From: Max Semenik <maxsem@users.mediawiki.org>
Date: Wed, 4 Nov 2009 15:18:30 +0000 (+0000)
Subject: (bug 18609) Search index was empty for some pages
X-Git-Tag: 1.31.0-rc.0~38977
X-Git-Url: http://git.cyclocoop.org/%40spipnet%40?a=commitdiff_plain;h=b86085163121be5469fa42530b81f2b8ea362591;p=lhc%2Fweb%2Fwiklou.git

(bug 18609) Search index was empty for some pages
---

diff --git a/RELEASE-NOTES b/RELEASE-NOTES
index 8c10923d02..a2316d23f5 100644
--- a/RELEASE-NOTES
+++ b/RELEASE-NOTES
@@ -619,6 +619,7 @@ Hopefully we will remove this configuration var soon)
 * Removed section edit links in edit conflict form
 * Allow SpecialActiveusers to work on non-MySQL databases
 * (bug 6579) Fixed protecting images from uploading only 
+* (bug 18609) Search index was empty for some pages
 
 == API changes in 1.16 ==
 
diff --git a/includes/search/SearchUpdate.php b/includes/search/SearchUpdate.php
index edf2bcbb31..b9c2335151 100644
--- a/includes/search/SearchUpdate.php
+++ b/includes/search/SearchUpdate.php
@@ -46,7 +46,7 @@ class SearchUpdate {
 		$text = $wgContLang->stripForSearch( $this->mText );
 
 		wfProfileIn( $fname.'-regexps' );
-		$text = preg_replace( "/<\\/?\\s*[A-Za-z][A-Za-z0-9]*\\s*([^>]*?)>/",
+		$text = preg_replace( "/<\\/?\\s*[A-Za-z][^>]*?>/",
 			' ', $wgContLang->lc( " " . $text . " " ) ); # Strip HTML markup
 		$text = preg_replace( "/(^|\\n)==\\s*([^\\n]+)\\s*==(\\s)/sD",
 		  "\\1\\2 \\2 \\2\\3", $text ); # Emphasize headings
diff --git a/t/DatabaseMock.inc b/t/DatabaseMock.inc
new file mode 100644
index 0000000000..59d380f5b0
--- /dev/null
+++ b/t/DatabaseMock.inc
@@ -0,0 +1,33 @@
+<?php
+/**
+ * Mock database class for tests, does nothing.
+ * Include after LocalSettings.php
+ */
+
+$wgDBtype = 'mock';
+
+class DatabaseMock extends DatabaseBase {
+	function __construct( $server = false, $user = false, $password = false, $dbName = false,
+		$failFunction = false, $flags = 0, $tablePrefix = 'get from global' )
+	{
+		$this->mConn = true;
+		$this->mOpened = true;
+	}
+
+	function open( $server, $user, $password, $dbName ) { return true; }
+	function doQuery( $sql ) {}
+	function fetchObject( $res ) {}
+	function fetchRow( $res ) {}
+	function numRows( $res ) {}
+	function numFields( $res ) {}
+	function fieldName( $res, $n ) {}
+	function insertId() {}
+	function dataSeek( $res, $row ) {}
+	function lastErrno() { return 0; }
+	function lastError() { return ''; }
+	function affectedRows() {}
+	function fieldInfo( $table, $field ) {}
+	function strencode( $s ) {}
+	function getSoftwareLink() {}
+	function getServerVersion() {}
+}
\ No newline at end of file
diff --git a/t/inc/SearchUpdate.t b/t/inc/SearchUpdate.t
new file mode 100644
index 0000000000..da061f7c02
--- /dev/null
+++ b/t/inc/SearchUpdate.t
@@ -0,0 +1,61 @@
+#!/usr/bin/env php
+<?php
+
+require 't/Test.php';
+
+plan( 4 );
+
+define( 'MEDIAWIKI', 1 );
+require 'includes/Defines.php';
+require 'includes/ProfilerStub.php';
+require 'includes/AutoLoader.php';
+require 'LocalSettings.php';
+
+require 't/DatabaseMock.inc';
+
+$wgSearchType = 'MockSearch';
+
+require 'includes/Setup.php';
+
+
+class MockSearch extends SearchEngine {
+	public static $id;
+	public static $title;
+	public static $text;
+
+	public function __construct( $db ) {
+	}
+
+	public function update( $id, $title, $text ) {
+		self::$id = $id;
+		self::$title = $title;
+		self::$text = $text;
+	}
+}
+
+function update( $text, $title = 'Test', $id = 1 ) {
+	$u = new SearchUpdate( $id, $title, $text );
+	$u->doUpdate();
+	return array( MockSearch::$title, MockSearch::$text );
+}
+
+function updateText( $text ) {
+	list( $title, $resultText ) = update( $text );
+	$resultText = trim( $resultText ); // abstract from some implementation details
+	return $resultText;
+}
+
+is( updateText( '<div>TeSt</div>' ), 'test', 'HTML stripped, text lowercased' );
+
+is( updateText( <<<EOT
+<table style="color:red; font-size:100px">
+	<tr class="scary"><td><div>foo</div></td><tr>bar</td></tr>
+	<tr><td>boz</td><tr>quux</td></tr>
+</table>
+EOT
+), 'foo bar boz quux', 'Stripping HTML tables' );
+
+is( updateText( 'a > b' ), 'a b', 'Handle unclosed tags' );
+
+$text = str_pad( "foo <barbarbar \n", 10000, 'x' );
+ok( updateText( $text ) != '', 'Bug 18609' );
\ No newline at end of file