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         /** @var string */
  36         public $prefix = '';
  37
  38         /** @var int[]|null */
  39         public $namespaces = [ NS_MAIN ];
  40
  41         /** @var int */
  42         protected $limit = 10;
  43
  44         /** @var int */
  45         protected $offset = 0;
  46
  47         /** @var array|string */
  48         protected $searchTerms = [];
  49
  50         /** @var bool */
  51         protected $showSuggestion = true;
  52         private $sort = 'relevance';
  53
  54         /** @var array Feature values */
  55         protected $features = [];
  56
  57         /** @const string profile type for completionSearch */
  58         const COMPLETION_PROFILE_TYPE = 'completionSearchProfile';
  59
  60         /** @const string profile type for query independent ranking features */
  61         const FT_QUERY_INDEP_PROFILE_TYPE = 'fulltextQueryIndepProfile';
  62
  63         /** @const int flag for legalSearchChars: includes all chars allowed in a search query */
  64         const CHARS_ALL = 1;
  65
  66         /** @const int flag for legalSearchChars: includes all chars allowed in a search term */
  67         const CHARS_NO_SYNTAX = 2;
  68
  69         /**
  70          * Perform a full text search query and return a result set.
  71          * If full text searches are not supported or disabled, return null.
  72          *
  73          * As of 1.32 overriding this function is deprecated. It will
  74          * be converted to final in 1.34. Override self::doSearchText().
  75          *
  76          * @param string $term Raw search term
  77          * @return SearchResultSet|Status|null
  78          */
  79         public function searchText( $term ) {
  80                 return $this->maybePaginate( function () use ( $term ) {
  81                         return $this->doSearchText( $term );
  82                 } );
  83         }
  84
  85         /**
  86          * Perform a full text search query and return a result set.
  87          *
  88          * @param string $term Raw search term
  89          * @return SearchResultSet|Status|null
  90          * @since 1.32
  91          */
  92         protected function doSearchText( $term ) {
  93                 return null;
  94         }
  95
  96         /**
  97          * Perform a title search in the article archive.
  98          * NOTE: these results still should be filtered by
  99          * matching against PageArchive, permissions checks etc
 100          * The results returned by this methods are only sugegstions and
 101          * may not end up being shown to the user.
 102          *
 103          * As of 1.32 overriding this function is deprecated. It will
 104          * be converted to final in 1.34. Override self::doSearchArchiveTitle().
 105          *
 106          * @param string $term Raw search term
 107          * @return Status<Title[]>
 108          * @since 1.29
 109          */
 110         public function searchArchiveTitle( $term ) {
 111                 return $this->doSearchArchiveTitle( $term );
 112         }
 113
 114         /**
 115          * Perform a title search in the article archive.
 116          *
 117          * @param string $term Raw search term
 118          * @return Status<Title[]>
 119          * @since 1.32
 120          */
 121         protected function doSearchArchiveTitle( $term ) {
 122                 return Status::newGood( [] );
 123         }
 124
 125         /**
 126          * Perform a title-only search query and return a result set.
 127          * If title searches are not supported or disabled, return null.
 128          * STUB
 129          *
 130          * As of 1.32 overriding this function is deprecated. It will
 131          * be converted to final in 1.34. Override self::doSearchTitle().
 132          *
 133          * @param string $term Raw search term
 134          * @return SearchResultSet|null
 135          */
 136         public function searchTitle( $term ) {
 137                 return $this->maybePaginate( function () use ( $term ) {
 138                         return $this->doSearchTitle( $term );
 139                 } );
 140         }
 141
 142         /**
 143          * Perform a title-only search query and return a result set.
 144          *
 145          * @param string $term Raw search term
 146          * @return SearchResultSet|null
 147          * @since 1.32
 148          */
 149         protected function doSearchTitle( $term ) {
 150                 return null;
 151         }
 152
 153         /**
 154          * Performs an overfetch and shrink operation to determine if
 155          * the next page is available for search engines that do not
 156          * explicitly implement their own pagination.
