feat(validate-data): validate MovingImageRepresentation (DEV-4333)

src/dsp_tools/commands/validate_data/make_data_rdf.py

@@ -34,11 +34,11 @@
 from dsp_tools.commands.validate_data.models.data_rdf import DataRDF
 from dsp_tools.commands.validate_data.models.data_rdf import DateValueRDF
 from dsp_tools.commands.validate_data.models.data_rdf import DecimalValueRDF
-from dsp_tools.commands.validate_data.models.data_rdf import GenericFileValueRDF
 from dsp_tools.commands.validate_data.models.data_rdf import GeonameValueRDF
 from dsp_tools.commands.validate_data.models.data_rdf import IntValueRDF
 from dsp_tools.commands.validate_data.models.data_rdf import LinkValueRDF
 from dsp_tools.commands.validate_data.models.data_rdf import ListValueRDF
+from dsp_tools.commands.validate_data.models.data_rdf import MovingImageFileValueRDF
 from dsp_tools.commands.validate_data.models.data_rdf import RDFTriples
 from dsp_tools.commands.validate_data.models.data_rdf import ResourceRDF
 from dsp_tools.commands.validate_data.models.data_rdf import RichtextRDF
@@ -65,7 +65,9 @@ def make_data_rdf(data_deserialised: DataDeserialised) -> DataRDF:
     all_triples: list[RDFTriples] = []
     for r in data_deserialised.resources:
         all_triples.extend(_transform_one_resource(r))
-    file_values: list[RDFTriples] = [_transform_file_value(x) for x in data_deserialised.file_values]
+    file_values: list[RDFTriples] = [
+        transformed for x in data_deserialised.file_values if (transformed := _transform_file_value(x))
+    ]
     all_triples.extend(file_values)
     return DataRDF(all_triples)
 
@@ -205,19 +207,25 @@ def _transform_uri_value(val: ValueDeserialised, res_iri: URIRef) -> ValueRDF:
     return UriValueRDF(URIRef(val.prop_name), content, res_iri)
 
 
-def _transform_file_value(val: AbstractFileValueDeserialised) -> AbstractFileValueRDF:
-    res_iri = DATA[val.res_id]
+def _transform_file_value(val: AbstractFileValueDeserialised) -> AbstractFileValueRDF | None:
     if isinstance(val, IIIFUriDeserialised):
-        return GenericFileValueRDF(res_iri=res_iri, value=Literal(str(val.value)))
+        return None
+    return _map_into_correct_file_value(val)
+
+
+def _map_into_correct_file_value(val: AbstractFileValueDeserialised) -> AbstractFileValueRDF | None:
+    res_iri = DATA[val.res_id]
+    file_literal = Literal(val.value)
     file_extension = _get_file_extension(val.value)
-    return GenericFileValueRDF(res_iri=res_iri, value=Literal(file_extension))
+    match file_extension:
+        case "mp4":
+            return MovingImageFileValueRDF(res_iri=res_iri, value=file_literal)
+        case _:
+            return None
 
 
 def _get_file_extension(value: str | None) -> str:
-    file_extension = "No file path was given."
-    if value:
-        if "." not in value:
-            file_extension = f"This file is missing a valid extension, actual value: {value}"
-        else:
-            file_extension = value.split(".")[-1].lower()
+    file_extension = ""
+    if value and "." in value:
+        file_extension = value.split(".")[-1].lower()
     return file_extension

src/dsp_tools/commands/validate_data/models/data_rdf.py

@@ -224,13 +224,11 @@ def make_graph(self) -> Graph:
 
 
 @dataclass
-class GenericFileValueRDF(AbstractFileValueRDF):
-    """This class is a placeholder for all types of file values that are not yet implemented."""
-
+class MovingImageFileValueRDF(AbstractFileValueRDF):
     def make_graph(self) -> Graph:
         g = Graph()
         val_iri = DATA[str(uuid4())]
-        g.add((val_iri, RDF.type, API_SHAPES.GenericFileValue))
-        g.add((val_iri, API_SHAPES.fileValueHasValue, self.value))
-        g.add((self.res_iri, API_SHAPES.hasGenericFileValue, val_iri))
+        g.add((val_iri, RDF.type, KNORA_API.MovingImageFileValue))
+        g.add((val_iri, KNORA_API.fileValueHasFilename, self.value))
+        g.add((self.res_iri, KNORA_API.hasMovingImageFileValue, val_iri))
         return g

src/dsp_tools/commands/validate_data/models/input_problems.py

@@ -383,12 +383,12 @@ def sort_value(self) -> str:
 
 
 @dataclass
-class MissingFileValueProblem(InputProblem):
+class FileValueProblem(InputProblem):
     expected: str
 
