1 |
dpavlin |
189 |
package Grep::Search::KinoSearch; |
2 |
dpavlin |
47 |
|
3 |
|
|
use strict; |
4 |
|
|
use warnings; |
5 |
|
|
|
6 |
|
|
use Data::Dump qw/dump/; |
7 |
dpavlin |
127 |
use KinoSearch::InvIndexer; |
8 |
|
|
use KinoSearch::Searcher; |
9 |
dpavlin |
47 |
use Jifty::Util; |
10 |
|
|
|
11 |
dpavlin |
190 |
my $debug = 1; |
12 |
dpavlin |
47 |
|
13 |
dpavlin |
109 |
=head1 NAME |
14 |
dpavlin |
47 |
|
15 |
dpavlin |
189 |
Grep::Search::KinoSearch - full text search using L<KinoSearch> |
16 |
dpavlin |
47 |
|
17 |
dpavlin |
109 |
=head1 METHODS |
18 |
dpavlin |
47 |
|
19 |
dpavlin |
185 |
=head2 invindexer |
20 |
|
|
|
21 |
|
|
Accessor to call any method defined on L<KinoSearch::InvIndexer> |
22 |
|
|
|
23 |
|
|
$search->invindexer->delete_by_term( 'id', 42 ); |
24 |
|
|
|
25 |
|
|
=cut |
26 |
|
|
|
27 |
dpavlin |
189 |
our $indexes; |
28 |
dpavlin |
185 |
|
29 |
|
|
sub invindexer { |
30 |
|
|
my $self = shift; |
31 |
|
|
my $invindexer; |
32 |
dpavlin |
190 |
my $index_path = Jifty::Util->app_root . '/var/invindex'; |
33 |
dpavlin |
185 |
|
34 |
|
|
if ( $invindexer = $indexes->{$index_path} ) { |
35 |
|
|
$self->log->debug("Using cached index $index_path"); |
36 |
dpavlin |
47 |
} else { |
37 |
dpavlin |
189 |
if ( $self->create || -e $index_path ) { |
38 |
dpavlin |
185 |
$self->log->debug("Creating new index $index_path"); |
39 |
dpavlin |
189 |
$invindexer = KinoSearch::InvIndexer->new( invindex => Grep::Search::KinoSearch::Schema->clobber( $index_path ) ) |
40 |
|
|
or die "can't create index $index_path: $!"; |
41 |
|
|
$self->create( 0 ); |
42 |
dpavlin |
185 |
} else { |
43 |
|
|
$self->log->debug("Opening index: $index_path"); |
44 |
dpavlin |
189 |
$invindexer = KinoSearch::InvIndexer->new( invindex => Grep::Search::KinoSearch::Schema->open( $index_path ) ) |
45 |
|
|
or die "can't open index $index_path: $!"; |
46 |
dpavlin |
185 |
} |
47 |
|
|
$indexes->{$index_path} = $invindexer; |
48 |
dpavlin |
47 |
} |
49 |
|
|
|
50 |
dpavlin |
189 |
return $invindexer; |
51 |
dpavlin |
49 |
} |
52 |
|
|
|
53 |
dpavlin |
47 |
=head2 add |
54 |
|
|
|
55 |
dpavlin |
190 |
$search->add( $doc_hash ); |
56 |
dpavlin |
47 |
|
57 |
|
|
=cut |
58 |
|
|
|
59 |
|
|
sub add { |
60 |
|
|
my $self = shift; |
61 |
dpavlin |
190 |
my $doc = shift; |
62 |
dpavlin |
127 |
$self->invindexer->add_doc( $doc ); |
63 |
dpavlin |
151 |
return 1; |
64 |
dpavlin |
47 |
} |
65 |
|
|
|
66 |
dpavlin |
190 |
=head2 search |
67 |
dpavlin |
47 |
|
68 |
dpavlin |
190 |
my $fetch_hit_coderef = $self->search('search query'); |
69 |
dpavlin |
141 |
|
70 |
dpavlin |
47 |
=cut |
71 |
|
|
|
72 |
dpavlin |
190 |
sub search { |
73 |
dpavlin |
47 |
my $self = shift; |
74 |
|
|
|
75 |
|
|
my $q = shift or die "no q?"; |
76 |
|
|
|
77 |
dpavlin |
192 |
my $full_q = "($q)"; |
78 |
|
|
my $uid = Jifty->web->current_user->id; |
79 |
|
|
$full_q .= ' AND _owner_id:' . $uid if (defined $uid); |
80 |
|
|
|
81 |
dpavlin |
190 |
my $index_path = Jifty::Util->app_root . '/var/invindex'; |
82 |
dpavlin |
127 |
my $searcher = KinoSearch::Searcher->new( |
83 |
dpavlin |
190 |
invindex => Grep::Search::KinoSearch::Schema->open( $index_path ), ); |
84 |
dpavlin |
98 |
$self->log->debug("$searcher created"); |
85 |
dpavlin |
47 |
|
86 |
dpavlin |
127 |
$self->log->debug("searching for '$q' using $full_q"); |
87 |
dpavlin |
153 |
|
88 |
|
|