 157          *
 158          * @param Closure $fn Takes no arguments
 159          * @return SearchResultSet|Status<SearchResultSet>|null Result of calling $fn
 160          */
 161         private function maybePaginate( Closure $fn ) {
 162                 if ( $this instanceof PaginatingSearchEngine ) {
 163                         return $fn();
 164                 }
 165                 $this->limit++;
 166                 try {
 167                         $resultSetOrStatus = $fn();
 168                 } finally {
 169                         $this->limit--;
 170                 }
 171
 172                 $resultSet = null;
 173                 if ( $resultSetOrStatus instanceof SearchResultSet ) {
 174                         $resultSet = $resultSetOrStatus;
 175                 } elseif ( $resultSetOrStatus instanceof Status &&
 176                         $resultSetOrStatus->getValue() instanceof SearchResultSet
 177                 ) {
 178                         $resultSet = $resultSetOrStatus->getValue();
 179                 }
 180                 if ( $resultSet ) {
 181                         $resultSet->shrink( $this->limit );
 182                 }
 183
 184                 return $resultSetOrStatus;
 185         }
 186
 187         /**
 188          * @since 1.18
 189          * @param string $feature
 190          * @return bool
 191          */
 192         public function supports( $feature ) {
 193                 switch ( $feature ) {
 194                         case 'search-update':
 195                                 return true;
 196                         case 'title-suffix-filter':
 197                         default:
 198                                 return false;
 199                 }
 200         }
 201
 202         /**
 203          * Way to pass custom data for engines
 204          * @since 1.18
 205          * @param string $feature
 206          * @param mixed $data
 207          */
 208         public function setFeatureData( $feature, $data ) {
 209                 $this->features[$feature] = $data;
 210         }
 211
 212         /**
 213          * Way to retrieve custom data set by setFeatureData
 214          * or by the engine itself.
 215          * @since 1.29
 216          * @param string $feature feature name
 217          * @return mixed the feature value or null if unset
 218          */
 219         public function getFeatureData( $feature ) {
 220                 if ( isset( $this->features[$feature] ) ) {
 221                         return $this->features[$feature];
 222                 }
 223                 return null;
 224         }
 225
 226         /**
 227          * When overridden in derived class, performs database-specific conversions
 228          * on text to be used for searching or updating search index.
 229          * Default implementation does nothing (simply returns $string).
 230          *
 231          * @param string $string String to process
 232          * @return string
 233          */
 234         public function normalizeText( $string ) {
 235                 global $wgContLang;
 236
 237                 // Some languages such as Chinese require word segmentation
 238                 return $wgContLang->segmentByWord( $string );
 239         }
 240
 241         /**
 242          * Transform search term in cases when parts of the query came as different
 243          * GET params (when supported), e.g. for prefix queries:
 244          * search=test&prefix=Main_Page/Archive -> test prefix:Main Page/Archive
 245          * @param string $term
 246          * @return string
 247          * @deprecated since 1.32 this should now be handled internally by the
 248          * search engine
 249          */
 250         public function transformSearchTerm( $term ) {
 251                 return $term;
 252         }
 253
 254         /**
 255          * Get service class to finding near matches.
 256          * @param Config $config Configuration to use for the matcher.
 257          * @return SearchNearMatcher
 258          */
 259         public function getNearMatcher( Config $config ) {
 260                 global $wgContLang;
 261                 return new SearchNearMatcher( $config, $wgContLang );
 262         }
 263
 264         /**
 265          * Get near matcher for default SearchEngine.
 266          * @return SearchNearMatcher
 267          */
 268         protected static function defaultNearMatcher() {
 269                 $config = MediaWikiServices::getInstance()->getMainConfig();
 270                 return MediaWikiServices::getInstance()->newSearchEngine()->getNearMatcher( $config );
 271         }
 272
 273         /**
 274          * If an exact title match can be found, or a very slightly close match,
 275          * return the title. If no match, returns NULL.
 276          * @deprecated since 1.27; Use SearchEngine::getNearMatcher()
 277          * @param string $searchterm
 278          * @return Title
 279          */
 280         public static function getNearMatch( $searchterm ) {
 281                 return static::defaultNearMatcher()->getNearMatch( $searchterm );
 282         }
 283
 284         /**
 285          * Do a near match (see SearchEngine::getNearMatch) and wrap it into a
 286          * SearchResultSet.
