Merge branch 'master' into patch-1

2025-02-08 12:42:51 -05:00
parent d881c82def 845a810599
commit d3383736e0
65 changed files with 561 additions and 369 deletions
--- a/resources/addresses/de.yaml
+++ b/resources/addresses/de.yaml
@@ -63,10 +63,23 @@ numbers:


 house_numbers:
+    gebaude: &gebaude
+        canonical: gebäude
+        abbreviated: geb
+        sample: true
+        canonical_probability: 0.5
+        abbreviated_probability: 0.5
+        sample_probability: 0.05
+        numeric:
+            direction: left
    alphanumeric:
        default: *nummer
+        probability: 0.95
+        alternatives:
+            - alternative: *gebaude
+              probability: 0.05

-    alphanumeric_phrase_probability: 0.0001
+    alphanumeric_phrase_probability: 0.05

 conscription_numbers:
    alphanumeric:
--- a/resources/addresses/uk.yaml
+++ b/resources/addresses/uk.yaml
@@ -49,7 +49,7 @@ numbers:


 house_numbers:
-    budnyok: &budnyok
+    budynok: &budynok
        canonical: будинок
        abbreviated: буд
        sample: true
@@ -58,8 +58,8 @@ house_numbers:
        sample_probability: 0.1
        numeric:
            direction: left
-    budnyok_latin: &budnyok_latin
-        canonical: budnyok
+    budynok_latin: &budynok_latin
+        canonical: budynok
        abbreviated: bud
        sample: true
        canonical_probability: 0.6
@@ -88,10 +88,10 @@ house_numbers:
            direction: left

    alphanumeric:
-        default: *budnyok
+        default: *budynok
        probability: 0.65
        alternatives:
-            - alternative: *budnyok_latin
+            - alternative: *budynok_latin
              probability: 0.05
            - alternative: *dom
              probability: 0.25
--- a/resources/boundaries/osm/ar.yaml
+++ b/resources/boundaries/osm/ar.yaml
@@ -11,8 +11,9 @@

    overrides:
        id:
-            # Buenos Aires (state boundary coterminous with city)
-            "3082668": null
+            relation:
+                # Buenos Aires (state boundary coterminous with city)
+                "3082668": null
        contained_by:
            relation:
                # Buenos Aires
--- a/resources/boundaries/osm/tw.yaml
+++ b/resources/boundaries/osm/tw.yaml
@@ -11,9 +11,9 @@
    "9": "suburb"

