Merge pull request #125 from openstates/rj_updates

6.18.3: Add other_names to Match Committees on Event Importer

CHANGELOG.md

@@ -1,5 +1,9 @@
 # Changelog
 
+## 6.18.3 - March 25, 2024
+
+* Add other_names to committee matching for Event Imports
+
 ## 6.18.2
 
 * Maine (ME): add a tribal representative district to the jurisdiction metadata

openstates/importers/base.py

@@ -593,3 +593,60 @@ def resolve_person(
 
         # return the newly-cached object
         return self.person_cache[cache_key]
+
+    def resolve_committee(
+        self, json_id: str, allow_no_match: bool = False
+    ) -> typing.Optional[_ID]:
+        if not json_id:
+            return None
+
+        if json_id.startswith("~"):
+            # keep caches of all the pseudo-ids to avoid doing 1000s of lookups during import
+            if json_id not in self.pseudo_id_cache:
+                spec = get_pseudo_id(json_id)
+                # need to include other_names in matching
+                if list(spec.keys()) == ["name"]:
+                    name = spec["name"]
+                    spec = (
+                        Q(name__iexact=name)
+                        | Q(other_names__name__iexact=name)
+                    )
+                else:
+                    spec = Q(**spec)
+
+                # continue with logic from resolve_json_id
+                if isinstance(spec, Q):
+                    objects = self.model_class.objects.filter(spec)
+                else:
+                    objects = self.model_class.objects.filter(**spec)
+                ids = {each.id for each in objects}
+                if len(ids) == 1:
+                    self.pseudo_id_cache[json_id] = ids.pop()
+                    errmsg = None
+                elif not ids:
+                    errmsg = "cannot resolve pseudo id to {}: {}".format(
+                        self.model_class.__name__, json_id
+                    )
+                else:
+                    errmsg = "multiple objects returned for {} pseudo id {}: {}".format(
+                        self.model_class.__name__, json_id, ids
+                    )
+
+                # either raise or log error
+                if errmsg:
+                    if not allow_no_match:
+                        raise UnresolvedIdError(errmsg)
+                    else:
+                        self.error(errmsg)
+                        self.pseudo_id_cache[json_id] = None
+
+            # return the cached object
+            return self.pseudo_id_cache[json_id]
+
+        # get the id that the duplicate points to, or use self
+        json_id = self.duplicates.get(json_id, json_id)
+
+        try:
+            return self.json_to_db_id[json_id]
+        except KeyError:
+            raise UnresolvedIdError("cannot resolve id: {}".format(json_id))

openstates/importers/events.py

@@ -90,7 +90,7 @@ def prepare_for_db(self, data: _JsonDict) -> _JsonDict:
             if "person_id" in participant:
                 participant["person_id"] = self.resolve_person(participant["person_id"])
             elif "organization_id" in participant:
-                participant["organization_id"] = self.org_importer.resolve_json_id(
+                participant["organization_id"] = self.resolve_committee(
                     participant["organization_id"], allow_no_match=True
                 )
 

pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "openstates"
-version = "6.18.2"
+version = "6.18.3"
 description = "core infrastructure for the openstates project"
 authors = ["James Turk <[email protected]>"]
 license = "MIT"