Change deprecated Gaussian/Laplacian pyramid args

[AlbertDominguez]

Hi,

Thanks for the great effort you're doing here :) This PR updates the arguments passed to skimage.transform.pyramid_gaussian and skimage.transform.pyramid_laplacian (without any behaviour change) used by the Scaler class, as the arg multichannel was deprecated in scikit-image 0.19.x. Using the latest version of scikit-image (0.25.x) causes a crash if either of these methods are used for downsampling.

Note that this causes the code not to work if scikit-image<0.19. Unsure if it's worth pinning in the deps 🙂

Here's a simple code to replicate:

from ome_zarr.io import parse_url
from ome_zarr.scale import Scaler
from ome_zarr.writer import write_image
import numpy as np
import zarr

if __name__ == "__main__":
    # Example usage
    store = parse_url("dbg.ome.zarr", "w").store
    group = zarr.group(store=store, overwrite=True)
    array = np.random.randn(1,256,256)

    scaler = Scaler(
        downscale=2,
        method="gaussian", # or "laplacian"
        max_layer=2,
    )
   
    # Next line currently breaks due to deprecated args being used downstream by the Scaler instance
    write_image(
        image=array,
        group=group,
        scaler=scaler,
        axes="cyx",
    )

[joshmoore]

Note that this causes the code not to work if scikit-image<0.19. Unsure if it's worth pinning in the deps 🙂

Yes, please, if you are still interested, @AlbertDominguez. With #413 finally out of the way, I imagine we can look at getting this in. Thanks!

[AlbertDominguez]

Hi @joshmoore, thanks! I've just merged the latest changes into the PR + pinned scikit-image. Also further tested locally and everything seems to work as expected.

[joshmoore]

Leaving this for @will-moore to look at, but I imagine we can roll it into the imminent 0.12.0.

[will-moore]

Great, thanks @AlbertDominguez

I turned your test example above into a test which your PR fixes.
Could you include this in tests/test_scaler.py?

+++ b/tests/test_scaler.py
@@ -1,9 +1,12 @@
 import dask.array as da
 import numpy as np
+import pathlib
 import pytest
+import zarr
 
 from ome_zarr.scale import Scaler
-
+from ome_zarr.io import parse_url
+from ome_zarr.writer import write_image
 
 class TestScaler:
     @pytest.fixture(
@@ -146,3 +149,25 @@ class TestScaler:
         # to zarr invokes compute
         data_dir = tmpdir.mkdir("test_big_dask_pyramid")
         da.to_zarr(level_1, str(data_dir))
+
+
+    @pytest.mark.parametrize("method", ["gaussian", "laplacian"])
+    def test_pyramid_args(self, shape, tmpdir, method):
+        path = pathlib.Path(tmpdir.mkdir("data"))
+        store = parse_url(path, mode="w").store
+        group = zarr.group(store=store, overwrite=True)
+        data = self.create_data(shape)
+
+        scaler = Scaler(
+            downscale=2,
+            method=method,
+            max_layer=2,
+        )
+
+        axes = "tczyx"[-len(shape) :]
+        write_image(
+            image=data,
+            group=group,
+            scaler=scaler,
+            axes=axes,
+        )

[AlbertDominguez]

Thanks @joshmoore and @will-moore! I've just added the suggested test

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 86.99%. Comparing base (bc34123) to head (13bab65).
⚠️ Report is 96 commits behind head on master.

Additional details and impacted files

@@            Coverage Diff             @@
##           master     #454      +/-   ##
==========================================
- Coverage   87.02%   86.99%   -0.03%     
==========================================
  Files          13       13              
  Lines        1772     1769       -3     
==========================================
- Hits         1542     1539       -3     
  Misses        230      230              

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[will-moore]

Great, thanks