1 package Koha::SearchEngine::Elasticsearch::Search;
3 # Copyright 2014 Catalyst IT
5 # This file is part of Koha.
7 # Koha is free software; you can redistribute it and/or modify it under the
8 # terms of the GNU General Public License as published by the Free Software
9 # Foundation; either version 3 of the License, or (at your option) any later
12 # Koha is distributed in the hope that it will be useful, but WITHOUT ANY
13 # WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR
14 # A PARTICULAR PURPOSE. See the GNU General Public License for more details.
16 # You should have received a copy of the GNU General Public License along
17 # with Koha; if not, write to the Free Software Foundation, Inc.,
18 # 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.
22 Koha::SearchEngine::Elasticsearch::Search - search functions for Elasticsearch
27 Koha::SearchEngine::Elasticsearch::Search->new( { index => $index } );
28 my $builder = Koha::SearchEngine::Elasticsearch::QueryBuilder->new(
29 { index => $index } );
30 my $query = $builder->build_query('perl');
31 my $results = $searcher->search($query);
32 print "There were " . $results->total . " results.\n";
43 use base qw(Koha::SearchEngine::Elasticsearch);
46 use Koha::AuthorisedValues;
47 use Koha::SearchEngine::QueryBuilder;
49 use Catmandu::Store::ElasticSearch;
51 use Data::Dumper; #TODO remove
54 Koha::SearchEngine::Elasticsearch::Search->mk_accessors(qw( store ));
58 my $results = $searcher->search($query, $page, $count, %options);
60 Run a search using the query. It'll return C<$count> results, starting at page
61 C<$page> (C<$page> counts from 1, anything less that, or C<undef> becomes 1.)
62 C<$count> is also the number of entries on a page.
64 C<%options> is a hash containing extra options:
70 If provided, this overrides the C<$page> value, and specifies the record as
71 an offset (i.e. the number of the record to start with), rather than a page.
80 my ($self, $query, $page, $count, %options) = @_;
82 my $params = $self->get_elasticsearch_params();
84 # 20 is the default number of results per page
85 $paging{limit} = $count || 20;
86 # ES/Catmandu doesn't want pages, it wants a record to start from.
87 if (exists $options{offset}) {
88 $paging{start} = $options{offset};
90 $page = (!defined($page) || ($page <= 0)) ? 0 : $page - 1;
91 $paging{start} = $page * $paging{limit};
94 Catmandu::Store::ElasticSearch->new(
97 ) unless $self->store;
99 $self->store->bag->search( %$query, %paging );
102 die $self->process_error($@);
109 my $count = $searcher->count($query);
111 This mimics a search request, but just gets the result count instead. That's
112 faster than pulling all the data in, usually.
117 my ( $self, $query ) = @_;
119 my $params = $self->get_elasticsearch_params();
121 Catmandu::Store::ElasticSearch->new( %$params, trace_calls => 0, ) )
124 my $search = $self->store->bag->search( %$query);
125 my $count = $search->total() || 0;
131 my ( $error, $results, $facets ) = $search->search_compat(
132 $query, $simple_query, \@sort_by, \@servers,
133 $results_per_page, $offset, $expanded_facet, $branches,
137 A search interface somewhat compatible with L<C4::Search->getRecords>. Anything
138 that is returned in the query created by build_query_compat will probably
139 get ignored here, along with some other things (like C<@servers>.)
145 $self, $query, $simple_query, $sort_by,
146 $servers, $results_per_page, $offset, $expanded_facet,
147 $branches, $query_type, $scan
151 $options{offset} = $offset;
152 $options{expanded_facet} = $expanded_facet;
153 my $results = $self->search($query, undef, $results_per_page, %options);
155 # Convert each result into a MARC::Record
156 my (@records, $index);
157 $index = $offset; # opac-search expects results to be put in the
158 # right place in the array, according to $offset
160 # The results come in an array for some reason
161 my $marc_json = $_[0]->{record};
162 my $marc = $self->json2marc($marc_json);
163 $records[$index++] = $marc;
165 # consumers of this expect a name-spaced result, we provide the default
168 $result{biblioserver}{hits} = $results->total;
169 $result{biblioserver}{RECORDS} = \@records;
170 return (undef, \%result, $self->_convert_facets($results->{aggregations}, $expanded_facet));
173 =head2 search_auth_compat
175 my ( $results, $total ) =
176 $searcher->search_auth_compat( $query, $page, $count, %options );
178 This has a similar calling convention to L<search>, however it returns its
179 results in a form the same as L<C4::AuthoritiesMarc::SearchAuthorities>.
