t/db_dependent/Search.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 utf8;
  21
  22 use YAML;
  23
  24 use C4::Debug;
  25 require C4::Context;
  26
  27 # work around spurious wide character warnings
  28 use open ':std', ':encoding(utf8)';
  29
  30 use Test::More tests => 4;
  31 use Test::MockModule;
  32 use Test::Warn;
  33
  34 use Koha::Caches;
  35
  36 use MARC::Record;
  37 use File::Spec;
  38 use File::Basename;
  39 use File::Find;
  40
  41 use File::Temp qw/ tempdir /;
  42 use File::Path;
  43
  44 our $child;
  45 our $datadir;
  46
  47 sub index_sample_records_and_launch_zebra {
  48     my ($datadir, $indexing_mode, $marc_type) = @_;
  49
  50     my $sourcedir = dirname(__FILE__) . "/data";
  51     unlink("$datadir/zebra.log");
  52     if (-f "$sourcedir/${marc_type}/zebraexport/biblio/exported_records") {
  53         my $zebra_bib_cfg = ($indexing_mode eq 'dom') ? 'zebra-biblios-dom.cfg' : 'zebra-biblios.cfg';
  54         system("zebraidx -c $datadir/etc/koha/zebradb/$zebra_bib_cfg  -v none,fatal -g iso2709 -d biblios init");
  55         system("zebraidx -c $datadir/etc/koha/zebradb/$zebra_bib_cfg  -v none,fatal -g iso2709 -d biblios update $sourcedir/${marc_type}/zebraexport/biblio");
  56         system("zebraidx -c $datadir/etc/koha/zebradb/$zebra_bib_cfg  -v none,fatal -g iso2709 -d biblios commit");
  57     }
  58     # ... and add large bib records, if present
  59     if (-f "$sourcedir/${marc_type}/zebraexport/large_biblio_${indexing_mode}/exported_records.xml") {
  60         my $zebra_bib_cfg = ($indexing_mode eq 'dom') ? 'zebra-biblios-dom.cfg' : 'zebra-biblios.cfg';
  61         system("zebraidx -c $datadir/etc/koha/zebradb/$zebra_bib_cfg  -v none,fatal -g marcxml -d biblios update $sourcedir/${marc_type}/zebraexport/large_biblio_${indexing_mode}");
  62         system("zebraidx -c $datadir/etc/koha/zebradb/$zebra_bib_cfg  -v none,fatal -g marcxml -d biblios commit");
  63     }
  64     if (-f "$sourcedir/${marc_type}/zebraexport/authority/exported_records") {
  65         my $zebra_auth_cfg = ($indexing_mode eq 'dom') ? 'zebra-authorities-dom.cfg' : 'zebra-authorities.cfg';
  66         system("zebraidx -c $datadir/etc/koha/zebradb/$zebra_auth_cfg  -v none,fatal -g iso2709 -d authorities init");
  67         system("zebraidx -c $datadir/etc/koha/zebradb/$zebra_auth_cfg  -v none,fatal -g iso2709 -d authorities update $sourcedir/${marc_type}/zebraexport/authority");
  68         system("zebraidx -c $datadir/etc/koha/zebradb/$zebra_auth_cfg  -v none,fatal -g iso2709 -d authorities commit");
  69     }
  70
  71     $child = fork();
  72     if ($child == 0) {
  73         exec("zebrasrv -f $datadir/etc/koha-conf.xml -v none,request -l $datadir/zebra.log");
  74         exit;
  75     }
  76
  77     sleep(1);
  78 }
  79
  80 sub cleanup {
  81     if ($child) {
  82         kill 9, $child;
  83
  84         # Clean up the Zebra files since the child process was just shot
  85         rmtree $datadir;
  86     }
  87 }
  88
  89 # Fall back to make sure that the Zebra process
  90 # and files get cleaned up
  91 END {
  92     cleanup();
  93 }
  94
  95 our $QueryStemming = 0;
  96 our $QueryAutoTruncate = 0;
  97 our $QueryWeightFields = 0;
  98 our $QueryFuzzy = 0;
  99 our $UseQueryParser = 0;
 100 our $SearchEngine = 'Zebra';
 101 our $marcflavour = 'MARC21';
 102 our $contextmodule = new Test::MockModule('C4::Context');
 103 $contextmodule->mock('preference', sub {
 104     my ($self, $pref) = @_;
 105     if ($pref eq 'marcflavour') {
 106         return $marcflavour;
 107     } elsif ($pref eq 'QueryStemming') {
 108         return $QueryStemming;
 109     } elsif ($pref eq 'QueryAutoTruncate') {
 110         return $QueryAutoTruncate;
 111     } elsif ($pref eq 'QueryWeightFields') {
 112         return $QueryWeightFields;
 113     } elsif ($pref eq 'QueryFuzzy') {
 114         return $QueryFuzzy;
 115     } elsif ($pref eq 'UseQueryParser') {
 116         return $UseQueryParser;
 117     } elsif ($pref eq 'SearchEngine') {
 118         return $SearchEngine;
 119     } elsif ($pref eq 'maxRecordsForFacets') {
 120         return 20;
 121     } elsif ($pref eq 'FacetLabelTruncationLength') {
 122         return 20;
 123     } elsif ($pref eq 'FacetMaxCount') {
 124         return 20;
 125     } elsif ($pref eq 'OpacHiddenItems') {
 126         return '';
 127     } elsif ($pref eq 'opacthemes') {
 128         return 'bootstrap';
 129     } elsif ($pref eq 'opaclanguages') {
 130         return 'en';
 131     } elsif ($pref eq 'AlternateHoldingsField') {
 132         return '490av';
 133     } elsif ($pref eq 'AuthoritySeparator') {
 134         return '--';
 135     } elsif ($pref eq 'DisplayLibraryFacets') {
 136         return 'holding';
 137     } elsif ($pref eq 'UNIMARCAuthorsFacetsSeparator') {
 138         return '--';
 139     } else {
 140         warn "The syspref $pref was requested but I don't know what to say; this indicates that the test requires updating"
 141             unless $pref =~ m/(XSLT|item|branch|holding|image)/i;
 142         return 0;
 143     }
 144 });
 145 $contextmodule->mock('queryparser', sub {
 146     my $QParser     = Koha::QueryParser::Driver::PQF->new();
 147     $QParser->load_config("$datadir/etc/searchengine/queryparser.yaml");
 148     return $QParser;
 149 });
 150
 151 our $bibliomodule = new Test::MockModule('C4::Biblio');
 152
 153 sub mock_GetMarcSubfieldStructure {
 154     my $marc_type = shift;
 155     if ($marc_type eq 'marc21') {
 156         $bibliomodule->mock('GetMarcSubfieldStructure', sub {
 157             return {
 158                     'biblio.biblionumber' => { tagfield =>  '999', tagsubfield => 'c' },
 159                     'biblio.isbn' => { tagfield => '020', tagsubfield => 'a' },
 160                     'biblio.title' => { tagfield => '245', tagsubfield => 'a' },
 161                     'biblio.notes' => { tagfield => '500', tagsubfield => 'a' },
 162                     'items.barcode' => { tagfield => '952', tagsubfield => 'p' },
 163                     'items.booksellerid' => { tagfield => '952', tagsubfield => 'e' },
 164                     'items.ccode' => { tagfield => '952', tagsubfield => '8' },
 165                     'items.cn_sort' => { tagfield => '952', tagsubfield => '6' },
 166                     'items.cn_source' => { tagfield => '952', tagsubfield => '2' },
 167                     'items.coded_location_qualifier' => { tagfield => '952', tagsubfield => 'f' },
 168                     'items.copynumber' => { tagfield => '952', tagsubfield => 't' },
 169                     'items.damaged' => { tagfield => '952', tagsubfield => '4' },
 170                     'items.dateaccessioned' => { tagfield => '952', tagsubfield => 'd' },
 171                     'items.datelastborrowed' => { tagfield => '952', tagsubfield => 's' },
 172                     'items.datelastseen' => { tagfield => '952', tagsubfield => 'r' },
 173                     'items.enumchron' => { tagfield => '952', tagsubfield => 'h' },
 174                     'items.holdingbranch' => { tagfield => '952', tagsubfield => 'b' },
 175                     'items.homebranch' => { tagfield => '952', tagsubfield => 'a' },
 176                     'items.issues' => { tagfield => '952', tagsubfield => 'l' },
 177                     'items.itemcallnumber' => { tagfield => '952', tagsubfield => 'o' },
 178                     'items.itemlost' => { tagfield => '952', tagsubfield => '1' },
