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
  40         /// Feature values
  41         protected $features = array();
  42
  43         /**
  44          * @var DatabaseBase
  45          */
  46         protected $db;
  47
  48         function __construct( $db = null ) {
  49                 if ( $db ) {
  50                         $this->db = $db;
  51                 } else {
  52                         $this->db = wfGetDB( DB_SLAVE );
  53                 }
  54         }
  55
  56         /**
  57          * Perform a full text 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|Status|null
  63          */
  64         function searchText( $term ) {
  65                 return null;
  66         }
  67
  68         /**
  69          * Perform a title-only search query and return a result set.
  70          * If title searches are not supported or disabled, return null.
  71          * STUB
  72          *
  73          * @param string $term raw search term
  74          * @return SearchResultSet|null
  75          */
  76         function searchTitle( $term ) {
  77                 return null;
  78         }
  79
  80         /**
  81          * If this search backend can list/unlist redirects
  82          * @deprecated since 1.18 Call supports( 'list-redirects' );
  83          * @return bool
  84          */
  85         function acceptListRedirects() {
  86                 wfDeprecated( __METHOD__, '1.18' );
  87                 return $this->supports( 'list-redirects' );
  88         }
  89
  90         /**
  91          * @since 1.18
  92          * @param $feature String
  93          * @return Boolean
  94          */
  95         public function supports( $feature ) {
  96                 switch ( $feature ) {
  97                 case 'list-redirects':
  98                 case 'search-update':
  99                         return true;
 100                 case 'title-suffix-filter':
 101                 default:
 102                         return false;
 103                 }
 104         }
 105
 106         /**
 107          * Way to pass custom data for engines
 108          * @since 1.18
 109          * @param $feature String
 110          * @param $data Mixed
 111          * @return bool
 112          */
 113         public function setFeatureData( $feature, $data ) {
 114                 $this->features[$feature] = $data;
 115         }
 116
 117         /**
 118          * When overridden in derived class, performs database-specific conversions
 119          * on text to be used for searching or updating search index.
 120          * Default implementation does nothing (simply returns $string).
 121          *
 122          * @param string $string String to process
 123          * @return string
 124          */
 125         public function normalizeText( $string ) {
 126                 global $wgContLang;
 127
 128                 // Some languages such as Chinese require word segmentation
 129                 return $wgContLang->segmentByWord( $string );
 130         }
 131
 132         /**
 133          * Transform search term in cases when parts of the query came as different GET params (when supported)
 134          * e.g. for prefix queries: search=test&prefix=Main_Page/Archive -> test prefix:Main Page/Archive
 135          */
 136         function transformSearchTerm( $term ) {
 137                 return $term;
 138         }
 139
 140         /**
 141          * If an exact title match can be found, or a very slightly close match,
 142          * return the title. If no match, returns NULL.
 143          *
 144          * @param $searchterm String
 145          * @return Title
 146          */
 147         public static function getNearMatch( $searchterm ) {
 148                 $title = self::getNearMatchInternal( $searchterm );
 149
 150                 wfRunHooks( 'SearchGetNearMatchComplete', array( $searchterm, &$title ) );
 151                 return $title;
 152         }
 153
 154         /**
 155          * Do a near match (see SearchEngine::getNearMatch) and wrap it into a
 156          * SearchResultSet.
 157          *
 158          * @param $searchterm string
 159          * @return SearchResultSet
 160          */
 161         public static function getNearMatchResultSet( $searchterm ) {
 162                 return new SearchNearMatchResultSet( self::getNearMatch( $searchterm ) );
 163         }
 164
 165         /**
 166          * Really find the title match.
 167          * @return null|Title
 168          */
 169         private static function getNearMatchInternal( $searchterm ) {
 170                 global $wgContLang, $wgEnableSearchContributorsByIP;
 171
 172                 $allSearchTerms = array( $searchterm );
 173
 174                 if ( $wgContLang->hasVariants() ) {
 175                         $allSearchTerms = array_merge( $allSearchTerms, $wgContLang->autoConvertToAllVariants( $searchterm ) );
 176                 }
 177
 178                 $titleResult = null;
 179                 if ( !wfRunHooks( 'SearchGetNearMatchBefore', array( $allSearchTerms, &$titleResult ) ) ) {
 180                         return $titleResult;
 181                 }
 182
 183                 foreach ( $allSearchTerms as $term ) {
 184
 185                         # Exact match? No need to look further.
 186                         $title = Title::newFromText( $term );
 187                         if ( is_null( $title ) ) {
 188                                 return null;
 189                         }
 190
 191                         # Try files if searching in the Media: namespace
 192                         if ( $title->getNamespace() == NS_MEDIA ) {
 193                                 $title = Title::makeTitle( NS_FILE, $title->getText() );
 194                         }
 195
 196                         if ( $title->isSpecialPage() || $title->isExternal() || $title->exists() ) {
 197                                 return $title;
 198                         }
 199
 200                         # See if it still otherwise has content is some sane sense
 201                         $page = WikiPage::factory( $title );
 202                         if ( $page->hasViewableContent() ) {
 203                                 return $title;
 204                         }
 205
 206                         if ( !wfRunHooks( 'SearchAfterNoDirectMatch', array( $term, &$title ) ) ) {
 207                                 return $title;
 208                         }
 209
 210                         # Now try all lower case (i.e. first letter capitalized)
 211                         $title = Title::newFromText( $wgContLang->lc( $term ) );
 212                         if ( $title && $title->exists() ) {
 213                                 return $title;
 214                         }
 215
 216                         # Now try capitalized string
 217                         $title = Title::newFromText( $wgContLang->ucwords( $term ) );
 218                         if ( $title && $title->exists() ) {
 219                                 return $title;
 220                         }
 221
 222                         # Now try all upper case
 223                         $title = Title::newFromText( $wgContLang->uc( $term ) );
 224                         if ( $title && $title->exists() ) {
 225                                 return $title;
 226                         }
 227
 228                         # Now try Word-Caps-Breaking-At-Word-Breaks, for hyphenated names etc
 229                         $title = Title::newFromText( $wgContLang->ucwordbreaks( $term ) );
 230                         if ( $title && $title->exists() ) {
 231                                 return $title;
 232                         }
 233
 234                         // Give hooks a chance at better match variants
 235                         $title = null;
 236                         if ( !wfRunHooks( 'SearchGetNearMatch', array( $term, &$title ) ) ) {
 237                                 return $title;
 238                         }
 239                 }
 240
 241                 $title = Title::newFromText( $searchterm );
 242
 243                 # Entering an IP address goes to the contributions page
 244                 if ( $wgEnableSearchContributorsByIP ) {
 245                         if ( ( $title->getNamespace() == NS_USER && User::isIP( $title->getText() ) )
 246                                 || User::isIP( trim( $searchterm ) ) ) {
 247                                 return SpecialPage::getTitleFor( 'Contributions', $title->getDBkey() );
 248                         }
 249                 }
 250
 251                 # Entering a user goes to the user page whether it's there or not
 252                 if ( $title->getNamespace() == NS_USER ) {
 253                         return $title;
 254                 }
 255
 256                 # Go to images that exist even if there's no local page.
