<?php
/**
- * Contain site class
- * See search.doc
- * @package MediaWiki
- */
-
-/**
- *
- */
-define( 'MW_SEARCH_OK', true );
-define( 'MW_SEARCH_BAD_QUERY', false );
-
-/**
- * @todo document
- * @package MediaWiki
+ * Contain a class for special pages
+ * @addtogroup Search
*/
class SearchEngine {
- /* private */ var $rawText, $filteredText, $searchTerms;
- /* private */ var $titleCond, $textCond;
-
- var $doSearchRedirects = true;
- var $addToQuery = array();
- var $namespacesToSearch = array();
- var $alternateTitle;
- var $allTitles = false;
-
- function SearchEngine( $text ) {
- $this->rawText = trim( $text );
-
- # We display the query, so let's strip it for safety
- #
- global $wgDBmysql4;
- $lc = SearchEngine::legalSearchChars() . '()';
- if( $wgDBmysql4 ) {
- $lc .= "\"~<>*+-";
- }
- $this->filteredText = trim( preg_replace( "/[^{$lc}]/", " ", $text ) );
- $this->searchTerms = array();
- $this->strictMatching = true; # Google-style, add '+' on all terms
-
- $this->db =& wfGetDB( DB_SLAVE );
- }
+ var $limit = 10;
+ var $offset = 0;
+ var $searchTerms = array();
+ var $namespaces = array( NS_MAIN );
+ var $showRedirects = false;
/**
- * Return a partial WHERE clause to limit the search to the given namespaces
+ * 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
+ * @return SearchResultSet
+ * @access public
+ * @abstract
*/
- function queryNamespaces() {
- $namespaces = implode( ',', $this->namespacesToSearch );
- if ($namespaces == '') {
- $namespaces = '0';
- }
- return "AND cur_namespace IN (" . $namespaces . ')';
+ function searchText( $term ) {
+ return null;
}
/**
- * Return a partial WHERE clause to include or exclude redirects from results
+ * 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
+ * @return SearchResultSet
+ * @access public
+ * @abstract
*/
- function searchRedirects() {
- if ( $this->doSearchRedirects ) {
- return '';
- } else {
- return 'AND cur_is_redirect=0 ';
- }
+ function searchTitle( $term ) {
+ return null;
}
-
- /**
- * @access private
- */ function initNamespaceCheckbox( $i ) {
- global $wgUser, $wgNamespacesToBeSearchedDefault;
-
- if ($wgUser->getID()) {
- // User is logged in so we retrieve his default namespaces
- return $wgUser->getOption( 'searchNs'.$i );
- } else {
- // User is not logged in so we give him the global default namespaces
- return !empty($wgNamespacesToBeSearchedDefault[ $i ]);
- }
- }
-
+
/**
- * Display the "power search" footer. Does not actually perform the search,
- * that is done by showResults()
+ * If an exact title match can be find, or a very slightly close match,
+ * return the title. If no match, returns NULL.
+ *
+ * @param string $term
+ * @return Title
*/
- function powersearch() {
- global $wgUser, $wgOut, $wgContLang, $wgTitle, $wgRequest;
- $sk =& $wgUser->getSkin();
-
- $search = $this->rawText;
- $searchx = $wgRequest->getVal( 'searchx' );
- $listredirs = $wgRequest->getVal( 'redirs' );
-
- $ret = wfMsg('powersearchtext'); # Text to be returned
- $tempText = ''; # Temporary text, for substitution into $ret
+ public static function getNearMatch( $searchterm ) {
+ global $wgContLang;
+
+ $allSearchTerms = array($searchterm);
- if( isset( $_REQUEST['searchx'] ) ) {
- $this->addToQuery['searchx'] = '1';
+ if($wgContLang->hasVariants()){
+ $allSearchTerms = array_merge($allSearchTerms,$wgContLang->convertLinkToAllVariants($searchterm));
}
-
- # Do namespace checkboxes
- $namespaces = $wgContLang->getNamespaces();
- foreach ( $namespaces as $i => $namespace ) {
- # Skip virtual namespaces
- if ( $i < 0 ) {
- continue;
+
+ foreach($allSearchTerms as $term){
+
+ # Exact match? No need to look further.