183 sub search_auth_compat {
187 my $database = Koha::Database->new();
188 my $schema = $database->schema();
189 my $res = $self->search(@_);
190 my $bib_searcher = Koha::SearchEngine::Elasticsearch::Search->new({index => 'biblios'});
196 my $marc_json = $record->{record};
198 # I wonder if these should be real values defined in the mapping
199 # rather than hard-coded conversions.
200 # Our results often come through as nested arrays, to fix this
201 # requires changes in catmandu.
202 my $authid = $record->{ 'Local-number' }[0][0];
203 $result{authid} = $authid;
205 # TODO put all this info into the record at index time so we
206 # don't have to go and sort it all out now.
207 my $authtypecode = $record->{authtype};
208 my $rs = $schema->resultset('AuthType')
209 ->search( { authtypecode => $authtypecode } );
211 # FIXME there's an assumption here that we will get a result.
212 # the original code also makes an assumption that some provided
213 # authtypecode may sometimes be used instead of the one stored
214 # with the record. It's not documented why this is the case, so
215 # it's not reproduced here yet.
216 my $authtype = $rs->single;
217 my $auth_tag_to_report = $authtype->auth_tag_to_report;
218 my $marc = $self->json2marc($marc_json);
219 my $mainentry = $marc->field($auth_tag_to_report);
222 foreach ( $mainentry->subfields() ) {
223 $reported_tag .= '$' . $_->[0] . $_->[1];
226 # Turn the resultset into a hash
228 foreach my $col ($authtype->result_source->columns) {
229 $authtype_cols{$col} = $authtype->get_column($col);
231 $result{authtype} = $authtype->authtypetext;
232 $result{reported_tag} = $reported_tag;
234 # Reimplementing BuildSummary is out of scope because it'll be hard
236 C4::AuthoritiesMarc::BuildSummary( $marc, $result{authid},
238 $result{used} = $self->count_auth_use($bib_searcher, $authid);
239 push @records, \%result;
242 return ( \@records, $res->total );
245 =head2 count_auth_use
247 my $count = $auth_searcher->count_auth_use($bib_searcher, $authid);
249 This runs a search to determine the number of records that reference the
250 specified authid. C<$bib_searcher> must be something compatible with
251 elasticsearch, as the query is built in this function.
256 my ($self, $bib_searcher, $authid) = @_;
261 # query => { match_all => {} },
262 filter => { term => { an => $authid } }
266 $bib_searcher->count($query);
269 =head2 simple_search_compat
271 my ( $error, $marcresults, $total_hits ) =
272 $searcher->simple_search( $query, $offset, $max_results );
274 This is a simpler interface to the searching, intended to be similar enough to
275 L<C4::Search::SimpleSearch>.
283 A thing to search for. It could be a simple string, or something constructed
284 with the appropriate QueryBuilder module.
288 How many results to skip from the start of the results.
290 =item C<$max_results>
292 The max number of results to return. The default is 100 (because unlimited
293 is a pretty terrible thing to do.)
303 if something went wrong, this'll contain some kind of error
306 =item C<$marcresults>
308 an arrayref of MARC::Records (note that this is different from the
309 L<C4::Search> version which will return plain XML, but too bad.)
313 the total number of results that this search could have returned.
319 sub simple_search_compat {
320 my ($self, $query, $offset, $max_results) = @_;
322 return ('No query entered', undef, undef) unless $query;
325 $options{offset} = $offset // 0;
326 $max_results //= 100;
328 unless (ref $query) {
329 # We'll push it through the query builder to sanitise everything.
