16473770414

Committed 23 Jul 2025 02:38PM UTC coverage: 80.988% (-0.07%) from 81.055%

Build # 16473770414

Build Type

push

github

Committed by

web-flow

Commit Message

chore[duckdb]: scan log info -> trace (#3989)

Signed-off-by: Joe Isaacs <joe.isaacs@live.co.uk>

Run Details

1 of 1 new or added line in 1 file covered. (100.0%)

35 existing lines in 7 files now uncovered.

42052 of 51924 relevant lines covered (80.99%)

173623.22 hits per line

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

82.85

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

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

use core::fmt;
use std::any::Any;
use std::fmt::{Display, Formatter};
use std::sync::LazyLock;

use arcref::ArcRef;
use arrow_buffer::BooleanBuffer;
use arrow_ord::cmp;
use vortex_dtype::{DType, NativePType, Nullability};
use vortex_error::{VortexError, VortexExpect, VortexResult, vortex_bail, vortex_err};
use vortex_scalar::Scalar;

use crate::arrays::ConstantArray;
use crate::arrow::{Datum, from_arrow_array_with_len};
use crate::compute::{ComputeFn, ComputeFnVTable, InvocationArgs, Kernel, Options, Output};
use crate::vtable::VTable;
use crate::{Array, ArrayRef, Canonical, IntoArray};

/// Compares two arrays and returns a new boolean array with the result of the comparison.
/// Or, returns None if comparison is not supported for these arrays.
pub fn compare(left: &dyn Array, right: &dyn Array, operator: Operator) -> VortexResult<ArrayRef> {
    COMPARE_FN
        .invoke(&InvocationArgs {
            inputs: &[left.into(), right.into()],
            options: &operator,
        })?
        .unwrap_array()
}

#[derive(Copy, Clone, Debug, Eq, PartialEq, PartialOrd, Hash)]
pub enum Operator {
    Eq,
    NotEq,
    Gt,
    Gte,
    Lt,
    Lte,
}

impl Display for Operator {
    fn fmt(&self, f: &mut Formatter) -> fmt::Result {
        let display = match &self {
            Operator::Eq => "=",
            Operator::NotEq => "!=",
            Operator::Gt => ">",
            Operator::Gte => ">=",
            Operator::Lt => "<",
            Operator::Lte => "<=",
        };
        Display::fmt(display, f)
    }
}

impl Operator {
    pub fn inverse(self) -> Self {
        match self {
            Operator::Eq => Operator::NotEq,
            Operator::NotEq => Operator::Eq,
            Operator::Gt => Operator::Lte,
            Operator::Gte => Operator::Lt,
            Operator::Lt => Operator::Gte,
            Operator::Lte => Operator::Gt,
        }
    }

    /// Change the sides of the operator, where changing lhs and rhs won't change the result of the operation
    pub fn swap(self) -> Self {
        match self {
            Operator::Eq => Operator::Eq,
            Operator::NotEq => Operator::NotEq,
            Operator::Gt => Operator::Lt,
            Operator::Gte => Operator::Lte,
            Operator::Lt => Operator::Gt,
            Operator::Lte => Operator::Gte,
        }
    }
}

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

pub trait CompareKernel: VTable {
    fn compare(
        &self,
        lhs: &Self::Array,
        rhs: &dyn Array,
        operator: Operator,
    ) -> VortexResult<Option<ArrayRef>>;
}

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

impl<V: VTable + CompareKernel> CompareKernelAdapter<V> {
    pub const fn lift(&'static self) -> CompareKernelRef {
        CompareKernelRef(ArcRef::new_ref(self))
    }
}

impl<V: VTable + CompareKernel> Kernel for CompareKernelAdapter<V> {
    fn invoke(&self, args: &InvocationArgs) -> VortexResult<Option<Output>> {
        let inputs = CompareArgs::try_from(args)?;
        let Some(array) = inputs.lhs.as_opt::<V>() else {
            return Ok(None);
        };
        Ok(V::compare(&self.0, array, inputs.rhs, inputs.operator)?.map(|array| array.into()))
    }
}