 257                 # There may have been a funny upload, or it may be on a shared
 258                 # file repository such as Wikimedia Commons.
 259                 if ( $title->getNamespace() == NS_FILE ) {
 260                         $image = wfFindFile( $title );
 261                         if ( $image ) {
 262                                 return $title;
 263                         }
 264                 }
 265
 266                 # MediaWiki namespace? Page may be "implied" if not customized.
 267                 # Just return it, with caps forced as the message system likes it.
 268                 if ( $title->getNamespace() == NS_MEDIAWIKI ) {
 269                         return Title::makeTitle( NS_MEDIAWIKI, $wgContLang->ucfirst( $title->getText() ) );
 270                 }
 271
 272                 # Quoted term? Try without the quotes...
 273                 $matches = array();
 274                 if ( preg_match( '/^"([^"]+)"$/', $searchterm, $matches ) ) {
 275                         return SearchEngine::getNearMatch( $matches[1] );
 276                 }
 277
 278                 return null;
 279         }
 280
 281         public static function legalSearchChars() {
 282                 return "A-Za-z_'.0-9\\x80-\\xFF\\-";
 283         }
 284
 285         /**
 286          * Set the maximum number of results to return
 287          * and how many to skip before returning the first.
 288          *
 289          * @param $limit Integer
 290          * @param $offset Integer
 291          */
 292         function setLimitOffset( $limit, $offset = 0 ) {
 293                 $this->limit = intval( $limit );
 294                 $this->offset = intval( $offset );
 295         }
 296
 297         /**
 298          * Set which namespaces the search should include.
 299          * Give an array of namespace index numbers.
 300          *
 301          * @param $namespaces Array
 302          */
 303         function setNamespaces( $namespaces ) {
 304                 $this->namespaces = $namespaces;
 305         }
 306
 307         /**
 308          * Parse some common prefixes: all (search everything)
 309          * or namespace names
 310          *
 311          * @param $query String
 312          * @return string
 313          */
 314         function replacePrefixes( $query ) {
 315                 global $wgContLang;
 316
 317                 $parsed = $query;
 318                 if ( strpos( $query, ':' ) === false ) { // nothing to do
 319                         wfRunHooks( 'SearchEngineReplacePrefixesComplete', array( $this, $query, &$parsed ) );
 320                         return $parsed;
 321                 }
 322
 323                 $allkeyword = wfMessage( 'searchall' )->inContentLanguage()->text() . ":";
 324                 if ( strncmp( $query, $allkeyword, strlen( $allkeyword ) ) == 0 ) {
 325                         $this->namespaces = null;
 326                         $parsed = substr( $query, strlen( $allkeyword ) );
 327                 } elseif ( strpos( $query, ':' ) !== false ) {
 328                         $prefix = substr( $query, 0, strpos( $query, ':' ) );
 329                         $index = $wgContLang->getNsIndex( $prefix );
 330                         if ( $index !== false ) {
 331                                 $this->namespaces = array( $index );
 332                                 $parsed = substr( $query, strlen( $prefix ) + 1 );
 333                         }
 334                 }
 335                 if ( trim( $parsed ) == '' ) {
 336                         $parsed = $query; // prefix was the whole query
 337                 }
 338
 339                 wfRunHooks( 'SearchEngineReplacePrefixesComplete', array( $this, $query, &$parsed ) );
 340
 341                 return $parsed;
 342         }
 343
 344         /**
 345          * Make a list of searchable namespaces and their canonical names.
 346          * @return Array
 347          */
 348         public static function searchableNamespaces() {
 349                 global $wgContLang;
 350                 $arr = array();
 351                 foreach ( $wgContLang->getNamespaces() as $ns => $name ) {
 352                         if ( $ns >= NS_MAIN ) {
 353                                 $arr[$ns] = $name;
 354                         }
 355                 }
 356
 357                 wfRunHooks( 'SearchableNamespaces', array( &$arr ) );
 358                 return $arr;
 359         }
 360
 361         /**
 362          * Extract default namespaces to search from the given user's
 363          * settings, returning a list of index numbers.