 287          * @deprecated since 1.27; Use SearchEngine::getNearMatcher()
 288          * @param string $searchterm
 289          * @return SearchResultSet
 290          */
 291         public static function getNearMatchResultSet( $searchterm ) {
 292                 return static::defaultNearMatcher()->getNearMatchResultSet( $searchterm );
 293         }
 294
 295         /**
 296          * Get chars legal for search
 297          * NOTE: usage as static is deprecated and preserved only as BC measure
 298          * @param int $type type of search chars (see self::CHARS_ALL
 299          * and self::CHARS_NO_SYNTAX). Defaults to CHARS_ALL
 300          * @return string
 301          */
 302         public static function legalSearchChars( $type = self::CHARS_ALL ) {
 303                 return "A-Za-z_'.0-9\\x80-\\xFF\\-";
 304         }
 305
 306         /**
 307          * Set the maximum number of results to return
 308          * and how many to skip before returning the first.
 309          *
 310          * @param int $limit
 311          * @param int $offset
 312          */
 313         function setLimitOffset( $limit, $offset = 0 ) {
 314                 $this->limit = intval( $limit );
 315                 $this->offset = intval( $offset );
 316         }
 317
 318         /**
 319          * Set which namespaces the search should include.
 320          * Give an array of namespace index numbers.
 321          *
 322          * @param int[]|null $namespaces
 323          */
 324         function setNamespaces( $namespaces ) {
 325                 if ( $namespaces ) {
 326                         // Filter namespaces to only keep valid ones
 327                         $validNs = $this->searchableNamespaces();
 328                         $namespaces = array_filter( $namespaces, function ( $ns ) use( $validNs ) {
 329                                 return $ns < 0 || isset( $validNs[$ns] );
 330                         } );
 331                 } else {
 332                         $namespaces = [];
 333                 }
 334                 $this->namespaces = $namespaces;
 335         }
 336
 337         /**
 338          * Set whether the searcher should try to build a suggestion.  Note: some searchers
 339          * don't support building a suggestion in the first place and others don't respect
 340          * this flag.
 341          *
 342          * @param bool $showSuggestion Should the searcher try to build suggestions
 343          */
 344         function setShowSuggestion( $showSuggestion ) {
 345                 $this->showSuggestion = $showSuggestion;
 346         }
 347
 348         /**
 349          * Get the valid sort directions.  All search engines support 'relevance' but others
 350          * might support more. The default in all implementations should be 'relevance.'
 351          *
 352          * @since 1.25
 353          * @return string[] the valid sort directions for setSort
 354          */
 355         public function getValidSorts() {
 356                 return [ 'relevance' ];
 357         }
 358
 359         /**
 360          * Set the sort direction of the search results. Must be one returned by
 361          * SearchEngine::getValidSorts()
 362          *
 363          * @since 1.25
 364          * @throws InvalidArgumentException
 365          * @param string $sort sort direction for query result
 366          */
 367         public function setSort( $sort ) {
 368                 if ( !in_array( $sort, $this->getValidSorts() ) ) {
 369                         throw new InvalidArgumentException( "Invalid sort: $sort. " .
 370                                 "Must be one of: " . implode( ', ', $this->getValidSorts() ) );
 371                 }
 372                 $this->sort = $sort;
 373         }
 374
 375         /**
 376          * Get the sort direction of the search results
 377          *
 378          * @since 1.25
 379          * @return string
 380          */
 381         public function getSort() {
 382                 return $this->sort;
 383         }
 384
 385         /**
 386          * Parse some common prefixes: all (search everything)
 387          * or namespace names and set the list of namespaces
 388          * of this class accordingly.
 389          *
 390          * @param string $query
 391          * @return string
 392          */
 393         function replacePrefixes( $query ) {
 394                 $queryAndNs = self::parseNamespacePrefixes( $query );
 395                 if ( $queryAndNs === false ) {
 396                         return $query;
 397                 }
 398                 $this->namespaces = $queryAndNs[1];
 399                 return $queryAndNs[0];
 400         }
 401
 402         /**
 403          * Parse some common prefixes: all (search everything)
 404          * or namespace names
 405          *
 406          * @param string $query
 407          * @return false|array false if no namespace was extracted, an array
 408          * with the parsed query at index 0 and an array of namespaces at index
 409          * 1 (or null for all namespaces).
