Win a copy of Design for the Mind this week in the Design forum!
  • Post Reply
  • Bookmark Topic Watch Topic
  • New Topic

XSL question: position() strangeness

 
Ron Newman
Ranch Hand
Posts: 1056
  • Mark post as helpful
  • send pies
  • Quote
  • Report post to moderator
Given the following XML input:
� <test>
� <scan>1</scan>
� <scan>2</scan>
� <scan>3</scan>
� <scan>4</scan>
� <scan>5</scan>
� </test>
and the following XSL stylesheet:
� <xsl:stylesheet xmlns:xsl="http://www.w3.org/1999/XSL/Transform"
� � � � � � � � version="1.0">
� <xslutput method="text"/>
� <xsl:template match="text()"/>
� <xsl:template match="test">
� � �<xsl:apply-templates/>
� </xsl:template>
� <xsl:template match="scan">
� � �content:<xsl:value-of select="."/>, position:<xsl:value-of select="position()"/>
� � �<xsl:if test="position() = last()"> I'M LAST!</xsl:if>
� </xsl:template>
� </xsl:stylesheet>
Why do I get this output?
� �content:1, position:2
� �content:2, position:4
� �content:3, position:6
� �content:4, position:8
� �content:5, position:10
It's acting as if there were other children of <test>, which there aren't.� It also doesn't seem to think the last child is really last.� This happens with both Xalan and Saxon.
If I change
� �<xsl:apply-templates/>
to either
� �<xsl:apply-templates select="*"/>
or
� �<xsl:apply-templates select="scan"/>
I get
� �content:1, position:1
� �content:2, position:2
� �content:3, position:3
� �content:4, position:4
� �content:5, position:5 I'M LAST!
which looks a lot more reasonable to me.
What's going on here?
 
Anonymous
Ranch Hand
Posts: 18944
  • Mark post as helpful
  • send pies
  • Quote
  • Report post to moderator
Hi Ron,
In a match template, the context is the whole input document, so the context position is equal to the numbered position of the node across all nodes (element, attribute, text node, comment, processing instruction)
The list of nodes is as follows:
1. test
2. scan
3. text()
4. scan
5. text()
etc.
If the template is applied without a select, you may use the preceding-sibling and following-sibling axis to achieve this result. The following template, for instance, would give the expected results:

An other way (cleaner) to do this is to use a for-each construct:

Cheers
[ January 11, 2003: Message edited by: Beno�t d'Oncieu ]
 
  • Post Reply
  • Bookmark Topic Watch Topic
  • New Topic