From 643a9e3290c34052de516dfdf10718710e082afe Mon Sep 17 00:00:00 2001
From: Maximilian Schwerin <maximilian.schwerin@tigris.de>
Date: Wed, 28 Oct 2015 11:38:32 +0100
Subject: [PATCH] Match unicode characters when comparing for alphanum

---
 HtmlDiff.php | 6 +++++-
 1 file changed, 5 insertions(+), 1 deletion(-)

diff --git a/HtmlDiff.php b/HtmlDiff.php
index 74a5d35..15dc7d0 100644
--- a/HtmlDiff.php
+++ b/HtmlDiff.php
@@ -103,7 +103,7 @@ private function ConvertHtmlToListOfWords( $characterString ) {
 							$current_word = $character;
 							$mode = 'whitespace';
 						} else {
-							if( ctype_alnum( $character ) && ( strlen($current_word) == 0 || ctype_alnum( $current_word ) ) ) {
+							if( $this->IsAlphaNum( $character ) && ( strlen($current_word) == 0 || $this->IsAlphaNum( $current_word ) ) ) {
 								$current_word .= $character;
 							} else {
 								$words[] = $current_word;
@@ -165,6 +165,10 @@ private function IsWhiteSpace( $value ) {
 			return !preg_match( '[^\s]', $value );
 		}
 
+		private function IsAlphaNum( $value ) {
+			return preg_match( '/[\p{L}\p{N}]+/u', $value );
+		}
+
 		private function Explode( $value ) {
 			// as suggested by @onassar
 			return preg_split( '//u', $value );