1 |
dpavlin |
47 |
package Grep::Search; |
2 |
|
|
|
3 |
|
|
use strict; |
4 |
|
|
use warnings; |
5 |
dpavlin |
112 |
use base qw( Class::Accessor ); |
6 |
dpavlin |
109 |
Grep::Search->mk_accessors( qw( analyzer store writer create index_path ) ); |
7 |
dpavlin |
47 |
|
8 |
|
|
use Data::Dump qw/dump/; |
9 |
|
|
use Lucene; |
10 |
|
|
use Jifty::Util; |
11 |
|
|
|
12 |
dpavlin |
109 |
my $debug = 0; |
13 |
dpavlin |
47 |
|
14 |
dpavlin |
109 |
=head1 NAME |
15 |
dpavlin |
47 |
|
16 |
dpavlin |
109 |
Grep::Search - full text search |
17 |
dpavlin |
47 |
|
18 |
dpavlin |
109 |
=head1 METHODS |
19 |
dpavlin |
47 |
|
20 |
dpavlin |
109 |
=head2 new |
21 |
dpavlin |
58 |
|
22 |
dpavlin |
109 |
my $search = Grep::Search->new(); |
23 |
|
|
|
24 |
|
|
=cut |
25 |
|
|
|
26 |
dpavlin |
112 |
sub log { Jifty->web->log } |
27 |
|
|
|
28 |
dpavlin |
109 |
sub new { |
29 |
|
|
my $class = shift; |
30 |
|
|
my $self = $class->SUPER::new(@_); |
31 |
|
|
|
32 |
|
|
my $index_path = Jifty::Util->app_root . '/var/lucene'; |
33 |
|
|
|
34 |
|
|
$self->index_path( $index_path ); |
35 |
|
|
|
36 |
dpavlin |
47 |
if (! -e "$index_path/segments") { |
37 |
dpavlin |
109 |
$self->create( 1 ); |
38 |
|
|
$self->log->debug("Creating new index $index_path"); |
39 |
dpavlin |
47 |
} else { |
40 |
dpavlin |
109 |
$self->create( 0 ); |
41 |
|
|
$self->log->debug("Opening index: $index_path"); |
42 |
dpavlin |
47 |
} |
43 |
|
|
|
44 |
dpavlin |
109 |
$self->analyzer( new Lucene::Analysis::Standard::StandardAnalyzer() ); |
45 |
|
|
$self->log->debug($self->analyzer . " created"); |
46 |
dpavlin |
49 |
|
47 |
dpavlin |
109 |
$self->store( Lucene::Store::FSDirectory->getDirectory( $index_path, $self->create ) ); |
48 |
|
|
$self->log->debug($self->store, " created"); |
49 |
dpavlin |
49 |
|
50 |
dpavlin |
109 |
return $self; |
51 |
dpavlin |
49 |
} |
52 |
|
|
|
53 |
dpavlin |
47 |
=head2 add |
54 |
|
|
|
55 |
dpavlin |
109 |
$search->add( $record, $owner_id ); |
56 |
dpavlin |
47 |
|
57 |
|
|
=cut |
58 |
|
|
|
59 |
|
|
sub add { |
60 |
|
|
my $self = shift; |
61 |
|
|
|
62 |
|
|
my $i = shift or die "no record to add"; |
63 |
dpavlin |
64 |
my $uid = shift; |
64 |
dpavlin |
47 |
|
65 |
|
|
die "record not Jifty::Record but ", ref $i unless ($i->isa('Jifty::Record')); |
66 |
|
|
|
67 |
|
|
my $pk = { $i->primary_keys }; |
68 |
|
|
|
69 |
|
|
my $doc = new Lucene::Document; |
70 |
|
|
|
71 |
|
|
my @columns = map { $_->name } $i->columns; |
72 |
|
|
|
73 |
|
|
foreach my $c ( @columns ) { |
74 |
|
|
|
75 |
|
|
my $v = $i->$c; |
76 |
|
|
|
77 |
|
|
if ( ref($v) ne '' ) { |
78 |
dpavlin |
53 |
|
79 |
|
|
foreach my $f_c ( qw/id name title/ ) { |
80 |
|
|
if ( $i->$c->can( $f_c ) ) { |
81 |
|
|
