[places] adding world_region tag and adding the phrase West Indies with small random probability for English-speaking Caribbean nations. Ref: #113

resources/places/countries/global.yaml

@@ -24,8 +24,23 @@ global:
             probability: 0.05
         country:
             probability: 0.8
+        world_region:
+            probability: 0.001
+
+
+add_west_indies: &add_west_indies
+    components:
+        world_region: &add_west_indies_world_region
+            probability: 0.1
+            value: West Indies
 
 countries:
+    # Anguilla
+    ai: *add_west_indies
+
+    # Antigua and Barbuda
+    ag: *add_west_indies
+
     # Australia
     au:
         components:
@@ -34,12 +49,21 @@ countries:
             country:
                 probability: 0.5
 
+    # Barbados
+    bb: *add_west_indies
+
+    # Bermuda
+    bm: *add_west_indies
+
     # Brazil
     br:
         components:
             state:
                 probability: 0.6
 
+    # Bahamas
+    bs: *add_west_indies
+
     # Canada
     ca:
         components:
@@ -48,6 +72,9 @@ countries:
             country:
                 probability: 0.2
 
+    # Dominica
+    dm: *add_west_indies
+
     # France
     fr:
         components:
@@ -64,6 +91,11 @@ countries:
             country:
                 probability: 0.6
 
+
+    # Grenada
+    gd: *add_west_indies
+
+
     # Indonesia
     id:
         components:
@@ -84,6 +116,9 @@ countries:
                       type: relation
                       probability: 0.0
 
+    # Jamaica
+    jm: *add_west_indies
+
     # Japan
     jp:
         components:
@@ -110,6 +145,16 @@ countries:
         components:
             island:
                 probability: 0.8
+            world_region: *add_west_indies_world_region
+
+    # Cayman Islands
+    ky: *add_west_indies
+
+    # Saint Lucia
+    lc: *add_west_indies
+
+    # Montserrat
+    ms: *add_west_indies
 
     # Mexico
     mx:
@@ -134,6 +179,15 @@ countries:
             island:
                 probability: 0.8
 
+    # Sint Maarten
+    sx: *add_west_indies
+
+    # Turks and Caicos
+    tc: *add_west_indies
+
+    # Trinidad and Tobago
+    tt: *add_west_indies
+
     # United States
     us:
         # Definitions
@@ -234,14 +288,19 @@ countries:
             island:
                 probability: 0.8
 
+    # Saint Vincent and the Grenadines
+    vc: *add_west_indies
+
     # US Virgin Islands
     vi:
         components:
             island:
                 probability: 0.8
+            world_region: *add_west_indies_world_region
 
     # British Virgin Islands
     vg:
         components:
             island:
                 probability: 0.8
+            world_region: *add_west_indies_world_region

scripts/geodata/address_formatting/formatter.py

@@ -75,6 +75,7 @@ class AddressFormatter(object):
     POSTCODE = 'postcode'
     COUNTRY_REGION = 'country_region'
     COUNTRY = 'country'
+    WORLD_REGION = 'world_region'
 
     component_order = {k: i for i, k in enumerate([
         CATEGORY,
@@ -102,6 +103,7 @@ class AddressFormatter(object):
         POSTCODE,
         COUNTRY_REGION,
         COUNTRY,
+        WORLD_REGION,
     ])}
 
     BOUNDARY_COMPONENTS_ORDERED = [
@@ -115,6 +117,7 @@ class AddressFormatter(object):
         STATE,
         COUNTRY_REGION,
         COUNTRY,
+        WORLD_REGION,
     ]
 
     BOUNDARY_COMPONENTS = set(BOUNDARY_COMPONENTS_ORDERED)
@@ -153,6 +156,7 @@ class AddressFormatter(object):
             ('county', STATE_DISTRICT),
             ('state_code', STATE),
             ('country_name', COUNTRY),
+            ('continent', WORLD_REGION),
             ('postal_code', POSTCODE),
             ('post_code', POSTCODE),
         ])

scripts/geodata/places/config.py

@@ -30,6 +30,7 @@ class PlaceConfig(object):
         AddressFormatter.STATE,
         AddressFormatter.COUNTRY_REGION,
         AddressFormatter.COUNTRY,
+        AddressFormatter.WORLD_REGION,
     }
 
     numeric_ops = (('lte', operator.le),
@@ -139,6 +140,13 @@ class PlaceConfig(object):
                     value = components[component]
                     for c in names[value]:
                         new_components.pop(c, None)
+
+        for component in self.ADMIN_COMPONENTS:
+            value = self.get_property(('components', component, 'value'), country=country, default=None)
+
+            if value is not None and component not in components and self.include_component(component, containing_ids, country=country, population=population):
+                new_components[component] = value
+
         return new_components