Python: How to load and use trained and pickled NLTK tagger to GAE?

开发者_开发知识库I have a trained and pickled NLTK tagger (Brill's transformational rule-based tagger). I want to use it on GAE. What the best way to do it?

If your NLTK tagger code and data is of limited size, then carry it along with your GAE code.

If you have to act upon it to retrain the set, then storing the content of the file as a BLOB in the datastore would be an option, so that you get, analyze, retrain and put.But that will limit size of dataitem to be less than 1 MB because of GAE hardlimit.