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          * @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                 # Now try Word-Caps-Breaking-At-Word-Breaks, for hyphenated names etc
  85                 $title = Title::newFromText( preg_replace_callback(
  86                         '/\b([\w\x80-\xff]+)\b/',
  87                         create_function( '$matches', '
  88                                 global $wgContLang;
  89                                 return $wgContLang->ucfirst($matches[1]);
  90                                 ' ),
  91                         $term ) );
  92                 if ( $title->exists() ) {
  93                         return $title;
  94                 }
  95
  96                 global $wgCapitalLinks, $wgContLang;
  97                 if( !$wgCapitalLinks ) {
  98                         // Catch differs-by-first-letter-case-only
  99                         $title = Title::newFromText( $wgContLang->ucfirst( $term ) );
 100                         if ( $title->exists() ) {
 101                                 return $title;
 102                         }
 103                         $title = Title::newFromText( $wgContLang->lcfirst( $term ) );
 104                         if ( $title->exists() ) {
 105                                 return $title;
 106                         }
 107                 }
 108
 109                 $title = Title::newFromText( $term );
 110
 111                 # Entering an IP address goes to the contributions page
 112                 if ( ( $title->getNamespace() == NS_USER && User::isIP($title->getText() ) )
 113                         || User::isIP( trim( $term ) ) ) {
 114                         return Title::makeTitle( NS_SPECIAL, "Contributions/" . $title->getDbkey() );
 115                 }
 116
 117
 118                 # Entering a user goes to the user page whether it's there or not
 119                 if ( $title->getNamespace() == NS_USER ) {
 120                         return $title;
 121                 }
 122
 123                 # Quoted term? Try without the quotes...
 124                 if( preg_match( '/^"([^"]+)"$/', $term, $matches ) ) {
 125                         return SearchEngine::getNearMatch( $matches[1] );
 126                 }
 127
 128                 return NULL;
 129         }
 130
 131         function legalSearchChars() {
 132                 return "A-Za-z_'0-9\\x80-\\xFF\\-";
 133         }
 134
 135         /**
 136          * Set the maximum number of results to return
 137          * and how many to skip before returning the first.
 138          *
 139          * @param int $limit
 140          * @param int $offset
 141          * @access public
 142          */
 143         function setLimitOffset( $limit, $offset = 0 ) {
 144                 $this->limit = intval( $limit );
 145                 $this->offset = intval( $offset );
 146         }
 147
 148         /**
 149          * Set which namespaces the search should include.
 150          * Give an array of namespace index numbers.
 151          *
 152          * @param array $namespaces
 153          * @access public
 154          */
 155         function setNamespaces( $namespaces ) {
 156                 $this->namespaces = $namespaces;
 157         }
 158
 159         /**
 160          * Make a list of searchable namespaces and their canonical names.
 161          * @return array
 162          * @access public
 163          */
 164         function searchableNamespaces() {
 165                 global $wgContLang;
 166                 $arr = array();
 167                 foreach( $wgContLang->getNamespaces() as $ns => $name ) {
 168                         if( $ns >= NS_MAIN ) {
 169                                 $arr[$ns] = $name;
 170                         }
 171                 }
 172                 return $arr;
 173         }
 174
 175         /**
 176          * Return a 'cleaned up' search string
 177          *
 178          * @return string
 179          * @access public
 180          */
 181         function filter( $text ) {
 182                 $lc = $this->legalSearchChars();
 183                 return trim( preg_replace( "/[^{$lc}]/", " ", $text ) );
 184         }
 185         /**
 186          * Load up the appropriate search engine class for the currently
 187          * active database backend, and return a configured instance.
 188          *
 189          * @return SearchEngine
 190          * @access private
 191          */
 192         function create() {
 193                 global $wgDBtype, $wgSearchType;
 194                 if( $wgSearchType ) {
 195                         $class = $wgSearchType;
 196                 } elseif( $wgDBtype == 'mysql' ) {
 197                         $class = 'SearchMySQL4';
 198                         require_once( 'SearchMySQL4.php' );
 199                 } else if ( $wgDBtype == 'PostgreSQL' ) {
 200                         $class = 'SearchTsearch2';
 201                         require_once( 'SearchTsearch2.php' );
 202                 } else {
 203                         $class = 'SearchEngineDummy';
 204                 }
 205                 $search = new $class( wfGetDB( DB_SLAVE ) );
 206                 $search->setLimitOffset(0,0);
 207                 return $search;
 208         }
 209
 210         /**
 211          * Create or update the search index record for the given page.
 212          * Title and text should be pre-processed.
 213          *
 214          * @param int $id
 215          * @param string $title
 216          * @param string $text
 217          * @abstract
 218          */
 219         function update( $id, $title, $text ) {
 220                 // no-op
 221         }
 222
 223         /**
 224          * Update a search index record's title only.
 225          * Title should be pre-processed.
 226          *
 227          * @param int $id
 228          * @param string $title
 229          * @abstract
 230          */
 231     function updateTitle( $id, $title ) {
 232                 // no-op
 233     }
 234 }
 235
 236 /** @package MediaWiki */
 237 class SearchResultSet {
 238         /**
 239          * Fetch an array of regular expression fragments for matching
 240          * the search terms as parsed by this engine in a text extract.
 241          *
 242          * @return array
 243          * @access public
 244          * @abstract
 245          */
 246         function termMatches() {
 247                 return array();
 248         }
 249
 250         function numRows() {
 251                 return 0;
 252         }
 253
 254         /**
 255          * Return true if results are included in this result set.
 256          * @return bool
 257          * @abstract
 258          */
 259         function hasResults() {
 260                 return false;
 261         }
 262
 263         /**
 264          * Some search modes return a total hit count for the query
 265          * in the entire article database. This may include pages
 266          * in namespaces that would not be matched on the given
 267          * settings.
 268          *
 269          * Return null if no total hits number is supported.
 270          *
 271          * @return int
 272          * @access public
 273          */
 274         function getTotalHits() {
 275                 return null;
 276         }
 277
 278         /**
 279          * Some search modes return a suggested alternate term if there are
 280          * no exact hits. Returns true if there is one on this set.
 281          *
 282          * @return bool
 283          * @access public
 284          */
 285         function hasSuggestion() {
 286                 return false;
 287         }
 288
 289         /**
 290          * Some search modes return a suggested alternate term if there are
 291          * no exact hits. Check hasSuggestion() first.
 292          *
 293          * @return string
 294          * @access public
 295          */
 296         function getSuggestion() {
 297                 return '';
 298         }
 299
 300         /**
 301          * Fetches next search result, or false.
 302          * @return SearchResult
 303          * @access public
 304          * @abstract
 305          */
 306         function next() {
 307                 return false;
 308         }
 309 }
 310
 311 /** @package MediaWiki */
 312 class SearchResult {
 313         function SearchResult( $row ) {
 314                 $this->mTitle = Title::makeTitle( $row->page_namespace, $row->page_title );
 315         }
 316
 317         /**
 318          * @return Title
 319          * @access public
 320          */
 321         function getTitle() {
 322                 return $this->mTitle;
 323         }
 324
 325         /**
 326          * @return double or null if not supported
 327          */
 328         function getScore() {
 329                 return null;
 330         }
 331 }
 332
 333 /**
 334  * @package MediaWiki
 335  */
 336 class SearchEngineDummy {
 337         function search( $term ) {
 338                 return null;
 339         }
 340         function setLimitOffset($l, $o) {}
 341         function legalSearchChars() {}
 342         function update() {}
 343         function setnamespaces() {}
 344         function searchtitle() {}
 345         function searchtext() {}
 346 }
 347 ?>