t/Koha/SearchEngine/Elasticsearch.t

   1 #!/usr/bin/perl
   2 #
   3 # This file is part of Koha.
   4 #
   5 # Koha is free software; you can redistribute it and/or modify it
   6 # under the terms of the GNU General Public License as published by
   7 # the Free Software Foundation; either version 3 of the License, or
   8 # (at your option) any later version.
   9 #
  10 # Koha is distributed in the hope that it will be useful, but
  11 # WITHOUT ANY WARRANTY; without even the implied warranty of
  12 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  13 # GNU General Public License for more details.
  14 #
  15 # You should have received a copy of the GNU General Public License
  16 # along with Koha; if not, see <http://www.gnu.org/licenses>.
  17
  18 use Modern::Perl;
  19
  20 use Test::More tests => 4;
  21 use Test::Exception;
  22
  23 use t::lib::Mocks;
  24
  25 use Test::MockModule;
  26
  27 use MARC::Record;
  28 use Try::Tiny;
  29
  30 use Koha::SearchEngine::Elasticsearch;
  31 use Koha::SearchEngine::Elasticsearch::Search;
  32
  33 subtest '_read_configuration() tests' => sub {
  34
  35     plan tests => 10;
  36
  37     my $configuration;
  38     t::lib::Mocks::mock_config( 'elasticsearch', undef );
  39
  40     # 'elasticsearch' missing in configuration
  41     throws_ok {
  42         $configuration = Koha::SearchEngine::Elasticsearch::_read_configuration;
  43     }
  44     'Koha::Exceptions::Config::MissingEntry',
  45       'Configuration problem, exception thrown';
  46     is(
  47         $@->message,
  48         "Missing 'elasticsearch' block in config file",
  49         'Exception message is correct'
  50     );
  51
  52     # 'elasticsearch' present but no 'server' entry
  53     t::lib::Mocks::mock_config( 'elasticsearch', {} );
  54     throws_ok {
  55         $configuration = Koha::SearchEngine::Elasticsearch::_read_configuration;
  56     }
  57     'Koha::Exceptions::Config::MissingEntry',
  58       'Configuration problem, exception thrown';
  59     is(
  60         $@->message,
  61         "Missing 'server' entry in config file for elasticsearch",
  62         'Exception message is correct'
  63     );
  64
  65     # 'elasticsearch' and 'server' entries present, but no 'index_name'
  66     t::lib::Mocks::mock_config( 'elasticsearch', { server => 'a_server' } );
  67     throws_ok {
  68         $configuration = Koha::SearchEngine::Elasticsearch::_read_configuration;
  69     }
  70     'Koha::Exceptions::Config::MissingEntry',
  71       'Configuration problem, exception thrown';
  72     is(
  73         $@->message,
  74         "Missing 'index_name' entry in config file for elasticsearch",
  75         'Exception message is correct'
  76     );
  77
  78     # Correct configuration, only one server
  79     t::lib::Mocks::mock_config( 'elasticsearch',  { server => 'a_server', index_name => 'index' } );
  80
  81     $configuration = Koha::SearchEngine::Elasticsearch::_read_configuration;
  82     is( $configuration->{index_name}, 'index', 'Index configuration parsed correctly' );
  83     is_deeply( $configuration->{nodes}, ['a_server'], 'Server configuration parsed correctly' );
  84
  85     # Correct configuration, two servers
  86     my @servers = ('a_server', 'another_server');
  87     t::lib::Mocks::mock_config( 'elasticsearch', { server => \@servers, index_name => 'index' } );
  88
  89     $configuration = Koha::SearchEngine::Elasticsearch::_read_configuration;
  90     is( $configuration->{index_name}, 'index', 'Index configuration parsed correctly' );
  91     is_deeply( $configuration->{nodes}, \@servers , 'Server configuration parsed correctly' );
  92 };
  93
  94 subtest 'get_elasticsearch_settings() tests' => sub {
  95
  96     plan tests => 1;
  97
  98     my $settings;
  99
 100     # test reading index settings
 101     my $es = Koha::SearchEngine::Elasticsearch->new( {index => $Koha::SearchEngine::Elasticsearch::BIBLIOS_INDEX} );
 102     $settings = $es->get_elasticsearch_settings();
 103     is( $settings->{index}{analysis}{analyzer}{analyser_phrase}{tokenizer}, 'keyword', 'Index settings parsed correctly' );
 104 };
 105
 106 subtest 'get_elasticsearch_mappings() tests' => sub {
 107
 108     plan tests => 1;
 109
 110     my $mappings;
 111
 112     # test reading mappings
 113     my $es = Koha::SearchEngine::Elasticsearch->new( {index => $Koha::SearchEngine::Elasticsearch::BIBLIOS_INDEX} );
 114     $mappings = $es->get_elasticsearch_mappings();
 115     is( $mappings->{data}{_all}{type}, 'string', 'Field mappings parsed correctly' );
 116 };
 117
 118 subtest 'Koha::SearchEngine::Elasticsearch::marc_records_to_documents () tests' => sub {
 119
 120     plan tests => 32;
 121
 122     t::lib::Mocks::mock_preference('marcflavour', 'MARC21');
 123
 124     my @mappings = (
 125         {
 126             name => 'control_number',
 127             type => 'string',
 128             facet => 0,
 129             suggestible => 0,
 130             sort => undef,
 131             marc_type => 'marc21',
 132             marc_field => '001',
 133         },
 134         {
 135             name => 'isbn',
 136             type => 'isbn',
 137             facet => 0,
 138             suggestible => 0,
 139             sort => 0,
 140             marc_type => 'marc21',
 141             marc_field => '020a',
 142         },
 143         {
 144             name => 'author',
 145             type => 'string',
 146             facet => 1,
 147             suggestible => 1,
 148             sort => undef,
 149             marc_type => 'marc21',
 150             marc_field => '100a',
 151         },
 152         {
 153             name => 'author',
 154             type => 'string',
 155             facet => 1,
 156             suggestible => 1,
 157             sort => 1,
 158             marc_type => 'marc21',
 159             marc_field => '110a',
 160         },
 161         {
 162             name => 'title',
 163             type => 'string',
 164             facet => 0,
 165             suggestible => 1,
 166             sort => 1,
 167             marc_type => 'marc21',
 168             marc_field => '245(ab)ab',
 169         },
 170         {
 171             name => 'unimarc_title',
 172             type => 'string',
 173             facet => 0,
 174             suggestible => 1,
 175             sort => 1,
 176             marc_type => 'unimarc',
 177             marc_field => '245a',
 178         },
 179         {
 180             name => 'title',
 181             type => 'string',
 182             facet => 0,
 183             suggestible => undef,
 184             sort => 0,
 185             marc_type => 'marc21',
 186             marc_field => '220',
 187         },
 188         {
 189             name => 'title_wildcard',
 190             type => 'string',
 191             facet => 0,
 192             suggestible => 0,
 193             sort => undef,
 194             marc_type => 'marc21',
 195             marc_field => '245',
 196         },
 197         {
 198             name => 'sum_item_price',
 199             type => 'sum',
 200             facet => 0,
 201             suggestible => 0,
 202             sort => 0,
 203             marc_type => 'marc21',
 204             marc_field => '952g',
 205         },
 206         {
 207             name => 'items_withdrawn_status',
 208             type => 'boolean',
 209             facet => 0,
 210             suggestible => 0,
 211             sort => 0,
 212             marc_type => 'marc21',
 213             marc_field => '9520',
 214         },
 215         {
 216             name => 'type_of_record',
 217             type => 'string',
 218             facet => 0,
