Add API for annotating year and update NLM catalog source (#268)

src/pyobo/sources/nlm_catalog.py

@@ -4,6 +4,7 @@
 from xml.etree import ElementTree
 
 from pyobo.struct import Obo, Reference, Term, TypeDef, default_reference
+from pyobo.struct.typedef import exact_match, has_end_date, has_start_date
 from pyobo.utils.path import ensure_df, ensure_path
 
 __all__ = [
@@ -14,8 +15,6 @@
 CATALOG_TO_PUBLISHER = "https://ftp.ncbi.nlm.nih.gov/pubmed/xmlprovidernames.txt"
 JOURNAL_INFO_PATH = "https://ftp.ncbi.nlm.nih.gov/pubmed/jourcache.xml"
 PUBLISHER = TypeDef.default(PREFIX, "has_publisher", name="has publisher")
-START_YEAR = TypeDef.default(PREFIX, "has_start_year", name="has start year")
-END_YEAR = TypeDef.default(PREFIX, "has_end_year", name="has end year")
 
 
 # TODO enrich with context from https://ftp.ncbi.nlm.nih.gov/pubmed/J_Entrez.txt and https://ftp.ncbi.nlm.nih.gov/pubmed/J_Medline.txt
@@ -26,7 +25,7 @@ class NLMCatalogGetter(Obo):
 
     bioversions_key = ontology = PREFIX
     dynamic_version = True
-    typedefs = [PUBLISHER, START_YEAR, END_YEAR]
+    typedefs = [PUBLISHER, has_end_date, has_start_date, exact_match]
     idspaces = {
         PREFIX: "https://www.ncbi.nlm.nih.gov/nlmcatalog/",
     }
@@ -70,9 +69,9 @@ def _process_journal(element, journal_id_to_publisher_key: dict[str, Reference])
         #  to determine a "canonical" one
         term.append_xref(Reference(prefix="issn", identifier=issn))
     if start_year := element.findtext("StartYear"):
-        term.annotate_integer(START_YEAR, start_year)
+        term.annotate_year(has_start_date, start_year)
     if end_year := element.findtext("EndYear"):
-        term.annotate_integer(END_YEAR, end_year)
+        term.annotate_year(has_end_date, end_year)
     if publisher_reference := journal_id_to_publisher_key.get(term.identifier):
         term.annotate_object(PUBLISHER, publisher_reference)
     return term

src/pyobo/struct/struct.py

@@ -649,9 +649,17 @@ def annotate_boolean(self, prop: ReferenceHint, value: bool) -> Self:
             prop, str(value).lower(), Reference(prefix="xsd", identifier="boolean")
         )
 
-    def annotate_integer(self, prop: ReferenceHint, value: str) -> Self:
+    def annotate_integer(self, prop: ReferenceHint, value: int | str) -> Self:
         """Append an object annotation."""
-        return self.annotate_literal(prop, value, Reference(prefix="xsd", identifier="integer"))
+        return self.annotate_literal(
+            prop, str(int(value)), Reference(prefix="xsd", identifier="integer")
+        )
+
+    def annotate_year(self, prop: ReferenceHint, value: int | str) -> Self:
+        """Append a year annotation."""
+        return self.annotate_literal(
+            prop, str(int(value)), Reference(prefix="xsd", identifier="gYear")
+        )
 
     def _definition_fp(self) -> str:
         definition = obo_escape_slim(self.definition) if self.definition else ""

src/pyobo/struct/typedef.py

@@ -25,6 +25,7 @@
     "gene_product_member_of",
     "has_contributor",
     "has_dbxref",
+    "has_end_date",
     "has_gene_product",
     "has_homepage",
     "has_inchi",
@@ -36,6 +37,7 @@
     "has_role",
     "has_salt",
     "has_smiles",
+    "has_start_date",
     "has_successor",
     "has_taxonomy_rank",
     "is_a",
@@ -402,6 +404,15 @@ def default(cls, prefix: str, identifier: str, *, name: str | None = None) -> Se
     is_metadata_tag=True,
 )
 
+has_start_date = TypeDef(
+    reference=Reference(prefix="dcat", identifier="startDate", name="has start date"),
+    is_metadata_tag=True,
+)
+has_end_date = TypeDef(
+    reference=Reference(prefix="dcat", identifier="endDate", name="has end date"),
+    is_metadata_tag=True,
+)
+
 default_typedefs: dict[ReferenceTuple, TypeDef] = {
     v.pair: v for v in locals().values() if isinstance(v, TypeDef)
 }

tests/test_struct.py

@@ -165,7 +165,49 @@ def test_property_literal(self) -> None:
             name: lysine dehydrogenase activity
             property_value: RO:1234567 "value" xsd:string
             """,
-            term.iterate_obo_lines(ontology_prefix="go", typedefs={}),
+            term.iterate_obo_lines(ontology_prefix="go", typedefs={RO_DUMMY.pair: RO_DUMMY}),
+        )
+
+    def test_property_integer(self) -> None:
+        """Test emitting property literals that were annotated as a boolean."""
+        term = Term(reference=LYSINE_DEHYDROGENASE_ACT)
+        term.annotate_integer(RO_DUMMY, 1234)
+        self.assert_lines(
+            """\
+            [Term]
+            id: GO:0050069
+            name: lysine dehydrogenase activity
+            property_value: RO:1234567 "1234" xsd:integer
+            """,
+            term.iterate_obo_lines(ontology_prefix="go", typedefs={RO_DUMMY.pair: RO_DUMMY}),
+        )
+
+    def test_property_bool(self) -> None:
+        """Test emitting property literals that were annotated as a boolean."""
+        term = Term(reference=LYSINE_DEHYDROGENASE_ACT)
+        term.annotate_boolean(RO_DUMMY, True)
+        self.assert_lines(
+            """\
+            [Term]
+            id: GO:0050069
+            name: lysine dehydrogenase activity
+            property_value: RO:1234567 "true" xsd:boolean
+            """,
+            term.iterate_obo_lines(ontology_prefix="go", typedefs={RO_DUMMY.pair: RO_DUMMY}),
+        )
+
+    def test_property_year(self) -> None:
+        """Test emitting property literals that were annotated as a year."""
+        term = Term(reference=LYSINE_DEHYDROGENASE_ACT)
+        term.annotate_year(RO_DUMMY, "1993")
+        self.assert_lines(
+            """\
+            [Term]
+            id: GO:0050069
+            name: lysine dehydrogenase activity
+            property_value: RO:1234567 "1993" xsd:gYear
+            """,
+            term.iterate_obo_lines(ontology_prefix="go", typedefs={RO_DUMMY.pair: RO_DUMMY}),
         )
 
     def test_property_object(self) -> None:
@@ -179,7 +221,7 @@ def test_property_object(self) -> None:
             name: lysine dehydrogenase activity
             property_value: RO:1234567 hgnc:123
             """,
-            term.iterate_obo_lines(ontology_prefix="go", typedefs={}),
+            term.iterate_obo_lines(ontology_prefix="go", typedefs={RO_DUMMY.pair: RO_DUMMY}),
         )
 
     def test_relation(self) -> None:
@@ -350,7 +392,7 @@ def test_property_default_reference(self) -> None:
             name: lysine dehydrogenase activity
             property_value: hey GO:1234569
             """,
-            term.iterate_obo_lines(ontology_prefix="go", typedefs={RO_DUMMY.pair: RO_DUMMY}),
+            term.iterate_obo_lines(ontology_prefix="go", typedefs={r.pair: r}),
         )
 
     def test_alt(self) -> None: