MAINT: Use explicitly public APIs (#521)

Co-authored-by: Daniel McCloy <[email protected]>

.github/workflows/macos_conda.yml

@@ -31,11 +31,12 @@ jobs:
       - uses: actions/checkout@v2
         with:
           fetch-depth: 0
-      - uses: conda-incubator/setup-miniconda@v2
+      - uses: mamba-org/setup-micromamba@v1
         with:
-          activate-environment: 'mne-nirs'
-          python-version: ${{ env.PYTHON_VERSION }}
           environment-file: ${{ env.CONDA_ENV }}
+          environment-name: 'mne-nirs'
+          create-args: >-
+            python=${{ env.PYTHON_VERSION }}
         name: 'Setup conda'
       - shell: bash -el {0}
         run: |

examples/general/plot_10_hrf_simulation.py

@@ -60,7 +60,7 @@
 design_matrix = make_first_level_design_matrix(raw, stim_dur=5.0,
                                                drift_order=1,
                                                drift_model='polynomial')
-fig, ax1 = plt.subplots(figsize=(10, 6), nrows=1, ncols=1)
+fig, ax1 = plt.subplots(figsize=(10, 6), constrained_layout=True)
 fig = plot_design_matrix(design_matrix, ax=ax1)
 
 
@@ -98,20 +98,20 @@ def print_results(glm_est, truth):
 # and plot the noisy data and the GLM fitted model.
 # We print the response estimate and see that is close, but not exactly correct,
 # we observe the mean square error is similar to the added noise.
-# Note that the clean data plot is so similar to the GLM estimate that it is hard to see unless zoomed in. 
+# Note that the clean data plot is so similar to the GLM estimate that it is hard to see unless zoomed in.
 
 # First take a copy of noise free data for comparison
 raw_noise_free = raw.copy()
 
 raw._data += np.random.normal(0, np.sqrt(1e-11), raw._data.shape)
 glm_est = run_glm(raw, design_matrix)
 
-plt.plot(raw.times, raw_noise_free.get_data().T * 1e6)
-plt.plot(raw.times, raw.get_data().T * 1e6, alpha=0.3)
-plt.plot(raw.times, glm_est.theta()[0][0] * design_matrix["A"].values * 1e6)
-plt.xlabel("Time (s)")
-plt.ylabel("Haemoglobin (uM)")
-plt.legend(["Clean Data", "Noisy Data", "GLM Estimate"])
+fig, ax = plt.subplots(constrained_layout=True)
+ax.plot(raw.times, raw_noise_free.get_data().T * 1e6)
+ax.plot(raw.times, raw.get_data().T * 1e6, alpha=0.3)
+ax.plot(raw.times, glm_est.theta()[0][0] * design_matrix["A"].values * 1e6)
+ax.set(xlabel="Time (s)", ylabel="Haemoglobin (μM)")
+ax.legend(["Clean Data", "Noisy Data", "GLM Estimate"])
 
 print_results(glm_est, amp)
 
@@ -138,12 +138,12 @@ def print_results(glm_est, truth):
                                                drift_model='polynomial')
 glm_est = run_glm(raw, design_matrix)
 
-plt.plot(raw.times, raw_noise_free.get_data().T * 1e6)
-plt.plot(raw.times, raw.get_data().T * 1e6, alpha=0.3)
-plt.plot(raw.times, glm_est.theta()[0][0] * design_matrix["A"].values * 1e6)
-plt.xlabel("Time (s)")
-plt.ylabel("Haemoglobin (uM)")
-plt.legend(["Clean Data", "Noisy Data", "GLM Estimate"])
+fig, ax = plt.subplots(constrained_layout=True)
+ax.plot(raw.times, raw_noise_free.get_data().T * 1e6)
+ax.plot(raw.times, raw.get_data().T * 1e6, alpha=0.3)
+ax.plot(raw.times, glm_est.theta()[0][0] * design_matrix["A"].values * 1e6)
+ax.set(xlabel="Time (s)", ylabel="Haemoglobin (μM)")
+ax.legend(["Clean Data", "Noisy Data", "GLM Estimate"])
 
 print_results(glm_est, amp)
 
