includes/SearchEngine.php

   1 <?php
   2 /**
   3  * Contain a class for special pages
   4  * @package MediaWiki
   5  * @subpackage Search
   6  */
   7
   8 /**
   9  * @package MediaWiki
  10  */
  11 class SearchEngine {
  12         var $limit = 10;
  13         var $offset = 0;
  14         var $searchTerms = array();
  15         var $namespaces = array( NS_MAIN );
  16         var $showRedirects = false;
  17
  18         /**
  19          * Perform a full text search query and return a result set.
  20          * If title searches are not supported or disabled, return null.
  21          *
  22          * @param string $term - Raw search term
  23          * @return SearchResultSet
  24          * @access public
  25          * @abstract
  26          */
  27         function searchText( $term ) {
  28                 return null;
  29         }
  30
  31         /**
  32          * Perform a title-only search query and return a result set.
  33          * If title searches are not supported or disabled, return null.
  34          *
  35          * @param string $term - Raw search term
  36          * @return SearchResultSet
  37          * @access public
  38          * @abstract
  39          */
  40         function searchTitle( $term ) {
  41                 return null;
  42         }
  43
  44         /**
  45          * If an exact title match can be find, or a very slightly close match,
  46          * return the title. If no match, returns NULL.
  47          *
  48          * @static
  49          * @param string $term
  50          * @return Title
  51          * @private
  52          */
  53         function getNearMatch( $searchterm ) {
  54                 global $wgContLang;
  55
  56                 $allSearchTerms = array($searchterm);
  57
  58                 if($wgContLang->hasVariants()){
  59                         $allSearchTerms = array_merge($allSearchTerms,$wgContLang->convertLinkToAllVariants($searchterm));
  60                 }
  61
  62                 foreach($allSearchTerms as $term){
  63
  64                         # Exact match? No need to look further.
  65                         $title = Title::newFromText( $term );
  66                         if (is_null($title))
  67                                 return NULL;
  68
  69                         if ( $title->getNamespace() == NS_SPECIAL || $title->exists() ) {
  70                                 return $title;
  71                         }
  72
  73                         # Now try all lower case (i.e. first letter capitalized)
  74                         #
  75                         $title = Title::newFromText( $wgContLang->lc( $term ) );
  76                         if ( $title->exists() ) {
  77                                 return $title;
  78                         }
  79
  80                         # Now try capitalized string
  81                         #
  82                         $title = Title::newFromText( $wgContLang->ucwords( $term ) );
  83                         if ( $title->exists() ) {
  84                                 return $title;
  85                         }
  86
  87                         # Now try all upper case
  88                         #
  89                         $title = Title::newFromText( $wgContLang->uc( $term ) );
  90                         if ( $title->exists() ) {
  91                                 return $title;
  92                         }
  93
  94                         # Now try Word-Caps-Breaking-At-Word-Breaks, for hyphenated names etc
  95                         $title = Title::newFromText( $wgContLang->ucwordbreaks($term) );
  96                         if ( $title->exists() ) {
  97                                 return $title;
  98                         }
  99
 100                         global $wgCapitalLinks, $wgContLang;
 101                         if( !$wgCapitalLinks ) {
 102                                 // Catch differs-by-first-letter-case-only
 103                                 $title = Title::newFromText( $wgContLang->ucfirst( $term ) );
 104                                 if ( $title->exists() ) {
 105                                         return $title;
 106                                 }
 107                                 $title = Title::newFromText( $wgContLang->lcfirst( $term ) );
 108                                 if ( $title->exists() ) {
 109                                         return $title;
 110                                 }
 111                         }
 112                 }
 113
 114                 $title = Title::newFromText( $searchterm );
 115
 116                 # Entering an IP address goes to the contributions page
 117                 if ( ( $title->getNamespace() == NS_USER && User::isIP($title->getText() ) )
 118                         || User::isIP( trim( $searchterm ) ) ) {
 119                         return SpecialPage::getTitleFor( 'Contributions', $title->getDbkey() );
 120                 }
 121
 122
 123                 # Entering a user goes to the user page whether it's there or not
 124                 if ( $title->getNamespace() == NS_USER ) {
 125                         return $title;
 126                 }
 127
 128                 # Quoted term? Try without the quotes...
 129                 $matches = array();
 130                 if( preg_match( '/^"([^"]+)"$/', $searchterm, $matches ) ) {
 131                         return SearchEngine::getNearMatch( $matches[1] );
 132                 }
 133
 134                 return NULL;
 135         }
 136
 137         function legalSearchChars() {
 138                 return "A-Za-z_'0-9\\x80-\\xFF\\-";
 139         }
 140
 141         /**
 142          * Set the maximum number of results to return
 143          * and how many to skip before returning the first.
 144          *
 145          * @param int $limit
 146          * @param int $offset
 147          * @access public
 148          */
 149         function setLimitOffset( $limit, $offset = 0 ) {
 150                 $this->limit = intval( $limit );
 151                 $this->offset = intval( $offset );
 152         }
 153
 154         /**
 155          * Set which namespaces the search should include.
