DomRecursion.sc

import java.io.ByteArrayInputStream
import javax.xml.parsers.DocumentBuilderFactory
import org.w3c.dom._

val factory = DocumentBuilderFactory.newInstance()
val builder = factory.newDocumentBuilder();
val content = """<start>
                   <div id="header_nav">
                    <div class="avatar">
                      <img width="40" />
                    </div>
                    <a class="logged-in">Name</a>
                      <ul>
                        <li>
                          <a>link</a>
                        </li>
                        <li>
                          <a>link2</a>
                        </li>
                      </ul>
                    <form id="search" />
                   </div>
                 </start>"""

// Convert to bytes
val input = new ByteArrayInputStream(content getBytes())

// Parse into Document
val document = builder.parse(input)

// Helper function to find the Node with the id (header_nav) we want.
def findDivWithId(d:Document, id:String):Option[Node] = {
  val list: NodeList = d.getElementsByTagName("div")
  for(x <- 0 to list.getLength)
  {
    if(list.item(x).hasAttributes){

      val attrs:NamedNodeMap = list.item(x).getAttributes

      for(y <- 0 to attrs.getLength)
      {
        val attr = attrs.item(y).asInstanceOf[Attr];
        if(attr.getName.toLowerCase == "id" && attr.getValue.toLowerCase == id)
        {
          return Some(list.item(x));
        }
      }
    }
  }

  return None;
}

// Recursive Function for counting descendants
def countSubNodes(n:Node):Int = {

  var current = 1

  // The W3C API will give text nodes in their implementation.
  // I'm assuming our theoretical Node class won't return these,
  // So I just don't count them.
  if(n.getNodeName.charAt(0) == '#')
  {
    current = 0;
  }

  if(n.getFirstChild != null){
    current += countSubNodes(n.getFirstChild)
  }

  if(n.getNextSibling != null){
    current += countSubNodes(n.getNextSibling)
  }

  current
}

def countDescendants(n:Node):Int = {

  if(n == null)
  {
    return 0
  }

  // Must have at least one child to begin DOM traversing
  if(n.getFirstChild == null) {
    0
  }
  else {
    countSubNodes(n.getFirstChild);
  }
}

// Find our node
// TODO: Add a null check, I have wrapped this in a Scala Option though,
// so the correct way to call would be .getOrElse(...)
val node = findDivWithId(document, "header_nav").get

// Count 'em up
countDescendants(node)