Repository navigation
How to open a NDArray for read and write? #177
Description
Hey,
I am having difficulties understanding the documentation on how to create a peristent NDArray on disk that is populated slice-wise with data over time. The same goes vice-versa with reading from the NDArray on the disk.
With Zarr I would approach it like this:
Writing:
import zarr
import numpy
array_zarr = zarr.open("tmp.zarr", shape=(10, ), dtype=np.float32, chunks=(2, ), mode='w')
array_zarr[:5] = [1, 2, 1, 2, 1]
array_zarr[5:] = [3, 4, 3, 4, 3]Reading:
import zarr
import numpy
array_zarr = zarr.open("tmp.zarr", mode='r')
array = np.zeros((10, ))
array[:5] = array_zarr[:5]
array[5:] = array_zarr[5:]How do I do the same with blosc2?
Best regards,
Karol
Activity
FrancescAlted commentedon Mar 27, 2024
Sure. Following the tutorial should help you grasping the basics for persistence and slicing: https://www.blosc.org/python-blosc2/getting_started/tutorials/02.ndarray-basics.html
For creating an empty array (the equivalent of open in zarr), you can use blosc2.empty() with a
urlpathas keyword argument.BTW, the documentation can be improved for sure. If you think that you can contribute to the cause, we would gladly accept PRs.