 364          *
 365          * @param $user User
 366          * @return Array
 367          */
 368         public static function userNamespaces( $user ) {
 369                 global $wgSearchEverythingOnlyLoggedIn;
 370
 371                 $searchableNamespaces = SearchEngine::searchableNamespaces();
 372
 373                 // get search everything preference, that can be set to be read for logged-in users
 374                 // it overrides other options
 375                 if ( !$wgSearchEverythingOnlyLoggedIn || $user->isLoggedIn() ) {
 376                         if ( $user->getOption( 'searcheverything' ) ) {
 377                                 return array_keys( $searchableNamespaces );
 378                         }
 379                 }
 380
 381                 $arr = array();
 382                 foreach ( $searchableNamespaces as $ns => $name ) {
 383                         if ( $user->getOption( 'searchNs' . $ns ) ) {
 384                                 $arr[] = $ns;
 385                         }
 386                 }
 387
 388                 return $arr;
 389         }
 390
 391         /**
 392          * Find snippet highlight settings for all users
 393          *
 394          * @return Array contextlines, contextchars
 395          */
 396         public static function userHighlightPrefs() {
 397                 $contextlines = 2; // Hardcode this. Old defaults sucked. :)
 398                 $contextchars = 75; // same as above.... :P
 399                 return array( $contextlines, $contextchars );
 400         }
 401
 402         /**
 403          * An array of namespaces indexes to be searched by default
 404          *
 405          * @return Array
 406          */
 407         public static function defaultNamespaces() {
 408                 global $wgNamespacesToBeSearchedDefault;
 409
 410                 return array_keys( $wgNamespacesToBeSearchedDefault, true );
 411         }
 412
 413         /**
 414          * Get a list of namespace names useful for showing in tooltips
 415          * and preferences
 416          *
 417          * @param $namespaces Array
 418          * @return array
 419          */
 420         public static function namespacesAsText( $namespaces ) {
 421                 global $wgContLang;
 422
 423                 $formatted = array_map( array( $wgContLang, 'getFormattedNsText' ), $namespaces );
 424                 foreach ( $formatted as $key => $ns ) {
 425                         if ( empty( $ns ) ) {
 426                                 $formatted[$key] = wfMessage( 'blanknamespace' )->text();
 427                         }
 428                 }
 429                 return $formatted;
 430         }
 431
 432         /**
 433          * Return the help namespaces to be shown on Special:Search
 434          *
 435          * @return Array
 436          */
 437         public static function helpNamespaces() {
 438                 global $wgNamespacesToBeSearchedHelp;
 439
 440                 return array_keys( $wgNamespacesToBeSearchedHelp, true );
 441         }
 442
 443         /**
 444          * Return a 'cleaned up' search string
 445          *
 446          * @param $text String
 447          * @return String
 448          */
 449         function filter( $text ) {
 450                 $lc = $this->legalSearchChars();
 451                 return trim( preg_replace( "/[^{$lc}]/", " ", $text ) );
 452         }
 453         /**
 454          * Load up the appropriate search engine class for the currently
 455          * active database backend, and return a configured instance.
 456          *
 457          * @param String $type Type of search backend, if not the default
 458          * @return SearchEngine
 459          */
 460         public static function create( $type = null ) {
 461                 global $wgSearchType;
 462                 $dbr = null;
 463
 464                 $alternatives = self::getSearchTypes();
 465
 466                 if ( $type && in_array( $type, $alternatives ) ) {
 467                         $class = $type;
 468                 } elseif ( $wgSearchType !== null ) {
 469                         $class = $wgSearchType;
 470                 } else {
 471                         $dbr = wfGetDB( DB_SLAVE );
 472                         $class = $dbr->getSearchEngine();
 473                 }
 474
 475                 $search = new $class( $dbr );
 476                 $search->setLimitOffset( 0, 0 );
 477                 return $search;
 478         }
 479
 480         /**
 481          * Return the search engines we support. If only $wgSearchType
 482          * is set, it'll be an array of just that one item.
 483          *
 484          * @return array
 485          */
 486         public static function getSearchTypes() {
 487                 global $wgSearchType, $wgSearchTypeAlternatives;
 488                 static $alternatives = null;
 489                 if ( $alternatives === null ) {
 490                         $alternatives = $wgSearchTypeAlternatives ?: array();
 491                         array_unshift( $alternatives, $wgSearchType );
 492                 }
 493                 return $alternatives;
 494         }
 495
 496         /**
 497          * Create or update the search index record for the given page.
 498          * Title and text should be pre-processed.
 499          * STUB
 500          *
 501          * @param $id Integer
 502          * @param $title String
 503          * @param $text String
 504          */
 505         function update( $id, $title, $text ) {
 506                 // no-op
 507         }
 508
 509         /**
 510          * Update a search index record's title only.
 511          * Title should be pre-processed.
 512          * STUB
 513          *
 514          * @param $id Integer
 515          * @param $title String
 516          */
 517         function updateTitle( $id, $title ) {
 518                 // no-op
 519         }
 520
 521         /**
 522          * Delete an indexed page
 523          * Title should be pre-processed.
 524          * STUB
 525          *
 526          * @param Integer $id Page id that was deleted
 527          * @param String $title Title of page that was deleted
 528          */
 529         function delete( $id, $title ) {
 530                 // no-op
 531         }
 532
 533         /**
 534          * Get OpenSearch suggestion template
 535          *
 536          * @return String
 537          */
 538         public static function getOpenSearchTemplate() {
 539                 global $wgOpenSearchTemplate, $wgCanonicalServer;
 540                 if ( $wgOpenSearchTemplate ) {
 541                         return $wgOpenSearchTemplate;
 542                 } else {
 543                         $ns = implode( '|', SearchEngine::defaultNamespaces() );
 544                         if ( !$ns ) {
 545                                 $ns = "0";
 546                         }
 547                         return $wgCanonicalServer . wfScript( 'api' ) . '?action=opensearch&search={searchTerms}&namespace=' . $ns;
 548                 }
 549         }
 550
 551         /**
 552          * Get the raw text for updating the index from a content object
 553          * Nicer search backends could possibly do something cooler than
 554          * just returning raw text
 555          *
 556          * @todo This isn't ideal, we'd really like to have content-specific handling here
 557          * @param Title $t Title we're indexing
 558          * @param Content $c Content of the page to index
 559          * @return string
 560          */
 561         public function getTextFromContent( Title $t, Content $c = null ) {
 562                 return $c ? $c->getTextForSearchIndex() : '';
 563         }
 564
 565         /**
 566          * If an implementation of SearchEngine handles all of its own text processing
 567          * in getTextFromContent() and doesn't require SearchUpdate::updateText()'s
 568          * rather silly handling, it should return true here instead.
 569          *
 570          * @return bool
 571          */
 572         public function textAlreadyUpdatedForIndex() {
 573                 return false;
 574         }
 575 }
 576
 577 /**
 578  * @ingroup Search
 579  */
 580 class SearchResultSet {
 581         /**
 582          * Fetch an array of regular expression fragments for matching
 583          * the search terms as parsed by this engine in a text extract.
 584          * STUB
 585          *
 586          * @return Array
 587          */
 588         function termMatches() {
 589                 return array();
 590         }
 591
 592         function numRows() {
 593                 return 0;
 594         }
 595
 596         /**
 597          * Return true if results are included in this result set.
 598          * STUB
 599          *
 600          * @return Boolean
 601          */
 602         function hasResults() {
 603                 return false;
 604         }
 605
 606         /**
 607          * Some search modes return a total hit count for the query
 608          * in the entire article database. This may include pages
 609          * in namespaces that would not be matched on the given
 610          * settings.
 611          *
 612          * Return null if no total hits number is supported.
 613          *
 614          * @return Integer
 615          */
 616         function getTotalHits() {
 617                 return null;
 618         }
 619
 620         /**
 621          * Some search modes return a suggested alternate term if there are
 622          * no exact hits. Returns true if there is one on this set.
