#843845 (C++14) No.2226 Hello, Forgotten World!

提出ソース
結果

問題	No.2226 Hello, Forgotten World!
ユーザー	maksim
提出日時	2023-02-24 23:13:57
言語	C++14 (gcc 12.3.0 + boost 1.83.0)
結果	AC
実行時間	57 ms / 2,000 ms
コード長	21,081 bytes
コンパイル時間	11,110 ms
コンパイル使用メモリ	505,736 KB
実行使用メモリ	28,072 KB
最終ジャッジ日時	2023-10-11 06:52:50
合計ジャッジ時間	11,330 ms
ジャッジサーバーID （参考情報）	judge14 / judge11
このコードへのチャレンジ
（要ログイン）
テストケース

テストケース表示
入力	結果	実行時間実行使用メモリ
testcase_00	AC	14 ms 27,992 KB
testcase_01	AC	42 ms 27,976 KB
testcase_02	AC	57 ms 28,024 KB
testcase_03	AC	40 ms 28,012 KB
testcase_04	AC	20 ms 27,988 KB
testcase_05	AC	45 ms 27,988 KB
testcase_06	AC	25 ms 28,020 KB
testcase_07	AC	38 ms 27,952 KB
testcase_08	AC	17 ms 28,048 KB
testcase_09	AC	35 ms 28,072 KB
権限があれば一括ダウンロードができます
ソースコード

raw source code
#include <bits/stdc++.h>

using namespace std;
const int maxn=1e6+5;
int sp[maxn][20];
template <typename T>
vector<int> suffix_array(int n, const T &s, int char_bound) {
  vector<int> a(n);
  if (n == 0) {
    return a;
  }
  if (char_bound != -1) {
    vector<int> aux(char_bound, 0);
    for (int i = 0; i < n; i++) {
      aux[s[i]]++;
    }
    int sum = 0;
    for (int i = 0; i < char_bound; i++) {
      int add = aux[i];
      aux[i] = sum;
      sum += add;
    }
    for (int i = 0; i < n; i++) {
      a[aux[s[i]]++] = i;
    }
  } else {
    iota(a.begin(), a.end(), 0);
    sort(a.begin(), a.end(), [&s](int i, int j) { return s[i] < s[j]; });
  }
  vector<int> sorted_by_second(n);
  vector<int> ptr_group(n);
  vector<int> new_group(n);
  vector<int> group(n);
  group[a[0]] = 0;
  for (int i = 1; i < n; i++) {
    group[a[i]] = group[a[i - 1]] + (!(s[a[i]] == s[a[i - 1]]));
  }
  int cnt = group[a[n - 1]] + 1;
  int step = 1;
  while (cnt < n) {
    int at = 0;
    for (int i = n - step; i < n; i++) {
      sorted_by_second[at++] = i;
    }
    for (int i = 0; i < n; i++) {
      if (a[i] - step >= 0) {
        sorted_by_second[at++] = a[i] - step;
      }
    }
    for (int i = n - 1; i >= 0; i--) {
      ptr_group[group[a[i]]] = i;
    }
    for (int i = 0; i < n; i++) {
      int x = sorted_by_second[i];
      a[ptr_group[group[x]]++] = x;
    }
    new_group[a[0]] = 0;
    for (int i = 1; i < n; i++) {
      if (group[a[i]] != group[a[i - 1]]) {
        new_group[a[i]] = new_group[a[i - 1]] + 1;
      } else {
        int pre = (a[i - 1] + step >= n ? -1 : group[a[i - 1] + step]);
        int cur = (a[i] + step >= n ? -1 : group[a[i] + step]);
        new_group[a[i]] = new_group[a[i - 1]] + (pre != cur);
      }
    }
    swap(group, new_group);
    cnt = group[a[n - 1]] + 1;
    step <<= 1;
  }
  return a;
}

template <typename T>
vector<int> suffix_array(const T &s, int char_bound) {
  return suffix_array((int) s.size(), s, char_bound);
}

