stouputils 1.3.3__tar.gz → 1.3.4__tar.gz

{stouputils-1.3.3 → stouputils-1.3.4}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: stouputils
-Version: 1.3.3
+Version: 1.3.4
 Summary: Stouputils is a collection of utility modules designed to simplify and enhance the development process. It includes a range of tools for tasks such as execution of doctests, display utilities, decorators, as well as context managers, and many more.
 Project-URL: Homepage, https://github.com/Stoupy51/stouputils
 Project-URL: Issues, https://github.com/Stoupy51/stouputils/issues

{stouputils-1.3.3 → stouputils-1.3.4}/pyproject.toml

@@ -5,7 +5,7 @@ build-backend = "hatchling.build"
 
 [project]
 name = "stouputils"
-version = "1.3.3"
+version = "1.3.4"
 description = "Stouputils is a collection of utility modules designed to simplify and enhance the development process. It includes a range of tools for tasks such as execution of doctests, display utilities, decorators, as well as context managers, and many more."
 readme = "README.md"
 requires-python = ">=3.10"

{stouputils-1.3.3 → stouputils-1.3.4}/stouputils/data_science/models/keras_utils/visualizations.py

@@ -4,6 +4,7 @@
 # pyright: reportUnknownVariableType=false
 # pyright: reportUnknownArgumentType=false
 # pyright: reportMissingTypeStubs=false
+# pyright: reportIndexIssue=false
 
 # Imports
 import os
@@ -67,7 +68,7 @@ def make_gradcam_heatmap(
 			error("Last convolutional layer not found. Please provide the name of the last convolutional layer.")
 
 	# Get the last convolutional layer
-	last_layer: Model | list[Model] = model.get_layer(last_conv_layer_name)
+	last_layer: list[Model] | Any = model.get_layer(last_conv_layer_name)
 	if isinstance(last_layer, list):
 		last_layer = last_layer[0]
 
@@ -79,10 +80,11 @@ def make_gradcam_heatmap(
 
 	# Record operations for automatic differentiation using GradientTape.
 	with tf.GradientTape() as tape:
+
 		# Forward pass: get the activations of the last conv layer and the predictions.
 		conv_outputs: tf.Tensor
 		predictions: tf.Tensor
-		conv_outputs, predictions = grad_model(img)
+		conv_outputs, predictions = grad_model(img) # pyright: ignore [reportGeneralTypeIssues]
 		# print("conv_outputs shape:", conv_outputs.shape)
 		# print("predictions shape:", predictions.shape)
 
@@ -110,7 +112,7 @@ def make_gradcam_heatmap(
 			activation_map: tf.Tensor = conv_outputs[:, :, :, i]
 
 			# Weight the activation map by the gradient
-			weighted_map: tf.Tensor = activation_map * pooled_grad
+			weighted_map: tf.Tensor = activation_map * pooled_grad # pyright: ignore [reportOperatorIssue]
 
 			# Ensure that the heatmap has non-negative values and normalize it
 			heatmap: tf.Tensor = tf.maximum(weighted_map, 0) / tf.math.reduce_max(weighted_map)
@@ -176,9 +178,9 @@ def make_saliency_map(
 	# Record operations for automatic differentiation
 	with tf.GradientTape() as tape:
 		tape.watch(img_tensor)
-		predictions = model(img_tensor)
+		predictions: tf.Tensor = model(img_tensor) # pyright: ignore [reportAssignmentType]
 		# Use class index for positive class prediction
-		loss = predictions[:, class_idx]
+		loss: tf.Tensor = predictions[:, class_idx]
 
 	# Compute gradients of loss with respect to input image
 	try:
@@ -214,7 +216,7 @@ def make_saliency_map(
 			else:
 				channel_saliency: tf.Tensor = channel_grads
 
-			saliency_maps.append(channel_saliency.numpy().squeeze())
+			saliency_maps.append(channel_saliency.numpy().squeeze()) # type: ignore
 	else:
 		# Take absolute value of gradients
 		abs_grads = tf.abs(grads)
@@ -324,6 +326,10 @@ def create_visualization_overlay(
 	else:
 		original_pil: Image.Image = original_img
 
+	# Force RGB mode
+	if original_pil.mode != "RGB":
+		original_pil = original_pil.convert("RGB")
+
 	# Convert heatmap to PIL Image
 	heatmap_pil: Image.Image = Image.fromarray(heatmap_rgb)
 
@@ -331,12 +337,9 @@ def create_visualization_overlay(
 	if heatmap_pil.size != original_pil.size:
 		heatmap_pil = heatmap_pil.resize(original_pil.size, Image.Resampling.LANCZOS)
 
