16581101742

Committed 28 Jul 2025 09:48PM UTC coverage: 82.251% (-0.02%) from 82.268%

Build # 16581101742

Build Type

push

github

Committed by

web-flow

Commit Message

zip compute function (#4031)

Signed-off-by: Onur Satici <onur@spiraldb.com>

Run Details

106 of 140 new or added lines in 1 file covered. (75.71%)

2 existing lines in 1 file now uncovered.

44340 of 53908 relevant lines covered (82.25%)

170128.99 hits per line

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

75.71

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

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

use std::sync::LazyLock;

use arcref::ArcRef;
use vortex_dtype::DType;
use vortex_error::{VortexError, VortexResult, vortex_bail, vortex_err};
use vortex_mask::{AllOr, Mask};

use super::{ComputeFnVTable, InvocationArgs, Output, cast};
use crate::builders::builder_with_capacity;
use crate::compute::{ComputeFn, Kernel};
use crate::vtable::VTable;
use crate::{Array, ArrayRef};

/// Performs element-wise conditional selection between two arrays based on a mask.
///
/// Returns a new array where `result[i] = if_true[i]` when `mask[i]` is true,
/// otherwise `result[i] = if_false[i]`.
pub fn zip(mask: &Mask, if_true: &dyn Array, if_false: &dyn Array) -> VortexResult<ArrayRef> {
    ZIP_FN
        .invoke(&InvocationArgs {
            inputs: &[mask.into(), if_true.into(), if_false.into()],
            options: &(),
        })?
        .unwrap_array()
}

pub static ZIP_FN: LazyLock<ComputeFn> = LazyLock::new(|| {
    let compute = ComputeFn::new("zip".into(), ArcRef::new_ref(&Zip));
    for kernel in inventory::iter::<ZipKernelRef> {
        compute.register_kernel(kernel.0.clone());
    }
    compute
});

struct Zip;

impl ComputeFnVTable for Zip {
    fn invoke(
        &self,
        args: &InvocationArgs,
        kernels: &[ArcRef<dyn Kernel>],
    ) -> VortexResult<Output> {
        let ZipArgs {
            mask,
            if_true,
            if_false,
        } = ZipArgs::try_from(args)?;

        if mask.all_true() {
            return Ok(cast(if_true, &zip_return_dtype(if_true, if_false))?.into());
        }

        if mask.all_false() {
            return Ok(cast(if_false, &zip_return_dtype(if_true, if_false))?.into());
        }

        if if_true.is_canonical() && if_false.is_canonical() {
            // skip kernel lookup if both arrays are canonical
            return Ok(zip_impl(mask, if_true, if_false)?.into());
        }

        // check if if_true supports zip directly
        for kernel in kernels {
            if let Some(output) = kernel.invoke(args)? {
                return Ok(output);
            }
        }

        if let Some(output) = if_true.invoke(&ZIP_FN, args)? {
            return Ok(output);
        }

        // TODO(os): add invert_mask opt and check if if_false has a kernel like:
        //           kernel.invoke(Args(if_false, if_true, mask, invert_mask = true))

        Ok(zip_impl(
            mask,
            if_true.to_canonical()?.as_ref(),
            if_false.to_canonical()?.as_ref(),
        )?
        .into())
    }

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

        if !if_true.dtype().eq_ignore_nullability(if_false.dtype()) {
            vortex_bail!("input arrays to zip must have the same dtype");
        }
        Ok(zip_return_dtype(if_true, if_false))
    }

    fn return_len(&self, args: &InvocationArgs) -> VortexResult<usize> {
        let ZipArgs { if_true, .. } = ZipArgs::try_from(args)?;
        // ComputeFn::invoke asserts if_true.len() == if_false.len(), because zip is elementwise
        Ok(if_true.len())
    }

