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          * @static
  44          * @param string $term
  45          * @return Title
  46          * @private
  47          */
  48         function getNearMatch( $searchterm ) {
  49                 global $wgContLang;
  50
  51                 $allSearchTerms = array($searchterm);
  52
  53                 if($wgContLang->hasVariants()){
  54                         $allSearchTerms = array_merge($allSearchTerms,$wgContLang->convertLinkToAllVariants($searchterm));
  55                 }
  56
  57                 foreach($allSearchTerms as $term){
  58
  59                         # Exact match? No need to look further.
  60                         $title = Title::newFromText( $term );
  61                         if (is_null($title))
  62                                 return NULL;
  63
  64                         if ( $title->getNamespace() == NS_SPECIAL || $title->exists() ) {
  65                                 return $title;
  66                         }
  67
  68                         # Now try all lower case (i.e. first letter capitalized)
  69                         #
  70                         $title = Title::newFromText( $wgContLang->lc( $term ) );
  71                         if ( $title->exists() ) {
  72                                 return $title;
  73                         }
  74
  75                         # Now try capitalized string
  76                         #
  77                         $title = Title::newFromText( $wgContLang->ucwords( $term ) );
  78                         if ( $title->exists() ) {
  79                                 return $title;
  80                         }
  81
  82                         # Now try all upper case
  83                         #
  84                         $title = Title::newFromText( $wgContLang->uc( $term ) );
  85                         if ( $title->exists() ) {
  86                                 return $title;
  87                         }
  88
  89                         # Now try Word-Caps-Breaking-At-Word-Breaks, for hyphenated names etc
  90                         $title = Title::newFromText( $wgContLang->ucwordbreaks($term) );
  91                         if ( $title->exists() ) {
  92                                 return $title;
  93                         }
  94
  95                         global $wgCapitalLinks, $wgContLang;
  96                         if( !$wgCapitalLinks ) {
  97                                 // Catch differs-by-first-letter-case-only
  98                                 $title = Title::newFromText( $wgContLang->ucfirst( $term ) );
  99                                 if ( $title->exists() ) {
 100                                         return $title;
 101                                 }
 102                                 $title = Title::newFromText( $wgContLang->lcfirst( $term ) );
 103                                 if ( $title->exists() ) {
 104                                         return $title;
 105                                 }
 106                         }
 107                 }
 108
 109                 $title = Title::newFromText( $searchterm );
 110
 111                 # Entering an IP address goes to the contributions page
 112                 if ( ( $title->getNamespace() == NS_USER && User::isIP($title->getText() ) )
 113                         || User::isIP( trim( $searchterm ) ) ) {
 114                         return SpecialPage::getTitleFor( 'Contributions', $title->getDbkey() );
 115                 }
 116
 117
 118                 # Entering a user goes to the user page whether it's there or not
 119                 if ( $title->getNamespace() == NS_USER ) {
 120                         return $title;
 121                 }
 122
 123                 # Go to images that exist even if there's no local page.
 124                 # There may have been a funny upload, or it may be on a shared
 125                 # file repository such as Wikimedia Commons.
 126                 if( $title->getNamespace() == NS_IMAGE ) {
 127                         $image = new Image( $title );
 128                         if( $image->exists() ) {
 129                                 return $title;
 130                         }
 131                 }
 132
 133                 # MediaWiki namespace? Page may be "implied" if not customized.
 134                 # Just return it, with caps forced as the message system likes it.
 135                 if( $title->getNamespace() == NS_MEDIAWIKI ) {
 136                         return Title::makeTitle( NS_MEDIAWIKI, $wgContLang->ucfirst( $title->getText() ) );
 137                 }
 138
 139                 # Quoted term? Try without the quotes...
 140                 $matches = array();
 141                 if( preg_match( '/^"([^"]+)"$/', $searchterm, $matches ) ) {
 142                         return SearchEngine::getNearMatch( $matches[1] );
 143                 }
 144
 145                 return NULL;
 146         }
 147
 148         public static function legalSearchChars() {
 149                 return "A-Za-z_'0-9\\x80-\\xFF\\-";
 150         }
 151
 152         /**
 153          * Set the maximum number of results to return
 154          * and how many to skip before returning the first.
 155          *
 156          * @param int $limit
 157          * @param int $offset
 158          * @access public
 159          */
 160         function setLimitOffset( $limit, $offset = 0 ) {
 161                 $this->limit = intval( $limit );
 162                 $this->offset = intval( $offset );
 163         }
 164
 165         /**
 166          * Set which namespaces the search should include.
 167          * Give an array of namespace index numbers.
 168          *
 169          * @param array $namespaces
 170          * @access public
 171          */
 172         function setNamespaces( $namespaces ) {
 173                 $this->namespaces = $namespaces;
 174         }
 175
 176         /**
 177          * Make a list of searchable namespaces and their canonical names.
