write to zarr fails when SpatialImage is created from a xarray.DataArray

[LucaMarconato]

When the SpatialImage is created from a np.ndarray all is good, when is created from a xarray.DataArray an error occurs (see traceback).
I have added tests for this. To reproduce uncomment the parts of _get_images() and _get_labels() (from tests/conftest.py) that start with out['....._xarray'] = ......

tests/_io/test_readwrite.py:14 (TestReadWrite.test_images)
self = <tests._io.test_readwrite.TestReadWrite object at 0x7f9f9490edd0>
tmp_path = PosixPath('/private/tmp/pytest-of-macbook/pytest-74/test_images0')
images = SpatialData object with:
└── Images
│     ├── 'image2d_numpy': SpatialImage[cyx] (3, 64, 64)
│     ├── 'image2d_multis...3, 64, 64)
│     └── 'image2d_multiscale_numpy_xarray': MultiscaleSpatialImage[cyx] (3, 64, 64), (3, 32, 32), (3, 8, 8)

    def test_images(self, tmp_path: str, images: SpatialData) -> None:
        """Test read/write."""
        tmpdir = Path(tmp_path) / "tmp.zarr"
>       images.write(tmpdir)

tests/_io/test_readwrite.py:18: 
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ 
spatialdata/_core/_spatialdata.py:147: in write
    write_image(
spatialdata/_io/write.py:95: in write_image
    write_image_ngff(
../../../../miniconda3/envs/ome/lib/python3.10/site-packages/ome_zarr/writer.py:455: in write_image
    _write_dask_image(
../../../../miniconda3/envs/ome/lib/python3.10/site-packages/ome_zarr/writer.py:565: in _write_dask_image
    da.compute(*delayed)
../../../../miniconda3/envs/ome/lib/python3.10/site-packages/dask/base.py:600: in compute
    results = schedule(dsk, keys, **kwargs)
../../../../miniconda3/envs/ome/lib/python3.10/site-packages/dask/threaded.py:89: in get
    results = get_async(
../../../../miniconda3/envs/ome/lib/python3.10/site-packages/dask/local.py:511: in get_async
    raise_exception(exc, tb)
../../../../miniconda3/envs/ome/lib/python3.10/site-packages/dask/local.py:319: in reraise
    raise exc
../../../../miniconda3/envs/ome/lib/python3.10/site-packages/dask/local.py:224: in execute_task
    result = _execute_task(task, data)
../../../../miniconda3/envs/ome/lib/python3.10/site-packages/dask/core.py:119: in _execute_task
    return func(*(_execute_task(a, cache) for a in args))
../../../../miniconda3/envs/ome/lib/python3.10/site-packages/dask/array/core.py:4366: in store_chunk
    return load_store_chunk(x, out, index, lock, return_stored, False)
../../../../miniconda3/envs/ome/lib/python3.10/site-packages/dask/array/core.py:4348: in load_store_chunk
    out[index] = x
../../../../miniconda3/envs/ome/lib/python3.10/site-packages/zarr/core.py:1373: in __setitem__
    self.set_basic_selection(pure_selection, value, fields=fields)
../../../../miniconda3/envs/ome/lib/python3.10/site-packages/zarr/core.py:1468: in set_basic_selection
    return self._set_basic_selection_nd(selection, value, fields=fields)
../../../../miniconda3/envs/ome/lib/python3.10/site-packages/zarr/core.py:1772: in _set_basic_selection_nd
    self._set_selection(indexer, value, fields=fields)
../../../../miniconda3/envs/ome/lib/python3.10/site-packages/zarr/core.py:1844: in _set_selection
    self._chunk_setitems(lchunk_coords, lchunk_selection, chunk_values,
../../../../miniconda3/envs/ome/lib/python3.10/site-packages/zarr/core.py:2045: in _chunk_setitems
    to_store = {k: self._encode_chunk(v) for k, v in cdatas.items()}
../../../../miniconda3/envs/ome/lib/python3.10/site-packages/zarr/core.py:2045: in <dictcomp>
    to_store = {k: self._encode_chunk(v) for k, v in cdatas.items()}
../../../../miniconda3/envs/ome/lib/python3.10/site-packages/zarr/core.py:2222: in _encode_chunk
    if ensure_ndarray_like(chunk).dtype == object:
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ 

buf = <xarray.DataArray (c: 3, y: 64, x: 64)>
array([[[-1.45060918e+00,  5.93752346e-01, -3.55374641e-02, ...,
         -5.1...053254e+00, ...,
         -2.45753512e-01,  1.03229240e+00, -8.63332273e-01]]])
Dimensions without coordinates: c, y, x

    def ensure_ndarray_like(buf) -> NDArrayLike:
        """Convenience function to coerce `buf` to ndarray-like array.
    
        Parameters
        ----------
        buf : ndarray-like, array-like, or bytes-like
            A numpy array like object such as numpy.ndarray, cupy.ndarray, or
            any object exporting a buffer interface.
    
        Returns
        -------
        arr : NDArrayLike
            A ndarray-like, sharing memory with `buf`.
    
        Notes
        -----
        This function will not create a copy under any circumstances, it is guaranteed to
        return a view on memory exported by `buf`.
        """
    
        if not is_ndarray_like(buf):
            if isinstance(buf, array.array) and buf.typecode in "cu":
                # Guard condition, do not support array.array with unicode type, this is
                # problematic because numpy does not support it on all platforms. Also do not
                # support char as it was removed in Python 3.
                raise TypeError("array.array with char or unicode type is not supported")
            else:
                # N.B., first take a memoryview to make sure that we subsequently create a
                # numpy array from a memory buffer with no copy
>               mem = memoryview(buf)
E               TypeError: memoryview: a bytes-like object is required, not 'DataArray'

../../../../miniconda3/envs/ome/lib/python3.10/site-packages/numcodecs/compat.py:42: TypeError

[giovp]

I believe it has to do with the data type (xarray with numpy array might be a problem). Will check!

[LucaMarconato]

In my previous more experimental branch (#48) I was converting everything to DataArray instead than to Dask arrays. This didn't induce the problem, and Dask arrays converted to DataArray where still backed. I tried changing this in the new code but then I noticed that the Schema asks for Dask arrays, so I didn't want to change that before checking with you. Maybe this could fix it.

[giovp]

I believe this is now tested and addressed in #52

[LucaMarconato]

I checked, the tests that failed are still commented in tests/conftest.py. The one added in test_models.py may still not cover those cases because maybe Dask doesn't reach the problem until things are computed/written to disk. I'll take care of this.