| line |
stmt |
bran |
cond |
sub |
pod |
time |
code |
|
1
|
|
|
|
|
|
|
package HTML::Microdata; |
|
2
|
|
|
|
|
|
|
|
|
3
|
2
|
|
|
2
|
|
25985
|
use strict; |
|
|
2
|
|
|
|
|
5
|
|
|
|
2
|
|
|
|
|
66
|
|
|
4
|
2
|
|
|
2
|
|
11
|
use warnings; |
|
|
2
|
|
|
|
|
3
|
|
|
|
2
|
|
|
|
|
181
|
|
|
5
|
|
|
|
|
|
|
|
|
6
|
2
|
|
|
2
|
|
2708
|
use HTML::TreeBuilder::LibXML; |
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
7
|
|
|
|
|
|
|
use Hash::MultiValue; |
|
8
|
|
|
|
|
|
|
use Scalar::Util qw(refaddr); |
|
9
|
|
|
|
|
|
|
use JSON; |
|
10
|
|
|
|
|
|
|
use URI; |
|
11
|
|
|
|
|
|
|
|
|
12
|
|
|
|
|
|
|
our $VERSION = '0.04'; |
|
13
|
|
|
|
|
|
|
|
|
14
|
|
|
|
|
|
|
sub new { |
|
15
|
|
|
|
|
|
|
my ($class, %args) = @_; |
|
16
|
|
|
|
|
|
|
bless { |
|
17
|
|
|
|
|
|
|
items => [], |
|
18
|
|
|
|
|
|
|
base => $args{base} ? URI->new($args{base}) : undef, |
|
19
|
|
|
|
|
|
|
}, $class; |
|
20
|
|
|
|
|
|
|
} |
|
21
|
|
|
|
|
|
|
|
|
22
|
|
|
|
|
|
|
sub extract { |
|
23
|
|
|
|
|
|
|
my ($class, $content, %opts) = @_; |
|
24
|
|
|
|
|
|
|
my $self = $class->new(%opts); |
|
25
|
|
|
|
|
|
|
$self->_parse($content); |
|
26
|
|
|
|
|
|
|
$self |
|
27
|
|
|
|
|
|
|
} |
|
28
|
|
|
|
|
|
|
|
|
29
|
|
|
|
|
|
|
sub as_json { |
|
30
|
|
|
|
|
|
|
my ($self) = @_; |
|
31
|
|
|
|
|
|
|
encode_json +{ |
|
32
|
|
|
|
|
|
|
items => $self->{items}, |
|
33
|
|
|
|
|
|
|
}; |
|
34
|
|
|
|
|
|
|
} |
|
35
|
|
|
|
|
|
|
|
|
36
|
|
|
|
|
|
|
sub items { |
|
37
|
|
|
|
|
|
|
my ($self) = @_; |
|
38
|
|
|
|
|
|
|
$self->{items}; |
|
39
|
|
|
|
|
|
|
} |
|
40
|
|
|
|
|
|
|
|
|
41
|
|
|
|
|
|
|
sub _parse { |
|
42
|
|
|
|
|
|
|
my ($self, $content) = @_; |
|
43
|
|
|
|
|
|
|
|
|
44
|
|
|
|
|
|
|
my $items = {}; |
|
45
|
|
|
|
|
|
|
my $tree = HTML::TreeBuilder::LibXML->new_from_content($content); |
|
46
|
|
|
|
|
|
|
my $scopes = $tree->findnodes('//*[@itemscope]'); |
|
47
|
|
|
|
|
|
|
my $number = 0; |
|
48
|
|
|
|
|
|
|
|
|
49
|
|
|
|
|
|
|
for my $scope (@$scopes) { |
|
50
|
|
|
|
|
|
|
my $type = $scope->attr('itemtype'); |
|
51
|
|
|
|
|
|
|
my $id = $scope->attr('itemid'); |
|
52
|
|
|
|
|
|
|
|
|
53
|
|
|
|
|
|
|
unless ($scope->id) { |
|
54
|
|
|
|
|
|
|
$scope->id($number++); |
|
55
|
|
|
|
|
|
|
} |
|
56
|
|
|
|
|
|
|
|
|
57
|
|
|
|
|
|
|
my $item = { |
|
58
|
|
|
|
|
|
|
($id ? (id => $id) : ()), |
|
59
|
|
|
|
|
|
|
($type ? (type => $type) : ()), |
|
60
|
|
|
|
|
|
|
properties => Hash::MultiValue->new, |
|
61
|
|
|
|
|
|
|
}; |
|
62
|
|
|
|
|
|
|
|
|
63
|
|
|
|
|
|
|
$items->{ $scope->id } = $item; |
|
64
|
|
|
|
|
|
|
|
|
65
|
|
|
|
|
|
|
unless ($scope->attr('itemprop')) { |
|
66
|
|
|
|
|
|
|
# This is top level item |
|
67
|
|
|
|
|
|
|
push @{ $self->{items} }, $item; |
|
68
|
|
|
|
|
|
|
} |
|
69
|
|
|
|
|
|
|
} |
|
70
|
|
|
|
|
|
|
|
|
71
|
|
|
|
|
|
|
for my $scope (@$scopes) { |
|
72
|
|
|
|
|
|
|
if (my $refs = $scope->attr('itemref')) { |
|
73
|
|
|
|
|
|
|
my $ids = [ split /\s+/, $refs ]; |
|
74
|
|
|
|
|
|
|
for my $id (@$ids) { |
|
75
|
|
|
|
|
|
|
my $props = $tree->findnodes('//*[@id="' . $id . '"]/descendant-or-self::*[@itemprop]'); |
|
76
|
|
|
|
|
|
|
for my $prop (@$props) { |
|
77
|
|
|
|
|
|
|
my $name = $prop->attr('itemprop'); |
|
78
|
|
|
|
|
|
|
my $value = $self->extract_value($prop, items => $items); |
|
79
|
|
|
|
|
|
|
$items->{ $scope->id }->{properties}->add($name => $value); |
|
80
|
|
|
|
|
|
|
$prop->delete; |
|
81
|
|
|
|
|
|
|
} |
|
82
|
|
|
|
|
|
|
} |
|
83
|
|
|
|
|
|
|
} |
|
84
|
|
|
|
|
|
|
} |
|
85
|
|
|
|
|
|
|
|
|
86
|
|
|
|
|
|
|
my $props = $tree->findnodes('//*[@itemscope]/descendant-or-self::*[@itemprop]'); |
|
87
|
|
|
|
|
|
|
for my $prop (@$props) { |
|
88
|
|
|
|
|
|
|
my $value = $self->extract_value($prop, items => $items); |
|
89
|
|
|
|
|
|
|
my $scope = $prop->findnodes('./ancestor::*[@itemscope]')->[-1]; |
|
90
|
|
|
|
|
|
|
for my $name (split /\s+/, $prop->attr('itemprop')) { |
|
91
|
|
|
|
|
|
|
$items->{ $scope->id }->{properties}->add($name => $value); |
|
92
|
|
|
|
|
|
|
} |
|
93
|
|
|
|
|
|
|
} |
|
94
|
|
|
|
|
|
|
|
|
95
|
|
|
|
|
|
|
for my $key (keys %$items) { |
|
96
|
|
|
|
|
|
|
my $item = $items->{$key}; |
|
97
|
|
|
|
|
|
|
$item->{properties} = $item->{properties}->multi; |
|
98
|
|
|
|
|
|
|
} |
|
99
|
|
|
|
|
|
|
|
|
100
|
|
|
|
|
|
|
} |
|
101
|
|
|
|
|
|
|
|
|
102
|
|
|
|
|
|
|
sub absolute { |
|
103
|
|
|
|
|
|
|
my ($self, $uri) = @_; |
|
104
|
|
|
|
|
|
|
if (defined $uri) { |
|
105
|
|
|
|
|
|
|
if ($self->{base}) { |
|
106
|
|
|
|
|
|
|
URI->new_abs($uri, $self->{base}).q(); |
|
107
|
|
|
|
|
|
|
} else { |
|
108
|
|
|
|
|
|
|
$uri; |
|
109
|
|
|
|
|
|
|
} |
|
110
|
|
|
|
|
|
|
} else { |
|
111
|
|
|
|
|
|
|
""; |
|
112
|
|
|
|
|
|
|
} |
|
113
|
|
|
|
|
|
|
} |
|
114
|
|
|
|
|
|
|
|
|
115
|
|
|
|
|
|
|
sub extract_value { |
|
116
|
|
|
|
|
|
|
my ($self, $prop, %opts) = @_; |
|
117
|
|
|
|
|
|
|
|
|
118
|
|
|
|
|
|
|
my $value; |
|
119
|
|
|
|
|
|
|
if (defined $prop->attr('itemscope')) { |
|
120
|
|
|
|
|
|
|
# XXX : inifinite loop |
|
121
|
|
|
|
|
|
|
$value = $opts{items}->{ $prop->id }; |
|
122
|
|
|
|
|
|
|
} elsif ($prop->tag eq 'meta') { |
|
123
|
|
|
|
|
|
|
$value = $prop->attr('content'); |
|
124
|
|
|
|
|
|
|
} elsif ($prop->tag =~ m{^audio|embed|iframe|img|source|video|track$}) { |
|
125
|
|
|
|
|
|
|
$value = $self->absolute($prop->attr('src')); |
|
126
|
|
|
|
|
|
|
} elsif ($prop->tag =~ m{^a|area|link$}) { |
|
127
|
|
|
|
|
|
|
$value = $self->absolute($prop->attr('href')); |
|
128
|
|
|
|
|
|
|
} elsif ($prop->tag eq 'object') { |
|
129
|
|
|
|
|
|
|
$value = $self->absolute($prop->attr('data')); |
|
130
|
|
|
|
|
|
|
} elsif ($prop->tag eq 'data') { |
|
131
|
|
|
|
|
|
|
$value = $prop->attr('value'); |
|
132
|
|
|
|
|
|
|
} elsif ($prop->tag eq 'time' && $prop->attr('datetime')) { |
|
133
|
|
|
|
|
|
|
$value = $prop->attr('datetime'); |
|
134
|
|
|
|
|
|
|
} else { |
|
135
|
|
|
|
|
|
|
$value = $prop->findvalue('normalize-space(.)'); |
|
136
|
|
|
|
|
|
|
} |
|
137
|
|
|
|
|
|
|
|
|
138
|
|
|
|
|
|
|
$value; |
|
139
|
|
|
|
|
|
|
} |
|
140
|
|
|
|
|
|
|
|
|
141
|
|
|
|
|
|
|
1; |
|
142
|
|
|
|
|
|
|
__END__ |