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( 0 );
  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          * @access private
  52          */
  53         function getNearMatch( $term ) {
  54                 # Exact match? No need to look further.
  55                 $title = Title::newFromText( $term );
  56                 if (is_null($title))
  57                         return NULL;
  58
  59                 if ( $title->getNamespace() == NS_SPECIAL || $title->exists() ) {
  60                         return $title;
  61                 }
  62
  63                 # Now try all lower case (i.e. first letter capitalized)
  64                 #
  65                 $title = Title::newFromText( strtolower( $term ) );
  66                 if ( $title->exists() ) {
  67                         return $title;
  68                 }
  69
  70                 # Now try capitalized string
  71                 #
  72                 $title = Title::newFromText( ucwords( strtolower( $term ) ) );
  73                 if ( $title->exists() ) {
  74                         return $title;
  75                 }
  76
  77                 # Now try all upper case
  78                 #
  79                 $title = Title::newFromText( strtoupper( $term ) );
  80                 if ( $title->exists() ) {
  81                         return $title;
  82                 }
  83
  84                 global $wgCapitalLinks, $wgContLang;
  85                 if( !$wgCapitalLinks ) {
  86                         // Catch differs-by-first-letter-case-only
  87                         $title = Title::newFromText( $wgContLang->ucfirst( $term ) );
  88                         if ( $title->exists() ) {
  89                                 return $title;
  90                         }
  91                         $title = Title::newFromText( $wgContLang->lcfirst( $term ) );
  92                         if ( $title->exists() ) {
  93                                 return $title;
  94                         }
  95                 }
  96
  97                 $title = Title::newFromText( $term );
  98
  99                 # Entering an IP address goes to the contributions page
 100                 if ( ( $title->getNamespace() == NS_USER && User::isIP($title->getText() ) )
 101                         || User::isIP( trim( $term ) ) ) {
 102                         return Title::makeTitle( NS_SPECIAL, "Contributions/" . $title->getDbkey() );
 103                 }
 104
 105
 106                 # Entering a user goes to the user page whether it's there or not
 107                 if ( $title->getNamespace() == NS_USER ) {
 108                         return $title;
 109                 }
 110
 111                 # Quoted term? Try without the quotes...
 112                 if( preg_match( '/^"([^"]+)"$/', $term, $matches ) ) {
 113                         return SearchEngine::getNearMatch( $matches[1] );
 114                 }
 115
 116                 return NULL;
 117         }
 118
 119         function legalSearchChars() {
 120                 return "A-Za-z_'0-9\\x80-\\xFF\\-";
 121         }
 122
 123         /**
 124          * Set the maximum number of results to return
 125          * and how many to skip before returning the first.
 126          *
 127          * @param int $limit
 128          * @param int $offset
 129          * @access public
 130          */
 131         function setLimitOffset( $limit, $offset = 0 ) {
 132                 $this->limit = IntVal( $limit );
 133                 $this->offset = IntVal( $offset );
 134         }
 135
 136         /**
 137          * Set which namespaces the search should include.
 138          * Give an array of namespace index numbers.
 139          *
 140          * @param array $namespaces
 141          * @access public
 142          */
 143         function setNamespaces( $namespaces ) {
 144                 $this->namespaces = $namespaces;
 145         }
 146
 147         /**
 148          * Make a list of searchable namespaces and their canonical names.
 149          * @return array
 150          * @access public
 151          */
 152         function searchableNamespaces() {
 153                 global $wgContLang;
 154                 $arr = array();
 155                 foreach( $wgContLang->getNamespaces() as $ns => $name ) {
 156                         if( $ns >= NS_MAIN ) {
 157                                 $arr[$ns] = $name;
 158                         }
 159                 }
 160                 return $arr;
 161         }
 162
 163         /**
 164          * Return a 'cleaned up' search string
 165          *
 166          * @return string
 167          * @access public
 168          */
 169         function filter( $text ) {
 170                 $lc = $this->legalSearchChars();
 171                 return trim( preg_replace( "/[^{$lc}]/", " ", $text ) );
 172         }
 173         /**
 174          * Load up the appropriate search engine class for the currently
 175          * active database backend, and return a configured instance.
 176          *
 177          * @return SearchEngine
 178          * @access private
 179          */
 180         function create() {
 181                 global $wgDBtype, $wgDBmysql4, $wgSearchType;
 182                 if( $wgSearchType ) {
 183                         $class = $wgSearchType;
 184                 } elseif( $wgDBtype == 'mysql' ) {
 185                         if( $wgDBmysql4 ) {
 186                                 $class = 'SearchMySQL4';
 187                                 require_once( 'SearchMySQL4.php' );
 188                         } else {
 189                                 $class = 'SearchMysql3';
 190                                 require_once( 'SearchMySQL3.php' );
 191                         }
 192                 } else if ( $wgDBtype == 'PostgreSQL' ) {
 193                         $class = 'SearchTsearch2';
 194                         require_once( 'SearchTsearch2.php' );
 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 }
 334