BUG: Squeeze when using interpolate_na with extra dim

[snowman2]

Possibly related to: pydata/xarray#9372

https://github.com/corteva/rioxarray/actions/runs/11151510847/job/30995110503

_______________ test_interpolate_na__all_nodata[open_rasterio] ________________

interpolate_na_nan = {'compare': '/home/runner/work/rioxarray/rioxarray/test/test_data/compare/MODIS_ARRAY_INTERPOLATE_NAN.nc', 'input': '/...nner/work/rioxarray/rioxarray/test/test_data/input/MODIS_ARRAY.nc', 'open': <function open_rasterio at 0x7f3fefa3b4c0>}

    def test_interpolate_na__all_nodata(interpolate_na_nan):
        rio_opened = "open_rasterio " in str(interpolate_na_nan["open"])
        with interpolate_na_nan["open"](
            interpolate_na_nan["input"], mask_and_scale=True
        ) as mda, interpolate_na_nan["open"](
            interpolate_na_nan["compare"], mask_and_scale=True
        ) as mdc:
            if hasattr(mda, "variables"):
                for var in mda.rio.vars:
                    mda[var].values[~numpy.isnan(mda[var].values)] = numpy.nan
            else:
                mda.values[~numpy.isnan(mda.values)] = numpy.nan
    
>           interpolated_ds = mda.rio.interpolate_na()

test/integration/test_integration_rioxarray.py:1517: 
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ 
rioxarray/raster_array.py:1063: in interpolate_na
    interp_array = xarray.DataArray(
../../../micromamba/envs/test/lib/python3.11/site-packages/xarray/core/dataarray.py:479: in __init__
    coords, dims = _infer_coords_and_dims(data.shape, coords, dims)
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ 

shape = (1, 1, 200, 200)
coords = Coordinates:
  * band         (band) int64 8B 1
  * x            (x) float64 2kB -7.274e+06 -7.274e+06 ... -7.228e+06 -7.228e+06
  * y            (y) float64 2kB 5.05e+06 5.05e+06 ... 5.004e+06 5.004e+06
    spatial_ref  int64 8B 0
dims = ('band', 'y', 'x')

    def _infer_coords_and_dims(
        shape: tuple[int, ...],
        coords: (
            Sequence[Sequence | pd.Index | DataArray | Variable | np.ndarray]
            | Mapping
            | None
        ),
        dims: str | Iterable[Hashable] | None,
    ) -> tuple[Mapping[Hashable, Any], tuple[Hashable, ...]]:
        """All the logic for creating a new DataArray"""
    
        if (
            coords is not None
            and not utils.is_dict_like(coords)
            and len(coords) != len(shape)
        ):
            raise ValueError(
                f"coords is not dict-like, but it has {len(coords)} items, "
                f"which does not match the {len(shape)} dimensions of the "
                "data"
            )
    
        if isinstance(dims, str):
            dims = (dims,)
        elif dims is None:
            dims = [f"dim_{n}" for n in range(len(shape))]
            if coords is not None and len(coords) == len(shape):
                # try to infer dimensions from coords
                if utils.is_dict_like(coords):
                    dims = list(coords.keys())
                else:
                    for n, (dim, coord) in enumerate(zip(dims, coords, strict=True)):
                        coord = as_variable(
                            coord, name=dims[n], auto_convert=False
                        ).to_index_variable()
                        dims[n] = coord.name
        dims_tuple = tuple(dims)
        if len(dims_tuple) != len(shape):
>           raise ValueError(
                "different number of dimensions on data "
                f"and dims: {len(shape)} vs {len(dims_tuple)}"
            )
E           ValueError: different number of dimensions on data and dims: 4 vs 3

../../../micromamba/envs/test/lib/python3.11/site-packages/xarray/core/dataarray.py:185: ValueError
_____________ test_interpolate_na__all_nodata[interpolate_na_nan3] _____________

interpolate_na_nan = {'compare': '/home/runner/work/rioxarray/rioxarray/test/test_data/compare/MODIS_ARRAY_INTERPOLATE_NAN.nc', 'input': '/...st_data/input/MODIS_ARRAY.nc', 'open': functools.partial(<function open_dataset at 0x7f3fefd97880>, engine='rasterio')}

    def test_interpolate_na__all_nodata(interpolate_na_nan):
        rio_opened = "open_rasterio " in str(interpolate_na_nan["open"])
        with interpolate_na_nan["open"](
            interpolate_na_nan["input"], mask_and_scale=True
        ) as mda, interpolate_na_nan["open"](
            interpolate_na_nan["compare"], mask_and_scale=True
        ) as mdc:
            if hasattr(mda, "variables"):
                for var in mda.rio.vars:
                    mda[var].values[~numpy.isnan(mda[var].values)] = numpy.nan
            else:
                mda.values[~numpy.isnan(mda.values)] = numpy.nan
    
>           interpolated_ds = mda.rio.interpolate_na()

test/integration/test_integration_rioxarray.py:1517: 
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ 
rioxarray/raster_dataset.py:441: in interpolate_na
    .rio.interpolate_na(method=method)
rioxarray/raster_array.py:1063: in interpolate_na
    interp_array = xarray.DataArray(
../../../micromamba/envs/test/lib/python3.11/site-packages/xarray/core/dataarray.py:479: in __init__
    coords, dims = _infer_coords_and_dims(data.shape, coords, dims)
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ 

shape = (1, 1, 200, 200)
coords = Coordinates:
  * band         (band) int64 8B 1
  * x            (x) float64 2kB -7.274e+06 -7.274e+06 ... -7.228e+06 -7.228e+06
  * y            (y) float64 2kB 5.05e+06 5.05e+06 ... 5.004e+06 5.004e+06
    spatial_ref  int64 8B 0
dims = ('band', 'y', 'x')

    def _infer_coords_and_dims(
        shape: tuple[int, ...],
        coords: (
            Sequence[Sequence | pd.Index | DataArray | Variable | np.ndarray]
            | Mapping
            | None
        ),
        dims: str | Iterable[Hashable] | None,
    ) -> tuple[Mapping[Hashable, Any], tuple[Hashable, ...]]:
        """All the logic for creating a new DataArray"""
    
        if (
            coords is not None
            and not utils.is_dict_like(coords)
            and len(coords) != len(shape)
        ):
            raise ValueError(
                f"coords is not dict-like, but it has {len(coords)} items, "
                f"which does not match the {len(shape)} dimensions of the "
                "data"
            )
    
        if isinstance(dims, str):
            dims = (dims,)
        elif dims is None:
            dims = [f"dim_{n}" for n in range(len(shape))]
            if coords is not None and len(coords) == len(shape):
                # try to infer dimensions from coords
                if utils.is_dict_like(coords):
                    dims = list(coords.keys())
                else:
                    for n, (dim, coord) in enumerate(zip(dims, coords, strict=True)):
                        coord = as_variable(
                            coord, name=dims[n], auto_convert=False
                        ).to_index_variable()
                        dims[n] = coord.name
        dims_tuple = tuple(dims)
        if len(dims_tuple) != len(shape):
>           raise ValueError(
                "different number of dimensions on data "
                f"and dims: {len(shape)} vs {len(dims_tuple)}"
            )
E           ValueError: different number of dimensions on data and dims: 4 vs 3

../../../micromamba/envs/test/lib/python3.11/site-packages/xarray/core/dataarray.py:185: ValueError

[snowman2]

Thanks @justingruca 👍