+ $title = Title::newFromText( $term );
+ if (is_null($title))
+ return NULL;
+
+ if ( $title->getNamespace() == NS_SPECIAL || $title->exists() ) {
+ return $title;
}
- $formVar = 'ns'.$i;
+ # Now try all lower case (i.e. first letter capitalized)
+ #
+ $title = Title::newFromText( $wgContLang->lc( $term ) );
+ if ( $title->exists() ) {
+ return $title;
+ }
- # Initialise checkboxValues, either from defaults or from
- # a previous invocation
- if ( !isset( $searchx ) ) {
- $checkboxValue = $this->initNamespaceCheckbox( $i );
- } else {
- $checkboxValue = $wgRequest->getVal( $formVar );
+ # Now try capitalized string
+ #
+ $title = Title::newFromText( $wgContLang->ucwords( $term ) );
+ if ( $title->exists() ) {
+ return $title;
}
- $checked = '';
- if ( $checkboxValue == 1 ) {
- $checked = ' checked="checked"';
- $this->addToQuery['ns'.$i] = 1;
- array_push( $this->namespacesToSearch, $i );
+ # Now try all upper case
+ #
+ $title = Title::newFromText( $wgContLang->uc( $term ) );
+ if ( $title->exists() ) {
+ return $title;
}
- $name = str_replace( '_', ' ', $namespaces[$i] );
- if ( '' == $name ) {
- $name = wfMsg( 'blanknamespace' );
+
+ # Now try Word-Caps-Breaking-At-Word-Breaks, for hyphenated names etc
+ $title = Title::newFromText( $wgContLang->ucwordbreaks($term) );
+ if ( $title->exists() ) {
+ return $title;
}
- if ( $tempText !== '' ) {
- $tempText .= ' ';
+ 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;
+ }
+ }
+
+ // Give hooks a chance at better match variants
+ $title = null;
+ if( !wfRunHooks( 'SearchGetNearMatch', array( $term, &$title ) ) ) {
+ return $title;
}
- $tempText .= "<input type='checkbox' value=\"1\" name=\"" .
- "ns{$i}\"{$checked} />{$name}\n";
}
- $ret = str_replace ( '$1', $tempText, $ret );
- # List redirects checkbox
+ $title = Title::newFromText( $searchterm );
- $checked = '';
- if ( $listredirs == 1 ) {
- $this->addToQuery['redirs'] = 1;
- $checked = ' checked="checked"';
+ # Entering an IP address goes to the contributions page
+ if ( ( $title->getNamespace() == NS_USER && User::isIP($title->getText() ) )
+ || User::isIP( trim( $searchterm ) ) ) {
+ return SpecialPage::getTitleFor( 'Contributions', $title->getDBkey() );
}
- $tempText = "<input type='checkbox' value='1' name=\"redirs\"{$checked} />\n";
- $ret = str_replace( '$2', $tempText, $ret );
- # Search field
- $tempText = "<input type='text' name=\"search\" value=\"" .
- htmlspecialchars( $search ) ."\" width=\"80\" />\n";
- $ret = str_replace( "$3", $tempText, $ret );
-
- # Searchx button
+ # Entering a user goes to the user page whether it's there or not
+ if ( $title->getNamespace() == NS_USER ) {
+ return $title;
+ }
- $tempText = '<input type="submit" name="searchx" value="' .
- wfMsg('powersearch') . "\" />\n";
- $ret = str_replace( '$9', $tempText, $ret );
+ # Go to images that exist even if there's no local page.
+ # There may have been a funny upload, or it may be on a shared
+ # file repository such as Wikimedia Commons.
+ if( $title->getNamespace() == NS_IMAGE ) {
+ $image = wfFindFile( $title );
+ if( $image ) {
+ return $title;
+ }
+ }
- $action = $sk->escapeSearchLink();
- $ret = "<br /><br />\n<form id=\"powersearch\" method=\"get\" " .
- "action=\"$action\">\n{$ret}\n</form>\n";
+ # MediaWiki namespace? Page may be "implied" if not customized.
+ # Just return it, with caps forced as the message system likes it.
+ if( $title->getNamespace() == NS_MEDIAWIKI ) {
+ return Title::makeTitle( NS_MEDIAWIKI, $wgContLang->ucfirst( $title->getText() ) );
+ }
- if ( isset ( $searchx ) ) {
- if ( ! $listredirs ) {
- $this->doSearchRedirects = false;
- }
+ # Quoted term? Try without the quotes...
+ $matches = array();
+ if( preg_match( '/^"([^"]+)"$/', $searchterm, $matches ) ) {
+ return SearchEngine::getNearMatch( $matches[1] );
}
- return $ret;
+
+ return NULL;
}
- function setupPage() {
- global $wgOut;
- $wgOut->setPageTitle( wfMsg( 'searchresults' ) );
- $wgOut->setSubtitle( wfMsg( 'searchquery', htmlspecialchars( $this->rawText ) ) );
- $wgOut->setArticleRelated( false );
- $wgOut->setRobotpolicy( 'noindex,nofollow' );
+ public static function legalSearchChars() {
+ return "A-Za-z_'0-9\\x80-\\xFF\\-";
}
/**
- * Perform the search and construct the results page
+ * Set the maximum number of results to return
+ * and how many to skip before returning the first.
