cudf-polars-cu12 25.6.0__py3-none-any.whl → 25.8.0__py3-none-any.whl

cudf_polars/VERSION

@@ -1 +1 @@
-25.06.00
+25.08.00

cudf_polars/callback.py

@@ -7,6 +7,7 @@ from __future__ import annotations
 
 import contextlib
 import os
+import textwrap
 import time
 import warnings
 from functools import cache, partial
@@ -21,7 +22,9 @@ import pylibcudf
 import rmm
 from rmm._cuda import gpu
 
+from cudf_polars.dsl.tracing import CUDF_POLARS_NVTX_DOMAIN
 from cudf_polars.dsl.translate import Translator
+from cudf_polars.utils.config import _env_get_int, get_total_device_memory
 from cudf_polars.utils.timer import Timer
 
 if TYPE_CHECKING:
@@ -45,13 +48,6 @@ _SUPPORTED_PREFETCHES = {
 }
 
 
-def _env_get_int(name: str, default: int) -> int:
-    try:
-        return int(os.getenv(name, default))
-    except (ValueError, TypeError):  # pragma: no cover
-        return default  # pragma: no cover
-
-
 @cache
 def default_memory_resource(
     device: int,
@@ -102,8 +98,7 @@ def default_memory_resource(
         ):
             raise ComputeError(
                 "GPU engine requested, but incorrect cudf-polars package installed. "
-                "If your system has a CUDA 11 driver, please uninstall `cudf-polars-cu12` "
-                "and install `cudf-polars-cu11`"
+                "cudf-polars requires CUDA 12.0+ to installed."
             ) from None
         else:
             raise
@@ -140,7 +135,11 @@ def set_memory_resource(
         mr = default_memory_resource(
             device=device,
             cuda_managed_memory=bool(
-                _env_get_int("POLARS_GPU_ENABLE_CUDA_MANAGED_MEMORY", default=1) != 0
+                _env_get_int(
+                    "POLARS_GPU_ENABLE_CUDA_MANAGED_MEMORY",
+                    default=1 if get_total_device_memory() is not None else 0,
+                )
+                != 0
             ),
         )
     rmm.mr.set_current_device_resource(mr)
@@ -222,7 +221,7 @@ def _callback(
     if timer is not None:
         assert should_time
     with (
-        nvtx.annotate(message="ExecuteIR", domain="cudf_polars"),
+        nvtx.annotate(message="ExecuteIR", domain=CUDF_POLARS_NVTX_DOMAIN),
         # Device must be set before memory resource is obtained.
         set_device(config_options.device),
         set_memory_resource(memory_resource),
@@ -236,6 +235,16 @@ def _callback(
         elif config_options.executor.name == "streaming":
             from cudf_polars.experimental.parallel import evaluate_streaming
 
+            if timer is not None:
+                msg = textwrap.dedent("""\
+                    LazyFrame.profile() is not supported with the streaming executor.
+                    To profile execution with the streaming executor, use:
+
+                    - NVIDIA NSight Systems with the 'streaming' scheduler.
+                    - Dask's built-in profiling tools with the 'distributed' scheduler.
+                    """)
+                raise NotImplementedError(msg)
+
             return evaluate_streaming(ir, config_options).to_polars()
         assert_never(f"Unknown executor '{config_options.executor}'")
 
@@ -277,7 +286,7 @@ def execute_with_cudf(
 
     memory_resource = config.memory_resource
 
-    with nvtx.annotate(message="ConvertIR", domain="cudf_polars"):
+    with nvtx.annotate(message="ConvertIR", domain=CUDF_POLARS_NVTX_DOMAIN):
         translator = Translator(nt, config)
         ir = translator.translate_ir()
         ir_translation_errors = translator.errors

cudf_polars/containers/__init__.py

@@ -1,11 +1,13 @@
-# SPDX-FileCopyrightText: Copyright (c) 2024 NVIDIA CORPORATION & AFFILIATES.
+# SPDX-FileCopyrightText: Copyright (c) 2024-2025, NVIDIA CORPORATION & AFFILIATES.
 # SPDX-License-Identifier: Apache-2.0
 
