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          * If this search backend can list/unlist redirects
  70          * @deprecated since 1.18 Call supports( 'list-redirects' );
  71          * @return bool
  72          */
  73         function acceptListRedirects() {
  74                 wfDeprecated( __METHOD__, '1.18' );
  75                 return $this->supports( 'list-redirects' );
  76         }
  77
  78         /**
  79          * @since 1.18
  80          * @param $feature String
  81          * @return Boolean
  82          */
  83         public function supports( $feature ) {
  84                 switch ( $feature ) {
  85                 case 'list-redirects':
  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 $feature String
  98          * @param $data Mixed
  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 GET params (when supported)
 122          * e.g. for prefix queries: search=test&prefix=Main_Page/Archive -> test prefix:Main Page/Archive
 123          */
 124         function transformSearchTerm( $term ) {
 125                 return $term;
 126         }
 127
 128         /**
 129          * If an exact title match can be found, or a very slightly close match,
 130          * return the title. If no match, returns NULL.
 131          *
 132          * @param $searchterm String
 133          * @return Title
 134          */
 135         public static function getNearMatch( $searchterm ) {
 136                 $title = self::getNearMatchInternal( $searchterm );
 137
 138                 wfRunHooks( 'SearchGetNearMatchComplete', array( $searchterm, &$title ) );
 139                 return $title;
 140         }
 141
 142         /**
 143          * Do a near match (see SearchEngine::getNearMatch) and wrap it into a
 144          * SearchResultSet.
 145          *
 146          * @param $searchterm string
 147          * @return SearchResultSet
 148          */
 149         public static function getNearMatchResultSet( $searchterm ) {
 150                 return new SearchNearMatchResultSet( self::getNearMatch( $searchterm ) );
 151         }
 152
 153         /**
 154          * Really find the title match.
 155          * @return null|Title
 156          */
 157         private static function getNearMatchInternal( $searchterm ) {
 158                 global $wgContLang, $wgEnableSearchContributorsByIP;
 159
 160                 $allSearchTerms = array( $searchterm );
 161
 162                 if ( $wgContLang->hasVariants() ) {
 163                         $allSearchTerms = array_merge( $allSearchTerms, $wgContLang->autoConvertToAllVariants( $searchterm ) );
 164                 }
 165
 166                 $titleResult = null;
 167                 if ( !wfRunHooks( 'SearchGetNearMatchBefore', array( $allSearchTerms, &$titleResult ) ) ) {
 168                         return $titleResult;
 169                 }
 170
 171                 foreach ( $allSearchTerms as $term ) {
 172
 173                         # Exact match? No need to look further.
 174                         $title = Title::newFromText( $term );
 175                         if ( is_null( $title ) ) {
 176                                 return null;
 177                         }
 178
 179                         # Try files if searching in the Media: namespace
 180                         if ( $title->getNamespace() == NS_MEDIA ) {
 181                                 $title = Title::makeTitle( NS_FILE, $title->getText() );
 182                         }
 183
 184                         if ( $title->isSpecialPage() || $title->isExternal() || $title->exists() ) {
 185                                 return $title;
 186                         }
 187
 188                         # See if it still otherwise has content is some sane sense
 189                         $page = WikiPage::factory( $title );
 190                         if ( $page->hasViewableContent() ) {
 191                                 return $title;
 192                         }
 193
 194                         if ( !wfRunHooks( 'SearchAfterNoDirectMatch', array( $term, &$title ) ) ) {
 195                                 return $title;
 196                         }
 197
 198                         # Now try all lower case (i.e. first letter capitalized)
 199                         $title = Title::newFromText( $wgContLang->lc( $term ) );
 200                         if ( $title && $title->exists() ) {
 201                                 return $title;
 202                         }
 203
 204                         # Now try capitalized string
 205                         $title = Title::newFromText( $wgContLang->ucwords( $term ) );
 206                         if ( $title && $title->exists() ) {
 207                                 return $title;
 208                         }
 209
 210                         # Now try all upper case
 211                         $title = Title::newFromText( $wgContLang->uc( $term ) );
 212                         if ( $title && $title->exists() ) {
 213                                 return $title;
 214                         }
 215
 216                         # Now try Word-Caps-Breaking-At-Word-Breaks, for hyphenated names etc
 217                         $title = Title::newFromText( $wgContLang->ucwordbreaks( $term ) );
 218                         if ( $title && $title->exists() ) {
 219                                 return $title;
 220                         }
 221
 222                         // Give hooks a chance at better match variants
 223                         $title = null;
 224                         if ( !wfRunHooks( 'SearchGetNearMatch', array( $term, &$title ) ) ) {
 225                                 return $title;
 226                         }
 227                 }
 228
 229                 $title = Title::newFromText( $searchterm );
 230
 231                 # Entering an IP address goes to the contributions page
 232                 if ( $wgEnableSearchContributorsByIP ) {
 233                         if ( ( $title->getNamespace() == NS_USER && User::isIP( $title->getText() ) )
 234                                 || User::isIP( trim( $searchterm ) ) ) {
 235                                 return SpecialPage::getTitleFor( 'Contributions', $title->getDBkey() );
 236                         }
 237                 }
 238
 239                 # Entering a user goes to the user page whether it's there or not
 240                 if ( $title->getNamespace() == NS_USER ) {
 241                         return $title;
 242                 }
 243
 244                 # Go to images that exist even if there's no local page.
 245                 # There may have been a funny upload, or it may be on a shared
 246                 # file repository such as Wikimedia Commons.
 247                 if ( $title->getNamespace() == NS_FILE ) {
 248                         $image = wfFindFile( $title );
 249                         if ( $image ) {
 250                                 return $title;
 251                         }
 252                 }
 253
 254                 # MediaWiki namespace? Page may be "implied" if not customized.
 255                 # Just return it, with caps forced as the message system likes it.
 256                 if ( $title->getNamespace() == NS_MEDIAWIKI ) {
 257                         return Title::makeTitle( NS_MEDIAWIKI, $wgContLang->ucfirst( $title->getText() ) );
 258                 }
 259
 260                 # Quoted term? Try without the quotes...
 261                 $matches = array();
 262                 if ( preg_match( '/^"([^"]+)"$/', $searchterm, $matches ) ) {
 263                         return SearchEngine::getNearMatch( $matches[1] );
 264                 }
 265
 266                 return null;
 267         }
 268
 269         public static function legalSearchChars() {
 270                 return "A-Za-z_'.0-9\\x80-\\xFF\\-";
 271         }
 272
 273         /**
 274          * Set the maximum number of results to return
 275          * and how many to skip before returning the first.
 276          *
 277          * @param $limit Integer
 278          * @param $offset Integer
 279          */
 280         function setLimitOffset( $limit, $offset = 0 ) {
 281                 $this->limit = intval( $limit );
 282                 $this->offset = intval( $offset );
 283         }
 284
 285         /**
 286          * Set which namespaces the search should include.
 287          * Give an array of namespace index numbers.
 288          *
 289          * @param $namespaces Array
 290          */
 291         function setNamespaces( $namespaces ) {
 292                 $this->namespaces = $namespaces;
 293         }
 294
 295         /**
 296          * Set whether the searcher should try to build a suggestion.  Note: some searchers
 297          * don't support building a suggestion in the first place and others don't respect
 298          * this flag.
 299          *
 300          * @param boolean $showSuggestion should the searcher try to build suggestions
 301          */
 302         function setShowSuggestion( $showSuggestion ) {
 303                 $this->showSuggestion = $showSuggestion;
 304         }
 305
 306         /**
 307          * Parse some common prefixes: all (search everything)
 308          * or namespace names
 309          *
 310          * @param $query String
 311          * @return string
 312          */
 313         function replacePrefixes( $query ) {
 314                 global $wgContLang;
 315
 316                 $parsed = $query;
 317                 if ( strpos( $query, ':' ) === false ) { // nothing to do
 318                         wfRunHooks( 'SearchEngineReplacePrefixesComplete', array( $this, $query, &$parsed ) );
 319                         return $parsed;
 320                 }
 321
 322                 $allkeyword = wfMessage( 'searchall' )->inContentLanguage()->text() . ":";
 323                 if ( strncmp( $query, $allkeyword, strlen( $allkeyword ) ) == 0 ) {
 324                         $this->namespaces = null;
 325                         $parsed = substr( $query, strlen( $allkeyword ) );
 326                 } elseif ( strpos( $query, ':' ) !== false ) {
 327                         $prefix = substr( $query, 0, strpos( $query, ':' ) );
 328                         $index = $wgContLang->getNsIndex( $prefix );
 329                         if ( $index !== false ) {
 330                                 $this->namespaces = array( $index );
 331                                 $parsed = substr( $query, strlen( $prefix ) + 1 );
 332                         }
 333                 }
 334                 if ( trim( $parsed ) == '' ) {
 335                         $parsed = $query; // prefix was the whole query
 336                 }
 337
 338                 wfRunHooks( 'SearchEngineReplacePrefixesComplete', array( $this, $query, &$parsed ) );
 339
 340                 return $parsed;
 341         }
 342
 343         /**
 344          * Make a list of searchable namespaces and their canonical names.
 345          * @return Array
 346          */
 347         public static function searchableNamespaces() {
 348                 global $wgContLang;
 349                 $arr = array();
 350                 foreach ( $wgContLang->getNamespaces() as $ns => $name ) {
 351                         if ( $ns >= NS_MAIN ) {
 352                                 $arr[$ns] = $name;
 353                         }
 354                 }
 355
 356                 wfRunHooks( 'SearchableNamespaces', array( &$arr ) );
 357                 return $arr;
 358         }
 359
 360         /**
 361          * Extract default namespaces to search from the given user's
 362          * settings, returning a list of index numbers.
 363          *
 364          * @param $user User
 365          * @return Array
 366          */
 367         public static function userNamespaces( $user ) {
 368                 global $wgSearchEverythingOnlyLoggedIn;
 369
 370                 $searchableNamespaces = SearchEngine::searchableNamespaces();
 371
 372                 // get search everything preference, that can be set to be read for logged-in users
 373                 // it overrides other options
 374                 if ( !$wgSearchEverythingOnlyLoggedIn || $user->isLoggedIn() ) {
 375                         if ( $user->getOption( 'searcheverything' ) ) {
 376                                 return array_keys( $searchableNamespaces );
 377                         }
 378                 }
 379
 380                 $arr = array();
 381                 foreach ( $searchableNamespaces as $ns => $name ) {
 382                         if ( $user->getOption( 'searchNs' . $ns ) ) {
 383                                 $arr[] = $ns;
 384                         }
 385                 }
 386
 387                 return $arr;
 388         }
 389
 390         /**
 391          * Find snippet highlight settings for all users
 392          *
 393          * @return Array contextlines, contextchars
 394          */
 395         public static function userHighlightPrefs() {
 396                 $contextlines = 2; // Hardcode this. Old defaults sucked. :)
 397                 $contextchars = 75; // same as above.... :P
 398                 return array( $contextlines, $contextchars );
 399         }
 400
 401         /**
 402          * An array of namespaces indexes to be searched by default
 403          *
 404          * @return Array
 405          */
 406         public static function defaultNamespaces() {
 407                 global $wgNamespacesToBeSearchedDefault;
 408
 409                 return array_keys( $wgNamespacesToBeSearchedDefault, true );
 410         }
 411
 412         /**
 413          * Get a list of namespace names useful for showing in tooltips
 414          * and preferences
 415          *
 416          * @param $namespaces Array
 417          * @return array
 418          */
 419         public static function namespacesAsText( $namespaces ) {
 420                 global $wgContLang;
 421
 422                 $formatted = array_map( array( $wgContLang, 'getFormattedNsText' ), $namespaces );
 423                 foreach ( $formatted as $key => $ns ) {
 424                         if ( empty( $ns ) ) {
 425                                 $formatted[$key] = wfMessage( 'blanknamespace' )->text();
 426                         }
 427                 }
 428                 return $formatted;
 429         }
 430
 431         /**
 432          * Return the help namespaces to be shown on Special:Search
 433          *
 434          * @return Array
 435          */
 436         public static function helpNamespaces() {
 437                 global $wgNamespacesToBeSearchedHelp;
 438
 439                 return array_keys( $wgNamespacesToBeSearchedHelp, true );
 440         }
 441
 442         /**
 443          * Return a 'cleaned up' search string
 444          *
 445          * @param $text String
 446          * @return String
 447          */
 448         function filter( $text ) {
 449                 $lc = $this->legalSearchChars();
 450                 return trim( preg_replace( "/[^{$lc}]/", " ", $text ) );
 451         }
 452         /**
 453          * Load up the appropriate search engine class for the currently
 454          * active database backend, and return a configured instance.
 455          *
 456          * @param String $type Type of search backend, if not the default
 457          * @return SearchEngine
 458          */
 459         public static function create( $type = null ) {
 460                 global $wgSearchType;
 461                 $dbr = null;
 462
 463                 $alternatives = self::getSearchTypes();
 464
 465                 if ( $type && in_array( $type, $alternatives ) ) {
 466                         $class = $type;
 467                 } elseif ( $wgSearchType !== null ) {
 468                         $class = $wgSearchType;
 469                 } else {
 470                         $dbr = wfGetDB( DB_SLAVE );
 471                         $class = $dbr->getSearchEngine();
 472                 }
 473
 474                 $search = new $class( $dbr );
 475                 $search->setLimitOffset( 0, 0 );
 476                 return $search;
 477         }
 478
 479         /**
 480          * Return the search engines we support. If only $wgSearchType
 481          * is set, it'll be an array of just that one item.
 482          *
 483          * @return array
 484          */
 485         public static function getSearchTypes() {
 486                 global $wgSearchType, $wgSearchTypeAlternatives;
 487
 488                 $alternatives = $wgSearchTypeAlternatives ?: array();
 489                 array_unshift( $alternatives, $wgSearchType );
 490
 491                 return $alternatives;
 492         }
 493
 494         /**
 495          * Create or update the search index record for the given page.
 496          * Title and text should be pre-processed.
 497          * STUB
 498          *
 499          * @param $id Integer
 500          * @param $title String
 501          * @param $text String
 502          */
 503         function update( $id, $title, $text ) {
 504                 // no-op
 505         }
 506
 507         /**
 508          * Update a search index record's title only.
 509          * Title should be pre-processed.
 510          * STUB
 511          *
 512          * @param $id Integer
 513          * @param $title String
 514          */
 515         function updateTitle( $id, $title ) {
 516                 // no-op
 517         }
 518
 519         /**
 520          * Delete an indexed page
 521          * Title should be pre-processed.
 522          * STUB
 523          *
 524          * @param Integer $id Page id that was deleted
 525          * @param String $title Title of page that was deleted
 526          */
 527         function delete( $id, $title ) {
 528                 // no-op
 529         }
 530
 531         /**
 532          * Get OpenSearch suggestion template
 533          *
 534          * @return String
 535          */
 536         public static function getOpenSearchTemplate() {
 537                 global $wgOpenSearchTemplate, $wgCanonicalServer;
 538                 if ( $wgOpenSearchTemplate ) {
 539                         return $wgOpenSearchTemplate;
 540                 } else {
 541                         $ns = implode( '|', SearchEngine::defaultNamespaces() );
 542                         if ( !$ns ) {
 543                                 $ns = "0";
 544                         }
 545                         return $wgCanonicalServer . wfScript( 'api' ) . '?action=opensearch&search={searchTerms}&namespace=' . $ns;
 546                 }
 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  * @ingroup Search
 577  */
 578 class SearchResultSet {
 579         /**
 580          * Fetch an array of regular expression fragments for matching
 581          * the search terms as parsed by this engine in a text extract.
 582          * STUB
 583          *
 584          * @return Array
 585          */
 586         function termMatches() {
 587                 return array();
 588         }
 589
 590         function numRows() {
 591                 return 0;
 592         }
 593
 594         /**
 595          * Return true if results are included in this result set.
 596          * STUB
 597          *
 598          * @return Boolean
 599          */
 600         function hasResults() {
 601                 return false;
 602         }
 603
 604         /**
 605          * Some search modes return a total hit count for the query
 606          * in the entire article database. This may include pages
 607          * in namespaces that would not be matched on the given
 608          * settings.
 609          *
 610          * Return null if no total hits number is supported.
 611          *
 612          * @return Integer
 613          */
 614         function getTotalHits() {
 615                 return null;
 616         }
 617
 618         /**
 619          * Some search modes return a suggested alternate term if there are
 620          * no exact hits. Returns true if there is one on this set.
 621          *
 622          * @return Boolean
 623          */
 624         function hasSuggestion() {
 625                 return false;
 626         }
 627
 628         /**
 629          * @return String: suggested query, null if none
 630          */
 631         function getSuggestionQuery() {
 632                 return null;
 633         }
 634
 635         /**
 636          * @return String: HTML highlighted suggested query, '' if none
 637          */
 638         function getSuggestionSnippet() {
 639                 return '';
 640         }
 641
 642         /**
 643          * Return information about how and from where the results were fetched,
 644          * should be useful for diagnostics and debugging
 645          *
 646          * @return String
 647          */
 648         function getInfo() {
 649                 return null;
 650         }
 651
 652         /**
 653          * Return a result set of hits on other (multiple) wikis associated with this one
 654          *
 655          * @return SearchResultSet
 656          */
 657         function getInterwikiResults() {
 658                 return null;
 659         }
 660
 661         /**
 662          * Check if there are results on other wikis
 663          *
 664          * @return Boolean
 665          */
 666         function hasInterwikiResults() {
 667                 return $this->getInterwikiResults() != null;
 668         }
 669
 670         /**
 671          * Fetches next search result, or false.
 672          * STUB
 673          *
 674          * @return SearchResult
 675          */
 676         function next() {
 677                 return false;
 678         }
 679
 680         /**
 681          * Frees the result set, if applicable.
 682          */
 683         function free() {
 684                 // ...
