Merge pull request #15 from EMMC-ASBL/14-persistent-rdflib-storage

Added support for simple persistent storage in the rdflib backend

tests/test_triplestore.py

@@ -97,6 +97,40 @@ def cost2(parameter):
     assert len(store.function_repo) == 4
 
 
+def test_backend_rdflib_base_iri(
+    get_ontology_path: "Callable[[str], Path]", tmp_path: "Path"
+) -> None:
+    """Test rdflib with `base_iri`.
+
+    Parameters:
+        get_ontology_path: Fixture from `conftest.py` to retrieve a `pathlib.Path`
+            object pointing to an ontology test file.
+        tmp_path: Built-in pytest fixture, which returns a `pathlib.Path` object
+            representing a temporary folder.
+
+    """
+    import shutil
+
+    from tripper.triplestore import RDF, Triplestore
+
+    ontopath_family = get_ontology_path("family")
+    tmp_onto = tmp_path / "family.ttl"
+    shutil.copy(ontopath_family, tmp_onto)
+
+    store = Triplestore(backend="rdflib", base_iri=f"file://{tmp_onto}")
+    FAM = store.bind(  # pylint: disable=invalid-name
+        "fam", "http://onto-ns.com/ontologies/examples/family#"
+    )
+    store.add_triples(
+        [
+            (":Nils", RDF.type, FAM.Father),
+            (":Anna", RDF.type, FAM.Dauther),
+            (":Nils", FAM.hasChild, ":Anna"),
+        ]
+    )
+    store.close()
+
+
 def test_backend_ontopy(get_ontology_path: "Callable[[str], Path]") -> None:
     """Specifically test the ontopy backend Triplestore.
 

tripper/backends/rdflib.py

@@ -10,6 +10,7 @@
 from rdflib import BNode, Graph
 from rdflib import Literal as rdflibLiteral
 from rdflib import URIRef
+from rdflib.util import guess_format
 
 from tripper.triplestore import Literal
 
@@ -38,14 +39,32 @@ def astriple(triple: "Triple"):
 
 
 class RdflibStrategy:
-    """Triplestore strategy for rdflib."""
-
-    def __init__(self, **kwargs) -> None:  # pylint: disable=unused-argument
+    """Triplestore strategy for rdflib.
+
+    Arguments:
+        base_iri: If given, initialise the triplestore from this storage.
+            When `close()` is called, the storage will be overwritten
+            with the current content of the triplestore.
+        triplestore_url: Alternative URL to the underlying ontology if
+            `base_iri` is not resolvable.  Defaults to `base_iri`.
+        format: Format of storage specified with `base_iri`.
+    """
+
+    def __init__(
+        self, base_iri: str = None, triplestore_url: str = None, format: str = None
+    ):  # pylint: disable=redefined-builtin
         self.graph = Graph()
+        self.base_iri = base_iri
+        self.triplestore_url = triplestore_url if triplestore_url else base_iri
+        if self.triplestore_url is not None:
+            if format is None:
+                format = guess_format(self.triplestore_url)
+            self.parse(location=self.triplestore_url, format=format)
+        self.base_format = format
 
     def triples(self, triple: "Triple") -> "Generator":
         """Returns a generator over matching triples."""
-        for (s, p, o,) in self.graph.triples(  # pylint: disable=not-an-iterable
+        for s, p, o in self.graph.triples(  # pylint: disable=not-an-iterable
             astriple(triple)
         ):
             yield (
@@ -66,6 +85,12 @@ def remove(self, triple: "Triple"):
         self.graph.remove(astriple(triple))
 
     # Optional methods
+    def close(self):
+        """Close the internal RDFLib graph."""
+        if self.triplestore_url:
+            self.serialize(destination=self.triplestore_url, format=self.base_format)
+        self.graph.close()
+
     def parse(
         self,
         source=None,

tripper/triplestore.py

@@ -111,7 +111,7 @@ def __init__(
 
         if need_triplestore and triplestore is None:
             url = triplestore_url if triplestore_url else iri
-            triplestore = Triplestore("rdflib", base_iri=iri)
+            triplestore = Triplestore("rdflib", base_iri=iri, triplestore_url=url)
             triplestore.parse(url)
 
         self._cache = {} if cachemode != Namespace.NO_CACHE else None
@@ -378,9 +378,10 @@ def __init__(self, backend: str, base_iri: str = None, **kwargs):
         module = import_module(
             backend if "." in backend else f"tripper.backends.{backend}"
         )
-        cls = getattr(module, backend.title() + "Strategy")
+        cls = getattr(module, f"{backend.title()}Strategy")
         self.base_iri = base_iri
         self.namespaces: "Dict[str, Namespace]" = {}
+        self.closed = False
         self.backend_name = backend
         self.backend = cls(base_iri=base_iri, **kwargs)
         # Keep functions in the triplestore for convienence even though
@@ -405,6 +406,16 @@ def remove(self, triple: "Triple"):
 
     # Methods optionally implemented by backend
     # -----------------------------------------
+    def close(self):
+        """Calls the backend close() method if it is implemented.
+        Otherwise, this method has no effect.
+        """
+        # It should be ok to call close() regardless of whether the backend
+        # implements this method or not.  Hence, don't call _check_method().
+        if not self.closed and hasattr(self.backend, "close"):
+            self.backend.close()
+        self.closed = True
+
     def parse(
         self, source=None, format=None, **kwargs  # pylint: disable=redefined-builtin
     ):
@@ -496,9 +507,9 @@ def add(self, triple: "Triple"):
         """Add `triple` to triplestore."""
         self.add_triples([triple])
 
-    def value(
+    def value(  # pylint: disable=redefined-builtin
         self, subject=None, predicate=None, object=None, default=None, any=False
-    ):  # pylint: disable=redefined-builtin
+    ):
         """Return the value for a pair of two criteria.
 
         Useful if one knows that there may only be one value.