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          */
 248         public function transformSearchTerm( $term ) {
 249                 return $term;
 250         }
 251
 252         /**
 253          * Get service class to finding near matches.
 254          * @param Config $config Configuration to use for the matcher.
 255          * @return SearchNearMatcher
 256          */
 257         public function getNearMatcher( Config $config ) {
 258                 global $wgContLang;
 259                 return new SearchNearMatcher( $config, $wgContLang );
 260         }
 261
 262         /**
 263          * Get near matcher for default SearchEngine.
 264          * @return SearchNearMatcher
 265          */
 266         protected static function defaultNearMatcher() {
 267                 $config = MediaWikiServices::getInstance()->getMainConfig();
 268                 return MediaWikiServices::getInstance()->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                 return static::defaultNearMatcher()->getNearMatchResultSet( $searchterm );
 291         }
 292
 293         /**
 294          * Get chars legal for search
 295          * NOTE: usage as static is deprecated and preserved only as BC measure
 296          * @param int $type type of search chars (see self::CHARS_ALL
 297          * and self::CHARS_NO_SYNTAX). Defaults to CHARS_ALL
 298          * @return string
 299          */
 300         public static function legalSearchChars( $type = self::CHARS_ALL ) {
 301                 return "A-Za-z_'.0-9\\x80-\\xFF\\-";
 302         }
 303
 304         /**
 305          * Set the maximum number of results to return
 306          * and how many to skip before returning the first.
 307          *
 308          * @param int $limit
 309          * @param int $offset
 310          */
 311         function setLimitOffset( $limit, $offset = 0 ) {
 312                 $this->limit = intval( $limit );
 313                 $this->offset = intval( $offset );
 314         }
 315
 316         /**
 317          * Set which namespaces the search should include.
 318          * Give an array of namespace index numbers.
 319          *
 320          * @param int[]|null $namespaces
 321          */
 322         function setNamespaces( $namespaces ) {
 323                 if ( $namespaces ) {
 324                         // Filter namespaces to only keep valid ones
 325                         $validNs = $this->searchableNamespaces();
 326                         $namespaces = array_filter( $namespaces, function ( $ns ) use( $validNs ) {
 327                                 return $ns < 0 || isset( $validNs[$ns] );
 328                         } );
 329                 } else {
 330                         $namespaces = [];
 331                 }
 332                 $this->namespaces = $namespaces;
 333         }
 334
 335         /**
 336          * Set whether the searcher should try to build a suggestion.  Note: some searchers
 337          * don't support building a suggestion in the first place and others don't respect
 338          * this flag.
 339          *
 340          * @param bool $showSuggestion Should the searcher try to build suggestions
 341          */
 342         function setShowSuggestion( $showSuggestion ) {
 343                 $this->showSuggestion = $showSuggestion;
 344         }
 345
 346         /**
 347          * Get the valid sort directions.  All search engines support 'relevance' but others
 348          * might support more. The default in all implementations should be 'relevance.'
 349          *
 350          * @since 1.25
 351          * @return string[] the valid sort directions for setSort
 352          */
 353         public function getValidSorts() {
 354                 return [ 'relevance' ];
 355         }
 356
 357         /**
 358          * Set the sort direction of the search results. Must be one returned by
 359          * SearchEngine::getValidSorts()
 360          *
 361          * @since 1.25
 362          * @throws InvalidArgumentException
 363          * @param string $sort sort direction for query result
 364          */
 365         public function setSort( $sort ) {
 366                 if ( !in_array( $sort, $this->getValidSorts() ) ) {
 367                         throw new InvalidArgumentException( "Invalid sort: $sort. " .
 368                                 "Must be one of: " . implode( ', ', $this->getValidSorts() ) );
 369                 }
 370                 $this->sort = $sort;
 371         }
 372
 373         /**
 374          * Get the sort direction of the search results
 375          *
 376          * @since 1.25
 377          * @return string
 378          */
 379         public function getSort() {
 380                 return $this->sort;
 381         }
 382
 383         /**
 384          * Parse some common prefixes: all (search everything)
 385          * or namespace names and set the list of namespaces
 386          * of this class accordingly.
