includes/search/SearchEngine.php

   1 <?php
   2 /**
   3  * Basic search engine
   4  *
   5  * This program is free software; you can redistribute it and/or modify
   6  * it under the terms of the GNU General Public License as published by
   7  * the Free Software Foundation; either version 2 of the License, or
   8  * (at your option) any later version.
   9  *
  10  * This program is distributed in the hope that it will be useful,
  11  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  12  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  13  * GNU General Public License for more details.
  14  *
  15  * You should have received a copy of the GNU General Public License along
  16  * with this program; if not, write to the Free Software Foundation, Inc.,
  17  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  18  * http://www.gnu.org/copyleft/gpl.html
  19  *
  20  * @file
  21  * @ingroup Search
  22  */
  23
  24 /**
  25  * @defgroup Search Search
  26  */
  27
  28 use MediaWiki\MediaWikiServices;
  29
  30 /**
  31  * Contain a class for special pages
  32  * @ingroup Search
  33  */
  34 abstract class SearchEngine {
  35         const DEFAULT_SORT = 'relevance';
  36
  37         /** @var string */
  38         public $prefix = '';
  39
  40         /** @var int[]|null */
  41         public $namespaces = [ NS_MAIN ];
  42
  43         /** @var int */
  44         protected $limit = 10;
  45
  46         /** @var int */
  47         protected $offset = 0;
  48
  49         /** @var array|string */
  50         protected $searchTerms = [];
  51
  52         /** @var bool */
  53         protected $showSuggestion = true;
  54         private $sort = self::DEFAULT_SORT;
  55
  56         /** @var array Feature values */
  57         protected $features = [];
  58
  59         /** @const string profile type for completionSearch */
  60         const COMPLETION_PROFILE_TYPE = 'completionSearchProfile';
  61
  62         /** @const string profile type for query independent ranking features */
  63         const FT_QUERY_INDEP_PROFILE_TYPE = 'fulltextQueryIndepProfile';
  64
  65         /** @const int flag for legalSearchChars: includes all chars allowed in a search query */
  66         const CHARS_ALL = 1;
  67
  68         /** @const int flag for legalSearchChars: includes all chars allowed in a search term */
  69         const CHARS_NO_SYNTAX = 2;
  70
  71         /**
  72          * Perform a full text search query and return a result set.
  73          * If full text searches are not supported or disabled, return null.
  74          *
  75          * As of 1.32 overriding this function is deprecated. It will
  76          * be converted to final in 1.34. Override self::doSearchText().
  77          *
  78          * @param string $term Raw search term
  79          * @return SearchResultSet|Status|null
  80          */
  81         public function searchText( $term ) {
  82                 return $this->maybePaginate( function () use ( $term ) {
  83                         return $this->doSearchText( $term );
  84                 } );
  85         }
  86
  87         /**
  88          * Perform a full text search query and return a result set.
  89          *
  90          * @param string $term Raw search term
  91          * @return SearchResultSet|Status|null
  92          * @since 1.32
  93          */
  94         protected function doSearchText( $term ) {
  95                 return null;
  96         }
  97
  98         /**
  99          * Perform a title search in the article archive.
 100          * NOTE: these results still should be filtered by
 101          * matching against PageArchive, permissions checks etc
 102          * The results returned by this methods are only sugegstions and
 103          * may not end up being shown to the user.
 104          *
 105          * As of 1.32 overriding this function is deprecated. It will
 106          * be converted to final in 1.34. Override self::doSearchArchiveTitle().
 107          *
 108          * @param string $term Raw search term
 109          * @return Status<Title[]>
 110          * @since 1.29
 111          */
 112         public function searchArchiveTitle( $term ) {
 113                 return $this->doSearchArchiveTitle( $term );
 114         }
 115
 116         /**
 117          * Perform a title search in the article archive.
 118          *
 119          * @param string $term Raw search term
 120          * @return Status<Title[]>
 121          * @since 1.32
 122          */
 123         protected function doSearchArchiveTitle( $term ) {
 124                 return Status::newGood( [] );
 125         }
 126
 127         /**
 128          * Perform a title-only search query and return a result set.
 129          * If title searches are not supported or disabled, return null.
