Merge pull request #3542 from lonvia/remove-legacy-tokenizer

[nominatim.git] / settings / country_settings.yaml

diff --git a/settings/country_settings.yaml b/settings/country_settings.yaml
index 684b0e44e949332e59c0d03e8e9c43332ed19c53..667684c670a0dacb46b1d052f4a4f82547c3ed83 100644 (file)
--- a/settings/country_settings.yaml
+++ b/settings/country_settings.yaml
@@ -3,6 +3,9 @@ ad:
     partition: 35
     languages: ca
     names: !include country-names/ad.yaml
+    postcode:
+      pattern: "(ddd)"
+      output: AD\1
 
 
 # United Arab Emirates (الإمارات العربية المتحدة)
@@ -35,6 +38,9 @@ ai:
     partition: 175
     languages: en
     names: !include country-names/ai.yaml
+    postcode:
+      pattern: "2640"
+      output: AI-2640
 
 
 # Albania (Shqipëria)
@@ -55,13 +61,6 @@ am:
       pattern: "dddd"
 
 
-# Netherlands Antilles (De Nederlandse Antillen)
-an:
-    partition: 58
-    languages: nl, en, pap
-    names: !include country-names/an.yaml
-
-
 # Angola (Angola)
 ao:
     partition: 85
@@ -70,13 +69,6 @@ ao:
     postcode: no
 
 
-#  (Antarctica)
-aq:
-    partition: 181
-    languages: en, es, fr, ru
-    names: !include country-names/aq.yaml
-
-
 # Argentina (Argentina)
 ar:
     partition: 39
@@ -86,13 +78,6 @@ ar:
       pattern: "l?dddd(?:lll)?"
 
 
-#  (American Samoa)
-as:
-    partition: 182
-    languages: en, sm
-    names: !include country-names/as.yaml
-
-
 # Austria (Österreich)
 at:
     partition: 245
@@ -111,26 +96,13 @@ au:
       pattern: "dddd"
 
 
-#  (Aruba)
-aw:
-    partition: 183
-    languages: nl, pap
-    names: !include country-names/aw.yaml
-    postcode: no
-
-
-#  (Aland Islands)
-ax:
-    partition: 184
-    languages: sv
-    names: !include country-names/ax.yaml
-
-
 # Azerbaijan (Azərbaycan)
 az:
     partition: 119
     languages: az
     names: !include country-names/az.yaml
+    postcode:
+      pattern: "dddd"
 
 
 # Bosnia and Herzegovina (Bosna i Hercegovina / Босна и Херцеговина)
@@ -147,6 +119,9 @@ bb:
     partition: 206
     languages: en
     names: !include country-names/bb.yaml
+    postcode:
+      pattern: "(ddddd)"
+      output: BB\1
 
 
 # Bangladesh (Bangladesh)
@@ -209,18 +184,14 @@ bj:
     postcode: no
 
 
-#  (Saint Barthélemy)
-bl:
-    partition: 204
-    languages: fr
-    names: !include country-names/bl.yaml
-
-
 # Bermuda (Bermuda)
 bm:
     partition: 176
     languages: en
     names: !include country-names/bm.yaml
+    postcode:
+      pattern: "(ll)[ -]?(dd)"
+      output: \1 \2
 
 
 # Brunei (Brunei)
@@ -228,6 +199,9 @@ bn:
     partition: 86
     languages: ms
     names: !include country-names/bn.yaml
+    postcode:
+      pattern: "(ll) ?(dddd)"
+      output: \1\2
 
 
 # Bolivia (Bolivia)
@@ -238,18 +212,14 @@ bo:
     postcode: no
 
 
-# Caribbean Netherlands (Caribisch Nederland)
-bq:
-    partition: 250
-    languages: nl
-    names: !include country-names/bq.yaml
-
-
 # Brazil (Brasil)
 br:
     partition: 121
     languages: pt
     names: !include country-names/br.yaml
+    postcode:
+      pattern: "(ddddd)-?(ddd)"
+      output: \1-\2
 
