tests/phpunit/includes/search/SearchEngineTest.php

   1 <?php
   2
   3 use Wikimedia\Rdbms\LoadBalancerSingle;
   4
   5 /**
   6  * @group Search
   7  * @group Database
   8  *
   9  * @covers SearchEngine<extended>
  10  * @note Coverage will only ever show one of on of the Search* classes
  11  */
  12 class SearchEngineTest extends MediaWikiLangTestCase {
  13
  14         /**
  15          * @var SearchEngine
  16          */
  17         protected $search;
  18
  19         /**
  20          * Checks for database type & version.
  21          * Will skip current test if DB does not support search.
  22          */
  23         protected function setUp() {
  24                 parent::setUp();
  25
  26                 // Search tests require MySQL or SQLite with FTS
  27                 $dbType = $this->db->getType();
  28                 $dbSupported = ( $dbType === 'mysql' )
  29                         || ( $dbType === 'sqlite' && $this->db->getFulltextSearchModule() == 'FTS3' );
  30
  31                 if ( !$dbSupported ) {
  32                         $this->markTestSkipped( "MySQL or SQLite with FTS3 only" );
  33                 }
  34
  35                 $searchType = SearchEngineFactory::getSearchEngineClass( $this->db );
  36                 $this->setMwGlobals( [
  37                         'wgSearchType' => $searchType,
  38                         'wgCapitalLinks' => true,
  39                         'wgCapitalLinkOverrides' => [
  40                                 NS_CATEGORY => false // for testCompletionSearchMustRespectCapitalLinkOverrides
  41                         ]
  42                 ] );
  43
  44                 $lb = LoadBalancerSingle::newFromConnection( $this->db );
  45                 $this->search = new $searchType( $lb );
  46         }
  47
  48         protected function tearDown() {
  49                 unset( $this->search );
  50
  51                 parent::tearDown();
  52         }
  53
  54         public function addDBDataOnce() {
  55                 if ( !$this->isWikitextNS( NS_MAIN ) ) {
  56                         // @todo cover the case of non-wikitext content in the main namespace
  57                         return;
  58                 }
  59
  60                 // Reset the search type back to default - some extensions may have
  61                 // overridden it.
  62                 $this->setMwGlobals( [
  63                         'wgSearchType' => null,
  64                         'wgCapitalLinks' => true,
  65                         'wgCapitalLinkOverrides' => [
  66                                 NS_CATEGORY => false // for testCompletionSearchMustRespectCapitalLinkOverrides
  67                         ]
  68                 ] );
  69
  70                 $this->insertPage( 'Not_Main_Page', 'This is not a main page' );
  71                 $this->insertPage(
  72                         'Talk:Not_Main_Page',
  73                         'This is not a talk page to the main page, see [[smithee]]'
  74                 );
  75                 $this->insertPage( 'Smithee', 'A smithee is one who smiths. See also [[Alan Smithee]]' );
  76                 $this->insertPage( 'Talk:Smithee', 'This article sucks.' );
  77                 $this->insertPage( 'Unrelated_page', 'Nothing in this page is about the S word.' );
  78                 $this->insertPage( 'Another_page', 'This page also is unrelated.' );
  79                 $this->insertPage( 'Help:Help', 'Help me!' );
  80                 $this->insertPage( 'Thppt', 'Blah blah' );
  81                 $this->insertPage( 'Alan_Smithee', 'yum' );
  82                 $this->insertPage( 'Pages', 'are\'food' );
  83                 $this->insertPage( 'HalfOneUp', 'AZ' );
  84                 $this->insertPage( 'FullOneUp', 'ＡＺ' );
  85                 $this->insertPage( 'HalfTwoLow', 'az' );
  86                 $this->insertPage( 'FullTwoLow', 'ａｚ' );
  87                 $this->insertPage( 'HalfNumbers', '1234567890' );
  88                 $this->insertPage( 'FullNumbers', '１２３４５６７８９０' );
  89                 $this->insertPage( 'DomainName', 'example.com' );
  90                 $this->insertPage( 'DomainName', 'example.com' );
  91                 $this->insertPage( 'Category:search is not Search', '' );
  92                 $this->insertPage( 'Category:Search is not search', '' );
  93         }
  94
  95         protected function fetchIds( $results ) {
  96                 if ( !$this->isWikitextNS( NS_MAIN ) ) {
  97                         $this->markTestIncomplete( __CLASS__ . " does no yet support non-wikitext content "
  98                                 . "in the main namespace" );
  99                 }
 100                 $this->assertTrue( is_object( $results ) );
 101
 102                 $matches = [];
 103                 foreach ( $results as $row ) {
 104                         $matches[] = $row->getTitle()->getPrefixedText();
 105                 }
 106                 $results->free();
 107                 # Search is not guaranteed to return results in a certain order;
 108                 # sort them numerically so we will compare simply that we received
 109                 # the expected matches.
