Merge pull request #17 from bioio-devs/bugfix/read-zarr-from-s3

(Not a fork) Bugfix: read zarr from s3

README.md

@@ -41,6 +41,15 @@ img = BioImage("my_file.zarr", reader=bioio_ome_zarr.Reader)
 img.data
 ```
 
+### Reading from AWS S3
+To read from private S3 buckets or public buckets using `s3://` paths, [credentials](https://docs.aws.amazon.com/cli/latest/userguide/cli-configure-files.html) must be configured. Public buckets can be accessed without credentials by using the `https://` path.
+```python
+from bioio import BioImage
+path = "https://allencell.s3.amazonaws.com/aics/nuc_morph_data/data_for_analysis/baseline_colonies/20200323_09_small/raw.ome.zarr"
+image = BioImage(path)
+print(image.get_image_dask_data())
+```
+
 ## Issues
 [_Click here to view all open issues in bioio-devs organization at once_](https://github.com/search?q=user%3Abioio-devs+is%3Aissue+is%3Aopen&type=issues&ref=advsearch) or check this repository's issue tab.
 

bioio_ome_zarr/reader.py

@@ -22,17 +22,10 @@ class Reader(reader.Reader):
 
     Parameters
     ----------
-    image: Any
-        Some type of object to read and follow the Reader specification.
+    image: types.PathLike
+        String or Path to the ZARR root
     fs_kwargs: Dict[str, Any]
-        Any specific keyword arguments to pass down to the fsspec created filesystem.
-        Default: {}
-
-    Notes
-    -----
-    It is up to the implementer of the Reader to decide which types they would like to
-    accept (certain readers may not support buffers for example).
-
+        Ignored
     """
 
     _xarray_dask_data: Optional["xr.DataArray"] = None
@@ -65,17 +58,19 @@ def __init__(
         # Enforce valid image
         if not self._is_supported_image(self._fs, self._path):
             raise exceptions.UnsupportedFileFormatError(
-                self.__class__.__name__, self._path
+                self.__class__.__name__,
+                self._path,
+                "Could not find a .zgroup or .zarray file at the provided path.",
             )
 
-        self._zarr = ZarrReader(parse_url(self._path, mode="r")).zarr
+        self._zarr = get_zarr_reader(self._fs, self._path).zarr
         self._physical_pixel_sizes: Optional[types.PhysicalPixelSizes] = None
         self._channel_names: Optional[List[str]] = None
 
     @staticmethod
     def _is_supported_image(fs: AbstractFileSystem, path: str, **kwargs: Any) -> bool:
         try:
-            ZarrReader(parse_url(path, mode="r"))
+            get_zarr_reader(fs, path)
             return True
 
         except AttributeError:
@@ -89,12 +84,7 @@ def is_supported_image(
         **kwargs: Any,
     ) -> bool:
         if isinstance(image, (str, Path)):
-            try:
-                ZarrReader(parse_url(image, mode="r"))
-                return True
-
-            except AttributeError:
-                return False
+            return cls._is_supported_image(None, str(image), **kwargs)
         else:
             return reader.Reader.is_supported_image(
                 cls, image, fs_kwargs=fs_kwargs, **kwargs
@@ -269,3 +259,9 @@ def _get_coords(
                 coords[dimensions.DimensionNames.Channel] = channel_names
 
         return coords
+
+
+def get_zarr_reader(fs: AbstractFileSystem, path: str) -> ZarrReader:
+    if fs is not None:
+        path = fs.unstrip_protocol(path)
+    return ZarrReader(parse_url(path, mode="r"))

bioio_ome_zarr/tests/test_s3_read.py

@@ -0,0 +1,37 @@
+import numpy as np
+
+from bioio_ome_zarr import Reader
+
+
+def test_ome_zarr_reader() -> None:
+    # ARRANGE
+    uri = (
+        # Cannot use s3:// URL due to ome-zarr issue #369
+        # "s3://allencell/aics/nuc_morph_data"
+        "https://allencell.s3.amazonaws.com/aics/nuc_morph_data"
+        "/data_for_analysis/baseline_colonies/20200323_09_small/raw.ome.zarr"
+    )
+    scene = "/"
+    resolution_level = 0
+
+    # ACT
+    image_container = Reader(uri, fs_kwargs=dict(anon=True))
+    image_container.set_scene(scene)
+    image_container.set_resolution_level(resolution_level)
+
+    # ASSERT
+    assert image_container.scenes == (scene,)
+    assert image_container.current_scene == scene
+    assert image_container.resolution_levels == (0, 1, 2, 3, 4)
+    assert image_container.shape == (570, 2, 42, 1248, 1824)
+    assert image_container.dtype == np.uint16
+    assert image_container.dims.order == "TCZYX"
+    assert image_container.dims.shape == (570, 2, 42, 1248, 1824)
+    assert image_container.channel_names == ["low_EGFP", "low_Bright"]
+    assert image_container.current_resolution_level == resolution_level
+    # pixel sized in (Z, Y, X) order
+    assert image_container.physical_pixel_sizes == (
+        0.53,
+        0.2708333333333333,
+        0.2708333333333333,
+    )