Ett lite snabbhack för att dra in alla städer med en viss storlek samt tillhörande län och spara dem till en databas.

city_importer.rb

#require "nokogiri"
#require "open-uri"

class CityImporter
  attr_reader :wikipedia_cities, :wikipedia_municipalities

  def import_wikipedia_cities
    @wikipedia_cities ||= []

    wikipedia_cities_page = Nokogiri::HTML(open('http://sv.wikipedia.org/wiki/Lista_%C3%B6ver_Sveriges_t%C3%A4torter'))

    wikipedia_cities_page.css("table.sortable.wikitable tr").each_with_index do |row, i|

      next if i == 0
      pop = row.css("td")[2].children.last.text.gsub(/[^0-9]+/, "").to_i

      if pop > 1999
        @wikipedia_cities << [row.css("td")[0].text, row.css("td")[1].text]
        puts "added: #{@wikipedia_cities.last.first}, #{@wikipedia_cities.last.last}\n"
      end

    end
    puts "All done!"
  end

  def import_wikipedia_municipalities
    @wikipedia_municipalities ||= []

    wikipedia_municipalities_page = Nokogiri::HTML(open('http://sv.wikipedia.org/wiki/Sveriges_kommuner_listade_per_l%C3%A4n'))

    wikipedia_municipalities_page.css("table.sortable.wikitable tr").each_with_index do |row, i|
      next if i == 0
      
      @wikipedia_municipalities << [row.css("td")[1].text, row.css("td")[2].text]
      puts "added: #{@wikipedia_municipalities.last.first}, #{@wikipedia_municipalities.last.last}\n"

    end
    puts "All done!"
  end

  def add_to_db
    states = wikipedia_municipalities.uniq { |m| m[1] }
    states.each do |state|
      puts "added: #{state.last} to database.\n" if State.create(name: state.last)
    end

    @wikipedia_cities.each do |city|
      state = State.find_by_slug(get_state_slug(city))
      puts "Added: #{city[0]}" if City.create(name: city[0], state: state)
    end
  end

  def initialize(autorun = true)
    if autorun
      import_wikipedia_cities
      import_wikipedia_municipalities
    end
    puts "#{@wikipedia_municipalities.count} municipalities and #{@wikipedia_cities.count} cities added.\nRun 'add_to_db' to add to database.\n"
  end

  private

  def get_state_slug(city)
    @wikipedia_municipalities[@wikipedia_municipalities.index { |value| value[0]==city[1] }][1].parameterize
  end
end