 110                 sort( $matches );
 111
 112                 return $matches;
 113         }
 114
 115         public function testFullWidth() {
 116                 $this->assertEquals(
 117                         [ 'FullOneUp', 'FullTwoLow', 'HalfOneUp', 'HalfTwoLow' ],
 118                         $this->fetchIds( $this->search->searchText( 'AZ' ) ),
 119                         "Search for normalized from Half-width Upper" );
 120                 $this->assertEquals(
 121                         [ 'FullOneUp', 'FullTwoLow', 'HalfOneUp', 'HalfTwoLow' ],
 122                         $this->fetchIds( $this->search->searchText( 'az' ) ),
 123                         "Search for normalized from Half-width Lower" );
 124                 $this->assertEquals(
 125                         [ 'FullOneUp', 'FullTwoLow', 'HalfOneUp', 'HalfTwoLow' ],
 126                         $this->fetchIds( $this->search->searchText( 'ＡＺ' ) ),
 127                         "Search for normalized from Full-width Upper" );
 128                 $this->assertEquals(
 129                         [ 'FullOneUp', 'FullTwoLow', 'HalfOneUp', 'HalfTwoLow' ],
 130                         $this->fetchIds( $this->search->searchText( 'ａｚ' ) ),
 131                         "Search for normalized from Full-width Lower" );
 132         }
 133
 134         public function testTextSearch() {
 135                 $this->assertEquals(
 136                         [ 'Smithee' ],
 137                         $this->fetchIds( $this->search->searchText( 'smithee' ) ),
 138                         "Plain search" );
 139         }
 140
 141         public function testWildcardSearch() {
 142                 $res = $this->search->searchText( 'smith*' );
 143                 $this->assertEquals(
 144                         [ 'Smithee' ],
 145                         $this->fetchIds( $res ),
 146                         "Search with wildcards" );
 147
 148                 $res = $this->search->searchText( 'smithson*' );
 149                 $this->assertEquals(
 150                         [],
 151                         $this->fetchIds( $res ),
 152                         "Search with wildcards must not find unrelated articles" );
 153
 154                 $res = $this->search->searchText( 'smith* smithee' );
 155                 $this->assertEquals(
 156                         [ 'Smithee' ],
 157                         $this->fetchIds( $res ),
 158                         "Search with wildcards can be combined with simple terms" );
 159
 160                 $res = $this->search->searchText( 'smith* "one who smiths"' );
 161                 $this->assertEquals(
 162                         [ 'Smithee' ],
 163                         $this->fetchIds( $res ),
 164                         "Search with wildcards can be combined with phrase search" );
 165         }
 166
 167         public function testPhraseSearch() {
 168                 $res = $this->search->searchText( '"smithee is one who smiths"' );
 169                 $this->assertEquals(
 170                         [ 'Smithee' ],
 171                         $this->fetchIds( $res ),
 172                         "Search a phrase" );
 173
 174                 $res = $this->search->searchText( '"smithee is who smiths"' );
 175                 $this->assertEquals(
 176                         [],
 177                         $this->fetchIds( $res ),
 178                         "Phrase search is not sloppy, search terms must be adjacent" );
 179
 180                 $res = $this->search->searchText( '"is smithee one who smiths"' );
 181                 $this->assertEquals(
 182                         [],
 183                         $this->fetchIds( $res ),
 184                         "Phrase search is ordered" );
 185         }
 186
 187         public function testPhraseSearchHighlight() {
 188                 $phrase = "smithee is one who smiths";
 189                 $res = $this->search->searchText( "\"$phrase\"" );
 190                 $match = $res->getIterator()->current();
 191                 $snippet = "A <span class='searchmatch'>" . $phrase . "</span>";
 192                 $this->assertStringStartsWith( $snippet,
 193                         $match->getTextSnippet( $res->termMatches() ),
 194                         "Highlight a phrase search" );
 195         }
 196
 197         public function testTextPowerSearch() {
 198                 $this->search->setNamespaces( [ 0, 1, 4 ] );
 199                 $this->assertEquals(
 200                         [
 201                                 'Smithee',
 202                                 'Talk:Not Main Page',
 203                         ],
 204                         $this->fetchIds( $this->search->searchText( 'smithee' ) ),
 205                         "Power search" );
 206         }
 207
 208         public function testTitleSearch() {
 209                 $this->assertEquals(
 210                         [
 211                                 'Alan Smithee',
 212                                 'Smithee',
 213                         ],
 214                         $this->fetchIds( $this->search->searchTitle( 'smithee' ) ),
 215                         "Title search" );
 216         }
 217