 """Containers of concrete data."""
 
 from __future__ import annotations
 
-__all__: list[str] = ["Column", "DataFrame"]
+__all__: list[str] = ["Column", "DataFrame", "DataType"]
 
+# dataframe.py & column.py imports DataType, so import in this order to avoid circular import
+from cudf_polars.containers.datatype import DataType  # noqa: I001
 from cudf_polars.containers.column import Column
 from cudf_polars.containers.dataframe import DataFrame

cudf_polars/containers/column.py

@@ -8,6 +8,8 @@ from __future__ import annotations
 import functools
 from typing import TYPE_CHECKING
 
+import polars as pl
+import polars.datatypes.convert
 from polars.exceptions import InvalidOperationError
 
 import pylibcudf as plc
@@ -19,19 +21,39 @@ from pylibcudf.strings.convert.convert_integers import (
 )
 from pylibcudf.traits import is_floating_point
 
+from cudf_polars.containers import DataType
 from cudf_polars.utils import conversion
 from cudf_polars.utils.dtypes import is_order_preserving_cast
 
 if TYPE_CHECKING:
     from typing_extensions import Self
 
-    import polars as pl
-
-    from cudf_polars.typing import ColumnHeader, ColumnOptions, Slice
+    from cudf_polars.typing import (
+        ColumnHeader,
+        ColumnOptions,
+        DeserializedColumnOptions,
+        Slice,
+    )
 
 __all__: list[str] = ["Column"]
 
 
+def _dtype_short_repr_to_dtype(dtype_str: str) -> pl.DataType:
+    """Convert a Polars dtype short repr to a Polars dtype."""
+    # limitations of dtype_short_repr_to_dtype described in
+    # py-polars/polars/datatypes/convert.py#L299
+    if dtype_str.startswith("list["):
+        stripped = dtype_str.removeprefix("list[").removesuffix("]")
+        return pl.List(_dtype_short_repr_to_dtype(stripped))
+    pl_type = polars.datatypes.convert.dtype_short_repr_to_dtype(dtype_str)
+    if pl_type is None:
+        raise ValueError(f"{dtype_str} was not able to be parsed by Polars.")
+    if isinstance(pl_type, polars.datatypes.DataTypeClass):
+        return pl_type()
+    else:
+        return pl_type
+
+
 class Column:
     """An immutable column with sortedness metadata."""
 
@@ -43,10 +65,12 @@ class Column:
     # Optional name, only ever set by evaluation of NamedExpr nodes
     # The internal evaluation should not care about the name.
     name: str | None
+    dtype: DataType
 
     def __init__(
         self,
         column: plc.Column,
+        dtype: DataType,
         *,
         is_sorted: plc.types.Sorted = plc.types.Sorted.NO,
         order: plc.types.Order = plc.types.Order.ASCENDING,
@@ -56,6 +80,7 @@ class Column:
         self.obj = column
         self.is_scalar = self.size == 1
         self.name = name
+        self.dtype = dtype
         self.set_sorted(is_sorted=is_sorted, order=order, null_order=null_order)
 
     @classmethod
@@ -81,7 +106,23 @@ class Column:
         (plc_column,) = plc.contiguous_split.unpack_from_memoryviews(
             packed_metadata, packed_gpu_data
         ).columns()
-        return cls(plc_column, **header["column_kwargs"])
+        return cls(plc_column, **cls.deserialize_ctor_kwargs(header["column_kwargs"]))
+
+    @staticmethod
+    def deserialize_ctor_kwargs(
+        column_kwargs: ColumnOptions,
+    ) -> DeserializedColumnOptions:
+        """Deserialize the constructor kwargs for a Column."""
+        dtype = DataType(  # pragma: no cover
+            _dtype_short_repr_to_dtype(column_kwargs["dtype"])
+        )
+        return {
+            "is_sorted": column_kwargs["is_sorted"],
+            "order": column_kwargs["order"],
+            "null_order": column_kwargs["null_order"],
+            "name": column_kwargs["name"],
+            "dtype": dtype,
+        }
 