 179                     'items.itemnotes' => { tagfield => '952', tagsubfield => 'z' },
 180                     'items.itemnumber' => { tagfield => '952', tagsubfield => '9' },
 181                     'items.itype' => { tagfield => '952', tagsubfield => 'y' },
 182                     'items.location' => { tagfield => '952', tagsubfield => 'c' },
 183                     'items.materials' => { tagfield => '952', tagsubfield => '3' },
 184                     'items.nonpublicnote' => { tagfield => '952', tagsubfield => 'x' },
 185                     'items.notforloan' => { tagfield => '952', tagsubfield => '7' },
 186                     'items.onloan' => { tagfield => '952', tagsubfield => 'q' },
 187                     'items.price' => { tagfield => '952', tagsubfield => 'g' },
 188                     'items.renewals' => { tagfield => '952', tagsubfield => 'm' },
 189                     'items.replacementprice' => { tagfield => '952', tagsubfield => 'v' },
 190                     'items.replacementpricedate' => { tagfield => '952', tagsubfield => 'w' },
 191                     'items.reserves' => { tagfield => '952', tagsubfield => 'n' },
 192                     'items.restricted' => { tagfield => '952', tagsubfield => '5' },
 193                     'items.stack' => { tagfield => '952', tagsubfield => 'j' },
 194                     'items.uri' => { tagfield => '952', tagsubfield => 'u' },
 195                     'items.withdrawn' => { tagfield => '952', tagsubfield => '0' },
 196                 };
 197         });
 198     }
 199 }
 200
 201 sub run_marc21_search_tests {
 202     my $indexing_mode = shift;
 203     $datadir = tempdir();
 204     system(dirname(__FILE__) . "/zebra_config.pl $datadir marc21 $indexing_mode");
 205
 206     Koha::Caches->get_instance('config')->flush_all;
 207
 208     mock_GetMarcSubfieldStructure('marc21');
 209     my $context = new C4::Context("$datadir/etc/koha-conf.xml");
 210     $context->set_context();
 211
 212     is($context->config('zebra_bib_index_mode'),$indexing_mode,
 213         "zebra_bib_index_mode is properly set to '$indexing_mode' in the created koha-conf.xml file (BZ11499)");
 214     is($context->config('zebra_auth_index_mode'),$indexing_mode,
 215         "zebra_auth_index_mode is properly set to '$indexing_mode' in the created koha-conf.xml file (BZ11499)");
 216
 217     use_ok('C4::Search');
 218
 219     # set search syspreferences to a known starting point
 220     $QueryStemming = 0;
 221     $QueryAutoTruncate = 0;
 222     $QueryWeightFields = 0;
 223     $QueryFuzzy = 0;
 224     $UseQueryParser = 0;
 225     $marcflavour = 'MARC21';
 226
 227     my $indexes = C4::Search::getIndexes();
 228     is(scalar(grep(/^ti$/, @$indexes)), 1, "Title index supported");
 229
 230     my $bibliomodule = new Test::MockModule('C4::Biblio');
 231
 232     my %branches = (
 233         'CPL' => { 'branchaddress1' => 'Jefferson Summit', 'branchcode' => 'CPL', 'branchname' => 'Centerville', },
 234         'FFL' => { 'branchaddress1' => 'River Station', 'branchcode' => 'FFL', 'branchname' => 'Fairfield', },
 235         'FPL' => { 'branchaddress1' => 'Hickory Squere', 'branchcode' => 'FPL', 'branchname' => 'Fairview', },
 236         'FRL' => { 'branchaddress1' => 'Smith Heights', 'branchcode' => 'FRL', 'branchname' => 'Franklin', },
 237         'IPT' => { 'branchaddress1' => '', 'branchcode' => 'IPT', 'branchname' => "Institut Protestant de Théologie", },
 238         'LPL' => { 'branchaddress1' => 'East Hills', 'branchcode' => 'LPL', 'branchname' => 'Liberty', },
 239         'MPL' => { 'branchaddress1' => '372 Forest Street', 'branchcode' => 'MPL', 'branchname' => 'Midway', },
 240         'PVL' => { 'branchaddress1' => 'Meadow Grove', 'branchcode' => 'PVL', 'branchname' => 'Pleasant Valley', },
 241         'RPL' => { 'branchaddress1' => 'Johnson Terrace', 'branchcode' => 'RPL', 'branchname' => 'Riverside', },
 242         'SPL' => { 'branchaddress1' => 'Highland Boulevard', 'branchcode' => 'SPL', 'branchname' => 'Springfield', },
 243         'S'   => { 'branchaddress1' => '', 'branchcode' => 'S', 'branchname' => 'Test', },
 244         'TPL' => { 'branchaddress1' => 'Valley Way', 'branchcode' => 'TPL', 'branchname' => 'Troy', },
 245         'UPL' => { 'branchaddress1' => 'Chestnut Hollow', 'branchcode' => 'UPL', 'branchname' => 'Union', },
 246     );
 247     my %itemtypes = (
 248         'BK' => { 'imageurl' => 'bridge/book.gif', 'summary' => '', 'itemtype' => 'BK', 'description' => 'Books' },
 249         'CF' => { 'imageurl' => 'bridge/computer_file.gif', 'summary' => '', 'itemtype' => 'CF', 'description' => 'Computer Files' },
 250         'CR' => { 'imageurl' => 'bridge/periodical.gif', 'summary' => '', 'itemtype' => 'CR', 'description' => 'Continuing Resources' },
 251         'MP' => { 'imageurl' => 'bridge/map.gif', 'summary' => '', 'itemtype' => 'MP', 'description' => 'Maps' },
 252         'MU' => { 'imageurl' => 'bridge/sound.gif', 'summary' => '', 'itemtype' => 'MU', 'description' => 'Music' },
 253         'MX' => { 'imageurl' => 'bridge/kit.gif', 'summary' => '', 'itemtype' => 'MX', 'description' => 'Mixed Materials' },
 254         'REF' => { 'imageurl' => '', 'summary' => '', 'itemtype' => 'REF', 'description' => 'Reference' },
 255         'VM' => { 'imageurl' => 'bridge/dvd.gif', 'summary' => '', 'itemtype' => 'VM', 'description' => 'Visual Materials' },
 256     );
 257
 258     index_sample_records_and_launch_zebra($datadir, $indexing_mode, 'marc21');
 259
 260     my ($biblionumber, $title);
 261     my $record = MARC::Record->new;
 262
 263     $record->add_fields(
 264             [ '020', ' ', ' ', a => '9788522421718' ],
 265             [ '245', '0', '0', a => 'Administração da produção /' ]
 266             );
 267     ($biblionumber,undef,$title) = FindDuplicate($record);
 268     is($biblionumber, 51, 'Found duplicate with ISBN');
 269
 270     $record = MARC::Record->new;
 271
 272     $record->add_fields(
 273             [ '100', '1', ' ', a => 'Carter, Philip J.' ],
 274             [ '245', '1', '4', a => 'Test your emotional intelligence :' ]
 275             );
 276     ($biblionumber,undef,$title) = FindDuplicate($record);
 277     is($biblionumber, 203, 'Found duplicate with author/title');
 278
 279     # Testing SimpleSearch
 280
 281     my ( $error, $marcresults, $total_hits ) = SimpleSearch("book", 0, 9);
 282
 283     is(scalar @$marcresults, 9, "SimpleSearch retrieved requested number of records");
 284     is($total_hits, 101, "SimpleSearch for 'book' matched right number of records");
 285     is($error, undef, "SimpleSearch does not return an error when successful");
 286
 287     my $marcresults2;
 288     ( $error, $marcresults2, $total_hits ) = SimpleSearch("book", 5, 5);
 289     is($marcresults->[5], $marcresults2->[0], "SimpleSearch cursor functions");
 290
 291     ( $error, $marcresults, $total_hits ) = SimpleSearch("kw=book", 0, 10);
 292     is($total_hits, 101, "SimpleSearch handles simple CCL");
 293
 294     ( $error, $marcresults, $total_hits ) = SimpleSearch("Music-number=49631-2", 0, 10);
 295     is($total_hits, 1, "SimpleSearch on music publisher number works (bug 8252)");
 296     ( $error, $marcresults, $total_hits ) = SimpleSearch("Identifier-publisher-for-music=49631-2", 0, 10);
 297     is($total_hits, 1, "SimpleSearch on music publisher number works using Identifier-publisher-for-music (bug 8252)");
 298
 299     # Testing getRecords
 300