     @property
     def problem(self) -> str:
-        return "Required file value is missing"
+        return self.expected
 
     def get_msg(self) -> str:
         return f"{self.problem}"

src/dsp_tools/commands/validate_data/reformat_validaton_result.py

@@ -11,13 +11,13 @@
 from dsp_tools.commands.validate_data.models.input_problems import AllProblems
 from dsp_tools.commands.validate_data.models.input_problems import ContentRegexProblem
 from dsp_tools.commands.validate_data.models.input_problems import DuplicateValueProblem
+from dsp_tools.commands.validate_data.models.input_problems import FileValueProblem
 from dsp_tools.commands.validate_data.models.input_problems import GenericProblem
 from dsp_tools.commands.validate_data.models.input_problems import InputProblem
 from dsp_tools.commands.validate_data.models.input_problems import LinkedResourceDoesNotExistProblem
 from dsp_tools.commands.validate_data.models.input_problems import LinkTargetTypeMismatchProblem
 from dsp_tools.commands.validate_data.models.input_problems import MaxCardinalityProblem
 from dsp_tools.commands.validate_data.models.input_problems import MinCardinalityProblem
-from dsp_tools.commands.validate_data.models.input_problems import MissingFileValueProblem
 from dsp_tools.commands.validate_data.models.input_problems import NonExistentCardinalityProblem
 from dsp_tools.commands.validate_data.models.input_problems import UnexpectedResults
 from dsp_tools.commands.validate_data.models.input_problems import ValueTypeProblem
@@ -358,9 +358,9 @@ def _reformat_one_validation_result(validation_result: ValidationResult) -> Inpu
 
 def _reformat_min_cardinality_validation_result(validation_result: ResultMinCardinalityViolation) -> InputProblem:
     iris = _reformat_main_iris(validation_result)
-    file_value_properties = ["shapes:hasGenericFileValue"]
+    file_value_properties = ["hasMovingImageFileValue"]
     if iris.prop_name in file_value_properties:
-        return MissingFileValueProblem(
+        return FileValueProblem(
             res_id=iris.res_id,
             res_type=iris.res_type,
             prop_name="bitstream / iiif-uri",

src/dsp_tools/commands/validate_data/sparql/file_value_shacl.py

@@ -14,12 +14,19 @@ def construct_file_value_cardinality(onto: Graph) -> Graph:
     Returns:
         Graph with file cardinalities
     """
+    val_prop_mapper = {"MovingImageRepresentation": "hasMovingImageFileValue"}
+
+    def as_class_type_and_shacl_shape(cls_name: str) -> tuple[str, str]:
+        return f"knora-api:{cls_name}", f"api-shapes:{val_prop_mapper[cls_name]}_PropShape"
+
     g = Graph()
-    g += _construct_generic_file_value_cardinality(onto)
+    for t in val_prop_mapper.keys():
+        representation_type, shacl_shape = as_class_type_and_shacl_shape(t)
+        g += _construct_one_representation_shape(onto, representation_type, shacl_shape)
     return g
 
 
-def _construct_generic_file_value_cardinality(onto: Graph) -> Graph:
+def _construct_one_representation_shape(onto: Graph, representation_type: str, shacl_shape: str) -> Graph:
     query_s = """
     PREFIX owl: <http://www.w3.org/2002/07/owl#> 
     PREFIX sh: <http://www.w3.org/ns/shacl#>
@@ -30,27 +37,12 @@ def _construct_generic_file_value_cardinality(onto: Graph) -> Graph:
     PREFIX dash: <http://datashapes.org/dash#>
 
     CONSTRUCT {
-        ?class sh:property [
-                              a sh:PropertyShape ;
-                              sh:path api-shapes:hasGenericFileValue ;
-                              sh:minCount 1 ;
-                              sh:maxCount 1 ;
-                              sh:severity sh:Violation ;
-                              sh:message "A file is required for this resource" ;
-                            ] .
+        ?class sh:property %(shacl_shape)s .
     } WHERE {
         ?class a owl:Class ;
-               rdfs:subClassOf ?superClass .
-        VALUES ?superClass {
-            knora-api:ArchiveRepresentation
-            knora-api:AudioRepresentation
-            knora-api:DocumentRepresentation
-            knora-api:MovingImageRepresentation
-            knora-api:StillImageRepresentation
-            knora-api:TextRepresentation
-        }
+               rdfs:subClassOf %(representation_type)s .
     }
-    """
+    """ % {"representation_type": representation_type, "shacl_shape": shacl_shape}  # noqa: UP031 (printf-string-formatting)
     if results_graph := onto.query(query_s).graph:
         return results_graph
     return Graph()

src/dsp_tools/commands/validate_data/validate_data.py

@@ -84,6 +84,7 @@ def _inform_about_experimental_feature() -> None:
         "The following information of your data is being validated:",
         "Cardinalities",
         "If the value type used matches the ontology",
+        "Content of the values",
     ]
     cprint(LIST_SEPARATOR.join(what_is_validated), color="magenta", attrs=["bold"])
 
@@ -117,7 +118,6 @@ def _get_all_onto_classes(ontos: Graph) -> tuple[set[str], set[str]]:
     is_value_iri = URIRef(KNORA_API + "isValueClass")
     value_classes = set(ontos.subjects(is_value_iri, Literal(True)))
     value_cls = {reformat_onto_iri(x) for x in value_classes}
-    value_cls.add("shapes:GenericFileValue")
     return res_cls, value_cls
 
