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          * @param string $term
  44          * @return Title
  45          */
  46         public static function getNearMatch( $searchterm ) {
  47                 global $wgContLang;
  48
  49                 $allSearchTerms = array($searchterm);
  50
  51                 if($wgContLang->hasVariants()){
  52                         $allSearchTerms = array_merge($allSearchTerms,$wgContLang->convertLinkToAllVariants($searchterm));
  53                 }
  54
  55                 foreach($allSearchTerms as $term){
  56
  57                         # Exact match? No need to look further.
  58                         $title = Title::newFromText( $term );
  59                         if (is_null($title))
  60                                 return NULL;
  61
  62                         if ( $title->getNamespace() == NS_SPECIAL || $title->exists() ) {
  63                                 return $title;
  64                         }
  65
  66                         # Now try all lower case (i.e. first letter capitalized)
  67                         #
  68                         $title = Title::newFromText( $wgContLang->lc( $term ) );
  69                         if ( $title->exists() ) {
  70                                 return $title;
  71                         }
  72
  73                         # Now try capitalized string
  74                         #
  75                         $title = Title::newFromText( $wgContLang->ucwords( $term ) );
  76                         if ( $title->exists() ) {
  77                                 return $title;
  78                         }
  79
  80                         # Now try all upper case
  81                         #
  82                         $title = Title::newFromText( $wgContLang->uc( $term ) );
  83                         if ( $title->exists() ) {
  84                                 return $title;
  85                         }
  86
  87                         # Now try Word-Caps-Breaking-At-Word-Breaks, for hyphenated names etc
  88                         $title = Title::newFromText( $wgContLang->ucwordbreaks($term) );
  89                         if ( $title->exists() ) {
  90                                 return $title;
  91                         }
  92
  93                         global $wgCapitalLinks, $wgContLang;
  94                         if( !$wgCapitalLinks ) {
  95                                 // Catch differs-by-first-letter-case-only
  96                                 $title = Title::newFromText( $wgContLang->ucfirst( $term ) );
  97                                 if ( $title->exists() ) {
  98                                         return $title;
  99                                 }
 100                                 $title = Title::newFromText( $wgContLang->lcfirst( $term ) );
 101                                 if ( $title->exists() ) {
 102                                         return $title;
 103                                 }
 104                         }
 105
 106                         // Give hooks a chance at better match variants
 107                         $title = null;
 108                         if( !wfRunHooks( 'SearchGetNearMatch', array( $term, &$title ) ) ) {
 109                                 return $title;
 110                         }
 111                 }
 112
 113                 $title = Title::newFromText( $searchterm );
 114
 115                 # Entering an IP address goes to the contributions page
 116                 if ( ( $title->getNamespace() == NS_USER && User::isIP($title->getText() ) )
 117                         || User::isIP( trim( $searchterm ) ) ) {
 118                         return SpecialPage::getTitleFor( 'Contributions', $title->getDBkey() );
 119                 }
 120
 121
 122                 # Entering a user goes to the user page whether it's there or not
 123                 if ( $title->getNamespace() == NS_USER ) {
 124                         return $title;
 125                 }
 126
 127                 # Go to images that exist even if there's no local page.
 128                 # There may have been a funny upload, or it may be on a shared
 129                 # file repository such as Wikimedia Commons.
 130                 if( $title->getNamespace() == NS_IMAGE ) {
 131                         $image = wfFindFile( $title );
 132                         if( $image ) {
 133                                 return $title;
 134                         }
 135                 }
 136
 137                 # MediaWiki namespace? Page may be "implied" if not customized.
 138                 # Just return it, with caps forced as the message system likes it.
 139                 if( $title->getNamespace() == NS_MEDIAWIKI ) {
 140                         return Title::makeTitle( NS_MEDIAWIKI, $wgContLang->ucfirst( $title->getText() ) );
 141                 }
 142
 143                 # Quoted term? Try without the quotes...
 144                 $matches = array();
 145                 if( preg_match( '/^"([^"]+)"$/', $searchterm, $matches ) ) {
 146                         return SearchEngine::getNearMatch( $matches[1] );
 147                 }
 148
 149                 return NULL;
 150         }
 151
 152         public static function legalSearchChars() {
 153                 return "A-Za-z_'0-9\\x80-\\xFF\\-";
 154         }
 155
 156         /**
 157          * Set the maximum number of results to return
 158          * and how many to skip before returning the first.
 159          *
 160          * @param int $limit
 161          * @param int $offset
 162          * @access public
 163          */
 164         function setLimitOffset( $limit, $offset = 0 ) {
 165                 $this->limit = intval( $limit );
 166                 $this->offset = intval( $offset );
 167         }
 168
 169         /**
 170          * Set which namespaces the search should include.
 171          * Give an array of namespace index numbers.
 172          *
 173          * @param array $namespaces
 174          * @access public
 175          */
 176         function setNamespaces( $namespaces ) {
 177                 $this->namespaces = $namespaces;
 178         }
 179
 180         /**
 181          * Make a list of searchable namespaces and their canonical names.
