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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250
//! Provides functionality to get the `n` largest items from a `&mut [T]`. //! //! ``` //! let mut v = [-5, 4, 1, -3, 2]; //! let max = out::max(&mut v, 3); //! assert_eq!(max, [1, 2, 4]); //! ``` //! //! It's worth noting that the largest item will be the last item in the slice that is returned. //! This library can provide significant performance increase compared to sorting or //! converting to a heap when `n` is relatively small compared to the length of the slice. //! In addition, the unstable methods can be used without //! the standard library by disabling the default features. //! //! n = `100`, len = `1_000_000`: //! ```text //! test binary_heap ... bench: 6,599,355 ns/iter (+/- 84,674) //! test max ... bench: 669,726 ns/iter (+/- 13,595) //! test max_unstable ... bench: 635,435 ns/iter (+/- 9,683) //! test sort ... bench: 62,585,547 ns/iter (+/- 1,361,258) //! test sort_unstable ... bench: 34,595,265 ns/iter (+/- 739,255) //! ``` //! //! n = `1000`, len = `10_000_000`: //! ```text //! test binary_heap ... bench: 78,343,761 ns/iter (+/- 2,579,283) //! test max ... bench: 14,970,673 ns/iter (+/- 580,660) //! test max_unstable ... bench: 14,870,344 ns/iter (+/- 573,286) //! test sort ... bench: 716,240,206 ns/iter (+/- 23,188,586) //! test sort_unstable ... bench: 385,681,859 ns/iter (+/- 19,847,519) //! ``` #![cfg_attr(not(feature = "std"), no_std)] #![doc(html_root_url = "https://docs.rs/out/2.0.1")] #![deny( bad_style, bare_trait_objects, missing_debug_implementations, missing_docs, unused_import_braces, unused_qualifications )] #[cfg(not(feature = "std"))] use core::{cmp::Ordering, mem, slice}; #[cfg(feature = "std")] use std::{cmp::Ordering, mem, slice}; /// Get the `n` largest items. /// /// This method is stable, i.e. it preserves the order of equal elements. /// /// # Panics /// Panics if `n > len`. /// /// # Examples /// ``` /// let mut v = [-5, 4, 1, -3, 2]; /// let max = out::max(&mut v, 3); /// assert_eq!(max, [1, 2, 4]); /// ``` #[inline] #[cfg(feature = "std")] pub fn max<T: Ord>(v: &mut [T], n: usize) -> &mut [T] { max_by(v, n, T::cmp) } /// Get the `n` largest items. /// /// This method is not stable, i.e. it may not preserve the order of equal elements. /// This method should be faster than `max` in most cases, and can be used with `no_std`. /// /// # Panics /// Panics if `n > len`. /// /// # Examples /// ``` /// let mut v = [-5, 4, 1, -3, 2]; /// let max = out::max_unstable(&mut v, 3); /// assert_eq!(max, [1, 2, 4]); /// ``` #[inline] pub fn max_unstable<T: Ord>(v: &mut [T], n: usize) -> &mut [T] { max_unstable_by(v, n, T::cmp) } /// Get the `n` largest items with a comparator function. /// /// This method is stable, i.e. it preserves the order of equal elements. /// /// # Panics /// Panics if `n > len`. /// /// # Examples /// ``` /// let mut v = [-5, 4, 1, -3, 2]; /// let min = out::max_by(&mut v, 3, |a, b| b.cmp(a)); /// assert_eq!(min, [1, -3, -5]); /// ``` #[inline] #[cfg(feature = "std")] pub fn max_by<T>(v: &mut [T], n: usize, mut cmp: impl FnMut(&T, &T) -> Ordering) -> &mut [T] { if n == 0 { return &mut []; } let (mut left, mut right) = v.split_at_mut(n); left.sort_by(&mut cmp); let mut i = 0; while i < right.len() { // Using `==` seems to be 10-30% faster than `!