includes/search/SearchEngine.php

   1 <?php
   2 /**
   3  * Basic search engine
   4  *
   5  * This program is free software; you can redistribute it and/or modify
   6  * it under the terms of the GNU General Public License as published by
   7  * the Free Software Foundation; either version 2 of the License, or
   8  * (at your option) any later version.
   9  *
  10  * This program is distributed in the hope that it will be useful,
  11  * but 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 along
  16  * with this program; if not, write to the Free Software Foundation, Inc.,
  17  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  18  * http://www.gnu.org/copyleft/gpl.html
  19  *
  20  * @file
  21  * @ingroup Search
  22  */
  23
  24 /**
  25  * @defgroup Search Search
  26  */
  27
  28 use MediaWiki\MediaWikiServices;
  29
  30 /**
  31  * Contain a class for special pages
  32  * @ingroup Search
  33  */
  34 abstract class SearchEngine {
  35         const DEFAULT_SORT = 'relevance';
  36
  37         /** @var string */
  38         public $prefix = '';
  39
  40         /** @var int[]|null */
  41         public $namespaces = [ NS_MAIN ];
  42
  43         /** @var int */
  44         protected $limit = 10;
  45
  46         /** @var int */
  47         protected $offset = 0;
  48
  49         /** @var array|string */
  50         protected $searchTerms = [];
  51
  52         /** @var bool */
  53         protected $showSuggestion = true;
  54         private $sort = self::DEFAULT_SORT;
  55
  56         /** @var array Feature values */
  57         protected $features = [];
  58
  59         /** @const string profile type for completionSearch */
  60         const COMPLETION_PROFILE_TYPE = 'completionSearchProfile';
  61
  62         /** @const string profile type for query independent ranking features */
  63         const FT_QUERY_INDEP_PROFILE_TYPE = 'fulltextQueryIndepProfile';
  64
  65         /** @const int flag for legalSearchChars: includes all chars allowed in a search query */
  66         const CHARS_ALL = 1;
  67
  68         /** @const int flag for legalSearchChars: includes all chars allowed in a search term */
  69         const CHARS_NO_SYNTAX = 2;
  70
  71         /**
  72          * Perform a full text search query and return a result set.
  73          * If full text searches are not supported or disabled, return null.
  74          *
  75          * As of 1.32 overriding this function is deprecated. It will
  76          * be converted to final in 1.34. Override self::doSearchText().
  77          *
  78          * @param string $term Raw search term
  79          * @return SearchResultSet|Status|null
  80          */
  81         public function searchText( $term ) {
  82                 return $this->maybePaginate( function () use ( $term ) {
  83                         return $this->doSearchText( $term );
  84                 } );
  85         }
  86
  87         /**
  88          * Perform a full text search query and return a result set.
  89          *
  90          * @param string $term Raw search term
  91          * @return SearchResultSet|Status|null
  92          * @since 1.32
  93          */
  94         protected function doSearchText( $term ) {
  95                 return null;
  96         }
  97
  98         /**
  99          * Perform a title search in the article archive.
 100          * NOTE: these results still should be filtered by
 101          * matching against PageArchive, permissions checks etc
 102          * The results returned by this methods are only sugegstions and
 103          * may not end up being shown to the user.
 104          *
 105          * As of 1.32 overriding this function is deprecated. It will
 106          * be converted to final in 1.34. Override self::doSearchArchiveTitle().
 107          *
 108          * @param string $term Raw search term
 109          * @return Status<Title[]>
 110          * @since 1.29
 111          */
 112         public function searchArchiveTitle( $term ) {
 113                 return $this->doSearchArchiveTitle( $term );
 114         }
 115
 116         /**
 117          * Perform a title search in the article archive.
 118          *
 119          * @param string $term Raw search term
 120          * @return Status<Title[]>
 121          * @since 1.32
 122          */
 123         protected function doSearchArchiveTitle( $term ) {
 124                 return Status::newGood( [] );
 125         }
 126
 127         /**
 128          * Perform a title-only search query and return a result set.
 129          * If title searches are not supported or disabled, return null.
 130          * STUB
 131          *
 132          * As of 1.32 overriding this function is deprecated. It will
 133          * be converted to final in 1.34. Override self::doSearchTitle().
