Add style check test (flake8)

.circleci/config.yml

@@ -83,6 +83,22 @@ jobs:
           paths:
             - src/coverage/.coverage.py310
 
+  style_check: 
+    docker:
+      - image: cimg/python:3.7
+    working_directory: /tmp/src/nigsp
+    resource_class: small
+    steps:
+      - checkout
+      - python/install-packages:
+          path-args: .[style]
+          pypi-cache: false
+          venv-cache: false
+          pkg-manager: pip-dist
+      - run:
+          name: Check style
+          command: flake8 ./nigsp
+
   merge_coverage:
     working_directory: /tmp/src/nigsp
     docker:
@@ -113,8 +129,13 @@ workflows:
   build_test: # This is the name of the workflow, feel free to change it to better match your workflow.
     # Inside the workflow, you define the jobs you want to run.
     jobs:
-      - test37
-      - test310
+      - style_check
+      - test37:
+          requires:
+            - style_check
+      - test310:
+          requires:
+            - style_check
       - merge_coverage:
           requires:
             - test37

nigsp/io.py

@@ -142,7 +142,7 @@ def check_mtx_dim(fname, data, shape=None):
     -------
     np.ndarray
         If `data.ndim` = 2, returns data.
-        If `data.ndim` = 1 and `shape` == 'rectangle', 
+        If `data.ndim` = 1 and `shape` == 'rectangle',
         Returns data with added empty axis.
 
     Raises
@@ -268,8 +268,8 @@ def load_mat(fname, shape=None):
     """
     Read files in matlab format.
 
-    Assumes the existence of a matrix/vector in the mat file, rendered as 
-    a numpy.ndarray. If there is more than a marix, the one with the largest 
+    Assumes the existence of a matrix/vector in the mat file, rendered as
+    a numpy.ndarray. If there is more than a marix, the one with the largest
     size will be selected.
 
     Parameters
@@ -313,7 +313,8 @@ def load_mat(fname, shape=None):
         # Check data key only if it's not hidden
         # (skip '__header__', '__version__', '__global__')
         if '__' not in k:
-            LGR.info(f'Checking {fname} key {str(k)} content for data (float array/matrices in MATLAB).')
+            LGR.info(f'Checking {fname} key {str(k)} content for data '
+                     '(float array/matrices in MATLAB).')
             if type(data[k]) is np.ndarray:
                 data_keys.append(k)
 

nigsp/objects.py

@@ -25,6 +25,7 @@
 
 
 class SCGraph():
+    """Main module object, containing all data representing the graph."""
 
     def __init__(self, mtx, timeseries, atlas=None, filename=None, img=None,
                  eigenval=None, eigenvec=None, energy=None, lapl_mtx=None,
@@ -170,12 +171,14 @@ def test_significance(self, method='Bernoulli', p=0.05,
         if self.sdi is not None:
             surr_sdi = operations.sdi(self.surr_split, mean, keys=None)
             self.sdi = operations.test_significance(surr=surr_sdi, data=self.sdi,
-                                                    method=method, p=p, return_masked=return_masked,
+                                                    method=method, p=p,
+                                                    return_masked=return_masked,
                                                     mean=mean)
         if self.gsdi is not None:
             surr_sdi = operations.gsdi(self.surr_split, mean, keys=None)
             self.gsdi = operations.test_significance(surr=surr_sdi, data=self.gsdi,
-                                                     method=method, p=p, return_masked=return_masked,
+                                                     method=method, p=p,
+                                                     return_masked=return_masked,
                                                      mean=mean)
         return self
 

nigsp/operations/__init__.py

@@ -8,7 +8,7 @@
 `nigsp` module.
 
 For example, calling `nigsp.operations.metrics.sdi` is equivalent to calling
-`nigsp.operations.sdi` or `nigsp.metrisc.sdi`. 
+`nigsp.operations.sdi` or `nigsp.metrisc.sdi`.
 """
 
 

nigsp/operations/laplacian.py

@@ -34,7 +34,7 @@ def symmetric_normalisation(mtx):
     --------
     https://en.wikipedia.org/wiki/Laplacian_matrix#Symmetrically_normalized_Laplacian_2
     """
-    d = np.diag(mtx.sum(axis=-1) ** (-1/2))
+    d = np.diag(mtx.sum(axis=-1) ** (-1 / 2))
 
     symm_norm = (d @ mtx @ d)
 
