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