 134          *
 135          * @param string $term Raw search term
 136          * @return SearchResultSet|null
 137          */
 138         public function searchTitle( $term ) {
 139                 return $this->maybePaginate( function () use ( $term ) {
 140                         return $this->doSearchTitle( $term );
 141                 } );
 142         }
 143
 144         /**
 145          * Perform a title-only search query and return a result set.
 146          *
 147          * @param string $term Raw search term
 148          * @return SearchResultSet|null
 149          * @since 1.32
 150          */
 151         protected function doSearchTitle( $term ) {
 152                 return null;
 153         }
 154
 155         /**
 156          * Performs an overfetch and shrink operation to determine if
 157          * the next page is available for search engines that do not
 158          * explicitly implement their own pagination.
 159          *
 160          * @param Closure $fn Takes no arguments
 161          * @return SearchResultSet|Status<SearchResultSet>|null Result of calling $fn
 162          */
 163         private function maybePaginate( Closure $fn ) {
 164                 if ( $this instanceof PaginatingSearchEngine ) {
 165                         return $fn();
 166                 }
 167                 $this->limit++;
 168                 try {
 169                         $resultSetOrStatus = $fn();
 170                 } finally {
 171                         $this->limit--;
 172                 }
 173
 174                 $resultSet = null;
 175                 if ( $resultSetOrStatus instanceof SearchResultSet ) {
 176                         $resultSet = $resultSetOrStatus;
 177                 } elseif ( $resultSetOrStatus instanceof Status &&
 178                         $resultSetOrStatus->getValue() instanceof SearchResultSet
 179                 ) {
 180                         $resultSet = $resultSetOrStatus->getValue();
 181                 }
 182                 if ( $resultSet ) {
 183                         $resultSet->shrink( $this->limit );
 184                 }
 185
 186                 return $resultSetOrStatus;
 187         }
 188
 189         /**
 190          * @since 1.18
 191          * @param string $feature
 192          * @return bool
 193          */
 194         public function supports( $feature ) {
 195                 switch ( $feature ) {
 196                         case 'search-update':
 197                                 return true;
 198                         case 'title-suffix-filter':
 199                         default:
 200                                 return false;
 201                 }
 202         }
 203
 204         /**
 205          * Way to pass custom data for engines
 206          * @since 1.18
 207          * @param string $feature
 208          * @param mixed $data
 209          */
 210         public function setFeatureData( $feature, $data ) {
 211                 $this->features[$feature] = $data;
 212         }
 213
 214         /**
 215          * Way to retrieve custom data set by setFeatureData
 216          * or by the engine itself.
 217          * @since 1.29
 218          * @param string $feature feature name
 219          * @return mixed the feature value or null if unset
 220          */
 221         public function getFeatureData( $feature ) {
 222                 return $this->features[$feature] ?? null;
 223         }
 224
 225         /**
 226          * When overridden in derived class, performs database-specific conversions
 227          * on text to be used for searching or updating search index.
 228          * Default implementation does nothing (simply returns $string).
 229          *
 230          * @param string $string String to process
 231          * @return string
 232          */
 233         public function normalizeText( $string ) {
 234                 // Some languages such as Chinese require word segmentation
 235                 return MediaWikiServices::getInstance()->getContentLanguage()->segmentByWord( $string );
 236         }
 237
 238         /**
 239          * Transform search term in cases when parts of the query came as different
 240          * GET params (when supported), e.g. for prefix queries:
 241          * search=test&prefix=Main_Page/Archive -> test prefix:Main Page/Archive
 242          * @param string $term
 243          * @return string
 244          * @deprecated since 1.32 this should now be handled internally by the
 245          * search engine
 246          */
 247         public function transformSearchTerm( $term ) {
 248                 return $term;
 249         }
 250
 251         /**
 252          * Get service class to finding near matches.
 253          * @param Config $config Configuration to use for the matcher.
 254          * @return SearchNearMatcher
 255          */
 256         public function getNearMatcher( Config $config ) {
 257                 return new SearchNearMatcher( $config,
 258                         MediaWikiServices::getInstance()->getContentLanguage() );
 259         }
 260
 261         /**
 262          * Get near matcher for default SearchEngine.
