[boundaries] adding use_admin_center to boundary configs right alongside other overrides

resources/boundaries/osm/co.yaml

@@ -1,17 +1,21 @@
 ---
-  admin_level: 
+    admin_level: 
-    "2": "country"
+        "2": "country"
-    "3": "country_region"
+        "3": "country_region"
-    "4": "state"
+        "4": "state"
-    "5": "state_district"
+        "5": "state_district"
-    "6": "city"
+        "6": "city"
-    "7": "city_district"
+        "7": "city_district"
-    "8": "city_district"
+        "8": "city_district"
-    "9": "suburb"
+        "9": "suburb"
-
-
     overrides:
         id:
             relation:
                 # Bogotá
                 "1387968": "city"
+
+        use_admin_center:
+            - id: 1387968 # Bogotá
+              type: relation
+              # Occasionally use properties/names from the state boundary
+              probability: 0.8

resources/boundaries/osm/lu.yaml

@@ -1,15 +1,21 @@
 ---
-  admin_level: 
+    admin_level: 
-    "2": "country"
+        "2": "country"
-    "4": "state"
+        "4": "state"
-    "5": "state_district"
+        "5": "state_district"
-    "6": "state_district"
+        "6": "state_district"
-    "8": "city"
+        "8": "city"
-    "9": "suburb"
+        "9": "suburb"
 
     overrides:
         contained_by:
-                    relation:
+            relation:
-                        "407489":
+                # Luxembourg City
-                            admin_level:
+                "407489":
-                                "9": "city_district"
+                    admin_level:
+                        "9": "city_district"
+
+        use_admin_center:
+            - id: 407489 # Luxembourg City
+              type: relation
+              probability: 0.6

resources/boundaries/osm/mx.yaml

@@ -1,20 +1,25 @@
 ---
-  admin_level: 
+    admin_level: 
-    "2": "country"
+        "2": "country"
-    "4": "state"
+        "4": "state"
-    "6": "state_district"
+        "6": "state_district"
-    "8": "city"
+        "8": "city"
-    "9": "city_district"
+        "9": "city_district"
-    "10": "city_district"
+        "10": "city_district"
-
 
     overrides:
         id:
             relation:
+                # Mexico City
                 "1376330": "city"
 
         contained_by:
             relation:
                 "1376330":
                     admin_level:
-                        "6": "city_district"
+                        "6": "city_district"
+
+        use_admin_center:
+            - id: 1376330 # Mexico City
+              type: relation
+              probability: 0.6

resources/boundaries/osm/ne.yaml

@@ -8,4 +8,4 @@
         id:
             relation:
                 # Niamey
-                "3218804": "city"
+                "3218804": "city"

resources/boundaries/osm/nz.yaml

@@ -1,8 +1,76 @@
 ---
-  admin_level: 
+    admin_level: 
-    "2": "country"
+        "2": "country"
-    "4": "state"
+        "4": "state"
-    "6": "state_district"
+        "6": "state_district"
-    "7": "city_district"
+        "8": "city_district"
-    "8": "city_district"
+
+    overrides:
+        id:
+            relation:
+                # Auckland
+                "2094141": "city"
+                # Wellington
+                "4266321": "city"
+                # Christchurch
+                "1640161": "city"
+                # Hamilton
+                "2595124": "city"
+                # Dunedin
+                "4426790": "city"
+                # Tauranga
+                "4266815": "city"
+                # Lower Hutt
+                "4266371": "city"
+                # Palmerston North
+                "3440209": "city"
+                # Napier
+                "2731991": "city"
+                # Porirua
+                "4266370": "city"
+                # Invercargill
+                "1656388": "city"
+                # Nelson
+                "4266962": "city"
+                # Upper Hutt
+                "4266375": "city"
+
+        use_admin_center:
+            - id: 2094141 # Auckland
+              type: relation
+            - id: 4266321 # Wellington
+              type: relation
+              probability: 0.7
+            - id: 1640161 # Christchurch
+              type: relation
+              probability: 0.7
+            - id: 2595124 # Hamilton
+              type: relation
+              probability: 0.7
+            - id: 4426790 # Dunedin
+              type: relation
+              probability: 0.7
+            - id: 4266815 # Tauranga
+              type: relation
+              probability: 0.7
+            - id: 4266371 # Lower Hutt
+              type: relation
+              probability: 0.7
+            - id: 3440209 # Palmerston North
+              type: relation
+              probability: 0.7
+            - id: 2731991 # Napier
+              type: relation
+              probability: 0.7
+            - id: 4266370 # Porirua
+              type: relation
+              probability: 0.7
+            - id: 1656388 # Invercargill
+              type: relation
+              probability: 0.7
+            - id: 4266962 # Nelson
+              type: relation
+            - id: 4266375 # Upper Hutt
+              type: relation
+              probability: 0.7
 

resources/boundaries/osm/th.yaml

@@ -9,10 +9,16 @@
     overrides:
         id:
             relation:
+                # Bangkok
                 "92277": "city"
         contained_by:
             relation:
                 "92277":
                     admin_level:
                         "6": "city_district"
-                        "8": "suburb"
+                        "8": "suburb"
+
+        use_admin_center:
+            - id: 92277 # Bangkok
+              type: relation
+              probability: 0.85

resources/parser/default.yaml

@@ -109,12 +109,6 @@ category:
 
 boundaries:
     abbreviate_toponym_probability: 0.35
-    # OSM relations should inherit properties from their admin_center
-    override_with_admin_center:
-        - id: 92277 # Bangkok, Thailand
-          type: relation
-        - id: 1387968 # Bogotá, Colombia
-          type: relation
 
 neighborhood:
     # Usually in Germany, may have e.g. name:prefix=Ortsteil

scripts/geodata/osm/components.py

@@ -8,6 +8,8 @@ from copy import deepcopy
 from geodata.address_formatting.formatter import AddressFormatter
 from geodata.configs.utils import recursive_merge
 
+from geodata.encoding import safe_encode
+
 this_dir = os.path.realpath(os.path.dirname(__file__))
 
 OSM_BOUNDARIES_DIR = os.path.join(this_dir, os.pardir, os.pardir, os.pardir,
@@ -61,18 +63,23 @@ class OSMAddressComponents(object):
     def __init__(self, boundaries_dir=OSM_BOUNDARIES_DIR):
         self.config = {}
 
+        self.use_admin_center = {}
+
         for filename in os.listdir(boundaries_dir):
             if not filename.endswith('.yaml'):
                 continue
 
             country_code = filename.rsplit('.yaml', 1)[0]
             data = yaml.load(open(os.path.join(boundaries_dir, filename)))
+
             for prop, values in six.iteritems(data):
                 for k, v in values.iteritems():
                     if isinstance(v, six.string_types) and v not in AddressFormatter.address_formatter_fields:
                         raise ValueError(u'Invalid value in {} for prop={}, key={}: {}'.format(filename, prop, k, v))
 
                 if prop == 'overrides':
+                    self.use_admin_center.update({(r['type'], safe_encode(r['id'])): r.get('probability', 1.0) for r in values.get('use_admin_center', [])})
+
                     containing_overrides = values.get('contained_by', {})
 
                     if not containing_overrides: