MAINT: Fix linting errors and bump numpy version

Increased min numpy version from 1.18 to 1.20 to allow using dtype within numpy.concatenate.

.github/workflows/python-test.yml

@@ -44,7 +44,7 @@ jobs:
     - name: Install required dependencies
       run: |
         python -m pip install --upgrade pip
-        python -m pip install "numpy>=1.18" "scipy>=1.5" pytest
+        python -m pip install "numpy>=1.20" "scipy>=1.5" pytest
 
     - name: Test with required dependencies
       run: pytest .
@@ -79,7 +79,7 @@ jobs:
     - name: Install minimum dependencies
       run: |
         python -m pip install --upgrade pip
-        python -m pip install numpy==1.18 scipy==1.5 pytest
+        python -m pip install numpy==1.20 scipy==1.5 pytest
 
     - name: Test with minimum required dependencies
       run: pytest .

docs/installation.rst

@@ -10,7 +10,7 @@ Dependencies
 
 pybaselines requires `Python <https://python.org>`_ version 3.8 or later and the following libraries:
 
-* `NumPy <https://numpy.org>`_ (>= 1.18)
+* `NumPy <https://numpy.org>`_ (>= 1.20)
 * `SciPy <https://www.scipy.org>`_ (>= 1.5)
 
 

examples/general/plot_algorithm_convergence.py

@@ -9,8 +9,8 @@
 the measured tolerance value at each iteration. The `tol_history` parameter
 can be helpful for determining appropriate `max_iter` or `tol` values.
 
-In this example, the convergence of the :meth:`~.Baseline.asls` and :meth:`~.Baseline.aspls` functions
-will be compared. asls is a relatively simple calculation that sets its weighting
+In this example, the convergence of the :meth:`~.Baseline.asls` and :meth:`~.Baseline.aspls`
+functions will be compared. asls is a relatively simple calculation that sets its weighting
 each iteration based on whether the current baseline is above or below the input data
 at each point. aspls has a much more intricate weighting based on the logistic distribution
 of the residuals (data minus baseline); further, aspls also updates an additional

examples/whittaker/plot_whittaker_solvers.py

@@ -113,7 +113,8 @@ def make_data(num_x):
 
     if not _banded_utils._HAS_PENTAPY:
         warnings.warn(
-            'pentapy is not installed so pentapy and scipy-banded timings will be identical'
+            'pentapy is not installed so pentapy and scipy-banded timings will be identical',
+            stacklevel=2
         )
 
     # equation obtained following similar procedure as `lam` vs data size example

pybaselines/classification.py

@@ -1028,9 +1028,9 @@ def _averaged_interp(x, y, mask, interp_half_window=0):
     mask_sum = mask.sum()
     if not mask_sum:  # all points belong to peaks
         # will just interpolate between first and last points
-        warnings.warn('there were no baseline points found', ParameterWarning)
+        warnings.warn('there were no baseline points found', ParameterWarning, stacklevel=2)
     elif mask_sum == mask.shape[0]:  # all points belong to baseline
-        warnings.warn('there were no peak points found', ParameterWarning)
+        warnings.warn('there were no peak points found', ParameterWarning, stacklevel=2)
         return output
 
     peak_starts, peak_ends = _find_peak_segments(mask)
@@ -1153,7 +1153,7 @@ def _iter_threshold(power, num_std=3.0):
         if masked_power.size < 2:  # need at least 2 points for std calculation
             warnings.warn(
                 'not enough baseline points found; "num_std" is likely too low',
-                ParameterWarning
+                ParameterWarning, stacklevel=2
             )
             break
         mask = power < np.mean(masked_power) + num_std * np.std(masked_power, ddof=1)

pybaselines/smooth.py

@@ -178,7 +178,7 @@ def snip(self, data, max_half_window=None, decreasing=False, smooth_half_window=
             if half_window > (self._len - 1) // 2:
                 warnings.warn(
                     'max_half_window values greater than (len(data) - 1) / 2 have no effect.',
-                    ParameterWarning
+                    ParameterWarning, stacklevel=2
                 )
                 half_windows[i] = (self._len - 1) // 2
 