 263          * @return SearchNearMatcher
 264          */
 265         protected static function defaultNearMatcher() {
 266                 $services = MediaWikiServices::getInstance();
 267                 $config = $services->getMainConfig();
 268                 return $services->newSearchEngine()->getNearMatcher( $config );
 269         }
 270
 271         /**
 272          * If an exact title match can be found, or a very slightly close match,
 273          * return the title. If no match, returns NULL.
 274          * @deprecated since 1.27; Use SearchEngine::getNearMatcher()
 275          * @param string $searchterm
 276          * @return Title
 277          */
 278         public static function getNearMatch( $searchterm ) {
 279                 return static::defaultNearMatcher()->getNearMatch( $searchterm );
 280         }
 281
 282         /**
 283          * Do a near match (see SearchEngine::getNearMatch) and wrap it into a
 284          * SearchResultSet.
 285          * @deprecated since 1.27; Use SearchEngine::getNearMatcher()
 286          * @param string $searchterm
 287          * @return SearchResultSet
 288          */
 289         public static function getNearMatchResultSet( $searchterm ) {
 290                 return static::defaultNearMatcher()->getNearMatchResultSet( $searchterm );
 291         }
 292
 293         /**
 294          * Get chars legal for search
 295          * NOTE: usage as static is deprecated and preserved only as BC measure
 296          * @param int $type type of search chars (see self::CHARS_ALL
 297          * and self::CHARS_NO_SYNTAX). Defaults to CHARS_ALL
 298          * @return string
 299          */
 300         public static function legalSearchChars( $type = self::CHARS_ALL ) {
 301                 return "A-Za-z_'.0-9\\x80-\\xFF\\-";
 302         }
 303
 304         /**
 305          * Set the maximum number of results to return
 306          * and how many to skip before returning the first.
 307          *
 308          * @param int $limit
 309          * @param int $offset
 310          */
 311         function setLimitOffset( $limit, $offset = 0 ) {
 312                 $this->limit = intval( $limit );
 313                 $this->offset = intval( $offset );
 314         }
 315
 316         /**
 317          * Set which namespaces the search should include.
 318          * Give an array of namespace index numbers.
 319          *
 320          * @param int[]|null $namespaces
 321          */
 322         function setNamespaces( $namespaces ) {
 323                 if ( $namespaces ) {
 324                         // Filter namespaces to only keep valid ones
 325                         $validNs = $this->searchableNamespaces();
 326                         $namespaces = array_filter( $namespaces, function ( $ns ) use( $validNs ) {
 327                                 return $ns < 0 || isset( $validNs[$ns] );
 328                         } );
 329                 } else {
 330                         $namespaces = [];
 331                 }
 332                 $this->namespaces = $namespaces;
 333         }
 334
 335         /**
 336          * Set whether the searcher should try to build a suggestion.  Note: some searchers
 337          * don't support building a suggestion in the first place and others don't respect
 338          * this flag.
 339          *
 340          * @param bool $showSuggestion Should the searcher try to build suggestions
 341          */
 342         function setShowSuggestion( $showSuggestion ) {
 343                 $this->showSuggestion = $showSuggestion;
 344         }
 345
 346         /**
 347          * Get the valid sort directions.  All search engines support 'relevance' but others
 348          * might support more. The default in all implementations must be 'relevance.'
 349          *
 350          * @since 1.25
 351          * @return string[] the valid sort directions for setSort
 352          */
 353         public function getValidSorts() {
 354                 return [ self::DEFAULT_SORT ];
 355         }
 356
 357         /**
 358          * Set the sort direction of the search results. Must be one returned by
 359          * SearchEngine::getValidSorts()
 360          *
 361          * @since 1.25
 362          * @throws InvalidArgumentException
 363          * @param string $sort sort direction for query result
 364          */
 365         public function setSort( $sort ) {
 366                 if ( !in_array( $sort, $this->getValidSorts() ) ) {
 367                         throw new InvalidArgumentException( "Invalid sort: $sort. " .