 
 # The Bahamas (The Bahamas)
@@ -269,13 +239,6 @@ bt:
       pattern: "ddddd"
 
 
-#  (Bouvet Island)
-bv:
-    partition: 185
-    languages: "no"
-    names: !include country-names/bv.yaml
-
-
 # Botswana (Botswana)
 bw:
     partition: 122
@@ -306,13 +269,9 @@ ca:
     partition: 244
     languages: en, fr
     names: !include country-names/ca.yaml
-
-
-# Cocos (Keeling) Islands (Cocos (Keeling) Islands)
-cc:
-    partition: 118
-    languages: en
-    names: !include country-names/cc.yaml
+    postcode:
+      pattern: "(ldl) ?(dld)"
+      output: \1 \2
 
 
 # Democratic Republic of the Congo (République démocratique du Congo)
@@ -426,20 +385,6 @@ cv:
       pattern: "dddd"
 
 
-# Curaçao (Curaçao)
-cw:
-    partition: 248
-    languages: nl, en
-    names: !include country-names/cw.yaml
-
-
-# Christmas Island (Christmas Island)
-cx:
-    partition: 177
-    languages: en
-    names: !include country-names/cx.yaml
-
-
 # Cyprus (Κύπρος - Kıbrıs)
 cy:
     partition: 114
@@ -454,6 +399,9 @@ cz:
     partition: 124
     languages: cs
     names: !include country-names/cz.yaml
+    postcode:
+      pattern: "(ddd) ?(dd)"
+      output: \1 \2
 
 
 # Germany (Deutschland)
@@ -513,6 +461,8 @@ ec:
     partition: 78
     languages: es
     names: !include country-names/ec.yaml
+    postcode:
+      pattern: "dddddd"
 
 
 # Estonia (Eesti)
@@ -606,6 +556,8 @@ fo:
     partition: 10
     languages: fo, da
     names: !include country-names/fo.yaml
+    postcode:
+      pattern: "ddd"
 
 
 # France (France)
@@ -630,6 +582,9 @@ gb:
     partition: 1
     languages: en
     names: !include country-names/gb.yaml
+    postcode:
+      pattern: "(l?ld[A-Z0-9]?) ?(dll)"
+      output: \1 \2
 
 
 # Grenada (Grenada)
@@ -649,18 +604,14 @@ ge:
       pattern: "dddd"
 
 
-# French Guiana (Guyane Française)
-gf:
-    partition: 231
-    languages: fr
-    names: !include country-names/gf.yaml
-
-
 # Guernsey (Guernsey)
 gg:
     partition: 77
     languages: en
     names: !include country-names/gg.yaml
+    postcode:
+      pattern: "(GYdd?) ?(dll)"
+      output: \1 \2
 
 
 # Ghana (Ghana)
@@ -677,6 +628,9 @@ gi:
     partition: 138
     languages: en
     names: !include country-names/gi.yaml
+    postcode:
+      pattern: "(GX11) ?(1AA)"
+      output: GX11 1AA
 
 
 # Greenland (Kalaallit Nunaat)
@@ -705,13 +659,6 @@ gn:
       pattern: "ddd"
 
 
-# Guadeloupe (Guadeloupe)
-gp:
-    partition: 232
-    languages: fr
-    names: !include country-names/gp.yaml
-
-
 # Equatorial Guinea (Guinea Ecuatorial)
 gq:
     partition: 12
@@ -725,6 +672,9 @@ gr:
     partition: 22
     languages: el
     names: !include country-names/gr.yaml
+    postcode:
+      pattern: "(ddd) ?(dd)"
+      output: \1 \2
 
 
 # South Georgia and the South Sandwich Islands (South Georgia and the South Sandwich Islands)
@@ -732,6 +682,9 @@ gs:
     partition: 44
     languages: en
     names: !include country-names/gs.yaml
+    postcode:
+      pattern: "(SIQQ) ?(1ZZ)"
+      output: \1 \2
 
 
 # Guatemala (Guatemala)
@@ -743,13 +696,6 @@ gt:
       pattern: "ddddd"
 
 
-# Guam (Guam)
-gu:
-    partition: 187
-    languages: en, ch
-    names: !include country-names/gu.yaml
-
-
 # Guinea-Bissau (Guiné-Bissau)
 gw:
     partition: 8
@@ -767,20 +713,6 @@ gy:
     postcode: no
 
