facebookresearch/fastText ? reverse-engineered prompt

Build me a fastText style library and command line tool for learning word vectors and training text classifiers from plain text data.

I want to be able to feed in a UTF 8 training file, train on it quickly, then save a model and use it later to print word vectors, including for words that were not in the training set. I also want a simple supervised mode for text classification, with labels written in the input text, plus commands to test accuracy and predict the most likely labels for new text.

Please include a clean way to build it from the source, and make sure it works as a reusable library with Python support too. If you need to check current docs or build steps online, go ahead. Use the examples and common defaults people would expect from a project like this, and keep it easy to run from the command line.