 685         }
 686 }
 687
 688 /**
 689  * This class is used for different SQL-based search engines shipped with MediaWiki
 690  */
 691 class SqlSearchResultSet extends SearchResultSet {
 692
 693         protected $mResultSet;
 694
 695         function __construct( $resultSet, $terms ) {
 696                 $this->mResultSet = $resultSet;
 697                 $this->mTerms = $terms;
 698         }
 699
 700         function termMatches() {
 701                 return $this->mTerms;
 702         }
 703
 704         function numRows() {
 705                 if ( $this->mResultSet === false ) {
 706                         return false;
 707                 }
 708
 709                 return $this->mResultSet->numRows();
 710         }
 711
 712         function next() {
 713                 if ( $this->mResultSet === false ) {
 714                         return false;
 715                 }
 716
 717                 $row = $this->mResultSet->fetchObject();
 718                 if ( $row === false ) {
 719                         return false;
 720                 }
 721
 722                 return SearchResult::newFromRow( $row );
 723         }
 724
 725         function free() {
 726                 if ( $this->mResultSet === false ) {
 727                         return false;
 728                 }
 729
 730                 $this->mResultSet->free();
 731         }
 732 }
 733
 734 /**
 735  * @ingroup Search
 736  */
 737 class SearchResultTooMany {
 738         # # Some search engines may bail out if too many matches are found
 739 }
 740
 741 /**
 742  * @todo FIXME: This class is horribly factored. It would probably be better to
 743  * have a useful base class to which you pass some standard information, then
 744  * let the fancy self-highlighters extend that.
 745  * @ingroup Search
 746  */
 747 class SearchResult {
 748
 749         /**
 750          * @var Revision
 751          */
 752         protected $mRevision = null;
 753
 754         /**
 755          * @var File
 756          */
 757         protected $mImage = null;
 758
 759         /**
 760          * @var Title
 761          */
 762         protected $mTitle;
 763
 764         /**
 765          * @var String
 766          */
 767         protected $mText;
 768
 769         /**
 770          * Return a new SearchResult and initializes it with a title.
 771          *
 772          * @param $title Title
 773          * @return SearchResult
 774          */
 775         public static function newFromTitle( $title ) {
 776                 $result = new self();
 777                 $result->initFromTitle( $title );
 778                 return $result;
 779         }
 780         /**
 781          * Return a new SearchResult and initializes it with a row.
 782          *
 783          * @param $row object
 784          * @return SearchResult
 785          */
 786         public static function newFromRow( $row ) {
 787                 $result = new self();
 788                 $result->initFromRow( $row );
 789                 return $result;
 790         }
 791
 792         public function __construct( $row = null ) {
 793                 if ( !is_null( $row ) ) {
 794                         // Backwards compatibility with pre-1.17 callers
 795                         $this->initFromRow( $row );
 796                 }
 797         }
 798
 799         /**
 800          * Initialize from a database row. Makes a Title and passes that to
 801          * initFromTitle.
 802          *
 803          * @param $row object
 804          */
 805         protected function initFromRow( $row ) {
 806                 $this->initFromTitle( Title::makeTitle( $row->page_namespace, $row->page_title ) );
 807         }
 808
 809         /**
 810          * Initialize from a Title and if possible initializes a corresponding
 811          * Revision and File.
 812          *
 813          * @param $title Title
 814          */
 815         protected function initFromTitle( $title ) {
 816                 $this->mTitle = $title;
 817                 if ( !is_null( $this->mTitle ) ) {
 818                         $id = false;
 819                         wfRunHooks( 'SearchResultInitFromTitle', array( $title, &$id ) );
 820                         $this->mRevision = Revision::newFromTitle(
 821                                 $this->mTitle, $id, Revision::READ_NORMAL );
 822                         if ( $this->mTitle->getNamespace() === NS_FILE ) {
 823                                 $this->mImage = wfFindFile( $this->mTitle );
 824                         }
 825                 }
 826         }
 827
 828         /**
 829          * Check if this is result points to an invalid title
 830          *
 831          * @return Boolean
 832          */
 833         function isBrokenTitle() {
 834                 return is_null( $this->mTitle );
 835         }
 836
 837         /**
 838          * Check if target page is missing, happens when index is out of date
 839          *
 840          * @return Boolean
 841          */
 842         function isMissingRevision() {
 843                 return !$this->mRevision && !$this->mImage;
 844         }
 845
 846         /**
 847          * @return Title
 848          */
 849         function getTitle() {
 850                 return $this->mTitle;
 851         }
 852
 853         /**
 854          * Get the file for this page, if one exists
 855          * @return File|null
 856          */
 857         function getFile() {
 858                 return $this->mImage;
 859         }
 860
 861         /**
 862          * @return float|null if not supported
 863          */
 864         function getScore() {
 865                 return null;
 866         }
 867
 868         /**
 869          * Lazy initialization of article text from DB
 870          */
 871         protected function initText() {
 872                 if ( !isset( $this->mText ) ) {
 873                         if ( $this->mRevision != null ) {
 874                                 $this->mText = SearchEngine::create()
 875                                         ->getTextFromContent( $this->mTitle, $this->mRevision->getContent() );
 876                         } else { // TODO: can we fetch raw wikitext for commons images?
 877                                 $this->mText = '';
 878                         }
 879                 }
 880         }
 881
 882         /**
 883          * @param array $terms terms to highlight
 884          * @return String: highlighted text snippet, null (and not '') if not supported
 885          */
 886         function getTextSnippet( $terms ) {
 887                 global $wgAdvancedSearchHighlighting;
 888                 $this->initText();
 889
 890                 // TODO: make highliter take a content object. Make ContentHandler a factory for SearchHighliter.
 891                 list( $contextlines, $contextchars ) = SearchEngine::userHighlightPrefs();
 892                 $h = new SearchHighlighter();
 893                 if ( $wgAdvancedSearchHighlighting ) {
 894                         return $h->highlightText( $this->mText, $terms, $contextlines, $contextchars );
 895                 } else {
 896                         return $h->highlightSimple( $this->mText, $terms, $contextlines, $contextchars );
 897                 }
 898         }
 899
 900         /**
 901          * @param array $terms terms to highlight
 902          * @return String: highlighted title, '' if not supported
 903          */
 904         function getTitleSnippet( $terms ) {
 905                 return '';
 906         }
 907
 908         /**
 909          * @param array $terms terms to highlight
 910          * @return String: highlighted redirect name (redirect to this page), '' if none or not supported
 911          */
 912         function getRedirectSnippet( $terms ) {
 913                 return '';
 914         }
 915
 916         /**
 917          * @return Title object for the redirect to this page, null if none or not supported
 918          */
 919         function getRedirectTitle() {
 920                 return null;
 921         }
 922
 923         /**
 924          * @return string highlighted relevant section name, null if none or not supported
 925          */
 926         function getSectionSnippet() {
 927                 return '';
 928         }
 929
 930         /**
 931          * @return Title object (pagename+fragment) for the section, null if none or not supported
 932          */
 933         function getSectionTitle() {
 934                 return null;
 935         }
 936
 937         /**
 938          * @return String: timestamp
 939          */
 940         function getTimestamp() {
 941                 if ( $this->mRevision ) {
 942                         return $this->mRevision->getTimestamp();
 943                 } elseif ( $this->mImage ) {
 944                         return $this->mImage->getTimestamp();
 945                 }
 946                 return '';
 947         }
 948
 949         /**
 950          * @return Integer: number of words
 951          */
 952         function getWordCount() {
 953                 $this->initText();
 954                 return str_word_count( $this->mText );
 955         }
 956
 957         /**
 958          * @return Integer: size in bytes
 959          */
 960         function getByteSize() {
 961                 $this->initText();
 962                 return strlen( $this->mText );
 963         }
 964
 965         /**
 966          * @return Boolean if hit has related articles
 967          */
 968         function hasRelated() {
 969                 return false;
 970         }
 971
 972         /**
 973          * @return String: interwiki prefix of the title (return iw even if title is broken)
 974          */
 975         function getInterwikiPrefix() {
 976                 return '';
 977         }
 978
 979         /**
 980          * Did this match file contents (eg: PDF/DJVU)?
 981          */
 982         function isFileMatch() {
 983                 return false;
 984         }
 985 }
 986 /**
 987  * A SearchResultSet wrapper for SearchEngine::getNearMatch
 988  */
 989 class SearchNearMatchResultSet extends SearchResultSet {
 990         private $fetched = false;
 991         /**
 992          * @param $match mixed Title if matched, else null
 993          */
 994         public function __construct( $match ) {
 995                 $this->result = $match;
 996         }
 997         public function hasResult() {
 998                 return (bool)$this->result;
 999         }
1000         public function numRows() {
1001                 return $this->hasResults() ? 1 : 0;
1002         }
1003         public function next() {
1004                 if ( $this->fetched || !$this->result ) {
1005                         return false;
1006                 }
1007                 $this->fetched = true;
1008                 return SearchResult::newFromTitle( $this->result );
1009         }
1010 }
1011
1012 /**
1013  * Highlight bits of wikitext
1014  *
1015  * @ingroup Search
1016  */
1017 class SearchHighlighter {
1018         var $mCleanWikitext = true;
1019
1020         function __construct( $cleanupWikitext = true ) {
1021                 $this->mCleanWikitext = $cleanupWikitext;
1022         }
1023
1024         /**
1025          * Default implementation of wikitext highlighting
1026          *
1027          * @param $text String
1028          * @param array $terms terms to highlight (unescaped)
1029          * @param $contextlines Integer
1030          * @param $contextchars Integer
1031          * @return String
1032          */
1033         public function highlightText( $text, $terms, $contextlines, $contextchars ) {
1034                 global $wgContLang;
1035                 global $wgSearchHighlightBoundaries;
1036                 $fname = __METHOD__;
1037
1038                 if ( $text == '' ) {
1039                         return '';
1040                 }
1041
1042                 // spli text into text + templates/links/tables
1043                 $spat = "/(\\{\\{)|(\\[\\[[^\\]:]+:)|(\n\\{\\|)";
1044                 // first capture group is for detecting nested templates/links/tables/references
1045                 $endPatterns = array(
1046                         1 => '/(\{\{)|(\}\})/', // template
1047                         2 => '/(\[\[)|(\]\])/', // image
1048                         3 => "/(\n\\{\\|)|(\n\\|\\})/" ); // table
1049
1050                 // @todo FIXME: This should prolly be a hook or something
1051                 if ( function_exists( 'wfCite' ) ) {
1052                         $spat .= '|(<ref>)'; // references via cite extension
1053                         $endPatterns[4] = '/(<ref>)|(<\/ref>)/';
1054                 }
1055                 $spat .= '/';
1056                 $textExt = array(); // text extracts
1057                 $otherExt = array(); // other extracts
1058                 wfProfileIn( "$fname-split" );
1059                 $start = 0;
1060                 $textLen = strlen( $text );
1061                 $count = 0; // sequence number to maintain ordering
1062                 while ( $start < $textLen ) {
1063                         // find start of template/image/table
1064                         if ( preg_match( $spat, $text, $matches, PREG_OFFSET_CAPTURE, $start ) ) {
1065                                 $epat = '';
1066                                 foreach ( $matches as $key => $val ) {
1067                                         if ( $key > 0 && $val[1] != - 1 ) {
1068                                                 if ( $key == 2 ) {
1069                                                         // see if this is an image link
1070                                                         $ns = substr( $val[0], 2, - 1 );
1071                                                         if ( $wgContLang->getNsIndex( $ns ) != NS_FILE ) {
1072                                                                 break;
1073                                                         }
1074
1075                                                 }
1076                                                 $epat = $endPatterns[$key];
1077                                                 $this->splitAndAdd( $textExt, $count, substr( $text, $start, $val[1] - $start ) );
1078                                                 $start = $val[1];
