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                 $allkeyword = wfMsg('searchall').":";
 194                 if( strncmp($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          * Return a 'cleaned up' search string
 228          *
 229          * @return string
 230          * @access public
 231          */
 232         function filter( $text ) {
 233                 $lc = $this->legalSearchChars();
 234                 return trim( preg_replace( "/[^{$lc}]/", " ", $text ) );
 235         }
 236         /**
 237          * Load up the appropriate search engine class for the currently
 238          * active database backend, and return a configured instance.
 239          *
 240          * @return SearchEngine
 241          */
 242         public static function create() {
 243                 global $wgDBtype, $wgSearchType;
 244                 if( $wgSearchType ) {
 245                         $class = $wgSearchType;
 246                 } elseif( $wgDBtype == 'mysql' ) {
 247                         $class = 'SearchMySQL';
 248                 } else if ( $wgDBtype == 'postgres' ) {
 249                         $class = 'SearchPostgres';
 250                 } else if ( $wgDBtype == 'oracle' ) {
 251                         $class = 'SearchOracle';
 252                 } else {
 253                         $class = 'SearchEngineDummy';
 254                 }
 255                 $search = new $class( wfGetDB( DB_SLAVE ) );
 256                 $search->setLimitOffset(0,0);
 257                 return $search;
 258         }
 259
 260         /**
 261          * Create or update the search index record for the given page.
 262          * Title and text should be pre-processed.
 263          *
 264          * @param int $id
 265          * @param string $title
 266          * @param string $text
 267          * @abstract
 268          */
 269         function update( $id, $title, $text ) {
 270                 // no-op
 271         }
 272
 273         /**
 274          * Update a search index record's title only.
 275          * Title should be pre-processed.
 276          *
 277          * @param int $id
 278          * @param string $title
 279          * @abstract
 280          */
 281         function updateTitle( $id, $title ) {
 282                 // no-op
 283         }
 284 }
 285
 286
 287 /**
 288  * @addtogroup Search
 289  */
 290 class SearchResultSet {
 291         /**
 292          * Fetch an array of regular expression fragments for matching
 293          * the search terms as parsed by this engine in a text extract.
 294          *
 295          * @return array
 296          * @access public
 297          * @abstract
 298          */
 299         function termMatches() {
 300                 return array();
 301         }
 302
 303         function numRows() {
 304                 return 0;
 305         }
 306
 307         /**
 308          * Return true if results are included in this result set.
 309          * @return bool
 310          * @abstract
 311          */
 312         function hasResults() {
 313                 return false;
 314         }
 315
 316         /**
 317          * Some search modes return a total hit count for the query
 318          * in the entire article database. This may include pages
 319          * in namespaces that would not be matched on the given
 320          * settings.
 321          *
 322          * Return null if no total hits number is supported.
 323          *
 324          * @return int
 325          * @access public
 326          */
 327         function getTotalHits() {
 328                 return null;
 329         }
 330
 331         /**
 332          * Some search modes return a suggested alternate term if there are
 333          * no exact hits. Returns true if there is one on this set.
 334          *
 335          * @return bool
 336          * @access public
 337          */
 338         function hasSuggestion() {
 339                 return false;
 340         }
 341
 342         /**
 343          * @return string suggested query, null if none
 344          */
 345         function getSuggestionQuery(){
 346                 return null;
 347         }
 348
 349         /**
 350          * @return string highlighted suggested query, '' if none
 351          */
 352         function getSuggestionSnippet(){
 353                 return '';
 354         }
 355
 356         /**
 357          * Fetches next search result, or false.
 358          * @return SearchResult
 359          * @access public
 360          * @abstract
 361          */
 362         function next() {
 363                 return false;
 364         }
 365
 366         /**
 367          * Frees the result set, if applicable.
 368          * @ access public
 369          */
 370         function free() {
 371                 // ...
 372         }
 373 }
 374
 375
 376 /**
 377  * @addtogroup Search
 378  */
 379 class SearchResultTooMany {
 380         ## Some search engines may bail out if too many matches are found
 381 }
 382
 383
 384 /**
 385  * @addtogroup Search
 386  */
 387 class SearchResult {
 388
 389         function SearchResult( $row ) {
 390                 $this->mTitle = Title::makeTitle( $row->page_namespace, $row->page_title );
 391         }
 392
 393         /**
 394          * @return Title
 395          * @access public
 396          */
 397         function getTitle() {
 398                 return $this->mTitle;
 399         }
 400
 401         /**
 402          * @return double or null if not supported
 403          */
 404         function getScore() {
 405                 return null;
 406         }
 407
 408         /**
 409          * @return string highlighted text snippet, null if not supported
 410          */
 411         function getTextSnippet(){
 412                 return null;
 413         }
 414
 415         /**
 416          * @return string highlighted title, '' if not supported
 417          */
 418         function getTitleSnippet(){
 419                 return '';
 420         }
 421
 422         /**
 423          * @return string highlighted redirect name (redirect to this page), '' if none or not supported
 424          */
 425         function getRedirectSnippet(){
 426                 return '';
 427         }
 428
 429         /**
 430          * @return Title object for the redirect to this page, null if none or not supported
 431          */
 432         function getRedirectTitle(){
 433                 return null;
 434         }
 435
 436         /**
 437          * @return string highlighted relevant section name, null if none or not supported
 438          */
 439         function getSectionSnippet(){
 440                 return '';
 441         }
 442
 443         /**
 444          * @return Title object (pagename+fragment) for the section, null if none or not supported
 445          */
 446         function getSectionTitle(){
 447                 return null;
 448         }
 449
 450         /**
 451          * @return string timestamp, null if not supported
 452          */
 453         function getTimestamp(){
 454                 return null;
 455         }
 456
 457         /**
 458          * @return int number of words, null if not supported
 459          */
 460         function getWordCount(){
 461                 return null;
 462         }
 463
 464         /**
 465          * @return int size in bytes, null if not supported
 466          */
 467         function getByteSize(){
 468                 return null;
 469         }
 470 }
 471
 472 /**
 473  * @addtogroup Search
 474  */
 475 class SearchEngineDummy {
 476         function search( $term ) {
 477                 return null;
 478         }
 479         function setLimitOffset($l, $o) {}
 480         function legalSearchChars() {}
 481         function update() {}
 482         function setnamespaces() {}
 483         function searchtitle() {}
 484         function searchtext() {}
 485 }
 486