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 /**
  29  * Contain a class for special pages
  30  * @ingroup Search
  31  */
  32 class SearchEngine {
  33         /** @var string */
  34         public $prefix = '';
  35
  36         /** @var int[]|null */
  37         public $namespaces = array( NS_MAIN );
  38
  39         /** @var int */
  40         protected $limit = 10;
  41
  42         /** @var int */
  43         protected $offset = 0;
  44
  45         /** @var array|string */
  46         protected $searchTerms = array();
  47
  48         /** @var bool */
  49         protected $showSuggestion = true;
  50         private $sort = 'relevance';
  51
  52         /** @var array Feature values */
  53         protected $features = array();
  54
  55         /**
  56          * Perform a full text search query and return a result set.
  57          * If title searches are not supported or disabled, return null.
  58          * STUB
  59          *
  60          * @param string $term Raw search term
  61          * @return SearchResultSet|Status|null
  62          */
  63         function searchText( $term ) {
  64                 return null;
  65         }
  66
  67         /**
  68          * Perform a title-only search query and return a result set.
  69          * If title searches are not supported or disabled, return null.
  70          * STUB
  71          *
  72          * @param string $term Raw search term
  73          * @return SearchResultSet|null
  74          */
  75         function searchTitle( $term ) {
  76                 return null;
  77         }
  78
  79         /**
  80          * @since 1.18
  81          * @param string $feature
  82          * @return bool
  83          */
  84         public function supports( $feature ) {
  85                 switch ( $feature ) {
  86                 case 'search-update':
  87                         return true;
  88                 case 'title-suffix-filter':
  89                 default:
  90                         return false;
  91                 }
  92         }
  93
  94         /**
  95          * Way to pass custom data for engines
  96          * @since 1.18
  97          * @param string $feature
  98          * @param mixed $data
  99          * @return bool
 100          */
 101         public function setFeatureData( $feature, $data ) {
 102                 $this->features[$feature] = $data;
 103         }
 104
 105         /**
 106          * When overridden in derived class, performs database-specific conversions
 107          * on text to be used for searching or updating search index.
 108          * Default implementation does nothing (simply returns $string).
 109          *
 110          * @param string $string String to process
 111          * @return string
 112          */
 113         public function normalizeText( $string ) {
 114                 global $wgContLang;
 115
 116                 // Some languages such as Chinese require word segmentation
 117                 return $wgContLang->segmentByWord( $string );
 118         }
 119
 120         /**
 121          * Transform search term in cases when parts of the query came as different
 122          * GET params (when supported), e.g. for prefix queries:
 123          * search=test&prefix=Main_Page/Archive -> test prefix:Main Page/Archive
 124          * @param string $term
 125          * @return string
 126          */
 127         function transformSearchTerm( $term ) {
 128                 return $term;
 129         }
 130
 131         /**
 132          * If an exact title match can be found, or a very slightly close match,
 133          * return the title. If no match, returns NULL.
 134          *
 135          * @param string $searchterm
 136          * @return Title
 137          */
 138         public static function getNearMatch( $searchterm ) {
 139                 $title = self::getNearMatchInternal( $searchterm );
 140
 141                 Hooks::run( 'SearchGetNearMatchComplete', array( $searchterm, &$title ) );
 142                 return $title;
 143         }
 144
 145         /**
 146          * Do a near match (see SearchEngine::getNearMatch) and wrap it into a
 147          * SearchResultSet.
 148          *
 149          * @param string $searchterm
 150          * @return SearchResultSet
 151          */
 152         public static function getNearMatchResultSet( $searchterm ) {
 153                 return new SearchNearMatchResultSet( self::getNearMatch( $searchterm ) );
 154         }
 155
 156         /**
 157          * Really find the title match.
 158          * @param string $searchterm
 159          * @return null|Title
 160          */
 161         private static function getNearMatchInternal( $searchterm ) {
 162                 global $wgContLang, $wgEnableSearchContributorsByIP;
 163
 164                 $allSearchTerms = array( $searchterm );
 165
 166                 if ( $wgContLang->hasVariants() ) {
 167                         $allSearchTerms = array_unique( array_merge(
 168                                 $allSearchTerms,
 169                                 $wgContLang->autoConvertToAllVariants( $searchterm )
 170                         ) );
 171                 }
 172
 173                 $titleResult = null;
 174                 if ( !Hooks::run( 'SearchGetNearMatchBefore', array( $allSearchTerms, &$titleResult ) ) ) {
 175                         return $titleResult;
 176                 }
 177
 178                 foreach ( $allSearchTerms as $term ) {
 179
 180                         # Exact match? No need to look further.
