lib/Grep/Search.pm

package Grep::Search;

use strict;
use warnings;
use base qw( Class::Accessor );
Grep::Search->mk_accessors( qw( invindexer create index_path hits ) );

use Data::Dump qw/dump/;
use KinoSearch::InvIndexer;
use KinoSearch::Searcher;
use Jifty::Util;

my $debug = 0;

=head1 NAME

Grep::Search - full text search using L<KinoSearch>

=head1 METHODS

=head2 new

  my $search = Grep::Search->new();

=cut

sub log { Jifty->web->log }

sub new {
        my $class = shift;        
        my $self = $class->SUPER::new(@_);

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

        $self->index_path( $index_path );

        if (! -e "$index_path") {
                $self->log->debug("Creating new index $index_path");
                $self->invindexer( KinoSearch::InvIndexer->new( invindex => Grep::Search::Schema->clobber( $index_path ) ) );
        } else {
                $self->log->debug("Opening index: $index_path");
                $self->invindexer( KinoSearch::InvIndexer->new( invindex => Grep::Search::Schema->open( $index_path ) ) );
        }

        return $self;
}

=head2 add

  $search->add( $record, $owner_id );

=cut

sub add {
        my $self = shift;

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

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

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

        my $doc;

        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->{ $col } = $f_v;
                                        } else {
                                                warn "  . $col is NULL\n" if ($debug);
                                        }
                                }
                        }

                        if ($v->isa('Jifty::DateTime')) {
                                warn "  d $c = $v\n" if ($debug);
                                $doc->{$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->{ $c } = $v;
                        warn "  * $c = $v\n" if ($debug);
                } else {
                        $doc->{ $c } = $v;
                        warn "  + $c = ", $self->snippet( 50, $v ), "\n" if ($debug);
                }
        }

        # add _owner_id to speed up filtering of search results
        $uid ||= Jifty->web->current_user->id;
        $doc->{ '_owner_id' } = $uid;

        $self->invindexer->add_doc( $doc );

        $self->log->debug("added ", $i->id, " for user $uid to index");
}

=head2 collection

Return C<Grep::Model::ItemCollection> which is result of C<search query>

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

=head2 hits

Return number of results from last C<collection> call

  my $num_results = $search->hits;

=cut

sub collection {
        my $self = shift;

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

        my $searcher = KinoSearch::Searcher->new(
                invindex => Grep::Search::Schema->open( $self->index_path ), );
        $self->log->debug("$searcher created");

        my $full_q = "($q) AND _owner_id:" . Jifty->web->current_user->id;

        $self->log->debug("searching for '$q' using $full_q");
        my $hits = $searcher->search(
                query           => $full_q,
#               offset          => $offset,
#               num_wanted      => $hits_per_page,
        );

        $self->hits( $hits->total_hits );

        $self->log->debug("found ", $self->hits, " results");

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

        my @results;

        my $i = 0;
        while ( my $hit = $hits->fetch_hit_hashref ) {

                my $score = $hit->{score};
                my $title = $hit->{title};
                my $id = $hit->{id};

                $self->log->debug("result $i $score $title");

                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";
                }

        }

        $self->log->debug("finished search");

        return $collection;
}

=head2 finish

  $search->finish

=cut

sub finish {
        my $self = shift;
        if ($self->invindexer) {
                $self->log->debug("closing index");
                $self->invindexer->finish;
        }

        $self->log->debug("finish");

        undef $self;

        return;
}

=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;
        }
}

package Grep::Search::KeywordField;
use base qw( KinoSearch::Schema::FieldSpec );
sub analyzed { 0 }

package Grep::Search::Schema;

=head1 NAME

Grep::Search::Schema - schema definition for full-text search

=cut

use base 'KinoSearch::Schema';
use KinoSearch::Analysis::PolyAnalyzer;

our %fields = (
        id                              => 'Grep::Search::KeywordField',

        in_feed_id              => 'Grep::Search::KeywordField',
        in_feed_url             => 'Grep::Search::KeywordField',
        in_feed_title   => 'KinoSearch::Schema::FieldSpec',
        in_feed_owner   => 'Grep::Search::KeywordField',
        in_feed_created_on      => 'Grep::Search::KeywordField',

        title                   => 'KinoSearch::Schema::FieldSpec',
        link                    => 'Grep::Search::KeywordField',
        content                 => 'KinoSearch::Schema::FieldSpec',
        summary                 => 'KinoSearch::Schema::FieldSpec',
        category                => 'KinoSearch::Schema::FieldSpec',
        author                  => 'KinoSearch::Schema::FieldSpec',
        issued                  => 'Grep::Search::KeywordField',
        modified                => 'Grep::Search::KeywordField',

