[addresses] Using YAML inheritance instead of baking it into the config parser

resources/addresses/en.yaml

@@ -10,6 +10,7 @@
 # country overrides section. Each country can create its own copy of the entire top-level
 # structure and it will be recursively merged with the defaults.
 
+default: &default
     # Number
     # ======
     # Number, No., #, etc. can be used in both floor and apartment numbers,
@@ -1175,6 +1176,7 @@ units:
 countries:
     # United States
     us:
+        <<: *default
         levels:
             storey: &story
                 canonical: story
@@ -1261,6 +1263,7 @@ countries:
     # Canada
     # Specifically Canadian English. If the address is in French it will use fr.yaml
     ca:
+        <<: *default
         levels:
             # Note: Canadian English uses "storey" keeping with the British convention, so no need to change that
 
@@ -1288,6 +1291,7 @@ countries:
             combined_probability: 0.1
     # Australia
     au:
+        <<: *default
         po_boxes: &australia_po_boxes
             alphanumeric:
                 default: *po_box
@@ -1330,6 +1334,7 @@ countries:
 
     # New Zealand - same rules as Australia
     nz:
+        <<: *default
         po_boxes: *australia_po_boxes
         units: *australia_unit_types
 

scripts/geodata/addresses/config.py

@@ -7,9 +7,9 @@ import yaml
 from collections import Mapping
 
 from geodata.address_expansions.address_dictionaries import address_phrase_dictionaries
+from geodata.configs.utils import nested_get, DoesNotExist, recursive_merge
 from geodata.math.sampling import cdf, check_probability_distribution
 
-
 this_dir = os.path.realpath(os.path.dirname(__file__))
 
 ADDRESS_CONFIG_DIR = os.path.join(this_dir, os.pardir, os.pardir, os.pardir,
@@ -19,35 +19,6 @@ DICTIONARIES_DIR = os.path.join(this_dir, os.pardir, os.pardir, os.pardir,
                                 'resources', 'dictionaries')
 
 
-def recursive_merge(a, b):
-    for k, v in six.iteritems(b):
-        if isinstance(v, Mapping):
-            existing = a.get(k, v)
-            merged = recursive_merge(existing, v)
-            a[k] = merged
-        else:
-            a[k] = b[k]
-    return a
-
-
-class DoesNotExist:
-    pass
-
-
-def nested_get(obj, keys):
-    if len(keys) == 0:
-        return obj
-    try:
-        for key in keys[:-1]:
-            obj = obj.get(key, {})
-            if not hasattr(obj, 'items'):
-                return DoesNotExist
-        key = keys[-1]
-        return obj.get(key, DoesNotExist)
-    except AttributeError:
-        return DoesNotExist
-
-
 class AddressConfig(object):
     def __init__(self, config_dir=ADDRESS_CONFIG_DIR, dictionaries_dir=DICTIONARIES_DIR):
         self.address_configs = {}
@@ -58,17 +29,14 @@ class AddressConfig(object):
                 continue
 
             config = yaml.load(open(os.path.join(ADDRESS_CONFIG_DIR, filename)))
+            default = config['default']
             countries = config.pop('countries', {})
 
-            for k in countries.keys():
+            if countries:
-                country_config = countries[k]
+                default['countries'] = countries
-                config_copy = copy.deepcopy(config)
-                countries[k] = recursive_merge(config_copy, country_config)
-
-            config['countries'] = countries
 
             lang = filename.strip('.yaml')
-            self.address_configs[lang] = config
+            self.address_configs[lang] = default
 
         self.sample_phrases = {}
 
@@ -87,6 +55,7 @@ class AddressConfig(object):
             if country_config:
                 config = country_config
 
+
         value = nested_get(config, keys)
         if value is not DoesNotExist:
             return value