Xml document validator partly accepts UTF lexical presentation of digit and words

The validator uses the following schema:

<xsd:element name="doc">
    <xsd:complexType>
            <xsd:choice>
                <xsd:element name="elem" type="Regex" minOccurs="1" maxOccurs="unbounded"/>
            </xsd:choice>
    </xsd:complexType>
</xsd:element>

<xsd:complexType name="Regex">
   <xsd:attribute name="att">
       <xsd:simpleType>
           <xsd:restriction base="xsd:string">
               <xsd:pattern value="\d"/>
           </xsd:restriction>
       </xsd:simpleType>
   </xsd:attribute>
</xsd:complexType>

and the xml document:

<doc xmlns:xsi='http://www.w3.org/2001/XMLSchema-instance&#39;
    xsi:noNamespaceSchemaLocation='reS17.xsd' >

<!--
base='string', pattern='\d', value='#x1369;', type='valid', RULE='37'
-->

      <elem att='&#x1369;'/>

</doc>

If value of the att attribute has a lexical presentation of digit starting with 0, the validator accept such xml document as valid.
There is a set of tested UTF symbols which validator accepts
U+0C66 TELUGU DIGIT ZERO
U+0CE6 KANNADA DIGIT ZERO
U+0D66 MALAYALAM DIGIT ZERO
U+0E50 THAI DIGIT ZERO
U+0ED0 LAO DIGIT ZERO
U+0F20 TIBETAN DIGIT ZERO

If the document has following digit presentation:
U+1040 MYANMAR DIGIT ZERO
U+1369 ETHIOPIC DIGIT ONE
U+17E0 KHMER DIGIT ZERO
U+1810 MONGOLIAN DIGIT ZERO
U+FF10 FULLWIDTH DIGIT ZERO
U+1049 MYANMAR DIGIT NINE
U+1371 ETHIOPIC DIGIT NINE
U+17E9 KHMER DIGIT NINE

The validator fails with exception:
SAX error: file:/devel/analysis/reS17.xml(9,29): cvc-pattern-valid: Value '¿' is not facet-valid with respect to pattern '\d' for type '#AnonType_attRegex'.