-<?
+<?php
# See search.doc
class SearchEngine {
var $addtoquery = array();
var $namespacesToSearch = array();
var $alternateTitle;
+ var $all_titles = false;
function SearchEngine( $text )
{
return $wgUser->getOption( "searchNs".$i );
} else {
// User is not logged in so we give him the global default namespaces
- return $wgNamespacesToBeSearchedDefault[ $i ];
+ return !empty($wgNamespacesToBeSearchedDefault[ $i ]);
}
}
# that is done by showResults()
function powersearch()
{
- global $wgUser, $wgOut, $wgLang, $wgTitle;
-
- $search = $_REQUEST['search'];
- $searchx = $_REQUEST['searchx'];
- $listredirs = $_REQUEST['redirs'];
+ global $wgUser, $wgOut, $wgLang, $wgTitle, $wgRequest;
+ $sk =& $wgUser->getSkin();
+
+ $search = $wgRequest->getText( 'search' );
+ $searchx = $wgRequest->getVal( 'searchx' );
+ $listredirs = $wgRequest->getVal( 'redirs' );
$ret = wfMsg("powersearchtext"); # Text to be returned
$tempText = ""; # Temporary text, for substitution into $ret
if ( !isset( $searchx ) ) {
$checkboxValue = $this->initNamespaceCheckbox( $i );
} else {
- $checkboxValue = $_REQUEST[$formVar];
+ $checkboxValue = $wgRequest->getVal( $formVar );
}
$checked = "";
if ( $checkboxValue == 1 ) {
- $checked = " checked";
+ $checked = " checked='checked'";
$this->addtoquery["ns{$i}"] = 1;
array_push( $this->namespacesToSearch, $i );
}
if ( $tempText !== "" ) {
$tempText .= " ";
}
- $tempText .= "<input type=checkbox value=\"1\" name=\"" .
- "ns{$i}\"{$checked}>{$name}\n";
+ $tempText .= "<input type='checkbox' value=\"1\" name=\"" .
+ "ns{$i}\"{$checked} />{$name}\n";
}
$ret = str_replace ( "$1", $tempText, $ret );
$checked = "";
if ( $listredirs == 1 ) {
$this->addtoquery["redirs"] = 1;
- $checked = " checked";
+ $checked = " checked='checked'";
}
- $tempText = "<input type=checkbox value=1 name=\"redirs\"{$checked}>\n";
+ $tempText = "<input type='checkbox' value='1' name=\"redirs\"{$checked} />\n";
$ret = str_replace( "$2", $tempText, $ret );
# Search field
- $tempText = "<input type=text name=\"search\" value=\"" .
- htmlspecialchars( $search ) ."\" width=80>\n";
+ $tempText = "<input type='text' name=\"search\" value=\"" .
+ htmlspecialchars( $search ) ."\" width='80' />\n";
$ret = str_replace( "$3", $tempText, $ret );
# Searchx button
- $tempText = "<input type=submit name=\"searchx\" value=\"" .
- wfMsg("powersearch") . "\">\n";
+ $tempText = "<input type='submit' name=\"searchx\" value=\"" .
+ wfMsg("powersearch") . "\" />\n";
$ret = str_replace( "$9", $tempText, $ret );
- $ret = "<br><br>\n<form id=\"powersearch\" method=\"get\" " .
- "action=\"" . wfLocalUrl( "" ) . "\">\n{$ret}\n</form>\n";
+ $action = $sk->escapeSearchLink();
+ $ret = "<br /><br />\n<form id=\"powersearch\" method=\"get\" " .
