Hi,

I've been given an XML file that I need to do some work on. I have very little control over the structure of the XML file (it was generated by Adobe FrameMaker).

The XML file uses Processing Instructions to flag certain aspects of the content (like marking it as conditional; it's a FrameMaker thing).

I'm trying to find the best way to find content that falls within a pair of Processing Instructions. Since they are not really start/end tag, I want to find a <? condstart ?> tag, then locate it's paired <? condend ?> tag and throw away everything in between.

To do this I need to know if preceding-sibling and following-sibling process nodes in document order, or reverse document order (for preceding-sibling).


------------------------
GnuPG Key fingerprint = 1AD4 726D E359 A31D 05BF&nbsp;&nbsp;ACE5 CA93 7AD5 D8E3 A876

Michael Hare

__________________
------------------------
Keep Moving Forward

GnuPG Key fingerprint = 1AD4 726D E359 A31D 05BF ACE5 CA93 7AD5 D8E3 A876

Michael Hare

Well, I may be hosed. Once I find a PI, is it possible to parse the text within the PI?

I have <?Fm Condition AccessControlLists Teal%20(med) STRIKETHROUGH show?>, I detected it with <xsl:template match="processing-instruction()">. But can I parse the PI to find it's content? Specifically, is it 'STRIKETHROUGH', is it 'show'? What is the Type ('Condition')? What is the Name ('AccessControlList')

Are these detectable in XPATH/XSLT?




------------------------
GnuPG Key fingerprint = 1AD4 726D E359 A31D 05BF&nbsp;&nbsp;ACE5 CA93 7AD5 D8E3 A876

Michael Hare

Instructions like xsl:for-each and xsl:apply-templates always process nodes in document order, regardless of the XPath expression that was used to locate the nodes. However, predicates like [1] consider the nodes in axis order, so preceding-sibling::*[1] is the immediately preceding sibling.

Parsing the content of your processing instructions will be much easier if you use XSLT 2.0, which gives you regular expression support. String manipulation can be done in XSLT 1.0, but it's a pain.

Michael Kay
http://www.saxonica.com/
Author, XSLT Programmer's Reference and XPath 2.0 Programmer's Reference

Thanks Michael.

I have both your XPATH 2.0 and XSLT 2.0 books. Looking through them, I'm missing how to access the resulting node text from the for-each or apply-template instruction.

------------------------
GnuPG Key fingerprint = 1AD4 726D E359 A31D 05BF&nbsp;&nbsp;ACE5 CA93 7AD5 D8E3 A876

Michael Hare

You can capture the result in a variable:

<xsl:variable name="t">
  <xsl:apply-templates select-"...."/>
</xsl:variable>

and then process the variable, e.g.

<xsl:value-of select="count($t//xyz)"/>

Michael Kay
http://www.saxonica.com/
Author, XSLT Programmer's Reference and XPath 2.0 Programmer's Reference

Thanks again.
It looks like the regex in XPath 2 is going to be the way to do this.

------------------------
GnuPG Key fingerprint = 1AD4 726D E359 A31D 05BF ACE5 CA93 7AD5 D8E3 A876

Michael Hare