@@ -171,11 +171,11 @@ def print_results(glm_est, truth):
                                                drift_model='polynomial')
 glm_est = run_glm(raw, design_matrix)
 
-plt.plot(raw.times, raw.get_data().T * 1e6, alpha=0.3)
-plt.plot(raw.times, glm_est.theta()[0][0] * design_matrix["A"].values * 1e6)
-plt.xlabel("Time (s)")
-plt.ylabel("Haemoglobin (uM)")
-plt.legend(["Noisy Data", "GLM Estimate"])
+fig, ax = plt.subplots(constrained_layout=True)
+ax.plot(raw.times, raw.get_data().T * 1e6, alpha=0.3)
+ax.plot(raw.times, glm_est.theta()[0][0] * design_matrix["A"].values * 1e6)
+ax.set(xlabel="Time (s)", ylabel="Haemoglobin (μM)")
+ax.legend(["Noisy Data", "GLM Estimate"])
 
 print_results(glm_est, amp)
 
@@ -193,12 +193,12 @@ def print_results(glm_est, truth):
 
 glm_est = run_glm(raw, design_matrix, noise_model='ar5')
 
-fig, axes = plt.subplots(nrows=1, ncols=1, figsize=(15, 6))
-plt.plot([-0.58853134, -0.29575669, -0.52246482, 0.38735476, 0.02428681],
-         axes=axes)  # actual values from model above
-plt.plot(glm_est.model()[0].rho * -1.0,  axes=axes)  # estimates
-plt.legend(["Simulation AR coefficients", "Estimated AR coefficients"])
-plt.xlabel("Coefficient")
+fig, ax = plt.subplots(figsize=(15, 6), constrained_layout=True)
+# actual values from model above
+ax.plot([-0.58853134, -0.29575669, -0.52246482, 0.38735476, 0.02428681])
+ax.plot(glm_est.model()[0].rho * -1.0)  # estimates
+ax.legend(["Simulation AR coefficients", "Estimated AR coefficients"])
+ax.set_xlabel("Coefficient")
 
 
 # %%

examples/general/plot_11_hrf_measured.py

@@ -143,10 +143,10 @@
 # line is raised for the duration of the stimulus/condition.
 
 s = mne_nirs.experimental_design.create_boxcar(raw_haemo)
-fig, axes = plt.subplots(nrows=1, ncols=1, figsize=(15, 6))
-plt.plot(raw_haemo.times, s, axes=axes)
-plt.legend(["Control", "Left", "Right"], loc="upper right")
-plt.xlabel("Time (s)");
+fig, ax = plt.subplots(figsize=(15, 6), constrained_layout=True)
+ax.plot(raw_haemo.times, s)
+ax.legend(["Control", "Left", "Right"], loc="upper right")
+ax.set_xlabel("Time (s)");
 
 
 # %%
@@ -203,7 +203,7 @@
 # The next columns illustrate the drift and constant components.
 # The last columns illustrate the short channel signals.
 
-fig, ax1 = plt.subplots(figsize=(10, 6), nrows=1, ncols=1)
+fig, ax1 = plt.subplots(figsize=(10, 6), constrained_layout=True)
 fig = plot_design_matrix(design_matrix, ax=ax1)
 
 
@@ -227,13 +227,12 @@
 # name. Note however, that this is just for visualisation and does not affect
 # the results below.
 
+fig, ax = plt.subplots(constrained_layout=True)
 s = mne_nirs.experimental_design.create_boxcar(raw_intensity, stim_dur=5.0)
-plt.plot(raw_intensity.times, s[:, 1])
-plt.plot(design_matrix['Tapping_Left'])
-plt.xlim(180, 300)
-plt.legend(["Stimulus", "Expected Response"])
-plt.xlabel("Time (s)")
-plt.ylabel("Amplitude")
+ax.plot(raw_intensity.times, s[:, 1])
+ax.plot(design_matrix['Tapping_Left'])
+ax.legend(["Stimulus", "Expected Response"])
+ax.set(xlim=(180, 300), xlabel="Time (s)", ylabel="Amplitude")
 
 
 # %%

mne_nirs/conftest.py

@@ -48,8 +48,15 @@ def pytest_configure(config):
     ignore:The register_cmap function.*:
     ignore:The get_cmap function.*:
     ignore:The figure layout has changed.*:UserWarning
-    # old MNE
-    ignore:The `pyvista.plotting.plotting` module.*:
+    # H5py
+    ignore:`product` is deprecated as of NumPy.*:DeprecationWarning
+    # seaborn
+    ignore:is_categorical_dtype is deprecated.*:FutureWarning
+    ignore:use_inf_as_na option is deprecated.*:FutureWarning
+    # nilearn
+    ignore:The provided callable <function sum.*:FutureWarning
+    # TODO: in an example (should fix eventually)
+    ignore:The behavior of DataFrame concatenation.*:FutureWarning
     """  # noqa: E501
     for warning_line in warning_lines.split('\n'):
         warning_line = warning_line.strip()

mne_nirs/preprocessing/tests/test_mayer.py

@@ -12,6 +12,9 @@
 pytest.importorskip('fooof')
 
 
+# in fooof->scipy.optimize.curve_fit
+@pytest.mark.filterwarnings("ignore:divide by zero encountered in divide.*:")
+@pytest.mark.filterwarnings("ignore:invalid value encountered in.*:")
 def test_mayer():
     fnirs_data_folder = mne.datasets.fnirs_motor.data_path()
     fnirs_raw_dir = os.path.join(fnirs_data_folder, 'Participant-1')

mne_nirs/statistics/_glm_level_first.py

@@ -16,14 +16,14 @@
     import nilearn.glm
     from nilearn.glm.first_level import run_glm as nilearn_glm
 
-try:  # remove once MNE 1.0 is required
-    from mne.io.meas_info import ContainsMixin
+try:  # remove once MNE 1.6 is required
+    from mne._fiff.meas_info import ContainsMixin
 except ImportError:
-    from mne.channels.channels import ContainsMixin
+    from mne.io.meas_info import ContainsMixin
 from mne.utils import fill_doc, warn, verbose, check_fname, _validate_type
-from mne.io.pick import _picks_to_idx, pick_info
+from mne.io.pick import _picks_to_idx
 from mne.io.constants import FIFF
-from mne import Info
+from mne import Info, pick_info
 
 from ..visualisation._plot_GLM_topo import _plot_glm_topo,\
     _plot_glm_contrast_topo

mne_nirs/statistics/_statsmodels.py

@@ -2,6 +2,8 @@
 #
 # License: BSD (3-clause)
 
+from io import StringIO
+
 import pandas as pd
 import numpy as np
 
@@ -14,7 +16,7 @@ def summary_to_dataframe(summary):
     '''
     results = summary.tables[1]
     if type(results) is not pd.core.frame.DataFrame:
-        results = summary.tables[1].as_html()
+        results = StringIO(summary.tables[1].as_html())
         results = pd.read_html(results, header=0, index_col=0)[0]
     return results
 

mne_nirs/statistics/tests/test_glm_type.py

@@ -298,6 +298,9 @@ def test_create_results_glm_contrast():
     df = res.to_dataframe()
     assert df.shape == (n_channels, 10)
 
+    src = np.asarray(df["Source"])
+    assert src.dtype.kind == "i"
+
 
 def test_results_glm_io():
 

mne_nirs/tests/test_examples.py

@@ -7,7 +7,6 @@
 import os
 import pytest
 import sys
-import warnings
 
 from mne.utils import check_version
 
@@ -26,14 +25,6 @@ def examples_path():
     return path
 
 
-def run_script_and_check(test_file_path):
-    with open(test_file_path) as fid, warnings.catch_warnings():
-        # Ignore deprecation warning caused by
-        # app.setAttribute(Qt.AA_UseHighDpiPixmaps) in mne-python
-        warnings.filterwarnings("ignore", category=DeprecationWarning)
-        return exec(fid.read(), locals(), locals())
-
-
 requires_mne_1p2 = pytest.mark.skipif(
     not check_version('mne', '1.2'), reason='Needs MNE-Python 1.2')
 # https://github.com/mne-tools/mne-bids/pull/406
@@ -53,6 +44,8 @@ def run_script_and_check(test_file_path):
 
