line |
stmt |
bran |
cond |
sub |
pod |
time |
code |
1
|
|
|
|
|
|
|
package Plucene::Search::Hits; |
2
|
|
|
|
|
|
|
|
3
|
|
|
|
|
|
|
=head1 NAME |
4
|
|
|
|
|
|
|
|
5
|
|
|
|
|
|
|
Plucene::Search::Hits - A list of ranked documents |
6
|
|
|
|
|
|
|
|
7
|
|
|
|
|
|
|
=head1 SYNOPSIS |
8
|
|
|
|
|
|
|
|
9
|
|
|
|
|
|
|
my $hits = Plucene::Search::Hits->new; |
10
|
|
|
|
|
|
|
|
11
|
|
|
|
|
|
|
my $doc = $hits->doc($n); |
12
|
|
|
|
|
|
|
my $score = $hits->score($n); |
13
|
|
|
|
|
|
|
my $hit_doc = $hits->hit_doc($n); |
14
|
|
|
|
|
|
|
|
15
|
|
|
|
|
|
|
=head1 DESCRIPTION |
16
|
|
|
|
|
|
|
|
17
|
|
|
|
|
|
|
This is a list of ranked documents, used to hold search results. |
18
|
|
|
|
|
|
|
|
19
|
|
|
|
|
|
|
=head1 METHODS |
20
|
|
|
|
|
|
|
|
21
|
|
|
|
|
|
|
=cut |
22
|
|
|
|
|
|
|
|
23
|
16
|
|
|
16
|
|
90
|
use strict; |
|
16
|
|
|
|
|
35
|
|
|
16
|
|
|
|
|
550
|
|
24
|
16
|
|
|
16
|
|
196
|
use warnings; |
|
16
|
|
|
|
|
37
|
|
|
16
|
|
|
|
|
460
|
|
25
|
|
|
|
|
|
|
|
26
|
16
|
|
|
16
|
|
106
|
use Carp qw/croak/; |
|
16
|
|
|
|
|
305
|
|
|
16
|
|
|
|
|
1299
|
|
27
|
|
|
|
|
|
|
|
28
|
16
|
|
|
16
|
|
91
|
use Plucene::Search::TopDocs; |
|
16
|
|
|
|
|
31
|
|
|
16
|
|
|
|
|
531
|
|
29
|
|
|
|
|
|
|
|
30
|
16
|
|
|
16
|
|
473
|
use base 'Class::Accessor::Fast'; |
|
16
|
|
|
|
|
31
|
|
|
16
|
|
|
|
|
10207
|
|
31
|
|
|
|
|
|
|
__PACKAGE__->mk_accessors( |
32
|
|
|
|
|
|
|
qw/ query searcher filter length hit_docs |
33
|
|
|
|
|
|
|
first last num_docs max_docs / |
34
|
|
|
|
|
|
|
); |
35
|
|
|
|
|
|
|
|
36
|
|
|
|
|
|
|
=head2 new |
37
|
|
|
|
|
|
|
|
38
|
|
|
|
|
|
|
my $hits = Plucene::Search::Hits->new; |
39
|
|
|
|
|
|
|
|
40
|
|
|
|
|
|
|
=head2 query / searcher / filter / length / hit_docs / first / |
41
|
|
|
|
|
|
|
last / num_docs / max_docs |
42
|
|
|
|
|
|
|
|
43
|
|
|
|
|
|
|
Get / set these attributes. |
44
|
|
|
|
|
|
|
|
45
|
|
|
|
|
|
|
=cut |
46
|
|
|
|
|
|
|
|
47
|
|
|
|
|
|
|
sub new { |
48
|
176
|
|
|
176
|
1
|
871
|
my $self = shift->SUPER::new(@_); |
49
|
176
|
|
|
|
|
2642
|
$self->num_docs(0); |
50
|
176
|
|
|
|
|
1483
|
$self->max_docs(200); |
51
|
176
|
|
|
|
|
1378
|
$self->hit_docs([]); |
52
|
176
|
|
|
|
|
1320
|
$self->_get_more_hits(50); |
53
|
176
|
|
|
|
|
4321
|
return $self; |
54
|
|
|
|
|
|
|
} |
55
|
|
|
|
|
|
|
|
56
|
|
|
|
|
|
|
sub _get_more_hits { |
57
|
176
|
|
|
176
|
|
451
|
my ($self, $min) = @_; |
58
|
176
|
50
|
|
|
|
547
|
if (@{ $self->{hit_docs} } > $min) { $min = @{ $self->{hit_docs} }; } |
|
176
|
|
|
|
|
962
|
|
|
0
|
|
|
|
|
0
|
|
|
0
|
|
|
|
|
0
|
|
59
|
176
|
|
|
|
|
501
|
my $n = $min * 2; |
60
|
176
|
|
|
|
|
785
|
my $top_docs = $self->searcher->search_top($self->query, $self->filter, $n); |
61
|
176
|
|
|
|
|
10990
|
$self->length($top_docs->total_hits); |
62
|
|
|
|
|
|
|
|
63
|
176
|
|
|
|
|
