vortex_array/aggregate_fn/
vtable.rs

1// SPDX-License-Identifier: Apache-2.0
2// SPDX-FileCopyrightText: Copyright the Vortex contributors
3
4use std::fmt;
5use std::fmt::Debug;
6use std::fmt::Display;
7use std::fmt::Formatter;
8use std::hash::Hash;
9
10use vortex_error::VortexResult;
11use vortex_error::vortex_bail;
12use vortex_session::VortexSession;
13
14use crate::ArrayRef;
15use crate::Columnar;
16use crate::ExecutionCtx;
17use crate::aggregate_fn::AggregateFn;
18use crate::aggregate_fn::AggregateFnId;
19use crate::aggregate_fn::AggregateFnRef;
20use crate::dtype::DType;
21use crate::scalar::Scalar;
22
23/// Defines the interface for aggregate function vtables.
24///
25/// This trait is non-object-safe and allows the implementer to make use of associated types
26/// for improved type safety, while allowing Vortex to enforce runtime checks on the inputs and
27/// outputs of each function.
28///
29/// The [`AggregateFnVTable`] trait should be implemented for a struct that holds global data across
30/// all instances of the aggregate. In almost all cases, this struct will be an empty unit
31/// struct, since most aggregates do not require any global state.
32pub trait AggregateFnVTable: 'static + Sized + Clone + Send + Sync {
33    /// Options for this aggregate function.
34    type Options: 'static + Send + Sync + Clone + Debug + Display + PartialEq + Eq + Hash;
35
36    /// The partial accumulator state for a single group.
37    type Partial: 'static + Send;
38
39    /// Returns the ID of the aggregate function vtable.
40    fn id(&self) -> AggregateFnId;
41
42    /// Serialize the options for this aggregate function.
43    ///
44    /// Should return `Ok(None)` if the function is not serializable, and `Ok(vec![])` if it is
45    /// serializable but has no metadata.
46    fn serialize(&self, options: &Self::Options) -> VortexResult<Option<Vec<u8>>> {
47        _ = options;
48        Ok(None)
49    }
50
51    /// Deserialize the options of this aggregate function.
52    fn deserialize(
53        &self,
54        _metadata: &[u8],
55        _session: &VortexSession,
56    ) -> VortexResult<Self::Options> {
57        vortex_bail!("Aggregate function {} is not deserializable", self.id());
58    }
59
60    /// The return [`DType`] of the aggregate.
61    ///
62    /// Returns `None` if the aggregate function cannot be applied to the input dtype.
63    fn return_dtype(&self, options: &Self::Options, input_dtype: &DType) -> Option<DType>;
64
65    /// DType of the intermediate partial accumulator state.
66    ///
67    /// Use a struct dtype when multiple fields are needed
68    /// (e.g., Mean: `Struct { sum: f64, count: u64 }`).
69    ///
70    /// Returns `None` if the aggregate function cannot be applied to the input dtype.
71    fn partial_dtype(&self, options: &Self::Options, input_dtype: &DType) -> Option<DType>;
72
73    /// Return the partial accumulator state for an empty group.
74    fn empty_partial(
75        &self,
76        options: &Self::Options,
77        input_dtype: &DType,
78    ) -> VortexResult<Self::Partial>;
79
80    /// Combine partial scalar state into the accumulator.
81    fn combine_partials(&self, partial: &mut Self::Partial, other: Scalar) -> VortexResult<()>;
82
83    /// Convert the partial state into a partial scalar.
84    ///
85    /// The returned scalar must have the same DType as specified by `partial_dtype` for the
86    /// options and input dtype used to construct the state.
87    fn to_scalar(&self, partial: &Self::Partial) -> VortexResult<Scalar>;
88
89    /// Reset the state of the accumulator to an empty group.
90    fn reset(&self, partial: &mut Self::Partial);
91
92    /// Is the partial accumulator state is "saturated", i.e. has it reached a state where the
93    /// final result is fully determined.
94    fn is_saturated(&self, state: &Self::Partial) -> bool;
95
96    /// Try to accumulate the raw array before decompression.
97    ///
98    /// Returns `true` if the array was handled, `false` to fall through to
99    /// the default kernel dispatch and canonicalization path.
100    ///
101    /// This is useful for aggregates that only depend on array metadata (e.g., validity)
102    /// rather than the encoded data, avoiding unnecessary decompression.
103    fn try_accumulate(
104        &self,
105        _state: &mut Self::Partial,
106        _batch: &ArrayRef,
107        _ctx: &mut ExecutionCtx,
108    ) -> VortexResult<bool> {
109        Ok(false)
110    }
111
112    /// Accumulate a new canonical array into the accumulator state.
113    fn accumulate(
114        &self,
115        state: &mut Self::Partial,
116        batch: &Columnar,
117        ctx: &mut ExecutionCtx,
118    ) -> VortexResult<()>;
119
120    /// Finalize an array of accumulator states into an array of aggregate results.
121    ///
122    /// The provides `states` array has dtype as specified by `state_dtype`, the result array
123    /// must have dtype as specified by `return_dtype`.
124    fn finalize(&self, states: ArrayRef) -> VortexResult<ArrayRef>;
125
126    /// Finalize a scalar accumulator state into an aggregate result.
127    ///
128    /// The provided `state` has dtype as specified by `state_dtype`, the result scalar must have
129    /// dtype as specified by `return_dtype`.
130    fn finalize_scalar(&self, partial: &Self::Partial) -> VortexResult<Scalar>;
131}
132
133#[derive(Clone, Debug, PartialEq, Eq, Hash)]
134pub struct EmptyOptions;
135impl Display for EmptyOptions {
136    fn fmt(&self, f: &mut Formatter<'_>) -> fmt::Result {
137        write!(f, "")
138    }
139}
140
141/// Factory functions for aggregate vtables.
142pub trait AggregateFnVTableExt: AggregateFnVTable {
143    /// Bind this vtable with the given options into an [`AggregateFnRef`].
144    fn bind(&self, options: Self::Options) -> AggregateFnRef {
145        AggregateFn::new(self.clone(), options).erased()
146    }
147}
148impl<V: AggregateFnVTable> AggregateFnVTableExt for V {}
vortex_array/aggregate_fn/vtable.rs

vortex_array/aggregate_fn/
vtable.rs
