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         /**
  58          * Perform a full text search query and return a result set.
  59          * If full text searches are not supported or disabled, return null.
  60          * STUB
  61          *
  62          * @param string $term Raw search term
  63          * @return SearchResultSet|Status|null
  64          */
  65         function searchText( $term ) {
  66                 return null;
  67         }
  68
  69         /**
  70          * Perform a title-only search query and return a result set.
  71          * If title searches are not supported or disabled, return null.
  72          * STUB
  73          *
  74          * @param string $term Raw search term
  75          * @return SearchResultSet|null
  76          */
  77         function searchTitle( $term ) {
  78                 return null;
  79         }
  80
  81         /**
  82          * @since 1.18
  83          * @param string $feature
  84          * @return bool
  85          */
  86         public function supports( $feature ) {
  87                 switch ( $feature ) {
  88                 case 'search-update':
  89                         return true;
  90                 case 'title-suffix-filter':
  91                 default:
  92                         return false;
  93                 }
  94         }
  95
  96         /**
  97          * Way to pass custom data for engines
  98          * @since 1.18
  99          * @param string $feature
 100          * @param mixed $data
 101          * @return bool
 102          */
 103         public function setFeatureData( $feature, $data ) {
 104                 $this->features[$feature] = $data;
 105         }
 106
 107         /**
 108          * When overridden in derived class, performs database-specific conversions
 109          * on text to be used for searching or updating search index.
 110          * Default implementation does nothing (simply returns $string).
 111          *
 112          * @param string $string String to process
 113          * @return string
 114          */
 115         public function normalizeText( $string ) {
 116                 global $wgContLang;
 117
 118                 // Some languages such as Chinese require word segmentation
 119                 return $wgContLang->segmentByWord( $string );
 120         }
 121
 122         /**
 123          * Transform search term in cases when parts of the query came as different
 124          * GET params (when supported), e.g. for prefix queries:
 125          * search=test&prefix=Main_Page/Archive -> test prefix:Main Page/Archive
 126          * @param string $term
 127          * @return string
 128          */
 129         public function transformSearchTerm( $term ) {
 130                 return $term;
 131         }
 132
 133         /**
 134          * Get service class to finding near matches.
 135          * @param Config $config Configuration to use for the matcher.
 136          * @return SearchNearMatcher
 137          */
 138         public function getNearMatcher( Config $config ) {
 139                 global $wgContLang;
 140                 return new SearchNearMatcher( $config, $wgContLang );
 141         }
 142
 143         /**
 144          * Get near matcher for default SearchEngine.
 145          * @return SearchNearMatcher
 146          */
 147         protected static function defaultNearMatcher() {
 148                 $config = MediaWikiServices::getInstance()->getMainConfig();
 149                 return MediaWikiServices::getInstance()->newSearchEngine()->getNearMatcher( $config );
 150         }
 151
 152         /**
 153          * If an exact title match can be found, or a very slightly close match,
 154          * return the title. If no match, returns NULL.
 155          * @deprecated since 1.27; Use SearchEngine::getNearMatcher()
 156          * @param string $searchterm
 157          * @return Title
 158          */
 159         public static function getNearMatch( $searchterm ) {
 160                 return static::defaultNearMatcher()->getNearMatch( $searchterm );
 161         }
 162
 163         /**
 164          * Do a near match (see SearchEngine::getNearMatch) and wrap it into a
 165          * SearchResultSet.
 166          * @deprecated since 1.27; Use SearchEngine::getNearMatcher()
 167          * @param string $searchterm
 168          * @return SearchResultSet
 169          */
 170         public static function getNearMatchResultSet( $searchterm ) {
 171                 return static::defaultNearMatcher()->getNearMatchResultSet( $searchterm );
 172         }
 173
 174         /**
 175          * Get chars legal for search.
 176          * NOTE: usage as static is deprecated and preserved only as BC measure
 177          * @return string
 178          */
 179         public static function legalSearchChars() {
 180                 return "A-Za-z_'.0-9\\x80-\\xFF\\-";
 181         }
 182
 183         /**
 184          * Set the maximum number of results to return
 185          * and how many to skip before returning the first.
