includes/SearchEngine.php

   1 <?php
   2 /**
   3  * Contain a class for special pages
   4  * @addtogroup Search
   5  */
   6 class SearchEngine {
   7         var $limit = 10;
   8         var $offset = 0;
   9         var $searchTerms = array();
  10         var $namespaces = array( NS_MAIN );
  11         var $showRedirects = false;
  12
  13         /**
  14          * Perform a full text search query and return a result set.
  15          * If title searches are not supported or disabled, return null.
  16          *
  17          * @param string $term - Raw search term
  18          * @return SearchResultSet
  19          * @access public
  20          * @abstract
  21          */
  22         function searchText( $term ) {
  23                 return null;
  24         }
  25
  26         /**
  27          * Perform a title-only search query and return a result set.
  28          * If title searches are not supported or disabled, return null.
  29          *
  30          * @param string $term - Raw search term
  31          * @return SearchResultSet
  32          * @access public
  33          * @abstract
  34          */
  35         function searchTitle( $term ) {
  36                 return null;
  37         }
  38
  39         /**
  40          * If an exact title match can be find, or a very slightly close match,
  41          * return the title. If no match, returns NULL.
  42          *
  43          * @param string $term
  44          * @return Title
  45          */
  46         public static function getNearMatch( $searchterm ) {
  47                 global $wgContLang;
  48
  49                 $allSearchTerms = array($searchterm);
  50
  51                 if($wgContLang->hasVariants()){
  52                         $allSearchTerms = array_merge($allSearchTerms,$wgContLang->convertLinkToAllVariants($searchterm));
  53                 }
  54
  55                 foreach($allSearchTerms as $term){
  56
  57                         # Exact match? No need to look further.
  58                         $title = Title::newFromText( $term );
  59                         if (is_null($title))
  60                                 return NULL;
  61
  62                         if ( $title->getNamespace() == NS_SPECIAL || $title->exists() ) {
  63                                 return $title;
  64                         }
  65
  66                         # Now try all lower case (i.e. first letter capitalized)
  67                         #
  68                         $title = Title::newFromText( $wgContLang->lc( $term ) );
  69                         if ( $title->exists() ) {
  70                                 return $title;
  71                         }
  72
  73                         # Now try capitalized string
  74                         #
  75                         $title = Title::newFromText( $wgContLang->ucwords( $term ) );
  76                         if ( $title->exists() ) {
  77                                 return $title;
  78                         }
  79
  80                         # Now try all upper case
  81                         #
  82                         $title = Title::newFromText( $wgContLang->uc( $term ) );
  83                         if ( $title->exists() ) {
  84                                 return $title;
  85                         }
  86
  87                         # Now try Word-Caps-Breaking-At-Word-Breaks, for hyphenated names etc
  88                         $title = Title::newFromText( $wgContLang->ucwordbreaks($term) );
  89                         if ( $title->exists() ) {
  90                                 return $title;
  91                         }
  92
  93                         global $wgCapitalLinks, $wgContLang;
  94                         if( !$wgCapitalLinks ) {
  95                                 // Catch differs-by-first-letter-case-only
  96                                 $title = Title::newFromText( $wgContLang->ucfirst( $term ) );
  97                                 if ( $title->exists() ) {
  98                                         return $title;
  99                                 }
 100                                 $title = Title::newFromText( $wgContLang->lcfirst( $term ) );
 101                                 if ( $title->exists() ) {
 102                                         return $title;
 103                                 }
 104                         }
 105
 106                         // Give hooks a chance at better match variants
 107                         $title = null;
 108                         if( !wfRunHooks( 'SearchGetNearMatch', array( $term, &$title ) ) ) {
 109                                 return $title;
 110                         }
 111                 }
 112
 113                 $title = Title::newFromText( $searchterm );
 114
 115                 # Entering an IP address goes to the contributions page
 116                 if ( ( $title->getNamespace() == NS_USER && User::isIP($title->getText() ) )
 117                         || User::isIP( trim( $searchterm ) ) ) {
 118                         return SpecialPage::getTitleFor( 'Contributions', $title->getDBkey() );
 119                 }
 120
 121
 122                 # Entering a user goes to the user page whether it's there or not
 123                 if ( $title->getNamespace() == NS_USER ) {
 124                         return $title;
 125                 }
 126
 127                 # Go to images that exist even if there's no local page.
