From e3ff109b37acf8b95130f18633b17adb44f209ef Mon Sep 17 00:00:00 2001
From: Aaron Parecki <aaron@parecki.com>
Date: Sun, 28 Feb 2016 15:56:52 -0800
Subject: [PATCH] restrict matching mf2 classes to only lowercase names

see http://microformats.org/wiki/microformats2-parsing-issues#ignore_u-camelCase_properties for background
---
 lib/Formats/HTMLPurifier_AttrDef_HTML_Microformats2.php | 2 +-
 tests/SanitizeTest.php                                  | 1 +
 tests/data/sanitize.example/entry-with-mf2-classes      | 2 ++
 3 files changed, 4 insertions(+), 1 deletion(-)

diff --git a/lib/Formats/HTMLPurifier_AttrDef_HTML_Microformats2.php b/lib/Formats/HTMLPurifier_AttrDef_HTML_Microformats2.php
index 5d19c8e..d3d506f 100644
--- a/lib/Formats/HTMLPurifier_AttrDef_HTML_Microformats2.php
+++ b/lib/Formats/HTMLPurifier_AttrDef_HTML_Microformats2.php
@@ -32,7 +32,7 @@ class HTMLPurifier_AttrDef_HTML_Microformats2 extends HTMLPurifier_AttrDef_HTML_
     {
         $ret = array();
         foreach ($tokens as $token) {
-            if(preg_match('/^([hpue]|dt)-.+/', $token)) {
+            if(preg_match('/^([hpue]|dt)-[a-z\-]+$/', $token)) {
                 $ret[] = $token;
             }
         }
diff --git a/tests/SanitizeTest.php b/tests/SanitizeTest.php
index c123ee0..1d1dbb9 100644
--- a/tests/SanitizeTest.php
+++ b/tests/SanitizeTest.php
@@ -79,6 +79,7 @@ class SanitizeTest extends PHPUnit_Framework_TestCase {
 
     $this->assertEquals('entry', $data['data']['type']);
     $this->assertContains('<h2 class="p-name">Hello World</h2>', $html);
+    $this->assertContains('<h3>Utility Class</h3>', $html);
   }
 
 }
diff --git a/tests/data/sanitize.example/entry-with-mf2-classes b/tests/data/sanitize.example/entry-with-mf2-classes
index 8579def..64e5e63 100644
--- a/tests/data/sanitize.example/entry-with-mf2-classes
+++ b/tests/data/sanitize.example/entry-with-mf2-classes
@@ -12,6 +12,8 @@ Connection: keep-alive
     <div class="e-content">
       <h2 class="p-name header">Hello World</h2>
       This content has a valid tags that contains microformats classes.
+      <h3 class="u-utilityClass">Utility Class</h3>
+      The u-utilityClass should not be returned according to http://microformats.org/wiki/microformats2-parsing-issues#ignore_u-camelCase_properties
     </div>
   </body>
 </html>