Implementation of string::find_first_of

问题

My confusion comes from a solution for problem 792. Number of Matching Subsequences in leetcode, the naive solution is to check all characters in S for each searching word. The time complexity is too high to pass all test cases as outlined in the official answer.

The first code piece listed below could beats 100% merely because it calls string::find_first_of function, while my hand written codes uses exactly the same routine(I believe) get TLE as expected.

So, why the first one is so fast?

// comes from the second solution post in leetcode-cn
class Solution {
public:
    int numMatchingSubseq(string S, vector<string>& words) {
        int res = 0, j;
        for (int i = 0; i < words.size(); i ++) {
            int position = -1;
            for (j = 0; j < words[i].size(); j ++) {
                position = S.find_first_of(words[i][j], position + 1); 
                if (position == -1) break; 
            }
            if (j == words[i].length()) res ++; 
        }
        return res;
    }
};

version 2 hand written Time limit exceed

static inline int find_first_of(const string & s, char & c, int st) {
    int  n = s.size() - st;
    const char * data = s.data(), * cur = data + st;
    for (std::size_t i = 0; i < n; ++i) {
        if (cur[i] == c)
            return cur + i - data;
    }
    return -1;
}

class Solution {
public:
    int numMatchingSubseq(string S, vector<string>& words) {
        int res = 0, j;
        for (int i = 0; i < words.size(); i ++) {
            int position = -1;
            for (j = 0; j < words[i].size(); j ++) {
                position = find_first_of(S, words[i][j], position + 1);
                if (position == -1) break; 
            }
            if (j == words[i].length()) res ++; 
        }
        return res;
    }
};

来源：https://stackoverflow.com/questions/63049548/implementation-of-stringfind-first-of