csurfer/rake-nltk ? reverse-engineered prompt

Build me a small Python library that extracts keywords and key phrases from text using the RAKE method, with NLTK handling the language stuff like stop words and tokenizing.

I want the main experience to be really simple, so someone can create a Rake object, pass in either a block of text or a list of sentences, and then get back ranked phrases, with and without scores. Please make it easy to change things like the stop words, sentence and word tokenizers, and the way phrases are ranked.

Also include good tests, a short README example that shows how to use it, and make sure it works cleanly as a reusable package. If you need to check current NLTK or packaging details, look up the latest docs online.