From 6c905411bac66caad5e220a70e3d561b8a648507 Mon Sep 17 00:00:00 2001
From: mark_story <mark@mark-story.com>
Date: Sat, 14 Jul 2012 11:53:37 -0400
Subject: [PATCH] Fix XML decoding attack via external entities.

---
 lib/Cake/Test/Case/Utility/XmlTest.php | 18 ++++++++++
 lib/Cake/Utility/Xml.php               | 46 ++++++++++++++++++--------
 2 files changed, 51 insertions(+), 13 deletions(-)

diff --git a/lib/Cake/Test/Case/Utility/XmlTest.php b/lib/Cake/Test/Case/Utility/XmlTest.php
index a2693713a..23517314d 100644
--- a/lib/Cake/Test/Case/Utility/XmlTest.php
+++ b/lib/Cake/Test/Case/Utility/XmlTest.php
@@ -1039,4 +1039,22 @@ XML;
 		$this->assertContains('mark &amp; mark', $result);
 	}
 
+/**
+ * Test that entity loading is disabled by default.
+ *
+ * @return void
+ */
+	public function testNoEntityLoading() {
+		$file = CAKE . 'VERSION.txt';
+		$xml = <<<XML
+<!DOCTYPE cakephp [
+  <!ENTITY payload SYSTEM "file://$file" >]>
+<request>
+  <xxe>&payload;</xxe>
+</request>
+XML;
+		$result = Xml::build($xml);
+		$this->assertEquals('', (string)$result->xxe);
+	}
+
 }
diff --git a/lib/Cake/Utility/Xml.php b/lib/Cake/Utility/Xml.php
index eb5c6d0f8..f8662b282 100644
--- a/lib/Cake/Utility/Xml.php
+++ b/lib/Cake/Utility/Xml.php
@@ -74,6 +74,8 @@ class Xml {
  * ### Options
  *
  * - `return` Can be 'simplexml' to return object of SimpleXMLElement or 'domdocument' to return DOMDocument.
+ * - `loadEntities` Defaults to false.  Set to true to enable loading of `<!ENTITY` definitions.  This
+ *   is disabled by default for security reasons.
  * - If using array as input, you can pass `options` from Xml::fromArray.
  *
  * @param string|array $input XML string, a path to a file, an URL or an array
@@ -86,32 +88,50 @@ class Xml {
 			$options = array('return' => (string)$options);
 		}
 		$defaults = array(
-			'return' => 'simplexml'
+			'return' => 'simplexml',
+			'loadEntities' => false,
 		);
 		$options = array_merge($defaults, $options);
 
 		if (is_array($input) || is_object($input)) {
 			return self::fromArray((array)$input, $options);
 		} elseif (strpos($input, '<') !== false) {
-			if ($options['return'] === 'simplexml' || $options['return'] === 'simplexmlelement') {
-				return new SimpleXMLElement($input, LIBXML_NOCDATA);
-			}
-			$dom = new DOMDocument();
-			$dom->loadXML($input);
-			return $dom;
+			return self::_loadXml($input, $options);
 		} elseif (file_exists($input) || strpos($input, 'http://') === 0 || strpos($input, 'https://') === 0) {
-			if ($options['return'] === 'simplexml' || $options['return'] === 'simplexmlelement') {
-				return new SimpleXMLElement($input, LIBXML_NOCDATA, true);
-			}
-			$dom = new DOMDocument();
-			$dom->load($input);
-			return $dom;
+			$input = file_get_contents($input);
+			return self::_loadXml($input, $options);
 		} elseif (!is_string($input)) {
 			throw new XmlException(__d('cake_dev', 'Invalid input.'));
 		}
 		throw new XmlException(__d('cake_dev', 'XML cannot be read.'));
 	}
 
+/**
+ * Parse the input data and create either a SimpleXmlElement object or a DOMDocument.
+ *
+ * @param string $input The input to load.
+ * @param array  $options The options to use. See Xml::build()
+ * @return SimpleXmlElement|DOMDocument.
+ */
+	protected static function _loadXml($input, $options) {
+		$hasDisable = function_exists('libxml_disable_entity_loader');
+		$internalErrors = libxml_use_internal_errors(true);
+		if ($hasDisable && !$options['loadEntities']) {
+			libxml_disable_entity_loader(true);
+		}
+		if ($options['return'] === 'simplexml' || $options['return'] === 'simplexmlelement') {
+			$xml = new SimpleXMLElement($input, LIBXML_NOCDATA);
+		} else {
+			$xml = new DOMDocument();
+			$xml->loadXML($input);
+		}
+		if ($hasDisable && !$options['loadEntities']) {
+			libxml_disable_entity_loader(false);
+		}
+		libxml_use_internal_errors($internalErrors);
+		return $xml;
+	}
+
 /**
  * Transform an array into a SimpleXMLElement
  *