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 MARC::Record;
  33 use File::Spec;
  34 use File::Basename;
  35 use File::Find;
  36 use Test::Warn;
  37 use File::Temp qw/ tempdir /;
  38 use File::Path;
  39 use DBI;
  40
  41 our $child;
  42 our $datadir;
  43
  44 sub index_sample_records_and_launch_zebra {
  45     my ($datadir, $indexing_mode, $marc_type) = @_;
  46
  47     my $sourcedir = dirname(__FILE__) . "/data";
  48     unlink("$datadir/zebra.log");
  49     if (-f "$sourcedir/${marc_type}/zebraexport/biblio/exported_records") {
  50         my $zebra_bib_cfg = ($indexing_mode eq 'dom') ? 'zebra-biblios-dom.cfg' : 'zebra-biblios.cfg';
  51         system("zebraidx -c $datadir/etc/koha/zebradb/$zebra_bib_cfg  -v none,fatal,warn  -g iso2709 -d biblios init");
  52         system("zebraidx -c $datadir/etc/koha/zebradb/$zebra_bib_cfg  -v none,fatal,warn   -g iso2709 -d biblios update $sourcedir/${marc_type}/zebraexport/biblio");
  53         system("zebraidx -c $datadir/etc/koha/zebradb/$zebra_bib_cfg  -v none,fatal,warn  -g iso2709 -d biblios commit");
  54     }
  55     # ... and add large bib records, if present
  56     if (-f "$sourcedir/${marc_type}/zebraexport/large_biblio_${indexing_mode}/exported_records.xml") {
  57         my $zebra_bib_cfg = ($indexing_mode eq 'dom') ? 'zebra-biblios-dom.cfg' : 'zebra-biblios.cfg';
  58         system("zebraidx -c $datadir/etc/koha/zebradb/$zebra_bib_cfg  -v none,fatal,warn   -g marcxml -d biblios update $sourcedir/${marc_type}/zebraexport/large_biblio_${indexing_mode}");
  59         system("zebraidx -c $datadir/etc/koha/zebradb/$zebra_bib_cfg  -v none,fatal,warn  -g marcxml -d biblios commit");
  60     }
  61     if (-f "$sourcedir/${marc_type}/zebraexport/authority/exported_records") {
  62         my $zebra_auth_cfg = ($indexing_mode eq 'dom') ? 'zebra-authorities-dom.cfg' : 'zebra-authorities.cfg';
  63         system("zebraidx -c $datadir/etc/koha/zebradb/$zebra_auth_cfg  -v none,fatal,warn  -g iso2709 -d authorities init");
  64         system("zebraidx -c $datadir/etc/koha/zebradb/$zebra_auth_cfg  -v none,fatal,warn   -g iso2709 -d authorities update $sourcedir/${marc_type}/zebraexport/authority");
  65         system("zebraidx -c $datadir/etc/koha/zebradb/$zebra_auth_cfg  -v none,fatal,warn  -g iso2709 -d authorities commit");
  66     }
  67
  68     $child = fork();
  69     if ($child == 0) {
  70         exec("zebrasrv -f $datadir/etc/koha-conf.xml -v none,request -l $datadir/zebra.log");
  71         exit;
  72     }
  73
  74     sleep(1);
  75 }
  76
  77 sub cleanup {
  78     if ($child) {
  79         kill 9, $child;
  80
  81         # Clean up the Zebra files since the child process was just shot
  82         rmtree $datadir;
  83     }
  84 }
  85
  86 # Fall back to make sure that the Zebra process
  87 # and files get cleaned up
  88 END {
  89     cleanup();
  90 }
  91
  92 our $QueryStemming = 0;
  93 our $QueryAutoTruncate = 0;
  94 our $QueryWeightFields = 0;
  95 our $QueryFuzzy = 0;
  96 our $QueryRemoveStopwords = 0;
  97 our $UseQueryParser = 0;
  98 our $marcflavour = 'MARC21';
  99 our $contextmodule = new Test::MockModule('C4::Context');
 100 $contextmodule->mock('_new_dbh', sub {
 101     my $dbh = DBI->connect( 'DBI:Mock:', '', '' )
 102     || die "Cannot create handle: $DBI::errstr\n";
 103     return $dbh });
 104 $contextmodule->mock('preference', sub {
 105     my ($self, $pref) = @_;
 106     if ($pref eq 'marcflavour') {
 107         return $marcflavour;
 108     } elsif ($pref eq 'QueryStemming') {
 109         return $QueryStemming;
 110     } elsif ($pref eq 'QueryAutoTruncate') {
 111         return $QueryAutoTruncate;
 112     } elsif ($pref eq 'QueryWeightFields') {
 113         return $QueryWeightFields;
 114     } elsif ($pref eq 'QueryFuzzy') {
 115         return $QueryFuzzy;
 116     } elsif ($pref eq 'QueryRemoveStopwords') {
 117         return $QueryRemoveStopwords;
 118     } elsif ($pref eq 'UseQueryParser') {
 119         return $UseQueryParser;
 120     } elsif ($pref eq 'maxRecordsForFacets') {
 121         return 20;
 122     } elsif ($pref eq 'FacetLabelTruncationLength') {
 123         return 20;
 124     } elsif ($pref eq 'FacetMaxCount') {
 125         return 20;
 126     } elsif ($pref eq 'OpacHiddenItems') {
 127         return '';
 128     } elsif ($pref eq 'AlternateHoldingsField') {
 129         return '490av';
 130     } elsif ($pref eq 'AuthoritySeparator') {
 131         return '--';
 132     } elsif ($pref eq 'DisplayLibraryFacets') {
 133         return 'holding';
 134     } elsif ($pref eq 'UNIMARCAuthorsFacetsSeparator') {
 135         return '--';
 136     } else {
 137         warn "The syspref $pref was requested but I don't know what to say; this indicates that the test requires updating"
 138             unless $pref =~ m/(XSLT|item|branch|holding|image)/i;
 139         return 0;
 140     }
 141 });
 142 $contextmodule->mock('queryparser', sub {
 143     my $QParser     = Koha::QueryParser::Driver::PQF->new();
 144     $QParser->load_config("$datadir/etc/searchengine/queryparser.yaml");
 145     return $QParser;
 146 });
 147
 148 sub mock_marcfromkohafield {
 149     my $marc_type = shift;
 150     if ($marc_type eq 'marc21') {
 151         $contextmodule->mock('marcfromkohafield', sub {
 152             return {
 153                 '' => {
 154                     'biblio.biblionumber' => [ '999', 'c' ],
 155                     'items.barcode' => ['952', 'p' ],
 156                     'items.booksellerid' => ['952', 'e' ],
 157                     'items.ccode' => ['952', '8' ],
 158                     'items.cn_sort' => ['952', '6' ],
 159                     'items.cn_source' => ['952', '2' ],
 160                     'items.coded_location_qualifier' => ['952', 'f' ],
 161                     'items.copynumber' => ['952', 't' ],
 162                     'items.damaged' => ['952', '4' ],
 163                     'items.dateaccessioned' => ['952', 'd' ],
 164                     'items.datelastborrowed' => ['952', 's' ],
 165                     'items.datelastseen' => ['952', 'r' ],
 166                     'items.enumchron' => ['952', 'h' ],
 167                     'items.holdingbranch' => ['952', 'b' ],
 168                     'items.homebranch' => ['952', 'a' ],
 169                     'items.issues' => ['952', 'l' ],
 170                     'items.itemcallnumber' => ['952', 'o' ],
 171                     'items.itemlost' => ['952', '1' ],
 172                     'items.itemnotes' => ['952', 'z' ],
 173                     'items.itemnumber' => ['952', '9' ],
 174                     'items.itype' => ['952', 'y' ],
 175                     'items.location' => ['952', 'c' ],
 176                     'items.materials' => ['952', '3' ],
 177                     'items.nonpublicnote' => ['952', 'x' ],
 178                     'items.notforloan' => ['952', '7' ],
 179                     'items.onloan' => ['952', 'q' ],
 180                     'items.price' => ['952', 'g' ],
 181                     'items.renewals' => ['952', 'm' ],
 182                     'items.replacementprice' => ['952', 'v' ],
 183                     'items.replacementpricedate' => ['952', 'w' ],
 184                     'items.reserves' => ['952', 'n' ],
 185                     'items.restricted' => ['952', '5' ],
 186                     'items.stack' => ['952', 'j' ],
 187                     'items.uri' => ['952', 'u' ],
 188                     'items.withdrawn' => ['952', '0' ]
 189                     }
 190                 };
 191         });
 192     }
 193 }
 194
 195 sub run_marc21_search_tests {
