Pass through SSSOM JSON-LD context when writing JSON (#479)

Fixes #477

---------

Co-authored-by: Charles Tapley Hoyt <[email protected]>

Author: matentzn

src/sssom/context.py

@@ -42,11 +42,15 @@ def _get_default_converter() -> Converter:
     return Converter(records)
 
 
+def _load_sssom_context():
+    with open(SSSOM_CONTEXT) as file:
+        return json.load(file, strict=False)
+
+
 @lru_cache(1)
 def _get_built_in_prefix_map() -> Converter:
     """Get URI prefixes for built-in prefixes."""
-    with open(SSSOM_CONTEXT) as file:
-        context = json.load(file, strict=False)
+    context = _load_sssom_context()
     prefix_map = {
         prefix: uri_prefix
         for prefix, uri_prefix in context["@context"].items()

src/sssom/writers.py

@@ -7,6 +7,7 @@
 
 import pandas as pd
 import yaml
+from curies import Converter
 from jsonasobj2 import JsonObj
 from linkml_runtime.dumpers import JSONDumper, rdflib_dumper
 from linkml_runtime.utils.schemaview import SchemaView
@@ -17,6 +18,7 @@
 from sssom.validators import check_all_prefixes_in_curie_map
 
 from .constants import CURIE_MAP, SCHEMA_YAML, SSSOM_URI_PREFIX
+from .context import _load_sssom_context
 from .parsers import to_mapping_set_document
 from .util import (
     RDF_FORMATS,
@@ -433,10 +435,24 @@ def to_fhir_json(msdf: MappingSetDataFrame) -> Dict:
     return json_obj
 
 
+def _update_sssom_context_with_prefixmap(converter: Converter):
+    """Prepare a JSON-LD context and dump to a string."""
+    context = _load_sssom_context()
+    for k, v in converter.bimap.items():
+        if k in context["@context"] and context["@context"][k] != v:
+            logging.info(
+                f"{k} namespace is already in the context, ({context['@context'][k]}, "
+                f"but with a different value than {v}. Overwriting!"
+            )
+        context["@context"][k] = v
+    return context
+
+
 def to_json(msdf: MappingSetDataFrame) -> JsonObj:
     """Convert a mapping set dataframe to a JSON object."""
     doc = to_mapping_set_document(msdf)
-    data = JSONDumper().dumps(doc.mapping_set, contexts={"@context": doc.prefix_map})
+    context = _update_sssom_context_with_prefixmap(doc.converter)
+    data = JSONDumper().dumps(doc.mapping_set, contexts=json.dumps(context))
     json_obj = json.loads(data)
     return json_obj
 

tests/test_writers.py

@@ -4,8 +4,23 @@
 import os
 import unittest
 
+import pandas as pd
+from curies import Converter
+
+from sssom import MappingSetDataFrame
+from sssom.constants import (
+    CREATOR_ID,
+    OBJECT_ID,
+    OBJECT_LABEL,
+    PREDICATE_ID,
+    SEMAPV,
+    SUBJECT_ID,
+    SUBJECT_LABEL,
+)
 from sssom.parsers import parse_sssom_json, parse_sssom_rdf, parse_sssom_table
 from sssom.writers import (
+    _update_sssom_context_with_prefixmap,
+    to_json,
     write_fhir_json,
     write_json,
     write_ontoportal_json,
@@ -67,6 +82,52 @@ def test_write_sssom_json(self):
             f"{path} has the wrong number of mappings.",
         )
 
+    def test_write_sssom_json_context(self):
+        """Test when writing to JSON, the context is correctly written as well."""
+        rows = [
+            (
+                "DOID:0050601",
+                "ADULT syndrome",
+                "skos:exactMatch",
+                "UMLS:C1863204",
+                "ADULT SYNDROME",
+                SEMAPV.ManualMappingCuration.value,
+                "orcid:0000-0003-4423-4370",
+            )
+        ]
+        columns = [
+            SUBJECT_ID,
+            SUBJECT_LABEL,
+            PREDICATE_ID,
+            OBJECT_ID,
+            OBJECT_LABEL,
+            SEMAPV.ManualMappingCuration.value,
+            CREATOR_ID,
+        ]
+        df = pd.DataFrame(rows, columns=columns)
+        msdf = MappingSetDataFrame(df)
+        msdf.clean_prefix_map()
+        json_object = to_json(msdf)
+        self.assertIn("@context", json_object)
+        self.assertIn("DOID", json_object["@context"])
+        self.assertIn("mapping_set_id", json_object["@context"])
+
+    def test_update_sssom_context_with_prefixmap(self):
+        """Test when writing to JSON, the context is correctly written as well."""
+        records = [
+            {
+                "prefix": "SCTID",
+                "prefix_synonyms": ["snomed"],
+                "uri_prefix": "http://snomed.info/id/",
+            },
+        ]
+        converter = Converter.from_extended_prefix_map(records)
+        context = _update_sssom_context_with_prefixmap(converter)
+        self.assertIn("@context", context)
+        self.assertIn("SCTID", context["@context"])
+        self.assertNotIn("snomed", context["@context"])
+        self.assertIn("mapping_set_id", context["@context"])
+
     def test_write_sssom_fhir(self):
         """Test writing as FHIR ConceptMap JSON."""
         path = os.path.join(test_out_dir, "test_write_sssom_fhir.json")