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 SearchResultSet|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 SearchResultSet|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 SearchResultSet|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 SearchResultSet|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 SearchResultSet|Status<SearchResultSet>|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 SearchResultSet ) {
 179                         $resultSet = $resultSetOrStatus;
 180                 } elseif ( $resultSetOrStatus instanceof Status &&
 181                         $resultSetOrStatus->getValue() instanceof SearchResultSet
 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          */
 437         public static function userHighlightPrefs() {
 438                 $contextlines = 2; // Hardcode this. Old defaults sucked. :)
 439                 $contextchars = 75; // same as above.... :P
 440                 return [ $contextlines, $contextchars ];
 441         }
 442
 443         /**
 444          * Create or update the search index record for the given page.
 445          * Title and text should be pre-processed.
 446          * STUB
 447          *
 448          * @param int $id
 449          * @param string $title
 450          * @param string $text
 451          */
 452         function update( $id, $title, $text ) {
 453                 // no-op
 454         }
 455
 456         /**
 457          * Update a search index record's title only.
 458          * Title should be pre-processed.
 459          * STUB
 460          *
 461          * @param int $id
 462          * @param string $title
 463          */
 464         function updateTitle( $id, $title ) {
 465                 // no-op
 466         }
 467
 468         /**
 469          * Delete an indexed page
 470          * Title should be pre-processed.
 471          * STUB
 472          *
 473          * @param int $id Page id that was deleted
 474          * @param string $title Title of page that was deleted
 475          */
 476         function delete( $id, $title ) {
 477                 // no-op
 478         }
 479
 480         /**
 481          * Get the raw text for updating the index from a content object
 482          * Nicer search backends could possibly do something cooler than
 483          * just returning raw text
 484          *
 485          * @todo This isn't ideal, we'd really like to have content-specific handling here
 486          * @param Title $t Title we're indexing
 487          * @param Content|null $c Content of the page to index
 488          * @return string
 489          */
 490         public function getTextFromContent( Title $t, Content $c = null ) {
 491                 return $c ? $c->getTextForSearchIndex() : '';
 492         }
 493
 494         /**
 495          * If an implementation of SearchEngine handles all of its own text processing
 496          * in getTextFromContent() and doesn't require SearchUpdate::updateText()'s
 497          * rather silly handling, it should return true here instead.
 498          *
 499          * @return bool
 500          */
 501         public function textAlreadyUpdatedForIndex() {
 502                 return false;
 503         }
 504
 505         /**
 506          * Makes search simple string if it was namespaced.
 507          * Sets namespaces of the search to namespaces extracted from string.
 508          * @param string $search
 509          * @return string Simplified search string
 510          */
 511         protected function normalizeNamespaces( $search ) {
 512                 $queryAndNs = self::parseNamespacePrefixes( $search, false, true );
 513                 if ( $queryAndNs !== false ) {
 514                         $this->setNamespaces( $queryAndNs[1] );
 515                         return $queryAndNs[0];
 516                 }
 517                 return $search;
 518         }
 519
 520         /**
 521          * Perform an overfetch of completion search results. This allows
 522          * determining if another page of results is available.
 523          *
 524          * @param string $search
 525          * @return SearchSuggestionSet
 526          */
 527         protected function completionSearchBackendOverfetch( $search ) {
 528                 $this->limit++;
 529                 try {
 530                         return $this->completionSearchBackend( $search );
 531                 } finally {
 532                         $this->limit--;
 533                 }
 534         }
 535
 536         /**
 537          * Perform a completion search.
 538          * Does not resolve namespaces and does not check variants.
 539          * Search engine implementations may want to override this function.
 540          * @param string $search
 541          * @return SearchSuggestionSet
 542          */
 543         protected function completionSearchBackend( $search ) {
 544                 $results = [];
 545
 546                 $search = trim( $search );
 547
 548                 if ( !in_array( NS_SPECIAL, $this->namespaces ) && // We do not run hook on Special: search
 549                          !Hooks::run( 'PrefixSearchBackend',
 550                                 [ $this->namespaces, $search, $this->limit, &$results, $this->offset ]
 551                 ) ) {
 552                         // False means hook worked.
 553                         // FIXME: Yes, the API is weird. That's why it is going to be deprecated.
 554
 555                         return SearchSuggestionSet::fromStrings( $results );
 556                 } else {
 557                         // Hook did not do the job, use default simple search
 558                         $results = $this->simplePrefixSearch( $search );
 559                         return SearchSuggestionSet::fromTitles( $results );
 560                 }
 561         }
 562
 563         /**
 564          * Perform a completion search.
 565          * @param string $search
 566          * @return SearchSuggestionSet
 567          */
 568         public function completionSearch( $search ) {
 569                 if ( trim( $search ) === '' ) {
 570                         return SearchSuggestionSet::emptySuggestionSet(); // Return empty result
 571                 }
 572                 $search = $this->normalizeNamespaces( $search );
 573                 $suggestions = $this->completionSearchBackendOverfetch( $search );
 574                 return $this->processCompletionResults( $search, $suggestions );
 575         }
 576
 577         /**
 578          * Perform a completion search with variants.
 579          * @param string $search
 580          * @return SearchSuggestionSet
 581          */
 582         public function completionSearchWithVariants( $search ) {
 583                 if ( trim( $search ) === '' ) {
 584                         return SearchSuggestionSet::emptySuggestionSet(); // Return empty result
 585                 }
 586                 $search = $this->normalizeNamespaces( $search );
 587
 588                 $results = $this->completionSearchBackendOverfetch( $search );
 589                 $fallbackLimit = 1 + $this->limit - $results->getSize();
 590                 if ( $fallbackLimit > 0 ) {
 591                         $fallbackSearches = MediaWikiServices::getInstance()->getContentLanguage()->
 592                                 autoConvertToAllVariants( $search );
 593                         $fallbackSearches = array_diff( array_unique( $fallbackSearches ), [ $search ] );
 594
 595                         foreach ( $fallbackSearches as $fbs ) {
 596                                 $this->setLimitOffset( $fallbackLimit );
 597                                 $fallbackSearchResult = $this->completionSearch( $fbs );
 598                                 $results->appendAll( $fallbackSearchResult );
 599                                 $fallbackLimit -= $fallbackSearchResult->getSize();
 600                                 if ( $fallbackLimit <= 0 ) {
 601                                         break;
 602                                 }
 603                         }
 604                 }
 605                 return $this->processCompletionResults( $search, $results );
 606         }
 607
 608         /**
 609          * Extract titles from completion results
 610          * @param SearchSuggestionSet $completionResults
 611          * @return Title[]
 612          */
 613         public function extractTitles( SearchSuggestionSet $completionResults ) {
 614                 return $completionResults->map( function ( SearchSuggestion $sugg ) {
 615                         return $sugg->getSuggestedTitle();
 616                 } );
 617         }
 618
 619         /**
 620          * Process completion search results.
 621          * Resolves the titles and rescores.
 622          * @param string $search
 623          * @param SearchSuggestionSet $suggestions
 624          * @return SearchSuggestionSet
 625          */
 626         protected function processCompletionResults( $search, SearchSuggestionSet $suggestions ) {
 627                 // We over-fetched to determine pagination. Shrink back down if we have extra results
 628                 // and mark if pagination is possible
 629                 $suggestions->shrink( $this->limit );
 630
 631                 $search = trim( $search );
 632                 // preload the titles with LinkBatch
 633                 $lb = new LinkBatch( $suggestions->map( function ( SearchSuggestion $sugg ) {
 634                         return $sugg->getSuggestedTitle();
 635                 } ) );
 636                 $lb->setCaller( __METHOD__ );
 637                 $lb->execute();
 638
 639                 $diff = $suggestions->filter( function ( SearchSuggestion $sugg ) {
 640                         return $sugg->getSuggestedTitle()->isKnown();
 641                 } );
 642                 if ( $diff > 0 ) {
 643                         MediaWikiServices::getInstance()->getStatsdDataFactory()
 644                                 ->updateCount( 'search.completion.missing', $diff );
 645                 }
 646
 647                 $results = $suggestions->map( function ( SearchSuggestion $sugg ) {
 648                         return $sugg->getSuggestedTitle()->getPrefixedText();
 649                 } );
 650
 651                 if ( $this->offset === 0 ) {
 652                         // Rescore results with an exact title match
 653                         // NOTE: in some cases like cross-namespace redirects
 654                         // (frequently used as shortcuts e.g. WP:WP on huwiki) some
 655                         // backends like Cirrus will return no results. We should still
 656                         // try an exact title match to workaround this limitation
 657                         $rescorer = new SearchExactMatchRescorer();
 658                         $rescoredResults = $rescorer->rescore( $search, $this->namespaces, $results, $this->limit );
 659                 } else {
 660                         // No need to rescore if offset is not 0
 661                         // The exact match must have been returned at position 0
 662                         // if it existed.
 663                         $rescoredResults = $results;
 664                 }
 665
 666                 if ( count( $rescoredResults ) > 0 ) {
 667                         $found = array_search( $rescoredResults[0], $results );
 668                         if ( $found === false ) {
 669                                 // If the first result is not in the previous array it
 670                                 // means that we found a new exact match
 671                                 $exactMatch = SearchSuggestion::fromTitle( 0, Title::newFromText( $rescoredResults[0] ) );
 672                                 $suggestions->prepend( $exactMatch );
 673                                 $suggestions->shrink( $this->limit );
 674                         } else {
 675                                 // if the first result is not the same we need to rescore
 676                                 if ( $found > 0 ) {
 677                                         $suggestions->rescore( $found );
 678                                 }
 679                         }
 680                 }
 681
 682                 return $suggestions;
 683         }
 684
 685         /**
 686          * Simple prefix search for subpages.
