Beginning of manifest validations using SHACL

[Tpt]

Only validates some fields of RDF tests for now

[rubensworks]

This is really useful, thanks for the effort @Tpt.

When validation fails, how will it be reported within the logs?
Does it show the filename of the file that was invalid (with line number)?

[Tpt]

@rubensworks Yes! Here is an example: https://github.com/w3c/rdf-tests/actions/runs/20309967842/job/58338033783?pr=260#step:4:11

[afs]

I tried on all the manifests and they were clean. 🎆

How about putting the shapes in a different place to /ns/ such as /build/ or /.github/shapes/?

[Tpt]

How about putting the shapes in a different place to /ns/ such as /build/ or /.github/shapes/?

@afs We can but I wanted to avoid yet an other directory and /.github/shapes/ is a bit burried for something that is not specific to the CI. What about putting the shapes into a manifest-shape.ttl file in the root directory?

[afs]

If you want, though I was allowing for future build related files. validate_manifest.py could go in there as well.

But I'd suggest not calling the file "manifest-shape.ttl".

Not all manifest files are "manifest.ttl". They don't even all start "manifest", they do all include "manifest" as a substring.

find . -name \*manifest\*.ttl

So maybe:
"validation-shapes.ttl"
"shapes-for-tests.ttl"

[Tpt]

@afs Done. aa30368

I am not sure to see the problem with naming the shape file "manifest". The main RDF node of the files listing tests uses the mf:Manifest class.