 196     my $indexing_mode = shift;
 197     $datadir = tempdir();
 198     system(dirname(__FILE__) . "/zebra_config.pl $datadir marc21 $indexing_mode");
 199
 200     mock_marcfromkohafield('marc21');
 201     my $context = new C4::Context("$datadir/etc/koha-conf.xml");
 202     $context->set_context();
 203
 204     is($context->config('zebra_bib_index_mode'),$indexing_mode,
 205         "zebra_bib_index_mode is properly set to '$indexing_mode' in the created koha-conf.xml file (BZ11499)");
 206     is($context->config('zebra_auth_index_mode'),$indexing_mode,
 207         "zebra_auth_index_mode is properly set to '$indexing_mode' in the created koha-conf.xml file (BZ11499)");
 208
 209     use_ok('C4::Search');
 210
 211     # set search syspreferences to a known starting point
 212     $QueryStemming = 0;
 213     $QueryAutoTruncate = 0;
 214     $QueryWeightFields = 0;
 215     $QueryFuzzy = 0;
 216     $QueryRemoveStopwords = 0;
 217     $UseQueryParser = 0;
 218     $marcflavour = 'MARC21';
 219
 220     foreach my $string ("Leçon","modèles") {
 221         my @results=C4::Search::_remove_stopwords($string,"kw");
 222         $debug && warn "$string ",Dump(@results);
 223         ok($results[0] eq $string,"$string is not modified");
 224     }
 225
 226     foreach my $string ("A book about the stars") {
 227         my @results=C4::Search::_remove_stopwords($string,"kw");
 228         $debug && warn "$string ",Dump(@results);
 229         ok($results[0] ne $string,"$results[0] from $string");
 230     }
 231
 232     my $indexes = C4::Search::getIndexes();
 233     is(scalar(grep(/^ti$/, @$indexes)), 1, "Title index supported");
 234
 235     my $bibliomodule = new Test::MockModule('C4::Biblio');
 236     $bibliomodule->mock('_get_inverted_marc_field_map', sub {
 237         my %hash = (
 238             '' => {
 239                 '245' => { 'sfs' => { 'a' => [ [ 'biblio', 'title' ] ], 'b' => [ [ 'bibliosubtitle', 'subtitle' ] ] },
 240                     'list' => [ [ 'a', 'biblio', 'title' ], [ 'b', 'bibliosubtitle', 'subtitle' ] ]
 241                 },
 242                 '100' => {
 243                     'sfs' => { 'a' => [ [ 'biblio', 'author' ] ] },
 244                     'list' => [ [ 'a', 'biblio', 'author' ] ]
 245                 },
 246                 '999' => {
 247                     'sfs' => { 'c' => [ [ 'biblio', 'biblionumber' ] ], 'd' => [ [ 'biblioitems', 'biblioitemnumber' ] ] },
 248                     'list' => [ [ 'd', 'biblioitems', 'biblioitemnumber' ], [ 'c', 'biblio', 'biblionumber' ] ]
 249                 },
 250                 '020' => {
 251                     'sfs' => { 'a' => [ [ 'biblioitems', 'isbn' ] ] },
 252                     'list' => [ [ 'a', 'biblioitems', 'isbn' ] ]
 253                 },
 254                 '500' => {
 255                     'sfs' => { 'a' => [ [ 'biblioitems', 'notes' ] ] },
 256                     'list' => [ [ 'a', 'biblioitems', 'notes' ] ]
 257                 },
 258             }
 259         );
 260         return \%hash;
 261     });
 262     my $dbh = C4::Context->dbh;
 263     $dbh->{mock_add_resultset} = {
 264         sql     => 'SHOW COLUMNS FROM items',
 265         results => [
 266             [ 'rows' ], # seems like $sth->rows is getting called
 267                         # implicitly, so we need this to make
 268                         # DBD::Mock return all of the results
 269             [ 'itemnumber' ], [ 'biblionumber' ], [ 'biblioitemnumber' ],
 270             [ 'barcode' ], [ 'dateaccessioned' ], [ 'booksellerid' ],
 271             [ 'homebranch' ], [ 'price' ], [ 'replacementprice' ],
 272             [ 'replacementpricedate' ], [ 'datelastborrowed' ], [ 'datelastseen' ],
 273             [ 'stack' ], [ 'notforloan' ], [ 'damaged' ],
 274             [ 'itemlost' ], [ 'withdrawn' ], [ 'itemcallnumber' ],
 275             [ 'issues' ], [ 'renewals' ], [ 'reserves' ],
 276             [ 'restricted' ], [ 'itemnotes' ], [ 'nonpublicnote' ],
 277             [ 'holdingbranch' ], [ 'paidfor' ], [ 'timestamp' ],
 278             [ 'location' ], [ 'permanent_location' ], [ 'onloan' ],
 279             [ 'cn_source' ], [ 'cn_sort' ], [ 'ccode' ],
 280             [ 'materials' ], [ 'uri' ], [ 'itype' ],
 281             [ 'more_subfields_xml' ], [ 'enumchron' ], [ 'copynumber' ],
 282             [ 'stocknumber' ],
 283         ]
 284     };
 285
 286     my %branches = (
 287         'CPL' => { 'branchaddress1' => 'Jefferson Summit', 'branchcode' => 'CPL', 'branchname' => 'Centerville', },
 288         'FFL' => { 'branchaddress1' => 'River Station', 'branchcode' => 'FFL', 'branchname' => 'Fairfield', },
 289         'FPL' => { 'branchaddress1' => 'Hickory Squere', 'branchcode' => 'FPL', 'branchname' => 'Fairview', },
 290         'FRL' => { 'branchaddress1' => 'Smith Heights', 'branchcode' => 'FRL', 'branchname' => 'Franklin', },
 291         'IPT' => { 'branchaddress1' => '', 'branchcode' => 'IPT', 'branchname' => "Institut Protestant de Théologie", },
 292         'LPL' => { 'branchaddress1' => 'East Hills', 'branchcode' => 'LPL', 'branchname' => 'Liberty', },
 293         'MPL' => { 'branchaddress1' => '372 Forest Street', 'branchcode' => 'MPL', 'branchname' => 'Midway', },
 294         'PVL' => { 'branchaddress1' => 'Meadow Grove', 'branchcode' => 'PVL', 'branchname' => 'Pleasant Valley', },
 295         'RPL' => { 'branchaddress1' => 'Johnson Terrace', 'branchcode' => 'RPL', 'branchname' => 'Riverside', },
 296         'SPL' => { 'branchaddress1' => 'Highland Boulevard', 'branchcode' => 'SPL', 'branchname' => 'Springfield', },
 297         'S'   => { 'branchaddress1' => '', 'branchcode' => 'S', 'branchname' => 'Test', },
 298         'TPL' => { 'branchaddress1' => 'Valley Way', 'branchcode' => 'TPL', 'branchname' => 'Troy', },
 299         'UPL' => { 'branchaddress1' => 'Chestnut Hollow', 'branchcode' => 'UPL', 'branchname' => 'Union', },
 300     );
 301     my %itemtypes = (
 302         'BK' => { 'imageurl' => 'bridge/book.gif', 'summary' => '', 'itemtype' => 'BK', 'description' => 'Books' },
 303         'CF' => { 'imageurl' => 'bridge/computer_file.gif', 'summary' => '', 'itemtype' => 'CF', 'description' => 'Computer Files' },
 304         'CR' => { 'imageurl' => 'bridge/periodical.gif', 'summary' => '', 'itemtype' => 'CR', 'description' => 'Continuing Resources' },
 305         'MP' => { 'imageurl' => 'bridge/map.gif', 'summary' => '', 'itemtype' => 'MP', 'description' => 'Maps' },
 306         'MU' => { 'imageurl' => 'bridge/sound.gif', 'summary' => '', 'itemtype' => 'MU', 'description' => 'Music' },
 307         'MX' => { 'imageurl' => 'bridge/kit.gif', 'summary' => '', 'itemtype' => 'MX', 'description' => 'Mixed Materials' },
 308         'REF' => { 'imageurl' => '', 'summary' => '', 'itemtype' => 'REF', 'description' => 'Reference' },
 309         'VM' => { 'imageurl' => 'bridge/dvd.gif', 'summary' => '', 'itemtype' => 'VM', 'description' => 'Visual Materials' },
 310     );
 311
 312     index_sample_records_and_launch_zebra($datadir, $indexing_mode, 'marc21');
 313
 314     my ($biblionumber, $title);
 315     my $record = MARC::Record->new;
 316
 317     $record->add_fields(
 318             [ '020', ' ', ' ', a => '9788522421718' ],
 319             [ '245', '0', '0', a => 'Administração da produção /' ]
 320             );
 321     ($biblionumber,undef,$title) = FindDuplicate($record);
 322     is($biblionumber, 51, 'Found duplicate with ISBN');
 323
 324     $record = MARC::Record->new;
 325
 326     $record->add_fields(
 327             [ '100', '1', ' ', a => 'Carter, Philip J.' ],