 186          *
 187          * @param int $limit
 188          * @param int $offset
 189          */
 190         function setLimitOffset( $limit, $offset = 0 ) {
 191                 $this->limit = intval( $limit );
 192                 $this->offset = intval( $offset );
 193         }
 194
 195         /**
 196          * Set which namespaces the search should include.
 197          * Give an array of namespace index numbers.
 198          *
 199          * @param int[]|null $namespaces
 200          */
 201         function setNamespaces( $namespaces ) {
 202                 if ( $namespaces ) {
 203                         // Filter namespaces to only keep valid ones
 204                         $validNs = $this->searchableNamespaces();
 205                         $namespaces = array_filter( $namespaces, function( $ns ) use( $validNs ) {
 206                                 return $ns < 0 || isset( $validNs[$ns] );
 207                         } );
 208                 } else {
 209                         $namespaces = [];
 210                 }
 211                 $this->namespaces = $namespaces;
 212         }
 213
 214         /**
 215          * Set whether the searcher should try to build a suggestion.  Note: some searchers
 216          * don't support building a suggestion in the first place and others don't respect
 217          * this flag.
 218          *
 219          * @param bool $showSuggestion Should the searcher try to build suggestions
 220          */
 221         function setShowSuggestion( $showSuggestion ) {
 222                 $this->showSuggestion = $showSuggestion;
 223         }
 224
 225         /**
 226          * Get the valid sort directions.  All search engines support 'relevance' but others
 227          * might support more. The default in all implementations should be 'relevance.'
 228          *
 229          * @since 1.25
 230          * @return array(string) the valid sort directions for setSort
 231          */
 232         public function getValidSorts() {
 233                 return [ 'relevance' ];
 234         }
 235
 236         /**
 237          * Set the sort direction of the search results. Must be one returned by
 238          * SearchEngine::getValidSorts()
 239          *
 240          * @since 1.25
 241          * @throws InvalidArgumentException
 242          * @param string $sort sort direction for query result
 243          */
 244         public function setSort( $sort ) {
 245                 if ( !in_array( $sort, $this->getValidSorts() ) ) {
 246                         throw new InvalidArgumentException( "Invalid sort: $sort. " .
 247                                 "Must be one of: " . implode( ', ', $this->getValidSorts() ) );
 248                 }
 249                 $this->sort = $sort;
 250         }
 251
 252         /**
 253          * Get the sort direction of the search results
 254          *
 255          * @since 1.25
 256          * @return string
 257          */
 258         public function getSort() {
 259                 return $this->sort;
 260         }
 261
 262         /**
 263          * Parse some common prefixes: all (search everything)
 264          * or namespace names
 265          *
 266          * @param string $query
 267          * @return string
 268          */
 269         function replacePrefixes( $query ) {
 270                 global $wgContLang;
 271
 272                 $parsed = $query;
 273                 if ( strpos( $query, ':' ) === false ) { // nothing to do
 274                         return $parsed;
 275                 }
 276
 277                 $allkeyword = wfMessage( 'searchall' )->inContentLanguage()->text() . ":";
 278                 if ( strncmp( $query, $allkeyword, strlen( $allkeyword ) ) == 0 ) {
 279                         $this->namespaces = null;
 280                         $parsed = substr( $query, strlen( $allkeyword ) );
 281                 } elseif ( strpos( $query, ':' ) !== false ) {
 282                         $prefix = str_replace( ' ', '_', substr( $query, 0, strpos( $query, ':' ) ) );
 283                         $index = $wgContLang->getNsIndex( $prefix );
 284                         if ( $index !== false ) {
 285                                 $this->namespaces = [ $index ];
 286                                 $parsed = substr( $query, strlen( $prefix ) + 1 );
 287                         }
 288                 }
 289                 if ( trim( $parsed ) == '' ) {
 290                         $parsed = $query; // prefix was the whole query
 291                 }
 292
 293                 return $parsed;
 294         }
 295
 296         /**
 297          * Find snippet highlight settings for all users
 298          * @return array Contextlines, contextchars
 299          */
 300         public static function userHighlightPrefs() {
 301                 $contextlines = 2; // Hardcode this. Old defaults sucked. :)
 302                 $contextchars = 75; // same as above.... :P
 303                 return [ $contextlines, $contextchars ];
 304         }
 305
 306         /**
 307          * Create or update the search index record for the given page.
 308          * Title and text should be pre-processed.
