Desktopography downloader

desktopography-dl.rb

#!/usr/bin/env ruby

# desktopography-dl.rb
# Requires: nokogiri
# 
# Downloads /everything/ from a given Desktopography exhibition. Assumes we're on
# something unixy, with mkdir and wget available for use.


require 'open-uri'
require 'nokogiri'

# pull down list of exhibitions
exhibitions_doc = Nokogiri::HTML(open 'http://www.desktopography.net/')
exhibitions = Hash[exhibitions_doc.css('#exhibitions .exhibition').map { |e| [e.text.strip.to_i, e['href']] }]

# which one, which one...
print "Exhibition year? (#{exhibitions.keys.join(', ')})\n> "
year = gets.strip.to_i

# grab wallpaper list for this exhibition
wallpapers_doc = Nokogiri::HTML(open "http://www.desktopography.net#{exhibitions[year]}")
wallpapers = wallpapers_doc.css('#wallpapers .wallpaper')
puts "#{wallpapers.count} wallpapers found."

# for each wallpaper...
wallpapers.each do |wallpaper|

  # requisite info
  wallpaper_doc = Nokogiri::HTML(open "http://www.desktopography.net/#{wallpaper['href']}")

  # name appropriately
  name = wallpaper_doc.css('#title').text.gsub(/[^\w]/, '_').gsub(/(^_+|_+$)/, '') + '.jpg'

  # download each resolution - stored in Desktopography/:year/:ratio/:resolution/:name.jpg
  wallpaper_doc.css('.download select option').each do |option|
    path = ['Desktopography', year, option.parent['label'], option.text, name].join('/')

    # skip if possible
    if File.exists? path
      puts "#{path} exists - skipping..."
      next
    end

    # proceed
    puts "Downloading #{path}..."
    `mkdir -p "#{File.dirname path}"`
    `wget -O "#{path}" "http://www.desktopography.net/#{option['value']}"`
  end

end