Improve usage of SSSOM-py (#666)

* Improve SSSOM-py usage

Relies on mapping-commons/sssom-py#440

* Update pyproject.toml

* Update sssom_utils.py

* Update sssom_utils.py

* Make code independent

* Fix typo

* Update lexical_indexer.py

* Update test_sssom_matcher.py

* Update test_sssom_matcher.py

* Update pyproject.toml

* Update lock

* Update lexical_indexer.py

* Update test_sssom_matcher.py

* Update pyproject.toml

* Lock

* Lock again

* Update reqs

pyproject.toml

@@ -15,7 +15,7 @@ SQLAlchemy = ">=1.4.32"
 linkml-runtime = ">=1.5.3"
 linkml-renderer = ">=0.3.0"
 networkx = ">=2.7.1"
-sssom = "<0.4.0"
+sssom = "^0.4.0"
 ratelimit = ">=2.2.1"
 appdirs = ">=1.4.4"
 semsql = ">=0.3.1"

src/oaklib/interfaces/basic_ontology_interface.py

@@ -209,9 +209,7 @@ def set_metamodel_mappings(self, mappings: Union[str, Path, List[Mapping]]) -> N
         :param mappings: Mappings for predicates such as rdfs:subClassOf
         :return:
         """
-        if isinstance(mappings, Path):
-            mappings = str
-        if isinstance(mappings, str):
+        if isinstance(mappings, (str, Path)):
             msdf = parse_sssom_table(mappings)
             msd = to_mapping_set_document(msdf)
             mappings = msd.mapping_set.mappings

src/oaklib/utilities/lexical/lexical_indexer.py

@@ -7,19 +7,20 @@
 """
 import logging
 import re
+import typing as t
 from collections import defaultdict
 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.sssom_document import MappingSetDocument
-from sssom.typehints import Metadata
-from sssom.util import MappingSetDataFrame, to_mapping_set_dataframe
-from sssom_schema import Mapping, MappingSet
+from sssom.constants import CURIE_MAP
+from sssom.context import ensure_converter
+from sssom.util import MappingSetDataFrame
+from sssom_schema import Mapping
 
 from oaklib.datamodels.lexical_index import (
     LexicalGrouping,
@@ -244,8 +245,8 @@ def lexical_index_to_sssom(
     oi: BasicOntologyInterface,
     lexical_index: LexicalIndex,
     ruleset: MappingRuleCollection = None,
-    meta: Metadata = None,
-    prefix_map: dict = None,
+    meta: Optional[Dict[str, t.Any]] = None,
+    prefix_map: Union[None, Converter, t.Mapping[str, str]] = None,
     subjects: Collection[CURIE] = None,
     objects: Collection[CURIE] = None,
     symmetric: bool = False,
@@ -310,24 +311,15 @@ 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]
-
-    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)
-    msdf = to_mapping_set_dataframe(doc)
-    num_mappings = len(msdf.df.index)
-    if ensure_strict_prefixes:
-        msdf.clean_prefix_map()
-        if len(msdf.df.index) < num_mappings:
-            raise ValueError("Mappings included prefixes that were not in the prefix map")
+    converter = curies.chain(
+        [
+            Converter.from_prefix_map((meta or {}).pop(CURIE_MAP, {})),
+            ensure_converter(prefix_map, use_defaults=False),
+            oi.converter,
+        ]
+    )
+    msdf = MappingSetDataFrame.from_mappings(mappings=mappings, metadata=meta, converter=converter)
+    msdf.clean_prefix_map(strict=ensure_strict_prefixes)
     return msdf
 
 

src/oaklib/utilities/mapping/sssom_utils.py

@@ -2,17 +2,14 @@
 from typing import Iterable, List, Optional, Tuple
 
 from linkml_runtime.utils.metamodelcore import URIorCURIE
-from sssom.sssom_document import MappingSetDocument
-from sssom.util import to_mapping_set_dataframe
+from sssom.util import MappingSetDataFrame
 from sssom.writers import write_table
-from sssom_schema import Mapping, MappingSet
+from sssom_schema import Mapping
 
 from oaklib.io.streaming_writer import StreamingWriter
 from oaklib.types import CURIE
 from oaklib.utilities.basic_utils import get_curie_prefix
 
-UNSPECIFIED_MAPPING_SET_ID = "https://w3id.org/sssom/license/unspecified"
-
 
 def create_sssom_mapping(
     subject_id: CURIE, object_id: CURIE, strict=False, **kwargs
@@ -74,14 +71,9 @@ def emit(self, obj: Mapping):
         self.mappings.append(obj)
 
     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)
-        msdf = to_mapping_set_dataframe(doc)
+        converter = self.ontology_interface.converter if self.ontology_interface else None
+        # default metadata, including an auto-generated mapping set URI and
+        # a license, are automatically added with this function
+        msdf = MappingSetDataFrame.from_mappings(self.mappings, converter=converter)
         msdf.clean_prefix_map(strict=False)
         write_table(msdf, self.file)

tests/test_utilities/test_sssom_matcher.py

@@ -1,7 +1,7 @@
 import unittest
 from pathlib import Path
 
-from sssom.io import get_metadata_and_prefix_map
+import yaml
 from sssom.parsers import to_mapping_set_document
 from sssom.writers import write_table
 from sssom_schema import Mapping
@@ -169,7 +169,7 @@ 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)
+        metadata = yaml.safe_load(MATCHER_META_YML.read_text())
         msdf = lexical_index_to_sssom(self.oi, lexical_index, ruleset=ruleset, meta=metadata)
         with open(MATCHER_TEST_SSSOM_OUT, "w", encoding="utf-8") as file:
             write_table(msdf, file)

tox.ini

@@ -28,6 +28,12 @@ deps =
     gilda
     linkml
 
+[testenv:lock]
+skip_install = true
+commands = poetry lock
+deps = poetry
+description = Lock dependencies with poetry
+
 [testenv:lint]
 deps =
     black