Update to latest SSSOM(v0.4.x)

pyproject.toml

@@ -16,12 +16,12 @@ linkml-runtime = ">=1.5.3"
 linkml-renderer = ">=0.3.0"
 networkx = ">=2.7.1"
 sssom-schema = ">=0.11.0"
-sssom = ">=0.3.38"
+sssom = "0.4.0rc1"
 ratelimit = ">=2.2.1"
 appdirs = ">=1.4.4"
 semsql = ">=0.3.1"
 lark = ">=1.1.2"
-kgcl-schema = "0.6.0"
+kgcl-schema = ">=0.6.0"
 funowl = ">=0.2.0"
 gilda = {version = ">=1.0.0", optional = true}
 kgcl-rdflib = "0.5.0"
@@ -37,7 +37,7 @@ pysolr = "^3.9.0"
 eutils = ">=0.6.0"
 requests-cache = "^1.0.1"
 click = "*"
-semsimian = "0.2.1"
+semsimian = ">=0.2.10"
 urllib3 = {version = "< 2", optional = true}
 pydantic = "*"
 

src/oaklib/datamodels/vocabulary.py

@@ -16,6 +16,8 @@
 URL_PREDICATE = "schema:url"
 PREFIX_PREDICATE = "sh:prefix"
 
+CURIE_MAP_KEY = "curie_map"
+
 # TODO: replace with oio vocab
 LABEL_PREDICATE = omd.slots.label.curie
 HAS_EXACT_SYNONYM = omd.slots.has_exact_synonym.curie

src/oaklib/utilities/lexical/lexical_indexer.py

@@ -11,13 +11,14 @@
 from pathlib import Path
 from typing import Collection, Dict, List, Optional, Tuple, Union
 
+import curies
+from curies import Converter
 from linkml_runtime.dumpers import json_dumper, yaml_dumper
 from linkml_runtime.loaders import json_loader, yaml_loader
 from linkml_runtime.utils.metamodelcore import URIorCURIE
-from sssom.constants import LICENSE, MAPPING_SET_ID
-from sssom.context import get_default_metadata
+from sssom.constants import LICENSE, MAPPING_SET_ID, MetadataType, get_default_metadata
+from sssom.context import ensure_converter
 from sssom.sssom_document import MappingSetDocument
-from sssom.typehints import Metadata
 from sssom.util import MappingSetDataFrame, to_mapping_set_dataframe
 from sssom_schema import Mapping, MappingSet
 
@@ -35,6 +36,7 @@
     Synonymizer,
 )
 from oaklib.datamodels.vocabulary import (
+    CURIE_MAP_KEY,
     IDENTIFIER_PREDICATE,
     SEMAPV,
     SKOS_BROAD_MATCH,
@@ -244,7 +246,7 @@ def lexical_index_to_sssom(
     oi: BasicOntologyInterface,
     lexical_index: LexicalIndex,
     ruleset: MappingRuleCollection = None,
-    meta: Metadata = None,
+    metadata: MetadataType = None,
     prefix_map: dict = None,
     subjects: Collection[CURIE] = None,
     objects: Collection[CURIE] = None,
@@ -310,18 +312,24 @@ def lexical_index_to_sssom(
         #            mappings.append(create_mapping(oi, term, r1, r2))
     logging.info("Done creating SSSOM mappings")
 
-    if meta is None:
-        meta = get_default_metadata()
-    if prefix_map:
-        meta.prefix_map.update(
-            {k: v for k, v in prefix_map.items() if k not in meta.prefix_map.keys()}
-        )
-    mapping_set_id = meta.metadata[MAPPING_SET_ID]
-    license = meta.metadata[LICENSE]
+    if metadata is None:
+        metadata = get_default_metadata()
+
+    converter = curies.chain(
+        [
+            Converter.from_prefix_map(metadata.pop(CURIE_MAP_KEY, {})),
+            ensure_converter(prefix_map, use_defaults=False),
+        ]
+    )
+    metadata.setdefault(CURIE_MAP_KEY, {}).update(converter.prefix_map)
+
+    mapping_set_id = metadata[MAPPING_SET_ID]
+    license = metadata[LICENSE]
 
     mset = MappingSet(mapping_set_id=mapping_set_id, mappings=mappings, license=license)
     # doc = MappingSetDocument(prefix_map=oi.prefix_map(), mapping_set=mset)
-    doc = MappingSetDocument(prefix_map=meta.prefix_map, mapping_set=mset)
+
+    doc = MappingSetDocument(converter=converter, mapping_set=mset)
     msdf = to_mapping_set_dataframe(doc)
     num_mappings = len(msdf.df.index)
     if ensure_strict_prefixes:

src/oaklib/utilities/mapping/sssom_utils.py

@@ -1,7 +1,9 @@
 from dataclasses import dataclass, field
 from typing import Iterable, List, Optional, Tuple
 
+import curies
 from linkml_runtime.utils.metamodelcore import URIorCURIE
+from sssom.context import ensure_converter
 from sssom.sssom_document import MappingSetDocument
 from sssom.util import to_mapping_set_dataframe
 from sssom.writers import write_table
@@ -77,11 +79,16 @@ def finish(self):
         mset = MappingSet(
             mapping_set_id="temp", mappings=self.mappings, license=UNSPECIFIED_MAPPING_SET_ID
         )
-        if self.ontology_interface:
-            prefix_map = self.ontology_interface.prefix_map()
-        else:
-            prefix_map = {}
-        doc = MappingSetDocument(prefix_map=prefix_map, mapping_set=mset)
+
+        converter = (
+            curies.chain(
+                [ensure_converter(self.ontology_interface.prefix_map(), use_defaults=False)]
+            )
+            if self.ontology_interface
+            else None
+        )
+
+        doc = MappingSetDocument(converter=converter, mapping_set=mset)
         msdf = to_mapping_set_dataframe(doc)
         msdf.clean_prefix_map(strict=False)
         write_table(msdf, self.file)

tests/test_utilities/test_sssom_matcher.py

@@ -169,8 +169,14 @@ def test_sssom_with_rules_file(self):
         lexical_index = create_lexical_index(oi)
         add_labels_from_uris(oi)
         ruleset = load_mapping_rules(str(RULES))
-        metadata = get_metadata_and_prefix_map(MATCHER_META_YML)
-        msdf = lexical_index_to_sssom(self.oi, lexical_index, ruleset=ruleset, meta=metadata)
+        converter, metadata = get_metadata_and_prefix_map(MATCHER_META_YML)
+        msdf = lexical_index_to_sssom(
+            self.oi,
+            lexical_index,
+            ruleset=ruleset,
+            metadata=metadata,
+            prefix_map=converter.prefix_map,
+        )
         with open(MATCHER_TEST_SSSOM_OUT, "w", encoding="utf-8") as file:
             write_table(msdf, file)
         msdoc = to_mapping_set_document(msdf)

tox.ini

@@ -26,6 +26,7 @@ deps =
     coverage
     semsimian
     gilda
+    linkml
 
 [testenv:lint]
 deps =