 128                 # There may have been a funny upload, or it may be on a shared
 129                 # file repository such as Wikimedia Commons.
 130                 if( $title->getNamespace() == NS_IMAGE ) {
 131                         $image = wfFindFile( $title );
 132                         if( $image ) {
 133                                 return $title;
 134                         }
 135                 }
 136
 137                 # MediaWiki namespace? Page may be "implied" if not customized.
 138                 # Just return it, with caps forced as the message system likes it.
 139                 if( $title->getNamespace() == NS_MEDIAWIKI ) {
 140                         return Title::makeTitle( NS_MEDIAWIKI, $wgContLang->ucfirst( $title->getText() ) );
 141                 }
 142
 143                 # Quoted term? Try without the quotes...
 144                 $matches = array();
 145                 if( preg_match( '/^"([^"]+)"$/', $searchterm, $matches ) ) {
 146                         return SearchEngine::getNearMatch( $matches[1] );
 147                 }
 148
 149                 return NULL;
 150         }
 151
 152         public static function legalSearchChars() {
 153                 return "A-Za-z_'0-9\\x80-\\xFF\\-";
 154         }
 155
 156         /**
 157          * Set the maximum number of results to return
 158          * and how many to skip before returning the first.
 159          *
 160          * @param int $limit
 161          * @param int $offset
 162          * @access public
 163          */
 164         function setLimitOffset( $limit, $offset = 0 ) {
 165                 $this->limit = intval( $limit );
 166                 $this->offset = intval( $offset );
 167         }
 168
 169         /**
 170          * Set which namespaces the search should include.
 171          * Give an array of namespace index numbers.
 172          *
 173          * @param array $namespaces
 174          * @access public
 175          */
 176         function setNamespaces( $namespaces ) {
 177                 $this->namespaces = $namespaces;
 178         }
 179
 180         /**
 181          * Parse some common prefixes: all (search everything)
 182          * or namespace names
 183          *
 184          * @param string $query
 185          */
 186         function replacePrefixes( $query ){
 187                 global $wgContLang;
 188
 189                 if( strpos($query,':') === false )
 190                         return $query; // nothing to do
 191
 192                 $parsed = $query;
 193                 $allkeyword = wfMsgForContent('searchall').":";
 194                 if( strncmp($query, $allkeyword, strlen($allkeyword)) == 0 ){
 195                         $this->namespaces = null;
 196                         $parsed = substr($query,strlen($allkeyword));
 197                 } else if( strpos($query,':') !== false ) {
 198                         $prefix = substr($query,0,strpos($query,':'));
 199                         $index = $wgContLang->getNsIndex($prefix);
 200                         if($index !== false){
 201                                 $this->namespaces = array($index);
 202                                 $parsed = substr($query,strlen($prefix)+1);
 203                         }
 204                 }
 205                 if(trim($parsed) == '')
 206                         return $query; // prefix was the whole query
 207
 208                 return $parsed;
 209         }
 210
 211         /**
 212          * Make a list of searchable namespaces and their canonical names.
 213          * @return array
 214          */
 215         public static function searchableNamespaces() {
 216                 global $wgContLang;
 217                 $arr = array();
 218                 foreach( $wgContLang->getNamespaces() as $ns => $name ) {
 219                         if( $ns >= NS_MAIN ) {
 220                                 $arr[$ns] = $name;
 221                         }
 222                 }
 223                 return $arr;
 224         }
 225
 226         /**
 227          * Extract default namespaces to search from the given user's
 228          * settings, returning a list of index numbers.
