On 11/3/2022 8:00 PM, Graydon Saunders wrote:

You can (I think) test if some attribute value is an RFC 4122 UUID by using a regular expression:

let $regexp as xs:string := '^[0-9a-fA-F]{8}-[0-9a-fA-F]{4}-[0-5][0-9a-fA-F]{3}-[089abAB][0-9a-fA-F]{3}-[0-9a-fA-F]{12}$'

for $uuid in //@GUID   let $value as xs:string := $uuid/string()   where not(matches($value,$regexp))   return $value

Is there a better way to do this?

I wondered whether Java has a class or method that could help and tried

declare namespace uuid = "java:java.util.UUID";

declare function local:isUUID($uuid as xs:string) as xs:boolean {   try {     let $uuidJava := uuid:fromString($uuid)     return true()   } catch * {     false()   } };

("123e4567-e89b-12d3-a456-9AC7CBDCEE52", "123e4567-h89b-12d3-a456-9AC7CBDCEE52") ! local:isUUID(.)

Note I haven't checked whether Java implements the RFC you mention.

On Thu, 2022-11-03 at 15:00 -0400, Graydon Saunders wrote:

Hello --

You can (I think) test if some attribute value is an RFC 4122 UUID by using a regular expression:

let $regexp as xs:string := '^[0-9a-fA-F]{8}-[0-9a-fA-F]{4}-[0-5][0-9a-fA-F]{3}-[089abAB][0-9a- fA-F]{3}-[0-9a-fA-F]{12}$'

You could also split this up into strings, maybe using a helper function:

let $hexdigit := '[0-9a-f]', $n := function($n as xs:integer) { $hexdigit || '{' || $n || '}' }, $matchuuid := '^' || $n(8) || '-' || $n(4) || '-' || '[0-5]' || $n($3) || '-' || '[098ab]' || $n{3} || '-' || $n{12}

and use it with the case insensitive "i" flag.

However, this isn't very clear.

Mike Kay gives a solution [1],

translate($in, '0123456789abcdefABCDEF', '000000000000000000000000') = '00000000-0000-0000-0000-000000000000'

which is probably enough in practice.

If you need to check further, see RFC 4122 for the components. Your regexp limits the UUID value space https://datatracker.ietf.org/doc/html/rfc4122#page-5

[1] https://p2p.wrox.com/xslt/68464-validate-guid-uuid.html