 687          * @param string $search
 688          * @return Title[]
 689          */
 690         public function defaultPrefixSearch( $search ) {
 691                 if ( trim( $search ) === '' ) {
 692                         return [];
 693                 }
 694
 695                 $search = $this->normalizeNamespaces( $search );
 696                 return $this->simplePrefixSearch( $search );
 697         }
 698
 699         /**
 700          * Call out to simple search backend.
 701          * Defaults to TitlePrefixSearch.
 702          * @param string $search
 703          * @return Title[]
 704          */
 705         protected function simplePrefixSearch( $search ) {
 706                 // Use default database prefix search
 707                 $backend = new TitlePrefixSearch;
 708                 return $backend->defaultSearchBackend( $this->namespaces, $search, $this->limit, $this->offset );
 709         }
 710
 711         /**
 712          * Get a list of supported profiles.
 713          * Some search engine implementations may expose specific profiles to fine-tune
 714          * its behaviors.
 715          * The profile can be passed as a feature data with setFeatureData( $profileType, $profileName )
 716          * The array returned by this function contains the following keys:
 717          * - name: the profile name to use with setFeatureData
 718          * - desc-message: the i18n description
 719          * - default: set to true if this profile is the default
 720          *
 721          * @since 1.28
 722          * @param string $profileType the type of profiles
 723          * @param User|null $user the user requesting the list of profiles
 724          * @return array|null the list of profiles or null if none available
 725          */
 726         public function getProfiles( $profileType, User $user = null ) {
 727                 return null;
 728         }
 729
 730         /**
 731          * Create a search field definition.
 732          * Specific search engines should override this method to create search fields.
 733          * @param string $name
 734          * @param string $type One of the types in SearchIndexField::INDEX_TYPE_*
 735          * @return SearchIndexField
 736          * @since 1.28
 737          */
 738         public function makeSearchFieldMapping( $name, $type ) {
 739                 return new NullIndexField();
 740         }
 741
 742         /**
 743          * Get fields for search index
 744          * @since 1.28
 745          * @return SearchIndexField[] Index field definitions for all content handlers
 746          */
 747         public function getSearchIndexFields() {
 748                 $models = ContentHandler::getContentModels();
 749                 $fields = [];
 750                 $seenHandlers = new SplObjectStorage();
 751                 foreach ( $models as $model ) {
 752                         try {
 753                                 $handler = ContentHandler::getForModelID( $model );
 754                         }
 755                         catch ( MWUnknownContentModelException $e ) {
 756                                 // If we can find no handler, ignore it
 757                                 continue;
 758                         }
 759                         // Several models can have the same handler, so avoid processing it repeatedly
 760                         if ( $seenHandlers->contains( $handler ) ) {
 761                                 // We already did this one
 762                                 continue;
 763                         }
 764                         $seenHandlers->attach( $handler );
 765                         $handlerFields = $handler->getFieldsForSearchIndex( $this );
 766                         foreach ( $handlerFields as $fieldName => $fieldData ) {
 767                                 if ( empty( $fields[$fieldName] ) ) {
 768                                         $fields[$fieldName] = $fieldData;
 769                                 } else {
 770                                         // TODO: do we allow some clashes with the same type or reject all of them?
 771                                         $mergeDef = $fields[$fieldName]->merge( $fieldData );
 772                                         if ( !$mergeDef ) {
 773                                                 throw new InvalidArgumentException( "Duplicate field $fieldName for model $model" );
 774                                         }
 775                                         $fields[$fieldName] = $mergeDef;
 776                                 }
 777                         }
 778                 }
 779                 // Hook to allow extensions to produce search mapping fields
 780                 Hooks::run( 'SearchIndexFields', [ &$fields, $this ] );
 781                 return $fields;
 782         }
 783
 784         /**
 785          * Augment search results with extra data.
 786          *
 787          * @param SearchResultSet $resultSet
 788          */
 789         public function augmentSearchResults( SearchResultSet $resultSet ) {
 790                 $setAugmentors = [];
 791                 $rowAugmentors = [];
 792                 Hooks::run( "SearchResultsAugment", [ &$setAugmentors, &$rowAugmentors ] );
 793                 if ( !$setAugmentors && !$rowAugmentors ) {
 794                         // We're done here
 795                         return;
 796                 }
 797
 798                 // Convert row augmentors to set augmentor
 799                 foreach ( $rowAugmentors as $name => $row ) {
 800                         if ( isset( $setAugmentors[$name] ) ) {
 801                                 throw new InvalidArgumentException( "Both row and set augmentors are defined for $name" );
 802                         }
 803                         $setAugmentors[$name] = new PerRowAugmentor( $row );
 804                 }
 805
 806                 foreach ( $setAugmentors as $name => $augmentor ) {
 807                         $data = $augmentor->augmentAll( $resultSet );
 808                         if ( $data ) {
 809                                 $resultSet->setAugmentedData( $name, $data );
 810                         }
 811                 }
 812         }
 813 }