From 849a89b1217b8a90c95bae85f378eeca81f87723 Mon Sep 17 00:00:00 2001
From: Haralan Dobrev <hkdobrev@gmail.com>
Date: Fri, 22 Nov 2013 23:06:20 +0200
Subject: [PATCH] Use UTF-8 encoding for htmlspecialchars. See #36.

Prior to PHP 5.4.0 the default encoding for `htmlentities()`
and `htmlspecialchars` is "ISO-8859-1". For PHP 5.4+ is "UTF-8".

This ensures always the right encoding is used no matter the PHP version
and the locale settings.
---
 Parsedown.php                      | 4 ++--
 tests/data/special_characters.html | 1 +
 tests/data/special_characters.md   | 2 ++
 3 files changed, 5 insertions(+), 2 deletions(-)

diff --git a/Parsedown.php b/Parsedown.php
index 8a93030..0ab3faf 100755
--- a/Parsedown.php
+++ b/Parsedown.php
@@ -568,7 +568,7 @@ class Parsedown
 				case 'code_block':
 				case 'fenced_code_block':
 
-					$text = htmlspecialchars($element['text'], ENT_NOQUOTES);
+					$text = htmlspecialchars($element['text'], ENT_NOQUOTES, 'UTF-8');
 
 					strpos($text, "\x1A\\") !== FALSE and $text = strtr($text, $this->escape_sequence_map);
 
@@ -634,7 +634,7 @@ class Parsedown
 			foreach ($matches as $matches)
 			{
 				$element_text = $matches[1];
-				$element_text = htmlspecialchars($element_text, ENT_NOQUOTES);
+				$element_text = htmlspecialchars($element_text, ENT_NOQUOTES, 'UTF-8');
 
 				# decodes escape sequences
 
diff --git a/tests/data/special_characters.html b/tests/data/special_characters.html
index 7fd3fbb..001d629 100644
--- a/tests/data/special_characters.html
+++ b/tests/data/special_characters.html
@@ -1,4 +1,5 @@
 <p>AT&amp;T has an ampersand in their name</p>
+<pre><code>Let's play some cards ♠ ♣ ♥ ♦</code></pre>
 <p>AT&amp;T is another way to write it</p>
 <p>this &amp; that</p>
 <p>4 &lt; 5 and 6 > 5</p>
diff --git a/tests/data/special_characters.md b/tests/data/special_characters.md
index 440c05e..94de13e 100644
--- a/tests/data/special_characters.md
+++ b/tests/data/special_characters.md
@@ -1,5 +1,7 @@
 AT&T has an ampersand in their name
 
+    Let's play some cards ♠ ♣ ♥ ♦
+
 AT&T is another way to write it
 
 this & that