 301     my $results_hashref;
 302     my $facets_loop;
 303     ( undef, $results_hashref, $facets_loop ) =
 304         getRecords('kw:book', 'book', [], [ 'biblioserver' ], '19', 0, undef, \%branches, \%itemtypes, 'ccl', undef);
 305     is($results_hashref->{biblioserver}->{hits}, 101, "getRecords keyword search for 'book' matched right number of records");
 306     is(scalar @{$results_hashref->{biblioserver}->{RECORDS}}, 19, "getRecords returned requested number of records");
 307     my $record5 = $results_hashref->{biblioserver}->{RECORDS}->[5];
 308     ( undef, $results_hashref, $facets_loop ) =
 309         getRecords('kw:book', 'book', [], [ 'biblioserver' ], '20', 5, undef, \%branches, \%itemtypes, 'ccl', undef);
 310     ok(!defined $results_hashref->{biblioserver}->{RECORDS}->[0] &&
 311         !defined $results_hashref->{biblioserver}->{RECORDS}->[1] &&
 312         !defined $results_hashref->{biblioserver}->{RECORDS}->[2] &&
 313         !defined $results_hashref->{biblioserver}->{RECORDS}->[3] &&
 314         !defined $results_hashref->{biblioserver}->{RECORDS}->[4] &&
 315         $results_hashref->{biblioserver}->{RECORDS}->[5] eq $record5, "getRecords cursor works");
 316
 317     ( undef, $results_hashref, $facets_loop ) =
 318         getRecords('ti:book', 'ti:book', [], [ 'biblioserver' ], '20', 0, undef, \%branches, \%itemtypes, 'ccl', undef);
 319     is($results_hashref->{biblioserver}->{hits}, 11, "getRecords title search for 'book' matched right number of records");
 320
 321     ( undef, $results_hashref, $facets_loop ) =
 322         getRecords('au:Lessig', 'au:Lessig', [], [ 'biblioserver' ], '20', 0, undef, \%branches, \%itemtypes, 'ccl', undef);
 323     is($results_hashref->{biblioserver}->{hits}, 4, "getRecords title search for 'Australia' matched right number of records");
 324
 325 if ( $indexing_mode eq 'dom' ) {
 326     ( undef, $results_hashref, $facets_loop ) =
 327         getRecords('salud', 'salud', [], [ 'biblioserver' ], '19', 0, undef, \%branches, \%itemtypes, 'ccl', undef);
 328     ok(MARC::Record::new_from_xml($results_hashref->{biblioserver}->{RECORDS}->[0],'UTF-8')->title_proper() =~ m/^Efectos del ambiente/ &&
 329         MARC::Record::new_from_xml($results_hashref->{biblioserver}->{RECORDS}->[7],'UTF-8')->title_proper() eq 'Salud y seguridad de los trabajadores del sector salud: manual para gerentes y administradores^ies' &&
 330         MARC::Record::new_from_xml($results_hashref->{biblioserver}->{RECORDS}->[18],'UTF-8')->title_proper() =~ m/^Indicadores de resultados identificados/
 331         , "Simple relevance sorting in getRecords matches old behavior");
 332
 333     ( undef, $results_hashref, $facets_loop ) =
 334         getRecords('salud', 'salud', [ 'author_az' ], [ 'biblioserver' ], '38', 0, undef, \%branches, \%itemtypes, 'ccl', undef);
 335     ok(MARC::Record::new_from_xml($results_hashref->{biblioserver}->{RECORDS}->[0],'UTF-8')->title_proper() =~ m/la enfermedad laboral\^ies$/ &&
 336         MARC::Record::new_from_xml($results_hashref->{biblioserver}->{RECORDS}->[6],'UTF-8')->title_proper() =~ m/^Indicadores de resultados identificados/ &&
 337         MARC::Record::new_from_xml($results_hashref->{biblioserver}->{RECORDS}->[18],'UTF-8')->title_proper() eq 'World health statistics 2009^ien'
 338         , "Simple ascending author sorting in getRecords matches old behavior");
 339
 340     ( undef, $results_hashref, $facets_loop ) =
 341         getRecords('salud', 'salud', [ 'author_za' ], [ 'biblioserver' ], '38', 0, undef, \%branches, \%itemtypes, 'ccl', undef);
 342     ok(MARC::Record::new_from_xml($results_hashref->{biblioserver}->{RECORDS}->[0],'UTF-8')->title_proper() eq 'World health statistics 2009^ien' &&
 343         MARC::Record::new_from_xml($results_hashref->{biblioserver}->{RECORDS}->[12],'UTF-8')->title_proper() =~ m/^Indicadores de resultados identificados/ &&
 344         MARC::Record::new_from_xml($results_hashref->{biblioserver}->{RECORDS}->[18],'UTF-8')->title_proper() =~ m/la enfermedad laboral\^ies$/
 345         , "Simple descending author sorting in getRecords matches old behavior");
 346
 347     ( undef, $results_hashref, $facets_loop ) =
 348         getRecords('salud', 'salud', [ 'pubdate_asc' ], [ 'biblioserver' ], '38', 0, undef, \%branches, \%itemtypes, 'ccl', undef);
 349     ok(MARC::Record::new_from_xml($results_hashref->{biblioserver}->{RECORDS}->[0],'UTF-8')->title_proper() eq 'Manual de higiene industrial^ies' &&
 350         MARC::Record::new_from_xml($results_hashref->{biblioserver}->{RECORDS}->[7],'UTF-8')->title_proper() =~ m/seguridad e higiene del trabajo\^ies$/ &&
 351         MARC::Record::new_from_xml($results_hashref->{biblioserver}->{RECORDS}->[18],'UTF-8')->title_proper() =~ m/^Indicadores de resultados identificados/
 352         , "Simple ascending publication date sorting in getRecords matches old behavior");
 353
 354     ( undef, $results_hashref, $facets_loop ) =
 355         getRecords('salud', 'salud', [ 'pubdate_dsc' ], [ 'biblioserver' ], '38', 0, undef, \%branches, \%itemtypes, 'ccl', undef);
 356     ok(MARC::Record::new_from_xml($results_hashref->{biblioserver}->{RECORDS}->[0],'UTF-8')->title_proper() =~ m/^Estado de salud/ &&
 357         MARC::Record::new_from_xml($results_hashref->{biblioserver}->{RECORDS}->[7],'UTF-8')->title_proper() eq 'World health statistics 2009^ien' &&
 358         MARC::Record::new_from_xml($results_hashref->{biblioserver}->{RECORDS}->[18],'UTF-8')->title_proper() eq 'Manual de higiene industrial^ies'
 359         , "Simple descending publication date sorting in getRecords matches old behavior");
 360
 361 } elsif ( $indexing_mode eq 'grs1' ){
 362     ( undef, $results_hashref, $facets_loop ) =
 363         getRecords('salud', 'salud', [], [ 'biblioserver' ], '19', 0, undef, \%branches, \%itemtypes, 'ccl', undef);
 364     ok(MARC::Record::new_from_usmarc($results_hashref->{biblioserver}->{RECORDS}->[0])->title_proper() =~ m/^Efectos del ambiente/ &&
 365         MARC::Record::new_from_usmarc($results_hashref->{biblioserver}->{RECORDS}->[7])->title_proper() eq 'Salud y seguridad de los trabajadores del sector salud: manual para gerentes y administradores^ies' &&
 366         MARC::Record::new_from_usmarc($results_hashref->{biblioserver}->{RECORDS}->[18])->title_proper() =~ m/^Indicadores de resultados identificados/
 367         , "Simple relevance sorting in getRecords matches old behavior");
 368
 369     ( undef, $results_hashref, $facets_loop ) =
 370         getRecords('salud', 'salud', [ 'author_az' ], [ 'biblioserver' ], '38', 0, undef, \%branches, \%itemtypes, 'ccl', undef);
 371     ok(MARC::Record::new_from_usmarc($results_hashref->{biblioserver}->{RECORDS}->[0])->title_proper() =~ m/la enfermedad laboral\^ies$/ &&
 372         MARC::Record::new_from_usmarc($results_hashref->{biblioserver}->{RECORDS}->[6])->title_proper() =~ m/^Indicadores de resultados identificados/ &&
 373         MARC::Record::new_from_usmarc($results_hashref->{biblioserver}->{RECORDS}->[18])->title_proper() eq 'World health statistics 2009^ien'
 374         , "Simple ascending author sorting in getRecords matches old behavior");
 375
 376     ( undef, $results_hashref, $facets_loop ) =
 377         getRecords('salud', 'salud', [ 'author_za' ], [ 'biblioserver' ], '38', 0, undef, \%branches, \%itemtypes, 'ccl', undef);
 378     ok(MARC::Record::new_from_usmarc($results_hashref->{biblioserver}->{RECORDS}->[0])->title_proper() eq 'World health statistics 2009^ien' &&