+ *
+ * @param int $limit
+ * @param int $offset
+ * @access public
*/
- function showResults() {
- global $wgUser, $wgTitle, $wgOut, $wgLang;
- global $wgDisableTextSearch, $wgInputEncoding;
- $fname = 'SearchEngine::showResults';
+ function setLimitOffset( $limit, $offset = 0 ) {
+ $this->limit = intval( $limit );
+ $this->offset = intval( $offset );
+ }
- $search = $this->rawText;
+ /**
+ * Set which namespaces the search should include.
+ * Give an array of namespace index numbers.
+ *
+ * @param array $namespaces
+ * @access public
+ */
+ function setNamespaces( $namespaces ) {
+ $this->namespaces = $namespaces;
+ }
- $powersearch = $this->powersearch(); /* Need side-effects here? */
+ /**
+ * Parse some common prefixes: all (search everything)
+ * or namespace names
+ *
+ * @param string $query
+ */
+ function replacePrefixes( $query ){
+ global $wgContLang;
- $this->setupPage();
+ if( strpos($query,':') === false )
+ return $query; // nothing to do
+
+ $parsed = $query;
+ $allkeyword = wfMsg('searchall').":";
+ if( strncmp($query, $allkeyword, strlen($allkeyword)) == 0 ){
+ $this->namespaces = null;
+ $parsed = substr($query,strlen($allkeyword));
+ } else if( strpos($query,':') !== false ) {
+ $prefix = substr($query,0,strpos($query,':'));
+ $index = $wgContLang->getNsIndex($prefix);
+ if($index !== false){
+ $this->namespaces = array($index);
+ $parsed = substr($query,strlen($prefix)+1);
+ }
+ }
+ if(trim($parsed) == '')
+ return $query; // prefix was the whole query
- $sk = $wgUser->getSkin();
- $wgOut->addWikiText( wfMsg( 'searchresulttext' ) );
+ return $parsed;
+ }
- if ( !$this->parseQuery() ) {
- $wgOut->addWikiText(
- '==' . wfMsg( 'badquery' ) . "==\n" .
- wfMsg( 'badquerytext' ) );
- return;
+ /**
+ * Make a list of searchable namespaces and their canonical names.
+ * @return array
+ */
+ public static function searchableNamespaces() {
+ global $wgContLang;
+ $arr = array();
+ foreach( $wgContLang->getNamespaces() as $ns => $name ) {
+ if( $ns >= NS_MAIN ) {
+ $arr[$ns] = $name;
+ }
}
- list( $limit, $offset ) = wfCheckLimits( 20, 'searchlimit' );
-
- if ( $wgDisableTextSearch ) {
- $wgOut->addHTML( wfMsg( 'searchdisabled' ) );
- $wgOut->addHTML( wfMsg( 'googlesearch',
- htmlspecialchars( $this->rawText ),
- htmlspecialchars( $wgInputEncoding ) ) );
- return;
+ return $arr;
+ }
+
+ /**
+ * Extract default namespaces to search from the given user's
+ * settings, returning a list of index numbers.
+ *
+ * @param User $user
+ * @return array
+ * @static
+ */
+ public static function userNamespaces( &$user ) {
+ $arr = array();
+ foreach( SearchEngine::searchableNamespaces() as $ns => $name ) {
+ if( $user->getOption( 'searchNs' . $ns ) ) {
+ $arr[] = $ns;
+ }
}
-
- $titleMatches = $this->getMatches( $this->titleCond, $limit, $offset );
- $textMatches = $this->getMatches( $this->textCond, $limit, $offset );
-
- $sk = $wgUser->getSkin();
+ return $arr;
+ }
+
+ /**
+ * Find snippet highlight settings for a given user
+ *
+ * @param User $user
+ * @return array contextlines, contextchars
+ * @static
+ */
+ public static function userHighlightPrefs( &$user ){
+ //$contextlines = $user->getOption( 'contextlines', 5 );
+ $contextlines = 2; // Hardcode this. Old defaults sucked. :)
+ $contextchars = $user->getOption( 'contextchars', 50 );
+ return array($contextlines, $contextchars);
+ }
+
+ /**
+ * An array of namespaces indexes to be searched by default
+ *
+ * @return array
+ * @static
+ */
+ public static function defaultNamespaces(){
+ global $wgNamespacesToBeSearchedDefault;
- $num = count( $titleMatches ) + count( $textMatches );
- if ( $num >= $limit ) {
- $top = wfShowingResults( $offset, $limit );
+ return array_keys($wgNamespacesToBeSearchedDefault, true);
+ }
+
+ /**
+ * Return a 'cleaned up' search string
+ *
+ * @return string
+ * @access public
+ */
+ function filter( $text ) {
+ $lc = $this->legalSearchChars();
+ return trim( preg_replace( "/[^{$lc}]/", " ", $text ) );
+ }
+ /**
+ * Load up the appropriate search engine class for the currently
+ * active database backend, and return a configured instance.
