Use natural sorting in imread(...) when globbing multiple files (#265)

* Add sortfunc argument to imread to customize glob order

* Fix flake8 complaints

* Refine docstring

* Remove custom sortfunc, default to natural sort

Co-authored-by: Volker Hilsenstein <volker.hilsenstein@embl.de>

dask_image/imread/__init__.py

@@ -6,6 +6,7 @@
 import dask.array as da
 import numpy as np
 import pims
+from tifffile import natural_sorted
 
 from . import _utils
 
@@ -21,6 +22,8 @@ def imread(fname, nframes=1, *, arraytype="numpy"):
     ----------
     fname : str or pathlib.Path
         A glob like string that may match one or multiple filenames.
+        Where multiple filenames match, they are sorted using
+        natural (as opposed to alphabetical) sort.
     nframes : int, optional
         Number of the frames to include in each chunk (default: 1).
     arraytype : str, optional
@@ -64,8 +67,8 @@ def imread(fname, nframes=1, *, arraytype="numpy"):
             RuntimeWarning
         )
 
-    # place source filenames into dask array
-    filenames = sorted(glob.glob(sfname))  # pims also does this
+    # place source filenames into dask array after sorting
+    filenames = natural_sorted(glob.glob(sfname))
     if len(filenames) > 1:
         ar = da.from_array(filenames, chunks=(nframes,))
         multiple_files = True
@@ -83,7 +86,6 @@ def imread(fname, nframes=1, *, arraytype="numpy"):
         arrayfunc=arrayfunc,
         meta=arrayfunc([]).astype(dtype),  # meta overwrites `dtype` argument
     )
-
     return a
 
 

tests/test_dask_image/test_imread/test_core.py

@@ -95,3 +95,11 @@ def test_tiff_imread(tmpdir, seed, nframes, shape, runtime_warning, dtype, is_pa
         assert (shape[0] % nframes) == d.chunks[0][-1]
 
     da.utils.assert_eq(a, d)
+
+
+def test_tiff_imread_glob_natural_sort(tmpdir):
+    dirpth = tmpdir.mkdir("test_imread")
+    tifffile.imwrite(dirpth.join("10.tif"), np.array([10]))
+    tifffile.imwrite(dirpth.join("9.tif"), np.array([9]))
+    actual = np.array(dask_image.imread.imread(dirpth.join("*.tif")))
+    assert np.all(actual == np.array([[9], [10]]))