pumi 0.16.0 → 0.18.0

data/data/provinces.yml

@@ -8,6 +8,8 @@ provinces:
       km: ខេត្ត
       latin: Khaet
       en: Province
+    links:
+      wikipedia: https://en.wikipedia.org/wiki/Banteay_Meanchey_province
   '02':
     name:
       km: បាត់ដំបង
@@ -16,6 +18,8 @@ provinces:
       km: ខេត្ត
       latin: Khaet
       en: Province
+    links:
+      wikipedia: https://en.wikipedia.org/wiki/Battambang_province
   '03':
     name:
       km: កំពង់ចាម
@@ -24,6 +28,8 @@ provinces:
       km: ខេត្ត
       latin: Khaet
       en: Province
+    links:
+      wikipedia: https://en.wikipedia.org/wiki/Kampong_Cham_province
   '04':
     name:
       km: កំពង់ឆ្នាំង
@@ -32,6 +38,8 @@ provinces:
       km: ខេត្ត
       latin: Khaet
       en: Province
+    links:
+      wikipedia: https://en.wikipedia.org/wiki/Kampong_Chhnang_province
   '05':
     name:
       km: កំពង់ស្ពឺ
@@ -40,6 +48,8 @@ provinces:
       km: ខេត្ត
       latin: Khaet
       en: Province
+    links:
+      wikipedia: https://en.wikipedia.org/wiki/Kampong_Speu_province
   '06':
     name:
       km: កំពង់ធំ
@@ -48,6 +58,8 @@ provinces:
       km: ខេត្ត
       latin: Khaet
       en: Province
+    links:
+      wikipedia: https://en.wikipedia.org/wiki/Kampong_Thom_province
   '07':
     name:
       km: កំពត
@@ -56,14 +68,18 @@ provinces:
       km: ខេត្ត
       latin: Khaet
       en: Province
+    links:
+      wikipedia: https://en.wikipedia.org/wiki/Kampot_province
   '08':
     name:
-      km: កណ្ដាល
+      km: កណ្តាល
       latin: Kandal
     administrative_unit:
       km: ខេត្ត
       latin: Khaet
       en: Province
+    links:
+      wikipedia: https://en.wikipedia.org/wiki/Kandal_province
   '09':
     name:
       km: កោះកុង
@@ -72,6 +88,8 @@ provinces:
       km: ខេត្ត
       latin: Khaet
       en: Province
+    links:
+      wikipedia: https://en.wikipedia.org/wiki/Koh_Kong_province
   '10':
     name:
       km: ក្រចេះ
@@ -80,6 +98,8 @@ provinces:
       km: ខេត្ត
       latin: Khaet
       en: Province
+    links:
+      wikipedia: https://en.wikipedia.org/wiki/Krati%C3%A9_province
   '11':
     name:
       km: មណ្ឌលគិរី
@@ -88,6 +108,8 @@ provinces:
       km: ខេត្ត
       latin: Khaet
       en: Province
+    links:
+      wikipedia: https://en.wikipedia.org/wiki/Mondulkiri_province
   '12':
     name:
       km: ភ្នំពេញ
@@ -96,6 +118,8 @@ provinces:
       km: រាជធានី
       latin: Reach Theani
       en: Capital
+    links:
+      wikipedia: https://en.wikipedia.org/wiki/Phnom_Penh
   '13':
     name:
       km: ព្រះវិហារ
@@ -104,6 +128,8 @@ provinces:
       km: ខេត្ត
       latin: Khaet
       en: Province
+    links:
+      wikipedia: https://en.wikipedia.org/wiki/Preah_Vihear_province
   '14':
     name:
       km: ព្រៃវែង
@@ -112,6 +138,8 @@ provinces:
       km: ខេត្ត
       latin: Khaet
       en: Province
+    links:
+      wikipedia: https://en.wikipedia.org/wiki/Prey_Veng_province
   '15':
     name:
       km: ពោធិ៍សាត់
@@ -120,6 +148,8 @@ provinces:
       km: ខេត្ត
       latin: Khaet
       en: Province
+    links:
+      wikipedia: https://en.wikipedia.org/wiki/Pursat_province
   '16':
     name:
       km: រតនគិរី
@@ -128,6 +158,8 @@ provinces:
       km: ខេត្ត
       latin: Khaet
       en: Province
+    links:
+      wikipedia: https://en.wikipedia.org/wiki/Ratanakiri_province
   '17':
     name:
       km: សៀមរាប
@@ -136,6 +168,8 @@ provinces:
       km: ខេត្ត
       latin: Khaet
       en: Province
+    links:
+      wikipedia: https://en.wikipedia.org/wiki/Siem_Reap_province
   '18':
     name:
       km: ព្រះសីហនុ
@@ -144,6 +178,8 @@ provinces:
       km: ខេត្ត
       latin: Khaet
       en: Province
+    links:
+      wikipedia: https://en.wikipedia.org/wiki/Sihanoukville_province
   '19':
     name:
       km: ស្ទឹងត្រែង
@@ -152,6 +188,8 @@ provinces:
       km: ខេត្ត
       latin: Khaet
       en: Province
+    links:
+      wikipedia: https://en.wikipedia.org/wiki/Stung_Treng_province
   '20':
     name:
       km: ស្វាយរៀង
@@ -160,6 +198,8 @@ provinces:
       km: ខេត្ត
       latin: Khaet
       en: Province
+    links:
+      wikipedia: https://en.wikipedia.org/wiki/Svay_Rieng_province
   '21':
     name:
       km: តាកែវ
@@ -168,14 +208,18 @@ provinces:
       km: ខេត្ត
       latin: Khaet
       en: Province
+    links:
+      wikipedia: https://en.wikipedia.org/wiki/Tak%C3%A9o_province
   '22':
     name:
-      km: ឧត្ដរមានជ័យ
+      km: ឧត្តរមានជ័យ
       latin: Oddar Meanchey
     administrative_unit:
       km: ខេត្ត
       latin: Khaet
       en: Province
+    links:
+      wikipedia: https://en.wikipedia.org/wiki/Oddar_Meanchey_province
   '23':
     name:
       km: កែប
@@ -184,6 +228,8 @@ provinces:
       km: ខេត្ត
       latin: Khaet
       en: Province
+    links:
+      wikipedia: https://en.wikipedia.org/wiki/Kep_province
   '24':
     name:
       km: ប៉ៃលិន
@@ -192,6 +238,8 @@ provinces:
       km: ខេត្ត
       latin: Khaet
       en: Province
+    links:
+      wikipedia: https://en.wikipedia.org/wiki/Pailin_province
   '25':
     name:
       km: ត្បូងឃ្មុំ
@@ -200,3 +248,5 @@ provinces:
       km: ខេត្ត
       latin: Khaet
       en: Province
+    links:
+      wikipedia: https://en.wikipedia.org/wiki/Tboung_Khmum_province

