include skos:definition in jsond-ld dump

docs/userguide/wrapper/DumpFile.ipynb

@@ -596,7 +596,7 @@
    "name": "python",
    "nbconvert_exporter": "python",
    "pygments_lexer": "ipython3",
-   "version": "3.8.18"
+   "version": "3.8.19"
   }
  },
  "nbformat": 4,

h5rdmtoolbox/__init__.py

@@ -153,13 +153,13 @@ def dump_jsonld(hdf_filename: Union[str, pathlib.Path],
     if structural and not semantic:
         return jsonld.dump_file(hdf_filename, skipND=skipND)
     with File(hdf_filename) as h5:
-        return jsonld.dumps(h5, indent=2, structural=structural, resolve_keys=resolve_keys, **kwargs)
+        return jsonld.dumps(h5, structural=structural, resolve_keys=resolve_keys, **kwargs)
 
 
 def register_dataset_decoder(decoder: Callable, decoder_name: str = None, overwrite: bool = False):
     """A decoder function takes a xarray.DataArray and a dataset as input and returns a xarray.DataArray
     It is called after the dataset is loaded into memory and before being returned to the user. Be careful:
-    Multiple decoders can be registered and they are called in the order of registration. Hence, your decoder
+    Multiple decoders can be registered, and they are called in the order of registration. Hence, your decoder
     may behave unexpectedly!
     """
     from .wrapper import ds_decoder

h5rdmtoolbox/convention/definition.py

@@ -21,6 +21,8 @@ def __getitem__(self, item) -> Optional[str]:
         return self.get(item)
 
     def __setitem__(self, key, value):
+        if key not in self._attr:
+            raise KeyError(f"Attribute {key} does not exist")
         attr_def = self._attr.get(DEFINITION_ATTR_NAME, {})
         attr_def.update({key: value})
         self._attr[DEFINITION_ATTR_NAME] = attr_def

h5rdmtoolbox/wrapper/core.py

@@ -1,24 +1,25 @@
 """Core wrapper module containing basic wrapper implementation of File, Dataset and Group
 """
 import datetime
-import h5py
 import json
 import logging
-import numpy as np
 import os
 import pathlib
-# noinspection PyUnresolvedReferences
-import pint
 import shutil
 import warnings
-import xarray as xr
 from collections.abc import Iterable
 from datetime import datetime, timezone
-from h5py._hl.base import phil, with_phil
-from h5py._objects import ObjectID
 from pathlib import Path
 from typing import List, Dict, Union, Tuple, Protocol, Optional, Generator
 
+import h5py
+import numpy as np
+# noinspection PyUnresolvedReferences
+import pint
+import xarray as xr
+from h5py._hl.base import phil, with_phil
+from h5py._objects import ObjectID
+
 from h5rdmtoolbox.database import ObjDB
 from h5rdmtoolbox.database.lazy import LHDFObject
 # noinspection PyUnresolvedReferences
@@ -29,8 +30,8 @@
 from .. import _repr, get_config, convention, utils, consts, protected_attributes
 from .. import get_ureg
 from .._repr import H5Repr, H5PY_SPECIAL_ATTRIBUTES
-from ..convention import rdf
 from ..convention import definition
+from ..convention import rdf
 from ..convention.consts import DefaultValue
 
 logger = logging.getLogger('h5rdmtoolbox')
@@ -1674,7 +1675,8 @@ def __setitem__(self, key, value):
             super().__setitem__(key, value)
 
     @dataset_value_decoder
-    def __getitem__(self, args, new_dtype=None, nparray=False, links_as_strings:bool=False) -> Union[xr.DataArray, np.ndarray]:
+    def __getitem__(self, args, new_dtype=None, nparray=False, links_as_strings: bool = False) -> Union[
+        xr.DataArray, np.ndarray]:
         """Return sliced HDF dataset. If global setting `return_xarray`
         is set to True, a `xr.DataArray` is returned, otherwise the default
         behaviour of the h5p-package is used and a np.ndarray is returned.
@@ -2212,6 +2214,21 @@ def open(filename: Union[str, pathlib.Path], mode: str = "r+") -> 'File':
         """
         return File(filename, mode)
 
+    def dump_jsonld(self,
+                    skipND: int = 1,
+                    structural: bool = True,
+                    semantic: bool = True,
+                    resolve_keys: bool = False,
+                    **kwargs):
+        """Dump the file content as JSON-LD"""
+        from .. import dump_jsonld
+        return dump_jsonld(self.hdf_filename,
+                           skipND=skipND,
+                           structural=structural,
+                           semantic=semantic,
+                           resolve_keys=resolve_keys,
+                           **kwargs)
+
 