 
 @pytest.mark.filterwarnings('ignore:No bad channels to interpolate.*:')
+@pytest.mark.filterwarnings('ignore:divide by zero encountered.*:')
+@pytest.mark.filterwarnings('ignore:invalid value encountered.*:')
 @pytest.mark.skipif(
     sys.platform.startswith('win'), reason='Unstable on Windows')
 @pytest.mark.examples
@@ -76,4 +69,6 @@ def run_script_and_check(test_file_path):
     "plot_99_bad.py"]))
 def test_examples(fname, requires_pyvista):
     test_file_path = examples_path() + fname
-    run_script_and_check(test_file_path)
+    with open(test_file_path) as fid:
+        code = fid.read()
+    exec(code, locals(), locals())

mne_nirs/utils/_io.py

@@ -110,7 +110,7 @@ def _tidy_RegressionResults(data, glm_est, design_matrix):
         mse_estimates[idx, :] = glm_est[name].MSE[0]
         for cond_idx, cond in enumerate(design_matrix.columns):
             t_estimates[idx, cond_idx] = glm_est[name].t(
-                column=cond_idx)
+                column=cond_idx).item()
             p_estimates[idx, cond_idx] = 2 * stats.t.cdf(
                 -1.0 * np.abs(t_estimates[idx, cond_idx]),
                 df=df_estimates[idx, cond_idx])
@@ -171,12 +171,18 @@ def _tidy_long_to_wide(d, expand_output=True):
 
     if expand_output:
         try:
-            d["Source"] = [re.search(r'S(\d+)_D(\d+) (\w+)', ch).group(1)
-                           for ch in d["ch_name"]]
-            d["Detector"] = [re.search(r'S(\d+)_D(\d+) (\w+)', ch).group(2)
-                             for ch in d["ch_name"]]
-            d["Chroma"] = [re.search(r'S(\d+)_D(\d+) (\w+)', ch).group(3)
-                           for ch in d["ch_name"]]
+            d["Source"] = [
+                int(re.search(r'S(\d+)_D(\d+) (\w+)', ch).group(1))
+                for ch in d["ch_name"]
+            ]
+            d["Detector"] = [
+                int(re.search(r'S(\d+)_D(\d+) (\w+)', ch).group(2))
+                for ch in d["ch_name"]
+            ]
+            d["Chroma"] = [
+                re.search(r'S(\d+)_D(\d+) (\w+)', ch).group(3)
+                for ch in d["ch_name"]
+            ]
         except AttributeError:
             warn("Non standard source detector names used")
         d["Significant"] = d["p_value"] < 0.05

mne_nirs/visualisation/_plot_3d_montage.py

@@ -7,10 +7,9 @@
 
 import numpy as np
 
-from mne import pick_info, pick_types
+from mne import pick_info, pick_types, Info
 from mne.channels import make_standard_montage
 from mne.channels.montage import transform_to_head
-from mne.io import Info
 from mne.transforms import _get_trans, apply_trans
 from mne.utils import _validate_type, _check_option, verbose, logger
 from mne.viz import Brain

mne_nirs/visualisation/_plot_GLM_topo.py

@@ -10,10 +10,10 @@
 import matplotlib as mpl
 from mpl_toolkits.axes_grid1.axes_divider import make_axes_locatable
 
-from mne import Info
+from mne import Info, pick_info
 from mne.utils import warn
 from mne.channels.layout import _merge_ch_data
-from mne.io.pick import _picks_to_idx, _get_channel_types, pick_info
+from mne.io.pick import _picks_to_idx
 from mne.viz import plot_topomap
 
 
@@ -38,7 +38,7 @@ def _plot_glm_topo(inst, glm_estimates, design_matrix, *,
     for idx, name in enumerate(glm_estimates.keys()):
         estimates[idx, :] = glm_estimates[name].theta.T
 
