[formatting] adding non-zero invert probabilities to all the former Soviet states. Other template insertions can still apply afterward for #125

resources/formatting/global.yaml

@@ -8,6 +8,9 @@ global:
         ko_rm: en
         zh_pinyin: en        
 
+    # Former Soviet states sometimes write addresses generai-to-specific
+    invert_probability: 0.0
+
     insertions:
         # For each component, insertions are mutually exclusive
         # They don't have to sum to 1 (especially for components
@@ -857,3 +860,63 @@ countries:
 
     kr_en:
         insertions: *us_template_insertions
+
+    # Russia
+    ru:
+        invert_probability: 0.2
+
+    # Ukraine
+    uk:
+        invert_probability: 0.2
+
+    # Belarus (using inverted by default)
+    by:
+        invert_probability: 0.4
+
+    # Kazakhstan (using inverted by default)
+    kz:
+        invert_probability: 0.4
+
+    # Kyrgyzstan (using inverted by default)
+    kg:
+        invert_probability: 0.4
+
+    # Latvia
+    lv:
+        invert_probability: 0.2
+
+    # Lithuania
+    lt:
+        invert_probability: 0.2
+
+    # Estonia
+    ee:
+        invert_probability: 0.2
+
+    # Armenia
+    am:
+        invert_probability: 0.2
+
+    # Azerbaijan
+    az:
+        invert_probability: 0.2
+
+    # Georgia
+    ge:
+        invert_probability: 0.2
+
+    # Moldova
+    md:
+        invert_probability: 0.2
+
+    # Uzbekistan
+    uz:
+        invert_probability: 0.2
+
+    # Tajikistan
+    tj:
+        invert_probability: 0.2
+
+    # Turkmenistan
+    tm:
+        invert_probability: 0.2

scripts/geodata/address_formatting/formatter.py

@@ -630,8 +630,13 @@ class AddressFormatter(object):
 
         invert_probability = self.country_invert_probabilities.get(country, self.global_invert_probability)
         if random.random() < invert_probability:
-            template = self.inverted(template)
             cache_keys.append('inverted')
+            cache_key = tuple(sorted(cache_keys))
+            if cache_key in self.template_cache:
+                template = self.template_cache[cache_key]
+            else:
+                template = self.inverted(template)
+                self.template_cache[cache_key] = template
 
         for component in sorted(components, key=self.component_order.get):
             scope = country