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);
43 use Koha::SearchEngine::QueryBuilder;
45 use Catmandu::Store::ElasticSearch;
47 use Data::Dumper; #TODO remove
50 Koha::SearchEngine::Elasticsearch::Search->mk_accessors(qw( store ));
54 my $results = $searcher->search($query, $page, $count, %options);
56 Run a search using the query. It'll return C<$count> results, starting at page
57 C<$page> (C<$page> counts from 1, anything less that, or C<undef> becomes 1.)
58 C<$count> is also the number of entries on a page.
60 C<%options> is a hash containing extra options:
66 If provided, this overrides the C<$page> value, and specifies the record as
67 an offset (i.e. the number of the record to start with), rather than a page.
76 my ($self, $query, $page, $count, %options) = @_;
78 my $params = $self->get_elasticsearch_params();
80 # 20 is the default number of results per page
81 $paging{limit} = $count || 20;
82 # ES/Catmandu doesn't want pages, it wants a record to start from.
83 if (exists $options{offset}) {
84 $paging{start} = $options{offset};
86 $page = (!defined($page) || ($page <= 0)) ? 0 : $page - 1;
87 $paging{start} = $page * $paging{limit};
90 Catmandu::Store::ElasticSearch->new(
93 ) unless $self->store;
96 $self->store->bag->search( %$query, %paging );
99 die $self->process_error($@);
106 my $count = $searcher->count($query);
108 This mimics a search request, but just gets the result count instead. That's
109 faster than pulling all the data in, ususally.
114 my ( $self, $query ) = @_;
116 my $params = $self->get_elasticsearch_params();
118 Catmandu::Store::ElasticSearch->new( %$params, trace_calls => 0, ) )
121 my $searcher = $self->store->bag->searcher(query => $query);
122 my $count = $searcher->count();
128 my ( $error, $results, $facets ) = $search->search_compat(
129 $query, $simple_query, \@sort_by, \@servers,
130 $results_per_page, $offset, $expanded_facet, $branches,
134 A search interface somewhat compatible with L<C4::Search->getRecords>. Anything
135 that is returned in the query created by build_query_compat will probably
136 get ignored here, along with some other things (like C<@servers>.)
142 $self, $query, $simple_query, $sort_by,
143 $servers, $results_per_page, $offset, $expanded_facet,
144 $branches, $query_type, $scan
148 $options{offset} = $offset;
149 my $results = $self->search($query, undef, $results_per_page, %options);
151 # Convert each result into a MARC::Record
152 my (@records, $index);
153 $index = $offset; # opac-search expects results to be put in the
154 # right place in the array, according to $offset
156 # The results come in an array for some reason
157 my $marc_json = @_[0]->{record};
158 my $marc = $self->json2marc($marc_json);
159 $records[$index++] = $marc;
161 # consumers of this expect a name-spaced result, we provide the default
164 $result{biblioserver}{hits} = $results->total;
165 $result{biblioserver}{RECORDS} = \@records;
166 return (undef, \%result, $self->_convert_facets($results->{facets}, $expanded_facet));
169 =head2 search_auth_compat
171 my ( $results, $total ) =
172 $searcher->search_auth_compat( $query, $page, $count, %options );
174 This has a similar calling convention to L<search>, however it returns its
175 results in a form the same as L<C4::AuthoritiesMarc::SearchAuthorities>.
179 sub search_auth_compat {
183 my $database = Koha::Database->new();
184 my $schema = $database->schema();
185 my $res = $self->search(@_);
186 my $bib_searcher = Koha::SearchEngine::Elasticsearch::Search->new({index => 'biblios'});
192 my $marc_json = $record->{record};
194 # I wonder if these should be real values defined in the mapping
195 # rather than hard-coded conversions.
196 # Our results often come through as nested arrays, to fix this
197 # requires changes in catmandu.
198 my $authid = $record->{ 'Local-number' }[0][0];
199 $result{authid} = $authid;
201 # TODO put all this info into the record at index time so we
202 # don't have to go and sort it all out now.
203 my $authtypecode = $record->{authtype};
204 my $rs = $schema->resultset('AuthType')
205 ->search( { authtypecode => $authtypecode } );
207 # FIXME there's an assumption here that we will get a result.
208 # the original code also makes an assumption that some provided
209 # authtypecode may sometimes be used instead of the one stored
210 # with the record. It's not documented why this is the case, so
211 # it's not reproduced here yet.
212 my $authtype = $rs->single;
213 my $auth_tag_to_report = $authtype->auth_tag_to_report;
214 my $marc = $self->json2marc($marc_json);
215 my $mainentry = $marc->field($auth_tag_to_report);
218 foreach ( $mainentry->subfields() ) {
219 $reported_tag .= '$' . $_->[0] . $_->[1];
222 # Turn the resultset into a hash
224 foreach my $col ($authtype->result_source->columns) {
225 $authtype_cols{$col} = $authtype->get_column($col);
227 $result{authtype} = $authtype->authtypetext;
228 $result{reported_tag} = $reported_tag;
230 # Reimplementing BuildSummary is out of scope because it'll be hard
232 C4::AuthoritiesMarc::BuildSummary( $marc, $result{authid},
234 $result{used} = $self->count_auth_use($bib_searcher, $authid);
235 push @records, \%result;
238 return ( \@records, $res->total );
241 =head2 count_auth_use
243 my $count = $auth_searcher->count_auth_use($bib_searcher, $authid);
245 This runs a search to determine the number of records that reference the
246 specified authid. C<$bib_searcher> must be something compatible with
247 elasticsearch, as the query is built in this function.
252 my ($self, $bib_searcher, $authid) = @_;
257 query => { match_all => {} },
258 filter => { term => { an => $authid } }
262 $bib_searcher->count($query);
265 =head2 simple_search_compat
267 my ( $error, $marcresults, $total_hits ) =
268 $searcher->simple_search( $query, $offset, $max_results );
270 This is a simpler interface to the searching, intended to be similar enough to
271 L<C4::Search::SimpleSearch>.
279 A thing to search for. It could be a simple string, or something constructed
280 with the appropriate QueryBuilder module.
284 How many results to skip from the start of the results.
286 =item C<$max_results>
288 The max number of results to return. The default is 100 (because unlimited
289 is a pretty terrible thing to do.)
299 if something went wrong, this'll contain some kind of error
302 =item C<$marcresults>
304 an arrayref of MARC::Records (note that this is different from the
305 L<C4::Search> version which will return plain XML, but too bad.)
309 the total number of results that this search could have returned.
315 sub simple_search_compat {
316 my ($self, $query, $offset, $max_results) = @_;
318 return ('No query entered', undef, undef) unless $query;
321 $options{offset} = $offset // 0;
322 $max_results //= 100;
324 unless (ref $query) {
325 # We'll push it through the query builder to sanitise everything.
326 my $qb = Koha::SearchEngine::QueryBuilder->new({index => $self->index});
327 (undef,$query) = $qb->build_query_compat(undef, [$query]);
329 my $results = $self->search($query, undef, $max_results, %options);
332 # The results come in an array for some reason
333 my $marc_json = @_[0]->{record};
334 my $marc = $self->json2marc($marc_json);
335 push @records, $marc;
337 return (undef, \@records, $results->total);
342 my $marc = $self->json2marc($marc_json);
344 Converts the form of marc (based on its JSON, but as a Perl structure) that
345 Catmandu stores into a MARC::Record object.
350 my ( $self, $marcjson ) = @_;
352 my $marc = MARC::Record->new();
353 $marc->encoding('UTF-8');
356 # [ '245', '1', '2', 'a' => 'Title', 'b' => 'Subtitle' ]
357 # conveniently, this is the form that MARC::Field->new() likes
358 foreach $field (@$marcjson) {
359 next if @$field < 5; # Shouldn't be possible, but...
360 if ( $field->[0] eq 'LDR' ) {
361 $marc->leader( $field->[4] );
364 my $marc_field = MARC::Field->new(@$field);
365 $marc->append_fields($marc_field);
371 =head2 _convert_facets
373 my $koha_facets = _convert_facets($es_facets, $expanded_facet);
375 Converts elasticsearch facets types to the form that Koha expects.
376 It expects the ES facet name to match the Koha type, for example C<itype>,
377 C<au>, C<su-to>, etc.
379 C<$expanded_facet> is the facet that we want to show 10 entries for, rather
380 than just 5 like normal.
384 sub _convert_facets {
385 my ( $self, $es, $exp_facet ) = @_;
387 return undef if !$es;
389 # These should correspond to the ES field names, as opposed to the CCL
390 # things that zebra uses.
391 my %type_to_label = (
393 location => 'Location',
394 itype => 'ItemTypes',
397 'su-geo' => 'Places',
400 # We also have some special cases, e.g. itypes that need to show the
401 # value rather than the code.
402 my @itypes = Koha::ItemTypes->search;
404 itype => { map { $_->itemtype => $_->description } @itypes },
408 while ( ( $type, $data ) = each %$es ) {
409 next if !exists( $type_to_label{$type} );
411 # We restrict to the most popular $limit results
412 my $limit = ( $type eq $exp_facet ) ? 10 : 5;
414 type_id => $type . '_id',
416 expandable => ( $type ne $exp_facet )
417 && ( @{ $data->{terms} } > $limit ),
418 "type_label_$type_to_label{$type}" => 1,
419 type_link_value => $type,
421 $limit = @{ $data->{terms} } if ( $limit > @{ $data->{terms} } );
422 foreach my $term ( @{ $data->{terms} }[ 0 .. $limit - 1 ] ) {
423 my $t = $term->{term};
424 my $c = $term->{count};
425 if ( exists( $special{$type} ) ) {
426 $label = $special{$type}->{$t} // $t;
431 push @{ $facet->{facets} }, {
433 facet_link_value => $t,
434 facet_title_value => $t . " ($c)",
435 facet_label_value => $label, # TODO either truncate this,
436 # or make the template do it like it should anyway
437 type_link_value => $type,
440 push @res, $facet if exists $facet->{facets};