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

struct Compare;

impl ComputeFnVTable for Compare {
    fn invoke(
        &self,
        args: &InvocationArgs,
        kernels: &[ArcRef<dyn Kernel>],
    ) -> VortexResult<Output> {
        let CompareArgs { lhs, rhs, operator } = CompareArgs::try_from(args)?;

        let return_dtype = self.return_dtype(args)?;

        if lhs.is_empty() {
            return Ok(Canonical::empty(&return_dtype).into_array().into());
        }

        let left_constant_null = lhs.as_constant().map(|l| l.is_null()).unwrap_or(false);
        let right_constant_null = rhs.as_constant().map(|r| r.is_null()).unwrap_or(false);
        if left_constant_null || right_constant_null {
            return Ok(ConstantArray::new(Scalar::null(return_dtype), lhs.len())
                .into_array()
                .into());
        }

        let right_is_constant = rhs.is_constant();

        // Always try to put constants on the right-hand side so encodings can optimise themselves.
        if lhs.is_constant() && !right_is_constant {
            return Ok(compare(rhs, lhs, operator.swap())?.into());
        }

        // First try lhs op rhs, then invert and try again.
        for kernel in kernels {
            if let Some(output) = kernel.invoke(args)? {
                return Ok(output);
            }
        }
        if let Some(output) = lhs.invoke(&COMPARE_FN, args)? {
            return Ok(output);
        }

        // Try inverting the operator and swapping the arguments
        let inverted_args = InvocationArgs {
            inputs: &[rhs.into(), lhs.into()],
            options: &operator.swap(),
        };
        for kernel in kernels {
            if let Some(output) = kernel.invoke(&inverted_args)? {
                return Ok(output);
            }
        }
        if let Some(output) = rhs.invoke(&COMPARE_FN, &inverted_args)? {
            return Ok(output);
        }

        // Only log missing compare implementation if there's possibly better one than arrow,
        // i.e. lhs isn't arrow or rhs isn't arrow or constant
        if !(lhs.is_arrow() && (rhs.is_arrow() || right_is_constant)) {
            log::debug!(
                "No compare implementation found for LHS {}, RHS {}, and operator {} (or inverse)",
                lhs.encoding_id(),
                rhs.encoding_id(),
                operator,
            );
        }

        // Fallback to arrow on canonical types
        Ok(arrow_compare(lhs, rhs, operator)?.into())
    }

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

        if !lhs.dtype().eq_ignore_nullability(rhs.dtype()) {
            vortex_bail!(
                "Cannot compare different DTypes {} and {}",
                lhs.dtype(),
                rhs.dtype()
            );
        }

        // TODO(ngates): no reason why not
        if lhs.dtype().is_struct() {
            vortex_bail!(
                "Compare does not support arrays with Struct DType, got: {} and {}",
                lhs.dtype(),
                rhs.dtype()
            )
        }

        Ok(DType::Bool(
            lhs.dtype().nullability() | rhs.dtype().nullability(),
        ))
    }

    fn return_len(&self, args: &InvocationArgs) -> VortexResult<usize> {
        let CompareArgs { lhs, rhs, .. } = CompareArgs::try_from(args)?;
        if lhs.len() != rhs.len() {
            vortex_bail!(
                "Compare operations only support arrays of the same length, got {} and {}",
                lhs.len(),
                rhs.len()
            );
        }
        Ok(lhs.len())
    }

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

struct CompareArgs<'a> {
    lhs: &'a dyn Array,
    rhs: &'a dyn Array,
    operator: Operator,
}

impl Options for Operator {
    fn as_any(&self) -> &dyn Any {
        self
    }
}

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

    fn try_from(value: &InvocationArgs<'a>) -> Result<Self, Self::Error> {
        if value.inputs.len() != 2 {
            vortex_bail!("Expected 2 inputs, found {}", value.inputs.len());
        }
        let lhs = value.inputs[0]
            .array()
            .ok_or_else(|| vortex_err!("Expected first input to be an array"))?;
        let rhs = value.inputs[1]
            .array()
            .ok_or_else(|| vortex_err!("Expected second input to be an array"))?;
        let operator = *value
            .options
            .as_any()
            .downcast_ref::<Operator>()
            .vortex_expect("Expected options to be an operator");

