LeetCode 930. Binary Subarrays With Sum LeetCode 前缀 双指针 坑

In an array A of 0s and 1s, how many non-empty subarrays have sum S?

 

Example 1:

Input: A = [1,0,1,0,1], S = 2
Output: 4
Explanation: 
The 4 subarrays are bolded below:
[1,0,1,0,1]
[1,0,1,0,1]
[1,0,1,0,1]
[1,0,1,0,1]

 

Note:

1. A.length <= 30000
2. 0 <= S <= A.length
3. A[i] is either 0 or 1.

----------------------------------------------------------------------------------

题目非常简单，前缀法很容易想到，但是前缀是否为空是前缀法的常见bug，第一次提交不幸入坑。正确代码是：

class Solution:
    def numSubarraysWithSum(self, A, S: int) -> int:
        dic, l, cursum, res = {0: 1}, len(A), 0, 0  # bug2:dic={}
        for i in range(l):
            cursum += A[i]
            # prefix + subarray(S) = cursum, so the number of prefixes dominate # of S
            res += dic.get(cursum - S, 0)
            dic[cursum] = dic.get(cursum, 0) + 1
        return res   # bug1: add dic.get(cursum-S,0)

但是前缀法并没有利用0，1的条件，有了0，1的限制，刚好搞一个非递减序列，因此可以用尺取法，但是由于是非递减，尺子前后都可能是0，堪称大坑，很容易写bug，好不容易写对了，复杂度还可能退化到O(n^2)。

一个比较好的思路是转换成两个小等于的差，以下是codes：

class Solution:
    def at_most(self, A, S):
        if (S < 0):
            return 0
        la,i,cursum,res = len(A),0,0,0
        for j in range(la):
            cursum += A[j]
            while (cursum > S):
                cursum -= A[i]
                i += 1
            res += (j-i+1)
        return res
    def numSubarraysWithSum(self, A, S: int) -> int:
        return self.at_most(A,S)-self.at_most(A,S-1)