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 Title::makeTitle( NS_SPECIAL, "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 $title;
 126                 }
 127
 128                 # Quoted term? Try without the quotes...
 129                 if( preg_match( '/^"([^"]+)"$/', $searchterm, $matches ) ) {
 130                         return SearchEngine::getNearMatch( $matches[1] );
 131                 }
 132
 133                 return NULL;
 134         }
 135
 136         function legalSearchChars() {
 137                 return "A-Za-z_'0-9\\x80-\\xFF\\-";
 138         }
 139
 140         /**
 141          * Set the maximum number of results to return
 142          * and how many to skip before returning the first.
 143          *
 144          * @param int $limit
 145          * @param int $offset
 146          * @access public
 147          */
 148         function setLimitOffset( $limit, $offset = 0 ) {
 149                 $this->limit = intval( $limit );
 150                 $this->offset = intval( $offset );
 151         }
 152
 153         /**
 154          * Set which namespaces the search should include.
 155          * Give an array of namespace index numbers.
 156          *
 157          * @param array $namespaces
 158          * @access public
 159          */
 160         function setNamespaces( $namespaces ) {
 161                 $this->namespaces = $namespaces;
 162         }
 163
 164         /**
 165          * Make a list of searchable namespaces and their canonical names.
 166          * @return array
 167          * @access public
 168          */
 169         function searchableNamespaces() {
 170                 global $wgContLang;
 171                 $arr = array();
 172                 foreach( $wgContLang->getNamespaces() as $ns => $name ) {
 173                         if( $ns >= NS_MAIN ) {
 174                                 $arr[$ns] = $name;
 175                         }
 176                 }
 177                 return $arr;
 178         }
 179
 180         /**
 181          * Return a 'cleaned up' search string
 182          *
 183          * @return string
 184          * @access public
 185          */
 186         function filter( $text ) {
 187                 $lc = $this->legalSearchChars();
 188                 return trim( preg_replace( "/[^{$lc}]/", " ", $text ) );
 189         }
 190         /**
 191          * Load up the appropriate search engine class for the currently
 192          * active database backend, and return a configured instance.
 193          *
 194          * @return SearchEngine
 195          * @private
 196          */
 197         function create() {
 198                 global $wgDBtype, $wgSearchType;
 199                 if( $wgSearchType ) {
 200                         $class = $wgSearchType;
 201                 } elseif( $wgDBtype == 'mysql' ) {
 202                         $class = 'SearchMySQL4';
 203                 } else if ( $wgDBtype == 'postgres' ) {
 204                         $class = 'SearchPostgres';
 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 /** @package MediaWiki */
 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 /** @package MediaWiki */
 315 class SearchResult {
 316         function SearchResult( $row ) {
 317                 $this->mTitle = Title::makeTitle( $row->page_namespace, $row->page_title );
 318         }
 319
 320         /**
 321          * @return Title
 322          * @access public
 323          */
 324         function getTitle() {
 325                 return $this->mTitle;
 326         }
 327
 328         /**
 329          * @return double or null if not supported
 330          */
 331         function getScore() {
 332                 return null;
 333         }
 334 }
 335
 336 /**
 337  * @package MediaWiki
 338  */
 339 class SearchEngineDummy {
 340         function search( $term ) {
 341                 return null;
 342         }
 343         function setLimitOffset($l, $o) {}
 344         function legalSearchChars() {}
 345         function update() {}
 346         function setnamespaces() {}
 347         function searchtitle() {}
 348         function searchtext() {}
 349 }
 350 ?>