 623          *
 624          * @return Boolean
 625          */
 626         function hasSuggestion() {
 627                 return false;
 628         }
 629
 630         /**
 631          * @return String: suggested query, null if none
 632          */
 633         function getSuggestionQuery() {
 634                 return null;
 635         }
 636
 637         /**
 638          * @return String: HTML highlighted suggested query, '' if none
 639          */
 640         function getSuggestionSnippet() {
 641                 return '';
 642         }
 643
 644         /**
 645          * Return information about how and from where the results were fetched,
 646          * should be useful for diagnostics and debugging
 647          *
 648          * @return String
 649          */
 650         function getInfo() {
 651                 return null;
 652         }
 653
 654         /**
 655          * Return a result set of hits on other (multiple) wikis associated with this one
 656          *
 657          * @return SearchResultSet
 658          */
 659         function getInterwikiResults() {
 660                 return null;
 661         }
 662
 663         /**
 664          * Check if there are results on other wikis
 665          *
 666          * @return Boolean
 667          */
 668         function hasInterwikiResults() {
 669                 return $this->getInterwikiResults() != null;
 670         }
 671
 672         /**
 673          * Fetches next search result, or false.
 674          * STUB
 675          *
 676          * @return SearchResult
 677          */
 678         function next() {
 679                 return false;
 680         }
 681
 682         /**
 683          * Frees the result set, if applicable.
 684          */
 685         function free() {
 686                 // ...
 687         }
 688 }
 689
 690 /**
 691  * This class is used for different SQL-based search engines shipped with MediaWiki
 692  */
 693 class SqlSearchResultSet extends SearchResultSet {
 694
 695         protected $mResultSet;
 696
 697         function __construct( $resultSet, $terms ) {
 698                 $this->mResultSet = $resultSet;
 699                 $this->mTerms = $terms;
 700         }
 701
 702         function termMatches() {
 703                 return $this->mTerms;
 704         }
 705
 706         function numRows() {
 707                 if ( $this->mResultSet === false ) {
 708                         return false;
 709                 }
 710
 711                 return $this->mResultSet->numRows();
 712         }
 713
 714         function next() {
 715                 if ( $this->mResultSet === false ) {
 716                         return false;
 717                 }
 718
 719                 $row = $this->mResultSet->fetchObject();
 720                 if ( $row === false ) {
 721                         return false;
 722                 }
 723
 724                 return SearchResult::newFromRow( $row );
 725         }
 726
 727         function free() {
 728                 if ( $this->mResultSet === false ) {
 729                         return false;
 730                 }
 731
 732                 $this->mResultSet->free();
 733         }
 734 }
 735
 736 /**
 737  * @ingroup Search
 738  */
 739 class SearchResultTooMany {
 740         # # Some search engines may bail out if too many matches are found
 741 }
 742
 743 /**
 744  * @todo FIXME: This class is horribly factored. It would probably be better to
 745  * have a useful base class to which you pass some standard information, then
 746  * let the fancy self-highlighters extend that.
 747  * @ingroup Search
 748  */
 749 class SearchResult {
 750
 751         /**
 752          * @var Revision
 753          */
 754         var $mRevision = null;
 755         var $mImage = null;
 756
 757         /**
 758          * @var Title
 759          */
 760         var $mTitle;
 761
 762         /**
 763          * @var String
 764          */
 765         var $mText;
 766
 767         /**
 768          * Return a new SearchResult and initializes it with a title.
 769          *
 770          * @param $title Title
 771          * @return SearchResult
 772          */
 773         public static function newFromTitle( $title ) {
 774                 $result = new self();
 775                 $result->initFromTitle( $title );
 776                 return $result;
 777         }
 778         /**
 779          * Return a new SearchResult and initializes it with a row.
 780          *
 781          * @param $row object
 782          * @return SearchResult
 783          */
 784         public static function newFromRow( $row ) {
 785                 $result = new self();
 786                 $result->initFromRow( $row );
 787                 return $result;
 788         }
 789
 790         public function __construct( $row = null ) {
 791                 if ( !is_null( $row ) ) {
 792                         // Backwards compatibility with pre-1.17 callers
 793                         $this->initFromRow( $row );
 794                 }
 795         }
 796
 797         /**
 798          * Initialize from a database row. Makes a Title and passes that to
 799          * initFromTitle.
 800          *
 801          * @param $row object
 802          */
 803         protected function initFromRow( $row ) {
 804                 $this->initFromTitle( Title::makeTitle( $row->page_namespace, $row->page_title ) );
 805         }
 806
 807         /**
 808          * Initialize from a Title and if possible initializes a corresponding
 809          * Revision and File.
 810          *
 811          * @param $title Title
 812          */
 813         protected function initFromTitle( $title ) {
 814                 $this->mTitle = $title;
 815                 if ( !is_null( $this->mTitle ) ) {
 816                         $id = false;
 817                         wfRunHooks( 'SearchResultInitFromTitle', array( $title, &$id ) );
 818                         $this->mRevision = Revision::newFromTitle(
 819                                 $this->mTitle, $id, Revision::READ_NORMAL );
 820                         if ( $this->mTitle->getNamespace() === NS_FILE ) {
 821                                 $this->mImage = wfFindFile( $this->mTitle );
 822                         }
 823                 }
 824         }
 825
 826         /**
 827          * Check if this is result points to an invalid title
 828          *
 829          * @return Boolean
 830          */
 831         function isBrokenTitle() {
 832                 if ( is_null( $this->mTitle ) ) {
 833                         return true;
 834                 }
 835                 return false;
 836         }
 837
 838         /**
 839          * Check if target page is missing, happens when index is out of date
 840          *
 841          * @return Boolean
 842          */
 843         function isMissingRevision() {
 844                 return !$this->mRevision && !$this->mImage;
 845         }
 846
 847         /**
 848          * @return Title
 849          */
 850         function getTitle() {
 851                 return $this->mTitle;
 852         }
 853
 854         /**
 855          * @return float|null if not supported
 856          */
 857         function getScore() {
 858                 return null;
 859         }
 860
 861         /**
 862          * Lazy initialization of article text from DB
 863          */
 864         protected function initText() {
 865                 if ( !isset( $this->mText ) ) {
 866                         if ( $this->mRevision != null ) {
 867                                 $this->mText = SearchEngine::create()
 868                                         ->getTextFromContent( $this->mTitle, $this->mRevision->getContent() );
 869                         } else { // TODO: can we fetch raw wikitext for commons images?
