| 1 |
package Grep::Search::KinoSearch; |
| 2 |
|
| 3 |
use strict; |
| 4 |
use warnings; |
| 5 |
|
| 6 |
use Data::Dump qw/dump/; |
| 7 |
use KinoSearch::InvIndexer; |
| 8 |
use KinoSearch::Searcher; |
| 9 |
use Jifty::Util; |
| 10 |
|
| 11 |
my $debug = 1; |
| 12 |
|
| 13 |
=head1 NAME |
| 14 |
|
| 15 |
Grep::Search::KinoSearch - full text search using L<KinoSearch> |
| 16 |
|
| 17 |
=head1 METHODS |
| 18 |
|
| 19 |
=head2 invindexer |
| 20 |
|
| 21 |
Accessor to call any method defined on L<KinoSearch::InvIndexer> |
| 22 |
|
| 23 |
$search->invindexer->delete_by_term( 'id', 42 ); |
| 24 |
|
| 25 |
=cut |
| 26 |
|
| 27 |
our $indexes; |
| 28 |
|
| 29 |
sub invindexer { |
| 30 |
my $self = shift; |
| 31 |
my $invindexer; |
| 32 |
my $index_path = Jifty::Util->app_root . '/var/invindex'; |
| 33 |
|
| 34 |
if ( $invindexer = $indexes->{$index_path} ) { |
| 35 |
$self->log->debug("Using cached index $index_path"); |
| 36 |
} else { |
| 37 |
if ( $self->create || -e $index_path ) { |
| 38 |
$self->log->debug("Creating new index $index_path"); |
| 39 |
$invindexer = KinoSearch::InvIndexer->new( invindex => Grep::Search::KinoSearch::Schema->clobber( $index_path ) ) |
| 40 |
or die "can't create index $index_path: $!"; |
| 41 |
$self->create( 0 ); |
| 42 |
} else { |
| 43 |
$self->log->debug("Opening index: $index_path"); |
| 44 |
$invindexer = KinoSearch::InvIndexer->new( invindex => Grep::Search::KinoSearch::Schema->open( $index_path ) ) |
| 45 |
or die "can't open index $index_path: $!"; |
| 46 |
} |
| 47 |
$indexes->{$index_path} = $invindexer; |
| 48 |
} |
| 49 |
|
| 50 |
return $invindexer; |
| 51 |
} |
| 52 |
|
| 53 |
=head2 add |
| 54 |
|
| 55 |
$search->add( $doc_hash ); |
| 56 |
|
| 57 |
=cut |
| 58 |
|
| 59 |
sub add { |
| 60 |
my $self = shift; |
| 61 |
my $doc = shift; |
| 62 |
$self->invindexer->add_doc( $doc ); |
| 63 |
return 1; |
| 64 |
} |
| 65 |
|
| 66 |
=head2 search |
| 67 |
|
| 68 |
my $fetch_hit_coderef = $self->search('search query'); |
| 69 |
|
| 70 |
=cut |
| 71 |
|
| 72 |
sub search { |
| 73 |
my $self = shift; |
| 74 |
|
| 75 |
my $q = shift or die "no q?"; |
| 76 |
|
| 77 |
my $index_path = Jifty::Util->app_root . '/var/invindex'; |
| 78 |
my $searcher = KinoSearch::Searcher->new( |
| 79 |
invindex => Grep::Search::KinoSearch::Schema->open( $index_path ), ); |
| 80 |
$self->log->debug("$searcher created"); |
| 81 |
|
| 82 |
my $full_q = "($q)"; |
| 83 |
|
| 84 |
my $uid = Jifty->web->current_user->id; |
| 85 |
$full_q .= ' AND _owner_id:' . $uid if (defined $uid); |
| 86 |
|
| 87 |
$self->log->debug("searching for '$q' using $full_q"); |
| 88 |
|
| 89 |
my $query_parser = KinoSearch::QueryParser->new( |
| 90 |