template <typename T>
vector<int> build_lcp(int n, const T &s, const vector<int> &sa) {
  assert((int) sa.size() == n);
  vector<int> pos(n);
  for (int i = 0; i < n; i++) {
    pos[sa[i]] = i;
  }
  vector<int> lcp(max(n - 1, 0));
  int k = 0;
  for (int i = 0; i < n; i++) {
    k = max(k - 1, 0);
    if (pos[i] == n - 1) {
      k = 0;
    } else {
      int j = sa[pos[i] + 1];
      while (i + k < n && j + k < n && s[i + k] == s[j + k]) {
        k++;
      }
      lcp[pos[i]] = k;
    }
  }
  return lcp;
}

template <typename T>
vector<int> build_lcp(const T &s, const vector<int> &sa) {
  return build_lcp((int) s.size(), s, sa);
}
typedef long long ll;
const int p=998244353;
int po(int a,int b) {if(b==0) return 1; if(b==1) return a; if(b%2==0) {int u=po(a,b/2);return (u*1LL*u)%p;} else {int u=po(a,b-1);return (a*1LL*u)%p;}}
int inv(int x) {return po(x,p-2);}
#pragma GCC target("avx2")
#pragma GCC optimize("O3")
#pragma GCC optimize("unroll-loops")

#include <immintrin.h>

#include <algorithm>
#include <cstdint>
#include <cstdio>
#include <cstring>
#include <vector>

constexpr std::uint32_t MOD = 998244353, PRIMITIVE_ROOT = 3;
const unsigned int MAX_N = 524288;

std::uint32_t mod_mul(std::uint32_t a, std::uint32_t b) { return static_cast<std::uint64_t>(a) * b % MOD; }

std::uint32_t pow(std::uint32_t a, std::uint32_t b) {
    std::uint32_t ret = 1;
    while (b) {
        if (b & 1) ret = mod_mul(ret, a);
        a = mod_mul(a, a);
        b >>= 1;
    }
    return ret;
}

__m256i mod_add_simd(__m256i a, __m256i b) {
    __m256i c = _mm256_add_epi32(a, b);
    __m256i d = _mm256_sub_epi32(c, _mm256_set1_epi32(MOD));
    __m256i mask = _mm256_cmpgt_epi32(_mm256_set1_epi32(0), d);
    return _mm256_blendv_epi8(d, c, mask);
}

__m256i mod_sub_simd(__m256i a, __m256i b) {
    __m256i c = _mm256_sub_epi32(a, b);
    __m256i d = _mm256_add_epi32(c, _mm256_set1_epi32(MOD));
    __m256i mask = _mm256_cmpgt_epi32(_mm256_set1_epi32(0), c);
    return _mm256_blendv_epi8(c, d, mask);
}

__m256i reduce_simd(__m256i a) {
    __m256i b = _mm256_sub_epi32(a, _mm256_set1_epi32(MOD));
    __m256i mask = _mm256_cmpgt_epi32(_mm256_set1_epi32(0), b);
    return _mm256_blendv_epi8(b, a, mask);
}

__m256i add_simd(__m256i a, __m256i b) { return _mm256_add_epi32(a, b); }

__m256i sub_simd(__m256i a, __m256i b) {
    __m256i c = _mm256_sub_epi32(a, b);
    return _mm256_add_epi32(c, _mm256_set1_epi32(MOD));
}

__m256i mulhi_simd(__m256i a, __m256i b) {
    __m256i a0 = _mm256_unpacklo_epi32(a, a);
    __m256i a1 = _mm256_unpackhi_epi32(a, a);
    __m256i b0 = _mm256_unpacklo_epi32(b, b);
    __m256i b1 = _mm256_unpackhi_epi32(b, b);

    __m256i c0 = _mm256_mul_epu32(a0, b0);
    __m256i c1 = _mm256_mul_epu32(a1, b1);

    return (__m256i)_mm256_shuffle_ps((__m256)c0, (__m256)c1, _MM_SHUFFLE(3, 1, 3, 1));
}

struct ConstMulSimd {
    __m256i a;
    __m256i a_div;

    explicit ConstMulSimd(std::uint32_t a) : a(_mm256_set1_epi32(a)) { a_div = _mm256_set1_epi32((std::uint64_t(a) << 32) / MOD); }