my $query_parser = KinoSearch::QueryParser->new( |
89 |
dpavlin |
189 |
schema => Grep::Search::KinoSearch::Schema->new, |
90 |
dpavlin |
153 |
fields => [ qw/ title link content summary category author / ], |
91 |
|
|
); |
92 |
|
|
$query_parser->set_heed_colons(1); # enable field:value AND/OR/NOT syntax |
93 |
|
|
my $query = $query_parser->parse( $full_q ); |
94 |
dpavlin |
127 |
my $hits = $searcher->search( |
95 |
dpavlin |
153 |
query => $query, |
96 |
dpavlin |
127 |
# offset => $offset, |
97 |
|
|
# num_wanted => $hits_per_page, |
98 |
|
|
); |
99 |
dpavlin |
64 |
|
100 |
dpavlin |
141 |
$self->hits( $hits->total_hits ); |
101 |
dpavlin |
47 |
|
102 |
dpavlin |
190 |
return sub { |
103 |
|
|
return $hits->fetch_hit; |
104 |
|
|
}; |
105 |
dpavlin |
47 |
} |
106 |
|
|
|
107 |
|
|
=head2 finish |
108 |
|
|
|
109 |
dpavlin |
109 |
$search->finish |
110 |
dpavlin |
47 |
|
111 |
|
|
=cut |
112 |
|
|
|
113 |
|
|
sub finish { |
114 |
|
|
my $self = shift; |
115 |
dpavlin |
127 |
if ($self->invindexer) { |
116 |
dpavlin |
102 |
$self->log->debug("closing index"); |
117 |
dpavlin |
127 |
$self->invindexer->finish; |
118 |
dpavlin |
47 |
} |
119 |
dpavlin |
58 |
|
120 |
dpavlin |
110 |
$self->log->debug("finish"); |
121 |
|
|
|
122 |
dpavlin |
112 |
undef $self; |
123 |
|
|
|
124 |
dpavlin |
151 |
return 1; |
125 |
dpavlin |
47 |
} |
126 |
|
|
|
127 |
dpavlin |
189 |
package Grep::Search::KinoSearch::KeywordField; |
128 |
dpavlin |
127 |
use base qw( KinoSearch::Schema::FieldSpec ); |
129 |
|
|
sub analyzed { 0 } |
130 |
dpavlin |
153 |
#sub indexed { 1 } |
131 |
|
|
#sub stored { 1 } |
132 |
|
|
sub vectorized { 0 } |
133 |
dpavlin |
127 |
|
134 |
dpavlin |
189 |
package Grep::Search::KinoSearch::Schema; |
135 |
dpavlin |
127 |
|
136 |
|
|
=head1 NAME |
137 |
|
|
|
138 |
dpavlin |
189 |
Grep::Search::KinoSearch::Schema - schema definition for full-text search |
139 |
dpavlin |
127 |
|
140 |
|
|
=cut |
141 |
|
|
|
142 |
|
|
use base 'KinoSearch::Schema'; |
143 |
|
|
use KinoSearch::Analysis::PolyAnalyzer; |
144 |
|
|
|
145 |
dpavlin |
131 |
our %fields = ( |
146 |
dpavlin |
189 |
id => 'Grep::Search::KinoSearch::KeywordField', |
147 |
dpavlin |
127 |
|
148 |
dpavlin |
189 |
in_feed_id => 'Grep::Search::KinoSearch::KeywordField', |
149 |
|
|
in_feed_url => 'Grep::Search::KinoSearch::KeywordField', |
150 |
dpavlin |
127 |
in_feed_title => 'KinoSearch::Schema::FieldSpec', |
151 |
dpavlin |
189 |
in_feed_owner => 'Grep::Search::KinoSearch::KeywordField', |
152 |
|
|
in_feed_created_on => 'Grep::Search::KinoSearch::KeywordField', |
153 |
dpavlin |
127 |
|
154 |
|
|
title => 'KinoSearch::Schema::FieldSpec', |
155 |
dpavlin |
189 |
link => 'Grep::Search::KinoSearch::KeywordField', |
156 |
dpavlin |
127 |
content => 'KinoSearch::Schema::FieldSpec', |
157 |
|
|
summary => 'KinoSearch::Schema::FieldSpec', |
158 |
|
|
category => 'KinoSearch::Schema::FieldSpec', |
159 |
|
|
author => 'KinoSearch::Schema::FieldSpec', |
160 |
dpavlin |
189 |
created_on => 'Grep::Search::KinoSearch::KeywordField', |
161 |
|
|
last_update => 'Grep::Search::KinoSearch::KeywordField', |
162 |
dpavlin |
127 |
|
163 |
dpavlin |
189 |
_owner_id => 'Grep::Search::KinoSearch::KeywordField', |
164 |
dpavlin |
127 |
); |
165 |
|
|
|
166 |
|
|
sub analyzer { |
167 |
|
|
return KinoSearch::Analysis::PolyAnalyzer->new( language => 'en' ); |
168 |
|
|
} |
169 |
|
|
|
170 |
dpavlin |
47 |
1; |