/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|  23.1M|  auto operator+(affine t) -> affine {
   13|  23.1M|    return {u32(u64(t.a) * a % P), u32((u64(t.a) * b + t.b) % P)};
   14|  23.1M|  }
   15|  23.1M|  auto operator+=(affine t) -> void { *this = *this + t; }
   16|       |};
   17|       |
   18|       |struct node {
   19|       |  u32 siz;
   20|       |  u32 sum;
   21|       |  affine aff;
   22|  23.1M|  auto operator+=(affine t) -> void {
   23|  23.1M|    aff += t;
   24|  23.1M|    sum = (u64(t.a) * sum + u64(t.b) * siz) % P;
   25|  23.1M|  }
   26|       |} a[N];
   27|       |
   28|   501k|void pushdown(int k) {
   29|  10.0M|  for (int i = log(k) - 1; i >= 0; --i) {
                                                 ^9.50M
  ------------------
  |  Branch (29:28): [True: 94.99%, False: 5.01%]
  ------------------
   30|  9.50M|    affine t = a[k >> i >> 1].aff;
   31|  9.50M|    a[k >> i >> 1].aff = affine{1, 0};
   32|  9.50M|    a[k >> i ^ 0] += t;
   33|  9.50M|    a[k >> i ^ 1] += t;
   34|  9.50M|  }
   35|   501k|}
   36|       |
   37|  9.75M|u32 mod(u32 x) { return x < P ? x : x - P; }
                                              ^4.87M^4.87M
  ------------------
  |  Branch (37:25): [True: 50.01%, False: 49.99%]
  ------------------
   38|       |
   39|   501k|void pushup(int k) {
   40|  9.50M|  for (k /= 2; k > 0; k /= 2) {
                                    ^9.00M
  ------------------
  |  Branch (40:16): [True: 94.72%, False: 5.28%]
  ------------------
   41|  9.00M|    a[k].sum = mod(a[k * 2].sum + a[k * 2 + 1].sum);
   42|  9.00M|  }
   43|   501k|}
   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|   500k|  for (int i = 0; i < n; ++i) a[n + i].siz = 1, a[n + i].sum = rd.uw();
                                       ^500k^500k
  ------------------
  |  Branch (52:19): [True: 100.00%, False: 0.00%]
  ------------------
   53|   500k|  for (int i = n - 1; i >= 0; --i) {
                                            ^500k
  ------------------
  |  Branch (53:23): [True: 100.00%, False: 0.00%]
  ------------------
   54|   500k|    a[i].aff = {1, 0};
   55|   500k|    a[i].siz = a[2 * i].siz + a[2 * i + 1].siz;
   56|   500k|    a[i].sum = mod(a[2 * i].sum + a[2 * i + 1].sum);
   57|   500k|  }
   58|   500k|  while (q--) {
  ------------------
  |  Branch (58:10): [True: 100.00%, False: 0.00%]
  ------------------
   59|   500k|    let t = rd.u1();
   60|   500k|    if (t == 0) {
  ------------------
  |  Branch (60:9): [True: 50.15%, False: 49.85%]
  ------------------
   61|   250k|      int l = n + rd.uh();
   62|   250k|      int r = n + rd.uh() - 1;
   63|   250k|      pushdown(l--);
   64|   250k|      pushdown(r++);
   65|   250k|      int k = log(l ^ r);
   66|   250k|      affine x = {rd.uw(), rd.uw()};
   67|   250k|      int t, i;
   68|  2.34M|      for (t = ~l & ~(-1 << k), i = 31; t > 0; t -= 1 << i) {
                                                             ^2.09M
  ------------------
  |  Branch (68:41): [True: 89.31%, False: 10.69%]
  ------------------
   69|  2.09M|        i = log(t);
   70|  2.09M|        a[l >> i ^ 1] += x;
   71|  2.09M|      }
   72|   250k|      pushup(l >> i);
   73|  2.34M|      for (t = +r & ~(-1 << k), i = 31; t > 0; t -= 1 << i) {
                                                             ^2.09M
  ------------------
  |  Branch (73:41): [True: 89.31%, False: 10.69%]
  ------------------
   74|  2.09M|        i = log(t);
   75|  2.09M|        a[r >> i ^ 1] += x;
   76|  2.09M|      }
   77|   250k|      pushup(r >> i);
   78|   250k|    }
   79|   500k|    if (t == 1) {
  ------------------
  |  Branch (79:9): [True: 49.85%, False: 50.15%]
  ------------------
   80|   249k|      int l = n + rd.uh();
   81|   249k|      int r = n + rd.uh() - 1;
   82|   249k|      u64 sizL = 0, sumL = 0;
   83|   249k|      u64 sizR = 0, sumR = 0;
   84|  4.52M|      for (--l, ++r; l ^ r ^ 1;) {
  ------------------
  |  Branch (84:22): [True: 94.49%, False: 5.51%]
  ------------------
   85|  4.27M|        if (~l & 1) sizL += a[l ^ 1].siz, sumL += a[l ^ 1].sum;
                                  ^2.08M
  ------------------
  |  Branch (85:13): [True: 48.70%, False: 51.30%]
  ------------------
   86|  4.27M|        if (+r & 1) sizR += a[r ^ 1].siz, sumR += a[r ^ 1].sum;
                                  ^2.08M
  ------------------
  |  Branch (86:13): [True: 48.82%, False: 51.18%]
  ------------------
   87|  4.27M|        l /= 2, r /= 2;
   88|  4.27M|        sumL = (a[l].aff.a * sumL + a[l].aff.b * sizL) % P;
   89|  4.27M|        sumR = (a[r].aff.a * sumR + a[r].aff.b * sizR) % P;
   90|  4.27M|      }
   91|   249k|      sumL = mod(u32(sumL + sumR));
   92|   249k|      sizL += sizR;
   93|   711k|      for (l /= 2; l > 0; l /= 2) {
                                        ^462k
  ------------------
  |  Branch (93:20): [True: 64.98%, False: 35.02%]
  ------------------
   94|   462k|        sumL = (a[l].aff.a * sumL + a[l].aff.b * sizL) % P;
   95|   462k|      }
   96|   249k|      wt.uw(u32(sumL));
   97|   249k|    }
   98|   500k|  }
   99|      1|  return 0;
  100|      1|}