1 files changed, 66 insertions, 16 deletions

diff --git a/library/HTMLPurifier/Injector/RemoveEmpty.php b/library/HTMLPurifier/Injector/RemoveEmpty.php
index 638bfca03..cd885722e 100644
--- a/library/HTMLPurifier/Injector/RemoveEmpty.php
+++ b/library/HTMLPurifier/Injector/RemoveEmpty.php
@@ -2,10 +2,44 @@
 
 class HTMLPurifier_Injector_RemoveEmpty extends HTMLPurifier_Injector
 {
+    /**
+     * @type HTMLPurifier_Context
+     */
+    private $context;
 
-    private $context, $config, $attrValidator, $removeNbsp, $removeNbspExceptions;
+    /**
+     * @type HTMLPurifier_Config
+     */
+    private $config;
 
-    public function prepare($config, $context) {
+    /**
+     * @type HTMLPurifier_AttrValidator
+     */
+    private $attrValidator;
+
+    /**
+     * @type bool
+     */
+    private $removeNbsp;
+
+    /**
+     * @type bool
+     */
+    private $removeNbspExceptions;
+
+    /**
+     * @type array
+     * TODO: make me configurable
+     */
+    private $_exclude = array('colgroup' => 1, 'th' => 1, 'td' => 1, 'iframe' => 1);
+
+    /**
+     * @param HTMLPurifier_Config $config
+     * @param HTMLPurifier_Context $context
+     * @return void
+     */
+    public function prepare($config, $context)
+    {
         parent::prepare($config, $context);
         $this->config = $config;
         $this->context = $context;
@@ -14,38 +48,54 @@ class HTMLPurifier_Injector_RemoveEmpty extends HTMLPurifier_Injector
         $this->attrValidator = new HTMLPurifier_AttrValidator();
     }
 
-    public function handleElement(&$token) {
-        if (!$token instanceof HTMLPurifier_Token_Start) return;
+    /**
+     * @param HTMLPurifier_Token $token
+     */
+    public function handleElement(&$token)
+    {
+        if (!$token instanceof HTMLPurifier_Token_Start) {
+            return;
+        }
         $next = false;
-        for ($i = $this->inputIndex + 1, $c = count($this->inputTokens); $i < $c; $i++) {
-            $next = $this->inputTokens[$i];
+        $deleted = 1; // the current tag
+        for ($i = count($this->inputZipper->back) - 1; $i >= 0; $i--, $deleted++) {
+            $next = $this->inputZipper->back[$i];
             if ($next instanceof HTMLPurifier_Token_Text) {
-                if ($next->is_whitespace) continue;
+                if ($next->is_whitespace) {
+                    continue;
+                }
                 if ($this->removeNbsp && !isset($this->removeNbspExceptions[$token->name])) {
                     $plain = str_replace("\xC2\xA0", "", $next->data);
                     $isWsOrNbsp = $plain === '' || ctype_space($plain);
-                    if ($isWsOrNbsp) continue;
+                    if ($isWsOrNbsp) {
+                        continue;
+                    }
                 }
             }
             break;
         }
         if (!$next || ($next instanceof HTMLPurifier_Token_End && $next->name == $token->name)) {
-            if ($token->name == 'colgroup') return;
+            if (isset($this->_exclude[$token->name])) {
+                return;
+            }
             $this->attrValidator->validateToken($token, $this->config, $this->context);
             $token->armor['ValidateAttributes'] = true;
-            if (isset($token->attr['id']) || isset($token->attr['name'])) return;
-            $token = $i - $this->inputIndex + 1;
-            for ($b = $this->inputIndex - 1; $b > 0; $b--) {
-                $prev = $this->inputTokens[$b];
-                if ($prev instanceof HTMLPurifier_Token_Text && $prev->is_whitespace) continue;
+            if (isset($token->attr['id']) || isset($token->attr['name'])) {
+                return;
+            }
+            $token = $deleted + 1;
+            for ($b = 0, $c = count($this->inputZipper->front); $b < $c; $b++) {
+                $prev = $this->inputZipper->front[$b];
+                if ($prev instanceof HTMLPurifier_Token_Text && $prev->is_whitespace) {
+                    continue;
+                }
                 break;
             }
             // This is safe because we removed the token that triggered this.
-            $this->rewind($b - 1);
+            $this->rewindOffset($b+$deleted);
             return;
         }
     }
-
 }
 
 // vim: et sw=4 sts=4