Update HTML Purifier to version 4.4.0.

3 files changed, 41 insertions, 7 deletions

diff --git a/lib/htmlpurifier/library/HTMLPurifier/URIFilter/HostBlacklist.php b/lib/htmlpurifier/library/HTMLPurifier/URIFilter/HostBlacklist.php
index 045aa0992..55fde3bf4 100644
--- a/lib/htmlpurifier/library/HTMLPurifier/URIFilter/HostBlacklist.php
+++ b/lib/htmlpurifier/library/HTMLPurifier/URIFilter/HostBlacklist.php
@@ -1,5 +1,9 @@
 <?php
 
+// It's not clear to me whether or not Punycode means that hostnames
+// do not have canonical forms anymore. As far as I can tell, it's
+// not a problem (punycoding should be identity when no Unicode
+// points are involved), but I'm not 100% sure
 class HTMLPurifier_URIFilter_HostBlacklist extends HTMLPurifier_URIFilter
 {
     public $name = 'HostBlacklist';
diff --git a/lib/htmlpurifier/library/HTMLPurifier/URIFilter/Munge.php b/lib/htmlpurifier/library/HTMLPurifier/URIFilter/Munge.php
index efa10a645..de695df14 100644
--- a/lib/htmlpurifier/library/HTMLPurifier/URIFilter/Munge.php
+++ b/lib/htmlpurifier/library/HTMLPurifier/URIFilter/Munge.php
@@ -20,13 +20,8 @@ class HTMLPurifier_URIFilter_Munge extends HTMLPurifier_URIFilter
 
         $scheme_obj = $uri->getSchemeObj($config, $context);
         if (!$scheme_obj) return true; // ignore unknown schemes, maybe another postfilter did it
-        if (is_null($uri->host) || empty($scheme_obj->browsable)) {
-            return true;
-        }
-        // don't redirect if target host is our host
-        if ($uri->host === $config->getDefinition('URI')->host) {
-            return true;
-        }
+        if (!$scheme_obj->browsable) return true; // ignore non-browseable schemes, since we can't munge those in a reasonable way
+        if ($uri->isBenign($config, $context)) return true; // don't redirect if a benign URL
 
         $this->makeReplace($uri, $config, $context);
         $this->replace = array_map('rawurlencode', $this->replace);
diff --git a/lib/htmlpurifier/library/HTMLPurifier/URIFilter/SafeIframe.php b/lib/htmlpurifier/library/HTMLPurifier/URIFilter/SafeIframe.php
new file mode 100644
index 000000000..284bb13de
--- /dev/null
+++ b/lib/htmlpurifier/library/HTMLPurifier/URIFilter/SafeIframe.php
@@ -0,0 +1,35 @@
+<?php
+
+/**
+ * Implements safety checks for safe iframes.
+ *
+ * @warning This filter is *critical* for ensuring that %HTML.SafeIframe
+ * works safely.
+ */
+class HTMLPurifier_URIFilter_SafeIframe extends HTMLPurifier_URIFilter
+{
+    public $name = 'SafeIframe';
+    public $always_load = true;
+    protected $regexp = NULL;
+    // XXX: The not so good bit about how this is all setup now is we
+    // can't check HTML.SafeIframe in the 'prepare' step: we have to
+    // defer till the actual filtering.
+    public function prepare($config) {
+        $this->regexp = $config->get('URI.SafeIframeRegexp');
+        return true;
+    }
+    public function filter(&$uri, $config, $context) {
+        // check if filter not applicable
+        if (!$config->get('HTML.SafeIframe')) return true;
+        // check if the filter should actually trigger
+        if (!$context->get('EmbeddedURI', true)) return true;
+        $token = $context->get('CurrentToken', true);
+        if (!($token && $token->name == 'iframe')) return true;
+        // check if we actually have some whitelists enabled
+        if ($this->regexp === null) return false;
+        // actually check the whitelists
+        return preg_match($this->regexp, $uri->toString());
+    }
+}
+
+// vim: et sw=4 sts=4