+ *
+ * @return SearchEngine
+ */
+ public static function create() {
+ global $wgDBtype, $wgSearchType;
+ if( $wgSearchType ) {
+ $class = $wgSearchType;
+ } elseif( $wgDBtype == 'mysql' ) {
+ $class = 'SearchMySQL';
+ } else if ( $wgDBtype == 'postgres' ) {
+ $class = 'SearchPostgres';
+ } else if ( $wgDBtype == 'oracle' ) {
+ $class = 'SearchOracle';
} else {
- $top = wfShowingResultsNum( $offset, $limit, $num );
+ $class = 'SearchEngineDummy';
}
- $wgOut->addHTML( "<p>{$top}</p>\n" );
+ $search = new $class( wfGetDB( DB_SLAVE ) );
+ $search->setLimitOffset(0,0);
+ return $search;
+ }
+
+ /**
+ * 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 update( $id, $title, $text ) {
+ // no-op
+ }
- # For powersearch
- $a2l = '';
- $akk = array_keys( $this->addToQuery );
- foreach ( $akk AS $ak ) {
- $a2l .= "&{$ak}={$this->addToQuery[$ak]}" ;
+ /**
+ * 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
+ }
+
+ /**
+ * Get OpenSearch suggestion template
+ *
+ * @return string
+ * @static
+ */
+ public static function getOpenSearchTemplate() {
+ global $wgOpenSearchTemplate, $wgServer, $wgScriptPath;
+ if($wgOpenSearchTemplate)
+ return $wgOpenSearchTemplate;
+ else{
+ $ns = implode(',',SearchEngine::defaultNamespaces());
+ if(!$ns) $ns = "0";
+ return $wgServer . $wgScriptPath . '/api.php?action=opensearch&search={searchTerms}&namespace='.$ns;
}
+ }
+
+ /**
+ * Get internal MediaWiki Suggest template
+ *
+ * @return string
+ * @static
+ */
+ public static function getMWSuggestTemplate() {
+ global $wgMWSuggestTemplate, $wgServer, $wgScriptPath;
+ if($wgMWSuggestTemplate)
+ return $wgMWSuggestTemplate;
+ else
+ return $wgServer . $wgScriptPath . '/api.php?action=opensearch&search={searchTerms}&namespace={namespaces}';
+ }
+}
- $prevnext = wfViewPrevNext( $offset, $limit, 'Special:Search',
- 'search=' . wfUrlencode( $this->filteredText ) . $a2l );
- $wgOut->addHTML( "<br />{$prevnext}\n" );
- $foundsome = $this->showMatches( $titleMatches, $offset, 'notitlematches', 'titlematches' )
- || $this->showMatches( $textMatches, $offset, 'notextmatches', 'textmatches' );
-
- if ( !$foundsome ) {
- $wgOut->addWikiText( wfMsg( 'nonefound' ) );
- }
- $wgOut->addHTML( "<p>{$prevnext}</p>\n" );
- $wgOut->addHTML( $powersearch );
+/**
+ * @addtogroup Search
+ */
+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 legalSearchChars() {
- $lc = "A-Za-z_'0-9\\x80-\\xFF\\-";
- return $lc;
+ function numRows() {
+ return 0;
}
- function parseQuery() {
- global $wgDBmysql4;
- if (strlen($this->filteredText) < 1)
- return MW_SEARCH_BAD_QUERY;
+ /**
+ * Return true if results are included in this result set.
+ * @return bool
+ * @abstract
+ */
+ function hasResults() {
+ return false;
+ }
- if( $wgDBmysql4 ) {
- # Use cleaner boolean search if available
- return $this->parseQuery4();
- } else {
- # Fall back to ugly hack with multiple search clauses
- return $this->parseQuery3();
- }
+ /**
+ * 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 hasSuggestion() {
+ return false;
}
-
- function parseQuery3() {
- global $wgDBminWordLen, $wgContLang;
-
- # on non mysql4 database: get list of words we don't want to search for
- require_once( 'FulltextStoplist.php' );
-
- $lc = SearchEngine::legalSearchChars() . '()';
- $q = preg_replace( "/([()])/", " \\1 ", $this->filteredText );
- $q = preg_replace( "/\\s+/", " ", $q );
- $w = explode( ' ', trim( $q ) );
-
- $last = $cond = '';
- foreach ( $w as $word ) {
- $word = $wgContLang->stripForSearch( $word );
- if ( 'and' == $word || 'or' == $word || 'not' == $word
- || '(' == $word || ')' == $word ) {
- $cond .= ' ' . strtoupper( $word );
- $last = '';
- } else if ( strlen( $word ) < $wgDBminWordLen ) {
- continue;
- } else if ( FulltextStoplist::inList( $word ) ) {
- continue;
- } else {
- if ( '' != $last ) { $cond .= ' AND'; }
- $cond .= " (MATCH (##field##) AGAINST ('" .
- $this->db->strencode( $word ). "'))";
- $last = $word;
- array_push( $this->searchTerms, "\\b" . $word . "\\b" );
- }
- }
- if ( 0 == count( $this->searchTerms ) ) {
- return MW_SEARCH_BAD_QUERY;
- }
- $this->titleCond = '(' . str_replace( '##field##',
- 'si_title', $cond ) . ' )';
+ /**
+ * @return string suggested query, null if none
+ */
+ function getSuggestionQuery(){
+ return null;
+ }
- $this->textCond = '(' . str_replace( '##field##',
- 'si_text', $cond ) . ' AND (cur_is_redirect=0) )';
-
- return MW_SEARCH_OK;
+ /**
+ * @return string highlighted suggested query, '' if none
+ */
+ function getSuggestionSnippet(){
+ return '';
}
- function parseQuery4() {
- global $wgContLang;
- $lc = SearchEngine::legalSearchChars();
- $searchon = '';
- $this->searchTerms = array();
-
- # FIXME: This doesn't handle parenthetical expressions.
- if( preg_match_all( '/([-+<>~]?)(([' . $lc . ']+)(\*?)|"[^"]*")/',
- $this->filteredText, $m, PREG_SET_ORDER ) ) {
- foreach( $m as $terms ) {
- if( $searchon !== '' ) $searchon .= ' ';
- if( $this->strictMatching && ($terms[1] == '') ) {
- $terms[1] = '+';
- }
- $searchon .= $terms[1] . $wgContLang->stripForSearch( $terms[2] );
- if( !empty( $terms[3] ) ) {
- $regexp = preg_quote( $terms[3] );
- if( $terms[4] ) $regexp .= "[0-9A-Za-z_]+";
- } else {
- $regexp = preg_quote( str_replace( '"', '', $terms[2] ) );
- }
- $this->searchTerms[] = $regexp;
- }
- wfDebug( "Would search with '$searchon'\n" );
- wfDebug( "Match with /\b" . implode( '\b|\b', $this->searchTerms ) . "\b/\n" );
- } else {
- wfDebug( "Can't understand search query '{$this->filteredText}'\n" );
- }
-
- $searchon = $this->db->strencode( $searchon );
- $this->titleCond = " MATCH(si_title) AGAINST('$searchon' IN BOOLEAN MODE)";
- $this->textCond = " (MATCH(si_text) AGAINST('$searchon' IN BOOLEAN MODE) AND cur_is_redirect=0)";
- return MW_SEARCH_OK;
+ /**
+ * Return information about how and from where the results were fetched,
+ * should be useful for diagnostics and debugging
+ *
+ * @return string
+ */
+ function getInfo() {
+ return null;
}
+
+ /**
+ * Return a result set of hits on other (multiple) wikis associated with this one
+ *
+ * @return SearchResultSet
+ */
+ function getInterwikiResults() {
+ return null;
+ }
+
+ /**
+ * Check if there are results on other wikis
+ *
+ * @return boolean
+ */
+ function hasInterwikiResults() {
+ return $this->getInterwikiResults() != null;
+ }
+
- function &getMatches( $cond, $limit, $offset = 0 ) {
- $searchindex = $this->db->tableName( 'searchindex' );
- $cur = $this->db->tableName( 'cur' );
- $searchnamespaces = $this->queryNamespaces();
- $redircond = $this->searchRedirects();
-
- $sql = "SELECT cur_id,cur_namespace,cur_title," .
- "cur_text FROM $cur,$searchindex " .
- "WHERE cur_id=si_page AND {$cond} " .
- "{$searchnamespaces} {$redircond} " .
- $this->db->limitResult( $limit, $offset );
-
- $res = $this->db->query( $sql, 'SearchEngine::getMatches' );
- $matches = array();
- while ( $row = $this->db->fetchObject( $res ) ) {
- $matches[] = $row;
- }
- $this->db->freeResult( $res );
-
- return $matches;
+ /**
+ * Fetches next search result, or false.
+ * @return SearchResult
+ * @access public
+ * @abstract
+ */
+ function next() {
+ return false;
}
- function showMatches( &$matches, $offset, $msgEmpty, $msgFound ) {
- global $wgOut;
- if ( 0 == count( $matches ) ) {
- $wgOut->addHTML( "<h2>" . wfMsg( $msgEmpty ) .
