Got rid of the MagicWord indexing constants (MAG_xxx), replaced them by string indexi...
[lhc/web/wiklou.git] / includes / SearchEngine.php
index f802aeb..c3b3851 100644 (file)
@@ -2,6 +2,7 @@
 /**
  * Contain a class for special pages
  * @package MediaWiki
+ * @subpackage Search
  */
 
 /**
@@ -11,89 +12,122 @@ class SearchEngine {
        var $limit = 10;
        var $offset = 0;
        var $searchTerms = array();
-       var $namespaces = array( 0 );
+       var $namespaces = array( NS_MAIN );
        var $showRedirects = false;
-       
+
        /**
         * Perform a full text search query and return a result set.
+        * If title searches are not supported or disabled, return null.
         *
         * @param string $term - Raw search term
-        * @param array $namespaces - List of namespaces to search
-        * @return ResultWrapper
+        * @return SearchResultSet
         * @access public
+        * @abstract
         */
        function searchText( $term ) {
-               return $this->db->resultObject( $this->db->query( $this->getQuery( $this->filter( $term ), true ) ) );
+               return null;
        }
 
        /**
         * Perform a title-only search query and return a result set.
+        * If title searches are not supported or disabled, return null.
         *
         * @param string $term - Raw search term
-        * @param array $namespaces - List of namespaces to search
-        * @return ResultWrapper
+        * @return SearchResultSet
         * @access public
+        * @abstract
         */
        function searchTitle( $term ) {
-               return $this->db->resultObject( $this->db->query( $this->getQuery( $this->filter( $term ), false ) ) );
+               return null;
        }
-       
+
        /**
         * If an exact title match can be find, or a very slightly close match,
         * return the title. If no match, returns NULL.
         *
+        * @static
         * @param string $term
         * @return Title
-        * @access private
+        * @private
         */
        function getNearMatch( $term ) {
-               # Eliminate Blanks at start
-               $term = ereg_replace('[[:blank:]]*', '', $term);
-
                # Exact match? No need to look further.
                $title = Title::newFromText( $term );
-               if ( $title->getNamespace() == NS_SPECIAL || 0 != $title->getArticleID() ) {
+               if (is_null($title))
+                       return NULL;
+
+               if ( $title->getNamespace() == NS_SPECIAL || $title->exists() ) {
                        return $title;
                }
 
                # Now try all lower case (i.e. first letter capitalized)
                #
                $title = Title::newFromText( strtolower( $term ) );
-               if ( 0 != $title->getArticleID() ) {
+               if ( $title->exists() ) {
                        return $title;
                }
 
                # Now try capitalized string
                #
                $title = Title::newFromText( ucwords( strtolower( $term ) ) );
-               if ( 0 != $title->getArticleID() ) {
+               if ( $title->exists() ) {
                        return $title;
                }
 
                # Now try all upper case
                #
                $title = Title::newFromText( strtoupper( $term ) );
-               if ( 0 != $title->getArticleID() ) {
+               if ( $title->exists() ) {
                        return $title;
                }
 
-               # Entering an IP address goes to the contributions page
-               if ( preg_match( '/^(user:)?\d{1,3}\.\d{1,3}\.\d{1,3}\.\d{1,3}$/', strtolower($term) ) ) {
-                       $title = Title::makeTitle( NS_SPECIAL, "Contributions/" . $term );
+               # Now try Word-Caps-Breaking-At-Word-Breaks, for hyphenated names etc
+               $title = Title::newFromText( preg_replace_callback(
+                       '/\b([\w\x80-\xff]+)\b/',
+                       create_function( '$matches', '
+                               global $wgContLang;
+                               return $wgContLang->ucfirst($matches[1]);
+                               ' ),
+                       $term ) );
+               if ( $title->exists() ) {
                        return $title;
                }
 
