lib/Grep/Search.pm

package Grep::Search;

use strict;
use warnings;

use Data::Dump qw/dump/;
use Lucene;
use Jifty::Util;

my $index_path = Jifty::Util->app_root . '/var/lucene';

my ( $analyzer, $store, $writer );

my $debug = 1;
my $create;

sub create {

        if (defined( $create )) {
                Jifty->log->debug("using previous create $create");
                return $create;
        }

        if (! -e "$index_path/segments") {
                $create = 1;
                Jifty->log->debug("create index $index_path");
        } else {
                $create = 0;
                Jifty->log->debug("open index: $index_path");
        }
        return $create;
}

sub analyzer {
        my $self = shift;
        $analyzer ||= new Lucene::Analysis::Standard::StandardAnalyzer();
        return $analyzer;
}

sub store {
        my $self = shift;

        $store ||= Lucene::Store::FSDirectory->getDirectory( $index_path, $self->create );
        return $store;
}

sub writer {
        my $self = shift;
        $writer ||= new Lucene::Index::IndexWriter( $self->store, $self->analyzer, $self->create );
        return $writer;
}

=head2 add

  Grep::Search->add( $record );

=cut

sub add {
        my $self = shift;

        my $i = shift or die "no record to add";

        die "record not Jifty::Record but ", ref $i unless ($i->isa('Jifty::Record'));

        my $pk = { $i->primary_keys };

        my $doc = new Lucene::Document;

        my @columns = map { $_->name } $i->columns;

        foreach my $c ( @columns ) {

                my $v = $i->$c;

                if ( ref($v) ne '' ) {

                        foreach my $f_c ( qw/id name title/ ) {
                                if ( $i->$c->can( $f_c ) ) {
                                        my $f_v = $i->$c->$f_c || $i->$c->{values}->{ $f_c };
                                        my $col = $c . '_' . $f_c;
                                        if ( $f_v ) {
                                                warn "  # $col = $f_v\n" if ($debug);
                                                $doc->add(Lucene::Document::Field->Text( $col, $f_v ));
                                        } else {
                                                warn "  . $col is NULL\n" if ($debug);
                                        }
                                }
                        }

                        if ($v->isa('Jifty::DateTime')) {
                                warn "  d $c = $v\n" if ($debug);
                                $doc->add(Lucene::Document::Field->Keyword( $c, "$v" ));
                        } else {
                                warn "  s $c = $v [",ref($v),"]\n" if ($debug);
                        }
                        next;
                }

                next if (! defined($v) || $v eq '');

                $v =~ s/<[^>]+>/ /gs;

                if ( defined( $pk->{$c} ) ) {
                        $doc->add(Lucene::Document::Field->Keyword( $c, $v ));
                        warn "  * $c = $v\n" if ($debug);
                } else {
                        $doc->add(Lucene::Document::Field->Text( $c, $v ));
                        warn "  + $c = ", $self->snippet( 50, $v ), "\n" if ($debug);
                }
        }

        $self->writer->addDocument($doc);

        Jifty->log->debug("added ", $i->id, " to index");
}

=head2

  my $ItemCollection = Grep::Search->collection( 'search query' );

=cut

sub collection {
        my $self = shift;

        my $q = shift or die "no q?";

        return if ( $self->create );

        my $searcher = new Lucene::Search::IndexSearcher($self->store);
        my $parser = new Lucene::QueryParser("content", $self->analyzer);
        my $query = $parser->parse( $q );

        Jifty->log->debug("searching for '$q'");

        my $hits = $searcher->search($query);
        my $num_hits = $hits->length();

        Jifty->log->debug("found $num_hits results");

        my $collection = Grep::Model::ItemCollection->new();

        my @results;

        for ( my $i = 0; $i < $num_hits; $i++ ) {

                my $doc = $hits->doc( $i );

                my $score = $hits->score($i);
                my $title = $doc->get("title");
                my $id = $doc->get("id");

                warn "## $i $score $title\n";

                my $item = Grep::Model::Item->new();
                my ($ok,$msg) = $item->load_by_cols( id => $id );

                if ( $ok ) {
                        $collection->add_record( $item );
                } else {
                        warn "can't load item $id\n";
                }

        }

        undef $hits;
        undef $query;
        undef $parser;
        undef $searcher;

        return $collection;
}

=head2 finish

  Grep::Search->finish

=cut

sub finish {
        my $self = shift;
        if ($writer) {
                warn "closing index\n";
                $writer->close;
        }
        undef $writer;
        undef $create;

        return;
}

=for TODO

sub _signal {
        my $s = shift;
        warn "catched SIG $s\n";
        finish();
        exit(0);
}

$SIG{'__DIE__'} = \&_signal;
$SIG{'INT'} = \&_signal;
$SIG{'QUIT'} = \&_signal;

=cut

=head2 snippet

  my $short = $self->snippet( 50, $text );


=cut

sub snippet {
        my $self = shift;

        my $len = shift or die "no len?";
        my $m = join(" ", @_);

        $m =~ s/\s+/ /gs;

        if (length($m) > $len) {
                return substr($m,0,$len) . '...';
        } else {
                return $m;
        }
}

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