        _owner_id               => 'Grep::Search::KeywordField',
);

sub analyzer {
        return KinoSearch::Analysis::PolyAnalyzer->new( language => 'en' );
}

1;
1	dpavlin	47	package Grep::Search;
2
3			use strict;
4			use warnings;
5	dpavlin	112	use base qw( Class::Accessor );
6	dpavlin	141	Grep::Search->mk_accessors( qw( invindexer create index_path hits ) );
7	dpavlin	47
8			use Data::Dump qw/dump/;
9	dpavlin	127	use KinoSearch::InvIndexer;
10			use KinoSearch::Searcher;
11	dpavlin	47	use Jifty::Util;
12
13	dpavlin	109	my $debug = 0;
14	dpavlin	47
15	dpavlin	109	=head1 NAME
16	dpavlin	47
17	dpavlin	127	Grep::Search - full text search using L<KinoSearch>
18	dpavlin	47
19	dpavlin	109	=head1 METHODS
20	dpavlin	47
21	dpavlin	109	=head2 new
22	dpavlin	58
23	dpavlin	109	my $search = Grep::Search->new();
24
25			=cut
26
27	dpavlin	112	sub log { Jifty->web->log }
28
29	dpavlin	109	sub new {
30			my $class = shift;
31			my $self = $class->SUPER::new(@_);
32
33	dpavlin	127	my $index_path = Jifty::Util->app_root . '/var/invindex';
34	dpavlin	109
35			$self->index_path( $index_path );
36
37	dpavlin	127	if (! -e "$index_path") {
38	dpavlin	109	$self->log->debug("Creating new index $index_path");
39	dpavlin	127	$self->invindexer( KinoSearch::InvIndexer->new( invindex => Grep::Search::Schema->clobber( $index_path ) ) );
40	dpavlin	47	} else {
41	dpavlin	109	$self->log->debug("Opening index: $index_path");
42	dpavlin	127	$self->invindexer( KinoSearch::InvIndexer->new( invindex => Grep::Search::Schema->open( $index_path ) ) );
43	dpavlin	47	}
44
45	dpavlin	109	return $self;
46	dpavlin	49	}
47
48	dpavlin	47	=head2 add
49
50	dpavlin	109	$search->add( $record, $owner_id );
51	dpavlin	47
52			=cut
53
54			sub add {
55			my $self = shift;
56
57			my $i = shift or die "no record to add";
58	dpavlin	64	my $uid = shift;
59	dpavlin	47
60			die "record not Jifty::Record but ", ref $i unless ($i->isa('Jifty::Record'));
61
62			my $pk = { $i->primary_keys };
63
64	dpavlin	127	my $doc;
65	dpavlin	47
66			my @columns = map { $_->name } $i->columns;
67
68			foreach my $c ( @columns ) {
69
70			my $v = $i->$c;
71
72			if ( ref($v) ne '' ) {
73	dpavlin	53
74			foreach my $f_c ( qw/id name title/ ) {
75			if ( $i->$c->can( $f_c ) ) {
76			my $f_v = $i->$c->$f_c \|\| $i->$c->{values}->{ $f_c };
77			my $col = $c . '_' . $f_c;
78			if ( $f_v ) {
79			warn " # $col = $f_v\n" if ($debug);
80	dpavlin	127	$doc->{ $col } = $f_v;
81	dpavlin	53	} else {
82			warn " . $col is NULL\n" if ($debug);
83			}
84			}
85			}
86
87			if ($v->isa('Jifty::DateTime')) {
88	dpavlin	47	warn " d $c = $v\n" if ($debug);
89	dpavlin	127	$doc->{$c} = $v;
90	dpavlin	47	} else {
91			warn " s $c = $v [",ref($v),"]\n" if ($debug);
92			}
93			next;
94			}
95
96			next if (! defined($v) \|\| $v eq '');
97
98			$v =~ s/<[^>]+>/ /gs;
99
100			if ( defined( $pk->{$c} ) ) {
101	dpavlin	127	$doc->{ $c } = $v;
102	dpavlin	47	warn " * $c = $v\n" if ($debug);
103			} else {
104	dpavlin	127	$doc->{ $c } = $v;
105	dpavlin	53	warn " + $c = ", $self->snippet( 50, $v ), "\n" if ($debug);
106	dpavlin	47	}
107			}
108
109	dpavlin	64	# add _owner_id to speed up filtering of search results
110			$uid \|\|= Jifty->web->current_user->id;
111	dpavlin	127	$doc->{ '_owner_id' } = $uid;
112	dpavlin	64
113	dpavlin	127	$self->invindexer->add_doc( $doc );
114	dpavlin	109
115	dpavlin	98	$self->log->debug("added ", $i->id, " for user $uid to index");
116	dpavlin	47	}
117
118	dpavlin	109	=head2 collection
119	dpavlin	47
120	dpavlin	141	Return C<Grep::Model::ItemCollection> which is result of C<search query>
121
122	dpavlin	109	my $ItemCollection = $search->collection( 'search query' );
123	dpavlin	47
124	dpavlin	141	=head2 hits
125
