populatecities.rake

namespace :pinow do
  desc 'Update the population field of all city locations with the help of a csv file'
  task :populate_cities => :environment do

    require 'fastercsv'

    #constants
    CODE = 0
    NAME = 5
    STATE_NAME = 6
    POP = 7

    #statistical
    good = 0

    #look at http://www.census.gov/popest/cities/files/SUB-EST2009.pdf for desc codes
    right_codes = [ "061" , "162" ]

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

      FasterCSV.foreach("SUB-EST2009-ALL.csv") do |city_row|

        if city_row[CODE].present? && right_codes.include?(city_row[CODE].to_s)

          location_type = "city"

          #create identifier-----------
          if (city_row[NAME] =~ /town/).present?
            location_type = "town"
          elsif (city_row[NAME] =~ /village/).present?
            location_type = "-village"
          elsif (city_row[NAME] =~ /borough/).present?
            location_type = "borough"
          elsif (city_row[NAME] =~ /township/).present?
            location_type = "township"
          end
          #------------------------------

          location_url = File.join("", city_row[STATE_NAME].parameterize, city_row[NAME].gsub(/#{location_type}$/, "").parameterize)

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

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

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