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