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                 # Quoted term? Try without the quotes...
 124                 $matches = array();
 125                 if( preg_match( '/^"([^"]+)"$/', $searchterm, $matches ) ) {
 126                         return SearchEngine::getNearMatch( $matches[1] );
 127                 }
 128
 129                 return NULL;
 130         }
 131
 132         public static function legalSearchChars() {
 133                 return "A-Za-z_'0-9\\x80-\\xFF\\-";
 134         }
 135
 136         /**
 137          * Set the maximum number of results to return
 138          * and how many to skip before returning the first.
 139          *
 140          * @param int $limit
 141          * @param int $offset
 142          * @access public
 143          */
 144         function setLimitOffset( $limit, $offset = 0 ) {
 145                 $this->limit = intval( $limit );
 146                 $this->offset = intval( $offset );
 147         }
 148
 149         /**
 150          * Set which namespaces the search should include.
 151          * Give an array of namespace index numbers.
 152          *
 153          * @param array $namespaces
 154          * @access public
 155          */
 156         function setNamespaces( $namespaces ) {
 157                 $this->namespaces = $namespaces;
 158         }
 159
 160         /**
 161          * Make a list of searchable namespaces and their canonical names.
 162          * @return array
 163          * @access public
 164          */
 165         function searchableNamespaces() {
 166                 global $wgContLang;
 167                 $arr = array();
 168                 foreach( $wgContLang->getNamespaces() as $ns => $name ) {
 169                         if( $ns >= NS_MAIN ) {
 170                                 $arr[$ns] = $name;
 171                         }
 172                 }
 173                 return $arr;
 174         }
 175
 176         /**
 177          * Return a 'cleaned up' search string
 178          *
 179          * @return string
 180          * @access public
 181          */
 182         function filter( $text ) {
 183                 $lc = $this->legalSearchChars();
 184                 return trim( preg_replace( "/[^{$lc}]/", " ", $text ) );
 185         }
 186         /**
 187          * Load up the appropriate search engine class for the currently
 188          * active database backend, and return a configured instance.
 189          *
 190          * @return SearchEngine
 191          */
 192         public static function create() {
 193                 global $wgDBtype, $wgSearchType;
 194                 if( $wgSearchType ) {
 195                         $class = $wgSearchType;
 196                 } elseif( $wgDBtype == 'mysql' ) {
 197                         $class = 'SearchMySQL4';
 198                 } else if ( $wgDBtype == 'postgres' ) {
 199                         $class = 'SearchPostgres';
 200                 } else if ( $wgDBtype == 'oracle' ) {
 201                         $class = 'SearchOracle';
 202                 } else {
 203                         $class = 'SearchEngineDummy';
 204                 }
 205                 $search = new $class( wfGetDB( DB_SLAVE ) );
 206                 $search->setLimitOffset(0,0);
 207                 return $search;
 208         }
 209
 210         /**
 211          * Create or update the search index record for the given page.
 212          * Title and text should be pre-processed.
 213          *
 214          * @param int $id
 215          * @param string $title
 216          * @param string $text
 217          * @abstract
 218          */
 219         function update( $id, $title, $text ) {
 220                 // no-op
 221         }
 222
 223         /**
 224          * Update a search index record's title only.
 225          * Title should be pre-processed.
 226          *
 227          * @param int $id
 228          * @param string $title
 229          * @abstract
 230          */
 231     function updateTitle( $id, $title ) {
 232                 // no-op
 233     }
 234 }
 235
 236
 237 /**
 238  * @addtogroup Search
 239  */
 240 class SearchResultSet {
 241         /**
 242          * Fetch an array of regular expression fragments for matching
 243          * the search terms as parsed by this engine in a text extract.
 244          *
 245          * @return array
 246          * @access public
 247          * @abstract
 248          */
 249         function termMatches() {
 250                 return array();
 251         }
 252
 253         function numRows() {
 254                 return 0;
 255         }
 256
 257         /**
 258          * Return true if results are included in this result set.
 259          * @return bool
 260          * @abstract
 261          */
 262         function hasResults() {
 263                 return false;
 264         }
 265
 266         /**
 267          * Some search modes return a total hit count for the query
 268          * in the entire article database. This may include pages
 269          * in namespaces that would not be matched on the given
 270          * settings.
 271          *
 272          * Return null if no total hits number is supported.
 273          *
 274          * @return int
 275          * @access public
 276          */
 277         function getTotalHits() {
 278                 return null;
 279         }
 280
 281         /**
 282          * Some search modes return a suggested alternate term if there are
 283          * no exact hits. Returns true if there is one on this set.
 284          *
 285          * @return bool
 286          * @access public
 287          */
 288         function hasSuggestion() {
 289                 return false;
 290         }
 291
 292         /**
 293          * Some search modes return a suggested alternate term if there are
 294          * no exact hits. Check hasSuggestion() first.
 295          *
 296          * @return string
 297          * @access public
 298          */
 299         function getSuggestion() {
 300                 return '';
 301         }
 302
 303         /**
 304          * Fetches next search result, or false.
 305          * @return SearchResult
 306          * @access public
 307          * @abstract
 308          */
 309         function next() {
 310                 return false;
 311         }
 312 }
 313
 314
 315 /**
 316  * @addtogroup Search
 317  */
 318 class SearchResult {
 319         function SearchResult( $row ) {
 320                 $this->mTitle = Title::makeTitle( $row->page_namespace, $row->page_title );
 321         }
 322
 323         /**
 324          * @return Title
 325          * @access public
 326          */
 327         function getTitle() {
 328                 return $this->mTitle;
 329         }
 330
 331         /**
 332          * @return double or null if not supported
 333          */
 334         function getScore() {
 335                 return null;
 336         }
 337 }
 338
 339 /**
 340  * @addtogroup Search
 341  */
 342 class SearchEngineDummy {
 343         function search( $term ) {
 344                 return null;
 345         }
 346         function setLimitOffset($l, $o) {}
 347         function legalSearchChars() {}
 348         function update() {}
 349         function setnamespaces() {}
 350         function searchtitle() {}
 351         function searchtext() {}
 352 }
 353 ?>