 130          * STUB
 131          *
 132          * As of 1.32 overriding this function is deprecated. It will
 133          * be converted to final in 1.34. Override self::doSearchTitle().
 134          *
 135          * @param string $term Raw search term
 136          * @return SearchResultSet|null
 137          */
 138         public function searchTitle( $term ) {
 139                 return $this->maybePaginate( function () use ( $term ) {
 140                         return $this->doSearchTitle( $term );
 141                 } );
 142         }
 143
 144         /**
 145          * Perform a title-only search query and return a result set.
 146          *
 147          * @param string $term Raw search term
 148          * @return SearchResultSet|null
 149          * @since 1.32
 150          */
 151         protected function doSearchTitle( $term ) {
 152                 return null;
 153         }
 154
 155         /**
 156          * Performs an overfetch and shrink operation to determine if
 157          * the next page is available for search engines that do not
 158          * explicitly implement their own pagination.
 159          *
 160          * @param Closure $fn Takes no arguments
 161          * @return SearchResultSet|Status<SearchResultSet>|null Result of calling $fn
 162          */
 163         private function maybePaginate( Closure $fn ) {
 164                 if ( $this instanceof PaginatingSearchEngine ) {
 165                         return $fn();
 166                 }
 167                 $this->limit++;
 168                 try {
 169                         $resultSetOrStatus = $fn();
 170                 } finally {
 171                         $this->limit--;
 172                 }
 173
 174                 $resultSet = null;
 175                 if ( $resultSetOrStatus instanceof SearchResultSet ) {
 176                         $resultSet = $resultSetOrStatus;
 177                 } elseif ( $resultSetOrStatus instanceof Status &&
 178                         $resultSetOrStatus->getValue() instanceof SearchResultSet
 179                 ) {
 180                         $resultSet = $resultSetOrStatus->getValue();
 181                 }
 182                 if ( $resultSet ) {
 183                         $resultSet->shrink( $this->limit );
 184                 }
 185
 186                 return $resultSetOrStatus;
 187         }
 188
 189         /**
 190          * @since 1.18
 191          * @param string $feature
 192          * @return bool
 193          */
 194         public function supports( $feature ) {
 195                 switch ( $feature ) {
 196                         case 'search-update':
 197                                 return true;
 198                         case 'title-suffix-filter':
 199                         default:
 200                                 return false;
 201                 }
 202         }
 203
 204         /**
 205          * Way to pass custom data for engines
 206          * @since 1.18
 207          * @param string $feature
 208          * @param mixed $data
 209          */
 210         public function setFeatureData( $feature, $data ) {
 211                 $this->features[$feature] = $data;
 212         }
 213
 214         /**
 215          * Way to retrieve custom data set by setFeatureData
 216          * or by the engine itself.
 217          * @since 1.29
 218          * @param string $feature feature name
 219          * @return mixed the feature value or null if unset
 220          */
 221         public function getFeatureData( $feature ) {
 222                 return $this->features[$feature] ?? null;
 223         }
 224
 225         /**
 226          * When overridden in derived class, performs database-specific conversions
 227          * on text to be used for searching or updating search index.
 228          * Default implementation does nothing (simply returns $string).
 229          *
 230          * @param string $string String to process
 231          * @return string
 232          */
 233         public function normalizeText( $string ) {
 234                 // Some languages such as Chinese require word segmentation
 235                 return MediaWikiServices::getInstance()->getContentLanguage()->segmentByWord( $string );
 236         }
 237
 238         /**
 239          * Transform search term in cases when parts of the query came as different
 240          * GET params (when supported), e.g. for prefix queries:
 241          * search=test&prefix=Main_Page/Archive -> test prefix:Main Page/Archive
 242          * @param string $term
 243          * @return string
 244          * @deprecated since 1.32 this should now be handled internally by the
 245          * search engine
 246          */
 247         public function transformSearchTerm( $term ) {
 248                 return $term;
 249         }
 250
 251         /**
 252          * Get service class to finding near matches.
 253          * @param Config $config Configuration to use for the matcher.
 254          * @return SearchNearMatcher
 255          */
 256         public function getNearMatcher( Config $config ) {
 257                 return new SearchNearMatcher( $config,
 258                         MediaWikiServices::getInstance()->getContentLanguage() );
 259         }
 260
 261         /**
 262          * Get near matcher for default SearchEngine.