schema => Grep::Search::KinoSearch::Schema->new, |
| 91 |
fields => [ qw/ title link content summary category author / ], |
| 92 |
); |
| 93 |
$query_parser->set_heed_colons(1); # enable field:value AND/OR/NOT syntax |
| 94 |
my $query = $query_parser->parse( $full_q ); |
| 95 |
my $hits = $searcher->search( |
| 96 |
query => $query, |
| 97 |
# offset => $offset, |
| 98 |
# num_wanted => $hits_per_page, |
| 99 |
); |
| 100 |
|
| 101 |
$self->hits( $hits->total_hits ); |
| 102 |
|
| 103 |
return sub { |
| 104 |
return $hits->fetch_hit; |
| 105 |
}; |
| 106 |
} |
| 107 |
|
| 108 |
=head2 finish |
| 109 |
|
| 110 |
$search->finish |
| 111 |
|
| 112 |
=cut |
| 113 |
|
| 114 |
sub finish { |
| 115 |
my $self = shift; |
| 116 |
if ($self->invindexer) { |
| 117 |
$self->log->debug("closing index"); |
| 118 |
$self->invindexer->finish; |
| 119 |
} |
| 120 |
|
| 121 |
$self->log->debug("finish"); |
| 122 |
|
| 123 |
undef $self; |
| 124 |
|
| 125 |
return 1; |
| 126 |
} |
| 127 |
|
| 128 |
package Grep::Search::KinoSearch::KeywordField; |
| 129 |
use base qw( KinoSearch::Schema::FieldSpec ); |
| 130 |
sub analyzed { 0 } |
| 131 |
#sub indexed { 1 } |
| 132 |
#sub stored { 1 } |
| 133 |
sub vectorized { 0 } |
| 134 |
|
| 135 |
package Grep::Search::KinoSearch::Schema; |
| 136 |
|
| 137 |
=head1 NAME |
| 138 |
|
| 139 |
Grep::Search::KinoSearch::Schema - schema definition for full-text search |
| 140 |
|
| 141 |
=cut |
| 142 |
|
| 143 |
use base 'KinoSearch::Schema'; |
| 144 |
use KinoSearch::Analysis::PolyAnalyzer; |
| 145 |
|
| 146 |
our %fields = ( |
| 147 |
id => 'Grep::Search::KinoSearch::KeywordField', |
| 148 |
|
| 149 |
in_feed_id => 'Grep::Search::KinoSearch::KeywordField', |
| 150 |
in_feed_url => 'Grep::Search::KinoSearch::KeywordField', |
| 151 |
in_feed_title => 'KinoSearch::Schema::FieldSpec', |
| 152 |
in_feed_owner => 'Grep::Search::KinoSearch::KeywordField', |
| 153 |
in_feed_created_on => 'Grep::Search::KinoSearch::KeywordField', |
| 154 |
|
| 155 |
title => 'KinoSearch::Schema::FieldSpec', |
| 156 |
link => 'Grep::Search::KinoSearch::KeywordField', |
| 157 |
content => 'KinoSearch::Schema::FieldSpec', |
| 158 |
summary => 'KinoSearch::Schema::FieldSpec', |
| 159 |
category => 'KinoSearch::Schema::FieldSpec', |
| 160 |
author => 'KinoSearch::Schema::FieldSpec', |
| 161 |
created_on => 'Grep::Search::KinoSearch::KeywordField', |
| 162 |
last_update => 'Grep::Search::KinoSearch::KeywordField', |
| 163 |
|
| 164 |
_owner_id => 'Grep::Search::KinoSearch::KeywordField', |
| 165 |
); |
| 166 |
|
| 167 |
sub analyzer { |
| 168 |
return KinoSearch::Analysis::PolyAnalyzer->new( language => 'en' ); |
| 169 |
} |
| 170 |
|
| 171 |
1; |
Parent Directory
| 
Revision Log