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

struct ZipArgs<'a> {
    mask: &'a Mask,
    if_true: &'a dyn Array,
    if_false: &'a dyn Array,
}

impl<'a> TryFrom<&InvocationArgs<'a>> for ZipArgs<'a> {
    type Error = VortexError;

    fn try_from(value: &InvocationArgs<'a>) -> Result<Self, Self::Error> {
        if value.inputs.len() != 3 {
            vortex_bail!("Expected 3 inputs for zip, found {}", value.inputs.len());
        }
        let mask = value.inputs[0]
            .mask()
            .ok_or_else(|| vortex_err!("Expected input 0 to be a mask"))?;

        let if_true = value.inputs[1]
            .array()
            .ok_or_else(|| vortex_err!("Expected input 1 to be an array"))?;

        let if_false = value.inputs[2]
            .array()
            .ok_or_else(|| vortex_err!("Expected input 2 to be an array"))?;
        Ok(Self {
            mask,
            if_true,
            if_false,
        })
    }
}

pub trait ZipKernel: VTable {
    fn zip(
        &self,
        mask: &Mask,
        if_true: &Self::Array,
        if_false: &dyn Array,
    ) -> VortexResult<ArrayRef>;
}

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

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

impl<V: VTable + ZipKernel> ZipKernelAdapter<V> {
    pub const fn lift(&'static self) -> ZipKernelRef {
        ZipKernelRef(ArcRef::new_ref(self))
    }
}

impl<V: VTable + ZipKernel> Kernel for ZipKernelAdapter<V> {
    fn invoke(&self, args: &InvocationArgs) -> VortexResult<Option<Output>> {
        let ZipArgs {
            mask,
            if_true,
            if_false,
        } = ZipArgs::try_from(args)?;
        let Some(if_true) = if_true.as_opt::<V>() else {
            return Ok(None);
        };
        Ok(Some(V::zip(&self.0, mask, if_true, if_false)?.into()))
    }
}

fn zip_return_dtype(if_true: &dyn Array, if_false: &dyn Array) -> DType {
    if_true
        .dtype()
        .union_nullability(if_false.dtype().nullability())
}

fn zip_impl(mask: &Mask, if_true: &dyn Array, if_false: &dyn Array) -> VortexResult<ArrayRef> {
    // if_true.len() == if_false.len() from ComputeFn::invoke
    let mut builder = builder_with_capacity(&zip_return_dtype(if_true, if_false), if_true.len());

    match mask.slices() {
        AllOr::All => Ok(if_true.to_array()),
        AllOr::None => Ok(if_false.to_array()),
        AllOr::Some(slices) => {
            for (start, end) in slices {
                builder.extend_from_array(&if_false.slice(builder.len(), *start)?)?;
                builder.extend_from_array(&if_true.slice(*start, *end)?)?;
            }
            if builder.len() < if_false.len() {
                builder.extend_from_array(&if_false.slice(builder.len(), if_false.len())?)?;
            }
            Ok(builder.finish())
        }
    }
}

#[cfg(test)]
mod tests {
    use vortex_array::arrays::{BoolArray, PrimitiveArray};
    use vortex_array::compute::zip;
    use vortex_array::{IntoArray, ToCanonical};
    use vortex_mask::Mask;

    #[test]
    fn test_zip_basic() {
        let mask =
            Mask::try_from(&BoolArray::from_iter([true, false, false, true, false])).unwrap();
        let if_true = PrimitiveArray::from_iter([10, 20, 30, 40, 50]).into_array();
        let if_false = PrimitiveArray::from_iter([1, 2, 3, 4, 5]).into_array();

        let result = zip(&mask, &if_true, &if_false).unwrap();
        let expected = PrimitiveArray::from_iter([10, 2, 3, 40, 5]);

        assert_eq!(
            result.to_primitive().unwrap().as_slice::<i32>(),
            expected.as_slice::<i32>()
        );
    }

