这是一个更快的解决方案，因为它不受字符串连接计算复杂度O(n^2)的影响。另一方面它是无循环的，完全递归的

public static void main(String[] args) {
    permutation("ABCDEFGHIJKLMNOPQRSTUVWXYZ");
}

private static void permutation(String str) {
    char[] stringArray = str.toCharArray();
    printPermutation(stringArray, 0, stringArray.length, 0, 1);
}

private static void printPermutation(char[] string, int loopCounter, int length, int indexFrom, int indexTo) {
    // Stop condition
    if (loopCounter == length)
        return;

    /* 
     When reaching the end of the array:
     1- Reset loop indices.
     2- Increase length counter. 
    */ 
    if (indexTo == length) {
        indexFrom = 0;
        indexTo = 1;
        ++loopCounter;
    }

    // Print.
    System.out.println(string);

    // Swap from / to indices.
    char temp = string[indexFrom];
    string[indexFrom] = string[indexTo];
    string[indexTo] = temp;

    // Go for next iteration.
    printPermutation(string, loopCounter, length, ++indexFrom, ++indexTo);
}

我们可以用阶乘来计算有多少字符串以某个字母开头。

示例:取输入abcd。(3!) == 6个字符串将以abcd中的每个字母开头。

static public int facts(int x){
    int sum = 1;
    for (int i = 1; i < x; i++) {
        sum *= (i+1);
    }
    return sum;
}

public static void permutation(String str) {
    char[] str2 = str.toCharArray();
    int n = str2.length;
    int permutation = 0;
    if (n == 1) {
        System.out.println(str2[0]);
    } else if (n == 2) {
        System.out.println(str2[0] + "" + str2[1]);
        System.out.println(str2[1] + "" + str2[0]);
    } else {
        for (int i = 0; i < n; i++) {
            if (true) {
                char[] str3 = str.toCharArray();
                char temp = str3[i];
                str3[i] = str3[0];
                str3[0] = temp;
                str2 = str3;
            }

            for (int j = 1, count = 0; count < facts(n-1); j++, count++) {
                if (j != n-1) {
                    char temp1 = str2[j+1];
                    str2[j+1] = str2[j];
                    str2[j] = temp1;
                } else {
                    char temp1 = str2[n-1];
                    str2[n-1] = str2[1];
                    str2[1] = temp1;
                    j = 1;
                } // end of else block
                permutation++;
                System.out.print("permutation " + permutation + " is   -> ");
                for (int k = 0; k < n; k++) {
                    System.out.print(str2[k]);
                } // end of loop k
                System.out.println();
            } // end of loop j
        } // end of loop i
    }
}

让我们以输入abc为例。

从集合(["c"])中的最后一个元素(c)开始，然后将最后第二个元素(b)添加到它的前面，末尾和中间的每个可能位置，使其["bc"， "cb"]，然后以同样的方式将后面的下一个元素(a)添加到集合中的每个字符串中，使其:

"a" + "bc" = ["abc", "bac", "bca"]  and  "a" + "cb" = ["acb" ,"cab", "cba"] 

因此整个排列:

["abc", "bac", "bca","acb" ,"cab", "cba"]

代码:

public class Test 
{
    static Set<String> permutations;
    static Set<String> result = new HashSet<String>();

    public static Set<String> permutation(String string) {
        permutations = new HashSet<String>();

        int n = string.length();
        for (int i = n - 1; i >= 0; i--) 
        {
            shuffle(string.charAt(i));
        }
        return permutations;
    }

    private static void shuffle(char c) {
        if (permutations.size() == 0) {
            permutations.add(String.valueOf(c));
        } else {
            Iterator<String> it = permutations.iterator();
            for (int i = 0; i < permutations.size(); i++) {

                String temp1;
                for (; it.hasNext();) {
                    temp1 = it.next();
                    for (int k = 0; k < temp1.length() + 1; k += 1) {
                        StringBuilder sb = new StringBuilder(temp1);

                        sb.insert(k, c);

                        result.add(sb.toString());
                    }
                }
            }
            permutations = result;
            //'result' has to be refreshed so that in next run it doesn't contain stale values.
            result = new HashSet<String>();
        }
    }

    public static void main(String[] args) {
        Set<String> result = permutation("abc");

        System.out.println("\nThere are total of " + result.size() + " permutations:");
        Iterator<String> it = result.iterator();
        while (it.hasNext()) {
            System.out.println(it.next());
        }
    }
}

下面是两个c#版本(仅供参考): 1. 打印所有排列 2. 返回所有排列

算法的基本要点是(可能下面的代码更直观-尽管如此，下面的代码是做什么的一些解释): -从当前索引到集合的其余部分，交换当前索引处的元素 -递归地获得下一个索引中剩余元素的排列 -恢复秩序，通过重新交换

注意:上述递归函数将从起始索引中调用。

private void PrintAllPermutations(int[] a, int index, ref int count)
        {
            if (index == (a.Length - 1))
            {
                count++;
                var s = string.Format("{0}: {1}", count, string.Join(",", a));
                Debug.WriteLine(s);
            }
            for (int i = index; i < a.Length; i++)
            {
                Utilities.swap(ref a[i], ref a[index]);
                this.PrintAllPermutations(a, index + 1, ref count);
                Utilities.swap(ref a[i], ref a[index]);
            }
        }
        private int PrintAllPermutations(int[] a)
        {
            a.ThrowIfNull("a");
            int count = 0;
            this.PrintAllPermutations(a, index:0, count: ref count);
            return count;
        }

