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         /** Profile type for completionSearch */
  60         const COMPLETION_PROFILE_TYPE = 'completionSearchProfile';
  61
  62         /** Profile type for query independent ranking features */
  63         const FT_QUERY_INDEP_PROFILE_TYPE = 'fulltextQueryIndepProfile';
  64
  65         /** Integer flag for legalSearchChars: includes all chars allowed in a search query */
  66         const CHARS_ALL = 1;
  67
  68         /** Integer 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                 wfDeprecated( __METHOD__, '1.34' );
 249                 return $term;
 250         }
 251
 252         /**
 253          * Get service class to finding near matches.
 254          * @param Config $config Configuration to use for the matcher.
 255          * @return SearchNearMatcher
 256          */
 257         public function getNearMatcher( Config $config ) {
 258                 return new SearchNearMatcher( $config,
 259                         MediaWikiServices::getInstance()->getContentLanguage() );
 260         }
 261
 262         /**
 263          * Get near matcher for default SearchEngine.
 264          * @return SearchNearMatcher
 265          */
 266         protected static function defaultNearMatcher() {
 267                 $services = MediaWikiServices::getInstance();
 268                 $config = $services->getMainConfig();
 269                 return $services->newSearchEngine()->getNearMatcher( $config );
 270         }
 271
 272         /**
 273          * Get chars legal for search
 274          * @param int $type type of search chars (see self::CHARS_ALL
 275          * and self::CHARS_NO_SYNTAX). Defaults to CHARS_ALL
 276          * @return string
 277          */
 278         public function legalSearchChars( $type = self::CHARS_ALL ) {
 279                 return "A-Za-z_'.0-9\\x80-\\xFF\\-";
 280         }
 281
 282         /**
 283          * Set the maximum number of results to return
 284          * and how many to skip before returning the first.
 285          *
 286          * @param int $limit
 287          * @param int $offset
 288          */
 289         function setLimitOffset( $limit, $offset = 0 ) {
 290                 $this->limit = intval( $limit );
 291                 $this->offset = intval( $offset );
 292         }
 293
 294         /**
 295          * Set which namespaces the search should include.
 296          * Give an array of namespace index numbers.
 297          *
 298          * @param int[]|null $namespaces
 299          */
 300         function setNamespaces( $namespaces ) {
 301                 if ( $namespaces ) {
 302                         // Filter namespaces to only keep valid ones
 303                         $validNs = MediaWikiServices::getInstance()->getSearchEngineConfig()->searchableNamespaces();
 304                         $namespaces = array_filter( $namespaces, function ( $ns ) use( $validNs ) {
 305                                 return $ns < 0 || isset( $validNs[$ns] );
 306                         } );
 307                 } else {
 308                         $namespaces = [];
 309                 }
 310                 $this->namespaces = $namespaces;
 311         }
 312
 313         /**
 314          * Set whether the searcher should try to build a suggestion.  Note: some searchers
 315          * don't support building a suggestion in the first place and others don't respect
 316          * this flag.
 317          *
 318          * @param bool $showSuggestion Should the searcher try to build suggestions
 319          */
 320         function setShowSuggestion( $showSuggestion ) {
 321                 $this->showSuggestion = $showSuggestion;
 322         }
 323
 324         /**
 325          * Get the valid sort directions.  All search engines support 'relevance' but others
 326          * might support more. The default in all implementations must be 'relevance.'
 327          *
 328          * @since 1.25
 329          * @return string[] the valid sort directions for setSort
 330          */
 331         public function getValidSorts() {
 332                 return [ self::DEFAULT_SORT ];
 333         }
 334
 335         /**
 336          * Set the sort direction of the search results. Must be one returned by
 337          * SearchEngine::getValidSorts()
 338          *
 339          * @since 1.25
 340          * @throws InvalidArgumentException
 341          * @param string $sort sort direction for query result
 342          */
 343         public function setSort( $sort ) {
 344                 if ( !in_array( $sort, $this->getValidSorts() ) ) {
 345                         throw new InvalidArgumentException( "Invalid sort: $sort. " .
 346                                 "Must be one of: " . implode( ', ', $this->getValidSorts() ) );
 347                 }
 348                 $this->sort = $sort;
 349         }
 350
 351         /**
 352          * Get the sort direction of the search results
 353          *
 354          * @since 1.25
 355          * @return string
 356          */
 357         public function getSort() {
 358                 return $this->sort;
 359         }
 360
 361         /**
 362          * Parse some common prefixes: all (search everything)
 363          * or namespace names and set the list of namespaces
 364          * of this class accordingly.
