| line |
stmt |
bran |
cond |
sub |
pod |
time |
code |
|
1
|
|
|
|
|
|
|
package WWW::Crawl4AI::StrategyChain; |
|
2
|
|
|
|
|
|
|
# ABSTRACT: ordered list of strategy objects, pluggable at construction time |
|
3
|
2
|
|
|
2
|
|
10
|
use Moo; |
|
|
2
|
|
|
|
|
3
|
|
|
|
2
|
|
|
|
|
9
|
|
|
4
|
2
|
|
|
2
|
|
1288
|
use WWW::Crawl4AI::Strategy::Plain (); |
|
|
2
|
|
|
|
|
5
|
|
|
|
2
|
|
|
|
|
49
|
|
|
5
|
2
|
|
|
2
|
|
741
|
use WWW::Crawl4AI::Strategy::Browser (); |
|
|
2
|
|
|
|
|
5
|
|
|
|
2
|
|
|
|
|
54
|
|
|
6
|
2
|
|
|
2
|
|
750
|
use WWW::Crawl4AI::Strategy::Stealth (); |
|
|
2
|
|
|
|
|
5
|
|
|
|
2
|
|
|
|
|
60
|
|
|
7
|
2
|
|
|
2
|
|
812
|
use WWW::Crawl4AI::Strategy::CloakBrowser (); |
|
|
2
|
|
|
|
|
6
|
|
|
|
2
|
|
|
|
|
47
|
|
|
8
|
2
|
|
|
2
|
|
727
|
use WWW::Crawl4AI::Strategy::Proxy (); |
|
|
2
|
|
|
|
|
5
|
|
|
|
2
|
|
|
|
|
47
|
|
|
9
|
2
|
|
|
2
|
|
782
|
use WWW::Crawl4AI::Strategy::Callback (); |
|
|
2
|
|
|
|
|
5
|
|
|
|
2
|
|
|
|
|
685
|
|
|
10
|
|
|
|
|
|
|
|
|
11
|
|
|
|
|
|
|
our $VERSION = '0.001'; |
|
12
|
|
|
|
|
|
|
|
|
13
|
|
|
|
|
|
|
|
|
14
|
|
|
|
|
|
|
# Default order, cheapest first. Override in subclass or at construction. |
|
15
|
|
|
|
|
|
|
my @CHAIN_CLASSES = qw( |
|
16
|
|
|
|
|
|
|
WWW::Crawl4AI::Strategy::Plain |
|
17
|
|
|
|
|
|
|
WWW::Crawl4AI::Strategy::Browser |
|
18
|
|
|
|
|
|
|
WWW::Crawl4AI::Strategy::Stealth |
|
19
|
|
|
|
|
|
|
WWW::Crawl4AI::Strategy::CloakBrowser |
|
20
|
|
|
|
|
|
|
WWW::Crawl4AI::Strategy::Proxy |
|
21
|
|
|
|
|
|
|
WWW::Crawl4AI::Strategy::Callback |
|
22
|
|
|
|
|
|
|
); |
|
23
|
|
|
|
|
|
|
|
|
24
|
|
|
|
|
|
|
|
|
25
|
16
|
|
|
16
|
1
|
40
|
sub chain_classes { @CHAIN_CLASSES } |
|
26
|
|
|
|
|
|
|
|
|
27
|
|
|
|
|
|
|
has strategies => ( |
|
28
|
|
|
|
|
|
|
is => 'ro', |
|
29
|
|
|
|
|
|
|
builder => 1, |
|
30
|
|
|
|
|
|
|
); |
|
31
|
|
|
|
|
|
|
|
|
32
|
|
|
|
|
|
|
|
|
33
|
|
|
|
|
|
|
sub _build_strategies { |
|
34
|
16
|
|
|
16
|
|
965
|
my ( $self ) = @_; |
|
35
|
16
|
|
|
|
|
29
|
return [ map { $_->new } $self->chain_classes ]; |
|
|
96
|
|
|
|
|
771
|
|
|
36
|
|
|
|
|
|
|
} |
|
37
|
|
|
|
|
|
|
|
|
38
|
|
|
|
|
|
|
sub add_strategy { |
|
39
|
0
|
|
|
0
|
0
|
0
|
my ( $self, $strategy ) = @_; |
|
40
|
0
|
|
|
|
|
0
|
push @{ $self->{strategies} }, $strategy; |
|
|
0
|
|
|
|
|
0
|
|
|
41
|
|
|
|
|
|
|
} |
|
42
|
|
|
|
|
|
|
|
|
43
|
|
|
|
|
|
|
sub remove_strategy { |
|
44
|
0
|
|
|
0
|
0
|
0
|
my ( $self, $name ) = @_; |
|
45
|
0
|
|
|
|
|
0
|
@{ $self->{strategies} } = grep { $_->name ne $name } @{ $self->{strategies} }; |
|
|
0
|
|
|
|
|
0
|
|
|
|
0
|
|
|
|
|
0
|
|
|
|
0
|
|
|
|
|
0
|
|
|
46
|
|
|
|
|
|
|
} |
|
47
|
|
|
|
|
|
|
|
|
48
|
|
|
|
|
|
|
sub replace_strategy { |
|
49
|
0
|
|
|
0
|
0
|
0
|
my ( $self, $name, $strategy ) = @_; |
|
50
|
0
|
|
|
|
|
0
|
my $s = $self->{strategies}; |
|
51
|
0
|
|
|
|
|
0
|
for my $i ( 0 .. $#$s ) { |
|
52
|
0
|
0
|
|
|
|
0
|
$s->[$i] = $strategy if $s->[$i]->name eq $name; |
|
53
|
|
|
|
|
|
|
} |
|
54
|
|
|
|
|
|
|
} |
|
55
|
|
|
|
|
|
|
|
|
56
|
|
|
|
|
|
|
# Filter strategies by applicable($crawler) for this crawler instance. |
|
57
|
|
|
|
|
|
|
sub applicable { |
|
58
|
6
|
|
|
6
|
0
|
36
|
my ( $self, $crawler ) = @_; |
|
59
|
6
|
|
|
|
|
7
|
return [ grep { $_->applicable($crawler) } @{ $self->strategies } ]; |
|
|
36
|
|
|
|
|
79
|
|
|
|
6
|
|
|
|
|
45
|
|
|
60
|
|
|
|
|
|
|
} |
|
61
|
|
|
|
|
|
|
|
|
62
|
|
|
|
|
|
|
1; |
|
63
|
|
|
|
|
|
|
|
|
64
|
|
|
|
|
|
|
__END__ |