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";
41 use base qw(Koha::ElasticSearch);
44 use Koha::AuthorisedValues;
45 use Koha::SearchEngine::QueryBuilder;
47 use Catmandu::Store::ElasticSearch;
49 use Data::Dumper; #TODO remove
52 Koha::SearchEngine::Elasticsearch::Search->mk_accessors(qw( store ));
56 my $results = $searcher->search($query, $page, $count, %options);
58 Run a search using the query. It'll return C<$count> results, starting at page
59 C<$page> (C<$page> counts from 1, anything less that, or C<undef> becomes 1.)
60 C<$count> is also the number of entries on a page.
62 C<%options> is a hash containing extra options:
68 If provided, this overrides the C<$page> value, and specifies the record as
69 an offset (i.e. the number of the record to start with), rather than a page.
78 my ($self, $query, $page, $count, %options) = @_;
80 my $params = $self->get_elasticsearch_params();
82 # 20 is the default number of results per page
83 $paging{limit} = $count || 20;
84 # ES/Catmandu doesn't want pages, it wants a record to start from.
85 if (exists $options{offset}) {
86 $paging{start} = $options{offset};
88 $page = (!defined($page) || ($page <= 0)) ? 0 : $page - 1;
89 $paging{start} = $page * $paging{limit};
92 Catmandu::Store::ElasticSearch->new(
95 ) unless $self->store;
98 $self->store->bag->search( %$query, %paging );
101 die $self->process_error($@);
108 my $count = $searcher->count($query);
110 This mimics a search request, but just gets the result count instead. That's
111 faster than pulling all the data in, ususally.
116 my ( $self, $query ) = @_;
118 my $params = $self->get_elasticsearch_params();
120 Catmandu::Store::ElasticSearch->new( %$params, trace_calls => 0, ) )
123 my $search = $self->store->bag->search( %$query);
124 my $count = $search->total() || 0;
130 my ( $error, $results, $facets ) = $search->search_compat(
131 $query, $simple_query, \@sort_by, \@servers,
132 $results_per_page, $offset, $expanded_facet, $branches,
136 A search interface somewhat compatible with L<C4::Search->getRecords>. Anything
137 that is returned in the query created by build_query_compat will probably
138 get ignored here, along with some other things (like C<@servers>.)
144 $self, $query, $simple_query, $sort_by,
145 $servers, $results_per_page, $offset, $expanded_facet,
146 $branches, $query_type, $scan
150 $options{offset} = $offset;
151 $options{expanded_facet} = $expanded_facet;
152 my $results = $self->search($query, undef, $results_per_page, %options);
154 # Convert each result into a MARC::Record
155 my (@records, $index);
156 $index = $offset; # opac-search expects results to be put in the
157 # right place in the array, according to $offset
159 # The results come in an array for some reason
160 my $marc_json = @_[0]->{record};
161 my $marc = $self->json2marc($marc_json);
162 $records[$index++] = $marc;
164 # consumers of this expect a name-spaced result, we provide the default
167 $result{biblioserver}{hits} = $results->total;
168 $result{biblioserver}{RECORDS} = \@records;
169 return (undef, \%result, $self->_convert_facets($results->{facets}, $expanded_facet));
172 =head2 search_auth_compat
174 my ( $results, $total ) =
175 $searcher->search_auth_compat( $query, $page, $count, %options );
177 This has a similar calling convention to L<search>, however it returns its
178 results in a form the same as L<C4::AuthoritiesMarc::SearchAuthorities>.
182 sub search_auth_compat {
186 my $database = Koha::Database->new();
187 my $schema = $database->schema();
188 my $res = $self->search(@_);
189 my $bib_searcher = Koha::SearchEngine::Elasticsearch::Search->new({index => 'biblios'});
195 my $marc_json = $record->{record};
197 # I wonder if these should be real values defined in the mapping
198 # rather than hard-coded conversions.
199 # Our results often come through as nested arrays, to fix this
200 # requires changes in catmandu.
201 my $authid = $record->{ 'Local-number' }[0][0];
202 $result{authid} = $authid;
204 # TODO put all this info into the record at index time so we
205 # don't have to go and sort it all out now.
206 my $authtypecode = $record->{authtype};
207 my $rs = $schema->resultset('AuthType')
208 ->search( { authtypecode => $authtypecode } );
210 # FIXME there's an assumption here that we will get a result.
211 # the original code also makes an assumption that some provided
212 # authtypecode may sometimes be used instead of the one stored
213 # with the record. It's not documented why this is the case, so
214 # it's not reproduced here yet.
215 my $authtype = $rs->single;
216 my $auth_tag_to_report = $authtype->auth_tag_to_report;
217 my $marc = $self->json2marc($marc_json);
218 my $mainentry = $marc->field($auth_tag_to_report);
221 foreach ( $mainentry->subfields() ) {
222 $reported_tag .= '$' . $_->[0] . $_->[1];
225 # Turn the resultset into a hash
227 foreach my $col ($authtype->result_source->columns) {
228 $authtype_cols{$col} = $authtype->get_column($col);
230 $result{authtype} = $authtype->authtypetext;
231 $result{reported_tag} = $reported_tag;
233 # Reimplementing BuildSummary is out of scope because it'll be hard
235 C4::AuthoritiesMarc::BuildSummary( $marc, $result{authid},
237 $result{used} = $self->count_auth_use($bib_searcher, $authid);
238 push @records, \%result;
241 return ( \@records, $res->total );
244 =head2 count_auth_use
246 my $count = $auth_searcher->count_auth_use($bib_searcher, $authid);
248 This runs a search to determine the number of records that reference the
249 specified authid. C<$bib_searcher> must be something compatible with
250 elasticsearch, as the query is built in this function.
