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