atlas_engine 0.1.0

data/app/countries/atlas_engine/au/synonyms.yml

@@ -0,0 +1,209 @@
+city_synonyms:
+- saint, st
+- mount, mt
+street_synonyms:
+- ANT, Antenna
+- APT, Apartment
+- ATM, Automated Teller Machine
+- BBQ, Barbecue
+- BTSD, Boatshed
+- BLDG, Building
+- BNGW, Bungalow
+- CAGE, Cage
+- CARP, Carpark
+- CARS, Carspace
+- CLUB, Club
+- COOL, Coolroom
+- CTGE, Cottage
+- DUPL, Duplex
+- FCTY, Factory
+- FLAT, Flat
+- GRGE, Garage
+- HALL, Hall
+- HSE, House
+- KSK, Kiosk
+- LSE, Lease
+- LBBY, Lobby
+- LOFT, Loft
+- LOT, Lot
+- MT, Mount
+- MSNT, Maisonette
+- MBTH, Marine Berth
+- OFFC, Office
+- RESV, Reserve
+- ROOM, Room
+- SHED, Shed
+- SHOP, Shop
+- SHRM, Showroom
+- SIGN, Sign
+- SITE, Site
+- STLL, Stall
+- STOR, Store
+- STR, Strata unit
+- STU, Studio/studio apartment
+- SUBS, Substation
+- SE, Suite
+- TNCY, Tenancy
+- TWR, Tower
+- TNHS, Townhouse
+- UNIT, Unit
+- VLT, Vault
+- VLLA, Villa
+- WARD, Ward
+- WHSE, Warehouse
+- WKSH, Workshop
+- B, Basement
+- FL, Floor
+- G, Ground
+- L, Level
+- LG, Lower ground floor
+- M, Mezzanine
+- OD, Observation deck
+- P, Parking
+- PTHS, Penthouse
+- PLF, Platform
+- PDM, Podium
+- RT, Rooftop
+- SB, Sub-basement
+- UG, Upper ground floor
+- CN, Central
+- E, East
+- EX, Extension
+- LR, Lower
+- N, North
+- NE, North East
+- NW, North West
+- S, South
+- SE, South East
+- SW, South West
+- UP, Upper
+- W, West
+- ACCS, Access
+- ALLY, Alley
+- ALWY, Alleyway
+- AMBL, Amble
+- APP, Approach
+- ARC, Arcade
+- ARTL, Arterial
+- ARTY, Artery
+- AV, Ave, Avenue
+- BA, Banan
+- BEND, Bend
+- BWLK, Boardwalk
+- BVD, Boulevard
+- BR, Brace
+- BRAE, Brae
+- BRK, Break
+- BROW, Brow
+- BYPA, Bypass
+- BYWY, Byway
+- CSWY, Causeway
+- CTR, Centre
+- CH, Chase
+- CIR, Circle
+- CCT, Circuit
+- CRCS, Circus
+- CL, Close
+- CON, Concourse
+- CPS, Copse
+- CNR, Corner
+- CT, Crt, Court
+- CTYD, Courtyard
+- COVE, Cove
+- CR, Cres, Crescent
+- CRST, Crest
+- CRSS, Cross
+- CSAC, Cul-de-sac
+- CUTT, Cutting
+- DALE, Dale
+- DIP, Dip
+- DR, Drive
+- DVWY, Driveway
+- EDGE, Edge
+- ELB, Elbow
+- END, End
+- ENT, Entrance
+- ESP, Esplanade
+- EXP, Expressway
+- FAWY, Fairway
+- FOLW, Follow
+- FTWY, Footway
+- FORM, Formation
+- FWY, Freeway
+- FRTG, Frontage
+- GAP, Gap
+- GDNS, Gardens
+- GTE, Gate
+- GLDE, Glade
+- GLEN, Glen
+- GRA, Grange
+- GRN, Green
+- GR, Grove
+- HTS, Heights
+- HIRD, Highroad
+- HWY, Highway
+- HILL, Hill
+- INTG, Interchange
+- JNC, Junction
+- KEY, Key
+- LANE, Ln, Lane
+- LNWY, Laneway
+- LINE, Line
+- LINK, Link
+- LKT, Lookout
+- LOOP, Loop
+- MALL, Mall
+- MNDR, Meander
+- MEWS, Mews
+- MTWY, Motorway
+- NOOK, Nook
+- OTLK, Outlook
+- PDE, Parade
+- PWY, Pkwy, Parkway
+- PASS, Pass
+- PSGE, Passage
+- PATH, Path
+- PWAY, Pathway
+- PIAZ, Piazza
+- PLZA, Plaza
+- Pl, Place
+- PKT, Pocket
+- PNT, Point
+- PORT, Port
+- PROM, Promenade
+- QDRT, Quadrant
+- QYS, Quays
+- RMBL, Ramble
+- REST, Rest
+- RTT, Retreat
+- RDGE, Ridge
+- RISE, Rise
+- RD, Road
+- RTY, Rotary
+- RTE, Route
+- ROW, Row
+- RUE, Rue
+- SVWY, Serviceway
+- SHUN, Shunt
+- SPUR, Spur
+- SQ, Square
+- ST, Street
+- SBWY, Subway
+- TARN, Tarn
+- TCE, Terrace
+- THFR, Thoroughfare
+- TLWY, Tollway
+- TOP, Top
+- TOR, Tor
+- TRK, Track
+- TRL, Trail
+- TURN, Turn
+- UPAS, Underpass
+- VALE, Vale
+- VIAD, Viaduct
+- VIEW, View
+- VSTA, Vista
+- WALK, Walk
+- WKWY, Walkway
+- WHRF, Wharf
+- WYND, Wynd

