Satisfy pre-commit

Author: BrianMichell

src/mdio/builder/dataset_builder.py

@@ -7,6 +7,8 @@
 from typing import Any
 
 from mdio import __version__
+from mdio.builder.schemas.chunk_grid import RegularChunkGrid
+from mdio.builder.schemas.chunk_grid import RegularChunkShape
 from mdio.builder.schemas.compressors import ZFP
 from mdio.builder.schemas.compressors import Blosc
 from mdio.builder.schemas.dimension import NamedDimension
@@ -102,17 +104,22 @@ def add_dimension(self, name: str, size: int) -> "MDIODatasetBuilder":
         return self
 
     def push_dimension(
-        self, dimension: NamedDimension, position: int, new_dim_chunk_size: int = 1, new_dim_size: int = 1
+        self, dimension: NamedDimension, position: int, new_dim_chunk_size: int = 1
     ) -> "MDIODatasetBuilder":
         """Pushes a dimension to all Coordiantes and Variables.
+
         The position argument is the domain index of the dimension to push.
-        If a Variable is within the position domain, it will be inserted at the position and all remaining dimensions will be shifted to the right.
+        If a Variable is within the position domain, it will be inserted at the position
+        and all remaining dimensions will be shifted to the right.
 
         Args:
             dimension: The dimension to push
             position: The position to push the dimension to
             new_dim_chunk_size: The chunk size for only the new dimension
 
+        Raises:
+            ValueError: If the position is invalid
+
         Returns:
             self: Returns self for method chaining
         """
@@ -123,7 +130,6 @@ def push_dimension(
             msg = "Position is greater than the number of dimensions"
             raise ValueError(msg)
         if new_dim_chunk_size <= 0:
-            # TODO(BrianMichell): Do we actually need to check this, or does Pydantic handle when we call?
             msg = "New dimension chunk size must be greater than 0"
             raise ValueError(msg)
 
@@ -132,9 +138,6 @@ def push_dimension(
 
         def propogate_dimension(variable: Variable, position: int, new_dim_chunk_size: int) -> Variable:
             """Propogates the dimension to the variable or coordinate."""
-            from mdio.builder.schemas.chunk_grid import RegularChunkGrid
-            from mdio.builder.schemas.chunk_grid import RegularChunkShape
-
             if len(variable.dimensions) + 1 <= position:
                 # Don't do anything if the new dimension is not within the Variable's domain
                 return variable
@@ -154,12 +157,11 @@ def propogate_dimension(variable: Variable, position: int, new_dim_chunk_size: i
             # Update metadata with new chunk grid
             new_metadata = variable.metadata.model_copy() if variable.metadata else VariableMetadata()
             new_metadata.chunk_grid = new_chunk_grid
-            ret = variable.model_copy(update={"dimensions": new_dimensions, "metadata": new_metadata})
-            return ret
+            return variable.model_copy(update={"dimensions": new_dimensions, "metadata": new_metadata})
 
         to_ignore = []
         for v in self._dimensions:
-            to_ignore.append(v.name)
+            to_ignore.append(v.name)  # noqa: PERF401
 
         for i in range(len(self._variables)):
             var = self._variables[i]

src/mdio/builder/templates/abstract_dataset_template.py

@@ -81,9 +81,9 @@ def build_dataset(
         if header_dtype:
             self._add_trace_headers(header_dtype)
 
-        # This seems to be breaking the dataset, but adds the trace dimension.
         for transform in self._queued_transforms:
-            logger.debug(f"Applying transform: {transform.__name__}")
+            stmnt = f"Applying transform: {transform.__name__} to dataset"
+            logger.debug(stmnt)
             transform(self._builder)
         self._queued_transforms = []
         return self._builder.build()

src/mdio/segy/geometry.py

@@ -258,6 +258,7 @@ def analyze_non_indexed_headers(
     Args:
         index_headers: numpy array with index headers
         dtype: numpy type for value of created trace header.
+        index_names: list of index names to analyze
 
     Returns:
         Dict container header name as key and numpy array of values as value
@@ -438,7 +439,7 @@ def transform(
         self,
         index_headers: HeaderArray,
         grid_overrides: dict[str, bool | int],
-        index_names=None,
+        index_names: Sequence[str] = None,  # noqa: ARG002
     ) -> NDArray:
         """Perform the grid transform."""
         self.validate(index_headers, grid_overrides)
@@ -476,7 +477,7 @@ def transform(
         self,
         index_headers: HeaderArray,
         grid_overrides: dict[str, bool | int],
-        index_names=None,
+        index_names: Sequence[str] = None,  # noqa: ARG002
     ) -> NDArray:
         """Perform the grid transform."""
         self.validate(index_headers, grid_overrides)

src/mdio/segy/utilities.py

@@ -11,6 +11,7 @@
 from dask.array.core import normalize_chunks
 
 from mdio.builder.schemas.dimension import NamedDimension
+from mdio.builder.schemas.dtype import ScalarType
 from mdio.core import Dimension
 from mdio.segy.geometry import GridOverrider
 from mdio.segy.parsers import parse_headers
@@ -20,6 +21,7 @@
     from segy import SegyFile
     from segy.arrays import HeaderArray
 
+    from mdio.builder.dataset_builder import MDIODatasetBuilder
     from mdio.builder.templates.abstract_dataset_template import AbstractDatasetTemplate
 
 
@@ -41,15 +43,14 @@ def _create_delayed_trace_dimension_transform(headers_subset: HeaderArray, posit
         A callable that can be used as a transform function
     """
 
-    def delayed_transform(builder):
-        from mdio.builder.schemas.dtype import ScalarType
-
+    def delayed_transform(builder: MDIODatasetBuilder) -> MDIODatasetBuilder:
         # Calculate the trace dimension size at execution time
         if "trace" in headers_subset.dtype.names:
             trace_size = int(np.max(headers_subset["trace"]))
         else:
             # Fallback: if trace field doesn't exist, we need to determine size differently
-            raise ValueError("Trace field not found in headers_subset when executing delayed transform")
+            msg = "Trace field not found in headers_subset when executing delayed transform"
+            raise ValueError(msg)
 
         # Add the trace dimension
         trace_dimension = NamedDimension(name="trace", size=trace_size)
@@ -110,8 +111,7 @@ def get_grid_plan(  # noqa:  C901
     )
 
     if grid_overrides.get("HasDuplicates", False):
-        pos = len(template.dimension_names) - 1  # TODO: Implement the negative position case...
-        # Use the delayed transform function instead of a simple lambda
+        pos = len(template.dimension_names) - 1
         template._queue_transform(_create_delayed_trace_dimension_transform(headers_subset, pos))
         horizontal_dimensions = (*horizontal_dimensions, "trace")