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