data/app/countries/atlas_engine/au/validation_transcriber/address_parser.rb

@@ -0,0 +1,121 @@
+# typed: true
+# frozen_string_literal: true
+
+module AtlasEngine
+  module Au
+    module ValidationTranscriber
+      class AddressParser < AtlasEngine::ValidationTranscriber::AddressParserBase
+        # Temporary till we fix constants loading for AddressParsers
+        # Source: https://meteor.aihw.gov.au/content/429004
+        UNIT_TYPE_KEYWORDS = T.let(
+          Set.new([
+            "ant",
+            "apt",
+            "atm",
+            "bbq",
+            "btsd",
+            "bldg",
+            "bngw",
+            "cage",
+            "carp",
+            "cars",
+            "club",
+            "cool",
+            "ctge",
+            "dupl",
+            "fcty",
+            "flat",
+            "grge",
+            "hall",
+            "hse",
+            "ksk",
+            "lse",
+            "lbby",
+            "loft",
+            "lot",
+            "msnt",
+            "mbth",
+            "offc",
+            "resv",
+            "room",
+            "shed",
+            "shop",
+            "shrm",
+            "sign",
+            "site",
+            "stll",
+            "stor",
+            "str",
+            "stu",
+            "subs",
+            "se",
+            "tncy",
+            "twr",
+            "tnhs",
+            "unit",
+            "vlt",
+            "vlla",
+            "ward",
+            "whse",
+            "wksh",
+          ]),
+          T::Set[String],
+        )
+
+        sig { override.returns(T::Array[Regexp]) }
+        def country_regex_formats
+          @country_regex_formats ||= [
+            /(#{unit_regex})?#{bldg_num_regex}\s+#{street_regex}/i,
+          ]
+        end
+
+        private
+
+        sig { returns(Regexp) }
+        def unit_regex
+          # examples
+          # 10
+          # 10/
+          # flat 10
+          # flat 10,
+          # flat 10/
+          # 10A
+          # A10
+          @unit_regex ||= T.let(
+            %r{
+                (?<unit_type>(#{unit_type_list}))?\s*
+                (?<unit_num>\d+[a-z]?|[a-z]?\d+)(\s|\/|,)+
+              }ix,
+            T.nilable(Regexp),
+          )
+        end
+
+        sig { returns(Regexp) }
+        def bldg_num_regex
+          # examples:
+          # 100
+          # 100B
+          # 100-102
+          @bldg_num_regex ||= T.let(/(?<building_num>\d+(\-\d+|[a-z]?))/i, T.nilable(Regexp))
+        end
+
+        sig { returns(Regexp) }
+        def street_regex
+          @street_regex ||= T.let(/(?<street>.+)/, T.nilable(Regexp))
+        end
+
+        sig { override.params(token: T.nilable(String)).returns(T::Boolean) }
+        def secondary_unit_designator?(token)
+          return false if token.blank?
+
+          UNIT_TYPE_KEYWORDS.include?(token)
+        end
+
+        sig { returns(String) }
+        def unit_type_list
+          UNIT_TYPE_KEYWORDS.join("|")
+        end
+      end
+    end
+  end
+end