    #[test]
    fn test_zip_all_true() {
        let mask = Mask::new_true(4);
        let if_true = PrimitiveArray::from_iter([10, 20, 30, 40]).into_array();
        let if_false =
            PrimitiveArray::from_option_iter([Some(1), Some(2), Some(3), None]).into_array();

        let result = zip(&mask, &if_true, &if_false).unwrap();

        assert_eq!(
            result.to_primitive().unwrap().as_slice::<i32>(),
            if_true.to_primitive().unwrap().as_slice::<i32>()
        );

        // result must be nullable even if_true was not
        assert_eq!(result.dtype(), if_false.dtype())
    }

    #[test]
    #[should_panic]
    fn test_invalid_lengths() {
        let mask = Mask::new_false(4);
        let if_true = PrimitiveArray::from_iter([10, 20, 30]).into_array();
        let if_false = PrimitiveArray::from_iter([1, 2, 3, 4]).into_array();

        zip(&mask, &if_true, &if_false).unwrap();
    }
}

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;
8	use vortex_error::{VortexError, VortexResult, vortex_bail, vortex_err};
9	use vortex_mask::{AllOr, Mask};
10
11	use super::{ComputeFnVTable, InvocationArgs, Output, cast};
12	use crate::builders::builder_with_capacity;
13	use crate::compute::{ComputeFn, Kernel};
14	use crate::vtable::VTable;
15	use crate::{Array, ArrayRef};
16
17	/// Performs element-wise conditional selection between two arrays based on a mask.
18	///
19	/// Returns a new array where `result[i] = if_true[i]` when `mask[i]` is true,
20	/// otherwise `result[i] = if_false[i]`.
21	pub fn zip(mask: &Mask, if_true: &dyn Array, if_false: &dyn Array) -> VortexResult<ArrayRef> {	114✔
22	ZIP_FN	114✔
23	.invoke(&InvocationArgs {	114✔
24	inputs: &[mask.into(), if_true.into(), if_false.into()],	114✔
25	options: &(),	114✔
26	})?	114✔
27	.unwrap_array()	76✔
28	}	114✔
29
30	pub static ZIP_FN: LazyLock<ComputeFn> = LazyLock::new(\|\| {	114✔
31	let compute = ComputeFn::new("zip".into(), ArcRef::new_ref(&Zip));	114✔
32	for kernel in inventory::iter::<ZipKernelRef> {	114✔
NEW 33	compute.register_kernel(kernel.0.clone());	×
NEW 34	}	×
35	compute	114✔
36	});	114✔
37
38	struct Zip;
39
40	impl ComputeFnVTable for Zip {
41	fn invoke(	76✔
42	&self,	76✔
43	args: &InvocationArgs,	76✔
44	kernels: &[ArcRef<dyn Kernel>],	76✔
45	) -> VortexResult<Output> {	76✔
46	let ZipArgs {
47	mask,	76✔
48	if_true,	76✔
49	if_false,	76✔
50	} = ZipArgs::try_from(args)?;	76✔
51
52	if mask.all_true() {	76✔
53	return Ok(cast(if_true, &zip_return_dtype(if_true, if_false))?.into());	38✔
54	}	38✔
55
56	if mask.all_false() {	38✔
NEW 57	return Ok(cast(if_false, &zip_return_dtype(if_true, if_false))?.into());	×
58	}	38✔
59
60	if if_true.is_canonical() && if_false.is_canonical() {	38✔
61	// skip kernel lookup if both arrays are canonical
62	return Ok(zip_impl(mask, if_true, if_false)?.into());	38✔
NEW 63	}	×
64
65	// check if if_true supports zip directly
NEW 66	for kernel in kernels {	×
NEW 67	if let Some(output) = kernel.invoke(args)? {	×