- "</h2>\n" );
- return false;
- } else {
- $off = $offset + 1;
- $wgOut->addHTML( "<h2>" . wfMsg( $msgFound ) .
- "</h2>\n<ol start='{$off}'>" );
+ /**
+ * Frees the result set, if applicable.
+ * @ access public
+ */
+ function free() {
+ // ...
+ }
+}
- foreach( $matches as $row ) {
- $this->showHit( $row );
- }
- $wgOut->addHTML( "</ol>\n" );
+
+/**
+ * @addtogroup Search
+ */
+class SearchResultTooMany {
+ ## Some search engines may bail out if too many matches are found
+}
+
+
+/**
+ * @addtogroup Search
+ */
+class SearchResult {
+
+ function SearchResult( $row ) {
+ $this->mTitle = Title::makeTitle( $row->page_namespace, $row->page_title );
+ if( !is_null($this->mTitle) )
+ $this->mRevision = Revision::newFromTitle( $this->mTitle );
+ }
+
+ /**
+ * Check if this is result points to an invalid title
+ *
+ * @return boolean
+ * @access public
+ */
+ function isBrokenTitle(){
+ if( is_null($this->mTitle) )
return true;
- }
+ return false;
+ }
+
+ /**
+ * Check if target page is missing, happens when index is out of date
+ *
+ * @return boolean
+ * @access public
+ */
+ function isMissingRevision(){
+ if( !$this->mRevision )
+ return true;
+ return false;
+ }
+
+ /**
+ * @return Title
+ * @access public
+ */
+ function getTitle() {
+ return $this->mTitle;
}
- function showHit( $row ) {
- global $wgUser, $wgOut, $wgContLang;
+ /**
+ * @return double or null if not supported
+ */
+ function getScore() {
+ return null;
+ }
- $t = Title::makeName( $row->cur_namespace, $row->cur_title );
- if( is_null( $t ) ) {
- $wgOut->addHTML( "<!-- Broken link in search result -->\n" );
- return;
+ /**
+ * Lazy initialization of article text from DB
+ */
+ protected function initText(){
+ if( !isset($this->mText) ){
+ $this->mText = $this->mRevision->getText();
}
- $sk = $wgUser->getSkin();
-
- $contextlines = $wgUser->getOption( 'contextlines' );
- if ( '' == $contextlines ) { $contextlines = 5; }
- $contextchars = $wgUser->getOption( 'contextchars' );
- if ( '' == $contextchars ) { $contextchars = 50; }
+ }
- $link = $sk->makeKnownLink( $t, '' );
- $size = wfMsg( 'nbytes', strlen( $row->cur_text ) );
- $wgOut->addHTML( "<li>{$link} ({$size})" );
+ /**
+ * @param array $terms terms to highlight
+ * @return string highlighted text snippet, null (and not '') if not supported
+ */
+ function getTextSnippet($terms){
+ global $wgUser;
+ $this->initText();
+ list($contextlines,$contextchars) = SearchEngine::userHighlightPrefs($wgUser);
+ return $this->extractText( $this->mText, $terms, $contextlines, $contextchars);
+ }
+
+ /**
+ * Default implementation of snippet extraction
+ *
+ * @param string $text
+ * @param array $terms
+ * @param int $contextlines
+ * @param int $contextchars
+ * @return string
+ */
+ protected function extractText( $text, $terms, $contextlines, $contextchars ) {
+ global $wgLang, $wgContLang;
+ $fname = __METHOD__;
+
+ $lines = explode( "\n", $text );
+
+ $terms = implode( '|', $terms );
+ $max = intval( $contextchars ) + 1;
+ $pat1 = "/(.*)($terms)(.{0,$max})/i";
- $lines = explode( "\n", $row->cur_text );
- $pat1 = "/(.*)(" . implode( "|", $this->searchTerms ) . ")(.*)/i";
$lineno = 0;
+ $extract = "";
+ wfProfileIn( "$fname-extract" );
foreach ( $lines as $line ) {
if ( 0 == $contextlines ) {
break;
}
- --$contextlines;
++$lineno;
+ $m = array();
if ( ! preg_match( $pat1, $line, $m ) ) {
continue;
}
-
- $pre = $wgContLang->truncate( $m[1], -$contextchars, '...' );
+ --$contextlines;
+ $pre = $wgContLang->truncate( $m[1], -$contextchars, ' ... ' );
if ( count( $m ) < 3 ) {
$post = '';
} else {
- $post = $wgContLang->truncate( $m[3], $contextchars, '...' );
+ $post = $wgContLang->truncate( $m[3], $contextchars, ' ... ' );
}
$found = $m[2];
$line = htmlspecialchars( $pre . $found . $post );
- $pat2 = '/(' . implode( '|', $this->searchTerms ) . ")/i";
+ $pat2 = '/(' . $terms . ")/i";
$line = preg_replace( $pat2,
"<span class='searchmatch'>\\1</span>", $line );
- $wgOut->addHTML( "<br /><small>{$lineno}: {$line}</small>\n" );
- }
- $wgOut->addHTML( "</li>\n" );
- }
-
- function getNearMatch() {
- # Exact match? No need to look further.