    ConstMulSimd(std::uint32_t a0, std::uint32_t a1, std::uint32_t a2, std::uint32_t a3, std::uint32_t a4, std::uint32_t a5, std::uint32_t a6, std::uint32_t a7) {
        a = _mm256_setr_epi32(a0, a1, a2, a3, a4, a5, a6, a7);
        std::uint32_t a_div0, a_div1, a_div2, a_div3, a_div4, a_div5, a_div6, a_div7;
        a_div0 = (std::uint64_t(a0) << 32) / MOD;
        a_div1 = (std::uint64_t(a1) << 32) / MOD;
        a_div2 = (std::uint64_t(a2) << 32) / MOD;
        a_div3 = (std::uint64_t(a3) << 32) / MOD;
        a_div4 = (std::uint64_t(a4) << 32) / MOD;
        a_div5 = (std::uint64_t(a5) << 32) / MOD;
        a_div6 = (std::uint64_t(a6) << 32) / MOD;
        a_div7 = (std::uint64_t(a7) << 32) / MOD;
        a_div = _mm256_setr_epi32(a_div0, a_div1, a_div2, a_div3, a_div4, a_div5, a_div6, a_div7);
    }

    __m256i mul(__m256i b) {
        __m256i q = mulhi_simd(a_div, b);
        __m256i c = _mm256_mullo_epi32(a, b);
        __m256i qm = _mm256_mullo_epi32(q, _mm256_set1_epi32(MOD));
        __m256i r = _mm256_sub_epi32(c, qm);
        return r;
    }

    __m256i mul_mod(__m256i b) {
        __m256i r = mul(b);
        return reduce_simd(r);
    }
};

void ntt(unsigned int n, std::uint32_t *a) {
    std::uint32_t w = pow(PRIMITIVE_ROOT, (MOD - 1) / n);

    unsigned int m = n;
    while (m >= 16) {
        unsigned int mh = m / 2;

        std::uint32_t w0, w1, w2, w3, w4, w5, w6, w7, w8;
        w0 = 1;
        w1 = w;
        w2 = mod_mul(w1, w1);
        w3 = mod_mul(w2, w1);
        w4 = mod_mul(w2, w2);
        w5 = mod_mul(w4, w1);
        w6 = mod_mul(w4, w2);
        w7 = mod_mul(w4, w3);
        w8 = mod_mul(w4, w4);

        ConstMulSimd wi_mul(w0, w1, w2, w3, w4, w5, w6, w7);

        for (unsigned int i = 0; i < n; i += m) {
            std::uint32_t wj = 1;

            for (unsigned int j = 0; j < mh; j += 8) {
                unsigned int j0 = i + j, j1 = j0 + mh;

                __m256i x0 = _mm256_load_si256(reinterpret_cast<__m256i *>(a + j0));
                __m256i x1 = _mm256_load_si256(reinterpret_cast<__m256i *>(a + j1));

                __m256i y0 = mod_add_simd(x0, x1);
                __m256i y1 = wi_mul.mul(sub_simd(x0, x1));

                ConstMulSimd wj_mul(wj);
                y1 = wj_mul.mul_mod(y1);

                _mm256_store_si256(reinterpret_cast<__m256i *>(a + j0), y0);
                _mm256_store_si256(reinterpret_cast<__m256i *>(a + j1), y1);

                wj = mod_mul(wj, w8);
            }
        }

        m = mh;
        w = w2;
    }

    std::uint32_t w0, w1, w2, w3;
    w0 = 1;
    w1 = w;
    w2 = mod_mul(w1, w1);
    w3 = mod_mul(w2, w1);
    ConstMulSimd mul_1(1, 1, 1, 1, w0, w1, w2, w3), mul_2(1, 1, w0, w2, 1, 1, w0, w2);