1079                                                 break;
1080                                         }
1081                                 }
1082                                 if ( $epat ) {
1083                                         // find end (and detect any nested elements)
1084                                         $level = 0;
1085                                         $offset = $start + 1;
1086                                         $found = false;
1087                                         while ( preg_match( $epat, $text, $endMatches, PREG_OFFSET_CAPTURE, $offset ) ) {
1088                                                 if ( array_key_exists( 2, $endMatches ) ) {
1089                                                         // found end
1090                                                         if ( $level == 0 ) {
1091                                                                 $len = strlen( $endMatches[2][0] );
1092                                                                 $off = $endMatches[2][1];
1093                                                                 $this->splitAndAdd( $otherExt, $count,
1094                                                                         substr( $text, $start, $off + $len - $start ) );
1095                                                                 $start = $off + $len;
1096                                                                 $found = true;
1097                                                                 break;
1098                                                         } else {
1099                                                                 // end of nested element
1100                                                                 $level -= 1;
1101                                                         }
1102                                                 } else {
1103                                                         // nested
1104                                                         $level += 1;
1105                                                 }
1106                                                 $offset = $endMatches[0][1] + strlen( $endMatches[0][0] );
1107                                         }
1108                                         if ( ! $found ) {
1109                                                 // couldn't find appropriate closing tag, skip
1110                                                 $this->splitAndAdd( $textExt, $count, substr( $text, $start, strlen( $matches[0][0] ) ) );
1111                                                 $start += strlen( $matches[0][0] );
1112                                         }
1113                                         continue;
1114                                 }
1115                         }
1116                         // else: add as text extract
1117                         $this->splitAndAdd( $textExt, $count, substr( $text, $start ) );
1118                         break;
1119                 }
1120
1121                 $all = $textExt + $otherExt; // these have disjunct key sets
1122
1123                 wfProfileOut( "$fname-split" );
1124
1125                 // prepare regexps
1126                 foreach ( $terms as $index => $term ) {
1127                         // manually do upper/lowercase stuff for utf-8 since PHP won't do it
1128                         if ( preg_match( '/[\x80-\xff]/', $term ) ) {
1129                                 $terms[$index] = preg_replace_callback( '/./us', array( $this, 'caseCallback' ), $terms[$index] );
1130                         } else {
1131                                 $terms[$index] = $term;
1132                         }
1133                 }
1134                 $anyterm = implode( '|', $terms );
1135                 $phrase = implode( "$wgSearchHighlightBoundaries+", $terms );
1136
1137                 // @todo FIXME: A hack to scale contextchars, a correct solution
1138                 // would be to have contextchars actually be char and not byte
1139                 // length, and do proper utf-8 substrings and lengths everywhere,
1140                 // but PHP is making that very hard and unclean to implement :(
1141                 $scale = strlen( $anyterm ) / mb_strlen( $anyterm );
1142                 $contextchars = intval( $contextchars * $scale );
1143
1144                 $patPre = "(^|$wgSearchHighlightBoundaries)";
1145                 $patPost = "($wgSearchHighlightBoundaries|$)";
1146
1147                 $pat1 = "/(" . $phrase . ")/ui";
1148                 $pat2 = "/$patPre(" . $anyterm . ")$patPost/ui";
1149
1150                 wfProfileIn( "$fname-extract" );
1151
1152                 $left = $contextlines;
1153
1154                 $snippets = array();
1155                 $offsets = array();
1156
1157                 // show beginning only if it contains all words
1158                 $first = 0;
1159                 $firstText = '';
1160                 foreach ( $textExt as $index => $line ) {
1161                         if ( strlen( $line ) > 0 && $line[0] != ';' && $line[0] != ':' ) {
1162                                 $firstText = $this->extract( $line, 0, $contextchars * $contextlines );
1163                                 $first = $index;
1164                                 break;
1165                         }
1166                 }
1167                 if ( $firstText ) {
1168                         $succ = true;
1169                         // check if first text contains all terms
1170                         foreach ( $terms as $term ) {
1171                                 if ( ! preg_match( "/$patPre" . $term . "$patPost/ui", $firstText ) ) {
1172                                         $succ = false;
1173                                         break;
1174                                 }
1175                         }
1176                         if ( $succ ) {
1177                                 $snippets[$first] = $firstText;
1178                                 $offsets[$first] = 0;
1179                         }
1180                 }
1181                 if ( ! $snippets ) {
1182                         // match whole query on text
1183                         $this->process( $pat1, $textExt, $left, $contextchars, $snippets, $offsets );
1184                         // match whole query on templates/tables/images
1185                         $this->process( $pat1, $otherExt, $left, $contextchars, $snippets, $offsets );
1186                         // match any words on text
1187                         $this->process( $pat2, $textExt, $left, $contextchars, $snippets, $offsets );
1188                         // match any words on templates/tables/images
1189                         $this->process( $pat2, $otherExt, $left, $contextchars, $snippets, $offsets );
1190
1191                         ksort( $snippets );
1192                 }
1193
1194                 // add extra chars to each snippet to make snippets constant size
1195                 $extended = array();
1196                 if ( count( $snippets ) == 0 ) {
1197                         // couldn't find the target words, just show beginning of article
1198                         if ( array_key_exists( $first, $all ) ) {
1199                                 $targetchars = $contextchars * $contextlines;
1200                                 $snippets[$first] = '';
1201                                 $offsets[$first] = 0;
1202                         }
1203                 } else {
1204                         // if begin of the article contains the whole phrase, show only that !!