 410          */
 411         public static function parseNamespacePrefixes( $query ) {
 412                 global $wgContLang;
 413
 414                 $parsed = $query;
 415                 if ( strpos( $query, ':' ) === false ) { // nothing to do
 416                         return false;
 417                 }
 418                 $extractedNamespace = null;
 419                 $allkeywords = [];
 420
 421                 $allkeywords[] = wfMessage( 'searchall' )->inContentLanguage()->text() . ":";
 422                 // force all: so that we have a common syntax for all the wikis
 423                 if ( !in_array( 'all:', $allkeywords ) ) {
 424                         $allkeywords[] = 'all:';
 425                 }
 426
 427                 $allQuery = false;
 428                 foreach ( $allkeywords as $kw ) {
 429                         if ( strncmp( $query, $kw, strlen( $kw ) ) == 0 ) {
 430                                 $extractedNamespace = null;
 431                                 $parsed = substr( $query, strlen( $kw ) );
 432                                 $allQuery = true;
 433                                 break;
 434                         }
 435                 }
 436
 437                 if ( !$allQuery && strpos( $query, ':' ) !== false ) {
 438                         // TODO: should we unify with PrefixSearch::extractNamespace ?
 439                         $prefix = str_replace( ' ', '_', substr( $query, 0, strpos( $query, ':' ) ) );
 440                         $index = $wgContLang->getNsIndex( $prefix );
 441                         if ( $index !== false ) {
 442                                 $extractedNamespace = [ $index ];
 443                                 $parsed = substr( $query, strlen( $prefix ) + 1 );
 444                         } else {
 445                                 return false;
 446                         }
 447                 }
 448
 449                 if ( trim( $parsed ) == '' ) {
 450                         $parsed = $query; // prefix was the whole query
 451                 }
 452
 453                 return [ $parsed, $extractedNamespace ];
 454         }
 455
 456         /**
 457          * Find snippet highlight settings for all users
 458          * @return array Contextlines, contextchars
 459          */
 460         public static function userHighlightPrefs() {
 461                 $contextlines = 2; // Hardcode this. Old defaults sucked. :)
 462                 $contextchars = 75; // same as above.... :P
 463                 return [ $contextlines, $contextchars ];
 464         }
 465
 466         /**
 467          * Create or update the search index record for the given page.
 468          * Title and text should be pre-processed.
 469          * STUB
 470          *
 471          * @param int $id
 472          * @param string $title
 473          * @param string $text
 474          */
 475         function update( $id, $title, $text ) {
 476                 // no-op
 477         }
 478
 479         /**
 480          * Update a search index record's title only.
 481          * Title should be pre-processed.
 482          * STUB
 483          *
 484          * @param int $id
 485          * @param string $title
 486          */
 487         function updateTitle( $id, $title ) {
 488                 // no-op
 489         }
 490
 491         /**
 492          * Delete an indexed page
 493          * Title should be pre-processed.
 494          * STUB
 495          *
 496          * @param int $id Page id that was deleted
 497          * @param string $title Title of page that was deleted
 498          */
 499         function delete( $id, $title ) {
 500                 // no-op
 501         }
 502
 503         /**
 504          * Get the raw text for updating the index from a content object
 505          * Nicer search backends could possibly do something cooler than
 506          * just returning raw text
 507          *
 508          * @todo This isn't ideal, we'd really like to have content-specific handling here
 509          * @param Title $t Title we're indexing
 510          * @param Content|null $c Content of the page to index
 511          * @return string
 512          */
 513         public function getTextFromContent( Title $t, Content $c = null ) {
 514                 return $c ? $c->getTextForSearchIndex() : '';
 515         }
 516
 517         /**
 518          * If an implementation of SearchEngine handles all of its own text processing
 519          * in getTextFromContent() and doesn't require SearchUpdate::updateText()'s
 520          * rather silly handling, it should return true here instead.
