Initial fork commit

resources/boundaries/names/global.yaml

@@ -0,0 +1,120 @@
+names:
+    keys:
+        default: name
+        probability: 0.75
+        alternatives:
+            - alternative: short_name # e.g. NYC
+              probability: 0.12
+            - alternative: alt_name # e.g. New York (instead of New York City)
+              probability: 0.12
+            - alternative: official_name # e.g. United Kingdom of Great Britain and Northern Ireland
+              probability: 0.01
+
+    components:
+        country:
+            keys:
+                default: name
+                probability: 0.87
+                alternatives:
+                    - alternative: ISO3166-1:alpha2
+                      probability: 0.02
+                    - alternative: ISO3166-1:alpha3
+                      probability: 0.01
+                    - alternative: short_name
+                      probability: 0.04
+                    - alternative: alt_name
+                      probability: 0.04
+                    - alternative: int_name
+                      probability: 0.01
+                    - alternative: official_name # e.g. United Kingdom of Great Britain and Northern Ireland
+                      probability: 0.01
+
+    regex_replacements:
+        - country: fr
+          pattern: "(?:lyon|paris|marseilles?) ([\\d]+(?:e|er|ème|eme) arrondissement)"
+          replace_with_group: 1
+          replace_probability: 0.5
+
+    prefixes:
+        language:
+            ru:
+                city:
+                    default:
+                        prefix: г.
+                    probability: 0.35
+                    alternatives:
+                        - alternative:
+                              prefix: г
+                          probability: 0.1
+                        - alternative:
+                              prefix: город
+                          probability: 0.05
+
+    # This section overrides place names
+    exceptions:
+        # Boroughs of New York City
+        - id: 2552485 # New York County (don't use Manhattan)
+          type: relation
+          default: name # New York County
+          probability: 1.0
+        - id: 369518 # Kings County (don't use Brooklyn)
+          type: relation
+          default: name # Kings County
+          probability: 1.0
+        - id: 369519 # Queens County (don't use Queens)
+          type: relation
+          default: name # Queens County
+          probability: 1.0
+        - id: 2552450 # Bronx County (don't use The Bronx)
+          type: relation
+          default: name # Bronx County
+          probability: 1.0
+        - id: 962876 # Richmond County (don't use Staten Island)
+          type: relation
+          default: name # Richmond County
+          probability: 1.0
+        - id: 6577227 # Kingston Parish (always use Kingston)
+          type: relation
+          default: name # Kingston
+          probability: 1.0
+        - id: 30674098 # Sao Paulo
+          type: node
+          default: name
+          probability: 0.9
+          alternatives:
+              - alternative: alt_name
+                probability: 0.09
+              - alternative: official_name
+                probability: 0.01
+        - id: 298285 # Sao Paulo (relation)
+          type: relation
+          default: name
+          probability: 0.9
+          alternatives:
+              - alternative: alt_name
+                probability: 0.09
+              - alternative: official_name
+                probability: 0.01
+        - id: 556706 # New Zealand
+          type: relation
+          default: name:en
+          probability: 0.77
+          alternatives:
+              - alternative: name
+                probability: 0.1
+              - alternative: ISO3166-1:alpha2
+                probability: 0.02
+              - alternative: ISO3166-1:alpha3
+                probability: 0.01
+              - alternative: short_name
+                probability: 0.04
+              - alternative: alt_name
+                probability: 0.04
+              - alternative: int_name
+                probability: 0.01
+              - alternative: official_name
+                probability: 0.01
+        - id: 2383266 # Melbourne (city center)
+          type: relation
+          default: alt_name # Melbourne
+          probability: 1.0

resources/boundaries/names/languages/de.yaml

@@ -0,0 +1,11 @@
+# Prefixes which can be stripped to normalize a place name
+prefixes:
+    - stadtteil
+    - stadtbezirk
+    - gemeinde
+    - landkreis
+    - kreis
+    - grenze
+    - freistaat
+    - regierungsbezirk
+    - gemeindefreies gebiet

resources/boundaries/names/languages/en.yaml

@@ -0,0 +1,66 @@
+# Prefixes which can be stripped to normalize a place name
+# Note these will be stripped off the front at random, not every time
+prefixes:
+    - city of
+    - city and borough of
+    - city and county of
+    - township of
+    - town of
+    - regional municipality of
+    - district municipality of
+    - rural city of
+    - municipality of
+    - borough of
+    - london borough of
+    - royal borough of
+    - shire of
+    - parish of
+    - free state of
+    - autonomous city of
+    - territorial waters of
+    - county of
+    - county
+
+# Suffixes which can be stripped to normalize a place name
+suffixes:
+    - township
+    - municipality
+    - local municipality
+    - metropolitan municipality
+    - district municipality
+    - regional municipality
+    - cp
+    - civil parish
+    - community development council
+    - cdc
+    - council
+    - borough council
+    - city council
+    - regional council
+
+countries:
+    au:
+        prefixes:
+            - district of
+
+    gb:
+        prefixes:
+            - district of
+
+        suffixes:
+            - city
+
+    ie:
+        suffixes:
+            - city
+            - municipal district
+
+    in:
+        prefixes:
+            - district of
+
+    nz:
+        prefixes:
+            - district of
+        suffixes:
+            - city

resources/boundaries/names/languages/es.yaml

@@ -0,0 +1,10 @@
+prefixes:
+    # Note: "ciudad de" should not be included as it's part of proper names
+    - colonia
+    - municipio nuestra senora de
+
+prefixes_similarity_only:
+    - ciudad de
+
+suffixes:
+    - colonia

resources/boundaries/names/languages/fr.yaml

@@ -0,0 +1,5 @@
+prefixes:
+    - commune de
+
+suffixes:
+    - (eaux territoriales)

resources/boundaries/names/languages/ja.yaml

@@ -0,0 +1,37 @@
+prefixes:
+    - ōaza
+    - ō-aza
+    - oaza
+    - aza
+    - koaza
+
+suffixes:
+    - ken
+    - to
+    - dō
+    - do
+    - fu
+    - gun
+    - shi
+    - machi
+    - chō
+    - cho
+    - mura
+    - ku
+    - aza
+    - oaza
+    - koaza
+
+suffixes_no_whitespace:
+    - 県 # ken
+    - 都 # to
+    - 道 # dō
+    - 府 # fu
+    - 郡 # gun
+    - 市 # shi
+    - 町 # machi
+    - 村 # mura
+    - 区 # ku
+    - 字 # aza
+    - 大字 # ōaza
+    - 小字 # koaza