seatgeek/thefuzz — reverse-engineered prompt

Build me a small Python library for fuzzy string matching. I want to be able to compare two pieces of text and get a score from 0 to 100 showing how similar they are, even when there are typos, extra words, missing punctuation, or words in a different order.

Please make it easy to use from Python, with imports like a normal package. Include functions for a basic similarity score, partial matching, matching after sorting words, matching repeated or shared words, and partial versions of those word based matches. I also want a helper that takes one search phrase and a list of choices, then returns the best matches with their scores, including a way to get only the single best match.

Use a reliable Levenshtein style approach, and it’s fine to use rapidfuzz underneath. Add clear examples in the README, plus tests that cover normal strings, messy strings, empty input, and finding the best choice from a list. Make it installable with pip and keep the API simple.