Téléchargement de toutes les conférences Paris Web 2012 au format audio. A écouter sans modération.

.gitignore

tmp
*part
*mp4
*m4a

download-parisweb-2012

#!/usr/bin/env ruby
## Encoding : UTF-8
require "shellwords"
require "digest/md5"
require "fileutils"
require "nokogiri"
require "open-uri"
require "addressable/uri"


class DownloadParisWeb

	def initialize
		check_dependencies
		@list_url = "http://www.paris-web.fr/2012/"

		FileUtils.mkdir_p('./tmp')
		@done_file = File.expand_path('./tmp/done')
		%x[touch #{@done_file.shellescape}] unless File.exists?(@done_file)
	end

	def check_dependencies
		if %x[which youtube-dl] == ""
			puts "You need to have youtube-dl installed to use this script"
			puts "Check https://github.com/rg3/youtube-dl"
			exit
		end
		if %x[youtube-dl --version] < "2013.09.07"
			puts "Your version of youtube-dl is too old."
			puts "Run youtube-dl -U to update to the latest"
			exit
		end
	end

	# Given a url, returns the cache location on disk
	def get_cache_file(url)
		File.expand_path(File.join("./", "tmp", Digest::MD5.hexdigest(url)+".url"))
	end

	# Returns a webpage html source, using cache if found
	def get_page_source(url)
		cache_file = get_cache_file(url)

		# Returning cached version if exists and young enough
		if File.exists?(cache_file)
			return File.open(cache_file, "r").read
		end

		# Downloading page
		# puts "Downloading #{url}"
		content = open(url, 'r').read
		File.open(cache_file, "w") do |cache|
			cache.write(content)
		end
		return content
	end

	def run
		# Getting list of all talk pages
		list_content = get_page_source(@list_url)
		doc = Nokogiri::HTML(list_content)
		links = doc.css('table.programme a.url').map(){|l| l['href']}

		done_links = File.readlines(@done_file).map(&:chomp)

		links.each do |link|
			content = get_page_source(link)
			subdoc = Nokogiri::HTML(content)
			# Getting title
			title = subdoc.css('title').text.gsub('Paris Web – ', '')
			# Gettin dailymotion iframes
			iframes = subdoc.css('iframe')
			iframes.each do |iframe|
				uri = Addressable::URI.parse(iframe['src'])
				next unless uri.host == "www.dailymotion.com"
				# Converting url to a format youtube-dl can parse
				uri.query = nil
				uri.path.gsub!('/embed', '')
				url = uri.to_s

				next if done_links.include?(url)

				# Downloading sound
				puts "Downloading #{title}"
				%x[youtube-dl -t -x "#{url}"]

				# Marking file as downloaded
				%x[echo "#{url}" >> #{@done_file}]

			end

		end
	end


end


DownloadParisWeb.new().run()