From 669f7301a550c6ef6fc529d7c3a7f27106c6cd3a Mon Sep 17 00:00:00 2001
From: Charles Tapley Hoyt <cthoyt@gmail.com>
Date: Sun, 19 Jan 2025 14:34:25 -0500
Subject: [PATCH] Add ROR organization classifications (#326)

---
 src/pyobo/sources/ror.py   | 50 +++++++++++++++++++++++---------------
 src/pyobo/struct/struct.py |  5 ++++
 2 files changed, 36 insertions(+), 19 deletions(-)

diff --git a/src/pyobo/sources/ror.py b/src/pyobo/sources/ror.py
index 3c003a47..7aac9175 100644
--- a/src/pyobo/sources/ror.py
+++ b/src/pyobo/sources/ror.py
@@ -3,6 +3,7 @@
 from __future__ import annotations
 
 import json
+import logging
 import zipfile
 from collections.abc import Iterable
 from typing import Any
@@ -12,7 +13,7 @@
 from tqdm.auto import tqdm
 
 from pyobo.struct import Obo, Reference, Term
-from pyobo.struct.struct import acronym
+from pyobo.struct.struct import CHARLIE_TERM, HUMAN_TERM, PYOBO_INJECTED, acronym
 from pyobo.struct.typedef import (
     has_homepage,
     has_part,
@@ -23,6 +24,7 @@
     see_also,
 )
 
+logger = logging.getLogger(__name__)
 PREFIX = "ror"
 ROR_ZENODO_RECORD_ID = "10086202"
 
@@ -81,26 +83,35 @@ def __post_init__(self):
 
     def iter_terms(self, force: bool = False) -> Iterable[Term]:
         """Iterate over terms in the ontology."""
-        return iterate_ror_terms(force=force)
-
-
-ROR_ORGANIZATION_TYPE_TO_OBI = {
-    "Education": ...,
-    "Facility": ...,
-    "Company": ...,
-    "Government": ...,
-    "Healthcare": ...,
-    "Other": ...,
-    "Archive": ...,
+        yield CHARLIE_TERM
+        yield HUMAN_TERM
+        yield Term(reference=ORG_CLASS)
+        yield Term(reference=CITY_CLASS)
+        yield from ROR_ORGANIZATION_TYPE_TO_OBI.values()
+        yield from iterate_ror_terms(force=force)
+
+
+ROR_ORGANIZATION_TYPE_TO_OBI: dict[str, Term] = {
+    "Education": Term.default(PREFIX, "education", "educational organization"),
+    "Facility": Term.default(PREFIX, "facility", "facility"),
+    "Company": Term.default(PREFIX, "company", "company"),
+    "Government": Term.default(PREFIX, "government", "government organization"),
+    "Healthcare": Term.default(PREFIX, "healthcare", "healthcare organization"),
+    "Archive": Term.default(PREFIX, "archive", "archival organization"),
+    "Nonprofit": Term.default(PREFIX, "healthcare", "nonprofit organization")
+    .append_xref(Reference(prefix="ICO", identifier="0000048"))
+    .append_xref(Reference(prefix="GSSO", identifier="004615")),
 }
+for _k, v in ROR_ORGANIZATION_TYPE_TO_OBI.items():
+    v.append_parent(ORG_CLASS)
+    v.append_contributor(CHARLIE_TERM)
+    v.append_comment(PYOBO_INJECTED)
+
 _MISSED_ORG_TYPES: set[str] = set()
 
 
 def iterate_ror_terms(*, force: bool = False) -> Iterable[Term]:
     """Iterate over terms in ROR."""
-    yield Term(reference=ORG_CLASS)
-    yield Term(reference=CITY_CLASS)
-
     _version, _source_uri, records = get_latest(force=force)
     unhandled_xref_prefixes = set()
 
@@ -120,10 +131,11 @@ def iterate_ror_terms(*, force: bool = False) -> Iterable[Term]:
             type="Instance",
             definition=description,
         )
-        term.append_parent(ORG_CLASS)
-        # TODO replace term.append_parent(ORG_CLASS) with:
-        # for organization_type in organization_types:
-        #     term.append_parent(ORG_PARENTS[organization_type])
+        for organization_type in organization_types:
+            if organization_type == "Other":
+                term.append_parent(ORG_CLASS)
+            else:
+                term.append_parent(ROR_ORGANIZATION_TYPE_TO_OBI[organization_type])
 
         for link in record.get("links", []):
             term.annotate_uri(has_homepage, link)
diff --git a/src/pyobo/struct/struct.py b/src/pyobo/struct/struct.py
index 40472bec..92aebefa 100644
--- a/src/pyobo/struct/struct.py
+++ b/src/pyobo/struct/struct.py
@@ -323,6 +323,11 @@ def auto(
             definition=get_definition(prefix, identifier),
         )
 
+    @classmethod
+    def default(cls, prefix, identifier, name=None) -> Self:
+        """Create a default term."""
+        return cls(reference=default_reference(prefix=prefix, identifier=identifier, name=name))
+
     def append_see_also_uri(self, uri: str) -> Self:
         """Add a see also property."""
         return self.annotate_uri(v.see_also, uri)