- $title = Title::newFromText( $this->rawText );
- if ( $title->getNamespace() == NS_SPECIAL || 0 != $title->getArticleID() ) {
- return $title;
- }
-
- # Now try all lower case (i.e. first letter capitalized)
- #
- $title = Title::newFromText( strtolower( $this->rawText ) );
- if ( 0 != $title->getArticleID() ) {
- return $title;
- }
-
- # Now try capitalized string
- #
- $title = Title::newFromText( ucwords( strtolower( $this->rawText ) ) );
- if ( 0 != $title->getArticleID() ) {
- return $title;
- }
-
- # Now try all upper case
- #
- $title = Title::newFromText( strtoupper( $this->rawText ) );
- if ( 0 != $title->getArticleID() ) {
- return $title;
- }
-
- # Entering an IP address goes to the contributions page
- if ( preg_match( '/^\d{1,3}\.\d{1,3}\.\d{1,3}\.\d{1,3}$/', $this->rawText ) ) {
- $title = Title::makeTitle( NS_SPECIAL, "Contributions/" . $this->rawText );
- return $title;
+ $extract .= "${line}\n";
}
+ wfProfileOut( "$fname-extract" );
- return NULL;
+ return $extract;
+ }
+
+ /**
+ * @param array $terms terms to highlight
+ * @return string highlighted title, '' if not supported
+ */
+ function getTitleSnippet($terms){
+ return '';
}
- function goResult() {
- global $wgOut, $wgGoToEdit;
- global $wgDisableTextSearch;
- $fname = 'SearchEngine::goResult';
-
- # Try to go to page as entered.
- #
- $t = Title::newFromText( $this->rawText );
-
- # If the string cannot be used to create a title
- if( is_null( $t ) ){
- $this->showResults();
- return;
- }
-
- # If there's an exact or very near match, jump right there.
- $t = $this->getNearMatch();
- if( !is_null( $t ) ) {
- $wgOut->redirect( $t->getFullURL() );
- return;
- }
-
- # No match, generate an edit URL
- $t = Title::newFromText( $this->rawText );
-
- # If the feature is enabled, go straight to the edit page
- if ( $wgGoToEdit ) {
- $wgOut->redirect( $t->getFullURL( 'action=edit' ) );
- return;
- }
-
- if( $t ) {
- $editurl = $t->escapeLocalURL( 'action=edit' );
- } else {
- $editurl = ''; # ??
- }
- $wgOut->addHTML( '<p>' . wfMsg('nogomatch', $editurl, htmlspecialchars( $this->rawText ) ) . "</p>\n" );
-
- # Try a fuzzy title search
- $anyhit = false;
- global $wgDisableFuzzySearch;
- if(! $wgDisableFuzzySearch ){
- foreach( array(NS_MAIN, NS_PROJECT, NS_USER, NS_IMAGE, NS_MEDIAWIKI) as $namespace){
- $anyhit |= SearchEngine::doFuzzyTitleSearch( $this->rawText, $namespace );
- }
- }
-
- if( ! $anyhit ){
- return $this->showResults();
- }
+ /**
+ * @param array $terms terms to highlight
+ * @return string highlighted redirect name (redirect to this page), '' if none or not supported
+ */
+ function getRedirectSnippet($terms){
+ return '';
}
/**
- * @static
+ * @return Title object for the redirect to this page, null if none or not supported
*/
- function doFuzzyTitleSearch( $search, $namespace ){
- global $wgContLang, $wgOut;
-
- $this->setupPage();
-
- $sstr = ucfirst($search);
- $sstr = str_replace(' ', '_', $sstr);
- $fuzzymatches = SearchEngine::fuzzyTitles( $sstr, $namespace );
- $fuzzymatches = array_slice($fuzzymatches, 0, 10);
- $slen = strlen( $search );
- $wikitext = '';
- foreach($fuzzymatches as $res){
- $t = str_replace('_', ' ', $res[1]);
- $tfull = $wgContLang->getNsText( $namespace ) . ":$t|$t";
- if( $namespace == NS_MAIN )
- $tfull = $t;
- $distance = $res[0];
- $closeness = (strlen( $search ) - $distance) / strlen( $search );
- $percent = intval( $closeness * 100 ) . '%';
- $stars = str_repeat('*', ceil(5 * $closeness) );
- $wikitext .= "* [[$tfull]] $percent ($stars)\n";
- }
- if( $wikitext ){
- if( $namespace != NS_MAIN )
- $wikitext = '=== ' . $wgContLang->getNsText( $namespace ) . " ===\n" . $wikitext;
- $wgOut->addWikiText( $wikitext );
- return true;
- }
- return false;
+ function getRedirectTitle(){
+ return null;
}
/**
- * @static
+ * @return string highlighted relevant section name, null if none or not supported
*/
- function fuzzyTitles( $sstr, $namespace = NS_MAIN ){
- $span = 0.10; // weed on title length before doing levenshtein.
