16598973893

Committed 29 Jul 2025 02:25PM UTC coverage: 82.692% (-0.01%) from 82.703%

Build # 16598973893

Build Type

push

github

Committed by

web-flow

Commit Message

Clean up stats propagation for slicing (#3356)

Reduces the amount we copy some stats (by removing into_iter that forces
a full stats copy)

---------

Signed-off-by: Nicholas Gates <nick@nickgates.com>
Signed-off-by: Robert Kruszewski <github@robertk.io>
Signed-off-by: Will Manning <will@willmanning.io>
Co-authored-by: Robert Kruszewski <github@robertk.io>
Co-authored-by: Will Manning <will@willmanning.io>

Run Details

130 of 157 new or added lines in 15 files covered. (82.8%)

30 existing lines in 13 files now uncovered.

45215 of 54679 relevant lines covered (82.69%)

184610.34 hits per line

Source File
Press 'n' to go to next uncovered line, 'b' for previous

94.71

/vortex-array/src/compute/min_max.rs

// SPDX-License-Identifier: Apache-2.0
// SPDX-FileCopyrightText: Copyright the Vortex contributors

use std::sync::LazyLock;

use arcref::ArcRef;
use vortex_dtype::{DType, Nullability, StructFields};
use vortex_error::{VortexExpect, VortexResult, vortex_bail};
use vortex_scalar::Scalar;

use crate::Array;
use crate::arrays::ConstantVTable;
use crate::compute::{ComputeFn, ComputeFnVTable, InvocationArgs, Kernel, Output, UnaryArgs};
use crate::stats::{Precision, Stat, StatsProviderExt};
use crate::vtable::VTable;

static MIN_MAX_FN: LazyLock<ComputeFn> = LazyLock::new(|| {
    let compute = ComputeFn::new("min_max".into(), ArcRef::new_ref(&MinMax));
    for kernel in inventory::iter::<MinMaxKernelRef> {
        compute.register_kernel(kernel.0.clone());
    }
    compute
});

/// The minimum and maximum non-null values of an array, or None if there are no non-null values.
///
/// The return value dtype is the non-nullable version of the array dtype.
///
/// This will update the stats set of this array (as a side effect).
pub fn min_max(array: &dyn Array) -> VortexResult<Option<MinMaxResult>> {
    let scalar = MIN_MAX_FN
        .invoke(&InvocationArgs {
            inputs: &[array.into()],
            options: &(),
        })?
        .unwrap_scalar()?;
    MinMaxResult::from_scalar(scalar)
}

#[derive(Debug, Clone, PartialEq, Eq)]
pub struct MinMaxResult {
    pub min: Scalar,
    pub max: Scalar,
}

impl MinMaxResult {
    pub fn from_scalar(scalar: Scalar) -> VortexResult<Option<Self>> {
        if scalar.is_null() {
            Ok(None)
        } else {
            let min = scalar
                .as_struct()
                .field_by_idx(0)
                .vortex_expect("missing min field");
            let max = scalar
                .as_struct()
                .field_by_idx(1)
                .vortex_expect("missing max field");
            Ok(Some(MinMaxResult { min, max }))
        }
    }
}

pub struct MinMax;

impl ComputeFnVTable for MinMax {
    fn invoke(
        &self,
        args: &InvocationArgs,
        kernels: &[ArcRef<dyn Kernel>],
    ) -> VortexResult<Output> {
        let UnaryArgs { array, .. } = UnaryArgs::<()>::try_from(args)?;

        let return_dtype = self.return_dtype(args)?;

        match min_max_impl(array, kernels)? {
            None => Ok(Scalar::null(return_dtype).into()),
            Some(MinMaxResult { min, max }) => {
                assert!(
                    min <= max,
                    "min > max: min={} max={} encoding={}",
                    min,
                    max,
                    array.encoding_id()
                );

                // Update the stats set with the computed min/max
                array
                    .statistics()
                    .set(Stat::Min, Precision::Exact(min.value().clone()));
                array
                    .statistics()
                    .set(Stat::Max, Precision::Exact(max.value().clone()));

                // Return the min/max as a struct scalar
                Ok(Scalar::struct_(return_dtype, vec![min, max]).into())
            }
        }
    }

    fn return_dtype(&self, args: &InvocationArgs) -> VortexResult<DType> {
        let UnaryArgs { array, .. } = UnaryArgs::<()>::try_from(args)?;

