Tifa's CP Library

:heavy_check_mark: rsort32 (src/code/fast/rsort32.hpp)

Depends on

Verified with

Code

#ifndef TIFALIBS_FAST_RSORT32
#define TIFALIBS_FAST_RSORT32

#include "../util/util.hpp"

namespace tifa_libs {

template <class T>
requires(sizeof(T) == 4)
void rsort32(vec<T>& a) {
  static u32 _0[256], _1[256], _2[256], _3[256];
  u32 n = (u32)a.size();
  vecu b(n);
  u32 *a_ = a.data(), *b_ = b.data();
  for (u32 *_ = a_ + n, *i = a_; i < _; ++i) {
    ++_0[*i & 255];
    ++_1[*i >> 8 & 255];
    ++_2[*i >> 16 & 255];
    ++_3[*i >> 24 & 255];
  }
  for (u32 i = 1; i < 256; ++i) {
    _0[i] += _0[i - 1];
    _1[i] += _1[i - 1];
    _2[i] += _2[i - 1];
    _3[i] += _3[i - 1];
  }
  for (u32* i = a_ + n; --i >= a_;) b_[--_0[*i & 255]] = *i;
  for (u32* i = b_ + n; --i >= b_;) a_[--_1[*i >> 8 & 255]] = *i;
  for (u32* i = a_ + n; --i >= a_;) b_[--_2[*i >> 16 & 255]] = *i;
  for (u32* i = b_ + n; --i >= b_;) a_[--_3[*i >> 24 & 255]] = *i;
  if constexpr (std::is_signed_v<T>) {
    u32 i = n;
    while (i && a[i - 1] < 0) --i;
    std::rotate(a_, a_ + i, a_ + n);
  }
}

}  // namespace tifa_libs

#endif
#line 1 "src/code/fast/rsort32.hpp"



#line 1 "src/code/util/util.hpp"



#include <bits/stdc++.h>

template <class T>
constexpr T abs(T x) { return x < 0 ? -x : x; }

using i8 = int8_t;
using i16 = int16_t;
using i32 = int32_t;
using i64 = int64_t;
using i128 = __int128_t;
using isz = ptrdiff_t;

using u8 = uint8_t;
using u16 = uint16_t;
using u32 = uint32_t;
using u64 = uint64_t;
using u128 = __uint128_t;
using usz = size_t;

using f32 = float;
using f64 = double;
using f128 = long double;

template <class T>
using ptt = std::pair<T, T>;
template <class T>
using pt3 = std::tuple<T, T, T>;
template <class T>
using pt4 = std::tuple<T, T, T, T>;

template <class T, usz N>
using arr = std::array<T, N>;
template <class T>
using vec = std::vector<T>;
template <class T>
using vvec = vec<vec<T>>;
template <class T>
using v3ec = vec<vvec<T>>;
template <class U, class T>
using vecp = vec<std::pair<U, T>>;
template <class U, class T>
using vvecp = vvec<std::pair<U, T>>;
template <class T>
using vecpt = vec<ptt<T>>;
template <class T>
using vvecpt = vvec<ptt<T>>;

template <class T, class C = std::less<T>>
using pq = std::priority_queue<T, vec<T>, C>;
template <class T>
using pqg = std::priority_queue<T, vec<T>, std::greater<T>>;

using strn = std::string;
using strnv = std::string_view;

using vecu = vec<u32>;
using vvecu = vvec<u32>;
using v3ecu = v3ec<u32>;
using vecu64 = vec<u64>;
using vecb = vec<bool>;
using vvecb = vvec<bool>;

#ifdef ONLINE_JUDGE
#undef assert
#define assert(x) 42
#endif

using namespace std::literals;

constexpr i8 operator""_i8(unsigned long long x) { return (i8)x; }
constexpr i16 operator""_i16(unsigned long long x) { return (i16)x; }
constexpr i32 operator""_i32(unsigned long long x) { return (i32)x; }
constexpr i64 operator""_i64(unsigned long long x) { return (i64)x; }
constexpr isz operator""_iz(unsigned long long x) { return (isz)x; }

constexpr u8 operator""_u8(unsigned long long x) { return (u8)x; }
constexpr u16 operator""_u16(unsigned long long x) { return (u16)x; }
constexpr u32 operator""_u32(unsigned long long x) { return (u32)x; }
constexpr u64 operator""_u64(unsigned long long x) { return (u64)x; }
constexpr usz operator""_uz(unsigned long long x) { return (usz)x; }

inline const auto fn_0 = [](auto&&...) {};


#line 5 "src/code/fast/rsort32.hpp"

namespace tifa_libs {

template <class T>
requires(sizeof(T) == 4)
void rsort32(vec<T>& a) {
  static u32 _0[256], _1[256], _2[256], _3[256];
  u32 n = (u32)a.size();
  vecu b(n);
  u32 *a_ = a.data(), *b_ = b.data();
  for (u32 *_ = a_ + n, *i = a_; i < _; ++i) {
    ++_0[*i & 255];
    ++_1[*i >> 8 & 255];
    ++_2[*i >> 16 & 255];
    ++_3[*i >> 24 & 255];
  }
  for (u32 i = 1; i < 256; ++i) {
    _0[i] += _0[i - 1];
    _1[i] += _1[i - 1];
    _2[i] += _2[i - 1];
    _3[i] += _3[i - 1];
  }
  for (u32* i = a_ + n; --i >= a_;) b_[--_0[*i & 255]] = *i;
  for (u32* i = b_ + n; --i >= b_;) a_[--_1[*i >> 8 & 255]] = *i;
  for (u32* i = a_ + n; --i >= a_;) b_[--_2[*i >> 16 & 255]] = *i;
  for (u32* i = b_ + n; --i >= b_;) a_[--_3[*i >> 24 & 255]] = *i;
  if constexpr (std::is_signed_v<T>) {
    u32 i = n;
    while (i && a[i - 1] < 0) --i;
    std::rotate(a_, a_ + i, a_ + n);
  }
}

}  // namespace tifa_libs


Back to top page