使用递归。

当输入是空字符串时，唯一的排列就是空字符串。尝试将字符串中的每个字母作为第一个字母，然后使用递归调用找到其余字母的所有排列。

import java.util.ArrayList;
import java.util.List;

class Permutation {
    private static List<String> permutation(String prefix, String str) {
        List<String> permutations = new ArrayList<>();
        int n = str.length();
        if (n == 0) {
            permutations.add(prefix);
        } else {
            for (int i = 0; i < n; i++) {
                permutations.addAll(permutation(prefix + str.charAt(i), str.substring(i + 1, n) + str.substring(0, i)));
            }
        }
        return permutations;
    }

    public static void main(String[] args) {
        List<String> perms = permutation("", "abcd");

        String[] array = new String[perms.size()];
        for (int i = 0; i < perms.size(); i++) {
            array[i] = perms.get(i);
        }

        int x = array.length;

        for (final String anArray : array) {
            System.out.println(anArray);
        }
    }
}

以下是我在《破解编程面试》(P54)一书中提出的解决方案:

/**
 * List permutations of a string.
 * 
 * @param s the input string
 * @return  the list of permutations
 */
public static ArrayList<String> permutation(String s) {
    // The result
    ArrayList<String> res = new ArrayList<String>();
    // If input string's length is 1, return {s}
    if (s.length() == 1) {
        res.add(s);
    } else if (s.length() > 1) {
        int lastIndex = s.length() - 1;
        // Find out the last character
        String last = s.substring(lastIndex);
        // Rest of the string
        String rest = s.substring(0, lastIndex);
        // Perform permutation on the rest string and
        // merge with the last character
        res = merge(permutation(rest), last);
    }
    return res;
}

/**
 * @param list a result of permutation, e.g. {"ab", "ba"}
 * @param c    the last character
 * @return     a merged new list, e.g. {"cab", "acb" ... }
 */
public static ArrayList<String> merge(ArrayList<String> list, String c) {
    ArrayList<String> res = new ArrayList<>();
    // Loop through all the string in the list
    for (String s : list) {
        // For each string, insert the last character to all possible positions
        // and add them to the new list
        for (int i = 0; i <= s.length(); ++i) {
            String ps = new StringBuffer(s).insert(i, c).toString();
            res.add(ps);
        }
    }
    return res;
}

字符串"abcd"的运行输出:

第一步:合并[a]和b: [ba, ab] 步骤2:Merge [ba, ab]和c: [cba, bca, bac, cab, acb, abc] 第三步:Merge [cba, bca, bac, cab, acb, abc]和d: [dcba, cdba, cbad, cbca, bdcad

作为Python生成器，带有现代类型提示:

from typing import Iterator


def permutations(string: str, prefix: str = '') -> Iterator[str]:
    if len(string) == 0:
        yield prefix
    for i, character in enumerate(string):
        yield from permutations(string[:i] + string[i + 1:], prefix + character)


for p in permutations('abcd'):
    print(p)

在这里和其他论坛给出的所有解决方案中，我最喜欢Mark Byers。这个描述实际上让我自己思考并编写了代码。 可惜我不能投票支持他的解决方案，因为我是新手。 无论如何，这是我对他的描述的实现

public class PermTest {

    public static void main(String[] args) throws Exception {
        String str = "abcdef";
        StringBuffer strBuf = new StringBuffer(str);
        doPerm(strBuf,0);
    }

    private static void doPerm(StringBuffer str, int index){

        if(index == str.length())
            System.out.println(str);            
        else { //recursively solve this by placing all other chars at current first pos
            doPerm(str, index+1);
            for (int i = index+1; i < str.length(); i++) {//start swapping all other chars with current first char
                swap(str,index, i);
                doPerm(str, index+1);
                swap(str,i, index);//restore back my string buffer
            }
        }
    }

    private  static void swap(StringBuffer str, int pos1, int pos2){
        char t1 = str.charAt(pos1);
        str.setCharAt(pos1, str.charAt(pos2));
        str.setCharAt(pos2, t1);
    }
}   

我更喜欢这个解决方案，而不是第一个解决方案，因为这个解决方案使用StringBuffer。我不会说我的解决方案没有创建任何临时字符串(它实际上在system.out.println中创建，其中调用StringBuffer的toString())。但我只是觉得这比第一个解决方案好太多的字符串字面值被创建。可能有些性能人员可以根据“内存”来评估这一点(对于“时间”来说，由于额外的“交换”，它已经滞后了)

//Rotate and create words beginning with all letter possible and push to stack 1

//Read from stack1 and for each word create words with other letters at the next location by rotation and so on 