        // We return a min/max struct scalar, where the overall struct is nullable in the case
        // that the array is all null or empty.
        Ok(DType::Struct(
            StructFields::new(
                ["min", "max"].into(),
                vec![array.dtype().clone(), array.dtype().clone()],
            ),
            Nullability::Nullable,
        ))
    }

    fn return_len(&self, _args: &InvocationArgs) -> VortexResult<usize> {
        Ok(1)
    }

    fn is_elementwise(&self) -> bool {
        false
    }
}

fn min_max_impl(
    array: &dyn Array,
    kernels: &[ArcRef<dyn Kernel>],
) -> VortexResult<Option<MinMaxResult>> {
    if array.is_empty() || array.valid_count()? == 0 {
        return Ok(None);
    }

    if let Some(array) = array.as_opt::<ConstantVTable>() {
        if !array.scalar().is_null() {
            return Ok(Some(MinMaxResult {
                min: array.scalar().clone(),
                max: array.scalar().clone(),
            }));
        }
    }

    let min = array
        .statistics()
        .get_scalar(Stat::Min, array.dtype())
        .and_then(Precision::as_exact);
    let max = array
        .statistics()
        .get_scalar(Stat::Max, array.dtype())
        .and_then(Precision::as_exact);

    if let Some((min, max)) = min.zip(max) {
        return Ok(Some(MinMaxResult { min, max }));
    }

    let args = InvocationArgs {
        inputs: &[array.into()],
        options: &(),
    };
    for kernel in kernels {
        if let Some(output) = kernel.invoke(&args)? {
            return MinMaxResult::from_scalar(output.unwrap_scalar()?);
        }
    }
    if let Some(output) = array.invoke(&MIN_MAX_FN, &args)? {
        return MinMaxResult::from_scalar(output.unwrap_scalar()?);
    }

    if !array.is_canonical() {
        let array = array.to_canonical()?;
        return min_max(array.as_ref());
    }

    vortex_bail!(NotImplemented: "min_max", array.encoding_id());
}

/// The minimum and maximum non-null values of an array, or None if there are no non-null values.
pub trait MinMaxKernel: VTable {
    fn min_max(&self, array: &Self::Array) -> VortexResult<Option<MinMaxResult>>;
}

pub struct MinMaxKernelRef(ArcRef<dyn Kernel>);
inventory::collect!(MinMaxKernelRef);

#[derive(Debug)]
pub struct MinMaxKernelAdapter<V: VTable>(pub V);

impl<V: VTable + MinMaxKernel> MinMaxKernelAdapter<V> {
    pub const fn lift(&'static self) -> MinMaxKernelRef {
        MinMaxKernelRef(ArcRef::new_ref(self))
    }
}

impl<V: VTable + MinMaxKernel> Kernel for MinMaxKernelAdapter<V> {
    fn invoke(&self, args: &InvocationArgs) -> VortexResult<Option<Output>> {
        let inputs = UnaryArgs::<()>::try_from(args)?;
        let Some(array) = inputs.array.as_opt::<V>() else {
            return Ok(None);
        };
        let dtype = DType::Struct(
            StructFields::new(
                ["min", "max"].into(),
                vec![array.dtype().clone(), array.dtype().clone()],
            ),
            Nullability::Nullable,
        );
        Ok(Some(match V::min_max(&self.0, array)? {
            None => Scalar::null(dtype).into(),
            Some(MinMaxResult { min, max }) => Scalar::struct_(dtype, vec![min, max]).into(),
        }))
    }
}

#[cfg(test)]
mod tests {
    use arrow_buffer::BooleanBuffer;
    use vortex_buffer::buffer;

    use crate::arrays::{BoolArray, NullArray, PrimitiveArray};
    use crate::compute::{MinMaxResult, min_max};
    use crate::validity::Validity;

    #[test]
    fn test_prim_max() {
        let p = PrimitiveArray::new(buffer![1, 2, 3], Validity::NonNullable);
        assert_eq!(
            min_max(p.as_ref()).unwrap(),
            Some(MinMaxResult {
                min: 1.into(),
                max: 3.into()
            })
        );
    }

