/[Grep]/lib/Grep/Search/KinoSearch.pm
This is repository of my old source code which isn't updated any more. Go to git.rot13.org for current projects!
ViewVC logotype

Annotation of /lib/Grep/Search/KinoSearch.pm

Parent Directory Parent Directory | Revision Log Revision Log


Revision 153 - (hide annotations)
Sun Jun 10 11:52:24 2007 UTC (16 years, 11 months ago) by dpavlin
Original Path: lib/Grep/Search.pm
File size: 5935 byte(s)
re-enable syntax field:value AND/OR/NOT using set_heed_colons against latest
KinoSearch trunk, don't vectorize keyword fields
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     $v =~ s/<[^>]+>/ /gs;
107    
108     if ( defined( $pk->{$c} ) ) {
109 dpavlin 127 $doc->{ $c } = $v;
110 dpavlin 47 warn " * $c = $v\n" if ($debug);
111     } else {
112 dpavlin 127 $doc->{ $c } = $v;
113 dpavlin 53 warn " + $c = ", $self->snippet( 50, $v ), "\n" if ($debug);
114 dpavlin 47 }
115     }
116    
117 dpavlin 64 # add _owner_id to speed up filtering of search results
118     $uid ||= Jifty->web->current_user->id;
119 dpavlin 127 $doc->{ '_owner_id' } = $uid;
120 dpavlin 64
121 dpavlin 127 $self->invindexer->add_doc( $doc );
122 dpavlin 109
123 dpavlin 98 $self->log->debug("added ", $i->id, " for user $uid to index");
124 dpavlin 151
125     return 1;
126 dpavlin 47 }
127    
128 dpavlin 109 =head2 collection
129 dpavlin 47
130 dpavlin 141 Return C<Grep::Model::ItemCollection> which is result of C<search query>
131    
132 dpavlin 109 my $ItemCollection = $search->collection( 'search query' );
133 dpavlin 47
134 dpavlin 141 =head2 hits
135    
136     Return number of results from last C<collection> call
137    
138     my $num_results = $search->hits;
139    
140 dpavlin 47 =cut
141    
142     sub collection {
143     my $self = shift;
144    
145     my $q = shift or die "no q?";
146    
147 dpavlin 127 my $searcher = KinoSearch::Searcher->new(
148     invindex => Grep::Search::Schema->open( $self->index_path ), );
149 dpavlin 98 $self->log->debug("$searcher created");
150 dpavlin 47
151 dpavlin 151 my $full_q = "($q)";
152 dpavlin 64
153 dpavlin 151 my $uid = Jifty->web->current_user->id;
154 dpavlin 153 $full_q .= ' AND _owner_id:' . $uid if (defined $uid);
155 dpavlin 151
156 dpavlin 127 $self->log->debug("searching for '$q' using $full_q");
157 dpavlin 153
158     my $query_parser = KinoSearch::QueryParser->new(
159     schema => Grep::Search::Schema->new,
160     fields => [ qw/ title link content summary category author / ],
161     );
162     $query_parser->set_heed_colons(1); # enable field:value AND/OR/NOT syntax
163     my $query = $query_parser->parse( $full_q );
164 dpavlin 127 my $hits = $searcher->search(
165 dpavlin 153 query => $query,
166 dpavlin 127 # offset => $offset,
167     # num_wanted => $hits_per_page,
168     );
169 dpavlin 64
170 dpavlin 141 $self->hits( $hits->total_hits );
171 dpavlin 47
172 dpavlin 141 $self->log->debug("found ", $self->hits, " results");
173 dpavlin 47
174     my $collection = Grep::Model::ItemCollection->new();
175    
176     my @results;
177    
178 dpavlin 127 my $i = 0;
179     while ( my $hit = $hits->fetch_hit_hashref ) {
180 dpavlin 47
181 dpavlin 127 my $score = $hit->{score};
182     my $title = $hit->{title};
183     my $id = $hit->{id};
184 dpavlin 47
185 dpavlin 144 $self->log->debug("result $i [$id] $title $score");
186 dpavlin 47
187     my $item = Grep::Model::Item->new();
188     my ($ok,$msg) = $item->load_by_cols( id => $id );
189    
190     if ( $ok ) {
191     $collection->add_record( $item );
192     } else {
193     warn "can't load item $id\n";
194     }
195    
196     }
197    
198 dpavlin 127 $self->log->debug("finished search");
199 dpavlin 47
200     return $collection;
201     }
202    
203     =head2 finish
204    
205 dpavlin 109 $search->finish
206 dpavlin 47
207     =cut
208    
209     sub finish {
210     my $self = shift;
211 dpavlin 127 if ($self->invindexer) {
212 dpavlin 102 $self->log->debug("closing index");
213 dpavlin 127 $self->invindexer->finish;
214 dpavlin 47 }
215 dpavlin 58
216 dpavlin 110 $self->log->debug("finish");
217    
218 dpavlin 112 undef $self;
219    
220 dpavlin 151 return 1;
221 dpavlin 47 }
222    
223 dpavlin 53 =head2 snippet
224    
225     my $short = $self->snippet( 50, $text );
226    
227     =cut
228    
229     sub snippet {
230     my $self = shift;
231    
232     my $len = shift or die "no len?";
233     my $m = join(" ", @_);
234    
235     $m =~ s/\s+/ /gs;
236    
237     if (length($m) > $len) {
238     return substr($m,0,$len) . '...';
239     } else {
240     return $m;
241     }
242     }
243    
244 dpavlin 127 package Grep::Search::KeywordField;
245     use base qw( KinoSearch::Schema::FieldSpec );
246     sub analyzed { 0 }
247 dpavlin 153 #sub indexed { 1 }
248     #sub stored { 1 }
249     sub vectorized { 0 }
250 dpavlin 127
251     package Grep::Search::Schema;
252    
253     =head1 NAME
254    
255 dpavlin 136 Grep::Search::Schema - schema definition for full-text search
256 dpavlin 127
257     =cut
258    
259     use base 'KinoSearch::Schema';
260     use KinoSearch::Analysis::PolyAnalyzer;
261    
262 dpavlin 131 our %fields = (
263 dpavlin 127 id => 'Grep::Search::KeywordField',
264    
265     in_feed_id => 'Grep::Search::KeywordField',
266     in_feed_url => 'Grep::Search::KeywordField',
267     in_feed_title => 'KinoSearch::Schema::FieldSpec',
268     in_feed_owner => 'Grep::Search::KeywordField',
269     in_feed_created_on => 'Grep::Search::KeywordField',
270    
271     title => 'KinoSearch::Schema::FieldSpec',
272     link => 'Grep::Search::KeywordField',
273     content => 'KinoSearch::Schema::FieldSpec',
274     summary => 'KinoSearch::Schema::FieldSpec',
275     category => 'KinoSearch::Schema::FieldSpec',
276     author => 'KinoSearch::Schema::FieldSpec',
277     issued => 'Grep::Search::KeywordField',
278     modified => 'Grep::Search::KeywordField',
279    
280     _owner_id => 'Grep::Search::KeywordField',
281     );
282    
283     sub analyzer {
284     return KinoSearch::Analysis::PolyAnalyzer->new( language => 'en' );
285     }
286    
287 dpavlin 47 1;

  ViewVC Help
Powered by ViewVC 1.1.26