 328             [ '245', '1', '4', a => 'Test your emotional intelligence :' ]
 329             );
 330     ($biblionumber,undef,$title) = FindDuplicate($record);
 331     is($biblionumber, 203, 'Found duplicate with author/title');
 332
 333     # Testing SimpleSearch
 334
 335     my ( $error, $marcresults, $total_hits ) = SimpleSearch("book", 0, 9);
 336
 337     is(scalar @$marcresults, 9, "SimpleSearch retrieved requested number of records");
 338     is($total_hits, 101, "SimpleSearch for 'book' matched right number of records");
 339     is($error, undef, "SimpleSearch does not return an error when successful");
 340
 341     my $marcresults2;
 342     ( $error, $marcresults2, $total_hits ) = SimpleSearch("book", 5, 5);
 343     is($marcresults->[5], $marcresults2->[0], "SimpleSearch cursor functions");
 344
 345     ( $error, $marcresults, $total_hits ) = SimpleSearch("kw=book", 0, 10);
 346     is($total_hits, 101, "SimpleSearch handles simple CCL");
 347
 348     ( $error, $marcresults, $total_hits ) = SimpleSearch("Music-number=49631-2", 0, 10);
 349     is($total_hits, 1, "SimpleSearch on music publisher number works (bug 8252)");
 350     ( $error, $marcresults, $total_hits ) = SimpleSearch("Identifier-publisher-for-music=49631-2", 0, 10);
 351     is($total_hits, 1, "SimpleSearch on music publisher number works using Identifier-publisher-for-music (bug 8252)");
 352
 353     # Testing getRecords
 354
 355     my $results_hashref;
 356     my $facets_loop;
 357     ( undef, $results_hashref, $facets_loop ) =
 358         getRecords('kw:book', 'book', [], [ 'biblioserver' ], '19', 0, undef, \%branches, \%itemtypes, 'ccl', undef);
 359     is($results_hashref->{biblioserver}->{hits}, 101, "getRecords keyword search for 'book' matched right number of records");
 360     is(scalar @{$results_hashref->{biblioserver}->{RECORDS}}, 19, "getRecords returned requested number of records");
 361     my $record5 = $results_hashref->{biblioserver}->{RECORDS}->[5];
 362     ( undef, $results_hashref, $facets_loop ) =
 363         getRecords('kw:book', 'book', [], [ 'biblioserver' ], '20', 5, undef, \%branches, \%itemtypes, 'ccl', undef);
 364     ok(!defined $results_hashref->{biblioserver}->{RECORDS}->[0] &&
 365         !defined $results_hashref->{biblioserver}->{RECORDS}->[1] &&
 366         !defined $results_hashref->{biblioserver}->{RECORDS}->[2] &&
 367         !defined $results_hashref->{biblioserver}->{RECORDS}->[3] &&
 368         !defined $results_hashref->{biblioserver}->{RECORDS}->[4] &&
 369         $results_hashref->{biblioserver}->{RECORDS}->[5] eq $record5, "getRecords cursor works");
 370
 371     ( undef, $results_hashref, $facets_loop ) =
 372         getRecords('ti:book', 'ti:book', [], [ 'biblioserver' ], '20', 0, undef, \%branches, \%itemtypes, 'ccl', undef);
 373     is($results_hashref->{biblioserver}->{hits}, 11, "getRecords title search for 'book' matched right number of records");
 374
 375     ( undef, $results_hashref, $facets_loop ) =
 376         getRecords('au:Lessig', 'au:Lessig', [], [ 'biblioserver' ], '20', 0, undef, \%branches, \%itemtypes, 'ccl', undef);
 377     is($results_hashref->{biblioserver}->{hits}, 4, "getRecords title search for 'Australia' matched right number of records");
 378
 379 if ( $indexing_mode eq 'dom' ) {
 380     ( undef, $results_hashref, $facets_loop ) =
 381         getRecords('salud', 'salud', [], [ 'biblioserver' ], '19', 0, undef, \%branches, \%itemtypes, 'ccl', undef);
 382     ok(MARC::Record::new_from_xml($results_hashref->{biblioserver}->{RECORDS}->[0],'UTF-8')->title_proper() =~ m/^Efectos del ambiente/ &&
 383         MARC::Record::new_from_xml($results_hashref->{biblioserver}->{RECORDS}->[7],'UTF-8')->title_proper() eq 'Salud y seguridad de los trabajadores del sector salud: manual para gerentes y administradores^ies' &&
 384         MARC::Record::new_from_xml($results_hashref->{biblioserver}->{RECORDS}->[18],'UTF-8')->title_proper() =~ m/^Indicadores de resultados identificados/
 385         , "Simple relevance sorting in getRecords matches old behavior");
 386
 387     ( undef, $results_hashref, $facets_loop ) =
 388         getRecords('salud', 'salud', [ 'author_az' ], [ 'biblioserver' ], '38', 0, undef, \%branches, \%itemtypes, 'ccl', undef);
 389     ok(MARC::Record::new_from_xml($results_hashref->{biblioserver}->{RECORDS}->[0],'UTF-8')->title_proper() =~ m/la enfermedad laboral\^ies$/ &&
 390         MARC::Record::new_from_xml($results_hashref->{biblioserver}->{RECORDS}->[6],'UTF-8')->title_proper() =~ m/^Indicadores de resultados identificados/ &&
 391         MARC::Record::new_from_xml($results_hashref->{biblioserver}->{RECORDS}->[18],'UTF-8')->title_proper() eq 'World health statistics 2009^ien'
 392         , "Simple ascending author sorting in getRecords matches old behavior");
 393
 394     ( undef, $results_hashref, $facets_loop ) =
 395         getRecords('salud', 'salud', [ 'author_za' ], [ 'biblioserver' ], '38', 0, undef, \%branches, \%itemtypes, 'ccl', undef);
 396     ok(MARC::Record::new_from_xml($results_hashref->{biblioserver}->{RECORDS}->[0],'UTF-8')->title_proper() eq 'World health statistics 2009^ien' &&
 397         MARC::Record::new_from_xml($results_hashref->{biblioserver}->{RECORDS}->[12],'UTF-8')->title_proper() =~ m/^Indicadores de resultados identificados/ &&
 398         MARC::Record::new_from_xml($results_hashref->{biblioserver}->{RECORDS}->[18],'UTF-8')->title_proper() =~ m/la enfermedad laboral\^ies$/
 399         , "Simple descending author sorting in getRecords matches old behavior");
 400
 401     ( undef, $results_hashref, $facets_loop ) =
 402         getRecords('salud', 'salud', [ 'pubdate_asc' ], [ 'biblioserver' ], '38', 0, undef, \%branches, \%itemtypes, 'ccl', undef);
 403     ok(MARC::Record::new_from_xml($results_hashref->{biblioserver}->{RECORDS}->[0],'UTF-8')->title_proper() eq 'Manual de higiene industrial^ies' &&
 404         MARC::Record::new_from_xml($results_hashref->{biblioserver}->{RECORDS}->[7],'UTF-8')->title_proper() =~ m/seguridad e higiene del trabajo\^ies$/ &&
 405         MARC::Record::new_from_xml($results_hashref->{biblioserver}->{RECORDS}->[18],'UTF-8')->title_proper() =~ m/^Indicadores de resultados identificados/
 406         , "Simple ascending publication date sorting in getRecords matches old behavior");
 407
 408     ( undef, $results_hashref, $facets_loop ) =
 409         getRecords('salud', 'salud', [ 'pubdate_dsc' ], [ 'biblioserver' ], '38', 0, undef, \%branches, \%itemtypes, 'ccl', undef);
 410     ok(MARC::Record::new_from_xml($results_hashref->{biblioserver}->{RECORDS}->[0],'UTF-8')->title_proper() =~ m/^Estado de salud/ &&
 411         MARC::Record::new_from_xml($results_hashref->{biblioserver}->{RECORDS}->[7],'UTF-8')->title_proper() eq 'World health statistics 2009^ien' &&
 412         MARC::Record::new_from_xml($results_hashref->{biblioserver}->{RECORDS}->[18],'UTF-8')->title_proper() eq 'Manual de higiene industrial^ies'
 413         , "Simple descending publication date sorting in getRecords matches old behavior");
 414
 415 } elsif ( $indexing_mode eq 'grs1' ){
 416     ( undef, $results_hashref, $facets_loop ) =
 417         getRecords('salud', 'salud', [], [ 'biblioserver' ], '19', 0, undef, \%branches, \%itemtypes, 'ccl', undef);
 418     ok(MARC::Record::new_from_usmarc($results_hashref->{biblioserver}->{RECORDS}->[0])->title_proper() =~ m/^Efectos del ambiente/ &&
 419         MARC::Record::new_from_usmarc($results_hashref->{biblioserver}->{RECORDS}->[7])->title_proper() eq 'Salud y seguridad de los trabajadores del sector salud: manual para gerentes y administradores^ies' &&