 387          *
 388          * @param string $query
 389          * @return string
 390          */
 391         function replacePrefixes( $query ) {
 392                 $queryAndNs = self::parseNamespacePrefixes( $query );
 393                 if ( $queryAndNs === false ) {
 394                         return $query;
 395                 }
 396                 $this->namespaces = $queryAndNs[1];
 397                 return $queryAndNs[0];
 398         }
 399
 400         /**
 401          * Parse some common prefixes: all (search everything)
 402          * or namespace names
 403          *
 404          * @param string $query
 405          * @return false|array false if no namespace was extracted, an array
 406          * with the parsed query at index 0 and an array of namespaces at index
 407          * 1 (or null for all namespaces).
 408          */
 409         public static function parseNamespacePrefixes( $query ) {
 410                 global $wgContLang;
 411
 412                 $parsed = $query;
 413                 if ( strpos( $query, ':' ) === false ) { // nothing to do
 414                         return false;
 415                 }
 416                 $extractedNamespace = null;
 417                 $allkeywords = [];
 418
 419                 $allkeywords[] = wfMessage( 'searchall' )->inContentLanguage()->text() . ":";
 420                 // force all: so that we have a common syntax for all the wikis
 421                 if ( !in_array( 'all:', $allkeywords ) ) {
 422                         $allkeywords[] = 'all:';
 423                 }
 424
 425                 $allQuery = false;
 426                 foreach ( $allkeywords as $kw ) {
 427                         if ( strncmp( $query, $kw, strlen( $kw ) ) == 0 ) {
 428                                 $extractedNamespace = null;
 429                                 $parsed = substr( $query, strlen( $kw ) );
 430                                 $allQuery = true;
 431                                 break;
 432                         }
 433                 }
 434
 435                 if ( !$allQuery && strpos( $query, ':' ) !== false ) {
 436                         // TODO: should we unify with PrefixSearch::extractNamespace ?
 437                         $prefix = str_replace( ' ', '_', substr( $query, 0, strpos( $query, ':' ) ) );
 438                         $index = $wgContLang->getNsIndex( $prefix );
 439                         if ( $index !== false ) {
 440                                 $extractedNamespace = [ $index ];
 441                                 $parsed = substr( $query, strlen( $prefix ) + 1 );
 442                         } else {
 443                                 return false;
 444                         }
 445                 }
 446
 447                 if ( trim( $parsed ) == '' ) {
 448                         $parsed = $query; // prefix was the whole query
 449                 }
 450
 451                 return [ $parsed, $extractedNamespace ];
 452         }
 453
 454         /**
 455          * Find snippet highlight settings for all users
 456          * @return array Contextlines, contextchars
 457          */
 458         public static function userHighlightPrefs() {
 459                 $contextlines = 2; // Hardcode this. Old defaults sucked. :)
 460                 $contextchars = 75; // same as above.... :P
 461                 return [ $contextlines, $contextchars ];
 462         }
 463
 464         /**
 465          * Create or update the search index record for the given page.
 466          * Title and text should be pre-processed.
 467          * STUB
 468          *
 469          * @param int $id
 470          * @param string $title
 471          * @param string $text
 472          */
 473         function update( $id, $title, $text ) {
 474                 // no-op
 475         }
 476
 477         /**
 478          * Update a search index record's title only.
 479          * Title should be pre-processed.
 480          * STUB
 481          *
 482          * @param int $id
 483          * @param string $title
 484          */
 485         function updateTitle( $id, $title ) {
 486                 // no-op
 487         }
 488
 489         /**
 490          * Delete an indexed page
 491          * Title should be pre-processed.
 492          * STUB
 493          *
 494          * @param int $id Page id that was deleted
 495          * @param string $title Title of page that was deleted
 496          */
 497         function delete( $id, $title ) {
 498                 // no-op
 499         }
 500
 501         /**
 502          * Get the raw text for updating the index from a content object
 503          * Nicer search backends could possibly do something cooler than
 504          * just returning raw text
 505          *
 506          * @todo This isn't ideal, we'd really like to have content-specific handling here
 507          * @param Title $t Title we're indexing
 508          * @param Content|null $c Content of the page to index
 509          * @return string
 510          */
 511         public function getTextFromContent( Title $t, Content $c = null ) {
 512                 return $c ? $c->getTextForSearchIndex() : '';
 513         }
 514
 515         /**
 516          * If an implementation of SearchEngine handles all of its own text processing
 517          * in getTextFromContent() and doesn't require SearchUpdate::updateText()'s
 518          * rather silly handling, it should return true here instead.
