Ruby script to import tiger/line 2011 data into geocommons geocoder.db

tiger2011_import.rb

#!/bin/env ruby

require 'tmpdir'

tmp      = Dir.tmpdir + "/TIGER2011"
base     = Dir.pwd + "/build"
shp2     = base + "/shp2sqlite"
sql      = base + "/sql"
lib      = Dir.pwd + "/lib/geocoder/us/sqlite3.so"
database = ARGV.shift
tiger    = ARGV.shift

STDERR.puts "Building in #{tmp}..."
system("mkdir -p '#{tmp}'")
system("cat '#{sql}/create.sql' '#{sql}/place.sql' | sqlite3 '#{database}'") unless File.exists? database

system("echo '.load #{lib}' > '#{tmp}/loading.sql'")
system("cat '#{sql}/setup.sql' >> '#{tmp}/loading.sql'")

['EDGES', 'FEATNAMES', 'ADDR'].each do |dir|
  path = tiger + '/' + dir
  Dir.open(path).each do |zip|
    if zip =~ /^(.*?)\.zip$/i
      STDERR.puts "Unzipping: "+path + "/" + zip+"..."
      # Skip unzipping the same zip twice
      full = tmp + "/" + $1
      system("/usr/bin/unzip", "-q", path + "/" + zip, "-d", tmp)
      if dir == 'EDGES'
        STDERR.puts "Importing: #{full}.shp..."
        system(shp2+" -aS '#{full}.shp' 'tiger_#{dir.downcase}' >> '#{tmp}/loading.sql'")
      else
        STDERR.puts "Importing: #{full}.dbf..."
        system(shp2+" -an '#{full}.dbf' 'tiger_#{dir.downcase}' >> '#{tmp}/loading.sql'")
      end
    end
  end
end

STDERR.puts "Adding conversion..."
system("cat '#{sql}/convert.sql' >> '#{tmp}/loading.sql'")
STDERR.puts "Doing the import..."
system("cat '#{tmp}/loading.sql' | sqlite3 '#{database}'")