    #[test]
    fn test_bool_max() {
        let p = BoolArray::new(
            BooleanBuffer::from([true, true, true].as_slice()),
            Validity::NonNullable,
        );
        assert_eq!(
            min_max(p.as_ref()).unwrap(),
            Some(MinMaxResult {
                min: true.into(),
                max: true.into()
            })
        );

        let p = BoolArray::new(
            BooleanBuffer::from([false, false, false].as_slice()),
            Validity::NonNullable,
        );
        assert_eq!(
            min_max(p.as_ref()).unwrap(),
            Some(MinMaxResult {
                min: false.into(),
                max: false.into()
            })
        );

        let p = BoolArray::new(
            BooleanBuffer::from([false, true, false].as_slice()),
            Validity::NonNullable,
        );
        assert_eq!(
            min_max(p.as_ref()).unwrap(),
            Some(MinMaxResult {
                min: false.into(),
                max: true.into()
            })
        );
    }

    #[test]
    fn test_null() {
        let p = NullArray::new(1);
        assert_eq!(min_max(p.as_ref()).unwrap(), None);
    }
}

1	// SPDX-License-Identifier: Apache-2.0
2	// SPDX-FileCopyrightText: Copyright the Vortex contributors
3
4	use std::sync::LazyLock;
5
6	use arcref::ArcRef;
7	use vortex_dtype::{DType, Nullability, StructFields};
8	use vortex_error::{VortexExpect, VortexResult, vortex_bail};
9	use vortex_scalar::Scalar;
10
11	use crate::Array;
12	use crate::arrays::ConstantVTable;
13	use crate::compute::{ComputeFn, ComputeFnVTable, InvocationArgs, Kernel, Output, UnaryArgs};
14	use crate::stats::{Precision, Stat, StatsProviderExt};
15	use crate::vtable::VTable;
16
17	static MIN_MAX_FN: LazyLock<ComputeFn> = LazyLock::new(\|\| {	5,036✔
18	let compute = ComputeFn::new("min_max".into(), ArcRef::new_ref(&MinMax));	5,036✔
19	for kernel in inventory::iter::<MinMaxKernelRef> {	63,876✔
20	compute.register_kernel(kernel.0.clone());	58,840✔
21	}	58,840✔
22	compute	5,036✔
23	});	5,036✔
24
25	/// The minimum and maximum non-null values of an array, or None if there are no non-null values.
26	///
27	/// The return value dtype is the non-nullable version of the array dtype.
28	///
29	/// This will update the stats set of this array (as a side effect).
30	pub fn min_max(array: &dyn Array) -> VortexResult<Option<MinMaxResult>> {	64,762✔
31	let scalar = MIN_MAX_FN	64,762✔
32	.invoke(&InvocationArgs {	64,762✔
33	inputs: &[array.into()],	64,762✔
34	options: &(),	64,762✔
35	})?	64,762✔
36	.unwrap_scalar()?;	64,762✔
37	MinMaxResult::from_scalar(scalar)	64,762✔
38	}	64,762✔
39
40	#[derive(Debug, Clone, PartialEq, Eq)]
41	pub struct MinMaxResult {
42	pub min: Scalar,
43	pub max: Scalar,
44	}
45
46	impl MinMaxResult {
47	pub fn from_scalar(scalar: Scalar) -> VortexResult<Option<Self>> {	126,537✔
48	if scalar.is_null() {	126,537✔
49	Ok(None)	11,282✔
50	} else {
51	let min = scalar	115,255✔
52	.as_struct()	115,255✔
53	.field_by_idx(0)	115,255✔
54	.vortex_expect("missing min field");	115,255✔
55	let max = scalar	115,255✔
56	.as_struct()	115,255✔
57	.field_by_idx(1)	115,255✔
58	.vortex_expect("missing max field");	115,255✔
59	Ok(Some(MinMaxResult { min, max }))	115,255✔
60	}
61	}	126,537✔
62	}
63
64	pub struct MinMax;
65
66	impl ComputeFnVTable for MinMax {
67	fn invoke(	64,762✔
68	&self,	64,762✔
69	args: &InvocationArgs,	64,762✔
70	kernels: &[ArcRef<dyn Kernel>],	64,762✔
71	) -> VortexResult<Output> {	64,762✔
72	let UnaryArgs { array, .. } = UnaryArgs::<()>::try_from(args)?;	64,762✔
73
74	let return_dtype = self.return_dtype(args)?;	64,762✔
75
76	match min_max_impl(array, kernels)? {	64,762✔
77	None => Ok(Scalar::null(return_dtype).into()),	5,870✔
78	Some(MinMaxResult { min, max }) => {	58,892✔
79	assert!(	58,892✔
80	min <= max,	58,892✔
81	"min > max: min={} max={} encoding={}",	×
82	min,
83	max,
84	array.encoding_id()	×
85	);
86
87	// Update the stats set with the computed min/max
88	array	58,892✔
89	.statistics()	58,892✔
90	.set(Stat::Min, Precision::Exact(min.value().clone()));	58,892✔
91	array	58,892✔
92	.statistics()	58,892✔
93	.set(Stat::Max, Precision::Exact(max.value().clone()));	58,892✔
94
95	// Return the min/max as a struct scalar
96	Ok(Scalar::struct_(return_dtype, vec![min, max]).into())	58,892✔
97	}
98	}
99	}	64,762✔
100
101	fn return_dtype(&self, args: &InvocationArgs) -> VortexResult<DType> {	129,524✔
102	let UnaryArgs { array, .. } = UnaryArgs::<()>::try_from(args)?;	129,524✔
103
104	// We return a min/max struct scalar, where the overall struct is nullable in the case
105	// that the array is all null or empty.
106	Ok(DType::Struct(	129,524✔
107	StructFields::new(	129,524✔
108	["min", "max"].into(),	129,524✔
109	vec![array.dtype().clone(), array.dtype().clone()],	129,524✔
110	),	129,524✔
111	Nullability::Nullable,	129,524✔
112	))	129,524✔
113	}	129,524✔
114
115	fn return_len(&self, _args: &InvocationArgs) -> VortexResult<usize> {	64,762✔
116	Ok(1)	64,762✔
117	}	64,762✔
118
119	fn is_elementwise(&self) -> bool {	64,762✔
120	false	64,762✔
121	}	64,762✔
122	}
123
124	fn min_max_impl(	64,762✔
125	array: &dyn Array,	64,762✔
126	kernels: &[ArcRef<dyn Kernel>],	64,762✔
127	) -> VortexResult<Option<MinMaxResult>> {	64,762✔
128	if array.is_empty() \|\| array.valid_count()? == 0 {	64,762✔
129	return Ok(None);	458✔
130	}	64,304✔
131
132	if let Some(array) = array.as_opt::<ConstantVTable>() {	64,304✔
133	if !array.scalar().is_null() {	2,053✔
134	return Ok(Some(MinMaxResult {	2,053✔
135	min: array.scalar().clone(),	2,053✔
136	max: array.scalar().clone(),	2,053✔
137	}));	2,053✔
NEW 138	}	×
139	}	62,251✔
140
141	let min = array	62,251✔
142	.statistics()	62,251✔
143	.get_scalar(Stat::Min, array.dtype())	62,251✔
144	.and_then(Precision::as_exact);	62,251✔
145	let max = array	62,251✔
146	.statistics()	62,251✔
147	.get_scalar(Stat::Max, array.dtype())	62,251✔
148	.and_then(Precision::as_exact);	62,251✔
149
150	if let Some((min, max)) = min.zip(max) {	62,251✔
151	return Ok(Some(MinMaxResult { min, max }));	400✔
152	}	61,851✔
153
154	let args = InvocationArgs {	61,851✔
155	inputs: &[array.into()],	61,851✔
156	options: &(),	61,851✔
157	};	61,851✔
158	for kernel in kernels {	170,987✔
159	if let Some(output) = kernel.invoke(&args)? {	170,911✔
160	return MinMaxResult::from_scalar(output.unwrap_scalar()?);	61,775✔
161	}	109,136✔
162	}
163	if let Some(output) = array.invoke(&MIN_MAX_FN, &args)? {	76✔
164	return MinMaxResult::from_scalar(output.unwrap_scalar()?);	×
165	}	76✔
166
167	if !array.is_canonical() {	76✔