     def serialize(
         self,
@@ -105,17 +146,21 @@ class Column:
             Two-tuple of frames suitable for passing to `plc.contiguous_split.unpack_from_memoryviews`
         """
         packed = plc.contiguous_split.pack(plc.Table([self.obj]))
-        column_kwargs: ColumnOptions = {
+        header: ColumnHeader = {
+            "column_kwargs": self.serialize_ctor_kwargs(),
+            "frame_count": 2,
+        }
+        return header, packed.release()
+
+    def serialize_ctor_kwargs(self) -> ColumnOptions:
+        """Serialize the constructor kwargs for self."""
+        return {
             "is_sorted": self.is_sorted,
             "order": self.order,
             "null_order": self.null_order,
             "name": self.name,
+            "dtype": pl.polars.dtype_str_repr(self.dtype.polars),
         }
-        header: ColumnHeader = {
-            "column_kwargs": column_kwargs,
-            "frame_count": 2,
-        }
-        return header, packed.release()
 
     @functools.cached_property
     def obj_scalar(self) -> plc.Scalar:
@@ -172,6 +217,7 @@ class Column:
         return type(self)(
             self.obj,
             name=self.name,
+            dtype=self.dtype,
             is_sorted=like.is_sorted,
             order=like.order,
             null_order=like.null_order,
@@ -202,11 +248,11 @@ class Column:
         If the sortedness flag is not set, this launches a kernel to
         check sortedness.
         """
-        if self.obj.size() <= 1 or self.obj.size() == self.obj.null_count():
+        if self.size <= 1 or self.size == self.null_count:
             return True
         if self.is_sorted == plc.types.Sorted.YES:
             return self.order == order and (
-                self.obj.null_count() == 0 or self.null_order == null_order
+                self.null_count == 0 or self.null_order == null_order
             )
         if plc.sorting.is_sorted(plc.Table([self.obj]), [order], [null_order]):
             self.sorted = plc.types.Sorted.YES
@@ -215,7 +261,7 @@ class Column:
             return True
         return False
 
-    def astype(self, dtype: plc.DataType) -> Column:
+    def astype(self, dtype: DataType) -> Column:
         """
         Cast the column to as the requested dtype.
 
@@ -238,14 +284,18 @@ class Column:
         This only produces a copy if the requested dtype doesn't match
         the current one.
         """
-        if self.obj.type() == dtype:
+        plc_dtype = dtype.plc
+        if self.obj.type() == plc_dtype:
             return self
 
-        if dtype.id() == plc.TypeId.STRING or self.obj.type().id() == plc.TypeId.STRING:
-            return Column(self._handle_string_cast(dtype))
+        if (
+            plc_dtype.id() == plc.TypeId.STRING
+            or self.obj.type().id() == plc.TypeId.STRING
+        ):
+            return Column(self._handle_string_cast(plc_dtype), dtype=dtype)
         else:
-            result = Column(plc.unary.cast(self.obj, dtype))
-            if is_order_preserving_cast(self.obj.type(), dtype):
+            result = Column(plc.unary.cast(self.obj, plc_dtype), dtype=dtype)
+            if is_order_preserving_cast(self.obj.type(), plc_dtype):
                 return result.sorted_like(self)
             return result
 
@@ -258,24 +308,20 @@ class Column:
         else:
             if is_floating_point(dtype):
                 floats = is_float(self.obj)
-                if not plc.interop.to_arrow(
-                    plc.reduce.reduce(
-                        floats,
-                        plc.aggregation.all(),
-                        plc.DataType(plc.TypeId.BOOL8),
-                    )
-                ).as_py():
+                if not plc.reduce.reduce(
+                    floats,
+                    plc.aggregation.all(),
+                    plc.DataType(plc.TypeId.BOOL8),
+                ).to_py():
                     raise InvalidOperationError("Conversion from `str` failed.")
                 return to_floats(self.obj, dtype)
             else:
                 integers = is_integer(self.obj)
-                if not plc.interop.to_arrow(
-                    plc.reduce.reduce(
-                        integers,
-                        plc.aggregation.all(),
-                        plc.DataType(plc.TypeId.BOOL8),
-                    )
-                ).as_py():
+                if not plc.reduce.reduce(
+                    integers,
+                    plc.aggregation.all(),
+                    plc.DataType(plc.TypeId.BOOL8),
+                ).to_py():
                     raise InvalidOperationError("Conversion from `str` failed.")
                 return to_integers(self.obj, dtype)
 
@@ -361,6 +407,7 @@ class Column:
             order=self.order,
             null_order=self.null_order,
             name=self.name,
+            dtype=self.dtype,
         )
 
     def mask_nans(self) -> Self:
@@ -368,7 +415,7 @@ class Column:
         if plc.traits.is_floating_point(self.obj.type()):
             old_count = self.null_count
             mask, new_count = plc.transform.nans_to_nulls(self.obj)
-            result = type(self)(self.obj.with_mask(mask, new_count))
+            result = type(self)(self.obj.with_mask(mask, new_count), self.dtype)
             if old_count == new_count:
                 return result.sorted_like(self)
             return result
@@ -377,14 +424,12 @@ class Column:
     @functools.cached_property
     def nan_count(self) -> int:
         """Return the number of NaN values in the column."""
-        if plc.traits.is_floating_point(self.obj.type()):
-            return plc.interop.to_arrow(
-                plc.reduce.reduce(
-                    plc.unary.is_nan(self.obj),
-                    plc.aggregation.sum(),
-                    plc.types.SIZE_TYPE,
-                )
-            ).as_py()
+        if self.size > 0 and plc.traits.is_floating_point(self.obj.type()):
+            return plc.reduce.reduce(
+                plc.unary.is_nan(self.obj),
+                plc.aggregation.sum(),
+                plc.types.SIZE_TYPE,
+            ).to_py()
         return 0
 
     @property
@@ -418,4 +463,4 @@ class Column:
             conversion.from_polars_slice(zlice, num_rows=self.size),
         )
         (column,) = table.columns()
-        return type(self)(column, name=self.name).sorted_like(self)
+        return type(self)(column, name=self.name, dtype=self.dtype).sorted_like(self)

cudf_polars/containers/dataframe.py

@@ -12,20 +12,51 @@ import polars as pl
 
 import pylibcudf as plc
 
-from cudf_polars.containers import Column
+from cudf_polars.containers import Column, DataType
 from cudf_polars.utils import conversion
 
 if TYPE_CHECKING:
     from collections.abc import Iterable, Mapping, Sequence, Set
 
-    from typing_extensions import Any, Self
+    from typing_extensions import Any, CapsuleType, Self
 
-    from cudf_polars.typing import ColumnOptions, DataFrameHeader, Slice
+    from cudf_polars.typing import ColumnOptions, DataFrameHeader, PolarsDataType, Slice
 