  overrides:
-    id:
-      relation:
-        # Taiwan Province
-        "3777248": "state"
-        # Fujian Province
-        "3777250": "state"
+      id:
+           relation:
+               # Taiwan Province
+               "3777248": "state"
+               # Fujian Province
+               "3777250": "state"
--- a/resources/dictionaries/en/street_types.txt
+++ b/resources/dictionaries/en/street_types.txt
@@ -132,6 +132,7 @@ falls|fls
 fare
 farm|frm
 farms|frms
+farm to market|fm|farm-to-market
 fern
 ferry|fry|fy
 field|fld|fd
@@ -407,4 +408,4 @@ well|wl
 wells|wls
 wharf|whrf|whf
 wynd|wyn
-yard|yd|yrd
+yard|yd|yrd
--- a/resources/dictionaries/hi/building_types.txt
+++ b/resources/dictionaries/hi/building_types.txt
@@ -0,0 +1,5 @@
+mandir|मन्दिर|मंदिर
+station
+police station
+post office
+office
--- a/resources/dictionaries/hi/directionals.txt
+++ b/resources/dictionaries/hi/directionals.txt
@@ -0,0 +1,4 @@
+dakshin|दक्षिण
+uttar|उत्तर
+poorva|poorav|पूर्व
+paschim|पश्चिम
--- a/resources/dictionaries/hi/people.txt
+++ b/resources/dictionaries/hi/people.txt
@@ -0,0 +1 @@
+mahatma gandhi|mg|m g
--- a/resources/dictionaries/hi/personal_titles.txt
+++ b/resources/dictionaries/hi/personal_titles.txt
@@ -0,0 +1,24 @@
+baba
+babu
+bhagat
+guru
+jagirdar
+maharaja|maharaj
+mahatma|महात्मा
+pandit
+raja
+rajarshi
+rajkumar
+rajkumari
+rani
+rishi
+sahib
+sant
+sardar
+senapati
+shah
+shrimati|smt|srimathi|श्रीमती
+shri|shree|sri|श्री
+sushri
+swami
+ustad
--- a/resources/dictionaries/hi/qualifiers.txt
+++ b/resources/dictionaries/hi/qualifiers.txt
@@ -0,0 +1,3 @@
+nagar|नगर
+colony
+cantonment|cantt
--- a/resources/dictionaries/hi/street_types.txt
+++ b/resources/dictionaries/hi/street_types.txt
@@ -1,3 +1,4 @@
 bazaar|bazar
 marg
-nagar
+flyover
+रोड
--- a/resources/dictionaries/hu/ambiguous_expansions.txt
+++ b/resources/dictionaries/hu/ambiguous_expansions.txt
@@ -3,4 +3,4 @@ d
 e
 k
 n
-u
+u
--- a/resources/dictionaries/hu/level_types_mezzanine.txt
+++ b/resources/dictionaries/hu/level_types_mezzanine.txt
@@ -1,2 +1,2 @@
-félemelet|felemelet
-magasföldszint|magasfoldszint
+félemelet|felemelet|félem|1/2 em|1/2em
+magasföldszint|magasfoldszint|mgfszt|mgfsz|mfszt|mfsz
--- a/resources/dictionaries/hu/personal_titles.txt
+++ b/resources/dictionaries/hu/personal_titles.txt
@@ -1 +1,2 @@
-szent|szt
+szent|szt
+idősebb|id
--- a/resources/dictionaries/hu/street_types.txt
+++ b/resources/dictionaries/hu/street_types.txt
@@ -1,21 +1,34 @@
 árok|arok
-dűlő|dulo
+dűlő|dulo|d.|d
 fasor
+fasora
+főközlekedési út|főút|fout
+határút|hatarut
+kapu
 kert
 körönd|korond|krnd
+körvasútsor|korvasutsor
 körút|korut|krt
 köz|koz
+lakótelep|lakotelep|ltp.|ltp
 lejtő|lejto
 lépcső|lepcso
 liget
 mező|mezo
+országút|orszagut
 park
-rakpart|rpt
-sétány|setany
-sor
-sugárút|sugarut
+parkja
+rakpart|rkpt|rkp|rpt
+sétány|setany|stny.|stny
+sor|s.|s
+sétány|setany|sét
+sugárút|sugarut|sgrt.|sgrt|srt.|srt|sgt.|sgt
+sziget
+telep
 tér|ter
 tere
-utca|u
-út|ut
-útja|utja
+tanya|t.|t
+udvar
+utca|u.|u
+út|ut|u.|u
