vortex_array/scalar_fn/fns/fill_null/
mod.rs

1// SPDX-License-Identifier: Apache-2.0
2// SPDX-FileCopyrightText: Copyright the Vortex contributors
3
4mod kernel;
5
6pub use kernel::*;
7use vortex_error::VortexResult;
8use vortex_error::vortex_bail;
9use vortex_error::vortex_ensure;
10use vortex_error::vortex_err;
11use vortex_session::VortexSession;
12use vortex_session::registry::CachedId;
13
14use crate::AnyColumnar;
15use crate::ArrayRef;
16use crate::CanonicalView;
17use crate::ColumnarView;
18use crate::ExecutionCtx;
19use crate::IntoArray;
20use crate::arrays::Bool;
21use crate::arrays::Decimal;
22use crate::arrays::Primitive;
23use crate::arrays::ScalarFnArray;
24use crate::dtype::DType;
25use crate::expr::BoundExpression;
26use crate::scalar::Scalar;
27use crate::scalar_fn::Arity;
28use crate::scalar_fn::ChildName;
29use crate::scalar_fn::EmptyOptions;
30use crate::scalar_fn::ExecutionArgs;
31use crate::scalar_fn::ReduceNode;
32use crate::scalar_fn::ReduceNodeValidity;
33use crate::scalar_fn::ScalarFnId;
34use crate::scalar_fn::ScalarFnVTable;
35use crate::scalar_fn::ScalarFnVTableExt;
36use crate::scalar_fn::is_not_null_node;
37
38/// An expression that replaces null values in the input with a fill value.
39#[derive(Clone)]
40pub struct FillNull;
41
42impl FillNull {
43    /// Creates a lazy operation that replaces null input values with `fill_value`.
44    ///
45    /// # Errors
46    ///
47    /// Returns an error if the children have different lengths or incompatible dtypes.
48    pub fn try_new(input: ArrayRef, fill_value: ArrayRef) -> VortexResult<ScalarFnArray> {
49        ScalarFnArray::try_new(FillNull.bind(EmptyOptions), [input, fill_value])
50    }
51}
52
53impl ScalarFnVTable for FillNull {
54    type Options = EmptyOptions;
55
56    fn id(&self) -> ScalarFnId {
57        static ID: CachedId = CachedId::new("vortex.fill_null");
58        *ID
59    }
60
61    fn serialize(&self, _options: &Self::Options) -> VortexResult<Option<Vec<u8>>> {
62        Ok(Some(vec![]))
63    }
64
65    fn deserialize(
66        &self,
67        _metadata: &[u8],
68        _session: &VortexSession,
69    ) -> VortexResult<Self::Options> {
70        Ok(EmptyOptions)
71    }
72
73    fn arity(&self, _options: &Self::Options) -> Arity {
74        Arity::Exact(2)
75    }
76
77    fn child_name(&self, _options: &Self::Options, child_idx: usize) -> ChildName {
78        match child_idx {
79            0 => ChildName::from("input"),
80            1 => ChildName::from("fill_value"),
81            _ => unreachable!("Invalid child index {} for FillNull expression", child_idx),
82        }
83    }
84
85    fn return_dtype(&self, _options: &Self::Options, arg_dtypes: &[DType]) -> VortexResult<DType> {
86        vortex_ensure!(
87            arg_dtypes[0].eq_ignore_nullability(&arg_dtypes[1]),
88            "fill_null requires input and fill value to have the same base type, got {} and {}",
89            arg_dtypes[0],
90            arg_dtypes[1]
91        );
92        // The result dtype takes the nullability of the fill value.
93        Ok(arg_dtypes[0]
94            .clone()
95            .with_nullability(arg_dtypes[1].nullability()))
96    }
97
98    fn execute(
99        &self,
100        _options: &Self::Options,
101        args: &dyn ExecutionArgs,
102        ctx: &mut ExecutionCtx,
103    ) -> VortexResult<ArrayRef> {
104        let input = args.get(0)?;
105        let fill_value = args.get(1)?;
106
107        let fill_scalar = fill_value
108            .as_constant()
109            .ok_or_else(|| vortex_err!("fill_null fill_value must be a constant/scalar"))?;
110
111        vortex_ensure!(
112            !fill_scalar.is_null(),
113            "fill_null requires a non-null fill value"
114        );
115
116        let Some(columnar) = input.as_opt::<AnyColumnar>() else {
117            let input = input.execute::<ArrayRef>(ctx)?;
118            return Ok(FillNull::try_new(input, fill_value)?.into_array());
119        };
120
121        match columnar {
122            ColumnarView::Canonical(canonical) => fill_null_canonical(canonical, &fill_scalar, ctx),
123            ColumnarView::Constant(constant) => fill_null_constant(constant, &fill_scalar),
124        }
125    }
126
127    fn simplify(
128        &self,
129        _options: &Self::Options,
130        expr: &BoundExpression,
131    ) -> VortexResult<Option<BoundExpression>> {
132        if !expr.child(0).dtype().is_nullable() {
133            return Ok(Some(expr.child(0).clone()));
134        }
135
136        Ok(None)
137    }
138
139    fn validity<T: ReduceNode>(
140        &self,
141        _options: &Self::Options,
142        node: &T,
143    ) -> VortexResult<ReduceNodeValidity<T>> {
144        // After fill_null, the result validity depends on the fill value's nullability.