 182          * @return array
 183          */
 184         public static function searchableNamespaces() {
 185                 global $wgContLang;
 186                 $arr = array();
 187                 foreach( $wgContLang->getNamespaces() as $ns => $name ) {
 188                         if( $ns >= NS_MAIN ) {
 189                                 $arr[$ns] = $name;
 190                         }
 191                 }
 192                 return $arr;
 193         }
 194
 195         /**
 196          * Return a 'cleaned up' search string
 197          *
 198          * @return string
 199          * @access public
 200          */
 201         function filter( $text ) {
 202                 $lc = $this->legalSearchChars();
 203                 return trim( preg_replace( "/[^{$lc}]/", " ", $text ) );
 204         }
 205         /**
 206          * Load up the appropriate search engine class for the currently
 207          * active database backend, and return a configured instance.
 208          *
 209          * @return SearchEngine
 210          */
 211         public static function create() {
 212                 global $wgDBtype, $wgSearchType;
 213                 if( $wgSearchType ) {
 214                         $class = $wgSearchType;
 215                 } elseif( $wgDBtype == 'mysql' ) {
 216                         $class = 'SearchMySQL4';
 217                 } else if ( $wgDBtype == 'postgres' ) {
 218                         $class = 'SearchPostgres';
 219                 } else if ( $wgDBtype == 'oracle' ) {
 220                         $class = 'SearchOracle';
 221                 } else {
 222                         $class = 'SearchEngineDummy';
 223                 }
 224                 $search = new $class( wfGetDB( DB_SLAVE ) );
 225                 $search->setLimitOffset(0,0);
 226                 return $search;
 227         }
 228
 229         /**
 230          * Create or update the search index record for the given page.
 231          * Title and text should be pre-processed.
 232          *
 233          * @param int $id
 234          * @param string $title
 235          * @param string $text
 236          * @abstract
 237          */
 238         function update( $id, $title, $text ) {
 239                 // no-op
 240         }
 241
 242         /**
 243          * Update a search index record's title only.
 244          * Title should be pre-processed.
 245          *
 246          * @param int $id
 247          * @param string $title
 248          * @abstract
 249          */
 250         function updateTitle( $id, $title ) {
 251                 // no-op
 252         }
 253 }
 254
 255
 256 /**
 257  * @addtogroup Search
 258  */
 259 class SearchResultSet {
 260         /**
 261          * Fetch an array of regular expression fragments for matching
 262          * the search terms as parsed by this engine in a text extract.
 263          *
 264          * @return array
 265          * @access public
 266          * @abstract
 267          */
 268         function termMatches() {
 269                 return array();
 270         }
 271
 272         function numRows() {
 273                 return 0;
 274         }
 275
 276         /**
 277          * Return true if results are included in this result set.
 278          * @return bool
 279          * @abstract
 280          */
 281         function hasResults() {
 282                 return false;
 283         }
 284
 285         /**
 286          * Some search modes return a total hit count for the query
 287          * in the entire article database. This may include pages
 288          * in namespaces that would not be matched on the given
 289          * settings.
 290          *
 291          * Return null if no total hits number is supported.
 292          *
 293          * @return int
 294          * @access public
 295          */
 296         function getTotalHits() {
 297                 return null;
 298         }
 299
 300         /**
 301          * Some search modes return a suggested alternate term if there are
 302          * no exact hits. Returns true if there is one on this set.
 303          *
 304          * @return bool
 305          * @access public
 306          */
 307         function hasSuggestion() {
 308                 return false;
 309         }
 310
 311         /**
 312          * Some search modes return a suggested alternate term if there are
 313          * no exact hits. Check hasSuggestion() first.
 314          *
 315          * @return string
 316          * @access public
 317          */
 318         function getSuggestion() {
 319                 return '';
 320         }
 321
 322         /**
 323          * Fetches next search result, or false.
 324          * @return SearchResult
 325          * @access public
 326          * @abstract
 327          */
 328         function next() {
 329                 return false;
 330         }
 331
 332         /**
 333          * Frees the result set, if applicable.
 334          * @ access public
 335          */
 336         function free() {
 337                 // ...
 338         }
 339 }
 340
 341
 342 /**
 343  * @addtogroup Search
 344  */
 345 class SearchResultTooMany {
 346         ## Some search engines may bail out if too many matches are found
 347 }
 348
 349
 350 /**
 351  * @addtogroup Search
 352  */
 353 class SearchResult {
 354
 355         function SearchResult( $row ) {
 356                 $this->mTitle = Title::makeTitle( $row->page_namespace, $row->page_title );
 357         }
 358
 359         /**
 360          * @return Title
 361          * @access public
 362          */
 363         function getTitle() {
 364                 return $this->mTitle;
 365         }
 366
 367         /**
 368          * @return double or null if not supported
 369          */
 370         function getScore() {
 371                 return null;
 372         }
 373 }
 374
 375 /**
 376  * @addtogroup Search
 377  */
 378 class SearchEngineDummy {
 379         function search( $term ) {
 380                 return null;
 381         }
 382         function setLimitOffset($l, $o) {}
 383         function legalSearchChars() {}
 384         function update() {}
 385         function setnamespaces() {}
 386         function searchtitle() {}
 387         function searchtext() {}
 388 }
 389