line |
stmt |
bran |
cond |
sub |
pod |
time |
code |
1
|
|
|
|
|
|
|
package KinoSearch1::Searcher; |
2
|
18
|
|
|
18
|
|
99107
|
use strict; |
|
18
|
|
|
|
|
262
|
|
|
18
|
|
|
|
|
778
|
|
3
|
18
|
|
|
18
|
|
103
|
use warnings; |
|
18
|
|
|
|
|
39
|
|
|
18
|
|
|
|
|
546
|
|
4
|
18
|
|
|
18
|
|
2053
|
use KinoSearch1::Util::ToolSet; |
|
18
|
|
|
|
|
42
|
|
|
18
|
|
|
|
|
2586
|
|
5
|
18
|
|
|
18
|
|
107
|
use base qw( KinoSearch1::Search::Searchable ); |
|
18
|
|
|
|
|
39
|
|
|
18
|
|
|
|
|
10699
|
|
6
|
|
|
|
|
|
|
|
7
|
|
|
|
|
|
|
BEGIN { |
8
|
18
|
|
|
18
|
|
215
|
__PACKAGE__->init_instance_vars( |
9
|
|
|
|
|
|
|
# params/members |
10
|
|
|
|
|
|
|
invindex => undef, |
11
|
|
|
|
|
|
|
analyzer => undef, |
12
|
|
|
|
|
|
|
# members |
13
|
|
|
|
|
|
|
reader => undef, |
14
|
|
|
|
|
|
|
close_reader => 0, # not implemented yet |
15
|
|
|
|
|
|
|
); |
16
|
18
|
|
|
|
|
170
|
__PACKAGE__->ready_get(qw( reader )); |
17
|
|
|
|
|
|
|
} |
18
|
|
|
|
|
|
|
|
19
|
18
|
|
|
18
|
|
4994
|
use KinoSearch1::Store::FSInvIndex; |
|
18
|
|
|
|
|
61
|
|
|
18
|
|
|
|
|
606
|
|
20
|
18
|
|
|
18
|
|
5386
|
use KinoSearch1::Index::IndexReader; |
|
18
|
|
|
|
|
60
|
|
|
18
|
|
|
|
|
552
|
|
21
|
18
|
|
|
18
|
|
12399
|
use KinoSearch1::Search::Hits; |
|
18
|
|
|
|
|
72
|
|
|
18
|
|
|
|
|
620
|
|
22
|
18
|
|
|
18
|
|
116
|
use KinoSearch1::Search::HitCollector; |
|
18
|
|
|
|
|
36
|
|
|
18
|
|
|
|
|
331
|
|
23
|
18
|
|
|
18
|
|
113
|
use KinoSearch1::Search::Similarity; |
|
18
|
|
|
|
|
30
|
|
|
18
|
|
|
|
|
389
|
|
24
|
18
|
|
|
18
|
|
11678
|
use KinoSearch1::QueryParser::QueryParser; |
|
18
|
|
|
|
|
64
|
|
|
18
|
|
|
|
|
776
|
|
25
|
18
|
|
|
18
|
|
140
|
use KinoSearch1::Search::BooleanQuery; |
|
18
|
|
|
|
|
45
|
|
|
18
|
|
|
|
|
405
|
|
26
|
18
|
|
|
18
|
|
107
|
use KinoSearch1::Analysis::Analyzer; |
|
18
|
|
|
|
|
44
|
|
|
18
|
|
|
|
|
16544
|
|
27
|
|
|
|
|
|
|
|
28
|
|
|
|
|
|
|
sub init_instance { |
29
|
39
|
|
|
39
|
1
|
78
|
my $self = shift; |
30
|
|
|
|
|
|
|
|
31
|
39
|
|
66
|
|
|
342
|
$self->{analyzer} ||= KinoSearch1::Analysis::Analyzer->new; |
32
|
39
|
|
|
|
|
547
|
$self->{similarity} = KinoSearch1::Search::Similarity->new; |
33
|
39
|
|
|
|
|
168
|
$self->{field_sims} = {}; |
34
|
|
|
|
|
|
|
|
35
|
39
|
50
|
|
|
|
162
|
if ( !defined $self->{reader} ) { |
36
|
|
|
|
|
|
|
# confirm or create an InvIndex object |
37
|
39
|
|
|
|
|
69
|
my $invindex; |
38
|
39
|
100
|
66
|
|
|
757
|
if ( blessed( $self->{invindex} ) |
|
|
50
|
|
|
|
|
|
39
|
|
|
|
|
|
|
and $self->{invindex}->isa('KinoSearch1::Store::InvIndex') ) |
40
|
|
|
|
|
|
|
{ |
41
|
36
|
|
|
|
|
79
|
$invindex = $self->{invindex}; |
42
|
|
|
|
|
|
|
} |
43
