pumi 0.19.0 → 0.20.1

data/lib/pumi/bot/wikipedia/templates/commune_list.wikitext.erb

@@ -0,0 +1,46 @@
+<div id="communes-list">
+<% provinces.each do |province| %>
+<% province_page = URI.parse(province.links[:wikipedia]).path.split("/").last %>
+==[[<%= province_page %>|<%= province.full_name_en %>]]==
+<div id=province-communes-<%= province.id %>>
+<%= province.name_en %> contains <%= province.communes_summary %>.<ref>{{cite web|url=http://db.ncdd.gov.kh/gazetteer/view/province.castle?pv=<%= province.id %> |title=<%= province.name_en %> |publisher=National Committee for Sub-National Democratic Development }}</ref>
+
+<% province.districts.each do |district| %>
+<% if district.links[:wikipedia] %>
+<% district_page = URI.parse(district.links[:wikipedia]).path.split("/").last %>
+===[[<%= district_page %>|<%= district.full_name_en %>]]===
+<% else %>
+===<%= district.full_name_en %>===
+<% end %>
+
+<div id=district-communes-<%= district.id %>>
+<%= district.name_en %> contains <%= district.communes_summary %>.<ref>{{cite web|url=http://db.ncdd.gov.kh/gazetteer/view/district.castle?ds=<%= district.id %> |title=<%= district.name_en %> |publisher=National Committee for Sub-National Democratic Development }}</ref>
+
+{| class="wikitable sortable"
+|-
+! #
+! Name
+! Khmer
+! Administrative Unit
+! Geocode
+|-
+
+<% district.communes.each_with_index do |commune, index| %>
+| <%= index + 1 %>
+<% if commune.links[:wikipedia] %>
+<% commune_page = URI.parse(commune.links[:wikipedia]).path.split("/").last.gsub("_", " ") %>
+| [[<%= commune_page %>|<%= commune.name_en %>]]
+<% else %>
+| <%= commune.name_en %>
+<% end %>
+| <%= commune.name_km %>
+| <%= "#{commune.administrative_unit.name_en} (#{commune.administrative_unit.name_km} #{commune.administrative_unit.name_latin})" %>
+| <%= commune.id %>
+|-
+<% end %>
+|}
+</div>
+<% end %>
+</div>
+<% end %>
+</div>

data/lib/pumi/bot/wikipedia/templates/district_list.wikitext.erb

@@ -0,0 +1,27 @@
+<div id=province-districts-<%= province.id %>>
+<%= province.name_en %> contains <%= districts_summary %>. <ref>{{cite web|url=http://db.ncdd.gov.kh/gazetteer/view/province.castle?pv=<%= province.id %> |title=<%= province.name_en %> |publisher=National Committee for Sub-National Democratic Development }}</ref>
+
+{| class="wikitable sortable"
+|-
+! #
+! Name
+! Khmer
+! Administrative Unit
+! Geocode
+|-
+
+<% districts.each_with_index do |district, index| %>
+| <%= index + 1 %>
+<% if district.links[:wikipedia] %>
+<% district_page = URI.parse(district.links[:wikipedia]).path.split("/").last.gsub("_", " ") %>
+| [[<%= district_page %>|<%= district.name_en %>]]
+<% else %>
+| <%= district.name_en %>
+<% end %>
+| <%= district.name_km %>
+| <%= "#{district.administrative_unit.name_en} (#{district.administrative_unit.name_km} #{district.administrative_unit.name_latin})" %>
+| <%= district.id %>
+|-
+<% end %>
+|}
+</div>

data/lib/pumi/bot/wikipedia.rb

@@ -0,0 +1,10 @@
+module Pumi
+  module Bot
+    module Wikipedia
+    end
+  end
+end
+
+require_relative "wikipedia/article"
+require_relative "wikipedia/communes_in_cambodia_article"
+require_relative "wikipedia/districts_in_cambodia_article"