 420         MARC::Record::new_from_usmarc($results_hashref->{biblioserver}->{RECORDS}->[18])->title_proper() =~ m/^Indicadores de resultados identificados/
 421         , "Simple relevance sorting in getRecords matches old behavior");
 422
 423     ( undef, $results_hashref, $facets_loop ) =
 424         getRecords('salud', 'salud', [ 'author_az' ], [ 'biblioserver' ], '38', 0, undef, \%branches, \%itemtypes, 'ccl', undef);
 425     ok(MARC::Record::new_from_usmarc($results_hashref->{biblioserver}->{RECORDS}->[0])->title_proper() =~ m/la enfermedad laboral\^ies$/ &&
 426         MARC::Record::new_from_usmarc($results_hashref->{biblioserver}->{RECORDS}->[6])->title_proper() =~ m/^Indicadores de resultados identificados/ &&
 427         MARC::Record::new_from_usmarc($results_hashref->{biblioserver}->{RECORDS}->[18])->title_proper() eq 'World health statistics 2009^ien'
 428         , "Simple ascending author sorting in getRecords matches old behavior");
 429
 430     ( undef, $results_hashref, $facets_loop ) =
 431         getRecords('salud', 'salud', [ 'author_za' ], [ 'biblioserver' ], '38', 0, undef, \%branches, \%itemtypes, 'ccl', undef);
 432     ok(MARC::Record::new_from_usmarc($results_hashref->{biblioserver}->{RECORDS}->[0])->title_proper() eq 'World health statistics 2009^ien' &&
 433         MARC::Record::new_from_usmarc($results_hashref->{biblioserver}->{RECORDS}->[12])->title_proper() =~ m/^Indicadores de resultados identificados/ &&
 434         MARC::Record::new_from_usmarc($results_hashref->{biblioserver}->{RECORDS}->[18])->title_proper() =~ m/la enfermedad laboral\^ies$/
 435         , "Simple descending author sorting in getRecords matches old behavior");
 436
 437     ( undef, $results_hashref, $facets_loop ) =
 438         getRecords('salud', 'salud', [ 'pubdate_asc' ], [ 'biblioserver' ], '38', 0, undef, \%branches, \%itemtypes, 'ccl', undef);
 439     ok(MARC::Record::new_from_usmarc($results_hashref->{biblioserver}->{RECORDS}->[0])->title_proper() eq 'Manual de higiene industrial^ies' &&
 440         MARC::Record::new_from_usmarc($results_hashref->{biblioserver}->{RECORDS}->[7])->title_proper() =~ m/seguridad e higiene del trabajo\^ies$/ &&
 441         MARC::Record::new_from_usmarc($results_hashref->{biblioserver}->{RECORDS}->[18])->title_proper() =~ m/^Indicadores de resultados identificados/
 442         , "Simple ascending publication date sorting in getRecords matches old behavior");
 443
 444     ( undef, $results_hashref, $facets_loop ) =
 445         getRecords('salud', 'salud', [ 'pubdate_dsc' ], [ 'biblioserver' ], '38', 0, undef, \%branches, \%itemtypes, 'ccl', undef);
 446     ok(MARC::Record::new_from_usmarc($results_hashref->{biblioserver}->{RECORDS}->[0])->title_proper() =~ m/^Estado de salud/ &&
 447         MARC::Record::new_from_usmarc($results_hashref->{biblioserver}->{RECORDS}->[7])->title_proper() eq 'World health statistics 2009^ien' &&
 448         MARC::Record::new_from_usmarc($results_hashref->{biblioserver}->{RECORDS}->[18])->title_proper() eq 'Manual de higiene industrial^ies'
 449         , "Simple descending publication date sorting in getRecords matches old behavior");
 450 }
 451
 452     ( undef, $results_hashref, $facets_loop ) =
 453         getRecords('books', 'books', [ 'relevance' ], [ 'biblioserver' ], '20', 0, undef, \%branches, \%itemtypes, undef, 1);
 454     $record = MARC::Record::new_from_usmarc($results_hashref->{biblioserver}->{RECORDS}->[0]);
 455     is($record->title_proper(), 'Books', "Scan returned requested item");
 456     is($record->subfield('100', 'a'), 2, "Scan returned correct number of records matching term");
 457     # Time to test buildQuery and searchResults too.
 458
 459     my ( $query, $simple_query, $query_cgi,
 460     $query_desc, $limit, $limit_cgi, $limit_desc,
 461     $stopwords_removed, $query_type );
 462     ( $error, $query, $simple_query, $query_cgi,
 463     $query_desc, $limit, $limit_cgi, $limit_desc,
 464     $stopwords_removed, $query_type ) = buildQuery([], [ 'salud' ], [], [], [], 0, 'en');
 465     like($query, qr/kw\W.*salud/, "Built CCL keyword query");
 466
 467     ($error, $results_hashref, $facets_loop) = getRecords($query,$simple_query,[ ], [ 'biblioserver' ],20,0,undef,\%branches,\%itemtypes,$query_type,0);
 468     is($results_hashref->{biblioserver}->{hits}, 19, "getRecords generated keyword search for 'salud' matched right number of records");
 469
 470     my @newresults = searchResults('opac', $query_desc, $results_hashref->{'biblioserver'}->{'hits'}, 18, 0, 0,
 471         $results_hashref->{'biblioserver'}->{"RECORDS"});
 472     is(scalar @newresults,18, "searchResults returns requested number of hits");
 473
 474     ( $error, $query, $simple_query, $query_cgi,
 475     $query_desc, $limit, $limit_cgi, $limit_desc,
 476     $stopwords_removed, $query_type ) = buildQuery([ 'and' ], [ 'salud', 'higiene' ], [], [], [], 0, 'en');
 477     like($query, qr/kw\W.*salud\W.*and.*kw\W.*higiene/, "Built composed explicit-and CCL keyword query");
 478
 479     ($error, $results_hashref, $facets_loop) = getRecords($query,$simple_query,[ ], [ 'biblioserver' ],20,0,undef,\%branches,\%itemtypes,$query_type,0);
 480     is($results_hashref->{biblioserver}->{hits}, 3, "getRecords generated composed keyword search for 'salud' explicit-and 'higiene' matched right number of records");
 481
 482     ( $error, $query, $simple_query, $query_cgi,
 483     $query_desc, $limit, $limit_cgi, $limit_desc,
 484     $stopwords_removed, $query_type ) = buildQuery([ 'or' ], [ 'salud', 'higiene' ], [], [], [], 0, 'en');
 485     like($query, qr/kw\W.*salud\W.*or.*kw\W.*higiene/, "Built composed explicit-or CCL keyword query");
 486
 487     ($error, $results_hashref, $facets_loop) = getRecords($query,$simple_query,[ ], [ 'biblioserver' ],20,0,undef,\%branches,\%itemtypes,$query_type,0);
 488     is($results_hashref->{biblioserver}->{hits}, 20, "getRecords generated composed keyword search for 'salud' explicit-or 'higiene' matched right number of records");
 489
 490     ( $error, $query, $simple_query, $query_cgi,
 491     $query_desc, $limit, $limit_cgi, $limit_desc,
 492     $stopwords_removed, $query_type ) = buildQuery([], [ 'salud', 'higiene' ], [], [], [], 0, 'en');
 493     like($query, qr/kw\W.*salud\W.*and.*kw\W.*higiene/, "Built composed implicit-and CCL keyword query");
 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}, 3, "getRecords generated composed keyword search for 'salud' implicit-and 'higiene' matched right number of records");
 497
 498     ( $error, $query, $simple_query, $query_cgi,
 499     $query_desc, $limit, $limit_cgi, $limit_desc,
 500     $stopwords_removed, $query_type ) = buildQuery([], [ 'salud' ], [ 'kw' ], [ 'su-to:Laboratorios' ], [], 0, 'en');
 501     like($query, qr/kw\W.*salud\W*and\W*su-to\W.*Laboratorios/, "Faceted query generated correctly");
 502     unlike($query_desc, qr/Laboratorios/, "Facets not included in query description");
 503
 504     ($error, $results_hashref, $facets_loop) = getRecords($query,$simple_query,[ ], [ 'biblioserver' ],20,0,undef,\%branches,\%itemtypes,$query_type,0);
 505     is($results_hashref->{biblioserver}->{hits}, 2, "getRecords generated faceted search matched right number of records");
 506
 507
 508     ( $error, $query, $simple_query, $query_cgi,
 509     $query_desc, $limit, $limit_cgi, $limit_desc,
 510     $stopwords_removed, $query_type ) = buildQuery([], [ '' ], [ 'kw' ], [ 'mc-itype:MP', 'mc-itype:MU' ], [], 0, 'en');
 511
 512     ($error, $results_hashref, $facets_loop) = getRecords($query,$simple_query,[ ], [ 'biblioserver' ],20,0,undef,\%branches,\%itemtypes,$query_type,0);
