RDF Containers (rdf:Alt/Bag/Seq) not supported

We were not sure whether RDF containers should be allowed, or whether
they are even useful. As it turns out, SHACL does not support these
types for constraints such as `sh:in`, thus, we can remove them
entirely.

If lists/sequences, alternatives/enums, sets/bags are needed, they can
be implemented by using the collection type `rdf:List`.

Author: zyzzyxdonta

docs/appendices/ontology.md

@@ -88,7 +88,7 @@ sc:AllowedOuterType a owl:Class ;
     rdfs:label "Allowed Outer Type"@en ;
     rdfs:comment """The set of container types that may be used as outer types for parameters.
                     This includes scalar (single value), ordered lists, and unordered bags."""@en ;
-    owl:oneOf ( sc:Scalar rdf:List rdf:Bag ) ;
+    owl:oneOf ( sc:Scalar rdf:List ) ;
     rdfs:isDefinedBy <https://schema.software-metadata.pub/software-card/2025-01/#> .
 
 # Class Definition with cardinality restrictions
@@ -143,7 +143,7 @@ sc:parameterOuterType a owl:ObjectProperty ;
     rdfs:label "parameter outer type"@en ;
     rdfs:comment """Specifies the container/cardinality type of the parameter value.
                     Determines whether the parameter represents a single value (sc:Scalar),
-                    an ordered collection (rdf:List), or an unordered collection (rdf:Bag)."""@en ;
+                    or a list of values (rdf:List)."""@en ;
     rdfs:domain sc:Parameter ;
     rdfs:range sc:AllowedOuterType ;
     rdfs:isDefinedBy <https://schema.software-metadata.pub/software-card/2025-01/#> .
@@ -171,7 +171,7 @@ sc:parameterDefaultValue a rdf:Property ;
                     parameter becomes required and must be provided in the configuration file.
                     When present, the default value must match the parameter's declared outer
                     and inner types. For scalar parameters, this is a single literal or resource.
-                    For list or bag parameters, this is an rdf:List or rdf:Bag."""@en ;
+                    For list parameters, this is an rdf:List."""@en ;
     rdfs:domain sc:Parameter ;
     rdfs:isDefinedBy <https://schema.software-metadata.pub/software-card/2025-01/#> .
 
@@ -241,8 +241,8 @@ sc:ParameterShape a sh:NodeShape ;
         sh:description "The container/cardinality type of the parameter."@en ;
         sh:minCount 1 ;
         sh:maxCount 1 ;
-        sh:in ( sc:Scalar rdf:List rdf:Bag ) ;
-        sh:message "Parameter must have exactly one outer type: sc:Scalar, rdf:List, or rdf:Bag."@en ;
+        sh:in ( sc:Scalar rdf:List ) ;
+        sh:message "Parameter must have exactly one outer type: sc:Scalar, or rdf:List."@en ;
     ] ;
 
     # Must have exactly one inner type from allowed set

docs/specification/parameterization.md

@@ -97,8 +97,8 @@ We mention dates as an example here but disallow them later.
 This separation allows the framework to correctly handle configurations like:
 
 - A minimum character requirement for descriptions (outer: `sc:Scalar`, inner: `xsd:int`)
-- A list of license URIs (outer: `rdf:List`, inner: `xsd:anyURI`)
-- A set of organization names (outer: `rdf:Bag`, inner: `xsd:string`)
+- A selection of license URIs (outer: `rdf:List`, inner: `xsd:anyURI`)
+- A list of organization names (outer: `rdf:List`, inner: `xsd:string`)
 
 
 ## `sc:parameterInnerType`
@@ -159,13 +159,13 @@ The outer type MUST be present and MUST specify how values are structured.
 
 These types MUST be supported by implementations:
 
-| Outer Type  | Meaning            | Use When                                     | Example                                    |
-| ----------- | ------------------ | -------------------------------------------- | ------------------------------------------ |
-| `sc:Scalar` | Single value       | Only one value is needed                     | Minimum description length, a boolean flag |
-| `rdf:List`  | Ordered sequence   | Order matters and multiple values are needed | Priority-ordered list of licenses          |
-| `rdf:Bag`   | Unordered sequence | Order does not matter for multiple values    | List of software dependencies              |
+| Outer Type  | Meaning                                       | Use When                   | Examples                                                         |
+| ----------- | --------------------------------------------- | -------------------------- | ---------------------------------------------------------------- |
+| `sc:Scalar` | Single value                                  | Only one value is needed   | Minimum text length, a boolean flag, a single email address, ... |
+| `rdf:List`  | Ordered and unordered sequences, enumerations | Multiple values are needed | A selection of allowed licenses, a list of persons, ...          |
 
-Ohter RDF collection types (like `rdf:Seq`, `rdf:Alt`) MAY be supported by a validation implementation.
+RDF container types (`rdf:Alt`, `rdf:Bag`, `rdf:Seq`) MAY be supported by validation implementations.
+However, these types are not supported by SHACL which means conversion has to be applied.
 