 368                                 "Must be one of: " . implode( ', ', $this->getValidSorts() ) );
 369                 }
 370                 $this->sort = $sort;
 371         }
 372
 373         /**
 374          * Get the sort direction of the search results
 375          *
 376          * @since 1.25
 377          * @return string
 378          */
 379         public function getSort() {
 380                 return $this->sort;
 381         }
 382
 383         /**
 384          * Parse some common prefixes: all (search everything)
 385          * or namespace names and set the list of namespaces
 386          * of this class accordingly.
 387          *
 388          * @deprecated since 1.32; should be handled internally by the search engine
 389          * @param string $query
 390          * @return string
 391          */
 392         function replacePrefixes( $query ) {
 393                 return $query;
 394         }
 395
 396         /**
 397          * Parse some common prefixes: all (search everything)
 398          * or namespace names
 399          *
 400          * @param string $query
 401          * @param bool $withAllKeyword activate support of the "all:" keyword and its
 402          * translations to activate searching on all namespaces.
 403          * @param bool $withPrefixSearchExtractNamespaceHook call the PrefixSearchExtractNamespace hook
 404          *  if classic namespace identification did not match.
 405          * @return false|array false if no namespace was extracted, an array
 406          * with the parsed query at index 0 and an array of namespaces at index
 407          * 1 (or null for all namespaces).
 408          * @throws FatalError
 409          * @throws MWException
 410          */
 411         public static function parseNamespacePrefixes(
 412                 $query,
 413                 $withAllKeyword = true,
 414                 $withPrefixSearchExtractNamespaceHook = false
 415         ) {
 416                 $parsed = $query;
 417                 if ( strpos( $query, ':' ) === false ) { // nothing to do
 418                         return false;
 419                 }
 420                 $extractedNamespace = null;
 421
 422                 $allQuery = false;
 423                 if ( $withAllKeyword ) {
 424                         $allkeywords = [];
 425
 426                         $allkeywords[] = wfMessage( 'searchall' )->inContentLanguage()->text() . ":";
 427                         // force all: so that we have a common syntax for all the wikis
 428                         if ( !in_array( 'all:', $allkeywords ) ) {
 429                                 $allkeywords[] = 'all:';
 430                         }
 431
 432                         foreach ( $allkeywords as $kw ) {
 433                                 if ( strncmp( $query, $kw, strlen( $kw ) ) == 0 ) {
 434                                         $extractedNamespace = null;
 435                                         $parsed = substr( $query, strlen( $kw ) );
 436                                         $allQuery = true;
 437                                         break;
 438                                 }
 439                         }
 440                 }
 441
 442                 if ( !$allQuery && strpos( $query, ':' ) !== false ) {
 443                         $prefix = str_replace( ' ', '_', substr( $query, 0, strpos( $query, ':' ) ) );
 444                         $index = MediaWikiServices::getInstance()->getContentLanguage()->getNsIndex( $prefix );
 445                         if ( $index !== false ) {
 446                                 $extractedNamespace = [ $index ];
 447                                 $parsed = substr( $query, strlen( $prefix ) + 1 );
 448                         } elseif ( $withPrefixSearchExtractNamespaceHook ) {
 449                                 $hookNamespaces = [ NS_MAIN ];
 450                                 $hookQuery = $query;
 451                                 Hooks::run( 'PrefixSearchExtractNamespace', [ &$hookNamespaces, &$hookQuery ] );
 452                                 if ( $hookQuery !== $query ) {
 453                                         $parsed = $hookQuery;
 454                                         $extractedNamespace = $hookNamespaces;
 455                                 } else {
 456                                         return false;
 457                                 }
 458                         } else {
 459                                 return false;
 460                         }
 461                 }
 462
 463                 return [ $parsed, $extractedNamespace ];
 464         }
 465
 466         /**
 467          * Find snippet highlight settings for all users
 468          * @return array Contextlines, contextchars
 469          */
 470         public static function userHighlightPrefs() {
 471                 $contextlines = 2; // Hardcode this. Old defaults sucked. :)
 472                 $contextchars = 75; // same as above.... :P
 473                 return [ $contextlines, $contextchars ];
 474         }
 475
 476         /**
 477          * Create or update the search index record for the given page.
 478          * Title and text should be pre-processed.
