vortex_mask/
intersect_by_rank.rs

1// SPDX-License-Identifier: Apache-2.0
2// SPDX-FileCopyrightText: Copyright the Vortex contributors
3
4use crate::{AllOr, Mask};
5
6impl Mask {
7    /// Take the intersection of the `mask` with the set of true values in `self`.
8    ///
9    /// We are more interested in low selectivity `self` (as indices) with a boolean buffer mask,
10    /// so we don't optimize for other cases, yet.
11    ///
12    /// Note: we might be able to accelerate this function on x86 with BMI, see:
13    /// <https://www.microsoft.com/en-us/research/uploads/prod/2023/06/parquet-select-sigmod23.pdf>
14    ///
15    /// # Examples
16    ///
17    /// Keep the third and fifth set values from mask `m1`:
18    /// ```
19    /// use vortex_mask::Mask;
20    ///
21    /// let m1 = Mask::from_iter([true, false, false, true, true, true, false, true]);
22    /// let m2 = Mask::from_iter([false, false, true, false, true]);
23    /// assert_eq!(
24    ///     m1.intersect_by_rank(&m2),
25    ///     Mask::from_iter([false, false, false, false, true, false, false, true])
26    /// );
27    /// ```
28    pub fn intersect_by_rank(&self, mask: &Mask) -> Mask {
29        assert_eq!(self.true_count(), mask.len());
30
31        match (self.indices(), mask.indices()) {
32            (AllOr::All, _) => mask.clone(),
33            (_, AllOr::All) => self.clone(),
34            (AllOr::None, _) => Self::new_false(0),
35            (_, AllOr::None) => Self::new_false(self.len()),
36            (AllOr::Some(self_indices), AllOr::Some(mask_indices)) => {
37                Self::from_indices(
38                    self.len(),
39                    mask_indices
40                        .iter()
41                        .map(|idx|
42                            // This is verified as safe because we know that the indices are less than the
43                            // mask.len() and we known mask.len() <= self.len(),
44                            // implied by `self.true_count() == mask.len()`.
45                            unsafe{*self_indices.get_unchecked(*idx)})
46                        .collect(),
47                )
48            }
49        }
50    }
51}
52
53#[cfg(test)]
54mod test {
55    use rstest::rstest;
56    use vortex_buffer::BitBuffer;
57
58    use crate::Mask;
59
60    #[test]
61    fn mask_bitand_all_as_bit_and() {
62        let this = Mask::from_buffer(BitBuffer::from_iter(vec![true, true, true, true, true]));
63        let mask = Mask::from_buffer(BitBuffer::from_iter(vec![false, true, false, true, true]));
64        assert_eq!(
65            this.intersect_by_rank(&mask),
66            Mask::from_indices(5, vec![1, 3, 4])
67        );
68    }
69
70    #[test]
71    fn mask_bitand_all_true() {
72        let this = Mask::from_buffer(BitBuffer::from_iter(vec![false, false, true, true, true]));
73        let mask = Mask::from_buffer(BitBuffer::from_iter(vec![true, true, true]));
74        assert_eq!(
75            this.intersect_by_rank(&mask),
76            Mask::from_indices(5, vec![2, 3, 4])
77        );
78    }
79
80    #[test]
81    fn mask_bitand_true() {
82        let this = Mask::from_buffer(BitBuffer::from_iter(vec![true, false, false, true, true]));
83        let mask = Mask::from_buffer(BitBuffer::from_iter(vec![true, false, true]));
84        assert_eq!(
85            this.intersect_by_rank(&mask),
86            Mask::from_indices(5, vec![0, 4])
87        );
88    }
89
90    #[test]
91    fn mask_bitand_false() {
92        let this = Mask::from_buffer(BitBuffer::from_iter(vec![true, false, false, true, true]));
93        let mask = Mask::from_buffer(BitBuffer::from_iter(vec![false, false, false]));
94        assert_eq!(this.intersect_by_rank(&mask), Mask::from_indices(5, vec![]));
95    }
96
97    #[rstest]
98    #[case::all_true_with_all_true(
99        Mask::new_true(5),
100        Mask::new_true(5),
101        vec![0, 1, 2, 3, 4]
102    )]
103    #[case::all_true_with_all_false(
104        Mask::new_true(5),
105        Mask::new_false(5),
106        vec![]
107    )]
108    #[case::all_false_with_any(
109        Mask::new_false(10),
110        Mask::new_true(0),
111        vec![]
112    )]
113    #[case::indices_with_all_true(
114        Mask::from_indices(10, vec![2, 5, 7, 9]),
115        Mask::new_true(4),
116        vec![2, 5, 7, 9]
117    )]
118    #[case::indices_with_all_false(
119        Mask::from_indices(10, vec![2, 5, 7, 9]),
120        Mask::new_false(4),
121        vec![]
122    )]
123    fn test_intersect_by_rank_special_cases(
124        #[case] base_mask: Mask,
125        #[case] rank_mask: Mask,
126        #[case] expected_indices: Vec<usize>,
127    ) {
128        let result = base_mask.intersect_by_rank(&rank_mask);
129
130        match result.indices() {
131            crate::AllOr::All => assert_eq!(expected_indices.len(), result.len()),
132            crate::AllOr::None => assert!(expected_indices.is_empty()),
133            crate::AllOr::Some(indices) => assert_eq!(indices, &expected_indices[..]),
134        }
135    }
136
137    #[test]
138    fn test_intersect_by_rank_example() {
139        // Example from the documentation
140        let m1 = Mask::from_iter([true, false, false, true, true, true, false, true]);
141        let m2 = Mask::from_iter([false, false, true, false, true]);
142        let result = m1.intersect_by_rank(&m2);
143        let expected = Mask::from_iter([false, false, false, false, true, false, false, true]);
144        assert_eq!(result, expected);
145    }
146
147    #[test]
148    #[should_panic]
149    fn test_intersect_by_rank_wrong_length() {
150        let m1 = Mask::from_indices(10, vec![2, 5, 7]); // 3 true values
151        let m2 = Mask::new_true(5); // 5 true values - doesn't match
152        m1.intersect_by_rank(&m2);
153    }
154
155    #[rstest]
156    #[case::single_element(
157        vec![3],
158        vec![true],
159        vec![3]
160    )]
161    #[case::single_element_masked(
162        vec![3],
163        vec![false],
164        vec![]
165    )]
166    #[case::alternating(
167        vec![0, 2, 4, 6, 8],
168        vec![true, false, true, false, true],
169        vec![0, 4, 8]
170    )]
171    #[case::consecutive(
172        vec![5, 6, 7, 8, 9],
173        vec![false, true, true, true, false],
174        vec![6, 7, 8]
175    )]
176    fn test_intersect_by_rank_patterns(
177        #[case] base_indices: Vec<usize>,
178        #[case] rank_pattern: Vec<bool>,
179        #[case] expected_indices: Vec<usize>,
180    ) {
181        let base = Mask::from_indices(10, base_indices);
182        let rank = Mask::from_iter(rank_pattern);
183        let result = base.intersect_by_rank(&rank);
184
185        match result.indices() {
186            crate::AllOr::Some(indices) => assert_eq!(indices, &expected_indices[..]),
187            crate::AllOr::None => assert!(expected_indices.is_empty()),
188            _ => panic!("Unexpected result"),
189        }
190    }
191}
vortex_mask/intersect_by_rank.rs

vortex_mask/
intersect_by_rank.rs