 870                                 $this->mText = '';
 871                         }
 872                 }
 873         }
 874
 875         /**
 876          * @param array $terms terms to highlight
 877          * @return String: highlighted text snippet, null (and not '') if not supported
 878          */
 879         function getTextSnippet( $terms ) {
 880                 global $wgAdvancedSearchHighlighting;
 881                 $this->initText();
 882
 883                 // TODO: make highliter take a content object. Make ContentHandler a factory for SearchHighliter.
 884                 list( $contextlines, $contextchars ) = SearchEngine::userHighlightPrefs();
 885                 $h = new SearchHighlighter();
 886                 if ( $wgAdvancedSearchHighlighting ) {
 887                         return $h->highlightText( $this->mText, $terms, $contextlines, $contextchars );
 888                 } else {
 889                         return $h->highlightSimple( $this->mText, $terms, $contextlines, $contextchars );
 890                 }
 891         }
 892
 893         /**
 894          * @param array $terms terms to highlight
 895          * @return String: highlighted title, '' if not supported
 896          */
 897         function getTitleSnippet( $terms ) {
 898                 return '';
 899         }
 900
 901         /**
 902          * @param array $terms terms to highlight
 903          * @return String: highlighted redirect name (redirect to this page), '' if none or not supported
 904          */
 905         function getRedirectSnippet( $terms ) {
 906                 return '';
 907         }
 908
 909         /**
 910          * @return Title object for the redirect to this page, null if none or not supported
 911          */
 912         function getRedirectTitle() {
 913                 return null;
 914         }
 915
 916         /**
 917          * @return string highlighted relevant section name, null if none or not supported
 918          */
 919         function getSectionSnippet() {
 920                 return '';
 921         }
 922
 923         /**
 924          * @return Title object (pagename+fragment) for the section, null if none or not supported
 925          */
 926         function getSectionTitle() {
 927                 return null;
 928         }
 929
 930         /**
 931          * @return String: timestamp
 932          */
 933         function getTimestamp() {
 934                 if ( $this->mRevision ) {
 935                         return $this->mRevision->getTimestamp();
 936                 } elseif ( $this->mImage ) {
 937                         return $this->mImage->getTimestamp();
 938                 }
 939                 return '';
 940         }
 941
 942         /**
 943          * @return Integer: number of words
 944          */
 945         function getWordCount() {
 946                 $this->initText();
 947                 return str_word_count( $this->mText );
 948         }
 949
 950         /**
 951          * @return Integer: size in bytes
 952          */
 953         function getByteSize() {
 954                 $this->initText();
 955                 return strlen( $this->mText );
 956         }
 957
 958         /**
 959          * @return Boolean if hit has related articles
 960          */
 961         function hasRelated() {
 962                 return false;
 963         }
 964
 965         /**
 966          * @return String: interwiki prefix of the title (return iw even if title is broken)
 967          */
 968         function getInterwikiPrefix() {
 969                 return '';
 970         }
 971 }
 972 /**
 973  * A SearchResultSet wrapper for SearchEngine::getNearMatch
 974  */
 975 class SearchNearMatchResultSet extends SearchResultSet {
 976         private $fetched = false;
 977         /**
 978          * @param $match mixed Title if matched, else null
 979          */
 980         public function __construct( $match ) {
 981                 $this->result = $match;
 982         }
 983         public function hasResult() {
 984                 return (bool)$this->result;
 985         }
 986         public function numRows() {
 987                 return $this->hasResults() ? 1 : 0;
 988         }
 989         public function next() {
 990                 if ( $this->fetched || !$this->result ) {
 991                         return false;
 992                 }
 993                 $this->fetched = true;
 994                 return SearchResult::newFromTitle( $this->result );
 995         }
 996 }
 997
 998 /**
 999  * Highlight bits of wikitext
1000  *
1001  * @ingroup Search
1002  */
1003 class SearchHighlighter {
1004         var $mCleanWikitext = true;
1005
1006         function __construct( $cleanupWikitext = true ) {
1007                 $this->mCleanWikitext = $cleanupWikitext;
1008         }
1009
1010         /**
1011          * Default implementation of wikitext highlighting
1012          *
1013          * @param $text String
1014          * @param array $terms terms to highlight (unescaped)
1015          * @param $contextlines Integer
1016          * @param $contextchars Integer
1017          * @return String
1018          */
1019         public function highlightText( $text, $terms, $contextlines, $contextchars ) {
1020                 global $wgContLang;
1021                 global $wgSearchHighlightBoundaries;
1022                 $fname = __METHOD__;
1023
1024                 if ( $text == '' ) {
1025                         return '';
1026                 }
1027
1028                 // spli text into text + templates/links/tables
1029                 $spat = "/(\\{\\{)|(\\[\\[[^\\]:]+:)|(\n\\{\\|)";
1030                 // first capture group is for detecting nested templates/links/tables/references
1031                 $endPatterns = array(
1032                         1 => '/(\{\{)|(\}\})/', // template
1033                         2 => '/(\[\[)|(\]\])/', // image
1034                         3 => "/(\n\\{\\|)|(\n\\|\\})/" ); // table
1035
1036                 // @todo FIXME: This should prolly be a hook or something
1037                 if ( function_exists( 'wfCite' ) ) {
1038                         $spat .= '|(<ref>)'; // references via cite extension
1039                         $endPatterns[4] = '/(<ref>)|(<\/ref>)/';
1040                 }
1041                 $spat .= '/';
1042                 $textExt = array(); // text extracts
1043                 $otherExt = array(); // other extracts
1044                 wfProfileIn( "$fname-split" );
1045                 $start = 0;
1046                 $textLen = strlen( $text );
1047                 $count = 0; // sequence number to maintain ordering
1048                 while ( $start < $textLen ) {
1049                         // find start of template/image/table
1050                         if ( preg_match( $spat, $text, $matches, PREG_OFFSET_CAPTURE, $start ) ) {
1051                                 $epat = '';
1052                                 foreach ( $matches as $key => $val ) {
1053                                         if ( $key > 0 && $val[1] != - 1 ) {
1054                                                 if ( $key == 2 ) {
1055                                                         // see if this is an image link
1056                                                         $ns = substr( $val[0], 2, - 1 );
1057                                                         if ( $wgContLang->getNsIndex( $ns ) != NS_FILE ) {
1058                                                                 break;
1059                                                         }
1060
1061                                                 }
1062                                                 $epat = $endPatterns[$key];
