| line |
stmt |
bran |
cond |
sub |
pod |
time |
code |
|
1
|
|
|
|
|
|
|
package Plucene::Search::BooleanScorer; |
|
2
|
|
|
|
|
|
|
|
|
3
|
|
|
|
|
|
|
=head1 NAME |
|
4
|
|
|
|
|
|
|
|
|
5
|
|
|
|
|
|
|
Plucene::Search::BooleanScorer - A boolean scorer |
|
6
|
|
|
|
|
|
|
|
|
7
|
|
|
|
|
|
|
=head1 SYNOPSIS |
|
8
|
|
|
|
|
|
|
|
|
9
|
|
|
|
|
|
|
# isa Plucene::Search::Scorer |
|
10
|
|
|
|
|
|
|
|
|
11
|
|
|
|
|
|
|
$bool_scorer->add($scorer, $required, $prohibited); |
|
12
|
|
|
|
|
|
|
$bool_scorer->score($results, $max_doc); |
|
13
|
|
|
|
|
|
|
|
|
14
|
|
|
|
|
|
|
=head1 DESCRIPTION |
|
15
|
|
|
|
|
|
|
|
|
16
|
|
|
|
|
|
|
This is a scoring class for boolean scorers. |
|
17
|
|
|
|
|
|
|
|
|
18
|
|
|
|
|
|
|
=head1 METHODS |
|
19
|
|
|
|
|
|
|
|
|
20
|
|
|
|
|
|
|
=cut |
|
21
|
|
|
|
|
|
|
|
|
22
|
3
|
|
|
3
|
|
16
|
use strict; |
|
|
3
|
|
|
|
|
6
|
|
|
|
3
|
|
|
|
|
98
|
|
|
23
|
3
|
|
|
3
|
|
34
|
use warnings; |
|
|
3
|
|
|
|
|
7
|
|
|
|
3
|
|
|
|
|
99
|
|
|
24
|
|
|
|
|
|
|
|
|
25
|
3
|
|
|
3
|
|
16
|
use List::Util qw(min); |
|
|
3
|
|
|
|
|
5
|
|
|
|
3
|
|
|
|
|
189
|
|
|
26
|
|
|
|
|
|
|
|
|
27
|
3
|
|
|
3
|
|
17
|
use Plucene::Search::Similarity; |
|
|
3
|
|
|
|
|
5
|
|
|
|
3
|
|
|
|
|
78
|
|
|
28
|
|
|
|
|
|
|
|
|
29
|
3
|
|
|
3
|
|
14
|
use base qw(Plucene::Search::Scorer Class::Accessor::Fast); |
|
|
3
|
|
|
|
|
7
|
|
|
|
3
|
|
|
|
|
1608
|
|
|
30
|
|
|
|
|
|
|
|
|
31
|
|
|
|
|
|
|
__PACKAGE__->mk_accessors( |
|
32
|
|
|
|
|
|
|
qw(next_mask required_mask prohibited_mask max_coord scorers bucket_table |
|
33
|
|
|
|
|
|
|
coord_factors current_doc) |
|
34
|
|
|
|
|
|
|
); |
|
35
|
|
|
|
|
|
|
|
|
36
|
|
|
|
|
|
|
=head2 new |
|
37
|
|
|
|
|
|
|
|
|
38
|
|
|
|
|
|
|
my $bool_scorer = Plucene::Search::BooleanScorer->new; |
|
39
|
|
|
|
|
|
|
|
|
40
|
|
|
|
|
|
|
Create a new Plucene::Search::BooleanScorer object. |
|
41
|
|
|
|
|
|
|
|
|
42
|
|
|
|
|
|
|
=head2 next_mask / required_mask / prohibited_mask max_coord / scorers / |
|
43
|
|
|
|
|
|
|
bucket_table / coord_factors / current_doc |
|
44
|
|
|
|
|
|
|
|
|
45
|
|
|
|
|
|
|
Get / set these attributes |
|
46
|
|
|
|
|
|
|
|
|
47
|
|
|
|
|
|
|
=cut |
|
48
|
|
|
|
|
|
|
|
|
49
|
|
|
|
|
|
|
sub new { |
|
50
|
16
|
|
|
16
|
1
|
123
|
my $self = shift->SUPER::new(@_); |
|
51
|
16
|
|
|
|
|
260
|
$self->max_coord(1); |
|
52
|
16
|
|
|
|
|
176
|
$self->next_mask(1); |
|
53
|
16
|
|
|
|
|
137
|
$self->current_doc(0); |
|
54
|
16
|
|
|
|
|
141
|
$self->required_mask(0); |
|
55
|
16
|
|
|
|
|
139
|
$self->prohibited_mask(0); |
|
56
|
16
|
|
|
|
|
142
|
$self->scorers([]); |
|
57
|
16
|
|
|
|
|
213
|
$self->bucket_table(Plucene::Search::BucketTable->new({ scorer => $self })); |
|
58
|
16
|
|
|
|
|
138
|
return $self; |
|
59
|
|
|
|
|
|
|
} |
|
60
|
|
|
|
|
|
|
|
|
61
|
|
|
|
|
|
|
=head2 add |
|
62
|
|
|
|
|
|
|
|
|
63
|
|
|
|
|
|
|
