Skip to main content

vortex_array/arrays/dict/compute/
min_max.rs

1// SPDX-License-Identifier: Apache-2.0
2// SPDX-FileCopyrightText: Copyright the Vortex contributors
3
4use vortex_error::VortexResult;
5
6use super::DictArray;
7use super::DictVTable;
8use crate::Array as _;
9use crate::IntoArray;
10use crate::arrays::BoolArray;
11use crate::builtins::ArrayBuiltins;
12use crate::compute::MinMaxKernel;
13use crate::compute::MinMaxKernelAdapter;
14use crate::compute::MinMaxResult;
15use crate::compute::min_max;
16use crate::register_kernel;
17use crate::validity::Validity;
18
19impl MinMaxKernel for DictVTable {
20    fn min_max(&self, array: &DictArray) -> VortexResult<Option<MinMaxResult>> {
21        let codes_validity = array.codes().validity_mask()?;
22        if codes_validity.all_false() {
23            return Ok(None);
24        }
25
26        // Fast path: if all values are referenced, directly compute min/max on values
27        if array.has_all_values_referenced() {
28            return min_max(array.values());
29        }
30
31        // Slow path: compute which values are unreferenced and mask them out
32        let unreferenced_mask = BoolArray::new(
33            array.compute_referenced_values_mask(true)?,
34            Validity::NonNullable,
35        )
36        .into_array();
37
38        min_max(&array.values().clone().mask(unreferenced_mask)?)
39    }
40}
41
42register_kernel!(MinMaxKernelAdapter(DictVTable).lift());
43
44#[cfg(test)]
45mod tests {
46    use rstest::rstest;
47    use vortex_buffer::buffer;
48
49    use super::DictArray;
50    use crate::Array;
51    use crate::IntoArray;
52    use crate::arrays::PrimitiveArray;
53    use crate::builders::dict::dict_encode;
54    use crate::compute::min_max;
55
56    fn assert_min_max(array: &dyn Array, expected: Option<(i32, i32)>) {
57        match (min_max(array).unwrap(), expected) {
58            (Some(result), Some((expected_min, expected_max))) => {
59                assert_eq!(i32::try_from(&result.min).unwrap(), expected_min);
60                assert_eq!(i32::try_from(&result.max).unwrap(), expected_max);
61            }
62            (None, None) => {}
63            (got, expected) => panic!(
64                "min_max mismatch: expected {:?}, got {:?}",
65                expected,
66                got.as_ref().map(|r| (
67                    i32::try_from(&r.min.clone()).ok(),
68                    i32::try_from(&r.max.clone()).ok()
69                ))
70            ),
71        }
72    }
73
74    #[rstest]
75    #[case::covering(
76        DictArray::try_new(
77            buffer![0u32, 1, 2, 3, 0, 1].into_array(),
78            buffer![10i32, 20, 30, 40].into_array(),
79        ).unwrap(),
80        (10, 40)
81    )]
82    #[case::non_covering_duplicates(
83        DictArray::try_new(
84            buffer![1u32, 1, 1, 3, 3].into_array(),
85            buffer![1i32, 2, 3, 4, 5].into_array(),
86        ).unwrap(),
87        (2, 4)
88    )]
89    // Non-covering: codes with gaps
90    #[case::non_covering_gaps(
91        DictArray::try_new(
92            buffer![0u32, 2, 4].into_array(),
93            buffer![1i32, 2, 3, 4, 5].into_array(),
94        ).unwrap(),
95        (1, 5)
96    )]
97    #[case::single(dict_encode(&buffer![42i32].into_array()).unwrap(), (42, 42))]
98    #[case::nullable_codes(
99        DictArray::try_new(
100            PrimitiveArray::from_option_iter([Some(0u32), None, Some(1), Some(2)]).into_array(),
101            buffer![10i32, 20, 30].into_array(),
102        ).unwrap(),
103        (10, 30)
104    )]
105    #[case::nullable_values(
106        dict_encode(
107            PrimitiveArray::from_option_iter([Some(1i32), None, Some(2), Some(1), None]).as_ref()
108        ).unwrap(),
109        (1, 2)
110    )]
111    fn test_min_max(#[case] dict: DictArray, #[case] expected: (i32, i32)) {
112        assert_min_max(dict.as_ref(), Some(expected));
113    }
114
115    #[test]
116    fn test_sliced_dict() {
117        let reference = PrimitiveArray::from_iter([1, 5, 10, 50, 100]);
118        let dict = dict_encode(reference.as_ref()).unwrap();
119        let sliced = dict.slice(1..3).unwrap();
120        assert_min_max(sliced.as_ref(), Some((5, 10)));
121    }
122
123    #[rstest]
124    #[case::empty(
125        DictArray::try_new(
126            PrimitiveArray::from_iter(Vec::<u32>::new()).into_array(),
127            buffer![10i32, 20, 30].into_array(),
128        ).unwrap()
129    )]
130    #[case::all_null_codes(
131        DictArray::try_new(
132            PrimitiveArray::from_option_iter([Option::<u32>::None, None, None]).into_array(),
133            buffer![10i32, 20, 30].into_array(),
134        ).unwrap()
135    )]
136    fn test_min_max_none(#[case] dict: DictArray) {
137        assert_min_max(dict.as_ref(), None);
138    }
139}