    for (unsigned int i = 0; i < n; i += 8) {
        __m256i x00, x01, x10, x11, x20, x21, x30, x31, x40, x41, x50, x51, x6;
        x00 = _mm256_load_si256(reinterpret_cast<__m256i *>(a + i));
        x01 = _mm256_permute4x64_epi64(x00, 0b01001110);
        x10 = add_simd(x00, x01);
        x11 = sub_simd(x01, x00);
        x20 = mul_1.mul_mod(_mm256_blend_epi32(x10, x11, 0b11110000));
        x21 = _mm256_shuffle_epi32(x20, 0b01001110);
        x30 = add_simd(x20, x21);
        x31 = sub_simd(x21, x20);
        x40 = mul_2.mul_mod(_mm256_blend_epi32(x30, x31, 0b11001100));
        x41 = _mm256_shuffle_epi32(x40, 0b10110001);
        x50 = add_simd(x40, x41);
        x51 = sub_simd(x41, x40);
        x6 = reduce_simd(_mm256_blend_epi32(x50, x51, 0b10101010));
        _mm256_store_si256(reinterpret_cast<__m256i *>(a + i), x6);
    }
}

void intt(unsigned int n, std::uint32_t *a) {
    std::vector<std::uint32_t> w_vec;
    std::uint32_t w = pow(PRIMITIVE_ROOT, MOD - 1 - (MOD - 1) / n);
    for (unsigned int i = n; i >= 16; i >>= 1) {
        w_vec.push_back(w);
        w = mod_mul(w, w);
    }

    std::uint32_t w0, w1, w2, w3;
    w0 = 1;
    w1 = w;
    w2 = mod_mul(w1, w1);
    w3 = mod_mul(w2, w1);
    ConstMulSimd mul_1(1, 1, 1, 1, w0, w1, w2, w3), mul_2(1, 1, w0, w2, 1, 1, w0, w2);

    for (unsigned int i = 0; i < n; i += 8) {
        __m256i x00, x01, x10, x11, x20, x21, x30, x31, x40, x41, x50, x51, x6;
        x00 = _mm256_load_si256(reinterpret_cast<__m256i *>(a + i));
        x01 = _mm256_shuffle_epi32(x00, 0b10110001);
        x10 = add_simd(x00, x01);
        x11 = sub_simd(x01, x00);
        x20 = mul_2.mul_mod(_mm256_blend_epi32(x10, x11, 0b10101010));
        x21 = _mm256_shuffle_epi32(x20, 0b01001110);
        x30 = add_simd(x20, x21);
        x31 = sub_simd(x21, x20);
        x40 = mul_1.mul_mod(_mm256_blend_epi32(x30, x31, 0b11001100));
        x41 = _mm256_permute4x64_epi64(x40, 0b01001110);
        x50 = add_simd(x40, x41);
        x51 = sub_simd(x41, x40);
        x6 = reduce_simd(_mm256_blend_epi32(x50, x51, 0b11110000));
        _mm256_store_si256(reinterpret_cast<__m256i *>(a + i), x6);
    }

    unsigned int m = 16;
    while (m <= n) {
        unsigned int mh = m / 2;

        std::uint32_t w = w_vec.back();
        w_vec.pop_back();
        std::uint32_t w0, w1, w2, w3, w4, w5, w6, w7, w8;
        w0 = 1;
        w1 = w;
        w2 = mod_mul(w1, w1);
        w3 = mod_mul(w2, w1);
        w4 = mod_mul(w2, w2);
        w5 = mod_mul(w4, w1);
        w6 = mod_mul(w4, w2);
        w7 = mod_mul(w4, w3);
        w8 = mod_mul(w4, w4);

        ConstMulSimd wi_mul(w0, w1, w2, w3, w4, w5, w6, w7);

        for (unsigned int i = 0; i < n; i += m) {
            std::uint32_t wj = 1;

            for (unsigned int j = 0; j < mh; j += 8) {
                unsigned int j0 = i + j, j1 = j0 + mh;

                __m256i x0 = _mm256_load_si256(reinterpret_cast<__m256i *>(a + j0));
                __m256i x1 = _mm256_load_si256(reinterpret_cast<__m256i *>(a + j1));

                x1 = wi_mul.mul_mod(x1);

                ConstMulSimd wj_mul(wj);
                x1 = wj_mul.mul_mod(x1);

