vortex-array 0.59.4

Vortex in memory columnar data format
Documentation 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160

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

use std::any::Any;

use vortex_error::VortexResult;

use crate::Array;
use crate::ArrayRef;
use crate::IntoArray;
use crate::arrays::ConstantArray;
use crate::arrow::Datum;
use crate::arrow::from_arrow_array_with_len;
use crate::compute::Options;
use crate::scalar::NumericOperator;
use crate::scalar::Scalar;

/// Point-wise add two numeric arrays.
///
/// Errs at runtime if the sum would overflow or underflow.
///
/// The result is null at any index that either input is null.
pub fn add(lhs: &dyn Array, rhs: &dyn Array) -> VortexResult<ArrayRef> {
    numeric(lhs, rhs, NumericOperator::Add)
}

/// Point-wise add a scalar value to this array on the right-hand-side.
pub fn add_scalar(lhs: &dyn Array, rhs: Scalar) -> VortexResult<ArrayRef> {
    numeric(
        lhs,
        &ConstantArray::new(rhs, lhs.len()).into_array(),
        NumericOperator::Add,
    )
}

/// Point-wise subtract two numeric arrays.
pub fn sub(lhs: &dyn Array, rhs: &dyn Array) -> VortexResult<ArrayRef> {
    numeric(lhs, rhs, NumericOperator::Sub)
}

/// Point-wise subtract a scalar value from this array on the right-hand-side.
pub fn sub_scalar(lhs: &dyn Array, rhs: Scalar) -> VortexResult<ArrayRef> {
    numeric(
        lhs,
        &ConstantArray::new(rhs, lhs.len()).into_array(),
        NumericOperator::Sub,
    )
}

/// Point-wise multiply two numeric arrays.
pub fn mul(lhs: &dyn Array, rhs: &dyn Array) -> VortexResult<ArrayRef> {
    numeric(lhs, rhs, NumericOperator::Mul)
}

/// Point-wise multiply a scalar value into this array on the right-hand-side.
pub fn mul_scalar(lhs: &dyn Array, rhs: Scalar) -> VortexResult<ArrayRef> {
    numeric(
        lhs,
        &ConstantArray::new(rhs, lhs.len()).into_array(),
        NumericOperator::Mul,
    )
}

/// Point-wise divide two numeric arrays.
pub fn div(lhs: &dyn Array, rhs: &dyn Array) -> VortexResult<ArrayRef> {
    numeric(lhs, rhs, NumericOperator::Div)
}

/// Point-wise divide a scalar value into this array on the right-hand-side.
pub fn div_scalar(lhs: &dyn Array, rhs: Scalar) -> VortexResult<ArrayRef> {
    numeric(
        lhs,
        &ConstantArray::new(rhs, lhs.len()).into_array(),
        NumericOperator::Mul,
    )
}

/// Point-wise numeric operation between two arrays of the same type and length.
pub fn numeric(lhs: &dyn Array, rhs: &dyn Array, op: NumericOperator) -> VortexResult<ArrayRef> {
    arrow_numeric(lhs, rhs, op)
}

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

/// Implementation of numeric operations using the Arrow crate.
pub(crate) fn arrow_numeric(
    lhs: &dyn Array,
    rhs: &dyn Array,
    operator: NumericOperator,
) -> VortexResult<ArrayRef> {
    let nullable = lhs.dtype().is_nullable() || rhs.dtype().is_nullable();
    let len = lhs.len();

    let left = Datum::try_new(lhs)?;
    let right = Datum::try_new_with_target_datatype(rhs, left.data_type())?;

    let array = match operator {
        NumericOperator::Add => arrow_arith::numeric::add(&left, &right)?,
        NumericOperator::Sub => arrow_arith::numeric::sub(&left, &right)?,
        NumericOperator::RSub => arrow_arith::numeric::sub(&right, &left)?,
        NumericOperator::Mul => arrow_arith::numeric::mul(&left, &right)?,
        NumericOperator::Div => arrow_arith::numeric::div(&left, &right)?,
        NumericOperator::RDiv => arrow_arith::numeric::div(&right, &left)?,
    };

    from_arrow_array_with_len(array.as_ref(), len, nullable)
}

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

    use crate::IntoArray;
    use crate::arrays::PrimitiveArray;
    use crate::assert_arrays_eq;
    use crate::compute::sub_scalar;

    #[test]
    fn test_scalar_subtract_unsigned() {
        let values = buffer![1u16, 2, 3].into_array();
        let result = sub_scalar(&values, 1u16.into()).unwrap();
        assert_arrays_eq!(result, PrimitiveArray::from_iter([0u16, 1, 2]));
    }

    #[test]
    fn test_scalar_subtract_signed() {
        let values = buffer![1i64, 2, 3].into_array();
        let result = sub_scalar(&values, (-1i64).into()).unwrap();
        assert_arrays_eq!(result, PrimitiveArray::from_iter([2i64, 3, 4]));
    }

    #[test]
    fn test_scalar_subtract_nullable() {
        let values = PrimitiveArray::from_option_iter([Some(1u16), Some(2), None, Some(3)]);
        let result = sub_scalar(values.as_ref(), Some(1u16).into()).unwrap();
        assert_arrays_eq!(
            result,
            PrimitiveArray::from_option_iter([Some(0u16), Some(1), None, Some(2)])
        );
    }

    #[test]
    fn test_scalar_subtract_float() {
        let values = buffer![1.0f64, 2.0, 3.0].into_array();
        let to_subtract = -1f64;
        let result = sub_scalar(&values, to_subtract.into()).unwrap();
        assert_arrays_eq!(result, PrimitiveArray::from_iter([2.0f64, 3.0, 4.0]));
    }

    #[test]
    fn test_scalar_subtract_float_underflow_is_ok() {
        let values = buffer![f32::MIN, 2.0, 3.0].into_array();
        let _results = sub_scalar(&values, 1.0f32.into()).unwrap();
        let _results = sub_scalar(&values, f32::MAX.into()).unwrap();
    }
}