Merge pull request #88 from discovery-unicamp/merge-from-dev

Merge from development - 12/04/2025

Author: GabrielBG0

.github/workflows/continuous-testing.yml

@@ -4,26 +4,26 @@ on:
   push:
     branches: ["main"]
     paths-ignore:
-      - "**/README.md"
-      - "**/CONTRIBUTING.md"
-      - "**/CODE_OF_CONDUCT.md"
-      - "**/SECURITY.md"
-      - "**/.github/ISSUE_TEMPLATE/*"
-      - "**/.github/PULL_REQUEST_TEMPLATE/*"
-      - "**/LICENSE"
+      - '**/README.md'
+      - '**/CONTRIBUTING.md'
+      - '**/CODE_OF_CONDUCT.md'
+      - '**/SECURITY.md'
+      - '**/.github/ISSUE_TEMPLATE/*'
+      - '**/.github/PULL_REQUEST_TEMPLATE/*'
+      - '**/LICENSE'
   pull_request:
     branches: ["main"]
     paths-ignore:
-      - "**/README.md"
-      - "**/CONTRIBUTING.md"
-      - "**/CODE_OF_CONDUCT.md"
-      - "**/SECURITY.md"
-      - "**/.github/ISSUE_TEMPLATE/*"
-      - "**/.github/PULL_REQUEST_TEMPLATE/*"
-      - "**/LICENSE"
+      - '**/README.md'
+      - '**/CONTRIBUTING.md'
+      - '**/CODE_OF_CONDUCT.md'
+      - '**/SECURITY.md'
+      - '**/.github/ISSUE_TEMPLATE/*'
+      - '**/.github/PULL_REQUEST_TEMPLATE/*'
+      - '**/LICENSE'
 
 permissions:
-  contents: read # No write permission needed
+  contents: read  # No write permission needed
 
 jobs:
   continuous-testing:
@@ -37,7 +37,7 @@ jobs:
         uses: actions/setup-python@v5
         with:
           python-version: "3.10"
-          cache: "pip" # Cache pip packages to improve workflow runtime
+          cache: "pip"  # Cache pip packages to improve workflow runtime
 
       - name: Install Dependencies
         run: |
@@ -56,18 +56,19 @@ jobs:
         uses: actions/upload-artifact@v4
         with:
           name: html-coverage-report
-          path: htmlcov/ # Upload the generated HTML report directory
-
+          path: htmlcov/  # Upload the generated HTML report directory
+      
       - name: Creating interrogate folder
         run: |
           mkdir interrogate
 
+
       - name: Docstring Coverage
         run: interrogate minerva -vv --fail-under=80 --generate-badge interrogate/interrogate_badge.svg --badge-format svg -o interrogate/simple-report.md
         continue-on-error: true
-
+      
       - name: Upload Docstring Coverage Report
         uses: actions/upload-artifact@v4
         with:
           name: docstring-coverage-files
-          path: interrogate/ # Upload the generated docstring coverage report
+          path: interrogate/  # Upload the generated docstring coverage report

.github/workflows/format-check.yaml

@@ -0,0 +1,35 @@
+name: Black Format Check
+
+on:
+  push:
+    branches: ["main"]
+    paths:
+      - '**.py'
+
+  pull_request:
+    branches: ["main"]
+    paths:
+      - '**.py'
+
+jobs:
+  black-check:
+    runs-on: ubuntu-latest
+    steps:
+      - name: Checkout code
+        uses: actions/checkout@v3
+
+      - name: Set up Python
+        uses: actions/setup-python@v4
+        with:
+          python-version: '3.10'
+
+      - name: Install dependencies
+        run: |
+          pip install black nbqa
+
+      - name: Check Python file formatting
+        run: |
+          echo "🔍 Checking Python files..."
+          black --check --diff . || (
+            echo "❌ Black found formatting issues in Python files." && exit 1
+          )

minerva/analysis/complexity_performance_analysis.py

@@ -17,7 +17,9 @@ class ComplexityPerformanceAnalysis(_ModelAnalysis):
     If necessary, random data can be employed.
     """
 
-    def __init__(self, path: Optional[PathLike] = None, custom_input_size: Optional[Tuple] = None):
+    def __init__(
+        self, path: Optional[PathLike] = None, custom_input_size: Optional[Tuple] = None
+    ):
         super().__init__()
         self._path = path
         self._custom_input_size = custom_input_size
@@ -41,7 +43,9 @@ def compute(self, model: L.LightningModule, data: L.LightningDataModule):
             evaluation_data = torch.tensor(evaluation_data, dtype=torch.float32)
         # Computing MACs, parameters, and energy consumption
         macs, params = profile(model, inputs=(evaluation_data,))
-        carbonTracker = EmissionsTracker(project_name="basic_measurement", measure_power_secs=10, save_to_file=False)
+        carbonTracker = EmissionsTracker(
+            project_name="basic_measurement", measure_power_secs=10, save_to_file=False
+        )
         try:
             carbonTracker.start_task("measure_inference")
             _ = model(evaluation_data)

minerva/analysis/metrics/__init__.py

@@ -1,4 +1,4 @@
 from .pixel_accuracy import PixelAccuracy
 from .balanced_accuracy import BalancedAccuracy
 