data/data/villages.yml

@@ -53016,6 +53016,22 @@ villages:
       km: ភូមិ
       latin: Phum
       en: Village
+  '08130602':
+    name:
+      km: ស្វាយជ្រុំ
+      latin: Svay Chrum
+    administrative_unit:
+      km: ភូមិ
+      latin: Phum
+      en: Village
+  '08130603':
+    name:
+      km: បារាជ
+      latin: Ba Reach
+    administrative_unit:
+      km: ភូមិ
+      latin: Phum
+      en: Village
   '08130701':
     name:
       km: អរិយក្សត្រ

data/lib/pumi/data_file.rb

@@ -0,0 +1,32 @@
+require "yaml"
+require "pathname"
+
+module Pumi
+  class DataFile
+    DEFAULT_DATA_DIRECTORY = File.join(File.expand_path("..", File.dirname(__dir__)), "data")
+    TYPES = %w[provinces districts communes villages].freeze
+
+    attr_reader :type
+
+    def initialize(type)
+      @type = type.to_s
+      raise ArgumentError, "#{type} is not included in #{TYPES}" unless TYPES.include?(@type)
+    end
+
+    def read(data_directory: DEFAULT_DATA_DIRECTORY)
+      YAML.load_file(data_file(data_directory)).fetch(type)
+    end
+
+    def write(data, data_directory: DEFAULT_DATA_DIRECTORY)
+      return if data.empty?
+
+      File.write(data_file(data_directory), { type => data.sort.to_h }.to_yaml)
+    end
+
+    private
+
+    def data_file(data_directory)
+      Pathname(data_directory).join("#{type}.yml")
+    end
+  end
+end

data/lib/pumi/data_source/ncdd.rb

@@ -0,0 +1,100 @@
+require "pathname"
+require "csv"
+require "yaml"
+
+# https://en.wikipedia.org/wiki/Administrative_divisions_of_Cambodia
+# https://en.wikipedia.org/wiki/Romanization_of_Khmer
+# https://en.wikipedia.org/wiki/United_Nations_Group_of_Experts_on_Geographical_Names
+
+module Pumi
+  module DataSource
+    class NCDD
+      CSV_HEADERS = %w[type code name_km name_latin reference note1 note2].freeze
+
+      AdministrativeUnit = Struct.new(:en, :km, :latin, :code_length, :group, :type, keyword_init: true)
+      Row = Struct.new(:code, :name_km, :name_latin, :type, keyword_init: true) do
+        def administrative_unit
+          ADMINISTRATIVE_UNITS.fetch(type)
+        end
+      end
+
+      ADMINISTRATIVE_UNITS = {
+        "ស្រុក" => AdministrativeUnit.new(en: "District", km: "ស្រុក", latin: "Srok", code_length: 4, group: "districts"),
+        "ខណ្ឌ" => AdministrativeUnit.new(en: "Section", km: "ខណ្ឌ", latin: "Khan", code_length: 4, group: "districts"),
+        "ក្រុង" => AdministrativeUnit.new(en: "Municipality", km: "ក្រុង", latin: "Krong", code_length: 4, group: "districts"),
+        "ឃុំ" => AdministrativeUnit.new(en: "Commune", km: "ឃុំ", latin: "Khum", code_length: 6, group: "communes"),
+        "សង្កាត់" => AdministrativeUnit.new(en: "Quarter", km: "សង្កាត់", latin: "Sangkat", code_length: 6, group: "communes"),
+        "ភូមិ" => AdministrativeUnit.new(en: "Village", km: "ភូមិ", latin: "Phum", code_length: 8, group: "villages")
+      }.freeze
+
+      def load_data!(source_dir: "tmp", output_dir: "data")
+        source_files(source_dir).each do |file|
+          parse_source_file(file)
+        end
+
+        write_data!(output_dir)
+      end
+
+      private
+
+      def parse_source_file(file)
+        CSV.read(file, headers: CSV_HEADERS).each do |csv_row|
+          row = build_row(csv_row)
+
+          next unless row.code
+          next if row.administrative_unit.code_length != row.code.length
+
+          write_location(row)
+        end
+      end
+
+      def data
+        @data ||= {}
+      end
+
+      def build_row(row)
+        Row.new(
+          code: parse_location_code(row),
+          name_km: row.fetch("name_km"),
+          name_latin: row.fetch("name_latin"),
+          type: row.fetch("type")
+        )
+      end
+
+      def parse_location_code(row)
+        code = row.fetch("code")
+        return if code.to_s.gsub(/\D/, "").empty?
+
+        code = code.rjust(code.length + 1, "0") if code.length.odd?
+        code
+      end
+
+      def write_location(row)
+        data[row.administrative_unit.group] ||= {}
+        data[row.administrative_unit.group][row.code] = {
+          "name" => {
+            "km" => row.name_km,
+            "latin" => row.name_latin
+          },
+          "administrative_unit" => {
+            "km" => row.administrative_unit.km,
+            "latin" => row.administrative_unit.latin,
+            "en" => row.administrative_unit.en
+          }
+        }
+      end
+
+      def source_files(source_dir)
+        Pathname.glob("#{source_dir}/*.csv").select(&:file?)
+      end
+
+      def write_data!(output_dir)
+        return if data.empty?
+
+        ADMINISTRATIVE_UNITS.values.map(&:group).uniq do |data_group|
+          DataFile.new(data_group).write(data.fetch(data_group), data_directory: output_dir)
+        end
+      end
+    end
+  end
+end