From 7bf6095d73d2117f4ebc60419a9cf809f9072738 Mon Sep 17 00:00:00 2001
From: Brion Vibber <brion@users.mediawiki.org>
Date: Sun, 14 Nov 2004 02:24:44 +0000
Subject: [PATCH] Fix UTF-8 validation bug where some cases didn't get
 replacement chars inserted correctly

---
 includes/normal/CleanUpTest.php | 26 ++++++++++++++++++++++++++
 includes/normal/UtfNormal.php   |  1 -
 2 files changed, 26 insertions(+), 1 deletion(-)

diff --git a/includes/normal/CleanUpTest.php b/includes/normal/CleanUpTest.php
index 64f7b63904..2dd8d5616f 100644
--- a/includes/normal/CleanUpTest.php
+++ b/includes/normal/CleanUpTest.php
@@ -1,4 +1,8 @@
 <?php
+/** */
+if( isset( $_SERVER['argv'] ) && in_array( '--icu', $_SERVER['argv'] ) ) {
+	dl( 'php_utfnormal.so' );
+}
 
 #ini_set( 'memory_limit', '40M' );
 
@@ -214,6 +218,28 @@ class CleanUpTest extends PHPUnit_TestCase {
 			}
 		}
 	}
+	
+	function testChunkRegression() {
+		# Check for regression against a chunking bug
+		$text   = "\x46\x55\xb8" .
+		          "\xdc\x96" . 
+		          "\xee" .
+		          "\xe7" .
+		          "\x44" .
+		          "\xaa" .
+		          "\x2f\x25";
+		$expect = "\x46\x55\xef\xbf\xbd" .
+		          "\xdc\x96" . 
+		          "\xef\xbf\xbd" .
+		          "\xef\xbf\xbd" .
+		          "\x44" .
+		          "\xef\xbf\xbd" .
+		          "\x2f\x25";
+
+		$this->assertEquals(
+			bin2hex( $expect ),
+			bin2hex( UtfNormal::cleanUp( $text ) ) );
+	}
 
 }
 
diff --git a/includes/normal/UtfNormal.php b/includes/normal/UtfNormal.php
index 2883342f8e..7dd9072185 100644
--- a/includes/normal/UtfNormal.php
+++ b/includes/normal/UtfNormal.php
@@ -350,7 +350,6 @@ class UtfNormal {
 								$replace[] = array( UTF8_REPLACEMENT,
 													$base + $i + 1 - strlen( $sequence ),
 													strlen( $sequence ) );
-								$base += $chunk;
 								break 2;
 							} else {
 								# Illegal tail byte; abandon the sequence.
-- 
2.20.1