168	let array = array.to_canonical()?;	76✔
169	return min_max(array.as_ref());	76✔
170	}	×
171
172	vortex_bail!(NotImplemented: "min_max", array.encoding_id());	×
173	}	64,762✔
174
175	/// The minimum and maximum non-null values of an array, or None if there are no non-null values.
176	pub trait MinMaxKernel: VTable {
177	fn min_max(&self, array: &Self::Array) -> VortexResult<Option<MinMaxResult>>;
178	}
179
180	pub struct MinMaxKernelRef(ArcRef<dyn Kernel>);
181	inventory::collect!(MinMaxKernelRef);
182
183	#[derive(Debug)]
184	pub struct MinMaxKernelAdapter<V: VTable>(pub V);
185
186	impl<V: VTable + MinMaxKernel> MinMaxKernelAdapter<V> {
187	pub const fn lift(&'static self) -> MinMaxKernelRef {	×
188	MinMaxKernelRef(ArcRef::new_ref(self))	×
189	}	×
190	}
191
192	impl<V: VTable + MinMaxKernel> Kernel for MinMaxKernelAdapter<V> {
193	fn invoke(&self, args: &InvocationArgs) -> VortexResult<Option<Output>> {	170,829✔
194	let inputs = UnaryArgs::<()>::try_from(args)?;	170,829✔
195	let Some(array) = inputs.array.as_opt::<V>() else {	170,829✔
196	return Ok(None);	109,091✔
197	};
198	let dtype = DType::Struct(	61,738✔
199	StructFields::new(	61,738✔
200	["min", "max"].into(),	61,738✔
201	vec![array.dtype().clone(), array.dtype().clone()],	61,738✔
202	),	61,738✔
203	Nullability::Nullable,	61,738✔
204	);	61,738✔
205	Ok(Some(match V::min_max(&self.0, array)? {	61,738✔
206	None => Scalar::null(dtype).into(),	5,412✔
207	Some(MinMaxResult { min, max }) => Scalar::struct_(dtype, vec![min, max]).into(),	56,326✔
208	}))
209	}	170,829✔
210	}
211
212	#[cfg(test)]
213	mod tests {
214	use arrow_buffer::BooleanBuffer;
215	use vortex_buffer::buffer;
216
217	use crate::arrays::{BoolArray, NullArray, PrimitiveArray};
218	use crate::compute::{MinMaxResult, min_max};
219	use crate::validity::Validity;
220
221	#[test]
222	fn test_prim_max() {	1✔
223	let p = PrimitiveArray::new(buffer![1, 2, 3], Validity::NonNullable);	1✔
224	assert_eq!(	1✔
225	min_max(p.as_ref()).unwrap(),	1✔
226	Some(MinMaxResult {	1✔
227	min: 1.into(),	1✔
228	max: 3.into()	1✔
229	})	1✔
230	);
231	}	1✔
232
233	#[test]
234	fn test_bool_max() {	1✔
235	let p = BoolArray::new(	1✔
236	BooleanBuffer::from([true, true, true].as_slice()),	1✔
237	Validity::NonNullable,	1✔
238	);
239	assert_eq!(	1✔
240	min_max(p.as_ref()).unwrap(),	1✔
241	Some(MinMaxResult {	1✔
242	min: true.into(),	1✔
243	max: true.into()	1✔
244	})	1✔
245	);
246
247	let p = BoolArray::new(	1✔
248	BooleanBuffer::from([false, false, false].as_slice()),	1✔
249	Validity::NonNullable,	1✔
250	);
251	assert_eq!(	1✔
252	min_max(p.as_ref()).unwrap(),	1✔
253	Some(MinMaxResult {	1✔
254	min: false.into(),	1✔
255	max: false.into()	1✔
256	})	1✔
257	);
258
259	let p = BoolArray::new(	1✔
260	BooleanBuffer::from([false, true, false].as_slice()),	1✔
261	Validity::NonNullable,	1✔
262	);
263	assert_eq!(	1✔
264	min_max(p.as_ref()).unwrap(),	1✔
265	Some(MinMaxResult {	1✔
266	min: false.into(),	1✔
267	max: true.into()	1✔
268	})	1✔
269	);
270	}	1✔
271
272	#[test]
273	fn test_null() {	1✔
274	let p = NullArray::new(1);	1✔
275	assert_eq!(min_max(p.as_ref()).unwrap(), None);	1✔
276	}	1✔
277	}

vortex-data / vortex / 16598973893

Source File Press 'n' to go to next uncovered line, 'b' for previous

Source File
Press 'n' to go to next uncovered line, 'b' for previous