Hi Greg,



The example on that web page is misguided. While it does effectively 
perform an identity transformation, you can't simply add a template rule 
to customize the processing of attributes. The rule is defined in such a 
way that attributes are *always* copied. You can't override this 
behavior, since templates are never applied to attributes. (The @* in 
the match pattern has no effect, so its presence is just misleading.)



Here's the true identity transform rule:



<xsl:template match="@* | node()">
 <xsl:copy>
   <xsl:apply-templates select="@* | node()"/>
 </xsl:copy>
</xsl:template>

The select="@* | node()" part ensures that templates will be applied to 
attributes to determine how to process them (rather than using 
<xsl:copy-of>).



When the rule gets invoked for an attribute, it copies the attribute, 
and, since attributes can't have attributes or child nodes, the 
xsl:apply-templates doesn't do anything, but it's perfectly fine (not an 
error) to have it there. Thus, a single rule can conveniently apply to 
both attributes and elements (and the other child nodes--text nodes, 
comments, and processing instructions).



Now that I re-read your email, it sounds like you may have already 
gotten this far.



To remove attributes that don't have any non-whitespace characters, just 
add this higher-priority rule to the mix:



<xsl:template match="@*[not(normalize-space(.))]"/>



The normalize-space() function takes a string and removes all leading 
and trailing whitespace characters and then converts contiguous 
sequences of whitespace characters to a single space. For a string that 
has all whitespace characters, it effectively returns the empty string. 
The not() function converts its argument to a boolean and returns the 
complement (true if false, false if true). An empty string converts to 
boolean false, so the predicate (everything inside []) returns true 
whenever the given attribute's value is empty or when it consists only 
of whitespace characters. Finally, the template rule is empty (it does 
nothing). This has the effect of stripping the matching attributes from 
the result.



I hope this helps!



Evan Lenz



P.S. You can find a succinct but comprehensive description of how the 
processing model for template rules works in the free PDF sample chapter 
for my XSLT 1.0 Pocket Reference, linked to on this page: 
http://www.oreilly.com/catalog/xsltpr/







floatingisland@xxxxxxx wrote:
Hi,



 I'm trying to copy everything in an XML file except unpopulated 
attributes or attributes containing only whitespace. So far, I have 
not been successful. Although I did manage to copy a file and delete 
all the attributes using a modified version the style sheet found 
here: 
http://www.abbeyworkshop.com/howto/xslt/copy_xml_document/index.html



 I've tried variations of other examples, but so far no success. As a 
side note, does it make a difference in this case if the attributes 
(empty or populated) have namespaces, beyond including the namespaces 
in the XSLT Style Sheet.



If it makes a difference, I'm using the XSLT parser in PHP5, which I 
believe is the libxml parser. Any assistance you could provide would 
be helpful.



Thanks,



 Greg