XSLT function to validate date in the format of YYYYMMDD

Question

I am looking for suggestions to validate an XML element which holds the date and i need a function in the XSLT to validate whether it is in YYYYMMDD format or not.

Answer 1

Note: The following applies to XSLT version 1.0, which has no date datatype. If your XSLT processor supports a never version, please use one of the built-in date functions.

---

It depends a bit on how far you want to push on "validation".

You could do this:

<xsl:if test="
  string-length(date) = 8 
  and translate(date, '0123456789', '') = ''
">
  <!-- date looks like it could be valid -->
</xsl:if>

You could also make a more thorough check:

<xsl:if test="
  string-length(date) = 8 
  and translate(date, '0123456789', '') = ''
  and number(substring(date, 1, 4)) &gt;= 1970
  and number(substring(date, 5, 2)) &lt;= 12
  and number(substring(date, 7, 2)) &lt;= 31
">
  <!-- date looks like it could really be valid -->
</xsl:if>

However, the latter would still allow 20090231. If you want to rule that out, invocation of an extension function of some sort might become inevitable.

Answer 2

You say that you're using Saxon. If this is the recent version (8.x, 9.x), then it's an XSLT 2.0 processor, and as such, it supports the xsl:analyze-string instruction for parsing strings using regular expressions, and XML Schema primitive datatypes, including xs:date. So you can use regex to split date into components, then convert the result to ISO 8601 date, try to convert it to xs:date for validation of month and day (this should handle leap years etc correctly):

<xsl:variable name="date-string" select="..."/>

...
<xsl:analyze-string select="$date-string" regex="^(\d{4})(\d{2})(\d{2})$">
  <xsl:matching-substring>
    <xsl:variable name="$year" select="xs:integer(regex-group(1))"/>
    <xsl:variable name="$month" select="xs:integer(regex-group(2))"/>
    <xsl:variable name="$day" select="xs:integer(regex-group(3))"/>
    <xsl:variable name="$date-iso" select="concat($year, '-', $month, '-', $day)" />
    <xsl:choose>
      <xsl:when test="$date-iso castable as xs:date">
        <xsl:variable name="$date" select="$date-iso cast as xs:date" />
        <!-- $date now contains an xs:date value, which you can work with using XPath 2.0 date functions -->
        ...
      </xsl:when>
      <xsl:otherwise>
        <!-- $date-string was in YYYYMMDD format, but values for some of components were incorrect (e.g. February 31). -->
        ...
      </xsl:otherwise>
    </xsl:choose>
  </xsl:matching-substring>
  <xsl:non-matching-substring>
    <!-- $date-string wasn't in YYYYMMDD format at all -->
    ...
  </xsl:non-matching-substring>
</xsl:analyze-string>