 219             suggestible => 0,
 220             sort => 0,
 221             marc_type => 'marc21',
 222             marc_field => 'leader_/6',
 223         },
 224         {
 225             name => 'type_of_record_and_bib_level',
 226             type => 'string',
 227             facet => 0,
 228             suggestible => 0,
 229             sort => 0,
 230             marc_type => 'marc21',
 231             marc_field => 'leader_/6-7',
 232         },
 233     );
 234
 235     my $se = Test::MockModule->new('Koha::SearchEngine::Elasticsearch');
 236     $se->mock('_foreach_mapping', sub {
 237         my ($self, $sub) = @_;
 238
 239         foreach my $map (@mappings) {
 240             $sub->(
 241                 $map->{name},
 242                 $map->{type},
 243                 $map->{facet},
 244                 $map->{suggestible},
 245                 $map->{sort},
 246                 $map->{marc_type},
 247                 $map->{marc_field}
 248             );
 249         }
 250     });
 251
 252     my $see = Koha::SearchEngine::Elasticsearch::Search->new({ index => $Koha::SearchEngine::Elasticsearch::BIBLIOS_INDEX });
 253
 254     my $marc_record_1 = MARC::Record->new();
 255     $marc_record_1->leader('     cam  22      a 4500');
 256     $marc_record_1->append_fields(
 257         MARC::Field->new('001', '123'),
 258         MARC::Field->new('020', '', '', a => '1-56619-909-3'),
 259         MARC::Field->new('100', '', '', a => 'Author 1'),
 260         MARC::Field->new('110', '', '', a => 'Corp Author'),
 261         MARC::Field->new('210', '', '', a => 'Title 1'),
 262         MARC::Field->new('245', '', '', a => 'Title:', b => 'first record'),
 263         MARC::Field->new('999', '', '', c => '1234567'),
 264         # '  ' for testing trimming of white space in boolean value callback:
 265         MARC::Field->new('952', '', '', 0 => '  ', g => '123.30'),
 266         MARC::Field->new('952', '', '', 0 => 0, g => '127.20'),
 267     );
 268     my $marc_record_2 = MARC::Record->new();
 269     $marc_record_2->leader('     cam  22      a 4500');
 270     $marc_record_2->append_fields(
 271         MARC::Field->new('100', '', '', a => 'Author 2'),
 272         # MARC::Field->new('210', '', '', a => 'Title 2'),
 273         # MARC::Field->new('245', '', '', a => 'Title: second record'),
 274         MARC::Field->new('999', '', '', c => '1234568'),
 275         MARC::Field->new('952', '', '', 0 => 1, g => 'string where should be numeric'),
 276     );
 277     my $records = [$marc_record_1, $marc_record_2];
 278
 279     $see->get_elasticsearch_mappings(); #sort_fields will call this and use the actual db values unless we call it first
 280
 281     my $docs = $see->marc_records_to_documents($records);
 282
 283     # First record:
 284     is(scalar @{$docs}, 2, 'Two records converted to documents');
 285
 286     is($docs->[0][0], '1234567', 'First document biblionumber should be set as first element in document touple');
 287
 288     is_deeply($docs->[0][1]->{control_number}, ['123'], 'First record control number should be set correctly');
 289
 290     is(scalar @{$docs->[0][1]->{author}}, 2, 'First document author field should contain two values');
 291     is_deeply($docs->[0][1]->{author}, ['Author 1', 'Corp Author'], 'First document author field should be set correctly');
 292
 293     is(scalar @{$docs->[0][1]->{author__sort}}, 2, 'First document author__sort field should have two values');
 294     is_deeply($docs->[0][1]->{author__sort}, ['Author 1', 'Corp Author'], 'First document author__sort field should be set correctly');
 295
 296     is(scalar @{$docs->[0][1]->{title__sort}}, 3, 'First document title__sort field should have three values');
 297     is_deeply($docs->[0][1]->{title__sort}, ['Title:', 'first record', 'Title: first record'], 'First document title__sort field should be set correctly');
 298
 299     is(scalar @{$docs->[0][1]->{title_wildcard}}, 2, 'First document title_wildcard field should have two values');
 300     is_deeply($docs->[0][1]->{title_wildcard}, ['Title:', 'first record'], 'First document title_wildcard field should be set correctly');
 301
 302     is(scalar @{$docs->[0][1]->{author__suggestion}}, 2, 'First document author__suggestion field should contain two values');
 303     is_deeply(
 304         $docs->[0][1]->{author__suggestion},
 305         [
 306             {
 307                 'input' => 'Author 1'
 308             },
 309             {
 310                 'input' => 'Corp Author'
 311             }
 312         ],
 313         'First document author__suggestion field should be set correctly'
 314     );
 315
 316     is(scalar @{$docs->[0][1]->{title__suggestion}}, 3, 'First document title__suggestion field should contain three values');
 317     is_deeply(
 318         $docs->[0][1]->{title__suggestion},
 319         [
 320             { 'input' => 'Title:' },
 321             { 'input' => 'first record' },
 322             { 'input' => 'Title: first record' }
 323         ],
 324         'First document title__suggestion field should be set correctly'
 325     );
 326
 327     ok(!(defined $docs->[0][1]->{title__facet}), 'First document should have no title__facet field');
 328
 329     is(scalar @{$docs->[0][1]->{author__facet}}, 2, 'First document author__facet field should have two values');
 330     is_deeply(
 331         $docs->[0][1]->{author__facet},
 332         ['Author 1', 'Corp Author'],
 333         'First document author__facet field should be set correctly'
 334     );
 335
 336     is(scalar @{$docs->[0][1]->{items_withdrawn_status}}, 2, 'First document items_withdrawn_status field should have two values');
 337     is_deeply(
 338         $docs->[0][1]->{items_withdrawn_status},
 339         ['false', 'false'],
 340         'First document items_withdrawn_status field should be set correctly'
 341     );
 342
 343     is(
 344         $docs->[0][1]->{sum_item_price},
 345         '250.5',
 346         'First document sum_item_price field should be set correctly'
 347     );
 348
 349     ok(defined $docs->[0][1]->{marc_data}, 'First document marc_data field should be set');
 350     ok(defined $docs->[0][1]->{marc_format}, 'First document marc_format field should be set');
 351     is($docs->[0][1]->{marc_format}, 'base64ISO2709', 'First document marc_format should be set correctly');
 352
 353     my $decoded_marc_record = $see->decode_record_from_result($docs->[0][1]);
 354
 355     ok($decoded_marc_record->isa('MARC::Record'), "base64ISO2709 record successfully decoded from result");
 356     is($decoded_marc_record->as_usmarc(), $marc_record_1->as_usmarc(), "Decoded base64ISO2709 record has same data as original record");
 357
 358     is(scalar @{$docs->[0][1]->{type_of_record}}, 1, 'First document type_of_record field should have one value');
 359     is_deeply(
 360         $docs->[0][1]->{type_of_record},
 361         ['a'],
 362         'First document type_of_record field should be set correctly'
 363     );
 364
 365     is(scalar @{$docs->[0][1]->{type_of_record_and_bib_level}}, 1, 'First document type_of_record_and_bib_level field should have one value');
 366     is_deeply(
 367         $docs->[0][1]->{type_of_record_and_bib_level},
 368         ['am'],
 369         'First document type_of_record_and_bib_level field should be set correctly'
 370     );
 371
 372     is(scalar @{$docs->[0][1]->{isbn}}, 4, 'First document isbn field should contain four values');
 373     is_deeply($docs->[0][1]->{isbn}, ['978-1-56619-909-4', '9781566199094', '1-56619-909-3', '1566199093'], 'First document isbn field should be set correctly');
 374
 375     # Second record:
 376
