Initial fork commit

This commit is contained in:
2025-09-06 22:03:29 -04:00
commit 2d238cd339
1748 changed files with 932506 additions and 0 deletions

View File

@@ -0,0 +1,2 @@
admin_codes:
admin1: state

View File

@@ -0,0 +1,3 @@
admin_codes:
admin1: state
admin2: state_district

View File

@@ -0,0 +1,3 @@
admin_codes:
admin1: state
# admin2 is a mix of state_district and city, need to list specifically

View File

@@ -0,0 +1,3 @@
admin_codes:
admin1: state
# admin2 is a mix of state_district and city, need to list specifically

View File

@@ -0,0 +1,3 @@
admin_codes:
admin1: state
admin2: city

View File

@@ -0,0 +1,3 @@
admin_codes:
admin1: state
# unclear what admin2 is, maybe city

View File

@@ -0,0 +1,3 @@
admin_codes:
admin1: state
admin2: state_district

View File

@@ -0,0 +1,3 @@
admin_codes:
admin1: state
admin2: city

View File

@@ -0,0 +1,3 @@
admin_codes:
admin1: state
admin2: city

View File

@@ -0,0 +1,3 @@
admin_codes:
admin1: state
admin2: state_district

View File

@@ -0,0 +1,3 @@
admin_codes:
admin1: state
admin2: state_district

View File

@@ -0,0 +1,4 @@
admin_codes:
# The GeoNames admin1 boundaries are admin_level=5 or 6 in OSM
# However, they do appear to be states, might need to update Czech OSM config
admin1: state_district

View File

@@ -0,0 +1,3 @@
admin_codes:
admin1: state
admin2: state_district

View File

@@ -0,0 +1,3 @@
admin_codes:
admin1: state
# admin2 is a mix of city and island, need to list specifically

View File

@@ -0,0 +1,3 @@
admin_codes:
admin1: state
admin2: city

View File

@@ -0,0 +1,3 @@
admin_codes:
admin1: state
admin2: state_district

View File

@@ -0,0 +1,3 @@
admin_codes:
admin1: state
admin2: state_district

View File

@@ -0,0 +1,5 @@
admin_codes:
# The GeoNames admin1 boundaries are admin_level=6 in OSM
# However, they do appear to be states, might need to update Finnish OSM config
admin1: state_district
admin2: state_district

View File

@@ -0,0 +1,3 @@
admin_codes:
admin1: state
admin2: city

View File

@@ -0,0 +1,3 @@
admin_codes:
admin1: state
admin2: state_district

View File

@@ -0,0 +1,3 @@
admin_codes:
admin1: state
admin2: state_district

View File

@@ -0,0 +1,3 @@
admin_codes:
admin1: state
admin2: state_district

View File

@@ -0,0 +1,2 @@
admin_codes:
admin1: city

View File

@@ -0,0 +1,3 @@
admin_codes:
admin1: state
# admin2 is a mix of city and city_district, need to list specifically

View File

@@ -0,0 +1,4 @@
admin_codes:
# The GeoNames admin1 boundaries are admin_level=6 in OSM
# However, they do appear to be states, might need to update Hungary OSM config
admin1: state_district

View File

@@ -0,0 +1,3 @@
admin_codes:
admin1: state
admin2: state_district

View File

@@ -0,0 +1,2 @@
admin_codes:
admin1: city

View File

@@ -0,0 +1,3 @@
admin_codes:
admin1: state
admin2: state_district

View File

@@ -0,0 +1,3 @@
admin_codes:
admin1: state
admin2: city

View File

@@ -0,0 +1,3 @@
admin_codes:
admin1: state
admin2: state_district

View File

@@ -0,0 +1,2 @@
admin_codes:
admin1: state

View File

@@ -0,0 +1,3 @@
admin_codes:
admin1: state
# admin2 is a mix of state_district and city, need to list specifically