my $f_v = $i->$c->$f_c || $i->$c->{values}->{ $f_c }; |
82 |
|
|
my $col = $c . '_' . $f_c; |
83 |
|
|
if ( $f_v ) { |
84 |
|
|
warn " # $col = $f_v\n" if ($debug); |
85 |
|
|
$doc->add(Lucene::Document::Field->Text( $col, $f_v )); |
86 |
|
|
} else { |
87 |
|
|
warn " . $col is NULL\n" if ($debug); |
88 |
|
|
} |
89 |
|
|
} |
90 |
|
|
} |
91 |
|
|
|
92 |
|
|
if ($v->isa('Jifty::DateTime')) { |
93 |
dpavlin |
47 |
warn " d $c = $v\n" if ($debug); |
94 |
|
|
$doc->add(Lucene::Document::Field->Keyword( $c, "$v" )); |
95 |
|
|
} else { |
96 |
|
|
warn " s $c = $v [",ref($v),"]\n" if ($debug); |
97 |
|
|
} |
98 |
|
|
next; |
99 |
|
|
} |
100 |
|
|
|
101 |
|
|
next if (! defined($v) || $v eq ''); |
102 |
|
|
|
103 |
|
|
$v =~ s/<[^>]+>/ /gs; |
104 |
|
|
|
105 |
|
|
if ( defined( $pk->{$c} ) ) { |
106 |
|
|
$doc->add(Lucene::Document::Field->Keyword( $c, $v )); |
107 |
|
|
warn " * $c = $v\n" if ($debug); |
108 |
|
|
} else { |
109 |
|
|
$doc->add(Lucene::Document::Field->Text( $c, $v )); |
110 |
dpavlin |
53 |
warn " + $c = ", $self->snippet( 50, $v ), "\n" if ($debug); |
111 |
dpavlin |
47 |
} |
112 |
|
|
} |
113 |
|
|
|
114 |
dpavlin |
64 |
# add _owner_id to speed up filtering of search results |
115 |
|
|
$uid ||= Jifty->web->current_user->id; |
116 |
|
|
$doc->add(Lucene::Document::Field->Keyword( '_owner_id', $uid )); |
117 |
|
|
|
118 |
dpavlin |
109 |
if (! defined( $self->writer )) { |
119 |
|
|
$self->writer( new Lucene::Index::IndexWriter( $self->store, $self->analyzer, $self->create ) ); |
120 |
|
|
$self->log->debug($self->writer, " created"); |
121 |
|
|
} |
122 |
|
|
|
123 |
dpavlin |
49 |
$self->writer->addDocument($doc); |
124 |
dpavlin |
47 |
|
125 |
dpavlin |
98 |
$self->log->debug("added ", $i->id, " for user $uid to index"); |
126 |
dpavlin |
47 |
} |
127 |
|
|
|
128 |
dpavlin |
109 |
=head2 collection |
129 |
dpavlin |
47 |
|
130 |
dpavlin |
109 |
my $ItemCollection = $search->collection( 'search query' ); |
131 |
dpavlin |
47 |
|
132 |
|
|
=cut |
133 |
|
|
|
134 |
|
|
sub collection { |
135 |
|
|
my $self = shift; |
136 |
|
|
|
137 |
|
|
my $q = shift or die "no q?"; |
138 |
|
|
|
139 |
dpavlin |
58 |
return if ( $self->create ); |
140 |
|
|
|
141 |
dpavlin |
49 |
my $searcher = new Lucene::Search::IndexSearcher($self->store); |
142 |
dpavlin |
98 |
$self->log->debug("$searcher created"); |
143 |
dpavlin |
49 |
my $parser = new Lucene::QueryParser("content", $self->analyzer); |
144 |
dpavlin |
98 |
$self->log->debug("$parser created"); |
145 |
dpavlin |
47 |
|
146 |
dpavlin |
64 |
my $full_q = "($q) AND _owner_id:" . Jifty->web->current_user->id; |
147 |
|
|
|
148 |
|
|
my $query = $parser->parse( $full_q ); |
149 |
|
|
|
150 |
dpavlin |
98 |
$self->log->debug("searching for '$q' using ", $query->toString); |
151 |
dpavlin |
47 |
|
152 |
|
|
my $hits = $searcher->search($query); |
153 |
|
|
my $num_hits = $hits->length(); |
154 |
|
|
|
155 |
dpavlin |
98 |
$self->log->debug("found $num_hits results"); |
156 |
dpavlin |
47 |
|
157 |
|
|
my $collection = Grep::Model::ItemCollection->new(); |
158 |
|
|
|
159 |
|
|
my @results; |
160 |
|
|
|
161 |
|
|
for ( my $i = 0; $i < $num_hits; $i++ ) { |
162 |
|
|
|
163 |
|
|
my $doc = $hits->doc( $i ); |
164 |
|
|
|
165 |
|
|
my $score = $hits->score($i); |
166 |
|
|
my $title = $doc->get("title"); |
167 |
|
|
my $id = $doc->get("id"); |
168 |
|
|
|
169 |
dpavlin |
102 |
$self->log->debug("result $i $score $title"); |
170 |
dpavlin |
47 |
|
171 |
|
|
my $item = Grep::Model::Item->new(); |
172 |
|
|
my ($ok,$msg) = $item->load_by_cols( id => $id ); |
173 |
|
|
|
174 |
|
|
if ( $ok ) { |
175 |
|
|
$collection->add_record( $item ); |
176 |
|
|
} else { |
177 |
|
|
warn "can't load item $id\n"; |
178 |
|
|
} |
179 |
|
|
|
180 |
|
|
} |
181 |
|
|
|
182 |
|
|
undef $hits; |
183 |
|
|
undef $query; |
184 |
|
|
undef $parser; |
185 |
dpavlin |
76 |
$searcher->close; |
186 |
dpavlin |
47 |
undef $searcher; |
187 |
|
|
|
188 |
|
|
return $collection; |
189 |
|
|
} |
190 |
|
|
|
191 |
|
|
=head2 finish |
192 |
|
|
|
193 |
dpavlin |
109 |
$search->finish |
194 |
dpavlin |
47 |
|
195 |
|
|
=cut |
196 |
|
|
|
197 |
|
|
sub finish { |
198 |
|
|
my $self = shift; |
199 |
dpavlin |
109 |
if ($self->writer) { |
200 |
dpavlin |
102 |
$self->log->debug("closing index"); |
201 |
dpavlin |
109 |
$self->writer->close; |
202 |
dpavlin |
47 |
} |
203 |
dpavlin |
58 |
|
204 |
dpavlin |
110 |
$self->log->debug("finish"); |
205 |
|
|
|
206 |
dpavlin |
112 |
undef $self; |
207 |
|
|
|
208 |
dpavlin |
58 |
return; |
209 |
dpavlin |
47 |
} |
210 |
|
|
|
211 |
dpavlin |
57 |
=for TODO |
212 |
|
|
|
213 |
dpavlin |
47 |
sub _signal { |
214 |
|
|
my $s = shift; |
215 |
|
|
warn "catched SIG $s\n"; |
216 |
|
|
finish(); |
217 |
|
|
exit(0); |
218 |
|
|
} |
219 |
|
|
|
220 |
|
|
$SIG{'__DIE__'} = \&_signal; |
221 |
|
|
$SIG{'INT'} = \&_signal; |
222 |
|
|
$SIG{'QUIT'} = \&_signal; |
223 |
|
|
|
224 |
dpavlin |
57 |
=cut |
225 |
dpavlin |
53 |
|
226 |
|
|
=head2 snippet |
227 |
|
|
|
228 |
|
|
my $short = $self->snippet( 50, $text ); |
229 |
|
|
|
230 |
|
|
=cut |
231 |
|
|
|
232 |
|
|
sub snippet { |
233 |
|
|
my $self = shift; |
234 |
|
|
|
235 |
|
|
my $len = shift or die "no len?"; |
236 |
|
|
my $m = join(" ", @_); |
237 |
|
|
|
238 |
|
|
$m =~ s/\s+/ /gs; |
239 |
|
|
|
240 |
|
|
if (length($m) > $len) { |
241 |
|
|
return substr($m,0,$len) . '...'; |
242 |
|
|
} else { |
243 |
|
|
return $m; |
244 |
|
|
} |
245 |
|
|
} |
246 |
|
|
|
247 |
dpavlin |
47 |
1; |