fix: Default to non nullable column for array and object (#326)

Co-authored-by: gabriel <gabriel.g.robin@airbus.com>

Author: gab23r

dataframely/columns/array.py

@@ -33,7 +33,7 @@ def __init__(
         inner: Column,
         shape: int | tuple[int, ...],
         *,
-        nullable: bool = True,
+        nullable: bool = False,
         # polars doesn't yet support grouping by arrays,
         # see https://github.com/pola-rs/polars/issues/22574
         primary_key: Literal[False] = False,

dataframely/columns/object.py

@@ -21,7 +21,7 @@ class Object(Column):
     def __init__(
         self,
         *,
-        nullable: bool = True,
+        nullable: bool = False,
         primary_key: bool = False,
         check: Check | None = None,
         alias: str | None = None,

tests/columns/test_pyarrow.py

@@ -90,7 +90,7 @@ def test_equal_polars_schema_list(inner: Column) -> None:
     ],
 )
 def test_equal_polars_schema_array(inner: Column, shape: int | tuple[int, ...]) -> None:
-    schema = create_schema("test", {"a": dy.Array(inner, shape)})
+    schema = create_schema("test", {"a": dy.Array(inner, shape, nullable=True)})
     actual = schema.to_pyarrow_schema()
     expected = schema.create_empty().to_arrow().schema
     assert actual == expected

tests/columns/test_sample.py

@@ -184,7 +184,7 @@ def test_sample_list(generator: Generator) -> None:
 
 
 def test_sample_array(generator: Generator) -> None:
-    column = dy.Array(dy.Bool(nullable=True), (2, 3))
+    column = dy.Array(dy.Bool(nullable=True), (2, 3), nullable=True)
     samples = sample_and_validate(column, generator, n=10_000)
     assert samples.is_null().any()
     assert set(samples.arr.len()) == {2, None}