Understand why last dimension has incorrect size

Author: BrianMichell

src/mdio/builder/dataset_builder.py

@@ -128,6 +128,9 @@ def push_dimension(self, dimension: NamedDimension, position: int, new_dim_chunk
         # In-place insertion of the dimension to the existing list of dimensions
         self._dimensions.insert(position, dimension)
 
+        # TODO: HARDCODED FOR DEBUGGING
+        self._dimensions[-1].size = 25
+
         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, RegularChunkShape

src/mdio/segy/geometry.py

@@ -247,7 +247,7 @@ def create_trace_index(
     return index_headers
 
 
-def analyze_non_indexed_headers(index_headers: HeaderArray, dtype: DTypeLike = np.int16) -> NDArray:
+def analyze_non_indexed_headers(index_headers: HeaderArray, dtype: DTypeLike = np.int16, index_names: Sequence[str] = None) -> NDArray:
     """Check input headers for SEG-Y input to help determine geometry.
 
     This function reads in trace_qc_count headers and finds the unique cable values. Then, it
@@ -266,7 +266,7 @@ def analyze_non_indexed_headers(index_headers: HeaderArray, dtype: DTypeLike = n
     total_depth = 0
     header_names = []
     for header_key in index_headers.dtype.names:
-        if header_key != "trace":
+        if header_key != "trace" and header_key in index_names:
             unique_headers[header_key] = np.sort(np.unique(index_headers[header_key]))
             header_names.append(header_key)
             total_depth += 1
@@ -302,6 +302,7 @@ def transform(
         self,
         index_headers: HeaderArray,
         grid_overrides: dict[str, bool | int],
+        index_names: Sequence[str] = None,
     ) -> NDArray:
         """Perform the grid transform."""
 
@@ -378,11 +379,12 @@ def transform(
         self,
         index_headers: HeaderArray,
         grid_overrides: dict[str, bool | int],
+        index_names: Sequence[str],
     ) -> NDArray:
         """Perform the grid transform."""
         self.validate(index_headers, grid_overrides)
 
-        return analyze_non_indexed_headers(index_headers)
+        return analyze_non_indexed_headers(index_headers, index_names=index_names)
 
     def transform_index_names(self, index_names: Sequence[str]) -> Sequence[str]:
         """Insert dimension "trace" to the sample-1 dimension."""
@@ -434,6 +436,7 @@ def transform(
         self,
         index_headers: HeaderArray,
         grid_overrides: dict[str, bool | int],
+        index_names = None,
     ) -> NDArray:
         """Perform the grid transform."""
         self.validate(index_headers, grid_overrides)
@@ -471,6 +474,7 @@ def transform(
         self,
         index_headers: HeaderArray,
         grid_overrides: dict[str, bool | int],
+        index_names = None,
     ) -> NDArray:
         """Perform the grid transform."""
         self.validate(index_headers, grid_overrides)
@@ -535,11 +539,6 @@ def run(
     ) -> tuple[HeaderArray, tuple[str], tuple[int]]:
         """Run grid overrides and return result."""
 
-        # print("="*100)
-        # print(index_headers.to_dict().keys())
-        # print(index_headers)
-        # print("="*100)
-
         for override in grid_overrides:
             if override in self.parameters:
                 continue
@@ -548,7 +547,7 @@ def run(
                 raise GridOverrideUnknownError(override)
 
             function = self.commands[override].transform
-            index_headers = function(index_headers, grid_overrides=grid_overrides)
+            index_headers = function(index_headers, grid_overrides=grid_overrides, index_names=index_names)
 
             function = self.commands[override].transform_index_names
             index_names = function(index_names)

src/mdio/segy/utilities.py

@@ -58,29 +58,19 @@ def get_grid_plan(  # noqa:  C901
     horizontal_dimensions = template.dimension_names[:-1]
     horizontal_coordinates = horizontal_dimensions + template.coordinate_names
     headers_subset = parse_headers(segy_file=segy_file, subset=horizontal_coordinates)
-    from segy.arrays import HeaderArray
-
-    # reduced_headers_subset = HeaderArray(h for h in headers_subset.to_dict().keys() if h in horizontal_dimensions)
-    # Get field names to keep
-    fields_to_keep = [h for h in headers_subset.dtype.names if h in horizontal_dimensions]
-
-    # Create filtered copy using numpy's field selection
-    reduced_headers_subset = HeaderArray(headers_subset[fields_to_keep])
 
     # Handle grid overrides.
     override_handler = GridOverrider()
     headers_subset, horizontal_coordinates, chunksize = override_handler.run(
-        # headers_subset,
-        reduced_headers_subset,
-        # horizontal_coordinates,
+        headers_subset,
         horizontal_dimensions,
         chunksize=chunksize,
         grid_overrides=grid_overrides,
     )
 
     if grid_overrides.get("HasDuplicates", False):
         pos = len(template.dimension_names) - 1  # TODO: Implement the negative position case...
-        template._queue_transform(lambda builder: builder.push_dimension(NamedDimension(name="trace", size=headers_subset["trace"].size), position=pos, new_dim_chunk_size=1))
+        template._queue_transform(lambda builder: builder.push_dimension(NamedDimension(name="trace", size=np.max(headers_subset["trace"])), position=pos, new_dim_chunk_size=1))
         horizontal_dimensions = (*horizontal_dimensions, "trace")
 
     dimensions = []