-<?
-# See search.doc
-
+<?php
+/**
+ * Contain a class for special pages
+ * @package MediaWiki
+ * @subpackage Search
+ */
+
+/**
+ * @package MediaWiki
+ */
class SearchEngine {
- /* private */ var $mUsertext, $mSearchterms;
- /* private */ var $mTitlecond, $mTextcond;
-
- var $doSearchRedirects = true;
- var $addtoquery = array();
- var $namespacesToSearch = array();
- var $alternateTitle;
+ var $limit = 10;
+ var $offset = 0;
+ var $searchTerms = array();
+ 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
+ * @return SearchResultSet
+ * @access public
+ * @abstract
+ */
+ function searchText( $term ) {
+ return null;
+ }
- function SearchEngine( $text )
- {
- # We display the query, so let's strip it for safety
- #
- global $wgDBmysql4;
- $lc = SearchEngine::legalSearchChars() . "()";
- if( $wgDBmysql4 ) $lc .= "\"~<>*+-";
- $this->mUsertext = trim( preg_replace( "/[^{$lc}]/", " ", $text ) );
- $this->mSearchterms = array();
+ /**
+ * 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 searchTitle( $term ) {
+ return null;
}
- function queryNamespaces()
- {
- $namespaces = implode( ",", $this->namespacesToSearch );
- if ($namespaces == "") {
- $namespaces = "0";
+ /**
+ * 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
+ */
+ function getNearMatch( $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;
}
- return "AND cur_namespace IN (" . $namespaces . ")";
- #return "1";
- }
- function searchRedirects()
- {
- if ( $this->doSearchRedirects ) return "";
- return "AND cur_is_redirect=0 ";
- }
+ # Now try all lower case (i.e. first letter capitalized)
+ #
+ $title = Title::newFromText( strtolower( $term ) );
+ if ( $title->exists() ) {
+ return $title;
+ }
- /* private */ function initNamespaceCheckbox( $i )
- {
- global $wgUser, $wgNamespacesToBeSearchedDefault;
-
+ # Now try capitalized string
+ #
+ $title = Title::newFromText( ucwords( strtolower( $term ) ) );
+ if ( $title->exists() ) {
+ return $title;
+ }
- if ($wgUser->getID()) {
- // User is logged in so we retrieve his default namespaces
- return $wgUser->getOption( "searchNs".$i );
+ # Now try all upper case
+ #
+ $title = Title::newFromText( strtoupper( $term ) );
+ if ( $title->exists() ) {
+ return $title;
}
- else {
- // User is not logged in so we give him the global default namespaces
- return $wgNamespacesToBeSearchedDefault[ $i ];
+
+ # 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;
}
- }
- # Display the "power search" footer. Does not actually perform the search,
- # that is done by showResults()
- function powersearch()
- {
- global $wgUser, $wgOut, $wgLang, $wgTitle;
-
- $search = $_REQUEST['search'];
- $searchx = $_REQUEST['searchx'];
- $listredirs = $_REQUEST['redirs'];
-
- $ret = wfMsg("powersearchtext"); # Text to be returned
- $tempText = ""; # Temporary text, for substitution into $ret
-
- # Do namespace checkboxes
- $namespaces = $wgLang->getNamespaces();
- foreach ( $namespaces as $i => $namespace ) {
- # Skip virtual namespaces
- if ( $i < 0 ) {
- continue;
+ global $wgCapitalLinks, $wgContLang;
+ if( !$wgCapitalLinks ) {
+ // Catch differs-by-first-letter-case-only
+ $title = Title::newFromText( $wgContLang->ucfirst( $term ) );
+ if ( $title->exists() ) {
+ return $title;
}
-
- $formVar = "ns$i";
-
- # Initialise checkboxValues, either from defaults or from
- # a previous invocation
- if ( !isset( $searchx ) ) {
- $checkboxValue = $this->initNamespaceCheckbox( $i );
- } else {
- $checkboxValue = $_REQUEST[$formVar];
+ $title = Title::newFromText( $wgContLang->lcfirst( $term ) );
+ if ( $title->exists() ) {
+ return $title;
}
+ }
- $checked = "";
- if ( $checkboxValue == 1 ) {
- $checked = " checked";
- $this->addtoquery["ns{$i}"] = 1;
- array_push( $this->namespacesToSearch, $i );
- }
- $name = str_replace( "_", " ", $namespaces[$i] );
- if ( "" == $name ) {
- $name = wfMsg( "blanknamespace" );
- }
+ $title = Title::newFromText( $term );
- if ( $tempText !== "" ) {
- $tempText .= " ";
- }
- $tempText .= "<input type=checkbox value=\"1\" name=\"" .
