字符串全排列

问题:给定字符串S,生成该字符串的全排列。
比如输入为abc,那么输出有以下几种:

1
abc acb bac bca cab cba

即如果输入字符串的长度为N的话,会输出N!个结果。
本文整理了求字符串全排列的若干方法。

方法1 使用STL库的标准函数

C++ STL中提供了std::next_permutation与std::prev_permutation可以获取数字或者是字符的全排列,其中std::next_permutation提供升序、std::prev_permutation提供降序。
1.std::next_permutation函数原型

1
2
3
4
  template <class BidirectionalIterator>
  bool next_permutation (BidirectionalIterator first, BidirectionalIterator last );
  template <class BidirectionalIterator, class Compare>
  bool next_permutation (BidirectionalIterator first,BidirectionalIterator last, Compare comp);

说明:next_permutation,重新排列范围内的元素[第一,最后一个)返回按照字典序排列的下一个值较大的组合。
返回值:如果有一个更高的排列,它重新排列元素,并返回true;如果这是不可能的(因为它已经在最大可能的排列),它按升序排列重新元素,并返回false。

1
2
3
4
5
6
7
8
9
10
11
12
#include <algorithm>
#include <string>
#include <iostream>
int main()
{
std::string s = "aba";
std::sort(s.begin(), s.end());
do {
std::cout << s << '\n';
} while(std::next_permutation(s.begin(), s.end()));
}

方法2 仿STL库的标准函数实现

1
2
 [例]字符集{1,2,3},较小的数字较先,这样按字典序生成的全排列是:
    123,132,213,231,312,321

※ 一个全排列可看做一个字符串,字符串可有前缀、后缀。
生成给定全排列的下一个排列.所谓一个的下一个就是这一个与下一个之间没有其他的。这就要求这一个与下一个有尽可能长的共同前缀,也即变化限制在尽可能短的后缀上。

1
[例]8396475211--9的排列。19的排列最前面的是123456789,最后面的987654321,从右向左扫描若都是增的,就到了987654321,也就没有下一个了。否则找出第一次出现下降的位置。

1
2
3
4
5
6
7
【例】 一般而言,设P是[1,n]的一个全排列。
      P=P1P2…Pn=P1P2…Pj-1PjPj+1…Pk-1PkPk+1…Pn
    find:  j=max{i|Pi<Pi+1}
         k=max{i|Pi>Pj}
      1, 对换Pj,Pk,
      2, 将Pj+1…Pk-1PjPk+1…Pn翻转
P’= P1P2…Pj-1PkPn…Pk+1PjPk-1…Pj+1即P的下一个

【例】 如何得到346987521的下一个

  1. 从尾部往前找第一个P(i-1) < P(i)的位置
    3 4 6 <- 9 <- 8 <- 7 <- 5 <- 2 <- 1
    最终找到6是第一个变小的数字,记录下6的位置i-1
  2. 从i位置往后找到最后一个大于6的数
    3 4 6 -> 9 -> 8 -> 7 5 2 1
    最终找到7的位置,记录位置为m
  3. 交换位置i-1和m的值
    3 4 7 9 8 6 5 2 1
  4. 倒序i位置后的所有数据
    3 4 7 1 2 5 6 8 9
    则347125689为346987521的下一个排列

依照上面的讲述不难将代码写出来,如下:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
private static void PermutationList()
{
int fromIndex, endIndex, changeIndex;
Sort(0, length - 1);
do
{
// 输出一种全排列
Output();
fromIndex = endIndex = length - 1;
// 向前查找第一个变小的元素
while (fromIndex > 0 && words[fromIndex] < words[fromIndex - 1]) --fromIndex;
changeIndex = fromIndex;
if (fromIndex == 0) break;
// 向后查找最后一个大于words[fromIndex-1]的元素
while (changeIndex + 1 < length && words[changeIndex + 1] > words[fromIndex - 1]) ++changeIndex;
Swap(fromIndex - 1, changeIndex); // 交换两个值
InvertArray(fromIndex, endIndex); // 对后面的所有值进行反向处理
} while (true);
}

方法3 递归方法求全排列

  递归方法很容易理解:分别将每个位置交换到最前面位,之后全排列剩下的位。

1
2
3
4
【例】递归全排列 1 2 3 4 5
1. for循环将每个位置的数据交换到第一位
swap(1,1~5)
2. 按相同的方式全排列剩余的位

递归全排列1

由于递归方法很容易理解,而且网上也有很多的资料,所以不过多讲述,代码如下:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
/// <summary>
/// 递归方式生成全排列的方法 3-1
/// </summary>
/// <param name="fromIndex">全排列的起始位置</param>
/// <param name="endIndex">全排列的终止位置</param>
private static void PermutationList(int fromIndex, int endIndex)
{
if (fromIndex == endIndex)
Output();
else
{
for (int index = fromIndex; index <= endIndex; ++index)
{
//此处排序主要是为了生成字典序全排列,否则递归会打乱字典序
sort(fromIndex, endIndex);
Swap(fromIndex, index);
PermutationList(fromIndex + 1, endIndex);
Swap(fromIndex, index);
}
}
}

不过需要注意,如果不加入第15行的sort(fromIndex, endIndex);那么生成的全排列就是非完全升序的.


outorder-permutation.jpg

同时可以对进行修改,使其生成的全排列是完全升序的

递归全排列2

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
#include <iostream>
using namespace std;
void p(char* str, int i, int size)
{
//到达字符串末尾则停止并输出
if(i == size)
{
cout << str << endl;
}
for(int j = i; j < size; j++)
{
//置换;
for(int k = j; k > i; k--)
{
swap(str[k], str[k - 1]);
}
p(str, i + 1, size);
//置换;
for(int k = i; k < j; k++)
{
swap(str[k], str[k + 1]);
}
}
return;
}
int main()
{
char str[7];
cin >> str;
int size = 0;
//判断字符长度
for(int i = 0; i < 7; i++)
{
if(str[i] == '\0')
{
size = i;
break;
}
}
p(str, 0, size);
}

方法4 递归方法求全排列2

思路是这样的:我们维护两个序列,一个序列是要进行全排列的序列,我们暂称之为源序列,另一个序列是全排列之后的结果序列,我们称其为结果序列。过程如下:

  1. 初始时源序列为输入的字符串序列,结果序列为空
  2. 如果源序列中的元素个数大于1,则对源序列中的每一个元素,进行如下操作:

    1. 以结果序列+该元素生成新的结果序列
    2. 将该元素从源序列中剔除并保持其他元素顺序不变生成新的源序列

    然后以I产生的结果序列和II产生的源序列为基础递归2)过程

  3. 如果源序列中元素个数不大于1,则打印结果序列+源序列

下面给出了该思路的C++实现,参考[1][2]

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
void permutation(string begin, string end)
{
int endLength = end.size(), beginLength = begin.size();
if(endLength == 1)
{
cout << begin << end << endl;
return ;
}
string newEnd;
for(int i = 0; i < endLength; i++)
{
newEnd = end.substr(0, i) + end.substr(i+1);
permutation(begin + end[i], newEnd);
}
}
int main()
{
string str;
string n = "";
cin >> str;
permutation(n, str);
}


inorder-permutation.png

综上,文章介绍了四种字符串全排列的方法:

  • 去重: 方法1,方法2
  • 不去重按序:方法3-2,方法4
  • 不去重不排序:方法3-1

参考:

  1. Recursive method to find all permutations of a String
  2. 关于全排列算法的思考
  3. 字典序全排列算法研究