overture-dataset-configs

Dataset configs for Overture providers, the per-theme license policy, and the classes that validate, load and render them.

from overture_dataset_configs import load_config, provider_names
from overture_dataset_configs.dataset import Dataset
from overture_dataset_configs.license_policy import load_policy

provider_names()  # ["alltheplaces", "birdeye", ...]
load_config("linz")  # raw dict from overture_dataset_configs/datasets/linz.json

ds = Dataset.from_name("osm", "planet")  # packaged configs by default
ds.dataset_id, ds.license_type, ds.download_url

load_policy().allowed("places")  # {"CDLA-Permissive-2.0", ...}

For a command line, overture-core's overture-datasets CLI validates and renders these configs.

Concepts

• Provider: an organization Overture takes data from, such as Land Information New Zealand.
• Dataset (a resources entry): one licensed collection of data from a provider, such as a single layer or file set. A provider can have many datasets, and each has its own license, coverage and refresh cadence.
• License policy: per theme, the license identifiers Overture accepts. Every dataset's license type should appear in it.

For how themes, attribution, and licensing work across Overture data, see the Overture documentation: the schema for themes, and the attribution page for the terms published with each release.

Each file in overture_dataset_configs/datasets describes one provider and its datasets. Files are named <provider_label>.json and looked up by file stem. The field reference is schema.py, which also rejects unknown keys.

Contributing

1. Edit or add a file, then run uv sync --extra dev and uv run pytest from this directory. The tests validate every file against the schema, plus semantic checks (license ids and URLs, attribution, URL pairs), and run the configs through the overture-datasets CLI (validate --policy, license).
2. A new field needs a matching change to schema.py in the same PR.
3. Provider and resource labels feed dataset IDs and S3 paths, so don't rename them.