 521          *
 522          * @return bool
 523          */
 524         public function textAlreadyUpdatedForIndex() {
 525                 return false;
 526         }
 527
 528         /**
 529          * Makes search simple string if it was namespaced.
 530          * Sets namespaces of the search to namespaces extracted from string.
 531          * @param string $search
 532          * @return string Simplified search string
 533          */
 534         protected function normalizeNamespaces( $search ) {
 535                 // Find a Title which is not an interwiki and is in NS_MAIN
 536                 $title = Title::newFromText( $search );
 537                 $ns = $this->namespaces;
 538                 if ( $title && !$title->isExternal() ) {
 539                         $ns = [ $title->getNamespace() ];
 540                         $search = $title->getText();
 541                         if ( $ns[0] == NS_MAIN ) {
 542                                 $ns = $this->namespaces; // no explicit prefix, use default namespaces
 543                                 Hooks::run( 'PrefixSearchExtractNamespace', [ &$ns, &$search ] );
 544                         }
 545                 } else {
 546                         $title = Title::newFromText( $search . 'Dummy' );
 547                         if ( $title && $title->getText() == 'Dummy'
 548                                         && $title->getNamespace() != NS_MAIN
 549                                         && !$title->isExternal()
 550                         ) {
 551                                 $ns = [ $title->getNamespace() ];
 552                                 $search = '';
 553                         } else {
 554                                 Hooks::run( 'PrefixSearchExtractNamespace', [ &$ns, &$search ] );
 555                         }
 556                 }
 557
 558                 $ns = array_map( function ( $space ) {
 559                         return $space == NS_MEDIA ? NS_FILE : $space;
 560                 }, $ns );
 561
 562                 $this->setNamespaces( $ns );
 563                 return $search;
 564         }
 565
 566         /**
 567          * Perform an overfetch of completion search results. This allows
 568          * determining if another page of results is available.
 569          *
 570          * @param string $search
 571          * @return SearchSuggestionSet
 572          */
 573         protected function completionSearchBackendOverfetch( $search ) {
 574                 $this->limit++;
 575                 try {
 576                         return $this->completionSearchBackend( $search );
 577                 } finally {
 578                         $this->limit--;
 579                 }
 580         }
 581
 582         /**
 583          * Perform a completion search.
 584          * Does not resolve namespaces and does not check variants.
 585          * Search engine implementations may want to override this function.
 586          * @param string $search
 587          * @return SearchSuggestionSet
 588          */
 589         protected function completionSearchBackend( $search ) {
 590                 $results = [];
 591
 592                 $search = trim( $search );
 593
 594                 if ( !in_array( NS_SPECIAL, $this->namespaces ) && // We do not run hook on Special: search
 595                          !Hooks::run( 'PrefixSearchBackend',
 596                                 [ $this->namespaces, $search, $this->limit, &$results, $this->offset ]
 597                 ) ) {
 598                         // False means hook worked.
 599                         // FIXME: Yes, the API is weird. That's why it is going to be deprecated.
 600
 601                         return SearchSuggestionSet::fromStrings( $results );
 602                 } else {
 603                         // Hook did not do the job, use default simple search
 604                         $results = $this->simplePrefixSearch( $search );
 605                         return SearchSuggestionSet::fromTitles( $results );
 606                 }
 607         }
 608
 609         /**
 610          * Perform a completion search.
 611          * @param string $search
 612          * @return SearchSuggestionSet
 613          */
 614         public function completionSearch( $search ) {
 615                 if ( trim( $search ) === '' ) {
 616                         return SearchSuggestionSet::emptySuggestionSet(); // Return empty result
 617                 }
 618                 $search = $this->normalizeNamespaces( $search );
 619                 $suggestions = $this->completionSearchBackendOverfetch( $search );
 620                 return $this->processCompletionResults( $search, $suggestions );
 621         }
 622
 623         /**
 624          * Perform a completion search with variants.
 625          * @param string $search
 626          * @return SearchSuggestionSet
 627          */
 628         public function completionSearchWithVariants( $search ) {
 629                 if ( trim( $search ) === '' ) {
 630                         return SearchSuggestionSet::emptySuggestionSet(); // Return empty result
 631                 }
 632                 $search = $this->normalizeNamespaces( $search );
 633
 634                 $results = $this->completionSearchBackendOverfetch( $search );
 635                 $fallbackLimit = 1 + $this->limit - $results->getSize();
 636                 if ( $fallbackLimit > 0 ) {
 637                         global $wgContLang;
 638
 639                         $fallbackSearches = $wgContLang->autoConvertToAllVariants( $search );
 640                         $fallbackSearches = array_diff( array_unique( $fallbackSearches ), [ $search ] );
 641
 642                         foreach ( $fallbackSearches as $fbs ) {
 643                                 $this->setLimitOffset( $fallbackLimit );
 644                                 $fallbackSearchResult = $this->completionSearch( $fbs );
 645                                 $results->appendAll( $fallbackSearchResult );
 646                                 $fallbackLimit -= $fallbackSearchResult->getSize();
 647                                 if ( $fallbackLimit <= 0 ) {
 648                                         break;
 649                                 }
 650                         }
 651                 }
 652                 return $this->processCompletionResults( $search, $results );
 653         }
 654
 655         /**
 656          * Extract titles from completion results
 657          * @param SearchSuggestionSet $completionResults
 658          * @return Title[]
 659          */
 660         public function extractTitles( SearchSuggestionSet $completionResults ) {
 661                 return $completionResults->map( function ( SearchSuggestion $sugg ) {
 662                         return $sugg->getSuggestedTitle();
 663                 } );
 664         }
 665
 666         /**
 667          * Process completion search results.
 668          * Resolves the titles and rescores.
 669          * @param string $search
 670          * @param SearchSuggestionSet $suggestions
 671          * @return SearchSuggestionSet
 672          */
 673         protected function processCompletionResults( $search, SearchSuggestionSet $suggestions ) {
 674                 // We over-fetched to determine pagination. Shrink back down if we have extra results
 675                 // and mark if pagination is possible
 676                 $suggestions->shrink( $this->limit );
 677
 678                 $search = trim( $search );
 679                 // preload the titles with LinkBatch
 680                 $lb = new LinkBatch( $suggestions->map( function ( SearchSuggestion $sugg ) {
 681                         return $sugg->getSuggestedTitle();
 682                 } ) );
 683                 $lb->setCaller( __METHOD__ );
 684                 $lb->execute();
 685
 686                 $diff = $suggestions->filter( function ( SearchSuggestion $sugg ) {
 687                         return $sugg->getSuggestedTitle()->isKnown();
 688                 } );
 689                 if ( $diff > 0 ) {
 690                         MediaWikiServices::getInstance()->getStatsdDataFactory()
 691                                 ->updateCount( 'search.completion.missing', $diff );
 692                 }
 693
 694                 $results = $suggestions->map( function ( SearchSuggestion $sugg ) {
 695                         return $sugg->getSuggestedTitle()->getPrefixedText();
 696                 } );
 697
 698                 if ( $this->offset === 0 ) {
 699                         // Rescore results with an exact title match
 700                         // NOTE: in some cases like cross-namespace redirects
 701                         // (frequently used as shortcuts e.g. WP:WP on huwiki) some
 702                         // backends like Cirrus will return no results. We should still
 703                         // try an exact title match to workaround this limitation
 704                         $rescorer = new SearchExactMatchRescorer();
 705                         $rescoredResults = $rescorer->rescore( $search, $this->namespaces, $results, $this->limit );
 706                 } else {
 707                         // No need to rescore if offset is not 0
 708                         // The exact match must have been returned at position 0
 709                         // if it existed.
