2 |
|
|
3 |
use strict; |
use strict; |
4 |
use warnings; |
use warnings; |
5 |
|
use base qw( Class::Accessor Jifty::Object ); |
6 |
|
Grep::Search->mk_accessors( qw( analyzer store writer create index_path ) ); |
7 |
|
|
8 |
use Data::Dump qw/dump/; |
use Data::Dump qw/dump/; |
9 |
use Lucene; |
use Lucene; |
10 |
use Jifty::Util; |
use Jifty::Util; |
11 |
|
|
12 |
my $index_path = Jifty::Util->app_root . '/var/lucene'; |
my $debug = 0; |
13 |
|
|
14 |
my ( $analyzer, $store, $writer ); |
=head1 NAME |
15 |
|
|
16 |
my $debug = 1; |
Grep::Search - full text search |
|
my $create; |
|
17 |
|
|
18 |
sub create { |
=head1 METHODS |
19 |
|
|
20 |
if (defined( $create )) { |
=head2 new |
21 |
Jifty->log->debug("using previous create $create"); |
|
22 |
return $create; |
my $search = Grep::Search->new(); |
23 |
} |
|
24 |
|
=cut |
25 |
|
|
26 |
|
sub new { |
27 |
|
my $class = shift; |
28 |
|
my $self = $class->SUPER::new(@_); |
29 |
|
|
30 |
|
my $index_path = Jifty::Util->app_root . '/var/lucene'; |
31 |
|
|
32 |
|
$self->index_path( $index_path ); |
33 |
|
|
34 |
if (! -e "$index_path/segments") { |
if (! -e "$index_path/segments") { |
35 |
$create = 1; |
$self->create( 1 ); |
36 |
Jifty->log->debug("create index $index_path"); |
$self->log->debug("Creating new index $index_path"); |
37 |
} else { |
} else { |
38 |
$create = 0; |
$self->create( 0 ); |
39 |
Jifty->log->debug("open index: $index_path"); |
$self->log->debug("Opening index: $index_path"); |
40 |
} |
} |
|
return $create; |
|
|
} |
|
41 |
|
|
42 |
sub analyzer { |
$self->analyzer( new Lucene::Analysis::Standard::StandardAnalyzer() ); |
43 |
my $self = shift; |
$self->log->debug($self->analyzer . " created"); |
|
$analyzer ||= new Lucene::Analysis::Standard::StandardAnalyzer(); |
|
|
return $analyzer; |
|
|
} |
|
44 |
|
|
45 |
sub store { |
$self->store( Lucene::Store::FSDirectory->getDirectory( $index_path, $self->create ) ); |
46 |
my $self = shift; |
$self->log->debug($self->store, " created"); |
47 |
|
|
48 |
$store ||= Lucene::Store::FSDirectory->getDirectory( $index_path, $self->create ); |
return $self; |
|
return $store; |
|
49 |
} |
} |
50 |
|
|
|
sub writer { |
|
|
my $self = shift; |
|
|
$writer ||= new Lucene::Index::IndexWriter( $self->store, $self->analyzer, $self->create ); |
|
|
return $writer; |
|
|
} |
|
51 |
|
|
52 |
=head2 add |
=head2 add |
53 |
|
|
54 |
Grep::Search->add( $record ); |
$search->add( $record, $owner_id ); |
55 |
|
|
56 |
=cut |
=cut |
57 |
|
|
59 |
my $self = shift; |
my $self = shift; |
60 |
|
|
61 |
my $i = shift or die "no record to add"; |
my $i = shift or die "no record to add"; |
62 |
|
my $uid = shift; |
63 |
|
|
64 |
die "record not Jifty::Record but ", ref $i unless ($i->isa('Jifty::Record')); |
die "record not Jifty::Record but ", ref $i unless ($i->isa('Jifty::Record')); |
65 |
|
|
110 |
} |
} |
111 |
} |
} |
112 |
|
|
113 |
|
# add _owner_id to speed up filtering of search results |
114 |
|
$uid ||= Jifty->web->current_user->id; |
115 |
|
$doc->add(Lucene::Document::Field->Keyword( '_owner_id', $uid )); |
116 |
|
|
117 |
|
if (! defined( $self->writer )) { |
118 |
|
$self->writer( new Lucene::Index::IndexWriter( $self->store, $self->analyzer, $self->create ) ); |
119 |
|
