Nokogiri search for word content controls

I want to find all Microsoft Word content control nodes in a document.xml. Because I am going to preprocess these controls, I need to find these guys recursively from a given xml node, BUT I want to stop at the first occurence.

Sample data : (The content control node is the <w:sdt> one)

Assume these are the nodes I can see from a node X :

<w:tc>
  <w:tcPr>
  ...
  </w:tcPr>
  <!-- A content control node I want to get -->
  <w:sdt>
    <w:sdtPr>
    ...
    </w:sdtPr>
    <w:sdtEndPr/>
    <w:sdtContent>
      ...
      <!--- Embedded in a content control. I don't want it now --->
      <w:sdt> 
        ...
      </w:sdt>
    </w:sdtContent>
  </w:sdt>
  ...
  <!-- Another content control ! It isn't embedded in a <w:sdt>, so I want this one -->
  <w:sdt>
  </w:sdt>
</w:tc>
...
<!--- Another one ! Not at the same depth, but not embedded so I want it ! -->
<w:sdt>
...
</w:sdt>

I want 3 out of the 4 nodes given above. If I just do a nodeX.search('.//sdt'), I will grab all 4 right ? How can I exclude the embedded one ?