View File

@@ -0,0 +1,2 @@
admin_codes:
admin1: city

View File

@@ -0,0 +1,3 @@
admin_codes:
admin1: state
admin2: state_district

View File

@@ -0,0 +1,3 @@
admin_codes:
admin1: state
# admin2 is a mix of state_district and city, need to list specifically

View File

@@ -0,0 +1,4 @@
admin_codes:
# The admin1 names don't appear to exist in OSM, but would be states otherwise
admin1: state
admin2: state_district

View File

@@ -0,0 +1,2 @@
admin_codes:
admin1: state_district

View File

@@ -0,0 +1,2 @@
admin_codes:
admin1: state_district

View File

@@ -0,0 +1,2 @@
admin_codes:
admin1: city

View File

@@ -0,0 +1,3 @@
admin_codes:
admin1: state
admin2: state_district

View File

@@ -0,0 +1,3 @@
admin_codes:
admin1: state
admin2: state_district

View File

@@ -0,0 +1,3 @@
admin_codes:
admin1: state
admin2: city

View File

@@ -0,0 +1,3 @@
admin_codes:
admin1: state
admin2: city

View File

@@ -0,0 +1,3 @@
admin_codes:
admin1: state
admin2: state_district

View File

@@ -0,0 +1,3 @@
admin_codes:
admin1: country_region
# admin2 is a mix of state_district and city, need to list specifically

View File

@@ -0,0 +1,3 @@
admin_codes:
admin1: state
admin2: state_district

View File

@@ -0,0 +1,3 @@
admin_codes:
admin1: state
admin2: state_district

View File

@@ -0,0 +1,6 @@
admin_codes:
admin1: state_district
# The notion of a "barrio" in the official sense in PR is not quite a
# municipality, and has no current official purpose, but might be useful
# to have the name + "barrio" version available in libpostal
admin2: city

View File

@@ -0,0 +1,8 @@
admin_codes:
admin1: state_district
admin2: city
overrides:
id:
"2593105": "state" # Madeira
"3411865": "state" # Azores

View File

@@ -0,0 +1,4 @@
admin_codes:
admin1: state
# These are mostly admin_level=6, which maps to city in OSM
admin2: city

View File

@@ -0,0 +1,3 @@
admin_codes:
admin1: state
admin2: state_district

View File

@@ -0,0 +1,3 @@
admin_codes:
admin1: state
admin2: city

View File

@@ -0,0 +1,22 @@
admin_codes:
admin1: city
overrides:
id:
# Districts of Ljubljana (suburbs in OSM)
"3196350": "suburb" # Opština Ljubljana-Vič-Rudnik
"3196352": "suburb" # Opština [historical] Ljubljana-Šiška
"3196355": "suburb" # Opština Ljubljana-Moste-Polje
"3196356": "suburb" # Opština Ljubljana-Center
"3196357": "suburb" # Opčina Ljubljana-Bežigrad
"9794374": "suburb" # Črnuče District
"9794375": "suburb" # Dravlje District
"9794376": "suburb" # Golovec District
"9794377": "suburb" # Jarše District
"9794378": "suburb" # Posavje District
"9794379": "suburb" # Rožnik District
"9794380": "suburb" # Sostro District
"9794381": "suburb" # Šentvid District
"9794382": "suburb" # Šmarna Gora District
"9794384": "suburb" # Trnovo District
"9794386": "suburb" # Vič District

View File

@@ -0,0 +1,17 @@
admin_codes:
admin1: state
# admin2 is a mix of state_district and city, need to list specifically
admin2: state_district
overrides:
id:
# Districts of Bratislava
"8986283": "city_district" # Okres Bratislava I
"8986339": "city_district" # Okres Bratislava II
"8986340": "city_district" # Okres Bratislava III
"8986341": "city_district" # Okres Bratislava IV
"8986342": "city_district" # Okres Bratislava V
# Districts of Košice
"8986335": "city_district" # Košice I
"8986336": "city_district" # Košice II
"8986337": "city_district" # Košice III
"8986338": "city_district" # Košice IV

