rename compiled ext to _cutils, explicit import in FFTKDE

KDEpy/FFTKDE.py

@@ -10,8 +10,10 @@
 from scipy.signal import convolve
 
 from KDEpy.BaseKDE import BaseKDE
-from KDEpy.binning import grid_is_sorted, linear_binning
+from KDEpy.binning import linear_binning
 from KDEpy.utils import cartesian
+# noinspection PyProtectedMember, PyUnresolvedReferences
+from KDEpy._cutils import grid_is_sorted
 
 
 class FFTKDE(BaseKDE):

KDEpy/binning.py

@@ -35,13 +35,12 @@
 
 import numpy as np
 
-# Import compiled cutils module locally (not declared in __init__)
+# Import compiled _cutils module (not declared in __init__)
 # and ignore unresolved reference (may not be build in dev environment)
-# noinspection PyUnresolvedReferences
-from . import cutils
+# noinspection PyProtectedMember, PyUnresolvedReferences
+from KDEpy import _cutils
 from KDEpy.utils import cartesian
 
-grid_is_sorted = cutils.grid_is_sorted
 
 # This parameter was in use when a NumPy implementation and a Cython
 # implementation were both used. Now only Cython is used.
@@ -110,10 +109,10 @@ def linbin_cython(data, grid_points, weights=None):
     # Two Cython functions are implemented, one for weighted data and one
     # for unweighted data, since creating equal weights is costly w.r.t time
     if weights is None:
-        result = cutils.iterate_data_1D(transformed_data, result)
+        result = _cutils.iterate_data_1D(transformed_data, result)
         return np.asfarray(result[:-1]) / transformed_data.shape[0]
     else:
-        res = cutils.iterate_data_1D_weighted(transformed_data, weights, result)
+        res = _cutils.iterate_data_1D_weighted(transformed_data, weights, result)
         return np.asfarray(res[:-1])  # Remove last, outside of grid
 
 
@@ -358,9 +357,9 @@ def linbin_Ndim(data, grid_points, weights=None):
     if weights is not None:
         if data_dims >= 3:
             binary_flgs = cartesian(([0, 1],) * dims)
-            result = cutils.iterate_data_ND_weighted(data, weights, result, grid_num, obs_tot, binary_flgs)
+            result = _cutils.iterate_data_ND_weighted(data, weights, result, grid_num, obs_tot, binary_flgs)
         else:
-            result = cutils.iterate_data_2D_weighted(data, weights, result, grid_num, obs_tot)
+            result = _cutils.iterate_data_2D_weighted(data, weights, result, grid_num, obs_tot)
         result = np.asarray_chkfinite(result, dtype=float)
 
     # Unweighted data has two specific routines too. This is because creating
@@ -369,9 +368,9 @@ def linbin_Ndim(data, grid_points, weights=None):
     else:
         if data_dims >= 3:
             binary_flgs = cartesian(([0, 1],) * dims)
-            result = cutils.iterate_data_ND(data, result, grid_num, obs_tot, binary_flgs)
+            result = _cutils.iterate_data_ND(data, result, grid_num, obs_tot, binary_flgs)
         else:
-            result = cutils.iterate_data_2D(data, result, grid_num, obs_tot)
+            result = _cutils.iterate_data_2D(data, result, grid_num, obs_tot)
         result = np.asarray_chkfinite(result, dtype=float)
         result = result / data_obs
 

KDEpy/tests/test_sorted_grid.py

@@ -6,7 +6,10 @@
 import numpy as np
 import pytest
 
-from KDEpy.binning import grid_is_sorted  # Imported from .pyx to binning.py, then here
+# Import compiled _cutils module (not declared in __init__)
+# and ignore unresolved reference (may not be build in dev environment)
+# noinspection PyProtectedMember, PyUnresolvedReferences
+from KDEpy._cutils import grid_is_sorted  # Imported from .pyx to binning.py, then here
 
 
 class TestGridSorted:

setup.py

@@ -15,5 +15,5 @@
     packages=["KDEpy"],
     cmdclass={"build_ext": build_ext},
     include_dirs=[np.get_include()],
-    ext_modules=[Extension("KDEpy.cutils", ["KDEpy/cutils_ext/cutils.pyx"])],
+    ext_modules=[Extension("KDEpy._cutils", ["KDEpy/cutils_ext/cutils.pyx"])],
 )