330 my $qb = Koha::SearchEngine::QueryBuilder->new({index => $self->index});
331 (undef,$query) = $qb->build_query_compat(undef, [$query]);
333 my $results = $self->search($query, undef, $max_results, %options);
336 # The results come in an array for some reason
337 my $marc_json = $_[0]->{record};
338 my $marc = $self->json2marc($marc_json);
339 push @records, $marc;
341 return (undef, \@records, $results->total);
346 my $marc = $self->json2marc($marc_json);
348 Converts the form of marc (based on its JSON, but as a Perl structure) that
349 Catmandu stores into a MARC::Record object.
354 my ( $self, $marcjson ) = @_;
356 my $marc = MARC::Record->new();
357 $marc->encoding('UTF-8');
360 # [ '245', '1', '2', 'a' => 'Title', 'b' => 'Subtitle' ]
362 # [ '001', undef, undef, '_', 'a value' ]
363 # conveniently, this is the form that MARC::Field->new() likes
364 foreach my $field (@$marcjson) {
366 if ( $field->[0] eq 'LDR' ) {
367 $marc->leader( $field->[4] );
370 my $tag = $field->[0];
372 if ( MARC::Field->is_controlfield_tag( $field->[0] ) ) {
373 $marc_field = MARC::Field->new($field->[0], $field->[4]);
375 $marc_field = MARC::Field->new(@$field);
377 $marc->append_fields($marc_field);
383 =head2 _convert_facets
385 my $koha_facets = _convert_facets($es_facets, $expanded_facet);
387 Converts elasticsearch facets types to the form that Koha expects.
388 It expects the ES facet name to match the Koha type, for example C<itype>,
389 C<au>, C<su-to>, etc.
391 C<$expanded_facet> is the facet that we want to show FacetMaxCount entries for, rather
392 than just 5 like normal.
396 sub _convert_facets {
397 my ( $self, $es, $exp_facet ) = @_;
401 # These should correspond to the ES field names, as opposed to the CCL
402 # things that zebra uses.
403 # TODO let the library define the order using the interface.
404 my %type_to_label = (
405 author => { order => 1, label => 'Authors', },
406 itype => { order => 2, label => 'ItemTypes', },
407 location => { order => 3, label => 'Location', },
408 'su-geo' => { order => 4, label => 'Places', },
409 se => { order => 5, label => 'Series', },
410 subject => { order => 6, label => 'Topics', },
411 holdingbranch => { order => 8, label => 'HoldingLibrary' },
412 homebranch => { order => 9, label => 'HomeLibrary' }
415 # We also have some special cases, e.g. itypes that need to show the
416 # value rather than the code.
417 my @itypes = Koha::ItemTypes->search;
418 my @libraries = Koha::Libraries->search;
419 my $library_names = { map { $_->branchcode => $_->branchname } @libraries };
420 my @locations = Koha::AuthorisedValues->search( { category => 'LOC' } );
421 my $opac = C4::Context->interface eq 'opac' ;
423 itype => { map { $_->itemtype => $_->description } @itypes },
424 location => { map { $_->authorised_value => ( $opac ? ( $_->lib_opac || $_->lib ) : $_->lib ) } @locations },
425 holdingbranch => $library_names,
426 homebranch => $library_names
430 while ( my ( $type, $data ) = each %$es ) {
431 next if !exists( $type_to_label{$type} );
433 # We restrict to the most popular $limit !results
434 my $limit = ( $type eq $exp_facet ) ? C4::Context->preference('FacetMaxCount') : 5;
436 type_id => $type . '_id',
438 expandable => ( $type ne $exp_facet )
439 && ( @{ $data->{buckets} } > $limit ),
440 "type_label_$type_to_label{$type}{label}" => 1,
441 type_link_value => $type,
442 order => $type_to_label{$type}{order},
444 $limit = @{ $data->{buckets} } if ( $limit > @{ $data->{buckets} } );
445 foreach my $term ( @{ $data->{buckets} }[ 0 .. $limit - 1 ] ) {
446 my $t = $term->{key};
447 my $c = $term->{doc_count};
449 if ( exists( $special{$type} ) ) {
450 $label = $special{$type}->{$t} // $t;
455 push @{ $facet->{facets} }, {
457 facet_link_value => $t,
458 facet_title_value => $t . " ($c)",
459 facet_label_value => $label, # TODO either truncate this,
460 # or make the template do it like it should anyway
461 type_link_value => $type,
464 push @facets, $facet if exists $facet->{facets};
467 @facets = sort { $a->{order} cmp $b->{order} } @facets;