+ "action=\"$action\">\n{$ret}\n</form>\n";
if ( isset ( $searchx ) ) {
if ( ! $listredirs ) {
return $ret;
}
+ function setupPage() {
+ global $wgOut;
+ $wgOut->setPageTitle( wfMsg( "searchresults" ) );
+ $q = wfMsg( "searchquery", htmlspecialchars( $this->mUsertext ) );
+ $wgOut->setSubtitle( $q );
+ $wgOut->setArticleRelated( false );
+ $wgOut->setRobotpolicy( "noindex,nofollow" );
+ }
+
# Perform the search and construct the results page
function showResults()
{
- global $wgUser, $wgTitle, $wgOut, $wgLang, $wgDisableTextSearch;
- global $wgInputEncoding;
+ global $wgUser, $wgTitle, $wgOut, $wgLang, $wgRequest;
+ global $wgDisableTextSearch, $wgInputEncoding;
$fname = "SearchEngine::showResults";
- $search = $_REQUEST['search'];
+ $search = $wgRequest->getText( '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" );
-
+ $this->setupPage();
+
$sk = $wgUser->getSkin();
$header = wfMsg( "searchresulttext", $sk->makeKnownLink(
wfMsg( "searchhelppage" ), wfMsg( "searchingwikipedia" ) ) );
$this->parseQuery();
if ( "" == $this->mTitlecond || "" == $this->mTextcond ) {
$wgOut->addHTML( "<h2>" . wfMsg( "badquery" ) . "</h2>\n" .
- "<p>" . wfMsg( "badquerytext" ) );
+ "<p>" . wfMsg( "badquerytext" ) . "</p>\n" );
return;
}
list( $limit, $offset ) = wfCheckLimits( 20, "searchlimit" );
$redircond = $this->searchRedirects();
if ( $wgDisableTextSearch ) {
- $wgOut->addHTML( wfMsg( "searchdisabled", htmlspecialchars( $search ), $wgInputEncoding ) );
+ $wgOut->addHTML( wfMsg( "searchdisabled" ) );
+ $wgOut->addHTML( wfMsg( "googlesearch", htmlspecialchars( $search ), $GLOBALS['wgInputEncoding'] ) );
} else {
$sql = "SELECT cur_id,cur_namespace,cur_title," .
"cur_text FROM cur,searchindex " .
$this->parseQuery();
if ( "" == $this->mTitlecond || "" == $this->mTextcond ) {
$wgOut->addHTML( "<h2>" . wfMsg( "badquery" ) . "</h2>\n" .
- "<p>" . wfMsg( "badquerytext" ) );
+ "<p>" . wfMsg( "badquerytext" ) . "</p>\n" );
return;
}
list( $limit, $offset ) = wfCheckLimits( 20, "searchlimit" );
} else {
$top = wfShowingResultsNum( $offset, $limit, $num );
}
- $wgOut->addHTML( "<p>{$top}\n" );
+ $wgOut->addHTML( "<p>{$top}</p>\n" );
# For powersearch
$sl = wfViewPrevNext( $offset, $limit, "",
"search=" . wfUrlencode( $this->mUsertext ) . $a2l );
- $wgOut->addHTML( "<br>{$sl}\n" );
+ $wgOut->addHTML( "<br />{$sl}\n" );
$foundsome = false;
$wgOut->addHTML( "</ol>\n" );
}
if ( ! $foundsome ) {
- $wgOut->addHTML( "<p>" . wfMsg( "nonefound" ) . "\n" );
+ $wgOut->addHTML( "<p>" . wfMsg( "nonefound" ) . "</p>\n" );
}
- $wgOut->addHTML( "<p>{$sl}\n" );
+ $wgOut->addHTML( "<p>{$sl}</p>\n" );
$wgOut->addHTML( $powersearch );
}
}
# Use cleaner boolean search if available
return $this->parseQuery4();
}
+ # 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->mUsertext );
$line = preg_replace( $pat2,
"<font color='red'>\\1</font>", $line );
- $wgOut->addHTML( "<br><small>{$lineno}: {$line}</small>\n" );
+ $wgOut->addHTML( "<br /><small>{$lineno}: {$line}</small>\n" );
}
$wgOut->addHTML( "</li>\n" );
}
function goResult()
{
- global $wgOut, $wgDisableTextSearch;
+ global $wgOut, $wgRequest, $wgGoToEdit;
+ global $wgDisableTextSearch;
$fname = "SearchEngine::goResult";
- $search = $_REQUEST['search'];
+ $search = trim( $wgRequest->getText( "search" ) );
- # First try to go to page as entered.
+ # Try to go to page as entered.
#
$t = Title::newFromText( $search );
return;
}
- if ( 0 != $t->getArticleID() ) {
- $wgOut->redirect( wfLocalUrl( $t->getPrefixedURL() ) );
+ # Exact match? No need to look further.