 513     is($results_hashref->{biblioserver}->{hits}, 2, "getRecords generated mc-faceted search matched right number of records");
 514
 515
 516     ( $error, $query, $simple_query, $query_cgi,
 517     $query_desc, $limit, $limit_cgi, $limit_desc,
 518     $stopwords_removed, $query_type ) = buildQuery([], [ '' ], [ 'kw' ], [ 'mc-loc:GEN', 'branch:FFL' ], [], 0, 'en');
 519
 520     ($error, $results_hashref, $facets_loop) = getRecords($query,$simple_query,[ ], [ 'biblioserver' ],20,0,undef,\%branches,\%itemtypes,$query_type,0);
 521     is($results_hashref->{biblioserver}->{hits}, 2, "getRecords generated multi-faceted search matched right number of records");
 522
 523     ( $error, $query, $simple_query, $query_cgi,
 524     $query_desc, $limit, $limit_cgi, $limit_desc,
 525     $stopwords_removed, $query_type ) = buildQuery([], [ 'NEKLS' ], [ 'Code-institution' ], [], [], 0, 'en');
 526     ($error, $results_hashref, $facets_loop) = getRecords($query,$simple_query,[ ], [ 'biblioserver' ],20,0,undef,\%branches,\%itemtypes,$query_type,0);
 527     is($results_hashref->{biblioserver}->{hits}, 12,
 528        'search using index whose name contains "ns" returns expected results (bug 10271)');
 529
 530     $UseQueryParser = 1;
 531     ( $error, $query, $simple_query, $query_cgi,
 532     $query_desc, $limit, $limit_cgi, $limit_desc,
 533     $stopwords_removed, $query_type ) = buildQuery([], [ 'book' ], [ 'kw' ], [], [], 0, 'en');
 534     ($error, $results_hashref, $facets_loop) = getRecords($query,$simple_query,[ ], [ 'biblioserver' ],20,0,undef,\%branches,\%itemtypes,$query_type,0);
 535     is($results_hashref->{biblioserver}->{hits}, 101, "Search for 'book' with index set to 'kw' returns 101 hits");
 536     ( $error, $query, $simple_query, $query_cgi,
 537     $query_desc, $limit, $limit_cgi, $limit_desc,
 538     $stopwords_removed, $query_type ) = buildQuery([ 'and' ], [ 'book', 'another' ], [ 'kw', 'kw' ], [], [], 0, 'en');
 539     ($error, $results_hashref, $facets_loop) = getRecords($query,$simple_query,[ ], [ 'biblioserver' ],20,0,undef,\%branches,\%itemtypes,$query_type,0);
 540     is($results_hashref->{biblioserver}->{hits}, 1, "Search for 'kw:book && kw:another' returns 1 hit");
 541     $UseQueryParser = 0;
 542
 543     # FIXME: the availability limit does not actually work, so for the moment we
 544     # are just checking that it behaves consistently
 545     ( $error, $query, $simple_query, $query_cgi,
 546     $query_desc, $limit, $limit_cgi, $limit_desc,
 547     $stopwords_removed, $query_type ) = buildQuery([], [ '' ], [ 'kw' ], [ 'available' ], [], 0, 'en');
 548
 549     ($error, $results_hashref, $facets_loop) = getRecords($query,$simple_query,[ ], [ 'biblioserver' ],20,0,undef,\%branches,\%itemtypes,$query_type,0);
 550     is($results_hashref->{biblioserver}->{hits}, 26, "getRecords generated availability-limited search matched right number of records");
 551
 552     @newresults = searchResults('opac', $query_desc, $results_hashref->{'biblioserver'}->{'hits'}, 17, 0, 0,
 553         $results_hashref->{'biblioserver'}->{"RECORDS"});
 554     my $allavailable = 'true';
 555     foreach my $result (@newresults) {
 556         $allavailable = 'false' unless $result->{availablecount} > 0;
 557     }
 558     is ($allavailable, 'true', 'All records have at least one item available');
 559
 560
 561     ( $error, $query, $simple_query, $query_cgi,
 562     $query_desc, $limit, $limit_cgi, $limit_desc,
 563     $stopwords_removed, $query_type ) = buildQuery([], [ 'pqf=@attr 1=_ALLRECORDS @attr 2=103 ""' ], [], [], [], 0, 'en');
 564
 565     ($error, $results_hashref, $facets_loop) = getRecords($query,$simple_query,[ ], [ 'biblioserver' ],20,0,undef,\%branches,\%itemtypes,$query_type,0);
 566     is($results_hashref->{biblioserver}->{hits}, 180, "getRecords on _ALLRECORDS PQF returned all records");
 567
 568     ( $error, $query, $simple_query, $query_cgi,
 569     $query_desc, $limit, $limit_cgi, $limit_desc,
 570     $stopwords_removed, $query_type ) = buildQuery([], [ 'pqf=@attr 1=1016 "Lessig"' ], [], [], [], 0, 'en');
 571
 572     ($error, $results_hashref, $facets_loop) = getRecords($query,$simple_query,[ ], [ 'biblioserver' ],20,0,undef,\%branches,\%itemtypes,$query_type,0);
 573     is($results_hashref->{biblioserver}->{hits}, 4, "getRecords PQF author search for Lessig returned proper number of matches");
 574
 575     ( $error, $query, $simple_query, $query_cgi,
 576     $query_desc, $limit, $limit_cgi, $limit_desc,
 577     $stopwords_removed, $query_type ) = buildQuery([], [ 'ccl=au:Lessig' ], [], [], [], 0, 'en');
 578
 579     ($error, $results_hashref, $facets_loop) = getRecords($query,$simple_query,[ ], [ 'biblioserver' ],20,0,undef,\%branches,\%itemtypes,$query_type,0);
 580     is($results_hashref->{biblioserver}->{hits}, 4, "getRecords CCL author search for Lessig returned proper number of matches");
 581
 582     ( $error, $query, $simple_query, $query_cgi,
 583     $query_desc, $limit, $limit_cgi, $limit_desc,
 584     $stopwords_removed, $query_type ) = buildQuery([], [ 'cql=dc.author any lessig' ], [], [], [], 0, 'en');
 585
 586     ($error, $results_hashref, $facets_loop) = getRecords($query,$simple_query,[ ], [ 'biblioserver' ],20,0,undef,\%branches,\%itemtypes,$query_type,0);
 587     is($results_hashref->{biblioserver}->{hits}, 4, "getRecords CQL author search for Lessig returned proper number of matches");
 588
 589     $QueryStemming = $QueryAutoTruncate = $QueryFuzzy = $QueryRemoveStopwords = 0;
 590     $QueryWeightFields = 1;
 591     ( $error, $query, $simple_query, $query_cgi,
 592     $query_desc, $limit, $limit_cgi, $limit_desc,
 593     $stopwords_removed, $query_type ) = buildQuery([], [ 'salud' ], [ 'kw' ], [], [], 0, 'en');
 594
 595     ($error, $results_hashref, $facets_loop) = getRecords($query,$simple_query,[ ], [ 'biblioserver' ],20,0,undef,\%branches,\%itemtypes,$query_type,0);
 596     is($results_hashref->{biblioserver}->{hits}, 19, "Weighted query returned correct number of results");
 597     if ($indexing_mode eq 'grs1') {
 598         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");
 599     } else {
 600         local $TODO = "Query weighting does not behave exactly the same in DOM vs. GRS";
 601         is(MARC::Record::new_from_xml($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");
 602     }
 603
 604     $QueryStemming = $QueryWeightFields = $QueryFuzzy = $QueryRemoveStopwords = 0;
 605     $QueryAutoTruncate = 1;
 606     ( $error, $query, $simple_query, $query_cgi,
 607     $query_desc, $limit, $limit_cgi, $limit_desc,
 608     $stopwords_removed, $query_type ) = buildQuery([], [ 'medic' ], [ 'kw' ], [], [], 0, 'en');
 609
 610     ($error, $results_hashref, $facets_loop) = getRecords($query,$simple_query,[ ], [ 'biblioserver' ],20,0,undef,\%branches,\%itemtypes,$query_type,0);
 611     is($results_hashref->{biblioserver}->{hits}, 5, "Search for 'medic' returns matches  with automatic truncation on");
 612
 613     ( $error, $query, $simple_query, $query_cgi,
 614     $query_desc, $limit, $limit_cgi, $limit_desc,
 615     $stopwords_removed, $query_type ) = buildQuery([], [ 'medic*' ], [ 'kw' ], [], [], 0, 'en');
 616
 617     ($error, $results_hashref, $facets_loop) = getRecords($query,$simple_query,[ ], [ 'biblioserver' ],20,0,undef,\%branches,\%itemtypes,$query_type,0);
 618     is($results_hashref->{biblioserver}->{hits}, 5, "Search for 'medic*' returns matches with automatic truncation on");
 619
 620     $QueryStemming = $QueryFuzzy = $QueryRemoveStopwords = $QueryAutoTruncate = 0;
 621     $QueryWeightFields = 1;
 622     ( $error, $query, $simple_query, $query_cgi,
 623     $query_desc, $limit, $limit_cgi, $limit_desc,
 624     $stopwords_removed, $query_type ) = buildQuery([], [ 'web application' ], [ 'kw' ], [], [], 0, 'en');
 625     ($error, $results_hashref, $facets_loop) = getRecords($query,$simple_query,[ ], [ 'biblioserver' ],20,0,undef,\%branches,\%itemtypes,$query_type,0);
 626     is($results_hashref->{biblioserver}->{hits}, 1, "Search for 'web application' returns one hit with QueryWeightFields on");
 627
 628     ( $error, $query, $simple_query, $query_cgi,
 629     $query_desc, $limit, $limit_cgi, $limit_desc,
 630     $stopwords_removed, $query_type ) = buildQuery([], [ 'web "application' ], [ 'kw' ], [], [], 0, 'en');
 631     ($error, $results_hashref, $facets_loop) = getRecords($query,$simple_query,[ ], [ 'biblioserver' ],20,0,undef,\%branches,\%itemtypes,$query_type,0);
 632     is($results_hashref->{biblioserver}->{hits}, 1, "Search for 'web \"application' returns one hit with QueryWeightFields on (bug 7518)");
 633
 634     $QueryStemming = $QueryWeightFields = $QueryFuzzy = $QueryRemoveStopwords = $QueryAutoTruncate = 0;
 635     ( $error, $query, $simple_query, $query_cgi,
 636     $query_desc, $limit, $limit_cgi, $limit_desc,
 637     $stopwords_removed, $query_type ) = buildQuery([], [ 'medic' ], [ 'kw' ], [], [], 0, 'en');
 638
 639     ($error, $results_hashref, $facets_loop) = getRecords($query,$simple_query,[ ], [ 'biblioserver' ],20,0,undef,\%branches,\%itemtypes,$query_type,0);
 640     is($results_hashref->{biblioserver}->{hits}, undef, "Search for 'medic' returns no matches with automatic truncation off");
 641
 642     ( $error, $query, $simple_query, $query_cgi,
 643     $query_desc, $limit, $limit_cgi, $limit_desc,
 644     $stopwords_removed, $query_type ) = buildQuery([], [ 'medic*' ], [ 'kw' ], [], [], 0, 'en');
 645
 646     ($error, $results_hashref, $facets_loop) = getRecords($query,$simple_query,[ ], [ 'biblioserver' ],20,0,undef,\%branches,\%itemtypes,$query_type,0);
 647     is($results_hashref->{biblioserver}->{hits}, 5, "Search for 'medic*' returns matches with automatic truncation off");
 648
 649     $QueryStemming = $QueryWeightFields = 1;
 650     $QueryFuzzy = $QueryRemoveStopwords = $QueryAutoTruncate = 0;
 651     ( $error, $query, $simple_query, $query_cgi,
 652     $query_desc, $limit, $limit_cgi, $limit_desc,
 653     $stopwords_removed, $query_type ) = buildQuery([], [ 'pressed' ], [ 'kw' ], [], [], 0, 'en');
 654
 655     ($error, $results_hashref, $facets_loop) = getRecords($query,$simple_query,[ ], [ 'biblioserver' ],20,0,undef,\%branches,\%itemtypes,$query_type,0);
 656     is($results_hashref->{biblioserver}->{hits}, 7, "Search for 'pressed' returns matches when stemming (and query weighting) is on");
 657
 658     $QueryStemming = $QueryWeightFields = $QueryFuzzy = $QueryRemoveStopwords = $QueryAutoTruncate = 0;
 659     ( $error, $query, $simple_query, $query_cgi,
 660     $query_desc, $limit, $limit_cgi, $limit_desc,
 661     $stopwords_removed, $query_type ) = buildQuery([], [ 'pressed' ], [ 'kw' ], [], [], 0, 'en');
 662
 663     ($error, $results_hashref, $facets_loop) = getRecords($query,$simple_query,[ ], [ 'biblioserver' ],20,0,undef,\%branches,\%itemtypes,$query_type,0);
 664     is($results_hashref->{biblioserver}->{hits}, undef, "Search for 'pressed' returns no matches when stemming is off");
 665
 666     # Let's see what happens when we pass bad data into these routines.
 667     # We have to catch warnings since we're not very good about returning errors.
 668
 669     warning_like { ( $error, $marcresults, $total_hits ) = SimpleSearch("@==ccl blah", 0, 9) } qr/CCL parsing error/,
 670         "SimpleSearch warns about CCL parsing error with nonsense query";
 671     isnt($error, undef, "SimpleSearch returns an error when passed gibberish");
 672
 673     warning_like {( undef, $results_hashref, $facets_loop ) =
 674         getRecords('kw:book', 'book', [], [ 'biblioserver' ], '19', 0, undef, \%branches, \%itemtypes, 'nonsense', undef) }
 675         qr/Unknown query_type/, "getRecords warns about unknown query type";
 676
 677     warning_like {( undef, $results_hashref, $facets_loop ) =
 678         getRecords('pqf=@attr 1=4 "title"', 'pqf=@attr 1=4 "title"', [], [ 'biblioserver' ], '19', 0, undef, \%branches, \%itemtypes, '', undef) }
 679         qr/WARNING: query problem/, "getRecords warns when query type is not specified for non-CCL query";
 680
 681     # Let's just test a few other bits and bobs, just for fun
 682
 683     ($error, $results_hashref, $facets_loop) = getRecords("Godzina pąsowej róży","Godzina pąsowej róży",[ ], [ 'biblioserver' ],20,0,undef,\%branches,\%itemtypes,$query_type,0);
 684     @newresults = searchResults('intranet', $query_desc, $results_hashref->{'biblioserver'}->{'hits'}, 17, 0, 0,
 685         $results_hashref->{'biblioserver'}->{"RECORDS"});
 686     is($newresults[0]->{'alternateholdings_count'}, 1, 'Alternate holdings filled in correctly');
 687
 688
 689     ## Regression test for Bug 10741
 690
 691     # make one of the test items appear to be in transit
 692     my $circ_module = new Test::MockModule('C4::Circulation');
 693     $circ_module->mock('GetTransfers', sub {
 694         my $itemnumber = shift // -1;
 695         if ($itemnumber == 11) {
 696             return ('2013-07-19', 'MPL', 'CPL');
 697         } else {
 698             return;
 699         }
 700     });
 701
 702     ($error, $results_hashref, $facets_loop) = getRecords("TEST12121212","TEST12121212",[ ], [ 'biblioserver' ],20,0,undef,\%branches,\%itemtypes,$query_type,0);
 703     @newresults = searchResults('intranet', $query_desc, $results_hashref->{'biblioserver'}->{'hits'}, 17, 0, 0,
 704         $results_hashref->{'biblioserver'}->{"RECORDS"});
 705     ok(!exists($newresults[0]->{norequests}), 'presence of a transit does not block hold request action (bug 10741)');
 706
 707     ## Regression test for bug 10684
 708     ( undef, $results_hashref, $facets_loop ) =
 709         getRecords('ti:punctuation', 'punctuation', [], [ 'biblioserver' ], '19', 0, undef, \%branches, \%itemtypes, 'ccl', undef);
 710     is($results_hashref->{biblioserver}->{hits}, 1, "search for ti:punctuation returned expected number of records");
 711     @newresults = searchResults('intranet', $query_desc, $results_hashref->{'biblioserver'}->{'hits'}, 20, 0, 0,
 712         $results_hashref->{'biblioserver'}->{"RECORDS"});
 713     is(scalar(@newresults), 0, 'a record that cannot be parsed by MARC::Record is simply skipped (bug 10684)');
 714
 715     # Testing exploding indexes
 716     my $term;
 717     my $searchmodule = new Test::MockModule('C4::Search');
 718     $searchmodule->mock('SimpleSearch', sub {
 719         my $query = shift;
 720
 721         is($query, "he:$term", "Searching for expected term '$term' for exploding") or return '', [], 0;
 722
 723         my $record = MARC::Record->new;
 724         if ($query =~ m/Arizona/) {
 725             $record->add_fields(
 726                 [ '001', '1234' ],
 727                 [ '151', ' ', ' ', a => 'Arizona' ],
 728                 [ '551', ' ', ' ', a => 'United States', w => 'g' ],
 729                 [ '551', ' ', ' ', a => 'Maricopa County', w => 'h' ],
 730                 [ '551', ' ', ' ', a => 'Navajo County', w => 'h' ],
 731                 [ '551', ' ', ' ', a => 'Pima County', w => 'h' ],
 732                 [ '551', ' ', ' ', a => 'New Mexico' ],
 733                 );
 734         }
 735         return '', [ $record->as_usmarc() ], 1;
 736     });
 737
 738     $UseQueryParser = 1;
 739     $term = 'Arizona';
 740     ( $error, $query, $simple_query, $query_cgi,