 710                         $rescoredResults = $results;
 711                 }
 712
 713                 if ( count( $rescoredResults ) > 0 ) {
 714                         $found = array_search( $rescoredResults[0], $results );
 715                         if ( $found === false ) {
 716                                 // If the first result is not in the previous array it
 717                                 // means that we found a new exact match
 718                                 $exactMatch = SearchSuggestion::fromTitle( 0, Title::newFromText( $rescoredResults[0] ) );
 719                                 $suggestions->prepend( $exactMatch );
 720                                 $suggestions->shrink( $this->limit );
 721                         } else {
 722                                 // if the first result is not the same we need to rescore
 723                                 if ( $found > 0 ) {
 724                                         $suggestions->rescore( $found );
 725                                 }
 726                         }
 727                 }
 728
 729                 return $suggestions;
 730         }
 731
 732         /**
 733          * Simple prefix search for subpages.
 734          * @param string $search
 735          * @return Title[]
 736          */
 737         public function defaultPrefixSearch( $search ) {
 738                 if ( trim( $search ) === '' ) {
 739                         return [];
 740                 }
 741
 742                 $search = $this->normalizeNamespaces( $search );
 743                 return $this->simplePrefixSearch( $search );
 744         }
 745
 746         /**
 747          * Call out to simple search backend.
 748          * Defaults to TitlePrefixSearch.
 749          * @param string $search
 750          * @return Title[]
 751          */
 752         protected function simplePrefixSearch( $search ) {
 753                 // Use default database prefix search
 754                 $backend = new TitlePrefixSearch;
 755                 return $backend->defaultSearchBackend( $this->namespaces, $search, $this->limit, $this->offset );
 756         }
 757
 758         /**
 759          * Make a list of searchable namespaces and their canonical names.
 760          * @deprecated since 1.27; use SearchEngineConfig::searchableNamespaces()
 761          * @return array
 762          */
 763         public static function searchableNamespaces() {
 764                 return MediaWikiServices::getInstance()->getSearchEngineConfig()->searchableNamespaces();
 765         }
 766
 767         /**
 768          * Extract default namespaces to search from the given user's
 769          * settings, returning a list of index numbers.
 770          * @deprecated since 1.27; use SearchEngineConfig::userNamespaces()
 771          * @param user $user
 772          * @return array
 773          */
 774         public static function userNamespaces( $user ) {
 775                 return MediaWikiServices::getInstance()->getSearchEngineConfig()->userNamespaces( $user );
 776         }
 777
 778         /**
 779          * An array of namespaces indexes to be searched by default
 780          * @deprecated since 1.27; use SearchEngineConfig::defaultNamespaces()
 781          * @return array
 782          */
 783         public static function defaultNamespaces() {
 784                 return MediaWikiServices::getInstance()->getSearchEngineConfig()->defaultNamespaces();
 785         }
 786
 787         /**
 788          * Get a list of namespace names useful for showing in tooltips
 789          * and preferences
 790          * @deprecated since 1.27; use SearchEngineConfig::namespacesAsText()
 791          * @param array $namespaces
 792          * @return array
 793          */
 794         public static function namespacesAsText( $namespaces ) {
 795                 return MediaWikiServices::getInstance()->getSearchEngineConfig()->namespacesAsText( $namespaces );
 796         }
 797
 798         /**
 799          * Load up the appropriate search engine class for the currently
 800          * active database backend, and return a configured instance.
 801          * @deprecated since 1.27; Use SearchEngineFactory::create
 802          * @param string $type Type of search backend, if not the default
 803          * @return SearchEngine
 804          */
 805         public static function create( $type = null ) {
 806                 return MediaWikiServices::getInstance()->getSearchEngineFactory()->create( $type );
 807         }
 808
 809         /**
 810          * Return the search engines we support. If only $wgSearchType
 811          * is set, it'll be an array of just that one item.
 812          * @deprecated since 1.27; use SearchEngineConfig::getSearchTypes()
 813          * @return array
 814          */
 815         public static function getSearchTypes() {
 816                 return MediaWikiServices::getInstance()->getSearchEngineConfig()->getSearchTypes();
 817         }
 818
 819         /**
 820          * Get a list of supported profiles.
 821          * Some search engine implementations may expose specific profiles to fine-tune
 822          * its behaviors.
