Improve root term handling (#266)

1. Fix bug to output correctly normalized "default" references 2. Add reading for property values with custom handling for root terms 3. Still TODO - handle reading "default" references
biopragmatics · Jan 1, 2025 · a036f9c · a036f9c
1 parent 8e5d739
commit a036f9c
Show file tree

Hide file tree

Showing 3 changed files with 34 additions and 3 deletions.
diff --git a/src/pyobo/reader.py b/src/pyobo/reader.py
@@ -35,7 +35,7 @@
 )
 from .struct.reference import _parse_identifier
 from .struct.struct import DEFAULT_SYNONYM_TYPE, LiteralProperty, ObjectProperty
-from .struct.typedef import default_typedefs
+from .struct.typedef import default_typedefs, has_ontology_root_term
 from .utils.misc import STATIC_VERSION_REWRITES, cleanup_version
 
 __all__ = [
@@ -125,6 +125,21 @@ def from_obonet(
             f"[{ontology_prefix}] slashes not allowed in data versions because of filesystem usage: {data_version}"
         )
 
+    root_terms: list[Reference] = []
+    for t in iterate_node_properties(
+        graph.graph,
+        ontology_prefix=ontology_prefix,
+        upgrade=upgrade,
+        node=Reference(prefix="obo", identifier=ontology_prefix),
+    ):
+        # TODO other custom handling can be put here
+        match t:
+            case LiteralProperty(predicate, value, datatype):
+                pass
+            case ObjectProperty(predicate, obj, _):
+                if predicate.pair == has_ontology_root_term.pair:
+                    root_terms.append(obj)
+
     #: CURIEs to typedefs
     typedefs: Mapping[ReferenceTuple, TypeDef] = {
         typedef.pair: typedef
@@ -241,7 +256,7 @@ def from_obonet(
             match t:
                 case LiteralProperty(predicate, value, datatype):
                     term.annotate_literal(predicate, value, datatype)
-                case ObjectProperty(predicate, obj, _datatype):
+                case ObjectProperty(predicate, obj, _):
                     term.annotate_object(predicate, obj)
         terms.append(term)
 
@@ -260,6 +275,7 @@ def from_obonet(
         _synonym_typedefs=list(synonym_typedefs.values()),
         _date=date,
         _data_version=data_version,
+        _root_terms=root_terms,
         terms=terms,
     )
 

diff --git a/src/pyobo/struct/struct.py b/src/pyobo/struct/struct.py
@@ -1105,7 +1105,7 @@ def iterate_obo_lines(
             yield f'property_value: http://purl.org/dc/terms/description "{description}" xsd:string'
 
         for root_term in self.root_terms or []:
-            yield f"property_value: {has_ontology_root_term.preferred_curie} {root_term.preferred_curie}"
+            yield f"property_value: {has_ontology_root_term.preferred_curie} {reference_escape(root_term, ontology_prefix=self.ontology)}"
 
         for typedef in sorted(self.typedefs or []):
             yield from typedef.iterate_obo_lines(ontology_prefix=self.ontology)

diff --git a/tests/test_reader.py b/tests/test_reader.py
@@ -984,6 +984,21 @@ def test_sssom_axiom(self) -> None:
         self.assertIsNotNone(context.contributor)
         self.assertEqual("0000-0003-4423-4370", context.contributor.identifier)
 
+    def test_root(self) -> None:
+        """Test root terms."""
+        ontology = _read("""\
+            ontology: go
+            property_value: IAO:0000700 GO:0050069
+
+            [Term]
+            id: GO:0050069
+        """)
+        # FIXME support default reference, like property_value: IAO:0000700 adhoc
+        self.assertEqual(
+            [Reference(prefix="GO", identifier="0050069")],
+            ontology.root_terms,
+        )
+
 
 class TestVersionHandling(unittest.TestCase):
     """Test version handling."""