+ if ( $t->getNamespace() == NS_SPECIAL || 0 != $t->getArticleID() ) {
+ $wgOut->redirect( $t->getFullURL() );
return;
}
#
$t = Title::newFromText( strtolower( $search ) );
if ( 0 != $t->getArticleID() ) {
- $wgOut->redirect( wfLocalUrl( $t->getPrefixedURL() ) );
+ $wgOut->redirect( $t->getFullURL() );
return;
}
#
$t = Title::newFromText( ucwords( strtolower( $search ) ) );
if ( 0 != $t->getArticleID() ) {
- $wgOut->redirect( wfLocalUrl( $t->getPrefixedURL() ) );
+ $wgOut->redirect( $t->getFullURL() );
return;
}
#
$t = Title::newFromText( strtoupper( $search ) );
if ( 0 != $t->getArticleID() ) {
- $wgOut->redirect( wfLocalUrl( $t->getPrefixedURL() ) );
+ $wgOut->redirect( $t->getFullURL() );
return;
}
- # Try a near match
- #
- if( !$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( $t->getPrefixedURL() ) );
- return;
+ # Entering an IP address goes to the contributions page
+ if ( preg_match( '/^\d{1,3}\.\d{1,3}\.\d{1,3}\.\d{1,3}$/', $search ) ) {
+ $title = Title::makeTitle( NS_SPECIAL, "Contributions" );
+ $wgOut->redirect( $title->getFullUrl( "target=$search" ) );
+ return;
+ }
+
+ # No match, generate an edit URL
+ $t = Title::newFromText( $this->mUsertext );
+
+ # 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 ) . "</p>\n" );
+
+ # Try a fuzzy title search
+ $anyhit = false;
+ global $wgDisableFuzzySearch;
+ if(! $wgDisableFuzzySearch ){
+ foreach( array(NS_MAIN, NS_WP, NS_USER, NS_IMAGE, NS_MEDIAWIKI) as $namespace){
+ $anyhit |= SearchEngine::doFuzzyTitleSearch( $search, $namespace );
+ }
+ }
+
+ if( ! $anyhit ){
+ return $this->showResults();
+ }
+ }
+
+ /* static */ function doFuzzyTitleSearch( $search, $namespace ){
+ global $wgLang, $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 = $wgLang->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 = "=== " . $wgLang->getNsText( $namespace ) . " ===\n" . $wikitext;
+ $wgOut->addWikiText( $wikitext );
+ return true;
+ }
+ return false;
+ }
+
+ /* static */ 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++;
}
}
- $wgOut->addHTML( wfMsg("nogomatch",
- htmlspecialchars( wfLocalUrl( ucfirst($this->mUsertext), "action=edit") ) )
- . "\n<p>" );
- $this->showResults();
+ usort($result, "SearchEngine_pcmp");
+ return $result;
+ }
+
+ /* static */ function getTitlesByLength($aLength, $aNamespace = 0){
+ global $wgMemc, $wgDBname;
+
+ // to avoid multiple costly SELECTs in case of no memcached
+ if( $this->all_titles ){
+ if( isset( $this->all_titles[$aLength][$aNamespace] ) ){
+ return $this->all_titles[$aLength][$aNamespace];
+ } else {
+ return array();
+ }
+ }
+
+ $mkey = "$wgDBname:titlesbylength:$aLength:$aNamespace";
+ $mkeyts = "$wgDBname:titlesbylength:createtime";
+ $ts = $wgMemc->get( $mkeyts );
+ $result = $wgMemc->get( $mkey );
+
+ 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();
+ }
+
+ $wgMemc->set( $mkeyts, time() );
+
+ $res = wfQuery("SELECT cur_title, cur_namespace FROM cur", DB_READ);
+ $titles = array(); // length, ns, [titles]
+ while( $obj = wfFetchObject( $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->all_titles = $titles;
+ if( isset( $titles[$aLength][$aNamespace] ) )
+ return $titles[$aLength][$aNamespace];
+ else
+ return array();
}
}
+/* private static */ function SearchEngine_pcmp($a, $b){ return $a[0] - $b[0]; }
+
?>
dépôts / lhc / web / wiklou.git / blobdiff