nullable struct

Author: gabriel.g.robin

dataframely/columns/struct.py

@@ -96,10 +96,14 @@ def pyarrow_dtype(self) -> pa.DataType:
         return pa.struct({name: col.pyarrow_dtype for name, col in self.inner.items()})
 
     def _sample_unchecked(self, generator: Generator, n: int) -> pl.Series:
-        return (
+        series = (
             pl.DataFrame(
                 {name: col.sample(generator, n) for name, col in self.inner.items()}
             )
             .select(pl.struct(pl.all()))
             .to_series()
         )
+        # Apply a null mask.
+        return generator._apply_null_mask(
+            series, null_probability=self._null_probability
+        )

tests/columns/test_sample.py

@@ -184,6 +184,7 @@ def test_sample_list(generator: Generator) -> None:
 def test_sample_struct(generator: Generator) -> None:
     column = dy.Struct({"a": dy.String(regex="[abc]"), "b": dy.String(regex="[a-z]xx")})
     samples = sample_and_validate(column, generator, n=10_000)
+    assert samples.is_null().any()
     assert len(samples) == 10_000