$bool_scorer->add($scorer, $required, $prohibited); |
|
64
|
|
|
|
|
|
|
|
|
65
|
|
|
|
|
|
|
=cut |
|
66
|
|
|
|
|
|
|
|
|
67
|
|
|
|
|
|
|
sub add { |
|
68
|
32
|
|
|
32
|
1
|
430
|
my ($self, $scorer, $required, $prohibited) = @_; |
|
69
|
32
|
|
|
|
|
74
|
my $mask = 0; |
|
70
|
32
|
100
|
100
|
|
|
192
|
if ($required || $prohibited) { |
|
71
|
15
|
|
|
|
|
76
|
$mask = $self->next_mask; |
|
72
|
15
|
|
|
|
|
86
|
$self->{next_mask} <<= 1; |
|
73
|
|
|
|
|
|
|
} |
|
74
|
|
|
|
|
|
|
|
|
75
|
32
|
100
|
|
|
|
118
|
$self->{max_coord}++ unless $prohibited; |
|
76
|
|
|
|
|
|
|
|
|
77
|
32
|
100
|
|
|
|
115
|
$self->{prohibited_mask} |= $mask if $prohibited; |
|
78
|
32
|
100
|
|
|
|
110
|
$self->{required_mask} |= $mask if $required; |
|
79
|
32
|
|
|
|
|
66
|
push @{ $self->{scorers} }, |
|
|
32
|
|
|
|
|
161
|
|
|
80
|
|
|
|
|
|
|
{ |
|
81
|
|
|
|
|
|
|
scorer => $scorer, |
|
82
|
|
|
|
|
|
|
required => $required, |
|
83
|
|
|
|
|
|
|
prohibited => $prohibited, |
|
84
|
|
|
|
|
|
|
collector => $self->bucket_table->new_collector($mask) }; |
|
85
|
|
|
|
|
|
|
} |
|
86
|
|
|
|
|
|
|
|
|
87
|
|
|
|
|
|
|
sub _compute_coord_factors { |
|
88
|
16
|
|
|
16
|
|
131
|
my $self = shift; |
|
89
|
16
|
|
|
|
|
76
|
$self->coord_factors([ |
|
90
|
|
|
|
|
|
|
map Plucene::Search::Similarity->coord($_, $self->max_coord), |
|
91
|
|
|
|
|
|
|
0 .. $self->max_coord |
|
92
|
|
|
|
|
|
|
]); |
|
93
|
|
|
|
|
|
|
} |
|
94
|
|
|
|
|
|
|
|
|
95
|
|
|
|
|
|
|
=head2 score |
|
96
|
|
|
|
|
|
|
|
|
97
|
|
|
|
|
|
|
$bool_scorer->score($results, $max_doc); |
|
98
|
|
|
|
|
|
|
|
|
99
|
|
|
|
|
|
|
=cut |
|
100
|
|
|
|
|
|
|
|
|
101
|
|
|
|
|
|
|
sub score { |
|
102
|
16
|
|
|
16
|
1
|
105
|
my ($self, $results, $max_doc) = @_; |
|
103
|
16
|
50
|
|
|
|
73
|
$self->_compute_coord_factors if not defined $self->coord_factors; |
|
104
|
16
|
|
|
|
|
174
|
while ($self->current_doc < $max_doc) { |
|
105
|
16
|
|
|
|
|
195
|
$self->current_doc( |
|
106
|
|
|
|
|
|
|
min( |
|
107
|
|
|
|
|
|
|
$self->{current_doc} + $Plucene::Search::BucketTable::SIZE, $max_doc |
|
108
|
|
|
|
|
|
|
)); |
|
109
|
16
|
|
|
|
|
87
|
for my $t (@{ $self->{scorers} }) { |
|
|
16
|
|
|
|
|
51
|
|
|
110
|
32
|
|
|
|
|
266
|
$t->{scorer}->score($t->{collector}, $self->current_doc); |
|
111
|
|
|
|
|
|
|
} |
|
112
|
16
|
|
|
|
|
239
|
$self->bucket_table->collect_hits($results); |
|
113
|
|
|
|
|
|
|
} |
|
114
|
|
|
|
|
|
|
} |
|
115
|
|
|
|
|
|
|
|
|
116
|
|
|
|
|
|
|
package Plucene::Search::BucketTable; |
|
117
|
|
|
|
|
|
|
our $SIZE = 1 << 10; |
|
118
|
|
|
|
|
|
|
our $MASK = $SIZE - 1; |
|
119
|
|
|
|
|
|
|
|
|
120
|
3
|
|
|
3
|
|
20
|
use base 'Class::Accessor::Fast'; |
|
|
3
|
|
|
|
|
7
|
|
|
|
3
|
|
|
|
|
936
|
|
|
121
|
|
|
|
|
|
|
__PACKAGE__->mk_accessors(qw(buckets first scorer)); |
|
122
|
|
|
|
|
|
|
|
|
123
|
|
|
|
|
|
|
sub new { |
|
124
|
16
|
|
|
16
|
|
89
|
