ensure non-dim coordinates are compressed

Author: tasansal

src/mdio/builder/templates/abstract_dataset_template.py

@@ -160,22 +160,17 @@ def _add_coordinates(self) -> None:
             )
 
         # Add non-dimension coordinates
-        # TODO(Dmitriy Repin): do chunked write for non-dimensional coordinates and trace_mask
-        # https://github.com/TGSAI/mdio-python/issues/587
-        # The chunk size used for trace mask will be different from the _var_chunk_shape
         for i in range(len(self._coord_names)):
             self._builder.add_coordinate(
                 self._coord_names[i],
                 dimensions=self._coord_dim_names,
                 data_type=ScalarType.FLOAT64,
+                compressor=compressors.Blosc(cname=compressors.BloscCname.zstd),
                 metadata=CoordinateMetadata(units_v1=self._horizontal_coord_unit),
             )
 
     def _add_trace_mask(self) -> None:
         """Add trace mask variables."""
-        # TODO(Dmitriy Repin): do chunked write for non-dimensional coordinates and trace_mask
-        # https://github.com/TGSAI/mdio-python/issues/587
-        # The chunk size used for trace mask will be different from the _var_chunk_shape
         self._builder.add_variable(
             name="trace_mask",
             dimensions=self._dim_names[:-1],  # All dimensions except vertical (the last one)

src/mdio/builder/templates/seismic_2d_prestack_shot.py

@@ -2,6 +2,7 @@
 
 from typing import Any
 
+from mdio.builder.schemas import compressors
 from mdio.builder.schemas.dtype import ScalarType
 from mdio.builder.schemas.v1.variable import CoordinateMetadata
 from mdio.builder.templates.abstract_dataset_template import AbstractDatasetTemplate
@@ -32,33 +33,39 @@ def _add_coordinates(self) -> None:
             self._builder.add_coordinate(name, dimensions=(name,), data_type=ScalarType.INT32)
 
         # Add non-dimension coordinates
+        compressor = compressors.Blosc(cname=compressors.BloscCname.zstd)
         coordinate_metadata = CoordinateMetadata(units_v1=self._horizontal_coord_unit)
         self._builder.add_coordinate(
             "gun",
             dimensions=("shot_point",),
             data_type=ScalarType.UINT8,
+            compressor=compressor,
         )
         self._builder.add_coordinate(
             "source_coord_x",
             dimensions=("shot_point",),
             data_type=ScalarType.FLOAT64,
+            compressor=compressor,
             metadata=coordinate_metadata,
         )
         self._builder.add_coordinate(
             "source_coord_y",
             dimensions=("shot_point",),
             data_type=ScalarType.FLOAT64,
+            compressor=compressor,
             metadata=coordinate_metadata,
         )
         self._builder.add_coordinate(
             "group_coord_x",
             dimensions=("shot_point", "channel"),
             data_type=ScalarType.FLOAT64,
+            compressor=compressor,
             metadata=coordinate_metadata,
         )
         self._builder.add_coordinate(
             "group_coord_y",
             dimensions=("shot_point", "channel"),
             data_type=ScalarType.FLOAT64,
+            compressor=compressor,
             metadata=coordinate_metadata,
         )

src/mdio/builder/templates/seismic_3d_prestack_coca.py

@@ -2,6 +2,7 @@
 
 from typing import Any
 
+from mdio.builder.schemas import compressors
 from mdio.builder.schemas.dtype import ScalarType
 from mdio.builder.schemas.v1.units import AngleUnitModel
 from mdio.builder.schemas.v1.variable import CoordinateMetadata
@@ -59,15 +60,18 @@ def _add_coordinates(self) -> None:
         )
 
         # Add non-dimension coordinates
+        compressor = compressors.Blosc(cname=compressors.BloscCname.zstd)
         self._builder.add_coordinate(
             "cdp_x",
             dimensions=("inline", "crossline"),
             data_type=ScalarType.FLOAT64,
+            compressor=compressor,
             metadata=CoordinateMetadata(units_v1=self._horizontal_coord_unit),
         )
         self._builder.add_coordinate(
             "cdp_y",
             dimensions=("inline", "crossline"),
             data_type=ScalarType.FLOAT64,
+            compressor=compressor,
             metadata=CoordinateMetadata(units_v1=self._horizontal_coord_unit),
         )

src/mdio/builder/templates/seismic_3d_prestack_shot.py

@@ -2,6 +2,7 @@
 
 from typing import Any
 
+from mdio.builder.schemas import compressors
 from mdio.builder.schemas.dtype import ScalarType
 from mdio.builder.schemas.v1.variable import CoordinateMetadata
 from mdio.builder.templates.abstract_dataset_template import AbstractDatasetTemplate
@@ -32,32 +33,38 @@ def _add_coordinates(self) -> None:
             self._builder.add_coordinate(name, dimensions=(name,), data_type=ScalarType.INT32)
 
         # Add non-dimension coordinates
+        compressor = compressors.Blosc(cname=compressors.BloscCname.zstd)
         self._builder.add_coordinate(
             "gun",
             dimensions=("shot_point",),
             data_type=ScalarType.UINT8,
+            compressor=compressor,
         )
         self._builder.add_coordinate(
             "source_coord_x",
             dimensions=("shot_point",),
             data_type=ScalarType.FLOAT64,
+            compressor=compressor,
             metadata=CoordinateMetadata(units_v1=self._horizontal_coord_unit),
         )
         self._builder.add_coordinate(
             "source_coord_y",
             dimensions=("shot_point",),
             data_type=ScalarType.FLOAT64,
+            compressor=compressor,
             metadata=CoordinateMetadata(units_v1=self._horizontal_coord_unit),
         )
         self._builder.add_coordinate(
             "group_coord_x",
             dimensions=("shot_point", "cable", "channel"),
             data_type=ScalarType.FLOAT64,
+            compressor=compressor,
             metadata=CoordinateMetadata(units_v1=self._horizontal_coord_unit),
         )
         self._builder.add_coordinate(
             "group_coord_y",
             dimensions=("shot_point", "cable", "channel"),
             data_type=ScalarType.FLOAT64,
+            compressor=compressor,
             metadata=CoordinateMetadata(units_v1=self._horizontal_coord_unit),
         )

src/mdio/core/utils_write.py

@@ -10,7 +10,7 @@
 
 
 MAX_SIZE_LIVE_MASK = 128 * 1024**2
-MAX_COORDINATES_BYTES = 16 * 1024**2
+MAX_COORDINATES_BYTES = 8 * 1024**2
 
 
 def get_constrained_chunksize(