includes/SearchEngine.php

   1 <?php
   2 /**
   3  * Contain a class for special pages
   4  * @addtogroup Search
   5  */
   6
   7 /**
   8  */
   9 class SearchEngine {
  10         var $limit = 10;
  11         var $offset = 0;
  12         var $searchTerms = array();
  13         var $namespaces = array( NS_MAIN );
  14         var $showRedirects = false;
  15
  16         /**
  17          * Perform a full text search query and return a result set.
  18          * If title searches are not supported or disabled, return null.
  19          *
  20          * @param string $term - Raw search term
  21          * @return SearchResultSet
  22          * @access public
  23          * @abstract
  24          */
  25         function searchText( $term ) {
  26                 return null;
  27         }
  28
  29         /**
  30          * Perform a title-only search query and return a result set.
  31          * If title searches are not supported or disabled, return null.
  32          *
  33          * @param string $term - Raw search term
  34          * @return SearchResultSet
  35          * @access public
  36          * @abstract
  37          */
  38         function searchTitle( $term ) {
  39                 return null;
  40         }
  41
  42         /**
  43          * If an exact title match can be find, or a very slightly close match,
  44          * return the title. If no match, returns NULL.
  45          *
  46          * @static
  47          * @param string $term
  48          * @return Title
  49          * @private
  50          */
  51         function getNearMatch( $searchterm ) {
  52                 global $wgContLang;
  53
  54                 $allSearchTerms = array($searchterm);
  55
  56                 if($wgContLang->hasVariants()){
  57                         $allSearchTerms = array_merge($allSearchTerms,$wgContLang->convertLinkToAllVariants($searchterm));
  58                 }
  59
  60                 foreach($allSearchTerms as $term){
  61
  62                         # Exact match? No need to look further.
  63                         $title = Title::newFromText( $term );
  64                         if (is_null($title))
  65                                 return NULL;
  66
  67                         if ( $title->getNamespace() == NS_SPECIAL || $title->exists() ) {
  68                                 return $title;
  69                         }
  70
  71                         # Now try all lower case (i.e. first letter capitalized)
  72                         #
  73                         $title = Title::newFromText( $wgContLang->lc( $term ) );
  74                         if ( $title->exists() ) {
  75                                 return $title;
  76                         }
  77
  78                         # Now try capitalized string
  79                         #
  80                         $title = Title::newFromText( $wgContLang->ucwords( $term ) );
  81                         if ( $title->exists() ) {
  82                                 return $title;
  83                         }
  84
  85                         # Now try all upper case
  86                         #
  87                         $title = Title::newFromText( $wgContLang->uc( $term ) );
  88                         if ( $title->exists() ) {
  89                                 return $title;
  90                         }
  91
  92                         # Now try Word-Caps-Breaking-At-Word-Breaks, for hyphenated names etc
  93                         $title = Title::newFromText( $wgContLang->ucwordbreaks($term) );
  94                         if ( $title->exists() ) {
  95                                 return $title;
  96                         }
  97
  98                         global $wgCapitalLinks, $wgContLang;
  99                         if( !$wgCapitalLinks ) {
 100                                 // Catch differs-by-first-letter-case-only
 101                                 $title = Title::newFromText( $wgContLang->ucfirst( $term ) );
 102                                 if ( $title->exists() ) {
 103                                         return $title;
 104                                 }
 105                                 $title = Title::newFromText( $wgContLang->lcfirst( $term ) );
 106                                 if ( $title->exists() ) {
 107                                         return $title;
 108                                 }
 109                         }
 110                 }
 111
 112                 $title = Title::newFromText( $searchterm );
 113
 114                 # Entering an IP address goes to the contributions page
 115                 if ( ( $title->getNamespace() == NS_USER && User::isIP($title->getText() ) )
 116                         || User::isIP( trim( $searchterm ) ) ) {
 117                         return SpecialPage::getTitleFor( 'Contributions', $title->getDbkey() );
 118                 }
 119
 120
 121                 # Entering a user goes to the user page whether it's there or not
 122                 if ( $title->getNamespace() == NS_USER ) {
 123                         return $title;
 124                 }
 125
 126                 # Quoted term? Try without the quotes...
 127                 $matches = array();
 128                 if( preg_match( '/^"([^"]+)"$/', $searchterm, $matches ) ) {
 129                         return SearchEngine::getNearMatch( $matches[1] );
 130                 }
 131
 132                 return NULL;
 133         }
 134
 135         public static function legalSearchChars() {
 136                 return "A-Za-z_'0-9\\x80-\\xFF\\-";
 137         }
 138
 139         /**
 140          * Set the maximum number of results to return
 141          * and how many to skip before returning the first.
 142          *
 143          * @param int $limit
 144          * @param int $offset
 145          * @access public
 146          */
 147         function setLimitOffset( $limit, $offset = 0 ) {
 148                 $this->limit = intval( $limit );
 149                 $this->offset = intval( $offset );
 150         }
 151
 152         /**
 153          * Set which namespaces the search should include.
 154          * Give an array of namespace index numbers.
 155          *
 156          * @param array $namespaces
 157          * @access public
 158          */
 159         function setNamespaces( $namespaces ) {
 160                 $this->namespaces = $namespaces;
 161         }
 162
 163         /**
 164          * Make a list of searchable namespaces and their canonical names.
 165          * @return array
 166          * @access public
 167          */
 168         function searchableNamespaces() {
 169                 global $wgContLang;
 170                 $arr = array();
 171                 foreach( $wgContLang->getNamespaces() as $ns => $name ) {
 172                         if( $ns >= NS_MAIN ) {
 173                                 $arr[$ns] = $name;
 174                         }
 175                 }
 176                 return $arr;
 177         }
 178
 179         /**
 180          * Return a 'cleaned up' search string
 181          *
 182          * @return string
 183          * @access public
 184          */
 185         function filter( $text ) {
 186                 $lc = $this->legalSearchChars();
 187                 return trim( preg_replace( "/[^{$lc}]/", " ", $text ) );
 188         }
 189         /**
 190          * Load up the appropriate search engine class for the currently
 191          * active database backend, and return a configured instance.
 192          *
 193          * @return SearchEngine
 194          */
 195         public static function create() {
 196                 global $wgDBtype, $wgSearchType;
 197                 if( $wgSearchType ) {
 198                         $class = $wgSearchType;
 199                 } elseif( $wgDBtype == 'mysql' ) {
 200                         $class = 'SearchMySQL4';
 201                 } else if ( $wgDBtype == 'postgres' ) {
 202                         $class = 'SearchPostgres';
 203                 } else if ( $wgDBtype == 'oracle' ) {
 204                         $class = 'SearchOracle';
 205                 } else {
 206                         $class = 'SearchEngineDummy';
 207                 }
 208                 $search = new $class( wfGetDB( DB_SLAVE ) );
 209                 $search->setLimitOffset(0,0);
 210                 return $search;
 211         }
 212
 213         /**
 214          * Create or update the search index record for the given page.
 215          * Title and text should be pre-processed.
 216          *
 217          * @param int $id
 218          * @param string $title
 219          * @param string $text
 220          * @abstract
 221          */
 222         function update( $id, $title, $text ) {
 223                 // no-op
 224         }
 225
 226         /**
 227          * Update a search index record's title only.
 228          * Title should be pre-processed.
 229          *
 230          * @param int $id
 231          * @param string $title
 232          * @abstract
 233          */
 234     function updateTitle( $id, $title ) {
 235                 // no-op
 236     }
 237 }
 238
 239 class SearchResultSet {
 240         /**
 241          * Fetch an array of regular expression fragments for matching
 242          * the search terms as parsed by this engine in a text extract.
 243          *
 244          * @return array
 245          * @access public
 246          * @abstract
 247          */
 248         function termMatches() {
 249                 return array();
 250         }
 251
 252         function numRows() {
 253                 return 0;
 254         }
 255
 256         /**
 257          * Return true if results are included in this result set.
 258          * @return bool
 259          * @abstract
 260          */
 261         function hasResults() {
 262                 return false;
 263         }
 264
 265         /**
 266          * Some search modes return a total hit count for the query
 267          * in the entire article database. This may include pages
 268          * in namespaces that would not be matched on the given
 269          * settings.
 270          *
 271          * Return null if no total hits number is supported.
 272          *
 273          * @return int
 274          * @access public
 275          */
 276         function getTotalHits() {
 277                 return null;
 278         }
 279
 280         /**
 281          * Some search modes return a suggested alternate term if there are
 282          * no exact hits. Returns true if there is one on this set.
 283          *
 284          * @return bool
 285          * @access public
 286          */
 287         function hasSuggestion() {
 288                 return false;
 289         }
 290
 291         /**
 292          * Some search modes return a suggested alternate term if there are
 293          * no exact hits. Check hasSuggestion() first.
 294          *
 295          * @return string
 296          * @access public
 297          */
 298         function getSuggestion() {
 299                 return '';
 300         }
 301
 302         /**
 303          * Fetches next search result, or false.
 304          * @return SearchResult
 305          * @access public
 306          * @abstract
 307          */
 308         function next() {
 309                 return false;
 310         }
 311 }
 312
 313 class SearchResult {
 314         function SearchResult( $row ) {
 315                 $this->mTitle = Title::makeTitle( $row->page_namespace, $row->page_title );
 316         }
 317
 318         /**
 319          * @return Title
 320          * @access public
 321          */
 322         function getTitle() {
 323                 return $this->mTitle;
 324         }
 325
 326         /**
 327          * @return double or null if not supported
 328          */
 329         function getScore() {
 330                 return null;
 331         }
 332 }
 333
 334 /**
 335  */
 336 class SearchEngineDummy {
 337         function search( $term ) {
 338                 return null;
 339         }
 340         function setLimitOffset($l, $o) {}
 341         function legalSearchChars() {}
 342         function update() {}
 343         function setnamespaces() {}
 344         function searchtitle() {}
 345         function searchtext() {}
 346 }
 347 ?>