-<?
-# See search.doc
-
+<?php
+/**
+ * Contain a class for special pages
+ * @package MediaWiki
+ */
+
+/**
+ * @package MediaWiki
+ */
class SearchEngine {
- /* private */ var $mUsertext, $mSearchterms;
- /* private */ var $mTitlecond, $mTextcond;
-
- var $doSearchRedirects = true;
- var $addtoquery = array();
- var $namespacesToSearch = array();
- var $alternateTitle;
-
- 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();
- }
-
- function queryNamespaces()
- {
- return "cur_namespace IN (" . implode( ",", $this->namespacesToSearch ) . ")";
- #return "1";
+ var $limit = 10;
+ var $offset = 0;
+ var $searchTerms = array();
+ var $namespaces = array( 0 );
+ var $showRedirects = false;
+
+ /**
+ * Perform a full text search query and return a result set.
+ *
+ * @param string $term - Raw search term
+ * @param array $namespaces - List of namespaces to search
+ * @return ResultWrapper
+ * @access public
+ */
+ function searchText( $term ) {
+ return $this->db->resultObject( $this->db->query( $this->getQuery( $this->filter( $term ), true ) ) );
}
- function searchRedirects()
- {
- if ( $this->doSearchRedirects ) return "";
- return "AND cur_is_redirect=0 ";
+ /**
+ * Perform a title-only search query and return a result set.
+ *
+ * @param string $term - Raw search term
+ * @param array $namespaces - List of namespaces to search
+ * @return ResultWrapper
+ * @access public
+ */
+ function searchTitle( $term ) {
+ return $this->db->resultObject( $this->db->query( $this->getQuery( $this->filter( $term ), false ) ) );
}
-
- function powersearch()
- {
- global $wgUser, $wgOut, $wgLang, $wgTitle;
- $nscb = array();
-
- $search = $_REQUEST['search'];
- $searchx = $_REQUEST['searchx'];
- $listredirs = $_REQUEST['redirs'];
- $nscb[0] = $_REQUEST['ns0'];
- $nscb[1] = $_REQUEST['ns1'];
- $nscb[2] = $_REQUEST['ns2'];
- $nscb[3] = $_REQUEST['ns3'];
- $nscb[4] = $_REQUEST['ns4'];
- $nscb[5] = $_REQUEST['ns5'];
- $nscb[6] = $_REQUEST['ns6'];
- $nscb[7] = $_REQUEST['ns7'];
-
- if ( ! isset ( $searchx ) ) { /* First time here */
- $nscb[0] = $listredirs = 1; /* All others should be unset */
- }
- $this->checkboxes["searchx"] = 1;
- $ret = wfMsg("powersearchtext");
-
- # Determine namespace checkboxes
-
- $ns = $wgLang->getNamespaces();
- array_shift( $ns ); /* Skip "Special" */
-
- $r1 = "";
- for ( $i = 0; $i < count( $ns ); ++$i ) {
- $checked = "";
- if ( $nscb[$i] == 1 ) {
- $checked = " checked";
- $this->addtoquery["ns{$i}"] = 1;
- array_push( $this->namespacesToSearch, $i );
- }
- $name = str_replace( "_", " ", $ns[$i] );
- if ( "" == $name ) { $name = "(Main)"; }
-
- if ( 0 != $i ) { $r1 .= " "; }
- $r1 .= "<input type=checkbox value=\"1\" name=\"" .
- "ns{$i}\"{$checked}>{$name}\n";
+
+ /**
+ * 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
+ * @access private
+ */
+ function getNearMatch( $term ) {
+ # Exact match? No need to look further.
+ $title = Title::newFromText( $term );
+ if ( $title->getNamespace() == NS_SPECIAL || 0 != $title->getArticleID() ) {
+ return $title;
}
- $ret = str_replace ( "$1", $r1, $ret );
-
- # List redirects checkbox
- $checked = "";
- if ( $listredirs == 1 ) {
- $this->addtoquery["redirs"] = 1;
- $checked = " checked";
+ # Now try all lower case (i.e. first letter capitalized)
+ #
+ $title = Title::newFromText( strtolower( $term ) );
+ if ( 0 != $title->getArticleID() ) {
+ return $title;
}
- $r2 = "<input type=checkbox value=1 name=\"redirs\"{$checked}>\n";
- $ret = str_replace( "$2", $r2, $ret );
-
- # Search field
- $r3 = "<input type=text name=\"search\" value=\"" .
