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();

  my $search = Grep::Search->new( create => 1 );

=head2 invindexer

Accessor to call any method defined on L<KinoSearch::InvIndexer>

  $search->invindexer->delete_by_term( 'id', 42 );

=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->create ) {
                $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 '');

                eval { $v =~ s/<[^>]+>/ /gs; };
                if ($@) {
                        Jifty->log->error("can't strip html from $c in item ", $i->id);
                        next;
                }

                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");

        return 1;
}

=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)";

        my $uid = Jifty->web->current_user->id;
        $full_q .= ' AND _owner_id:' . $uid if (defined $uid);

        $self->log->debug("searching for '$q' using $full_q");

        my $query_parser = KinoSearch::QueryParser->new(
                schema => Grep::Search::Schema->new,
                fields => [ qw/ title link content summary category author / ],
        );
        $query_parser->set_heed_colons(1);       # enable field:value AND/OR/NOT syntax
        my $query = $query_parser->parse( $full_q );
        my $hits = $searcher->search(
                query           => $query,
#               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 ) {

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

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

                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 1;
}

=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 }
#sub indexed { 1 }
#sub stored { 1 }
sub vectorized { 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',
        created_on              => 'Grep::Search::KeywordField',
        last_update             => '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	dpavlin	145	my $search = Grep::Search->new( create => 1 );
26	dpavlin	144
27	dpavlin	145	=head2 invindexer
28
29			Accessor to call any method defined on L<KinoSearch::InvIndexer>
30
31			$search->invindexer->delete_by_term( 'id', 42 );
32
33	dpavlin	109	=cut
34
35	dpavlin	112	sub log { Jifty->web->log }
36
37	dpavlin	109	sub new {
38			my $class = shift;
39			my $self = $class->SUPER::new(@_);
40
41	dpavlin	127	my $index_path = Jifty::Util->app_root . '/var/invindex';
42	dpavlin	109
43			$self->index_path( $index_path );
44
45	dpavlin	144	if ( ! -e "$index_path" \|\| $self->create ) {
46	dpavlin	109	$self->log->debug("Creating new index $index_path");
47	dpavlin	127	$self->invindexer( KinoSearch::InvIndexer->new( invindex => Grep::Search::Schema->clobber( $index_path ) ) );
48	dpavlin	47	} else {
49	dpavlin	109	$self->log->debug("Opening index: $index_path");
50	dpavlin	127	$self->invindexer( KinoSearch::InvIndexer->new( invindex => Grep::Search::Schema->open( $index_path ) ) );
51	dpavlin	47	}
52
53	dpavlin	109	return $self;
54	dpavlin	49	}
55
56	dpavlin	47	=head2 add
57
58	dpavlin	109	$search->add( $record, $owner_id );
59	dpavlin	47
60			=cut
61
62			sub add {
63			my $self = shift;
64
65			my $i = shift or die "no record to add";
66	dpavlin	64	my $uid = shift;
67	dpavlin	47
68			die "record not Jifty::Record but ", ref $i unless ($i->isa('Jifty::Record'));
69
70			my $pk = { $i->primary_keys };
71
72	dpavlin	127	my $doc;
73	dpavlin	47
74			my @columns = map { $_->name } $i->columns;
75
76			foreach my $c ( @columns ) {
77
78			my $v = $i->$c;
79
80			if ( ref($v) ne '' ) {
81	dpavlin	53
82			foreach my $f_c ( qw/id name title/ ) {
83			if ( $i->$c->can( $f_c ) ) {
84			my $f_v = $i->$c->$f_c \|\| $i->$c->{values}->{ $f_c };
85			my $col = $c . '_' . $f_c;
86			if ( $f_v ) {
87			warn " # $col = $f_v\n" if ($debug);
88	dpavlin	127	$doc->{ $col } = $f_v;
89	dpavlin	53	} else {
90			warn " . $col is NULL\n" if ($debug);
91			}
92			}
93			}
94
95			if ($v->isa('Jifty::DateTime')) {
96	dpavlin	47	warn " d $c = $v\n" if ($debug);
97	dpavlin	127	$doc->{$c} = $v;
98	dpavlin	47	} else {
99			warn " s $c = $v [",ref($v),"]\n" if ($debug);
100			}
101			next;
102			}
103
104			next if (! defined($v) \|\| $v eq '');
105
106	dpavlin	160	eval { $v =~ s/<[^>]+>/ /gs; };
107			if ($@) {
108			Jifty->log->error("can't strip html from $c in item ", $i->id);
109			next;
110			}
111	dpavlin	47
112			if ( defined( $pk->{$c} ) ) {
113	dpavlin	127	$doc->{ $c } = $v;
114	dpavlin	47	warn " * $c = $v\n" if ($debug);
115			} else {
116	dpavlin	127	$doc->{ $c } = $v;
117	dpavlin	53	warn " + $c = ", $self->snippet( 50, $v ), "\n" if ($debug);
118	dpavlin	47	}
119			}
120
121	dpavlin	64	# add _owner_id to speed up filtering of search results
122			$uid \|\|= Jifty->web->current_user->id;
123	dpavlin	127	$doc->{ '_owner_id' } = $uid;