 379         MARC::Record::new_from_usmarc($results_hashref->{biblioserver}->{RECORDS}->[12])->title_proper() =~ m/^Indicadores de resultados identificados/ &&
 380         MARC::Record::new_from_usmarc($results_hashref->{biblioserver}->{RECORDS}->[18])->title_proper() =~ m/la enfermedad laboral\^ies$/
 381         , "Simple descending author sorting in getRecords matches old behavior");
 382
 383     ( undef, $results_hashref, $facets_loop ) =
 384         getRecords('salud', 'salud', [ 'pubdate_asc' ], [ 'biblioserver' ], '38', 0, undef, \%branches, \%itemtypes, 'ccl', undef);
 385     ok(MARC::Record::new_from_usmarc($results_hashref->{biblioserver}->{RECORDS}->[0])->title_proper() eq 'Manual de higiene industrial^ies' &&
 386         MARC::Record::new_from_usmarc($results_hashref->{biblioserver}->{RECORDS}->[7])->title_proper() =~ m/seguridad e higiene del trabajo\^ies$/ &&
 387         MARC::Record::new_from_usmarc($results_hashref->{biblioserver}->{RECORDS}->[18])->title_proper() =~ m/^Indicadores de resultados identificados/
 388         , "Simple ascending publication date sorting in getRecords matches old behavior");
 389
 390     ( undef, $results_hashref, $facets_loop ) =
 391         getRecords('salud', 'salud', [ 'pubdate_dsc' ], [ 'biblioserver' ], '38', 0, undef, \%branches, \%itemtypes, 'ccl', undef);
 392     ok(MARC::Record::new_from_usmarc($results_hashref->{biblioserver}->{RECORDS}->[0])->title_proper() =~ m/^Estado de salud/ &&
 393         MARC::Record::new_from_usmarc($results_hashref->{biblioserver}->{RECORDS}->[7])->title_proper() eq 'World health statistics 2009^ien' &&
 394         MARC::Record::new_from_usmarc($results_hashref->{biblioserver}->{RECORDS}->[18])->title_proper() eq 'Manual de higiene industrial^ies'
 395         , "Simple descending publication date sorting in getRecords matches old behavior");
 396 }
 397
 398     ( undef, $results_hashref, $facets_loop ) =
 399         getRecords('books', 'books', [ 'relevance' ], [ 'biblioserver' ], '20', 0, undef, \%branches, \%itemtypes, undef, 1);
 400     $record = MARC::Record::new_from_usmarc($results_hashref->{biblioserver}->{RECORDS}->[0]);
 401     is($record->title_proper(), 'Books', "Scan returned requested item");
 402     is($record->subfield('100', 'a'), 2, "Scan returned correct number of records matching term");
 403     # Time to test buildQuery and searchResults too.
 404
 405     my ( $query, $simple_query, $query_cgi,
 406     $query_desc, $limit, $limit_cgi, $limit_desc,
 407     $query_type );
 408     ( $error, $query, $simple_query, $query_cgi,
 409     $query_desc, $limit, $limit_cgi, $limit_desc,
 410     $query_type ) = buildQuery([], [ 'salud' ], [], [], [], 0, 'en');
 411     like($query, qr/kw\W.*salud/, "Built CCL keyword query");
 412
 413     ($error, $results_hashref, $facets_loop) = getRecords($query,$simple_query,[ ], [ 'biblioserver' ],20,0,undef,\%branches,\%itemtypes,$query_type,0);
 414     is($results_hashref->{biblioserver}->{hits}, 19, "getRecords generated keyword search for 'salud' matched right number of records");
 415
 416     my @newresults = searchResults('opac', $query_desc, $results_hashref->{'biblioserver'}->{'hits'}, 18, 0, 0,
 417         $results_hashref->{'biblioserver'}->{"RECORDS"});
 418     is(scalar @newresults,18, "searchResults returns requested number of hits");
 419
 420     ( $error, $query, $simple_query, $query_cgi,
 421     $query_desc, $limit, $limit_cgi, $limit_desc,
 422     $query_type ) = buildQuery([ 'and' ], [ 'salud', 'higiene' ], [], [], [], 0, 'en');
 423     like($query, qr/kw\W.*salud\W.*and.*kw\W.*higiene/, "Built composed explicit-and CCL keyword query");
 424
 425     ($error, $results_hashref, $facets_loop) = getRecords($query,$simple_query,[ ], [ 'biblioserver' ],20,0,undef,\%branches,\%itemtypes,$query_type,0);
 426     is($results_hashref->{biblioserver}->{hits}, 3, "getRecords generated composed keyword search for 'salud' explicit-and 'higiene' matched right number of records");
 427
 428     ( $error, $query, $simple_query, $query_cgi,
 429     $query_desc, $limit, $limit_cgi, $limit_desc,
 430     $query_type ) = buildQuery([ 'or' ], [ 'salud', 'higiene' ], [], [], [], 0, 'en');
 431     like($query, qr/kw\W.*salud\W.*or.*kw\W.*higiene/, "Built composed explicit-or CCL keyword query");
 432
 433     ($error, $results_hashref, $facets_loop) = getRecords($query,$simple_query,[ ], [ 'biblioserver' ],20,0,undef,\%branches,\%itemtypes,$query_type,0);
 434     is($results_hashref->{biblioserver}->{hits}, 20, "getRecords generated composed keyword search for 'salud' explicit-or 'higiene' matched right number of records");
 435
 436     ( $error, $query, $simple_query, $query_cgi,
 437     $query_desc, $limit, $limit_cgi, $limit_desc,
 438     $query_type ) = buildQuery([], [ 'salud', 'higiene' ], [], [], [], 0, 'en');
 439     like($query, qr/kw\W.*salud\W.*and.*kw\W.*higiene/, "Built composed implicit-and CCL keyword query");
 440
 441     ($error, $results_hashref, $facets_loop) = getRecords($query,$simple_query,[ ], [ 'biblioserver' ],20,0,undef,\%branches,\%itemtypes,$query_type,0);
 442     is($results_hashref->{biblioserver}->{hits}, 3, "getRecords generated composed keyword search for 'salud' implicit-and 'higiene' matched right number of records");
 443
 444     ( $error, $query, $simple_query, $query_cgi,
 445     $query_desc, $limit, $limit_cgi, $limit_desc,
 446     $query_type ) = buildQuery([], [ 'salud' ], [ 'kw' ], [ 'su-to:Laboratorios' ], [], 0, 'en');
 447     like($query, qr/kw\W.*salud\W*and\W*su-to\W.*Laboratorios/, "Faceted query generated correctly");
 448     unlike($query_desc, qr/Laboratorios/, "Facets not included in query description");
 449
 450     ($error, $results_hashref, $facets_loop) = getRecords($query,$simple_query,[ ], [ 'biblioserver' ],20,0,undef,\%branches,\%itemtypes,$query_type,0);
 451     is($results_hashref->{biblioserver}->{hits}, 2, "getRecords generated faceted search matched right number of records");
 452
 453
 454     ( $error, $query, $simple_query, $query_cgi,
 455     $query_desc, $limit, $limit_cgi, $limit_desc,
 456     $query_type ) = buildQuery([], [ '' ], [ 'kw' ], [ 'mc-itype:MP', 'mc-itype:MU' ], [], 0, 'en');
 457
 458     ($error, $results_hashref, $facets_loop) = getRecords($query,$simple_query,[ ], [ 'biblioserver' ],20,0,undef,\%branches,\%itemtypes,$query_type,0);
 459     is($results_hashref->{biblioserver}->{hits}, 2, "getRecords generated mc-faceted search matched right number of records");
 460
 461
 462     ( $error, $query, $simple_query, $query_cgi,
 463     $query_desc, $limit, $limit_cgi, $limit_desc,
 464     $query_type ) = buildQuery([], [ '' ], [ 'kw' ], [ 'mc-loc:GEN', 'branch:FFL' ], [], 0, 'en');
 465
 466     ($error, $results_hashref, $facets_loop) = getRecords($query,$simple_query,[ ], [ 'biblioserver' ],20,0,undef,\%branches,\%itemtypes,$query_type,0);
 467     is($results_hashref->{biblioserver}->{hits}, 2, "getRecords generated multi-faceted search matched right number of records");
 468
 469     ( $error, $query, $simple_query, $query_cgi,
 470     $query_desc, $limit, $limit_cgi, $limit_desc,
 471     $query_type ) = buildQuery([], [ 'NEKLS' ], [ 'Code-institution' ], [], [], 0, 'en');
 472     ($error, $results_hashref, $facets_loop) = getRecords($query,$simple_query,[ ], [ 'biblioserver' ],20,0,undef,\%branches,\%itemtypes,$query_type,0);
 473     is($results_hashref->{biblioserver}->{hits}, 12,
