fix(csr): Set type of node_props to float64 (#235)

`dtype` of `node_props` passed from `csr_to_nbgraph()` to `NBGraph` was
`float32`. This sets the type to explicitly be `np.float64` which
matches the expectation of `csr_spec_float` passed to
`numba.experimental.jitclass()`.

The tests that were previously failing in
[TopoStats](https://github.com/AFM-SPM/TopoStats/) (see #234 for
details) now pass locally. 🙂

Closes #234

---------

Co-authored-by: Juan Nunez-Iglesias <[email protected]>

src/skan/csr.py

@@ -207,7 +207,7 @@ def csr_to_nbgraph(csr, node_props=None):
             csr.indices,
             csr.data,
             np.array(csr.shape, dtype=np.int32),
-            node_props,
+            node_props.astype(np.float64),
             )
 
 
@@ -525,7 +525,7 @@ def __init__(
         if np.issubdtype(skeleton_image.dtype, np.floating):
             self.pixel_values = skeleton_image[coords]
         elif np.issubdtype(skeleton_image.dtype, np.integer):
-            self.pixel_values = skeleton_image.astype(float)[coords]
+            self.pixel_values = skeleton_image.astype(np.float64)[coords]
         else:
             self.pixel_values = None
         self.graph = graph

src/skan/test/test_csr.py

@@ -10,7 +10,11 @@
 from numpy.testing import assert_equal, assert_almost_equal
 import pandas as pd
 import pytest
+import scipy
+from scipy import ndimage as ndi
+from skimage import data
 from skimage.draw import line
+from skimage.morphology import skeletonize
 
 from skan import csr
 from skan._testdata import (
@@ -357,6 +361,16 @@ def test_skeleton_integer_dtype(dtype):
     assert stats['mean_pixel_value'].max() > 1
 
 
+@pytest.mark.parametrize('dtype', [np.float32, np.float64])
+def test_skeleton_all_float_dtypes(dtype):
+    """Test that skeleton data types can be both float32 and float64."""
+    horse = ~data.horse()
+    skeleton_image = skeletonize(horse)
+    dt = ndi.distance_transform_edt(horse)
+    float_skel = (dt * skeleton_image).astype(dtype)
+    _ = csr.Skeleton(float_skel)
+
+
 def test_default_summarize_separator():
     with pytest.warns(np.exceptions.VisibleDeprecationWarning,
                       match='separator in column name'):
@@ -523,7 +537,7 @@ def test_nx_to_skeleton(
 
 
 @pytest.mark.parametrize(
-        'wrong_skeleton',
+        ('wrong_skeleton'),
         [
                 pytest.param(skeleton0, id='Numpy Array.'),
                 pytest.param(csr.Skeleton(skeleton0), id='Skeleton.'),
@@ -538,3 +552,36 @@ def test_nx_to_skeleton_attribute_error(wrong_skeleton: Any) -> None:
     """Test various errors are raised by nx_to_skeleton()."""
     with pytest.raises(Exception):
         csr.nx_to_skeleton(wrong_skeleton)
+
+
+@pytest.mark.parametrize(
+        ('skeleton'),
+        [
+                pytest.param(skeleton0, id='Numpy Array'),
+                pytest.param(csr.Skeleton(skeleton0), id='Skeleton'),
+                pytest.param(nx_graph, id='NetworkX Graph without edges.'),
+                ],
+        )
+def test_csr_to_nbgraph_attribute_error(skeleton: Any) -> None:
+    """Raise AttributeError if csr_to_nbgraph() passed incomplete objects."""
+    with pytest.raises(AttributeError):
+        csr.csr_to_nbgraph(skeleton)
+
+
+@pytest.mark.parametrize(
+        ('graph'),
+        [
+                pytest.param(
+                        scipy.sparse.csr_matrix(skeleton0),
+                        id='Sparse matrix directly from Numpy Array',
+                        ),
+                pytest.param(
+                        scipy.sparse.csr_matrix(csr.Skeleton(skeleton0)),
+                        id='Sparse matrix from csr.Skeleton',
+                        ),
+                ],
+        )
+def test_csr_to_nbgraph_type_error(graph: scipy.sparse.csr_matrix) -> None:
+    """Test TypeError is raised by csr_to_nbgraph() if wrong type is passed."""
+    with pytest.raises(TypeError):
+        csr.csr_to_nbgraph(graph)