gpq
gpq copied to clipboard
GeoParquet 1.1 support
It'd be great to more broadly support GeoParquet 1.1. There's a range of what could be done with GPQ, in rough order of importance / effort:
- Validate the new 1.1 features - recognize the new encodings, and check for the bbox (perhaps recommend adding if it's not there)
- Write support for bounding box column - writing without trying to sort would be a good first step, and just assume/hope the file has a decent sort order
- Option to sort by r-tree or some other nice spatial option when writing bounding box column
- Read a subset of data leveraging the bbox column. Reading with bbox is less important right now as GPQ doesn't have a notion of filtering by bounds / getting a subset, but if it added it then it could efficiently grab overture data and other large datasets.
- native encoding / geoarrow support, for both read and write. Perhaps would not be a huge amount of work as there's likely good arrow support in the underlying parquet/arrow libraries?