fix: pillow vulnerability and other + formatting (#89)

## 📥 Pull Request Description

Fix vulnerabilities and run some pre-commit hooks for reformatting
etc...

## 📝 Checklist

Please make sure you've completed the following tasks before submitting
this pull request:

- [x] Pre-commit hooks were executed (not all green)
- [x] Changes have been reviewed by at least one other developer
- [ ] Tests have been added or updated to cover the changes (only
necessary if the changes affect the executable code)
- [x] All tests ran successfully
- [x] All merge conflicts are resolved
- [ ] Documentation has been updated to reflect the changes
- [ ] Any necessary migrations have been run

---------

Co-authored-by: Nils <[email protected]>

configs/ops/image_to_tabular_data/op_image_to_tabular_data.yaml

@@ -5,4 +5,4 @@ target_image_size:
   _target_: niceml.utilities.imagesize.ImageSize
   width: 10
   height: 10
-use_dirs_as_subsets: True
+use_dirs_as_subsets: True

docs/hydra-dagster.md

@@ -137,4 +137,4 @@ configured you can find either in the source code or at our ops reference page.
 ### Conclusion
 
 In this tutorial, we've seen how to use Hydra with Dagster to configure a machine learning pipeline. Hydra allows for a hierarchical set of configuration files that can be used to define settings for different parts of the pipeline. 
-The `defaults` section in the YAML file allows us to import the content of other YAML files, making it easy to reuse common settings across different parts of the pipeline.
+The `defaults` section in the YAML file allows us to import the content of other YAML files, making it easy to reuse common settings across different parts of the pipeline.

niceml/dagster/ops/splitdata.py

@@ -44,7 +44,6 @@
     }
 )
 def split_data(context: OpExecutionContext, input_location: dict):
-
     """Splits the data in input_location into subsets (set_infos)"""
     op_config = json.loads(json.dumps(context.op_config))
 

niceml/dashboard/cam.py

@@ -12,7 +12,9 @@
     load_img,
 )
 
-from niceml.experiments.expdatalocalstorageloader import create_expdata_from_local_storage
+from niceml.experiments.expdatalocalstorageloader import (
+    create_expdata_from_local_storage,
+)
 from niceml.experiments.experimentdata import ExperimentData
 
 

niceml/dashboard/imagenetdataloggerviscomponent.py

@@ -56,7 +56,6 @@ def _render(
         def _load_net_data(
             *args,  # pylint: disable = unused-argument
         ) -> List[np.ndarray]:
-
             return [
                 self.image_loader(filepath=net_data_path)
                 for net_data_path in net_data_paths[: self.max_output]

niceml/data/netdataloggers/netdatalogger.py

@@ -17,6 +17,7 @@ class NetDataLogger(ABC):
     """Abstract implementation of an NetDataLogger"""
 
     def __init__(self):
+        """Initializes the NetDataLogger with default values"""
         self.data_description = None
         self.exp_context = None
         self.set_name = None
@@ -28,7 +29,6 @@ def initialize(
         exp_context: ExperimentContext,
         set_name: str,
     ):
-
         """Method to initialize the NetDataLogger"""
         self.data_description = data_description
         self.exp_context = exp_context

niceml/dlframeworks/tensorflow/losses/categoricalfocalloss.py

@@ -1,3 +1,4 @@
+"""module for categorical focal loss"""
 from typing import List, Union
 
 import numpy as np
@@ -7,6 +8,8 @@
 
 @tf.keras.utils.register_keras_serializable()
 class CategoricalFocalLoss(tf.keras.losses.Loss):
+    """class for categorical focal loss"""
+
     def __init__(self, alpha: Union[float, List[float]], gamma: float = 2.0, **kwargs):
         """
         Inspired by https://github.com/umbertogriffo/focal-loss-keras
@@ -28,7 +31,7 @@ def __init__(self, alpha: Union[float, List[float]], gamma: float = 2.0, **kwarg
         self.gamma = gamma
 
     def __call__(self, y_true, y_pred, sample_weight=None):
-
+        """Call method for loss"""
         # Clip the prediction value to prevent NaN's and Inf's
         epsilon = kb.epsilon()
         y_pred = kb.clip(y_pred, epsilon, 1.0 - epsilon)

niceml/dlframeworks/tensorflow/metrics/objdetmetrics.py

@@ -13,10 +13,11 @@ class AvgPosPredObjDet:  # pylint: disable=too-few-public-methods
     """Positive Classification Values for object detection"""
 
     def __init__(self, name: str = "avg_pos_pred"):
-
+        """Initializes the AvgPosPredObjDet with the given name"""
         self.__name__ = name
 