版本2(与上面相同-但返回排列而不是打印)

private int[][] GetAllPermutations(int[] a, int index)
        {
            List<int[]> permutations = new List<int[]>();
            if (index == (a.Length - 1))
            {
                permutations.Add(a.ToArray());
            }

            for (int i = index; i < a.Length; i++)
            {
                Utilities.swap(ref a[i], ref a[index]);
                var r = this.GetAllPermutations(a, index + 1);
                permutations.AddRange(r);
                Utilities.swap(ref a[i], ref a[index]);
            }
            return permutations.ToArray();
        }
        private int[][] GetAllPermutations(int[] p)
        {
            p.ThrowIfNull("p");
            return this.GetAllPermutations(p, 0);
        }

单元测试

[TestMethod]
        public void PermutationsTests()
        {
            List<int> input = new List<int>();
            int[] output = { 0, 1, 2, 6, 24, 120 };
            for (int i = 0; i <= 5; i++)
            {
                if (i != 0)
                {
                    input.Add(i);
                }
                Debug.WriteLine("================PrintAllPermutations===================");
                int count = this.PrintAllPermutations(input.ToArray());
                Assert.IsTrue(count == output[i]);
                Debug.WriteLine("=====================GetAllPermutations=================");
                var r = this.GetAllPermutations(input.ToArray());
                Assert.IsTrue(count == r.Length);
                for (int j = 1; j <= r.Length;j++ )
                {
                    string s = string.Format("{0}: {1}", j,
                        string.Join(",", r[j - 1]));
                    Debug.WriteLine(s);
                }
                Debug.WriteLine("No.OfElements: {0}, TotalPerms: {1}", i, count);
            }
        }

另一种简单的方法是遍历字符串，选择尚未使用的字符并将其放入缓冲区，继续循环，直到缓冲区大小等于字符串长度。我更喜欢这个回溯跟踪解决方案，因为:

容易理解 容易避免重复 输出是排序的

下面是java代码:

List<String> permute(String str) {
  if (str == null) {
    return null;
  }

  char[] chars = str.toCharArray();
  boolean[] used = new boolean[chars.length];

  List<String> res = new ArrayList<String>();
  StringBuilder sb = new StringBuilder();

  Arrays.sort(chars);

  helper(chars, used, sb, res);

  return res;
}

void helper(char[] chars, boolean[] used, StringBuilder sb, List<String> res) {
  if (sb.length() == chars.length) {
    res.add(sb.toString());
    return;
  }

  for (int i = 0; i < chars.length; i++) {
    // avoid duplicates
    if (i > 0 && chars[i] == chars[i - 1] && !used[i - 1]) {
      continue;
    }

    // pick the character that has not used yet
    if (!used[i]) {
      used[i] = true;
      sb.append(chars[i]);

      helper(chars, used, sb, res);

      // back tracking
      sb.deleteCharAt(sb.length() - 1);
      used[i] = false;
    }
  }
}

str输入:1231

输出列表:{1123,1132,1213,1231,1312,1321,2113,2131,2311,3112,3121,3211}

注意，输出是排序的，没有重复的结果。

使用Set操作建模“依赖于其他选择的选择”更容易理解相关排列 使用依赖排列，可用的选择减少，因为位置被从左到右的选定字符填充。递归调用的终端条件是测试可用选择集是否为空。当满足终端条件时，置换完成，并存储到“结果”列表中。

public static List<String> stringPermutation(String s) {
    List<String> results = new ArrayList<>();
    Set<Character> charSet = s.chars().mapToObj(m -> (char) m).collect(Collectors.toSet());
    stringPermutation(charSet, "", results);
    return results;
}

private static void stringPermutation(Set<Character> charSet, 
        String prefix, List<String> results) {
    if (charSet.isEmpty()) {
        results.add(prefix);
        return;
    }
    for (Character c : charSet) {
        Set<Character> newSet = new HashSet<>(charSet);
        newSet.remove(c);
        stringPermutation(newSet, prefix + c, results);
    }
} 

该代码可以泛化为一组对象查找排列。在本例中，我使用了一组颜色。

public enum Color{
    ORANGE,RED,BULE,GREEN,YELLOW;
}

public static List<List<Color>> colorPermutation(Set<Color> colors) {
    List<List<Color>> results = new ArrayList<>();
    List<Color> prefix = new ArrayList<>();
    permutation(colors, prefix, results);
    return results;
}

private static <T> void permutation(Set<T> set, List<T> prefix, List<List<T>> results) {
    if (set.isEmpty()) {
        results.add(prefix);
        return;
    }
    for (T t : set) {
        Set<T> newSet = new HashSet<>(set);
        List<T> newPrefix = new ArrayList<>(prefix);
        newSet.remove(t);
        newPrefix.add(t);
        permutation(newSet, newPrefix, results);
    }
} 

测试代码。

public static void main(String[] args) {
    List<String> stringPerm = stringPermutation("abcde");
    System.out.println("# of permutations:" + stringPerm.size());
    stringPerm.stream().forEach(e -> System.out.println(e));

    Set<Color> colorSet = Arrays.stream(Color.values()).collect(Collectors.toSet());
    List<List<Color>> colorPerm = colorPermutation(colorSet);
    System.out.println("# of permutations:" + colorPerm.size());
    colorPerm.stream().forEach(e -> System.out.println(e));
}