KMP算法C++代码

2. //by baihacker 
4. #include <iostream>  
5. #include <vector>  
6. #include <functional>  
7. #include <algorithm>  
9. using namespace std;
10. template<typename FType, typename SType>
11. struct equal_cmp
12. {
13.   bool operator ()(const FType& _F, const SType& _S)const
14.   {return _F == _S;}  
15. };  
18. template<typename RandAccessConcept,
19.          typename EqualCmp>
20. inline void     
21. CalFailFunction(
22.                 const RandAccessConcept& in,
23.                 int len,
24.                 vector<int>& ff,
25.                 EqualCmp cmp
26.                 )
27. { 
28.     ff.clear();
29.     ff.reserve(len);
30.     ff.push_back(-1);
31.     for (int curr = 1, k = -1; curr < len; ++curr)
32.     {
33.         for (;k >= 0 && !cmp(in[k+1], in[curr]);
34.                 k = ff[k]);
35.         if (cmp(in[k+1], in[curr])) ++k;
36.         ff.push_back(k);
37.     }   
38. }  
40. template<typename FRandAccessConcept,
41.          typename SRandAccessConcept,
42.          typename  EqualCmp>
43. inline int      
44. KMPSearchFirst(
45.                 const FRandAccessConcept& d,
46.                 int   dlength,
47.                 const SRandAccessConcept& p,
48.                 int   plength,
49.                 EqualCmp cmp,
50.                 int start = 0
51.                 )
52. {
53.     vector<int> ff;
54.     CalFailFunction(p, plength, ff, cmp);
55.     for (int i = start, j = -1; i < dlength; ++i)
56.     {
57.         while (j >= 0 && !cmp(d[i], p[j+1])) j = ff[j];
58.         if (cmp(d[i], p[j+1])) ++j;
59.         if (j == plength - 1) return i - plength + 1;
60.     }
61.     return -1; 
62. }    
64. template<typename FRandAccessConcept,
65.          typename SRandAccessConcept,
66.          typename EqualCmp>
67. inline int      
68. KMPSearchALL(
69.                 const FRandAccessConcept& d,
70.                 int   dlength,
71.                 const SRandAccessConcept& p,
72.                 int   plength,
73.                 vector<int>& result,
74.                 EqualCmp cmp
75.                 )
76. {
77.     vector<int> ff;
78.     CalFailFunction(p, plength, ff, cmp);
79.     result.clear();
80.     for (int i = 0, j = -1; i < dlength; ++i)
81.     {
82.         while (j >= 0 && !cmp(d[i], p[j+1])) j = ff[j];
83.         if (cmp(d[i], p[j+1])) ++j;
84.         if (j == plength - 1)
85.         {
86.              result.push_back(i - plength + 1);
87.              j = ff[j];
88.         }    
89.     }
90.     return result.size();
91. }      
93. int main()
94. {
95.     int d[] = {1,2,3,1,2,3,1,2};
96.     char p[] = "/1/2/3/1/2";
97.     vector<int> result;
98.     int found = KMPSearchALL(d, 8, p, 5, result, equal_cmp<int, char>());
99.     
100.     for (int i = 0; i < found; ++i)
101.         cout << result[i] << " ";
102.     cout << endl;
103.     
104.     return 0;
105. }
        

 2013/06/29 : 更新一些能直接使用并用于较短模式串的代码。推荐一本叫《柔性字符串匹配》的书

template<typename ElemType>
int MatchSO(ElemType* t, int n, ElemType* p, int m, vector<int>& result)
{
	typedef unsigned long long NumberType;
	const int CharSet = 128;
	NumberType	B[CharSet];
	NumberType	D = ~0;
	NumberType	Check = 1 << (m-1);
	NumberType	BIT = 1;
	for (int i = 0; i < CharSet; ++i) B[i] = ~0;
	for (int i = 0; i < m; ++i, BIT <<= 1) B[p[i]] &= ~BIT;
	result.clear();
	for (int i = 0; i < n; ++i)
	{
		D = (D << 1) | B[t[i]];
		if ((D & Check) == 0) result.push_back(i-m+1);
	}
	return result.size();
}

template<typename ElemType>
int MatchSA(ElemType* t, int n, ElemType* p, int m, vector<int>& result)
{
	typedef unsigned long long NumberType;
	const int CharSet = 128;
	NumberType	B[CharSet];
	NumberType	D = 0;
	NumberType	Check = 1 << (m-1);
	NumberType	BIT = 1;
	for (int i = 0; i < CharSet; ++i) B[i] = 0;
	for (int i = 0; i < m; ++i, BIT <<= 1) B[p[i]] |= BIT;
	result.clear();
	for (int i = 0; i < n; ++i)
	{
		D = ((D << 1) | 1) & B[t[i]];
		if ((D & Check) != 0) result.push_back(i-m+1);
	}
	return result.size();
}

template<typename ElemType>
int MatchKMP(ElemType* t, int n, ElemType* p, int m, vector<int>& result)
{
	int	* Prefix = new int[m];
	{
		Prefix[0] = -1;
		for (int i = 1, F = -1; i < m; ++i)
		{
			while (F >= 0 && p[F+1] != p[i]) F = Prefix[F];
			if (p[F+1] == p[i]) ++F;
			Prefix[i] = F;
		}
	}
	{
		result.clear();
		for (int i = 0, F = -1; i < n; ++i)
		{
			while (F >= 0 && p[F+1] != t[i]) F = Prefix[F];
			if (p[F+1] == t[i]) ++F;
			if (F+1 == m)
			{
				result.push_back(i-m+1);
				F = Prefix[F];
			}
		}
	}
	delete[] Prefix;
	return result.size();
}