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 ( $title->getNamespace() == NS_SPECIAL || $title->exists() ) {
  57                         return $title;
  58                 }
  59
  60                 # Now try all lower case (i.e. first letter capitalized)
  61                 #
  62                 $title = Title::newFromText( strtolower( $term ) );
  63                 if ( $title->exists() ) {
  64                         return $title;
  65                 }
  66
  67                 # Now try capitalized string
  68                 #
  69                 $title = Title::newFromText( ucwords( strtolower( $term ) ) );
  70                 if ( $title->exists() ) {
  71                         return $title;
  72                 }
  73
  74                 # Now try all upper case
  75                 #
  76                 $title = Title::newFromText( strtoupper( $term ) );
  77                 if ( $title->exists() ) {
  78                         return $title;
  79                 }
  80
  81                 $title = Title::newFromText( $term );
  82
  83                 # Entering an IP address goes to the contributions page
  84                 if ( ( $title->getNamespace() == NS_USER && User::isIP($title->getText() ) )
  85                         || User::isIP( trim( $term ) ) ) {
  86                         return Title::makeTitle( NS_SPECIAL, "Contributions/" . $title->getDbkey() );
  87                 }
  88
  89
  90                 # Entering a user goes to the user page whether it's there or not
  91                 if ( $title->getNamespace() == NS_USER ) {
  92                         return $title;
  93                 }
  94
  95                 # Quoted term? Try without the quotes...
  96                 if( preg_match( '/^"([^"]+)"$/', $term, $matches ) ) {
  97                         return SearchEngine::getNearMatch( $matches[1] );
  98                 }
  99
 100                 return NULL;
 101         }
 102
 103         function legalSearchChars() {
 104                 return "A-Za-z_'0-9\\x80-\\xFF\\-";
 105         }
 106
 107         /**
 108          * Set the maximum number of results to return
 109          * and how many to skip before returning the first.
 110          *
 111          * @param int $limit
 112          * @param int $offset
 113          * @access public
 114          */
 115         function setLimitOffset( $limit, $offset = 0 ) {
 116                 $this->limit = IntVal( $limit );
 117                 $this->offset = IntVal( $offset );
 118         }
 119
 120         /**
 121          * Set which namespaces the search should include.
 122          * Give an array of namespace index numbers.
 123          *
 124          * @param array $namespaces
 125          * @access public
 126          */
 127         function setNamespaces( $namespaces ) {
 128                 $this->namespaces = $namespaces;
 129         }
 130
 131         /**
 132          * Make a list of searchable namespaces and their canonical names.
 133          * @return array
 134          * @access public
 135          */
 136         function searchableNamespaces() {
 137                 global $wgContLang;
 138                 $arr = array();
 139                 foreach( $wgContLang->getNamespaces() as $ns => $name ) {
 140                         if( $ns >= NS_MAIN ) {
 141                                 $arr[$ns] = $name;
 142                         }
 143                 }
 144                 return $arr;
 145         }
 146
 147         /**
 148          * Return a 'cleaned up' search string
 149          *
 150          * @return string
 151          * @access public
 152          */
 153         function filter( $text ) {
 154                 $lc = $this->legalSearchChars();
 155                 return trim( preg_replace( "/[^{$lc}]/", " ", $text ) );
 156         }
 157         /**
 158          * Load up the appropriate search engine class for the currently
 159          * active database backend, and return a configured instance.
 160          *
 161          * @return SearchEngine
 162          * @access private
 163          */
 164         function create() {
 165                 global $wgDBtype, $wgDBmysql4, $wgSearchType;
 166                 if( $wgSearchType ) {
 167                         $class = $wgSearchType;
 168                 } elseif( $wgDBtype == 'mysql' ) {
 169                         if( $wgDBmysql4 ) {
 170                                 $class = 'SearchMySQL4';
 171                                 require_once( 'SearchMySQL4.php' );
 172                         } else {
 173                                 $class = 'SearchMysql3';
 174                                 require_once( 'SearchMySQL3.php' );
 175                         }
 176                 } else if ( $wgDBtype == 'PostgreSQL' ) {
 177                         $class = 'SearchTsearch2';
 178                         require_once( 'SearchTsearch2.php' );
 179                 } else {
 180                         $class = 'SearchEngineDummy';
 181                 }
 182                 $search = new $class( wfGetDB( DB_SLAVE ) );
 183                 $search->setLimitOffset(0,0);
 184                 return $search;
 185         }
 186
 187         /**
 188          * Create or update the search index record for the given page.
 189          * Title and text should be pre-processed.
 190          *
 191          * @param int $id
 192          * @param string $title
 193          * @param string $text
 194          * @abstract
 195          */
 196         function update( $id, $title, $text ) {
 197                 // no-op
 198         }
 199
 200         /**
 201          * Update a search index record's title only.
 202          * Title should be pre-processed.
 203          *
 204          * @param int $id
 205          * @param string $title
 206          * @abstract
 207          */
 208     function updateTitle( $id, $title ) {
 209                 // no-op
 210     }
 211 }
 212
 213 class SearchResultSet {
 214         /**
 215          * Fetch an array of regular expression fragments for matching
 216          * the search terms as parsed by this engine in a text extract.
 217          *
 218          * @return array
 219          * @access public
 220          * @abstract
 221          */
 222         function termMatches() {
 223                 return array();
 224         }
 225
 226         function numRows() {
 227                 return 0;
 228         }
 229
 230         /**
 231          * Return true if results are included in this result set.
 232          * @return bool
 233          * @abstract
 234          */
 235         function hasResults() {
 236                 return false;
 237         }
 238
 239         /**
 240          * Some search modes return a total hit count for the query
 241          * in the entire article database. This may include pages
 242          * in namespaces that would not be matched on the given
 243          * settings.
 244          *
 245          * Return null if no total hits number is supported.
 246          *
 247          * @return int
 248          * @access public
 249          */
 250         function getTotalHits() {
 251                 return null;
 252         }
 253
 254         /**
 255          * Some search modes return a suggested alternate term if there are
 256          * no exact hits. Returns true if there is one on this set.
 257          *
 258          * @return bool
 259          * @access public
 260          */
 261         function hasSuggestion() {
 262                 return false;
 263         }
 264
 265         /**
 266          * Some search modes return a suggested alternate term if there are
 267          * no exact hits. Check hasSuggestion() first.
 268          *
 269          * @return string
 270          * @access public
 271          */
 272         function getSuggestion() {
 273                 return '';
 274         }
 275
 276         /**
 277          * Fetches next search result, or false.
 278          * @return SearchResult
 279          * @access public
 280          * @abstract
 281          */
 282         function next() {
 283                 return false;
 284         }
 285 }
 286
 287 class SearchResult {
 288         function SearchResult( $row ) {
 289                 $this->mTitle = Title::makeTitle( $row->page_namespace, $row->page_title );
 290         }
 291
 292         /**
 293          * @return Title
 294          * @access public
 295          */
 296         function getTitle() {
 297                 return $this->mTitle;
 298         }
 299
 300         /**
 301          * @return double or null if not supported
 302          */
 303         function getScore() {
 304                 return null;
 305         }
 306 }
 307
 308 /**
 309  * @package MediaWiki
 310  */
 311 class SearchEngineDummy {
 312         function search( $term ) {
 313                 return null;
 314         }
 315 }
 316