- "ns{$i}\"{$checked}>{$name}\n";
+ # 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() );
}
- $ret = str_replace ( "$1", $tempText, $ret );
- # List redirects checkbox
- $checked = "";
- if ( $listredirs == 1 ) {
- $this->addtoquery["redirs"] = 1;
- $checked = " checked";
+ # Entering a user goes to the user page whether it's there or not
+ if ( $title->getNamespace() == NS_USER ) {
+ return $title;
}
- $tempText = "<input type=checkbox value=1 name=\"redirs\"{$checked}>\n";
- $ret = str_replace( "$2", $tempText, $ret );
- # Search field
+ # Quoted term? Try without the quotes...
+ if( preg_match( '/^"([^"]+)"$/', $term, $matches ) ) {
+ return SearchEngine::getNearMatch( $matches[1] );
+ }
- $tempText = "<input type=text name=\"search\" value=\"" .
- htmlspecialchars( $search ) ."\" width=80>\n";
- $ret = str_replace( "$3", $tempText, $ret );
+ return NULL;
+ }
- # Searchx button
+ function legalSearchChars() {
+ return "A-Za-z_'0-9\\x80-\\xFF\\-";
+ }
- $tempText = "<input type=submit name=\"searchx\" value=\"" .
- wfMsg("powersearch") . "\">\n";
- $ret = str_replace( "$9", $tempText, $ret );
+ /**
+ * 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 setLimitOffset( $limit, $offset = 0 ) {
+ $this->limit = intval( $limit );
+ $this->offset = intval( $offset );
+ }
- $ret = "<br><br>\n<form id=\"powersearch\" method=\"get\" " .
- "action=\"" . wfLocalUrl( "" ) . "\">\n{$ret}\n</form>\n";
+ /**
+ * 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;
+ }
- if ( isset ( $searchx ) ) {
- if ( ! $listredirs ) {
- $this->doSearchRedirects = false;
+ /**
+ * Make a list of searchable namespaces and their canonical names.
+ * @return array
+ * @access public
+ */
+ function searchableNamespaces() {
+ global $wgContLang;
+ $arr = array();
+ foreach( $wgContLang->getNamespaces() as $ns => $name ) {
+ if( $ns >= NS_MAIN ) {
+ $arr[$ns] = $name;
}
}
- return $ret;
+ return $arr;
}
- # Perform the search and construct the results page
- function showResults()
- {
- global $wgUser, $wgTitle, $wgOut, $wgLang, $wgDisableTextSearch;
- global $wgInputEncoding;
- $fname = "SearchEngine::showResults";
-
- $search = $_REQUEST['search'];
-
- $powersearch = $this->powersearch(); /* Need side-effects here? */
-
- $wgOut->setPageTitle( wfMsg( "searchresults" ) );
- $q = wfMsg( "searchquery", htmlspecialchars( $this->mUsertext ) );
- $wgOut->setSubtitle( $q );
- $wgOut->setArticleFlag( false );
- $wgOut->setRobotpolicy( "noindex,nofollow" );
-
- $sk = $wgUser->getSkin();
- $text = wfMsg( "searchresulttext", $sk->makeKnownLink(
- wfMsg( "searchhelppage" ), wfMsg( "searchingwikipedia" ) ) );
- $wgOut->addHTML( $text );
-
- $this->parseQuery();
- if ( "" == $this->mTitlecond || "" == $this->mTextcond ) {
- $wgOut->addHTML( "<h2>" . wfMsg( "badquery" ) . "</h2>\n" .
- "<p>" . wfMsg( "badquerytext" ) );
- return;
- }
- list( $limit, $offset ) = wfCheckLimits( 20, "searchlimit" );
-
- $searchnamespaces = $this->queryNamespaces();
- $redircond = $this->searchRedirects();
-
- $sql = "SELECT cur_id,cur_namespace,cur_title," .
