make uri_to_curie slightly more robust

eva_cttv_pipeline/clinvar_xml_io/clinvar_xml_io/ontology_uri.py

@@ -1,3 +1,10 @@
+import logging
+
+logging.basicConfig()
+logger = logging.getLogger(__name__)
+logger.setLevel(logging.INFO)
+
+
 class OntologyUri:
     # ClinVar stores cross-references in very different formats. This provides their conversion to full IRIs, along with
     # some examples of how this looks like in ClinVar data.
@@ -47,6 +54,13 @@ def uri_to_curie(uri):
             db = "OMIM"
             id_ = uri_list[-1]
         else:
-            db, id_ = uri_list[-1].split("_")
+            last_component = uri_list[-1]
+            if ":" in last_component:
+                return last_component
+            elif "_" in last_component:
+                db, id_ = last_component.split("_")
+            else:
+                logger.warning(f"Could not convert URI to CURIE: {uri}")
+                return None
         db = uri_db_to_curie_db[db.lower()]
         return "{}:{}".format(db, id_)

eva_cttv_pipeline/clinvar_xml_io/tests/test_ontology_uri.py

@@ -0,0 +1,11 @@
+from eva_cttv_pipeline.clinvar_xml_io.clinvar_xml_io.ontology_uri import OntologyUri
+
+
+def test_uri_to_curie():
+    assert OntologyUri.uri_to_curie('http://www.orpha.net/ORDO/Orphanet_713') == 'Orphanet:713'
+    assert OntologyUri.uri_to_curie('http://purl.obolibrary.org/obo/HP_0002930') == 'HP:0002930'
+    assert OntologyUri.uri_to_curie('https://omim.org/entry/300377') == 'OMIM:300377'
+    # If for some reason we're already passed a CURIE (probably a mistake in OLS response), return it as-is
+    assert OntologyUri.uri_to_curie('HP:0002505') == 'HP:0002505'
+    # Not a supported db
+    assert OntologyUri.uri_to_curie('http://purl.obolibrary.org/obo/DOID_10652') == None