AttributeError: 'Dataset' object has no attribute 'store'; open_ds

[Mohinta2892]

Hello!

When trying to load hdf via open_ds it throws AttributeError: 'Dataset' object has no attribute 'store' in _read_attrs(ds, order) (line 288) under datasets.py.

It's checking for a zarr object, when loading a hdf file.

Full Error traceback:

In [2]: f = open_ds("/Users/sam/Documents/random_codebases/ssTEM_data_Acardona/Seg2Link/data/cremi/hdf/sample_A_20160501.hdf")
---------------------------------------------------------------------------
TypeError                                 Traceback (most recent call last)
Cell In[2], line 1
----> 1 f = open_ds("/Users/sam/Documents/random_codebases/ssTEM_data_Acardona/Seg2Link/data/cremi/hdf/sample_A_20160501.hdf")

TypeError: open_ds() missing 1 required positional argument: 'ds_name'

In [3]: f = open_ds("/Users/sam/Documents/random_codebases/ssTEM_data_Acardona/Seg2Link/data/cremi/hdf/sample_A_20160501.hdf", "volumes/raw")
---------------------------------------------------------------------------
AttributeError                            Traceback (most recent call last)
Cell In[3], line 1
----> 1 f = open_ds("/Users/sam/Documents/random_codebases/ssTEM_data_Acardona/Seg2Link/data/cremi/hdf/sample_A_20160501.hdf", "volumes/raw")

File ~/opt/anaconda3/envs/nglance/lib/python3.10/site-packages/funlib/persistence/arrays/datasets.py:446, in open_ds(filename, ds_name, mode)
    443 logger.debug("opening H5 dataset %s in %s", ds_name, filename)
    444 ds = h5py.File(filename, mode=mode)[ds_name]
--> 446 voxel_size, offset = _read_voxel_size_offset(ds, "C")
    447 shape = Coordinate(ds.shape[-len(voxel_size) :])
    448 roi = Roi(offset, voxel_size * shape)

File ~/opt/anaconda3/envs/nglance/lib/python3.10/site-packages/funlib/persistence/arrays/datasets.py:378, in _read_voxel_size_offset(ds, order)
    376 def _read_voxel_size_offset(ds, order="C"):
--> 378     voxel_size, offset, units = _read_attrs(ds, order)
    380     return regularize_offset(voxel_size, offset)

File ~/opt/anaconda3/envs/nglance/lib/python3.10/site-packages/funlib/persistence/arrays/datasets.py:288, in _read_attrs(ds, order)
    284 multiscales = None
    286 if  not isinstance(ds, zarr.core.Array):
    287     raise TypeError(
--> 288         f"{os.path.join(ds.store.path, ds.path)} is not zarr.core.Array"
    289     )
    290 # check recursively for multiscales attribute in the zarr store tree
    291 multiscales, multiscale_group = check_for_multiscale(group=access_parent(ds))

AttributeError: 'Dataset' object has no attribute 'store'

Best,
Samia

[mzouink]

(perhaps this is a very late answer)

to read hdf file you need to specify the store:

try:

import h5py
from funlib.persistence import open_ds
source = h5py.File("/Users/sam/Documents/random_codebases/ssTEM_data_Acardona/Seg2Link/data/cremi/hdf/sample_A_20160501.hdf", mode='r')
f = open_ds(source)