abstract syntax for SHACL

In order to tie the surface syntax of ShEx to SHACL, I wrote up an
abstract syntax for SHACL. I tried to use the same types as described
in the SHACL document but I'd like folks to check it for consistency.

Notation
  TypeName := label1:Type1, label2:Set[Type2]
means TypeName has two component addressed by label1 and label2.
label1 is a Type1 (defined elsewhere in the abstract syntax or in
RDF standards) and label2 is a set of things matching Type2.


Schema     := shapes:Set[Shape]
  A Schema is a set of Shapes. (Do we care about schema boundaries?)

Shape      := label:IRI|BNode, scopes:Set[Scope],
              filters:Set[Shape], constraints:Set[Constraint]
  A Shape has a label of some IRI or BNode, and sets of scopes, filter
  shapes and constraints.
  The "in-scope" nodes are the nodes in the data graph which
    1 match any of the scopes.
    2 pass all of the filter Shapes. (Scopes in filter Shapes have no
      effect.)
  Validation returns the set of errors returned from testing each
  constraint against each "focus node" in the in-scope node.

# Scopes
Scope      := ScopeNode|ScopeClass|PropertyScope|InversePropertyScope
  A Scope is either a ScopeNode or a ScopeClass or a PropertyScope.

ScopeNode  := node:IRI|Bnode # later demands told bnodes.
  A ScopeNode matches node in the data graph.

ScopeClass := type:IRI
  A ScopeClass matches any node in the data graph with the triple
  (node rdf:type type).

PropertyScope := predicate:IRI
  A PropertyScope matches any node in the data graph with a triple
  (node predicate _) where "_" is any node.

InversePropertyScope := predicate:IRI
  A PropertyScope matches any node in the data graph with a triple
  (_ predicate node) where "_" is any node.

# Constraints
Constraint := NodeConstraint|PropertyConstraint|
              InversePropertyConstraint|Constraint

  A Constraint is either PropertyConstraint or InversePropertyConstraint
  or a NodeConstraint. Each type of Constraint selects a value node
  and tests it against a set of Paramenters. Parameters take
  arguments, including "numeric" which is the set of SPARQL numeric
  types enumerated in SPARQL Operand Data Types
  <http://www.w3.org/TR/sparql11-query/#operandDataTypes>.

  Exception: three Parameters, MinCount, MaxCount and UniqueLang, are
  evaluated against the set of all value nodes matching a
  PropertyConstraint or InversePropertyConstraint.

NodeConstraint := parms:Set[NCParameter]
  Testing a NodeConstraint against a focus node returns all of the
  errors from testing each NCParameter against the the value node
  where the value node is the focus node.

PropertyConstraint := predicate:IRI, parms:Set[PCParameter]
  Testing a NodeConstraint against a focus node returns all of the errors
  from testing each NCParameter against each value node in the triple
  (focus node, predicate, value node).

InversePropertyConstraint := predicate:IRI, parms:Set[IPCParameter]
  Testing a NodeConstraint against a focus node returns all of the errors
  from testing each NCParameter against each value node in the triple
  (value node, predicate, focus node).

Constraint := @@still fuzzy on what's permitted in a general constraint.
  Extrapolate from Closed?


NP, PC and IPC Parameters are listed in the first table in
<http://w3c.github.io/data-shapes/shacl/#h-constraints>.

# Nary Parameters - evaluate a set of value nodes.
## Cardinality: PC, ICP
MinCount   := ref:numeric
  Testing a MinCount parameter against a set of value nodes returns an
  error if the number of value nodes is less than ref.

MaxCount   := ref:numeric
  Testing a MaxCount parameter against a set of value nodes returns an
  error if the number of value nodes is greater than ref.

UniqueLang := b:boolean
  Testing a UniqueLang parameter against a set of value nodes returns an
  error if two or more of the value nodes are RDFLiterals with the same
  language tag.


# Unary Parameters - evaluate a single value node.
## Logical operators: NC, PC, IPC
And        := shapes:Set[Shape]
  Testing an And parameter against a value node returns the errors from
  testing each of shapes against the value node.

Or         := shapes:Set[Shape]
  Testing an Or parameter against a value node returns an unspecified
  error if testing each of shapes against the value node returned an
  error.
  
Not        := shape:Shape
  Testing an Or parameter against a value node returns an unspecified
  error if testing shape against the value node returned no
  error.

## RDF term type of value node: NC, PC, IPC
NodeKind   := kind:"IRI"|"blank node"|"literal"
  Testing a NodeKind parameter against a value node returns an error if
    • kind = "IRI" and value node is not an IRI
    • kind = "blank node" and value node is not an blank node
    • kind = "literal" and value node is not an RDFLiteral

## Datatype: NC, PC
Datatype     := dt:IRI
  Testing a Datatype parameter against dt returns an error if the
  value node is not an RDFLiteral or of the datatype of the value node
  is not the same RDF term as dt.

DatatypeIn   := dts:Set[IRI]
  Testing a Datatype parameter against dt returns an error if the
  value node is not an RDFLiteral or of the datatype of the value node
  is not the set dts.

## XML Schema facets:
### XML Schema string facets: NC, PC, ICP
MinLength:   := ref:numeric
  Testing a MinLength paramenter against ref returns an error if the
  lexical form of the value node is longer than ref.

MaxLength    := ref:numeric
  Testing a MinLength paramenter against ref returns an error if the
  lexical form of the value node is shorter than ref.

Pattern      := pat:RDFLiteral
  Testing a Pattern paramenter against pat returns an error if evaluating
  the SPARQL regex function with the lexical form of value node and pat
  is false or results in a type error.

### XML Schema numeric facets: NC, PC
MaxExclusive := ref:RDFLiteral
  Testing a MaxExclusive parameter against ref returns an error if the
  evaluation of (value node > ref) in SPARQL1.1 Operator Mapping returns
  false or results in a type error.

MaxInclusive := ref:RDFLiteral
  Testing a MaxExclusive parameter against ref returns an error if the
  evaluation of (value node >= ref) in SPARQL1.1 Operator Mapping returns
  false or results in a type error.

MinExclusive := ref:RDFLiteral
  Testing a MaxExclusive parameter against ref returns an error if the
  evaluation of (value node > ref) in SPARQL1.1 Operator Mapping returns
  false or results in a type error.

MinInclusive := ref:RDFLiteral
  Testing a MaxExclusive parameter against ref returns an error if the
  evaluation of (value node >= ref) in SPARQL1.1 Operator Mapping returns
  false or results in a type error.

@@ more Parameters to go, but you get the idea.

-- 
-ericP

office: +1.617.599.3509
mobile: +33.6.80.80.35.59

(eric@w3.org)
Feel free to forward this message to any list for any purpose other than
email address distribution.

There are subtle nuances encoded in font variation and clever layout
which can only be seen by printing this message on high-clay paper.

Received on Wednesday, 29 June 2016 15:30:18 UTC