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                 $title = Title::newFromText( $term );
  85
  86                 # Entering an IP address goes to the contributions page
  87                 if ( ( $title->getNamespace() == NS_USER && User::isIP($title->getText() ) )
  88                         || User::isIP( trim( $term ) ) ) {
  89                         return Title::makeTitle( NS_SPECIAL, "Contributions/" . $title->getDbkey() );
  90                 }
  91
  92
  93                 # Entering a user goes to the user page whether it's there or not
  94                 if ( $title->getNamespace() == NS_USER ) {
  95                         return $title;
  96                 }
  97
  98                 # Quoted term? Try without the quotes...
  99                 if( preg_match( '/^"([^"]+)"$/', $term, $matches ) ) {
 100                         return SearchEngine::getNearMatch( $matches[1] );
 101                 }
 102
 103                 return NULL;
 104         }
 105
 106         function legalSearchChars() {
 107                 return "A-Za-z_'0-9\\x80-\\xFF\\-";
 108         }
 109
 110         /**
 111          * Set the maximum number of results to return
 112          * and how many to skip before returning the first.
 113          *
 114          * @param int $limit
 115          * @param int $offset
 116          * @access public
 117          */
 118         function setLimitOffset( $limit, $offset = 0 ) {
 119                 $this->limit = IntVal( $limit );
 120                 $this->offset = IntVal( $offset );
 121         }
 122
 123         /**
 124          * Set which namespaces the search should include.
 125          * Give an array of namespace index numbers.
 126          *
 127          * @param array $namespaces
 128          * @access public
 129          */
 130         function setNamespaces( $namespaces ) {
 131                 $this->namespaces = $namespaces;
 132         }
 133
 134         /**
 135          * Make a list of searchable namespaces and their canonical names.
 136          * @return array
 137          * @access public
 138          */
 139         function searchableNamespaces() {
 140                 global $wgContLang;
 141                 $arr = array();
 142                 foreach( $wgContLang->getNamespaces() as $ns => $name ) {
 143                         if( $ns >= NS_MAIN ) {
 144                                 $arr[$ns] = $name;
 145                         }
 146                 }
 147                 return $arr;
 148         }
 149
 150         /**
 151          * Return a 'cleaned up' search string
 152          *
 153          * @return string
 154          * @access public
 155          */
 156         function filter( $text ) {
 157                 $lc = $this->legalSearchChars();
 158                 return trim( preg_replace( "/[^{$lc}]/", " ", $text ) );
 159         }
 160         /**
 161          * Load up the appropriate search engine class for the currently
 162          * active database backend, and return a configured instance.
 163          *
 164          * @return SearchEngine
 165          * @access private
 166          */
 167         function create() {
 168                 global $wgDBtype, $wgDBmysql4, $wgSearchType;
 169                 if( $wgSearchType ) {
 170                         $class = $wgSearchType;
 171                 } elseif( $wgDBtype == 'mysql' ) {
 172                         if( $wgDBmysql4 ) {
 173                                 $class = 'SearchMySQL4';
 174                                 require_once( 'SearchMySQL4.php' );
 175                         } else {
 176                                 $class = 'SearchMysql3';
 177                                 require_once( 'SearchMySQL3.php' );
 178                         }
 179                 } else if ( $wgDBtype == 'PostgreSQL' ) {
 180                         $class = 'SearchTsearch2';
 181                         require_once( 'SearchTsearch2.php' );
 182                 } else {
 183                         $class = 'SearchEngineDummy';
 184                 }
 185                 $search = new $class( wfGetDB( DB_SLAVE ) );
 186                 $search->setLimitOffset(0,0);
 187                 return $search;
 188         }
 189
 190         /**
 191          * Create or update the search index record for the given page.
 192          * Title and text should be pre-processed.
 193          *
 194          * @param int $id
 195          * @param string $title
 196          * @param string $text
 197          * @abstract
 198          */
 199         function update( $id, $title, $text ) {
 200                 // no-op
 201         }
 202
 203         /**
 204          * Update a search index record's title only.
 205          * Title should be pre-processed.
 206          *
 207          * @param int $id
 208          * @param string $title
 209          * @abstract
 210          */
 211     function updateTitle( $id, $title ) {
 212                 // no-op
 213     }
 214 }
 215
 216 class SearchResultSet {
 217         /**
 218          * Fetch an array of regular expression fragments for matching
 219          * the search terms as parsed by this engine in a text extract.
 220          *
 221          * @return array
 222          * @access public
 223          * @abstract
 224          */
 225         function termMatches() {
 226                 return array();
 227         }
 228
 229         function numRows() {
 230                 return 0;
 231         }
 232
 233         /**
 234          * Return true if results are included in this result set.
 235          * @return bool
 236          * @abstract
 237          */
 238         function hasResults() {
 239                 return false;
 240         }
 241
 242         /**
 243          * Some search modes return a total hit count for the query
 244          * in the entire article database. This may include pages
 245          * in namespaces that would not be matched on the given
 246          * settings.
 247          *
 248          * Return null if no total hits number is supported.
 249          *
 250          * @return int
 251          * @access public
 252          */
 253         function getTotalHits() {
 254                 return null;
 255         }
 256
 257         /**
 258          * Some search modes return a suggested alternate term if there are
 259          * no exact hits. Returns true if there is one on this set.
 260          *
 261          * @return bool
 262          * @access public
 263          */
 264         function hasSuggestion() {
 265                 return false;
 266         }
 267
 268         /**
 269          * Some search modes return a suggested alternate term if there are
 270          * no exact hits. Check hasSuggestion() first.
 271          *
 272          * @return string
 273          * @access public
 274          */
 275         function getSuggestion() {
 276                 return '';
 277         }
 278
 279         /**
 280          * Fetches next search result, or false.
 281          * @return SearchResult
 282          * @access public
 283          * @abstract
 284          */
 285         function next() {
 286                 return false;
 287         }
 288 }
 289
 290 class SearchResult {
 291         function SearchResult( $row ) {
 292                 $this->mTitle = Title::makeTitle( $row->page_namespace, $row->page_title );
 293         }
 294
 295         /**
 296          * @return Title
 297          * @access public
 298          */
 299         function getTitle() {
 300                 return $this->mTitle;
 301         }
 302
 303         /**
 304          * @return double or null if not supported
 305          */
 306         function getScore() {
 307                 return null;
 308         }
 309 }
 310
 311 /**
 312  * @package MediaWiki
 313  */
 314 class SearchEngineDummy {
 315         function search( $term ) {
 316                 return null;
 317         }
 318 }
 319