 181                         $title = Title::newFromText( $term );
 182                         if ( is_null( $title ) ) {
 183                                 return null;
 184                         }
 185
 186                         # Try files if searching in the Media: namespace
 187                         if ( $title->getNamespace() == NS_MEDIA ) {
 188                                 $title = Title::makeTitle( NS_FILE, $title->getText() );
 189                         }
 190
 191                         if ( $title->isSpecialPage() || $title->isExternal() || $title->exists() ) {
 192                                 return $title;
 193                         }
 194
 195                         # See if it still otherwise has content is some sane sense
 196                         $page = WikiPage::factory( $title );
 197                         if ( $page->hasViewableContent() ) {
 198                                 return $title;
 199                         }
 200
 201                         if ( !Hooks::run( 'SearchAfterNoDirectMatch', array( $term, &$title ) ) ) {
 202                                 return $title;
 203                         }
 204
 205                         # Now try all lower case (i.e. first letter capitalized)
 206                         $title = Title::newFromText( $wgContLang->lc( $term ) );
 207                         if ( $title && $title->exists() ) {
 208                                 return $title;
 209                         }
 210
 211                         # Now try capitalized string
 212                         $title = Title::newFromText( $wgContLang->ucwords( $term ) );
 213                         if ( $title && $title->exists() ) {
 214                                 return $title;
 215                         }
 216
 217                         # Now try all upper case
 218                         $title = Title::newFromText( $wgContLang->uc( $term ) );
 219                         if ( $title && $title->exists() ) {
 220                                 return $title;
 221                         }
 222
 223                         # Now try Word-Caps-Breaking-At-Word-Breaks, for hyphenated names etc
 224                         $title = Title::newFromText( $wgContLang->ucwordbreaks( $term ) );
 225                         if ( $title && $title->exists() ) {
 226                                 return $title;
 227                         }
 228
 229                         // Give hooks a chance at better match variants
 230                         $title = null;
 231                         if ( !Hooks::run( 'SearchGetNearMatch', array( $term, &$title ) ) ) {
 232                                 return $title;
 233                         }
 234                 }
 235
 236                 $title = Title::newFromText( $searchterm );
 237
 238                 # Entering an IP address goes to the contributions page
 239                 if ( $wgEnableSearchContributorsByIP ) {
 240                         if ( ( $title->getNamespace() == NS_USER && User::isIP( $title->getText() ) )
 241                                 || User::isIP( trim( $searchterm ) ) ) {
 242                                 return SpecialPage::getTitleFor( 'Contributions', $title->getDBkey() );
 243                         }
 244                 }
 245
 246                 # Entering a user goes to the user page whether it's there or not
 247                 if ( $title->getNamespace() == NS_USER ) {
 248                         return $title;
 249                 }
 250
 251                 # Go to images that exist even if there's no local page.
 252                 # There may have been a funny upload, or it may be on a shared
 253                 # file repository such as Wikimedia Commons.
 254                 if ( $title->getNamespace() == NS_FILE ) {
 255                         $image = wfFindFile( $title );
 256                         if ( $image ) {
 257                                 return $title;
 258                         }
 259                 }
 260
 261                 # MediaWiki namespace? Page may be "implied" if not customized.
 262                 # Just return it, with caps forced as the message system likes it.
 263                 if ( $title->getNamespace() == NS_MEDIAWIKI ) {
 264                         return Title::makeTitle( NS_MEDIAWIKI, $wgContLang->ucfirst( $title->getText() ) );
 265                 }
 266
 267                 # Quoted term? Try without the quotes...
 268                 $matches = array();
 269                 if ( preg_match( '/^"([^"]+)"$/', $searchterm, $matches ) ) {
 270                         return SearchEngine::getNearMatch( $matches[1] );
 271                 }
 272
 273                 return null;
 274         }
 275
 276         public static function legalSearchChars() {
 277                 return "A-Za-z_'.0-9\\x80-\\xFF\\-";
 278         }
 279
 280         /**
 281          * Set the maximum number of results to return
 282          * and how many to skip before returning the first.
 283          *
 284          * @param int $limit
 285          * @param int $offset
 286          */
 287         function setLimitOffset( $limit, $offset = 0 ) {
 288                 $this->limit = intval( $limit );
 289                 $this->offset = intval( $offset );
 290         }
 291
 292         /**
 293          * Set which namespaces the search should include.
 294          * Give an array of namespace index numbers.
 295          *
 296          * @param int[]|null $namespaces
 297          */
 298         function setNamespaces( $namespaces ) {
 299                 $this->namespaces = $namespaces;
 300         }
 301
 302         /**
 303          * Set whether the searcher should try to build a suggestion.  Note: some searchers
 304          * don't support building a suggestion in the first place and others don't respect
 305          * this flag.