 
-#  (Hong Kong)
-hk:
-    partition: 188
-    languages: zh-hant, en
-    names: !include country-names/hk.yaml
-
-
-#  (Heard Island and MaxDonald Islands)
-hm:
-    partition: 189
-    languages: en
-    names: !include country-names/hm.yaml
-
-
 # Honduras (Honduras)
 hn:
     partition: 56
@@ -804,6 +736,8 @@ ht:
     partition: 29
     languages: fr, ht
     names: !include country-names/ht.yaml
+    postcode:
+      pattern: "dddd"
 
 
 # Hungary (Magyarország)
@@ -829,6 +763,9 @@ ie:
     partition: 46
     languages: en, ga
     names: !include country-names/ie.yaml
+    postcode:
+      pattern: "(ldd) ?([0123456789ACDEFHKNPRTVWXY]{4})"
+      output: \1 \2
 
 
 # Israel (ישראל)
@@ -845,6 +782,9 @@ im:
     partition: 190
     languages: en
     names: !include country-names/im.yaml
+    postcode:
+      pattern: "(IMdd?) ?(dll)"
+      output: \1 \2
 
 
 # India (India)
@@ -852,6 +792,9 @@ in:
     partition: 128
     languages: hi, en
     names: !include country-names/in.yaml
+    postcode:
+      pattern: "(ddd) ?(ddd)"
+      output: \1\2
 
 
 # British Indian Ocean Territory (British Indian Ocean Territory)
@@ -859,6 +802,9 @@ io:
     partition: 13
     languages: en
     names: !include country-names/io.yaml
+    postcode:
+      pattern: "(BBND) ?(1ZZ)"
+      output: \1 \2
 
 
 # Iraq (العراق)
@@ -875,6 +821,9 @@ ir:
     partition: 80
     languages: fa
     names: !include country-names/ir.yaml
+    postcode:
+      pattern: "(ddddd)[-_ ]?(ddddd)"
+      output: \1-\2
 
 
 # Iceland (Ísland)
@@ -900,6 +849,9 @@ je:
     partition: 123
     languages: en
     names: !include country-names/je.yaml
+    postcode:
+      pattern: "(JEdd?) ?(dll)"
+      output: \1 \2
 
 
 # Jamaica (Jamaica)
@@ -907,6 +859,7 @@ jm:
     partition: 214
     languages: en
     names: !include country-names/jm.yaml
+    postcode: no
 
 
 # Jordan (الأردن)
@@ -923,6 +876,9 @@ jp:
     partition: 11
     languages: ja
     names: !include country-names/jp.yaml
+    postcode:
+      pattern: "(ddd)-?(dddd)"
+      output: \1-\2
 
 
 # Kenya (Kenya)
@@ -1008,6 +964,9 @@ ky:
     partition: 38
     languages: en
     names: !include country-names/ky.yaml
+    postcode:
+      pattern: "(d)-(dddd)"
+      output: KY\1-\2
 
 
 # Kazakhstan (Қазақстан)
@@ -1033,6 +992,8 @@ lb:
     partition: 66
     languages: ar, fr
     names: !include country-names/lb.yaml
+    postcode:
+      pattern: "(dddd)(?: ?dddd)?"
 
 
 # Saint Lucia (Saint Lucia)
@@ -1040,6 +1001,9 @@ lc:
     partition: 146
     languages: en
     names: !include country-names/lc.yaml
+    postcode:
+      pattern: "(dd) ?(ddd)"
+      output: LC\1 \2
 
 
 # Liechtenstein (Liechtenstein)
@@ -1083,6 +1047,8 @@ lt:
     partition: 67
     languages: lt
     names: !include country-names/lt.yaml
+    postcode:
+      pattern: "ddddd"
 
 
 # Luxembourg (Lëtzebuerg)
@@ -1099,6 +1065,9 @@ lv:
     partition: 162
     languages: lv
     names: !include country-names/lv.yaml
+    postcode:
+      pattern: "(dddd)"
+      output: LV-\1
 
 
 # Libya (ليبيا)
@@ -1132,6 +1101,9 @@ md:
     partition: 147
     languages: ro, ru, uk
     names: !include country-names/md.yaml
+    postcode:
+      pattern: "(dddd)"
+      output: MD-\1
 
 
 # Montenegro (Crna Gora / Црна Гора)
@@ -1143,13 +1115,6 @@ me:
       pattern: "ddddd"
 
 
-# Saint Martin (Saint Martin)
-mf:
-    partition: 203
-    languages: fr
-    names: !include country-names/mf.yaml
-
-
 # Madagascar (Madagasikara)
 mg:
     partition: 164
@@ -1203,28 +1168,6 @@ mn:
       pattern: "ddddd"
 