 218         public function testTextTitlePowerSearch() {
 219                 $this->search->setNamespaces( [ 0, 1, 4 ] );
 220                 $this->assertEquals(
 221                         [
 222                                 'Alan Smithee',
 223                                 'Smithee',
 224                                 'Talk:Smithee',
 225                         ],
 226                         $this->fetchIds( $this->search->searchTitle( 'smithee' ) ),
 227                         "Title power search" );
 228         }
 229
 230         public function provideCompletionSearchMustRespectCapitalLinkOverrides() {
 231                 return [
 232                         'Searching for "smithee" finds Smithee on NS_MAIN' => [
 233                                 'smithee',
 234                                 'Smithee',
 235                                 [ NS_MAIN ],
 236                         ],
 237                         'Searching for "search is" will finds "search is not Search" on NS_CATEGORY' => [
 238                                 'search is',
 239                                 'Category:search is not Search',
 240                                 [ NS_CATEGORY ],
 241                         ],
 242                         'Searching for "Search is" will finds "search is not Search" on NS_CATEGORY' => [
 243                                 'Search is',
 244                                 'Category:Search is not search',
 245                                 [ NS_CATEGORY ],
 246                         ],
 247                 ];
 248         }
 249
 250         /**
 251          * Test that the search query is not munged using wrong CapitalLinks setup
 252          * (in other test that the default search backend can benefit from wgCapitalLinksOverride)
 253          * Guard against regressions like T208255
 254          * @dataProvider provideCompletionSearchMustRespectCapitalLinkOverrides
 255          * @covers SearchEngine::completionSearch
 256          * @covers PrefixSearch::defaultSearchBackend
 257          * @param string $search
 258          * @param string $expectedSuggestion
 259          * @param int[] $namespaces
 260          */
 261         public function testCompletionSearchMustRespectCapitalLinkOverrides(
 262                 $search,
 263                 $expectedSuggestion,
 264                 array $namespaces
 265         ) {
 266                 $this->search->setNamespaces( $namespaces );
 267                 $results = $this->search->completionSearch( $search );
 268                 $this->assertEquals( 1, $results->getSize() );
 269                 $this->assertEquals( $expectedSuggestion, $results->getSuggestions()[0]->getText() );
 270         }
 271
 272         /**
 273          * @covers SearchEngine::getSearchIndexFields
 274          */
 275         public function testSearchIndexFields() {
 276                 /**
 277                  * @var SearchEngine $mockEngine
 278                  */
 279                 $mockEngine = $this->getMockBuilder( SearchEngine::class )
 280                         ->setMethods( [ 'makeSearchFieldMapping' ] )->getMock();
 281
 282                 $mockFieldBuilder = function ( $name, $type ) {
 283                         $mockField =
 284                                 $this->getMockBuilder( SearchIndexFieldDefinition::class )->setConstructorArgs( [
 285                                         $name,
 286                                         $type
 287                                 ] )->getMock();
 288
 289                         $mockField->expects( $this->any() )->method( 'getMapping' )->willReturn( [
 290                                 'testData' => 'test',
 291                                 'name' => $name,
 292                                 'type' => $type,
 293                         ] );
 294
 295                         $mockField->expects( $this->any() )
 296                                 ->method( 'merge' )
 297                                 ->willReturn( $mockField );
 298
 299                         return $mockField;
 300                 };
 301
 302                 $mockEngine->expects( $this->atLeastOnce() )
 303                         ->method( 'makeSearchFieldMapping' )
 304                         ->willReturnCallback( $mockFieldBuilder );
 305
 306                 // Not using mock since PHPUnit mocks do not work properly with references in params
 307                 $this->setTemporaryHook( 'SearchIndexFields',
 308                         function ( &$fields, SearchEngine $engine ) use ( $mockFieldBuilder ) {
 309                                 $fields['testField'] =
 310                                         $mockFieldBuilder( "testField", SearchIndexField::INDEX_TYPE_TEXT );
 311                                 return true;
 312                         } );
 313
 314                 $fields = $mockEngine->getSearchIndexFields();
 315                 $this->assertArrayHasKey( 'language', $fields );
 316                 $this->assertArrayHasKey( 'category', $fields );
 317                 $this->assertInstanceOf( SearchIndexField::class, $fields['testField'] );
 318
 319                 $mapping = $fields['testField']->getMapping( $mockEngine );