 309          * STUB
 310          *
 311          * @param int $id
 312          * @param string $title
 313          * @param string $text
 314          */
 315         function update( $id, $title, $text ) {
 316                 // no-op
 317         }
 318
 319         /**
 320          * Update a search index record's title only.
 321          * Title should be pre-processed.
 322          * STUB
 323          *
 324          * @param int $id
 325          * @param string $title
 326          */
 327         function updateTitle( $id, $title ) {
 328                 // no-op
 329         }
 330
 331         /**
 332          * Delete an indexed page
 333          * Title should be pre-processed.
 334          * STUB
 335          *
 336          * @param int $id Page id that was deleted
 337          * @param string $title Title of page that was deleted
 338          */
 339         function delete( $id, $title ) {
 340                 // no-op
 341         }
 342
 343         /**
 344          * Get OpenSearch suggestion template
 345          *
 346          * @deprecated since 1.25
 347          * @return string
 348          */
 349         public static function getOpenSearchTemplate() {
 350                 wfDeprecated( __METHOD__, '1.25' );
 351                 return ApiOpenSearch::getOpenSearchTemplate( 'application/x-suggestions+json' );
 352         }
 353
 354         /**
 355          * Get the raw text for updating the index from a content object
 356          * Nicer search backends could possibly do something cooler than
 357          * just returning raw text
 358          *
 359          * @todo This isn't ideal, we'd really like to have content-specific handling here
 360          * @param Title $t Title we're indexing
 361          * @param Content $c Content of the page to index
 362          * @return string
 363          */
 364         public function getTextFromContent( Title $t, Content $c = null ) {
 365                 return $c ? $c->getTextForSearchIndex() : '';
 366         }
 367
 368         /**
 369          * If an implementation of SearchEngine handles all of its own text processing
 370          * in getTextFromContent() and doesn't require SearchUpdate::updateText()'s
 371          * rather silly handling, it should return true here instead.
 372          *
 373          * @return bool
 374          */
 375         public function textAlreadyUpdatedForIndex() {
 376                 return false;
 377         }
 378
 379         /**
 380          * Makes search simple string if it was namespaced.
 381          * Sets namespaces of the search to namespaces extracted from string.
 382          * @param string $search
 383          * @return $string Simplified search string
 384          */
 385         protected function normalizeNamespaces( $search ) {
 386                 // Find a Title which is not an interwiki and is in NS_MAIN
 387                 $title = Title::newFromText( $search );
 388                 $ns = $this->namespaces;
 389                 if ( $title && !$title->isExternal() ) {
 390                         $ns = [ $title->getNamespace() ];
 391                         $search = $title->getText();
 392                         if ( $ns[0] == NS_MAIN ) {
 393                                 $ns = $this->namespaces; // no explicit prefix, use default namespaces
 394                                 Hooks::run( 'PrefixSearchExtractNamespace', [ &$ns, &$search ] );
 395                         }
 396                 } else {
 397                         $title = Title::newFromText( $search . 'Dummy' );
 398                         if ( $title && $title->getText() == 'Dummy'
 399                                         && $title->getNamespace() != NS_MAIN
 400                                         && !$title->isExternal() )
 401                         {
 402                                 $ns = [ $title->getNamespace() ];
 403                                 $search = '';
 404                         } else {
 405                                 Hooks::run( 'PrefixSearchExtractNamespace', [ &$ns, &$search ] );
 406                         }
 407                 }
 408
 409                 $ns = array_map( function( $space ) {
 410                         return $space == NS_MEDIA ? NS_FILE : $space;
 411                 }, $ns );
 412
 413                 $this->setNamespaces( $ns );
 414                 return $search;
 415         }
 416
 417         /**
 418          * Perform a completion search.
 419          * Does not resolve namespaces and does not check variants.
 420          * Search engine implementations may want to override this function.
 421          * @param string $search
 422          * @return SearchSuggestionSet
 423          */
 424         protected function completionSearchBackend( $search ) {
 425                 $results = [];
 426
 427                 $search = trim( $search );
 428
 429                 if ( !in_array( NS_SPECIAL, $this->namespaces ) && // We do not run hook on Special: search
 430                          !Hooks::run( 'PrefixSearchBackend',
 431                                 [ $this->namespaces, $search, $this->limit, &$results, $this->offset ]
 432                 ) ) {
 433                         // False means hook worked.
