hankcs/HanLP ? reverse-engineered prompt

Build me a Python based multilingual NLP library like HanLP that can analyze text in English, Chinese, Japanese, and other languages. I want it to let users quickly load a pretrained model and get tokenization, part of speech tags, named entities, dependency parsing, constituency parsing, semantic role labeling, and document classification from simple text input.

Please make it easy to use from both a native Python API and a lightweight REST style client, with a simple quick start that feels friendly for researchers and developers. It should work well out of the box, but also allow training custom models on your own data and saving them for later use. If you need current docs or model details, look them up online.

Focus on making the library clean, efficient, and practical, with examples, tests, and a basic docs structure so someone can understand how to use it right away.