145        // If fill_value is non-nullable, the result is always valid.
146        Ok(ReduceNodeValidity::Reduced(is_not_null_node(
147            &node.child(1),
148        )?))
149    }
150
151    fn is_strict(&self, _options: &Self::Options) -> bool {
152        // This function replaces null input values instead of propagating them.
153        false
154    }
155
156    fn is_infallible(&self, _options: &Self::Options) -> bool {
157        true
158    }
159}
160
161/// Fill nulls on a canonical array by directly dispatching to the appropriate kernel.
162///
163/// Returns the filled array, or bails if no kernel is registered for the canonical type.
164fn fill_null_canonical(
165    canonical: CanonicalView<'_>,
166    fill_value: &Scalar,
167    ctx: &mut ExecutionCtx,
168) -> VortexResult<ArrayRef> {
169    let arr = canonical.to_array_ref();
170    if let Some(result) = short_circuit(&arr, fill_value)? {
171        return Ok(result);
172    }
173    match canonical {
174        CanonicalView::Bool(a) => <Bool as FillNullKernel>::fill_null(a, fill_value, ctx)?
175            .ok_or_else(|| vortex_err!("FillNullKernel for BoolArray returned None")),
176        CanonicalView::Primitive(a) => {
177            <Primitive as FillNullKernel>::fill_null(a, fill_value, ctx)?
178                .ok_or_else(|| vortex_err!("FillNullKernel for PrimitiveArray returned None"))
179        }
180        CanonicalView::Decimal(a) => <Decimal as FillNullKernel>::fill_null(a, fill_value, ctx)?
181            .ok_or_else(|| vortex_err!("FillNullKernel for DecimalArray returned None")),
182        other => vortex_bail!(
183            "No FillNullKernel for canonical array {}",
184            other.to_array_ref().encoding_id()
185        ),
186    }
187}
188
189#[cfg(test)]
190mod tests {
191    use vortex_buffer::buffer;
192    use vortex_error::VortexExpect;
193
194    use crate::IntoArray;
195    use crate::VortexSessionExecute;
196    use crate::array_session;
197    use crate::arrays::PrimitiveArray;
198    use crate::arrays::StructArray;
199    use crate::assert_arrays_eq;
200    use crate::dtype::DType;
201    use crate::dtype::Nullability;
202    use crate::dtype::PType;
203    use crate::expr::fill_null;
204    use crate::expr::get_item;
205    use crate::expr::lit;
206    use crate::expr::root;
207
208    #[test]
209    fn dtype() {
210        let dtype = DType::Primitive(PType::I32, Nullability::Nullable);
211        assert_eq!(
212            fill_null(root(), lit(0i32)).return_dtype(&dtype).unwrap(),
213            DType::Primitive(PType::I32, Nullability::NonNullable)
214        );
215    }
216
217    #[test]
218    fn replace_children() {
219        let expr = fill_null(root(), lit(0i32));
220        expr.with_children(vec![root(), lit(0i32)])
221            .vortex_expect("operation should succeed in test");
222    }
223
224    #[test]
225    fn evaluate() {
226        let mut ctx = array_session().create_execution_ctx();
227        let test_array =
228            PrimitiveArray::from_option_iter([Some(1i32), None, Some(3), None, Some(5)])
229                .into_array();
230
231        let expr = fill_null(root(), lit(42i32));
232        let result = test_array.apply(&expr).unwrap();
233
234        assert_eq!(
235            result.dtype(),
236            &DType::Primitive(PType::I32, Nullability::NonNullable)
237        );
238        assert_arrays_eq!(
239            result,
240            PrimitiveArray::from_iter([1i32, 42, 3, 42, 5]),
241            &mut ctx
242        );
243    }
244
245    #[test]
246    fn evaluate_struct_field() {
247        let mut ctx = array_session().create_execution_ctx();
248        let test_array = StructArray::from_fields(&[(
249            "a",
250            PrimitiveArray::from_option_iter([Some(1i32), None, Some(3)]).into_array(),
251        )])
252        .unwrap()
253        .into_array();
254
255        let expr = fill_null(get_item("a", root()), lit(0i32));
256        let result = test_array.apply(&expr).unwrap();
257
258        assert_eq!(
259            result.dtype(),
260            &DType::Primitive(PType::I32, Nullability::NonNullable)
261        );
262        assert_arrays_eq!(result, PrimitiveArray::from_iter([1i32, 0, 3]), &mut ctx);
263    }
264
265    #[test]
266    fn evaluate_non_nullable_input() {
267        let mut ctx = array_session().create_execution_ctx();
268        let test_array = buffer![1i32, 2, 3].into_array();
269        let expr = fill_null(root(), lit(0i32));
270        let result = test_array.apply(&expr).unwrap();
271        assert_arrays_eq!(result, PrimitiveArray::from_iter([1i32, 2, 3]), &mut ctx);
272    }
273
274    #[test]
275    fn test_display() {
276        let expr = fill_null(get_item("value", root()), lit(0i32));
277        assert_eq!(expr.to_string(), "vortex.fill_null($.value, 0i32)");
278    }
279}
vortex_array/scalar_fn/fns/fill_null/mod.rs

vortex_array/scalar_fn/fns/fill_null/
mod.rs
