/tmp/solutions/build/range_affine_range_sum-main.cpp:
    1|       |#include <common.h>
    2|       |#include <toy/bit.h>
    3|       |prelude;
    4|       |
    5|       |namespace {
    6|       |
    7|       |constexpr int N = 1e6;
    8|       |constexpr int P = 998244353;
    9|       |
   10|       |struct affine {
   11|       |  u32 a, b;
   12|  14.6M|  auto operator+(affine t) -> affine {
   13|  14.6M|    return {u32(u64(t.a) * a % P), u32((u64(t.a) * b + t.b) % P)};
   14|  14.6M|  }
   15|  14.6M|  auto operator+=(affine t) -> void { *this = *this + t; }
   16|       |};
   17|       |
   18|       |struct node {
   19|       |  u32 siz;
   20|       |  u32 sum;
   21|       |  affine aff;
   22|  14.6M|  auto operator+=(affine t) -> void {
   23|  14.6M|    aff += t;
   24|  14.6M|    sum = (u64(t.a) * sum + u64(t.b) * siz) % P;
   25|  14.6M|  }
   26|       |} a[N];
   27|       |
   28|   382k|void pushdown(int k) {
   29|  6.41M|  for (int i = log(k) - 1; i >= 0; --i) {
                                                 ^6.03M
  ------------------
  |  Branch (29:28): [True: 94.04%, False: 5.96%]
  ------------------
   30|  6.03M|    affine t = a[k >> i >> 1].aff;
   31|  6.03M|    a[k >> i >> 1].aff = affine{1, 0};
   32|  6.03M|    a[k >> i ^ 0] += t;
   33|  6.03M|    a[k >> i ^ 1] += t;
   34|  6.03M|  }
   35|   382k|}
   36|       |
   37|  5.89M|u32 mod(u32 x) { return x < P ? x : x - P; }
                                              ^2.94M^2.94M
  ------------------
  |  Branch (37:25): [True: 49.98%, False: 50.02%]
  ------------------
   38|       |
   39|   382k|void pushup(int k) {
   40|  6.03M|  for (k /= 2; k > 0; k /= 2) {
                                    ^5.65M
  ------------------
  |  Branch (40:16): [True: 93.66%, False: 6.34%]
  ------------------
   41|  5.65M|    a[k].sum = mod(a[k * 2].sum + a[k * 2 + 1].sum);
   42|  5.65M|  }
   43|   382k|}
   44|       |
   45|       |}; // namespace
   46|       |
   47|      1|int main() {
   48|      1|  rd rd;
   49|      1|  wt wt;
   50|      1|  int n = rd.uh();
   51|      1|  int q = rd.uh();
   52|  53.3k|  for (int i = 0; i < n; ++i) a[n + i].siz = 1, a[n + i].sum = rd.uw();
                                       ^53.3k^53.3k
  ------------------
  |  Branch (52:19): [True: 100.00%, False: 0.00%]
  ------------------
   53|  53.3k|  for (int i = n - 1; i >= 0; --i) {
                                            ^53.3k
  ------------------
  |  Branch (53:23): [True: 100.00%, False: 0.00%]
  ------------------
   54|  53.3k|    a[i].aff = {1, 0};
   55|  53.3k|    a[i].siz = a[2 * i].siz + a[2 * i + 1].siz;
   56|  53.3k|    a[i].sum = mod(a[2 * i].sum + a[2 * i + 1].sum);
   57|  53.3k|  }
   58|   382k|  while (q--) {
  ------------------
  |  Branch (58:10): [True: 100.00%, False: 0.00%]
  ------------------
   59|   382k|    let t = rd.u1();
   60|   382k|    if (t == 0) {
  ------------------
  |  Branch (60:9): [True: 50.04%, False: 49.96%]
  ------------------
   61|   191k|      int l = n + rd.uh();
   62|   191k|      int r = n + rd.uh() - 1;
   63|   191k|      pushdown(l--);
   64|   191k|      pushdown(r++);
   65|   191k|      int k = log(l ^ r);
   66|   191k|      affine x = {rd.uw(), rd.uw()};
   67|   191k|      int t, i;
   68|  1.46M|      for (t = ~l & ~(-1 << k), i = 31; t > 0; t -= 1 << i) {
                                                             ^1.27M
  ------------------
  |  Branch (68:41): [True: 86.96%, False: 13.04%]
  ------------------
   69|  1.27M|        i = log(t);
   70|  1.27M|        a[l >> i ^ 1] += x;
   71|  1.27M|      }
   72|   191k|      pushup(l >> i);
   73|  1.48M|      for (t = +r & ~(-1 << k), i = 31; t > 0; t -= 1 << i) {
                                                             ^1.29M
  ------------------
  |  Branch (73:41): [True: 87.14%, False: 12.86%]
  ------------------
   74|  1.29M|        i = log(t);
   75|  1.29M|        a[r >> i ^ 1] += x;
   76|  1.29M|      }
   77|   191k|      pushup(r >> i);
   78|   191k|    }
   79|   382k|    if (t == 1) {
  ------------------
  |  Branch (79:9): [True: 49.96%, False: 50.04%]
  ------------------
   80|   191k|      int l = n + rd.uh();
   81|   191k|      int r = n + rd.uh() - 1;
   82|   191k|      u64 sizL = 0, sumL = 0;
   83|   191k|      u64 sizR = 0, sumR = 0;
   84|  2.93M|      for (--l, ++r; l ^ r ^ 1;) {
  ------------------
  |  Branch (84:22): [True: 93.48%, False: 6.52%]
  ------------------
   85|  2.73M|        if (~l & 1) sizL += a[l ^ 1].siz, sumL += a[l ^ 1].sum;
                                  ^1.27M
  ------------------
  |  Branch (85:13): [True: 46.60%, False: 53.40%]
  ------------------
   86|  2.73M|        if (+r & 1) sizR += a[r ^ 1].siz, sumR += a[r ^ 1].sum;
                                  ^1.29M
  ------------------
  |  Branch (86:13): [True: 47.30%, False: 52.70%]
  ------------------
   87|  2.73M|        l /= 2, r /= 2;
   88|  2.73M|        sumL = (a[l].aff.a * sumL + a[l].aff.b * sizL) % P;
   89|  2.73M|        sumR = (a[r].aff.a * sumR + a[r].aff.b * sizR) % P;
   90|  2.73M|      }
   91|   191k|      sumL = mod(u32(sumL + sumR));
   92|   191k|      sizL += sizR;
   93|   498k|      for (l /= 2; l > 0; l /= 2) {
                                        ^307k
  ------------------
  |  Branch (93:20): [True: 61.67%, False: 38.33%]
  ------------------
   94|   307k|        sumL = (a[l].aff.a * sumL + a[l].aff.b * sizL) % P;
   95|   307k|      }
   96|   191k|      wt.uw(u32(sumL));
   97|   191k|    }
   98|   382k|  }
   99|      1|  return 0;
  100|      1|}