vortex_array/compute/
mask.rs

1// SPDX-License-Identifier: Apache-2.0
2// SPDX-FileCopyrightText: Copyright the Vortex contributors
3
4use std::sync::LazyLock;
5
6use arcref::ArcRef;
7use arrow_array::BooleanArray;
8use vortex_dtype::DType;
9use vortex_error::{VortexError, VortexResult, vortex_bail, vortex_err};
10use vortex_mask::Mask;
11use vortex_scalar::Scalar;
12
13use crate::arrays::ConstantArray;
14use crate::arrow::{FromArrowArray, IntoArrowArray};
15use crate::compute::{ComputeFn, ComputeFnVTable, InvocationArgs, Kernel, Output, cast};
16use crate::vtable::VTable;
17use crate::{Array, ArrayRef, IntoArray};
18
19static MASK_FN: LazyLock<ComputeFn> = LazyLock::new(|| {
20    let compute = ComputeFn::new("mask".into(), ArcRef::new_ref(&MaskFn));
21    for kernel in inventory::iter::<MaskKernelRef> {
22        compute.register_kernel(kernel.0.clone());
23    }
24    compute
25});
26
27/// Replace values with null where the mask is true.
28///
29/// The returned array is nullable but otherwise has the same dtype and length as `array`.
30///
31/// # Examples
32///
33/// ```
34/// use vortex_array::IntoArray;
35/// use vortex_array::arrays::{BoolArray, PrimitiveArray};
36/// use vortex_array::compute::{ mask};
37/// use vortex_mask::Mask;
38/// use vortex_scalar::Scalar;
39///
40/// let array =
41///     PrimitiveArray::from_option_iter([Some(0i32), None, Some(1i32), None, Some(2i32)]);
42/// let mask_array = Mask::try_from(
43///     &BoolArray::from_iter([true, false, false, false, true]),
44/// )
45/// .unwrap();
46///
47/// let masked = mask(array.as_ref(), &mask_array).unwrap();
48/// assert_eq!(masked.len(), 5);
49/// assert!(!masked.is_valid(0).unwrap());
50/// assert!(!masked.is_valid(1).unwrap());
51/// assert_eq!(masked.scalar_at(2), Scalar::from(Some(1)));
52/// assert!(!masked.is_valid(3).unwrap());
53/// assert!(!masked.is_valid(4).unwrap());
54/// ```
55///
56pub fn mask(array: &dyn Array, mask: &Mask) -> VortexResult<ArrayRef> {
57    MASK_FN
58        .invoke(&InvocationArgs {
59            inputs: &[array.into(), mask.into()],
60            options: &(),
61        })?
62        .unwrap_array()
63}
64
65pub struct MaskKernelRef(ArcRef<dyn Kernel>);
66inventory::collect!(MaskKernelRef);
67
68pub trait MaskKernel: VTable {
69    /// Replace masked values with null in array.
70    fn mask(&self, array: &Self::Array, mask: &Mask) -> VortexResult<ArrayRef>;
71}
72
73#[derive(Debug)]
74pub struct MaskKernelAdapter<V: VTable>(pub V);
75
76impl<V: VTable + MaskKernel> MaskKernelAdapter<V> {
77    pub const fn lift(&'static self) -> MaskKernelRef {
78        MaskKernelRef(ArcRef::new_ref(self))
79    }
80}
81
82impl<V: VTable + MaskKernel> Kernel for MaskKernelAdapter<V> {
83    fn invoke(&self, args: &InvocationArgs) -> VortexResult<Option<Output>> {
84        let inputs = MaskArgs::try_from(args)?;
85        let Some(array) = inputs.array.as_opt::<V>() else {
86            return Ok(None);
87        };
88        Ok(Some(V::mask(&self.0, array, inputs.mask)?.into()))
89    }
90}
91
92struct MaskFn;
93
94impl ComputeFnVTable for MaskFn {
95    fn invoke(
96        &self,
97        args: &InvocationArgs,
98        kernels: &[ArcRef<dyn Kernel>],
99    ) -> VortexResult<Output> {
100        let MaskArgs { array, mask } = MaskArgs::try_from(args)?;
101
102        if matches!(mask, Mask::AllFalse(_)) {
103            // Fast-path for empty mask
104            return Ok(cast(array, &array.dtype().as_nullable())?.into());
105        }
106
107        if matches!(mask, Mask::AllTrue(_)) {
108            // Fast-path for full mask.
109            return Ok(ConstantArray::new(
110                Scalar::null(array.dtype().clone().as_nullable()),
111                array.len(),
112            )
113            .into_array()
114            .into());
115        }
116
117        for kernel in kernels {
118            if let Some(output) = kernel.invoke(args)? {
119                return Ok(output);
120            }
121        }
122        if let Some(output) = array.invoke(&MASK_FN, args)? {
123            return Ok(output);
124        }
125
126        // Fallback: implement using Arrow kernels.
127        log::debug!("No mask implementation found for {}", array.encoding_id());
128
129        let array_ref = array.to_array().into_arrow_preferred()?;
130        let mask = BooleanArray::new(mask.to_boolean_buffer(), None);
131
132        let masked = arrow_select::nullif::nullif(array_ref.as_ref(), &mask)?;
133
134        Ok(ArrayRef::from_arrow(masked.as_ref(), true).into())
135    }
136
137    fn return_dtype(&self, args: &InvocationArgs) -> VortexResult<DType> {
138        let MaskArgs { array, .. } = MaskArgs::try_from(args)?;
139        Ok(array.dtype().as_nullable())
140    }
141
142    fn return_len(&self, args: &InvocationArgs) -> VortexResult<usize> {
143        let MaskArgs { array, mask } = MaskArgs::try_from(args)?;
144
145        if mask.len() != array.len() {
146            vortex_bail!(
147                "mask.len() is {}, does not equal array.len() of {}",
148                mask.len(),
149                array.len()
150            );
151        }
152
153        Ok(mask.len())
154    }
155
156    fn is_elementwise(&self) -> bool {
157        true
158    }
159}
160
161struct MaskArgs<'a> {
162    array: &'a dyn Array,
163    mask: &'a Mask,
164}
165
166impl<'a> TryFrom<&InvocationArgs<'a>> for MaskArgs<'a> {
167    type Error = VortexError;
168
169    fn try_from(value: &InvocationArgs<'a>) -> Result<Self, Self::Error> {
170        if value.inputs.len() != 2 {
171            vortex_bail!("Mask function requires 2 arguments");
172        }
173        let array = value.inputs[0]
174            .array()
175            .ok_or_else(|| vortex_err!("Expected input 0 to be an array"))?;
176        let mask = value.inputs[1]
177            .mask()
178            .ok_or_else(|| vortex_err!("Expected input 1 to be a mask"))?;
179
180        Ok(MaskArgs { array, mask })
181    }
182}
vortex_array/compute/mask.rs

vortex_array/compute/
mask.rs
