openworld_radio_twin.providers.overture_reader#
Bounded Overture GeoParquet reads that share one S3 client and one STAC index.
overturemaps.record_batch_reader downloads the release’s STAC index and opens a
new S3 client on every call, and it reads every column of the matching row groups.
The compiler issues six such reads per scene, so this module keeps the client for
the life of the process, keeps the index of each release in memory and on disk
(releases are immutable), and projects each read to the columns the providers
decode. The row filter is the same bounding-box predicate. The index is the only
way objects are selected; when it cannot be fetched after a few retries, the query
fails and the calling provider reports that.
Functions
|
|
Directory holding one cached STAC index per release. |
|
|
Parquet objects of |
The current Overture release. |
|
|
The non-empty record batches inside |
- openworld_radio_twin.providers.overture_reader.latest_release()[source]#
The current Overture release.
- Return type:
- openworld_radio_twin.providers.overture_reader.index_root()[source]#
Directory holding one cached STAC index per release.
- Return type:
- openworld_radio_twin.providers.overture_reader.intersecting_files(overture_type, bbox, release)[source]#
Parquet objects of
overture_typewhose extent crossesbbox.Index files fetched before late September 2026 name each object’s type in
collection. The files served since then, those of earlier releases included, leave that column empty and name the type only in the object path (.../theme=buildings/type=building/part-...), so both forms are accepted.