 434                         // FIXME: Yes, the API is weird. That's why it is going to be deprecated.
 435
 436                         return SearchSuggestionSet::fromStrings( $results );
 437                 } else {
 438                         // Hook did not do the job, use default simple search
 439                         $results = $this->simplePrefixSearch( $search );
 440                         return SearchSuggestionSet::fromTitles( $results );
 441                 }
 442         }
 443
 444         /**
 445          * Perform a completion search.
 446          * @param string $search
 447          * @return SearchSuggestionSet
 448          */
 449         public function completionSearch( $search ) {
 450                 if ( trim( $search ) === '' ) {
 451                         return SearchSuggestionSet::emptySuggestionSet(); // Return empty result
 452                 }
 453                 $search = $this->normalizeNamespaces( $search );
 454                 return $this->processCompletionResults( $search, $this->completionSearchBackend( $search ) );
 455         }
 456
 457         /**
 458          * Perform a completion search with variants.
 459          * @param string $search
 460          * @return SearchSuggestionSet
 461          */
 462         public function completionSearchWithVariants( $search ) {
 463                 if ( trim( $search ) === '' ) {
 464                         return SearchSuggestionSet::emptySuggestionSet(); // Return empty result
 465                 }
 466                 $search = $this->normalizeNamespaces( $search );
 467
 468                 $results = $this->completionSearchBackend( $search );
 469                 $fallbackLimit = $this->limit - $results->getSize();
 470                 if ( $fallbackLimit > 0 ) {
 471                         global $wgContLang;
 472
 473                         $fallbackSearches = $wgContLang->autoConvertToAllVariants( $search );
 474                         $fallbackSearches = array_diff( array_unique( $fallbackSearches ), [ $search ] );
 475
 476                         foreach ( $fallbackSearches as $fbs ) {
 477                                 $this->setLimitOffset( $fallbackLimit );
 478                                 $fallbackSearchResult = $this->completionSearch( $fbs );
 479                                 $results->appendAll( $fallbackSearchResult );
 480                                 $fallbackLimit -= count( $fallbackSearchResult );
 481                                 if ( $fallbackLimit <= 0 ) {
 482                                         break;
 483                                 }
 484                         }
 485                 }
 486                 return $this->processCompletionResults( $search, $results );
 487         }
 488
 489         /**
 490          * Extract titles from completion results
 491          * @param SearchSuggestionSet $completionResults
 492          * @return Title[]
 493          */
 494         public function extractTitles( SearchSuggestionSet $completionResults ) {
 495                 return $completionResults->map( function( SearchSuggestion $sugg ) {
 496                         return $sugg->getSuggestedTitle();
 497                 } );
 498         }
 499
 500         /**
 501          * Process completion search results.
 502          * Resolves the titles and rescores.
 503          * @param SearchSuggestionSet $suggestions
 504          * @return SearchSuggestionSet
 505          */
 506         protected function processCompletionResults( $search, SearchSuggestionSet $suggestions ) {
 507                 $search = trim( $search );
 508                 // preload the titles with LinkBatch
 509                 $titles = $suggestions->map( function( SearchSuggestion $sugg ) {
 510                         return $sugg->getSuggestedTitle();
 511                 } );
 512                 $lb = new LinkBatch( $titles );
 513                 $lb->setCaller( __METHOD__ );
 514                 $lb->execute();
 515
 516                 $results = $suggestions->map( function( SearchSuggestion $sugg ) {
 517                         return $sugg->getSuggestedTitle()->getPrefixedText();
 518                 } );
 519
 520                 // Rescore results with an exact title match
 521                 // NOTE: in some cases like cross-namespace redirects
 522                 // (frequently used as shortcuts e.g. WP:WP on huwiki) some
 523                 // backends like Cirrus will return no results. We should still
 524                 // try an exact title match to workaround this limitation
 525                 $rescorer = new SearchExactMatchRescorer();
 526                 $rescoredResults = $rescorer->rescore( $search, $this->namespaces, $results, $this->limit );
 527
 528                 if ( count( $rescoredResults ) > 0 ) {
 529                         $found = array_search( $rescoredResults[0], $results );
 530                         if ( $found === false ) {
 531                                 // If the first result is not in the previous array it
 532                                 // means that we found a new exact match
 533                                 $exactMatch = SearchSuggestion::fromTitle( 0, Title::newFromText( $rescoredResults[0] ) );
 534                                 $suggestions->prepend( $exactMatch );
 535                                 $suggestions->shrink( $this->limit );
 536                         } else {
 537                                 // if the first result is not the same we need to rescore
 538                                 if ( $found > 0 ) {
 539                                         $suggestions->rescore( $found );
 540                                 }
 541                         }
 542                 }
 543
 544                 return $suggestions;
 545         }
 546
 547         /**
 548          * Simple prefix search for subpages.