 306          *
 307          * @param bool $showSuggestion Should the searcher try to build suggestions
 308          */
 309         function setShowSuggestion( $showSuggestion ) {
 310                 $this->showSuggestion = $showSuggestion;
 311         }
 312
 313         /**
 314          * Get the valid sort directions.  All search engines support 'relevance' but others
 315          * might support more. The default in all implementations should be 'relevance.'
 316          *
 317          * @since 1.25
 318          * @return array(string) the valid sort directions for setSort
 319          */
 320         public function getValidSorts() {
 321                 return array( 'relevance' );
 322         }
 323
 324         /**
 325          * Set the sort direction of the search results. Must be one returned by
 326          * SearchEngine::getValidSorts()
 327          *
 328          * @since 1.25
 329          * @throws InvalidArgumentException
 330          * @param string $sort sort direction for query result
 331          */
 332         public function setSort( $sort ) {
 333                 if ( !in_array( $sort, $this->getValidSorts() ) ) {
 334                         throw new InvalidArgumentException( "Invalid sort: $sort. " .
 335                                 "Must be one of: " . implode( ', ', $this->getValidSorts() ) );
 336                 }
 337                 $this->sort = $sort;
 338         }
 339
 340         /**
 341          * Get the sort direction of the search results
 342          *
 343          * @since 1.25
 344          * @return string
 345          */
 346         public function getSort() {
 347                 return $this->sort;
 348         }
 349
 350         /**
 351          * Parse some common prefixes: all (search everything)
 352          * or namespace names
 353          *
 354          * @param string $query
 355          * @return string
 356          */
 357         function replacePrefixes( $query ) {
 358                 global $wgContLang;
 359
 360                 $parsed = $query;
 361                 if ( strpos( $query, ':' ) === false ) { // nothing to do
 362                         return $parsed;
 363                 }
 364
 365                 $allkeyword = wfMessage( 'searchall' )->inContentLanguage()->text() . ":";
 366                 if ( strncmp( $query, $allkeyword, strlen( $allkeyword ) ) == 0 ) {
 367                         $this->namespaces = null;
 368                         $parsed = substr( $query, strlen( $allkeyword ) );
 369                 } elseif ( strpos( $query, ':' ) !== false ) {
 370                         $prefix = str_replace( ' ', '_', substr( $query, 0, strpos( $query, ':' ) ) );
 371                         $index = $wgContLang->getNsIndex( $prefix );
 372                         if ( $index !== false ) {
 373                                 $this->namespaces = array( $index );
 374                                 $parsed = substr( $query, strlen( $prefix ) + 1 );
 375                         }
 376                 }
 377                 if ( trim( $parsed ) == '' ) {
 378                         $parsed = $query; // prefix was the whole query
 379                 }
 380
 381                 return $parsed;
 382         }
 383
 384         /**
 385          * Make a list of searchable namespaces and their canonical names.
 386          * @return array
 387          */
 388         public static function searchableNamespaces() {
 389                 global $wgContLang;
 390                 $arr = array();
 391                 foreach ( $wgContLang->getNamespaces() as $ns => $name ) {
 392                         if ( $ns >= NS_MAIN ) {
 393                                 $arr[$ns] = $name;
 394                         }
 395                 }
 396
 397                 Hooks::run( 'SearchableNamespaces', array( &$arr ) );
 398                 return $arr;
 399         }
 400
 401         /**
 402          * Extract default namespaces to search from the given user's
 403          * settings, returning a list of index numbers.
 404          *
 405          * @param user $user
 406          * @return array
 407          */
 408         public static function userNamespaces( $user ) {
 409                 $arr = array();
 410                 foreach ( SearchEngine::searchableNamespaces() as $ns => $name ) {
 411                         if ( $user->getOption( 'searchNs' . $ns ) ) {
 412                                 $arr[] = $ns;
 413                         }
 414                 }
 415
 416                 return $arr;
 417         }
 418
 419         /**
 420          * Find snippet highlight settings for all users
 421          *
 422          * @return array Contextlines, contextchars
 423          */
 424         public static function userHighlightPrefs() {
 425                 $contextlines = 2; // Hardcode this. Old defaults sucked. :)
 426                 $contextchars = 75; // same as above.... :P
 427                 return array( $contextlines, $contextchars );
 428         }
 429
 430         /**
 431          * An array of namespaces indexes to be searched by default
 432          *
 433          * @return array
 434          */
 435         public static function defaultNamespaces() {
 436                 global $wgNamespacesToBeSearchedDefault;
 437
 438                 return array_keys( $wgNamespacesToBeSearchedDefault, true );
 439         }
 440
 441         /**
 442          * Get a list of namespace names useful for showing in tooltips
 443          * and preferences
 444          *
 445          * @param array $namespaces
 446          * @return array
 447          */
 448         public static function namespacesAsText( $namespaces ) {
 449                 global $wgContLang;
 450
 451                 $formatted = array_map( array( $wgContLang, 'getFormattedNsText' ), $namespaces );
 452                 foreach ( $formatted as $key => $ns ) {
 453                         if ( empty( $ns ) ) {
 454                                 $formatted[$key] = wfMessage( 'blanknamespace' )->text();
 455                         }
 456                 }
 457                 return $formatted;
 458         }
 459
 460         /**
 461          * Load up the appropriate search engine class for the currently
 462          * active database backend, and return a configured instance.
