This doesn't feel like a grouping problem to me. It's just tokenizing, using
a newline character as the token separator. Use any of the usual
tokenization methods:

tokenize() in XPath 2.0
str:tokenize() in EXSLT
tokenize in FXSL
recursive named templates using substring-before/substring-after in XSLT 1.0

Michael Kay
http://www.saxonica.com/ 

> -----Original Message-----
> From: Bruce D'Arcus [mailto:bdarcus@xxxxxxxxxxxxx] 
> Sent: 11 December 2004 15:09
> To: XSL-List@xxxxxxxxxxxxxxxxxxxxxx
> Subject:  cdata, grouping and upconversion
> 
> I'm converting presentation documents out of the (really 
> ugly!) Keynote 
> format into something cleaner using XSLT 2.0.  How I do convert the 
> following into proper paragraphs???
> 
>        <notes font-color="0 0 0" paragraph-head-indent="18" 
> paragraph-alignment="left"
>          tab-stops="L 18 L 28 L 56 L 84 L 112 L 140 L 168 L 
> 196 L 224 L 
> 252 L 280 L 308 L 336"
>          font-size="12" paragraph-line-spacing="0.4" 
> font-name="Times-Roman">
>          <div paragraph-spacing="2.32"><![CDATA[Here's one paragraph 
> stuck in CDATA.
> 
> Here's another; separated by a newline.]]></div>
>        </notes>
> 
> I realize I want to use grouping, but how should I use it here?
> 
> Output should be:
> 
> <notes>
>    <para>Here's one paragraph stuck in CDATA.</para>
>    <para>Here's another; separated by a newline.</para>
> </notes>
> 
> Bruce