jp_local_gov 0.1.0

data/data/exporter.rb

@@ -0,0 +1,65 @@
+# frozen_string_literal: true
+
+require "json"
+require_relative "base"
+
+module JpLocalGov
+  module Data
+    class Exporter
+      SELECT_SQL = <<-SQL
+        SELECT
+          code,
+          prefecture_code,
+          prefecture,
+          prefecture_kana,
+          city,
+          city_kana,
+          prefecture_capital
+        FROM local_governments
+        ORDER BY code
+      SQL
+
+      def initialize
+        @db = JpLocalGov::Data::Base.new.db
+      end
+
+      def execute
+        local_governments.each do |prefix, value|
+          file_path = File.expand_path("../../data/json/#{prefix}.json", __FILE__)
+          File.open(file_path, "wb") do |file|
+            file.write JSON.pretty_generate(value)
+          end
+        end
+      end
+
+      private
+
+      def local_governments
+        local_governments = {}
+
+        @db.execute(SELECT_SQL) do |row|
+          lg_hash = to_hash(row)
+          local_governments[lg_hash[:prefecture_code]] ||= {}
+          local_governments[lg_hash[:prefecture_code]] =
+            local_governments[lg_hash[:prefecture_code]].merge(lg_hash[:code] => lg_hash)
+        end
+
+        local_governments
+      end
+
+      def to_hash(row)
+        local_government = {}
+        local_government[:code],
+        local_government[:prefecture_code],
+        local_government[:prefecture],
+        local_government[:prefecture_kana],
+        local_government[:city],
+        local_government[:city_kana],
+        local_government[:prefecture_capital] = row
+
+        local_government[:prefecture_capital] = local_government[:prefecture_capital] == "1"
+        local_government
+      end
+    end
+  end
+end

data/data/importer.rb

@@ -0,0 +1,65 @@
+# frozen_string_literal: true
+
+require_relative "base"
+require "open-uri"
+require "pdf-reader"
+require "json"
+require "nokogiri"
+
+module JpLocalGov
+  module Data
+    class Importer < Base
+      private
+
+      URL_DOMAIN = "https://www.soumu.go.jp"
+      SEARCH_URL = "https://www.soumu.go.jp/denshijiti/code.html"
+      VALID_COLUMN_COUNT = 5
+
+      def retrieve
+        local_governments = []
+        PDF::Reader.new(OpenURI.open_uri(pdf_url)).pages.each do |page|
+          page.text.split("\n").each do |row|
+            items = row.split("\s")
+            next if header?(items) || items.length != VALID_COLUMN_COUNT
+
+            local_governments << to_hash(items)
+          end
+        end
+        local_governments.uniq # to exclude special areas header(ex: Hokkaido Sapporo)
+      end
+
+      def to_hash(items)
+        {
+          code: items[0],
+          prefecture_code: items[0][0..1],
+          prefecture: items[1],
+          prefecture_kana: covert_half_char_to_full_char(items[3]),
+          city: items[2],
+          city_kana: covert_half_char_to_full_char(items[4]),
+          prefecture_capital: prefecture_capital?(items[1], items[2]) ? 1 : 0
+        }
+      end
+
+      def pdf_url
+        html = Nokogiri::HTML(OpenURI.open_uri(SEARCH_URL))
+        url = html.css('[href$=".pdf"]').first.attributes["href"]
+        "#{URL_DOMAIN}#{url}"
+      end
+
+      def header?(row)
+        (row[0] =~ /[0-9]/).nil?
+      end
+
+      def covert_half_char_to_full_char(text)
+        text.gsub(/[\uFF61-\uFF9F]+/) { |str| str.unicode_normalize(:nfkc) }
+      end
+
+      def prefecture_capital?(prefecture, city)
+        prefecture_capital_list = JSON.parse(File.open(File.expand_path("prefecture_capital.json", __dir__)).read)
+        prefecture_capital_list.find do |prefecture_capital|
+          prefecture_capital["prefecture"] == prefecture && prefecture_capital["city"] == city
+        end
+      end
+    end
+  end
+end