Initial commit

libs/neo/validator.py

@@ -0,0 +1,70 @@
+"""SHACL validation using pySHACL"""
+
+import asyncio
+from typing import Any
+
+import structlog
+
+logger = structlog.get_logger()
+
+
+# pyright: ignore[reportAttributeAccessIssue]
+class SHACLValidator:  # pylint: disable=too-few-public-methods
+    """SHACL validation using pySHACL"""
+
+    def __init__(self, shapes_file: str) -> None:
+        self.shapes_file = shapes_file
+
+    async def validate_graph(self, rdf_data: str) -> dict[str, Any]:
+        """Validate RDF data against SHACL shapes"""
+
+        def _validate() -> dict[str, Any]:
+            try:
+                # pylint: disable=import-outside-toplevel
+                from pyshacl import validate
+                from rdflib import Graph
+
+                # Load data graph
+                data_graph = Graph()
+                data_graph.parse(data=rdf_data, format="turtle")
+
+                # Load shapes graph
+                shapes_graph = Graph()
+                shapes_graph.parse(self.shapes_file, format="turtle")
+
+                # Run validation
+                conforms, results_graph, results_text = validate(
+                    data_graph=data_graph,
+                    shacl_graph=shapes_graph,
+                    inference="rdfs",
+                    abort_on_first=False,
+                    allow_infos=True,
+                    allow_warnings=True,
+                )
+
+                return {
+                    "conforms": conforms,
+                    "results_text": results_text,
+                    "violations_count": len(
+                        list(
+                            results_graph.subjects() # pyright: ignore[reportAttributeAccessIssue]
+                        )  # fmt: skip # pyright: ignore[reportAttributeAccessIssue]
+                    ),
+                }
+
+            except ImportError:
+                logger.warning("pySHACL not available, skipping validation")
+                return {
+                    "conforms": True,
+                    "results_text": "SHACL validation skipped (pySHACL not installed)",
+                    "violations_count": 0,
+                }
+            except Exception as e:  # pylint: disable=broad-exception-caught
+                logger.error("SHACL validation failed", error=str(e))
+                return {
+                    "conforms": False,
+                    "results_text": f"Validation error: {str(e)}",
+                    "violations_count": -1,
+                }
+
+        return await asyncio.get_event_loop().run_in_executor(None, _validate)