 479          * STUB
 480          *
 481          * @param int $id
 482          * @param string $title
 483          * @param string $text
 484          */
 485         function update( $id, $title, $text ) {
 486                 // no-op
 487         }
 488
 489         /**
 490          * Update a search index record's title only.
 491          * Title should be pre-processed.
 492          * STUB
 493          *
 494          * @param int $id
 495          * @param string $title
 496          */
 497         function updateTitle( $id, $title ) {
 498                 // no-op
 499         }
 500
 501         /**
 502          * Delete an indexed page
 503          * Title should be pre-processed.
 504          * STUB
 505          *
 506          * @param int $id Page id that was deleted
 507          * @param string $title Title of page that was deleted
 508          */
 509         function delete( $id, $title ) {
 510                 // no-op
 511         }
 512
 513         /**
 514          * Get the raw text for updating the index from a content object
 515          * Nicer search backends could possibly do something cooler than
 516          * just returning raw text
 517          *
 518          * @todo This isn't ideal, we'd really like to have content-specific handling here
 519          * @param Title $t Title we're indexing
 520          * @param Content|null $c Content of the page to index
 521          * @return string
 522          */
 523         public function getTextFromContent( Title $t, Content $c = null ) {
 524                 return $c ? $c->getTextForSearchIndex() : '';
 525         }
 526
 527         /**
 528          * If an implementation of SearchEngine handles all of its own text processing
 529          * in getTextFromContent() and doesn't require SearchUpdate::updateText()'s
 530          * rather silly handling, it should return true here instead.
 531          *
 532          * @return bool
 533          */
 534         public function textAlreadyUpdatedForIndex() {
 535                 return false;
 536         }
 537
 538         /**
 539          * Makes search simple string if it was namespaced.
 540          * Sets namespaces of the search to namespaces extracted from string.
 541          * @param string $search
 542          * @return string Simplified search string
 543          */
 544         protected function normalizeNamespaces( $search ) {
 545                 $queryAndNs = self::parseNamespacePrefixes( $search, false, true );
 546                 if ( $queryAndNs !== false ) {
 547                         $this->setNamespaces( $queryAndNs[1] );
 548                         return $queryAndNs[0];
 549                 }
 550                 return $search;
 551         }
 552
 553         /**
 554          * Perform an overfetch of completion search results. This allows
 555          * determining if another page of results is available.
 556          *
 557          * @param string $search
 558          * @return SearchSuggestionSet
 559          */
 560         protected function completionSearchBackendOverfetch( $search ) {
 561                 $this->limit++;
 562                 try {
 563                         return $this->completionSearchBackend( $search );
 564                 } finally {
 565                         $this->limit--;
 566                 }
 567         }
 568
 569         /**
 570          * Perform a completion search.
 571          * Does not resolve namespaces and does not check variants.
 572          * Search engine implementations may want to override this function.
 573          * @param string $search
 574          * @return SearchSuggestionSet
 575          */
 576         protected function completionSearchBackend( $search ) {
 577                 $results = [];
 578
 579                 $search = trim( $search );
 580
 581                 if ( !in_array( NS_SPECIAL, $this->namespaces ) && // We do not run hook on Special: search
 582                          !Hooks::run( 'PrefixSearchBackend',
 583                                 [ $this->namespaces, $search, $this->limit, &$results, $this->offset ]
 584                 ) ) {
 585                         // False means hook worked.
 586                         // FIXME: Yes, the API is weird. That's why it is going to be deprecated.
 587
 588                         return SearchSuggestionSet::fromStrings( $results );
 589                 } else {
 590                         // Hook did not do the job, use default simple search
 591                         $results = $this->simplePrefixSearch( $search );
 592                         return SearchSuggestionSet::fromTitles( $results );
 593                 }
 594         }
 595
 596         /**
 597          * Perform a completion search.
 598          * @param string $search
 599          * @return SearchSuggestionSet
 600          */
 601         public function completionSearch( $search ) {
 602                 if ( trim( $search ) === '' ) {
 603                         return SearchSuggestionSet::emptySuggestionSet(); // Return empty result
 604                 }
 605                 $search = $this->normalizeNamespaces( $search );
 606                 $suggestions = $this->completionSearchBackendOverfetch( $search );
 607                 return $this->processCompletionResults( $search, $suggestions );
 608         }
 609
 610         /**
 611          * Perform a completion search with variants.
