Re: SHACL constraint on set of URI (sub)patterns

Here is a solution with a slightly different namespace but the same 
rules. Turtle file attached, TBC screenshot below.

In this solution I have attached a SPARQL query to a shape that gets 
executed over the class itself, which then becomes $this. This has the 
advantage that the inner SELECT to compute the number of instances only 
needs to be made once. The alternative to attach it via sh:targetClass 
would have caused repeated computation.

(Just noticed that rdf:type is not really the predicate of the triple in 
the validation report, so you may want to just remove the ?path result 
from the SELECT).

HTH
Holger

On 18/06/2019 16:31, Felix Sasaki wrote:
> Hi Holger,
>
> the rules applies to subjects that are instance of a given Class, e.g. 
> skos:Concepts.
>
> - Felix
>
> On Tue, 18 Jun 2019 at 07:55, Holger Knublauch <holger@topquadrant.com 
> <mailto:holger@topquadrant.com>> wrote:
>
>     Hi Felix,
>
>     could you clarify one thing please: how do you define "subject" in
>     these graphs. Does the rule apply to all subjects in the whole
>     data graph, including for example owl:Ontology resources or
>     classes, or is there a relation that you only want to apply to
>     sh:targetSubjectsOf <p> ?
>
>     Holger
>
>
>     On 18/06/2019 3:31 pm, Felix Sasaki wrote:
>>     Hi all,
>>
>>     I have a dataset with URIs in the subject position of triples as
>>     follows:
>>
>>     <http://example.com/mydata/1> p o
>>     <http://example.com/mydata/2> p o
>>     <http://example.com/mydata/3> p o
>>     ...
>>
>>     I would like to express the following constraints on all subject
>>     URIs:
>>
>>     1) The value of the URI pattern after
>>     "<http://example.com/mydata/" must always be an integer (in the
>>     example 1, 2 or 3
>>     2) The integer must be more than zero and less than the number of
>>     the subject URIs. That is, the third triple from the below would
>>     be invald:
>>     <http://example.com/mydata/1> p o
>>     <http://example.com/mydata/2> p o
>>     <http://example.com/mydata/4> p o # this should  have been 3
>>
>>     I tried to express this via  SPARQL constraint, but did not
>>     succeed. Any hints on how to  express via SHACL would be nice -
>>     thanks.
>>
>>     As some background for the use case: I am generating the URIs in
>>     an automated fashion and want to assure that the generation
>>     always follows the above constraints, to assure that after the
>>     generation no existing URI is "lost" and to assure that new URis
>>     follow the pattern described above.
>>
>>     Regards,
>>
>>     Felix
>