                __m256i y0 = mod_add_simd(x0, x1);
                __m256i y1 = mod_sub_simd(x0, x1);

                _mm256_store_si256(reinterpret_cast<__m256i *>(a + j0), y0);
                _mm256_store_si256(reinterpret_cast<__m256i *>(a + j1), y1);

                wj = mod_mul(wj, w8);
            }
        }

        m *= 2;
    }

    std::uint32_t n_inv = pow(n, MOD - 2);
    ConstMulSimd mul_n_inv(n_inv);
    for (unsigned int i = 0; i < n; i += 8) {
        __m256i x0 = _mm256_load_si256(reinterpret_cast<__m256i *>(a + i));
        _mm256_store_si256(reinterpret_cast<__m256i *>(a + i), mul_n_inv.mul_mod(x0));
    }
}

void convolution(unsigned int n, std::uint32_t *a, std::uint32_t *b) {
    ntt(n, a);
    ntt(n, b);

    for (unsigned int i = 0; i < n; i++) {
        a[i] = mod_mul(a[i], b[i]);
    }

    intt(n, a);
}

struct IO {
    const static unsigned int MAX_STDIO_SIZE = 10 * 2 * MAX_N + 14;
    char stdio_buf[MAX_STDIO_SIZE];

    std::uint32_t char_arr_to_int(std::uint64_t x) {
        x = (x >> 8) + x * 10;
        x = ((x >> 16) & 0x000000FF000000FFull) + (x & 0x000000FF000000FFull) * 100;
        x = (x >> 32) + x * 10000;
        return x & 0xFFFFFFFFull;
    }

    std::uint64_t int_to_char_arr(std::uint32_t x) {
        std::uint64_t r = x, q;
        q = ((r * 109951163) & 0xFFFFFF0000000000ull) >> 40;
        r = (r << 32) - q * 42949672959999ull;
        q = ((r * 10486) & 0xFFF00000FFF00000ull) >> 20;
        r = (r << 16) - q * 6553599ull;
        q = ((r * 103) & 0xFC00FC00FC00FC00ull) >> 10;
        r = (r << 8) - q * 2559ull;
        return r;
    }

    std::uint32_t read_int(char *&buf) {
        std::uint64_t x = *reinterpret_cast<std::uint64_t *>(buf);
        std::uint64_t space = ((x + 0x1010101010101010ull) & 0x4040404040404040) ^ 0x4040404040404040;

        if (space == 0) {
            std::uint32_t res = char_arr_to_int(x - 0x3030303030303030ull);
            if (buf[8] <= ' ') {
                buf += 9;
            } else {
                res = res * 10 + (buf[8] - '0');
                buf += 10;
            }
            return res;
        }

        unsigned int space_pos = __builtin_ffsll(space);
        std::uint64_t mask = (1ULL << (space_pos - 7)) - 1;
        std::uint32_t res = char_arr_to_int(((x - 0x3030303030303030ull) & mask) << (71 - space_pos));
        buf += (space_pos + 1) >> 3;
        return res;
    }

    void read(unsigned int &n, unsigned int &m, std::uint32_t *a, std::uint32_t *b) {
        std::fread(stdio_buf, 1, MAX_STDIO_SIZE, stdin);

        char *buf = stdio_buf;
        n = read_int(buf);
        m = read_int(buf);
        for (unsigned int i = 0; i < n; i++) {
            a[i] = read_int(buf);
        }
        for (unsigned int i = 0; i < m; i++) {
            b[i] = read_int(buf);
        }
    }

    void write_int(char *&buf, std::uint32_t x) {
        constexpr std::uint32_t eight_digit = 100000000;
        if (x >= eight_digit) [[likely]] {
            std::uint32_t upper = x / eight_digit, lower = x % eight_digit;
            *buf = '0' + upper;
            buf++;
            *reinterpret_cast<std::uint64_t *>(buf) = int_to_char_arr(lower) + 0x3030303030303030ull;
            buf += 8;
        } else {
            std::uint64_t char_arr = int_to_char_arr(x);
            unsigned int padding = __builtin_ctzll(char_arr + (1ull << 63)) >> 3, len = 8 - padding;
            char_arr += 0x3030303030303030ull;
            *reinterpret_cast<std::uint64_t *>(buf) = char_arr >> (padding << 3);
            buf += len;
        }
        *buf = ' ';
        buf++;
    }

