My code is in Scala.js, but I think the gist of it should be easy to understand from a JavaScript perspective:

  def htmlToXHTML(input: String)
  (implicit parser: DOMParser, serializer: XMLSerializer): String = {
    val doc = parser.parseFromString(input, "text/html")
    val body = getElementByXpath("/html/body", doc).singleNodeValue
    val bodyXmlString = serializer.serializeToString(body)
    val xmldoc = parser.parseFromString(bodyXmlString, "application/xml")
    val xmlDocElems: NodeList = xmldoc.getElementsByTagName("*")
    xmlDocElems.foreach{
      case elem: Element =>
        elem.removeAttribute("xmlns")
        println(s"Found element $elem with html: ${elem.outerHTML}")
      case node => println(s"Warning: found unexpected non-element node: $node.")
    }
    xmldoc.firstElementChild.innerHTML
  }

This is used above, so including it for pleteness ():

  def getElementByXpath(xpath: String, doc: Document): XPathResult =
    doc.evaluate(
      xpath, doc, null.asInstanceOf[XPathNSResolver],
      XPathResult.FIRST_ORDERED_NODE_TYPE, null
    )

In short, this function reads an HTML string, converts it to an HTML document, serializes to XML, reparses as XML, and finds all the elements in the doc and loops over them (foreach), and then removes the xmlns attribute. It seems that the resulting innerHTML, however, still has the xmlns attributes on elements, even though the first println (aka console.log) indicates we are finding the elements in question, but not removing the xmlns attributes.

The problem may derive from default values specified in a DTD:

If a default value for the attribute is defined in a DTD, a new attribute immediately appears with the default value

My code is in Scala.js, but I think the gist of it should be easy to understand from a JavaScript perspective:

  def htmlToXHTML(input: String)
  (implicit parser: DOMParser, serializer: XMLSerializer): String = {
    val doc = parser.parseFromString(input, "text/html")
    val body = getElementByXpath("/html/body", doc).singleNodeValue
    val bodyXmlString = serializer.serializeToString(body)
    val xmldoc = parser.parseFromString(bodyXmlString, "application/xml")
    val xmlDocElems: NodeList = xmldoc.getElementsByTagName("*")
    xmlDocElems.foreach{
      case elem: Element =>
        elem.removeAttribute("xmlns")
        println(s"Found element $elem with html: ${elem.outerHTML}")
      case node => println(s"Warning: found unexpected non-element node: $node.")
    }
    xmldoc.firstElementChild.innerHTML
  }

This is used above, so including it for pleteness (https://stackoverflow./a/14284815/3096687):

  def getElementByXpath(xpath: String, doc: Document): XPathResult =
    doc.evaluate(
      xpath, doc, null.asInstanceOf[XPathNSResolver],
      XPathResult.FIRST_ORDERED_NODE_TYPE, null
    )

In short, this function reads an HTML string, converts it to an HTML document, serializes to XML, reparses as XML, and finds all the elements in the doc and loops over them (foreach), and then removes the xmlns attribute. It seems that the resulting innerHTML, however, still has the xmlns attributes on elements, even though the first println (aka console.log) indicates we are finding the elements in question, but not removing the xmlns attributes.

The problem may derive from default values specified in a DTD:

If a default value for the attribute is defined in a DTD, a new attribute immediately appears with the default value

• javascript
• dom
• scala.js

Share Improve this question Follow edited Jan 19, 2018 at 15:26 bbarker asked Jan 19, 2018 at 14:37 bbarkerbbarker 13.2k1111 gold badges4747 silver badges7070 bronze badges 3

• I just tried elem.setAttribute("xmlns", "") instead of remove, but the result is weird: <a0:br xmlns:a0="http://www.w3/1999/xhtml" xmlns="" /> - I guess it doesn't deal well with the lack of namespace. – bbarker Commented Jan 19, 2018 at 14:50
• 1 You misunderstand how the DOM API works. The element's qualified name still includes its namespace URI, even if you remove the xmlns attribute. Think of the xmlns attribute as a cross-check of the namespace URI, rather than the namespace URI itself. – Alohci Commented Jan 19, 2018 at 17:23
• @Alohci, thanks, yeah I guess I understood that at some level, and that this is ultimately more an issue of serialization format (but XMLSerializer does not appear to be configurable, so I was trying to work around that) – bbarker Commented Jan 19, 2018 at 18:26

Add a ment  | 

3 Answers 3

Sorted by: Reset to default 6

As mentioned, the easiest is to remove it from the result string:

xmls.serializeToString(domNode).replace(/xmlns="[^"]+"/, '')

I probably would cheat and remove the xmlns from the resulting string, as it's a huge pain to make the elements lose their namespace.

If you insist on doing that, you could try building a document from scratch while walking the original DOM -- pedantically copying everything but the namespaces (i.e. using createElementNS with an empty namespace?)

If you only want to remove the xmlns="" from <html>, you can use this specific regexp:

xmls.serializeToString(domNode)
  .replace(/^<html xmlns="[^"]+">/, "<html>");