1205                         if ( array_key_exists( $first, $snippets ) && preg_match( $pat1, $snippets[$first] )
1206                                 && $offsets[$first] < $contextchars * 2 ) {
1207                                 $snippets = array( $first => $snippets[$first] );
1208                         }
1209
1210                         // calc by how much to extend existing snippets
1211                         $targetchars = intval( ( $contextchars * $contextlines ) / count ( $snippets ) );
1212                 }
1213
1214                 foreach ( $snippets as $index => $line ) {
1215                         $extended[$index] = $line;
1216                         $len = strlen( $line );
1217                         if ( $len < $targetchars - 20 ) {
1218                                 // complete this line
1219                                 if ( $len < strlen( $all[$index] ) ) {
1220                                         $extended[$index] = $this->extract( $all[$index], $offsets[$index], $offsets[$index] + $targetchars, $offsets[$index] );
1221                                         $len = strlen( $extended[$index] );
1222                                 }
1223
1224                                 // add more lines
1225                                 $add = $index + 1;
1226                                 while ( $len < $targetchars - 20
1227                                                 && array_key_exists( $add, $all )
1228                                                 && !array_key_exists( $add, $snippets ) ) {
1229                                         $offsets[$add] = 0;
1230                                         $tt = "\n" . $this->extract( $all[$add], 0, $targetchars - $len, $offsets[$add] );
1231                                         $extended[$add] = $tt;
1232                                         $len += strlen( $tt );
1233                                         $add++;
1234                                 }
1235                         }
1236                 }
1237
1238                 // $snippets = array_map( 'htmlspecialchars', $extended );
1239                 $snippets = $extended;
1240                 $last = - 1;
1241                 $extract = '';
1242                 foreach ( $snippets as $index => $line ) {
1243                         if ( $last == - 1 ) {
1244                                 $extract .= $line; // first line
1245                         } elseif ( $last + 1 == $index && $offsets[$last] + strlen( $snippets[$last] ) >= strlen( $all[$last] ) ) {
1246                                 $extract .= " " . $line; // continous lines
1247                         } else {
1248                                 $extract .= '<b> ... </b>' . $line;
1249                         }
1250
1251                         $last = $index;
1252                 }
1253                 if ( $extract ) {
1254                         $extract .= '<b> ... </b>';
1255                 }
1256
1257                 $processed = array();
1258                 foreach ( $terms as $term ) {
1259                         if ( ! isset( $processed[$term] ) ) {
1260                                 $pat3 = "/$patPre(" . $term . ")$patPost/ui"; // highlight word
1261                                 $extract = preg_replace( $pat3,
1262                                         "\\1<span class='searchmatch'>\\2</span>\\3", $extract );
1263                                 $processed[$term] = true;
1264                         }
1265                 }
1266
1267                 wfProfileOut( "$fname-extract" );
1268
1269                 return $extract;
1270         }
1271
1272         /**
1273          * Split text into lines and add it to extracts array
1274          *
1275          * @param array $extracts index -> $line
1276          * @param $count Integer
1277          * @param $text String
1278          */
1279         function splitAndAdd( &$extracts, &$count, $text ) {
1280                 $split = explode( "\n", $this->mCleanWikitext ? $this->removeWiki( $text ) : $text );
1281                 foreach ( $split as $line ) {
1282                         $tt = trim( $line );
1283                         if ( $tt ) {
1284                                 $extracts[$count++] = $tt;
1285                         }
1286                 }
1287         }
1288
1289         /**
1290          * Do manual case conversion for non-ascii chars
1291          *
1292          * @param $matches Array
1293          * @return string
1294          */
1295         function caseCallback( $matches ) {
1296                 global $wgContLang;
1297                 if ( strlen( $matches[0] ) > 1 ) {
1298                         return '[' . $wgContLang->lc( $matches[0] ) . $wgContLang->uc( $matches[0] ) . ']';
1299                 } else {
1300                         return $matches[0];
1301                 }
1302         }
1303
1304         /**
1305          * Extract part of the text from start to end, but by
1306          * not chopping up words
1307          * @param $text String
1308          * @param $start Integer
1309          * @param $end Integer
1310          * @param $posStart Integer: (out) actual start position
1311          * @param $posEnd Integer: (out) actual end position
1312          * @return String
1313          */
1314         function extract( $text, $start, $end, &$posStart = null, &$posEnd = null ) {
1315                 if ( $start != 0 ) {
1316                         $start = $this->position( $text, $start, 1 );
1317                 }
1318                 if ( $end >= strlen( $text ) ) {
1319                         $end = strlen( $text );
1320                 } else {
1321                         $end = $this->position( $text, $end );
1322                 }
1323
1324                 if ( !is_null( $posStart ) ) {
1325                         $posStart = $start;
1326                 }
1327                 if ( !is_null( $posEnd ) ) {
1328                         $posEnd = $end;
1329                 }
1330
1331                 if ( $end > $start ) {
1332                         return substr( $text, $start, $end - $start );
1333                 } else {
1334                         return '';
1335                 }
1336         }
1337
1338         /**
1339          * Find a nonletter near a point (index) in the text
1340          *
1341          * @param $text String
1342          * @param $point Integer
1343          * @param $offset Integer: offset to found index
1344          * @return Integer: nearest nonletter index, or beginning of utf8 char if none
1345          */
1346         function position( $text, $point, $offset = 0 ) {
1347                 $tolerance = 10;
1348                 $s = max( 0, $point - $tolerance );
1349                 $l = min( strlen( $text ), $point + $tolerance ) - $s;
1350                 $m = array();
1351                 if ( preg_match( '/[ ,.!?~!@#$%^&*\(\)+=\-\\\|\[\]"\'<>]/', substr( $text, $s, $l ), $m, PREG_OFFSET_CAPTURE ) ) {
1352                         return $m[0][1] + $s + $offset;
1353                 } else {
1354                         // check if point is on a valid first UTF8 char
1355                         $char = ord( $text[$point] );
1356                         while ( $char >= 0x80 && $char < 0xc0 ) {
1357                                 // skip trailing bytes
1358                                 $point++;
1359                                 if ( $point >= strlen( $text ) ) {
1360                                         return strlen( $text );
1361                                 }
1362                                 $char = ord( $text[$point] );
