/[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

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

Parent Directory Parent Directory | Revision Log Revision Log


Revision 141 - (show annotations)
Tue May 8 12:37:28 2007 UTC (16 years, 9 months ago) by dpavlin
Original Path: lib/Grep/Search.pm
File size: 5318 byte(s)
added hits accessor to Grep::Search which returns original number of search results
1 package Grep::Search;
2
3 use strict;
4 use warnings;
5 use base qw( Class::Accessor );
6 Grep::Search->mk_accessors( qw( invindexer 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 = 0;
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 =cut
26
27 sub log { Jifty->web->log }
28
29 sub new {
30 my $class = shift;
31 my $self = $class->SUPER::new(@_);
32
33 my $index_path = Jifty::Util->app_root . '/var/invindex';
34
35 $self->index_path( $index_path );
36
37 if (! -e "$index_path") {
38 $self->log->debug("Creating new index $index_path");
39 $self->invindexer( KinoSearch::InvIndexer->new( invindex => Grep::Search::Schema->clobber( $index_path ) ) );
40 } else {
41 $self->log->debug("Opening index: $index_path");
42 $self->invindexer( KinoSearch::InvIndexer->new( invindex => Grep::Search::Schema->open( $index_path ) ) );
43 }
44
45 return $self;
46 }
47
48 =head2 add
49
50 $search->add( $record, $owner_id );
51
52 =cut
53
54 sub add {
55 my $self = shift;
56
57 my $i = shift or die "no record to add";
58 my $uid = shift;
59
60 die "record not Jifty::Record but ", ref $i unless ($i->isa('Jifty::Record'));
61
62 my $pk = { $i->primary_keys };
63
64 my $doc;
65
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
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 $doc->{ $col } = $f_v;
81 } else {
82 warn " . $col is NULL\n" if ($debug);
83 }
84 }
85 }
86
87 if ($v->isa('Jifty::DateTime')) {
88 warn " d $c = $v\n" if ($debug);
89 $doc->{$c} = $v;
90 } 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 $doc->{ $c } = $v;
102 warn " * $c = $v\n" if ($debug);
103 } else {
104 $doc->{ $c } = $v;
105 warn " + $c = ", $self->snippet( 50, $v ), "\n" if ($debug);
106 }
107 }
108
109 # add _owner_id to speed up filtering of search results
110 $uid ||= Jifty->web->current_user->id;
111 $doc->{ '_owner_id' } = $uid;
112
113 $self->invindexer->add_doc( $doc );
114
115 $self->log->debug("added ", $i->id, " for user $uid to index");
116 }
117
118 =head2 collection
119
120 Return C<Grep::Model::ItemCollection> which is result of C<search query>
121
122 my $ItemCollection = $search->collection( 'search query' );
123
124 =head2 hits
125
126 Return number of results from last C<collection> call
127
128 my $num_results = $search->hits;
129
130 =cut
131
132 sub collection {
133 my $self = shift;
134
135 my $q = shift or die "no q?";
136
137 my $searcher = KinoSearch::Searcher->new(
138 invindex => Grep::Search::Schema->open( $self->index_path ), );
139 $self->log->debug("$searcher created");
140
141 my $full_q = "($q) AND _owner_id:" . Jifty->web->current_user->id;
142
143 $self->log->debug("searching for '$q' using $full_q");
144 my $hits = $searcher->search(
145 query => $full_q,
146 # offset => $offset,
147 # num_wanted => $hits_per_page,
148 );
149
150 $self->hits( $hits->total_hits );
151
152 $self->log->debug("found ", $self->hits, " results");
153
154 my $collection = Grep::Model::ItemCollection->new();
155
156 my @results;
157
158 my $i = 0;
159 while ( my $hit = $hits->fetch_hit_hashref ) {
160
161 my $score = $hit->{score};
162 my $title = $hit->{title};
163 my $id = $hit->{id};
164
165 $self->log->debug("result $i $score $title");
166
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 $self->log->debug("finished search");
179
180 return $collection;
181 }
182
183 =head2 finish
184
185 $search->finish
186
187 =cut
188
189 sub finish {
190 my $self = shift;
191 if ($self->invindexer) {
192 $self->log->debug("closing index");
193 $self->invindexer->finish;
194 }
195
196 $self->log->debug("finish");
197
198 undef $self;
199
200 return;
201 }
202
203 =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 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 Grep::Search::Schema - schema definition for full-text search
233
234 =cut
235
236 use base 'KinoSearch::Schema';
237 use KinoSearch::Analysis::PolyAnalyzer;
238
239 our %fields = (
240 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 1;

  ViewVC Help
Powered by ViewVC 1.1.26