- "cur_text FROM cur,searchindex " .
- "WHERE cur_id=si_page AND {$this->mTitlecond} " .
- "{$searchnamespaces} {$redircond}" .
- "LIMIT {$offset}, {$limit}";
- $res1 = wfQuery( $sql, DB_READ, $fname );
- $num = wfNumRows($res1);
-
- if ( $wgDisableTextSearch ) {
- $wgOut->addHTML( wfMsg( "searchdisabled", $search, $wgInputEncoding ) );
+ /**
+ * 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
+ * @access private
+ */
+ function create() {
+ global $wgDBtype, $wgSearchType;
+ if( $wgSearchType ) {
+ $class = $wgSearchType;
+ } elseif( $wgDBtype == 'mysql' ) {
+ $class = 'SearchMySQL4';
+ require_once( 'SearchMySQL4.php' );
+ } else if ( $wgDBtype == 'PostgreSQL' ) {
+ $class = 'SearchTsearch2';
+ require_once( 'SearchTsearch2.php' );
} else {
- $sk = $wgUser->getSkin();
- $text = wfMsg( "searchresulttext", $sk->makeKnownLink(
- wfMsg( "searchhelppage" ), wfMsg( "searchingwikipedia" ) ) );
- $wgOut->addHTML( $text );
-
- $this->parseQuery();
- if ( "" == $this->mTitlecond || "" == $this->mTextcond ) {
- $wgOut->addHTML( "<h2>" . wfMsg( "badquery" ) . "</h2>\n" .
- "<p>" . wfMsg( "badquerytext" ) );
- return;
- }
- list( $limit, $offset ) = wfCheckLimits( 20, "searchlimit" );
-
- $searchnamespaces = $this->queryNamespaces();
- $redircond = $this->searchRedirects();
-
- $sql = "SELECT cur_id,cur_namespace,cur_title," .
- "cur_text FROM cur,searchindex " .
- "WHERE cur_id=si_page AND {$this->mTitlecond} " .
- "{$searchnamespaces} {$redircond}" .
- "LIMIT {$offset}, {$limit}";
- $res1 = wfQuery( $sql, DB_READ, $fname );
- $num = wfNumRows($res1);
-
- $sql = "SELECT cur_id,cur_namespace,cur_title," .
- "cur_text FROM cur,searchindex " .
- "WHERE cur_id=si_page AND {$this->mTextcond} " .
- "{$searchnamespaces} {$redircond} " .
- "LIMIT {$offset}, {$limit}";
- $res2 = wfQuery( $sql, DB_READ, $fname );
- $num = $num + wfNumRows($res2);
-
- if ( $num == $limit ) {
- $top = wfShowingResults( $offset, $limit);
- } else {
- $top = wfShowingResultsNum( $offset, $limit, $num );
- }
- $wgOut->addHTML( "<p>{$top}\n" );
-
- # For powersearch
-
- $a2l = "" ;
- $akk = array_keys( $this->addtoquery ) ;
- foreach ( $akk AS $ak ) {
- $a2l .= "&{$ak}={$this->addtoquery[$ak]}" ;
- }
-
- $sl = wfViewPrevNext( $offset, $limit, "",
- "search=" . wfUrlencode( $this->mUsertext ) . $a2l );
- $wgOut->addHTML( "<br>{$sl}\n" );
-
- $foundsome = false;
-
- if ( 0 == wfNumRows( $res1 ) ) {
- $wgOut->addHTML( "<h2>" . wfMsg( "notitlematches" ) .
- "</h2>\n" );
- } else {
- $foundsome = true;
- $off = $offset + 1;
- $wgOut->addHTML( "<h2>" . wfMsg( "titlematches" ) .
- "</h2>\n<ol start='{$off}'>" );
-
- while ( $row = wfFetchObject( $res1 ) ) {
- $this->showHit( $row );
- }
- wfFreeResult( $res1 );
- $wgOut->addHTML( "</ol>\n" );
- }
- }
-
- if ( $wgDisableTextSearch ) {
- $wgOut->addHTML( wfMsg( "searchdisabled", $search, $wgInputEncoding ) );
- } else {
- if ( 0 == wfNumRows( $res2 ) ) {
- $wgOut->addHTML( "<h2>" . wfMsg( "notextmatches" ) .
- "</h2>\n" );
- } else {
- $foundsome = true;
- $off = $offset + 1;
- $wgOut->addHTML( "<h2>" . wfMsg( "textmatches" ) . "</h2>\n" .
- "<ol start='{$off}'>" );
- while ( $row = wfFetchObject( $res2 ) ) {
- $this->showHit( $row );
- }
- wfFreeResult( $res2 );
- $wgOut->addHTML( "</ol>\n" );
- }
- if ( ! $foundsome ) {
- $wgOut->addHTML( "<p>" . wfMsg( "nonefound" ) . "\n" );
- }
- $wgOut->addHTML( "<p>{$sl}\n" );
- $wgOut->addHTML( $powersearch );
+ $class = 'SearchEngineDummy';
}
+ $search = new $class( wfGetDB( DB_SLAVE ) );
+ $search->setLimitOffset(0,0);
+ return $search;
}
- function legalSearchChars()
- {
- $lc = "A-Za-z_'0-9\\x80-\\xFF\\-";
- return $lc;
+ /**
+ * 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
}
- function parseQuery()
- {
- global $wgDBminWordLen, $wgLang, $wgDBmysql4;
-
- if( $wgDBmysql4 ) {
- # Use cleaner boolean search if available
- return $this->parseQuery4();
- }
-
- $lc = SearchEngine::legalSearchChars() . "()";
- $q = preg_replace( "/([()])/", " \\1 ", $this->mUsertext );
- $q = preg_replace( "/\\s+/", " ", $q );
- $w = explode( " ", strtolower( trim( $q ) ) );
-
- $last = $cond = "";
- foreach ( $w as $word ) {
- $word = $wgLang->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 ('" .
- wfStrencode( $word ). "'))";
- $last = $word;
- array_push( $this->mSearchterms, "\\b" . $word . "\\b" );
- }
- }
- if ( 0 == count( $this->mSearchterms ) ) { return; }
-
- $this->mTitlecond = "(" . str_replace( "##field##",
- "si_title", $cond ) . " )";
+ /**
+ * 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
+ }
+}
- $this->mTextcond = "(" . str_replace( "##field##",
- "si_text", $cond ) . " AND (cur_is_redirect=0) )";
- }
-
- function parseQuery4()
- {
- # FIXME: not ready yet! Do not use.
-
- global $wgLang;
- $lc = SearchEngine::legalSearchChars();
- #$q = preg_replace( "/([+-]?)([$lc]+)/e",
- # "\"$1\" . \$wgLang->stripForSearch(\"$2\")",
- # $this->mUsertext );
-
- $q = $this->mUsertext;
- $qq = wfStrencode( $wgLang->stripForSearch( $q ) );
- $this->mSearchterms = preg_split( '/\s+/', $q );
- $this->mTitlecond = " MATCH(si_title) AGAINST('$qq' IN BOOLEAN MODE)";
- $this->mTextcond = " (MATCH(si_text) AGAINST('$qq' IN BOOLEAN MODE) AND cur_is_redirect=0)";
+/** @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 showHit( $row )
- {
- global $wgUser, $wgOut;
-
- $t = Title::makeName( $row->cur_namespace, $row->cur_title );
- $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})" );
-
- $lines = explode( "\n", $row->cur_text );
- $pat1 = "/(.*)(" . implode( "|", $this->mSearchterms ) . ")(.*)/i";
- $lineno = 0;
-
- foreach ( $lines as $line ) {
- if ( 0 == $contextlines ) { break; }
- --$contextlines;
- ++$lineno;
- if ( ! preg_match( $pat1, $line, $m ) ) { continue; }
-
- $pre = $m[1];
- if ( 0 == $contextchars ) { $pre = "..."; }
- else {
- if ( strlen( $pre ) > $contextchars ) {
- $pre = "..." . substr( $pre, -$contextchars );
- }
- }
- $pre = wfEscapeHTML( $pre );
-
- if ( count( $m ) < 3 ) { $post = ""; }
- else { $post = $m[3]; }
-
- if ( 0 == $contextchars ) { $post = "..."; }
- else {
- if ( strlen( $post ) > $contextchars ) {
- $post = substr( $post, 0, $contextchars ) . "...";
- }
- }
- $post = wfEscapeHTML( $post );
- $found = wfEscapeHTML( $m[2] );
+ function numRows() {
+ return 0;
+ }
- $line = "{$pre}{$found}{$post}";
- $pat2 = "/(" . implode( "|", $this->mSearchterms ) . ")/i";
- $line = preg_replace( $pat2,
- "<font color='red'>\\1</font>", $line );
+ /**
+ * Return true if results are included in this result set.
+ * @return bool
+ * @abstract
+ */
+ function hasResults() {
+ return false;
+ }
- $wgOut->addHTML( "<br><small>{$lineno}: {$line}</small>\n" );
- }
- $wgOut->addHTML( "</li>\n" );
+ /**
+ * 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;
}
- function goResult()
- {
- global $wgOut;
- $fname = "SearchEngine::goResult";
-
- $search = $_REQUEST['search'];
+ /**
+ * 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;
+ }
- # First try to go to page as entered
- #
- $t = Title::newFromText( $search );
+ /**
+ * Some search modes return a suggested alternate term if there are
+ * no exact hits. Check hasSuggestion() first.
+ *
+ * @return string
+ * @access public
+ */
+ function getSuggestion() {
+ return '';
+ }
- if ( 0 != $t->getArticleID() ) {
- $wgOut->redirect( wfLocalUrl( $t->getPrefixedURL() ) );
- return;
- }
+ /**
+ * Fetches next search result, or false.
+ * @return SearchResult
+ * @access public
+ * @abstract
+ */
+ function next() {
+ return false;
+ }
+}
- # Now try all lower case (i.e. first letter capitalized)
- #
- $t = Title::newFromText( strtolower( $search ) );
- if ( 0 != $t->getArticleID() ) {
- $wgOut->redirect( wfLocalUrl( $t->getPrefixedURL() ) );
- return;
- }
+/** @package MediaWiki */
+class SearchResult {
+ function SearchResult( $row ) {
+ $this->mTitle = Title::makeTitle( $row->page_namespace, $row->page_title );
+ }
- # Now try capitalized string
- #
- $t = Title::newFromText( ucwords( strtolower( $search ) ) );
- if ( 0 != $t->getArticleID() ) {
- $wgOut->redirect( wfLocalUrl( $t->getPrefixedURL() ) );
- return;
- }
+ /**
+ * @return Title
+ * @access public
+ */
+ function getTitle() {
+ return $this->mTitle;
+ }
- # Now try all upper case
- #
- $t = Title::newFromText( strtoupper( $search ) );
- if ( 0 != $t->getArticleID() ) {
- $wgOut->redirect( wfLocalUrl( $t->getPrefixedURL() ) );
- return;
- }
+ /**
+ * @return double or null if not supported
+ */
+ function getScore() {
+ return null;
+ }
+}
- # Try a near match
- #
- if( !$wgDisableTextSearch ) {
- $this->parseQuery();
- $sql = "SELECT cur_id,cur_title,cur_namespace,si_page FROM cur,searchindex " .
- "WHERE cur_id=si_page AND {$this->mTitlecond} ORDER BY cur_namespace LIMIT 1";
-
- if ( "" != $this->mTitlecond ) {
- $res = wfQuery( $sql, DB_READ, $fname );
- }
- if ( isset( $res ) && 0 != wfNumRows( $res ) ) {
- $s = wfFetchObject( $res );
-
- $t = Title::makeTitle( $s->cur_namespace, $s->cur_title );
- $wgOut->redirect( wfLocalUrl( $t->getPrefixedURL() ) );
- return;
- }
- }
- $wgOut->addHTML( wfMsg("nogomatch",
- htmlspecialchars( wfLocalUrl( ucfirst($this->mUsertext), "action=edit") ) )
- . "\n<p>" );
- $this->showResults();
+/**
+ * @package MediaWiki
+ */
+class SearchEngineDummy {
+ function search( $term ) {
+ return null;
}
+ function setLimitOffset($l, $o) {}
+ function legalSearchChars() {}
+ function update() {}
+ function setnamespaces() {}
+ function searchtitle() {}
+ function searchtext() {}
}
-?>