[osm] Adding method for overriding per-country boundary types (either by id or containing polygon)

scripts/geodata/neighborhoods/reverse_geocode.py

@@ -339,7 +339,7 @@ class NeighborhoodReverseGeocoder(RTreePolygonIndex):
                     if name and name == attrs.get('name'):
                         continue
 
-                    containing_component = osm_components.get_first_component(country, props)
+                    containing_component = osm_components.component_from_properties(country, props)
 
                     if containing_component != AddressFormatter.SUBURB:
                         skip_node = True

scripts/geodata/osm/components.py

@@ -21,7 +21,7 @@ class OSMAddressComponents(object):
 
     ADMIN_LEVEL = 'admin_level'
 
-    # These keys are country-independent
+    # These keys override country-level 
     global_keys = {
         'place': {
             'country': AddressFormatter.COUNTRY,
@@ -55,24 +55,54 @@ class OSMAddressComponents(object):
             data = yaml.load(open(os.path.join(boundaries_dir, filename)))
             for prop, values in six.iteritems(data):
                 for k, v in values.iteritems():
-                    if v not in AddressFormatter.address_formatter_fields:
+                    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))
             self.config[country_code] = data
 
-    def get_component(self, country, prop, value):
-        if prop in self.global_keys:
-            props = self.global_keys[prop]
-        else:
+    def component(self, country, prop, value):
+        props = self.global_keys.get(prop, {})
+        if not props:
             props = self.config.get(country, {}).get(prop, {})
         return props.get(value, None)
 
-    def get_first_component(self, country, properties):
-        for k, v in six.iteritems(props):
-            containing_component = self.get_component(country, k, v)
-            break
-        else:
-            containing_component = None
-        return containing_component
+    def component_from_properties(self, country, properties, containing=()):
+        country_config = self.config.get(country, {})
 
+        config = country_config
+
+        overrides = country_config.get('overrides')
+        if overrides:
+            id_overrides = overrides.get('id', {})
+            element_type = properties.get('type')
+            element_id = properties.get('id')
+
+            override_value = id_overrides.get(element_type, {}).get(six.binary_type(element_id or ''), None)
+            if override_value:
+                return override_value
+
+            contained_by_overrides = overrides.get('contained_by')
+            if contained_by_overrides and containing:
+                # Note, containing should be passed in from smallest to largest
+                for containing_type, containing_id in containing:
+                    config_updates = contained_by_overrides.get(containing_type, {}).get(six.binary_type(containing_id or ''), None)
+                    print contained_by_overrides
+                    if config_updates:
+                        config.update(config_updates)
+                        break
+
+        for k, v in six.iteritems(properties):
+            containing_component = self.global_keys.get(k, {}).get(v, None)
+
+            if containing_component is not None:
+                return containing_component
+
+        for k, v in six.iteritems(properties):
+            if containing_component is None:
+                containing_component = config.get(k, {}).get(v, None)
+
+            if containing_component is not None:
+                return containing_component
+
+        return None
 
 osm_address_components = OSMAddressComponents()