data/app/countries/atlas_engine/be/country_profile.yml

@@ -0,0 +1,12 @@
+id: BE
+ingestion:
+  settings:
+    min_zip_edge_ngram: "1"
+    max_zip_edge_ngram: "4"
+validation:
+  enabled: true
+  default_matching_strategy: local
+  index_locales:
+  - fr
+  - nl
+  - de

data/app/countries/atlas_engine/bm/address_importer/corrections/open_address/city_alias_corrector.rb

@@ -0,0 +1,38 @@
+# typed: true
+# frozen_string_literal: true
+
+module AtlasEngine
+  module Bm
+    module AddressImporter
+      module Corrections
+        module OpenAddress
+          class CityAliasCorrector
+            class << self
+              extend T::Sig
+
+              BM_PARISH_AND_CITY_NAMES = {
+                "Devonshire" => "Devonshire Parish",
+                "Hamilton" => "Hamilton Parish",
+                "Paget" => "Paget Parish",
+                "Pembroke" => "Pembroke Parish",
+                "Sandys" => "Sandys Parish",
+                "Smiths" => "Smiths Parish",
+                "Southampton" => "Southampton Parish",
+                "St. George's" => "St. George's Parish",
+                "Town of St. George" => "St. George",
+                "Warwick" => "Warwick Parish",
+              }.freeze
+
+              sig { params(address: Hash).void }
+              def apply(address)
+                if BM_PARISH_AND_CITY_NAMES.key?(address[:city][0])
+                  address[:city] << BM_PARISH_AND_CITY_NAMES[address[:city][0]]
+                end
+              end
+            end
+          end
+        end
+      end
+    end
+  end
+end

data/app/countries/atlas_engine/bm/address_importer/open_address/mapper.rb

@@ -0,0 +1,40 @@
+# typed: true
+# frozen_string_literal: true
+
+module AtlasEngine
+  module Bm
+    module AddressImporter
+      module OpenAddress
+        class Mapper < AtlasEngine::AddressImporter::OpenAddress::DefaultMapper
+          sig do
+            params(feature: AtlasEngine::AddressImporter::OpenAddress::Feature).returns(T::Hash[Symbol, T.untyped])
+          end
+          def map(feature)
+            city, street, number, unit, postcode = feature["properties"].values_at(
+              "district",
+              "street",
+              "number",
+              "unit",
+              "postcode",
+            )
+            result = {
+              source_id: openaddress_source_id(feature),
+              locale: @locale,
+              country_code: "BM",
+              province_code: nil,
+              # Omitted: region1..4
+              city: [city], # city names are already titleized
+              suburb: nil,
+              zip: normalize_zip(postcode),
+              street: street.titleize,
+              building_and_unit_ranges: housenumber_and_unit(number, unit),
+              latitude: geometry(feature)&.at(1),
+              longitude: geometry(feature)&.at(0),
+            }
+            result
+          end
+        end
+      end
+    end
+  end
+end

data/app/countries/atlas_engine/bm/country_profile.yml

@@ -0,0 +1,12 @@
+id: BM
+ingestion:
+  correctors:
+    open_address:
+      - AtlasEngine::Bm::AddressImporter::Corrections::OpenAddress::CityAliasCorrector
+  open_address:
+    feature_mapper: AtlasEngine::Bm::AddressImporter::OpenAddress::Mapper
+validation:
+  enabled: true
+  has_provinces: true
+  default_matching_strategy: es
+  address_parser: AtlasEngine::ValidationTranscriber::AddressParserNorthAmerica

data/app/countries/atlas_engine/br/country_profile.yml

@@ -0,0 +1,4 @@
+id: BR
+ingestion:
+  settings:
+    number_of_shards: "2"

data/app/countries/atlas_engine/ca/country_profile.yml

@@ -0,0 +1,7 @@
+id: CA
+ingestion:
+  settings:
+    min_zip_edge_ngram: "1"
+    max_zip_edge_ngram: "7"
+validation:
+  address_parser: AtlasEngine::ValidationTranscriber::AddressParserNorthAmerica