 229          *
 230          * @param User $user
 231          * @return array
 232          * @static
 233          */
 234         public static function userNamespaces( &$user ) {
 235                 $arr = array();
 236                 foreach( SearchEngine::searchableNamespaces() as $ns => $name ) {
 237                         if( $user->getOption( 'searchNs' . $ns ) ) {
 238                                 $arr[] = $ns;
 239                         }
 240                 }
 241                 return $arr;
 242         }
 243
 244         /**
 245          * Find snippet highlight settings for a given user
 246          *
 247          * @param User $user
 248          * @return array contextlines, contextchars
 249          * @static
 250          */
 251         public static function userHighlightPrefs( &$user ){
 252                 //$contextlines = $user->getOption( 'contextlines',  5 );
 253                 $contextlines = 2; // Hardcode this. Old defaults sucked. :)
 254                 $contextchars = $user->getOption( 'contextchars', 50 );
 255                 return array($contextlines, $contextchars);
 256         }
 257
 258         /**
 259          * An array of namespaces indexes to be searched by default
 260          *
 261          * @return array
 262          * @static
 263          */
 264         public static function defaultNamespaces(){
 265                 global $wgNamespacesToBeSearchedDefault;
 266
 267                 return array_keys($wgNamespacesToBeSearchedDefault, true);
 268         }
 269
 270         /**
 271          * Return a 'cleaned up' search string
 272          *
 273          * @return string
 274          * @access public
 275          */
 276         function filter( $text ) {
 277                 $lc = $this->legalSearchChars();
 278                 return trim( preg_replace( "/[^{$lc}]/", " ", $text ) );
 279         }
 280         /**
 281          * Load up the appropriate search engine class for the currently
 282          * active database backend, and return a configured instance.
 283          *
 284          * @return SearchEngine
 285          */
 286         public static function create() {
 287                 global $wgDBtype, $wgSearchType;
 288                 if( $wgSearchType ) {
 289                         $class = $wgSearchType;
 290                 } elseif( $wgDBtype == 'mysql' ) {
 291                         $class = 'SearchMySQL';
 292                 } else if ( $wgDBtype == 'postgres' ) {
 293                         $class = 'SearchPostgres';
 294                 } else if ( $wgDBtype == 'oracle' ) {
 295                         $class = 'SearchOracle';
 296                 } else {
 297                         $class = 'SearchEngineDummy';
 298                 }
 299                 $search = new $class( wfGetDB( DB_SLAVE ) );
 300                 $search->setLimitOffset(0,0);
 301                 return $search;
 302         }
 303
 304         /**
 305          * Create or update the search index record for the given page.
 306          * Title and text should be pre-processed.
 307          *
 308          * @param int $id
 309          * @param string $title
 310          * @param string $text
 311          * @abstract
 312          */
 313         function update( $id, $title, $text ) {
 314                 // no-op
 315         }
 316
 317         /**
 318          * Update a search index record's title only.
 319          * Title should be pre-processed.
 320          *
 321          * @param int $id
 322          * @param string $title
 323          * @abstract
 324          */
 325         function updateTitle( $id, $title ) {
 326                 // no-op
 327         }
 328
 329         /**
 330          * Get OpenSearch suggestion template
 331          *
 332          * @return string
 333          * @static
 334          */
 335         public static function getOpenSearchTemplate() {
 336                 global $wgOpenSearchTemplate, $wgServer, $wgScriptPath;
 337                 if($wgOpenSearchTemplate)
 338                         return $wgOpenSearchTemplate;
 339                 else{
 340                         $ns = implode(',',SearchEngine::defaultNamespaces());
 341                         if(!$ns) $ns = "0";
 342                         return $wgServer . $wgScriptPath . '/api.php?action=opensearch&search={searchTerms}&namespace='.$ns;
 343                 }
 344         }
 345
 346         /**
 347          * Get internal MediaWiki Suggest template
 348          *
 349          * @return string
 350          * @static
 351          */
 352         public static function getMWSuggestTemplate() {
 353                 global $wgMWSuggestTemplate, $wgServer, $wgScriptPath;
 354                 if($wgMWSuggestTemplate)
 355                         return $wgMWSuggestTemplate;
 356                 else
 357                         return $wgServer . $wgScriptPath . '/api.php?action=opensearch&search={searchTerms}&namespace={namespaces}';
 358         }
 359 }
 360
 361
 362 /**
 363  * @addtogroup Search
 364  */
 365 class SearchResultSet {
 366         /**
 367          * Fetch an array of regular expression fragments for matching
 368          * the search terms as parsed by this engine in a text extract.
 369          *
 370          * @return array
 371          * @access public
 372          * @abstract
 373          */
 374         function termMatches() {
 375                 return array();
 376         }
 377
 378         function numRows() {
 379                 return 0;
 380         }
 381
 382         /**
 383          * Return true if results are included in this result set.
