govtrack_kml.rb

## Methods dealing with Govtrack's KML exporter for states and congressional districts. 
#  Used to generate map files for a couple of Sunlight's projects, Politiwidgets and the 
#  Congress app for Android.

## You can also just download individual maps at GovTrack's page about all this:
#  http://www.govtrack.us/embed/googlemaps.xpd
#  This script is useful if you want to generate many programmatically. 
#  It also swaps out the default '0' placemark name for something minimally descriptive.

## This file isn't runnable, and just contains some methods. 
#  Most useful by opening up IRB in the same directory as this file, requiring this file, 
#  then running whichever methods are useful.

require 'rubygems'
require 'nokogiri'

@export = "http://www.govtrack.us/perl/wms/export.cgi"

@state_dataset = "http://www.rdfabout.com/rdf/usgov/us/states"
@state_region = "http://www.rdfabout.com/rdf/usgov/geo/us/%s" # state (e.g. CA)

@cd_dataset = "http://www.rdfabout.com/rdf/usgov/congress/house/%s" # session (e.g. 110)
@cd_region = "http://www.rdfabout.com/rdf/usgov/geo/us/%s/cd/%s/%s" # [state, session, district (e.g. 16)]


def get_state(state, filename = nil)
  name = filename || "states/#{state}"
  get_kml name, @state_dataset, (@state_region % state)
  change_placemark_names name
end

def get_cd(session, state, district)
  name = "cds/#{session}/#{state}-#{district}"
  get_kml name, (@cd_dataset % session), (@cd_region % [state, session, district])
  change_placemark_names name
end

def get_kml(path, dataset, region)
  system "wget -O \"#{path}.kml\" \"#{@export}?dataset=#{dataset}&region=#{region}&format=kml&maxpoints=1000\""
end


# update the default '0' placemark name to be the base name of the file (e.g. AK, CA-9, etc.)
def change_placemark_names(filename)
  # filename will be like "states/AK" or "cds/CA-16"
  place_name = File.basename filename
  
  # read file into Nokogiri
  file = File.open "#{filename}.kml"
  doc = Nokogiri::XML file
  file.close
  
  # find all the placemark name items and switch their names
  doc.css("Placemark name").each do |name|
    name.inner_html = place_name
  end
  
  # write over file
  out = File.open "#{filename}.kml", "w"
  out.write doc.to_xml
  out.close
end


# all 56 states, districts, and territories
def get_states
  @states.keys.each {|state| get_state state}
end

# all 441 districts for a given session of Congress
def get_districts(session)
  @states.each do |state, districts|
    if districts == 1
      get_state state, "cds/#{session}/#{state}-0"
    else # if districts > 1
      districts.times do |i| # 0 indexed
        get_cd session, state, i+1
        #sleep 1
      end
    end
  end
end

# state codes and number of districts apiece
@states = {
  "AL" => 7,
  "AK" => 1,
  "AZ" => 8,
  "AR" => 4,
  "AS" => 1,
  "CA" => 53,
  "CO" => 7,
  "CT" => 5,
  "DE" => 1,
  "DC" => 1,
  "FL" => 25,
  "GA" => 13,
  "GU" => 1,
  "HI" => 2,
  "ID" => 2,
  "IL" => 19,
  "IN" => 9,
  "IA" => 5,
  "KS" => 4,
  "KY" => 6,
  "LA" => 7,
  "ME" => 2,
  "MD" => 8,
  "MA" => 10,
  "MI" => 15,
  "MN" => 8,
  "MP" => 1,
  "MS" => 4,
  "MO" => 9,
  "MT" => 1,
  "NE" => 3,
  "NV" => 3,
  "NH" => 2,
  "NJ" => 13,
  "NM" => 3,
  "NY" => 29,
  "NC" => 13,
  "ND" => 1,
  "OH" => 18,
  "OK" => 5,
  "OR" => 5,
  "PA" => 19,
  "PR" => 1,
  "RI" => 2,
  "SC" => 6,
  "SD" => 1,
  "TN" => 9,
  "TX" => 32,
  "UT" => 3,
  "VT" => 1,
  "VA" => 11,
  "VI" => 1,
  "WA" => 9,
  "WV" => 3,
  "WI" => 8,
  "WY" => 1
}