pybaselines/spline.py

@@ -638,7 +638,7 @@ def pspline_airpls(self, data, lam=1e3, num_knots=100, spline_degree=3,
                 warnings.warn(
                     ('error occurred during fitting, indicating that "tol"'
                      ' is too low, "max_iter" is too high, or "lam" is too high'),
-                    ParameterWarning
+                    ParameterWarning, stacklevel=2
                 )
                 i -= 1  # reduce i so that output tol_history indexing is correct
                 break
@@ -653,7 +653,7 @@ def pspline_airpls(self, data, lam=1e3, num_knots=100, spline_degree=3,
                 # point would get a weight of 0, which fails the solver
                 warnings.warn(
                     ('almost all baseline points are below the data, indicating that "tol"'
-                     ' is too low and/or "max_iter" is too high'), ParameterWarning
+                     ' is too low and/or "max_iter" is too high'), ParameterWarning, stacklevel=2
                 )
                 i -= 1  # reduce i so that output tol_history indexing is correct
                 break
@@ -851,7 +851,8 @@ def pspline_drpls(self, data, lam=1e3, eta=0.5, num_knots=100, spline_degree=3,
                 # checking a scalar is faster; cannot use np.errstate since it is not 100% reliable
                 warnings.warn(
                     ('nan and/or +/- inf occurred in weighting calculation, likely meaning '
-                     '"tol" is too low and/or "max_iter" is too high'), ParameterWarning
+                     '"tol" is too low and/or "max_iter" is too high'), ParameterWarning,
+                     stacklevel=2
                 )
                 break
             elif calc_difference < tol:
@@ -938,7 +939,8 @@ def pspline_iarpls(self, data, lam=1e3, num_knots=100, spline_degree=3, diff_ord
                 # checking a scalar is faster; cannot use np.errstate since it is not 100% reliable
                 warnings.warn(
                     ('nan and/or +/- inf occurred in weighting calculation, likely meaning '
-                     '"tol" is too low and/or "max_iter" is too high'), ParameterWarning
+                     '"tol" is too low and/or "max_iter" is too high'), ParameterWarning,
+                     stacklevel=2
                 )
                 break
             elif calc_difference < tol:

pybaselines/two_d/_whittaker_utils.py

@@ -608,7 +608,9 @@ def basis(self):
     def _calc_dof(self, weights, assume_a='pos'):
         if not self._using_svd:
             # Could maybe just output a matrix of ones?
-            raise ValueError('Cannot calculate degrees of freedom when not using eigendecomposition')
+            raise ValueError(
+                'Cannot calculate degrees of freedom when not using eigendecomposition'
+            )
         lhs = self._make_btwb(weights)
         rhs = lhs.copy()
         np.fill_diagonal(lhs, lhs.diagonal() + self.penalty)

pybaselines/two_d/spline.py

@@ -557,7 +557,7 @@ def pspline_airpls(self, data, lam=1e3, num_knots=25, spline_degree=3,
                 warnings.warn(
                     ('error occurred during fitting, indicating that "tol"'
                      ' is too low, "max_iter" is too high, or "lam" is too high'),
-                    ParameterWarning
+                    ParameterWarning, stacklevel=2
                 )
                 i -= 1  # reduce i so that output tol_history indexing is correct
                 break
@@ -572,7 +572,8 @@ def pspline_airpls(self, data, lam=1e3, num_knots=25, spline_degree=3,
                 # point would get a weight of 0, which fails the solver
                 warnings.warn(
                     ('almost all baseline points are below the data, indicating that "tol"'
-                     ' is too low and/or "max_iter" is too high'), ParameterWarning
+                     ' is too low and/or "max_iter" is too high'), ParameterWarning,
+                    stacklevel=2
                 )
                 i -= 1  # reduce i so that output tol_history indexing is correct
                 break
@@ -744,7 +745,8 @@ def pspline_iarpls(self, data, lam=1e3, num_knots=25, spline_degree=3, diff_orde
                 # checking a scalar is faster; cannot use np.errstate since it is not 100% reliable
                 warnings.warn(
                     ('nan and/or +/- inf occurred in weighting calculation, likely meaning '
-                     '"tol" is too low and/or "max_iter" is too high'), ParameterWarning
+                     '"tol" is too low and/or "max_iter" is too high'), ParameterWarning,
+                    stacklevel=2
                 )
                 break
             elif calc_difference < tol:

