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
26 my $searcher = Koha::SearchEngine::ElasticSearch::Search->new();
27 my $builder = Koha::SearchEngine::Elasticsearch::QueryBuilder->new();
28 my $query = $builder->build_query('perl');
29 my $results = $searcher->search($query);
30 print "There were " . $results->total . " results.\n";
39 use base qw(Koha::ElasticSearch);
42 use Catmandu::Store::ElasticSearch;
44 use Data::Dumper; #TODO remove
47 Koha::SearchEngine::Elasticsearch::Search->mk_accessors(qw( store ));
51 my $results = $searcher->search($query, $page, $count, %options);
53 Run a search using the query. It'll return C<$count> results, starting at page
54 C<$page> (C<$page> counts from 1, anything less that, or C<undef> becomes 1.)
55 C<$count> is also the number of entries on a page.
57 C<%options> is a hash containing extra options:
63 If provided, this overrides the C<$page> value, and specifies the record as
64 an offset (i.e. the number of the record to start with), rather than a page.
73 my ($self, $query, $page, $count, %options) = @_;
75 my $params = $self->get_elasticsearch_params();
77 # 20 is the default number of results per page
78 $paging{limit} = $count || 20;
79 # ES/Catmandu doesn't want pages, it wants a record to start from.
80 if (exists $options{offset}) {
81 $paging{start} = $options{offset};
83 $page = (!defined($page) || ($page <= 0)) ? 0 : $page - 1;
84 $paging{start} = $page * $paging{limit};
87 Catmandu::Store::ElasticSearch->new(
92 my $results = $self->store->bag->search( %$query, %paging );
98 my ( $error, $results, $facets ) = $search->search_compat(
99 $query, $simple_query, \@sort_by, \@servers,
100 $results_per_page, $offset, $expanded_facet, $branches,
104 A search interface somewhat compatible with L<C4::Search->getRecords>. Anything
105 that is returned in the query created by build_query_compat will probably
112 $self, $query, $simple_query, $sort_by,
113 $servers, $results_per_page, $offset, $expanded_facet,
114 $branches, $query_type, $scan
118 $options{offset} = $offset;
119 my $results = $self->search($query, undef, $results_per_page, %options);
121 # Convert each result into a MARC::Record
122 my (@records, $index);
123 $index = $offset; # opac-search expects results to be put in the
124 # right place in the array, according to $offset
126 # The results come in an array for some reason
127 my $marc_json = @_[0]->{record};
128 my $marc = $self->json2marc($marc_json);
129 $records[$index++] = $marc;
131 # consumers of this expect a name-spaced result, we provide the default
134 $result{biblioserver}{hits} = $results->total;
135 $result{biblioserver}{RECORDS} = \@records;
136 return (undef, \%result, $self->_convert_facets($results->{facets}));
139 =head2 search_auth_compat
141 my ( $results, $total ) =
142 $searcher->search_auth_compat( $query, $page, $count, %options );
144 This has a similar calling convention to L<search>, however it returns its
145 results in a form the same as L<C4::AuthoritiesMarc::SearchAuthorities>.
149 sub search_auth_compat {
152 my $database = Koha::Database->new();
153 my $schema = $database->schema();
154 my $res = $self->search(@_);
160 my $marc_json = $record->{record};
162 # I wonder if these should be real values defined in the mapping
163 # rather than hard-coded conversions.
164 $result{authid} = $record->{ Local-Number };
166 # TODO put all this info into the record at index time so we
167 # don't have to go and sort it all out now.
168 my $authtypecode = $record->{authtype};
169 my $rs = $schema->resultset('AuthType')
170 ->search( { authtypecode => $authtypecode } );
172 # FIXME there's an assumption here that we will get a result.
173 # the original code also makes an assumption that some provided
174 # authtypecode may sometimes be used instead of the one stored
175 # with the record. It's not documented why this is the case, so
176 # it's not reproduced here.
177 my $authtype = $rs->single;
178 my $auth_tag_to_report = $authtype->auth_tag_to_report;
179 my $marc = $self->json2marc($marc_json);
180 my $mainentry = $marc->field($auth_tag_to_report);
183 foreach ( $mainentry->subfields() ) {
184 $reported_tag .= '$' . $_->[0] . $_->[1];
187 # Turn the resultset into a hash
189 foreach my $col (@{ $authtype->result_source->columns }) {
190 $authtype_cols{$col} = $authtype->get_column($col);
192 $result{authtype} = $authtype_cols;
193 $result{reported_tag} = $reported_tag;
195 # Reimplementing BuildSummary is out of scope because it'll be hard
197 C4::AuthoritiesMarc::BuildSummary( $marc, $result{authid},
199 push @records, $marc;
202 return ( \@records, $res->total );
207 my $marc = $self->json2marc($marc_json);
209 Converts the form of marc (based on its JSON, but as a Perl structure) that
210 Catmandu stores into a MARC::Record object.
215 my ( $self, $marcjson ) = @_;
217 my $marc = MARC::Record->new();
218 $marc->encoding('UTF-8');
221 # [ '245', '1', '2', 'a' => 'Title', 'b' => 'Subtitle' ]
222 # conveniently, this is the form that MARC::Field->new() likes
223 foreach $field (@$marcjson) {
224 next if @$field < 5; # Shouldn't be possible, but...
225 if ( $field->[0] eq 'LDR' ) {
226 $marc->leader( $field->[4] );
229 my $marc_field = MARC::Field->new(@$field);
230 $marc->append_fields($marc_field);
236 =head2 _convert_facets
238 my $koha_facets = _convert_facets($es_facets);
240 Converts elasticsearch facets types to the form that Koha expects.
241 It expects the ES facet name to match the Koha type, for example C<itype>,
242 C<au>, C<su-to>, etc.
246 sub _convert_facets {
247 my ( $self, $es ) = @_;
249 return undef if !$es;
251 # These should correspond to the ES field names, as opposed to the CCL
252 # things that zebra uses.
253 my %type_to_label = (
255 location => 'Location',
256 itype => 'ItemTypes',
259 'su-geo' => 'Places',
262 # We also have some special cases, e.g. itypes that need to show the
263 # value rather than the code.
264 my $itypes = Koha::ItemTypes->new();
265 my %special = ( itype => sub { $itypes->get_description_for_code(@_) }, );
267 while ( ( $type, $data ) = each %$es ) {
268 next if !exists( $type_to_label{$type} );
270 type_id => $type . '_id',
272 expandable => 1, # TODO figure how that's supposed to work
273 "type_label_$type_to_label{$type}" => 1,
274 type_link_value => $type,
276 foreach my $term ( @{ $data->{terms} } ) {
277 my $t = $term->{term};
278 my $c = $term->{count};
279 if ( exists( $special{$type} ) ) {
280 $label = $special{$type}->($t);
285 push @{ $facet->{facets} }, {
287 facet_link_value => $t,
288 facet_title_value => $t . " ($c)",
289 facet_label_value => $label, # TODO either truncate this,
290 # or make the template do it like it should anyway
291 type_link_value => $type,
294 push @res, $facet if exists $facet->{facets};