e-codices downloader - http://www.e-codices.unifr.ch/

ecodices-dl.rb

#!/usr/bin/env ruby

require 'nokogiri'
require 'open-uri'

def download_pages(uri, id)
	doc = Nokogiri::HTML(open(uri))
	doc.xpath('//select[@id = "page_select"]/option/@value').each do |value|
		page = value.to_s.split('/').last
		page_uri = "#{uri}/#{page}/x-large"
		page_doc = Nokogiri::HTML(open(page_uri))
		page_doc.xpath('//img[@id = "msImage"]/@src').each do |image|
			`wget "#{image.to_s}"`
		end
	end
end

# e.g. 10.5076/e-codices-kba-Ms-FZ-0018
doi = ARGV[0]
# kba-Ms-FZ-0018
id = doi[/e-codices-(.*)/,1]
# kba
collection = id.split('-').first
# Ms-FZ-0018
document = id[/#{collection}-(.*)/,1]

prefix = "http://www.e-codices.unifr.ch/en/"
pages_uri = prefix + "#{collection}/#{document}"
binding_uri = prefix + "binding/#{collection}/#{document}"

`wget "http://www.e-codices.unifr.ch/xml_descriptions/#{id}.xml"`
[binding_uri, pages_uri].each {|uri| download_pages(uri, id)}