1 |
package Grep::Search; |
2 |
|
3 |
use strict; |
4 |
use warnings; |
5 |
|
6 |
use Data::Dump qw/dump/; |
7 |
use Lucene; |
8 |
use Jifty::Util; |
9 |
|
10 |
my $index_path = Jifty::Util->app_root . '/var/lucene'; |
11 |
|
12 |
my ( $analyzer, $store, $writer ); |
13 |
|
14 |
my $debug = 0; |
15 |
|
16 |
sub create { |
17 |
|
18 |
my $create = 0; |
19 |
if (! -e "$index_path/segments") { |
20 |
$create = 1; |
21 |
Jifty->log->debug("create index $index_path") unless ($store); |
22 |
} else { |
23 |
Jifty->log->debug("open index: $index_path") unless ($store); |
24 |
} |
25 |
return $create; |
26 |
} |
27 |
|
28 |
sub analyzer { |
29 |
my $self = shift; |
30 |
$analyzer ||= new Lucene::Analysis::Standard::StandardAnalyzer(); |
31 |
return $analyzer; |
32 |
} |
33 |
|
34 |
sub store { |
35 |
my $self = shift; |
36 |
|
37 |
$store ||= Lucene::Store::FSDirectory->getDirectory( $index_path, $self->create ); |
38 |
return $store; |
39 |
} |
40 |
|
41 |
sub writer { |
42 |
my $self = shift; |
43 |
$writer ||= new Lucene::Index::IndexWriter( $self->store, $self->analyzer, $self->create ); |
44 |
return $writer; |
45 |
} |
46 |
|
47 |
=head2 add |
48 |
|
49 |
Grep::Search->add( $record ); |
50 |
|
51 |
=cut |
52 |
|
53 |
sub add { |
54 |
my $self = shift; |
55 |
|
56 |
my $i = shift or die "no record to add"; |
57 |
|
58 |
die "record not Jifty::Record but ", ref $i unless ($i->isa('Jifty::Record')); |
59 |
|
60 |
my $pk = { $i->primary_keys }; |
61 |
|
62 |
my $doc = new Lucene::Document; |
63 |
|
64 |
my @columns = map { $_->name } $i->columns; |
65 |
|
66 |
foreach my $c ( @columns ) { |
67 |
|
68 |
my $v = $i->$c; |
69 |
|
70 |
if ( ref($v) ne '' ) { |
71 |
if ($i->$c->can('id')) { |
72 |
$v = $i->$c->id; |
73 |
warn " # $c = $v [id]\n" if ($debug); |
74 |
$doc->add(Lucene::Document::Field->Keyword( $c, $v )); |
75 |
} elsif ($v->isa('Jifty::DateTime')) { |
76 |
warn " d $c = $v\n" if ($debug); |
77 |
$doc->add(Lucene::Document::Field->Keyword( $c, "$v" )); |
78 |
} else { |
79 |
warn " s $c = $v [",ref($v),"]\n" if ($debug); |
80 |
} |
81 |
next; |
82 |
} |
83 |
|
84 |
next if (! defined($v) || $v eq ''); |
85 |
|
86 |
$v =~ s/<[^>]+>/ /gs; |
87 |
|
88 |
if ( defined( $pk->{$c} ) ) { |
89 |
$doc->add(Lucene::Document::Field->Keyword( $c, $v )); |
90 |
warn " * $c = $v\n" if ($debug); |
91 |
} else { |
92 |
$doc->add(Lucene::Document::Field->Text( $c, $v )); |
93 |
warn " + $c = $v\n" if ($debug); |
94 |
} |
95 |
} |
96 |
|
97 |
$self->writer->addDocument($doc); |
98 |
|
99 |
Jifty->log->debug("added ", $i->id, " to index"); |
100 |
} |
101 |
|
102 |
=head2 |
103 |
|
104 |
my $ItemCollection = Grep::Search->collection( 'search query' ); |
105 |
|
106 |
=cut |
107 |
|
108 |
sub collection { |
109 |
my $self = shift; |
110 |
|
111 |
my $q = shift or die "no q?"; |
112 |
|
113 |
my $searcher = new Lucene::Search::IndexSearcher($self->store); |
114 |
my $parser = new Lucene::QueryParser("content", $self->analyzer); |
115 |
my $query = $parser->parse( $q ); |
116 |
|
117 |
Jifty->log->debug("searching for '$q'"); |
118 |
|
119 |
my $hits = $searcher->search($query); |
120 |
my $num_hits = $hits->length(); |
121 |
|
122 |
Jifty->log->debug("found $num_hits results"); |
123 |
|
124 |
my $collection = Grep::Model::ItemCollection->new(); |
125 |
|
126 |
my @results; |
127 |
|
128 |
for ( my $i = 0; $i < $num_hits; $i++ ) { |
129 |
|
130 |
my $doc = $hits->doc( $i ); |
131 |
|
132 |
my $score = $hits->score($i); |
133 |
my $title = $doc->get("title"); |
134 |
my $id = $doc->get("id"); |
135 |
|
136 |
warn "## $i $score $title\n"; |
137 |
|
138 |
my $item = Grep::Model::Item->new(); |
139 |
my ($ok,$msg) = $item->load_by_cols( id => $id ); |
140 |
|
141 |
if ( $ok ) { |
142 |
$collection->add_record( $item ); |
143 |
} else { |
144 |
warn "can't load item $id\n"; |
145 |
} |
146 |
|
147 |
} |
148 |
|
149 |
undef $hits; |
150 |
undef $query; |
151 |
undef $parser; |
152 |
undef $searcher; |
153 |
|
154 |
return $collection; |
155 |
} |
156 |
|
157 |
=head2 finish |
158 |
|
159 |
Grep::Search->finish |
160 |
|
161 |
=cut |
162 |
|
163 |
sub finish { |
164 |
my $self = shift; |
165 |
if ($writer) { |
166 |
warn "closing index\n"; |
167 |
$writer->close; |
168 |
} |
169 |
undef $writer; |
170 |
} |
171 |
|
172 |
sub _signal { |
173 |
my $s = shift; |
174 |
warn "catched SIG $s\n"; |
175 |
finish(); |
176 |
exit(0); |
177 |
} |
178 |
|
179 |
$SIG{'__DIE__'} = \&_signal; |
180 |
$SIG{'INT'} = \&_signal; |
181 |
$SIG{'QUIT'} = \&_signal; |
182 |
|
183 |
1; |