 
@@ -144,12 +144,15 @@ def _create_graphs(onto_client: OntologyClient, list_client: ListClient, data_rd
     shapes = construct_shapes_graphs(onto_for_construction, all_lists)
     api_shapes = Graph()
     api_shapes.parse("src/dsp_tools/resources/validate_data/api-shapes.ttl")
+    file_shapes = Graph()
+    file_shapes.parse("src/dsp_tools/resources/validate_data/file_value_cardinalities.ttl")
     content_shapes = shapes.content + api_shapes
+    card_shapes = shapes.cardinality + file_shapes
     data = data_rdf.make_graph()
     return RDFGraphs(
         data=data,
         ontos=ontologies,
-        cardinality_shapes=shapes.cardinality,
+        cardinality_shapes=card_shapes,
         content_shapes=content_shapes,
         knora_api=knora_api,
     )

src/dsp_tools/resources/validate_data/api-shapes.ttl

@@ -254,55 +254,6 @@ api-shapes:UriValue_ClassShape
 #########################################
 
 
-#########################################
-# RESOURCES WITH FILES
-#########################################
-
-###########################
-# ArchiveRepresentation
-###########################
-
-# knora-api:ArchiveFileValue -> knora-api:hasArchiveFileValue
-
-
-###########################
-# AudioRepresentation
-###########################
-
-# knora-api:AudioFileValue -> knora-api:hasAudioFileValue
-
-
-###########################
-# DocumentRepresentation
-###########################
-
-# knora-api:DocumentFileValue -> knora-api:hasDocumentFileValue
-
-
-###########################
-# MovingImageRepresentation
-###########################
-
-# knora-api:MovingImageFileValue -> knora-api:hasMovingImageFileValue
-
-
-###########################
-# StillImageRepresentation
-###########################
-
-# knora-api:StillImageExternalFileValue -> knora-api:fileValueHasExternalUrl
-
-
-# knora-api:StillImageFileValue -> knora-api:hasStillImageFileValue
-
-
-###########################
-# TextRepresentation
-###########################
-
-# knora-api:TextFileValue -> knora-api:hasTextFileValue
-
-
 #########################################
 # DSP BUILT IN RESOURCES
 #########################################

src/dsp_tools/resources/validate_data/file_value_cardinalities.ttl

@@ -0,0 +1,77 @@
+@prefix sh:         <http://www.w3.org/ns/shacl#> .
+@prefix rdfs:       <http://www.w3.org/2000/01/rdf-schema#> .
+@prefix xsd:        <http://www.w3.org/2001/XMLSchema#> .
+@prefix owl:        <http://www.w3.org/2002/07/owl#> .
+@prefix knora-api:  <http://api.knora.org/ontology/knora-api/v2#> .
+
+@prefix api-shapes: <http://api.knora.org/ontology/knora-api/shapes/v2#> .
+
+
+#########################################
+# RESOURCES WITH FILES
+#########################################
+
+###########################
+# ArchiveRepresentation
+###########################
+
+# knora-api:ArchiveFileValue -> knora-api:hasArchiveFileValue
+
+
+###########################
+# AudioRepresentation
+###########################
+
+# knora-api:AudioFileValue -> knora-api:hasAudioFileValue
+
+
+###########################
+# DocumentRepresentation
+###########################
+
+# knora-api:DocumentFileValue -> knora-api:hasDocumentFileValue
+
+
+###########################
+# MovingImageRepresentation
+###########################
+
+api-shapes:hasMovingImageFileValue_PropShape
+  a           sh:PropertyShape ;
+  sh:path     knora-api:hasMovingImageFileValue ;
+  sh:minCount 1 ;
+  sh:maxCount 1 ;
+  sh:message  "A MovingImageRepresentation requires a file with the extension 'mp4'." ;
+  sh:node     api-shapes:MovingImageFileValue_ClassShape .
+
+api-shapes:MovingImageFileValue_ClassShape
+  a           sh:NodeShape ;
+  sh:name     "Validates the class type" ;
+  sh:message  "MovingImageFileValue" ;
+  sh:class    knora-api:MovingImageFileValue ;
+  sh:property [
+                a           sh:PropertyShape ;
+                sh:path     knora-api:fileValueHasFilename ;
+                sh:minCount 1 ;
+                sh:maxCount 1 ;
+                sh:pattern  ".+\\.mp4$" ;
+                sh:severity sh:Violation ;
+                sh:message  "A MovingImageRepresentation requires a file with the extension 'mp4'." ;
+              ] ;
+  sh:severity sh:Violation .
+
+###########################
+# StillImageRepresentation
+###########################
+
+# knora-api:StillImageExternalFileValue -> knora-api:fileValueHasExternalUrl
+
+
+# knora-api:StillImageFileValue -> knora-api:hasStillImageFileValue
+
+
+###########################
+# TextRepresentation
+###########################
+
+# knora-api:TextFileValue -> knora-api:hasTextFileValue