vortex_array/aggregate_fn/fns/all_non_nan/
mod.rs

1// SPDX-License-Identifier: Apache-2.0
2// SPDX-FileCopyrightText: Copyright the Vortex contributors
3
4use vortex_error::VortexResult;
5use vortex_session::VortexSession;
6use vortex_session::registry::CachedId;
7
8use crate::ArrayRef;
9use crate::Columnar;
10use crate::ExecutionCtx;
11use crate::IntoArray;
12use crate::aggregate_fn::AggregateArgs;
13use crate::aggregate_fn::AggregateFnId;
14use crate::aggregate_fn::AggregateFnVTable;
15use crate::aggregate_fn::EmptyOptions;
16use crate::aggregate_fn::fns::nan_count::nan_count;
17use crate::dtype::DType;
18use crate::dtype::Nullability;
19use crate::scalar::Scalar;
20
21/// Compute whether every value in an array is not NaN.
22///
23/// Like other `all` aggregates, this is vacuously true for empty input.
24///
25/// This is a pruning aggregate, not just a convenience wrapper around
26/// [`NanCount`][crate::aggregate_fn::fns::nan_count::NanCount]. Pruning aggregates must prove a
27/// row-wise fact for every value in the scope, so their partials remain valid when a stats column is
28/// sliced or concatenated alongside the data. [`NanCount`][crate::aggregate_fn::fns::nan_count::NanCount]
29/// carries cross-row count information instead, so it is useful as a legacy storage format but not
30/// as the pruning expression itself.
31#[derive(Clone, Debug)]
32pub struct AllNonNan;
33
34impl AggregateFnVTable for AllNonNan {
35    type Options = EmptyOptions;
36    type Partial = bool;
37
38    fn id(&self) -> AggregateFnId {
39        static ID: CachedId = CachedId::new("vortex.all_non_nan");
40        *ID
41    }
42
43    fn serialize(&self, _options: &Self::Options) -> VortexResult<Option<Vec<u8>>> {
44        Ok(Some(vec![]))
45    }
46
47    fn deserialize(
48        &self,
49        _metadata: &[u8],
50        _session: &VortexSession,
51    ) -> VortexResult<Self::Options> {
52        Ok(EmptyOptions)
53    }
54
55    fn return_dtype(&self, _options: &Self::Options, input_dtype: &DType) -> Option<DType> {
56        matches!(input_dtype, DType::Primitive(ptype, _) if ptype.is_float())
57            .then_some(DType::Bool(Nullability::Nullable))
58    }
59
60    fn partial_dtype(&self, options: &Self::Options, input_dtype: &DType) -> Option<DType> {
61        self.return_dtype(options, input_dtype)
62    }
63
64    fn empty_partial(
65        &self,
66        _args: AggregateArgs<'_, Self::Options>,
67    ) -> VortexResult<Self::Partial> {
68        Ok(true)
69    }
70
71    fn partial_from_scalar(
72        &self,
73        _args: AggregateArgs<'_, Self::Options>,
74        scalar: &Scalar,
75    ) -> VortexResult<Self::Partial> {
76        bool::try_from(scalar)
77    }
78
79    fn merge_partials(
80        &self,
81        _args: AggregateArgs<'_, Self::Options>,
82        first: Self::Partial,
83        second: Self::Partial,
84    ) -> VortexResult<Self::Partial> {
85        Ok(first && second)
86    }
87
88    fn to_scalar(
89        &self,
90        _args: AggregateArgs<'_, Self::Options>,
91        partial: &Self::Partial,
92    ) -> VortexResult<Scalar> {
93        Ok(Scalar::bool(*partial, Nullability::Nullable))
94    }
95
96    fn is_saturated(
97        &self,
98        _args: AggregateArgs<'_, Self::Options>,
99        partial: &Self::Partial,
100    ) -> bool {
101        !*partial
102    }
103
104    fn try_accumulate(
105        &self,
106        _args: AggregateArgs<'_, Self::Options>,
107        state: &mut Self::Partial,
108        batch: &ArrayRef,
109        ctx: &mut ExecutionCtx,
110    ) -> VortexResult<bool> {
111        *state &= nan_count(batch, ctx)? == 0;
112        Ok(true)
113    }
114
115    fn accumulate(
116        &self,
117        _args: AggregateArgs<'_, Self::Options>,
118        partial: &mut Self::Partial,
119        batch: &Columnar,
120        ctx: &mut ExecutionCtx,
121    ) -> VortexResult<()> {
122        // Normal array dispatch is handled by `try_accumulate`, which always short-circuits.