 263          * @return SearchNearMatcher
 264          */
 265         protected static function defaultNearMatcher() {
 266                 $services = MediaWikiServices::getInstance();
 267                 $config = $services->getMainConfig();
 268                 return $services->newSearchEngine()->getNearMatcher( $config );
 269         }
 270
 271         /**
 272          * If an exact title match can be found, or a very slightly close match,
 273          * return the title. If no match, returns NULL.
 274          * @deprecated since 1.27; Use SearchEngine::getNearMatcher()
 275          * @param string $searchterm
 276          * @return Title
 277          */
 278         public static function getNearMatch( $searchterm ) {
 279                 return static::defaultNearMatcher()->getNearMatch( $searchterm );
 280         }
 281
 282         /**
 283          * Do a near match (see SearchEngine::getNearMatch) and wrap it into a
 284          * SearchResultSet.
 285          * @deprecated since 1.27; Use SearchEngine::getNearMatcher()
 286          * @param string $searchterm
 287          * @return SearchResultSet
 288          */
 289         public static function getNearMatchResultSet( $searchterm ) {
 290                 wfDeprecated( __METHOD__, '1.27' );
 291                 return static::defaultNearMatcher()->getNearMatchResultSet( $searchterm );
 292         }
 293
 294         /**
 295          * Get chars legal for search
 296          * NOTE: usage as static is deprecated and preserved only as BC measure
 297          * @param int $type type of search chars (see self::CHARS_ALL
 298          * and self::CHARS_NO_SYNTAX). Defaults to CHARS_ALL
 299          * @return string
 300          */
 301         public static function legalSearchChars( $type = self::CHARS_ALL ) {
 302                 return "A-Za-z_'.0-9\\x80-\\xFF\\-";
 303         }
 304
 305         /**
 306          * Set the maximum number of results to return
 307          * and how many to skip before returning the first.
 308          *
 309          * @param int $limit
 310          * @param int $offset
 311          */
 312         function setLimitOffset( $limit, $offset = 0 ) {
 313                 $this->limit = intval( $limit );
 314                 $this->offset = intval( $offset );
 315         }
 316
 317         /**
 318          * Set which namespaces the search should include.
 319          * Give an array of namespace index numbers.
 320          *
 321          * @param int[]|null $namespaces
 322          */
 323         function setNamespaces( $namespaces ) {
 324                 if ( $namespaces ) {
 325                         // Filter namespaces to only keep valid ones
 326                         $validNs = $this->searchableNamespaces();
 327                         $namespaces = array_filter( $namespaces, function ( $ns ) use( $validNs ) {
 328                                 return $ns < 0 || isset( $validNs[$ns] );
 329                         } );
 330                 } else {
 331                         $namespaces = [];
 332                 }
 333                 $this->namespaces = $namespaces;
 334         }
 335
 336         /**
 337          * Set whether the searcher should try to build a suggestion.  Note: some searchers
 338          * don't support building a suggestion in the first place and others don't respect
 339          * this flag.
 340          *
 341          * @param bool $showSuggestion Should the searcher try to build suggestions
 342          */
 343         function setShowSuggestion( $showSuggestion ) {
 344                 $this->showSuggestion = $showSuggestion;
 345         }
 346
 347         /**
 348          * Get the valid sort directions.  All search engines support 'relevance' but others
 349          * might support more. The default in all implementations must be 'relevance.'
 350          *
 351          * @since 1.25
 352          * @return string[] the valid sort directions for setSort
 353          */
 354         public function getValidSorts() {
 355                 return [ self::DEFAULT_SORT ];
 356         }
 357
 358         /**
 359          * Set the sort direction of the search results. Must be one returned by
 360          * SearchEngine::getValidSorts()
 361          *
 362          * @since 1.25
 363          * @throws InvalidArgumentException
 364          * @param string $sort sort direction for query result
 365          */
 366         public function setSort( $sort ) {
 367                 if ( !in_array( $sort, $this->getValidSorts() ) ) {
 368                         throw new InvalidArgumentException( "Invalid sort: $sort. " .