 ## `sc:parameterConfigKey`
 
@@ -323,18 +323,3 @@ scex:requiredKeywords a sc:Parameter ;
     sc:parameterConfigKey "required_keywords" ;
     sc:parameterDefaultValue ( "research-software" "open-source" ) .
 ```
-
-Configuration:
-```toml
-[policies.keywords.parameters]
-required_keywords = ["fair-software", "reproducible-research", "scientific-computing"]
-```
-
-**Bag of Strings (Unordered Collection):**
-```turtle
-scex:allowedLanguages a sc:Parameter ;
-    sc:parameterOuterType rdf:Bag ;
-    sc:parameterInnerType xsd:string ;
-    sc:parameterConfigKey "allowed_programming_languages" ;
-    sc:parameterDefaultValue ( "Python" "Java" "C++" ) .
-```

docs/specification/validation.md

@@ -90,8 +90,7 @@ Implementations MUST validate that configured parameter values match the declare
 
 1. **Outer Type Validation:**
    - `sc:Scalar`: parameters MUST receive a single value (not an array/list)
-   - `rdf:List`: parameters MUST receive an (ordered) array/list of values
-   - `rdf:Bag`: parameters MUST receive an (unorderd) array/list of values
+   - `rdf:List`: parameters MUST receive an array/list of values
 
 2. **Inner Type Validation:**
    - Values MUST be compatible with the declared XSD datatype

src/software_card_policies/data_model.py

@@ -7,6 +7,7 @@
 from enum import Enum
 from functools import reduce
 from pathlib import Path
+from types import NoneType
 from typing import Any, Dict, List, Tuple
 
 from pyshacl import validate
@@ -146,9 +147,6 @@ def from_graph(cls, reference: URIRef, graph: Graph):
 _ALLOWED_OUTER_TYPES = (
     SC.Scalar,
     RDF.List,
-    RDF.Seq,
-    RDF.Bag,
-    RDF.Alt,
 )
 
 _ALLOWED_INNER_TYPES = (
@@ -172,11 +170,10 @@ def read_rdf_resource(source: Path | str) -> Graph:
     return graph
 
 
-def _create_rdf_list_parameter(
+def _create_list_parameter(
     parameter: Parameter, graph: Graph, config_parameter: Any
 ) -> Node:
     assert parameter.outer_type == RDF.List
-    # assert (parameter.default_value, RDF.type, RDF.List) in graph
     assert (parameter.default_value, RDF.first, None) in graph
     assert (parameter.default_value, RDF.rest, None) in graph
 
@@ -190,12 +187,11 @@ def _create_rdf_list_parameter(
     ).uri
 
 
-def _create_sc_scalar_parameter(
+def _create_scalar_parameter(
     parameter: Parameter, graph: Graph, config_parameter: Any
 ) -> Node:
     assert parameter.outer_type == SC.Scalar
-
-    assert isinstance(config_parameter, (str, int, float, type(None)))
+    assert isinstance(config_parameter, (str, int, float, NoneType))
 
     if config_parameter:
         return Literal(config_parameter)
@@ -228,17 +224,10 @@ def parameterize_graph(graph: Graph, config_parameters: Dict[str, Any]) -> Graph
             )
 
         if parameter.outer_type == SC.Scalar:
-            o = _create_sc_scalar_parameter(parameter, graph, config_parameter)
-
-        elif parameter.outer_type == RDF.List:
-            o = _create_rdf_list_parameter(parameter, graph, config_parameter)
+            o = _create_scalar_parameter(parameter, graph, config_parameter)
 
         else:
-            raise NotImplementedError(
-                f"Parameter '{parameter.uri}' has outer type "
-                f"'{parameter.outer_type}', the handling of which "
-                "is currently not implemented"
-            )
+            o = _create_list_parameter(parameter, graph, config_parameter)
 
         # Add replacements for all occurences of the parameter. Again, extract
         # occurences before modifying the graph!