    void write(unsigned int n, std::uint32_t *a) {
        char *buf = stdio_buf;
        for (unsigned int i = 0; i < n; i++) {
            write_int(buf, a[i]);
        }
        unsigned int len = buf - stdio_buf;
        std::fwrite(stdio_buf, 1, len, stdout);
    }
} io;
alignas(32) static std::uint32_t v1[MAX_N * 2], v2[MAX_N * 2];
template<int M, int K, int G> struct Fft {
  // 1, 1/4, 1/8, 3/8, 1/16, 5/16, 3/16, 7/16, ...
  int g[1 << (K - 1)];
  constexpr Fft() : g() { //if tl constexpr...
    static_assert(K >= 2, "Fft: K >= 2 must hold");
    g[0] = 1;
    g[1 << (K - 2)] = G;
    for (int l = 1 << (K - 2); l >= 2; l >>= 1) {
      g[l >> 1] = (static_cast<long long>(g[l]) * g[l]) % M;
    }
    assert((static_cast<long long>(g[1]) * g[1]) % M == M - 1);
    for (int l = 2; l <= 1 << (K - 2); l <<= 1) {
      for (int i = 1; i < l; ++i) {
        g[l + i] = (static_cast<long long>(g[l]) * g[i]) % M;
      }
    }
  }
  void fft(vector<int> &x) const {
    const int n = x.size();
    assert(!(n & (n - 1)) && n <= 1 << K);
    for (int h = __builtin_ctz(n); h--; ) {
      const int l = 1 << h;
      for (int i = 0; i < n >> 1 >> h; ++i) {
        for (int j = i << 1 << h; j < ((i << 1) + 1) << h; ++j) {
          const int t = (static_cast<long long>(g[i]) * x[j | l]) % M;
          if ((x[j | l] = x[j] - t) < 0) x[j | l] += M;
          if ((x[j] += t) >= M) x[j] -= M;
        }
      }
    }
    for (int i = 0, j = 0; i < n; ++i) {
      if (i < j) std::swap(x[i], x[j]);
      for (int l = n; (l >>= 1) && !((j ^= l) & l); ) {}
    }
  }
  vector<int> convolution(const vector<int> &a, const vector<int> &b) const {
    if(a.empty() || b.empty()) return {};
    const int na = a.size(), nb = b.size();
    int n, invN = 1;
    for (n = 1; n < na + nb - 1; n <<= 1) invN = ((invN & 1) ? (invN + M) : invN) >> 1;
    vector<int> x(n, 0), y(n, 0);
    std::copy(a.begin(), a.end(), x.begin());
    std::copy(b.begin(), b.end(), y.begin());
    fft(x);
    fft(y);
    for (int i = 0; i < n; ++i) x[i] = (((static_cast<long long>(x[i]) * y[i]) % M) * invN) % M;
    std::reverse(x.begin() + 1, x.end());
    fft(x);
    x.resize(na + nb - 1);
    return x;
  }
};
Fft<998244353,23,31> muls;
vector<int> operator *(vector<int> v1,vector<int> v2)
{
    return muls.convolution(v1,v2);
}
int32_t main()
{
    ios_base::sync_with_stdio(false);cin.tie(0);cout.tie(0);
    /*string s="abbaaaf";
    vector<int> sa=suffix_array(s,256);
    vector<int> lc=build_lcp(s,sa);
    for(int i:sa) cout<<i<<' '; cout<<endl;
    for(int i:lc) cout<<i<<' '; cout<<endl;*/
    int t;cin>>t;
    while(t--)
    {
        int n,m;cin>>n;
        string s1,s2;cin>>s1;
        s2="helloworld";
        m=s2.size();
        string s0=s1;
        bool ok[n-m+1];for(int i=0;i<=n-m;++i) ok[i]=true;
        int h=1;while(h<=n+m) h*=2;
        for(char c='a';c<='z';++c)
        {
            int cnt=0;
            for(int i=0;i<h;++i) {v1[i]=0;v2[i]=0;}
            for(int i=0;i<n;++i) v1[n-i-1]=(s1[i]==c || s1[i]=='?');
            for(int i=0;i<m;++i) {v2[i]=(s2[i]==c);cnt+=v2[i];}
            if(h>=2000) convolution(h, v1, v2);
            else
            {
                vector<int> v11;for(int i=0;i<n;++i) v11.push_back(v1[i]);
                vector<int> v21;for(int i=0;i<m;++i) v21.push_back(v2[i]);