 474        'search using index whose name contains "ns" returns expected results (bug 10271)');
 475
 476     $UseQueryParser = 1;
 477     ( $error, $query, $simple_query, $query_cgi,
 478     $query_desc, $limit, $limit_cgi, $limit_desc,
 479     $query_type ) = buildQuery([], [ 'book' ], [ 'kw' ], [], [], 0, 'en');
 480     ($error, $results_hashref, $facets_loop) = getRecords($query,$simple_query,[ ], [ 'biblioserver' ],20,0,undef,\%branches,\%itemtypes,$query_type,0);
 481     is($results_hashref->{biblioserver}->{hits}, 101, "Search for 'book' with index set to 'kw' returns 101 hits");
 482     ( $error, $query, $simple_query, $query_cgi,
 483     $query_desc, $limit, $limit_cgi, $limit_desc,
 484     $query_type ) = buildQuery([ 'and' ], [ 'book', 'another' ], [ 'kw', 'kw' ], [], [], 0, 'en');
 485     ($error, $results_hashref, $facets_loop) = getRecords($query,$simple_query,[ ], [ 'biblioserver' ],20,0,undef,\%branches,\%itemtypes,$query_type,0);
 486     is($results_hashref->{biblioserver}->{hits}, 1, "Search for 'kw:book && kw:another' returns 1 hit");
 487     $UseQueryParser = 0;
 488
 489     # FIXME: the availability limit does not actually work, so for the moment we
 490     # are just checking that it behaves consistently
 491     ( $error, $query, $simple_query, $query_cgi,
 492     $query_desc, $limit, $limit_cgi, $limit_desc,
 493     $query_type ) = buildQuery([], [ '' ], [ 'kw' ], [ 'available' ], [], 0, 'en');
 494
 495     ($error, $results_hashref, $facets_loop) = getRecords($query,$simple_query,[ ], [ 'biblioserver' ],20,0,undef,\%branches,\%itemtypes,$query_type,0);
 496     is($results_hashref->{biblioserver}->{hits}, 26, "getRecords generated availability-limited search matched right number of records");
 497
 498     @newresults = searchResults('opac', $query_desc, $results_hashref->{'biblioserver'}->{'hits'}, 17, 0, 0,
 499         $results_hashref->{'biblioserver'}->{"RECORDS"});
 500     my $allavailable = 'true';
 501     foreach my $result (@newresults) {
 502         $allavailable = 'false' unless $result->{availablecount} > 0;
 503     }
 504     is ($allavailable, 'true', 'All records have at least one item available');
 505
 506
 507     ( $error, $query, $simple_query, $query_cgi,
 508     $query_desc, $limit, $limit_cgi, $limit_desc,
 509     $query_type ) = buildQuery([], [ 'pqf=@attr 1=_ALLRECORDS @attr 2=103 ""' ], [], [], [], 0, 'en');
 510
 511     ($error, $results_hashref, $facets_loop) = getRecords($query,$simple_query,[ ], [ 'biblioserver' ],20,0,undef,\%branches,\%itemtypes,$query_type,0);
 512     is($results_hashref->{biblioserver}->{hits}, 180, "getRecords on _ALLRECORDS PQF returned all records");
 513
 514     ( $error, $query, $simple_query, $query_cgi,
 515     $query_desc, $limit, $limit_cgi, $limit_desc,
 516     $query_type ) = buildQuery([], [ 'pqf=@attr 1=1016 "Lessig"' ], [], [], [], 0, 'en');
 517
 518     ($error, $results_hashref, $facets_loop) = getRecords($query,$simple_query,[ ], [ 'biblioserver' ],20,0,undef,\%branches,\%itemtypes,$query_type,0);
 519     is($results_hashref->{biblioserver}->{hits}, 4, "getRecords PQF author search for Lessig returned proper number of matches");
 520
 521     ( $error, $query, $simple_query, $query_cgi,
 522     $query_desc, $limit, $limit_cgi, $limit_desc,
 523     $query_type ) = buildQuery([], [ 'ccl=au:Lessig' ], [], [], [], 0, 'en');
 524
 525     ($error, $results_hashref, $facets_loop) = getRecords($query,$simple_query,[ ], [ 'biblioserver' ],20,0,undef,\%branches,\%itemtypes,$query_type,0);
 526     is($results_hashref->{biblioserver}->{hits}, 4, "getRecords CCL author search for Lessig returned proper number of matches");
 527
 528     ( $error, $query, $simple_query, $query_cgi,
 529     $query_desc, $limit, $limit_cgi, $limit_desc,
 530     $query_type ) = buildQuery([], [ 'cql=dc.author any lessig' ], [], [], [], 0, 'en');
 531
 532     ($error, $results_hashref, $facets_loop) = getRecords($query,$simple_query,[ ], [ 'biblioserver' ],20,0,undef,\%branches,\%itemtypes,$query_type,0);
 533     is($results_hashref->{biblioserver}->{hits}, 4, "getRecords CQL author search for Lessig returned proper number of matches");
 534
 535     $QueryStemming = $QueryAutoTruncate = $QueryFuzzy = 0;
 536     $QueryWeightFields = 1;
 537     ( $error, $query, $simple_query, $query_cgi,
 538     $query_desc, $limit, $limit_cgi, $limit_desc,
 539     $query_type ) = buildQuery([], [ 'salud' ], [ 'kw' ], [], [], 0, 'en');
 540
 541     ($error, $results_hashref, $facets_loop) = getRecords($query,$simple_query,[ ], [ 'biblioserver' ],20,0,undef,\%branches,\%itemtypes,$query_type,0);
 542     is($results_hashref->{biblioserver}->{hits}, 19, "Weighted query returned correct number of results");
 543     if ($indexing_mode eq 'grs1') {
 544         is(MARC::Record::new_from_usmarc($results_hashref->{biblioserver}->{RECORDS}->[0])->title_proper(), 'Salud y seguridad de los trabajadores del sector salud: manual para gerentes y administradores^ies', "Weighted query returns best match first");
 545     } else {
 546         local $TODO = "Query weighting does not behave exactly the same in DOM vs. GRS";
 547         is(MARC::Record::new_from_xml($results_hashref->{biblioserver}->{RECORDS}->[0],'UTF-8')->title_proper(), 'Salud y seguridad de los trabajadores del sector salud: manual para gerentes y administradores^ies', "Weighted query returns best match first");
 548     }
 549
 550     $QueryStemming = $QueryWeightFields = $QueryFuzzy = 0;
 551     $QueryAutoTruncate = 1;
 552     ( $error, $query, $simple_query, $query_cgi,
 553     $query_desc, $limit, $limit_cgi, $limit_desc,
 554     $query_type ) = buildQuery([], [ 'medic' ], [ 'kw' ], [], [], 0, 'en');
 555
 556     ($error, $results_hashref, $facets_loop) = getRecords($query,$simple_query,[ ], [ 'biblioserver' ],20,0,undef,\%branches,\%itemtypes,$query_type,0);
 557     is($results_hashref->{biblioserver}->{hits}, 5, "Search for 'medic' returns matches  with automatic truncation on");
 558
 559     ( $error, $query, $simple_query, $query_cgi,
 560     $query_desc, $limit, $limit_cgi, $limit_desc,
 561     $query_type ) = buildQuery([], [ 'medic*' ], [ 'kw' ], [], [], 0, 'en');
 562
 563     ($error, $results_hashref, $facets_loop) = getRecords($query,$simple_query,[ ], [ 'biblioserver' ],20,0,undef,\%branches,\%itemtypes,$query_type,0);
 564     is($results_hashref->{biblioserver}->{hits}, 5, "Search for 'medic*' returns matches with automatic truncation on");
 565
 566     $QueryStemming = $QueryFuzzy = $QueryAutoTruncate = 0;
 567     $QueryWeightFields = 1;
 568     ( $error, $query, $simple_query, $query_cgi,
 569     $query_desc, $limit, $limit_cgi, $limit_desc,
 570     $query_type ) = buildQuery([], [ 'web application' ], [ 'kw' ], [], [], 0, 'en');
 571     ($error, $results_hashref, $facets_loop) = getRecords($query,$simple_query,[ ], [ 'biblioserver' ],20,0,undef,\%branches,\%itemtypes,$query_type,0);
 572     is($results_hashref->{biblioserver}->{hits}, 1, "Search for 'web application' returns one hit with QueryWeightFields on");
 573
 574     ( $error, $query, $simple_query, $query_cgi,
 575     $query_desc, $limit, $limit_cgi, $limit_desc,
 576     $query_type ) = buildQuery([], [ 'web "application' ], [ 'kw' ], [], [], 0, 'en');
 577     ($error, $results_hashref, $facets_loop) = getRecords($query,$simple_query,[ ], [ 'biblioserver' ],20,0,undef,\%branches,\%itemtypes,$query_type,0);
