populatecities.rake

namespace :locations do
  
  desc 'Update the population field of all city locations with the help of a csv file'

  task :populate_cities, :pop_file, :no_match_file, :needs => :environment do | t , args |

    raise "Please send a valid CSV File, example: rake locations:populate_cities[pathCVSFile,noMatchFilePath*]" if args[:pop_file].nil?

    csv_file = args[:pop_file]
    unmatch_file = args[:no_match_file] || "no_match_location.csv"

    require 'fastercsv'

    #constants
    NAME = 3
    STATE_NAME = 4
    POP = 5

    #statistical
    good = 0

    #Before begin reset all population
    Location.update_all("population = NULL","location_type_id = 5 AND active = true")

    #Save locations that didn't match
    FasterCSV.open(unmatch_file, "w") do |csv|
      csv << ["location_name", "state_name","population", "url_for_attempt"]

      FasterCSV.foreach(csv_file) do |city_row|
        csv_city_name = city_row[NAME].gsub(/\(.*\)/, "").squeeze(" ").strip
        location_type = "city"

        if (csv_city_name =~ /town$/).present?
          location_type = "town"
        elsif (csv_city_name =~ /village/).present?
          location_type = "village"
        elsif (csv_city_name =~ /borough/).present?
          location_type = "borough"
        elsif (csv_city_name =~ /township/).present?
          location_type = "township"
        elsif (csv_city_name =~ /CDP/).present?
          location_type = "CDP"
        elsif (csv_city_name =~ /municipality/).present?
          location_type = "municipality"
        elsif (csv_city_name =~ /comunidad/).present?
          location_type = "comunidad"
        elsif (csv_city_name =~ /zona urbana/).present?
          location_type = "zona urbana"
        end

        csv_city_name = csv_city_name.gsub(/#{location_type}/, "").squeeze(" ")

        location_url = File.join("", city_row[STATE_NAME].parameterize, csv_city_name.parameterize)

        print "Find for: " + location_url + "\n"

        location = Location.find_active_cities(:first, :conditions => ["locations.url_for = ?", location_url])      

        if location.present? && ( location.population.nil? || location.population.to_i < city_row[POP].to_i )
          location.update_attributes(:population => city_row[POP] )
          good += 1
        else
          csv << [city_row[NAME], city_row[STATE_NAME],city_row[POP], location_url]
        end 
      end
    end

    print "Location populated correctly: #{good}"
  end
end