        Ok(CompareArgs { lhs, rhs, operator })
    }
}

/// Helper function to compare empty values with arrays that have external value length information
/// like `VarBin`.
pub fn compare_lengths_to_empty<P, I>(lengths: I, op: Operator) -> BooleanBuffer
where
    P: NativePType,
    I: Iterator<Item = P>,
{
    // All comparison can be expressed in terms of equality. "" is the absolute min of possible value.
    let cmp_fn = match op {
        Operator::Eq | Operator::Lte => |v| v == P::zero(),
        Operator::NotEq | Operator::Gt => |v| v != P::zero(),
        Operator::Gte => |_| true,
        Operator::Lt => |_| false,
    };

    lengths.map(cmp_fn).collect::<BooleanBuffer>()
}

/// Implementation of `CompareFn` using the Arrow crate.
fn arrow_compare(
    left: &dyn Array,
    right: &dyn Array,
    operator: Operator,
) -> VortexResult<ArrayRef> {
    let nullable = left.dtype().is_nullable() || right.dtype().is_nullable();
    let lhs = Datum::try_new(left)?;
    let rhs = Datum::try_new(right)?;

    let array = match operator {
        Operator::Eq => cmp::eq(&lhs, &rhs)?,
        Operator::NotEq => cmp::neq(&lhs, &rhs)?,
        Operator::Gt => cmp::gt(&lhs, &rhs)?,
        Operator::Gte => cmp::gt_eq(&lhs, &rhs)?,
        Operator::Lt => cmp::lt(&lhs, &rhs)?,
        Operator::Lte => cmp::lt_eq(&lhs, &rhs)?,
    };
    from_arrow_array_with_len(&array, left.len(), nullable)
}

pub fn scalar_cmp(lhs: &Scalar, rhs: &Scalar, operator: Operator) -> Scalar {
    if lhs.is_null() | rhs.is_null() {
        Scalar::null(DType::Bool(Nullability::Nullable))
    } else {
        let b = match operator {
            Operator::Eq => lhs == rhs,
            Operator::NotEq => lhs != rhs,
            Operator::Gt => lhs > rhs,
            Operator::Gte => lhs >= rhs,
            Operator::Lt => lhs < rhs,
            Operator::Lte => lhs <= rhs,
        };

        Scalar::bool(b, lhs.dtype().nullability() | rhs.dtype().nullability())
    }
}

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

    use super::*;
    use crate::ToCanonical;
    use crate::arrays::{BoolArray, ConstantArray, VarBinArray, VarBinViewArray};
    use crate::test_harness::to_int_indices;
    use crate::validity::Validity;

    #[test]
    fn test_bool_basic_comparisons() {
        let arr = BoolArray::new(
            BooleanBuffer::from_iter([true, true, false, true, false]),
            Validity::from_iter([false, true, true, true, true]),
        );

        let matches = compare(arr.as_ref(), arr.as_ref(), Operator::Eq)
            .unwrap()
            .to_bool()
            .unwrap();

        assert_eq!(to_int_indices(matches).unwrap(), [1u64, 2, 3, 4]);

        let matches = compare(arr.as_ref(), arr.as_ref(), Operator::NotEq)
            .unwrap()
            .to_bool()
            .unwrap();
        let empty: [u64; 0] = [];
        assert_eq!(to_int_indices(matches).unwrap(), empty);

        let other = BoolArray::new(
            BooleanBuffer::from_iter([false, false, false, true, true]),
            Validity::from_iter([false, true, true, true, true]),
        );

        let matches = compare(arr.as_ref(), other.as_ref(), Operator::Lte)
            .unwrap()
            .to_bool()
            .unwrap();
        assert_eq!(to_int_indices(matches).unwrap(), [2u64, 3, 4]);

        let matches = compare(arr.as_ref(), other.as_ref(), Operator::Lt)
            .unwrap()
            .to_bool()
            .unwrap();
        assert_eq!(to_int_indices(matches).unwrap(), [4u64]);

        let matches = compare(other.as_ref(), arr.as_ref(), Operator::Gte)
            .unwrap()
            .to_bool()
            .unwrap();
        assert_eq!(to_int_indices(matches).unwrap(), [2u64, 3, 4]);

        let matches = compare(other.as_ref(), arr.as_ref(), Operator::Gt)
            .unwrap()
            .to_bool()
            .unwrap();
        assert_eq!(to_int_indices(matches).unwrap(), [4u64]);
    }

    #[test]
    fn constant_compare() {
        let left = ConstantArray::new(Scalar::from(2u32), 10);
        let right = ConstantArray::new(Scalar::from(10u32), 10);

        let compare = compare(left.as_ref(), right.as_ref(), Operator::Gt).unwrap();
        let res = compare.as_constant().unwrap();
        assert_eq!(res.as_bool().value(), Some(false));
        assert_eq!(compare.len(), 10);

        let compare = arrow_compare(&left.into_array(), &right.into_array(), Operator::Gt).unwrap();
        let res = compare.as_constant().unwrap();
        assert_eq!(res.as_bool().value(), Some(false));
        assert_eq!(compare.len(), 10);
    }

    #[rstest]
    #[case(Operator::Eq, vec![false, false, false, true])]
    #[case(Operator::NotEq, vec![true, true, true, false])]
    #[case(Operator::Gt, vec![true, true, true, false])]
    #[case(Operator::Gte, vec![true, true, true, true])]
    #[case(Operator::Lt, vec![false, false, false, false])]
    #[case(Operator::Lte, vec![false, false, false, true])]
    fn test_cmp_to_empty(#[case] op: Operator, #[case] expected: Vec<bool>) {
        let lengths: Vec<i32> = vec![1, 5, 7, 0];

        let output = compare_lengths_to_empty(lengths.iter().copied(), op);
        assert_eq!(Vec::from_iter(output.iter()), expected);
    }

    #[rstest]
    #[case(VarBinArray::from(vec!["a", "b"]).into_array(), VarBinViewArray::from_iter_str(["a", "b"]).into_array())]
    #[case(VarBinViewArray::from_iter_str(["a", "b"]).into_array(), VarBinArray::from(vec!["a", "b"]).into_array())]
    #[case(VarBinArray::from(vec!["a".as_bytes(), "b".as_bytes()]).into_array(), VarBinViewArray::from_iter_bin(["a".as_bytes(), "b".as_bytes()]).into_array())]
    #[case(VarBinViewArray::from_iter_bin(["a".as_bytes(), "b".as_bytes()]).into_array(), VarBinArray::from(vec!["a".as_bytes(), "b".as_bytes()]).into_array())]
    fn arrow_compare_different_encodings(#[case] left: ArrayRef, #[case] right: ArrayRef) {
        let res = compare(&left, &right, Operator::Eq).unwrap();
        assert_eq!(
            res.to_bool().unwrap().boolean_buffer().count_set_bits(),
            left.len()
        );
    }
}