=`. if cmp(&right[i], &left[0]) == Ordering::Less { i += 1; } else if cmp(&right[i], &left[n / 2]) == Ordering::Greater { right.swap(i, 0); let mut j = n - 1; if cmp(&left[j], &right[0]) == Ordering::Greater { mem::swap(&mut left[j], &mut right[0]); while cmp(&left[j], &left[j - 1]) == Ordering::Less { left.swap(j, j - 1); j -= 1; } } unsafe { shift_slice_right(&mut left, &mut right, 1); } } else { let mut j = 0; mem::swap(&mut right[i], &mut left[j]); while j < n - 1 && cmp(&left[j], &left[j + 1]) != Ordering::Less { left.swap(j, j + 1); j += 1; } i += 1; } } left } /// Get the `n` largest items with a comparator function. /// /// This method is not stable, i.e. it may not preserve the order of equal elements. /// This method should be faster than `max_unstable_by` in most cases, and can be used with `no_std`. /// /// # Panics /// Panics if `n > len`. /// /// # Examples /// ``` /// let mut v = [-5, 4, 1, -3, 2]; /// let min = out::max_unstable_by(&mut v, 3, |a, b| b.cmp(a)); /// assert_eq!(min, [1, -3, -5]); /// ``` #[inline] pub fn max_unstable_by<T>( v: &mut [T], n: usize, mut cmp: impl FnMut(&T, &T) -> Ordering, ) -> &mut [T] { if n == 0 { return &mut []; } let (mut left, mut right) = v.split_at_mut(n); left.sort_unstable_by(&mut cmp); let mut i = 0; while i < right.len() { // Using `==` seems to be 10-30% faster than `!=`. if cmp(&left[0], &right[i]) == Ordering::Greater { i += 1; } else if cmp(&right[i], &left[n / 2]) == Ordering::Greater { right.swap(i, 0); let mut j = n - 1; if cmp(&left[j], &right[0]) == Ordering::Greater { mem::swap(&mut left[j], &mut right[0]); while cmp(&left[j], &left[j - 1]) == Ordering::Less { left.swap(j, j - 1); j -= 1; } } unsafe { shift_slice_right(&mut left, &mut right, 1); } } else { let mut j = 0; mem::swap(&mut right[i], &mut left[j]); while j < n - 1 && cmp(&left[j], &left[j + 1]) == Ordering::Greater { left.swap(j, j + 1); j += 1; } i += 1; } } left } /// Get the `n` largest items with a key extraction function. /// /// This method is stable, i.e. it preserves the order of equal elements. /// /// # Panics /// Panics if `n > len`. /// /// # Examples /// ``` /// let mut v = [-5_i32, 4, 1, -3, 2]; /// let max = out::max_by_key(&mut v, 3, |a| a.abs()); /// assert_eq!(max, [-3, 4, -5]); /// ``` #[inline] #[cfg(feature = "std")] pub fn max_by_key<T, K: Ord>(v: &mut [T], n: usize, mut cmp: impl FnMut(&T) -> K) -> &mut [T] { max_by(v, n, |a, b| cmp(a).cmp(&cmp(b))) } /// Get the `n` largest items with a key extraction function. /// /// This method is not stable, i.e. it may not preserve the order of equal elements. /// This method should be faster than `max_unstable_by_key` in most cases, and can be used with `no_std`. /// /// # Panics /// Panics if `n > len`. /// /// # Examples /// ``` /// let mut v = [-5_i32, 4, 1, -3, 2]; /// let max = out::max_unstable_by_key(&mut v, 3, |a| a.abs()); /// assert_eq!(max, [-3, 4, -5]); /// ``` #[inline] pub fn max_unstable_by_key<T, K: Ord>( v: &mut [T], n: usize, mut cmp: impl FnMut(&T) -> K, ) -> &mut [T] { max_unstable_by(v, n, |a, b| cmp(a).cmp(&cmp(b))) } /// Shift the left slice to the right while shrinking the right slice by `count`. /// /// ```text /// [a, b][c, d, e] -> a [b, c][d, e] /// ``` #[inline] unsafe fn shift_slice_right<T>(left: &mut &mut [T], right: &mut &mut [T], count: usize) { let len = left.len(); let ptr = left.as_mut_ptr(); *left = slice::from_raw_parts_mut(ptr.add(count), len); let len = right.len(); let ptr = right.as_mut_ptr(); *right = slice::from_raw_parts_mut(ptr.add(count), len - count); }