- htmlspecialchars( $search ) ."\" width=80>\n";
- $ret = str_replace( "$3", $r3, $ret );
-
- # Searchx button
-
- $r9 = "<input type=submit name=\"searchx\" value=\"" .
- wfMsg("powersearch") . "\">\n";
- $ret = str_replace( "$9", $r9, $ret );
-
- $ret = "<br><br>\n<form id=\"powersearch\" method=\"get\" " .
- "action=\"" . wfLocalUrl( "" ) . "\">\n{$ret}\n</form>\n";
-
- if ( isset ( $searchx ) ) {
- if ( ! $listredirs ) { $this->doSearchRedirects = false; }
+ # Now try capitalized string
+ #
+ $title = Title::newFromText( ucwords( strtolower( $term ) ) );
+ if ( 0 != $title->getArticleID() ) {
+ return $title;
}
- return $ret;
- }
- function showResults()
- {
- global $wgUser, $wgTitle, $wgOut, $wgLang, $wgDisableTextSearch;
- $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;
+ # Now try all upper case
+ #
+ $title = Title::newFromText( strtoupper( $term ) );
+ if ( 0 != $title->getArticleID() ) {
+ return $title;
}
- list( $limit, $offset ) = wfCheckLimits( 20, "searchlimit" );
- $searchnamespaces = $this->queryNamespaces();
- $redircond = $this->searchRedirects();
+ $title = Title::newFromText( $term );
- $sql = "SELECT cur_id,cur_namespace,cur_title," .
- "cur_text FROM cur,searchindex " .
- "WHERE cur_id=si_page AND {$this->mTitlecond} " .
- "AND {$searchnamespaces} {$redircond}" .
- "LIMIT {$offset}, {$limit}";
- $res1 = wfQuery( $sql, $fname );
- $num = wfNumRows($res1);
-
- if ( $wgDisableTextSearch ) {
- $res2 = 0;
- } else {
- $sql = "SELECT cur_id,cur_namespace,cur_title," .
- "cur_text FROM cur,searchindex " .
- "WHERE cur_id=si_page AND {$this->mTextcond} " .
- "AND {$searchnamespaces} {$redircond} " .
- "LIMIT {$offset}, {$limit}";
- $res2 = wfQuery( $sql, $fname );
- $num = $num + wfNumRows($res2);
+ # 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() );
}
- 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]}" ;
+ # Entering a user goes to the user page whether it's there or not
+ if ( $title->getNamespace() == NS_USER ) {
+ return $title;
}
+
+ return NULL;
+ }
+
+ function legalSearchChars() {
+ return "A-Za-z_'0-9\\x80-\\xFF\\-";
+ }
- $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 );
+ /**
+ * 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 );
+ }
+
+ /**
+ * 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;
+ }
+
+ /**
+ * 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;
}
- wfFreeResult( $res1 );
- $wgOut->addHTML( "</ol>\n" );
}
-
- if ( $wgDisableTextSearch ) {
- $wgOut->addHTML( str_replace( "$1",
- htmlspecialchars( $search ), wfMsg( "searchdisabled" ) ) );
+ 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
+ *
+ * @return string
+ * @access public
+ */
+ function filter( $text ) {
+ $lc = $this->legalSearchChars();
+ return trim( preg_replace( "/[^{$lc}]/", " ", $text ) );
+ }
+
+ /**
+ * Return a partial WHERE clause to exclude redirects, if so set
+ * @return string
+ * @access private
+ */
+ function queryRedirect() {
+ if( $this->showRedirects ) {
+ return 'AND cur_is_redirect=0';
} 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" );
- }
+ return '';
}
- if ( ! $foundsome ) {
- $wgOut->addHTML( "<p>" . wfMsg( "nonefound" ) . "\n" );
+ }
+
+ /**
+ * Return a partial WHERE clause to limit the search to the given namespaces
+ * @return string
+ * @access private
+ */
+ function queryNamespaces() {
+ $namespaces = implode( ',', $this->namespaces );
+ if ($namespaces == '') {
+ $namespaces = '0';
}
- $wgOut->addHTML( "<p>{$sl}\n" );
- $wgOut->addHTML( $powersearch );
+ return 'AND page_namespace IN (' . $namespaces . ')';
}
-
- function legalSearchChars()
- {
- $lc = "A-Za-z_'0-9\\x80-\\xFF\\-";
- return $lc;
+
+ /**
+ * Return a LIMIT clause to limit results on the query.