 578     is($results_hashref->{biblioserver}->{hits}, 1, "Search for 'web \"application' returns one hit with QueryWeightFields on (bug 7518)");
 579
 580     $QueryStemming = $QueryWeightFields = $QueryFuzzy = $QueryAutoTruncate = 0;
 581     ( $error, $query, $simple_query, $query_cgi,
 582     $query_desc, $limit, $limit_cgi, $limit_desc,
 583     $query_type ) = buildQuery([], [ 'medic' ], [ 'kw' ], [], [], 0, 'en');
 584
 585     ($error, $results_hashref, $facets_loop) = getRecords($query,$simple_query,[ ], [ 'biblioserver' ],20,0,undef,\%branches,\%itemtypes,$query_type,0);
 586     is($results_hashref->{biblioserver}->{hits}, undef, "Search for 'medic' returns no matches with automatic truncation off");
 587
 588     ( $error, $query, $simple_query, $query_cgi,
 589     $query_desc, $limit, $limit_cgi, $limit_desc,
 590     $query_type ) = buildQuery([], [ 'medic*' ], [ 'kw' ], [], [], 0, 'en');
 591
 592     ($error, $results_hashref, $facets_loop) = getRecords($query,$simple_query,[ ], [ 'biblioserver' ],20,0,undef,\%branches,\%itemtypes,$query_type,0);
 593     is($results_hashref->{biblioserver}->{hits}, 5, "Search for 'medic*' returns matches with automatic truncation off");
 594
 595     $QueryStemming = $QueryWeightFields = 1;
 596     $QueryFuzzy = $QueryAutoTruncate = 0;
 597     ( $error, $query, $simple_query, $query_cgi,
 598     $query_desc, $limit, $limit_cgi, $limit_desc,
 599     $query_type ) = buildQuery([], [ 'pressed' ], [ 'kw' ], [], [], 0, 'en');
 600
 601     ($error, $results_hashref, $facets_loop) = getRecords($query,$simple_query,[ ], [ 'biblioserver' ],20,0,undef,\%branches,\%itemtypes,$query_type,0);
 602     is($results_hashref->{biblioserver}->{hits}, 7, "Search for 'pressed' returns matches when stemming (and query weighting) is on");
 603
 604     $QueryStemming = $QueryWeightFields = $QueryFuzzy = $QueryAutoTruncate = 0;
 605     ( $error, $query, $simple_query, $query_cgi,
 606     $query_desc, $limit, $limit_cgi, $limit_desc,
 607     $query_type ) = buildQuery([], [ 'pressed' ], [ 'kw' ], [], [], 0, 'en');
 608
 609     ($error, $results_hashref, $facets_loop) = getRecords($query,$simple_query,[ ], [ 'biblioserver' ],20,0,undef,\%branches,\%itemtypes,$query_type,0);
 610     is($results_hashref->{biblioserver}->{hits}, undef, "Search for 'pressed' returns no matches when stemming is off");
 611
 612     ( $error, $query, $simple_query, $query_cgi,
 613     $query_desc, $limit, $limit_cgi, $limit_desc,
 614     $query_type ) = buildQuery([], [ 'ccl=an:42' ], [], ['available'], [], 0, 'en');
 615     is( $query, "an:42 and ( ( allrecords,AlwaysMatches:'' not onloan,AlwaysMatches:'') and (lost,st-numeric=0) )", 'buildQuery should add the available part to the query if requested with ccl' );
 616     is( $query_desc, 'an:42', 'buildQuery should remove the available part from the query' );
 617
 618     # Let's see what happens when we pass bad data into these routines.
 619     # We have to catch warnings since we're not very good about returning errors.
 620
 621     warning_like { ( $error, $marcresults, $total_hits ) = SimpleSearch("@==ccl blah", 0, 9) } qr/CCL parsing error/,
 622         "SimpleSearch warns about CCL parsing error with nonsense query";
 623     isnt($error, undef, "SimpleSearch returns an error when passed gibberish");
 624
 625     warning_like {( undef, $results_hashref, $facets_loop ) =
 626         getRecords('kw:book', 'book', [], [ 'biblioserver' ], '19', 0, undef, \%branches, \%itemtypes, 'nonsense', undef) }
 627         qr/Unknown query_type/, "getRecords warns about unknown query type";
 628
 629     warning_like {( undef, $results_hashref, $facets_loop ) =
 630         getRecords('pqf=@attr 1=4 "title"', 'pqf=@attr 1=4 "title"', [], [ 'biblioserver' ], '19', 0, undef, \%branches, \%itemtypes, '', undef) }
 631         qr/WARNING: query problem/, "getRecords warns when query type is not specified for non-CCL query";
 632
 633     # Let's just test a few other bits and bobs, just for fun
 634
 635     ($error, $results_hashref, $facets_loop) = getRecords("Godzina pąsowej róży","Godzina pąsowej róży",[ ], [ 'biblioserver' ],20,0,undef,\%branches,\%itemtypes,$query_type,0);
 636     @newresults = searchResults('intranet', $query_desc, $results_hashref->{'biblioserver'}->{'hits'}, 17, 0, 0,
 637         $results_hashref->{'biblioserver'}->{"RECORDS"});
 638     is($newresults[0]->{'alternateholdings_count'}, 1, 'Alternate holdings filled in correctly');
 639
 640
 641     ## Regression test for Bug 10741
 642
 643     # make one of the test items appear to be in transit
 644     my $circ_module = new Test::MockModule('C4::Circulation');
 645     $circ_module->mock('GetTransfers', sub {
 646         my $itemnumber = shift // -1;
 647         if ($itemnumber == 11) {
 648             return ('2013-07-19', 'MPL', 'CPL');
 649         } else {
 650             return;
 651         }
 652     });
 653
 654     ($error, $results_hashref, $facets_loop) = getRecords("TEST12121212","TEST12121212",[ ], [ 'biblioserver' ],20,0,undef,\%branches,\%itemtypes,$query_type,0);
 655     @newresults = searchResults('intranet', $query_desc, $results_hashref->{'biblioserver'}->{'hits'}, 17, 0, 0,
 656         $results_hashref->{'biblioserver'}->{"RECORDS"});
 657     ok(!exists($newresults[0]->{norequests}), 'presence of a transit does not block hold request action (bug 10741)');
 658
 659     ## Regression test for bug 10684
 660     ( undef, $results_hashref, $facets_loop ) =
 661         getRecords('ti:punctuation', 'punctuation', [], [ 'biblioserver' ], '19', 0, undef, \%branches, \%itemtypes, 'ccl', undef);
 662     is($results_hashref->{biblioserver}->{hits}, 1, "search for ti:punctuation returned expected number of records");
 663     warning_like { @newresults = searchResults('intranet', $query_desc,
 664                     $results_hashref->{'biblioserver'}->{'hits'}, 20, 0, 0,
 665                     $results_hashref->{'biblioserver'}->{"RECORDS"}) }
 666                 qr/^ERROR DECODING RECORD - Tag "50%" is not a valid tag/,
 667                 "Warning is raised correctly for invalid tags in MARC::Record";
 668     is(scalar(@newresults), 0, 'a record that cannot be parsed by MARC::Record is simply skipped (bug 10684)');
 669
 670     # Testing exploding indexes
 671     my $term;
 672     my $searchmodule = new Test::MockModule('C4::Search');
 673     $searchmodule->mock('SimpleSearch', sub {
 674         my $query = shift;
 675
 676         is($query, "he:$term", "Searching for expected term '$term' for exploding") or return '', [], 0;
 677
 678         my $record = MARC::Record->new;
 679         if ($query =~ m/Arizona/) {
 680             $record->add_fields(
 681                 [ '001', '1234' ],
 682                 [ '151', ' ', ' ', a => 'Arizona' ],
 683                 [ '551', ' ', ' ', a => 'United States', w => 'g' ],
 684                 [ '551', ' ', ' ', a => 'Maricopa County', w => 'h' ],
 685                 [ '551', ' ', ' ', a => 'Navajo County', w => 'h' ],
 686                 [ '551', ' ', ' ', a => 'Pima County', w => 'h' ],
 687                 [ '551', ' ', ' ', a => 'New Mexico' ],
 688                 );
 689         }
 690         return '', [ $record->as_usmarc() ], 1;
 691     });
 692
 693     $UseQueryParser = 1;
 694     $term = 'Arizona';
 695     ( $error, $query, $simple_query, $query_cgi,
 696     $query_desc, $limit, $limit_cgi, $limit_desc,
 697     $query_type ) = buildQuery([], [ $term ], [ 'su-br' ], [  ], [], 0, 'en');
 698     matchesExplodedTerms("Advanced search for broader subjects", $query, 'Arizona', 'United States');
 699
 700     ( $error, $query, $simple_query, $query_cgi,
 701     $query_desc, $limit, $limit_cgi, $limit_desc,
 702     $query_type ) = buildQuery([], [ $term ], [ 'su-na' ], [  ], [], 0, 'en');
 703     matchesExplodedTerms("Advanced search for narrower subjects", $query, 'Arizona', 'Maricopa County', 'Navajo County', 'Pima County');
 704
 705     ( $error, $query, $simple_query, $query_cgi,
 706     $query_desc, $limit, $limit_cgi, $limit_desc,
 707     $query_type ) = buildQuery([], [ $term ], [ 'su-rl' ], [  ], [], 0, 'en');
 708     matchesExplodedTerms("Advanced search for related subjects", $query, 'Arizona', 'United States', 'Maricopa County', 'Navajo County', 'Pima County');
 709
 710     ( $error, $query, $simple_query, $query_cgi,
 711     $query_desc, $limit, $limit_cgi, $limit_desc,
 712     $query_type ) = buildQuery([], [ "$term", 'history' ], [ 'su-rl', 'kw' ], [  ], [], 0, 'en');
 713     matchesExplodedTerms("Advanced search for related subjects and keyword 'history' searches related subjects", $query, 'Arizona', 'United States', 'Maricopa County', 'Navajo County', 'Pima County');
 714     like($query, qr/history/, "Advanced search for related subjects and keyword 'history' searches for 'history'");
 715
 716     ( $error, $query, $simple_query, $query_cgi,
 717     $query_desc, $limit, $limit_cgi, $limit_desc,
 718     $query_type ) = buildQuery([], [ 'history', "$term" ], [ 'kw', 'su-rl' ], [  ], [], 0, 'en');
 719     matchesExplodedTerms("Order of terms doesn't matter for advanced search", $query, 'Arizona', 'United States', 'Maricopa County', 'Navajo County', 'Pima County');
 720     like($query, qr/history/, "Order of terms doesn't matter for advanced search");
 721
 722     ( $error, $query, $simple_query, $query_cgi,
 723     $query_desc, $limit, $limit_cgi, $limit_desc,
 724     $query_type ) = buildQuery([], [ "su-br($term)" ], [  ], [  ], [], 0, 'en');
 725     matchesExplodedTerms("Simple search for broader subjects", $query, 'Arizona', 'United States');
 726
 727     ( $error, $query, $simple_query, $query_cgi,
 728     $query_desc, $limit, $limit_cgi, $limit_desc,
 729     $query_type ) = buildQuery([], [ "su-na($term)" ], [  ], [  ], [], 0, 'en');
 730     matchesExplodedTerms("Simple search for narrower subjects", $query, 'Arizona', 'Maricopa County', 'Navajo County', 'Pima County');
 731
 732     ( $error, $query, $simple_query, $query_cgi,
 733     $query_desc, $limit, $limit_cgi, $limit_desc,
 734     $query_type ) = buildQuery([], [ "su-rl($term)" ], [  ], [  ], [], 0, 'en');
 735     matchesExplodedTerms("Simple search for related subjects", $query, 'Arizona', 'United States', 'Maricopa County', 'Navajo County', 'Pima County');
 736
 737     ( $error, $query, $simple_query, $query_cgi,
 738     $query_desc, $limit, $limit_cgi, $limit_desc,
 739     $query_type ) = buildQuery([], [ "history && su-rl($term)" ], [  ], [  ], [], 0, 'en');
 740     matchesExplodedTerms("Simple search for related subjects and keyword 'history' searches related subjects", $query, 'Arizona', 'United States', 'Maricopa County', 'Navajo County', 'Pima County');
 741     like($query, qr/history/, "Simple search for related subjects and keyword 'history' searches for 'history'");
 742
 743     sub matchesExplodedTerms {
 744         my ($message, $query, @terms) = @_;
 745         my $match = '(' . join ('|', map { " \@attr 1=Subject \@attr 4=1 \"$_\"" } @terms) . "){" . scalar(@terms) . "}";
 746         like($query, qr/$match/, $message);
 747     }
 748
 749     # authority records
 750     use_ok('C4::AuthoritiesMarc');
 751     $UseQueryParser = 0;
 752
 753     my ($auths, $count) = SearchAuthorities(
 754         ['mainentry'], ['and'], [''], ['starts'],
 755         ['shakespeare'], 0, 10, '', '', 1
 756     );
 757     is($count, 1, 'MARC21 authorities: one hit on mainentry starts with "shakespeare"');
 758     ($auths, $count) = SearchAuthorities(
 759         ['mainentry'], ['and'], [''], ['starts'],
 760         ['shakespeare'], 0, 10, '', 'HeadingAsc', 1
 761     );
 762     is($count, 1, 'MARC21 authorities: one hit on mainentry starts with "shakespeare" sorted by heading ascending');
 763     ($auths, $count) = SearchAuthorities(
 764         ['mainentry'], ['and'], [''], ['starts'],
 765         ['shakespeare'], 0, 10, '', 'HeadingDsc', 1
 766     );
 767     is($count, 1, 'MARC21 authorities: one hit on mainentry starts with "shakespeare" sorted by heading descending');
 768     ($auths, $count) = SearchAuthorities(
 769         ['match'], ['and'], [''], ['contains'],
 770         ['沙士北亞威廉姆'], 0, 10, '', '', 1
 771     );
 772     is($count, 1, 'MARC21 authorities: one hit on match contains "沙士北亞威廉姆"');
 773
 774     $UseQueryParser = 1;
 775
 776     ($auths, $count) = SearchAuthorities(
 777         ['mainentry'], ['and'], [''], ['starts'],
 778         ['shakespeare'], 0, 10, '', '', 1
 779     );
 780     is($count, 1, 'MARC21 authorities: one hit on mainentry starts with "shakespeare" (QP)');
 781     ($auths, $count) = SearchAuthorities(
 782         ['mainentry'], ['and'], [''], ['starts'],
 783         ['shakespeare'], 0, 10, '', 'HeadingAsc', 1
 784     );
 785     is($count, 1, 'MARC21 authorities: one hit on mainentry starts with "shakespeare" sorted by heading ascending (QP)');
 786     ($auths, $count) = SearchAuthorities(
 787         ['mainentry'], ['and'], [''], ['starts'],
 788         ['shakespeare'], 0, 10, '', 'HeadingDsc', 1
 789     );
 790     is($count, 1, 'MARC21 authorities: one hit on mainentry starts with "shakespeare" sorted by heading descending (QP)');
 791     ($auths, $count) = SearchAuthorities(
 792         ['match'], ['and'], [''], ['contains'],
 793         ['沙士北亞威廉姆'], 0, 10, '', '', 1
 794     );
 795     is($count, 1, 'MARC21 authorities: one hit on match contains "沙士北亞威廉姆" (QP)');
 796
 797     # retrieve records that are larger than the MARC limit of 99,999 octets
 798     ( undef, $results_hashref, $facets_loop ) =
 799         getRecords('ti:marc the large record', '', [], [ 'biblioserver' ], '20', 0, undef, \%branches, \%itemtypes, 'ccl', undef);
 800     is($results_hashref->{biblioserver}->{hits}, 1, "Can do a search that retrieves an over-large bib record (bug 11096)");
 801     @newresults = searchResults('opac', $query_desc, $results_hashref->{'biblioserver'}->{'hits'}, 10, 0, 0,
 802         $results_hashref->{'biblioserver'}->{"RECORDS"});
 803     is($newresults[0]->{title}, 'Marc the Large Record', 'Able to render the title for over-large bib record (bug 11096)');
 804     is($newresults[0]->{biblionumber}, '300', 'Over-large bib record has the correct biblionumber (bug 11096)');
 805     like($newresults[0]->{notes}, qr/This is large note #550/, 'Able to render the notes field for over-large bib record (bug 11096)');
 806
 807     # notforloancount should be returned as part of searchResults output
 808     ok( defined $newresults[0]->{notforloancount},
 809         '\'notforloancount\' defined in searchResults output (Bug 12419)');
 810     is( $newresults[0]->{notforloancount}, 2,
 811         '\'notforloancount\' == 2 (Bug 12419)');
 812
 813     # verify that we don't attempt to sort if no results were returned
 814     # because of a query error
 815     warning_like {( undef, $results_hashref, $facets_loop ) =
 816         getRecords('ccl=( AND )', '', ['title_az'], [ 'biblioserver' ], '20', 0, undef, \%branches, \%itemtypes, 'ccl', undef)