1	// SPDX-License-Identifier: Apache-2.0
2	// SPDX-FileCopyrightText: Copyright the Vortex contributors
3
4	use core::fmt;
5	use std::any::Any;
6	use std::fmt::{Display, Formatter};
7	use std::sync::LazyLock;
8
9	use arcref::ArcRef;
10	use arrow_buffer::BooleanBuffer;
11	use arrow_ord::cmp;
12	use vortex_dtype::{DType, NativePType, Nullability};
13	use vortex_error::{VortexError, VortexExpect, VortexResult, vortex_bail, vortex_err};
14	use vortex_scalar::Scalar;
15
16	use crate::arrays::ConstantArray;
17	use crate::arrow::{Datum, from_arrow_array_with_len};
18	use crate::compute::{ComputeFn, ComputeFnVTable, InvocationArgs, Kernel, Options, Output};
19	use crate::vtable::VTable;
20	use crate::{Array, ArrayRef, Canonical, IntoArray};
21
22	/// Compares two arrays and returns a new boolean array with the result of the comparison.
23	/// Or, returns None if comparison is not supported for these arrays.
24	pub fn compare(left: &dyn Array, right: &dyn Array, operator: Operator) -> VortexResult<ArrayRef> {	15,762✔
25	COMPARE_FN	15,762✔
26	.invoke(&InvocationArgs {	15,762✔
27	inputs: &[left.into(), right.into()],	15,762✔
28	options: &operator,	15,762✔
29	})?	15,762✔
30	.unwrap_array()	15,762✔
31	}	15,762✔
32
33	#[derive(Copy, Clone, Debug, Eq, PartialEq, PartialOrd, Hash)]
34	pub enum Operator {
35	Eq,
36	NotEq,
37	Gt,
38	Gte,
39	Lt,
40	Lte,
41	}
42
43	impl Display for Operator {
UNCOV 44	fn fmt(&self, f: &mut Formatter) -> fmt::Result {	×
UNCOV 45	let display = match &self {	×
46	Operator::Eq => "=",	×
47	Operator::NotEq => "!=",	×
48	Operator::Gt => ">",	×
49	Operator::Gte => ">=",	×
UNCOV 50	Operator::Lt => "<",	×
UNCOV 51	Operator::Lte => "<=",	×
52	};
UNCOV 53	Display::fmt(display, f)	×
UNCOV 54	}	×
55	}
56
57	impl Operator {
58	pub fn inverse(self) -> Self {	×
59	match self {	×
60	Operator::Eq => Operator::NotEq,	×
61	Operator::NotEq => Operator::Eq,	×
62	Operator::Gt => Operator::Lte,	×
63	Operator::Gte => Operator::Lt,	×
64	Operator::Lt => Operator::Gte,	×
65	Operator::Lte => Operator::Gt,	×
66	}
67	}	×
68
69	/// Change the sides of the operator, where changing lhs and rhs won't change the result of the operation
70	pub fn swap(self) -> Self {	10,421✔
71	match self {	10,421✔
72	Operator::Eq => Operator::Eq,	4,499✔
73	Operator::NotEq => Operator::NotEq,	967✔
74	Operator::Gt => Operator::Lt,	1,386✔
75	Operator::Gte => Operator::Lte,	1,209✔
76	Operator::Lt => Operator::Gt,	1,020✔
77	Operator::Lte => Operator::Gte,	1,340✔
78	}
79	}	10,421✔
80	}
81
82	pub struct CompareKernelRef(ArcRef<dyn Kernel>);
83	inventory::collect!(CompareKernelRef);
84
85	pub trait CompareKernel: VTable {
86	fn compare(
87	&self,
88	lhs: &Self::Array,
89	rhs: &dyn Array,
90	operator: Operator,
91	) -> VortexResult<Option<ArrayRef>>;
92	}
93
94	#[derive(Debug)]
95	pub struct CompareKernelAdapter<V: VTable>(pub V);
96
97	impl<V: VTable + CompareKernel> CompareKernelAdapter<V> {
98	pub const fn lift(&'static self) -> CompareKernelRef {	×
99	CompareKernelRef(ArcRef::new_ref(self))	×
100	}	×
101	}
102
103	impl<V: VTable + CompareKernel> Kernel for CompareKernelAdapter<V> {
104	fn invoke(&self, args: &InvocationArgs) -> VortexResult<Option<Output>> {	157,132✔
105	let inputs = CompareArgs::try_from(args)?;	157,132✔
106	let Some(array) = inputs.lhs.as_opt::<V>() else {	157,132✔
107	return Ok(None);	142,913✔
108	};
109	Ok(V::compare(&self.0, array, inputs.rhs, inputs.operator)?.map(\|array\| array.into()))	14,219✔
110	}	157,132✔
111	}
112
113	pub static COMPARE_FN: LazyLock<ComputeFn> = LazyLock::new(\|\| {	2,763✔
114	let compute = ComputeFn::new("compare".into(), ArcRef::new_ref(&Compare));	2,763✔
115	for kernel in inventory::iter::<CompareKernelRef> {	22,376✔
116	compute.register_kernel(kernel.0.clone());	19,613✔