 365          *
 366          * @deprecated since 1.32; should be handled internally by the search engine
 367          * @param string $query
 368          * @return string
 369          */
 370         function replacePrefixes( $query ) {
 371                 return $query;
 372         }
 373
 374         /**
 375          * Parse some common prefixes: all (search everything)
 376          * or namespace names
 377          *
 378          * @param string $query
 379          * @param bool $withAllKeyword activate support of the "all:" keyword and its
 380          * translations to activate searching on all namespaces.
 381          * @param bool $withPrefixSearchExtractNamespaceHook call the PrefixSearchExtractNamespace hook
 382          *  if classic namespace identification did not match.
 383          * @return false|array false if no namespace was extracted, an array
 384          * with the parsed query at index 0 and an array of namespaces at index
 385          * 1 (or null for all namespaces).
 386          * @throws FatalError
 387          * @throws MWException
 388          */
 389         public static function parseNamespacePrefixes(
 390                 $query,
 391                 $withAllKeyword = true,
 392                 $withPrefixSearchExtractNamespaceHook = false
 393         ) {
 394                 $parsed = $query;
 395                 if ( strpos( $query, ':' ) === false ) { // nothing to do
 396                         return false;
 397                 }
 398                 $extractedNamespace = null;
 399
 400                 $allQuery = false;
 401                 if ( $withAllKeyword ) {
 402                         $allkeywords = [];
 403
 404                         $allkeywords[] = wfMessage( 'searchall' )->inContentLanguage()->text() . ":";
 405                         // force all: so that we have a common syntax for all the wikis
 406                         if ( !in_array( 'all:', $allkeywords ) ) {
 407                                 $allkeywords[] = 'all:';
 408                         }
 409
 410                         foreach ( $allkeywords as $kw ) {
 411                                 if ( strncmp( $query, $kw, strlen( $kw ) ) == 0 ) {
 412                                         $extractedNamespace = null;
 413                                         $parsed = substr( $query, strlen( $kw ) );
 414                                         $allQuery = true;
 415                                         break;
 416                                 }
 417                         }
 418                 }
 419
 420                 if ( !$allQuery && strpos( $query, ':' ) !== false ) {
 421                         $prefix = str_replace( ' ', '_', substr( $query, 0, strpos( $query, ':' ) ) );
 422                         $index = MediaWikiServices::getInstance()->getContentLanguage()->getNsIndex( $prefix );
 423                         if ( $index !== false ) {
 424                                 $extractedNamespace = [ $index ];
 425                                 $parsed = substr( $query, strlen( $prefix ) + 1 );
 426                         } elseif ( $withPrefixSearchExtractNamespaceHook ) {
 427                                 $hookNamespaces = [ NS_MAIN ];
 428                                 $hookQuery = $query;
 429                                 Hooks::run( 'PrefixSearchExtractNamespace', [ &$hookNamespaces, &$hookQuery ] );
 430                                 if ( $hookQuery !== $query ) {
 431                                         $parsed = $hookQuery;
 432                                         $extractedNamespace = $hookNamespaces;
 433                                 } else {
 434                                         return false;
 435                                 }
 436                         } else {
 437                                 return false;
 438                         }
 439                 }
 440
 441                 return [ $parsed, $extractedNamespace ];
 442         }
 443
 444         /**
 445          * Find snippet highlight settings for all users
 446          * @return array Contextlines, contextchars
 447          */
 448         public static function userHighlightPrefs() {
 449                 $contextlines = 2; // Hardcode this. Old defaults sucked. :)
 450                 $contextchars = 75; // same as above.... :P
 451                 return [ $contextlines, $contextchars ];
 452         }
 453
 454         /**
 455          * Create or update the search index record for the given page.
 456          * Title and text should be pre-processed.
 457          * STUB
 458          *
 459          * @param int $id
 460          * @param string $title
 461          * @param string $text
 462          */
 463         function update( $id, $title, $text ) {
 464                 // no-op
 465         }
 466
 467         /**
 468          * Update a search index record's title only.
 469          * Title should be pre-processed.
 470          * STUB
 471          *
 472          * @param int $id
 473          * @param string $title
 474          */
 475         function updateTitle( $id, $title ) {
 476                 // no-op
 477         }
 478
 479         /**
 480          * Delete an indexed page
 481          * Title should be pre-processed.
