how to create a case-insensitive pattern facet in XML Schema?

[FC]

Hello folks,

I'd like to make the following simple type definition case insensitive:


<xsd:simpleType name="MyType">

<xsd:restriction base="xsd:string">

<xsdattern value="(TRUE|FALSE|('([a-zA-Z0-9])+'))
?(NOOVERRIDE)?"/>

</xsd:restriction>

</xsd:simpleType>

As you can easily guess, if I later use an attribute of type MyType in an
instance document:

MyAttribute="TRUE NOOVERRIDE"

will be ok, but

MyAttribute="true NoOverride"

won't validate.

Even if I supply more combinations of the keywords like "True", "true" and
so on, this doesn't solve the problem in an elegant fashion.

Is there a better solution?

Bye,

Flavio

 

[Keith Davies]

Hello folks,

I'd like to make the following simple type definition case insensitive:


<xsd:simpleType name="MyType">

<xsd:restriction base="xsd:string">

<xsdattern value="(TRUE|FALSE|('([a-zA-Z0-9])+'))
?(NOOVERRIDE)?"/>

</xsd:restriction>

</xsd:simpleType>

As you can easily guess, if I later use an attribute of type MyType in an
instance document:

MyAttribute="TRUE NOOVERRIDE"

will be ok, but

MyAttribute="true NoOverride"

won't validate.

Your pattern, if I understand it correctly, allows one of
"TRUE"
"TRUE NOOVERRIDE"
"FALSE"
"FALSE NOOVERRIDE"
"[a-zA-Z0-9]+"
"[a-zA-Z0-9]+ NOOVERRIDE"

*IF* I understand it correctly (I'm not an expert in XSD, but I often
use patterns in other places) "true NOOVERRIDE" will fit last rule, not
the second.


Keith

 

[FC]

Hello folks,

I'd like to make the following simple type definition case insensitive:


<xsd:simpleType name="MyType">

<xsd:restriction base="xsd:string">

<xsdattern value="(TRUE|FALSE|('([a-zA-Z0-9])+'))
?(NOOVERRIDE)?"/>

</xsd:restriction>

</xsd:simpleType>

As you can easily guess, if I later use an attribute of type MyType in an
instance document:

MyAttribute="TRUE NOOVERRIDE"

will be ok, but

MyAttribute="true NoOverride"

won't validate.

Your pattern, if I understand it correctly, allows one of
"TRUE"
"TRUE NOOVERRIDE"
"FALSE"
"FALSE NOOVERRIDE"
"[a-zA-Z0-9]+"
"[a-zA-Z0-9]+ NOOVERRIDE"

*IF* I understand it correctly (I'm not an expert in XSD, but I often
use patterns in other places) "true NOOVERRIDE" will fit last rule, not
the second.


Keith

Well, unfortunately you didn't understand it correctly
The expression in square parentesys is enclosed by tick marks and they must
be present, moreover the meaning of a string inside tick marks is completely
different from a "token" like TRUE from the application standpoint, which
ultimately means that TRUE is not equal to 'true'.
So, back to my original posting, I found later a w3c document explaining why
case-insensitive facets are a bad idea so I restrained myself to using only
uppercase tokens.

Bye,
Flavio

 

[FC]

Hello folks,

I'd like to make the following simple type definition case insensitive:


<xsd:simpleType name="MyType">

<xsd:restriction base="xsd:string">

<xsdattern value="(TRUE|FALSE|('([a-zA-Z0-9])+'))
?(NOOVERRIDE)?"/>

</xsd:restriction>

</xsd:simpleType>

As you can easily guess, if I later use an attribute of type MyType in an
instance document:

MyAttribute="TRUE NOOVERRIDE"

will be ok, but

MyAttribute="true NoOverride"

won't validate.

Your pattern, if I understand it correctly, allows one of
"TRUE"
"TRUE NOOVERRIDE"
"FALSE"
"FALSE NOOVERRIDE"
"[a-zA-Z0-9]+"
"[a-zA-Z0-9]+ NOOVERRIDE"

*IF* I understand it correctly (I'm not an expert in XSD, but I often
use patterns in other places) "true NOOVERRIDE" will fit last rule, not
the second.


Keith

Well, unfortunately you didn't understand it correctly
The expression in square parentesys is enclosed by tick marks and they must
be present, moreover the meaning of a string inside tick marks is completely
different from a "token" like TRUE from the application standpoint, which
ultimately means that TRUE is not equal to 'true'.
So, back to my original posting, I found later a w3c document explaining why
case-insensitive facets are a bad idea so I restrained myself to using only
uppercase tokens.

Bye,
Flavio

As an additional remark, if one wants to create a truly case insensitive
tokens, then the only way is to "spell" each word like this:
([tT][rR][uU][eE])|([fF][aA][lL][sS][eE])... and so on.

Bye,
Flavio

 

[Keith Davies]

Hello folks,

I'd like to make the following simple type definition case insensitive:


<xsd:simpleType name="MyType">

<xsd:restriction base="xsd:string">

<xsdattern value="(TRUE|FALSE|('([a-zA-Z0-9])+'))
?(NOOVERRIDE)?"/>

</xsd:restriction>

</xsd:simpleType>

As you can easily guess, if I later use an attribute of type MyType in an
instance document:

MyAttribute="TRUE NOOVERRIDE"

will be ok, but

MyAttribute="true NoOverride"

won't validate.

Your pattern, if I understand it correctly, allows one of
"TRUE"
"TRUE NOOVERRIDE"
"FALSE"
"FALSE NOOVERRIDE"
"[a-zA-Z0-9]+"
"[a-zA-Z0-9]+ NOOVERRIDE"

*IF* I understand it correctly (I'm not an expert in XSD, but I often
use patterns in other places) "true NOOVERRIDE" will fit last rule, not
the second.


Keith

Well, unfortunately you didn't understand it correctly
The expression in square parentesys is enclosed by tick marks and they must
be present,

However, they are part of a selection. If I read your pattern
correctly, the relevant part is:

TRUE | FALSE | ('([a-zA-Z0-9])+')

Now, I'm more familiar with Perl RE than XSD patterns, but so far as I
understand, this says "TRUE, or FALSE, or some series of alphanumerics
with at least one character". I infer that the '?(NOOVERRIDE)?'
portion indicates that the string "NOOVERRIDE" is optional.

moreover the meaning of a string inside tick marks is completely
different from a "token" like TRUE from the application standpoint, which
ultimately means that TRUE is not equal to 'true'.

Which is what I said, is it not? Well, more or less; I said that it
would match the series of alphanumerics, not the TRUE portion.

So, back to my original posting, I found later a w3c document explaining why
case-insensitive facets are a bad idea so I restrained myself to using only
uppercase tokens.

That's what I'd do. Case insensitivity, IME, generally causes pain.


Keith