 178          * @return array
 179          * @access public
 180          */
 181         function searchableNamespaces() {
 182                 global $wgContLang;
 183                 $arr = array();
 184                 foreach( $wgContLang->getNamespaces() as $ns => $name ) {
 185                         if( $ns >= NS_MAIN ) {
 186                                 $arr[$ns] = $name;
 187                         }
 188                 }
 189                 return $arr;
 190         }
 191
 192         /**
 193          * Return a 'cleaned up' search string
 194          *
 195          * @return string
 196          * @access public
 197          */
 198         function filter( $text ) {
 199                 $lc = $this->legalSearchChars();
 200                 return trim( preg_replace( "/[^{$lc}]/", " ", $text ) );
 201         }
 202         /**
 203          * Load up the appropriate search engine class for the currently
 204          * active database backend, and return a configured instance.
 205          *
 206          * @return SearchEngine
 207          */
 208         public static function create() {
 209                 global $wgDBtype, $wgSearchType;
 210                 if( $wgSearchType ) {
 211                         $class = $wgSearchType;
 212                 } elseif( $wgDBtype == 'mysql' ) {
 213                         $class = 'SearchMySQL4';
 214                 } else if ( $wgDBtype == 'postgres' ) {
 215                         $class = 'SearchPostgres';
 216                 } else if ( $wgDBtype == 'oracle' ) {
 217                         $class = 'SearchOracle';
 218                 } else {
 219                         $class = 'SearchEngineDummy';
 220                 }
 221                 $search = new $class( wfGetDB( DB_SLAVE ) );
 222                 $search->setLimitOffset(0,0);
 223                 return $search;
 224         }
 225
 226         /**
 227          * Create or update the search index record for the given page.
 228          * Title and text should be pre-processed.
 229          *
 230          * @param int $id
 231          * @param string $title
 232          * @param string $text
 233          * @abstract
 234          */
 235         function update( $id, $title, $text ) {
 236                 // no-op
 237         }
 238
 239         /**
 240          * Update a search index record's title only.
 241          * Title should be pre-processed.
 242          *
 243          * @param int $id
 244          * @param string $title
 245          * @abstract
 246          */
 247         function updateTitle( $id, $title ) {
 248                 // no-op
 249         }
 250 }
 251
 252
 253 /**
 254  * @addtogroup Search
 255  */
 256 class SearchResultSet {
 257         /**
 258          * Fetch an array of regular expression fragments for matching
 259          * the search terms as parsed by this engine in a text extract.
 260          *
 261          * @return array
 262          * @access public
 263          * @abstract
 264          */
 265         function termMatches() {
 266                 return array();
 267         }
 268
 269         function numRows() {
 270                 return 0;
 271         }
 272
 273         /**
 274          * Return true if results are included in this result set.
 275          * @return bool
 276          * @abstract
 277          */
 278         function hasResults() {
 279                 return false;
 280         }
 281
 282         /**
 283          * Some search modes return a total hit count for the query
 284          * in the entire article database. This may include pages
 285          * in namespaces that would not be matched on the given
 286          * settings.
 287          *
 288          * Return null if no total hits number is supported.
 289          *
 290          * @return int
 291          * @access public
 292          */
 293         function getTotalHits() {
 294                 return null;
 295         }
 296
 297         /**
 298          * Some search modes return a suggested alternate term if there are
 299          * no exact hits. Returns true if there is one on this set.
 300          *
 301          * @return bool
 302          * @access public
 303          */
 304         function hasSuggestion() {
 305                 return false;
 306         }
 307
 308         /**
 309          * Some search modes return a suggested alternate term if there are
 310          * no exact hits. Check hasSuggestion() first.
 311          *
 312          * @return string
 313          * @access public
 314          */
 315         function getSuggestion() {
 316                 return '';
 317         }
 318
 319         /**
 320          * Fetches next search result, or false.
 321          * @return SearchResult
 322          * @access public
 323          * @abstract
 324          */
 325         function next() {
 326                 return false;
 327         }
 328 }
 329
 330
 331 /**
 332  * @addtogroup Search
 333  */
 334 class SearchResult {
 335         function SearchResult( $row ) {
 336                 $this->mTitle = Title::makeTitle( $row->page_namespace, $row->page_title );
 337         }
 338
 339         /**
 340          * @return Title
 341          * @access public
 342          */
 343         function getTitle() {
 344                 return $this->mTitle;
 345         }
 346
 347         /**
 348          * @return double or null if not supported
 349          */
 350         function getScore() {
 351                 return null;
 352         }
 353 }
 354
 355 /**
 356  * @addtogroup Search
 357  */
 358 class SearchEngineDummy {
 359         function search( $term ) {
 360                 return null;
 361         }
 362         function setLimitOffset($l, $o) {}
 363         function legalSearchChars() {}
 364         function update() {}
 365         function setnamespaces() {}
 366         function searchtitle() {}
 367         function searchtext() {}
 368 }
 369 ?>