I would guess the intent behind using vanilla python instead of libraries was to help learners understand exactly what's going on instead of just plugging stuff into a black box.
He says at the end:
"Naive Bayes is great because it's fairly easy to see what's going on under the hood. It's a great way to start any text analysis and it can easily scale out of core to work in a distributed environment. There are some excellent implementations in the Python community you can use as well, so if you don't want to roll your own, have no fear! The scikit-learn and nltk versions are great places to start."