View File

@@ -0,0 +1,2 @@
admin_codes:
admin1: city

View File

@@ -0,0 +1,8 @@
admin_codes:
admin1: state
overrides:
id:
# Bangkok the state is treated as a city
# Note: we do this in OSM to get the boundary, so duplicate in GeoNames
"1609348": "city"

View File

@@ -0,0 +1,3 @@
admin_codes:
admin1: state
admin2: state_district

View File

@@ -0,0 +1,16 @@
admin_codes:
admin1: state
admin2: state_district
overrides:
id:
# Manhattan (Island)
"8479493": "city_district"
# Brooklyn
"5110300": "city_district"
# Bronx
"5110266": "city_district"
# Queens
"5133266": "city_district"
# Staten Island
"5139568": "city_district"

View File

@@ -0,0 +1,2 @@
admin_codes:
admin1: city

View File

@@ -0,0 +1,3 @@
admin_codes:
admin1: state
admin2: state_district

View File

@@ -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

View File

@@ -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

View File

@@ -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

View File

@@ -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

View File

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

View File

@@ -0,0 +1,37 @@
prefixes:
- ōaza
- ō-aza
- oaza
- aza
- koaza
suffixes:
- ken
- to
-
- do
- fu
- gun
- shi
- machi
- chō
- cho
- mura
- ku
- aza
- oaza
- koaza
suffixes_no_whitespace:
- # ken
- # to
- # dō
- # fu
- # gun
- # shi
- # machi
- # mura
- # ku
- # aza
- 大字 # ōaza
- 小字 # koaza

View File

@@ -0,0 +1,5 @@
---
admin_level:
"2": "country"
"7": "state"
"8": "city"

View File

@@ -0,0 +1,5 @@
---
admin_level:
"2": "country"
"4": "state"

View File

@@ -0,0 +1,5 @@
---
admin_level:
"2": "country"
"4": "state"
"6": "state_district"

View File

@@ -0,0 +1,3 @@
---
admin_level:
"2": "country"

View File

@@ -0,0 +1,3 @@
---
admin_level:
"2": "country"

View File

@@ -0,0 +1,6 @@
---
admin_level:
"2": "country"
"6": "state"
"7": "city"
"8": "city"

View File

@@ -0,0 +1,5 @@
---
admin_level:
"2": "country"
"4": "state"
"8": "city"

View File

@@ -0,0 +1,4 @@
---
admin_level:
"2": "country"
"4": "state"

View File

@@ -0,0 +1,27 @@
---
admin_level:
"2": "country"
"4": "state"
"5": "state_district"
"6": "state_district"
"7": "city"
"8": "city"
"9": "suburb"
"10": "suburb"
overrides:
id:
relation:
# Buenos Aires (state boundary coterminous with city)
"3082668": null
contained_by:
relation:
# Buenos Aires
"1224652":
admin_level:
"5": "city_district"
"9": "suburb"
# Rosario
"3594027":
admin_level:
"9": "city_district"

View File

@@ -0,0 +1,5 @@
---
admin_level:
"4": "country"
"6": "state_district"
"8": "city"

View File

@@ -0,0 +1,15 @@
---
admin_level:
"2": "country"
"4": "state"
"6": "state_district"
"8": "city"
"9": "city_district"
"10": "suburb"
overrides:
id:
relation:
# Vienna
"109166": "city"

View File

@@ -0,0 +1,43 @@
---
admin_level:
"2": "country"
"4": "state"
"5": "state_district"
"6": "state_district"
"7": "city"
"9": "suburb"
"10": "suburb"
overrides:
id:
relation:
# City of Sydney
"1251066": "city"
# Sydney (suburb)
"5729534": "city"
# City of Melbourne
"2404870": "city"
# Melbourne (suburb)
"2383266": "city"
# City of Adelaide
"3093825": "city"
# Adelaide (suburb)
"3114051": "city"
# City of Darwin
"5515884": "city"
# Canberra City
"5342034": "city"
# City of Wollongong
"6303077": "city"
# Wollongong (locality)
"6076763": "city"
# City of Newcastle
"6189831": "city"
# Newcastle (suburb)
"5989933": "city"
# city of Hobart
"6269038": "city"
# City of Greater Geelong
"2646730": "city"
# Geelong (suburb)
"2456176": "city"

