catalog_to_xpublish is now on xpublish-experiments

[xaviernogueira]

I'm happy to say catalog_to_xpublish is now fully open-sourced and xpublish-experiments.
See: https://github.com/xpublish-experiments/Catalog-To-Xpublish

It can be pip installed via

pip install git+https://github.com/xpublish-experiments/Catalog-To-Xpublish.git

Check out our example notebook to see what this enabled (the example uses a STAC catalog).

Features

• Serves cataloged .zarr or .nc data over a REST API without pre-loading.
• Parses catalog hierarchy and creates matching URL paths/prefixes (i.e., "https://server.com/catalog1/sub_catalog2/sub_catalog_5/datasets/{dataset_id}").
• Currently supports both Intake and STAC catalogs.
• Fully compatible with any 3rd xpublish dataset router, or app router plugins. For example, xpublish-opendap can be included to provide key OPeNDAP capabilities (i.e., slicing).
• Leverages a factory pattern architecture and Python Abstract Base Classes (ABCs) to streamline the addition of more catalog schema as needs develop.
• Flexible logging configuration. Log into existing streams, or files.

Reliability

• Currently test coverage is at 86%. Both the intake and pySTAC implementations are tested.
• See Codecov report here: https://app.codecov.io/gh/xpublish-experiments/Catalog-To-Xpublish
• A known xpublish bug can be hit using the example catalogs (see Issue 207).

Future

• A Dockerized deployment of catalog_to_xpublish will be used to replace the backend of the THREDDs data server!
• Dask integration and async support will likely be added in the near future.
• Funding exists to continue to maintain, expand, and improve this framework.