Java context parsing sax / stax

I would like to extract all text elements that display directly as a child node for the root node . I took a look at the standard java sax features using DefaultHandler; but that doesn't seem to be the way to go.

The problem is getting the first-level onlynodes, not just extracting the text nodes .

Is there any non-DOM based approach for this? (Note that node names are not known in advance)

[EDIT]

Input example

<root>
   <a>text1</a>
   <b>text2</b>
   <c>text3</c>
   <nested>
       <d>not_text4</d>
       ...
   <nested>
   ...
</root>

Sample output

Map<String, String> map := {
    {a, text1}
    {b, text2}
    {c, text3}
}

It is currently being addressed as a DOM-oriented method. Although there are libraries that offer a subset of xpath expressions for SAX / STAX.