|
|
|
|
|
|
|
elsif ( defined $self->{invindex} ) { |
44
|
3
|
|
|
|
|
39
|
$invindex = $self->{invindex} |
45
|
|
|
|
|
|
|
= KinoSearch1::Store::FSInvIndex->new( |
46
|
|
|
|
|
|
|
create => $self->{create}, |
47
|
|
|
|
|
|
|
path => $self->{invindex}, |
48
|
|
|
|
|
|
|
); |
49
|
|
|
|
|
|
|
} |
50
|
|
|
|
|
|
|
else { |
51
|
0
|
|
|
|
|
0
|
croak("valid 'reader' or 'invindex' must be supplied"); |
52
|
|
|
|
|
|
|
} |
53
|
|
|
|
|
|
|
|
54
|
|
|
|
|
|
|
# now that we have an invindex, get a reader for it |
55
|
39
|
|
|
|
|
397
|
$self->{reader} = KinoSearch1::Index::IndexReader->new( |
56
|
|
|
|
|
|
|
invindex => $self->{invindex} ); |
57
|
|
|
|
|
|
|
} |
58
|
|
|
|
|
|
|
} |
59
|
|
|
|
|
|
|
|
60
|
|
|
|
|
|
|
my %search_args = ( |
61
|
|
|
|
|
|
|
query => undef, |
62
|
|
|
|
|
|
|
filter => undef, |
63
|
|
|
|
|
|
|
num_docs => undef, |
64
|
|
|
|
|
|
|
); |
65
|
|
|
|
|
|
|
|
66
|
|
|
|
|
|
|
sub search { |
67
|
282
|
|
|
282
|
1
|
13425
|
my $self = shift; |
68
|
282
|
100
|
|
|
|
2247
|
my %args |
69
|
|
|
|
|
|
|
= @_ == 1 |
70
|
|
|
|
|
|
|
? ( %search_args, query => $_[0] ) |
71
|
|
|
|
|
|
|
: ( %search_args, @_ ); |
72
|
282
|
50
|
|
|
|
1395
|
confess kerror() unless verify_args( \%search_args, %args ); |
73
|
|
|
|
|
|
|
|
74
|
|
|
|
|
|
|
# turn a query string into a query against all fields |
75
|
282
|
100
|
|
|
|
1235
|
if ( !a_isa_b( $args{query}, 'KinoSearch1::Search::Query' ) ) { |
76
|
41
|
|
|
|
|
221
|
$args{query} = $self->_prepare_simple_search( $args{query} ); |
77
|
|
|
|
|
|
|
} |
78
|
|
|
|
|
|
|
|
79
|
282
|
|
|
|
|
2226
|
return KinoSearch1::Search::Hits->new( searcher => $self, %args ); |
80
|
|
|
|
|
|
|
} |
81
|
|
|
|
|
|
|
|
82
|
|
|
|
|
|
|
sub get_field_names { |
83
|
42
|
|
|
42
|
0
|
85
|
my $self = shift; |
84
|
42
|
|
|
|
|
230
|
return $self->{reader}->get_field_names(@_); |
85
|
|
|
|
|
|
|
} |
86
|
|
|
|
|
|
|
|
87
|
|
|
|
|
|
|
# Search for the query string against all indexed fields |
88
|
|
|
|
|
|
|
sub _prepare_simple_search { |
89
|
41
|
|
|
41
|
|
96
|
my ( $self, $query_string ) = @_; |
90
|
|
|
|
|
|
|
|
91
|
41
|
|
|
|
|
190
|
my $indexed_field_names = $self->get_field_names( indexed => 1 ); |
92
|
41
|
|
|
|
|
533
|
my $query_parser = KinoSearch1::QueryParser::QueryParser->new( |
93
|
|
|
|
|
|
|
fields => $indexed_field_names, |
94
|
|
|
|
|
|
|
analyzer => $self->{analyzer}, |
95
|
|
|
|
|
|
|
); |
96
|
41
|
|
|
|
|
219
|
return $query_parser->parse($query_string); |
97
|
|
|
|
|
|
|
} |
98
|
|
|
|
|
|
|
|
99
|
|
|
|
|
|
|
my %search_hit_collector_args = ( |
100
|
|
|
|
|
|
|
hit_collector => undef, |
101
|
|
|
|
|
|
|
weight => undef, |
102
|
|
|
|
|
|
|