 
 __all__: list[str] = ["DataFrame"]
 
 
+def _create_polars_column_metadata(
+    name: str, dtype: PolarsDataType
+) -> plc.interop.ColumnMetadata:
+    """Create ColumnMetadata preserving pl.Struct field names."""
+    if isinstance(dtype, pl.Struct):
+        children_meta = [
+            _create_polars_column_metadata(field.name, field.dtype)
+            for field in dtype.fields
+        ]
+    else:
+        children_meta = []
+    timezone = dtype.time_zone if isinstance(dtype, pl.Datetime) else None
+    return plc.interop.ColumnMetadata(
+        name=name, timezone=timezone or "", children_meta=children_meta
+    )
+
+
+# This is also defined in pylibcudf.interop
+class _ObjectWithArrowMetadata:
+    def __init__(
+        self, obj: plc.Table, metadata: list[plc.interop.ColumnMetadata]
+    ) -> None:
+        self.obj = obj
+        self.metadata = metadata
+
+    def __arrow_c_array__(
+        self, requested_schema: None = None
+    ) -> tuple[CapsuleType, CapsuleType]:
+        return self.obj._to_schema(self.metadata), self.obj._to_host_array()
+
+
 # Pacify the type checker. DataFrame init asserts that all the columns
 # have a string name, so let's narrow the type.
 class NamedColumn(Column):
@@ -44,6 +75,7 @@ class DataFrame:
         if any(c.name is None for c in columns):
             raise ValueError("All columns must have a name")
         self.columns = [cast(NamedColumn, c) for c in columns]
+        self.dtypes = [c.dtype for c in self.columns]
         self.column_map = {c.name: c for c in self.columns}
         self.table = plc.Table([c.obj for c in self.columns])
 
@@ -60,11 +92,12 @@ class DataFrame:
         # To guarantee we produce correct names, we therefore
         # serialise with names we control and rename with that map.
         name_map = {f"column_{i}": name for i, name in enumerate(self.column_map)}
-        table = plc.interop.to_arrow(
-            self.table,
-            [plc.interop.ColumnMetadata(name=name) for name in name_map],
-        )
-        df: pl.DataFrame = pl.from_arrow(table)
+        metadata = [
+            _create_polars_column_metadata(name, dtype.polars)
+            for name, dtype in zip(name_map, self.dtypes, strict=True)
+        ]
+        table_with_metadata = _ObjectWithArrowMetadata(self.table, metadata)
+        df = pl.DataFrame(table_with_metadata)
         return df.rename(name_map).with_columns(
             pl.col(c.name).set_sorted(descending=c.order == plc.types.Order.DESCENDING)
             if c.is_sorted
@@ -106,16 +139,18 @@ class DataFrame:
         -------
         New dataframe representing the input.
         """
-        plc_table = plc.Table(df)
+        plc_table = plc.Table.from_arrow(df)
         return cls(
-            Column(d_col, name=name).copy_metadata(h_col)
+            Column(d_col, name=name, dtype=DataType(h_col.dtype)).copy_metadata(h_col)
             for d_col, h_col, name in zip(
                 plc_table.columns(), df.iter_columns(), df.columns, strict=True
             )
         )
 
     @classmethod
-    def from_table(cls, table: plc.Table, names: Sequence[str]) -> Self:
+    def from_table(
+        cls, table: plc.Table, names: Sequence[str], dtypes: Sequence[DataType]
+    ) -> Self:
         """
         Create from a pylibcudf table.
 
@@ -125,6 +160,8 @@ class DataFrame:
             Pylibcudf table to obtain columns from
         names
             Names for the columns
+        dtypes
+            Dtypes for the columns
 
         Returns
         -------
@@ -139,7 +176,8 @@ class DataFrame:
         if table.num_columns() != len(names):
             raise ValueError("Mismatching name and table length.")
         return cls(
-            Column(c, name=name) for c, name in zip(table.columns(), names, strict=True)
+            Column(c, name=name, dtype=dtype)
+            for c, name, dtype in zip(table.columns(), names, dtypes, strict=True)
         )
 
     @classmethod
@@ -166,7 +204,7 @@ class DataFrame:
             packed_metadata, packed_gpu_data
         )
         return cls(
-            Column(c, **kw)
+            Column(c, **Column.deserialize_ctor_kwargs(kw))
             for c, kw in zip(table.columns(), header["columns_kwargs"], strict=True)
         )
 
