Re: [xsl] analyze-string gotcha/reminder

[John Lumley <John.Lumley@xxxxxxxxxxxxxx>]

On 18/11/2012 18:42, Andrew Welch wrote:
> Yep, or the usual way is to put the regex in the content of a variable
> with as="xs:string":
>
> <xsl:variable name="regex" as="xs:string"> no need to worry about
> anything here </xsl:variable>
A useful technique I've found with a multi-case xsl:analyze-string is to 
use a set of variables for each of the regex alternatives and then join 
them for a composite alternative expression, viz:

<xsl:variable name="r1">\d+</xsl:variable>
<xsl:variable name="r2">[A-Za-z]+</xsl:variable>
....
<xsl:analyse-string select="." regex="{string-join(($r1,$r2),'|')}">
  <xsl:matching-substring>
    <xsl:choose>
      <xsl:when test="matches(.,$r1)">It's a number</xsl:when>
      <xsl:when test="matches(.,$r2)">A word this time....</xsl:when>
.......
This has three advantages:
i) you only define each case once,
ii) dealing with entities such as quotes is much easier and
iii) it's much clearer what the alternatives are.

The only minor niggle is with embedded bracketed substructures '(...)' 
where the numbering to access them, regex-group(n), depends upon the 
position in the composite sequence.

--
John Lumley MA PhD CEng FIEE
john@xxxxxxxxxxxxxx