 156          * Give an array of namespace index numbers.
 157          *
 158          * @param array $namespaces
 159          * @access public
 160          */
 161         function setNamespaces( $namespaces ) {
 162                 $this->namespaces = $namespaces;
 163         }
 164
 165         /**
 166          * Make a list of searchable namespaces and their canonical names.
 167          * @return array
 168          * @access public
 169          */
 170         function searchableNamespaces() {
 171                 global $wgContLang;
 172                 $arr = array();
 173                 foreach( $wgContLang->getNamespaces() as $ns => $name ) {
 174                         if( $ns >= NS_MAIN ) {
 175                                 $arr[$ns] = $name;
 176                         }
 177                 }
 178                 return $arr;
 179         }
 180
 181         /**
 182          * Return a 'cleaned up' search string
 183          *
 184          * @return string
 185          * @access public
 186          */
 187         function filter( $text ) {
 188                 $lc = $this->legalSearchChars();
 189                 return trim( preg_replace( "/[^{$lc}]/", " ", $text ) );
 190         }
 191         /**
 192          * Load up the appropriate search engine class for the currently
 193          * active database backend, and return a configured instance.
 194          *
 195          * @return SearchEngine
 196          * @private
 197          */
 198         function create() {
 199                 global $wgDBtype, $wgSearchType;
 200                 if( $wgSearchType ) {
 201                         $class = $wgSearchType;
 202                 } elseif( $wgDBtype == 'mysql' ) {
 203                         $class = 'SearchMySQL4';
 204                 } else if ( $wgDBtype == 'postgres' ) {
 205                         $class = 'SearchPostgres';
 206                 } else {
 207                         $class = 'SearchEngineDummy';
 208                 }
 209                 $search = new $class( wfGetDB( DB_SLAVE ) );
 210                 $search->setLimitOffset(0,0);
 211                 return $search;
 212         }
 213
 214         /**
 215          * Create or update the search index record for the given page.
 216          * Title and text should be pre-processed.
 217          *
 218          * @param int $id
 219          * @param string $title
 220          * @param string $text
 221          * @abstract
 222          */
 223         function update( $id, $title, $text ) {
 224                 // no-op
 225         }
 226
 227         /**
 228          * Update a search index record's title only.
 229          * Title should be pre-processed.
 230          *
 231          * @param int $id
 232          * @param string $title
 233          * @abstract
 234          */
 235     function updateTitle( $id, $title ) {
 236                 // no-op
 237     }
 238 }
 239
 240 /** @package MediaWiki */
 241 class SearchResultSet {
 242         /**
 243          * Fetch an array of regular expression fragments for matching
 244          * the search terms as parsed by this engine in a text extract.
 245          *
 246          * @return array
 247          * @access public
 248          * @abstract
 249          */
 250         function termMatches() {
 251                 return array();
 252         }
 253
 254         function numRows() {
 255                 return 0;
 256         }
 257
 258         /**
 259          * Return true if results are included in this result set.
 260          * @return bool
 261          * @abstract
 262          */
 263         function hasResults() {
 264                 return false;
 265         }
 266
 267         /**
 268          * Some search modes return a total hit count for the query
 269          * in the entire article database. This may include pages
 270          * in namespaces that would not be matched on the given
 271          * settings.
 272          *
 273          * Return null if no total hits number is supported.
 274          *
 275          * @return int
 276          * @access public
 277          */
 278         function getTotalHits() {
 279                 return null;
 280         }
 281
 282         /**
 283          * Some search modes return a suggested alternate term if there are
 284          * no exact hits. Returns true if there is one on this set.
 285          *
 286          * @return bool
 287          * @access public
 288          */
 289         function hasSuggestion() {
 290                 return false;
 291         }
 292
 293         /**
 294          * Some search modes return a suggested alternate term if there are
 295          * no exact hits. Check hasSuggestion() first.
 296          *
 297          * @return string
 298          * @access public
 299          */
 300         function getSuggestion() {
 301                 return '';
 302         }
 303
 304         /**
 305          * Fetches next search result, or false.
 306          * @return SearchResult
 307          * @access public
 308          * @abstract
 309          */
 310         function next() {
 311                 return false;
 312         }
 313 }
 314
 315 /** @package MediaWiki */
 316 class SearchResult {
 317         function SearchResult( $row ) {
 318                 $this->mTitle = Title::makeTitle( $row->page_namespace, $row->page_title );
 319         }
 320
 321         /**
 322          * @return Title
 323          * @access public
 324          */
 325         function getTitle() {
 326                 return $this->mTitle;
 327         }
 328
 329         /**
 330          * @return double or null if not supported
 331          */
 332         function getScore() {
 333                 return null;
 334         }
 335 }
 336
 337 /**
 338  * @package MediaWiki
 339  */
 340 class SearchEngineDummy {
 341         function search( $term ) {
 342                 return null;
 343         }
 344         function setLimitOffset($l, $o) {}
 345         function legalSearchChars() {}
 346         function update() {}
 347         function setnamespaces() {}
 348         function searchtitle() {}
 349         function searchtext() {}
 350 }
 351 ?>