Re: [xsl] one for the regex gods

["Imsieke, Gerrit, le-tex" <gerrit.imsieke@xxxxxxxxx>]

On 09.07.2013 01:05, Ihe Onwuka wrote:
> I have this regex
>
> <!-- remove commas embedded in amounts-->
> <!-- also replace multiple hyphens with asterisks -->
> <xsl:variable name="line"
> select="replace(replace(.,'-{2,}','**********'),'(,(\d{3})+(\D|$))','$2')"/>

Have you considered using '$2$3' as the outer replacement?

But Ibm wondering whether the regex is really intended as it is written 
here. If you are trying to replace repeating groups of (,(\d{3})) with 
each $2, it should be more like (,(\d{3}))+ instead of (,(\d{3})+b&)

You canbt, however, refer to repeated occurrences of a captured 
substring. $2 will give you only the last, iirc.

If there were lookaheads in XPath 2, it might be feasible for repeated 
,\d{3} groups.

Maybe you should just tokenize or xsl:analyze-string your lines on a 
tab, determine whether a given token represents a number and if it does, 
strip all commas. Then reassemble the line if needed.

Gerrit