 Dataset._h5grp = Group
 Dataset._h5ds = Dataset

h5rdmtoolbox/wrapper/jsonld.py

@@ -10,7 +10,7 @@
 import ontolutils
 import rdflib
 from ontolutils.classes.utils import split_URIRef
-from rdflib import Graph, URIRef, Literal, BNode, XSD, RDF
+from rdflib import Graph, URIRef, Literal, BNode, XSD, RDF, SKOS
 from rdflib.plugins.shared.jsonld.context import Context
 
 from h5rdmtoolbox.convention import hdf_ontology
@@ -579,10 +579,6 @@ def _add_hdf_node(name, obj, ctx):
                 # _add_node(g, (root_group, RDF.type, HDF5.Group))
                 # _add_node(g, (root_group, HDF5.name, rdflib.Literal(name)))
 
-            # return
-        # else:
-        #     root_group = rdflib.BNode()
-
         obj_node = iri_dict.get(obj.name, None)
         if obj_node is None:
             obj_node = _get_id(obj, local=local)
@@ -643,7 +639,17 @@ def _add_hdf_node(name, obj, ctx):
 
             if structural:  # add hdf type and name nodes
                 _add_node(g, (attr_node, RDF.type, HDF5.Attribute))
-                _add_node(g, (attr_node, HDF5.name, rdflib.Literal(ak)))
+                attr_def: str = obj.attrsdef.get(ak, None)
+                if attr_def:
+                    _add_node(g, (attr_node, HDF5.name, rdflib.Literal(ak)))
+                    _add_node(g, (attr_node, SKOS.definition, rdflib.Literal(attr_def)))
+                    if 'skos' not in ctx:
+                        ctx['skos'] = 'http://www.w3.org/2004/02/skos/core#'
+                    # def_node = rdflib.BNode()
+                    # _add_node(g, (def_node, SCHEMA.comment, rdflib.Literal(attr_def)))
+                    # _add_node(g, (attr_node, HDF5.name, def_node))
+                else:
+                    _add_node(g, (attr_node, HDF5.name, rdflib.Literal(ak)))
 
             list_node = None
             attr_literal = None
@@ -818,7 +824,7 @@ def dumps(grp,
           structural: bool = True,
           resolve_keys: bool = False,
           **kwargs) -> str:
-    """Dump a group or a dataset to to string."""
+    """Dump a group or a dataset to string."""
     return json.dumps(dumpd(
         grp=grp,
         iri_only=iri_only,

tests/wrapper/test_jsonld.py

@@ -1,9 +1,10 @@
 import json
+import pathlib
+import unittest
+
 import numpy as np
 import ontolutils
-import pathlib
 import rdflib
-import unittest
 from ontolutils import M4I
 from ontolutils import namespaces, urirefs, Thing
 
@@ -472,3 +473,22 @@ def test_codemeta_to_hdf(self):
         h5tbx.dumps('test.hdf')
 
         h5.hdf_filename.unlink(missing_ok=True)
+
+    def test_jsonld_with_attrs_definition(self):
+        with h5tbx.File() as h5:
+            h5.attrs['name'] = h5tbx.Attribute('Matthias', definition='My first name')
+            jstr = h5.dump_jsonld()
+
+        sparql_str = """SELECT ?n ?v ?d
+{
+    ?id <http://purl.allotrope.org/ontologies/hdf5/1.8#attribute> ?a .
+    ?a <http://purl.allotrope.org/ontologies/hdf5/1.8#name> ?n .
+    ?a <http://purl.allotrope.org/ontologies/hdf5/1.8#value> ?v .
+    ?a <http://www.w3.org/2004/02/skos/core#definition> ?d .
+}"""
+        g = rdflib.Graph().parse(data=jstr, format='json-ld')
+        qres = g.query(sparql_str)
+        for row in qres:
+            self.assertEqual(str(row[0]), 'name')
+            self.assertEqual(str(row[1]), 'Matthias')
+            self.assertEqual(str(row[2]), 'My first name')

tests/wrapper/test_rdf.py

@@ -272,3 +272,6 @@ def test_definition(self):
             self.assertEqual(h5.attrsdef['name'], 'This is the name of the person to contact')
 
             h5.dumps()
+
+            with self.assertRaises(KeyError):
+                h5.attrsdef['test'] = 'This should not work!'