     def __call__(self, y_true, y_pred):
+        """Call method is used as a default interface for the metric"""
         y_pred = tf.cast(y_pred, dtype=tf.float32)
 
         cls_predictions = y_pred[:, :, 4:]
@@ -35,10 +36,11 @@ class AvgNegPredObjDet:  # pylint: disable=too-few-public-methods
     """Negative Classification Values for object detection"""
 
     def __init__(self, name: str = "avg_neg_pred"):
-
+        """Initializes the AvgNegPredObjDet with the given name"""
         self.__name__ = name
 
     def __call__(self, y_true, y_pred):
+        """Call method is used as a default interface for the metric"""
         y_pred = tf.cast(y_pred, dtype=tf.float32)
 
         cls_predictions = y_pred[:, :, 4:]
@@ -67,9 +69,11 @@ class AvgPosTargetCountObjDet:  # pylint: disable=too-few-public-methods
     """Average positive target count for one image in object detection"""
 
     def __init__(self, name: str = "avg_pos_target_count"):
+        """Initializes the AvgPosTargetCountObjDet"""
         self.__name__ = name
 
     def __call__(self, y_true, y_pred):
+        """Call method is used as a default interface for the metric"""
         positive_mask = tf.cast(
             tf.equal(y_true[:, :, 4], POSITIVE_MASK_VALUE), dtype=tf.float32
         )
@@ -82,9 +86,11 @@ class AvgNegTargetCountObjDet:  # pylint: disable=too-few-public-methods
     """Average negative target count for one image in object detection"""
 
     def __init__(self, name: str = "avg_neg_target_count"):
+        """Initializes the AvgNegTargetCountObjDet"""
         self.__name__ = name
 
     def __call__(self, y_true, y_pred):
+        """Call method is used as a default interface for the metric"""
         negative_mask = tf.cast(
             tf.equal(y_true[:, :, 4], NEGATIVE_MASK_VALUE), dtype=tf.float32
         )

niceml/dlframeworks/tensorflow/models/mlp.py

@@ -4,7 +4,9 @@
 from tensorflow.keras import Sequential, layers  # pylint: disable=import-error
 
 from niceml.data.datadescriptions.datadescription import DataDescription
-from niceml.data.datadescriptions.inputdatadescriptions import InputVectorDataDescription
+from niceml.data.datadescriptions.inputdatadescriptions import (
+    InputVectorDataDescription,
+)
 from niceml.data.datadescriptions.outputdatadescriptions import (
     OutputVectorDataDescription,
 )
@@ -22,12 +24,14 @@ def __init__(
         final_activation: str = "linear",
         do_summary: bool = True,
     ):
+        """Initializes the OwnMLP model factory"""
         self.hidden_layers = hidden_layers
         self.activation = activation
         self.do_summary = do_summary
         self.final_activation = final_activation
 
     def create_model(self, data_description: DataDescription) -> Any:
+        """Creates the mlp model"""
         input_dd: InputVectorDataDescription = check_instance(
             data_description, InputVectorDataDescription
         )

niceml/experiments/experimenttests/validateexps.py

@@ -7,7 +7,9 @@
 
 from niceml.data.storages.fsfilesystemstorage import FsFileSystemStorage
 from niceml.data.storages.fsspecstorage import FSSpecStorage
-from niceml.experiments.expdatalocalstorageloader import create_expdata_from_local_storage
+from niceml.experiments.expdatalocalstorageloader import (
+    create_expdata_from_local_storage,
+)
 from niceml.experiments.experimenterrors import EmptyExperimentError
 from niceml.experiments.experimenttests.exptests import (
     ExperimentTest,

niceml/experiments/expoutinitializer.py

@@ -35,7 +35,6 @@ def __init__(
         self.exp_type: str = exp_type or ""
 
     def __call__(self, exp_context: ExperimentContext):
-
         produce_git_version_yaml(
             exp_context,
             ExperimentFilenames.GIT_VERSIONS,

niceml/mlcomponents/objdet/anchorencoding.py

@@ -53,13 +53,11 @@ def encode_anchors(  # pylint: disable=too-many-locals
         encoded_feature_list: List[List[float]] = []
 
         for anchor in anchor_list:
-
             if len(gt_labels) == 0:
                 target_bbox = anchor
                 target_label = None
                 prediction_flag = NEGATIVE_MASK_VALUE
             else:
-
                 max_iou = 0
                 target_bbox = gt_labels[0].bounding_box
                 target_label = None

niceml/mlcomponents/predictionhandlers/objdetpredictionhandler.py

@@ -29,15 +29,15 @@
 class ObjDetPredictionHandler(PredictionHandler):
     """Prediction handler for object detection predictions (BoundingBox, class prediction)"""
 
-    def __init__(  # pylint: disable=too-many-arguments
+    def __init__(  # noqa: PLR0913
         self,
         prediction_filter: PredictionFilter,
         prediction_prefix: str = "pred",
         pred_identifier: str = "image_location",
         detection_idx_col: str = DETECTION_INDEX_COLUMN_NAME,
         apply_sigmoid: bool = True,
     ):
-
+        """Initializes the ObjDetPredictionHandler"""
         super().__init__()
         self.prediction_filter = prediction_filter
         self.prediction_prefix = prediction_prefix
@@ -53,6 +53,7 @@ def __init__(  # pylint: disable=too-many-arguments
         self.anchor_array = None
 
     def initialize(self):
+        """Initializes the prediction handler"""
         self.anchors: List[BoundingBox] = self.anchor_generator.generate_anchors(
             data_description=self.data_description
         )

niceml/mlcomponents/predictionhandlers/tensorpredictionhandler.py

@@ -74,7 +74,6 @@ def __exit__(self, exc_type, exc_value, exc_traceback):
                     ),
                     "wb",
                 ) as file:
-
                     np.savez_compressed(file, **self.data)
 
 

niceml/mlcomponents/resultanalyzers/instancefinders/maskinstance.py

@@ -5,7 +5,9 @@
 
 import numpy as np
 
-from niceml.data.datadescriptions.outputdatadescriptions import OutputImageDataDescription
+from niceml.data.datadescriptions.outputdatadescriptions import (
+    OutputImageDataDescription,
+)
 from niceml.mlcomponents.resultanalyzers.instancefinders.instancecontour import (
     InstanceContour,
 )

niceml/mlcomponents/resultanalyzers/instancefinders/multichannelinstancefinder.py

@@ -11,7 +11,9 @@
 from niceml.mlcomponents.resultanalyzers.instancefinders.instancefinder import (
     InstanceFinder,
 )
-from niceml.mlcomponents.resultanalyzers.instancefinders.maskinstance import MaskInstance
+from niceml.mlcomponents.resultanalyzers.instancefinders.maskinstance import (
+    MaskInstance,
+)
 from niceml.mlcomponents.resultanalyzers.tensors.semsegdataiterator import (
     SemSegPredictionContainer,
 )

niceml/mlcomponents/targettransformer/objdettargettransformer.py

@@ -27,7 +27,6 @@ def __init__(
         self.anchors = None
 
     def get_net_targets(self, data_list: List[ObjDetData]) -> np.ndarray:
-
         if self.anchors is None:
             if isinstance(self.data_description, OutputObjDetDataDescription):
                 self.anchors = self.anchor_generator.generate_anchors(

niceml/mlcomponents/targettransformer/semsegtargettransformer.py

@@ -4,7 +4,9 @@
 import numpy as np
 
 from niceml.data.datadescriptions.inputdatadescriptions import InputImageDataDescription
-from niceml.data.datadescriptions.outputdatadescriptions import OutputImageDataDescription
+from niceml.data.datadescriptions.outputdatadescriptions import (
+    OutputImageDataDescription,
+)
 from niceml.data.datainfos.semsegdatainfo import SemSegData
 from niceml.mlcomponents.targettransformer.targettransformer import NetTargetTransformer
 from niceml.utilities.commonutils import check_instance

niceml/utilities/boundingboxes/bboxdrawing.py

@@ -42,7 +42,9 @@ def draw_bounding_box_on_image(
     font = get_font("OpenSans-Regular.ttf", font_size=font_size)
 
     text = f"{label.class_name}: {label.score:.2f}" if label.score else label.class_name
-    text_width, text_height = font.getsize(text)
+    (left, top, right, bottom) = font.getbbox(text)
+    text_width = right - left
+    text_height = bottom - top
     x_1, y_1, x_2, y_2 = label.bounding_box.get_absolute_ullr()
     draw.rectangle(
         (x_1 - line_width, y_1 - line_width, x_2 + line_width, y_2 + line_width),
@@ -69,7 +71,7 @@ def draw_bounding_box_on_image(
     return image
 