123        // Keep this fallback in sync for direct Columnar accumulation paths.
124        let array = match batch {
125            Columnar::Constant(c) => c.clone().into_array(),
126            Columnar::Canonical(c) => c.clone().into_array(),
127        };
128        *partial &= nan_count(&array, ctx)? == 0;
129        Ok(())
130    }
131
132    fn finalize(
133        &self,
134        _args: AggregateArgs<'_, Self::Options>,
135        partials: ArrayRef,
136    ) -> VortexResult<ArrayRef> {
137        Ok(partials)
138    }
139
140    fn finalize_scalar(
141        &self,
142        args: AggregateArgs<'_, Self::Options>,
143        partial: &Self::Partial,
144    ) -> VortexResult<Scalar> {
145        self.to_scalar(args, partial)
146    }
147}
148
149#[cfg(test)]
150mod tests {
151    use vortex_error::VortexResult;
152
153    use crate::IntoArray;
154    use crate::VortexSessionExecute;
155    use crate::aggregate_fn::Accumulator;
156    use crate::aggregate_fn::DynAccumulator;
157    use crate::aggregate_fn::EmptyOptions;
158    use crate::aggregate_fn::fns::all_non_nan::AllNonNan;
159    use crate::array_session;
160    use crate::arrays::PrimitiveArray;
161    use crate::dtype::DType;
162    use crate::dtype::Nullability;
163    use crate::dtype::PType;
164
165    #[test]
166    fn all_non_nan_aggregate_fn() -> VortexResult<()> {
167        let mut ctx = array_session().create_execution_ctx();
168        let dtype = DType::Primitive(PType::F32, Nullability::Nullable);
169        let mut acc = Accumulator::try_new(AllNonNan, EmptyOptions, dtype)?;
170
171        let batch = PrimitiveArray::from_option_iter([Some(1.0f32), None, Some(3.0)]).into_array();
172        acc.accumulate(&batch, &mut ctx)?;
173
174        assert!(bool::try_from(&acc.finish()?)?);
175        Ok(())
176    }
177
178    #[test]
179    fn all_non_nan_false_with_nan() -> VortexResult<()> {
180        let mut ctx = array_session().create_execution_ctx();
181        let dtype = DType::Primitive(PType::F32, Nullability::Nullable);
182        let mut acc = Accumulator::try_new(AllNonNan, EmptyOptions, dtype)?;
183
184        let batch = PrimitiveArray::from_option_iter([Some(1.0f32), Some(f32::NAN)]).into_array();
185        acc.accumulate(&batch, &mut ctx)?;
186
187        assert!(!bool::try_from(&acc.finish()?)?);
188        Ok(())
189    }
190
191    #[test]
192    fn all_non_nan_unsupported_for_non_float() -> VortexResult<()> {
193        let dtype = DType::Primitive(PType::I32, Nullability::Nullable);
194        assert!(Accumulator::try_new(AllNonNan, EmptyOptions, dtype).is_err());
195        Ok(())
196    }
197
198    #[test]
199    fn all_non_nan_true_for_empty_float() -> VortexResult<()> {
200        let dtype = DType::Primitive(PType::F32, Nullability::Nullable);
201        let mut acc = Accumulator::try_new(AllNonNan, EmptyOptions, dtype)?;
202
203        assert!(bool::try_from(&acc.finish()?)?);
204        Ok(())
205    }
206
207    #[test]
208    fn all_non_nan_true_with_nulls() -> VortexResult<()> {
209        let mut ctx = array_session().create_execution_ctx();
210        let dtype = DType::Primitive(PType::F32, Nullability::Nullable);
211        let mut acc = Accumulator::try_new(AllNonNan, EmptyOptions, dtype)?;
212
213        let batch = PrimitiveArray::from_option_iter([Some(1.0f32), None]).into_array();
214        acc.accumulate(&batch, &mut ctx)?;
215
216        assert!(bool::try_from(&acc.finish()?)?);
217        Ok(())
218    }
219}
vortex_array/aggregate_fn/fns/all_non_nan/mod.rs

vortex_array/aggregate_fn/fns/all_non_nan/
mod.rs
