Re: [xsl] Move leading/trailing spaces outside (XSLT 2.0)

[Abel Braaksma <abel.online@xxxxxxxxx>]

Michael Kay wrote:
> > > <xsl:template match="e">
> > >   <xsl:analyze-string select="." regex="^\s*.*?\s*$">
> > >       
> > I think that needs to be
> >     <xsl:analyze-string select="." regex="^(\s*)(.+?)(\s*)$"> 
> > as the posted regular expression lacks parentheses to allow 
> > the later use of regex-group and as the posted regular 
> > expression matches the empty string which is not allowed.
> >     
>
> Quite right, thanks.

Yves,

Please note the subtle differences between the above mentioned solution 
and the one I proposed. Both have their pros and cons. Depending on your 
final needs, one or the other may suit you best. Using

<xsl:analyze-string select="." regex="^(\s*)(.+?)(\s*)$">

will match the total string, so no need for using 
non-matching-substring. This has the advantage that you can do 
string-manipulation on the whole match at once, using regex-group(1), 
(2) and (3). If you only do string-replacement, your can optimize this 
with the replace() function.

Doing

<xsl:analyze-string select="." regex="^\s+|\s+$">

is quicker, has the advantage of separating logic, but has the 
disadvantage that you cannot intermix the matching and non-matching 
strings (meaning, when you are in non-matching-substring, it contains 
the non-space content of your match, and you cannot get the 
matching-substring from that instruction and vice versa).

If you need to create a node where the match is part of a childnode of 
the non-match, of vv, use the first solution with regex-group, if the 
nodes are siblings, you can use either (whichever you find clearest), if 
the nodes are only text-nodes, you can use replace() instead, depending 
a bit on the complexity of the match. If you mix text with nodes (mixed 
content), it depends a bit (but you cannot use replace()).

-- Abel