-__all__ = ["PixelAccuracy","BalancedAccuracy"]
+__all__ = ["PixelAccuracy", "BalancedAccuracy"]

minerva/analysis/metrics/balanced_accuracy.py

@@ -4,53 +4,54 @@
 from torchmetrics.functional import confusion_matrix
 import warnings
 
+
 class BalancedAccuracy(Metric):
-    def __init__(
-            self,
-            num_classes: int,
-            task: str,
-            adjusted:
-            bool = False
-    ):
+    def __init__(self, num_classes: int, task: str, adjusted: bool = False):
         """
-    Compute the balanced accuracy.
+        Compute the balanced accuracy.
+
+        The balanced accuracy in binary, multiclass, and multilabel classification problems
+        deals with imbalanced datasets. It is defined as the average of recall obtained on each class.
+
+        Parameters
+        ----------
+        num_classes : int
+            The number of classes in the target data.
 
-    The balanced accuracy in binary, multiclass, and multilabel classification problems
-    deals with imbalanced datasets. It is defined as the average of recall obtained on each class.
+        task : str
+            The type of classification task, should be one of 'binary' or 'multiclass'
 
-    Parameters
-    ----------
-    num_classes : int
-        The number of classes in the target data.
-    
-    task : str
-        The type of classification task, should be one of 'binary' or 'multiclass'
-    
-    adjusted : bool, optional (default=False)
-        When true, the result is adjusted for chance, so that random performance would score 0,
-        while keeping perfect performance at a score of 1.
+        adjusted : bool, optional (default=False)
+            When true, the result is adjusted for chance, so that random performance would score 0,
+            while keeping perfect performance at a score of 1.
 
-    Attributes
-    ----------
-    confmat : torch.Tensor
-        Confusion matrix to keep track of true positives, false positives, true negatives, and false negatives.
+        Attributes
+        ----------
+        confmat : torch.Tensor
+            Confusion matrix to keep track of true positives, false positives, true negatives, and false negatives.
 
-    Examples
-    --------
-    >>> y_true = torch.tensor([0, 1, 0, 0, 1, 0])
-    >>> y_pred = torch.tensor([0, 1, 0, 0, 0, 1])
-    >>> metric = BalancedAccuracy(num_classes=2, task='binary')
-    >>> metric(y_pred, y_true)
-    0.625
+        Examples
+        --------
+        >>> y_true = torch.tensor([0, 1, 0, 0, 1, 0])
+        >>> y_pred = torch.tensor([0, 1, 0, 0, 0, 1])
+        >>> metric = BalancedAccuracy(num_classes=2, task='binary')
+        >>> metric(y_pred, y_true)
+        0.625
         """
         super().__init__()
         self.num_classes = num_classes
         self.adjusted = adjusted
         self.task = task
-        self.add_state("confmat", default=torch.zeros((num_classes, num_classes)), dist_reduce_fx="sum")
+        self.add_state(
+            "confmat",
+            default=torch.zeros((num_classes, num_classes)),
+            dist_reduce_fx="sum",
+        )
 
     def update(self, preds: torch.Tensor, target: torch.Tensor):
-        self.confmat += confusion_matrix(preds, target, num_classes=self.num_classes, task=self.task)
+        self.confmat += confusion_matrix(
+            preds, target, num_classes=self.num_classes, task=self.task
+        )
 
     def compute(self):
         with torch.no_grad():

minerva/callback/embedding_logger_callback.py

@@ -7,16 +7,16 @@
 class EmbeddingLoggerCallback(Callback):
 
     def __init__(
-            self,
-            data_X: Tensor,
-            logger: CSVLogger,
-            data_Y: Tensor = None,
-            feature_preffix: str = 'EMB-',
-            backbone_names_list: List[str] = ['backbone', 'encoder'],
-            ) -> None:
-        '''
+        self,
+        data_X: Tensor,
+        logger: CSVLogger,
+        data_Y: Tensor = None,
+        feature_preffix: str = "EMB-",
+        backbone_names_list: List[str] = ["backbone", "encoder"],
+    ) -> None:
+        """
         Callback to extract and log embeddings from some data using the model's backbone.
-        
+
         Parameters
         ----------
         data_X : torch.Tensor
@@ -29,29 +29,34 @@ def __init__(
             The preffix to use for the feature names, by default 'EMB-'.
         backbone_names_list : List[str], optional
             List with the names of the backbones in the model, by default ['backbone', 'encoder'].
-        '''
+        """
         super().__init__()
         self.data_X = data_X
         self.data_Y = data_Y
         self.logger = logger
         self.feature_preffix = feature_preffix
         self.backbone_names_list = backbone_names_list
-    
+
     def on_train_start(self, trainer: Trainer, pl_module: LightningModule) -> None:
-        filtered_names_list = [name for name in self.backbone_names_list if hasattr(pl_module, name)]
+        filtered_names_list = [
+            name for name in self.backbone_names_list if hasattr(pl_module, name)
+        ]
         if len(filtered_names_list) == 0:
-            raise ValueError('No backbone found in the model')
+            raise ValueError("No backbone found in the model")
         self.backbone = getattr(pl_module, filtered_names_list[0])
-        
+
     def on_train_epoch_end(self, trainer: Trainer, pl_module: LightningModule) -> None:
         # Obtaining the embeddings
         self.backbone.eval()
         embeddings = self.backbone(self.data_X).detach().cpu().numpy()
         self.backbone.train()
         # Logging the embeddings
         for row_index, row in enumerate(embeddings):
-            data_dict = {f'{self.feature_preffix}{str(index).zfill(3)}': value for index, value in enumerate(row)}
-            data_dict['epoch'] = trainer.current_epoch
+            data_dict = {
+                f"{self.feature_preffix}{str(index).zfill(3)}": value
+                for index, value in enumerate(row)
+            }
+            data_dict["epoch"] = trainer.current_epoch
             if self.data_Y is not None:
-                data_dict['y'] = self.data_Y[row_index].item()
-            self.logger.log_metrics(data_dict)
+                data_dict["y"] = self.data_Y[row_index].item()
+            self.logger.log_metrics(data_dict)