 519          *
 520          * @return bool
 521          */
 522         public function textAlreadyUpdatedForIndex() {
 523                 return false;
 524         }
 525
 526         /**
 527          * Makes search simple string if it was namespaced.
 528          * Sets namespaces of the search to namespaces extracted from string.
 529          * @param string $search
 530          * @return string Simplified search string
 531          */
 532         protected function normalizeNamespaces( $search ) {
 533                 // Find a Title which is not an interwiki and is in NS_MAIN
 534                 $title = Title::newFromText( $search );
 535                 $ns = $this->namespaces;
 536                 if ( $title && !$title->isExternal() ) {
 537                         $ns = [ $title->getNamespace() ];
 538                         $search = $title->getText();
 539                         if ( $ns[0] == NS_MAIN ) {
 540                                 $ns = $this->namespaces; // no explicit prefix, use default namespaces
 541                                 Hooks::run( 'PrefixSearchExtractNamespace', [ &$ns, &$search ] );
 542                         }
 543                 } else {
 544                         $title = Title::newFromText( $search . 'Dummy' );
 545                         if ( $title && $title->getText() == 'Dummy'
 546                                         && $title->getNamespace() != NS_MAIN
 547                                         && !$title->isExternal()
 548                         ) {
 549                                 $ns = [ $title->getNamespace() ];
 550                                 $search = '';
 551                         } else {
 552                                 Hooks::run( 'PrefixSearchExtractNamespace', [ &$ns, &$search ] );
 553                         }
 554                 }
 555
 556                 $ns = array_map( function ( $space ) {
 557                         return $space == NS_MEDIA ? NS_FILE : $space;
 558                 }, $ns );
 559
 560                 $this->setNamespaces( $ns );
 561                 return $search;
 562         }
 563
 564         /**
 565          * Perform an overfetch of completion search results. This allows
 566          * determining if another page of results is available.
 567          *
 568          * @param string $search
 569          * @return SearchSuggestionSet
 570          */
 571         protected function completionSearchBackendOverfetch( $search ) {
 572                 $this->limit++;
 573                 try {
 574                         return $this->completionSearchBackend( $search );
 575                 } finally {
 576                         $this->limit--;
 577                 }
 578         }
 579
 580         /**
 581          * Perform a completion search.
 582          * Does not resolve namespaces and does not check variants.
 583          * Search engine implementations may want to override this function.
 584          * @param string $search
 585          * @return SearchSuggestionSet
 586          */
 587         protected function completionSearchBackend( $search ) {
 588                 $results = [];
 589
 590                 $search = trim( $search );
 591
 592                 if ( !in_array( NS_SPECIAL, $this->namespaces ) && // We do not run hook on Special: search
 593                          !Hooks::run( 'PrefixSearchBackend',
 594                                 [ $this->namespaces, $search, $this->limit, &$results, $this->offset ]
 595                 ) ) {
 596                         // False means hook worked.
 597                         // FIXME: Yes, the API is weird. That's why it is going to be deprecated.
 598
 599                         return SearchSuggestionSet::fromStrings( $results );
 600                 } else {
 601                         // Hook did not do the job, use default simple search
 602                         $results = $this->simplePrefixSearch( $search );
 603                         return SearchSuggestionSet::fromTitles( $results );
 604                 }
 605         }
 606
 607         /**
 608          * Perform a completion search.
 609          * @param string $search
 610          * @return SearchSuggestionSet
 611          */
 612         public function completionSearch( $search ) {
 613                 if ( trim( $search ) === '' ) {
 614                         return SearchSuggestionSet::emptySuggestionSet(); // Return empty result
 615                 }
 616                 $search = $this->normalizeNamespaces( $search );
 617                 $suggestions = $this->completionSearchBackendOverfetch( $search );
 618                 return $this->processCompletionResults( $search, $suggestions );
 619         }
 620
 621         /**
 622          * Perform a completion search with variants.