 549          * @param string $search
 550          * @return Title[]
 551          */
 552         public function defaultPrefixSearch( $search ) {
 553                 if ( trim( $search ) === '' ) {
 554                         return [];
 555                 }
 556
 557                 $search = $this->normalizeNamespaces( $search );
 558                 return $this->simplePrefixSearch( $search );
 559         }
 560
 561         /**
 562          * Call out to simple search backend.
 563          * Defaults to TitlePrefixSearch.
 564          * @param string $search
 565          * @return Title[]
 566          */
 567         protected function simplePrefixSearch( $search ) {
 568                 // Use default database prefix search
 569                 $backend = new TitlePrefixSearch;
 570                 return $backend->defaultSearchBackend( $this->namespaces, $search, $this->limit, $this->offset );
 571         }
 572
 573         /**
 574          * Make a list of searchable namespaces and their canonical names.
 575          * @deprecated since 1.27; use SearchEngineConfig::searchableNamespaces()
 576          * @return array
 577          */
 578         public static function searchableNamespaces() {
 579                 return MediaWikiServices::getInstance()->getSearchEngineConfig()->searchableNamespaces();
 580         }
 581
 582         /**
 583          * Extract default namespaces to search from the given user's
 584          * settings, returning a list of index numbers.
 585          * @deprecated since 1.27; use SearchEngineConfig::userNamespaces()
 586          * @param user $user
 587          * @return array
 588          */
 589         public static function userNamespaces( $user ) {
 590                 return MediaWikiServices::getInstance()->getSearchEngineConfig()->userNamespaces( $user );
 591         }
 592
 593         /**
 594          * An array of namespaces indexes to be searched by default
 595          * @deprecated since 1.27; use SearchEngineConfig::defaultNamespaces()
 596          * @return array
 597          */
 598         public static function defaultNamespaces() {
 599                 return MediaWikiServices::getInstance()->getSearchEngineConfig()->defaultNamespaces();
 600         }
 601
 602         /**
 603          * Get a list of namespace names useful for showing in tooltips
 604          * and preferences
 605          * @deprecated since 1.27; use SearchEngineConfig::namespacesAsText()
 606          * @param array $namespaces
 607          * @return array
 608          */
 609         public static function namespacesAsText( $namespaces ) {
 610                 return MediaWikiServices::getInstance()->getSearchEngineConfig()->namespacesAsText();
 611         }
 612
 613         /**
 614          * Load up the appropriate search engine class for the currently
 615          * active database backend, and return a configured instance.
 616          * @deprecated since 1.27; Use SearchEngineFactory::create
 617          * @param string $type Type of search backend, if not the default
 618          * @return SearchEngine
 619          */
 620         public static function create( $type = null ) {
 621                 return MediaWikiServices::getInstance()->getSearchEngineFactory()->create( $type );
 622         }
 623
 624         /**
 625          * Return the search engines we support. If only $wgSearchType
 626          * is set, it'll be an array of just that one item.
 627          * @deprecated since 1.27; use SearchEngineConfig::getSearchTypes()
 628          * @return array
 629          */
 630         public static function getSearchTypes() {
 631                 return MediaWikiServices::getInstance()->getSearchEngineConfig()->getSearchTypes();
 632         }
 633
 634 }
 635
 636 /**
 637  * Dummy class to be used when non-supported Database engine is present.
 638  * @todo FIXME: Dummy class should probably try something at least mildly useful,
 639  * such as a LIKE search through titles.
 640  * @ingroup Search
 641  */
 642 class SearchEngineDummy extends SearchEngine {
 643         // no-op
 644 }