KMP算法是一种改进的字符串匹配算法,由D.E.Knuth与V.R.Pratt和J.H.Morris同时发现,因此人们称它为克努特——莫里斯——普拉特操作(简称KMP算法).
KMP算法的关键是根据给定的模式串W1,m,定义一个next函数,next函数包含了模式串本身局部匹配的信息.
#include <iostream> #include <cstring> #include <string> #include <set> #include <map> using namespace std; void BuildPatchMatchTable(int *partMatchTable, char *findstr) { if(findstr == NULL) return; partMatchTable[0] = 0; int sizefind = strlen(findstr); for(int i = 1; i < sizefind; ++i) { set<string> preset; string tmppre = ""; tmppre = findstr[0]; preset.insert(tmppre); for(int j = 1; j < i; ++j) { tmppre = tmppre + findstr[j]; preset.insert(tmppre); } set<string> postset; string tmppost = ""; tmppost = findstr[i]; postset.insert(tmppost); for(int j = i - 1; j > 0; --j) { tmppost = findstr[j] + tmppost; postset.insert(tmppost); } set<string> comset; for(set<string>::iterator beg = preset.begin(); beg != preset.end(); ++beg) { if(postset.count(*beg) > 0) comset.insert(*beg); } int maxlen = 0; for(set<string>::iterator beg = comset.begin(); beg != comset.end(); ++beg) { if((*beg).size() > maxlen) maxlen = (*beg).size(); } partMatchTable[i] = maxlen; } } int kmp(char *srcstr, char *findstr) { if(srcstr == NULL || findstr == NULL) return -1; int lensrc = strlen(srcstr); int lenfind = strlen(findstr); int *partMatchTable = new int[lenfind]; BuildPatchMatchTable(partMatchTable, findstr); for(int i = 0; i < lenfind; ++i) cout << findstr[i] << "\t" << partMatchTable[i] << endl; int curFind = 0; for(int i = 0; i < lensrc; ) { if(findstr[curFind] == srcstr[i]) { ++i; ++curFind; } else { if(curFind == 0) ++i; else { int movestep = curFind - partMatchTable[curFind-1]; i += movestep; curFind = 0; } } if(curFind == lenfind) { delete []partMatchTable; return i - lenfind; } } return -1; delete []partMatchTable; } int main() { char srcStr[] = "bbcabcdababcdabcdabde"; char findStr[] = "abcdabd"; cout << "pos:" << kmp(srcStr, findStr) << endl; char srcStr2[] = "substring searching algorithm search"; char findStr2[] = "search"; cout << "pos:" << kmp(srcStr2, findStr2) << endl; }
原文地址:http://blog.csdn.net/sunmc1204953974/article/details/43371541