Ome zarr v0.5 reading and writing

[will-moore]

NB: this is on top of #404 (reading OME-Zarr v0.5), which needs to be reviewed & merged first

This PR adds support for writing OME-Zarr v0.5, which becomes the default CurrentFormat().

If you want to choose e.g. v0.4, this is specified in parse_url():

    store = parse_url(path, mode="w", fmt=FormatV04()).store
    root = zarr.group(store=store)
    write_image(image=data, group=root)

This creates a zarr v2 store and group. So when we write_image() with that group, we know that we want to use v0.4 rather than v0.5. The default fmt for write_image() and other write... methods is now None rather than CurrentFormat(), so that we can detect when a user hasn't specified a format and pick the right one.

if you try to use a mix of formats, e.g. v04 and v05, this will throw an Exception:

    store = parse_url(path, mode="w", fmt=FormatV04()).store
    root = zarr.group(store=store)
    write_image(image=data, group=root, fmt=FormatV05())

Same with this, since parse_url() will use CurrentFormat():

    store = parse_url(path, mode="w").store
    root = zarr.group(store=store)
    write_image(image=data, group=root, fmt=FormatV04())

Known issues:

• compressor option fails when writing dask arrays - see 1cf5273 . da.to_zarr() uses AsyncArray._create() which doesn't support compressor. See https://github.com/zarr-developers/zarr-python/blob/b877f89fa2959e15d3d49bf027792fb23ce97b84/src/zarr/core/array.py#L609

../ome_zarr/writer.py:589: in write_image
    dask_delayed_jobs = _write_dask_image(
../ome_zarr/writer.py:709: in _write_dask_image
    da.to_zarr(
../../../../opt/anaconda3/envs/omeroweb_zarrv3/lib/python3.12/site-packages/dask/array/core.py:3934: in to_zarr
    z = zarr.create(
../../../../opt/anaconda3/envs/omeroweb_zarrv3/lib/python3.12/site-packages/zarr/api/synchronous.py:712: in create
    sync(
../../../../opt/anaconda3/envs/omeroweb_zarrv3/lib/python3.12/site-packages/zarr/core/sync.py:163: in sync
    raise return_result
../../../../opt/anaconda3/envs/omeroweb_zarrv3/lib/python3.12/site-packages/zarr/core/sync.py:119: in _runner
    return await coro
../../../../opt/anaconda3/envs/omeroweb_zarrv3/lib/python3.12/site-packages/zarr/api/asynchronous.py:1055: in create
    return await AsyncArray._create(
...
            if compressor != "auto":
>               raise ValueError(
                    "compressor cannot be used for arrays with zarr_format 3. Use bytes-to-bytes codecs instead."
                )
E               ValueError: compressor cannot be used for arrays with zarr_format 3. Use bytes-to-bytes codecs instead.

Testing:

• Reading via napari-ome-zarr... images and plates, v0.1, 0.3, 0.4, 0.5...

  • $ napari --plugin napari-ome-zarr https://uk1s3.embassy.ebi.ac.uk/idr/zarr/v0.1/9836844.zarr
  • $ napari --plugin napari-ome-zarr https://uk1s3.embassy.ebi.ac.uk/idr/zarr/v0.3/idr0079A/9836998.zarr
  • $ napari --plugin napari-ome-zarr https://uk1s3.embassy.ebi.ac.uk/idr/zarr/v0.4/idr0072B/9512.zarr
  • $ napari --plugin napari-ome-zarr https://uk1s3.embassy.ebi.ac.uk/idr/zarr/v0.5/idr0083/9822152.zarr
  • $ napari --plugin napari-ome-zarr https://uk1s3.embassy.ebi.ac.uk/idr/zarr/v0.5/idr0010/76-45.ome.zarr

• See the changes in docs/source/python.rst and test the various code samples. By default, the examples will write OME-Zarr v0.5. Also try writing v0.4 as described.

[imagesc-bot]

This pull request has been mentioned on Image.sc Forum. There might be relevant details there:

https://forum.image.sc/t/reading-ome-zarr-v0-5-with-python/114280/4

[ralfox]

Consider updating documentation:
https://github.com/will-moore/ome-zarr-py/blob/65acce283f1fca38442ae619e4900840f22a7dc4/docs/source/python.rst?plain=1#L194

[will-moore]

Thanks @ralfox - I'm away now and all next week but will get back to this on return...

[will-moore]

It looks like the docs in this PR are working OK based on my testing and the comment and the last comment on the forum discussion above: https://forum.image.sc/t/reading-ome-zarr-v0-5-with-python/114280/4.

[will-moore]

Apologies @ralfox - the confusing and unnecessary line: store = parse_url(url, mode="r").store should be removed from the example code...

[joshmoore]

Really only one main thought re: --version, otherwise super excited to get this out. ❤️

[joshmoore]

--version is kinda universally "tell me the current version of the library". I could see calling this --format and then if need be, it could even take a class name.

[will-moore]

done in 1a5b773

[joshmoore]

Just adding a written reminder of our chat: 👍 for having a reproducer for this.

[dyf]

Really only one main thought re: --version, otherwise super excited to get this out. ❤️

Nothing to add, just very excited for this to land!

[joshmoore]

Thanks, @will-moore! Moving forward with the RC.