 612          * @param string $search
 613          * @return SearchSuggestionSet
 614          */
 615         public function completionSearchWithVariants( $search ) {
 616                 if ( trim( $search ) === '' ) {
 617                         return SearchSuggestionSet::emptySuggestionSet(); // Return empty result
 618                 }
 619                 $search = $this->normalizeNamespaces( $search );
 620
 621                 $results = $this->completionSearchBackendOverfetch( $search );
 622                 $fallbackLimit = 1 + $this->limit - $results->getSize();
 623                 if ( $fallbackLimit > 0 ) {
 624                         $fallbackSearches = MediaWikiServices::getInstance()->getContentLanguage()->
 625                                 autoConvertToAllVariants( $search );
 626                         $fallbackSearches = array_diff( array_unique( $fallbackSearches ), [ $search ] );
 627
 628                         foreach ( $fallbackSearches as $fbs ) {
 629                                 $this->setLimitOffset( $fallbackLimit );
 630                                 $fallbackSearchResult = $this->completionSearch( $fbs );
 631                                 $results->appendAll( $fallbackSearchResult );
 632                                 $fallbackLimit -= $fallbackSearchResult->getSize();
 633                                 if ( $fallbackLimit <= 0 ) {
 634                                         break;
 635                                 }
 636                         }
 637                 }
 638                 return $this->processCompletionResults( $search, $results );
 639         }
 640
 641         /**
 642          * Extract titles from completion results
 643          * @param SearchSuggestionSet $completionResults
 644          * @return Title[]
 645          */
 646         public function extractTitles( SearchSuggestionSet $completionResults ) {
 647                 return $completionResults->map( function ( SearchSuggestion $sugg ) {
 648                         return $sugg->getSuggestedTitle();
 649                 } );
 650         }
 651
 652         /**
 653          * Process completion search results.
 654          * Resolves the titles and rescores.
 655          * @param string $search
 656          * @param SearchSuggestionSet $suggestions
 657          * @return SearchSuggestionSet
 658          */
 659         protected function processCompletionResults( $search, SearchSuggestionSet $suggestions ) {
 660                 // We over-fetched to determine pagination. Shrink back down if we have extra results
 661                 // and mark if pagination is possible
 662                 $suggestions->shrink( $this->limit );
 663
 664                 $search = trim( $search );
 665                 // preload the titles with LinkBatch
 666                 $lb = new LinkBatch( $suggestions->map( function ( SearchSuggestion $sugg ) {
 667                         return $sugg->getSuggestedTitle();
 668                 } ) );
 669                 $lb->setCaller( __METHOD__ );
 670                 $lb->execute();
 671
 672                 $diff = $suggestions->filter( function ( SearchSuggestion $sugg ) {
 673                         return $sugg->getSuggestedTitle()->isKnown();
 674                 } );
 675                 if ( $diff > 0 ) {
 676                         MediaWikiServices::getInstance()->getStatsdDataFactory()
 677                                 ->updateCount( 'search.completion.missing', $diff );
 678                 }
 679
 680                 $results = $suggestions->map( function ( SearchSuggestion $sugg ) {
 681                         return $sugg->getSuggestedTitle()->getPrefixedText();
 682                 } );
 683
 684                 if ( $this->offset === 0 ) {
 685                         // Rescore results with an exact title match
 686                         // NOTE: in some cases like cross-namespace redirects
 687                         // (frequently used as shortcuts e.g. WP:WP on huwiki) some
 688                         // backends like Cirrus will return no results. We should still
 689                         // try an exact title match to workaround this limitation
 690                         $rescorer = new SearchExactMatchRescorer();
 691                         $rescoredResults = $rescorer->rescore( $search, $this->namespaces, $results, $this->limit );
 692                 } else {
 693                         // No need to rescore if offset is not 0
 694                         // The exact match must have been returned at position 0
 695                         // if it existed.