 623          * @param string $search
 624          * @return SearchSuggestionSet
 625          */
 626         public function completionSearchWithVariants( $search ) {
 627                 if ( trim( $search ) === '' ) {
 628                         return SearchSuggestionSet::emptySuggestionSet(); // Return empty result
 629                 }
 630                 $search = $this->normalizeNamespaces( $search );
 631
 632                 $results = $this->completionSearchBackendOverfetch( $search );
 633                 $fallbackLimit = 1 + $this->limit - $results->getSize();
 634                 if ( $fallbackLimit > 0 ) {
 635                         global $wgContLang;
 636
 637                         $fallbackSearches = $wgContLang->autoConvertToAllVariants( $search );
 638                         $fallbackSearches = array_diff( array_unique( $fallbackSearches ), [ $search ] );
 639
 640                         foreach ( $fallbackSearches as $fbs ) {
 641                                 $this->setLimitOffset( $fallbackLimit );
 642                                 $fallbackSearchResult = $this->completionSearch( $fbs );
 643                                 $results->appendAll( $fallbackSearchResult );
 644                                 $fallbackLimit -= $fallbackSearchResult->getSize();
 645                                 if ( $fallbackLimit <= 0 ) {
 646                                         break;
 647                                 }
 648                         }
 649                 }
 650                 return $this->processCompletionResults( $search, $results );
 651         }
 652
 653         /**
 654          * Extract titles from completion results
 655          * @param SearchSuggestionSet $completionResults
 656          * @return Title[]
 657          */
 658         public function extractTitles( SearchSuggestionSet $completionResults ) {
 659                 return $completionResults->map( function ( SearchSuggestion $sugg ) {
 660                         return $sugg->getSuggestedTitle();
 661                 } );
 662         }
 663
 664         /**
 665          * Process completion search results.
 666          * Resolves the titles and rescores.
 667          * @param string $search
 668          * @param SearchSuggestionSet $suggestions
 669          * @return SearchSuggestionSet
 670          */
 671         protected function processCompletionResults( $search, SearchSuggestionSet $suggestions ) {
 672                 // We over-fetched to determine pagination. Shrink back down if we have extra results
 673                 // and mark if pagination is possible
 674                 $suggestions->shrink( $this->limit );
 675
 676                 $search = trim( $search );
 677                 // preload the titles with LinkBatch
 678                 $lb = new LinkBatch( $suggestions->map( function ( SearchSuggestion $sugg ) {
 679                         return $sugg->getSuggestedTitle();
 680                 } ) );
 681                 $lb->setCaller( __METHOD__ );
 682                 $lb->execute();
 683
 684                 $diff = $suggestions->filter( function ( SearchSuggestion $sugg ) {
 685                         return $sugg->getSuggestedTitle()->isKnown();
 686                 } );
 687                 if ( $diff > 0 ) {
 688                         MediaWikiServices::getInstance()->getStatsdDataFactory()
 689                                 ->updateCount( 'search.completion.missing', $diff );
 690                 }
 691
 692                 $results = $suggestions->map( function ( SearchSuggestion $sugg ) {
 693                         return $sugg->getSuggestedTitle()->getPrefixedText();
 694                 } );
 695
 696                 if ( $this->offset === 0 ) {
 697                         // Rescore results with an exact title match
 698                         // NOTE: in some cases like cross-namespace redirects
 699                         // (frequently used as shortcuts e.g. WP:WP on huwiki) some
 700                         // backends like Cirrus will return no results. We should still
 701                         // try an exact title match to workaround this limitation
 702                         $rescorer = new SearchExactMatchRescorer();
 703                         $rescoredResults = $rescorer->rescore( $search, $this->namespaces, $results, $this->limit );
 704                 } else {
 705                         // No need to rescore if offset is not 0
 706                         // The exact match must have been returned at position 0
 707                         // if it existed.