 369                                 "Must be one of: " . implode( ', ', $this->getValidSorts() ) );
 370                 }
 371                 $this->sort = $sort;
 372         }
 373
 374         /**
 375          * Get the sort direction of the search results
 376          *
 377          * @since 1.25
 378          * @return string
 379          */
 380         public function getSort() {
 381                 return $this->sort;
 382         }
 383
 384         /**
 385          * Parse some common prefixes: all (search everything)
 386          * or namespace names and set the list of namespaces
 387          * of this class accordingly.
 388          *
 389          * @deprecated since 1.32; should be handled internally by the search engine
 390          * @param string $query
 391          * @return string
 392          */
 393         function replacePrefixes( $query ) {
 394                 return $query;
 395         }
 396
 397         /**
 398          * Parse some common prefixes: all (search everything)
 399          * or namespace names
 400          *
 401          * @param string $query
 402          * @param bool $withAllKeyword activate support of the "all:" keyword and its
 403          * translations to activate searching on all namespaces.
 404          * @param bool $withPrefixSearchExtractNamespaceHook call the PrefixSearchExtractNamespace hook
 405          *  if classic namespace identification did not match.
 406          * @return false|array false if no namespace was extracted, an array
 407          * with the parsed query at index 0 and an array of namespaces at index
 408          * 1 (or null for all namespaces).
 409          * @throws FatalError
 410          * @throws MWException
 411          */
 412         public static function parseNamespacePrefixes(
 413                 $query,
 414                 $withAllKeyword = true,
 415                 $withPrefixSearchExtractNamespaceHook = false
 416         ) {
 417                 $parsed = $query;
 418                 if ( strpos( $query, ':' ) === false ) { // nothing to do
 419                         return false;
 420                 }
 421                 $extractedNamespace = null;
 422
 423                 $allQuery = false;
 424                 if ( $withAllKeyword ) {
 425                         $allkeywords = [];
 426
 427                         $allkeywords[] = wfMessage( 'searchall' )->inContentLanguage()->text() . ":";
 428                         // force all: so that we have a common syntax for all the wikis
 429                         if ( !in_array( 'all:', $allkeywords ) ) {
 430                                 $allkeywords[] = 'all:';
 431                         }
 432
 433                         foreach ( $allkeywords as $kw ) {
 434                                 if ( strncmp( $query, $kw, strlen( $kw ) ) == 0 ) {
 435                                         $extractedNamespace = null;
 436                                         $parsed = substr( $query, strlen( $kw ) );
 437                                         $allQuery = true;
 438                                         break;
 439                                 }
 440                         }
 441                 }
 442
 443                 if ( !$allQuery && strpos( $query, ':' ) !== false ) {
 444                         $prefix = str_replace( ' ', '_', substr( $query, 0, strpos( $query, ':' ) ) );
 445                         $index = MediaWikiServices::getInstance()->getContentLanguage()->getNsIndex( $prefix );
 446                         if ( $index !== false ) {
 447                                 $extractedNamespace = [ $index ];
 448                                 $parsed = substr( $query, strlen( $prefix ) + 1 );
 449                         } elseif ( $withPrefixSearchExtractNamespaceHook ) {
 450                                 $hookNamespaces = [ NS_MAIN ];
 451                                 $hookQuery = $query;
 452                                 Hooks::run( 'PrefixSearchExtractNamespace', [ &$hookNamespaces, &$hookQuery ] );
 453                                 if ( $hookQuery !== $query ) {
 454                                         $parsed = $hookQuery;
 455                                         $extractedNamespace = $hookNamespaces;
 456                                 } else {
 457                                         return false;
 458                                 }
 459                         } else {
 460                                 return false;
 461                         }
 462                 }
 463
 464                 return [ $parsed, $extractedNamespace ];
 465         }
 466
 467         /**
 468          * Find snippet highlight settings for all users
 469          * @return array Contextlines, contextchars
 470          */
 471         public static function userHighlightPrefs() {
 472                 $contextlines = 2; // Hardcode this. Old defaults sucked. :)
 473                 $contextchars = 75; // same as above.... :P
 474                 return [ $contextlines, $contextchars ];
 475         }
 476
 477         /**
 478          * Create or update the search index record for the given page.
 479          * Title and text should be pre-processed.
