| 1 |
package Grep::Search; |
| 2 |
|
| 3 |
use strict; |
| 4 |
use warnings; |
| 5 |
use base qw( Class::Accessor ); |
| 6 |
Grep::Search->mk_accessors( qw( create index_path hits ) ); |
| 7 |
|
| 8 |
use Data::Dump qw/dump/; |
| 9 |
use KinoSearch::InvIndexer; |
| 10 |
use KinoSearch::Searcher; |
| 11 |
use Jifty::Util; |
| 12 |
|
| 13 |
my $debug = 1; |
| 14 |
|
| 15 |
=head1 NAME |
| 16 |
|
| 17 |
Grep::Search - full text search using L<KinoSearch> |
| 18 |
|
| 19 |
=head1 METHODS |
| 20 |
|
| 21 |
=head2 new |
| 22 |
|
| 23 |
my $search = Grep::Search->new(); |
| 24 |
|
| 25 |
my $search = Grep::Search->new( create => 1 ); |
| 26 |
|
| 27 |
=cut |
| 28 |
|
| 29 |
sub log { Jifty->web->log } |
| 30 |
|
| 31 |
sub new { |
| 32 |
my $class = shift; |
| 33 |
my $self = $class->SUPER::new(@_); |
| 34 |
|
| 35 |
my $index_path = Jifty::Util->app_root . '/var/invindex'; |
| 36 |
|
| 37 |
$self->index_path( $index_path ); |
| 38 |
|
| 39 |
return $self; |
| 40 |
} |
| 41 |
|
| 42 |
=head2 invindexer |
| 43 |
|
| 44 |
Accessor to call any method defined on L<KinoSearch::InvIndexer> |
| 45 |
|
| 46 |
$search->invindexer->delete_by_term( 'id', 42 ); |
| 47 |
|
| 48 |
=cut |
| 49 |
|
| 50 |
my $indexes; |
| 51 |
|
| 52 |
sub invindexer { |
| 53 |
my $self = shift; |
| 54 |
my $invindexer; |
| 55 |
my $index_path = $self->index_path or die "no index_path?"; |
| 56 |
|
| 57 |
if ( $invindexer = $indexes->{$index_path} ) { |
| 58 |
$self->log->debug("Using cached index $index_path"); |
| 59 |
} else { |
| 60 |
if ( ! -e "$index_path" || $self->create ) { |
| 61 |
$self->log->debug("Creating new index $index_path"); |
| 62 |
my $invindexer = KinoSearch::InvIndexer->new( invindex => Grep::Search::Schema->clobber( $index_path ) ); |
| 63 |
} else { |
| 64 |
$self->log->debug("Opening index: $index_path"); |
| 65 |
$invindexer = KinoSearch::InvIndexer->new( invindex => Grep::Search::Schema->open( $index_path ) ); |
| 66 |
} |
| 67 |
$indexes->{$index_path} = $invindexer; |
| 68 |
} |
| 69 |
|
| 70 |
$self->invindexer( $invindexer ); |
| 71 |
|
| 72 |
} |
| 73 |
|
| 74 |
=head2 add |
| 75 |
|
| 76 |
$search->add( $record, $owner_id ); |
| 77 |
|
| 78 |
=cut |
| 79 |
|
| 80 |
sub add { |
| 81 |
my $self = shift; |
| 82 |
|
| 83 |
my $i = shift or die "no record to add"; |
| 84 |
my $uid = shift; |
| 85 |
|
| 86 |
die "record not Jifty::Record but ", ref $i unless ($i->isa('Jifty::Record')); |
| 87 |
|
| 88 |
my $pk = { $i->primary_keys }; |
| 89 |
|
| 90 |
my $doc; |
| 91 |
|
| 92 |
my @columns = map { $_->name } $i->columns; |
| 93 |
|
| 94 |
foreach my $c ( @columns ) { |
| 95 |
|
| 96 |
my $v = $i->$c; |
| 97 |
|
| 98 |
if ( ref($v) ne '' ) { |
| 99 |
|
| 100 |
foreach my $f_c ( qw/id name title/ ) { |
| 101 |
if ( $i->$c->can( $f_c ) ) { |
| 102 |
my $f_v = $i->$c->$f_c || $i->$c->{values}->{ $f_c }; |
| 103 |
my $col = $c . '_' . $f_c; |
| 104 |
if ( $f_v ) { |
| 105 |
warn " # $col = $f_v\n" if ($debug); |
| 106 |
$doc->{ $col } = $f_v; |