@@ -61,7 +61,7 @@ def decomposition(mtx):
 
     idx = np.argsort(eigenval)
     eigenval = eigenval[idx]
-    # #!# Check that eigenvec has the right index and not inverted 
+    # #!# Check that eigenvec has the right index and not inverted
     eigenvec = eigenvec[:, idx]
 
     return eigenval, eigenvec

nigsp/operations/metrics.py

@@ -37,7 +37,7 @@ def sdi(ts_split, mean=False, keys=None):
         A dictionary containing two entries. If the two entries are "low" and
         "high", then SDI will be computed as the norm of the high vs the norm
         of the low, oterwise as the ratio between the second (second key in
-        sorted keys) and the first. 
+        sorted keys) and the first.
     mean : bool, optional
         If True, compute mean over the last axis (e.g. between subjects)
     keys : None or list of strings, optional

nigsp/operations/nifti.py

@@ -97,7 +97,7 @@ def apply_mask(data, mask):
     if data.shape[:mask.ndim] != mask.shape:
         raise ValueError(f'Cannot mask data with shape {data.shape} using mask '
                          f'with shape {mask.shape}')
-    if (data.ndim-mask.ndim) > 1:
+    if (data.ndim - mask.ndim) > 1:
         LGR.warning(f'Returning volume with {data.ndim-mask.ndim+1} dimensions.')
     else:
         LGR.info(f'Returning {data.ndim-mask.ndim+1}D array.')
@@ -131,7 +131,7 @@ def unmask(data, mask, shape=None, asdata=None):
     ------
     ValueError
         If both `shape` and `asdata` are empty
-        If the first dimension of `data` and the number of available voxels in 
+        If the first dimension of `data` and the number of available voxels in
         mask do not match.
         If the mask shape does not match the first (mask)
     """
@@ -193,21 +193,22 @@ def apply_atlas(data, atlas, mask=None):
     # #!# Add nilearn's fetching atlases utility
 
     if atlas.ndim > 3:
-        raise NotImplementedError(f'Files with {atlas.ndim} dimensions are not supported as atlases.')
+        raise NotImplementedError(f'Files with {atlas.ndim} dimensions are not '
+                                  'supported as atlases.')
     if data.shape[:mask.ndim] != mask.shape:
         raise ValueError(f'Cannot mask data with shape {data.shape} using mask '
                          f'with shape {mask.shape}')
     if data.shape[:atlas.ndim] != atlas.shape:
         raise ValueError(f'Cannot apply atlas with shape {atlas.shape} on data '
                          f'with shape {data.shape}')
-    if (data.ndim-atlas.ndim) > 1:
+    if (data.ndim - atlas.ndim) > 1:
         LGR.warning(f'returning volume with {data.ndim-atlas.ndim+1} dimensions.')
     else:
         LGR.info(f'Returning {data.ndim-atlas.ndim+1}D array of signal averages '
                  f'in atlas {atlas}.')
 
     # Mask data and atlas first
-    atlas = atlas*mask
+    atlas = atlas * mask
     labels = np.unique(atlas)
     labels = labels[labels > 0]
     LGR.info(f'Labels: {labels}, numbers: {len(labels)}')
@@ -268,7 +269,7 @@ def unfold_atlas(data, atlas, mask=None):
     if atlas.shape[:mask.ndim] != mask.shape:
         raise ValueError(f'Cannot mask atlas with shape {atlas.shape} using mask '
                          f'with shape {mask.shape}')
-    atlas = atlas*mask
+    atlas = atlas * mask
 
     labels = np.unique(atlas)
     labels = labels[labels > 0]

nigsp/operations/surrogates.py

@@ -365,17 +365,17 @@ def test_significance(surr,
     # the real data index (real_idx) is at the extremes of the matrix last axis
     # (with tolerance on the extremes depending on p).
     # real_idx serendipitously is the number of surrogates.
-    stat_mask = (reord_surr[..., :floor(real_idx * p) + 1].any(axis=-1) +
-                 reord_surr[..., -floor(real_idx * p) - 1:].any(axis=-1))
+    stat_mask = (reord_surr[..., :floor(real_idx * p) + 1].any(axis=-1)
+                 + reord_surr[..., -floor(real_idx * p) - 1:].any(axis=-1))
 
     if method == 'Bernoulli' and surr.shape[1] > 1 and surr.ndim >= 3:
         # The following computes the CDF of a binomial distribution
         # Difference with scipy's binom.cdf (100 samples) is: 5.066394802133445e-06
         # #!# See if there is a quicker way to get this (probably invert testing)
 
         def _pmf(x, n, p):
-            f = ((factorial(n) / (factorial(x) * factorial(n - x))) * p**x *
-                 (1 - p)**(n - x))
+            f = ((factorial(n) / (factorial(x) * factorial(n - x))) * p**x
+                 * (1 - p)**(n - x))
             return f
 
         x = np.arange(0, 100, 1)

nigsp/operations/timeseries.py

@@ -42,8 +42,8 @@ def normalise_ts(timeseries):
                     'Returning it as is.')
         return timeseries
 
