Merge pull request #240 from DerwenAI/oxrdflib-integration

allow configurable rdflib.Store plugins, e.g., Oxrdflib

.pre-commit-config.yaml

@@ -33,7 +33,7 @@ repos:
     rev: v2.0.0
     hooks:
     -   id: codespell # spell-check source code
-        args: ["-L", "wit,fils"] # comma separated list of words to ignore.
+        args: ["-L", "wit,fils,basf"] # comma separated list of words to ignore.
         exclude: ^examples\/|.*\.ipynb
         language: python
         types: [text]

changelog.txt

@@ -1,10 +1,12 @@
 # `kglab` changelog
 
-## 0.5.?
+## 0.5.2
 
 2022-03-??
 
-  * bumping up versions of dependencies to fit Py 3.7+
+  * bump up versions of dependencies to fit Py 3.7+
+  * improve testing based on pytest; kudos @Mec-iS
+  * allow configuration for rdflib.Store plugins
 
 
 ## 0.5.1

docs/depend.md

@@ -12,6 +12,7 @@ include:
 
 - [chocolate](https://github.com/seungjaeryanlee/chocolate)
 - [csvwlib](https://github.com/DerwenAI/csvwlib)
+- [cryptography](https://cryptography.io/)
 - [decorator](https://github.com/micheles/decorator)
 - [fsspec](https://filesystem-spec.readthedocs.io/)[^1]
 - [gcsfs](https://gcsfs.readthedocs.io/)
@@ -20,6 +21,7 @@ include:
 - [NetworkX](https://networkx.org/)
 - [NumPy](https://numpy.org/)[^2]
 - [OWL-RL](https://owl-rl.readthedocs.io/)
+- [Oxrdflib](https://github.com/oxigraph/oxrdflib)
 - [pandas](https://pandas.pydata.org/)
 - [pslpython](https://psl.linqs.org/)[^3]
 - [pyarrow](https://arrow.apache.org/)

kglab/kglab.py

@@ -1,11 +1,12 @@
-"""
-kglab main class definition.
+#!/usr/bin/env python
+# -*- coding: utf-8 -*-
+# see license https://github.com/DerwenAI/kglab#license-and-copyright
 
- see license https://github.com/DerwenAI/kglab#license-and-copyright
 """
-######################################################################
-## kglab - core classes
+Main class definitions for `kglab`
+"""
 
+## Python standard libraries
 import codecs
 import datetime
 import io
@@ -14,8 +15,7 @@
 import traceback
 import typing
 
-### third-parties bindings
-
+### third-parties libraries
 from icecream import ic  # type: ignore  # pylint: disable=E0401
 import chocolate  # type: ignore  # pylint: disable=E0401
 import csvwlib  # type: ignore  # pylint: disable=E0401
@@ -30,20 +30,16 @@
 import rdflib  # type: ignore  # pylint: disable=E0401
 import rdflib.plugin  # type: ignore  # pylint: disable=E0401
 import rdflib.plugins.parsers.notation3 as rdf_n3  # type: ignore  # pylint: disable=E0401
-#rdflib.plugin.register("json-ld", rdflib.plugin.Parser, "rdflib_jsonld.parser", "JsonLDParser")
-#rdflib.plugin.register("json-ld", rdflib.plugin.Serializer, "rdflib_jsonld.serializer", "JsonLDSerializer")
 
-
-######################################################################
 ## kglab - core classes
-
 from kglab.decorators import multifile
 from kglab.pkg_types import PathLike, IOPathLike, GraphLike, RDF_Node
 from kglab.gpviz import GPViz
 from kglab.util import get_gpu_count
 from kglab.version import _check_version
 
 
+## pre-constructor set-up
 _check_version()
 
 if get_gpu_count() > 0:
@@ -75,7 +71,7 @@ class KnowledgeGraph:
         "sh":     "http://www.w3.org/ns/shacl#",
         "skos":   "http://www.w3.org/2004/02/skos/core#",
         "xsd":    "http://www.w3.org/2001/XMLSchema#",
-        }
+    }
 
 
     def __init__ (
@@ -84,6 +80,7 @@ def __init__ (
         name: str = "generic",
         base_uri: str = None,
         language: str = "en",
+        store: str = None,
         use_gpus: bool = True,
         import_graph: typing.Optional[GraphLike] = None,
         namespaces: dict = None,
@@ -100,6 +97,9 @@ def __init__ (
     language:
 the default [*language tag*](https://www.w3.org/TR/rdf11-concepts/#dfn-language-tag), e.g., used for [*language indexing*](https://www.w3.org/TR/json-ld11/#language-indexing)
 
+    store:
+optionally, string representing an `rdflib.Store` plugin to use.
+
     use_gpus:
 optionally, use the NVidia GPU devices with [RAPIDS](https://rapids.ai/) if these libraries have been installed and the devices are available; defaults to `True`
 
@@ -112,19 +112,20 @@ def __init__ (
         self.name = name
         self.base_uri = base_uri
         self.language = language
+        self.store = store
 
         # use NVidia GPU devices if available and the libraries
         # have been installed and the flag is not disabled
-        if use_gpus and get_gpu_count() > 0:
+        if use_gpus is not None and get_gpu_count() > 0:
             self.use_gpus = True
         else:
             self.use_gpus = False
 
         # import relations from another RDF graph, or start from blank
-        if import_graph:
+        if import_graph is not None:
             self._g = import_graph
         else:
-            self._g = rdflib.Graph()
+            self._g = self.build_blank_graph()
 
