Bug in regex called by validatingreade r

[Colin Reid]

Hey MS, here's an apparent problem in the base class library. I pulled
the email validation pattern
"^([0-9a-zA-Z]([-.\w]*[0-9a-zA-Z])*@(([0-9a-zA-Z])+([-\w]*[0-9a-zA-Z])*\.)+[a-zA-Z]{2,9})"
from http://regexlib.com.

If I validate the email address "te**@someveryl ongemailaddress .com"
against it by just creating a RegEx and calling IsMatch it works fine,
but if I create a schema defining a simple type restricting an
xs:string by the regex pattern, it takes over full minute at 100% cpu
to match. This is true either when validating from the vs.net ide or
from an executable instantiating a validatingreade r. The time to match
seems to be exponential against the length of the text.

I'm no regex expert, so please let me know if this is just a problem
with the expression, and if so what a good email address validating
expression would be. We need full unicode support in the email
addresses.

Here's a doc and some schema to reproduce. Just open both in the ide
and then use the vs.net menu to validate the xml.

<?xml version="1.0" encoding="utf-8" ?>
<email xmlns="http://tempuri.org/XMLSchema1.xsd"
value="te**@som everylongemaila ddress.com" />
<?xml version="1.0" encoding="utf-8" ?>
<xs:schema id="XMLSchema1 "
targetNamespace ="http://tempuri.org/XMLSchema1.xsd"
elementFormDefa ult="qualified"
xmlns="http://tempuri.org/XMLSchema1.xsd"
xmlns:mstns="ht tp://tempuri.org/XMLSchema1.xsd"
xmlns:xs="http://www.w3.org/2001/XMLSchema">
<xs:element name="email">
<xs:complexType >
<xs:attribute name="value">
<xs:simpleTyp e>
<xs:restricti on base="xs:string ">
<xs:pattern value="^([0-9a-zA-Z]([-.\w]*[0-9a-zA-Z])*@(([0-9a-zA-Z])+([-\w]*[0-9a-zA-Z])*\.)+[a-zA-Z]{2,9})"
/>
</xs:restriction>
</xs:simpleType>
</xs:attribute>
</xs:complexType>
</xs:element>
</xs:schema>

[Derek Harmon]

"Colin Reid" <cr***@projux.c om> wrote in message news:10******** *************** **@posting.goog le.com...

If I validate the email address "te**@someveryl ongemailaddress .com"
against it by just creating a RegEx and calling IsMatch it works fine,
but if I create a schema defining a simple type restricting an
xs:string by the regex pattern, it takes over full minute at 100% cpu
to match. : : The time to match seems to be exponential against the length
of the text.

I confirm seeing the same behavior.

I'd expect the validating reader would just use the same Regex,
since as you've stated, Regex::IsMatch( ) handles this expression
just fine. Perhaps something else involved in the context of an
attribute value is responsible (space-separated list, the quote
delimiter character, checking for entity references).

What can be done as a workaround in an application is to just
plant a simple, valid e-mail address expression in the schema
document you're using, and handle the match against Rob
Eberhardt's e-mail address regular expression in the Validation-
EventHandler.

Change the xsd from this,

<xs:pattern value="^([0-9a-zA-Z]([-.\w]*[0-9a-zA-Z])*@(([0-9a-zA-Z])+([-\w]*[0-9a-zA-Z])*\.)+[a-zA-Z]{2,9})$" />

to this,

<xs:pattern value="^[e-mail] $" />

and initialize the XmlValidatingRe ader like so,

// . . .
ValidationEvent Handler veh = new ValidationEvent Handler(
this.Validation CallBack );

this.mValidatin gReader = new XmlValidatingRe ader(
new XmlTextReader( "XMLSchema1.xsd ") );
this.mValidatin gReader.Validat ionEventHandler += veh;

this.mEmailExpr = new Regex( "^([0-9a-zA-Z]([-.\w]*[0-9a-zA-Z])*@(([0-9a-zA-Z])+([-\w]*[0-9a-zA-Z])*\.)+[a-zA-Z]{2,9})$");

this.mDocIsVali d = true;
// . . .

then in the validation callback message that XmlValidatingRe ader will
invoke when it reads an invalid attribute value,

private void ValidationCallB ack(object sender, ValidationEvent Args e)
{
if ( this.mDocIsVali d && e.Message.Start sWith( "The 'value' attribute") )
{
string attr = this.mValidatin gReader.GetAttr ibute( "value");
this.mDocIsVali d = this.mEmailExpr .IsMatch( attr);
}
else
{
this.mDocIsVali d = false;
}
}

Here when the value attribute fails to match [e-mail] it goes on
to use a pre-compiled Regex, mEmailExpr, to validate the e-mail
address. Should this pass validation, the flag mDocIsValid continues
to indicate the document as being valid.

If you're concerned about changing the XML schema document (i.e.,
you need it to validate in other XML environments), you can apply an
XslTransform to it prior to loading it into the XmlSchema you attach
to the XmlValidatingRe ader. This XslTransform would replace Rob
Eberhardt's email address regular expression with the simplified
e-mail address pattern, and the change would only affect the schema
as its seen within the application.
Derek Harmon