Issue 414: list constraints

[jeswr]

Closes #414. This PR implements sh:minListLength, sh:maxListLength and sh:uniqList.

• See this document rendered online here

---

My interpretation of the following from the definition of 2.1.1 Constraints, Parameters and Constraint Components is that because the sh:MemberShapeConstraintComponent now has multiple parameters; there can only be one occurrence of sh:memberShape in a given shape - and I have updated the sh:maxCount accordingly.

Some constraint components declare only a single parameter. For example sh:ClassConstraintComponent has the single parameter sh:class. These parameters may be used multiple times in the same shape, and each value of such a parameter declares an individual constraint. The interpretation of such declarations is conjunction, i.e. all constraints apply. The following example specifies that the values of ex:customer have to be SHACL instances of both ex:Customer and ex:Person.

---

⚠️⚠️ ⚠️
This is currently targeted to #403 branch for a smaller diff in the GH UI. #403 should be merged and this targeted to gh-pages rather than merging this branch into #403

[HolgerKnublauch]

I think you have linked the new constraint parameters to the same constraint component for memberShape. I believe it would be better to introduce separate constraint components for each of them. It makes the definition more modular but, more importantly, it is perfectly sensible to use for example sh:minListLength without a sh:memberShape constraint.

[HolgerKnublauch]

Note that FAILURE is not the right word here. It should be "validation result". Failure is used when a node shape fails to evaluate due to some low-level problem. Normal constraint violations are validation results.

[HolgerKnublauch]

I still don't like the name uniqList, and would prefer uniqueMembers. The abbreviation uniq is not really necessary for such a short word, and not the list is unique but its members are.

[jeswr]

I think you have linked the new constraint parameters to the same constraint component for memberShape. I believe it would be better to introduce separate constraint components for each of them. It makes the definition more modular but, more importantly, it is perfectly sensible to use for example sh:minListLength without a sh:memberShape constraint.

Makes sense! Following this approach; I expect similarly to sh:memberShape - there should be an error if sh:minListLength is declared and the focus node is not a valid SHACL list. This would mean that if both sh:memberShape and sh:minListLength are declared - and the focus node is not a valid list then both constraints would independently trigger validation results which look like

data-shapes/shacl12-test-suite/tests/core/node/memberShape-001.ttl

Lines 121 to 138 in 536ce96

	sh:result [
	rdf:type sh:ValidationResult ;
	sh:focusNode _:b4 ;
	sh:resultMessage "Value does not conform to Shape ex:IRIListShape. See details for more information." ;
	sh:resultSeverity sh:Violation ;
	sh:sourceConstraintComponent sh:MemberShapeConstraintComponent ;
	sh:sourceShape ex:IRIListShape ;
	sh:value _:b4 ;
	sh:detail [
	rdf:type sh:ValidationResult ;
	sh:focusNode _:b4 ;
	sh:resultMessage "Value is a malformed SHACL List. _:b4 is missing rdf:rest property." ;
	sh:resultSeverity sh:Violation ;
	sh:sourceConstraintComponent sh:MemberShapeConstraintComponent ;
	sh:sourceShape ex:IRIListShape ;
	sh:value _:b4 ;
	] ;
	] ;

I guess we can expect similar behaviour if there are multiple sh:memberShape declarations in #403

Is this the behaviour you would expect for validation results?

[HolgerKnublauch]

I wouldn't worry too much about malformed lists. Usually people use tools to produce them, including editing TTL source code. So if they are reported twice, this shouldn't be the end of the world.

[jeswr]

I also saw a comment about lists not being valid objects of sh:targetNode but it's now not appearing in the GH UI - was this comment deleted because the statement doesn't hold, or is GH doing weird things?

[HolgerKnublauch]

Yeah my fault. I deleted the comment because I understood later what you did.