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