SHACL validation Roadmap

[angelip2303]

SHACL features based on the SHACL recommendation.

Target selection

Feature	Rust	SPARQL	Comments
Target Node sh:targetNode	✔️	❌
Class-based targets sh:targetClass	✔️	❌
Implicit class targets	✔️	❌
Subjects of targets sh:targetSubjectsOf	✔️	❌
Objects of targets sh:targetObjectsOf	✔️	❌

Shapes

Feature	Supported	Comments
sh:name	❌
sh:description	❌
sh:group	❌
sh:order	❌
owl:imports	❌
sh:entailment	❌
sh:shapesGraph	❌	The sh:shapesGraph declaration is part of the data graph passed as a parameter

Validation report

We generate a SHACL validation report.

Feature	Supported
sh:deactivated	✔️
sh:message	✔️
sh:severity	✔️

Constraints

Feature	Supported
Node Shapes	✔️
Property Shapes	✔️

Paths

Feature	Supported
Predicate paths	✔️
Sequence paths	✔️
Alternative paths	✔️
Inverse paths	✔️
Zero-or-more paths	✔️
One-or-more paths	✔️
Zero-or-one paths	✔️

Constraint components

Feature	Generic	Comments
sh:datatype	✔️	We cannot check ill-typed literals
sh:minCount	✔️
sh:maxCount	✔️
sh:languageIn	✔️
sh:uniqueLang	✔️
sh:equals	✔️
sh:not	✔️
sh:and	✔️
sh:or	✔️
sh:xone	✔️
sh:node	✔️
sh:qualifiedValueshape	✔️
sh:closed	✔️
sh:ignoredProperties	✔️
sh:hasValue	✔️
sh:in	✔️

Feature	Native	SPARQL	Comments
sh:class	✔️	❌
sh:nodeKind	✔️	❌
sh:minInclusive	✔️	❌
sh:maxInclusive	✔️	❌
sh:minExclusive	✔️	❌
sh:maxExclusive	✔️	❌
sh:maxLength	✔️	❌
sh:pattern	✔️	❌
sh:disjoint	✔️	❌
sh:lessThan	✔️	❌
sh:lessThanOrEquals	✔️	❌

[ensaremirerol]

If no one is working on this, I would like to help with it.

[angelip2303]

If no one is working on this, I would like to help with it.

Hi! We are currently working on the new architecture of the tool, as we are moving from SRDF to RDF.rs. Hence, the implementation of the SHACL validation is currently stalled for a bit. However, we are planning to continue with the development of this soon!

[ensaremirerol]

Oh, I see! Looking forward to it.

Just a quick question—what’s the reason behind the move to rdf.rs? On their repo, it mentions that it’s heavily under construction.

Additionally, I’d like to understand its function. From what I can tell by quickly looking at their code, it seems more like an interface to oxrdf and sophia than a standalone package. Please correct me if I’m wrong—this is just my initial impression.

[angelip2303]

The thing its that our core, SRDF, introduces a bit of technical debt, as it was implemented by demand: when a feature was required we implemented it, without a proper design of the whole library. Thus, when we wanted to implement the subsetting capabilities, or some of the SPARQL-related functionalities, we found that the design wasn't maintainable, and we started with the refactoring of the tool, which led to a dead end, as we had to rewrite more than what we could manage.

However, the idea behind the core of the tool (SRDF) is to define a set of traits and to implement rudof's functionalities on top of that, e.g. SHACL validation or RDF parsing, and this is exactly what RDF.rs tries to pursue, but with a better design at its core.

As a whole, we can understand rudof as an implementation agnostic framework for working with RDF, as long as a implementors provide the required functionalities (SRDF). Thus, if you implement the SRDF traits for Oxigraph, then you have all the functionality that is bundled in rudof.

[ensaremirerol]

Thanks for the explanation. Good luck!

(Also, I’ll be keeping an eye out for any issues I can help with. I was really impressed with the speed when I tried it, and I’d love to see this project become stable and fully conformant with the specs!)