Bug fix for previous mappings during alignment (#477)

* Add fix for obo

* Update utils.py

* Update struct.py

* Update utils.py

* Update utils.py

* Update utils.py

* Update utils.py

* clean up

* Update utils.py

* Clean up

* Update utils.py

src/bioregistry/align/obofoundry.py

@@ -17,13 +17,13 @@ class OBOFoundryAligner(Aligner):
 
     key = "obofoundry"
     getter = get_obofoundry
-    curation_header = ("name", "description")
+    curation_header = ("deprecated", "name", "description")
     include_new = True
+    normalize_invmap = True
 
     def get_skip(self) -> Mapping[str, str]:
         """Get the prefixes in the OBO Foundry that should be skipped."""
         return {
-            "bila": "website is not longer active",
             "obo_rel": "replaced",
         }
 

src/bioregistry/align/utils.py

@@ -47,6 +47,8 @@ class Aligner(ABC):
 
     subkey: ClassVar[str] = "prefix"
 
+    normalize_invmap: ClassVar[bool] = False
+
     def __init__(self):
         """Instantiate the aligner."""
         if self.key not in read_metaregistry():
@@ -60,7 +62,10 @@ def __init__(self):
         self.skip_external = self.get_skip()
 
         # Get all of the pre-curated mappings from the Bioregistry
-        self.external_id_to_bioregistry_id = self.manager.get_registry_invmap(self.key)
+        self.external_id_to_bioregistry_id = self.manager.get_registry_invmap(
+            self.key,
+            normalize=self.normalize_invmap,
+        )
 
         # Run lexical alignment
         self._align()
@@ -81,46 +86,53 @@ def _align(self):
                 continue
 
             bioregistry_id = self.external_id_to_bioregistry_id.get(external_id)
+            # There's already a mapping for this external ID to a bioregistry
+            # entry. Just add all of the latest metadata and move on
+            if bioregistry_id is not None:
+                self._align_action(bioregistry_id, external_id, external_entry)
+                continue
 
             # try to lookup with lexical match
-            if bioregistry_id is None:
-                if not self.alt_key_match:
-                    bioregistry_id = self.manager.normalize_prefix(external_id)
-                else:
-                    alt_match = external_entry.get(self.alt_key_match)
-                    if alt_match:
-                        bioregistry_id = self.manager.normalize_prefix(alt_match)
+            if not self.alt_key_match:
+                bioregistry_id = self.manager.normalize_prefix(external_id)
+            else:
+                alt_match = external_entry.get(self.alt_key_match)
+                if alt_match:
+                    bioregistry_id = self.manager.normalize_prefix(alt_match)
+
+            # A lexical match was possible
+            if bioregistry_id is not None:
+                # check this external ID for curated mismatches, and move
+                # on if one has already been curated
+                if is_mismatch(bioregistry_id, self.key, external_id):
+                    continue
+                if self.skip_deprecated and self.manager.is_deprecated(bioregistry_id):
+                    continue
+                self._align_action(bioregistry_id, external_id, external_entry)
+                continue
 
             # add the identifier from an external resource if it's been marked as high quality
-            if bioregistry_id is None and self.include_new:
+            elif self.include_new:
                 bioregistry_id = norm(external_id)
+                if is_mismatch(bioregistry_id, self.key, external_id):
+                    continue
                 self.internal_registry[bioregistry_id] = Resource(prefix=bioregistry_id)
-
-            if self._do_align_action(bioregistry_id):
                 self._align_action(bioregistry_id, external_id, external_entry)
+                continue
 
-    def _do_align_action(self, prefix: Optional[str]) -> bool:
-        # a match was found if the prefix is not None
-        return prefix is not None and (
-            not self.skip_deprecated or not self.manager.is_deprecated(prefix)
-        )
-
-    def _align_action(self, bioregistry_id, external_id, external_entry):
-        # skip mismatches
-        if is_mismatch(bioregistry_id, self.key, external_id):
-            return
-
-        # Add mapping
+    def _align_action(
+        self, bioregistry_id: str, external_id: str, external_entry: Dict[str, Any]
+    ) -> None:
         if self.internal_registry[bioregistry_id].mappings is None:
             self.internal_registry[bioregistry_id].mappings = {}
-        self.internal_registry[bioregistry_id].mappings[self.key] = external_id
+        self.internal_registry[bioregistry_id].mappings[self.key] = external_id  # type:ignore
 
         _entry = self.prepare_external(external_id, external_entry)
         _entry[self.subkey] = external_id
         self.internal_registry[bioregistry_id][self.key] = _entry
         self.external_id_to_bioregistry_id[external_id] = bioregistry_id
 
-    def prepare_external(self, external_id, external_entry) -> Dict[str, Any]:
+    def prepare_external(self, external_id: str, external_entry: Dict[str, Any]) -> Dict[str, Any]:
         """Prepare a dictionary to be added to the bioregistry for each external registry entry.
 
         The default implementation returns `external_entry` unchanged.

src/bioregistry/data/bioregistry.json

@@ -4351,6 +4351,7 @@
       "uri_format": "http://purl.obolibrary.org/obo/BILA_$1"
     },
     "deprecated": true,
+    "example": "0000000",
     "fairsharing": {
       "abbreviation": "BILA",
       "description": "This is a record of the Bilateria anatomy terminology and ontology. We have been unable to find any further information about this resource.",
@@ -4364,6 +4365,11 @@
       "obofoundry": "bila"
     },
     "obofoundry": {
+      "contact": "[email protected]",
+      "contact.label": "Thorsten Heinrich",
+      "deprecated": true,
+      "download.owl": "http://purl.obolibrary.org/obo/bila.owl",
+      "homepage": "http://4dx.embl.de/4DXpress",
       "name": "Bilateria anatomy",
       "prefix": "bila"
     }
@@ -12968,6 +12974,7 @@
       "prefix": "GRO-CPGA"
     },
     "deprecated": true,
+    "description": "A structured controlled vocabulary for the anatomy of Graminae. Please note that this ontology has now been superseded by the Plant Ontology.",
     "mappings": {
       "biocontext": "gro.cpga",
       "bioportal": "GRO-CPGA",
@@ -12978,7 +12985,6 @@
       "contact.label": "Plant Ontology Administrators",
       "deprecated": true,
       "homepage": "http://www.gramene.org/plant_ontology/",
-      "inactive": true,
       "name": "Cereal Plant Gross Anatomy",
       "prefix": "gro"
     }
@@ -25291,14 +25297,19 @@
     "namespace_in_lui": true,
     "obofoundry": {
       "contact": "[email protected]",
+      "contact.github": "hoganwr",
       "contact.label": "Bill Hogan",
+      "contact.orcid": "0000-0002-9881-1017",
       "deprecated": false,
       "description": "An ontology of geographical entities",
       "download.owl": "http://purl.obolibrary.org/obo/geo.owl",
       "homepage": "https://github.com/ufbmi/geographical-entity-ontology/wiki",
-      "license": "CC-BY 4.0",
+      "license": "CC BY 4.0",
+      "license.url": "https://creativecommons.org/licenses/by/4.0/",
       "name": "Geographical Entity Ontology",
-      "prefix": "geo"
+      "preferredPrefix": "GEO",
+      "prefix": "geo",
+      "repository": "https://github.com/ufbmi/geographical-entity-ontology"
     },
     "ols": {
       "description": "An ontology of geographical entities",

src/bioregistry/resource_manager.py

@@ -170,7 +170,19 @@ def get_registry_map(self, metaprefix: str) -> Dict[str, str]:
 
     @lru_cache(maxsize=None)  # noqa:B019
     def get_registry_invmap(self, metaprefix: str, normalize: bool = False) -> Dict[str, str]:
-        """Get a mapping from prefixes in another registry to Bioregistry prefixes."""
+        """Get a mapping from prefixes in another registry to Bioregistry prefixes.
+
+        :param metaprefix: Which external registry should be used?
+        :param normalize: Should the external prefixes be normalized?
+        :returns: A mapping of external prefixes to bioregistry prefies
+
+        >>> from bioregistry import manager
+        >>> obofoundry_to_bioregistry = manager.get_registry_invmap("obofoundry", normalize=True)
+        >>> obofoundry_to_bioregistry["go"]
+        'go'
+        >>> obofoundry_to_bioregistry["geo"]
+        'geogeo'
+        """
         if normalize:
             return {
                 _norm(external_prefix): prefix