Coursera Downloader

Coursera is arguably the leader in massive open online courses (MOOC) with a selection of more than 300 classes from 62 different institutions as of February 2013. Generous contributions by educators and institutions are making excellent education available to many who could not afford it otherwise. There are even non-profits with "feet on the ground" in remote areas of the world who are helping spread the wealth (see the feedback below from Tunapanda).

This script makes it easier to batch download lecture resources (e.g., videos, ppt, etc) for Coursera classes. Given one or more class names and account credentials, it obtains week and class names from the lectures page, and then downloads the related materials into appropriately named files and directories.

Why is this helpful? A utility like wget can work, but has the following limitations:

1. Video names have a number in them, but this does not correspond to the actual order. Manually renaming them is a pain.
2. Using names from the syllabus page provides more informative names.
3. Using a wget in a for loop picks up extra videos which are not posted/linked, and these are sometimes duplicates.

DownloadThemAll is another possiblity, but this script provides more features such as appropriately named files.

This work was originally inspired in part by youtube-dl by which I've downloaded many other good videos such as those from Khan Academy.

Features

• Intentionally detailed names, so that it will display and sort properly on most interfaces (e.g., MX Video on Android phone).
• Regex-based section (week) and lecture name filters to download only certain resources.
• File format extension filter to grab resource types you want.
• Login credentials accepted on command-line or from .netrc file
• Core functionality tested on Linux, Mac and Windows.

Directions

Requires Python 2.6 (or newer) and a free Coursera account enrolled in the class of interest.

1. Install any missing dependencies.

• Beautiful Soup 3
  • Ubuntu/Debian: sudo apt-get install python-beautifulsoup
  • Mac OSX + MacPorts: sudo port install py-beautifulsoup
  • Other: easy_install beautifulsoup
• Beautiful Soup 4: An alternative to Beautiful Soup 3. See also html5lib below.
  • Ubuntu/Debian: sudo apt-get install python-bs4
  • Mac OSX + MacPorts: sudo port install py-beautifulsoup4
  • Other: easy_install beautifulsoup4
• Argparse: Only necessary if using Python 2.6.
  • Ubuntu/Debian: sudo apt-get install python-argparse
  • Other: easy_install argparse
• easy_install: Only necessary if not using prepackaged dependencies.
  • Ubuntu/Debian: sudo apt-get install python-setuptools
• html5lib: Recommended if using Beautiful Soup 4.
  • Ubuntu/Debian: sudo apt-get install python-html5lib
  • Mac OSX + MacPorts: sudo port install py-html5lib
  • Other: easy_install html5lib

On Mac OSX using MacPorts, the following may be used:

port
> install py-beautifulsoup
> install py-argparse
> install py24-distribute  # for "py-setuptools", the obsolete name

If you are using pip, you can directly install all the dependencies from the requirements file using pip install -r requirements.txt.

2. Create a Coursera.org account and enroll in a class. e.g. http://saas-class.org

3. Run the script to download the materials by providing your Coursera account (e.g., email address), password (or a ~/.netrc file), the class names

General:                     coursera-dl -u <user> -p <pass> saas
Multiple classes:            coursera-dl -u <user> -p <pass> saas nlp proglang-2012-001
Filter by section name:      coursera-dl -u <user> -p <pass> -sf "Chapter_Four" saas
Filter by lecture name:      coursera-dl -u <user> -p <pass> -lf "3.1_" saas
Download only ppt files:     coursera-dl -u <user> -p <pass> -f "ppt" saas
Use a ~/.netrc file:         coursera-dl -n saas
Get the preview classes:     coursera-dl -b saas
Specify download path:       coursera-dl -n --path=C:\Coursera\Classes\ saas

Maintain a list of classes in a dir:
  Initialize:              mkdir -p CURRENT/{class1,class2,..classN}
  Update:                  coursera-dl -n --path CURRENT `ls CURRENT`

On *nix platforms*, the use of a ~/.netrc file is a good alternative to specifying both your username and password every time on the command line. To use it, simply add a line like the one below to a file named .netrc in your home directory (or the equivalent, if you are using Windows) with contents like:

machine coursera-dl login <user> password <pass>

Create the file if it doesn't exist yet. From then on, you can switch from using -u and -p to simply call coursera-dl with the option -n instead. This is especially convenient, as typing usernames and passwords directly on the command line can get tiresome (even more if you happened to choose a "strong" password).

* if this works on Windows, please add additional instructions for it if any are needed.

Troubleshooting

• When reporting bugs against coursera-dl, please don't forget to include enough information so that you can help us help you:

  • Is the problem happening with the latest version of the script?
  • What is the course that you are trying to access:
  • What is the precise command line that you are using (feel free to hide your username and password with asterisks, but leave all other information untouched).
  • What are the precise messages that you get? Please, copy and past them. Don't reword the messages.

• Make sure the classname you are using corresponds to the resource name used in the URL for that class: https://class.coursera.org/<CLASS_NAME>/class/index

• One can export a Netscape-style cookies file with a browser extension (1, 2) and use it with the -c option. This comes in handy when the authentication via password is not working (the authentication process changes now and then).

• If results show 0 sections, you most likely have provided invalid credentials (username and/or password in the command line or in your .netrc file).

• For courses that have not started yet, but have had a previous iteration sometimes a preview is available, containing all the classes from the last course. These files can be downloaded by passing the -b parameter.

• If you are using Beautiful Soup 4, make sure you have installed html5lib:

    $ python
    >>> import html5lib
    >>> print(html5lib.__version__)
    0.95-dev
  

• If you get an error like Could not find class: <CLASS_NAME>:

  • Verify that the name of the course is correct. Current class names in coursera are composed by a short course name e.g. class and the current version of the course e.g. class-001, class-002, etc.
  • Second, verify that you are enrolled in the course. You won't be able to access the course materials if you are not officially enrolled and agreed to the honor course via the website.

Feedback

I enjoy getting feedback. Here are a few of the comments I've received:

• "Thanks for the good job! Knowledge will flood the World a little more thanks to your script!"
  Guillaume V. 11/8/2012

• "Just wanted to send you props for your Python script to download Coursera courses. I've been using it in Kenya for my non-profit to get online courses to places where internet is really expensive and unreliable. Mostly kids here can't afford high school, and downloading one of these classes by the usual means would cost more than the average family earns in one week. Thanks!"
  Jay L., Tunapanda 3/20/2013

• "I am a big fan of Coursera and attend lots of different courses. Time constraints don't allow me to attend all the courses I want at the same time. I came across your script, and I am very happily using it! Great stuff and thanks for making this available on Github - well done!"
  William G. 2/18/2013

• "This script is awesome! I was painstakingly downloading each and every video and ppt by hand -- looked into wget but ran into wildcard issues with HTML, and then.. I came across your script. Can't tell you how many hours you've just saved me :) If you're ever in Paris / Stockholm, it is absolutely mandatory that I buy you a beer :)"
  Razvan T. 11/26/2012

• "Thanks a lot! :)"
  Viktor V. 24/04/2013

Contact

Post bugs and issues on github. Send other comments to John Lehmann: first last at geemail dotcom or @jplehmann