+útja|utja
--- a/resources/dictionaries/ms/toponyms.txt
+++ b/resources/dictionaries/ms/toponyms.txt
@@ -0,0 +1,3 @@
+kuala lumpur|federal territory kuala lumpur|federal territory of kuala lumpur|wilayah persekutuan kuala lumpur|kl
+labuan|federal territory labuan|federal territory of labuan|wilayah persekutuan labuan
+putrajaya|federal territory putrajaya|federal territory of putrajaya|wilayah persekutuan putrajaya
--- a/resources/dictionaries/pl/street_types.txt
+++ b/resources/dictionaries/pl/street_types.txt
@@ -1,12 +1,18 @@
 aleja|al
 autostrada
 boczna
-bulwar
+bulwar|bulw
 droga
 obwodnica
+ogród
+osiedle|os
+park
 plac|pl
 rondo
 rynek
+skwer
 szosa
 ulica|ul
-zaulek
+wybrzeże|wyb
+wyspa
+zaulek
--- a/resources/dictionaries/pt/street_types.txt
+++ b/resources/dictionaries/pt/street_types.txt
@@ -10,10 +10,10 @@ calçada|calcada|cc
 calçadinha|caclcadinha|ccnh
 câmara municipal|camara municipal|cm|c.m.|c. m.
 caminho|cam|camno
-direito|dto
+direito|dto|dt
 esquerdo|esq
 estrada|estr
-astrada marginal|estr marg
+estrada marginal|estr marg
 estrada municipal|em|e m|estr m
 estrada nacional|en|e n|estr n
 estrada regional|er|e r|estr r
@@ -50,4 +50,4 @@ viaduto|vd|vdto
 viela|ve
 vila|vl
 volta
-zona|zn
+zona|zn
--- a/resources/dictionaries/ro/building_types.txt
+++ b/resources/dictionaries/ro/building_types.txt
@@ -0,0 +1,6 @@
+anexa
+bloc|blc|bl
+casa
+cladirea|cladire
+complex
+garaj
--- a/resources/dictionaries/ro/company_types.txt
+++ b/resources/dictionaries/ro/company_types.txt
@@ -0,0 +1,5 @@
+banca
+organizatie neguvernamentala|ong
+societate comerciala|sc
+societate cu raspundere limitata|srl
+societate pe actiuni|sa
--- a/resources/dictionaries/ro/cross_streets.txt
+++ b/resources/dictionaries/ro/cross_streets.txt
@@ -1,5 +1,5 @@
 &
-colț|colt
+colț|colt|colț cu|colt cu
 între|intre
 la colțul de pe|la coltul de pe
-și|si
+și|si
--- a/resources/dictionaries/ro/entrances.txt
+++ b/resources/dictionaries/ro/entrances.txt
@@ -1 +1 @@
-intrare
+intrare|intrarea
--- a/resources/dictionaries/ro/near.txt
+++ b/resources/dictionaries/ro/near.txt
@@ -4,4 +4,4 @@ din
 in apropiere de
 în apropiere|in apropiere
 în jurul aici|in jurul aici
-lângă mine|langa mine
+lângă mine|langa mine|lângă|langa
--- a/resources/dictionaries/ro/number.txt
+++ b/resources/dictionaries/ro/number.txt
@@ -1 +1 @@
-număr|numar|nr|nº|n°|#|№|no
+număr|numar|nr|nº|n°|#|№|no|numarul|numărul
--- a/resources/dictionaries/ro/personal_titles.txt
+++ b/resources/dictionaries/ro/personal_titles.txt
@@ -8,7 +8,8 @@ general|gen
 major|maj
 locotenent
 locotenent colonel
+pictor
 profesor|prof
 sergent
 sublocotenent
-vice amiral
+vice amiral
--- a/resources/dictionaries/ro/place_names.txt
+++ b/resources/dictionaries/ro/place_names.txt
@@ -0,0 +1,3 @@
+cinema
+cafenea
+fabrica
--- a/resources/dictionaries/ro/qualifiers.txt
+++ b/resources/dictionaries/ro/qualifiers.txt
@@ -1 +1,7 @@
-bloc|bl
+bloc|bl
+cartier|cartierul
+comuna|comunā
+kilometrul|kilometru|km
+sat|satul
+sector|sectorul|sect
+zona
--- a/resources/dictionaries/ro/stopwords.txt
+++ b/resources/dictionaries/ro/stopwords.txt
@@ -1,2 +1,3 @@
 și|si|&
-cel
+cel
+intre
--- a/resources/dictionaries/ro/street_types.txt
+++ b/resources/dictionaries/ro/street_types.txt
@@ -1,13 +1,13 @@
 aleea|ale|alea|al
 bulevardul|bd|bul|bdul|blv|blvd|b-dul|b.dul|bulev|bulevardu|bulevard
-calea|cal
-drumul
+calea|cale|cal
+drumul|drum
 fundătura|fundatura|fnd
 fundacul|fdc
 intrarea|int|intr
 piaţa|piata|piață|pta|pţa|p-ta|p-ţa
-strada|str
+strada|str|st
 stradela|str-la|sdla
 șoseaua|soseaua|sos|șos
 splaiul|sp|spl
-vârful|varful|virful|vîrful|varf|vf
+vârful|varful|virful|vîrful|varf|vf
--- a/resources/dictionaries/ro/synonyms.txt
+++ b/resources/dictionaries/ro/synonyms.txt
@@ -0,0 +1 @@
+decembrie|dec
--- a/resources/dictionaries/ro/unit_types_numbered.txt
+++ b/resources/dictionaries/ro/unit_types_numbered.txt
@@ -1,4 +1,8 @@
-apartament|ap|apt|apart
+apartamentul|apartament|ap|apt|apart
 birou
+cladire|cladirea|clădire|clădirea
+corp|corpul
+complex
+interior|int
 lotul
-sală|sala
+sală|sala
--- a/resources/dictionaries/ru/entrances.txt
+++ b/resources/dictionaries/ru/entrances.txt
@@ -1,2 +1,4 @@
 вход
-vkhod
+vkhod
+подъезд
+pod'ezd
--- a/resources/dictionaries/uk/qualifiers.txt
+++ b/resources/dictionaries/uk/qualifiers.txt
@@ -6,3 +6,5 @@ kvartal|kvart|kv|kv-l
 oblast|obl
 район|р-н
 raion|r-n
+місто|міс|м
+misto|mis|m
--- a/resources/states/my.yaml
+++ b/resources/states/my.yaml
@@ -0,0 +1,93 @@
+"KL":
+    en: Kuala Lumpur
+    ms: Kuala Lumpur
+"federal territory kuala lumpur":
+    en: Kuala Lumpur
+    ms: Kuala Lumpur
+"federal territory of kuala lumpur":
+    en: Kuala Lumpur
+    ms: Kuala Lumpur
+"wilayah persekutuan kuala lumpur":
+    en: Kuala Lumpur
+    ms: Kuala Lumpur
+"federal territory labuan":
+    en: Labuan
+    ms: Labuan
+"federal territory of labuan":
+    en: Labuan
+    ms: Labuan
+"wilayah persekutuan labuan":
+    en: Labuan
+    ms: Labuan
+"federal territory putrajaya":
+    en: Putrajaya
+    ms: Putrajaya
+"federal territory of putrajaya":
+    en: Putrajaya
+    ms: Putrajaya
+"wilayah persekutuan putrajaya":
+    en: Putrajaya
+    ms: Putrajaya
+"pulau pinang":
+    en: Penang
+    ms: Pulau Pinang
+"penang":
+    en: Penang
+    ms: Pulau Pinang
+JHR:
+    en: Johor
+    ms: Johor
+KDH:
+    en: Kedah
+    ms: Kedah
+KTN:
+    en: Kelantan
+    ms: Kelantan
+MLK:
+    en: Melaka
+    ms: Melaka
+NSN:
+    en: Negeri Sembilan
+    ms: Negeri Sembilan
+PHG:
+    en: Pahang
+    ms: Pahang
+PRK:
+    en: Perak
+    ms: Perak
+PLS:
+    en: Perlis
+    ms: Perlis
+PNG:
+    en: Penang
+    ms: Pulau Pinang
+SBH:
+    en: Sabah
+    ms: Sabah
+SWK:
+    en: Sarawak
+    ms: Sarawak
+SGR:
+    en: Selangor
+    ms: Selangor
+TRG:
+    en: Terengganu
+    ms: Terengganu
+KUL:
+    en: Kuala Lumpur
+    ms: Kuala Lumpur
+LBN:
+    en: Labuan
+    ms: Labuan
+PJY:
+    en: Putrajaya
+    ms: Putrajaya
+KL:
+    en: Kuala Lumpur
+    ms: Kuala Lumpur
+LB:
+    en: Labuan
+    ms: Labuan
+PY:
+    en: Putrajaya
+    ms: Putrajaya