 741     $query_desc, $limit, $limit_cgi, $limit_desc,
 742     $stopwords_removed, $query_type ) = buildQuery([], [ $term ], [ 'su-br' ], [  ], [], 0, 'en');
 743     matchesExplodedTerms("Advanced search for broader subjects", $query, 'Arizona', 'United States');
 744
 745     ( $error, $query, $simple_query, $query_cgi,
 746     $query_desc, $limit, $limit_cgi, $limit_desc,
 747     $stopwords_removed, $query_type ) = buildQuery([], [ $term ], [ 'su-na' ], [  ], [], 0, 'en');
 748     matchesExplodedTerms("Advanced search for narrower subjects", $query, 'Arizona', 'Maricopa County', 'Navajo County', 'Pima County');
 749
 750     ( $error, $query, $simple_query, $query_cgi,
 751     $query_desc, $limit, $limit_cgi, $limit_desc,
 752     $stopwords_removed, $query_type ) = buildQuery([], [ $term ], [ 'su-rl' ], [  ], [], 0, 'en');
 753     matchesExplodedTerms("Advanced search for related subjects", $query, 'Arizona', 'United States', 'Maricopa County', 'Navajo County', 'Pima County');
 754
 755     ( $error, $query, $simple_query, $query_cgi,
 756     $query_desc, $limit, $limit_cgi, $limit_desc,
 757     $stopwords_removed, $query_type ) = buildQuery([], [ "$term", 'history' ], [ 'su-rl', 'kw' ], [  ], [], 0, 'en');
 758     matchesExplodedTerms("Advanced search for related subjects and keyword 'history' searches related subjects", $query, 'Arizona', 'United States', 'Maricopa County', 'Navajo County', 'Pima County');
 759     like($query, qr/history/, "Advanced search for related subjects and keyword 'history' searches for 'history'");
 760
 761     ( $error, $query, $simple_query, $query_cgi,
 762     $query_desc, $limit, $limit_cgi, $limit_desc,
 763     $stopwords_removed, $query_type ) = buildQuery([], [ 'history', "$term" ], [ 'kw', 'su-rl' ], [  ], [], 0, 'en');
 764     matchesExplodedTerms("Order of terms doesn't matter for advanced search", $query, 'Arizona', 'United States', 'Maricopa County', 'Navajo County', 'Pima County');
 765     like($query, qr/history/, "Order of terms doesn't matter for advanced search");
 766
 767     ( $error, $query, $simple_query, $query_cgi,
 768     $query_desc, $limit, $limit_cgi, $limit_desc,
 769     $stopwords_removed, $query_type ) = buildQuery([], [ "su-br($term)" ], [  ], [  ], [], 0, 'en');
 770     matchesExplodedTerms("Simple search for broader subjects", $query, 'Arizona', 'United States');
 771
 772     ( $error, $query, $simple_query, $query_cgi,
 773     $query_desc, $limit, $limit_cgi, $limit_desc,
 774     $stopwords_removed, $query_type ) = buildQuery([], [ "su-na($term)" ], [  ], [  ], [], 0, 'en');
 775     matchesExplodedTerms("Simple search for narrower subjects", $query, 'Arizona', 'Maricopa County', 'Navajo County', 'Pima County');
 776
 777     ( $error, $query, $simple_query, $query_cgi,
 778     $query_desc, $limit, $limit_cgi, $limit_desc,
 779     $stopwords_removed, $query_type ) = buildQuery([], [ "su-rl($term)" ], [  ], [  ], [], 0, 'en');
 780     matchesExplodedTerms("Simple search for related subjects", $query, 'Arizona', 'United States', 'Maricopa County', 'Navajo County', 'Pima County');
 781
 782     ( $error, $query, $simple_query, $query_cgi,
 783     $query_desc, $limit, $limit_cgi, $limit_desc,
 784     $stopwords_removed, $query_type ) = buildQuery([], [ "history && su-rl($term)" ], [  ], [  ], [], 0, 'en');
 785     matchesExplodedTerms("Simple search for related subjects and keyword 'history' searches related subjects", $query, 'Arizona', 'United States', 'Maricopa County', 'Navajo County', 'Pima County');
 786     like($query, qr/history/, "Simple search for related subjects and keyword 'history' searches for 'history'");
 787
 788     sub matchesExplodedTerms {
 789         my ($message, $query, @terms) = @_;
 790         my $match = '(' . join ('|', map { " \@attr 1=Subject \@attr 4=1 \"$_\"" } @terms) . "){" . scalar(@terms) . "}";
 791         like($query, qr/$match/, $message);
 792     }
 793
 794     # authority records
 795     use_ok('C4::AuthoritiesMarc');
 796     $UseQueryParser = 0;
 797
 798     my ($auths, $count) = SearchAuthorities(
 799         ['mainentry'], ['and'], [''], ['starts'],
 800         ['shakespeare'], 0, 10, '', '', 1
 801     );
 802     is($count, 1, 'MARC21 authorities: one hit on mainentry starts with "shakespeare"');
 803     ($auths, $count) = SearchAuthorities(
 804         ['mainentry'], ['and'], [''], ['starts'],
 805         ['shakespeare'], 0, 10, '', 'HeadingAsc', 1
 806     );
 807     is($count, 1, 'MARC21 authorities: one hit on mainentry starts with "shakespeare" sorted by heading ascending');
 808     ($auths, $count) = SearchAuthorities(
 809         ['mainentry'], ['and'], [''], ['starts'],
 810         ['shakespeare'], 0, 10, '', 'HeadingDsc', 1
 811     );
 812     is($count, 1, 'MARC21 authorities: one hit on mainentry starts with "shakespeare" sorted by heading descending');
 813     ($auths, $count) = SearchAuthorities(
 814         ['match'], ['and'], [''], ['contains'],
 815         ['沙士北亞威廉姆'], 0, 10, '', '', 1
 816     );
 817     is($count, 1, 'MARC21 authorities: one hit on match contains "沙士北亞威廉姆"');
 818
 819     $UseQueryParser = 1;
 820
 821     ($auths, $count) = SearchAuthorities(
 822         ['mainentry'], ['and'], [''], ['starts'],
 823         ['shakespeare'], 0, 10, '', '', 1
 824     );
 825     is($count, 1, 'MARC21 authorities: one hit on mainentry starts with "shakespeare" (QP)');
 826     ($auths, $count) = SearchAuthorities(
 827         ['mainentry'], ['and'], [''], ['starts'],
 828         ['shakespeare'], 0, 10, '', 'HeadingAsc', 1
 829     );
 830     is($count, 1, 'MARC21 authorities: one hit on mainentry starts with "shakespeare" sorted by heading ascending (QP)');
 831     ($auths, $count) = SearchAuthorities(
 832         ['mainentry'], ['and'], [''], ['starts'],
 833         ['shakespeare'], 0, 10, '', 'HeadingDsc', 1
 834     );
 835     is($count, 1, 'MARC21 authorities: one hit on mainentry starts with "shakespeare" sorted by heading descending (QP)');
 836     ($auths, $count) = SearchAuthorities(
 837         ['match'], ['and'], [''], ['contains'],
 838         ['沙士北亞威廉姆'], 0, 10, '', '', 1
 839     );
 840     is($count, 1, 'MARC21 authorities: one hit on match contains "沙士北亞威廉姆" (QP)');
 841
 842     # retrieve records that are larger than the MARC limit of 99,999 octets
 843     ( undef, $results_hashref, $facets_loop ) =
 844         getRecords('ti:marc the large record', '', [], [ 'biblioserver' ], '20', 0, undef, \%branches, \%itemtypes, 'ccl', undef);
 845     is($results_hashref->{biblioserver}->{hits}, 1, "Can do a search that retrieves an over-large bib record (bug 11096)");
 846     @newresults = searchResults('opac', $query_desc, $results_hashref->{'biblioserver'}->{'hits'}, 10, 0, 0,
 847         $results_hashref->{'biblioserver'}->{"RECORDS"});
 848     is($newresults[0]->{title}, 'Marc the Large Record', 'Able to render the title for over-large bib record (bug 11096)');
 849     is($newresults[0]->{biblionumber}, '300', 'Over-large bib record has the correct biblionumber (bug 11096)');
 850     like($newresults[0]->{notes}, qr/This is large note #550/, 'Able to render the notes field for over-large bib record (bug 11096)');
 851
 852     # notforloancount should be returned as part of searchResults output
 853     ok( defined $newresults[0]->{notforloancount},
 854         '\'notforloancount\' defined in searchResults output (Bug 12419)');
 855     is( $newresults[0]->{notforloancount}, 2,
 856         '\'notforloancount\' == 2 (Bug 12419)');
 857
 858     # verify that we don't attempt to sort if no results were returned