                vector<int> v31=v11*v21;
                for(int i=m-1;i<=n-1;++i) v1[i]=v31[i];
            }
            for(int i=m-1;i<=n-1;++i)
            {
                ok[n-i-1]&=(v1[i]==cnt);
            }
        }
        for(int i=0;i<s1.size();++i) if(s1[i]=='?') s1[i]='a';
        string s=s1;s.push_back('$');s+=s2;
        if(accumulate(ok,ok+n-m+1,0LL)==0)
        {
            cout<<(-1)<<'\n';
            continue;
        }
        vector<int> sa=suffix_array(s,256);
        vector<int> lc=build_lcp(s,sa);
        vector<int> pos(s.size());
        for(int i=0;i<s.size();++i)
        {
            pos[sa[i]]=i;
        }
        int lcsz=lc.size();
        for(int i=0;i<lcsz;++i)
        {
            sp[i][0]=lc[i];
        }
        for(int j=1;j<20;++j)
        {
            for(int i=0;i<=lcsz-(1<<j);++i)
            {
                sp[i][j]=min(sp[i][j-1],sp[i+(1<<(j-1))][j-1]);
            }
        }
        int l1=0;int l2=s1.size()+1;
        int inf=1e9;
        /*string stupid;
        {
            for(int i=0;i<=n-m;++i) if(ok[i])
            {
                 string ans=s1.substr(0,i);ans+=s2;ans+=s1.substr(i+m,n-m-i);
                 if(stupid.empty()) stupid=ans;
                 else stupid=min(stupid,ans);
            }
        }*/
        auto get=[&](int i,int j)
        {
            int i1=i;int j1=j;
            i=pos[i];j=pos[j];
            if(i>j) swap(i,j);
            if(i==j) return inf;
            int o=31-__builtin_clz(j-i);
            return min(sp[i][o],sp[j-(1<<o)][o]);
        };
        auto cmp=[&](int i,int j)->bool
        {
            if(i==j) return false;
            bool sw=false;
            if(i>j) {swap(i,j);sw=true;}
            {
                if(j-i<s2.size())
                {
                int pos1=get(i,l2);
                if(pos1<j-i)
                {
                    return (s[l2+pos1]<s[i+pos1])^sw;
                }
                int pos2=get(l2,l2+j-i);
                if(pos2<m-(j-i))
                {
                    return (s[l2+j-i+pos2]<s[l2+pos2])^sw;
                }
                int pos3=get(i+m,l2+m-(j-i));
                if(pos3<(j-i))
                {
                    return (s[i+m+pos3]<s[l2+m-(j-i)+pos3])^sw;
                }
                return false;
                }
                else
                {
                int pos1=get(i,l2);
                if(pos1<m)
                {
                    return (s[l2+pos1]<s[i+pos1])^sw;
                }
                int pos2=get(j,l2);
                if(pos1<m)
                {
                    return (s[j+pos1]<s[l2+pos1])^sw;
                }
                return false;
                }
            }
        };
        vector<int> v;
        for(int i=0;i<=n-m;++i) if(ok[i]) v.push_back(i);
        int pos1=(*min_element(v.begin(),v.end(),cmp));
        string ans=s1.substr(0,pos1);
        ans+=s2;
        ans+=s1.substr(pos1+m,n-m-pos1);
        /*if(ans!=stupid)
        {
            cout<<s0<<' '<<s1<<' '<<s2<<endl;
            cout<<ans<<' '<<stupid<<endl;
        }
        assert(ans==stupid);*/
        cout<<ans<<'\n';
    }
    return 0;
}
/*
1
12 3
t?t?r???g?s?
kog
*/
yukicoder

結果

テストケース

ソースコード
