vortex_array/arrays/primitive/
top_value.rs

1use std::hash::Hash;
2
3use rustc_hash::FxBuildHasher;
4use vortex_dtype::{NativePType, match_each_native_ptype};
5use vortex_error::{VortexExpect, VortexResult};
6use vortex_mask::{AllOr, Mask};
7use vortex_scalar::PValue;
8
9use crate::aliases::hash_map::HashMap;
10use crate::arrays::{NativeValue, PrimitiveArray};
11
12impl PrimitiveArray {
13    /// Compute most common present value of this array
14    pub fn top_value(&self) -> VortexResult<Option<(PValue, usize)>> {
15        if self.is_empty() {
16            return Ok(None);
17        }
18
19        if self.all_invalid()? {
20            return Ok(None);
21        }
22
23        match_each_native_ptype!(self.ptype(), |$P| {
24            let (top, count) = typed_top_value(self.as_slice::<$P>(), self.validity_mask()?);
25            Ok(Some((top.into(), count)))
26        })
27    }
28}
29
30fn typed_top_value<T>(values: &[T], mask: Mask) -> (T, usize)
31where
32    T: NativePType,
33    NativeValue<T>: Eq + Hash,
34{
35    let mut distinct_values: HashMap<NativeValue<T>, usize, FxBuildHasher> =
36        HashMap::with_hasher(FxBuildHasher);
37    match mask.indices() {
38        AllOr::All => {
39            for value in values.iter().copied() {
40                *distinct_values.entry(NativeValue(value)).or_insert(0) += 1;
41            }
42        }
43        AllOr::None => unreachable!("All invalid arrays should be handled earlier"),
44        AllOr::Some(idxs) => {
45            for &i in idxs {
46                *distinct_values
47                    .entry(NativeValue(unsafe { *values.get_unchecked(i) }))
48                    .or_insert(0) += 1
49            }
50        }
51    }
52
53    let (&top_value, &top_count) = distinct_values
54        .iter()
55        .max_by_key(|&(_, &count)| count)
56        .vortex_expect("non-empty");
57    (top_value.0, top_count)
58}
vortex_array/arrays/primitive/top_value.rs

vortex_array/arrays/primitive/
top_value.rs