 859     # because of a query error
 860     warning_like {( undef, $results_hashref, $facets_loop ) =
 861         getRecords('ccl=( AND )', '', ['title_az'], [ 'biblioserver' ], '20', 0, undef, \%branches, \%itemtypes, 'ccl', undef)
 862     } qr/WARNING: query problem with/, 'got warning instead of crash when attempting to run invalid query (bug 9578)';
 863
 864     # Test facet calculation
 865     my $facets_counter = {};
 866     my $facets         = C4::Koha::getFacets();
 867     # Create a record with a 100$z field
 868     my $marc_record    = MARC::Record->new;
 869     $marc_record->add_fields(
 870         [ '001', '1234' ],
 871         [ '100', ' ', ' ', a => 'Cohen Arazi, Tomas' ],
 872         [ '100', 'z', ' ', a => 'Tomasito' ],
 873         [ '245', ' ', ' ', a => 'First try' ]
 874     );
 875     C4::Search::_get_facets_data_from_record( $marc_record, $facets, $facets_counter );
 876     is_deeply( { au => { 'Cohen Arazi, Tomas' => 1 } },  $facets_counter,
 877         "_get_facets_data_from_record doesn't count 100\$z (Bug 12788)");
 878     $marc_record    = MARC::Record->new;
 879     $marc_record->add_fields(
 880         [ '001', '1234' ],
 881         [ '100', ' ', ' ', a => 'Cohen Arazi, Tomas' ],
 882         [ '100', 'z', ' ', a => 'Tomasito' ],
 883         [ '245', ' ', ' ', a => 'Second try' ]
 884     );
 885     C4::Search::_get_facets_data_from_record( $marc_record, $facets, $facets_counter );
 886     is_deeply( { au => { 'Cohen Arazi, Tomas' => 2 } },  $facets_counter,
 887         "_get_facets_data_from_record correctly counts author facet twice");
 888
 889     # Test _get_facets_info
 890     my $facets_info = C4::Search::_get_facets_info( $facets );
 891     my $expected_facets_info_marc21 = {
 892                    'au' => { 'expanded'    => undef,
 893                              'label_value' => "Authors" },
 894         'holdingbranch' => { 'expanded'    => undef,
 895                              'label_value' => "HoldingLibrary" },
 896                 'itype' => { 'expanded'    => undef,
 897                              'label_value' => "ItemTypes" },
 898              'location' => { 'expanded'    => undef,
 899                              'label_value' => "Location" },
 900                    'se' => { 'expanded'    => undef,
 901                              'label_value' => "Series" },
 902                'su-geo' => { 'expanded'    => undef,
 903                              'label_value' => "Places" },
 904                 'su-to' => { 'expanded'    => undef,
 905                              'label_value' => "Topics" },
 906                 'su-ut' => { 'expanded'    => undef,
 907                              'label_value' => "Titles" }
 908     };
 909     is_deeply( $facets_info, $expected_facets_info_marc21,
 910         "_get_facets_info returns the correct data");
 911
 912     cleanup();
 913 }
 914
 915 sub run_unimarc_search_tests {
 916     my $indexing_mode = shift;
 917     $datadir = tempdir();
 918     system(dirname(__FILE__) . "/zebra_config.pl $datadir unimarc $indexing_mode");
 919
 920     mock_marcfromkohafield('unimarc');
 921     my $context = new C4::Context("$datadir/etc/koha-conf.xml");
 922     $context->set_context();
 923
 924     use_ok('C4::Search');
 925
 926     # set search syspreferences to a known starting point
 927     $QueryStemming = 0;
 928     $QueryAutoTruncate = 0;
 929     $QueryWeightFields = 0;
 930     $QueryFuzzy = 0;
 931     $QueryRemoveStopwords = 0;
 932     $UseQueryParser = 0;
 933     $marcflavour = 'UNIMARC';
 934
 935     index_sample_records_and_launch_zebra($datadir, $indexing_mode, 'unimarc');
 936
 937     my ( $error, $marcresults, $total_hits ) = SimpleSearch("ti=Järnvägarnas efterfrågan och den svenska industrin", 0, 10);
 938     is($total_hits, 1, 'UNIMARC title search');
 939     ( $error, $marcresults, $total_hits ) = SimpleSearch("ta=u", 0, 10);
 940     is($total_hits, 1, 'UNIMARC target audience = u');
 941     ( $error, $marcresults, $total_hits ) = SimpleSearch("ta=k", 0, 10);
 942     is($total_hits, 4, 'UNIMARC target audience = k');
 943     ( $error, $marcresults, $total_hits ) = SimpleSearch("ta=m", 0, 10);
 944     is($total_hits, 3, 'UNIMARC target audience = m');
 945     ( $error, $marcresults, $total_hits ) = SimpleSearch("item=EXCLU DU PRET", 0, 10);
 946     is($total_hits, 1, 'UNIMARC generic item index (bug 10037)');
 947
 948     # authority records
 949     use_ok('C4::AuthoritiesMarc');
 950     $UseQueryParser = 0;
 951
 952     my ($auths, $count) = SearchAuthorities(
 953         ['mainentry'], ['and'], [''], ['contains'],
 954         ['wil'], 0, 10, '', '', 1
 955     );
 956     is($count, 11, 'UNIMARC authorities: hits on mainentry contains "wil"');
 957     ($auths, $count) = SearchAuthorities(
 958         ['match'], ['and'], [''], ['contains'],
 959         ['wil'], 0, 10, '', '', 1
 960     );
 961     is($count, 11, 'UNIMARC authorities: hits on match contains "wil"');
 962     ($auths, $count) = SearchAuthorities(
 963         ['mainentry'], ['and'], [''], ['contains'],
 964         ['michel'], 0, 20, '', '', 1
 965     );
 966     is($count, 14, 'UNIMARC authorities: hits on mainentry contains "michel"');
 967     ($auths, $count) = SearchAuthorities(
 968         ['mainmainentry'], ['and'], [''], ['exact'],
 969         ['valley'], 0, 20, '', '', 1
 970     );
 971     is($count, 1, 'UNIMARC authorities: hits on mainmainentry = "valley"');
 972     ($auths, $count) = SearchAuthorities(
 973         ['mainmainentry'], ['and'], [''], ['exact'],
 974         ['vall'], 0, 20, '', '', 1
 975     );
 976     is($count, 0, 'UNIMARC authorities: no hits on mainmainentry = "vall"');
 977     ($auths, $count) = SearchAuthorities(
 978         ['Any'], ['and'], [''], ['starts'],
 979         ['jean'], 0, 30, '', '', 1
 980     );
 981     is($count, 24, 'UNIMARC authorities: hits on any starts with "jean"');
 982
 983     # Test _get_facets_info
 984     my $facets      = C4::Koha::getFacets();
 985     my $facets_info = C4::Search::_get_facets_info( $facets );
 986     my $expected_facets_info_unimarc = {
 987                    'au' => { 'expanded'    => undef,
 988                              'label_value' => "Authors" },
 989         'holdingbranch' => { 'expanded'    => undef,
 990                              'label_value' => "HoldingLibrary" },
 991              'location' => { 'expanded'    => undef,
 992                              'label_value' => "Location" },
 993                    'se' => { 'expanded'    => undef,
 994                              'label_value' => "Series" },
 995                'su-geo' => { 'expanded'    => undef,
 996                              'label_value' => "Places" },
 997                 'su-to' => { 'expanded'    => undef,
 998                              'label_value' => "Topics" },
 999                 'su-ut' => { 'expanded'    => undef,
1000                              'label_value' => "Titles" }
1001     };
1002     is_deeply( $facets_info, $expected_facets_info_unimarc,
1003         "_get_facets_info returns the correct data");
1004
1005     cleanup();
1006 }
1007
1008 subtest 'MARC21 + GRS-1' => sub {
1009     plan tests => 109;
1010     run_marc21_search_tests('grs1');
1011 };
1012
1013 subtest 'MARC21 + DOM' => sub {
1014     plan tests => 109;
1015     run_marc21_search_tests('dom');
1016 };
1017
1018 subtest 'UNIMARC + GRS-1' => sub {
1019     plan tests => 14;
1020     run_unimarc_search_tests('grs1');
1021 };
1022
1023 subtest 'UNIMARC + DOM' => sub {
1024     plan tests => 14;
1025     run_unimarc_search_tests('dom');
1026 };
1027
1028 1;