 
-def draw_labels_on_image(  # pylint: disable=too-many-arguments
+def draw_labels_on_image(  # noqa: PLR0913
     image: ImageType,
     pred_bbox_label_list: List[ObjDetInstanceLabel],
     gt_bbox_label_list: List[ObjDetInstanceLabel],

niceml/utilities/boundingboxes/bboxencoding.py

@@ -5,7 +5,6 @@
 def encode_boxes(
     anchor_boxes_xywh: np.ndarray, gt_boxes_xywh: np.ndarray, box_variances: np.ndarray
 ) -> np.ndarray:
-
     """
     Encodes the anchor boxes to a numpy array
 
@@ -48,7 +47,6 @@ def decode_boxes(
     encoded_array_xywh: np.ndarray,
     box_variances: np.ndarray,
 ) -> np.ndarray:
-
     """
     Decodes the incoming array to target boxes
 

niceml/utilities/boundingboxes/bboxlabeling.py

@@ -4,7 +4,10 @@
 from attr import define, field
 
 from niceml.utilities.boundingboxes.bboxconversion import dict_to_bounding_box
-from niceml.utilities.boundingboxes.boundingbox import BoundingBox, bounding_box_from_ullr
+from niceml.utilities.boundingboxes.boundingbox import (
+    BoundingBox,
+    bounding_box_from_ullr,
+)
 from niceml.utilities.imagesize import ImageSize
 from niceml.utilities.instancelabeling import InstanceLabel
 

niceml/utilities/boundingboxes/filtering/nmsfilter.py

@@ -4,7 +4,10 @@
 import numpy as np
 from attrs import define
 
-from niceml.utilities.boundingboxes.bboxconversion import convert_to_ullr, convert_to_xywh
+from niceml.utilities.boundingboxes.bboxconversion import (
+    convert_to_ullr,
+    convert_to_xywh,
+)
 from niceml.utilities.boundingboxes.filtering.predictionfilter import PredictionFilter
 from niceml.utilities.ioumatrix import compute_iou_matrix
 

niceml/utilities/boundingboxes/filtering/unifiedboxfilter.py

@@ -4,7 +4,10 @@
 import numpy as np
 from attrs import define
 
-from niceml.utilities.boundingboxes.bboxconversion import convert_to_ullr, convert_to_xywh
+from niceml.utilities.boundingboxes.bboxconversion import (
+    convert_to_ullr,
+    convert_to_xywh,
+)
 from niceml.utilities.boundingboxes.filtering.predictionfilter import PredictionFilter
 from niceml.utilities.ioumatrix import compute_iou_matrix
 

niceml/utilities/commonutils.py

@@ -97,12 +97,12 @@ def human_readable_size(obj: Any) -> str:
     size = asizeof.asizeof(obj)
     if size < 1024:
         return f"{size} bytes"
-    if 1024 <= size < 1024 ** 2:
+    if 1024 <= size < 1024**2:
         size_kb = size / 1024
         return f"{size_kb:.2f} KB"
-    if 1024 ** 2 <= size < 1024 ** 3:
-        size_mb = size / 1024 ** 2
+    if 1024**2 <= size < 1024**3:
+        size_mb = size / 1024**2
         return f"{size_mb:.2f} MB"
 
-    size_gb = size / 1024 ** 3
+    size_gb = size / 1024**3
     return f"{size_gb:.2f} GB"

niceml/utilities/imageutils.py

@@ -121,7 +121,12 @@ def binarize_multichannel_image(
             mask including information where the prediction is above threshold
     """
     binary_multichannel_images: Dict[str, np.ndarray] = {}
-    scores_mask = cv2.threshold(image_scores, threshold, 1, cv2.THRESH_BINARY,)[
+    scores_mask = cv2.threshold(
+        image_scores,
+        threshold,
+        1,
+        cv2.THRESH_BINARY,
+    )[
         1
     ].astype(np.uint8)
 

niceml/utilities/instancelabelmatching.py

@@ -45,7 +45,6 @@ def get_kind_of_label_match(
                 pred_label.color = Color.BLUE
                 pred_label.active = True
             for gt_label in gt_label_list:
-
                 iou = pred_label.calc_iou(other=gt_label)
 
                 pred_label, gt_label = get_kind_of_instance_label_match(

niceml/utilities/splitutils.py

@@ -114,7 +114,7 @@ def create_copy_files_container(
                 identifier = "".join(
                     [char for char in cur_basename if char in ALPHANUMERICLIST]
                 )
-                cur_seed = int(identifier, base=len(ALPHANUMERICLIST)) % (2 ** 32 - 1)
+                cur_seed = int(identifier, base=len(ALPHANUMERICLIST)) % (2**32 - 1)
                 rng = np.random.default_rng(seed=cur_seed)
                 drawn_set = rng.choice(set_list, 1, p=prob_list)[0]
                 output_file = join(drawn_set, file)