 480          * STUB
 481          *
 482          * @param int $id
 483          * @param string $title
 484          * @param string $text
 485          */
 486         function update( $id, $title, $text ) {
 487                 // no-op
 488         }
 489
 490         /**
 491          * Update a search index record's title only.
 492          * Title should be pre-processed.
 493          * STUB
 494          *
 495          * @param int $id
 496          * @param string $title
 497          */
 498         function updateTitle( $id, $title ) {
 499                 // no-op
 500         }
 501
 502         /**
 503          * Delete an indexed page
 504          * Title should be pre-processed.
 505          * STUB
 506          *
 507          * @param int $id Page id that was deleted
 508          * @param string $title Title of page that was deleted
 509          */
 510         function delete( $id, $title ) {
 511                 // no-op
 512         }
 513
 514         /**
 515          * Get the raw text for updating the index from a content object
 516          * Nicer search backends could possibly do something cooler than
 517          * just returning raw text
 518          *
 519          * @todo This isn't ideal, we'd really like to have content-specific handling here
 520          * @param Title $t Title we're indexing
 521          * @param Content|null $c Content of the page to index
 522          * @return string
 523          */
 524         public function getTextFromContent( Title $t, Content $c = null ) {
 525                 return $c ? $c->getTextForSearchIndex() : '';
 526         }
 527
 528         /**
 529          * If an implementation of SearchEngine handles all of its own text processing
 530          * in getTextFromContent() and doesn't require SearchUpdate::updateText()'s
 531          * rather silly handling, it should return true here instead.
 532          *
 533          * @return bool
 534          */
 535         public function textAlreadyUpdatedForIndex() {
 536                 return false;
 537         }
 538
 539         /**
 540          * Makes search simple string if it was namespaced.
 541          * Sets namespaces of the search to namespaces extracted from string.
 542          * @param string $search
 543          * @return string Simplified search string
 544          */
 545         protected function normalizeNamespaces( $search ) {
 546                 $queryAndNs = self::parseNamespacePrefixes( $search, false, true );
 547                 if ( $queryAndNs !== false ) {
 548                         $this->setNamespaces( $queryAndNs[1] );
 549                         return $queryAndNs[0];
 550                 }
 551                 return $search;
 552         }
 553
 554         /**
 555          * Perform an overfetch of completion search results. This allows
 556          * determining if another page of results is available.
 557          *
 558          * @param string $search
 559          * @return SearchSuggestionSet
 560          */
 561         protected function completionSearchBackendOverfetch( $search ) {
 562                 $this->limit++;
 563                 try {
 564                         return $this->completionSearchBackend( $search );
 565                 } finally {
 566                         $this->limit--;
 567                 }
 568         }
 569
 570         /**
 571          * Perform a completion search.
 572          * Does not resolve namespaces and does not check variants.
 573          * Search engine implementations may want to override this function.
 574          * @param string $search
 575          * @return SearchSuggestionSet
 576          */
 577         protected function completionSearchBackend( $search ) {
 578                 $results = [];
 579
 580                 $search = trim( $search );
 581
 582                 if ( !in_array( NS_SPECIAL, $this->namespaces ) && // We do not run hook on Special: search
 583                          !Hooks::run( 'PrefixSearchBackend',
 584                                 [ $this->namespaces, $search, $this->limit, &$results, $this->offset ]
 585                 ) ) {
 586                         // False means hook worked.
 587                         // FIXME: Yes, the API is weird. That's why it is going to be deprecated.
 588
 589                         return SearchSuggestionSet::fromStrings( $results );
 590                 } else {
 591                         // Hook did not do the job, use default simple search
 592                         $results = $this->simplePrefixSearch( $search );
 593                         return SearchSuggestionSet::fromTitles( $results );
 594                 }
 595         }
 596
 597         /**
 598          * Perform a completion search.
 599          * @param string $search
 600          * @return SearchSuggestionSet
 601          */
 602         public function completionSearch( $search ) {
 603                 if ( trim( $search ) === '' ) {
 604                         return SearchSuggestionSet::emptySuggestionSet(); // Return empty result
 605                 }
 606                 $search = $this->normalizeNamespaces( $search );
 607                 $suggestions = $this->completionSearchBackendOverfetch( $search );
 608                 return $this->processCompletionResults( $search, $suggestions );
 609         }
 610
 611         /**
 612          * Perform a completion search with variants.
