| line |
stmt |
bran |
cond |
sub |
pod |
time |
code |
|
1
|
|
|
|
|
|
|
package Duadua::Parser; |
|
2
|
11
|
|
|
11
|
|
1146
|
use strict; |
|
|
11
|
|
|
|
|
20
|
|
|
|
11
|
|
|
|
|
400
|
|
|
3
|
11
|
|
|
11
|
|
60
|
use warnings; |
|
|
11
|
|
|
|
|
21
|
|
|
|
11
|
|
|
|
|
564
|
|
|
4
|
11
|
|
|
11
|
|
4306
|
use Duadua::Util; |
|
|
11
|
|
|
|
|
36
|
|
|
|
11
|
|
|
|
|
11590
|
|
|
5
|
|
|
|
|
|
|
|
|
6
|
|
|
|
|
|
|
my $BLANK_UA = { |
|
7
|
|
|
|
|
|
|
name => 'UNKNOWN', |
|
8
|
|
|
|
|
|
|
}; |
|
9
|
|
|
|
|
|
|
|
|
10
|
|
|
|
|
|
|
sub parse { |
|
11
|
444
|
|
|
444
|
1
|
1093
|
my ($class, $d) = @_; |
|
12
|
|
|
|
|
|
|
|
|
13
|
444
|
|
|
|
|
1157
|
my $index_mozilla = index($d->ua, 'Mozilla/'); |
|
14
|
444
|
100
|
|
|
|
1651
|
if ($index_mozilla == 0) { |
|
|
|
100
|
|
|
|
|
|
|
15
|
253
|
|
|
|
|
620
|
$d->{_contain_mozilla_top} = 1; |
|
16
|
253
|
|
|
|
|
537
|
$d->{_contain_mozilla} = 1; |
|
17
|
|
|
|
|
|
|
} elsif ($index_mozilla > 0) { |
|
18
|
4
|
|
|
|
|
14
|
$d->{_contain_mozilla} = 1; |
|
19
|
|
|
|
|
|
|
} |
|
20
|
|
|
|
|
|
|
|
|
21
|
444
|
|
|
|
|
731
|
for my $m (@{$d->parsers}) { |
|
|
444
|
|
|
|
|
1219
|
|
|
22
|
13115
|
100
|
|
|
|
57034
|
if ( my $res = $m->try($d) ) { |
|
23
|
410
|
|
|
|
|
1883
|
return $res; |
|
24
|
|
|
|
|
|
|
} |
|
25
|
|
|
|
|
|
|
} |
|
26
|
|
|
|
|
|
|
|
|
27
|
|
|
|
|
|
|
# Blank or '-' |
|
28
|
34
|
100
|
100
|
|
|
94
|
if ($d->ua eq '' || $d->ua eq '-') { |
|
29
|
3
|
|
|
|
|
15
|
return $BLANK_UA; |
|
30
|
|
|
|
|
|
|
} |
|
31
|
|
|
|
|
|
|
|
|
32
|
31
|
100
|
|
|
|
199
|
if ( my $browser = $class->_detect_general_browser($d) ) { |
|
33
|
2
|
|
|
|
|
9
|
return $browser; |
|
34
|
|
|
|
|
|
|
} |
|
35
|
|
|
|
|
|
|
|
|
36
|
29
|
100
|
|
|
|
92
|
if ( my $bot = $class->_detect_general_bot($d) ) { |
|
37
|
24
|
|
|
|
|
159
|
return $bot; |
|
38
|
|
|
|
|
|
|
} |
|
39
|
|
|
|
|
|
|
|
|
40
|
5
|
|
|
|
|
58
|
return $BLANK_UA; |
|
41
|
|
|
|
|
|
|
} |
|
42
|
|
|
|
|
|
|
|
|
43
|
|
|
|
|
|
|
sub _detect_general_browser { |
|
44
|
31
|
|
|
31
|
|
84
|
my ($class, $d) = @_; |
|
45
|
|
|
|
|
|
|
|
|
46
|
31
|
100
|
66
|
|
|
92
|
if ( index($d->ua, 'rowser') > 0 && $d->_prefix('Mozilla/') ) { |
|
47
|
2
|
50
|
|
|
|
7
|
if ( $d->ua =~ m/([\d.]+)! ) { |
|
48
|
2
|
|
|
|
|
5
|
my %h = %{$BLANK_UA}; |
|
|
2
|
|
|
|
|
11
|
|
|
49
|
2
|
|
|
|
|
14
|
($h{name}, $h{version}) = ($1, $2); |
|
50
|
2
|
|
|
|
|
15
|
return Duadua::Util->set_os($d, \%h); |
|
51
|
|
|
|
|
|
|
} |
|
52
|
|
|
|
|
|
|
} |
|
53
|
|
|
|
|
|
|
} |
|
54
|
|
|
|
|
|
|
|
|
55
|
|
|
|
|
|
|
sub _detect_general_bot { |
|
56
|
29
|
|
|
29
|
|
70
|
my ($class, $d) = @_; |
|
57
|
|
|
|
|
|
|
|
|
58
|
29
|
|
|
|
|
60
|
my %h = %{$BLANK_UA}; |
|
|
29
|
|
|
|
|
148
|
|
|
59
|
|
|
|
|
|
|
|
|
60
|
29
|
100
|
100
|
|
|
101
|
if ( $d->_contain('https://') || $d->_contain('http://') ) { |
|
61
|
24
|
|
|
|
|
71
|
$h{is_bot} = 1; |
|
62
|
24
|
100
|
100
|
|
|
63
|
if ( !$d->_contain('Mozilla/') && $d->ua =~ m!^([^/;]+)/(v?[\d.]+)! ) { |
|
|
|
100
|
|
|
|
|
|
|
|
|
50
|
|
|
|
|
|
|
63
|
9
|
|
|
|
|
43
|
my ($name, $version) = ($1, $2); |
|
64
|
9
|
|
|
|
|
26
|
$h{name} = $name; |
|
65
|
9
|
|
|
|
|
32
|
$h{version} = $version; |
|
66
|
|
|
|
|
|
|
} |
|
67
|
|
|
|
|
|
|
elsif ( $d->ua =~ m)/(v?[\d.]+)!i ) { |
|
68
|
14
|
|
|
|
|
69
|
my ($name, $version) = ($1, $2); |
|
69
|
14
|
|
|
|
|
42
|
$h{name} = $1; |
|
70
|
14
|
|
|
|
|
49
|
$h{version} = $version; |
|
71
|
|
|
|
|
|
|
} |
|
72
|
|
|
|
|
|
|
elsif ( $d->ua =~ m!([a-zA-Z0-9\-\_\.\!]+(?:bot|crawler|spider))!i ) { |
|
73
|
1
|
|
|
|
|
5
|
$h{name} = $1; |
|
74
|
|
|
|
|
|
|
} |
|
75
|
|
|
|
|
|
|
|
|
76
|
24
|
|
|
|
|
112
|
return \%h; |
|
77
|
|
|
|
|
|
|
} |
|
78
|
|
|
|
|
|
|
} |
|
79
|
|
|
|
|
|
|
|
|
80
|
|
|
|
|
|
|
1; |
|
81
|
|
|
|
|
|
|
|
|
82
|
|
|
|
|
|
|
__END__ |