Download NASA Astronomy Picture Of Day media. (Ruby version)

apod.rb

#!/usr/bin/env ruby
## 
#  This script will grab NASA's Astronomy Picture of the Day Archive
#
#    It will create 2 directories in current directory: big and small
#    to save pictures. 
#  
#  Requirements:
#    1. Ruby 1.8.x with Rubygems
#    2. nokogiri gem: (sudo) gem install nokogiri
#      2.1 It need libxml2 and libxslt and their source to build
#    3. curb gem: (sudo) gem install curb
#      3.1 curb need curl and its source to build
#

require 'rubygems'
require 'open-uri'
require 'nokogiri'
require 'curl'
require 'fileutils'

BASE = 'http://antwrp.gsfc.nasa.gov/apod/'
FileUtils.mkdir('small') unless File.exist?('small')
FileUtils.mkdir('big') unless File.exist?('big')

f = open 'http://antwrp.gsfc.nasa.gov/apod/archivepix.html'
html_doc = Nokogiri::HTML(f.read)
html_doc.xpath('//b//a').each do |element|
  imgurl = BASE + element.attributes['href'].value
  doc = Nokogiri::HTML(open(imgurl).read)
  doc.xpath('//p//a//img').each do |elem|
    small_img = BASE + elem.attributes['src'].value
    big_img = BASE + elem.parent.attributes['href'].value
    s_img_f = open("small/#{File.basename(small_img)}", 'wb')
    b_img_f = open("big/#{File.basename(big_img)}", 'wb')
    rs_img = Curl::Easy.new(small_img)
    rb_img = Curl::Easy.new(big_img)
    rs_img.perform
    s_img_f.write(rs_img.body_str) 
    rb_img.perform
    b_img_f.write(rb_img.body_str)
    s_img_f.close
    puts "Download #{File.basename(small_img)} finished."
    b_img_f.close
    puts "Download #{File.basename(big_img)} finished."
    rs_img.close
    rb_img.close
  end
end

puts "All done."