 482          * STUB
 483          *
 484          * @param int $id Page id that was deleted
 485          * @param string $title Title of page that was deleted
 486          */
 487         function delete( $id, $title ) {
 488                 // no-op
 489         }
 490
 491         /**
 492          * Get the raw text for updating the index from a content object
 493          * Nicer search backends could possibly do something cooler than
 494          * just returning raw text
 495          *
 496          * @todo This isn't ideal, we'd really like to have content-specific handling here
 497          * @param Title $t Title we're indexing
 498          * @param Content|null $c Content of the page to index
 499          * @return string
 500          */
 501         public function getTextFromContent( Title $t, Content $c = null ) {
 502                 return $c ? $c->getTextForSearchIndex() : '';
 503         }
 504
 505         /**
 506          * If an implementation of SearchEngine handles all of its own text processing
 507          * in getTextFromContent() and doesn't require SearchUpdate::updateText()'s
 508          * rather silly handling, it should return true here instead.
 509          *
 510          * @return bool
 511          */
 512         public function textAlreadyUpdatedForIndex() {
 513                 return false;
 514         }
 515
 516         /**
 517          * Makes search simple string if it was namespaced.
 518          * Sets namespaces of the search to namespaces extracted from string.
 519          * @param string $search
 520          * @return string Simplified search string
 521          */
 522         protected function normalizeNamespaces( $search ) {
 523                 $queryAndNs = self::parseNamespacePrefixes( $search, false, true );
 524                 if ( $queryAndNs !== false ) {
 525                         $this->setNamespaces( $queryAndNs[1] );
 526                         return $queryAndNs[0];
 527                 }
 528                 return $search;
 529         }
 530
 531         /**
 532          * Perform an overfetch of completion search results. This allows
 533          * determining if another page of results is available.
 534          *
 535          * @param string $search
 536          * @return SearchSuggestionSet
 537          */
 538         protected function completionSearchBackendOverfetch( $search ) {
 539                 $this->limit++;
 540                 try {
 541                         return $this->completionSearchBackend( $search );
 542                 } finally {
 543                         $this->limit--;
 544                 }
 545         }
 546
 547         /**
 548          * Perform a completion search.
 549          * Does not resolve namespaces and does not check variants.
 550          * Search engine implementations may want to override this function.
 551          * @param string $search
 552          * @return SearchSuggestionSet
 553          */
 554         protected function completionSearchBackend( $search ) {
 555                 $results = [];
 556
 557                 $search = trim( $search );
 558
 559                 if ( !in_array( NS_SPECIAL, $this->namespaces ) && // We do not run hook on Special: search
 560                          !Hooks::run( 'PrefixSearchBackend',
 561                                 [ $this->namespaces, $search, $this->limit, &$results, $this->offset ]
 562                 ) ) {
 563                         // False means hook worked.
 564                         // FIXME: Yes, the API is weird. That's why it is going to be deprecated.
 565
 566                         return SearchSuggestionSet::fromStrings( $results );
 567                 } else {
 568                         // Hook did not do the job, use default simple search
 569                         $results = $this->simplePrefixSearch( $search );
 570                         return SearchSuggestionSet::fromTitles( $results );
 571                 }
 572         }
 573
 574         /**
 575          * Perform a completion search.
 576          * @param string $search
 577          * @return SearchSuggestionSet
 578          */
 579         public function completionSearch( $search ) {
 580                 if ( trim( $search ) === '' ) {
 581                         return SearchSuggestionSet::emptySuggestionSet(); // Return empty result
 582                 }
 583                 $search = $this->normalizeNamespaces( $search );
 584                 $suggestions = $this->completionSearchBackendOverfetch( $search );
 585                 return $this->processCompletionResults( $search, $suggestions );
 586         }
 587
 588         /**
 589          * Perform a completion search with variants.
