ENH: Added some minimal test-cases to the Affine class

nitransforms/base.py

@@ -154,6 +154,10 @@ def __eq__(self, other):
         return (np.allclose(self.affine, other.affine, rtol=EQUALITY_TOL) and
                 self.shape == other.shape)
 
+    def __ne__(self, other):
+        """Overload not equal operator."""
+        return not self == other
+
 
 class TransformBase(object):
     """Abstract image class to represent transforms."""

nitransforms/conftest.py

@@ -19,7 +19,7 @@ def doctest_autoimport(doctest_namespace):
     doctest_namespace['nb'] = nb
     doctest_namespace['os'] = os
     doctest_namespace['Path'] = Path
-    doctest_namespace['datadir'] = os.path.join(os.path.dirname(__file__), 'tests/data')
+    doctest_namespace['datadir'] = Path(__file__).parent / 'tests' / 'data'
 
     tmpdir = tempfile.TemporaryDirectory()
     doctest_namespace['tmpdir'] = tmpdir.name

nitransforms/linear.py

@@ -7,8 +7,9 @@
 #
 ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ##
 """Linear transforms."""
-import numpy as np
 from pathlib import Path
+import warnings
+import numpy as np
 
 from nibabel.loadsave import load as loadimg
 from nibabel.affines import from_matvec, voxel_sizes, obliquity
@@ -40,6 +41,13 @@ def __init__(self, matrix=None, reference=None):
 
         Examples
         --------
+        >>> xfm = Affine(reference=datadir / 'someones_anatomy.nii.gz')
+        >>> xfm.matrix  # doctest: +NORMALIZE_WHITESPACE
+        array([[[1., 0., 0., 0.],
+                [0., 1., 0., 0.],
+                [0., 0., 1., 0.],
+                [0., 0., 0., 1.]]])
+
         >>> xfm = Affine([[1, 0, 0, 4], [0, 1, 0, 0], [0, 0, 1, 0], [0, 0, 0, 1]])
         >>> xfm.matrix  # doctest: +NORMALIZE_WHITESPACE
         array([[[1, 0, 0, 4],
@@ -62,8 +70,6 @@ def __init__(self, matrix=None, reference=None):
         ), 'affine matrix is not square'
 
         if reference:
-            if isinstance(reference, str):
-                reference = loadimg(reference)
             self.reference = reference
 
     def __eq__(self, other):
@@ -78,9 +84,10 @@ def __eq__(self, other):
         True
 
         """
-        if not self._reference == other._reference:
-            return False
-        return np.allclose(self.matrix, other.matrix, rtol=EQUALITY_TOL)
+        _eq = np.allclose(self.matrix, other.matrix, rtol=EQUALITY_TOL)
+        if _eq and self._reference != other._reference:
+            warnings.warn('Affines are equal, but references do not match.')
+        return _eq
 
     @property
     def matrix(self):
@@ -257,9 +264,7 @@ def load(filename, fmt='X5', reference=None):
     else:
         raise NotImplementedError
 
-    if reference and isinstance(reference, str):
-        reference = loadimg(reference)
-    return Affine(matrix, reference)
+    return Affine(matrix, reference=reference)
 
 
 def _fsl_aff_adapt(space):

nitransforms/tests/test_base.py

@@ -46,6 +46,7 @@ def test_ImageGrid(get_testdata, image_orientation):
 
     img2 = ImageGrid(img)
     assert img2 == img
+    assert (img2 != img) is False
 
 
 def test_ImageGrid_utils(tmpdir, data_path, get_testdata):

nitransforms/tests/test_linear.py

@@ -6,6 +6,7 @@
 import numpy as np
 from subprocess import check_call
 import shutil
+import h5py
 
 import nibabel as nb
 from nibabel.eulerangles import euler2mat
@@ -155,3 +156,9 @@ def test_apply_linear_transform(
     diff = sw_moved.get_fdata() - nt_moved.get_fdata()
     # A certain tolerance is necessary because of resampling at borders
     assert (np.abs(diff) > 1e-3).sum() / diff.size < TESTS_BORDER_TOLERANCE
+
+
+def test_Affine(tmpdir):
+    """Test affine's operations."""
+    with h5py.File('xfm.x5', 'w') as f:
+        nbl.Affine()._to_hdf5(f.create_group('Affine'))