I have an xml file where I need and XPath expression to find all elements that are a copy of a previous element in the same file.

I came up with the following:

*let* *$xml* := <a>
<b>
<c>1</c>
<d>1</d>
<d>2</d>
</b>
<b>
<c>2</c>
<d>1</d>
<d>3</d>
</b>
<b>
<c>3</c>
<d>2</d>
<d>3</d>
</b>
<b>
<c>3</c>
<d>2</d>
<d>3</d>
</b>
</a>
*return*

*$xml*//*[*let**$node1*:=.*return**some**$node2**in**$xml*//*[.<<*$node1*] *satisfies*/deep-equal/(*$node1*,*$node2*)]/(ancestor-or-self::/node/()//concat/(/node-name/(.),1+/count/(preceding-sibling::*)),'&#10;')

which gives

 B a1 b2 d2
 B a1 b3 d2
 B a1 b3 d3
 B a1 b4
 B a1 b4 c1
 B a1 b4 d2
 B a1 b4 d3

This is actually not to bad. However, I would like the correct position of the copy. Instead of preceding-sibling::/* /in the count() function,B I would like to write something like preceding-sibling::/element(node-name(.)). /Unfortunately the processor does not allow that ;-(b&