124	dpavlin	64
125	dpavlin	127	$self->invindexer->add_doc( $doc );
126	dpavlin	109
127	dpavlin	98	$self->log->debug("added ", $i->id, " for user $uid to index");
128	dpavlin	151
129			return 1;
130	dpavlin	47	}
131
132	dpavlin	109	=head2 collection
133	dpavlin	47
134	dpavlin	141	Return C<Grep::Model::ItemCollection> which is result of C<search query>
135
136	dpavlin	109	my $ItemCollection = $search->collection( 'search query' );
137	dpavlin	47
138	dpavlin	141	=head2 hits
139
140			Return number of results from last C<collection> call
141
142			my $num_results = $search->hits;
143
144	dpavlin	47	=cut
145
146			sub collection {
147			my $self = shift;
148
149			my $q = shift or die "no q?";
150
151	dpavlin	127	my $searcher = KinoSearch::Searcher->new(
152			invindex => Grep::Search::Schema->open( $self->index_path ), );
153	dpavlin	98	$self->log->debug("$searcher created");
154	dpavlin	47
155	dpavlin	151	my $full_q = "($q)";
156	dpavlin	64
157	dpavlin	151	my $uid = Jifty->web->current_user->id;
158	dpavlin	153	$full_q .= ' AND _owner_id:' . $uid if (defined $uid);
159	dpavlin	151
160	dpavlin	127	$self->log->debug("searching for '$q' using $full_q");
161	dpavlin	153
162			my $query_parser = KinoSearch::QueryParser->new(
163			schema => Grep::Search::Schema->new,
164			fields => [ qw/ title link content summary category author / ],
165			);
166			$query_parser->set_heed_colons(1); # enable field:value AND/OR/NOT syntax
167			my $query = $query_parser->parse( $full_q );
168	dpavlin	127	my $hits = $searcher->search(
169	dpavlin	153	query => $query,
170	dpavlin	127	# offset => $offset,
171			# num_wanted => $hits_per_page,
172			);
173	dpavlin	64
174	dpavlin	141	$self->hits( $hits->total_hits );
175	dpavlin	47
176	dpavlin	141	$self->log->debug("found ", $self->hits, " results");
177	dpavlin	47
178			my $collection = Grep::Model::ItemCollection->new();
179
180			my @results;
181
182	dpavlin	127	my $i = 0;
183	dpavlin	180	while ( my $hit = $hits->fetch_hit ) {
184	dpavlin	47
185	dpavlin	127	my $score = $hit->{score};
186			my $title = $hit->{title};
187			my $id = $hit->{id};
188	dpavlin	47
189	dpavlin	144	$self->log->debug("result $i [$id] $title $score");
190	dpavlin	47
191			my $item = Grep::Model::Item->new();
192			my ($ok,$msg) = $item->load_by_cols( id => $id );
193
194			if ( $ok ) {
195			$collection->add_record( $item );
196			} else {
197			warn "can't load item $id\n";
198			}
199
200			}
201
202	dpavlin	127	$self->log->debug("finished search");
203	dpavlin	47
204			return $collection;
205			}
206
207			=head2 finish
208
209	dpavlin	109	$search->finish
210	dpavlin	47
211			=cut
212
213			sub finish {
214			my $self = shift;
215	dpavlin	127	if ($self->invindexer) {
216	dpavlin	102	$self->log->debug("closing index");
217	dpavlin	127	$self->invindexer->finish;
218	dpavlin	47	}
219	dpavlin	58
220	dpavlin	110	$self->log->debug("finish");
221
222	dpavlin	112	undef $self;
223
224	dpavlin	151	return 1;
225	dpavlin	47	}
226
227	dpavlin	53	=head2 snippet
228
229			my $short = $self->snippet( 50, $text );
230
231			=cut
232
233			sub snippet {
234			my $self = shift;
235
236			my $len = shift or die "no len?";
237			my $m = join(" ", @_);
238
239			$m =~ s/\s+/ /gs;
240
241			if (length($m) > $len) {
242			return substr($m,0,$len) . '...';
243			} else {
244			return $m;
245			}
246			}
247
248	dpavlin	127	package Grep::Search::KeywordField;
249			use base qw( KinoSearch::Schema::FieldSpec );
250			sub analyzed { 0 }
251	dpavlin	153	#sub indexed { 1 }
252			#sub stored { 1 }
253			sub vectorized { 0 }
254	dpavlin	127
255			package Grep::Search::Schema;
256
257			=head1 NAME
258
259	dpavlin	136	Grep::Search::Schema - schema definition for full-text search
260	dpavlin	127
261			=cut
262
263			use base 'KinoSearch::Schema';
264			use KinoSearch::Analysis::PolyAnalyzer;
265
266	dpavlin	131	our %fields = (
267	dpavlin	127	id => 'Grep::Search::KeywordField',
268
269			in_feed_id => 'Grep::Search::KeywordField',
270			in_feed_url => 'Grep::Search::KeywordField',
271			in_feed_title => 'KinoSearch::Schema::FieldSpec',
272			in_feed_owner => 'Grep::Search::KeywordField',
273			in_feed_created_on => 'Grep::Search::KeywordField',
274
275			title => 'KinoSearch::Schema::FieldSpec',
276			link => 'Grep::Search::KeywordField',
277			content => 'KinoSearch::Schema::FieldSpec',
278			summary => 'KinoSearch::Schema::FieldSpec',
279			category => 'KinoSearch::Schema::FieldSpec',
280			author => 'KinoSearch::Schema::FieldSpec',
281	dpavlin	169	created_on => 'Grep::Search::KeywordField',
282			last_update => 'Grep::Search::KeywordField',
283	dpavlin	127
284			_owner_id => 'Grep::Search::KeywordField',
285			);
286
287			sub analyzer {
288			return KinoSearch::Analysis::PolyAnalyzer->new( language => 'en' );
289			}
290
291	dpavlin	47	1;