 463          *
 464          * @param string $type Type of search backend, if not the default
 465          * @return SearchEngine
 466          */
 467         public static function create( $type = null ) {
 468                 global $wgSearchType;
 469                 $dbr = null;
 470
 471                 $alternatives = self::getSearchTypes();
 472
 473                 if ( $type && in_array( $type, $alternatives ) ) {
 474                         $class = $type;
 475                 } elseif ( $wgSearchType !== null ) {
 476                         $class = $wgSearchType;
 477                 } else {
 478                         $dbr = wfGetDB( DB_SLAVE );
 479                         $class = $dbr->getSearchEngine();
 480                 }
 481
 482                 $search = new $class( $dbr );
 483                 return $search;
 484         }
 485
 486         /**
 487          * Return the search engines we support. If only $wgSearchType
 488          * is set, it'll be an array of just that one item.
 489          *
 490          * @return array
 491          */
 492         public static function getSearchTypes() {
 493                 global $wgSearchType, $wgSearchTypeAlternatives;
 494
 495                 $alternatives = $wgSearchTypeAlternatives ?: array();
 496                 array_unshift( $alternatives, $wgSearchType );
 497
 498                 return $alternatives;
 499         }
 500
 501         /**
 502          * Create or update the search index record for the given page.
 503          * Title and text should be pre-processed.
 504          * STUB
 505          *
 506          * @param int $id
 507          * @param string $title
 508          * @param string $text
 509          */
 510         function update( $id, $title, $text ) {
 511                 // no-op
 512         }
 513
 514         /**
 515          * Update a search index record's title only.
 516          * Title should be pre-processed.
 517          * STUB
 518          *
 519          * @param int $id
 520          * @param string $title
 521          */
 522         function updateTitle( $id, $title ) {
 523                 // no-op
 524         }
 525
 526         /**
 527          * Delete an indexed page
 528          * Title should be pre-processed.
 529          * STUB
 530          *
 531          * @param int $id Page id that was deleted
 532          * @param string $title Title of page that was deleted
 533          */
 534         function delete( $id, $title ) {
 535                 // no-op
 536         }
 537
 538         /**
 539          * Get OpenSearch suggestion template
 540          *
 541          * @deprecated since 1.25
 542          * @return string
 543          */
 544         public static function getOpenSearchTemplate() {
 545                 wfDeprecated( __METHOD__, '1.25' );
 546                 return ApiOpenSearch::getOpenSearchTemplate( 'application/x-suggestions+json' );
 547         }
 548
 549         /**
 550          * Get the raw text for updating the index from a content object
 551          * Nicer search backends could possibly do something cooler than
 552          * just returning raw text
 553          *
 554          * @todo This isn't ideal, we'd really like to have content-specific handling here
 555          * @param Title $t Title we're indexing
 556          * @param Content $c Content of the page to index
 557          * @return string
 558          */
 559         public function getTextFromContent( Title $t, Content $c = null ) {
 560                 return $c ? $c->getTextForSearchIndex() : '';
 561         }
 562
 563         /**
 564          * If an implementation of SearchEngine handles all of its own text processing
 565          * in getTextFromContent() and doesn't require SearchUpdate::updateText()'s
 566          * rather silly handling, it should return true here instead.
 567          *
 568          * @return bool
 569          */
 570         public function textAlreadyUpdatedForIndex() {
 571                 return false;
 572         }
 573 }
 574
 575 /**
 576  * Dummy class to be used when non-supported Database engine is present.
 577  * @todo FIXME: Dummy class should probably try something at least mildly useful,
 578  * such as a LIKE search through titles.
 579  * @ingroup Search
 580  */
 581 class SearchEngineDummy extends SearchEngine {
 582         // no-op
 583 }