The xsl:analyze-string element is new in XSLT 2.0. It applies a regular expression to a supplied string value. The string is split into a sequence of substrings, each of which is classified as either a matching substring (if it matches the regular expression) or a non-matching substring (if it doesn't). The substrings are then processed individually: the matching substrings by a xsl:matching-substring element that appears as a child of the xsl:analyze-string instruction, the non-matching substrings by a similar xsl:non-matching-substring element. If either of these is omitted, the relevant substrings are not processed.

The element has three attributes: select is an XPath expression whose value is the string to be analyzed; regex is the regular expression (which may be given as an attribute value template), and flags provides one or more Perl-like flags to control the way in which regular expression matching is performed, for example the value "m" indicates multi-line mode.

When processing matching substrings, it is possible to call the regex-group() function to find the parts of the matching substring that matched particular parenthesized groups within the regular expression.

There are examples of this element in the XSLT 2.0 Recommendation.

XSLT 3.0 is proposing to remove the restriction that the regular expression must not be one that matches a zero-length string. Saxon 9.5 still has this restriction.