Cloud fix remote embl datasets

src/spatialdata/_io/_utils.py

@@ -409,12 +409,8 @@ def _open_zarr_store(path: StoreLike, **kwargs: Any) -> zarr.storage.BaseStore:
             # but extend the path of the store with that of the zarr.Group
             return FSStore(path.store.path + "/" + path.path, fs=path.store.fs, **kwargs)
         if isinstance(path.store, zarr.storage.ConsolidatedMetadataStore):
-            # TODO: find a way to check if the consolidated metadata is still used. Probably best to wait for Zarr v3.
-            # if the store is a ConsolidatedMetadataStore, just return it
-            # get the FSStore url path from store and append it with the path from the Group StoreLike object
-            url = UPath(path.store.store.path + path.path)
-            # same as UPath option
-            return FSStore(url.path, fs=url.fs, **kwargs)
+            # if the store is a ConsolidatedMetadataStore, just return the underlying FSSpec store
+            return path.store.store
         raise ValueError(f"Unsupported store type or zarr.Group: {type(path.store)}")
     if isinstance(path, zarr.storage.StoreLike):
         # if the input already a store, wrap it in an FSStore

src/spatialdata/_io/format.py

@@ -257,7 +257,7 @@ def validate_table(
 def format_implementations() -> Iterator[Format]:
     """Return an instance of each format implementation, newest to oldest."""
     yield RasterFormatV02()
-    # yield RasterFormatV01()  # same format string as FormatV04
+    yield RasterFormatV01()  # same format string as FormatV04
     yield FormatV04()
     yield FormatV03()
     yield FormatV02()

tests/io/test_remote.py

@@ -1,32 +1,33 @@
 import pytest
 import zarr
+from upath import UPath
 
 from spatialdata import SpatialData
 
 
 class TestRemote:
     # Test actual remote datasets from https://spatialdata.scverse.org/en/latest/tutorials/notebooks/datasets/README.html
-    # TODO: mark these tests so they are disabled by default because they require internet access
+
     @pytest.fixture(params=["merfish", "mibitof", "mibitof_alt"])
     def s3_address(self, request):
         urls = {
-            "merfish": "https://s3.embl.de/spatialdata/spatialdata-sandbox/merfish.zarr/",
-            "mibitof": "https://s3.embl.de/spatialdata/spatialdata-sandbox/mibitof.zarr/",
+            "merfish": UPath(
+                "s3://spatialdata/spatialdata-sandbox/merfish.zarr", endpoint_url="https://s3.embl.de", anon=True
+            ),
+            "mibitof": UPath(
+                "s3://spatialdata/spatialdata-sandbox/mibitof.zarr", endpoint_url="https://s3.embl.de", anon=True
+            ),
             "mibitof_alt": "https://dl01.irc.ugent.be/spatial/mibitof/data.zarr/",
         }
         return urls[request.param]
 
-    # TODO: does not work for EMBL datasets, problem with opening version 0.6-dev
-    @pytest.mark.skip(reason="Problem with ome_zarr on test datasets: ValueError: Version 0.6-dev not recognized")
     def test_remote(self, s3_address):
         # TODO: remove selection once support for points, shapes and tables is added
         sdata = SpatialData.read(s3_address, selection=("images", "labels"))
         assert len(list(sdata.gen_elements())) > 0
 
-    @pytest.mark.skip(reason="Problem with ome_zarr on test datasets: ValueError: Version 0.6-dev not recognized")
     def test_remote_consolidated(self, s3_address):
-        # TODO: find a way to check if the consolidated metadata is actually used. Probably best to wait for Zarr v3.
-        root = zarr.open_consolidated(s3_address, mode="r", metadata_key="zmetadata")
-        # TODO: remove selection once support for points, shapes and tables is added
+        urlpath, storage_options = str(s3_address), getattr(s3_address, "storage_options", {})
+        root = zarr.open_consolidated(urlpath, mode="r", metadata_key="zmetadata", storage_options=storage_options)
         sdata = SpatialData.read(root, selection=("images", "labels"))
         assert len(list(sdata.gen_elements())) > 0