Finding recurring schemes in a code base

Our company has a number of tests written, mostly in Python. My task is to examine the tests, find recurring schemes that will eventually be made into a library whose objects the testers will use in upcoming automated test scripts.

My idea is to read the whole code base into memory, omit Python keywords and use a collections.Counter object to count each occurence of a word, and inspect the top X most frequent words. I will experiment with splitting the text using characters like ".". Maybe I could also try grouping words of 2…4 together to find recurring phrases in code.

Do you have other ideas?