@@ -195,13 +233,7 @@ class DataFrame:
 
         # Keyword arguments for `Column.__init__`.
         columns_kwargs: list[ColumnOptions] = [
-            {
-                "is_sorted": col.is_sorted,
-                "order": col.order,
-                "null_order": col.null_order,
-                "name": col.name,
-            }
-            for col in self.columns
+            col.serialize_ctor_kwargs() for col in self.columns
         ]
         header: DataFrameHeader = {
             "columns_kwargs": columns_kwargs,
@@ -288,7 +320,11 @@ class DataFrame:
     def filter(self, mask: Column) -> Self:
         """Return a filtered table given a mask."""
         table = plc.stream_compaction.apply_boolean_mask(self.table, mask.obj)
-        return type(self).from_table(table, self.column_names).sorted_like(self)
+        return (
+            type(self)
+            .from_table(table, self.column_names, self.dtypes)
+            .sorted_like(self)
+        )
 
     def slice(self, zlice: Slice | None) -> Self:
         """
@@ -309,4 +345,8 @@ class DataFrame:
         (table,) = plc.copying.slice(
             self.table, conversion.from_polars_slice(zlice, num_rows=self.num_rows)
         )
-        return type(self).from_table(table, self.column_names).sorted_like(self)
+        return (
+            type(self)
+            .from_table(table, self.column_names, self.dtypes)
+            .sorted_like(self)
+        )