NEW 68	return Ok(output);	×
NEW 69	}	×
70	}
71
NEW 72	if let Some(output) = if_true.invoke(&ZIP_FN, args)? {	×
NEW 73	return Ok(output);	×
NEW 74	}	×
75
76	// TODO(os): add invert_mask opt and check if if_false has a kernel like:
77	// kernel.invoke(Args(if_false, if_true, mask, invert_mask = true))
78
NEW 79	Ok(zip_impl(	×
NEW 80	mask,	×
NEW 81	if_true.to_canonical()?.as_ref(),	×
NEW 82	if_false.to_canonical()?.as_ref(),	×
NEW 83	)?	×
NEW 84	.into())	×
85	}	76✔
86
87	fn return_dtype(&self, args: &InvocationArgs) -> VortexResult<DType> {	76✔
88	let ZipArgs {
89	if_true, if_false, ..	76✔
90	} = ZipArgs::try_from(args)?;	76✔
91
92	if !if_true.dtype().eq_ignore_nullability(if_false.dtype()) {	76✔
NEW 93	vortex_bail!("input arrays to zip must have the same dtype");	×
94	}	76✔
95	Ok(zip_return_dtype(if_true, if_false))	76✔
96	}	76✔
97
98	fn return_len(&self, args: &InvocationArgs) -> VortexResult<usize> {	76✔
99	let ZipArgs { if_true, .. } = ZipArgs::try_from(args)?;	76✔
100	// ComputeFn::invoke asserts if_true.len() == if_false.len(), because zip is elementwise
101	Ok(if_true.len())	76✔
102	}	76✔
103
104	fn is_elementwise(&self) -> bool {	114✔
105	true	114✔
106	}	114✔
107	}
108
109	struct ZipArgs<'a> {
110	mask: &'a Mask,
111	if_true: &'a dyn Array,
112	if_false: &'a dyn Array,
113	}
114
115	impl<'a> TryFrom<&InvocationArgs<'a>> for ZipArgs<'a> {
116	type Error = VortexError;
117
118	fn try_from(value: &InvocationArgs<'a>) -> Result<Self, Self::Error> {	228✔
119	if value.inputs.len() != 3 {	228✔
NEW 120	vortex_bail!("Expected 3 inputs for zip, found {}", value.inputs.len());	×
121	}	228✔
122	let mask = value.inputs[0]	228✔
123	.mask()	228✔
124	.ok_or_else(\|\| vortex_err!("Expected input 0 to be a mask"))?;	228✔
125
126	let if_true = value.inputs[1]	228✔
127	.array()	228✔
128	.ok_or_else(\|\| vortex_err!("Expected input 1 to be an array"))?;	228✔
129
130	let if_false = value.inputs[2]	228✔
131	.array()	228✔
132	.ok_or_else(\|\| vortex_err!("Expected input 2 to be an array"))?;	228✔
133	Ok(Self {	228✔
134	mask,	228✔
135	if_true,	228✔
136	if_false,	228✔
137	})	228✔
138	}	228✔
139	}
140
141	pub trait ZipKernel: VTable {
142	fn zip(
143	&self,
144	mask: &Mask,
145	if_true: &Self::Array,
146	if_false: &dyn Array,
147	) -> VortexResult<ArrayRef>;
148	}
149
150	pub struct ZipKernelRef(pub ArcRef<dyn Kernel>);
151	inventory::collect!(ZipKernelRef);
152
153	#[derive(Debug)]
154	pub struct ZipKernelAdapter<V: VTable>(pub V);
155
156	impl<V: VTable + ZipKernel> ZipKernelAdapter<V> {
NEW 157	pub const fn lift(&'static self) -> ZipKernelRef {	×
NEW 158	ZipKernelRef(ArcRef::new_ref(self))	×
NEW 159	}	×
160	}
161
162	impl<V: VTable + ZipKernel> Kernel for ZipKernelAdapter<V> {
NEW 163	fn invoke(&self, args: &InvocationArgs) -> VortexResult<Option<Output>> {	×
164	let ZipArgs {
NEW 165	mask,	×
NEW 166	if_true,	×
NEW 167	if_false,	×
NEW 168	} = ZipArgs::try_from(args)?;	×
NEW 169	let Some(if_true) = if_true.as_opt::<V>() else {	×