 384          * @return bool
 385          * @abstract
 386          */
 387         function hasResults() {
 388                 return false;
 389         }
 390
 391         /**
 392          * Some search modes return a total hit count for the query
 393          * in the entire article database. This may include pages
 394          * in namespaces that would not be matched on the given
 395          * settings.
 396          *
 397          * Return null if no total hits number is supported.
 398          *
 399          * @return int
 400          * @access public
 401          */
 402         function getTotalHits() {
 403                 return null;
 404         }
 405
 406         /**
 407          * Some search modes return a suggested alternate term if there are
 408          * no exact hits. Returns true if there is one on this set.
 409          *
 410          * @return bool
 411          * @access public
 412          */
 413         function hasSuggestion() {
 414                 return false;
 415         }
 416
 417         /**
 418          * @return string suggested query, null if none
 419          */
 420         function getSuggestionQuery(){
 421                 return null;
 422         }
 423
 424         /**
 425          * @return string highlighted suggested query, '' if none
 426          */
 427         function getSuggestionSnippet(){
 428                 return '';
 429         }
 430
 431         /**
 432          * Return information about how and from where the results were fetched,
 433          * should be useful for diagnostics and debugging
 434          *
 435          * @return string
 436          */
 437         function getInfo() {
 438                 return null;
 439         }
 440
 441         /**
 442          * Return a result set of hits on other (multiple) wikis associated with this one
 443          *
 444          * @return SearchResultSet
 445          */
 446         function getInterwikiResults() {
 447                 return null;
 448         }
 449
 450         /**
 451          * Check if there are results on other wikis
 452          *
 453          * @return boolean
 454          */
 455         function hasInterwikiResults() {
 456                 return $this->getInterwikiResults() != null;
 457         }
 458
 459
 460         /**
 461          * Fetches next search result, or false.
 462          * @return SearchResult
 463          * @access public
 464          * @abstract
 465          */
 466         function next() {
 467                 return false;
 468         }
 469
 470         /**
 471          * Frees the result set, if applicable.
 472          * @ access public
 473          */
 474         function free() {
 475                 // ...
 476         }
 477 }
 478
 479
 480 /**
 481  * @addtogroup Search
 482  */
 483 class SearchResultTooMany {
 484         ## Some search engines may bail out if too many matches are found
 485 }
 486
 487
 488 /**
 489  * @addtogroup Search
 490  */
 491 class SearchResult {
 492         var $mRevision = null;
 493
 494         function SearchResult( $row ) {
 495                 $this->mTitle = Title::makeTitle( $row->page_namespace, $row->page_title );
 496                 if( !is_null($this->mTitle) )
 497                         $this->mRevision = Revision::newFromTitle( $this->mTitle );
 498         }
 499
 500         /**
 501          * Check if this is result points to an invalid title
 502          *
 503          * @return boolean
 504          * @access public
 505          */
 506         function isBrokenTitle(){
 507                 if( is_null($this->mTitle) )
 508                         return true;
 509                 return false;
 510         }
 511
 512         /**
 513          * Check if target page is missing, happens when index is out of date
 514          *
 515          * @return boolean
 516          * @access public
 517          */
 518         function isMissingRevision(){
 519                 if( !$this->mRevision )
 520                         return true;
 521                 return false;
 522         }
 523
 524         /**
 525          * @return Title
 526          * @access public
 527          */
 528         function getTitle() {
 529                 return $this->mTitle;
 530         }
 531
 532         /**
 533          * @return double or null if not supported
 534          */
 535         function getScore() {
 536                 return null;
 537         }
 538
 539         /**
 540          * Lazy initialization of article text from DB
 541          */
 542         protected function initText(){
 543                 if( !isset($this->mText) ){
 544                         $this->mText = $this->mRevision->getText();
 545                 }
 546         }
 547
 548         /**
 549          * @param array $terms terms to highlight
 550          * @return string highlighted text snippet, null (and not '') if not supported
 551          */
 552         function getTextSnippet($terms){
 553                 global $wgUser;
 554                 $this->initText();
 555                 list($contextlines,$contextchars) = SearchEngine::userHighlightPrefs($wgUser);