 
-# Macao (Macao)
-mo:
-    partition: 191
-    languages: zh-hant, pt
-    names: !include country-names/mo.yaml
-    postcode: no
-
-
-# Northern Mariana Islands (Northern Mariana Islands)
-mp:
-    partition: 192
-    languages: ch, en
-    names: !include country-names/mp.yaml
-
-
-# Martinique (Martinique)
-mq:
-    partition: 233
-    languages: fr
-    names: !include country-names/mq.yaml
-
-
 # Mauritania (موريتانيا)
 mr:
     partition: 149
@@ -1245,6 +1188,9 @@ mt:
     partition: 165
     languages: mt, en
     names: !include country-names/mt.yaml
+    postcode:
+      pattern: "(lll) ?(dddd)"
+      output: \1 \2
 
 
 # Mauritius (Mauritius)
@@ -1296,6 +1242,8 @@ mz:
     partition: 98
     languages: pt
     names: !include country-names/mz.yaml
+    postcode:
+      pattern: "(dddd)(?:-dd)?"
 
 
 # Namibia (Namibia)
@@ -1307,13 +1255,6 @@ na:
       pattern: "ddddd"
 
 
-# New Caledonia (Nouvelle-Calédonie)
-nc:
-    partition: 234
-    languages: fr
-    names: !include country-names/nc.yaml
-
-
 # Niger (Niger)
 ne:
     partition: 226
@@ -1323,13 +1264,6 @@ ne:
       pattern: "dddd"
 
 
-# Norfolk Island (Norfolk Island)
-nf:
-    partition: 100
-    languages: en, pih
-    names: !include country-names/nf.yaml
-
-
 # Nigeria (Nigeria)
 ng:
     partition: 218
@@ -1353,6 +1287,9 @@ nl:
     partition: 63
     languages: nl
     names: !include country-names/nl.yaml
+    postcode:
+      pattern: "(dddd) ?(ll)"
+      output: \1 \2
 
 
 # Norway (Norge)
@@ -1425,13 +1362,6 @@ pe:
       pattern: "ddddd"
 
 
-# French Polynesia (Polynésie française)
-pf:
-    partition: 202
-    languages: fr
-    names: !include country-names/pf.yaml
-
-
 # Papua New Guinea (Papua Niugini)
 pg:
     partition: 71
@@ -1464,13 +1394,9 @@ pl:
     partition: 168
     languages: pl
     names: !include country-names/pl.yaml
-
-
-# Saint Pierre and Miquelon (Saint-Pierre-et-Miquelon)
-pm:
-    partition: 236
-    languages: fr
-    names: !include country-names/pm.yaml
+    postcode:
+      pattern: "(dd)[ -]?(ddd)"
+      output: \1-\2
 
 
 # Pitcairn Islands (Pitcairn Islands)
@@ -1478,13 +1404,9 @@ pn:
     partition: 113
     languages: en, pih
     names: !include country-names/pn.yaml
-
-
-# Puerto Rico (Puerto Rico)
-pr:
-    partition: 193
-    languages: es, en
-    names: !include country-names/pr.yaml
+    postcode:
+      pattern: "(PCRN) ?(1ZZ)"
+      output: \1 \2
 
 
 # Palestinian Territory (Palestinian Territory)
@@ -1501,6 +1423,8 @@ pt:
     partition: 34
     languages: pt
     names: !include country-names/pt.yaml
+    postcode:
+      pattern: "dddd(?:-ddd)?"
 
 
 # Palau (Belau)
@@ -1529,13 +1453,6 @@ qa:
     postcode: no
 
 
-#  (Réunion)
-re:
-    partition: 235
-    languages: fr
-    names: !include country-names/re.yaml
-
-
 # Romania (România)
 ro:
     partition: 170
@@ -1576,6 +1493,8 @@ sa:
     partition: 52
     languages: ar
     names: !include country-names/sa.yaml
+    postcode:
+      pattern: "ddddd(?:-dddd)?"
 
