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