| 107 |
} else { |
| 108 |
warn " . $col is NULL\n" if ($debug); |
| 109 |
} |
| 110 |
} |
| 111 |
} |
| 112 |
|
| 113 |
if ($v->isa('Jifty::DateTime')) { |
| 114 |
warn " d $c = $v\n" if ($debug); |
| 115 |
$doc->{$c} = $v; |
| 116 |
} else { |
| 117 |
warn " s $c = $v [",ref($v),"]\n" if ($debug); |
| 118 |
} |
| 119 |
next; |
| 120 |
} |
| 121 |
|
| 122 |
next if (! defined($v) || $v eq ''); |
| 123 |
|
| 124 |
eval { $v =~ s/<[^>]+>/ /gs; }; |
| 125 |
if ($@) { |
| 126 |
Jifty->log->error("can't strip html from $c in item ", $i->id); |
| 127 |
next; |
| 128 |
} |
| 129 |
|
| 130 |
if ( defined( $pk->{$c} ) ) { |
| 131 |
$doc->{ $c } = $v; |
| 132 |
warn " * $c = $v\n" if ($debug); |
| 133 |
} else { |
| 134 |
$doc->{ $c } = $v; |
| 135 |
warn " + $c = ", $self->snippet( 50, $v ), "\n" if ($debug); |
| 136 |
} |
| 137 |
} |
| 138 |
|
| 139 |
# add _owner_id to speed up filtering of search results |
| 140 |
$uid ||= Jifty->web->current_user->id; |
| 141 |
$doc->{ '_owner_id' } = $uid; |
| 142 |
|
| 143 |
$self->invindexer->add_doc( $doc ); |
| 144 |
|
| 145 |
$self->log->debug("added ", $i->id, " for user $uid to index"); |
| 146 |
|
| 147 |
return 1; |
| 148 |
} |
| 149 |
|
| 150 |
=head2 collection |
| 151 |
|
| 152 |
Return C<Grep::Model::ItemCollection> which is result of C<search query> |
| 153 |
|
| 154 |
my $ItemCollection = $search->collection( 'search query' ); |
| 155 |
|
| 156 |
=head2 hits |
| 157 |
|
| 158 |
Return number of results from last C<collection> call |
| 159 |
|
| 160 |
my $num_results = $search->hits; |
| 161 |
|
| 162 |
=cut |
| 163 |
|
| 164 |
sub collection { |
| 165 |
my $self = shift; |
| 166 |
|
| 167 |
my $q = shift or die "no q?"; |
| 168 |
|
| 169 |
my $searcher = KinoSearch::Searcher->new( |
| 170 |
invindex => Grep::Search::Schema->open( $self->index_path ), ); |
| 171 |
$self->log->debug("$searcher created"); |
| 172 |
|
| 173 |
my $full_q = "($q)"; |
| 174 |
|
| 175 |
my $uid = Jifty->web->current_user->id; |
| 176 |
$full_q .= ' AND _owner_id:' . $uid if (defined $uid); |
| 177 |
|
| 178 |
$self->log->debug("searching for '$q' using $full_q"); |
| 179 |
|
| 180 |
my $query_parser = KinoSearch::QueryParser->new( |
| 181 |
schema => Grep::Search::Schema->new, |
| 182 |
fields => [ qw/ title link content summary category author / ], |
| 183 |
); |
| 184 |
$query_parser->set_heed_colons(1); # enable field:value AND/OR/NOT syntax |
| 185 |
my $query = $query_parser->parse( $full_q ); |
| 186 |
my $hits = $searcher->search( |
| 187 |
query => $query, |
| 188 |
# offset => $offset, |
| 189 |
# num_wanted => $hits_per_page, |
| 190 |
); |
| 191 |
|
| 192 |
$self->hits( $hits->total_hits ); |
| 193 |
|
| 194 |
$self->log->debug("found ", $self->hits, " results"); |
| 195 |
|
| 196 |
my $collection = Grep::Model::ItemCollection->new(); |
| 197 |
|
| 198 |
my @results; |
| 199 |
|
| 200 |
my $i = 0; |
| 201 |
while ( my $hit = $hits->fetch_hit ) { |
| 202 |
|
| 203 |
my $score = $hit->{score}; |
| 204 |
