Last active
November 24, 2016 08:14
-
-
Save tokland/4968851 to your computer and use it in GitHub Desktop.
Download video lectures from Coursera courses.
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
#!/usr/bin/ruby | |
# | |
# Download lecture videos of courses from Coursera (http://www.coursera.org). | |
# | |
# Install requirements: | |
# | |
# $ gem install curb trollop nokogiri capybara ruby-progressbar | |
# | |
# Example -- Download all video lectures of courses "Calculus: Single Variable" | |
# and "Introduction to Astronomy": | |
# | |
# $ ruby coursera-download-videos.rb | | |
# --email=user@server.org --password=123456 \ | |
# "Calculus: Single Variable" "Introduction to Astronomy" | |
# | |
# Contact: tokland@gmail.com | |
require 'rubygems' | |
require 'fileutils' | |
require 'logger' | |
require 'set' | |
require 'progressbar' | |
require 'capybara' | |
require 'nokogiri' | |
require 'trollop' | |
require 'curl' | |
### Extensions | |
class Hash | |
# Keep only given keys in hash | |
def slice(*keys) | |
keys_set = keys.to_set | |
select { |k, v| keys_set.include?(k) } | |
end | |
# Reverse update hash with a given default hash (performs key checking) | |
def defaults(default_hash) | |
unknown_options = self.keys - default_hash.keys | |
if unknown_options.empty? | |
replace(default_hash.merge(self)) | |
else | |
raise ArgumentError.new("unknown key(s): #{unknown_options.join(', ')}") | |
end | |
end | |
end | |
class MaybeWrapper | |
instance_methods.each { |m| undef_method(m) unless m == :object_id || m =~ /^__/ } | |
def method_missing(*args, &block) | |
nil | |
end | |
end | |
class Object | |
def maybe(&block) | |
if block_given? | |
nil? ? nil : yield(self) | |
else | |
nil? ? MaybeWrapper.new : self | |
end | |
end | |
end | |
module Curl | |
def self.download_to_file(url, destination, options = {}) | |
options.defaults(:progressbar => false, :cookies => []) | |
open(destination, "wb") do |fd| | |
curl = Curl::Easy.new(url).tap do |c| | |
c.follow_location = true | |
c.enable_cookies = true | |
if options[:cookies] | |
c.headers["Cookie"] = options[:cookies].join("; ") | |
end | |
c.on_body { |data| fd.write(data) } | |
end | |
if options[:progressbar] | |
title = File.basename(destination, File.extname(destination)) | |
state = {:pbar => nil, :dl_total => nil} | |
curl.on_progress do |dl_total, dl_now, ul_total, ul_now| | |
if dl_total > 0 | |
if !state[:pbar] || (!state[:dl_total] || dl_total > state[:pbar]) | |
state[:dl_total] = dl_total | |
state[:pbar] = ProgressBar.new(title, dl_total) | |
state[:pbar].format_arguments = | |
[:title, :percentage, :bar, :stat_for_file_transfer] | |
end | |
state[:pbar].set(dl_now) | |
end | |
true | |
end | |
curl.perform | |
state[:pbar].finish if state[:pbar] | |
else | |
curl.perform | |
end | |
end | |
end | |
end | |
class CapybaraBrowser | |
include Capybara::DSL | |
def initialize(options = {}) | |
options.defaults(:driver => :selenium, :default_wait_time => 60) | |
Capybara.current_driver = options.fetch(:driver) | |
Capybara.default_wait_time = options.fetch(:default_wait_time) | |
Capybara.run_server = false | |
end | |
def cookies | |
driver = Capybara.current_session.driver | |
case driver.class.name | |
when "Capybara::Webkit::Driver" | |
driver.browser.get_cookies.map { |s| s.split(";").first } | |
when "Capybara::Selenium::Driver" | |
driver.browser.manage.all_cookies.map { |c| [c[:name], c[:value]].join("=") } | |
else | |
raise ArgumentError.new("Unsupported driver: #{driver.class.name}") | |
end | |
end | |
end | |
### Application | |
class Coursera | |
attr_reader :browser, :logger | |
SessionError = Class.new(StandardError) | |
ParserError = Class.new(StandardError) | |
def initialize(options = {}) | |
@browser = CapybaraBrowser.new | |
@logger = Logger.new(STDERR) | |
@logger.formatter = proc { |severity, datetime, progname, msg| "#{msg}\n" } | |
end | |
def login(email, password) | |
logger.debug("Login: email='#{email}' password='#{'*' * password.size}'") | |
browser.visit("https://www.coursera.org/account/signin") | |
browser.fill_in("signin-email", :with => email) | |
browser.fill_in("signin-password", :with => password) | |
browser.click_button("Sign In") | |
browser.find(".coursera-header-account-name") | |
end | |
def download_videos(course_name, options = {}) | |
options.defaults(:destination_directory => nil) | |
browser.visit("https://www.coursera.org/") | |
if !browser.find(".coursera-header-account-name") | |
raise SessionError.new("Not logged in") | |
else | |
browser.click_link(course_name) | |
browser.find(:xpath, | |
'//*[@class="course-navbar-item"]/a[contains(@href, "lecture/index")]').click | |
browser.find("#spark") | |
get_videos_from_course_page(browser.html).map do |info| | |
course, section, index, lecture, url = info | |
directory = File.join([options[:destination_directory], course, section].compact) | |
path = File.join(directory, "%02d - %s.mp4" % [index+1, lecture]) | |
logger.debug("Download video: #{path}") | |
safe_download(url, path, browser.cookies) | |
path | |
end | |
end | |
end | |
private | |
def get_videos_from_course_page(html) | |
doc = Nokogiri::HTML(html) | |
course_name = doc.at_css("h1").maybe.text or | |
raise ParserError.new("Cannot find course name") | |
logger.debug("Course '#{course_name}'") | |
# Some courses have the videos in reverse order, detect this case | |
# with a simple heuristics (check if the first integers in the section | |
# titles are in descending order). | |
lis = doc.css(".course-item-list-header") | |
ns = lis.map { |li| li.text.match(/\d+/).maybe[0].maybe.to_i }.compact | |
is_reversed = !ns.empty? && ns.each_cons(2).all? { |x, y| x > y } | |
ordered_lis = is_reversed ? lis.reverse : lis | |
ordered_lis.flat_map.with_index do |section, section_index| | |
h3 = section.at_css("h3") or | |
raise ParserError.new("Cannot find h3") | |
section_title = h3.text.gsub(/[[:space:]]+/, ' ').strip | |
section_name = "%02d - %s" % [section_index+1, section_title] | |
lecture_video_list = section.next or | |
raise ParserError.new("Cannt find lecture video list") | |
lecture_video_list.css("li").map.with_index do |lecture, index| | |
link = lecture.at_css("a.lecture-link") or | |
raise ParserError.new("Cannot find lecture link") | |
lecture_title = link.text.gsub("/", "-").strip | |
url = lecture.at_css("a[title='Video (MP4)']").maybe["href"] or | |
raise ParserError.new("Cannot find video link") | |
[course_name, section_name, index, lecture_title, url] | |
end | |
end | |
end | |
def safe_download(url, path, cookies) | |
unless File.exists?(path) | |
FileUtils.mkpath(File.dirname(path)) | |
temp_path = path + ".partial" | |
begin | |
Curl.download_to_file(url, temp_path, :progressbar => true, :cookies => cookies) | |
FileUtils.copy(temp_path, path) | |
ensure | |
FileUtils.safe_unlink(temp_path) | |
end | |
end | |
end | |
end | |
if __FILE__ == $0 | |
options = Trollop.options do | |
banner "Usage: download-coursera-videos [OPTIONS] COURSENAME [COURSENAME2 ...]" | |
opt :destination_directory, "Directory destination", :type => :string | |
opt :email, "Email", :type => :string, :required => true | |
opt :password, "Password", :type => :string, :required => true | |
end | |
if ARGV.empty? | |
Trollop.die("Specify at least one course to download") | |
else | |
coursera = Coursera.new | |
coursera.login(options[:email], options[:password]) | |
download_options = options.slice(:destination_directory) | |
ARGV.each do |course_name| | |
coursera.download_videos(course_name, download_options) | |
end | |
end | |
end |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment