More clean up

varats/varats/data/databases/feature_perf_precision_database.py

@@ -224,7 +224,8 @@ def _is_feature_relevant(
         old_mean = np.mean(old_measurements)
         new_mean = np.mean(new_measurements)
 
-        if old_mean < self.absolute_cut_off and new_mean < self.absolute_cut_off:
+        if old_mean < self.absolute_cut_off and \
+                new_mean < self.absolute_cut_off:
             return False
 
         old_rel_cut_off = old_mean * self.relative_cut_off
@@ -251,22 +252,21 @@ def _precise_pim_regression_check(
 
                 new_values = current_pim[feature]
 
-                # Skip features that seem not to be relevant for regressions testing
+                # Skip features that seem not to be relevant
+                # for regressions testing
                 if not self._is_feature_relevant(old_values, new_values):
                     continue
 
-                is_regression = is_regression or self._is_significantly_different(
-                    old_values, new_values
-                )
+                is_regression = is_regression or \
+                    self._is_significantly_different(
+                        old_values, new_values
+                    )
             else:
                 if np.mean(old_values) > self.absolute_cut_off:
                     print(
                         f"Could not find feature {feature} in new trace. "
                         f"({np.mean(old_values)}us lost)"
                     )
-                # TODO: how to handle this?
-                # raise NotImplementedError()
-                # is_regression = True
 
         return is_regression
 
@@ -302,8 +302,8 @@ def _sum_pim_regression_check(
 
         mean_baseline = np.mean(baseline_pim_total)
         mean_diff = abs(mean_baseline - np.mean(current_pim_total))
-        if mean_diff < self.absolute_cut_off \
-                or mean_diff < mean_baseline * self.relative_cut_off:
+        if mean_diff < self.absolute_cut_off or \
+                mean_diff < mean_baseline * self.relative_cut_off:
             return False
 
         return self._is_significantly_different(
@@ -596,14 +596,13 @@ def compute_profiler_predictions(
                 report_files[0], patch_name
             )
         except Exception as exception:  # pylint: disable=W0718
+            # Print exception information but continue working on the plot/table
             print(
                 f"FAILURE: Skipping {config_id=} of {project_name=}, "
                 f"profiler={profiler.name}"
             )
             print(exception)
             print(traceback.format_exc())
-            # TODO: clean up
-            # raise exception
 
     return result_dict
 

varats/varats/tables/feature_perf_precision.py

@@ -36,15 +36,18 @@
 
 
 def cmap_map(
-    function, cmap: colors.LinearSegmentedColormap
+    function: tp.Callable[[np.ndarray[np.float64]], np.float64],
+    cmap: colors.LinearSegmentedColormap
 ) -> colors.LinearSegmentedColormap:
     """
     Applies function (which should operate on vectors of shape 3: [r, g, b]), on
     colormap cmap.
 
     This routine will break any discontinuous points in a colormap.
     """
-    c_dict = cmap._segmentdata  # pylint: disable=protected-access  # type: ignore
+    # pylint: disable=protected-access,attr-defined
+    c_dict = cmap._segmentdata  # type: ignore
+    # pylint: enable=protected-access,attr-defined
     step_dict: tp.Dict[str, tp.List[tp.Any]] = {}
 
     # First get the list of points where the segments start or end
@@ -83,7 +86,7 @@ class FeaturePerfPrecisionTable(Table, table_name="fperf_precision"):
     @staticmethod
     def _prepare_data_table(
         case_studies: tp.List[CaseStudy], profilers: tp.List[Profiler]
-    ):
+    ) -> pd.DataFrame:
         df = pd.DataFrame()
         table_rows = []
 
@@ -109,7 +112,6 @@ def _prepare_data_table(
                 }
 
                 for profiler in profilers:
-                    # TODO: multiple patch cycles
                     predicted = compute_profiler_predictions(
                         profiler, project_name, case_study,
                         case_study.get_config_ids_for_revision(rev), patch_name
@@ -285,8 +287,6 @@ def tabulate(self, table_format: TableFormat, wrap_table: bool) -> str:
             "CaseStudy", "Profiler", "time", "memory", "overhead_time",
             "overhead_memory"
         ]]
-        # print(f"{overhead_df=}")
-        # TODO: double check and refactor
         overhead_df['overhead_time_rel'] = overhead_df['time'] / (
             overhead_df['time'] - overhead_df['overhead_time']
         ) * 100 - 100
@@ -354,7 +354,10 @@ def add_extras(doc: Document) -> None:
             ]
             style.format({col: "{:.0f}" for col in mv_columns}, precision=2)
 
-            ryg_map = cmap_map(lambda x: x / 1.2 + 0.2, plt.get_cmap('RdYlGn'))
+            ryg_map = cmap_map(
+                lambda x: x / 1.2 + 0.2,
+                tp.cast(colors.LinearSegmentedColormap, plt.get_cmap('RdYlGn'))
+            )
 
             style.background_gradient(
                 cmap=ryg_map,
@@ -462,9 +465,7 @@ def _calc_folder_locs_dune(repo_path: Path, rev_range: str) -> int:
 
         for sub_project in dune_sub_projects:
             sub_project_path = repo_path / sub_project
-            # TODO: get sub_rpoject hashes
             locs = calc_repo_loc(sub_project_path, "HEAD")
-            # print(f"Calculated {locs} for {sub_project_path}")
             total_locs += locs
 
         return total_locs