data/lib/pumi/bot.rb

@@ -0,0 +1,6 @@
+module Pumi
+  module Bot
+  end
+end
+
+require_relative "bot/wikipedia"

data/lib/pumi/data_source/geocoder.rb

@@ -0,0 +1,251 @@
+require "geocoder"
+
+module Pumi
+  module DataSource
+    class Geocoder
+      Result = Struct.new(:code, :lat, :long, :bounding_box, keyword_init: true)
+
+      class AbstractGeocoder
+        Result = Struct.new(
+          :lat, :long, :bounding_box, :country_code,
+          :types, :iso3166_2, :district_name_en,
+          :name,
+          keyword_init: true
+        )
+
+        Misspelling = Struct.new(:incorrect_text, :correct_text, keyword_init: true)
+
+        MISSPELLINGS = []
+
+        class AbstractProvider
+          attr_reader :geocoder, :name
+
+          def initialize(geocoder:, name:)
+            @geocoder = geocoder
+            @name = name
+          end
+
+          def search(term)
+            geocoder.search(term, lookup: name).map do |result|
+              build_result(result.data)
+            end
+          end
+        end
+
+        class Google < AbstractProvider
+          private
+
+          def build_result(data)
+            province_name_en = find_address_component(
+              data,
+              "administrative_area_level_1"
+            )&.fetch("long_name")
+            province = Pumi::Province.where(full_name_en: province_name_en).first
+            Result.new(
+              name: data.dig("address_components", 0, "long_name"),
+              lat: data.dig("geometry", "location", "lat"),
+              long: data.dig("geometry", "location", "lng"),
+              bounding_box: [
+                data.dig("geometry", "bounds", "northeast", "lat"),
+                data.dig("geometry", "bounds", "northeast", "lng"),
+                data.dig("geometry", "bounds", "southwest", "lat"),
+                data.dig("geometry", "bounds", "southwest", "lng")
+              ],
+              country_code: find_address_component(data, "country").fetch("short_name").upcase,
+              district_name_en: find_address_component(
+                data,
+                "administrative_area_level_2"
+              )&.fetch("long_name"),
+              types: data["types"],
+              iso3166_2: province&.iso3166_2
+            )
+          end
+
+          def find_address_component(data, type)
+            data.fetch("address_components").find do |c|
+              c.fetch("types").include?(type)
+            end
+          end
+        end
+
+        class Nominatim < AbstractProvider
+          private
+
+          def build_result(data)
+            Result.new(
+              name: nil,
+              lat: data["lat"],
+              long: data["lon"],
+              bounding_box: data["boundingbox"],
+              types: Array(data["type"]),
+              iso3166_2: data.dig("address", "ISO3166-2-lvl4"),
+              country_code: data.dig("address", "country_code")&.upcase,
+              district_name_en: data.dig("address", "county")
+            )
+          end
+        end
+
+        PROVIDERS = {
+          nominatim: Nominatim,
+          google: Google
+        }.freeze
+
+        attr_reader :providers, :options
+
+        def initialize(geocoder: ::Geocoder, providers: PROVIDERS.keys, **options)
+          @options = options
+
+          geocoder.configure(
+            google: {
+              api_key: ENV["GOOGLE_API_KEY"]
+            }
+          )
+
+          @providers = Array(providers).map do |name|
+            PROVIDERS.fetch(name).new(geocoder:, name:)
+          end
+        end
+
+        def geocode_all
+          locations.each_with_object([]).with_index do |(location, results), _index|
+            next if !options[:regeocode] && !location.geodata.nil?
+
+            geocoder_result = geocode(location)
+
+            if geocoder_result.nil?
+              ungeocoded_locations << location
+              next
+            end
+
+            results << build_result(code: location.id, geocoder_result:)
+          end
+        end
+
+        private
+
+        def geocode(location)
+          providers.each do |provider|
+            Array(build_search_term(location)).each do |search_term|
+              all_results = provider.search(search_term)
+              geocoder_result = filter(location, all_results)
+
+              return geocoder_result unless geocoder_result.nil?
+            end
+          end
+
+          nil
+        end
+
+        def build_result(code:, geocoder_result:)
+          Geocoder::Result.new(
+            code:,
+            lat: geocoder_result.lat,
+            long: geocoder_result.long,
+            bounding_box: geocoder_result.bounding_box
+          )
+        end
+
+        def build_search_term(location)
+          [location.full_name_km, location.name_km].map do |term|
+            MISSPELLINGS.find { |m| m.correct_text == term }&.incorrect_text || term
+          end
+        end
+
+        def ungeocoded_locations
+          @ungeocoded_locations ||= []
+        end
+      end
+
+      class CambodianProvinces < AbstractGeocoder
+        private
+
+        def locations
+          @locations ||= Pumi::Province.all
+        end
+
+        def build_search_term(province)
+          province.iso3166_2
+        end
+
+        def filter(province, geocoder_results)
+          geocoder_results.find do |r|
+            r.iso3166_2 == province.iso3166_2 && r.types.include?("administrative")
+          end
+        end
+      end
+
+      class CambodianDistricts < AbstractGeocoder
+        private
+
+        def locations
+          @locations ||= Pumi::District.all
+        end
+
+        def filter(district, geocoder_results)
+          geocoder_results.find do |r|
+            r.country_code == "KH" &&
+              r.iso3166_2 == district.province.iso3166_2 && (
+                %w[administrative_area_level_2 town city administrative].any? do |type|
+                  r.types.include?(type)
+                end || (%w[locality political].sort == r.types.sort)
+              )
+          end
+        end
+      end
+
+      class CambodianCommunes < AbstractGeocoder
+        private
+
+        def locations
+          @locations ||= Pumi::Commune.all
+        end
+
+        def filter(commune, geocoder_results)
+          geocoder_results.find do |r|
+            r.country_code == "KH" &&
+              (r.iso3166_2 == commune.province.iso3166_2 || r.district_name_en.to_s.downcase.include?(commune.district.name_en.downcase)) &&
+              %w[administrative_area_level_3 village suburb neighbourhood].any? do |type|
+                r.types.include?(type)
+              end
+          end
+        end
+      end
+
+      attr_reader :data_file, :geocoder
+
+      def initialize(data_file:, geocoder:)
+        @data_file = data_file
+        @geocoder = geocoder
+      end
+
+      def load_data!(output_dir: "data")
+        data.each do |code, attributes|
+          geocoded_result = geocoded_results.find { |r| r.code == code }
+
+          next if geocoded_result.nil?
+
+          attributes["geodata"] ||= {}
+          attributes["geodata"]["lat"] = geocoded_result.lat
+          attributes["geodata"]["long"] = geocoded_result.long
+          attributes["geodata"]["bounding_box"] = geocoded_result.bounding_box
+        end
+
+        write_data!(output_dir)
+      end
+
+      private
+
+      def data
+        @data ||= data_file.read
+      end
+
+      def write_data!(data_directory)
+        data_file.write(data, data_directory:)
+      end
+
+      def geocoded_results
+        @geocoded_results ||= geocoder.geocode_all
+      end
+    end
+  end
+end

data/lib/pumi/data_source/iso31662.rb

@@ -0,0 +1,29 @@
+module Pumi
+  module DataSource
+    class ISO31662
+      attr_reader :data_file
+
+      def initialize(data_file: Pumi::DataFile.new(:provinces))
+        @data_file = data_file
+      end
+
+      def load_data!(output_dir: "data")
+        data.each do |code, attributes|
+          attributes["iso3166_2"] = "KH-#{code.to_i}"
+        end
+
+        write_data!(output_dir)
+      end
+
+      private
+
+      def data
+        @data ||= data_file.read
+      end
+
+      def write_data!(data_directory)
+        data_file.write(data, data_directory:)
+      end
+    end
+  end
+end