 556                 return $this->extractText( $this->mText, $terms, $contextlines, $contextchars);
 557         }
 558
 559         /**
 560          * Default implementation of snippet extraction
 561          *
 562          * @param string $text
 563          * @param array $terms
 564          * @param int $contextlines
 565          * @param int $contextchars
 566          * @return string
 567          */
 568         protected function extractText( $text, $terms, $contextlines, $contextchars ) {
 569                 global $wgLang, $wgContLang;
 570                 $fname = __METHOD__;
 571
 572                 $lines = explode( "\n", $text );
 573
 574                 $terms = implode( '|', $terms );
 575                 $terms = str_replace( '/', "\\/", $terms);
 576                 $max = intval( $contextchars ) + 1;
 577                 $pat1 = "/(.*)($terms)(.{0,$max})/i";
 578
 579                 $lineno = 0;
 580
 581                 $extract = "";
 582                 wfProfileIn( "$fname-extract" );
 583                 foreach ( $lines as $line ) {
 584                         if ( 0 == $contextlines ) {
 585                                 break;
 586                         }
 587                         ++$lineno;
 588                         $m = array();
 589                         if ( ! preg_match( $pat1, $line, $m ) ) {
 590                                 continue;
 591                         }
 592                         --$contextlines;
 593                         $pre = $wgContLang->truncate( $m[1], -$contextchars, ' ... ' );
 594
 595                         if ( count( $m ) < 3 ) {
 596                                 $post = '';
 597                         } else {
 598                                 $post = $wgContLang->truncate( $m[3], $contextchars, ' ... ' );
 599                         }
 600
 601                         $found = $m[2];
 602
 603                         $line = htmlspecialchars( $pre . $found . $post );
 604                         $pat2 = '/(' . $terms . ")/i";
 605                         $line = preg_replace( $pat2,
 606                           "<span class='searchmatch'>\\1</span>", $line );
 607
 608                         $extract .= "${line}\n";
 609                 }
 610                 wfProfileOut( "$fname-extract" );
 611
 612                 return $extract;
 613         }
 614
 615         /**
 616          * @param array $terms terms to highlight
 617          * @return string highlighted title, '' if not supported
 618          */
 619         function getTitleSnippet($terms){
 620                 return '';
 621         }
 622
 623         /**
 624          * @param array $terms terms to highlight
 625          * @return string highlighted redirect name (redirect to this page), '' if none or not supported
 626          */
 627         function getRedirectSnippet($terms){
 628                 return '';
 629         }
 630
 631         /**
 632          * @return Title object for the redirect to this page, null if none or not supported
 633          */
 634         function getRedirectTitle(){
 635                 return null;
 636         }
 637
 638         /**
 639          * @return string highlighted relevant section name, null if none or not supported
 640          */
 641         function getSectionSnippet(){
 642                 return '';
 643         }
 644
 645         /**
 646          * @return Title object (pagename+fragment) for the section, null if none or not supported
 647          */
 648         function getSectionTitle(){
 649                 return null;
 650         }
 651
 652         /**
 653          * @return string timestamp
 654          */
 655         function getTimestamp(){
 656                 return $this->mRevision->getTimestamp();
 657         }
 658
 659         /**
 660          * @return int number of words
 661          */
 662         function getWordCount(){
 663                 $this->initText();
 664                 return str_word_count( $this->mText );
 665         }
 666
 667         /**
 668          * @return int size in bytes
 669          */
 670         function getByteSize(){
 671                 $this->initText();
 672                 return strlen( $this->mText );
 673         }
 674
 675         /**
 676          * @return boolean if hit has related articles
 677          */
 678         function hasRelated(){
 679                 return false;
 680         }
 681
 682         /**
 683          * @return interwiki prefix of the title (return iw even if title is broken)
 684          */
 685         function getInterwikiPrefix(){
 686                 return '';
 687         }
 688 }
 689
 690 /**
 691  * @addtogroup Search
 692  */
 693 class SearchEngineDummy {
 694         function search( $term ) {
 695                 return null;
 696         }
 697         function setLimitOffset($l, $o) {}
 698         function legalSearchChars() {}
 699         function update() {}
 700         function setnamespaces() {}
 701         function searchtitle() {}
 702         function searchtext() {}
 703 }