handle bad xml

2 files changed, 10 insertions, 0 deletions

diff --git a/it_xml.class b/it_xml.class
index 654042b..6013b52 100644
--- a/it_xml.class
+++ b/it_xml.class
@@ -88,6 +88,9 @@ function from_xml($xmldata, $p)
 		if (!preg_match('/^(<\?xml|\xEF\xBB\xBF|\xFE\xFF|\xFF\xFE|\x00\x00\xFE\xFF|\x00\x00\xFF\xFE)/', $xmldata))
 			$xmldata = '<?xml version="1.0" encoding="' . $this->_p['encoding'] . '"?>' . $xmldata;
 
+		# decode illegal entities but protect semantically important ones
+		$xmldata = html_entity_decode(preg_replace('/&(amp|lt|gt|#38|#60|#62|#x26|#x3C|#3E);/i', '&amp;$1;', $xmldata));
+
 		$result = xml_parse($parser, $xmldata);
 	}
 
diff --git a/tests/it_xml.t b/tests/it_xml.t
index 7e555bd..a0922c5 100755
--- a/tests/it_xml.t
+++ b/tests/it_xml.t
@@ -46,6 +46,13 @@ match(
 	'Character data with entities'
 );
 
+match(
+	'<foo>&amp;&lt;&gt;&#38;&#60;&#62;&#x26;&#x3C;&#x3E; &uuml;</foo>',
+	'foo Object ( [val] => &<>&<>&<> ü ) ',
+	'Predecode illegal entities',
+);
+
+
 # Test inheritance
 class my_xml extends it_xml
 {