Overlapping structures

I have an XML document with two overlapping structures.  To get round this,
for one of the structures I use empty "start" and "end" tags as follows:

<a>I said <z.start/>I will watch my ways</a>
<a><x/>and keep my tongue from sin<z.end/></a>

In my XSL I want to test, from any node <z.start/> if there is the
additional empty element <x/> before the next <z.end/> (i.e. if the
imaginary "z" element "contains" x).  I have not found any way I can achieve
this -- any pointers please?

Secondly, if I want to invert the structures, so that the "a" tags become
the imaginary empty tags and the z tags "real" elements, I am currently
cheating to overcome the well-formed constraint as follows (ignoring the x
element above):

<xsl:template match="a">
  <a.start/><xsl:apply-templates/><a.end/>
</xsl:template>

<xsl:template match="z.start">
  <z>!!DELETE_CLOSE_TAG</z>
</xsl:template>

<xsl:template match="z.end">
  <z>!!DELETE_OPEN_TAG</z>
</xsl:template>

thus generating:

<a.start/>I said <z>!!DELETE_CLOSE_TAG</z>I will watch my ways<a.end/>
<a.start/>and keep my tongue from sin<z>!!DELETE_OPEN_TAG</z><a.end/>

and then running a simple Perl script to get rid of the unwanted tags amd
text to result in:

<a.start/>I said <z>I will watch my ways<a.end/>
<a.start/>and keep my tongue from sin</z><a.end/>

However, I would far rather handle this entirely within the XSL stylesheet.
Is there any way I can cheat using CDATA sections to overcome the
well-formed constraint and directly match <z> to <z.start/> and </z> to
<z.end/>?

Thanks,

Stuart

 XSL-List info and archive:  http://www.mulberrytech.com/xsl/xsl-list