16593958537

Committed 29 Jul 2025 10:48AM UTC coverage: 82.285% (+0.5%) from 81.796%

Build # 16593958537

Build Type

Pull #4036

github

Committed by

web-flow

Commit Message

Merge 04147cb0f into 348079fc3

Pull Request Pull Request #4036: varbinview builder buffer deduplication

Run Details

146 of 154 new or added lines in 2 files covered. (94.81%)

348 existing lines in 26 files now uncovered.

44470 of 54044 relevant lines covered (82.28%)

169522.95 hits per line

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

95.06

/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::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);
    }

    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::compute::{ComputeFn, ComputeFnVTable, InvocationArgs, Kernel, Output, UnaryArgs};
13	use crate::stats::{Precision, Stat, StatsProviderExt};
14	use crate::vtable::VTable;
15
16	static MIN_MAX_FN: LazyLock<ComputeFn> = LazyLock::new(\|\| {	3,895✔
17	let compute = ComputeFn::new("min_max".into(), ArcRef::new_ref(&MinMax));	3,895✔
18	for kernel in inventory::iter::<MinMaxKernelRef> {	50,983✔
19	compute.register_kernel(kernel.0.clone());	47,088✔
20	}	47,088✔
21	compute	3,895✔
22	});	3,895✔
23
24	/// The minimum and maximum non-null values of an array, or None if there are no non-null values.
25	///
26	/// The return value dtype is the non-nullable version of the array dtype.
27	///
28	/// This will update the stats set of this array (as a side effect).
29	pub fn min_max(array: &dyn Array) -> VortexResult<Option<MinMaxResult>> {	58,419✔
30	let scalar = MIN_MAX_FN	58,419✔
31	.invoke(&InvocationArgs {	58,419✔
32	inputs: &[array.into()],	58,419✔
33	options: &(),	58,419✔
34	})?	58,419✔
35	.unwrap_scalar()?;	58,419✔
36	MinMaxResult::from_scalar(scalar)	58,419✔
37	}	58,419✔
38
39	#[derive(Debug, Clone, PartialEq, Eq)]
40	pub struct MinMaxResult {
41	pub min: Scalar,
42	pub max: Scalar,
43	}
44
45	impl MinMaxResult {
46	pub fn from_scalar(scalar: Scalar) -> VortexResult<Option<Self>> {	115,866✔
47	if scalar.is_null() {	115,866✔
48	Ok(None)	11,282✔
49	} else {
50	let min = scalar	104,584✔
51	.as_struct()	104,584✔
52	.field_by_idx(0)	104,584✔
53	.vortex_expect("missing min field");	104,584✔
54	let max = scalar	104,584✔
55	.as_struct()	104,584✔
56	.field_by_idx(1)	104,584✔
57	.vortex_expect("missing max field");	104,584✔
58	Ok(Some(MinMaxResult { min, max }))	104,584✔
59	}
60	}	115,866✔
61	}
62
63	pub struct MinMax;
64
65	impl ComputeFnVTable for MinMax {
66	fn invoke(	58,419✔
67	&self,	58,419✔
68	args: &InvocationArgs,	58,419✔
69	kernels: &[ArcRef<dyn Kernel>],	58,419✔
70	) -> VortexResult<Output> {	58,419✔
71	let UnaryArgs { array, .. } = UnaryArgs::<()>::try_from(args)?;	58,419✔
72
73	let return_dtype = self.return_dtype(args)?;	58,419✔
74
75	match min_max_impl(array, kernels)? {	58,419✔
76	None => Ok(Scalar::null(return_dtype).into()),	5,870✔
77	Some(MinMaxResult { min, max }) => {	52,549✔
78	assert!(	52,549✔
79	min <= max,	52,549✔
UNCOV 80	"min > max: min={} max={} encoding={}",	×
81	min,
82	max,
UNCOV 83	array.encoding_id()	×
84	);
85
86	// Update the stats set with the computed min/max
87	array	52,549✔
88	.statistics()	52,549✔
89	.set(Stat::Min, Precision::Exact(min.value().clone()));	52,549✔
90	array	52,549✔
91	.statistics()	52,549✔
92	.set(Stat::Max, Precision::Exact(max.value().clone()));	52,549✔
93
94	// Return the min/max as a struct scalar
95	Ok(Scalar::struct_(return_dtype, vec![min, max]).into())	52,549✔
96	}
97	}
98	}	58,419✔
99
100	fn return_dtype(&self, args: &InvocationArgs) -> VortexResult<DType> {	116,838✔
101	let UnaryArgs { array, .. } = UnaryArgs::<()>::try_from(args)?;	116,838✔
102
103	// We return a min/max struct scalar, where the overall struct is nullable in the case
104	// that the array is all null or empty.
105	Ok(DType::Struct(	116,838✔
106	StructFields::new(	116,838✔
107	["min", "max"].into(),	116,838✔
108	vec![array.dtype().clone(), array.dtype().clone()],	116,838✔
109	),	116,838✔
110	Nullability::Nullable,	116,838✔
111	))	116,838✔
112	}	116,838✔
113
114	fn return_len(&self, _args: &InvocationArgs) -> VortexResult<usize> {	58,419✔
115	Ok(1)	58,419✔
116	}	58,419✔
117
118	fn is_elementwise(&self) -> bool {	58,419✔
119	false	58,419✔
120	}	58,419✔
121	}
122
123	fn min_max_impl(	58,419✔
124	array: &dyn Array,	58,419✔