/*  eg : man

    1. push1 - man, anm, nma
    2. pop1 - nma ,  push2 - nam,nma
       pop1 - anm ,  push2 - amn,anm
       pop1 - man ,  push2 - mna,man
*/

public class StringPermute {

    static String str;
    static String word;
    static int top1 = -1;
    static int top2 = -1;
    static String[] stringArray1;
    static String[] stringArray2;
    static int strlength = 0;

    public static void main(String[] args) throws IOException {
        System.out.println("Enter String : ");
        InputStreamReader isr = new InputStreamReader(System.in);
        BufferedReader bfr = new BufferedReader(isr);
        str = bfr.readLine();
        word = str;
        strlength = str.length();
        int n = 1;
        for (int i = 1; i <= strlength; i++) {
            n = n * i;
        }
        stringArray1 = new String[n];
        stringArray2 = new String[n];
        push(word, 1);
        doPermute();
        display();
    }

    public static void push(String word, int x) {
        if (x == 1)
            stringArray1[++top1] = word;
        else
            stringArray2[++top2] = word;
    }

    public static String pop(int x) {
        if (x == 1)
            return stringArray1[top1--];
        else
            return stringArray2[top2--];
    }

    public static void doPermute() {

        for (int j = strlength; j >= 2; j--)
            popper(j);

    }

    public static void popper(int length) {
        // pop from stack1 , rotate each word n times and push to stack 2
        if (top1 > -1) {
            while (top1 > -1) {
                word = pop(1);
                for (int j = 0; j < length; j++) {
                    rotate(length);
                    push(word, 2);
                }
            }
        }
        // pop from stack2 , rotate each word n times w.r.t position and push to
        // stack 1
        else {
            while (top2 > -1) {
                word = pop(2);
                for (int j = 0; j < length; j++) {
                    rotate(length);
                    push(word, 1);
                }
            }
        }

    }

    public static void rotate(int position) {
        char[] charstring = new char[100];
        for (int j = 0; j < word.length(); j++)
            charstring[j] = word.charAt(j);

        int startpos = strlength - position;
        char temp = charstring[startpos];
        for (int i = startpos; i < strlength - 1; i++) {
            charstring[i] = charstring[i + 1];
        }
        charstring[strlength - 1] = temp;
        word = new String(charstring).trim();
    }

    public static void display() {
        int top;
        if (top1 > -1) {
            while (top1 > -1)
                System.out.println(stringArray1[top1--]);
        } else {
            while (top2 > -1)
                System.out.println(stringArray2[top2--]);
        }
    }
}

递归是不必要的，甚至你可以直接计算任何排列，这个解决方案使用泛型来排列任何数组。

这里有关于这个algorihtm的很好的信息。

对于c#开发人员来说，这里有更有用的实现。

public static void main(String[] args) {
    String word = "12345";

    Character[] array = ArrayUtils.toObject(word.toCharArray());
    long[] factorials = Permutation.getFactorials(array.length + 1);

    for (long i = 0; i < factorials[array.length]; i++) {
        Character[] permutation = Permutation.<Character>getPermutation(i, array, factorials);
        printPermutation(permutation);
    }
}

private static void printPermutation(Character[] permutation) {
    for (int i = 0; i < permutation.length; i++) {
        System.out.print(permutation[i]);
    }
    System.out.println();
}

该算法计算每个排列的时间和空间复杂度为O(N)。

public class Permutation {
    public static <T> T[] getPermutation(long permutationNumber, T[] array, long[] factorials) {
        int[] sequence = generateSequence(permutationNumber, array.length - 1, factorials);
        T[] permutation = generatePermutation(array, sequence);

        return permutation;
    }

    public static <T> T[] generatePermutation(T[] array, int[] sequence) {
        T[] clone = array.clone();

        for (int i = 0; i < clone.length - 1; i++) {
            swap(clone, i, i + sequence[i]);
        }

        return clone;
    }

    private static int[] generateSequence(long permutationNumber, int size, long[] factorials) {
        int[] sequence = new int[size];

        for (int j = 0; j < sequence.length; j++) {
            long factorial = factorials[sequence.length - j];
            sequence[j] = (int) (permutationNumber / factorial);
            permutationNumber = (int) (permutationNumber % factorial);
        }

        return sequence;
    }

    private static <T> void swap(T[] array, int i, int j) {
        T t = array[i];
        array[i] = array[j];
        array[j] = t;
    }

    public static long[] getFactorials(int length) {
        long[] factorials = new long[length];
        long factor = 1;

        for (int i = 0; i < length; i++) {
            factor *= i <= 1 ? 1 : i;
            factorials[i] = factor;
        }

        return factorials;
    }
}