 817     } qr/WARNING: query problem with/, 'got warning instead of crash when attempting to run invalid query (bug 9578)';
 818
 819     # Test facet calculation
 820     my $facets_counter = {};
 821     my $facets         = C4::Koha::getFacets();
 822     # Create a record with a 100$z field
 823     my $marc_record    = MARC::Record->new;
 824     $marc_record->add_fields(
 825         [ '001', '1234' ],
 826         [ '100', ' ', ' ', a => 'Cohen Arazi, Tomas' ],
 827         [ '100', 'z', ' ', a => 'Tomasito' ],
 828         [ '245', ' ', ' ', a => 'First try' ]
 829     );
 830     C4::Search::_get_facets_data_from_record( $marc_record, $facets, $facets_counter );
 831     is_deeply( { au => { 'Cohen Arazi, Tomas' => 1 } },  $facets_counter,
 832         "_get_facets_data_from_record doesn't count 100\$z (Bug 12788)");
 833     $marc_record    = MARC::Record->new;
 834     $marc_record->add_fields(
 835         [ '001', '1234' ],
 836         [ '100', ' ', ' ', a => 'Cohen Arazi, Tomas' ],
 837         [ '100', 'z', ' ', a => 'Tomasito' ],
 838         [ '245', ' ', ' ', a => 'Second try' ]
 839     );
 840     C4::Search::_get_facets_data_from_record( $marc_record, $facets, $facets_counter );
 841     is_deeply( { au => { 'Cohen Arazi, Tomas' => 2 } },  $facets_counter,
 842         "_get_facets_data_from_record correctly counts author facet twice");
 843
 844     # Test _get_facets_info
 845     my $facets_info = C4::Search::_get_facets_info( $facets );
 846     my $expected_facets_info_marc21 = {
 847                    'au' => { 'expanded'    => undef,
 848                              'label_value' => "Authors" },
 849         'holdingbranch' => { 'expanded'    => undef,
 850                              'label_value' => "HoldingLibrary" },
 851                 'itype' => { 'expanded'    => undef,
 852                              'label_value' => "ItemTypes" },
 853              'location' => { 'expanded'    => undef,
 854                              'label_value' => "Location" },
 855                    'se' => { 'expanded'    => undef,
 856                              'label_value' => "Series" },
 857                'su-geo' => { 'expanded'    => undef,
 858                              'label_value' => "Places" },
 859                 'su-to' => { 'expanded'    => undef,
 860                              'label_value' => "Topics" },
 861                 'su-ut' => { 'expanded'    => undef,
 862                              'label_value' => "Titles" }
 863     };
 864     is_deeply( $facets_info, $expected_facets_info_marc21,
 865         "_get_facets_info returns the correct data");
 866
 867     cleanup();
 868 }
 869
 870 sub run_unimarc_search_tests {
 871     my $indexing_mode = shift;
 872     $datadir = tempdir();
 873     system(dirname(__FILE__) . "/zebra_config.pl $datadir unimarc $indexing_mode");
 874
 875     Koha::Caches->get_instance('config')->flush_all;
 876
 877     mock_GetMarcSubfieldStructure('unimarc');
 878     my $context = new C4::Context("$datadir/etc/koha-conf.xml");
 879     $context->set_context();
 880
 881     use_ok('C4::Search');
 882
 883     # set search syspreferences to a known starting point
 884     $QueryStemming = 0;
 885     $QueryAutoTruncate = 0;
 886     $QueryWeightFields = 0;
 887     $QueryFuzzy = 0;
 888     $UseQueryParser = 0;
 889     $marcflavour = 'UNIMARC';
 890
 891     index_sample_records_and_launch_zebra($datadir, $indexing_mode, 'unimarc');
 892
 893     my ( $error, $marcresults, $total_hits ) = SimpleSearch("ti=Järnvägarnas efterfrågan och den svenska industrin", 0, 10);
 894     is($total_hits, 1, 'UNIMARC title search');
 895     ( $error, $marcresults, $total_hits ) = SimpleSearch("ta=u", 0, 10);
 896     is($total_hits, 1, 'UNIMARC target audience = u');
 897     ( $error, $marcresults, $total_hits ) = SimpleSearch("ta=k", 0, 10);
 898     is($total_hits, 4, 'UNIMARC target audience = k');
 899     ( $error, $marcresults, $total_hits ) = SimpleSearch("ta=m", 0, 10);
 900     is($total_hits, 3, 'UNIMARC target audience = m');
 901     ( $error, $marcresults, $total_hits ) = SimpleSearch("item=EXCLU DU PRET", 0, 10);
 902     is($total_hits, 1, 'UNIMARC generic item index (bug 10037)');
 903
 904     # authority records
 905     use_ok('C4::AuthoritiesMarc');
 906     $UseQueryParser = 0;
 907
 908     my ($auths, $count) = SearchAuthorities(
 909         ['mainentry'], ['and'], [''], ['contains'],
 910         ['wil'], 0, 10, '', '', 1
 911     );
 912     is($count, 11, 'UNIMARC authorities: hits on mainentry contains "wil"');
 913     ($auths, $count) = SearchAuthorities(
 914         ['match'], ['and'], [''], ['contains'],
 915         ['wil'], 0, 10, '', '', 1
 916     );
 917     is($count, 11, 'UNIMARC authorities: hits on match contains "wil"');
 918     ($auths, $count) = SearchAuthorities(
 919         ['mainentry'], ['and'], [''], ['contains'],
 920         ['michel'], 0, 20, '', '', 1
 921     );
 922     is($count, 14, 'UNIMARC authorities: hits on mainentry contains "michel"');
 923     ($auths, $count) = SearchAuthorities(
 924         ['mainmainentry'], ['and'], [''], ['exact'],
 925         ['valley'], 0, 20, '', '', 1
 926     );
 927     is($count, 1, 'UNIMARC authorities: hits on mainmainentry = "valley"');
 928     ($auths, $count) = SearchAuthorities(
 929         ['mainmainentry'], ['and'], [''], ['exact'],
 930         ['vall'], 0, 20, '', '', 1
 931     );
 932     is($count, 0, 'UNIMARC authorities: no hits on mainmainentry = "vall"');
 933     ($auths, $count) = SearchAuthorities(
 934         ['Any'], ['and'], [''], ['starts'],
 935         ['jean'], 0, 30, '', '', 1
 936     );
 937     is($count, 24, 'UNIMARC authorities: hits on any starts with "jean"');
 938
 939     # Test _get_facets_info
 940     my $facets      = C4::Koha::getFacets();
 941     my $facets_info = C4::Search::_get_facets_info( $facets );
 942     my $expected_facets_info_unimarc = {
 943                    'au' => { 'expanded'    => undef,
 944                              'label_value' => "Authors" },
 945         'holdingbranch' => { 'expanded'    => undef,
 946                              'label_value' => "HoldingLibrary" },
 947              'location' => { 'expanded'    => undef,
 948                              'label_value' => "Location" },
 949                    'se' => { 'expanded'    => undef,
 950                              'label_value' => "Series" },
 951                'su-geo' => { 'expanded'    => undef,
 952                              'label_value' => "Places" },
 953                 'su-to' => { 'expanded'    => undef,
 954                              'label_value' => "Topics" },
 955                 'su-ut' => { 'expanded'    => undef,
 956                              'label_value' => "Titles" }
 957     };
 958     is_deeply( $facets_info, $expected_facets_info_unimarc,
 959         "_get_facets_info returns the correct data");
 960
 961     cleanup();
 962 }
 963
 964 subtest 'MARC21 + GRS-1' => sub {
 965     plan tests => 109;
 966     run_marc21_search_tests('grs1');
 967 };
 968
 969 subtest 'MARC21 + DOM' => sub {
 970     plan tests => 109;
 971     run_marc21_search_tests('dom');
 972 };
 973
 974 subtest 'UNIMARC + GRS-1' => sub {
 975     plan tests => 14;
 976     run_unimarc_search_tests('grs1');
 977 };
 978
 979 subtest 'UNIMARC + DOM' => sub {
 980     plan tests => 14;
 981     run_unimarc_search_tests('dom');
 982 };
 983
 984 1;