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