Feature request: transforms read array data

Propose an Improvement
1

I would like to make a feature request to the foundry-transforms-lib-python library.

Current it only supports tabular data and I would like it to support array data.

I would like the API to be like Dataset.get("air_temperature").read_array(format="xarray")

I think it should be easy to copy the read_table code to a read_array version.

For examples, you could have files
_array_dataset/_array_dataset.py and _array_dataset/_xarray_dataset.py

which would look like

class ArrayDataset(ABC, Generic[T]):
    def open(self) -> T:
        schema_to_readers = {
            # could expand to anything xarray.open_dataset can read e.g.
            # netcdf, grib etc.
            FileType.ZARR: "zarr",
        }

    @abstractmethod
    def open_dataset(self, engine: str = "zarr") -> T:
        pass

class XarrayDataset(ArrayDataset):
    def open_dataset(self, engine: str = "zarr) -> "xarray.Dataset":
        return self._xarray().open_dataset(self._path, engine=self._engine)
1 Like
2

Seconded. This feature would be helpful.