 377     is(scalar @{$docs->[1][1]->{author}}, 1, 'Second document author field should contain one value');
 378     is_deeply($docs->[1][1]->{author}, ['Author 2'], 'Second document author field should be set correctly');
 379
 380     is(scalar @{$docs->[1][1]->{items_withdrawn_status}}, 1, 'Second document items_withdrawn_status field should have one value');
 381     is_deeply(
 382         $docs->[1][1]->{items_withdrawn_status},
 383         ['true'],
 384         'Second document items_withdrawn_status field should be set correctly'
 385     );
 386
 387     is(
 388         $docs->[1][1]->{sum_item_price},
 389         0,
 390         'Second document sum_item_price field should be set correctly'
 391     );
 392
 393     # Mappings marc_type:
 394
 395     ok(!(defined $docs->[0][1]->{unimarc_title}), "No mapping when marc_type doesn't match marc flavour");
 396
 397     # Marc serialization format fallback for records exceeding ISO2709 max record size
 398
 399     my $large_marc_record = MARC::Record->new();
 400     $large_marc_record->leader('     cam  22      a 4500');
 401
 402     $large_marc_record->append_fields(
 403         MARC::Field->new('100', '', '', a => 'Author 1'),
 404         MARC::Field->new('110', '', '', a => 'Corp Author'),
 405         MARC::Field->new('210', '', '', a => 'Title 1'),
 406         MARC::Field->new('245', '', '', a => 'Title:', b => 'large record'),
 407         MARC::Field->new('999', '', '', c => '1234567'),
 408     );
 409
 410     my $item_field = MARC::Field->new('952', '', '', o => '123456789123456789123456789', p => '123456789', z => 'test');
 411     my $items_count = 1638;
 412     while(--$items_count) {
 413         $large_marc_record->append_fields($item_field);
 414     }
 415
 416     $docs = $see->marc_records_to_documents([$large_marc_record]);
 417
 418     is($docs->[0][1]->{marc_format}, 'MARCXML', 'For record exceeding max record size marc_format should be set correctly');
 419
 420     $decoded_marc_record = $see->decode_record_from_result($docs->[0][1]);
 421
 422     ok($decoded_marc_record->isa('MARC::Record'), "MARCXML record successfully decoded from result");
 423     is($decoded_marc_record->as_xml_record(), $large_marc_record->as_xml_record(), "Decoded MARCXML record has same data as original record");
 424
 425     push @mappings, {
 426         name => 'title',
 427         type => 'string',
 428         facet => 0,
 429         suggestible => 1,
 430         sort => 1,
 431         marc_type => 'marc21',
 432         marc_field => '245((ab)ab',
 433     };
 434
 435     my $exception = try {
 436         $see->marc_records_to_documents($records);
 437     }
 438     catch {
 439         return $_;
 440     };
 441
 442     ok(defined $exception, "Exception has been thrown when processing mapping with unmatched opening parenthesis");
 443     ok($exception->isa("Koha::Exceptions::Elasticsearch::MARCFieldExprParseError"), "Exception is of correct class");
 444     ok($exception->message =~ /Unmatched opening parenthesis/, "Exception has the correct message");
 445
 446     pop @mappings;
 447     push @mappings, {
 448         name => 'title',
 449         type => 'string',
 450         facet => 0,
 451         suggestible => 1,
 452         sort => 1,
 453         marc_type => 'marc21',
 454         marc_field => '245(ab))ab',
 455     };
 456
 457     $exception = try {
 458         $see->marc_records_to_documents($records);
 459     }
 460     catch {
 461         return $_;
 462     };
 463
 464     ok(defined $exception, "Exception has been thrown when processing mapping with unmatched closing parenthesis");
 465     ok($exception->isa("Koha::Exceptions::Elasticsearch::MARCFieldExprParseError"), "Exception is of correct class");
 466     ok($exception->message =~ /Unmatched closing parenthesis/, "Exception has the correct message");
 467 };