 696                         $rescoredResults = $results;
 697                 }
 698
 699                 if ( count( $rescoredResults ) > 0 ) {
 700                         $found = array_search( $rescoredResults[0], $results );
 701                         if ( $found === false ) {
 702                                 // If the first result is not in the previous array it
 703                                 // means that we found a new exact match
 704                                 $exactMatch = SearchSuggestion::fromTitle( 0, Title::newFromText( $rescoredResults[0] ) );
 705                                 $suggestions->prepend( $exactMatch );
 706                                 $suggestions->shrink( $this->limit );
 707                         } else {
 708                                 // if the first result is not the same we need to rescore
 709                                 if ( $found > 0 ) {
 710                                         $suggestions->rescore( $found );
 711                                 }
 712                         }
 713                 }
 714
 715                 return $suggestions;
 716         }
 717
 718         /**
 719          * Simple prefix search for subpages.
 720          * @param string $search
 721          * @return Title[]
 722          */
 723         public function defaultPrefixSearch( $search ) {
 724                 if ( trim( $search ) === '' ) {
 725                         return [];
 726                 }
 727
 728                 $search = $this->normalizeNamespaces( $search );
 729                 return $this->simplePrefixSearch( $search );
 730         }
 731
 732         /**
 733          * Call out to simple search backend.
 734          * Defaults to TitlePrefixSearch.
 735          * @param string $search
 736          * @return Title[]
 737          */
 738         protected function simplePrefixSearch( $search ) {
 739                 // Use default database prefix search
 740                 $backend = new TitlePrefixSearch;
 741                 return $backend->defaultSearchBackend( $this->namespaces, $search, $this->limit, $this->offset );
 742         }
 743
 744         /**
 745          * Make a list of searchable namespaces and their canonical names.
 746          * @deprecated since 1.27; use SearchEngineConfig::searchableNamespaces()
 747          * @return array
 748          */
 749         public static function searchableNamespaces() {
 750                 return MediaWikiServices::getInstance()->getSearchEngineConfig()->searchableNamespaces();
 751         }
 752
 753         /**
 754          * Extract default namespaces to search from the given user's
 755          * settings, returning a list of index numbers.
 756          * @deprecated since 1.27; use SearchEngineConfig::userNamespaces()
 757          * @param user $user
 758          * @return array
 759          */
 760         public static function userNamespaces( $user ) {
 761                 return MediaWikiServices::getInstance()->getSearchEngineConfig()->userNamespaces( $user );
 762         }
 763
 764         /**
 765          * An array of namespaces indexes to be searched by default
 766          * @deprecated since 1.27; use SearchEngineConfig::defaultNamespaces()
 767          * @return array
 768          */
 769         public static function defaultNamespaces() {
 770                 return MediaWikiServices::getInstance()->getSearchEngineConfig()->defaultNamespaces();
 771         }
 772
 773         /**
 774          * Get a list of namespace names useful for showing in tooltips
 775          * and preferences
 776          * @deprecated since 1.27; use SearchEngineConfig::namespacesAsText()
 777          * @param array $namespaces
 778          * @return array
 779          */
 780         public static function namespacesAsText( $namespaces ) {
 781                 return MediaWikiServices::getInstance()->getSearchEngineConfig()->namespacesAsText( $namespaces );
 782         }
 783
 784         /**
 785          * Load up the appropriate search engine class for the currently
 786          * active database backend, and return a configured instance.
 787          * @deprecated since 1.27; Use SearchEngineFactory::create
 788          * @param string $type Type of search backend, if not the default
 789          * @return SearchEngine
 790          */
 791         public static function create( $type = null ) {
 792                 return MediaWikiServices::getInstance()->getSearchEngineFactory()->create( $type );
 793         }
 794
 795         /**
 796          * Return the search engines we support. If only $wgSearchType
 797          * is set, it'll be an array of just that one item.
 798          * @deprecated since 1.27; use SearchEngineConfig::getSearchTypes()
 799          * @return array
 800          */
 801         public static function getSearchTypes() {
 802                 return MediaWikiServices::getInstance()->getSearchEngineConfig()->getSearchTypes();
 803         }
 804
 805         /**
 806          * Get a list of supported profiles.
 807          * Some search engine implementations may expose specific profiles to fine-tune
 808          * its behaviors.
