Issue when selecting an node with a specific name - xpath

Question

I am trying to select the node Prp[#name='node name'] which has a parent name item20 using the XPath expression //Prp[#name='node name'and ../../../*[#name='item20']] but this works only if my file contains only this part of XML:
<Node name="item20">
<Node name="config">
<Node name="runmodeparams">
<Node name="simple">
<Prp name="filename" type="S" value="p"/>
<Prp name="filepath" type="S" value="r"/>
</Node>
<Prp name="activerunmode" type="S" value="Simple"/>
</Node>
<Prp name="node name" type="S" value="lastversion"/>
</Node>
If it also contains another part of the XML file like the following one, then XPath returns an empty result.
<Node name="item20">
<Node name="config">
<Node name="runmodeparams">
<Node name="simple">
<Prp name="filename" type="S" value="p"/>
<Prp name="filepath" type="S" value="r"/>
</Node>
<Prp name="activerunmode" type="S" value="Simple"/>
</Node>
<Prp name="node name" type="S" value="lastversion"/>
</Node>
</Node>
<Node name="item21">
<Node name="config">
<Node name="runmodeparams">
<Node name="simple">
<Prp name="filename" type="S" value="p"/>
<Prp name="filepath" type="S" value="r"/>
</Node>
<Prp name="activerunmode" type="S" value="Simple"/>
</Node>
<Prp name="node name" type="S" value="lastversion"/>
</Node>
</Node>
How can I properly select the node?

Answer 1

The second XML snippet you gave is no valid XML as it contains two root nodes. If this really is your full XML input, you should
fix it if possible, or somewhat wrap it in a single root node and
try to fetch some error message from your XPath engine.
I wrapped it in another element and your second XPath somewhat worked - but probably didn't return the expected result; both node name elements of item20 and item21 are returned as you're stepping out too far.
Anyway, you'd better check for "item20" in a predicate when stepping down the XML tree:
//Node[#name='item20']//Prp[#name='node name']
This not only limits to the node you're looking for, but also should be faster for most cases.
If performance really matters and the <Prp/> element you're looking for is always at the same position, try to avoid the descendant-or-self-steps // and provide a full distinct path, here it would be
//Node[#name='item20']/Prp[#name='node name']