+ * @return string
+ * @access private
+ */
+ function queryLimit() {
+ return $this->db->limitResult( $this->limit, $this->offset );
}
- 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 ) . " )";
-
- $this->mTextcond = "(" . str_replace( "##field##",
- "si_text", $cond ) . " AND (cur_is_redirect=0) )";
+ /**
+ * Does not do anything for generic search engine
+ * subclasses may define this though
+ * @return string
+ * @access private
+ */
+ function queryRanking($filteredTerm,$fulltext) {
+ return "";
}
- 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( $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)";
+ /**
+ * Construct the full SQL query to do the search.
+ * The guts shoulds be constructed in queryMain()
+ * @param string $filteredTerm
+ * @param bool $fulltext
+ * @access private
+ */
+ function getQuery( $filteredTerm, $fulltext ) {
+ return $this->queryMain( $filteredTerm, $fulltext ) . ' ' .
+ $this->queryRedirect() . ' ' .
+ $this->queryNamespaces() . ' ' .
+ $this->queryRanking($filteredTerm, $fulltext) . ' ' .
+ $this->queryLimit();
}
- 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 = str_replace( "$1", strlen( $row->cur_text ), WfMsg( "nbytes" ) );
- $wgOut->addHTML( "<li>{$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] );
-
- $line = "{$pre}{$found}{$post}";
- $pat2 = "/(" . implode( "|", $this->mSearchterms ) . ")/i";
- $line = preg_replace( $pat2,
- "<font color='red'>\\1</font>", $line );
-
- $wgOut->addHTML( "<br><small>{$lineno}: {$line}</small>\n" );
- }
- $wgOut->addHTML( "</li>\n" );
+ /**
+ * 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;
}
- function goResult()
- {
- global $wgOut, $wgArticle, $wgTitle;
- $fname = "SearchEngine::goResult";
-
- $search = $_REQUEST['search'];
-
- # First try to go to page as entered
- #
- $wgArticle = new Article();
- $wgTitle = Title::newFromText( $search );
-
- if ( 0 != $wgArticle->getID() ) {
- $wgArticle->view();
- return;
- }
-
- # Now try all lower case (i.e. first letter capitalized)
- #
- $wgTitle = Title::newFromText( strtolower( $search ) );
- if ( 0 != $wgArticle->getID() ) {
- $wgArticle->view();
- return;
- }
-
- # Now try capitalized string
- #
- $wgTitle=Title::newFromText( ucwords( strtolower( $search ) ) );
- if ( 0 != $wgArticle->getID() ) {
- $wgArticle->view();
- return;
- }
-
- # Now try all upper case
- #
- $wgTitle = Title::newFromText( strtoupper( $search ) );
- if ( 0 != $wgArticle->getID() ) {
- $wgArticle->view();
- return;
- }
-
- # Try a near match
- #
- $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, $fname );
- }
- if ( isset( $res ) && 0 != wfNumRows( $res ) ) {
- $s = wfFetchObject( $res );
+
+}
- $wgTitle = Title::newFromDBkey( $s->cur_title );
- $wgTitle->setNamespace( $s->cur_namespace );
- $wgArticle->view();
- return;
- }
- $wgOut->addHTML( str_replace( "$1",
- wfLocalUrl( ucfirst($this->mUsertext) . "&action=edit"),
- wfMsg("nogomatch")) . "\n<p>" );
- $this->showResults();
+/**
+ * @package MediaWiki
+ */
+class SearchEngineDummy {
+ function search( $term ) {
+ return null;
}
}