class Solution {
public:
vector<int> findSubstring(string S, vector<string> &L) {
vector<int> res;
unordered_map<string, int> stat;
unordered_map<string, int> run;
int len = S.size();
int num = L.size();
int per = 0;
if (num == 0 || !(per = L[0].size())) return res;
int part= num * per;
if (part > len) return res;
int end = len - part;
unordered_map<string, int>::iterator iter;
pair<unordered_map<string, int>::iterator, bool> ir;
for (int i=0; i<num; i++) {
ir = stat.insert(pair<string, int>(L[i], 1));
if (ir.second == false){
ir.first->second++;
}
}
int i, j, pos, wc;
string pre;
for (i=0; i<=end; i++) {
pos = i;
for (j=0; j<num; j++, pos += per) {
string seg = S.substr(pos, per);
if (j == 0 || seg != pre) {
iter = stat.find(seg);
if (iter == stat.end()) break;
wc = iter->second;
ir = run.insert(pair<string, int>(seg, 1));
iter = ir.first;
if (ir.second) {
pre = seg;
continue;
}
}
iter->second++;
if (wc < iter->second) break;
}
if (j == num) res.push_back(i);
run.clear();
}
return res;
}
};
暴力解,跑了500+ms,如果每次都这样把题目应付过去,显然做题没有意义了。于是看Leetcode上关于那题的discuss,有人说可以用解决Minimum Window Substring的方法来解这一题,的确是可以。又搜了份Java实现的代码,一跑竟然也可以达到500ms,看来两种方法效率的确差很多。下面给出自己实现的代码
1 class Solution3 { 2 public: 3 vector<int> findSubstring(string S, vector<string> &L) { 4 vector<int> res; 5 unordered_map<string, int> stat; 6 unordered_map<string, int> run; 7 int len = S.size(); 8 int num = L.size(); 9 int per = 0; 10 11 if (num == 0 || !(per = L[0].size())) return res; 12 13 int part = num * per; 14 if (part > len) return res; 15 16 unordered_map<string, int>::iterator iter; 17 pair<unordered_map<string, int>::iterator, bool> ir; 18 19 for (int i = 0; i < num; i++) { 20 ir = stat.insert(pair<string, int>(L[i], 0)); 21 ir.first->second++; 22 } 23 int wc; 24 for (int i = 0; i < per; i++) { 25 int step = 0; 26 run.clear(); 27 // scan like a worm, string[spos, epos] is the candidate 28 int spos=i, epos = i + per - 1; 29 for (; epos < len; epos += per) { 30 string seg = S.substr(epos - per + 1, per); 31 iter = stat.find(seg); 32 // encounter some word not in L 33 if (iter == stat.end()) { 34 spos = epos + 1; 35 step = 0; 36 run.clear(); 37 continue; 38 } 39 40 wc = iter->second; 41 step++; 42 43 ir = run.insert(pair<string, int>(seg, 0)); 44 iter = ir.first; 45 iter->second++; 46 47 // string[spos, epos] is matched 48 if (iter->second == wc && step == num) { 49 res.push_back(spos); 50 run.find(S.substr(spos, per))->second--; 51 step--; 52 spos += per; 53 continue; 54 } 55 56 // number of duplicated word exceeds needed 57 if (iter->second > wc) { 58 string tmp = S.substr(spos, per); 59 // find the first duplicated one 60 while(seg != tmp) { 61 run.find(tmp)->second--; 62 step--; 63 spos += per; 64 tmp = S.substr(spos, per); 65 } 66 // then skip it 67 iter->second--; 68 spos += per; 69 step--; 70 } 71 } 72 } 73 return res; 74 } 75 };
跑了一下在100+ms左右,又习得一策略技能!
参考:
zhuli题解 http://www.cnblogs.com/zhuli19901106/p/3570539.html
java实现 https://github.com/shengmin/coding-problem/blob/master/leetcode/substring-with-concatenation-of-all-words/Solution.java
leetcode资料 http://leetcode.com/2010/11/finding-minimum-window-in-s-which.html