Add city field to research_organizations table

src/alexandria3k/data_sources/ror.py

@@ -190,9 +190,6 @@ def __init__(self, name, **kwargs):
             ColumnMeta("name", lambda row: row["name"]),
             ColumnMeta("status", lambda row: row["status"]),
             ColumnMeta("established", lambda row: row["established"]),
-            ColumnMeta(
-                "country_code", lambda row: row["country"]["country_code"]
-            ),
             # Although deprecated, we are adding it as an additional organization identifier, as
             # it provides useful to determine the ground truth data. Some organizations may not
             # have a GRID identifier, so we need to make sure it doesn't raise any errors.
@@ -202,6 +199,23 @@ def __init__(self, name, **kwargs):
                 .get("GRID", {})
                 .get("all"),
             ),
+            # Each research organization has only 1 address. They have been folded into this table.
+            # This is a simplified address schema. Add more field when ROR settles it.
+            ColumnMeta(
+                "address_city", lambda row: row["addresses"][0]["city"]
+            ),
+            ColumnMeta(
+                "address_state", lambda row: row["addresses"][0]["state"]
+            ),
+            ColumnMeta(
+                "address_postcode", lambda row: row["addresses"][0]["postcode"]
+            ),
+            ColumnMeta(
+                "address_country_code",
+                lambda row: row["country"]["country_code"],
+            ),
+            ColumnMeta("address_lat", lambda row: row["addresses"][0]["lat"]),
+            ColumnMeta("address_lng", lambda row: row["addresses"][0]["lng"]),
         ],
     ),
     RorDetailsTableMeta(
@@ -240,19 +254,6 @@ def __init__(self, name, **kwargs):
             ColumnMeta("ror_path", lambda row: row["id"][16:]),
         ],
     ),
-    RorDetailsTableMeta(
-        "ror_addresses",
-        extract_multiple=lambda row: row["addresses"],
-        columns=[
-            # ROR will simplify the current address schema.
-            # Add more fields when ROR settles it.
-            ColumnMeta("lat", lambda row: row["lat"]),
-            ColumnMeta("lng", lambda row: row["lng"]),
-            ColumnMeta("city", lambda row: row["city"]),
-            ColumnMeta("state", lambda row: row["state"]),
-            ColumnMeta("postcode", lambda row: row["postcode"]),
-        ],
-    ),
     # OrgRef is deprecated, so we are not supporting this field
     RorDetailsTableMeta(
         "ror_funder_ids",

tests/data_sources/test_ror.py

@@ -70,16 +70,26 @@ def test_organization(self):
             name,
             status,
             established,
-            country_code,
             grid,
+            city,
+            state,
+            postcode,
+            country_code,
+            latitude,
+            longitude,
         ) = result.fetchone()
         self.assertEqual(id, 0)
         self.assertEqual(ror_path, "019wvm592")
         self.assertEqual(name, "Australian National University")
         self.assertEqual(status, "active")
         self.assertEqual(established, 1946)
-        self.assertEqual(country_code, "AU")
         self.assertEqual(grid, "grid.1001.0")
+        self.assertEqual(city, "Canberra")
+        self.assertEqual(state, "Australian Capital Territory")
+        self.assertEqual(postcode, None)
+        self.assertEqual(country_code, "AU")
+        self.assertEqual(latitude, -35.2778)
+        self.assertEqual(longitude, 149.1205)
 
     def test_blank_external_ids(self):
         result = TestRorPopulate.cursor.execute(
@@ -92,16 +102,26 @@ def test_blank_external_ids(self):
             name,
             status,
             established,
-            country_code,
             grid,
+            city,
+            state,
+            postcode,
+            country_code,
+            latitude,
+            longitude,
         ) = result.fetchone()
         self.assertEqual(id, 28)
         self.assertEqual(ror_path, "02f4ks689")
         self.assertEqual(name, "Axiom Data Science")
         self.assertEqual(status, "active")
         self.assertEqual(established, 2007)
-        self.assertEqual(country_code, "US")
         self.assertEqual(grid, None)
+        self.assertEqual(city, "Anchorage")
+        self.assertEqual(state, None)
+        self.assertEqual(postcode, None)
+        self.assertEqual(country_code, "US")
+        self.assertEqual(latitude, 61.21806)
+        self.assertEqual(longitude, -149.90028)
 
     def test_funder_ids(self):
         result = TestRorPopulate.cursor.execute(
@@ -179,27 +199,6 @@ def test_ror_relationships(self):
             in rows
         )
 
-    def test_ror_addresses(self):
-        result = TestRorPopulate.cursor.execute(
-            """SELECT * FROM ror_addresses WHERE ror_id=(
-                    SELECT id FROM research_organizations WHERE
-                        ror_path='02bfwt286')"""
-        )
-        rows = list(result)
-        self.assertEqual(len(rows), 1)
-        self.assertTrue(
-            (
-                16384,
-                1,
-                -37.9083,
-                145.138,
-                "Melbourne",
-                "Victoria",
-                None,
-            )
-            in rows
-        )
-
     def test_ror_wikidata_ids(self):
         result = TestRorPopulate.cursor.execute(
             """SELECT wikidata_id FROM ror_wikidata_ids WHERE ror_id=(