117	}	19,613✔
118	compute	2,763✔
119	});	2,763✔
120
121	struct Compare;
122
123	impl ComputeFnVTable for Compare {
124	fn invoke(	15,762✔
125	&self,	15,762✔
126	args: &InvocationArgs,	15,762✔
127	kernels: &[ArcRef<dyn Kernel>],	15,762✔
128	) -> VortexResult<Output> {	15,762✔
129	let CompareArgs { lhs, rhs, operator } = CompareArgs::try_from(args)?;	15,762✔
130
131	let return_dtype = self.return_dtype(args)?;	15,762✔
132
133	if lhs.is_empty() {	15,762✔
134	return Ok(Canonical::empty(&return_dtype).into_array().into());	1✔
135	}	15,761✔
136
137	let left_constant_null = lhs.as_constant().map(\|l\| l.is_null()).unwrap_or(false);	15,761✔
138	let right_constant_null = rhs.as_constant().map(\|r\| r.is_null()).unwrap_or(false);	15,761✔
139	if left_constant_null \|\| right_constant_null {	15,761✔
140	return Ok(ConstantArray::new(Scalar::null(return_dtype), lhs.len())	112✔
141	.into_array()	112✔
142	.into());	112✔
143	}	15,649✔
144
145	let right_is_constant = rhs.is_constant();	15,649✔
146
147	// Always try to put constants on the right-hand side so encodings can optimise themselves.
148	if lhs.is_constant() && !right_is_constant {	15,649✔
149	return Ok(compare(rhs, lhs, operator.swap())?.into());	1,320✔
150	}	14,329✔
151
152	// First try lhs op rhs, then invert and try again.
153	for kernel in kernels {	111,373✔
154	if let Some(output) = kernel.invoke(args)? {	102,272✔
155	return Ok(output);	5,228✔
156	}	97,044✔
157	}
158	if let Some(output) = lhs.invoke(&COMPARE_FN, args)? {	9,101✔
159	return Ok(output);	×
160	}	9,101✔
161
162	// Try inverting the operator and swapping the arguments
163	let inverted_args = InvocationArgs {	9,101✔
164	inputs: &[rhs.into(), lhs.into()],	9,101✔
165	options: &operator.swap(),	9,101✔
166	};	9,101✔
167	for kernel in kernels {	75,525✔
168	if let Some(output) = kernel.invoke(&inverted_args)? {	69,423✔
169	return Ok(output);	2,999✔
170	}	66,424✔
171	}
172	if let Some(output) = rhs.invoke(&COMPARE_FN, &inverted_args)? {	6,102✔
173	return Ok(output);	×
174	}	6,102✔
175
176	// Only log missing compare implementation if there's possibly better one than arrow,
177	// i.e. lhs isn't arrow or rhs isn't arrow or constant
178	if !(lhs.is_arrow() && (rhs.is_arrow() \|\| right_is_constant)) {	6,102✔
179	log::debug!(	3,701✔
180	"No compare implementation found for LHS {}, RHS {}, and operator {} (or inverse)",	×
181	lhs.encoding_id(),	×
182	rhs.encoding_id(),	×
183	operator,
184	);
185	}	2,401✔
186
187	// Fallback to arrow on canonical types
188	Ok(arrow_compare(lhs, rhs, operator)?.into())	6,102✔
189	}	15,762✔
190
191	fn return_dtype(&self, args: &InvocationArgs) -> VortexResult<DType> {	31,524✔
192	let CompareArgs { lhs, rhs, .. } = CompareArgs::try_from(args)?;	31,524✔
193
194	if !lhs.dtype().eq_ignore_nullability(rhs.dtype()) {	31,524✔
195	vortex_bail!(	×
196	"Cannot compare different DTypes {} and {}",	×
197	lhs.dtype(),	×
198	rhs.dtype()	×
199	);
200	}	31,524✔
201
202	// TODO(ngates): no reason why not
203	if lhs.dtype().is_struct() {	31,524✔
204	vortex_bail!(	×
205	"Compare does not support arrays with Struct DType, got: {} and {}",	×
206	lhs.dtype(),	×
207	rhs.dtype()	×
208	)
209	}	31,524✔
210
211	Ok(DType::Bool(	31,524✔
212	lhs.dtype().nullability() \| rhs.dtype().nullability(),	31,524✔
213	))	31,524✔
214	}	31,524✔
215
216	fn return_len(&self, args: &InvocationArgs) -> VortexResult<usize> {	15,762✔
217	let CompareArgs { lhs, rhs, .. } = CompareArgs::try_from(args)?;	15,762✔
218	if lhs.len() != rhs.len() {	15,762✔
219	vortex_bail!(	×
220	"Compare operations only support arrays of the same length, got {} and {}",	×
221	lhs.len(),	×
222	rhs.len()	×
223	);
224	}	15,762✔
225	Ok(lhs.len())	15,762✔
226	}	15,762✔
227
228	fn is_elementwise(&self) -> bool {	15,762✔
229	true	15,762✔
230	}	15,762✔
231	}
232
233	struct CompareArgs<'a> {
234	lhs: &'a dyn Array,
235	rhs: &'a dyn Array,
236	operator: Operator,
237	}
238
239	impl Options for Operator {
240	fn as_any(&self) -> &dyn Any {	234,743✔
241	self	234,743✔
242	}	234,743✔
243	}
244
245	impl<'a> TryFrom<&InvocationArgs<'a>> for CompareArgs<'a> {
246	type Error = VortexError;
247
248	fn try_from(value: &InvocationArgs<'a>) -> Result<Self, Self::Error> {	234,743✔
249	if value.inputs.len() != 2 {	234,743✔
250	vortex_bail!("Expected 2 inputs, found {}", value.inputs.len());	×
251	}	234,743✔
252	let lhs = value.inputs[0]	234,743✔
253	.array()	234,743✔
254	.ok_or_else(\|\| vortex_err!("Expected first input to be an array"))?;	234,743✔
255	let rhs = value.inputs[1]	234,743✔
256	.array()	234,743✔
257	.ok_or_else(\|\| vortex_err!("Expected second input to be an array"))?;	234,743✔
258	let operator = *value	234,743✔
259	.options	234,743✔
260	.as_any()	234,743✔
261	.downcast_ref::<Operator>()	234,743✔
262	.vortex_expect("Expected options to be an operator");	234,743✔
263
264	Ok(CompareArgs { lhs, rhs, operator })	234,743✔
265	}	234,743✔
266	}
267
268	/// Helper function to compare empty values with arrays that have external value length information
269	/// like `VarBin`.
270	pub fn compare_lengths_to_empty<P, I>(lengths: I, op: Operator) -> BooleanBuffer	44✔
271	where	44✔
272	P: NativePType,	44✔
273	I: Iterator<Item = P>,	44✔
274	{
275	// All comparison can be expressed in terms of equality. "" is the absolute min of possible value.
276	let cmp_fn = match op {	44✔
277	Operator::Eq \| Operator::Lte => \|v\| v == P::zero(),	120✔
278	Operator::NotEq \| Operator::Gt => \|v\| v != P::zero(),	8✔
279	Operator::Gte => \|_\| true,
280	Operator::Lt => \|_\| false,
281	};
282
283	lengths.map(cmp_fn).collect::<BooleanBuffer>()	44✔
284	}	44✔
285
286	/// Implementation of `CompareFn` using the Arrow crate.
287	fn arrow_compare(	6,103✔
288	left: &dyn Array,	6,103✔
289	right: &dyn Array,	6,103✔
290	operator: Operator,	6,103✔
291	) -> VortexResult<ArrayRef> {	6,103✔
292	let nullable = left.dtype().is_nullable() \|\| right.dtype().is_nullable();	6,103✔
293	let lhs = Datum::try_new(left)?;	6,103✔
294	let rhs = Datum::try_new(right)?;	6,103✔
295
296	let array = match operator {	6,103✔
297	Operator::Eq => cmp::eq(&lhs, &rhs)?,	2,610✔
298	Operator::NotEq => cmp::neq(&lhs, &rhs)?,	227✔
299	Operator::Gt => cmp::gt(&lhs, &rhs)?,	793✔
300	Operator::Gte => cmp::gt_eq(&lhs, &rhs)?,	981✔
301	Operator::Lt => cmp::lt(&lhs, &rhs)?,	674✔
302	Operator::Lte => cmp::lt_eq(&lhs, &rhs)?,	818✔
303	};
304	from_arrow_array_with_len(&array, left.len(), nullable)	6,103✔
305	}	6,103✔
306
307	pub fn scalar_cmp(lhs: &Scalar, rhs: &Scalar, operator: Operator) -> Scalar {	3,768✔
308	if lhs.is_null() \| rhs.is_null() {	3,768✔
309	Scalar::null(DType::Bool(Nullability::Nullable))	×
310	} else {
311	let b = match operator {	3,768✔
312	Operator::Eq => lhs == rhs,	713✔
313	Operator::NotEq => lhs != rhs,	740✔
314	Operator::Gt => lhs > rhs,	855✔
315	Operator::Gte => lhs >= rhs,	376✔
316	Operator::Lt => lhs < rhs,	856✔
317	Operator::Lte => lhs <= rhs,	228✔
318	};
319
320	Scalar::bool(b, lhs.dtype().nullability() \| rhs.dtype().nullability())	3,768✔
321	}
322	}	3,768✔
323
324	#[cfg(test)]
325	mod tests {
326	use arrow_buffer::BooleanBuffer;
327	use rstest::rstest;
328
329	use super::*;
330	use crate::ToCanonical;
331	use crate::arrays::{BoolArray, ConstantArray, VarBinArray, VarBinViewArray};
332	use crate::test_harness::to_int_indices;
333	use crate::validity::Validity;
334
335	#[test]
336	fn test_bool_basic_comparisons() {	1✔
337	let arr = BoolArray::new(	1✔
338	BooleanBuffer::from_iter([true, true, false, true, false]),	1✔
339	Validity::from_iter([false, true, true, true, true]),	1✔
340	);
341
342	let matches = compare(arr.as_ref(), arr.as_ref(), Operator::Eq)	1✔
343	.unwrap()	1✔
344	.to_bool()	1✔
345	.unwrap();	1✔
346
347	assert_eq!(to_int_indices(matches).unwrap(), [1u64, 2, 3, 4]);	1✔
348
349	let matches = compare(arr.as_ref(), arr.as_ref(), Operator::NotEq)	1✔
350	.unwrap()	1✔
351	.to_bool()	1✔
352	.unwrap();	1✔
353	let empty: [u64; 0] = [];	1✔
354	assert_eq!(to_int_indices(matches).unwrap(), empty);	1✔
355
356	let other = BoolArray::new(	1✔
357	BooleanBuffer::from_iter([false, false, false, true, true]),	1✔
358	Validity::from_iter([false, true, true, true, true]),	1✔
359	);
360
361	let matches = compare(arr.as_ref(), other.as_ref(), Operator::Lte)	1✔
362	.unwrap()	1✔
363	.to_bool()	1✔
364	.unwrap();	1✔
365	assert_eq!(to_int_indices(matches).unwrap(), [2u64, 3, 4]);	1✔
366
367	let matches = compare(arr.as_ref(), other.as_ref(), Operator::Lt)	1✔
368	.unwrap()	1✔
369	.to_bool()	1✔
370	.unwrap();	1✔
371	assert_eq!(to_int_indices(matches).unwrap(), [4u64]);	1✔
372
373	let matches = compare(other.as_ref(), arr.as_ref(), Operator::Gte)	1✔
374	.unwrap()	1✔
375	.to_bool()	1✔
376	.unwrap();	1✔
377	assert_eq!(to_int_indices(matches).unwrap(), [2u64, 3, 4]);	1✔
378
379	let matches = compare(other.as_ref(), arr.as_ref(), Operator::Gt)	1✔
380	.unwrap()	1✔
381	.to_bool()	1✔
382	.unwrap();	1✔
383	assert_eq!(to_int_indices(matches).unwrap(), [4u64]);	1✔
384	}	1✔
385
386	#[test]
387	fn constant_compare() {	1✔
388	let left = ConstantArray::new(Scalar::from(2u32), 10);	1✔
389	let right = ConstantArray::new(Scalar::from(10u32), 10);	1✔
390
391	let compare = compare(left.as_ref(), right.as_ref(), Operator::Gt).unwrap();	1✔
392	let res = compare.as_constant().unwrap();	1✔
393	assert_eq!(res.as_bool().value(), Some(false));	1✔
394	assert_eq!(compare.len(), 10);	1✔
395
396	let compare = arrow_compare(&left.into_array(), &right.into_array(), Operator::Gt).unwrap();	1✔
397	let res = compare.as_constant().unwrap();	1✔
398	assert_eq!(res.as_bool().value(), Some(false));	1✔
399	assert_eq!(compare.len(), 10);	1✔
400	}	1✔
401
402	#[rstest]
403	#[case(Operator::Eq, vec![false, false, false, true])]
404	#[case(Operator::NotEq, vec![true, true, true, false])]
405	#[case(Operator::Gt, vec![true, true, true, false])]
406	#[case(Operator::Gte, vec![true, true, true, true])]
407	#[case(Operator::Lt, vec![false, false, false, false])]
408	#[case(Operator::Lte, vec![false, false, false, true])]
409	fn test_cmp_to_empty(#[case] op: Operator, #[case] expected: Vec<bool>) {
410	let lengths: Vec<i32> = vec![1, 5, 7, 0];
411
412	let output = compare_lengths_to_empty(lengths.iter().copied(), op);
413	assert_eq!(Vec::from_iter(output.iter()), expected);
414	}
415
416	#[rstest]
417	#[case(VarBinArray::from(vec!["a", "b"]).into_array(), VarBinViewArray::from_iter_str(["a", "b"]).into_array())]
418	#[case(VarBinViewArray::from_iter_str(["a", "b"]).into_array(), VarBinArray::from(vec!["a", "b"]).into_array())]
419	#[case(VarBinArray::from(vec!["a".as_bytes(), "b".as_bytes()]).into_array(), VarBinViewArray::from_iter_bin(["a".as_bytes(), "b".as_bytes()]).into_array())]
420	#[case(VarBinViewArray::from_iter_bin(["a".as_bytes(), "b".as_bytes()]).into_array(), VarBinArray::from(vec!["a".as_bytes(), "b".as_bytes()]).into_array())]
421	fn arrow_compare_different_encodings(#[case] left: ArrayRef, #[case] right: ArrayRef) {
422	let res = compare(&left, &right, Operator::Eq).unwrap();
423	assert_eq!(
424	res.to_bool().unwrap().boolean_buffer().count_set_bits(),
425	left.len()
426	);
427	}
428	}

vortex-data / vortex / 16473770414

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