 590          * @param string $search
 591          * @return SearchSuggestionSet
 592          */
 593         public function completionSearchWithVariants( $search ) {
 594                 if ( trim( $search ) === '' ) {
 595                         return SearchSuggestionSet::emptySuggestionSet(); // Return empty result
 596                 }
 597                 $search = $this->normalizeNamespaces( $search );
 598
 599                 $results = $this->completionSearchBackendOverfetch( $search );
 600                 $fallbackLimit = 1 + $this->limit - $results->getSize();
 601                 if ( $fallbackLimit > 0 ) {
 602                         $fallbackSearches = MediaWikiServices::getInstance()->getContentLanguage()->
 603                                 autoConvertToAllVariants( $search );
 604                         $fallbackSearches = array_diff( array_unique( $fallbackSearches ), [ $search ] );
 605
 606                         foreach ( $fallbackSearches as $fbs ) {
 607                                 $this->setLimitOffset( $fallbackLimit );
 608                                 $fallbackSearchResult = $this->completionSearch( $fbs );
 609                                 $results->appendAll( $fallbackSearchResult );
 610                                 $fallbackLimit -= $fallbackSearchResult->getSize();
 611                                 if ( $fallbackLimit <= 0 ) {
 612                                         break;
 613                                 }
 614                         }
 615                 }
 616                 return $this->processCompletionResults( $search, $results );
 617         }
 618
 619         /**
 620          * Extract titles from completion results
 621          * @param SearchSuggestionSet $completionResults
 622          * @return Title[]
 623          */
 624         public function extractTitles( SearchSuggestionSet $completionResults ) {
 625                 return $completionResults->map( function ( SearchSuggestion $sugg ) {
 626                         return $sugg->getSuggestedTitle();
 627                 } );
 628         }
 629
 630         /**
 631          * Process completion search results.
 632          * Resolves the titles and rescores.
 633          * @param string $search
 634          * @param SearchSuggestionSet $suggestions
 635          * @return SearchSuggestionSet
 636          */
 637         protected function processCompletionResults( $search, SearchSuggestionSet $suggestions ) {
 638                 // We over-fetched to determine pagination. Shrink back down if we have extra results
 639                 // and mark if pagination is possible
 640                 $suggestions->shrink( $this->limit );
 641
 642                 $search = trim( $search );
 643                 // preload the titles with LinkBatch
 644                 $lb = new LinkBatch( $suggestions->map( function ( SearchSuggestion $sugg ) {
 645                         return $sugg->getSuggestedTitle();
 646                 } ) );
 647                 $lb->setCaller( __METHOD__ );
 648                 $lb->execute();
 649
 650                 $diff = $suggestions->filter( function ( SearchSuggestion $sugg ) {
 651                         return $sugg->getSuggestedTitle()->isKnown();
 652                 } );
 653                 if ( $diff > 0 ) {
 654                         MediaWikiServices::getInstance()->getStatsdDataFactory()
 655                                 ->updateCount( 'search.completion.missing', $diff );
 656                 }
 657
 658                 $results = $suggestions->map( function ( SearchSuggestion $sugg ) {
 659                         return $sugg->getSuggestedTitle()->getPrefixedText();
 660                 } );
 661
 662                 if ( $this->offset === 0 ) {
 663                         // Rescore results with an exact title match
 664                         // NOTE: in some cases like cross-namespace redirects
 665                         // (frequently used as shortcuts e.g. WP:WP on huwiki) some
 666                         // backends like Cirrus will return no results. We should still
 667                         // try an exact title match to workaround this limitation
 668                         $rescorer = new SearchExactMatchRescorer();
 669                         $rescoredResults = $rescorer->rescore( $search, $this->namespaces, $results, $this->limit );
 670                 } else {
 671                         // No need to rescore if offset is not 0
 672                         // The exact match must have been returned at position 0
 673                         // if it existed.
 674                         $rescoredResults = $results;
 675                 }
 676
 677                 if ( count( $rescoredResults ) > 0 ) {
 678                         $found = array_search( $rescoredResults[0], $results );
 679                         if ( $found === false ) {
 680                                 // If the first result is not in the previous array it
 681                                 // means that we found a new exact match
 682                                 $exactMatch = SearchSuggestion::fromTitle( 0, Title::newFromText( $rescoredResults[0] ) );
 683                                 $suggestions->prepend( $exactMatch );
 684                                 $suggestions->shrink( $this->limit );
 685                         } else {
 686                                 // if the first result is not the same we need to rescore
 687                                 if ( $found > 0 ) {
 688                                         $suggestions->rescore( $found );
 689                                 }
 690                         }
 691                 }
 692
 693                 return $suggestions;
 694         }
 695
 696         /**
 697          * Simple prefix search for subpages.
 698          * @param string $search
 699          * @return Title[]
 700          */
 701         public function defaultPrefixSearch( $search ) {
 702                 if ( trim( $search ) === '' ) {
 703                         return [];
 704                 }
 705
 706                 $search = $this->normalizeNamespaces( $search );
 707                 return $this->simplePrefixSearch( $search );
 708         }
 709
 710         /**
 711          * Call out to simple search backend.
 712          * Defaults to TitlePrefixSearch.
