Use custom code for RDF (de)serialisation

.github/workflows/qc.yml

@@ -60,4 +60,5 @@ jobs:
         run: |
           poetry add "pandas<2.0.0"
           poetry add "numpy<1.24"
-          poetry run tox -e py
+          poetry add "scipy<1.16.3"
+          poetry run tox -e py

pyproject.toml

@@ -18,7 +18,7 @@ dependencies = [
     "linkml-runtime>=1.7.5",
     "linkml>1.7.10",
     "pandas>1.0.3",
-    "sssom-schema==1.1.0a2",
+    "sssom-schema==1.1.0a3",
     "sparqlwrapper>=2.0.0",
     "validators>=0.20.0",
     "deprecation>=2.1.0",

src/sssom/constants.py

@@ -82,6 +82,7 @@
 PREDICATE_MODIFIER_NOT = "Not"
 PREDICATE_LABEL = "predicate_label"
 PREDICATE_TYPE = "predicate_type"
+PUBLICATION_DATE = "publication_date"
 OBJECT_ID = "object_id"
 OBJECT_LABEL = "object_label"
 OBJECT_CATEGORY = "object_category"
@@ -120,6 +121,7 @@
 MATCH_STRING = "match_string"
 SUBJECT_PREPROCESSING = "subject_preprocessing"
 OBJECT_PREPROCESSING = "object_preprocessing"
+RECORD_ID = "record_id"
 SEMANTIC_SIMILARITY_SCORE = "semantic_similarity_score"
 SEMANTIC_SIMILARITY_MEASURE = "semantic_similarity_measure"
 SEE_ALSO = "see_also"
@@ -136,6 +138,8 @@
 # see <https://mapping-commons.github.io/sssom/spec-model/#representing-unmapped-entities>
 NO_TERM_FOUND = "sssom:NoTermFound"
 
+ENTITY_TYPE_RDFS_LITERAL = "rdfs literal"
+
 # PREDICATES
 OWL_EQUIVALENT_CLASS = "owl:equivalentClass"
 OWL_EQUIVALENT_PROPERTY = "owl:equivalentProperty"

src/sssom/parsers.py

@@ -39,7 +39,6 @@
     get_filter_df_by_prefixes_index,
 )
 from linkml_runtime.loaders.json_loader import JSONLoader
-from linkml_runtime.loaders.rdflib_loader import RDFLibLoader
 from pandas.errors import EmptyDataError
 from rdflib import Graph
 from sssom_schema import EntityTypeEnum, Mapping, MappingSet
@@ -80,6 +79,7 @@
 )
 
 from .context import ConverterHint, _get_built_in_prefix_map, ensure_converter
+from .rdf_internal import MappingSetRDFConverter
 from .sssom_document import MappingSetDocument
 from .util import (
     SSSOM_DEFAULT_RDF_SERIALISATION,
@@ -617,34 +617,23 @@ def from_sssom_rdf(
     :returns: MappingSetDataFrame object
     """
     converter = ensure_converter(prefix_map)
-    mapping_set = cast(
-        MappingSet,
-        RDFLibLoader().load(
-            source=g,
-            target_class=MappingSet,
-            schemaview=_get_sssom_schema_object().view,
-            prefix_map=converter.bimap,
-            ignore_unmapped_predicates=True,
-        ),
-    )
 
     # The priority order for combining metadata is:
     #  1. Metadata appearing in the SSSOM document
     #  2. Metadata passed through ``meta`` to this function
     #  3. Default metadata
 
-    # As the Metadata appearing in the SSSOM document is already parsed by LinkML
-    # we only need to overwrite the metadata from 2 and 3 if it is not present
+    # We prepare a default dictionary with (2) and (3), which will be
+    # automatically combined with the metadata extracted from the RDF
+    # graph by the RDF converter.
     combine_meta = dict(
         ChainMap(
             meta or {},
             get_default_metadata(),
         )
     )
 
-    _set_metadata_in_mapping_set(mapping_set, metadata=combine_meta, overwrite=False)
-    mdoc = MappingSetDocument(mapping_set=mapping_set, converter=converter)
-    return to_mapping_set_dataframe(mdoc)
+    return MappingSetRDFConverter().msdf_from_rdf(g, cc=converter, meta=combine_meta)
 
 
 def from_sssom_json(