 320                 $this->assertArrayHasKey( 'testData', $mapping );
 321                 $this->assertEquals( 'test', $mapping['testData'] );
 322         }
 323
 324         public function hookSearchIndexFields( $mockFieldBuilder, &$fields, SearchEngine $engine ) {
 325                 $fields['testField'] = $mockFieldBuilder( "testField", SearchIndexField::INDEX_TYPE_TEXT );
 326                 return true;
 327         }
 328
 329         public function testAugmentorSearch() {
 330                 $this->search->setNamespaces( [ 0, 1, 4 ] );
 331                 $resultSet = $this->search->searchText( 'smithee' );
 332                 // Not using mock since PHPUnit mocks do not work properly with references in params
 333                 $this->mergeMwGlobalArrayValue( 'wgHooks',
 334                         [ 'SearchResultsAugment' => [ [ $this, 'addAugmentors' ] ] ] );
 335                 $this->search->augmentSearchResults( $resultSet );
 336                 foreach ( $resultSet as $result ) {
 337                         $id = $result->getTitle()->getArticleID();
 338                         $augmentData = "Result:$id:" . $result->getTitle()->getText();
 339                         $augmentData2 = "Result2:$id:" . $result->getTitle()->getText();
 340                         $this->assertEquals( [ 'testSet' => $augmentData, 'testRow' => $augmentData2 ],
 341                                 $result->getExtensionData() );
 342                 }
 343         }
 344
 345         public function addAugmentors( &$setAugmentors, &$rowAugmentors ) {
 346                 $setAugmentor = $this->createMock( ResultSetAugmentor::class );
 347                 $setAugmentor->expects( $this->once() )
 348                         ->method( 'augmentAll' )
 349                         ->willReturnCallback( function ( SearchResultSet $resultSet ) {
 350                                 $data = [];
 351                                 foreach ( $resultSet as $result ) {
 352                                         $id = $result->getTitle()->getArticleID();
 353                                         $data[$id] = "Result:$id:" . $result->getTitle()->getText();
 354                                 }
 355                                 return $data;
 356                         } );
 357                 $setAugmentors['testSet'] = $setAugmentor;
 358
 359                 $rowAugmentor = $this->createMock( ResultAugmentor::class );
 360                 $rowAugmentor->expects( $this->exactly( 2 ) )
 361                         ->method( 'augment' )
 362                         ->willReturnCallback( function ( SearchResult $result ) {
 363                                 $id = $result->getTitle()->getArticleID();
 364                                 return "Result2:$id:" . $result->getTitle()->getText();
 365                         } );
 366                 $rowAugmentors['testRow'] = $rowAugmentor;
 367         }
 368
 369         public function testFiltersMissing() {
 370                 $availableResults = [];
 371                 foreach ( range( 0, 11 ) as $i ) {
 372                         $title = "Search_Result_$i";
 373                         $availableResults[] = $title;
 374                         // pages not created must be filtered
 375                         if ( $i % 2 == 0 ) {
 376                                 $this->editSearchResultPage( $title );
 377                         }
 378                 }
 379                 MockCompletionSearchEngine::addMockResults( 'foo', $availableResults );
 380
 381                 $engine = new MockCompletionSearchEngine();
 382                 $engine->setLimitOffset( 10, 0 );
 383                 $results = $engine->completionSearch( 'foo' );
 384                 $this->assertEquals( 5, $results->getSize() );
 385                 $this->assertTrue( $results->hasMoreResults() );
 386
 387                 $engine->setLimitOffset( 10, 10 );
 388                 $results = $engine->completionSearch( 'foo' );
 389                 $this->assertEquals( 1, $results->getSize() );
 390                 $this->assertFalse( $results->hasMoreResults() );
 391         }
 392
 393         private function editSearchResultPage( $title ) {
 394                 $page = WikiPage::factory( Title::newFromText( $title ) );
 395                 $page->doEditContent(
 396                         new WikitextContent( 'UTContent' ),
 397                         'UTPageSummary',
 398                         EDIT_NEW | EDIT_SUPPRESS_RC
 399                 );
 400         }
 401
 402         public function provideDataForParseNamespacePrefix() {
 403                 return [
 404                         'noop' => [
 405                                 [
 406                                         'query' => 'foo',
 407                                 ],
 408                                 false
 409                         ],
 410                         'empty' => [
 411                                 [
 412                                         'query' => '',
 413                                 ],
 414                                 false,
 415                         ],
 416                         'namespace prefix' => [
 417                                 [
 418                                         'query' => 'help:test',
 419                                 ],
 420                                 [ 'test', [ NS_HELP ] ],
 421                         ],
 422                         'accented namespace prefix with hook' => [
 423                                 [
 424                                         'query' => 'hélp:test',
 425                                         'withHook' => true,
 426                                 ],
 427                                 [ 'test', [ NS_HELP ] ],
 428                         ],
 429                         'accented namespace prefix without hook' => [
 430                                 [
 431                                         'query' => 'hélp:test',
 432                                         'withHook' => false,
 433                                 ],
 434                                 false,
 435                         ],
 436                         'all with all keyword allowed' => [
 437                                 [
 438                                         'query' => 'all:test',
 439                                         'withAll' => true,
 440                                 ],
 441                                 [ 'test', null ],
 442                         ],
 443                         'all with all keyword disallowed' => [
 444                                 [
 445                                         'query' => 'all:test',
 446                                         'withAll' => false,
 447                                 ],
 448                                 false
 449                         ],
 450                         'ns only' => [
 451                                 [
 452                                         'query' => 'help:',
 453                                 ],
 454                                 [ '', [ NS_HELP ] ]
 455                         ],
 456                         'all only' => [
 457                                 [
 458                                         'query' => 'all:',
 459                                         'withAll' => true,
 460                                 ],
 461                                 [ '', null ]
 462                         ],
 463                         'all wins over namespace when first' => [
 464                                 [
 465                                         'query' => 'all:help:test',
 466                                         'withAll' => true,
 467                                 ],
 468                                 [ 'help:test', null ]
 469                         ],
 470                         'ns wins over all when first' => [
 471                                 [
 472                                         'query' => 'help:all:test',
 473                                         'withAll' => true,
 474                                 ],
 475                                 [ 'all:test', [ NS_HELP ] ]
 476                         ],
 477                 ];
 478         }
 479
 480         /**
 481          * @dataProvider provideDataForParseNamespacePrefix
 482          * @param array $params
 483          * @param  array|false $expected
 484          * @throws FatalError
 485          * @throws MWException
 486          */
 487         public function testParseNamespacePrefix( array $params, $expected ) {
 488                 $this->setTemporaryHook( 'PrefixSearchExtractNamespace', function ( &$namespaces, &$query ) {
 489                         if ( strpos( $query, 'hélp:' ) === 0 ) {
 490                                 $namespaces = [ NS_HELP ];
 491                                 $query = substr( $query, strlen( 'hélp:' ) );
 492                         }
 493                         return false;
 494                 } );
 495                 $testSet = [];
 496                 if ( isset( $params['withAll'] ) && isset( $params['withHook'] ) ) {
 497                         $testSet[] = $params;
 498                 } elseif ( isset( $params['withAll'] ) ) {
 499                         $testSet[] = $params + [ 'withHook' => true ];
 500                         $testSet[] = $params + [ 'withHook' => false ];
 501                 } elseif ( isset( $params['withHook'] ) ) {
 502                         $testSet[] = $params + [ 'withAll' => true ];
 503                         $testSet[] = $params + [ 'withAll' => false ];
 504                 } else {
 505                         $testSet[] = $params + [ 'withAll' => true, 'withHook' => true ];
 506                         $testSet[] = $params + [ 'withAll' => true, 'withHook' => false ];
 507                         $testSet[] = $params + [ 'withAll' => false, 'withHook' => false ];
 508                         $testSet[] = $params + [ 'withAll' => true, 'withHook' => false ];
 509                 }
 510
 511                 foreach ( $testSet as $test ) {
 512                         $actual = SearchEngine::parseNamespacePrefixes( $test['query'],
 513                                 $test['withAll'], $test['withHook'] );
 514                         $this->assertEquals( $expected, $actual, 'with params: ' . print_r( $test, true ) );
 515                 }
 516         }
 517 }