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 |
my $search = Grep::Search->new( create => 1 ); |
26 |
|
27 |
=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 |
=cut |
34 |
|
35 |
sub log { Jifty->web->log } |
36 |
|
37 |
sub new { |
38 |
my $class = shift; |
39 |
my $self = $class->SUPER::new(@_); |
40 |
|
41 |
my $index_path = Jifty::Util->app_root . '/var/invindex'; |
42 |
|
43 |
$self->index_path( $index_path ); |
44 |
|
45 |
if ( ! -e "$index_path" || $self->create ) { |
46 |
$self->log->debug("Creating new index $index_path"); |
47 |
$self->invindexer( KinoSearch::InvIndexer->new( invindex => Grep::Search::Schema->clobber( $index_path ) ) ); |
48 |
} else { |
49 |
$self->log->debug("Opening index: $index_path"); |
50 |
$self->invindexer( KinoSearch::InvIndexer->new( invindex => Grep::Search::Schema->open( $index_path ) ) ); |
51 |
} |
52 |
|
53 |
return $self; |
54 |
} |
55 |
|
56 |
=head2 add |
57 |
|
58 |
$search->add( $record, $owner_id ); |
59 |
|
60 |
=cut |
61 |
|
62 |
sub add { |
63 |
my $self = shift; |
64 |
|
65 |
my $i = shift or die "no record to add"; |
66 |
my $uid = shift; |
67 |
|
68 |
die "record not Jifty::Record but ", ref $i unless ($i->isa('Jifty::Record')); |
69 |
|
70 |
my $pk = { $i->primary_keys }; |
71 |
|
72 |
my $doc; |
73 |
|
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 |
|
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 |
$doc->{ $col } = $f_v; |
89 |
} else { |
90 |
warn " . $col is NULL\n" if ($debug); |
91 |
} |
92 |
} |
93 |
} |
94 |
|
95 |
if ($v->isa('Jifty::DateTime')) { |
96 |
warn " d $c = $v\n" if ($debug); |
97 |
$doc->{$c} = $v; |
98 |
} 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 |
eval { $v =~ s/<[^>]+>/ /gs; }; |
107 |
if ($@) { |
108 |
Jifty->log->error("can't strip html from $c in item ", $i->id); |
109 |
next; |
110 |
} |
111 |
|
112 |
if ( defined( $pk->{$c} ) ) { |
113 |
$doc->{ $c } = $v; |
114 |
warn " * $c = $v\n" if ($debug); |
115 |
} else { |
116 |
$doc->{ $c } = $v; |
117 |
warn " + $c = ", $self->snippet( 50, $v ), "\n" if ($debug); |
118 |
} |
119 |
} |
120 |
|
121 |
# add _owner_id to speed up filtering of search results |
122 |
$uid ||= Jifty->web->current_user->id; |
123 |
$doc->{ '_owner_id' } = $uid; |
124 |
|
125 |
$self->invindexer->add_doc( $doc ); |
126 |
|
127 |
$self->log->debug("added ", $i->id, " for user $uid to index"); |
128 |
|
129 |
return 1; |
130 |
} |
131 |
|
132 |
=head2 collection |
133 |
|
134 |
Return C<Grep::Model::ItemCollection> which is result of C<search query> |
135 |
|
136 |
my $ItemCollection = $search->collection( 'search query' ); |
137 |
|
138 |
=head2 hits |
139 |
|
140 |
Return number of results from last C<collection> call |
141 |
|
142 |
my $num_results = $search->hits; |
143 |
|
144 |
=cut |
145 |
|
146 |
sub collection { |
147 |
my $self = shift; |
148 |
|
149 |
my $q = shift or die "no q?"; |
150 |
|
151 |
my $searcher = KinoSearch::Searcher->new( |
152 |
invindex => Grep::Search::Schema->open( $self->index_path ), ); |
153 |
$self->log->debug("$searcher created"); |
154 |
|
155 |
my $full_q = "($q)"; |
156 |
|
157 |
my $uid = Jifty->web->current_user->id; |
158 |
$full_q .= ' AND _owner_id:' . $uid if (defined $uid); |
159 |
|
160 |
$self->log->debug("searching for '$q' using $full_q"); |
161 |
|
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 |
my $hits = $searcher->search( |
169 |
query => $query, |
170 |
# offset => $offset, |
171 |
# num_wanted => $hits_per_page, |
172 |
); |
173 |
|
174 |
$self->hits( $hits->total_hits ); |
175 |
|
176 |
$self->log->debug("found ", $self->hits, " results"); |
177 |
|
178 |
my $collection = Grep::Model::ItemCollection->new(); |
179 |
|
180 |
my @results; |
181 |
|
182 |
my $i = 0; |
183 |
while ( my $hit = $hits->fetch_hit ) { |
184 |
|
185 |
my $score = $hit->{score}; |
186 |
my $title = $hit->{title}; |
187 |
my $id = $hit->{id}; |
188 |
|
189 |
$self->log->debug("result $i [$id] $title $score"); |
190 |
|
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 |
$self->log->debug("finished search"); |
203 |
|
204 |
return $collection; |
205 |
} |
206 |
|
207 |
=head2 finish |
208 |
|
209 |
$search->finish |
210 |
|
211 |
=cut |
212 |
|
213 |
sub finish { |
214 |
my $self = shift; |
215 |
if ($self->invindexer) { |
216 |
$self->log->debug("closing index"); |
217 |
$self->invindexer->finish; |
218 |
} |
219 |
|
220 |
$self->log->debug("finish"); |
221 |
|
222 |
undef $self; |
223 |
|
224 |
return 1; |
225 |
} |
226 |
|
227 |
=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 |
package Grep::Search::KeywordField; |
249 |
use base qw( KinoSearch::Schema::FieldSpec ); |
250 |
sub analyzed { 0 } |
251 |
#sub indexed { 1 } |
252 |
#sub stored { 1 } |
253 |
sub vectorized { 0 } |
254 |
|
255 |
package Grep::Search::Schema; |
256 |
|
257 |
=head1 NAME |
258 |
|
259 |
Grep::Search::Schema - schema definition for full-text search |
260 |
|
261 |
=cut |
262 |
|
263 |
use base 'KinoSearch::Schema'; |
264 |
use KinoSearch::Analysis::PolyAnalyzer; |
265 |
|
266 |
our %fields = ( |
267 |
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 |
created_on => 'Grep::Search::KeywordField', |
282 |
last_update => 'Grep::Search::KeywordField', |
283 |
|
284 |
_owner_id => 'Grep::Search::KeywordField', |
285 |
); |
286 |
|
287 |
sub analyzer { |
288 |
return KinoSearch::Analysis::PolyAnalyzer->new( language => 'en' ); |
289 |
} |
290 |
|
291 |
1; |