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
6use std::fmt::Formatter;
7
8pub use kernel::*;
9use vortex_error::VortexResult;
10use vortex_error::vortex_bail;
11use vortex_error::vortex_ensure;
12use vortex_error::vortex_err;
13use vortex_session::VortexSession;
14
15use crate::AnyColumnar;
16use crate::ArrayRef;
17use crate::CanonicalView;
18use crate::ColumnarView;
19use crate::ExecutionCtx;
20use crate::arrays::BoolVTable;
21use crate::arrays::DecimalVTable;
22use crate::arrays::PrimitiveVTable;
23use crate::builtins::ArrayBuiltins;
24use crate::dtype::DType;
25use crate::expr::Expression;
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::ScalarFnId;
32use crate::scalar_fn::ScalarFnVTable;
33
34/// An expression that replaces null values in the input with a fill value.
35#[derive(Clone)]
36pub struct FillNull;
37
38impl ScalarFnVTable for FillNull {
39    type Options = EmptyOptions;
40
41    fn id(&self) -> ScalarFnId {
42        ScalarFnId::from("vortex.fill_null")
43    }
44
45    fn serialize(&self, _options: &Self::Options) -> VortexResult<Option<Vec<u8>>> {
46        Ok(Some(vec![]))
47    }
48
49    fn deserialize(
50        &self,
51        _metadata: &[u8],
52        _session: &VortexSession,
53    ) -> VortexResult<Self::Options> {
54        Ok(EmptyOptions)
55    }
56
57    fn arity(&self, _options: &Self::Options) -> Arity {
58        Arity::Exact(2)
59    }
60
61    fn child_name(&self, _options: &Self::Options, child_idx: usize) -> ChildName {
62        match child_idx {
63            0 => ChildName::from("input"),
64            1 => ChildName::from("fill_value"),
65            _ => unreachable!("Invalid child index {} for FillNull expression", child_idx),
66        }
67    }
68
69    fn fmt_sql(
70        &self,
71        _options: &Self::Options,
72        expr: &Expression,
73        f: &mut Formatter<'_>,
74    ) -> std::fmt::Result {
75        write!(f, "fill_null(")?;
76        expr.child(0).fmt_sql(f)?;
77        write!(f, ", ")?;
78        expr.child(1).fmt_sql(f)?;
79        write!(f, ")")
80    }
81
82    fn return_dtype(&self, _options: &Self::Options, arg_dtypes: &[DType]) -> VortexResult<DType> {
83        vortex_ensure!(
84            arg_dtypes[0].eq_ignore_nullability(&arg_dtypes[1]),
85            "fill_null requires input and fill value to have the same base type, got {} and {}",
86            arg_dtypes[0],
87            arg_dtypes[1]
88        );
89        // The result dtype takes the nullability of the fill value.
90        Ok(arg_dtypes[0]
91            .clone()
92            .with_nullability(arg_dtypes[1].nullability()))
93    }
94
95    fn execute(
96        &self,
97        _options: &Self::Options,
98        args: &dyn ExecutionArgs,
99        ctx: &mut ExecutionCtx,
100    ) -> VortexResult<ArrayRef> {
101        let input = args.get(0)?;
102        let fill_value = args.get(1)?;
103
104        let fill_scalar = fill_value
105            .as_constant()
106            .ok_or_else(|| vortex_err!("fill_null fill_value must be a constant/scalar"))?;
107
108        let Some(columnar) = input.as_opt::<AnyColumnar>() else {
109            return input.execute::<ArrayRef>(ctx)?.fill_null(fill_scalar);
110        };
111
112        match columnar {
113            ColumnarView::Canonical(canonical) => fill_null_canonical(canonical, &fill_scalar, ctx),
114            ColumnarView::Constant(constant) => fill_null_constant(constant, &fill_scalar),
115        }
116    }
117
118    fn simplify(
119        &self,
120        _options: &Self::Options,
121        expr: &Expression,
122        ctx: &dyn crate::scalar_fn::SimplifyCtx,
123    ) -> VortexResult<Option<Expression>> {
124        let input_dtype = ctx.return_dtype(expr.child(0))?;
125
126        if !input_dtype.is_nullable() {
127            return Ok(Some(expr.child(0).clone()));
128        }
129
130        Ok(None)
131    }
132
133    fn validity(
134        &self,
135        _options: &Self::Options,
136        expression: &Expression,
137    ) -> VortexResult<Option<Expression>> {
138        // After fill_null, the result validity depends on the fill value's nullability.
139        // If fill_value is non-nullable, the result is always valid.
140        Ok(Some(expression.child(1).validity()?))
141    }
142
143    fn is_null_sensitive(&self, _options: &Self::Options) -> bool {
144        true
145    }
146
147    fn is_fallible(&self, _options: &Self::Options) -> bool {
148        false
149    }
150}
151
152/// Fill nulls on a canonical array by directly dispatching to the appropriate kernel.