NEW 170	return Ok(None);	×
171	};
NEW 172	Ok(Some(V::zip(&self.0, mask, if_true, if_false)?.into()))	×
NEW 173	}	×
174	}
175
176	fn zip_return_dtype(if_true: &dyn Array, if_false: &dyn Array) -> DType {	152✔
177	if_true	152✔
178	.dtype()	152✔
179	.union_nullability(if_false.dtype().nullability())	152✔
180	}	152✔
181
182	fn zip_impl(mask: &Mask, if_true: &dyn Array, if_false: &dyn Array) -> VortexResult<ArrayRef> {	38✔
183	// if_true.len() == if_false.len() from ComputeFn::invoke
184	let mut builder = builder_with_capacity(&zip_return_dtype(if_true, if_false), if_true.len());	38✔
185
186	match mask.slices() {	38✔
NEW 187	AllOr::All => Ok(if_true.to_array()),	×
NEW 188	AllOr::None => Ok(if_false.to_array()),	×
189	AllOr::Some(slices) => {	38✔
190	for (start, end) in slices {	114✔
191	builder.extend_from_array(&if_false.slice(builder.len(), *start)?)?;	76✔
192	builder.extend_from_array(&if_true.slice(start, end)?)?;	76✔
193	}
194	if builder.len() < if_false.len() {	38✔
195	builder.extend_from_array(&if_false.slice(builder.len(), if_false.len())?)?;	38✔
NEW 196	}	×
197	Ok(builder.finish())	38✔
198	}
199	}
200	}	38✔
201
202	#[cfg(test)]
203	mod tests {
204	use vortex_array::arrays::{BoolArray, PrimitiveArray};
205	use vortex_array::compute::zip;
206	use vortex_array::{IntoArray, ToCanonical};
207	use vortex_mask::Mask;
208
209	#[test]
210	fn test_zip_basic() {	1✔
211	let mask =	1✔
212	Mask::try_from(&BoolArray::from_iter([true, false, false, true, false])).unwrap();	1✔
213	let if_true = PrimitiveArray::from_iter([10, 20, 30, 40, 50]).into_array();	1✔
214	let if_false = PrimitiveArray::from_iter([1, 2, 3, 4, 5]).into_array();	1✔
215
216	let result = zip(&mask, &if_true, &if_false).unwrap();	1✔
217	let expected = PrimitiveArray::from_iter([10, 2, 3, 40, 5]);	1✔
218
219	assert_eq!(	1✔
220	result.to_primitive().unwrap().as_slice::<i32>(),	1✔
221	expected.as_slice::<i32>()	1✔
222	);
223	}	1✔
224
225	#[test]
226	fn test_zip_all_true() {	1✔
227	let mask = Mask::new_true(4);	1✔
228	let if_true = PrimitiveArray::from_iter([10, 20, 30, 40]).into_array();	1✔
229	let if_false =	1✔
230	PrimitiveArray::from_option_iter([Some(1), Some(2), Some(3), None]).into_array();	1✔
231
232	let result = zip(&mask, &if_true, &if_false).unwrap();	1✔
233
234	assert_eq!(	1✔
235	result.to_primitive().unwrap().as_slice::<i32>(),	1✔
236	if_true.to_primitive().unwrap().as_slice::<i32>()	1✔
237	);
238
239	// result must be nullable even if_true was not
240	assert_eq!(result.dtype(), if_false.dtype())	1✔
241	}	1✔
242
243	#[test]
244	#[should_panic]
245	fn test_invalid_lengths() {	1✔
246	let mask = Mask::new_false(4);	1✔
247	let if_true = PrimitiveArray::from_iter([10, 20, 30]).into_array();	1✔
248	let if_false = PrimitiveArray::from_iter([1, 2, 3, 4]).into_array();	1✔
249
250	zip(&mask, &if_true, &if_false).unwrap();	1✔
251	}	1✔
252	}

vortex-data / vortex / 16581101742

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