 
 # Solomon Islands (Solomon Islands)
@@ -1608,6 +1527,9 @@ se:
     partition: 112
     languages: sv
     names: !include country-names/se.yaml
+    postcode:
+      pattern: "(ddd) ?(dd)"
+      output: \1 \2
 
 
 # Singapore (Singapore)
@@ -1624,6 +1546,9 @@ sh:
     partition: 196
     languages: en
     names: !include country-names/sh.yaml
+    postcode:
+      pattern: "(ASCN|STHL|TDCU) ?(1ZZ)"
+      output: \1 \2
 
 
 # Slovenia (Slovenija)
@@ -1635,18 +1560,14 @@ si:
       pattern: "dddd"
 
 
-#  (Svalbard and Jan Mayen)
-sj:
-    partition: 197
-    languages: "no"
-    names: !include country-names/sj.yaml
-
-
 # Slovakia (Slovensko)
 sk:
     partition: 172
     languages: sk
     names: !include country-names/sk.yaml
+    postcode:
+      pattern: "(ddd) ?(dd)"
+      output: \1 \2
 
 
 # Sierra Leone (Sierra Leone)
@@ -1680,6 +1601,9 @@ so:
     partition: 154
     languages: so, ar
     names: !include country-names/so.yaml
+    postcode:
+      pattern: "(ll) ?(ddddd)"
+      output: \1 \2
 
 
 # Suriname (Suriname)
@@ -1715,13 +1639,6 @@ sv:
       pattern: "dddd"
 
 
-#  (Sint Maarten)
-sx:
-    partition: 249
-    languages: nl, en
-    names: !include country-names/sx.yaml
-
-
 # Syria (سوريا)
 sy:
     partition: 104
@@ -1744,6 +1661,9 @@ tc:
     partition: 106
     languages: en
     names: !include country-names/tc.yaml
+    postcode:
+      pattern: "(TKCA) ?(1ZZ)"
+      output: \1 \2
 
 
 # Chad (Tchad تشاد)
@@ -1754,13 +1674,6 @@ td:
     postcode: no
 
 
-# French Southern Lands (Terres australes et antarctiques françaises)
-tf:
-    partition: 132
-    languages: fr
-    names: !include country-names/tf.yaml
-
-
 # Togo (Togo)
 tg:
     partition: 243
@@ -1860,6 +1773,8 @@ tw:
     partition: 25
     languages: zh-hant
     names: !include country-names/tw.yaml
+    postcode:
+      pattern: "ddd(?:ddd?)?"
 
 
 # Tanzania (Tanzania)
@@ -1877,7 +1792,7 @@ ua:
     languages: uk
     names: !include country-names/ua.yaml
     postcode:
-      pattern: "ddddd"
+      pattern: "d?ddddd"
 
 
 # Uganda (Uganda)
@@ -1888,20 +1803,13 @@ ug:
     postcode: no
 
 
-#  (United States Minor Outlying Islands)
-um:
-    partition: 198
-    languages: en
-    names: !include country-names/um.yaml
-    postcode:
-      pattern: "96898"
-
-
 # United States (United States)
 us:
     partition: 2
     languages: en
     names: !include country-names/us.yaml
+    postcode:
+      pattern: "ddddd"
 
 
 # Uruguay (Uruguay)
@@ -1936,6 +1844,9 @@ vc:
     partition: 171
     languages: en
     names: !include country-names/vc.yaml
+    postcode:
+      pattern: "(dddd)"
+      output: VC\1
 
 
 # Venezuela (Venezuela)
@@ -1952,13 +1863,9 @@ vg:
     partition: 109
     languages: en
     names: !include country-names/vg.yaml
-
-
-#  (United States Virgin Islands)
-vi:
-    partition: 199
-    languages: en
-    names: !include country-names/vi.yaml
+    postcode:
+      pattern: "(dddd)"
+      output: VG\1
 
 
 # Vietnam (Việt Nam)
@@ -1978,13 +1885,6 @@ vu:
     postcode: no
 
 
-# Wallis and Futuna Islands (Wallis-et-Futuna)
-wf:
-    partition: 238
-    languages: fr
-    names: !include country-names/wf.yaml
-
-
 # Samoa (Sāmoa)
 ws:
     partition: 131
@@ -2009,13 +1909,6 @@ ye:
     postcode: no
 
 
-# Mayotte (Mayotte)
-yt:
-    partition: 200
-    languages: fr
-    names: !include country-names/yt.yaml
-
-
 # South Africa (South Africa)
 za:
     partition: 76