| line |
stmt |
bran |
cond |
sub |
pod |
time |
code |
|
1
|
|
|
|
|
|
|
package WWW::RobotRules::AnyDBM_File; |
|
2
|
1
|
|
|
1
|
|
119690
|
use strict; |
|
|
1
|
|
|
|
|
2
|
|
|
|
1
|
|
|
|
|
30
|
|
|
3
|
|
|
|
|
|
|
|
|
4
|
1
|
|
|
1
|
|
455
|
use WWW::RobotRules (); |
|
|
1
|
|
|
|
|
3
|
|
|
|
1
|
|
|
|
|
45
|
|
|
5
|
|
|
|
|
|
|
our @ISA = qw(WWW::RobotRules); |
|
6
|
|
|
|
|
|
|
our $VERSION = '6.03'; |
|
7
|
|
|
|
|
|
|
|
|
8
|
1
|
|
|
1
|
|
11
|
use Carp (); |
|
|
1
|
|
|
|
|
1
|
|
|
|
1
|
|
|
|
|
15
|
|
|
9
|
1
|
|
|
1
|
|
660
|
use AnyDBM_File; |
|
|
1
|
|
|
|
|
4422
|
|
|
|
1
|
|
|
|
|
60
|
|
|
10
|
1
|
|
|
1
|
|
8
|
use Fcntl qw( O_CREAT O_RDWR ); |
|
|
1
|
|
|
|
|
6
|
|
|
|
1
|
|
|
|
|
926
|
|
|
11
|
|
|
|
|
|
|
|
|
12
|
|
|
|
|
|
|
sub new { |
|
13
|
4
|
|
|
4
|
1
|
400916
|
my ($class, $ua, $file) = @_; |
|
14
|
4
|
50
|
|
|
|
16
|
Carp::croak('WWW::RobotRules::AnyDBM_File filename required') unless $file; |
|
15
|
|
|
|
|
|
|
|
|
16
|
4
|
|
|
|
|
12
|
my $self = bless {}, $class; |
|
17
|
4
|
|
|
|
|
21
|
$self->{'filename'} = $file; |
|
18
|
4
|
50
|
|
|
|
8
|
tie %{$self->{'dbm'}}, 'AnyDBM_File', $file, O_CREAT | O_RDWR, 0600 |
|
|
4
|
|
|
|
|
564
|
|
|
19
|
|
|
|
|
|
|
or Carp::croak("Can't open $file: $!"); |
|
20
|
|
|
|
|
|
|
|
|
21
|
4
|
100
|
|
|
|
19
|
if ($ua) { |
|
22
|
2
|
|
|
|
|
10
|
$self->agent($ua); |
|
23
|
|
|
|
|
|
|
} |
|
24
|
|
|
|
|
|
|
else { |
|
25
|
|
|
|
|
|
|
# Try to obtain name from DBM file |
|
26
|
2
|
|
|
|
|
45
|
$ua = $self->{'dbm'}{"|ua-name|"}; |
|
27
|
2
|
100
|
|
|
|
295
|
Carp::croak("No agent name specified") unless $ua; |
|
28
|
|
|
|
|
|
|
} |
|
29
|
|
|
|
|
|
|
|
|
30
|
3
|
|
|
|
|
12
|
$self; |
|
31
|
|
|
|
|
|
|
} |
|
32
|
|
|
|
|
|
|
|
|
33
|
|
|
|
|
|
|
sub agent { |
|
34
|
5
|
|
|
5
|
1
|
14
|
my ($self, $newname) = @_; |
|
35
|
5
|
|
|
|
|
58
|
my $old = $self->{'dbm'}{"|ua-name|"}; |
|
36
|
5
|
100
|
|
|
|
18
|
if (defined $newname) { |
|
37
|
2
|
|
|
|
|
21
|
$newname =~ s!/?\s*\d+.\d+\s*$!!; # loose version |
|
38
|
2
|
50
|
66
|
|
|
13
|
unless ($old && $old eq $newname) { |
|
39
|
|
|
|
|
|
|
|
|
40
|
|
|
|
|
|
|
# Old info is now stale. Clear all keys through the tied |
|
41
|
|
|
|
|
|
|
# interface rather than untie+tie(O_TRUNC), which is a |
|
42
|
|
|
|
|
|
|
# symlink-follow TOCTOU on the DBM-backing file(s). |
|
43
|
2
|
|
|
|
|
5
|
%{$self->{'dbm'}} = (); |
|
|
2
|
|
|
|
|
15
|
|
|
44
|
2
|
|
|
|
|
197
|
$self->{'dbm'}{"|ua-name|"} = $newname; |
|
45
|
|
|
|
|
|
|
} |
|
46
|
|
|
|
|
|
|
} |
|
47
|
5
|
|
|
|
|
17
|
$old; |
|
48
|
|
|
|
|
|
|
} |
|
49
|
|
|
|
|
|
|
|
|
50
|
|
|
|
|
|
|
sub no_visits { |
|
51
|
5
|
|
|
5
|
0
|
28
|
my ($self, $netloc) = @_; |
|
52
|
5
|
|
|
|
|
23
|
my $t = $self->{'dbm'}{"$netloc|vis"}; |
|
53
|
5
|
100
|
|
|
|
19
|
return 0 unless $t; |
|
54
|
4
|
|
|
|
|
36
|
(split(/;\s*/, $t))[0]; |
|
55
|
|
|
|
|
|
|
} |
|
56
|
|
|
|
|
|
|
|
|
57
|
|
|
|
|
|
|
sub last_visit { |
|
58
|
2
|
|
|
2
|
0
|
4
|
my ($self, $netloc) = @_; |
|
59
|
2
|
|
|
|
|
14
|
my $t = $self->{'dbm'}{"$netloc|vis"}; |
|
60
|
2
|
50
|
|
|
|
7
|
return undef unless $t; |
|
61
|
2
|
|
|
|
|
21
|
(split(/;\s*/, $t))[1]; |
|
62
|
|
|
|
|
|
|
} |
|
63
|
|
|
|
|
|
|
|
|
64
|
|
|
|
|
|
|
sub fresh_until { |
|
65
|
6
|
|
|
6
|
0
|
22
|
my ($self, $netloc, $fresh) = @_; |
|
66
|
6
|
|
|
|
|
60
|
my $old = $self->{'dbm'}{"$netloc|exp"}; |
|
67
|
6
|
100
|
|
|
|
24
|
if ($old) { |
|
68
|
3
|
|
|
|
|
18
|
$old =~ s/;.*//; # remove cleartext |
|
69
|
|
|
|
|
|
|
} |
|
70
|
6
|
100
|
|
|
|
24
|
if (defined $fresh) { |
|
71
|
2
|
|
|
|
|
68
|
$fresh .= "; " . localtime($fresh); |
|
72
|
2
|
|
|
|
|
90
|
$self->{'dbm'}{"$netloc|exp"} = $fresh; |
|
73
|
|
|
|
|
|
|
} |
|
74
|
6
|
|
|
|
|
25
|
$old; |
|
75
|
|
|
|
|
|
|
} |
|
76
|
|
|
|
|
|
|
|
|
77
|
|
|
|
|
|
|
sub visit { |
|
78
|
4
|
|
|
4
|
0
|
587
|
my ($self, $netloc, $time) = @_; |
|
79
|
4
|
|
66
|
|
|
17
|
$time ||= time; |
|
80
|
|
|
|
|
|
|
|
|
81
|
4
|
|
|
|
|
4
|
my $count = 0; |
|
82
|
4
|
|
|
|
|
16
|
my $old = $self->{'dbm'}{"$netloc|vis"}; |
|
83
|
4
|
100
|
|
|
|
11
|
if ($old) { |
|
84
|
2
|
|
|
|
|
3
|
my $last; |
|
85
|
2
|
|
|
|
|
12
|
($count, $last) = split(/;\s*/, $old); |
|
86
|
2
|
100
|
|
|
|
7
|
$time = $last if $last > $time; |
|
87
|
|
|
|
|
|
|
} |
|
88
|
4
|
|
|
|
|
6
|
$count++; |
|
89
|
4
|
|
|
|
|
173
|
$self->{'dbm'}{"$netloc|vis"} = "$count; $time; " . localtime($time); |
|
90
|
|
|
|
|
|
|
} |
|
91
|
|
|
|
|
|
|
|
|
92
|
|
|
|
|
|
|
sub push_rules { |
|
93
|
4
|
|
|
4
|
0
|
12
|
my ($self, $netloc, @rules) = @_; |
|
94
|
4
|
|
|
|
|
6
|
my $cnt = 1; |
|
95
|
4
|
|
|
|
|
35
|
$cnt++ while $self->{'dbm'}{"$netloc|r$cnt"}; |
|
96
|
|
|
|
|
|
|
|
|
97
|
4
|
|
|
|
|
20
|
foreach (@rules) { |
|
98
|
6
|
|
|
|
|
130
|
$self->{'dbm'}{"$netloc|r$cnt"} = $_; |
|
99
|
6
|
|
|
|
|
24
|
$cnt++; |
|
100
|
|
|
|
|
|
|
} |
|
101
|
|
|
|
|
|
|
} |
|
102
|
|
|
|
|
|
|
|
|
103
|
|
|
|
|
|
|
sub clear_rules { |
|
104
|
4
|
|
|
4
|
0
|
632
|
my ($self, $netloc) = @_; |
|
105
|
4
|
|
|
|
|
6
|
my $cnt = 1; |
|
106
|
4
|
|
|
|
|
64
|
while ($self->{'dbm'}{"$netloc|r$cnt"}) { |
|
107
|
4
|
|
|
|
|
66
|
delete $self->{'dbm'}{"$netloc|r$cnt"}; |
|
108
|
4
|
|
|
|
|
18
|
$cnt++; |
|
109
|
|
|
|
|
|
|
} |
|
110
|
|
|
|
|
|
|
} |
|
111
|
|
|
|
|
|
|
|
|
112
|
|
|
|
|
|
|
sub rules { |
|
113
|
4
|
|
|
4
|
0
|
12
|
my ($self, $netloc) = @_; |
|
114
|
4
|
|
|
|
|
9
|
my @rules = (); |
|
115
|
4
|
|
|
|
|
7
|
my $cnt = 1; |
|
116
|
4
|
|
|
|
|
7
|
while (1) { |
|
117
|
12
|
|
|
|
|
68
|
my $rule = $self->{'dbm'}{"$netloc|r$cnt"}; |
|
118
|
12
|
100
|
|
|
|
34
|
last unless $rule; |
|
119
|
8
|
|
|
|
|
14
|
push(@rules, $rule); |
|
120
|
8
|
|
|
|
|
14
|
$cnt++; |
|
121
|
|
|
|
|
|
|
} |
|
122
|
4
|
|
|
|
|
15
|
@rules; |
|
123
|
|
|
|
|
|
|
} |
|
124
|
|
|
|
|
|
|
|
|
125
|
|
|
|
0
|
0
|
|
sub dump { } |
|
126
|
|
|
|
|
|
|
|
|
127
|
|
|
|
|
|
|
1; |
|
128
|
|
|
|
|
|
|
__END__ |