This relies on CMU's new schedule of classes site at https://enr-apps.as.cmu.edu/open/SOC/SOCServlet/search. This works as of March 2014.
First, install the necessary packages into your Python 3 installation. You can use make install-scrape to do so. You may end up installing them into Python 2, in which case, use pip3 manually.
Second, are you sure you really want to redownload all the data? You can try the pre-downloaded data stored in raw.tgz by running make unpack-raw. Afterward, to select only the pages you want to redownload, delete the relevant files.
Third, scrape the data by running python3 pipeline.py. If you are editing the scraper, you may want to set force_scrape=True in get_all() so you can test it on all the courses.
You should end up with a folder processed with a bunch of .json files. If there are any errors, you may review them in pipeline.log.
Install MongoDB. This will vary depending on your development environment.
Install the necessary packages with make install-site.
If you haven't already, unpack the course data with make unpack-processed.
Then, convert the course data into a format required by the website by running python3 construct.py.
If you want, set a port number with export PORT=<number>. The default port is 8080.
For development, just run python site/main.py and go to http://localhost:<port number>.
Sometimes, on a force quit with Ctrl+C, the bottle server doesn't actually stop and it doesn't release its port. On Mac OS X, you can run lsof -i :<port number> and kill <pid> to kill the python process.