vortex_array/arrays/varbin/compute/
min_max.rs

1// SPDX-License-Identifier: Apache-2.0
2// SPDX-FileCopyrightText: Copyright the Vortex contributors
3
4use itertools::Itertools;
5use vortex_error::VortexResult;
6use vortex_error::vortex_panic;
7
8use crate::accessor::ArrayAccessor;
9use crate::arrays::VarBinArray;
10use crate::arrays::VarBinVTable;
11use crate::compute::MinMaxKernel;
12use crate::compute::MinMaxKernelAdapter;
13use crate::compute::MinMaxResult;
14use crate::dtype::DType;
15use crate::dtype::Nullability::NonNullable;
16use crate::register_kernel;
17use crate::scalar::Scalar;
18
19impl MinMaxKernel for VarBinVTable {
20    fn min_max(&self, array: &VarBinArray) -> VortexResult<Option<MinMaxResult>> {
21        Ok(varbin_compute_min_max(array, array.dtype()))
22    }
23}
24
25register_kernel!(MinMaxKernelAdapter(VarBinVTable).lift());
26
27/// Compute the min and max of VarBin like array.
28pub(crate) fn varbin_compute_min_max<T: ArrayAccessor<[u8]>>(
29    array: &T,
30    dtype: &DType,
31) -> Option<MinMaxResult> {
32    array.with_iterator(|iter| match iter.flatten().minmax() {
33        itertools::MinMaxResult::NoElements => None,
34        itertools::MinMaxResult::OneElement(value) => {
35            let scalar = make_scalar(dtype, value);
36            Some(MinMaxResult {
37                min: scalar.clone(),
38                max: scalar,
39            })
40        }
41        itertools::MinMaxResult::MinMax(min, max) => Some(MinMaxResult {
42            min: make_scalar(dtype, min),
43            max: make_scalar(dtype, max),
44        }),
45    })
46}
47
48/// Helper function to make sure that min/max has the right [`Scalar`] type.
49fn make_scalar(dtype: &DType, value: &[u8]) -> Scalar {
50    match dtype {
51        DType::Binary(_) => Scalar::binary(value.to_vec(), NonNullable),
52        DType::Utf8(_) => {
53            // SAFETY: We only call `compute_min_max` within `varbin/`, in which we always validate
54            // the arrays, and we always pass `array.dtype()` in as the `dtype` argument.
55            let value = unsafe { str::from_utf8_unchecked(value) };
56            Scalar::utf8(value, NonNullable)
57        }
58        _ => vortex_panic!("cannot make Scalar from bytes with dtype {dtype}"),
59    }
60}
61
62#[cfg(test)]
63mod tests {
64    use vortex_buffer::BufferString;
65
66    use crate::IntoArray;
67    use crate::arrays::VarBinArray;
68    use crate::compute::MinMaxResult;
69    use crate::compute::min_max;
70    use crate::dtype::DType::Utf8;
71    use crate::dtype::Nullability::NonNullable;
72    use crate::dtype::Nullability::Nullable;
73    use crate::expr::stats::Stat;
74    use crate::expr::stats::StatsProvider;
75    use crate::scalar::Scalar;
76
77    #[test]
78    fn some_nulls() {
79        let array = VarBinArray::from_iter(
80            vec![
81                Some("hello world"),
82                None,
83                Some("hello world this is a long string"),
84                None,
85            ],
86            Utf8(Nullable),
87        );
88        let MinMaxResult { min, max } = min_max(&array.into_array()).unwrap().unwrap();
89
90        assert_eq!(
91            min,
92            Scalar::try_new(
93                Utf8(NonNullable),
94                Some(BufferString::from("hello world".to_string()).into()),
95            )
96            .unwrap()
97        );
98        assert_eq!(
99            max,
100            Scalar::try_new(
101                Utf8(NonNullable),
102                Some(BufferString::from("hello world this is a long string".to_string()).into()),
103            )
104            .unwrap()
105        );
106    }
107
108    #[test]
109    fn all_nulls() {
110        let array = VarBinArray::from_iter(vec![Option::<&str>::None, None, None], Utf8(Nullable));
111        let stats = array.statistics();
112        assert!(stats.get(Stat::Min).is_none());
113        assert!(stats.get(Stat::Max).is_none());
114    }
115}
vortex_array/arrays/varbin/compute/min_max.rs

vortex_array/arrays/varbin/compute/
min_max.rs