126			Return number of results from last C<collection> call
127
128			my $num_results = $search->hits;
129
130	dpavlin	47	=cut
131
132			sub collection {
133			my $self = shift;
134
135			my $q = shift or die "no q?";
136
137	dpavlin	127	my $searcher = KinoSearch::Searcher->new(
138			invindex => Grep::Search::Schema->open( $self->index_path ), );
139	dpavlin	98	$self->log->debug("$searcher created");
140	dpavlin	47
141	dpavlin	64	my $full_q = "($q) AND _owner_id:" . Jifty->web->current_user->id;
142
143	dpavlin	127	$self->log->debug("searching for '$q' using $full_q");
144			my $hits = $searcher->search(
145			query => $full_q,
146			# offset => $offset,
147			# num_wanted => $hits_per_page,
148			);
149	dpavlin	64
150	dpavlin	141	$self->hits( $hits->total_hits );
151	dpavlin	47
152	dpavlin	141	$self->log->debug("found ", $self->hits, " results");
153	dpavlin	47
154			my $collection = Grep::Model::ItemCollection->new();
155
156			my @results;
157
158	dpavlin	127	my $i = 0;
159			while ( my $hit = $hits->fetch_hit_hashref ) {
160	dpavlin	47
161	dpavlin	127	my $score = $hit->{score};
162			my $title = $hit->{title};
163			my $id = $hit->{id};
164	dpavlin	47
165	dpavlin	102	$self->log->debug("result $i $score $title");
166	dpavlin	47
167			my $item = Grep::Model::Item->new();
168			my ($ok,$msg) = $item->load_by_cols( id => $id );
169
170			if ( $ok ) {
171			$collection->add_record( $item );
172			} else {
173			warn "can't load item $id\n";
174			}
175
176			}
177
178	dpavlin	127	$self->log->debug("finished search");
179	dpavlin	47
180			return $collection;
181			}
182
183			=head2 finish
184
185	dpavlin	109	$search->finish
186	dpavlin	47
187			=cut
188
189			sub finish {
190			my $self = shift;
191	dpavlin	127	if ($self->invindexer) {
192	dpavlin	102	$self->log->debug("closing index");
193	dpavlin	127	$self->invindexer->finish;
194	dpavlin	47	}
195	dpavlin	58
196	dpavlin	110	$self->log->debug("finish");
197
198	dpavlin	112	undef $self;
199
200	dpavlin	58	return;
201	dpavlin	47	}
202
203	dpavlin	53	=head2 snippet
204
205			my $short = $self->snippet( 50, $text );
206
207			=cut
208
209			sub snippet {
210			my $self = shift;
211
212			my $len = shift or die "no len?";
213			my $m = join(" ", @_);
214
215			$m =~ s/\s+/ /gs;
216
217			if (length($m) > $len) {
218			return substr($m,0,$len) . '...';
219			} else {
220			return $m;
221			}
222			}
223
224	dpavlin	127	package Grep::Search::KeywordField;
225			use base qw( KinoSearch::Schema::FieldSpec );
226			sub analyzed { 0 }
227
228			package Grep::Search::Schema;
229
230			=head1 NAME
231
232	dpavlin	136	Grep::Search::Schema - schema definition for full-text search
233	dpavlin	127
234			=cut
235
236			use base 'KinoSearch::Schema';
237			use KinoSearch::Analysis::PolyAnalyzer;
238
239	dpavlin	131	our %fields = (
240	dpavlin	127	id => 'Grep::Search::KeywordField',
241
242			in_feed_id => 'Grep::Search::KeywordField',
243			in_feed_url => 'Grep::Search::KeywordField',
244			in_feed_title => 'KinoSearch::Schema::FieldSpec',
245			in_feed_owner => 'Grep::Search::KeywordField',
246			in_feed_created_on => 'Grep::Search::KeywordField',
247
248			title => 'KinoSearch::Schema::FieldSpec',
249			link => 'Grep::Search::KeywordField',
250			content => 'KinoSearch::Schema::FieldSpec',
251			summary => 'KinoSearch::Schema::FieldSpec',
252			category => 'KinoSearch::Schema::FieldSpec',
253			author => 'KinoSearch::Schema::FieldSpec',
254			issued => 'Grep::Search::KeywordField',
255			modified => 'Grep::Search::KeywordField',
256
257			_owner_id => 'Grep::Search::KeywordField',
258			);
259
260			sub analyzer {
261			return KinoSearch::Analysis::PolyAnalyzer->new( language => 'en' );
262			}
263
264	dpavlin	47	1;