fix: Ignore valid dype in cast

Author: gabriel

dataframely/schema.py

@@ -38,7 +38,6 @@
     ParquetStorageBackend,
 )
 from ._typing import DataFrame, LazyFrame, Validation
-from .columns import Any as AnyColumn
 from .columns import Column, column_from_dict
 from .config import Config
 from .exc import (
@@ -814,11 +813,19 @@ def cast(
             the lazy frame's schema but also means that a call to :meth:`polars.LazyFrame.collect`
             further down the line might fail because of the cast and/or missing columns.
         """
-        lf = df.lazy().select(
-            # Skip casting for Any columns since they accept any type
-            pl.col(name) if isinstance(col, AnyColumn) else pl.col(name).cast(col.dtype)
-            for name, col in cls.columns().items()
-        )
+
+        def _cast_to_schema(
+            lf: pl.LazyFrame, schema: dict[str, pl.DataType]
+        ) -> pl.LazyFrame:
+            return lf.select(cls.column_names()).cast(
+                {
+                    name: col.dtype
+                    for name, col in cls.columns().items()
+                    if name in schema and not col.validate_dtype(schema[name])
+                }
+            )
+
+        lf = df.lazy().pipe_with_schema(_cast_to_schema)
         if isinstance(df, pl.DataFrame):
             return lf.collect()  # type: ignore
         return lf  # type: ignore

tests/schema/test_cast.py

@@ -43,3 +43,18 @@ def test_cast_invalid_schema_lazy() -> None:
     lf = MySchema.cast(lf)
     with pytest.raises(plexc.ColumnNotFoundError):
         lf.collect()
+
+
+class IntegerSchema(dy.Schema):
+    a = dy.Integer()
+
+
+@pytest.mark.parametrize("df_type", [pl.DataFrame, pl.LazyFrame])
+def test_cast_preserves_valid_dtype(
+    df_type: type[pl.DataFrame] | type[pl.LazyFrame],
+) -> None:
+    """Test that cast doesn't change already valid dtypes (issue #318)."""
+    df = df_type({"a": [1, 2, 3]}, schema={"a": pl.Int32})
+    result = IntegerSchema.cast(df)
+    # Int32 is valid for dy.Integer, so it should NOT be cast to Int64
+    assert result.lazy().collect_schema()["a"] == pl.Int32