153///
154/// Returns the filled array, or bails if no kernel is registered for the canonical type.
155fn fill_null_canonical(
156    canonical: CanonicalView<'_>,
157    fill_value: &Scalar,
158    ctx: &mut ExecutionCtx,
159) -> VortexResult<ArrayRef> {
160    let arr = canonical.as_ref().to_array();
161    if let Some(result) = precondition(&arr, fill_value)? {
162        // The result of precondition may return another ScalarFn, in which case we should
163        // apply it immediately.
164        // TODO(aduffy): Remove this once we have better driver check. We're also implicitly
165        //  relying on the fact that Cast execution will do an optimize on its result.
166        return result.execute::<ArrayRef>(ctx);
167    }
168    match canonical {
169        CanonicalView::Bool(a) => <BoolVTable as FillNullKernel>::fill_null(a, fill_value, ctx)?
170            .ok_or_else(|| vortex_err!("FillNullKernel for BoolArray returned None")),
171        CanonicalView::Primitive(a) => {
172            <PrimitiveVTable as FillNullKernel>::fill_null(a, fill_value, ctx)?
173                .ok_or_else(|| vortex_err!("FillNullKernel for PrimitiveArray returned None"))
174        }
175        CanonicalView::Decimal(a) => {
176            <DecimalVTable as FillNullKernel>::fill_null(a, fill_value, ctx)?
177                .ok_or_else(|| vortex_err!("FillNullKernel for DecimalArray returned None"))
178        }
179        other => vortex_bail!(
180            "No FillNullKernel for canonical array {}",
181            other.as_ref().encoding_id()
182        ),
183    }
184}
185
186#[cfg(test)]
187mod tests {
188    use vortex_buffer::buffer;
189    use vortex_error::VortexExpect;
190
191    use crate::IntoArray;
192    use crate::arrays::PrimitiveArray;
193    use crate::arrays::StructArray;
194    use crate::assert_arrays_eq;
195    use crate::dtype::DType;
196    use crate::dtype::Nullability;
197    use crate::dtype::PType;
198    use crate::expr::fill_null;
199    use crate::expr::get_item;
200    use crate::expr::lit;
201    use crate::expr::root;
202
203    #[test]
204    fn dtype() {
205        let dtype = DType::Primitive(PType::I32, Nullability::Nullable);
206        assert_eq!(
207            fill_null(root(), lit(0i32)).return_dtype(&dtype).unwrap(),
208            DType::Primitive(PType::I32, Nullability::NonNullable)
209        );
210    }
211
212    #[test]
213    fn replace_children() {
214        let expr = fill_null(root(), lit(0i32));
215        expr.with_children(vec![root(), lit(0i32)])
216            .vortex_expect("operation should succeed in test");
217    }
218
219    #[test]
220    fn evaluate() {
221        let test_array =
222            PrimitiveArray::from_option_iter([Some(1i32), None, Some(3), None, Some(5)])
223                .into_array();
224
225        let expr = fill_null(root(), lit(42i32));
226        let result = test_array.apply(&expr).unwrap();
227
228        assert_eq!(
229            result.dtype(),
230            &DType::Primitive(PType::I32, Nullability::NonNullable)
231        );
232        assert_arrays_eq!(result, PrimitiveArray::from_iter([1i32, 42, 3, 42, 5]));
233    }
234
235    #[test]
236    fn evaluate_struct_field() {
237        let test_array = StructArray::from_fields(&[(
238            "a",
239            PrimitiveArray::from_option_iter([Some(1i32), None, Some(3)]).into_array(),
240        )])
241        .unwrap()
242        .into_array();
243
244        let expr = fill_null(get_item("a", root()), lit(0i32));
245        let result = test_array.apply(&expr).unwrap();
246
247        assert_eq!(
248            result.dtype(),
249            &DType::Primitive(PType::I32, Nullability::NonNullable)
250        );
251        assert_arrays_eq!(result, PrimitiveArray::from_iter([1i32, 0, 3]));
252    }
253
254    #[test]
255    fn evaluate_non_nullable_input() {
256        let test_array = buffer![1i32, 2, 3].into_array();
257        let expr = fill_null(root(), lit(0i32));
258        let result = test_array.apply(&expr).unwrap();
259        assert_arrays_eq!(result, PrimitiveArray::from_iter([1i32, 2, 3]));
260    }
261
262    #[test]
263    fn test_display() {
264        let expr = fill_null(get_item("value", root()), lit(0i32));
265        assert_eq!(expr.to_string(), "fill_null($.value, 0i32)");
266    }
267}
vortex_array/scalar_fn/fns/fill_null/mod.rs

vortex_array/scalar_fn/fns/fill_null/
mod.rs