2973
|
my @score_docs = $top_docs->score_docs; |
64
|
|
|
|
|
|
|
|
65
|
176
|
|
|
|
|
420
|
my $score_norm = 1.0; |
66
|
176
|
100
|
100
|
|
|
556
|
$score_norm = 1 / $score_docs[0]->{score} |
67
|
|
|
|
|
|
|
if $self->length > 0 |
68
|
|
|
|
|
|
|
and $score_docs[0]->{score} > 1.0; |
69
|
|
|
|
|
|
|
|
70
|
176
|
50
|
|
|
|
2679
|
my $end = $#score_docs < $self->length ? $#score_docs : $self->length; |
71
|
176
|
|
|
|
|
1376
|
for my $score_doc (@score_docs[ @{ $self->{hit_docs} } .. $end ]) { |
|
176
|
|
|
|
|
974
|
|
72
|
259
|
|
|
|
|
1182
|
push @{ $self->{hit_docs} }, |
|
259
|
|
|
|
|
2392
|
|
73
|
|
|
|
|
|
|
Plucene::Search::HitDoc->new({ |
74
|
|
|
|
|
|
|
score => $score_doc->{score} * $score_norm, |
75
|
|
|
|
|
|
|
id => $score_doc->{doc}, |
76
|
|
|
|
|
|
|
}); |
77
|
|
|
|
|
|
|
} |
78
|
|
|
|
|
|
|
} |
79
|
|
|
|
|
|
|
|
80
|
|
|
|
|
|
|
=head2 doc |
81
|
|
|
|
|
|
|
|
82
|
|
|
|
|
|
|
my $doc = $hits->doc($n); |
83
|
|
|
|
|
|
|
|
84
|
|
|
|
|
|
|
Returns the nth document. |
85
|
|
|
|
|
|
|
|
86
|
|
|
|
|
|
|
=cut |
87
|
|
|
|
|
|
|
|
88
|
|
|
|
|
|
|
sub doc { |
89
|
0
|
|
|
0
|
1
|
0
|
my ($self, $n) = @_; |
90
|
0
|
|
|
|
|
0
|
my $hit = $self->hit_doc($n); |
91
|
|
|
|
|
|
|
|
92
|
|
|
|
|
|
|
# Not sure we need the LRU for now |
93
|
|
|
|
|
|
|
|
94
|
0
|
|
0
|
|
|
0
|
return $hit->doc || $hit->doc($self->searcher->doc($hit->id)); |
95
|
|
|
|
|
|
|
} |
96
|
|
|
|
|
|
|
|
97
|
|
|
|
|
|
|
=head2 score |
98
|
|
|
|
|
|
|
|
99
|
|
|
|
|
|
|
my $score = $hits->score($n); |
100
|
|
|
|
|
|
|
|
101
|
|
|
|
|
|
|
The score of the nth document. |
102
|
|
|
|
|
|
|
|
103
|
|
|
|
|
|
|
=cut |
104
|
|
|
|
|
|
|
|
105
|
|
|
|
|
|
|
sub score { |
106
|
1
|
|
|
1
|
1
|
549
|
my ($self, $n) = @_; |
107
|
1
|
|
|
|
|
4
|
return $self->hit_doc($n)->score; |
108
|
|
|
|
|
|
|
} |
109
|
|
|
|
|
|
|
|
110
|
|
|
|
|
|
|
=head2 hit_doc |
111
|
|
|
|
|
|
|
|
112
|
|
|
|
|
|
|
my $hit_doc = $hits->hit_doc($n); |
113
|
|
|
|
|
|
|
|
114
|
|
|
|
|
|
|
Returns the nth hit document. |
115
|
|
|
|
|
|
|
|
116
|
|
|
|
|
|
|
=cut |
117
|
|
|
|
|
|
|
|
118
|
|
|
|
|
|
|
sub hit_doc { |
119
|
2
|
|
|
2
|
1
|
556
|
my ($self, $n) = @_; |
120
|
2
|
100
|
|
|
|
8
|
if ($n >= $self->length) { |
121
|
1
|
|
|
|
|
229
|
croak("Not a valid hit number: $n"); |
122
|
|
|
|
|
|
|
} |
123
|
1
|
50
|
|
|
|
6
|
$self->_get_more_hits($n) if $n >= @{ $self->{hit_docs} }; |
|
1
|
|
|
|
|
5
|
|
124
|
1
|
|
|
|
|
7
|
return $self->{hit_docs}[$n]; |
125
|
|
|
|
|
|
|
} |
126
|
|
|
|
|
|
|
|
127
|
|
|
|
|
|
|
package Plucene::Search::HitDoc; |
128
|
|
|
|
|
|
|
|
129
|
16
|
|
|
16
|
|
103
|
use base 'Class::Accessor::Fast'; |
|
16
|
|
|
|
|
41
|
|
|
16
|
|
|
|
|
1611
|
|
130
|
|
|
|
|
|
|
|
131
|
|
|
|
|
|
|
__PACKAGE__->mk_accessors(qw/score id doc/); |
132
|
|
|
|
|
|
|
|
133
|
|
|
|
|
|
|
1; |