- $tolerance = 0.35; // allowed percentage of erronous characters
- $slen = strlen($sstr);
- $tolerance_count = ceil($tolerance * $slen);
- $spanabs = ceil($slen * (1 + $span)) - $slen;
- # print "Word: $sstr, len = $slen, range = [$min, $max], tolerance_count = $tolerance_count<BR>\n";
- $result = array();
- $cnt = 0;
- for( $i=0; $i <= $spanabs; $i++ ){
- $titles = SearchEngine::getTitlesByLength( $slen + $i, $namespace );
- if( $i != 0) {
- $titles = array_merge($titles, SearchEngine::getTitlesByLength( $slen - $i, $namespace ) );
- }
- foreach($titles as $t){
- $d = levenshtein($sstr, $t);
- if($d < $tolerance_count)
- $result[] = array($d, $t);
- $cnt++;
- }
- }
- usort($result, 'SearchEngine_pcmp');
- return $result;
+ function getSectionSnippet(){
+ return '';
}
/**
- * static
+ * @return Title object (pagename+fragment) for the section, null if none or not supported
*/
- function getTitlesByLength($aLength, $aNamespace = 0){
- global $wgMemc, $wgDBname;
- $fname = 'SearchEngin::getTitlesByLength';
-
- // to avoid multiple costly SELECTs in case of no memcached
- if( $this->allTitles ){
- if( isset( $this->allTitles[$aLength][$aNamespace] ) ){
- return $this->allTitles[$aLength][$aNamespace];
- } else {
- return array();
- }
- }
+ function getSectionTitle(){
+ return null;
+ }
- $mkey = "$wgDBname:titlesbylength:$aLength:$aNamespace";
- $mkeyts = "$wgDBname:titlesbylength:createtime";
- $ts = $wgMemc->get( $mkeyts );
- $result = $wgMemc->get( $mkey );
+ /**
+ * @return string timestamp
+ */
+ function getTimestamp(){
+ return $this->mRevision->getTimestamp();
+ }
- if( time() - $ts < 3600 ){
- // note: in case of insufficient memcached space, we return
- // an empty list instead of starting to hit the DB.
- return is_array( $result ) ? $result : array();
- }
+ /**
+ * @return int number of words
+ */
+ function getWordCount(){
+ $this->initText();
+ return str_word_count( $this->mText );
+ }
- $wgMemc->set( $mkeyts, time() );
-
- $res = $this->db->select( 'cur', array( 'cur_title', 'cur_namespace' ), false, $fname );
- $titles = array(); // length, ns, [titles]
- while( $obj = $this->db->fetchObject( $res ) ){
- $title = $obj->cur_title;
- $ns = $obj->cur_namespace;
- $len = strlen( $title );
- $titles[$len][$ns][] = $title;
- }
- foreach($titles as $length => $length_arr){
- foreach($length_arr as $ns => $title_arr){
- $mkey = "$wgDBname:titlesbylength:$length:$ns";
- $wgMemc->set( $mkey, $title_arr, 3600 * 24 );
- }
- }
- $this->allTitles = $titles;
- if( isset( $titles[$aLength][$aNamespace] ) )
- return $titles[$aLength][$aNamespace];
- else
- return array();
+ /**
+ * @return int size in bytes
+ */
+ function getByteSize(){
+ $this->initText();
+ return strlen( $this->mText );
+ }
+
+ /**
+ * @return boolean if hit has related articles
+ */
+ function hasRelated(){
+ return false;
+ }
+
+ /**
+ * @return interwiki prefix of the title (return iw even if title is broken)
+ */
+ function getInterwikiPrefix(){
+ return '';
}
}
/**
- * @access private
- * @static
+ * @addtogroup Search
*/
-function SearchEngine_pcmp($a, $b){ return $a[0] - $b[0]; }
-
-?>
+class SearchEngineDummy {
+ function search( $term ) {
+ return null;
+ }
+ function setLimitOffset($l, $o) {}
+ function legalSearchChars() {}
+ function update() {}
+ function setnamespaces() {}
+ function searchtitle() {}
+ function searchtext() {}
+}