$self->log->debug($self->writer, " created"); |
120 |
|
} |
121 |
|
|
122 |
$self->writer->addDocument($doc); |
$self->writer->addDocument($doc); |
123 |
|
|
124 |
Jifty->log->debug("added ", $i->id, " to index"); |
$self->log->debug("added ", $i->id, " for user $uid to index"); |
125 |
} |
} |
126 |
|
|
127 |
=head2 |
=head2 collection |
128 |
|
|
129 |
my $ItemCollection = Grep::Search->collection( 'search query' ); |
my $ItemCollection = $search->collection( 'search query' ); |
130 |
|
|
131 |
=cut |
=cut |
132 |
|
|
138 |
return if ( $self->create ); |
return if ( $self->create ); |
139 |
|
|
140 |
my $searcher = new Lucene::Search::IndexSearcher($self->store); |
my $searcher = new Lucene::Search::IndexSearcher($self->store); |
141 |
|
$self->log->debug("$searcher created"); |
142 |
my $parser = new Lucene::QueryParser("content", $self->analyzer); |
my $parser = new Lucene::QueryParser("content", $self->analyzer); |
143 |
my $query = $parser->parse( $q ); |
$self->log->debug("$parser created"); |
144 |
|
|
145 |
Jifty->log->debug("searching for '$q'"); |
my $full_q = "($q) AND _owner_id:" . Jifty->web->current_user->id; |
146 |
|
|
147 |
|
my $query = $parser->parse( $full_q ); |
148 |
|
|
149 |
|
$self->log->debug("searching for '$q' using ", $query->toString); |
150 |
|
|
151 |
my $hits = $searcher->search($query); |
my $hits = $searcher->search($query); |
152 |
my $num_hits = $hits->length(); |
my $num_hits = $hits->length(); |
153 |
|
|
154 |
Jifty->log->debug("found $num_hits results"); |
$self->log->debug("found $num_hits results"); |
155 |
|
|
156 |
my $collection = Grep::Model::ItemCollection->new(); |
my $collection = Grep::Model::ItemCollection->new(); |
157 |
|
|
165 |
my $title = $doc->get("title"); |
my $title = $doc->get("title"); |
166 |
my $id = $doc->get("id"); |
my $id = $doc->get("id"); |
167 |
|
|
168 |
warn "## $i $score $title\n"; |
$self->log->debug("result $i $score $title"); |
169 |
|
|
170 |
my $item = Grep::Model::Item->new(); |
my $item = Grep::Model::Item->new(); |
171 |
my ($ok,$msg) = $item->load_by_cols( id => $id ); |
my ($ok,$msg) = $item->load_by_cols( id => $id ); |
181 |
undef $hits; |
undef $hits; |
182 |
undef $query; |
undef $query; |
183 |
undef $parser; |
undef $parser; |
184 |
|
$searcher->close; |
185 |
undef $searcher; |
undef $searcher; |
186 |
|
|
187 |
return $collection; |
return $collection; |
189 |
|
|
190 |
=head2 finish |
=head2 finish |
191 |
|
|
192 |
Grep::Search->finish |
$search->finish |
193 |
|
|
194 |
=cut |
=cut |
195 |
|
|
196 |
sub finish { |
sub finish { |
197 |
my $self = shift; |
my $self = shift; |
198 |
if ($writer) { |
if ($self->writer) { |
199 |
warn "closing index\n"; |
$self->log->debug("closing index"); |
200 |
$writer->close; |
$self->writer->close; |
201 |
} |
} |
202 |
undef $writer; |
|
203 |
undef $create; |
$self->writer( undef ); |
204 |
|
$self->store( undef ); |
205 |
|
$self->create( undef ); |
206 |
|
$self->analyzer( undef ); |
207 |
|
|
208 |
|
$self->log->debug("finish"); |
209 |
|
|
210 |
return; |
return; |
211 |
} |
} |
229 |
|
|
230 |
my $short = $self->snippet( 50, $text ); |
my $short = $self->snippet( 50, $text ); |
231 |
|
|
|
|
|
232 |
=cut |
=cut |
233 |
|
|
234 |
sub snippet { |
sub snippet { |