cudf_polars/containers/datatype.py

@@ -0,0 +1,135 @@
+# SPDX-FileCopyrightText: Copyright (c) 2024-2025, NVIDIA CORPORATION & AFFILIATES.
+# SPDX-License-Identifier: Apache-2.0
+
+"""A datatype, preserving polars metadata."""
+
+from __future__ import annotations
+
+from functools import cache
+
+from typing_extensions import assert_never
+
+import polars as pl
+
+import pylibcudf as plc
+
+__all__ = ["DataType"]
+
+
+@cache
+def _from_polars(dtype: pl.DataType) -> plc.DataType:
+    """
+    Convert a polars datatype to a pylibcudf one.
+
+    Parameters
+    ----------
+    dtype
+        Polars dtype to convert
+
+    Returns
+    -------
+    Matching pylibcudf DataType object.
+
+    Raises
+    ------
+    NotImplementedError
+        For unsupported conversions.
+    """
+    if isinstance(dtype, pl.Boolean):
+        return plc.DataType(plc.TypeId.BOOL8)
+    elif isinstance(dtype, pl.Int8):
+        return plc.DataType(plc.TypeId.INT8)
+    elif isinstance(dtype, pl.Int16):
+        return plc.DataType(plc.TypeId.INT16)
+    elif isinstance(dtype, pl.Int32):
+        return plc.DataType(plc.TypeId.INT32)
+    elif isinstance(dtype, pl.Int64):
+        return plc.DataType(plc.TypeId.INT64)
+    if isinstance(dtype, pl.UInt8):
+        return plc.DataType(plc.TypeId.UINT8)
+    elif isinstance(dtype, pl.UInt16):
+        return plc.DataType(plc.TypeId.UINT16)
+    elif isinstance(dtype, pl.UInt32):
+        return plc.DataType(plc.TypeId.UINT32)
+    elif isinstance(dtype, pl.UInt64):
+        return plc.DataType(plc.TypeId.UINT64)
+    elif isinstance(dtype, pl.Float32):
+        return plc.DataType(plc.TypeId.FLOAT32)
+    elif isinstance(dtype, pl.Float64):
+        return plc.DataType(plc.TypeId.FLOAT64)
+    elif isinstance(dtype, pl.Date):
+        return plc.DataType(plc.TypeId.TIMESTAMP_DAYS)
+    elif isinstance(dtype, pl.Time):
+        raise NotImplementedError("Time of day dtype not implemented")
+    elif isinstance(dtype, pl.Datetime):
+        if dtype.time_unit == "ms":
+            return plc.DataType(plc.TypeId.TIMESTAMP_MILLISECONDS)
+        elif dtype.time_unit == "us":
+            return plc.DataType(plc.TypeId.TIMESTAMP_MICROSECONDS)
+        elif dtype.time_unit == "ns":
+            return plc.DataType(plc.TypeId.TIMESTAMP_NANOSECONDS)
+        assert dtype.time_unit is not None  # pragma: no cover
+        assert_never(dtype.time_unit)
+    elif isinstance(dtype, pl.Duration):
+        if dtype.time_unit == "ms":
+            return plc.DataType(plc.TypeId.DURATION_MILLISECONDS)
+        elif dtype.time_unit == "us":
+            return plc.DataType(plc.TypeId.DURATION_MICROSECONDS)
+        elif dtype.time_unit == "ns":
+            return plc.DataType(plc.TypeId.DURATION_NANOSECONDS)
+        assert dtype.time_unit is not None  # pragma: no cover
+        assert_never(dtype.time_unit)
+    elif isinstance(dtype, pl.String):
+        return plc.DataType(plc.TypeId.STRING)
+    elif isinstance(dtype, pl.Null):
+        # TODO: Hopefully
+        return plc.DataType(plc.TypeId.EMPTY)
+    elif isinstance(dtype, pl.List):
+        # Recurse to catch unsupported inner types
+        _ = _from_polars(dtype.inner)
+        return plc.DataType(plc.TypeId.LIST)
+    elif isinstance(dtype, pl.Struct):
+        # Recurse to catch unsupported field types
+        for field in dtype.fields:
+            _ = _from_polars(field.dtype)
+        return plc.DataType(plc.TypeId.STRUCT)
+    else:
+        raise NotImplementedError(f"{dtype=} conversion not supported")
+
+
+class DataType:
+    """A datatype, preserving polars metadata."""
+
+    polars: pl.datatypes.DataType
+    plc: plc.DataType
+
+    def __init__(self, polars_dtype: pl.DataType) -> None:
+        self.polars = polars_dtype
+        self.plc = _from_polars(polars_dtype)
+
+    def id(self) -> plc.TypeId:
+        """The pylibcudf.TypeId of this DataType."""
+        return self.plc.id()
+
+    @property
+    def children(self) -> list[DataType]:
+        """The children types of this DataType."""
+        if self.plc.id() == plc.TypeId.STRUCT:
+            return [DataType(field.dtype) for field in self.polars.fields]
+        elif self.plc.id() == plc.TypeId.LIST:
+            return [DataType(self.polars.inner)]
+        return []
+
+    def __eq__(self, other: object) -> bool:
+        """Equality of DataTypes."""
+        if not isinstance(other, DataType):
+            return False
+        return self.polars == other.polars
+
+    def __hash__(self) -> int:
+        """Hash of the DataType."""
+        return hash(self.polars)
+
+    def __repr__(self) -> str:
+        """Representation of the DataType."""
+        return f"<DataType(polars={self.polars}, plc={self.id()!r})>"

cudf_polars/dsl/expr.py

@@ -33,6 +33,7 @@ from cudf_polars.dsl.expressions.selection import Filter, Gather
 from cudf_polars.dsl.expressions.slicing import Slice
 from cudf_polars.dsl.expressions.sorting import Sort, SortBy
 from cudf_polars.dsl.expressions.string import StringFunction
+from cudf_polars.dsl.expressions.struct import StructFunction
 from cudf_polars.dsl.expressions.ternary import Ternary
 from cudf_polars.dsl.expressions.unary import Cast, Len, UnaryFunction
 
@@ -58,6 +59,7 @@ __all__ = [
     "Sort",
     "SortBy",
     "StringFunction",
+    "StructFunction",
     "TemporalFunction",
     "Ternary",
     "UnaryFunction",