 809          * The profile can be passed as a feature data with setFeatureData( $profileType, $profileName )
 810          * The array returned by this function contains the following keys:
 811          * - name: the profile name to use with setFeatureData
 812          * - desc-message: the i18n description
 813          * - default: set to true if this profile is the default
 814          *
 815          * @since 1.28
 816          * @param string $profileType the type of profiles
 817          * @param User|null $user the user requesting the list of profiles
 818          * @return array|null the list of profiles or null if none available
 819          */
 820         public function getProfiles( $profileType, User $user = null ) {
 821                 return null;
 822         }
 823
 824         /**
 825          * Create a search field definition.
 826          * Specific search engines should override this method to create search fields.
 827          * @param string $name
 828          * @param int $type One of the types in SearchIndexField::INDEX_TYPE_*
 829          * @return SearchIndexField
 830          * @since 1.28
 831          */
 832         public function makeSearchFieldMapping( $name, $type ) {
 833                 return new NullIndexField();
 834         }
 835
 836         /**
 837          * Get fields for search index
 838          * @since 1.28
 839          * @return SearchIndexField[] Index field definitions for all content handlers
 840          */
 841         public function getSearchIndexFields() {
 842                 $models = ContentHandler::getContentModels();
 843                 $fields = [];
 844                 $seenHandlers = new SplObjectStorage();
 845                 foreach ( $models as $model ) {
 846                         try {
 847                                 $handler = ContentHandler::getForModelID( $model );
 848                         }
 849                         catch ( MWUnknownContentModelException $e ) {
 850                                 // If we can find no handler, ignore it
 851                                 continue;
 852                         }
 853                         // Several models can have the same handler, so avoid processing it repeatedly
 854                         if ( $seenHandlers->contains( $handler ) ) {
 855                                 // We already did this one
 856                                 continue;
 857                         }
 858                         $seenHandlers->attach( $handler );
 859                         $handlerFields = $handler->getFieldsForSearchIndex( $this );
 860                         foreach ( $handlerFields as $fieldName => $fieldData ) {
 861                                 if ( empty( $fields[$fieldName] ) ) {
 862                                         $fields[$fieldName] = $fieldData;
 863                                 } else {
 864                                         // TODO: do we allow some clashes with the same type or reject all of them?
 865                                         $mergeDef = $fields[$fieldName]->merge( $fieldData );
 866                                         if ( !$mergeDef ) {
 867                                                 throw new InvalidArgumentException( "Duplicate field $fieldName for model $model" );
 868                                         }
 869                                         $fields[$fieldName] = $mergeDef;
 870                                 }
 871                         }
 872                 }
 873                 // Hook to allow extensions to produce search mapping fields
 874                 Hooks::run( 'SearchIndexFields', [ &$fields, $this ] );
 875                 return $fields;
 876         }
 877
 878         /**
 879          * Augment search results with extra data.
 880          *
 881          * @param SearchResultSet $resultSet
 882          */
 883         public function augmentSearchResults( SearchResultSet $resultSet ) {
 884                 $setAugmentors = [];
 885                 $rowAugmentors = [];
 886                 Hooks::run( "SearchResultsAugment", [ &$setAugmentors, &$rowAugmentors ] );
 887                 if ( !$setAugmentors && !$rowAugmentors ) {
 888                         // We're done here
 889                         return;
 890                 }
 891
 892                 // Convert row augmentors to set augmentor
 893                 foreach ( $rowAugmentors as $name => $row ) {
 894                         if ( isset( $setAugmentors[$name] ) ) {
 895                                 throw new InvalidArgumentException( "Both row and set augmentors are defined for $name" );
 896                         }
 897                         $setAugmentors[$name] = new PerRowAugmentor( $row );
 898                 }
 899
 900                 foreach ( $setAugmentors as $name => $augmentor ) {
 901                         $data = $augmentor->augmentAll( $resultSet );
 902                         if ( $data ) {
 903                                 $resultSet->setAugmentedData( $name, $data );
 904                         }
 905                 }
 906         }
 907 }
 908
 909 /**
 910  * Dummy class to be used when non-supported Database engine is present.
 911  * @todo FIXME: Dummy class should probably try something at least mildly useful,
 912  * such as a LIKE search through titles.
 913  * @ingroup Search
 914  */
 915 class SearchEngineDummy extends SearchEngine {
 916         // no-op
 917 }