1063                                                 $this->splitAndAdd( $textExt, $count, substr( $text, $start, $val[1] - $start ) );
1064                                                 $start = $val[1];
1065                                                 break;
1066                                         }
1067                                 }
1068                                 if ( $epat ) {
1069                                         // find end (and detect any nested elements)
1070                                         $level = 0;
1071                                         $offset = $start + 1;
1072                                         $found = false;
1073                                         while ( preg_match( $epat, $text, $endMatches, PREG_OFFSET_CAPTURE, $offset ) ) {
1074                                                 if ( array_key_exists( 2, $endMatches ) ) {
1075                                                         // found end
1076                                                         if ( $level == 0 ) {
1077                                                                 $len = strlen( $endMatches[2][0] );
1078                                                                 $off = $endMatches[2][1];
1079                                                                 $this->splitAndAdd( $otherExt, $count,
1080                                                                         substr( $text, $start, $off + $len - $start ) );
1081                                                                 $start = $off + $len;
1082                                                                 $found = true;
1083                                                                 break;
1084                                                         } else {
1085                                                                 // end of nested element
1086                                                                 $level -= 1;
1087                                                         }
1088                                                 } else {
1089                                                         // nested
1090                                                         $level += 1;
1091                                                 }
1092                                                 $offset = $endMatches[0][1] + strlen( $endMatches[0][0] );
1093                                         }
1094                                         if ( ! $found ) {
1095                                                 // couldn't find appropriate closing tag, skip
1096                                                 $this->splitAndAdd( $textExt, $count, substr( $text, $start, strlen( $matches[0][0] ) ) );
1097                                                 $start += strlen( $matches[0][0] );
1098                                         }
1099                                         continue;
1100                                 }
1101                         }
1102                         // else: add as text extract
1103                         $this->splitAndAdd( $textExt, $count, substr( $text, $start ) );
1104                         break;
1105                 }
1106
1107                 $all = $textExt + $otherExt; // these have disjunct key sets
1108
1109                 wfProfileOut( "$fname-split" );
1110
1111                 // prepare regexps
1112                 foreach ( $terms as $index => $term ) {
1113                         // manually do upper/lowercase stuff for utf-8 since PHP won't do it
1114                         if ( preg_match( '/[\x80-\xff]/', $term ) ) {
1115                                 $terms[$index] = preg_replace_callback( '/./us', array( $this, 'caseCallback' ), $terms[$index] );
1116                         } else {
1117                                 $terms[$index] = $term;
1118                         }
1119                 }
1120                 $anyterm = implode( '|', $terms );
1121                 $phrase = implode( "$wgSearchHighlightBoundaries+", $terms );
1122
1123                 // @todo FIXME: A hack to scale contextchars, a correct solution
1124                 // would be to have contextchars actually be char and not byte
1125                 // length, and do proper utf-8 substrings and lengths everywhere,
1126                 // but PHP is making that very hard and unclean to implement :(
1127                 $scale = strlen( $anyterm ) / mb_strlen( $anyterm );
1128                 $contextchars = intval( $contextchars * $scale );
1129
1130                 $patPre = "(^|$wgSearchHighlightBoundaries)";
1131                 $patPost = "($wgSearchHighlightBoundaries|$)";
1132
1133                 $pat1 = "/(" . $phrase . ")/ui";
1134                 $pat2 = "/$patPre(" . $anyterm . ")$patPost/ui";
1135
1136                 wfProfileIn( "$fname-extract" );
1137
1138                 $left = $contextlines;
1139
1140                 $snippets = array();
1141                 $offsets = array();
1142
1143                 // show beginning only if it contains all words
1144                 $first = 0;
1145                 $firstText = '';
1146                 foreach ( $textExt as $index => $line ) {
1147                         if ( strlen( $line ) > 0 && $line[0] != ';' && $line[0] != ':' ) {
1148                                 $firstText = $this->extract( $line, 0, $contextchars * $contextlines );
1149                                 $first = $index;
1150                                 break;
1151                         }
1152                 }
1153                 if ( $firstText ) {
1154                         $succ = true;
1155                         // check if first text contains all terms
1156                         foreach ( $terms as $term ) {
1157                                 if ( ! preg_match( "/$patPre" . $term . "$patPost/ui", $firstText ) ) {
1158                                         $succ = false;
1159                                         break;
1160                                 }
1161                         }
1162                         if ( $succ ) {
1163                                 $snippets[$first] = $firstText;
1164                                 $offsets[$first] = 0;
1165                         }
1166                 }
1167                 if ( ! $snippets ) {
1168                         // match whole query on text
1169                         $this->process( $pat1, $textExt, $left, $contextchars, $snippets, $offsets );
1170                         // match whole query on templates/tables/images
1171                         $this->process( $pat1, $otherExt, $left, $contextchars, $snippets, $offsets );
1172                         // match any words on text
1173                         $this->process( $pat2, $textExt, $left, $contextchars, $snippets, $offsets );
1174                         // match any words on templates/tables/images
1175                         $this->process( $pat2, $otherExt, $left, $contextchars, $snippets, $offsets );
1176
1177                         ksort( $snippets );
1178                 }
1179
1180                 // add extra chars to each snippet to make snippets constant size
1181                 $extended = array();
1182                 if ( count( $snippets ) == 0 ) {
1183                         // couldn't find the target words, just show beginning of article
1184                         if ( array_key_exists( $first, $all ) ) {
1185                                 $targetchars = $contextchars * $contextlines;
1186                                 $snippets[$first] = '';
1187                                 $offsets[$first] = 0;
1188                         }
1189                 } else {
1190                         // if begin of the article contains the whole phrase, show only that !!
1191                         if ( array_key_exists( $first, $snippets ) && preg_match( $pat1, $snippets[$first] )
1192                                 && $offsets[$first] < $contextchars * 2 ) {
1193                                 $snippets = array( $first => $snippets[$first] );
1194                         }
1195
1196                         // calc by how much to extend existing snippets
1197                         $targetchars = intval( ( $contextchars * $contextlines ) / count ( $snippets ) );
1198                 }
1199
1200                 foreach ( $snippets as $index => $line ) {
1201                         $extended[$index] = $line;
1202                         $len = strlen( $line );
1203                         if ( $len < $targetchars - 20 ) {
1204                                 // complete this line
1205                                 if ( $len < strlen( $all[$index] ) ) {
1206                                         $extended[$index] = $this->extract( $all[$index], $offsets[$index], $offsets[$index] + $targetchars, $offsets[$index] );
1207                                         $len = strlen( $extended[$index] );
1208                                 }
1209
1210                                 // add more lines
1211                                 $add = $index + 1;
1212                                 while ( $len < $targetchars - 20
1213                                                 && array_key_exists( $add, $all )
1214                                                 && !array_key_exists( $add, $snippets ) ) {
1215                                         $offsets[$add] = 0;
1216                                         $tt = "\n" . $this->extract( $all[$add], 0, $targetchars - $len, $offsets[$add] );
1217                                         $extended[$add] = $tt;
1218                                         $len += strlen( $tt );
1219                                         $add++;
1220                                 }
1221                         }
1222                 }
1223
1224                 // $snippets = array_map( 'htmlspecialchars', $extended );
1225                 $snippets = $extended;
1226                 $last = - 1;
1227                 $extract = '';
1228                 foreach ( $snippets as $index => $line ) {
1229                         if ( $last == - 1 ) {
1230                                 $extract .= $line; // first line
1231                         } elseif ( $last + 1 == $index && $offsets[$last] + strlen( $snippets[$last] ) >= strlen( $all[$last] ) ) {
1232                                 $extract .= " " . $line; // continous lines
1233                         } else {
1234                                 $extract .= '<b> ... </b>' . $line;
1235                         }
1236
1237                         $last = $index;
1238                 }
1239                 if ( $extract ) {
1240                         $extract .= '<b> ... </b>';
1241                 }
1242
1243                 $processed = array();
1244                 foreach ( $terms as $term ) {
1245                         if ( ! isset( $processed[$term] ) ) {
1246                                 $pat3 = "/$patPre(" . $term . ")$patPost/ui"; // highlight word
1247                                 $extract = preg_replace( $pat3,
1248                                         "\\1<span class='searchmatch'>\\2</span>\\3", $extract );
1249                                 $processed[$term] = true;
1250                         }
1251                 }
1252
1253                 wfProfileOut( "$fname-extract" );
1254
1255                 return $extract;
1256         }
1257
1258         /**
1259          * Split text into lines and add it to extracts array
1260          *
1261          * @param array $extracts index -> $line
1262          * @param $count Integer
1263          * @param $text String
1264          */
1265         function splitAndAdd( &$extracts, &$count, $text ) {
1266                 $split = explode( "\n", $this->mCleanWikitext ? $this->removeWiki( $text ) : $text );
1267                 foreach ( $split as $line ) {
1268                         $tt = trim( $line );
1269                         if ( $tt ) {
1270                                 $extracts[$count++] = $tt;
1271                         }
1272                 }
1273         }
1274
1275         /**
1276          * Do manual case conversion for non-ascii chars
1277          *
1278          * @param $matches Array
1279          * @return string
1280          */
1281         function caseCallback( $matches ) {
1282                 global $wgContLang;
1283                 if ( strlen( $matches[0] ) > 1 ) {
1284                         return '[' . $wgContLang->lc( $matches[0] ) . $wgContLang->uc( $matches[0] ) . ']';
1285                 } else {
1286                         return $matches[0];
1287                 }
1288         }
1289
1290         /**
1291          * Extract part of the text from start to end, but by
1292          * not chopping up words
1293          * @param $text String
1294          * @param $start Integer
1295          * @param $end Integer
1296          * @param $posStart Integer: (out) actual start position
1297          * @param $posEnd Integer: (out) actual end position
1298          * @return String
1299          */
1300         function extract( $text, $start, $end, &$posStart = null, &$posEnd = null ) {
1301                 if ( $start != 0 ) {
1302                         $start = $this->position( $text, $start, 1 );
1303                 }
1304                 if ( $end >= strlen( $text ) ) {
1305                         $end = strlen( $text );
1306                 } else {
1307                         $end = $this->position( $text, $end );
1308                 }
1309
1310                 if ( !is_null( $posStart ) ) {
1311                         $posStart = $start;
1312                 }
1313                 if ( !is_null( $posEnd ) ) {
1314                         $posEnd = $end;
1315                 }
1316
1317                 if ( $end > $start ) {
1318                         return substr( $text, $start, $end - $start );
1319                 } else {
1320                         return '';
1321                 }
1322         }
1323
1324         /**
1325          * Find a nonletter near a point (index) in the text
1326          *
1327          * @param $text String
1328          * @param $point Integer
1329          * @param $offset Integer: offset to found index
1330          * @return Integer: nearest nonletter index, or beginning of utf8 char if none
1331          */
1332         function position( $text, $point, $offset = 0 ) {
1333                 $tolerance = 10;
1334                 $s = max( 0, $point - $tolerance );
1335                 $l = min( strlen( $text ), $point + $tolerance ) - $s;
1336                 $m = array();
1337                 if ( preg_match( '/[ ,.!?~!@#$%^&*\(\)+=\-\\\|\[\]"\'<>]/', substr( $text, $s, $l ), $m, PREG_OFFSET_CAPTURE ) ) {
1338                         return $m[0][1] + $s + $offset;
1339                 } else {
1340                         // check if point is on a valid first UTF8 char
1341                         $char = ord( $text[$point] );
1342                         while ( $char >= 0x80 && $char < 0xc0 ) {
1343                                 // skip trailing bytes
1344                                 $point++;
1345                                 if ( $point >= strlen( $text ) ) {
1346                                         return strlen( $text );
1347                                 }
1348                                 $char = ord( $text[$point] );
1349                         }
1350                         return $point;
1351
1352                 }
1353         }
1354
1355         /**
1356          * Search extracts for a pattern, and return snippets
1357          *
1358          * @param string $pattern regexp for matching lines
1359          * @param array $extracts extracts to search
1360          * @param $linesleft Integer: number of extracts to make
1361          * @param $contextchars Integer: length of snippet
1362          * @param array $out map for highlighted snippets
1363          * @param array $offsets map of starting points of snippets
1364          * @protected
1365          */
1366         function process( $pattern, $extracts, &$linesleft, &$contextchars, &$out, &$offsets ) {
1367                 if ( $linesleft == 0 ) {
1368                         return; // nothing to do
1369                 }
1370                 foreach ( $extracts as $index => $line ) {
1371                         if ( array_key_exists( $index, $out ) ) {
1372                                 continue; // this line already highlighted
1373                         }
1374
1375                         $m = array();
1376                         if ( !preg_match( $pattern, $line, $m, PREG_OFFSET_CAPTURE ) ) {
1377                                 continue;
1378                         }
1379
1380                         $offset = $m[0][1];
1381                         $len = strlen( $m[0][0] );
1382                         if ( $offset + $len < $contextchars ) {
1383                                 $begin = 0;
1384                         } elseif ( $len > $contextchars ) {
1385                                 $begin = $offset;
1386                         } else {
1387                                 $begin = $offset + intval( ( $len - $contextchars ) / 2 );
1388                         }
1389
1390                         $end = $begin + $contextchars;
1391
1392                         $posBegin = $begin;
1393                         // basic snippet from this line
1394                         $out[$index] = $this->extract( $line, $begin, $end, $posBegin );
1395                         $offsets[$index] = $posBegin;
1396                         $linesleft--;
1397                         if ( $linesleft == 0 ) {
1398                                 return;
1399                         }
1400                 }
1401         }
1402
1403         /**
1404          * Basic wikitext removal
1405          * @protected
1406          * @return mixed
1407          */
1408         function removeWiki( $text ) {
1409                 $fname = __METHOD__;
1410                 wfProfileIn( $fname );
1411
1412                 // $text = preg_replace( "/'{2,5}/", "", $text );
1413                 // $text = preg_replace( "/\[[a-z]+:\/\/[^ ]+ ([^]]+)\]/", "\\2", $text );
1414                 // $text = preg_replace( "/\[\[([^]|]+)\]\]/", "\\1", $text );
1415                 // $text = preg_replace( "/\[\[([^]]+\|)?([^|]]+)\]\]/", "\\2", $text );
1416                 // $text = preg_replace( "/\\{\\|(.*?)\\|\\}/", "", $text );
1417                 // $text = preg_replace( "/\\[\\[[A-Za-z_-]+:([^|]+?)\\]\\]/", "", $text );
1418                 $text = preg_replace( "/\\{\\{([^|]+?)\\}\\}/", "", $text );
1419                 $text = preg_replace( "/\\{\\{([^|]+\\|)(.*?)\\}\\}/", "\\2", $text );
1420                 $text = preg_replace( "/\\[\\[([^|]+?)\\]\\]/", "\\1", $text );
1421                 $text = preg_replace_callback( "/\\[\\[([^|]+\\|)(.*?)\\]\\]/", array( $this, 'linkReplace' ), $text );
1422                 // $text = preg_replace("/\\[\\[([^|]+\\|)(.*?)\\]\\]/", "\\2", $text);
1423                 $text = preg_replace( "/<\/?[^>]+>/", "", $text );
1424                 $text = preg_replace( "/'''''/", "", $text );
1425                 $text = preg_replace( "/('''|<\/?[iIuUbB]>)/", "", $text );
1426                 $text = preg_replace( "/''/", "", $text );
1427
1428                 wfProfileOut( $fname );
1429                 return $text;
1430         }
1431
1432         /**
1433          * callback to replace [[target|caption]] kind of links, if
1434          * the target is category or image, leave it
1435          *
1436          * @param $matches Array
1437          */
1438         function linkReplace( $matches ) {
1439                 $colon = strpos( $matches[1], ':' );
1440                 if ( $colon === false ) {
1441                         return $matches[2]; // replace with caption
1442                 }
1443                 global $wgContLang;
1444                 $ns = substr( $matches[1], 0, $colon );
1445                 $index = $wgContLang->getNsIndex( $ns );
1446                 if ( $index !== false && ( $index == NS_FILE || $index == NS_CATEGORY ) ) {
1447                         return $matches[0]; // return the whole thing
1448                 } else {
1449                         return $matches[2];
1450                 }
1451         }
1452
1453         /**
1454          * Simple & fast snippet extraction, but gives completely unrelevant
1455          * snippets
1456          *
1457          * @param $text String
1458          * @param $terms Array
1459          * @param $contextlines Integer
1460          * @param $contextchars Integer
1461          * @return String
1462          */
1463         public function highlightSimple( $text, $terms, $contextlines, $contextchars ) {
1464                 global $wgContLang;
1465                 $fname = __METHOD__;
1466
1467                 $lines = explode( "\n", $text );
1468
1469                 $terms = implode( '|', $terms );
1470                 $max = intval( $contextchars ) + 1;
1471                 $pat1 = "/(.*)($terms)(.{0,$max})/i";
1472
1473                 $lineno = 0;
1474
1475                 $extract = "";
1476                 wfProfileIn( "$fname-extract" );
1477                 foreach ( $lines as $line ) {
1478                         if ( 0 == $contextlines ) {
1479                                 break;
1480                         }
1481                         ++$lineno;
1482                         $m = array();
1483                         if ( ! preg_match( $pat1, $line, $m ) ) {
1484                                 continue;
1485                         }
1486                         --$contextlines;
1487                         // truncate function changes ... to relevant i18n message.
1488                         $pre = $wgContLang->truncate( $m[1], - $contextchars, '...', false );
1489
1490                         if ( count( $m ) < 3 ) {
1491                                 $post = '';
1492                         } else {
1493                                 $post = $wgContLang->truncate( $m[3], $contextchars, '...', false );
1494                         }
1495
1496                         $found = $m[2];
1497
1498                         $line = htmlspecialchars( $pre . $found . $post );
1499                         $pat2 = '/(' . $terms . ")/i";
1500                         $line = preg_replace( $pat2, "<span class='searchmatch'>\\1</span>", $line );
1501
1502                         $extract .= "${line}\n";
1503                 }
1504                 wfProfileOut( "$fname-extract" );
1505
1506                 return $extract;
1507         }
1508
1509 }
1510
1511 /**
1512  * Dummy class to be used when non-supported Database engine is present.
1513  * @todo FIXME: Dummy class should probably try something at least mildly useful,
1514  * such as a LIKE search through titles.
1515  * @ingroup Search
1516  */
1517 class SearchEngineDummy extends SearchEngine {
1518         // no-op
1519 }