minerva/data/data_module_tools.py

@@ -63,7 +63,6 @@ def __init__(
         else:
             delattr(self, "test_dataloader")
 
-
     def _generate_data(self, num_samples, data_shape, label_shape, num_classes):
         data = torch.rand((num_samples, *data_shape), dtype=self.data_dtype)
         label = None
@@ -73,9 +72,9 @@ def _generate_data(self, num_samples, data_shape, label_shape, num_classes):
             label = torch.rand((num_samples, *label_shape))
         elif num_classes is not None:
             label = torch.randint(0, num_classes, (num_samples,))
-            
+
         label = label.to(dtype=self.label_dtype)
-        
+
         return data, label
 
     def setup(self, stage):

minerva/data/data_modules/base.py

@@ -66,7 +66,7 @@ class when creating the dataloaders, by default DataLoader
             Name of the data module, by default ""
         """
         super().__init__()
-        
+
         self._name = name
         self._train_dataset = train_dataset
         self._val_dataset = val_dataset
@@ -124,11 +124,11 @@ class when creating the dataloaders, by default DataLoader
         # Monkey patch the dataloaders if the datasets are not provided
         # It is applyed at instance level to avoid breaking the class signature
         if not self._train_dataset:
-            self.train_dataloader = None    # type: ignore
+            self.train_dataloader = None  # type: ignore
         if not self._val_dataset:
-            self.val_dataloader = None      # type: ignore
+            self.val_dataloader = None  # type: ignore
         if not self._test_dataset:
-            self.test_dataloader = None     # type: ignore
+            self.test_dataloader = None  # type: ignore
         if not self._predict_dataset:
             self.predict_dataloader = None  # type: ignore
 
@@ -168,19 +168,13 @@ def predict_dataset(self):
         return self._predict_dataset
 
     def train_dataloader(self):
-        return self._dataloader_cls(
-            self.train_dataset, **self._train_dataloader_kwargs
-        )
+        return self._dataloader_cls(self.train_dataset, **self._train_dataloader_kwargs)
 
     def val_dataloader(self):
-        return self._dataloader_cls(
-            self.val_dataset, **self._val_dataloader_kwargs
-        )
+        return self._dataloader_cls(self.val_dataset, **self._val_dataloader_kwargs)
 
     def test_dataloader(self):
-        return self._dataloader_cls(
-            self.test_dataset, **self._test_dataloader_kwargs
-        )
+        return self._dataloader_cls(self.test_dataset, **self._test_dataloader_kwargs)
 
     def predict_dataloader(self):
         return self._dataloader_cls(
@@ -194,7 +188,12 @@ def indent_text(text, spaces=6, add_line_breaks=True):
                 return "No data."
 
             return "\n".join(
-                (" " * int(spaces//2) + "│" if add_line_breaks else " " * int(spaces//2) + " ") + (" " * spaces + line if line.strip() else line)
+                (
+                    " " * int(spaces // 2) + "│"
+                    if add_line_breaks
+                    else " " * int(spaces // 2) + " "
+                )
+                + (" " * spaces + line if line.strip() else line)
                 for line in text.split("\n")
             )
 

minerva/data/datasets/base.py

@@ -171,7 +171,9 @@ def __getitem__(self, idx: int) -> Union[Any, Tuple[Any, ...]]:
 
     def __str__(self) -> str:
         readers = self.readers if isinstance(self.readers, list) else [self.readers]
-        transforms = self.transforms if isinstance(self.transforms, list) else [self.transforms]
+        transforms = (
+            self.transforms if isinstance(self.transforms, list) else [self.transforms]
+        )
 
         readers_info = "\n".join(
             [
@@ -192,4 +194,4 @@ def __str__(self) -> str:
         )
 
     def __repr__(self) -> str:
-        return self.__str__()
+        return self.__str__()