-    z = ((timeseries - timeseries.mean(axis=1)[:, np.newaxis, ...]) /
-         timeseries.std(axis=1, ddof=1)[:, np.newaxis, ...])
+    z = ((timeseries - timeseries.mean(axis=1)[:, np.newaxis, ...])
+         / timeseries.std(axis=1, ddof=1)[:, np.newaxis, ...])
     z[np.isnan(z)] = 0
 
     return z
@@ -133,7 +133,7 @@ def median_cutoff_frequency_idx(energy):
 
     if energy.ndim == 2:
         energy = energy.mean(axis=-1)
-    half_tot_auc = np.trapz(energy, axis=0)/2
+    half_tot_auc = np.trapz(energy, axis=0) / 2
     LGR.debug(f'Total AUC = {half_tot_auc*2}, targetting half of total AUC')
 
     # Compute the AUC from first to one to last frequency,
@@ -200,17 +200,17 @@ def graph_filter(timeseries, eigenvec, freq_idx, keys=['low', 'high']):
     LGR.info(f'Splitting graph into {len(freq_idx)+1} parts')
 
     # Check that there is the right amount of keys
-    if len(keys) > len(freq_idx)+1:
+    if len(keys) > len(freq_idx) + 1:
         LGR.warning(f'The declared keys list ({keys}) has {len(keys)} elements. '
                     f'Since the frequency index list ({freq_idx}) has {len(freq_idx)}, '
                     f'any keys after {keys[len(freq_idx)]} will be ignored.')
-        keys = keys[:len(freq_idx)+1]
-    elif len(keys) < len(freq_idx)+1:
+        keys = keys[:len(freq_idx) + 1]
+    elif len(keys) < len(freq_idx) + 1:
         LGR.warning(f'The declared keys list ({keys}) has {len(keys)} elements. '
                     f'Since the frequency index list ({freq_idx}) has {len(freq_idx)}, '
                     f'more keys will be created after {keys[len(freq_idx)]} .')
 
-        for i in range(len(keys), len(freq_idx)+1):
+        for i in range(len(keys), len(freq_idx) + 1):
             keys = keys + [f'key-{i+1:03d}']
 
     # Add 0 and None to freq_idx to have full indexes
@@ -287,7 +287,7 @@ def _fc(timeseries, mean=False):
         elif timeseries.ndim == 2:
             return np.corrcoef(timeseries)
         else:
-            # reshape the array to allow reiteration on unknown dimensions of timeseries 
+            # reshape the array to allow reiteration on unknown dimensions of timeseries
             temp_ts, _ = prepare_ndim_iteration(timeseries, 2)
             fcorr = np.empty(([temp_ts.shape[0]] * 2 + [temp_ts.shape[-1]]), dtype='float32')
             for i in range(temp_ts.shape[-1]):

nigsp/utils.py

@@ -19,7 +19,7 @@
 
 def pairwise(iterable):
     """
-    Recreate `itertools.pairwise()` behaviour for python < 3.10 compatibility.
+    Recreate `itertools.pairwise` behaviour for python < 3.10 compatibility.
 
     Parameters
     ----------
@@ -125,7 +125,7 @@ def prepare_ndim_iteration(data, idx):
     np.ndarray, np.ndarray
         The reshaped data and an empty array like it.
     """
-    if data.ndim > idx+1:
+    if data.ndim > idx + 1:
         new_shape = list(data.shape[:idx]) + [prod(data.shape[idx:])]
         data = data.reshape(new_shape)
 

setup.cfg

@@ -69,12 +69,13 @@ console_scripts =
 [flake8]
 doctest = True
 exclude=
-    *build/
+    _version.py
+    ./nigsp/cli/__init__.py
     tests
-ignore = E126, E402, W503
+ignore = E126, E402, W503, F401, F811
 max-line-length = 99
 per-file-ignores =
-    */__init__.py:F401
+    workflow.py:D401
 
 [tool:pytest]
 doctest_optionflags = NORMALIZE_WHITESPACE