my $title = $hit->{title}; |
| 205 |
my $id = $hit->{id}; |
| 206 |
|
| 207 |
$self->log->debug("result $i [$id] $title $score"); |
| 208 |
|
| 209 |
my $item = Grep::Model::Item->new(); |
| 210 |
my ($ok,$msg) = $item->load_by_cols( id => $id ); |
| 211 |
|
| 212 |
if ( $ok ) { |
| 213 |
$collection->add_record( $item ); |
| 214 |
} else { |
| 215 |
warn "can't load item $id\n"; |
| 216 |
} |
| 217 |
|
| 218 |
} |
| 219 |
|
| 220 |
$self->log->debug("finished search"); |
| 221 |
|
| 222 |
return $collection; |
| 223 |
} |
| 224 |
|
| 225 |
=head2 finish |
| 226 |
|
| 227 |
$search->finish |
| 228 |
|
| 229 |
=cut |
| 230 |
|
| 231 |
sub finish { |
| 232 |
my $self = shift; |
| 233 |
if ($self->invindexer) { |
| 234 |
$self->log->debug("closing index"); |
| 235 |
$self->invindexer->finish; |
| 236 |
} |
| 237 |
|
| 238 |
$self->log->debug("finish"); |
| 239 |
|
| 240 |
undef $self; |
| 241 |
|
| 242 |
return 1; |
| 243 |
} |
| 244 |
|
| 245 |
=head2 snippet |
| 246 |
|
| 247 |
my $short = $self->snippet( 50, $text ); |
| 248 |
|
| 249 |
=cut |
| 250 |
|
| 251 |
sub snippet { |
| 252 |
my $self = shift; |
| 253 |
|
| 254 |
my $len = shift or die "no len?"; |
| 255 |
my $m = join(" ", @_); |
| 256 |
|
| 257 |
$m =~ s/\s+/ /gs; |
| 258 |
|
| 259 |
if (length($m) > $len) { |
| 260 |
return substr($m,0,$len) . '...'; |
| 261 |
} else { |
| 262 |
return $m; |
| 263 |
} |
| 264 |
} |
| 265 |
|
| 266 |
package Grep::Search::KeywordField; |
| 267 |
use base qw( KinoSearch::Schema::FieldSpec ); |
| 268 |
sub analyzed { 0 } |
| 269 |
#sub indexed { 1 } |
| 270 |
#sub stored { 1 } |
| 271 |
sub vectorized { 0 } |
| 272 |
|
| 273 |
package Grep::Search::Schema; |
| 274 |
|
| 275 |
=head1 NAME |
| 276 |
|
| 277 |
Grep::Search::Schema - schema definition for full-text search |
| 278 |
|
| 279 |
=cut |
| 280 |
|
| 281 |
use base 'KinoSearch::Schema'; |
| 282 |
use KinoSearch::Analysis::PolyAnalyzer; |
| 283 |
|
| 284 |
our %fields = ( |
| 285 |
id => 'Grep::Search::KeywordField', |
| 286 |
|
| 287 |
in_feed_id => 'Grep::Search::KeywordField', |
| 288 |
in_feed_url => 'Grep::Search::KeywordField', |
| 289 |
in_feed_title => 'KinoSearch::Schema::FieldSpec', |
| 290 |
in_feed_owner => 'Grep::Search::KeywordField', |
| 291 |
in_feed_created_on => 'Grep::Search::KeywordField', |
| 292 |
|
| 293 |
title => 'KinoSearch::Schema::FieldSpec', |
| 294 |
link => 'Grep::Search::KeywordField', |
| 295 |
content => 'KinoSearch::Schema::FieldSpec', |
| 296 |
summary => 'KinoSearch::Schema::FieldSpec', |
| 297 |
category => 'KinoSearch::Schema::FieldSpec', |
| 298 |
author => 'KinoSearch::Schema::FieldSpec', |
| 299 |
created_on => 'Grep::Search::KeywordField', |
| 300 |
last_update => 'Grep::Search::KeywordField', |
| 301 |
|
| 302 |
_owner_id => 'Grep::Search::KeywordField', |
| 303 |
); |
| 304 |
|
| 305 |
sub analyzer { |
| 306 |
return KinoSearch::Analysis::PolyAnalyzer->new( language => 'en' ); |
| 307 |
} |
| 308 |
|
| 309 |
1; |
Parent Directory
| 
Revision Log