class FullyIndexableDictionary(): def __init__(self, size): self.size = size self.block = (size + 31) >> 5 self.bit = [0] * self.block self.sum = [0] * self.block def popcount(self, x): x = x - ((x >> 1) & 0x55555555) x = (x & 0x33333333) + ((x >> 2) & 0x33333333) x = x + (x >> 4) & 0x0f0f0f0f x = x + (x >> 8) x = x + (x >> 16) return x & 0x0000007f def set(self, k): self.bit[k >> 5] |= 1 << (k & 31) def build(self): for i in range(1, self.block): self.sum[i] = self.sum[i - 1] + self.popcount(self.bit[i - 1]) def access(self, k): return (self.bit[k >> 5] >> (k & 31)) & 1 def rank(self, k, v): r = self.sum[k >> 5] + self.popcount(self.bit[k >> 5] & ((1 << (k & 31)) - 1)) return r if v else k - r def select(self, k, v): if k < 0 or self.rank(self.size, v) <= k: return -1 l, r = 0, self.size while r - l > 1: m = (l + r) // 2 if self.rank(m, v) >= k + 1: r = m else: l = m return l class WaveletMatrix(): def __init__(self, log=32): self.size = 0 self.log = log self.mat = [None] * log self.mid = [None] * log def build(self, arr): self.size = len(arr) for lv in range(self.log)[::-1]: self.mat[lv] = FullyIndexableDictionary(self.size + 1) lt = [] rt = [] for i, a in enumerate(arr): if (a >> lv) & 1: self.mat[lv].set(i) rt.append(a) else: lt.append(a) self.mid[lv] = len(lt) self.mat[lv].build() arr = lt + rt def access(self, k): res = 0 for lv in range(self.log)[::-1]: if self.mat[lv].access(k): res |= 1 << lv k = self.mat[lv].rank(k, 1) + self.mid[lv] else: k = self.mat[lv].rank(k, 0) return res def rank(self, x, r): l = 0 for lv in range(self.log)[::-1]: if (x >> lv) & 1: l = self.mat[lv].rank(l, 1) + self.mid[lv] r = self.mat[lv].rank(r, 1) + self.mid[lv] else: l = self.mat[lv].rank(l, 0) r = self.mat[lv].rank(r, 0) return r - l def select(self, x, k): idx = 0 for lv in range(self.log)[::-1]: if (x >> lv) & 1: idx = self.mat[lv].rank(self.size, 0) + self.mat[lv].rank(idx, 1) else: idx = self.mat[lv].rank(idx, 0) idx += k for lv in range(self.log): if (x >> lv) & 1: idx = self.mat[lv].select(idx - self.mat[lv].rank(self.size, 0), 1) else: idx = self.mat[lv].select(idx, 0) return idx def freq(self, l, r, x): res = 0 for lv in range(self.log)[::-1]: if (x >> lv) & 1: res += self.mat[lv].rank(r, 0) - self.mat[lv].rank(l, 0) l = self.mat[lv].rank(l, 1) + self.mid[lv] r = self.mat[lv].rank(r, 1) + self.mid[lv] else: l = self.mat[lv].rank(l, 0) r = self.mat[lv].rank(r, 0) return res def quantile(self, l, r, k): res = 0 for lv in range(self.log)[::-1]: cnt = self.mat[lv].rank(r, 0) - self.mat[lv].rank(l, 0) if cnt <= k: res |= 1 << lv k -= cnt l = self.mat[lv].rank(l, 1) + self.mid[lv] r = self.mat[lv].rank(r, 1) + self.mid[lv] else: l = self.mat[lv].rank(l, 0) r = self.mat[lv].rank(r, 0) return res N, Q = map(int, input().split()) S = input() arr = [ord(s) for s in S] wm = WaveletMatrix() wm.build(arr) res = [] for _ in range(Q): l, r, x = map(int, input().split()) l -= 1; r -= 1; x -= 1 x = wm.quantile(l, r + 1, x) res.append(chr(x)) print('\n'.join(res))