feat: export shacl report summary

cubicweb_rodolf/import_data.py

@@ -21,7 +21,46 @@
 
 from cubicweb import Binary
 from cubicweb_rq.rq import rqjob
-from rdflib import Graph
+from rdflib import Graph, URIRef
+from rdflib.namespace import NamespaceManager
+
+
+def summarize_validation_report(shacl_report_graph):
+    summary_query = """
+PREFIX sh: <http://www.w3.org/ns/shacl#>
+select distinct ?property ?inverseProperty ?severity
+?constraint ?shape ?message
+(count(?x) as ?shcount)
+(sample(?fmessage) as ?message)
+(sample(?fnode) as ?node)
+where{
+  ?x a sh:ValidationResult.
+  ?x sh:resultPath ?property.
+  ?x sh:resultSeverity ?severity.
+  ?x sh:sourceShape ?shape.
+  ?x sh:resultMessage ?fmessage.
+  ?x sh:focusNode ?fnode.
+  OPTIONAL{?x sh:sourceConstraintComponent ?constraint.}
+  OPTIONAL{?property sh:inversePath ?inverseProperty}
+}
+GROUP BY ?property ?severity ?shape ?inverseProperty ?constraint
+    """
+    qres = shacl_report_graph.query(summary_query)
+    summary_report = ""
+    nb_violations = 0
+    g = Graph()
+    nm = NamespaceManager(g, bind_namespaces="rdflib")
+    for row in qres:
+        nb_violations += int(row.shcount)
+        sh_property = (
+            row.property.n3(nm)
+            if isinstance(row.shape, URIRef)
+            else f"^{row.inverseProperty.n3(nm)}"
+        )
+        sh_shape = row.shape.n3(nm) if isinstance(row.shape, URIRef) else ""
+        summary_report += f"{row.severity.n3(nm)} : {row.shcount} {row.message} ({row.constraint.n3(nm)} on {sh_property}) {sh_shape}\n\n"  # noqa
+
+    return f"{nb_violations} SHACL Validation problems detected\n\n" + summary_report
 
 
 def check_rdf_graph(graph, import_procedure):
@@ -32,9 +71,9 @@
             data=file.data.getvalue().decode("utf8"),
             format=file.data_format,
         )
-        conforms, _graph_reports, text_reports = pyshacl.validate(
+        conforms, graph_reports, text_reports = pyshacl.validate(
             graph,
             shacl_graph=shacl_shapes_graph,
         )
         if not conforms:
             everything_ok = False
@@ -36,9 +75,9 @@
             graph,
             shacl_graph=shacl_shapes_graph,
         )
         if not conforms:
             everything_ok = False
-            errors[file.eid] = text_reports
+            errors[file.eid] = summarize_validation_report(graph_reports)
     return everything_ok, errors