 613          * @param string $search
 614          * @return SearchSuggestionSet
 615          */
 616         public function completionSearchWithVariants( $search ) {
 617                 if ( trim( $search ) === '' ) {
 618                         return SearchSuggestionSet::emptySuggestionSet(); // Return empty result
 619                 }
 620                 $search = $this->normalizeNamespaces( $search );
 621
 622                 $results = $this->completionSearchBackendOverfetch( $search );
 623                 $fallbackLimit = 1 + $this->limit - $results->getSize();
 624                 if ( $fallbackLimit > 0 ) {
 625                         $fallbackSearches = MediaWikiServices::getInstance()->getContentLanguage()->
 626                                 autoConvertToAllVariants( $search );
 627                         $fallbackSearches = array_diff( array_unique( $fallbackSearches ), [ $search ] );
 628
 629                         foreach ( $fallbackSearches as $fbs ) {
 630                                 $this->setLimitOffset( $fallbackLimit );
 631                                 $fallbackSearchResult = $this->completionSearch( $fbs );
 632                                 $results->appendAll( $fallbackSearchResult );
 633                                 $fallbackLimit -= $fallbackSearchResult->getSize();
 634                                 if ( $fallbackLimit <= 0 ) {
 635                                         break;
 636                                 }
 637                         }
 638                 }
 639                 return $this->processCompletionResults( $search, $results );
 640         }
 641
 642         /**
 643          * Extract titles from completion results
 644          * @param SearchSuggestionSet $completionResults
 645          * @return Title[]
 646          */
 647         public function extractTitles( SearchSuggestionSet $completionResults ) {
 648                 return $completionResults->map( function ( SearchSuggestion $sugg ) {
 649                         return $sugg->getSuggestedTitle();
 650                 } );
 651         }
 652
 653         /**
 654          * Process completion search results.
 655          * Resolves the titles and rescores.
 656          * @param string $search
 657          * @param SearchSuggestionSet $suggestions
 658          * @return SearchSuggestionSet
 659          */
 660         protected function processCompletionResults( $search, SearchSuggestionSet $suggestions ) {
 661                 // We over-fetched to determine pagination. Shrink back down if we have extra results
 662                 // and mark if pagination is possible
 663                 $suggestions->shrink( $this->limit );
 664
 665                 $search = trim( $search );
 666                 // preload the titles with LinkBatch
 667                 $lb = new LinkBatch( $suggestions->map( function ( SearchSuggestion $sugg ) {
 668                         return $sugg->getSuggestedTitle();
 669                 } ) );
 670                 $lb->setCaller( __METHOD__ );
 671                 $lb->execute();
 672
 673                 $diff = $suggestions->filter( function ( SearchSuggestion $sugg ) {
 674                         return $sugg->getSuggestedTitle()->isKnown();
 675                 } );
 676                 if ( $diff > 0 ) {
 677                         MediaWikiServices::getInstance()->getStatsdDataFactory()
 678                                 ->updateCount( 'search.completion.missing', $diff );
 679                 }
 680
 681                 $results = $suggestions->map( function ( SearchSuggestion $sugg ) {
 682                         return $sugg->getSuggestedTitle()->getPrefixedText();
 683                 } );
 684
 685                 if ( $this->offset === 0 ) {
 686                         // Rescore results with an exact title match
 687                         // NOTE: in some cases like cross-namespace redirects
 688                         // (frequently used as shortcuts e.g. WP:WP on huwiki) some
 689                         // backends like Cirrus will return no results. We should still
 690                         // try an exact title match to workaround this limitation
 691                         $rescorer = new SearchExactMatchRescorer();
 692                         $rescoredResults = $rescorer->rescore( $search, $this->namespaces, $results, $this->limit );
 693                 } else {
 694                         // No need to rescore if offset is not 0
 695                         // The exact match must have been returned at position 0
 696                         // if it existed.
