected, returns last element |
192
|
|
|
|
|
|
|
of C attribute, otherwise returns C attribute. |
193
|
|
|
|
|
|
|
|
194
|
|
|
|
|
|
|
$job1->redirect_history([$url1, $url2, $url3]); |
195
|
|
|
|
|
|
|
my $url4 = $job1->original_url; # $url4 is $url3 |
196
|
|
|
|
|
|
|
|
197
|
|
|
|
|
|
|
=head2 upgrade |
198
|
|
|
|
|
|
|
|
199
|
|
|
|
|
|
|
Instanciates a job with string or a L instance. |
200
|
|
|
|
|
|
|
|
201
|
|
|
|
|
|
|
=head1 AUTHOR |
202
|
|
|
|
|
|
|
|
203
|
|
|
|
|
|
|
Keita Sugama, Esugama@jamadam.comE |
204
|
|
|
|
|
|
|
|
205
|
|
|
|
|
|
|
=head1 COPYRIGHT AND LICENSE |
206
|
|
|
|
|
|
|
|
207
|
|
|
|
|
|
|
Copyright (C) Keita Sugama. |
208
|
|
|
|
|
|
|
|
209
|
|
|
|
|
|
|
This program is free software; you can redistribute it and/or |
210
|
|
|
|
|
|
|
modify it under the same terms as Perl itself. |
211
|
|
|
|
|
|
|
|
212
|
|
|
|
|
|
|
=cut |