125	kernels: &[ArcRef<dyn Kernel>],	58,419✔
126	) -> VortexResult<Option<MinMaxResult>> {	58,419✔
127	if array.is_empty() \|\| array.valid_count()? == 0 {	58,419✔
128	return Ok(None);	458✔
129	}	57,961✔
130
131	let min = array	57,961✔
132	.statistics()	57,961✔
133	.get_scalar(Stat::Min, array.dtype())	57,961✔
134	.and_then(Precision::as_exact);	57,961✔
135	let max = array	57,961✔
136	.statistics()	57,961✔
137	.get_scalar(Stat::Max, array.dtype())	57,961✔
138	.and_then(Precision::as_exact);	57,961✔
139
140	if let Some((min, max)) = min.zip(max) {	57,961✔
141	return Ok(Some(MinMaxResult { min, max }));	438✔
142	}	57,523✔
143
144	let args = InvocationArgs {	57,523✔
145	inputs: &[array.into()],	57,523✔
146	options: &(),	57,523✔
147	};	57,523✔
148	for kernel in kernels {	205,384✔
149	if let Some(output) = kernel.invoke(&args)? {	205,308✔
150	return MinMaxResult::from_scalar(output.unwrap_scalar()?);	57,447✔
151	}	147,861✔
152	}
153	if let Some(output) = array.invoke(&MIN_MAX_FN, &args)? {	76✔
154	return MinMaxResult::from_scalar(output.unwrap_scalar()?);	×
155	}	76✔
156
157	if !array.is_canonical() {	76✔
158	let array = array.to_canonical()?;	76✔
159	return min_max(array.as_ref());	76✔
UNCOV 160	}	×
161
UNCOV 162	vortex_bail!(NotImplemented: "min_max", array.encoding_id());	×
163	}	58,419✔
164
165	/// The minimum and maximum non-null values of an array, or None if there are no non-null values.
166	pub trait MinMaxKernel: VTable {
167	fn min_max(&self, array: &Self::Array) -> VortexResult<Option<MinMaxResult>>;
168	}
169
170	pub struct MinMaxKernelRef(ArcRef<dyn Kernel>);
171	inventory::collect!(MinMaxKernelRef);
172
173	#[derive(Debug)]
174	pub struct MinMaxKernelAdapter<V: VTable>(pub V);
175
176	impl<V: VTable + MinMaxKernel> MinMaxKernelAdapter<V> {
UNCOV 177	pub const fn lift(&'static self) -> MinMaxKernelRef {	×
UNCOV 178	MinMaxKernelRef(ArcRef::new_ref(self))	×
UNCOV 179	}	×
180	}
181
182	impl<V: VTable + MinMaxKernel> Kernel for MinMaxKernelAdapter<V> {
183	fn invoke(&self, args: &InvocationArgs) -> VortexResult<Option<Output>> {	205,226✔
184	let inputs = UnaryArgs::<()>::try_from(args)?;	205,226✔
185	let Some(array) = inputs.array.as_opt::<V>() else {	205,226✔
186	return Ok(None);	147,816✔
187	};
188	let dtype = DType::Struct(	57,410✔
189	StructFields::new(	57,410✔
190	["min", "max"].into(),	57,410✔
191	vec![array.dtype().clone(), array.dtype().clone()],	57,410✔
192	),	57,410✔
193	Nullability::Nullable,	57,410✔
194	);	57,410✔
195	Ok(Some(match V::min_max(&self.0, array)? {	57,410✔
196	None => Scalar::null(dtype).into(),	5,412✔
197	Some(MinMaxResult { min, max }) => Scalar::struct_(dtype, vec![min, max]).into(),	51,998✔
198	}))
199	}	205,226✔
200	}
201
202	#[cfg(test)]
203	mod tests {
204	use arrow_buffer::BooleanBuffer;
205	use vortex_buffer::buffer;
206
207	use crate::arrays::{BoolArray, NullArray, PrimitiveArray};
208	use crate::compute::{MinMaxResult, min_max};
209	use crate::validity::Validity;
210
211	#[test]
212	fn test_prim_max() {	1✔
213	let p = PrimitiveArray::new(buffer![1, 2, 3], Validity::NonNullable);	1✔
214	assert_eq!(	1✔
215	min_max(p.as_ref()).unwrap(),	1✔
216	Some(MinMaxResult {	1✔
217	min: 1.into(),	1✔
218	max: 3.into()	1✔
219	})	1✔
220	);
221	}	1✔
222
223	#[test]
224	fn test_bool_max() {	1✔
225	let p = BoolArray::new(	1✔
226	BooleanBuffer::from([true, true, true].as_slice()),	1✔
227	Validity::NonNullable,	1✔
228	);
229	assert_eq!(	1✔
230	min_max(p.as_ref()).unwrap(),	1✔
231	Some(MinMaxResult {	1✔
232	min: true.into(),	1✔
233	max: true.into()	1✔
234	})	1✔
235	);
236
237	let p = BoolArray::new(	1✔
238	BooleanBuffer::from([false, false, false].as_slice()),	1✔
239	Validity::NonNullable,	1✔
240	);
241	assert_eq!(	1✔
242	min_max(p.as_ref()).unwrap(),	1✔
243	Some(MinMaxResult {	1✔
244	min: false.into(),	1✔
245	max: false.into()	1✔
246	})	1✔
247	);
248
249	let p = BoolArray::new(	1✔
250	BooleanBuffer::from([false, true, false].as_slice()),	1✔
251	Validity::NonNullable,	1✔
252	);
253	assert_eq!(	1✔
254	min_max(p.as_ref()).unwrap(),	1✔
255	Some(MinMaxResult {	1✔
256	min: false.into(),	1✔
257	max: true.into()	1✔
258	})	1✔
259	);
260	}	1✔
261
262	#[test]
263	fn test_null() {	1✔
264	let p = NullArray::new(1);	1✔
265	assert_eq!(min_max(p.as_ref()).unwrap(), None);	1✔
266	}	1✔
267	}

vortex-data / vortex / 16593958537

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