-	# Ensure both images have the same mode before blending
-	if original_pil.mode != heatmap_pil.mode:
-		if original_pil.mode == "RGB" and heatmap_pil.mode == "RGBA":
-			heatmap_pil = heatmap_pil.convert("RGB")
-		else:
-			original_pil = original_pil.convert(heatmap_pil.mode)
+	# Ensure heatmap is also in RGB mode
+	if heatmap_pil.mode != "RGB":
+		heatmap_pil = heatmap_pil.convert("RGB")
 
 	# Blend images
 	overlaid_img: Image.Image = Image.blend(original_pil, heatmap_pil, alpha=alpha)
@@ -389,7 +392,7 @@ def all_visualizations_for_image(
 	# Perform prediction to determine correctness
 	# Ensure img is in batch format (add dimension if needed)
 	img_batch: NDArray[Any] = np.expand_dims(img, axis=0) if img.ndim == 3 else img
-	predicted_class_idx: int = int(np.argmax(model.predict(img_batch, verbose=0)[0])) # pyright: ignore [reportArgumentType]
+	predicted_class_idx: int = int(np.argmax(model.predict(img_batch, verbose=0)[0])) # type: ignore
 	prediction_correct: bool = (predicted_class_idx == class_idx)
 	status_suffix: str = "_correct" if prediction_correct else "_missed"
 

{stouputils-1.3.3 → stouputils-1.3.4}/stouputils/data_science/scripts/exhaustive_process.py

@@ -118,8 +118,13 @@ def exhaustive_process(
 					commands.append(f"{base_cmd} {kfold_arg}")
 
 	# Run all commands
+	def runner(cmd: str) -> None:
+		info(f"Executing command: '{cmd}'")
+		sys.stdout.flush()
+		sys.stderr.flush()
+		os.system(cmd)
 	multithreading(
-		os.system,
+		runner,
 		commands,
 		desc="Processing all datasets",
 		max_workers=args.max_workers,

{stouputils-1.3.3 → stouputils-1.3.4}/stouputils/io.py

@@ -169,25 +169,56 @@ def super_open(file_path: str, mode: str, encoding: str = "utf-8") -> IO[Any]:
 
 
 # For easy file copy
-def super_copy(src: str, dst: str, create_dir: bool = True) -> str:
+def super_copy(src: str, dst: str, create_dir: bool = True, symlink: bool = False) -> str:
 	""" Copy a file (or a folder) from the source to the destination
 
 	Args:
 		src	(str): The source path
 		dst	(str): The destination path
 		create_dir (bool): Whether to create the directory if it doesn't exist (default: True)
+		symlink (bool): Whether to create a symlink instead of copying (default: False)
 	Returns:
 		str: The destination path
 	"""
-	# Make directory
+	# Create destination directory if needed
 	if create_dir:
 		os.makedirs(os.path.dirname(dst), exist_ok=True)
 
-	# If source is a folder, copy it recursively
+	# Handle directory copying
 	if os.path.isdir(src):
-		return shutil.copytree(src, dst, dirs_exist_ok = True)
+		if symlink:
+
+			# Remove existing destination if it's different from source
+			if os.path.exists(dst):
+				if os.path.samefile(src, dst) is False:
+					if os.path.isdir(dst):
+						shutil.rmtree(dst)
+					else:
+						os.remove(dst)
+					return os.symlink(src, dst, target_is_directory=True) or dst
+			else:
+				return os.symlink(src, dst, target_is_directory=True) or dst
+
+		# Regular directory copy
+		else:
+			return shutil.copytree(src, dst, dirs_exist_ok = True)
+
+	# Handle file copying
 	else:
-		return shutil.copy(src, dst)
+		if symlink:
+
+			# Remove existing destination if it's different from source
+			if os.path.exists(dst):
+				if os.path.samefile(src, dst) is False:
+					os.remove(dst)
+					return os.symlink(src, dst, target_is_directory=False) or dst
+			else:
+				return os.symlink(src, dst, target_is_directory=False) or dst
+
+		# Regular file copy
+		else:
+			return shutil.copy(src, dst)
+	return ""
 
 
 
@@ -228,7 +259,12 @@ def super_json_dump(data: Any, file: IO[Any]|None = None, max_level: int = 2, in
 	# Dump content with 2-space indent and replace it with the desired indent
 	content: str = orjson.dumps(data, option=orjson.OPT_INDENT_2).decode("utf-8")
 	if indent != "  ":
-		content = content.replace("  ", indent)
+		content = re.sub(
+			pattern=r'^(\s{2})+',  # Match groups of 2 spaces at start of lines
+			repl=lambda match: indent * (len(match.group(0)) // 2),  # Convert to desired indent
+			string=content,
+			flags=re.MULTILINE
+		)
 
 	# Limit max depth of indentation
 	if max_level > -1: