Passing all tests

Author: BrianMichell

pyproject.toml

@@ -26,7 +26,7 @@ dependencies = [
     "psutil>=7.0.0",
     "pydantic>=2.11.9",
     "rich>=14.1.0",
-    "segy>=0.5.1",
+    "segy>=0.5.2",
     "tqdm>=4.67.1",
     "universal-pathlib>=0.2.6",
     "xarray>=2025.9.1",

src/mdio/builder/dataset_builder.py

@@ -125,17 +125,8 @@ def push_dimension(self, dimension: NamedDimension, position: int, new_dim_chunk
             msg = "New dimension chunk size must be greater than 0"
             raise ValueError(msg)
 
-        print(f"Setting position {position} to size {new_dim_size}")
-        print(f"Before: {self._dimensions}")
-
         # In-place insertion of the dimension to the existing list of dimensions
         self._dimensions.insert(position, dimension)
-        print(f"After: {self._dimensions}")
-
-        # print(f"Setting position {position} to size {new_dim_size}")
-        # print(self._dimensions)
-        # self._dimensions[position].size = new_dim_size
-        # print(self._dimensions)
 
         def propogate_dimension(variable: Variable, position: int, new_dim_chunk_size: int) -> Variable:
             """Propogates the dimension to the variable or coordinate."""
@@ -165,8 +156,6 @@ def propogate_dimension(variable: Variable, position: int, new_dim_chunk_size: i
         to_ignore = []
         for v in self._dimensions:
             to_ignore.append(v.name)
-        for c in self._coordinates:
-            to_ignore.append(c.name)
 
         for i in range(len(self._variables)):
             var = self._variables[i]

src/mdio/builder/templates/abstract_dataset_template.py

@@ -68,7 +68,6 @@ def build_dataset(
             Dataset: The constructed dataset
         """
         self._dim_sizes = sizes
-        print(f"Sizes: {self._dim_sizes}")
         self._horizontal_coord_unit = horizontal_coord_unit
 
         attributes = self._load_dataset_attributes() or {}
@@ -161,6 +160,10 @@ def _add_dimensions(self) -> None:
         for i in range(len(self._dim_names)):
             self._builder.add_dimension(self._dim_names[i], self._dim_sizes[i])
 
+        for i in range(len(self._dim_sizes) - len(self._dim_names)):
+            self._builder.add_dimension(f"dim_{i}", self._dim_sizes[len(self._dim_names) + i])
+
+
     def _add_coordinates(self) -> None:
         """Add custom coordinates.
 

src/mdio/converters/segy.py

@@ -262,8 +262,9 @@ def populate_dim_coordinates(
 ) -> tuple[xr_Dataset, list[str]]:
     """Populate the xarray dataset with dimension coordinate variables."""
     for dim in grid.dims:
-        dataset[dim.name].values[:] = dim.coords
-        drop_vars_delayed.append(dim.name)
+        if dim.name in dataset.dims:
+            dataset[dim.name].values[:] = dim.coords
+            drop_vars_delayed.append(dim.name)
     return dataset, drop_vars_delayed
 
 
@@ -508,9 +509,18 @@ def segy_to_mdio(  # noqa PLR0913
 
     print(grid)
 
+    # When HasDuplicates is True, the grid includes a "trace" dimension that will be
+    # added to the dataset via a queued transform. We need to exclude it from the sizes
+    # passed to build_dataset to avoid dimension size mismatches.
+    dataset_sizes = grid.shape
+    if "trace" in grid.dim_names:
+        # Find the trace dimension index and remove it from sizes
+        trace_idx = grid.dim_names.index("trace")
+        dataset_sizes = grid.shape[:trace_idx] + grid.shape[trace_idx + 1:]
+
     mdio_ds: Dataset = mdio_template.build_dataset(
         name=mdio_template.name,
-        sizes=grid.shape,
+        sizes=dataset_sizes,
         horizontal_coord_unit=_get_horizontal_coordinate_unit(segy_dimensions),
         header_dtype=to_structured_type(segy_spec.trace.header.dtype),
     )