Wildcard Matching
2015-08-22 00:00
211 查看
摘要: Implement wildcard pattern matching with support for '?' and '*'.
The reason that the iterative solution is much faster for this case is we only need to save (and deal with) the positions (iStar for s, jStar for p) of the last "" we met. We only need to do traceback using iStar and jStar and all the previous "" can be ignored since the last "" will cover all the traceback cases for the previous "". What we need to do are
if the current p character is '' (i.e. p[j]==''), then we update iStar and jStar with the cureent i and j values. iStar/jStar will be used for traceback. Also we do --i to start the depth first search with the case that '*' represents a null string.
if p[j]!='', then we check if mismatch occurs (i.e. p[j]!=s[i] and p[j]!='?'), if so we check if we met a '' before (iStar>=0), if not, then we return false since no match can achieve. Otherwise, we traceback to the positions at which the last '*' happens and do the next possible dfs search (i.e. i = iStar++; j = jStar; remember to update iStar too to save the i position to try in the next traceback).
The loop will quit when we reach the end of s. At last, we need to skip all the '*' in p to see if we can reach the end of p. if so, match, otherwise mismatch
A DP solution is also given here. It has O(N^2) time complexity and O(N) space
A recursion version. A typical recursion version will give us TLE due to too many unnecessary recursive calls. As we explained, all the traceback recursive calls at the '' we met (except the last '') are unneccessary and should be avoided. In the below version, we use recLevel to track the recursion level (i.e the total '' we met) and we also use curLevel to save the order of '' we currently process. If it is not the last '' we met (i.e if(recLevel>curLevel+1) ), then we will return false directly ( if(recLevel>curLevel+1) return false;) to skip all unneccessary recursion call at the '' before the last '*'.
https://leetcode.com/discuss/52274/three-solutions-iterative-16ms-180ms-modified-recursion-88ms
'?' Matches any single character. '*' Matches any sequence of characters (including the empty sequence). The matching should cover the entire input string (not partial). The function prototype should be: bool isMatch(const char *s, const char *p) Some examples: isMatch("aa","a") → false isMatch("aa","aa") → true isMatch("aaa","aa") → false isMatch("aa", "*") → true isMatch("aa", "a*") → true isMatch("ab", "?*") → true isMatch("aab", "c*a*b") → false
The reason that the iterative solution is much faster for this case is we only need to save (and deal with) the positions (iStar for s, jStar for p) of the last "" we met. We only need to do traceback using iStar and jStar and all the previous "" can be ignored since the last "" will cover all the traceback cases for the previous "". What we need to do are
if the current p character is '' (i.e. p[j]==''), then we update iStar and jStar with the cureent i and j values. iStar/jStar will be used for traceback. Also we do --i to start the depth first search with the case that '*' represents a null string.
if p[j]!='', then we check if mismatch occurs (i.e. p[j]!=s[i] and p[j]!='?'), if so we check if we met a '' before (iStar>=0), if not, then we return false since no match can achieve. Otherwise, we traceback to the positions at which the last '*' happens and do the next possible dfs search (i.e. i = iStar++; j = jStar; remember to update iStar too to save the i position to try in the next traceback).
The loop will quit when we reach the end of s. At last, we need to skip all the '*' in p to see if we can reach the end of p. if so, match, otherwise mismatch
class Solution { public: bool isMatch(string s, string p) { int slen = s.size(), plen = p.size(), i, j, iStar=-1, jStar=-1; for(i=0,j=0 ; i<slen; ++i, ++j) { if(p[j]=='*') { //meet a new '*', update traceback i/j info iStar = i; jStar = j; --i; } else { if(p[j]!=s[i] && p[j]!='?') { // mismatch happens if(iStar >=0) { // met a '*' before, then do traceback i = iStar++; j = jStar; } else return false; // otherwise fail } } } while(p[j]=='*') ++j; return j==plen; } };
A DP solution is also given here. It has O(N^2) time complexity and O(N) space
class Solution { public: bool isMatch(string s, string p) { int pLen = p.size(), sLen = s.size(), i, j, k, cur, prev; if(!pLen) return sLen == 0; bool matched[2][sLen+1]; fill_n(&matched[0][0], 2*(sLen+1), false); matched[0][0] = true; for(i=1; i<=pLen; ++i) { cur = i%2, prev= 1-cur; matched[cur][0]= matched[prev][0] && p[i-1]=='*'; if(p[i-1]=='*') for(j=1; j<=sLen; ++j) matched[cur][j] = matched[cur][j-1] || matched[prev][j]; else for(j=1; j<=sLen; ++j) matched[cur][j] = matched[prev][j-1] && (p[i-1]=='?' || p[i-1]==s[j-1]) ; } return matched[cur][sLen]; } };
A recursion version. A typical recursion version will give us TLE due to too many unnecessary recursive calls. As we explained, all the traceback recursive calls at the '' we met (except the last '') are unneccessary and should be avoided. In the below version, we use recLevel to track the recursion level (i.e the total '' we met) and we also use curLevel to save the order of '' we currently process. If it is not the last '' we met (i.e if(recLevel>curLevel+1) ), then we will return false directly ( if(recLevel>curLevel+1) return false;) to skip all unneccessary recursion call at the '' before the last '*'.
class Solution { private: bool helper(const string &s, const string &p, int si, int pi, int &recLevel) { int sSize = s.size(), pSize = p.size(), i, curLevel = recLevel; bool first=true; while(si<sSize && (p[pi]==s[si] || p[pi]=='?')) {++pi; ++si;} //match as many as possible if(pi == pSize) return si == sSize; // if p reaches the end, return if(p[pi]=='*') { // if a star is met while(p[++pi]=='*'); //skip all the following stars if(pi>=pSize) return true; // if the rest of p are all star, return true for(i=si; i<sSize; ++i) { // then do recursion if(p[pi]!= '?' && p[pi]!=s[i]) continue; if(first) {++recLevel; first = false;} if(helper(s, p, i, pi, recLevel)) return true; if(recLevel>curLevel+1) return false; // if the currently processed star is not the last one, return } } return false; } public: bool isMatch(string s, string p) { int recLevel = 0; return helper(s, p, 0, 0, recLevel); } };
https://leetcode.com/discuss/52274/three-solutions-iterative-16ms-180ms-modified-recursion-88ms
相关文章推荐
- 装 VisualStudio 报“安装时发生严重错误”,查看日志有 Error 1402 字样
- 我的第一篇博客文章
- git
- Java并发学习笔记
- 单链表环问题
- 使用oh-my-zsh时报错compdef
- Android事件分发处理
- Android中使用内置的轻量级数据库SQLite存储信息
- window7安装django
- 1045 access denied…… 错误的解决方法
- MyBatis集合查询示例
- 2015年史玉柱演讲实录
- 为什么你应该写博客
- 为什么你应该写博客
- 吃一堑长一智,这次亏损让我的头脑开始清醒,理财没有那么容易,不是瞎买买股票、基金啥的就叫会理财了,我要学习的还有很多
- 【完全开源】百度地图Web service API C#.NET版,带地图显示控件、导航控件、POI查找控件
- Java common-fileupload 文件上传
- 显式intent和隐式intent
- Linux下部署多个Tomcat
- 时间time_t和string(char*)格式互转