255 my ($self, $bib_searcher, $authid) = @_;
260 query => { match_all => {} },
261 filter => { term => { an => $authid } }
265 $bib_searcher->count($query);
268 =head2 simple_search_compat
270 my ( $error, $marcresults, $total_hits ) =
271 $searcher->simple_search( $query, $offset, $max_results );
273 This is a simpler interface to the searching, intended to be similar enough to
274 L<C4::Search::SimpleSearch>.
282 A thing to search for. It could be a simple string, or something constructed
283 with the appropriate QueryBuilder module.
287 How many results to skip from the start of the results.
289 =item C<$max_results>
291 The max number of results to return. The default is 100 (because unlimited
292 is a pretty terrible thing to do.)
302 if something went wrong, this'll contain some kind of error
305 =item C<$marcresults>
307 an arrayref of MARC::Records (note that this is different from the
308 L<C4::Search> version which will return plain XML, but too bad.)
312 the total number of results that this search could have returned.
318 sub simple_search_compat {
319 my ($self, $query, $offset, $max_results) = @_;
321 return ('No query entered', undef, undef) unless $query;
324 $options{offset} = $offset // 0;
325 $max_results //= 100;
327 unless (ref $query) {
328 # We'll push it through the query builder to sanitise everything.
329 my $qb = Koha::SearchEngine::QueryBuilder->new({index => $self->index});
330 (undef,$query) = $qb->build_query_compat(undef, [$query]);
332 my $results = $self->search($query, undef, $max_results, %options);
335 # The results come in an array for some reason
336 my $marc_json = @_[0]->{record};
337 my $marc = $self->json2marc($marc_json);
338 push @records, $marc;
340 return (undef, \@records, $results->total);
345 my $marc = $self->json2marc($marc_json);
347 Converts the form of marc (based on its JSON, but as a Perl structure) that
348 Catmandu stores into a MARC::Record object.
353 my ( $self, $marcjson ) = @_;
355 my $marc = MARC::Record->new();
356 $marc->encoding('UTF-8');
359 # [ '245', '1', '2', 'a' => 'Title', 'b' => 'Subtitle' ]
360 # conveniently, this is the form that MARC::Field->new() likes
361 foreach $field (@$marcjson) {
362 next if @$field < 5; # Shouldn't be possible, but...
363 if ( $field->[0] eq 'LDR' ) {
364 $marc->leader( $field->[4] );
367 my $marc_field = MARC::Field->new(@$field);
368 $marc->append_fields($marc_field);
374 =head2 _convert_facets
376 my $koha_facets = _convert_facets($es_facets, $expanded_facet);
378 Converts elasticsearch facets types to the form that Koha expects.
379 It expects the ES facet name to match the Koha type, for example C<itype>,
380 C<au>, C<su-to>, etc.
382 C<$expanded_facet> is the facet that we want to show FacetMaxCount entries for, rather
383 than just 5 like normal.
387 sub _convert_facets {
388 my ( $self, $es, $exp_facet ) = @_;
390 return undef if !$es;
392 # These should correspond to the ES field names, as opposed to the CCL
393 # things that zebra uses.
394 # TODO let the library define the order using the interface.
395 my %type_to_label = (
396 author => { order => 1, label => 'Authors', },
397 itype => { order => 2, label => 'ItemTypes', },
398 location => { order => 3, label => 'Location', },
399 'su-geo' => { order => 4, label => 'Places', },
400 se => { order => 5, label => 'Series', },
401 subject => { order => 6, label => 'Topics', },
404 # We also have some special cases, e.g. itypes that need to show the
405 # value rather than the code.
406 my @itypes = Koha::ItemTypes->search;
407 my @locations = Koha::AuthorisedValues->search( { category => 'LOC' } );
408 my $opac = C4::Context->interface eq 'opac' ;
410 itype => { map { $_->itemtype => $_->description } @itypes },
411 location => { map { $_->authorised_value => ( $opac ? ( $_->lib_opac || $_->lib ) : $_->lib ) } @locations },
415 while ( ( $type, $data ) = each %$es ) {
416 next if !exists( $type_to_label{$type} );
418 # We restrict to the most popular $limit !results
419 my $limit = ( $type eq $exp_facet ) ? C4::Context->preference('FacetMaxCount') : 5;
421 type_id => $type . '_id',
423 expandable => ( $type ne $exp_facet )
424 && ( @{ $data->{terms} } > $limit ),
425 "type_label_$type_to_label{$type}{label}" => 1,
426 type_link_value => $type,
427 order => $type_to_label{$type}{order},
429 $limit = @{ $data->{terms} } if ( $limit > @{ $data->{terms} } );
430 foreach my $term ( @{ $data->{terms} }[ 0 .. $limit - 1 ] ) {
431 my $t = $term->{term};
432 my $c = $term->{count};
433 if ( exists( $special{$type} ) ) {
434 $label = $special{$type}->{$t} // $t;
439 push @{ $facet->{facets} }, {
441 facet_link_value => $t,
442 facet_title_value => $t . " ($c)",
443 facet_label_value => $label, # TODO either truncate this,
444 # or make the template do it like it should anyway
445 type_link_value => $type,
448 push @facets, $facet if exists $facet->{facets};
451 @facets = sort { $a->{order} cmp $b->{order} } @facets;