 713          * @param string $search
 714          * @return Title[]
 715          */
 716         protected function simplePrefixSearch( $search ) {
 717                 // Use default database prefix search
 718                 $backend = new TitlePrefixSearch;
 719                 return $backend->defaultSearchBackend( $this->namespaces, $search, $this->limit, $this->offset );
 720         }
 721
 722         /**
 723          * Get a list of supported profiles.
 724          * Some search engine implementations may expose specific profiles to fine-tune
 725          * its behaviors.
 726          * The profile can be passed as a feature data with setFeatureData( $profileType, $profileName )
 727          * The array returned by this function contains the following keys:
 728          * - name: the profile name to use with setFeatureData
 729          * - desc-message: the i18n description
 730          * - default: set to true if this profile is the default
 731          *
 732          * @since 1.28
 733          * @param string $profileType the type of profiles
 734          * @param User|null $user the user requesting the list of profiles
 735          * @return array|null the list of profiles or null if none available
 736          */
 737         public function getProfiles( $profileType, User $user = null ) {
 738                 return null;
 739         }
 740
 741         /**
 742          * Create a search field definition.
 743          * Specific search engines should override this method to create search fields.
 744          * @param string $name
 745          * @param string $type One of the types in SearchIndexField::INDEX_TYPE_*
 746          * @return SearchIndexField
 747          * @since 1.28
 748          */
 749         public function makeSearchFieldMapping( $name, $type ) {
 750                 return new NullIndexField();
 751         }
 752
 753         /**
 754          * Get fields for search index
 755          * @since 1.28
 756          * @return SearchIndexField[] Index field definitions for all content handlers
 757          */
 758         public function getSearchIndexFields() {
 759                 $models = ContentHandler::getContentModels();
 760                 $fields = [];
 761                 $seenHandlers = new SplObjectStorage();
 762                 foreach ( $models as $model ) {
 763                         try {
 764                                 $handler = ContentHandler::getForModelID( $model );
 765                         }
 766                         catch ( MWUnknownContentModelException $e ) {
 767                                 // If we can find no handler, ignore it
 768                                 continue;
 769                         }
 770                         // Several models can have the same handler, so avoid processing it repeatedly
 771                         if ( $seenHandlers->contains( $handler ) ) {
 772                                 // We already did this one
 773                                 continue;
 774                         }
 775                         $seenHandlers->attach( $handler );
 776                         $handlerFields = $handler->getFieldsForSearchIndex( $this );
 777                         foreach ( $handlerFields as $fieldName => $fieldData ) {
 778                                 if ( empty( $fields[$fieldName] ) ) {
 779                                         $fields[$fieldName] = $fieldData;
 780                                 } else {
 781                                         // TODO: do we allow some clashes with the same type or reject all of them?
 782                                         $mergeDef = $fields[$fieldName]->merge( $fieldData );
 783                                         if ( !$mergeDef ) {
 784                                                 throw new InvalidArgumentException( "Duplicate field $fieldName for model $model" );
 785                                         }
 786                                         $fields[$fieldName] = $mergeDef;
 787                                 }
 788                         }
 789                 }
 790                 // Hook to allow extensions to produce search mapping fields
 791                 Hooks::run( 'SearchIndexFields', [ &$fields, $this ] );
 792                 return $fields;
 793         }
 794
 795         /**
 796          * Augment search results with extra data.
 797          *
 798          * @param SearchResultSet $resultSet
 799          */
 800         public function augmentSearchResults( SearchResultSet $resultSet ) {
 801                 $setAugmentors = [];
 802                 $rowAugmentors = [];
 803                 Hooks::run( "SearchResultsAugment", [ &$setAugmentors, &$rowAugmentors ] );
 804                 if ( !$setAugmentors && !$rowAugmentors ) {
 805                         // We're done here
 806                         return;
 807                 }
 808
 809                 // Convert row augmentors to set augmentor
 810                 foreach ( $rowAugmentors as $name => $row ) {
 811                         if ( isset( $setAugmentors[$name] ) ) {
 812                                 throw new InvalidArgumentException( "Both row and set augmentors are defined for $name" );
 813                         }
 814                         $setAugmentors[$name] = new PerRowAugmentor( $row );
 815                 }
 816
 817                 foreach ( $setAugmentors as $name => $augmentor ) {
 818                         $data = $augmentor->augmentAll( $resultSet );
 819                         if ( $data ) {
 820                                 $resultSet->setAugmentedData( $name, $data );
 821                         }
 822                 }
 823         }
 824 }