+               global $wgCapitalLinks, $wgContLang;
+               if( !$wgCapitalLinks ) {
+                       // Catch differs-by-first-letter-case-only
+                       $title = Title::newFromText( $wgContLang->ucfirst( $term ) );
+                       if ( $title->exists() ) {
+                               return $title;
+                       }
+                       $title = Title::newFromText( $wgContLang->lcfirst( $term ) );
+                       if ( $title->exists() ) {
+                               return $title;
+                       }
+               }
+
+               $title = Title::newFromText( $term );
+
+               # Entering an IP address goes to the contributions page
+               if ( ( $title->getNamespace() == NS_USER && User::isIP($title->getText() ) )
+                       || User::isIP( trim( $term ) ) ) {
+                       return Title::makeTitle( NS_SPECIAL, "Contributions/" . $title->getDbkey() );
+               }
+
+
                # Entering a user goes to the user page whether it's there or not
-               if ( preg_match( '/^user:/', strtolower($term) ) ) {
-                        if (User::idFromName($term)) {
-                         $title = Title::newFromURL( $term );
-                         return $title;
-                        }
+               if ( $title->getNamespace() == NS_USER ) {
+                       return $title;
                }
-               
+
+               # Quoted term? Try without the quotes...
+               if( preg_match( '/^"([^"]+)"$/', $term, $matches ) ) {
+                       return SearchEngine::getNearMatch( $matches[1] );
+               }
+
                return NULL;
        }
-       
+
        function legalSearchChars() {
                return "A-Za-z_'0-9\\x80-\\xFF\\-";
        }
@@ -107,10 +141,10 @@ class SearchEngine {
         * @access public
         */
        function setLimitOffset( $limit, $offset = 0 ) {
-               $this->limit = IntVal( $limit );
-               $this->offset = IntVal( $offset );
+               $this->limit = intval( $limit );
+               $this->offset = intval( $offset );
        }
-       
+
        /**
         * Set which namespaces the search should include.
         * Give an array of namespace index numbers.
@@ -121,7 +155,7 @@ class SearchEngine {
        function setNamespaces( $namespaces ) {
                $this->namespaces = $namespaces;
        }
-       
+
        /**
         * Make a list of searchable namespaces and their canonical names.
         * @return array
@@ -137,18 +171,7 @@ class SearchEngine {
                }
                return $arr;
        }
-       
-       /**
-        * Fetch an array of regular expression fragments for matching
-        * the search terms as parsed by this engine in a text extract.
-        *
-        * @return array
-        * @access public
-        */
-       function termMatches() {
-               return $this->searchTerms;
-       }
-       
+
        /**
         * Return a 'cleaned up' search string
         *
@@ -159,96 +182,150 @@ class SearchEngine {
                $lc = $this->legalSearchChars();
                return trim( preg_replace( "/[^{$lc}]/", " ", $text ) );
        }
-       
        /**
-        * Return a partial WHERE clause to exclude redirects, if so set
-        * @return string
-        * @access private
+        * Load up the appropriate search engine class for the currently
+        * active database backend, and return a configured instance.
+        *
+        * @return SearchEngine
+        * @private
         */
-       function queryRedirect() {
-               if( $this->showRedirects ) {
-                       return 'AND cur_is_redirect=0';
+       function create() {
+               global $wgDBtype, $wgSearchType;
+               if( $wgSearchType ) {
+                       $class = $wgSearchType;
+               } elseif( $wgDBtype == 'mysql' ) {
+                       $class = 'SearchMySQL4';
+               } else if ( $wgDBtype == 'postgres' ) {
+                       $class = 'SearchPostgres';
                } else {
-                       return '';
+                       $class = 'SearchEngineDummy';
                }
+               $search = new $class( wfGetDB( DB_SLAVE ) );
+               $search->setLimitOffset(0,0);
+               return $search;
        }
-       
+
        /**
-        * Return a partial WHERE clause to limit the search to the given namespaces
-        * @return string
-        * @access private
+        * Create or update the search index record for the given page.
+        * Title and text should be pre-processed.
+        *
+        * @param int $id
+        * @param string $title
+        * @param string $text
+        * @abstract
         */
-       function queryNamespaces() {
-               $namespaces = implode( ',', $this->namespaces );
-               if ($namespaces == '') {
-                       $namespaces = '0';
-               }
-               return 'AND page_namespace IN (' . $namespaces . ')';
+       function update( $id, $title, $text ) {
+               // no-op
        }
-       
+
        /**
-        * Return a LIMIT clause to limit results on the query.
-        * @return string
-        * @access private
+        * Update a search index record's title only.
+        * Title should be pre-processed.
+        *
+        * @param int $id
+        * @param string $title
+        * @abstract
+        */
+    function updateTitle( $id, $title ) {
+               // no-op
+    }
+}
+
+/** @package MediaWiki */
+class SearchResultSet {
+       /**
+        * Fetch an array of regular expression fragments for matching
+        * the search terms as parsed by this engine in a text extract.
+        *
+        * @return array
+        * @access public
+        * @abstract
+        */
+       function termMatches() {
+               return array();
+       }
+
+       function numRows() {
+               return 0;
+       }
+
+       /**
+        * Return true if results are included in this result set.
+        * @return bool
+        * @abstract
+        */
+       function hasResults() {
+               return false;
+       }
+
+       /**
+        * Some search modes return a total hit count for the query
+        * in the entire article database. This may include pages
+        * in namespaces that would not be matched on the given
+        * settings.
+        *
+        * Return null if no total hits number is supported.
+        *
+        * @return int
+        * @access public
+        */
+       function getTotalHits() {
+               return null;
+       }
+
+       /**
+        * Some search modes return a suggested alternate term if there are
+        * no exact hits. Returns true if there is one on this set.
+        *
+        * @return bool
+        * @access public
         */
-       function queryLimit() {
-               return $this->db->limitResult( $this->limit, $this->offset );
+       function hasSuggestion() {
+               return false;
        }
 
