Table of contents
If something looks wrong, or if you have a question, please fille out a git issue
Many questions have already been asked and answered, so please do search all issues, including closed issues, before filing a new issue. Also, please upgrade to the latest version first.
If you find yourself inspired to fix an issue or add a feature, you can send us a pull request! Please direct the pull request to the
Stanza has a large suite of unit tests which you can run as part of the pull request. Ideally, you would even add testing for a new feature you add.
The tests use some models several times, along with a CoreNLP distribution, so those pieces are downloaded in a setup script:
python3 -m stanza.tests.setup
By default, the data objects will be downloaded to
stanza_test in whatever directory you run the setup script from. This can be changed with the
TEST_HOME_VAR environment variable.
The tests use
pytest, which you may need to
pip install. After that, you can run them with
pytest -s stanza/tests in your git clone. (You probably won’t need to run the tests unless you are about to make a PR, anyway)
Generally speaking, we are happy to include new models in our distributions. Even better is when we have access to the data and can rebuild the models in the future when the underlying data format or the model structure changes.
The best thing in general is to file a new git issue, and we will discuss the specific case in question.