How to select elements containing special characters using XSL?

I have an ascii-encoded XML-file (in which the various special characters are encoded as &#x..;). Here is a simplified example:

<?xml version="1.0" encoding="ascii"?>
<data>
    <element1>Some regular text</element1>
    <element2>Text containing special characters: 1º-2ª</element2>
    <element3>Again regular text, but with the special charactre prefix: #x</element3>
</data>

Now what I want to do is to pick all the leaf elements containing special characters. The output should look like

The following elements in the input file contain special characters:
<element2>Text containing special characters: 1º-2ª</element2>

I tried with this XSL:

<?xml version="1.0"?>
<xsl:stylesheet xmlns:xsl="http://www.w3.org/1999/XSL/Transform" version="3.0">
    <xsl:output omit-xml-declaration="yes"/>
    <xsl:template match="/">
        <xsl:text>The following elements in the input file contain special characters:
        </xsl:text>
        <xsl:for-each select="//*">
            <xsl:if test="not(*) and contains(., '&#x')">
                <xsl:copy-of select="."></xsl:copy-of>
            </xsl:if>
        </xsl:for-each>
    </xsl:template>
</xsl:stylesheet>

But it only gives me:

The following elements in the input file contain special characters:

If I try to search for just "#x" with this XSL:

<?xml version="1.0"?>
<xsl:stylesheet xmlns:xsl="http://www.w3.org/1999/XSL/Transform" version="3.0">
    <xsl:output omit-xml-declaration="yes"/>
    <xsl:template match="/">
        <xsl:text>The following elements in the input file contain special characters:
        </xsl:text>
        <xsl:for-each select="//*">
            <xsl:if test="not(*) and contains(., '#x')">
                <xsl:copy-of select="."></xsl:copy-of>
            </xsl:if>
        </xsl:for-each>
    </xsl:template>
</xsl:stylesheet>

I get:

The following elements in the input file contain special characters:
        <element3>Again regular text, but with the special character prefix: #x</element3>

So the question is: is there any way to find those elements which contain special characters encoded as "&#x..;"?

I know I can do this with grep etc:

grep '&#x' simpletest.xml
    <element2>Text containing special characters: 1º-2ª</element2>

but the ultimate goal is to generate a pretty output with information about parent elements etc that can be sent as email notification, and using XSLT would make that part so much easier.

In XSLT/XPath you can't know whether any Unicode character was literally in the input document or as a character reference but in XSLT 2 or 3 you can certainly check with matches and Unicode ranges whether certain characters occur (e.g. with \P{IsBasicLatin} for anything not ASCII/Latin):

<xsl:stylesheet xmlns:xsl="http://www.w3.org/1999/XSL/Transform" version="3.0">
    <xsl:output omit-xml-declaration="yes"/>
    <xsl:template match="/">
        <xsl:text>The following elements in the input file contain special characters:
        </xsl:text>
        <xsl:for-each select="//*[not(*) and matches(., '\P{IsBasicLatin}')]">
            <xsl:copy-of select="."></xsl:copy-of>
        </xsl:for-each>
    </xsl:template>
</xsl:stylesheet>

Output:

The following elements in the input file contain special characters:
    <element2>Text containing special characters: 1º-2ª</element2>

As Martin said, character entity references like ª are resolved by XML parsers so when the XML is passed to your XSLT they will have already been converted to regular Unicode characters, with no sign that they were encoded specially.

If you want to find characters which are "special" in some way (i.e. Unicode characters with particular code points), then Martin's solution using regular expressions is what you want. That will find those characters, irrespective of whether they were encoded with character entity references or not.

However, if you are actually trying to find character entity references, then your XSLT would need to read the XML file as plain text (without parsing it as XML), e.g. using the unparsed-text XPath function. Note though, that if you do that, then you won't be able to see which particular XML elements contains the characters, because the XML element markup will also not have been parsed.