        /**
-        * Does not do anything for generic search engine
-        * subclasses may define this though
+        * Some search modes return a suggested alternate term if there are
+        * no exact hits. Check hasSuggestion() first.
+        *
         * @return string
-        * @access private
+        * @access public
         */
-       function queryRanking($filteredTerm,$fulltext) {
-               return "";
+       function getSuggestion() {
+               return '';
        }
-       
+
        /**
-        * Construct the full SQL query to do the search.
-        * The guts shoulds be constructed in queryMain()
-        * @param string $filteredTerm
-        * @param bool $fulltext
-        * @access private
+        * Fetches next search result, or false.
+        * @return SearchResult
+        * @access public
+        * @abstract
         */
-       function getQuery( $filteredTerm, $fulltext ) {
-               return $this->queryMain( $filteredTerm, $fulltext ) . ' ' .
-                       $this->queryRedirect() . ' ' .
-                       $this->queryNamespaces() . ' ' .
-                       $this->queryRanking($filteredTerm, $fulltext) . ' ' .
-                       $this->queryLimit();
+       function next() {
+               return false;
+       }
+}
+
+/** @package MediaWiki */
+class SearchResult {
+       function SearchResult( $row ) {
+               $this->mTitle = Title::makeTitle( $row->page_namespace, $row->page_title );
        }
 
        /**
-        * Load up the appropriate search engine class for the currently
-          * active database backend, and return a configured instance.
-          *
-          * @return SearchEngine
-          * @access private
-       */
-       function create() {
-                global $wgDBtype, $wgDBmysql4, $wgSearchType;
-                if( $wgDBtype == 'mysql' ) {
-                        if( $wgDBmysql4 ) {
-                                $class = 'SearchMySQL4';
-                                require_once( 'SearchMySQL4.php' );
-                        } else {
-                                $class = 'SearchMysql3';
-                                require_once( 'SearchMySQL3.php' );
-                        }
-                } else if ( $wgDBtype == 'PostgreSQL' ) {
-                        $class = 'SearchTsearch2';
-                        require_once( 'SearchTsearch2.php' );
-                } else {
-                        $class = 'SearchEngineDummy';
-                }
-                $search = new $class( wfGetDB( DB_SLAVE ) );
-                $search->setLimitOffset(0,0);
-                return $search;
+        * @return Title
+        * @access public
+        */
+       function getTitle() {
+               return $this->mTitle;
        }
 
-       
+       /**
+        * @return double or null if not supported
+        */
+       function getScore() {
+               return null;
+       }
 }
 
 /**
@@ -258,5 +335,11 @@ class SearchEngineDummy {
        function search( $term ) {
                return null;
        }
+       function setLimitOffset($l, $o) {}
+       function legalSearchChars() {}
+       function update() {}
+       function setnamespaces() {}
+       function searchtitle() {}
+       function searchtext() {}
 }
-
+?>