View File

@@ -0,0 +1,3 @@
---
admin_level:
"3": "country"

View File

@@ -0,0 +1,5 @@
---
admin_level:
"3": "country"
"8": "state"

View File

@@ -0,0 +1,5 @@
---
admin_level:
"2": "country"
"4": "state"

View File

@@ -0,0 +1,10 @@
---
admin_level:
"2": "country"
"4": "state"
"5": "state_district"
"6": "city"
"7": "city"
"8": "city"
"9": "city_district"

View File

@@ -0,0 +1,4 @@
---
admin_level:
"2": "country"
"6": "state"

View File

@@ -0,0 +1,11 @@
---
admin_level:
"2": "country"
"3": "state"
"4": "state"
"5": "state_district"
"6": "state_district"
"7": "city"
"8": "city"
"9": "city"

View File

@@ -0,0 +1,21 @@
---
admin_level:
"2": "country"
"4": "country_region"
"6": "state"
"7": "state_district"
"8": "city"
"9": "city"
overrides:
id:
relation:
# Brussels-Capital Region
"54094": "state"
contained_by:
relation:
# Antwerpen
"59518":
admin_level:
"9": "city_district"

View File

@@ -0,0 +1,8 @@
---
admin_level:
"2": "country"
"4": "state"
"5": "state_district"
"6": "state_district"
"8": "city"
"9": "city"

View File

@@ -0,0 +1,9 @@
---
admin_level:
"2": "country"
"6": "state"
"7": "city"
"8": "city"
"9": "city_district"
"10": "suburb"

View File

@@ -0,0 +1,4 @@
---
admin_level:
"2": "country"
"4": "state"

View File

@@ -0,0 +1,5 @@
---
admin_level:
"2": "country"
"4": "state"
"6": "state_district"

View File

@@ -0,0 +1,5 @@
---
admin_level:
"2": "country"
"4": "state"
"6": "city"

View File

@@ -0,0 +1,3 @@
---
admin_level:
"3": "country"

View File

@@ -0,0 +1,6 @@
---
admin_level:
"2": "country"
"6": "state"
"8": "city"

View File

@@ -0,0 +1,6 @@
---
admin_level:
"2": "country"
"6": "state"
"9": "city"
"10": "city"

View File

@@ -0,0 +1,12 @@
---
admin_level:
"2": "country"
"4": "state"
"5": "state_district"
"6": "state_district"
"7": "state_district"
"8": "city"
"9": "city_district"
"10": "city_district"
"11": "suburb"

View File

@@ -0,0 +1,3 @@
---
admin_level:
"4": "country"

View File

@@ -0,0 +1,21 @@
---
admin_level:
"2": "country"
"3": "country_region"
"4": "state"
"5": "state_district"
"6": "state_district"
"7": "state_district"
"8": "city"
"9": "city_district"
"10": "suburb"
overrides:
contained_by:
relation:
# Brasilia
"2758138":
place:
"neighbourhood": null
"neighborhood": null
global_overrides_last: true

View File

@@ -0,0 +1,5 @@
---
admin_level:
"2": "country"
"6": "state"
"8": "state"

View File

@@ -0,0 +1,4 @@
---
admin_level:
"2": "country"
"4": "state"

View File

@@ -0,0 +1,3 @@
---
admin_level:
"4": "country"

View File

@@ -0,0 +1,4 @@
---
admin_level:
"2": "country"
"4": "state"

Some files were not shown because too many files have changed in this diff Show More