 708                         $rescoredResults = $results;
 709                 }
 710
 711                 if ( count( $rescoredResults ) > 0 ) {
 712                         $found = array_search( $rescoredResults[0], $results );
 713                         if ( $found === false ) {
 714                                 // If the first result is not in the previous array it
 715                                 // means that we found a new exact match
 716                                 $exactMatch = SearchSuggestion::fromTitle( 0, Title::newFromText( $rescoredResults[0] ) );
 717                                 $suggestions->prepend( $exactMatch );
 718                                 $suggestions->shrink( $this->limit );
 719                         } else {
 720                                 // if the first result is not the same we need to rescore
 721                                 if ( $found > 0 ) {
 722                                         $suggestions->rescore( $found );
 723                                 }
 724                         }
 725                 }
 726
 727                 return $suggestions;
 728         }
 729
 730         /**
 731          * Simple prefix search for subpages.
 732          * @param string $search
 733          * @return Title[]
 734          */
 735         public function defaultPrefixSearch( $search ) {
 736                 if ( trim( $search ) === '' ) {
 737                         return [];
 738                 }
 739
 740                 $search = $this->normalizeNamespaces( $search );
 741                 return $this->simplePrefixSearch( $search );
 742         }
 743
 744         /**
 745          * Call out to simple search backend.
 746          * Defaults to TitlePrefixSearch.
 747          * @param string $search
 748          * @return Title[]
 749          */
 750         protected function simplePrefixSearch( $search ) {
 751                 // Use default database prefix search
 752                 $backend = new TitlePrefixSearch;
 753                 return $backend->defaultSearchBackend( $this->namespaces, $search, $this->limit, $this->offset );
 754         }
 755
 756         /**
 757          * Make a list of searchable namespaces and their canonical names.
 758          * @deprecated since 1.27; use SearchEngineConfig::searchableNamespaces()
 759          * @return array
 760          */
 761         public static function searchableNamespaces() {
 762                 return MediaWikiServices::getInstance()->getSearchEngineConfig()->searchableNamespaces();
 763         }
 764
 765         /**
 766          * Extract default namespaces to search from the given user's
 767          * settings, returning a list of index numbers.
 768          * @deprecated since 1.27; use SearchEngineConfig::userNamespaces()
 769          * @param user $user
 770          * @return array
 771          */
 772         public static function userNamespaces( $user ) {
 773                 return MediaWikiServices::getInstance()->getSearchEngineConfig()->userNamespaces( $user );
 774         }
 775
 776         /**
 777          * An array of namespaces indexes to be searched by default
 778          * @deprecated since 1.27; use SearchEngineConfig::defaultNamespaces()
 779          * @return array
 780          */
 781         public static function defaultNamespaces() {
 782                 return MediaWikiServices::getInstance()->getSearchEngineConfig()->defaultNamespaces();
 783         }
 784
 785         /**
 786          * Get a list of namespace names useful for showing in tooltips
 787          * and preferences
 788          * @deprecated since 1.27; use SearchEngineConfig::namespacesAsText()
 789          * @param array $namespaces
 790          * @return array
 791          */
 792         public static function namespacesAsText( $namespaces ) {
 793                 return MediaWikiServices::getInstance()->getSearchEngineConfig()->namespacesAsText( $namespaces );
 794         }
 795
 796         /**
 797          * Load up the appropriate search engine class for the currently
 798          * active database backend, and return a configured instance.
 799          * @deprecated since 1.27; Use SearchEngineFactory::create
 800          * @param string $type Type of search backend, if not the default
 801          * @return SearchEngine
 802          */
 803         public static function create( $type = null ) {
 804                 return MediaWikiServices::getInstance()->getSearchEngineFactory()->create( $type );
 805         }
 806
 807         /**
 808          * Return the search engines we support. If only $wgSearchType
 809          * is set, it'll be an array of just that one item.
 810          * @deprecated since 1.27; use SearchEngineConfig::getSearchTypes()
 811          * @return array
 812          */
 813         public static function getSearchTypes() {
 814                 return MediaWikiServices::getInstance()->getSearchEngineConfig()->getSearchTypes();
 815         }
 816
 817         /**
 818          * Get a list of supported profiles.
 819          * Some search engine implementations may expose specific profiles to fine-tune
 820          * its behaviors.
 821          * The profile can be passed as a feature data with setFeatureData( $profileType, $profileName )
 822          * The array returned by this function contains the following keys:
 823          * - name: the profile name to use with setFeatureData
 824          * - desc-message: the i18n description
 825          * - default: set to true if this profile is the default
 826          *
 827          * @since 1.28
 828          * @param string $profileType the type of profiles
 829          * @param User|null $user the user requesting the list of profiles
 830          * @return array|null the list of profiles or null if none available
 831          */
 832         public function getProfiles( $profileType, User $user = null ) {
 833                 return null;
 834         }
 835
 836         /**
 837          * Create a search field definition.
 838          * Specific search engines should override this method to create search fields.
 839          * @param string $name
 840          * @param int $type One of the types in SearchIndexField::INDEX_TYPE_*
 841          * @return SearchIndexField
 842          * @since 1.28
 843          */
 844         public function makeSearchFieldMapping( $name, $type ) {
 845                 return new NullIndexField();
 846         }
 847
 848         /**
 849          * Get fields for search index
 850          * @since 1.28
 851          * @return SearchIndexField[] Index field definitions for all content handlers
 852          */
 853         public function getSearchIndexFields() {
 854                 $models = ContentHandler::getContentModels();
 855                 $fields = [];
 856                 $seenHandlers = new SplObjectStorage();
 857                 foreach ( $models as $model ) {
 858                         try {
 859                                 $handler = ContentHandler::getForModelID( $model );
 860                         }
 861                         catch ( MWUnknownContentModelException $e ) {
 862                                 // If we can find no handler, ignore it
 863                                 continue;
 864                         }
 865                         // Several models can have the same handler, so avoid processing it repeatedly
 866                         if ( $seenHandlers->contains( $handler ) ) {
 867                                 // We already did this one
 868                                 continue;
 869                         }
 870                         $seenHandlers->attach( $handler );
 871                         $handlerFields = $handler->getFieldsForSearchIndex( $this );
 872                         foreach ( $handlerFields as $fieldName => $fieldData ) {
 873                                 if ( empty( $fields[$fieldName] ) ) {
 874                                         $fields[$fieldName] = $fieldData;
 875                                 } else {
 876                                         // TODO: do we allow some clashes with the same type or reject all of them?
 877                                         $mergeDef = $fields[$fieldName]->merge( $fieldData );
 878                                         if ( !$mergeDef ) {
 879                                                 throw new InvalidArgumentException( "Duplicate field $fieldName for model $model" );
 880                                         }
 881                                         $fields[$fieldName] = $mergeDef;
 882                                 }
 883                         }
 884                 }
 885                 // Hook to allow extensions to produce search mapping fields
 886                 Hooks::run( 'SearchIndexFields', [ &$fields, $this ] );
 887                 return $fields;
 888         }
 889
 890         /**
 891          * Augment search results with extra data.
 892          *
 893          * @param SearchResultSet $resultSet
 894          */
 895         public function augmentSearchResults( SearchResultSet $resultSet ) {
 896                 $setAugmentors = [];
 897                 $rowAugmentors = [];
 898                 Hooks::run( "SearchResultsAugment", [ &$setAugmentors, &$rowAugmentors ] );
 899                 if ( !$setAugmentors && !$rowAugmentors ) {
 900                         // We're done here
 901                         return;
 902                 }
 903
 904                 // Convert row augmentors to set augmentor
 905                 foreach ( $rowAugmentors as $name => $row ) {
 906                         if ( isset( $setAugmentors[$name] ) ) {
 907                                 throw new InvalidArgumentException( "Both row and set augmentors are defined for $name" );
 908                         }
 909                         $setAugmentors[$name] = new PerRowAugmentor( $row );
 910                 }
 911
 912                 foreach ( $setAugmentors as $name => $augmentor ) {
 913                         $data = $augmentor->augmentAll( $resultSet );
 914                         if ( $data ) {
 915                                 $resultSet->setAugmentedData( $name, $data );
 916                         }
 917                 }
 918         }
 919 }
 920
 921 /**
 922  * Dummy class to be used when non-supported Database engine is present.
 923  * @todo FIXME: Dummy class should probably try something at least mildly useful,
 924  * such as a LIKE search through titles.
 925  * @ingroup Search
 926  */
 927 class SearchEngineDummy extends SearchEngine {
 928         // no-op
 929 }