 697                         $rescoredResults = $results;
 698                 }
 699
 700                 if ( count( $rescoredResults ) > 0 ) {
 701                         $found = array_search( $rescoredResults[0], $results );
 702                         if ( $found === false ) {
 703                                 // If the first result is not in the previous array it
 704                                 // means that we found a new exact match
 705                                 $exactMatch = SearchSuggestion::fromTitle( 0, Title::newFromText( $rescoredResults[0] ) );
 706                                 $suggestions->prepend( $exactMatch );
 707                                 $suggestions->shrink( $this->limit );
 708                         } else {
 709                                 // if the first result is not the same we need to rescore
 710                                 if ( $found > 0 ) {
 711                                         $suggestions->rescore( $found );
 712                                 }
 713                         }
 714                 }
 715
 716                 return $suggestions;
 717         }
 718
 719         /**
 720          * Simple prefix search for subpages.
 721          * @param string $search
 722          * @return Title[]
 723          */
 724         public function defaultPrefixSearch( $search ) {
 725                 if ( trim( $search ) === '' ) {
 726                         return [];
 727                 }
 728
 729                 $search = $this->normalizeNamespaces( $search );
 730                 return $this->simplePrefixSearch( $search );
 731         }
 732
 733         /**
 734          * Call out to simple search backend.
 735          * Defaults to TitlePrefixSearch.
 736          * @param string $search
 737          * @return Title[]
 738          */
 739         protected function simplePrefixSearch( $search ) {
 740                 // Use default database prefix search
 741                 $backend = new TitlePrefixSearch;
 742                 return $backend->defaultSearchBackend( $this->namespaces, $search, $this->limit, $this->offset );
 743         }
 744
 745         /**
 746          * Make a list of searchable namespaces and their canonical names.
 747          * @deprecated since 1.27; use SearchEngineConfig::searchableNamespaces()
 748          * @return array
 749          */
 750         public static function searchableNamespaces() {
 751                 return MediaWikiServices::getInstance()->getSearchEngineConfig()->searchableNamespaces();
 752         }
 753
 754         /**
 755          * Extract default namespaces to search from the given user's
 756          * settings, returning a list of index numbers.
 757          * @deprecated since 1.27; use SearchEngineConfig::userNamespaces()
 758          * @param user $user
 759          * @return array
 760          */
 761         public static function userNamespaces( $user ) {
 762                 return MediaWikiServices::getInstance()->getSearchEngineConfig()->userNamespaces( $user );
 763         }
 764
 765         /**
 766          * An array of namespaces indexes to be searched by default
 767          * @deprecated since 1.27; use SearchEngineConfig::defaultNamespaces()
 768          * @return array
 769          */
 770         public static function defaultNamespaces() {
 771                 return MediaWikiServices::getInstance()->getSearchEngineConfig()->defaultNamespaces();
 772         }
 773
 774         /**
 775          * Get a list of namespace names useful for showing in tooltips
 776          * and preferences
 777          * @deprecated since 1.27; use SearchEngineConfig::namespacesAsText()
 778          * @param array $namespaces
 779          * @return array
 780          */
 781         public static function namespacesAsText( $namespaces ) {
 782                 return MediaWikiServices::getInstance()->getSearchEngineConfig()->namespacesAsText( $namespaces );
 783         }
 784
 785         /**
 786          * Load up the appropriate search engine class for the currently
 787          * active database backend, and return a configured instance.
 788          * @deprecated since 1.27; Use SearchEngineFactory::create
 789          * @param string $type Type of search backend, if not the default
 790          * @return SearchEngine
 791          */
 792         public static function create( $type = null ) {
 793                 return MediaWikiServices::getInstance()->getSearchEngineFactory()->create( $type );
 794         }
 795
 796         /**
 797          * Return the search engines we support. If only $wgSearchType
 798          * is set, it'll be an array of just that one item.
 799          * @deprecated since 1.27; use SearchEngineConfig::getSearchTypes()
 800          * @return array
 801          */
 802         public static function getSearchTypes() {
 803                 return MediaWikiServices::getInstance()->getSearchEngineConfig()->getSearchTypes();
 804         }
 805
 806         /**
 807          * Get a list of supported profiles.
 808          * Some search engine implementations may expose specific profiles to fine-tune
 809          * its behaviors.
 810          * The profile can be passed as a feature data with setFeatureData( $profileType, $profileName )
 811          * The array returned by this function contains the following keys:
 812          * - name: the profile name to use with setFeatureData
 813          * - desc-message: the i18n description
 814          * - default: set to true if this profile is the default
 815          *
 816          * @since 1.28
 817          * @param string $profileType the type of profiles
 818          * @param User|null $user the user requesting the list of profiles
 819          * @return array|null the list of profiles or null if none available
 820          */
 821         public function getProfiles( $profileType, User $user = null ) {
 822                 return null;
 823         }
 824
 825         /**
 826          * Create a search field definition.
 827          * Specific search engines should override this method to create search fields.
 828          * @param string $name
 829          * @param int $type One of the types in SearchIndexField::INDEX_TYPE_*
 830          * @return SearchIndexField
 831          * @since 1.28
 832          */
 833         public function makeSearchFieldMapping( $name, $type ) {
 834                 return new NullIndexField();
 835         }
 836
 837         /**
 838          * Get fields for search index
 839          * @since 1.28
 840          * @return SearchIndexField[] Index field definitions for all content handlers
 841          */
 842         public function getSearchIndexFields() {
 843                 $models = ContentHandler::getContentModels();
 844                 $fields = [];
 845                 $seenHandlers = new SplObjectStorage();
 846                 foreach ( $models as $model ) {
 847                         try {
 848                                 $handler = ContentHandler::getForModelID( $model );
 849                         }
 850                         catch ( MWUnknownContentModelException $e ) {
 851                                 // If we can find no handler, ignore it
 852                                 continue;
 853                         }
 854                         // Several models can have the same handler, so avoid processing it repeatedly
 855                         if ( $seenHandlers->contains( $handler ) ) {
 856                                 // We already did this one
 857                                 continue;
 858                         }
 859                         $seenHandlers->attach( $handler );
 860                         $handlerFields = $handler->getFieldsForSearchIndex( $this );
 861                         foreach ( $handlerFields as $fieldName => $fieldData ) {
 862                                 if ( empty( $fields[$fieldName] ) ) {
 863                                         $fields[$fieldName] = $fieldData;
 864                                 } else {
 865                                         // TODO: do we allow some clashes with the same type or reject all of them?
 866                                         $mergeDef = $fields[$fieldName]->merge( $fieldData );
 867                                         if ( !$mergeDef ) {
 868                                                 throw new InvalidArgumentException( "Duplicate field $fieldName for model $model" );
 869                                         }
 870                                         $fields[$fieldName] = $mergeDef;
 871                                 }
 872                         }
 873                 }
 874                 // Hook to allow extensions to produce search mapping fields
 875                 Hooks::run( 'SearchIndexFields', [ &$fields, $this ] );
 876                 return $fields;
 877         }
 878
 879         /**
 880          * Augment search results with extra data.
 881          *
 882          * @param SearchResultSet $resultSet
 883          */
 884         public function augmentSearchResults( SearchResultSet $resultSet ) {
 885                 $setAugmentors = [];
 886                 $rowAugmentors = [];
 887                 Hooks::run( "SearchResultsAugment", [ &$setAugmentors, &$rowAugmentors ] );
 888                 if ( !$setAugmentors && !$rowAugmentors ) {
 889                         // We're done here
 890                         return;
 891                 }
 892
 893                 // Convert row augmentors to set augmentor
 894                 foreach ( $rowAugmentors as $name => $row ) {
 895                         if ( isset( $setAugmentors[$name] ) ) {
 896                                 throw new InvalidArgumentException( "Both row and set augmentors are defined for $name" );
 897                         }
 898                         $setAugmentors[$name] = new PerRowAugmentor( $row );
 899                 }
 900
 901                 foreach ( $setAugmentors as $name => $augmentor ) {
 902                         $data = $augmentor->augmentAll( $resultSet );
 903                         if ( $data ) {
 904                                 $resultSet->setAugmentedData( $name, $data );
 905                         }
 906                 }
 907         }
 908 }
 909
 910 /**
 911  * Dummy class to be used when non-supported Database engine is present.
 912  * @todo FIXME: Dummy class should probably try something at least mildly useful,
 913  * such as a LIKE search through titles.
 914  * @ingroup Search
 915  */
 916 class SearchEngineDummy extends SearchEngine {
 917         // no-op
 918 }