-    types = np.unique(_get_channel_types(info))
+    types = np.unique(info.get_channel_types())
 
     if requested_conditions is None:
         requested_conditions = design_matrix.columns
@@ -100,7 +100,7 @@ def _plot_glm_contrast_topo(inst, contrast, figsize=(12, 7), sphere=None):
     info = deepcopy(inst if isinstance(inst, Info) else inst.info)
 
     # Extract types. One subplot is created per type (hbo/hbr)
-    types = np.unique(_get_channel_types(info))
+    types = np.unique(info.get_channel_types())
 
     # Extract values to plot and rescale to uM
     estimates = contrast.effect[0]

tools/circleci_dependencies.sh

@@ -2,5 +2,5 @@
 
 
 python -m pip install --upgrade pip setuptools wheel
-python -m pip install --upgrade --progress-bar off -r requirements.txt -r requirements_testing.txt -r requirements_doc.txt git+https://github.com/mne-tools/mne-python.git@main
+python -m pip install --upgrade --progress-bar off -r requirements.txt -r requirements_testing.txt -r requirements_doc.txt git+https://github.com/mne-tools/mne-python.git@main "mne-qt-browser[opengl] @ git+https://github.com/mne-tools/mne-qt-browser.git@main"
 python -m pip install -e .

tools/github_actions_dependencies.sh

@@ -7,24 +7,15 @@ if [ ! -z "$CONDA_ENV" ]; then
 else
 	# Changes here should also go in the interactive_test CircleCI job
 	python -m pip install $STD_ARGS pip setuptools wheel
-	echo "Numpy"
-	pip uninstall -yq numpy
-	echo "Date utils"
-	# https://pip.pypa.io/en/latest/user_guide/#possible-ways-to-reduce-backtracking-occurring
-	pip install $STD_ARGS --pre --only-binary ":all:" python-dateutil pytz joblib threadpoolctl six
-    echo "pyside6"
-	pip install $STD_ARGS --pre --only-binary ":all:" pyside6
 	echo "NumPy/SciPy/pandas etc."
-	# TODO: Currently missing dipy for 3.10 https://github.com/dipy/dipy/issues/2489
-	pip install $STD_ARGS --pre --only-binary ":all:" --no-deps  --default-timeout=60 -i "https://pypi.anaconda.org/scipy-wheels-nightly/simple" numpy scipy pandas scikit-learn statsmodels dipy
-	echo "H5py, pillow, matplotlib"
-	pip install $STD_ARGS --pre --only-binary ":all:" --no-deps -f "https://7933911d6844c6c53a7d-47bd50c35cd79bd838daf386af554a83.ssl.cf2.rackcdn.com" h5py pillow matplotlib
-	echo "Numba, nilearn"
-	pip install $STD_ARGS --pre --only-binary ":all:" numba llvmlite nilearn
+	pip install $STD_ARGS --pre --only-binary ":all:" --default-timeout=60 --extra-index-url "https://pypi.anaconda.org/scientific-python-nightly-wheels/simple" numpy scipy scikit-learn pandas matplotlib pillow statsmodels
+	pip install $STD_ARGS --pre --only-binary ":all:" --no-deps  --default-timeout=60 -i "https://pypi.anaconda.org/scipy-wheels-nightly/simple" dipy
+	echo "H5py"
+	pip install $STD_ARGS --pre --only-binary ":all:" --no-deps -f "https://7933911d6844c6c53a7d-47bd50c35cd79bd838daf386af554a83.ssl.cf2.rackcdn.com" h5py
+	echo "nilearn"
+	pip install $STD_ARGS --pre --only-binary ":all:" nilearn
 	echo "VTK"
-	# Have to use our own version until VTK releases a 3.10 build
-	wget -q https://osf.io/ajder/download -O vtk-9.1.20220406.dev0-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl
-	pip install $STD_ARGS --pre --only-binary ":all:" vtk-9.1.20220406.dev0-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl
+	pip install $STD_ARGS --pre --only-binary ":all:" vtk
 	python -c "import vtk"
 	echo "PyVista"
 	pip install --progress-bar off git+https://github.com/pyvista/pyvista