pybaselines/whittaker.py

@@ -264,7 +264,7 @@ def airpls(self, data, lam=1e6, diff_order=2, max_iter=50, tol=1e-3, weights=Non
                 warnings.warn(
                     ('error occurred during fitting, indicating that "tol"'
                      ' is too low, "max_iter" is too high, or "lam" is too high'),
-                    ParameterWarning
+                    ParameterWarning, stacklevel=2
                 )
                 i -= 1  # reduce i so that output tol_history indexing is correct
                 break
@@ -278,7 +278,8 @@ def airpls(self, data, lam=1e6, diff_order=2, max_iter=50, tol=1e-3, weights=Non
                 # point would get a weight of 0, which fails the solver
                 warnings.warn(
                     ('almost all baseline points are below the data, indicating that "tol"'
-                     ' is too low and/or "max_iter" is too high'), ParameterWarning
+                     ' is too low and/or "max_iter" is too high'), ParameterWarning,
+                     stacklevel=2
                 )
                 i -= 1  # reduce i so that output tol_history indexing is correct
                 break
@@ -454,7 +455,8 @@ def drpls(self, data, lam=1e5, eta=0.5, max_iter=50, tol=1e-3, weights=None, dif
                 # checking a scalar is faster; cannot use np.errstate since it is not 100% reliable
                 warnings.warn(
                     ('nan and/or +/- inf occurred in weighting calculation, likely meaning '
-                     '"tol" is too low and/or "max_iter" is too high'), ParameterWarning
+                     '"tol" is too low and/or "max_iter" is too high'), ParameterWarning,
+                    stacklevel=2
                 )
                 break
             elif calc_difference < tol:
@@ -530,7 +532,8 @@ def iarpls(self, data, lam=1e5, diff_order=2, max_iter=50, tol=1e-3, weights=Non
                 # checking a scalar is faster; cannot use np.errstate since it is not 100% reliable
                 warnings.warn(
                     ('nan and/or +/- inf occurred in weighting calculation, likely meaning '
-                     '"tol" is too low and/or "max_iter" is too high'), ParameterWarning
+                     '"tol" is too low and/or "max_iter" is too high'), ParameterWarning,
+                    stacklevel=2
                 )
                 break
             elif calc_difference < tol:

pyproject.toml

@@ -42,9 +42,8 @@ keywords = [
 ]
 requires-python = ">=3.8"
 dependencies = [
-    # lowest versions supported for python 3.8
-    "numpy>=1.18",
-    "scipy>=1.5",
+    "numpy>=1.20",  # lowest version to allow dtype for np.concatenate
+    "scipy>=1.5",  # lowest versions supported for python 3.8
 ]
 
 [project.urls]

tests/conftest.py

@@ -441,10 +441,10 @@ def test_output(self, additional_keys=None, **kwargs):
         # check all entries in output param dictionary
         for key in total_keys:
             if key not in output[1]:
-                assert False, f'key "{key}" missing from param dictionary'
+                raise AssertionError(f'key "{key}" missing from param dictionary')
             output[1].pop(key)
         if output[1]:
-            assert False, f'unchecked keys in param dictionary: {output[1]}'
+            raise AssertionError(f'unchecked keys in param dictionary: {output[1]}')
 
     def test_x_ordering(self, assertion_kwargs=None, **kwargs):
         """Ensures arrays are correctly sorted within the function."""
@@ -688,10 +688,10 @@ def test_output(self, additional_keys=None, **kwargs):
         # check all entries in output param dictionary
         for key in total_keys:
             if key not in output[1]:
-                assert False, f'key "{key}" missing from param dictionary'
+                raise AssertionError(f'key "{key}" missing from param dictionary')
             output[1].pop(key)
         if output[1]:
-            assert False, f'unchecked keys in param dictionary: {output[1]}'
+            raise AssertionError(f'unchecked keys in param dictionary: {output[1]}')
 
     def test_xz_ordering(self, assertion_kwargs=None, **kwargs):
         """Ensures arrays are correctly sorted within the function."""

tests/test_spline_utils.py

@@ -422,7 +422,7 @@ def test_pspline_tck_none(data_fixture):
 
     assert pspline.coef is None
     with pytest.raises(ValueError):
-        pspline.tck
+        tck = pspline.tck
 
 
 def test_pspline_tck_readonly(data_fixture):

tests/two_d/test_spline_utils.py

@@ -263,7 +263,7 @@ def test_pspline_tck_none(data_fixture2d):
 
     assert pspline.coef is None
     with pytest.raises(ValueError):
-        pspline.tck
+        tck = pspline.tck
 
 
 def test_pspline_tck_readonly(data_fixture2d):