filter => undef, |
103
|
|
|
|
|
|
|
sort_spec => undef, |
104
|
|
|
|
|
|
|
); |
105
|
|
|
|
|
|
|
|
106
|
|
|
|
|
|
|
sub search_hit_collector { |
107
|
285
|
|
|
285
|
0
|
456
|
my $self = shift; |
108
|
285
|
50
|
|
|
|
1038
|
confess kerror() unless verify_args( \%search_hit_collector_args, @_ ); |
109
|
285
|
|
|
|
|
2043
|
my %args = ( %search_hit_collector_args, @_ ); |
110
|
|
|
|
|
|
|
|
111
|
|
|
|
|
|
|
# wrap the collector if there's a filter |
112
|
285
|
|
|
|
|
656
|
my $collector = $args{hit_collector}; |
113
|
285
|
100
|
|
|
|
940
|
if ( defined $args{filter} ) { |
114
|
1
|
|
|
|
|
7
|
$collector = KinoSearch1::Search::FilteredCollector->new( |
115
|
|
|
|
|
|
|
filter_bits => $args{filter}->bits($self), |
116
|
|
|
|
|
|
|
hit_collector => $args{hit_collector}, |
117
|
|
|
|
|
|
|
); |
118
|
|
|
|
|
|
|
} |
119
|
|
|
|
|
|
|
|
120
|
|
|
|
|
|
|
# accumulate hits into the HitCollector if the query is valid |
121
|
285
|
|
|
|
|
1548
|
my $scorer = $args{weight}->scorer( $self->{reader} ); |
122
|
285
|
100
|
|
|
|
1050
|
if ( defined $scorer ) { |
123
|
253
|
|
|
|
|
1508
|
$scorer->score_batch( |
124
|
|
|
|
|
|
|
hit_collector => $collector, |
125
|
|
|
|
|
|
|
end => $self->{reader}->max_doc, |
126
|
|
|
|
|
|
|
); |
127
|
|
|
|
|
|
|
} |
128
|
|
|
|
|
|
|
} |
129
|
|
|
|
|
|
|
|
130
|
48
|
|
|
48
|
0
|
254
|
sub fetch_doc { $_[0]->{reader}->fetch_doc( $_[1] ) } |
131
|
557
|
|
|
557
|
0
|
2457
|
sub max_doc { shift->{reader}->max_doc } |
132
|
|
|
|
|
|
|
|
133
|
|
|
|
|
|
|
sub doc_freq { |
134
|
1273
|
|
|
1273
|
0
|
1979
|
my ( $self, $term ) = @_; |
135
|
1273
|
|
|
|
|
4312
|
return $self->{reader}->doc_freq($term); |
136
|
|
|
|
|
|
|
} |
137
|
|
|
|
|
|
|
|
138
|
|
|
|
|
|
|
sub create_weight { |
139
|
278
|
|
|
278
|
0
|
484
|
my ( $self, $query ) = @_; |
140
|
278
|
|
|
|
|
1541
|
return $query->to_weight($self); |
141
|
|
|
|
|
|
|
} |
142
|
|
|
|
|
|
|
|
143
|
|
|
|
|
|
|
sub rewrite { |
144
|
279
|
|
|
279
|
0
|
434
|
my ( $self, $query ) = @_; |
145
|
279
|
|
|
|
|
650
|
my $reader = $self->{reader}; |
146
|
279
|
|
|
|
|
764
|
while (1) { |
147
|
279
|
|
|
|
|
1096
|
my $rewritten = $query->rewrite($reader); |
148
|
279
|
50
|
|
|
|
1177
|
last if ( 0 + $rewritten == 0 + $query ); |
149
|
0
|
|
|
|
|
0
|
$query = $rewritten; |
150
|
|
|
|
|
|
|
} |
151
|
279
|
|
|
|
|
738
|
return $query; |
152
|
|
|
|
|
|
|
} |
153
|
|
|
|
|
|
|
|
154
|
|
|
|
|
|
|
sub close { |
155
|
0
|
|
|
0
|
0
|
|
my $self = shift; |
156
|
0
|
0
|
|
|
|
|
$self->{reader}->close if $self->{close_reader}; |
157
|
|
|
|
|
|
|
} |
158
|
|
|
|
|
|
|
|
159
|
|
|
|
|
|
|
1; |
160
|
|
|
|
|
|
|
|
161
|
|
|
|
|
|
|
__END__ |