 823          * The profile can be passed as a feature data with setFeatureData( $profileType, $profileName )
 824          * The array returned by this function contains the following keys:
 825          * - name: the profile name to use with setFeatureData
 826          * - desc-message: the i18n description
 827          * - default: set to true if this profile is the default
 828          *
 829          * @since 1.28
 830          * @param string $profileType the type of profiles
 831          * @param User|null $user the user requesting the list of profiles
 832          * @return array|null the list of profiles or null if none available
 833          */
 834         public function getProfiles( $profileType, User $user = null ) {
 835                 return null;
 836         }
 837
 838         /**
 839          * Create a search field definition.
 840          * Specific search engines should override this method to create search fields.
 841          * @param string $name
 842          * @param int $type One of the types in SearchIndexField::INDEX_TYPE_*
 843          * @return SearchIndexField
 844          * @since 1.28
 845          */
 846         public function makeSearchFieldMapping( $name, $type ) {
 847                 return new NullIndexField();
 848         }
 849
 850         /**
 851          * Get fields for search index
 852          * @since 1.28
 853          * @return SearchIndexField[] Index field definitions for all content handlers
 854          */
 855         public function getSearchIndexFields() {
 856                 $models = ContentHandler::getContentModels();
 857                 $fields = [];
 858                 $seenHandlers = new SplObjectStorage();
 859                 foreach ( $models as $model ) {
 860                         try {
 861                                 $handler = ContentHandler::getForModelID( $model );
 862                         }
 863                         catch ( MWUnknownContentModelException $e ) {
 864                                 // If we can find no handler, ignore it
 865                                 continue;
 866                         }
 867                         // Several models can have the same handler, so avoid processing it repeatedly
 868                         if ( $seenHandlers->contains( $handler ) ) {
 869                                 // We already did this one
 870                                 continue;
 871                         }
 872                         $seenHandlers->attach( $handler );
 873                         $handlerFields = $handler->getFieldsForSearchIndex( $this );
 874                         foreach ( $handlerFields as $fieldName => $fieldData ) {
 875                                 if ( empty( $fields[$fieldName] ) ) {
 876                                         $fields[$fieldName] = $fieldData;
 877                                 } else {
 878                                         // TODO: do we allow some clashes with the same type or reject all of them?
 879                                         $mergeDef = $fields[$fieldName]->merge( $fieldData );
 880                                         if ( !$mergeDef ) {
 881                                                 throw new InvalidArgumentException( "Duplicate field $fieldName for model $model" );
 882                                         }
 883                                         $fields[$fieldName] = $mergeDef;
 884                                 }
 885                         }
 886                 }
 887                 // Hook to allow extensions to produce search mapping fields
 888                 Hooks::run( 'SearchIndexFields', [ &$fields, $this ] );
 889                 return $fields;
 890         }
 891
 892         /**
 893          * Augment search results with extra data.
 894          *
 895          * @param SearchResultSet $resultSet
 896          */
 897         public function augmentSearchResults( SearchResultSet $resultSet ) {
 898                 $setAugmentors = [];
 899                 $rowAugmentors = [];
 900                 Hooks::run( "SearchResultsAugment", [ &$setAugmentors, &$rowAugmentors ] );
 901                 if ( !$setAugmentors && !$rowAugmentors ) {
 902                         // We're done here
 903                         return;
 904                 }
 905
 906                 // Convert row augmentors to set augmentor
 907                 foreach ( $rowAugmentors as $name => $row ) {
 908                         if ( isset( $setAugmentors[$name] ) ) {
 909                                 throw new InvalidArgumentException( "Both row and set augmentors are defined for $name" );
 910                         }
 911                         $setAugmentors[$name] = new PerRowAugmentor( $row );
 912                 }
 913
 914                 foreach ( $setAugmentors as $name => $augmentor ) {
 915                         $data = $augmentor->augmentAll( $resultSet );
 916                         if ( $data ) {
 917                                 $resultSet->setAugmentedData( $name, $data );
 918                         }
 919                 }
 920         }
 921 }
 922
 923 /**
 924  * Dummy class to be used when non-supported Database engine is present.
 925  * @todo FIXME: Dummy class should probably try something at least mildly useful,
 926  * such as a LIKE search through titles.
 927  * @ingroup Search
 928  */
 929 class SearchEngineDummy extends SearchEngine {
 930         // no-op
 931 }