1363                         }
1364                         return $point;
1365
1366                 }
1367         }
1368
1369         /**
1370          * Search extracts for a pattern, and return snippets
1371          *
1372          * @param string $pattern regexp for matching lines
1373          * @param array $extracts extracts to search
1374          * @param $linesleft Integer: number of extracts to make
1375          * @param $contextchars Integer: length of snippet
1376          * @param array $out map for highlighted snippets
1377          * @param array $offsets map of starting points of snippets
1378          * @protected
1379          */
1380         function process( $pattern, $extracts, &$linesleft, &$contextchars, &$out, &$offsets ) {
1381                 if ( $linesleft == 0 ) {
1382                         return; // nothing to do
1383                 }
1384                 foreach ( $extracts as $index => $line ) {
1385                         if ( array_key_exists( $index, $out ) ) {
1386                                 continue; // this line already highlighted
1387                         }
1388
1389                         $m = array();
1390                         if ( !preg_match( $pattern, $line, $m, PREG_OFFSET_CAPTURE ) ) {
1391                                 continue;
1392                         }
1393
1394                         $offset = $m[0][1];
1395                         $len = strlen( $m[0][0] );
1396                         if ( $offset + $len < $contextchars ) {
1397                                 $begin = 0;
1398                         } elseif ( $len > $contextchars ) {
1399                                 $begin = $offset;
1400                         } else {
1401                                 $begin = $offset + intval( ( $len - $contextchars ) / 2 );
1402                         }
1403
1404                         $end = $begin + $contextchars;
1405
1406                         $posBegin = $begin;
1407                         // basic snippet from this line
1408                         $out[$index] = $this->extract( $line, $begin, $end, $posBegin );
1409                         $offsets[$index] = $posBegin;
1410                         $linesleft--;
1411                         if ( $linesleft == 0 ) {
1412                                 return;
1413                         }
1414                 }
1415         }
1416
1417         /**
1418          * Basic wikitext removal
1419          * @protected
1420          * @return mixed
1421          */
1422         function removeWiki( $text ) {
1423                 $fname = __METHOD__;
1424                 wfProfileIn( $fname );
1425
1426                 // $text = preg_replace( "/'{2,5}/", "", $text );
1427                 // $text = preg_replace( "/\[[a-z]+:\/\/[^ ]+ ([^]]+)\]/", "\\2", $text );
1428                 // $text = preg_replace( "/\[\[([^]|]+)\]\]/", "\\1", $text );
1429                 // $text = preg_replace( "/\[\[([^]]+\|)?([^|]]+)\]\]/", "\\2", $text );
1430                 // $text = preg_replace( "/\\{\\|(.*?)\\|\\}/", "", $text );
1431                 // $text = preg_replace( "/\\[\\[[A-Za-z_-]+:([^|]+?)\\]\\]/", "", $text );
1432                 $text = preg_replace( "/\\{\\{([^|]+?)\\}\\}/", "", $text );
1433                 $text = preg_replace( "/\\{\\{([^|]+\\|)(.*?)\\}\\}/", "\\2", $text );
1434                 $text = preg_replace( "/\\[\\[([^|]+?)\\]\\]/", "\\1", $text );
1435                 $text = preg_replace_callback( "/\\[\\[([^|]+\\|)(.*?)\\]\\]/", array( $this, 'linkReplace' ), $text );
1436                 // $text = preg_replace("/\\[\\[([^|]+\\|)(.*?)\\]\\]/", "\\2", $text);
1437                 $text = preg_replace( "/<\/?[^>]+>/", "", $text );
1438                 $text = preg_replace( "/'''''/", "", $text );
1439                 $text = preg_replace( "/('''|<\/?[iIuUbB]>)/", "", $text );
1440                 $text = preg_replace( "/''/", "", $text );
1441
1442                 wfProfileOut( $fname );
1443                 return $text;
1444         }
1445
1446         /**
1447          * callback to replace [[target|caption]] kind of links, if
1448          * the target is category or image, leave it
1449          *
1450          * @param $matches Array
1451          */
1452         function linkReplace( $matches ) {
1453                 $colon = strpos( $matches[1], ':' );
1454                 if ( $colon === false ) {
1455                         return $matches[2]; // replace with caption
1456                 }
1457                 global $wgContLang;
1458                 $ns = substr( $matches[1], 0, $colon );
1459                 $index = $wgContLang->getNsIndex( $ns );
1460                 if ( $index !== false && ( $index == NS_FILE || $index == NS_CATEGORY ) ) {
1461                         return $matches[0]; // return the whole thing
1462                 } else {
1463                         return $matches[2];
1464                 }
1465         }
1466
1467         /**
1468          * Simple & fast snippet extraction, but gives completely unrelevant
1469          * snippets
1470          *
1471          * @param $text String
1472          * @param $terms Array
1473          * @param $contextlines Integer
1474          * @param $contextchars Integer
1475          * @return String
1476          */
1477         public function highlightSimple( $text, $terms, $contextlines, $contextchars ) {
1478                 global $wgContLang;
1479                 $fname = __METHOD__;
1480
1481                 $lines = explode( "\n", $text );
1482
1483                 $terms = implode( '|', $terms );
1484                 $max = intval( $contextchars ) + 1;
1485                 $pat1 = "/(.*)($terms)(.{0,$max})/i";
1486
1487                 $lineno = 0;
1488
1489                 $extract = "";
1490                 wfProfileIn( "$fname-extract" );
1491                 foreach ( $lines as $line ) {
1492                         if ( 0 == $contextlines ) {
1493                                 break;
1494                         }
1495                         ++$lineno;
1496                         $m = array();
1497                         if ( ! preg_match( $pat1, $line, $m ) ) {
1498                                 continue;
1499                         }
1500                         --$contextlines;
1501                         // truncate function changes ... to relevant i18n message.
1502                         $pre = $wgContLang->truncate( $m[1], - $contextchars, '...', false );
1503
1504                         if ( count( $m ) < 3 ) {
1505                                 $post = '';
1506                         } else {
1507                                 $post = $wgContLang->truncate( $m[3], $contextchars, '...', false );
1508                         }
1509
1510                         $found = $m[2];
1511
1512                         $line = htmlspecialchars( $pre . $found . $post );
1513                         $pat2 = '/(' . $terms . ")/i";
1514                         $line = preg_replace( $pat2, "<span class='searchmatch'>\\1</span>", $line );
1515
1516                         $extract .= "${line}\n";
1517                 }
1518                 wfProfileOut( "$fname-extract" );
1519
1520                 return $extract;
1521         }
1522
1523 }
1524
1525 /**
1526  * Dummy class to be used when non-supported Database engine is present.
1527  * @todo FIXME: Dummy class should probably try something at least mildly useful,
1528  * such as a LIKE search through titles.
1529  * @ingroup Search
1530  */
1531 class SearchEngineDummy extends SearchEngine {
1532         // no-op
1533 }