Strip unwanted classNames from all nodes

java/org/chromium/distiller/DomUtil.java

Index: java/org/chromium/distiller/DomUtil.java
diff --git a/java/org/chromium/distiller/DomUtil.java b/java/org/chromium/distiller/DomUtil.java
index 98d95f5e87d7fffaf63622b46c06501c8983af6a..79ed52ff4d6ea72756610464861dd9dc9512dc6e 100644
--- a/java/org/chromium/distiller/DomUtil.java
+++ b/java/org/chromium/distiller/DomUtil.java
@@ -418,6 +418,7 @@ public class DomUtil {
     /**
      * Strips some attribute from certain tags in the tree rooted at |rootNode|, including root.
      * @param tagNames The tag names to be processed. ["*"] means all.
+     * TODO(crbug.com/654108): We should convert to whitelisting for all the tags.
      */
     @SuppressWarnings("unused")
     public static void stripAttributeFromTags(Node rootNode, String attribute, String[] tagNames) {
@@ -474,6 +475,32 @@ public class DomUtil {
     }
 
     /**
+     * Strips unwanted classNames from all nodes in the tree rooted at |root|.
+     * TODO(crbug.com/654109): "caption" is essential for styling, but all classNames should
+     *                         be removed eventually.
+     */
+    public static void stripUnwantedClassNames(Node root) {
+        if (root.getNodeType() == Node.ELEMENT_NODE) {
+            Element element = Element.as(root);
+            if (element.hasAttribute("class")) {
+                stripUnwantedClassName(element);
+            }
+        }
+        NodeList<Element> elems = DomUtil.querySelectorAll(root, "[class]");
+        for (int i = 0; i < elems.getLength(); i++) {
+            stripUnwantedClassName(elems.getItem(i));
+        }
+    }
+
+    private static void stripUnwantedClassName(Element elem) {
+        if (elem.getClassName().contains("caption")) {
+            elem.setClassName("caption");
+        } else {
+            elem.removeAttribute("class");
+        }
+    }
+
+    /**
      * Get a list of relevant nodes from a subtree.
      * @param root The root of the subtree.
      * @return A list of relevant nodes.