get_city_data

require 'rubygems'
require 'hpricot'
require 'open-uri'
require 'yaml'

doc = Hpricot.parse(open("http://www.lastminute.com"))
destinations = doc.search("select#destination option")

puts "destinations.size #{destinations.size}"
cities = {}

destinations.each do |link|
   unless link[:value].nil? or link[:value] == 'None'
	cities[link.inner_text] = { :name => link.inner_text, :IATA => link[:value] }
   end
end

puts "cities.size - #{cities.size}"
puts "cities.sort.first.inspect - #{cities.sort.first.inspect}"

clean_cities = []

cities.sort.each do | key,value |
  clean_cities.push(value)
end

puts clean_cities.size

if false
city = cities.sort.first
puts city[0]
wiki_page = "http://en.wikipedia.org/wiki/#{city[0]}"

puts "Getting wiki page #{wiki_page}"
wiki = Hpricot.parse(open(wiki_page))

geohack_page = wiki.search('span.geo-dms').first.parent.parent[:href]

puts "Getting geohack page #{geohack_page}"
geohack = Hpricot.parse(open(geohack_page))

latitude = geohack.search('span.latitude').first.inner_text
longitude = geohack.search('span.longitude').first.inner_text

puts latitude
puts longitude
end

File.open('cities.yml', 'w') {|f| f.write(YAML::dump(clean_cities)) }