         # initialize the namespaces
         self._ns: dict = {}
@@ -137,6 +138,20 @@ def __init__ (
                 self.add_ns(prefix, iri)
 
 
+    def build_blank_graph (
+        self,
+        ) -> rdflib.Graph:
+        """
+Build a new `rdflib.Graph` object, based on storage plugin configuration.
+        """
+        if self.store is not None:
+            g = rdflib.Graph(store=self.store)
+        else:
+            g = rdflib.Graph()
+
+        return g
+
+
     def rdf_graph (
         self
         ) -> rdflib.Graph:
@@ -472,7 +487,7 @@ def _get_filename (
             filename = str(path)
         elif isinstance(path, pathlib.Path):
             filename = path.as_posix()
-        elif type(path) == str:
+        elif isinstance(path, str):
             filename = path
         else:
             raise TypeError(f"path variable not recognised {type(path)}")
@@ -1225,7 +1240,7 @@ def validate (
             **chocolate.filter_args(kwargs, pyshacl.validate),
             )
 
-        g = rdflib.Graph()
+        g = self.build_blank_graph()
 
         g.parse(
             data=report_graph_data,

kglab/version.py

@@ -11,7 +11,7 @@
 
 
 MIN_PY_VERSION: typing.Tuple = (3, 7,)
-__version__: str = "0.5.1"
+__version__: str = "0.5.2"
 
 
 def _versify (

requirements.txt

@@ -1,7 +1,7 @@
 aiohttp >= 3.8
 chocolate >= 0.0.2
-cryptography >= 35.0
 csvwlib >= 0.3.2
+cryptography >= 35.0
 decorator >= 5.1
 fsspec[gs,s3] >= 2022.2
 gcsfs >= 2022.2
@@ -10,6 +10,7 @@ morph-kgc >= 1.6
 networkx >= 2.7
 numpy >= 1.22
 owlrl >= 6.0.2
+oxrdflib >= 0.2
 pandas >= 1.4
 pslpython >= 2.2.2
 pyarrow >= 7.0

scripts/oxrdf.py

@@ -0,0 +1,57 @@
+#!/usr/bin/env python
+# -*- coding: utf-8 -*-
+# type: ignore
+
+from os.path import abspath, dirname
+import pathlib
+import sys
+import time
+
+import oxrdflib
+import rdflib
+
+sys.path.insert(0, str(pathlib.Path(dirname(dirname(abspath(__file__))))))
+import kglab
+
+FILENAME = "../ffurf_fforms/dat/fiirm.12.owl"
+
+# get_items_interface:
+QUDT_QUERY9 = """
+PREFIX BASF_EC_RaMPO: <https://ontology.basf.net/ontology/BASF_EC_RaMPO/>
+PREFIX owl: <http://www.w3.org/2002/07/owl#>
+PREFIX rdfs: <http://www.w3.org/2000/01/rdf-schema#>
+PREFIX rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#>
+
+SELECT DISTINCT ?item WHERE { ?item rdfs:subClassOf ?item_class . FILTER(?item != owl:Nothing) . FILTER(?item != ?item_class) . FILTER(?item != BASF_EC_RaMPO:CoatingLayer) . FILTER(?interfaceClass != BASF_EC_RaMPO:CoatingLayer) . ?item rdfs:subClassOf ?restriction1 . ?restriction1 rdf:type owl:Restriction . ?restriction1 owl:onProperty BASF_EC_RaMPO:basedOn . ?restriction1 owl:onClass ?basedOn . { ?restriction1 owl:onClass ?basedOn . ?restriction1 owl:onClass ?base_layer . ?item rdfs:subClassOf ?restriction2 . ?restriction2 rdf:type owl:Restriction . ?restriction2 owl:onProperty BASF_EC_RaMPO:interactsWith . ?restriction2 owl:onClass ?interfaceClass } UNION { ?restriction1 owl:onClass ?basedOn . ?basedOn owl:unionOf/rdf:rest*/rdf:first ?base_layer . ?item rdfs:subClassOf ?restriction2 . ?restriction2 rdf:type owl:Restriction . ?restriction2 owl:onProperty BASF_EC_RaMPO:interactsWith . ?restriction2 owl:onClass ?interactsWith . ?interactsWith owl:unionOf/rdf:rest*/rdf:first ?interactsWithUnion . ?interactsWithUnion rdfs:subClassOf ?interfaceRestriction . ?interfaceRestriction rdf:type owl:Restriction . ?interfaceRestriction owl:onProperty BASF_EC_RaMPO:hasEquivalentCoatingLayer . ?interfaceRestriction owl:someValuesFrom ?interfaceClass . FILTER (?interfaceClass != ?base_layer) } OPTIONAL { ?item rdfs:label ?label } }
+"""
+
+
+def run_query (g) -> None:
+    """measure the timing for a SPARQL query"""
+    print(f"using graph: {g}")
+
+    # query init
+    init_time = time.time()
+    query_iter = g.query(QUDT_QUERY9)
+
+    duration = time.time() - init_time
+    print(f"query init time: {duration:10.3f}")
+
+    # query exec
+    init_time = time.time()
+
+    for row in query_iter:
+        print(row.item)
+
+    duration = time.time() - init_time
+    print(f"query exec time: {duration:10.3f}", "\n")
+
+
+if __name__ == "__main__":
+    g = rdflib.Graph(store="OxMemory")
+    g.parse(FILENAME, format="xml")
+    run_query(g)
+
+    g = kglab.KnowledgeGraph(store="OxMemory")
+    g.load_rdf(FILENAME, format="xml")
+    run_query(g)