my $self = shift->SUPER::new(@_); |
|
125
|
16
|
|
|
|
|
211
|
$self->buckets([]); |
|
126
|
16
|
|
|
|
|
159
|
$self; |
|
127
|
|
|
|
|
|
|
} |
|
128
|
|
|
|
|
|
|
|
|
129
|
|
|
|
|
|
|
sub collect_hits { |
|
130
|
16
|
|
|
16
|
|
150
|
my ($self, $results) = @_; |
|
131
|
16
|
|
|
|
|
87
|
my $scorer = $self->scorer; |
|
132
|
16
|
|
|
|
|
139
|
my $required = $scorer->required_mask; |
|
133
|
16
|
|
|
|
|
136
|
my $prohibited = $scorer->prohibited_mask; |
|
134
|
16
|
|
|
|
|
87
|
my @coord = @{ $scorer->coord_factors }; |
|
|
16
|
|
|
|
|
81
|
|
|
135
|
|
|
|
|
|
|
|
|
136
|
16
|
|
|
|
|
174
|
for (my $bucket = $self->{first} ; $bucket ; $bucket = $bucket->{next}) { |
|
137
|
136
|
100
|
100
|
|
|
1101
|
if ( ($bucket->{bits} & $prohibited) == 0 |
|
138
|
|
|
|
|
|
|
and ($bucket->{bits} & $required) == $required) { |
|
139
|
56
|
|
|
|
|
298
|
$results->collect($bucket->{doc}, |
|
140
|
|
|
|
|
|
|
$bucket->{score} * $coord[ $bucket->{coord} ]); |
|
141
|
|
|
|
|
|
|
} |
|
142
|
|
|
|
|
|
|
} |
|
143
|
16
|
|
|
|
|
203
|
undef $self->{first}; |
|
144
|
|
|
|
|
|
|
} |
|
145
|
|
|
|
|
|
|
|
|
146
|
|
|
|
|
|
|
sub new_collector { |
|
147
|
32
|
|
|
32
|
|
225
|
my ($self, $mask) = @_; |
|
148
|
32
|
|
|
|
|
311
|
return Plucene::Search::BucketCollector->new({ |
|
149
|
|
|
|
|
|
|
bucket_table => $self, |
|
150
|
|
|
|
|
|
|
mask => $mask |
|
151
|
|
|
|
|
|
|
}); |
|
152
|
|
|
|
|
|
|
} |
|
153
|
|
|
|
|
|
|
|
|
154
|
|
|
|
|
|
|
package Plucene::Search::BucketCollector; |
|
155
|
3
|
|
|
3
|
|
324
|
use base (qw(Class::Accessor::Fast Plucene::Search::HitCollector)); |
|
|
3
|
|
|
|
|
6
|
|
|
|
3
|
|
|
|
|
1145
|
|
|
156
|
|
|
|
|
|
|
|
|
157
|
|
|
|
|
|
|
__PACKAGE__->mk_accessors(qw(bucket_table mask)); |
|
158
|
|
|
|
|
|
|
|
|
159
|
|
|
|
|
|
|
sub collect { |
|
160
|
164
|
|
|
164
|
|
285
|
my ($self, $doc, $score) = @_; |
|
161
|
164
|
|
|
|
|
302
|
my $table = $self->{bucket_table}; |
|
162
|
164
|
|
|
|
|
240
|
my $i = $doc & $Plucene::Search::BucketTable::MASK; |
|
163
|
164
|
|
|
|
|
487
|
my $bucket = $table->buckets->[$i]; |
|
164
|
164
|
100
|
|
|
|
1189
|
$table->buckets->[$i] = $bucket = {} unless $bucket; |
|
165
|
|
|
|
|
|
|
|
|
166
|
164
|
100
|
66
|
|
|
1083
|
if (not defined $bucket->{doc} or $bucket->{doc} != $doc) { |
|
167
|
136
|
|
|
|
|
252
|
@{$bucket}{qw(doc score bits coord)} = |
|
|
136
|
|
|
|
|
585
|
|
|
168
|
|
|
|
|
|
|
($doc, $score, $self->{mask}, 1); |
|
169
|
136
|
|
|
|
|
477
|
$bucket->{next} = $table->first; |
|
170
|
136
|
|
|
|
|
846
|
$table->first($bucket); |
|
171
|
|
|
|
|
|
|
} else { |
|
172
|
28
|
|
|
|
|
68
|
$bucket->{score} += $score; |
|
173
|
28
|
|
|
|
|
68
|
$bucket->{bits} |= $self->{mask}; |
|
174
|
28
|
|
|
|
|
108
|
$bucket->{coord}++; |
|
175
|
|
|
|
|
|
|
} |
|
176
|
|
|
|
|
|
|
} |
|
177
|
|
|
|
|
|
|
|
|
178
|
|
|
|
|
|
|
1; |
|
179
|
|
|
|
|
|
|
|