Test for empty nodes, including only whitespace

mphare · December 17th, 2008, 10:59 AM

I am using Saxon9 on Windows

Here's my problem

I am given an XML document that is formatted such as:

Code:

<A>
  <B>
    <C>xyzzy</C>
  <D>
  ...
  </D>
  </B>
  <B>
    <C> </C>
  <D>
   ...
   </D>
   </B>
  <B>
    <C></C>
  <D>
   ...
   </D>
   </B>
  <B>
    <C>abc 123</C>
  </B>
</A>

My job is to translate this to another XML format that is basically the same structure, but different tags. No problem.

Later they come back a want me to only translate the 'C' tags if there is meaningful data in them. So the blank one and the one with only a whitespace need to be left out of the result.

So the result would look like:

Code:

<A>
  <B>
    <C>xyzzy</C>
  <D>
   ...
   </D>
   </B>
  <B>
<!-- dropped the C with whitespace content -->
  <D>
    ...
    </D>
    </B>
   <B>
<!-- dropped the C with empty content -->
  <D>
    ...
    </D>
    </B>
   <B>
     <C>abc 123</C>
  </B>
</A>

I've tried various analyze-string regex's, but they are not giving me the result I'm after.

<xsl:analyze-string select="string()" regex="^[\S*]$">

seemed like a good place to start, if there is a non-whitespace character anywhere in the string, then copy it to the result.

Does anyone have any suggestions?

Thanks!

mhkay · December 17th, 2008, 11:24 AM

You just need two template rules like this:

<xsl:template match="C[normalize-space(.)]">
<xsl:copy-of select="."/>
</xsl:template>

<xsl:template match="C"/>

mphare · December 17th, 2008, 03:40 PM

Thanks!

I believe this also solves the other problem I had with ignoring leading and trailing whitespace during a comparison.

What I ended up doing for the problem at hand ended up looking like this

Code:

    <xsl:template match="C[normalize-space(.)]" mode="rqC">
        <xsl:if test="string() != ''">
            <Cmod>
                <xsl:apply-templates/>
            </Cmod>
        </xsl:if>
    </xsl:template>

<!--
==================
-->
    <xsl:template match="C"/>

I found if I didn't eat the 'C' node alone, I would get some undesired content in the result.