Added integrity checks for the cached html

Author: paquettg

src/PHPHtmlParser/Dom/HtmlNode.php

@@ -162,4 +162,15 @@ public function text()
 
 		return '';
 	}
+
+	/**
+	 * Call this when something in the node tree has changed. Like a child has been added
+	 * or a parent has been changed.
+	 */
+    protected function clear()
+    {
+		$this->innerHtml = null;
+		$this->outerHtml = null;
+		$this->text      = null;
+    }
 }

src/PHPHtmlParser/Dom/Node.php

@@ -83,11 +83,14 @@ public function __get($key)
     }
 
 	/**
-	 * @todo Remove the need for this to be called.
+	 * Attempts to clear out any object references.
 	 */
     public function __destruct()
     {
-        $this->clear();
+		$this->tag      = null;
+		$this->attr     = [];
+        $this->parent   = null;
+        $this->children = [];
     }
 
 	/**
@@ -150,6 +153,9 @@ public function setParent(Node $parent)
         // assign child to parent
         $this->parent->addChild($this);
 
+		//clear any cache
+        $this->clear();
+
         return $this;
     }
 
@@ -243,6 +249,9 @@ public function addChild(Node $child)
     	// tell child I am the new parent
     	$child->setParent($this);
 
+		//clear any cache
+        $this->clear();
+
     	return true;
     }
 
@@ -272,6 +281,9 @@ public function removeChild($id)
 		// remove the child
 		unset($this->children[$id]);
 
+		//clear any cache
+        $this->clear();
+
 		return $this;
 	}
 
@@ -351,6 +363,26 @@ public function isAncestor($id)
 		return false;
 	}
 
+	/**
+	 * Attempts to get an ancestor node by the given id.
+	 *
+	 * @param int $id
+	 * @return null|Node
+	 */
+	public function getAncestor($id)
+	{
+		if ( ! is_null($this->parent))
+		{
+			if ($this->parent->id() == $id)
+			{
+				return $this->parent;
+			}
+			return $this->parent->getAncestor($id);
+		}
+
+		return null;
+	}
+
     /**
      * Shortcut to return the first child.
      *
@@ -589,16 +621,9 @@ public function get_display_size()
         return $result;
     }
 
-    /**
-     * clean up memory due to php5 circular references memory leak...
-     *
-     * @todo Remove the need for this. (Remove circular references)
-     */
-    protected function clear()
-    {
-        $this->nodes    = null;
-        $this->parent   = null;
-        $this->children = null;
-    }
-
+	/**
+	 * Call this when something in the node tree has changed. Like a child has been added
+	 * or a parent has been changed.
+	 */
+    protected function clear() {}
 }

src/PHPHtmlParser/Dom/TextNode.php

@@ -19,6 +19,13 @@ class TextNode extends Node {
 	 */
 	protected $text;
 
+	/**
+	 * This is the converted version of the text.
+	 *
+	 * @var string
+	 */
+	protected $convertedText = null;
+
 	/**
 	 * Sets the text for this node.
 	 *
@@ -30,7 +37,7 @@ public function __construct($text)
 		$text = preg_replace('/\s+/', ' ', $text);
 
 		$this->text = $text;
-		$this->tag  = new Tag('text');
+		$this->tag	= new Tag('text');
 		parent::__construct();
 	}
 
@@ -44,14 +51,30 @@ public function text()
 		// convert charset
 		if ( ! is_null($this->encode))
 		{
+			if ( ! is_null($this->convertedText))
+			{
+				// we already know the converted value
+				return $this->convertedText;
+			}
 			$text = $this->encode->convert($this->text);
+			
+			// remember the conversion
+			$this->convertedText = $text;
+
+			return $text;
 		}
 		else
 		{
-			$text = $this->text;
+			return $this->text;
 		}
-
-		return $text;
 	}
 
+	/**
+	 * Call this when something in the node tree has changed. Like a child has been added
+	 * or a parent has been changed.
+	 */
+	protected function clear()
+	{
+		$this->convertedText = null;
+	}
 }