| line |
stmt |
bran |
cond |
sub |
pod |
time |
code |
|
1
|
|
|
|
|
|
|
package WWW::Crawl4AI::Request; |
|
2
|
|
|
|
|
|
|
# ABSTRACT: builds Crawl4AI /crawl and /md request payloads |
|
3
|
5
|
|
|
5
|
|
125880
|
use Moo; |
|
|
5
|
|
|
|
|
5814
|
|
|
|
5
|
|
|
|
|
33
|
|
|
4
|
5
|
|
|
5
|
|
2671
|
use Carp qw( croak ); |
|
|
5
|
|
|
|
|
7
|
|
|
|
5
|
|
|
|
|
2570
|
|
|
5
|
|
|
|
|
|
|
|
|
6
|
|
|
|
|
|
|
our $VERSION = '0.001'; |
|
7
|
|
|
|
|
|
|
|
|
8
|
|
|
|
|
|
|
|
|
9
|
|
|
|
|
|
|
has urls => ( |
|
10
|
|
|
|
|
|
|
is => 'ro', |
|
11
|
|
|
|
|
|
|
required => 1, |
|
12
|
|
|
|
|
|
|
coerce => sub { ref $_[0] eq 'ARRAY' ? $_[0] : [ $_[0] ] }, |
|
13
|
|
|
|
|
|
|
); |
|
14
|
|
|
|
|
|
|
|
|
15
|
|
|
|
|
|
|
|
|
16
|
|
|
|
|
|
|
has browser_params => ( |
|
17
|
|
|
|
|
|
|
is => 'ro', |
|
18
|
|
|
|
|
|
|
default => sub { {} }, |
|
19
|
|
|
|
|
|
|
); |
|
20
|
|
|
|
|
|
|
|
|
21
|
|
|
|
|
|
|
|
|
22
|
|
|
|
|
|
|
has crawler_params => ( |
|
23
|
|
|
|
|
|
|
is => 'ro', |
|
24
|
|
|
|
|
|
|
default => sub { {} }, |
|
25
|
|
|
|
|
|
|
); |
|
26
|
|
|
|
|
|
|
|
|
27
|
|
|
|
|
|
|
|
|
28
|
|
|
|
|
|
|
# /md endpoint extras |
|
29
|
|
|
|
|
|
|
has filter => ( is => 'ro' ); # fit | raw | bm25 | llm |
|
30
|
|
|
|
|
|
|
|
|
31
|
|
|
|
|
|
|
|
|
32
|
|
|
|
|
|
|
has query => ( is => 'ro' ); |
|
33
|
|
|
|
|
|
|
|
|
34
|
|
|
|
|
|
|
|
|
35
|
|
|
|
|
|
|
has cache => ( is => 'ro' ); |
|
36
|
|
|
|
|
|
|
|
|
37
|
|
|
|
|
|
|
|
|
38
|
|
|
|
|
|
|
sub BUILD { |
|
39
|
44
|
|
|
44
|
0
|
307
|
my ( $self ) = @_; |
|
40
|
44
|
100
|
|
|
|
41
|
croak "WWW::Crawl4AI::Request needs at least one url" unless @{ $self->urls }; |
|
|
44
|
|
|
|
|
252
|
|
|
41
|
43
|
|
|
|
|
166
|
return; |
|
42
|
|
|
|
|
|
|
} |
|
43
|
|
|
|
|
|
|
|
|
44
|
|
|
|
|
|
|
sub _default_browser_params { |
|
45
|
23
|
|
|
23
|
|
52
|
return { headless => JSON_true() }; |
|
46
|
|
|
|
|
|
|
} |
|
47
|
|
|
|
|
|
|
|
|
48
|
|
|
|
|
|
|
sub _default_crawler_params { |
|
49
|
23
|
|
|
23
|
|
67
|
return { stream => JSON_false(), cache_mode => 'bypass' }; |
|
50
|
|
|
|
|
|
|
} |
|
51
|
|
|
|
|
|
|
|
|
52
|
|
|
|
|
|
|
# We avoid a hard JSON::PP::Boolean dependency at build time by deferring to |
|
53
|
|
|
|
|
|
|
# JSON::MaybeXS only where booleans are actually serialized. These helpers keep |
|
54
|
|
|
|
|
|
|
# the payload hash JSON-true/false rather than 1/0 (Crawl4AI is strict). |
|
55
|
|
|
|
|
|
|
my ( $TRUE, $FALSE ); |
|
56
|
61
|
|
66
|
61
|
1
|
5657
|
sub JSON_true { $TRUE ||= do { require JSON::MaybeXS; JSON::MaybeXS::true() } } |
|
|
3
|
|
|
|
|
18
|
|
|
|
3
|
|
|
|
|
73
|
|
|
57
|
23
|
|
33
|
23
|
1
|
43
|
sub JSON_false { $FALSE ||= do { require JSON::MaybeXS; JSON::MaybeXS::false() } } |
|
|
23
|
|
|
|
|
163
|
|
|
|
23
|
|
|
|
|
44
|
|
|
58
|
|
|
|
|
|
|
|
|
59
|
|
|
|
|
|
|
|
|
60
|
|
|
|
|
|
|
sub to_crawl_payload { |
|
61
|
23
|
|
|
23
|
1
|
147
|
my ( $self ) = @_; |
|
62
|
23
|
|
|
|
|
27
|
my %browser = ( %{ $self->_default_browser_params }, %{ $self->browser_params } ); |
|
|
23
|
|
|
|
|
37
|
|
|
|
23
|
|
|
|
|
200
|
|
|
63
|
23
|
|
|
|
|
43
|
my %crawler = ( %{ $self->_default_crawler_params }, %{ $self->crawler_params } ); |
|
|
23
|
|
|
|
|
38
|
|
|
|
23
|
|
|
|
|
155
|
|
|
64
|
|
|
|
|
|
|
return { |
|
65
|
23
|
|
|
|
|
177
|
urls => $self->urls, |
|
66
|
|
|
|
|
|
|
browser_config => { type => 'BrowserConfig', params => \%browser }, |
|
67
|
|
|
|
|
|
|
crawler_config => { type => 'CrawlerRunConfig', params => \%crawler }, |
|
68
|
|
|
|
|
|
|
}; |
|
69
|
|
|
|
|
|
|
} |
|
70
|
|
|
|
|
|
|
|
|
71
|
|
|
|
|
|
|
|
|
72
|
|
|
|
|
|
|
sub to_md_payload { |
|
73
|
1
|
|
|
1
|
1
|
5
|
my ( $self ) = @_; |
|
74
|
1
|
|
|
|
|
4
|
my %p = ( url => $self->urls->[0] ); |
|
75
|
1
|
50
|
|
|
|
6
|
$p{f} = $self->filter if defined $self->filter; |
|
76
|
1
|
50
|
|
|
|
4
|
$p{q} = $self->query if defined $self->query; |
|
77
|
1
|
50
|
|
|
|
3
|
$p{c} = $self->cache if defined $self->cache; |
|
78
|
1
|
|
|
|
|
6
|
return \%p; |
|
79
|
|
|
|
|
|
|
} |
|
80
|
|
|
|
|
|
|
|
|
81
|
|
|
|
|
|
|
|
|
82
|
|
|
|
|
|
|
1; |
|
83
|
|
|
|
|
|
|
|
|
84
|
|
|
|
|
|
|
__END__ |