without Generate list of all possible permutations of a string

print all permutations of a string without duplicates (24)

There are a lot of good answers here. I also suggest a very simple recursive solution in C++.

#include <string>
#include <iostream>

template<typename Consume>
void permutations(std::string s, Consume consume, std::size_t start = 0) {
    if (start == s.length()) consume(s);
    for (std::size_t i = start; i < s.length(); i++) {
        std::swap(s[start], s[i]);
        permutations(s, consume, start + 1);

int main(void) {
    std::string s = "abcd";
    permutations(s, [](std::string s) {
        std::cout << s << std::endl;

Note: strings with repeated characters will not produce unique permutations.

How would I go about generating a list of all possible permutations of a string between x and y characters in length, containing a variable list of characters.

Any language would work, but it should be portable.

It's better to use backtracking

#include <stdio.h>
#include <string.h>

void swap(char *a, char *b) {
    char temp;
    temp = *a;
    *a = *b;
    *b = temp;

void print(char *a, int i, int n) {
    int j;
    if(i == n) {
        printf("%s\n", a);
    } else {
        for(j = i; j <= n; j++) {
            swap(a + i, a + j);
            print(a, i + 1, n);
            swap(a + i, a + j);

int main(void) {
    char a[100];
    print(a, 0, strlen(a) - 1);
    return 0;

Here is a simple word C# recursive solution:


public ArrayList CalculateWordPermutations(string[] letters, ArrayList words, int index)
            bool finished = true;
            ArrayList newWords = new ArrayList();
            if (words.Count == 0)
                foreach (string letter in letters)

            for(int j=index; j<words.Count; j++)
                string word = (string)words[j];
                for(int i =0; i<letters.Length; i++)
                        finished = false;
                        string newWord = (string)word.Clone();
                        newWord += letters[i];

            foreach (string newWord in newWords)

            if(finished  == false)
                CalculateWordPermutations(letters, words, words.Count - newWords.Count);
            return words;


string[] letters = new string[]{"a","b","c"};
ArrayList words = CalculateWordPermutations(letters, new ArrayList(), 0);

I just whipped this up quick in Ruby:

def perms(x, y, possible_characters)
  all = [""]
  current_array = all.clone
  1.upto(y) { |iteration|
    next_array = []
    current_array.each { |string|
      possible_characters.each { |c|
        value = string + c
        next_array.insert next_array.length, value
        all.insert all.length, value
    current_array = next_array
  all.delete_if { |string| string.length < x }

You might look into language API for built in permutation type functions, and you might be able to write more optimized code, but if the numbers are all that high, I'm not sure there is much of a way around having a lot of results.

Anyways, the idea behind the code is start with string of length 0, then keep track of all the strings of length Z where Z is the current size in the iteration. Then, go through each string and append each character onto each string. Finally at the end, remove any that were below the x threshold and return the result.

I didn't test it with potentially meaningless input (null character list, weird values of x and y, etc).

I'm not sure why you would want to do this in the first place. The resulting set for any moderately large values of x and y will be huge, and will grow exponentially as x and/or y get bigger.

Lets say your set of possible characters is the 26 lowercase letters of the alphabet, and you ask your application to generate all permutations where length = 5. Assuming you don't run out of memory you'll get 11,881,376 (i.e. 26 to the power of 5) strings back. Bump that length up to 6, and you'll get 308,915,776 strings back. These numbers get painfully large, very quickly.

Here's a solution I put together in Java. You'll need to provide two runtime arguments (corresponding to x and y). Have fun.

public class GeneratePermutations {
    public static void main(String[] args) {
        int lower = Integer.parseInt(args[0]);
        int upper = Integer.parseInt(args[1]);

        if (upper < lower || upper == 0 || lower == 0) {

        for (int length = lower; length <= upper; length++) {
            generate(length, "");

    private static void generate(int length, String partial) {
        if (length <= 0) {
        } else {
            for (char c = 'a'; c <= 'z'; c++) {
                generate(length - 1, partial + c);

This is a translation of Mike's Ruby version, into Common Lisp:

(defun perms (x y original-string)
  (loop with all = (list "")
        with current-array = (list "")
        for iteration from 1 to y
        do (loop with next-array = nil
                 for string in current-array
                 do (loop for c across original-string
                          for value = (concatenate 'string string (string c))
                          do (push value next-array)
                             (push value all))
                    (setf current-array (reverse next-array)))
        finally (return (nreverse (delete-if #'(lambda (el) (< (length el) x)) all)))))

And another version, slightly shorter and using more loop facility features:

(defun perms (x y original-string)
  (loop repeat y
        collect (loop for string in (or (car (last sets)) (list ""))
                      append (loop for c across original-string
                                   collect (concatenate 'string string (string c)))) into sets
        finally (return (loop for set in sets
                              append (loop for el in set when (>= (length el) x) collect el)))))

c# iterative:

public List<string> Permutations(char[] chars)
        List<string> words = new List<string>();
        for (int i = 1; i < chars.Length; ++i)
            int currLen = words.Count;
            for (int j = 0; j < currLen; ++j)
                var w = words[j];
                for (int k = 0; k <= w.Length; ++k)
                    var nstr = w.Insert(k, chars[i].ToString());
                    if (k == 0)
                        words[j] = nstr;
        return words;

def gen( x,y,list): #to generate all strings inserting y at different positions
list = []
list.append( y+x )
for i in range( len(x) ):
    list.append( func(x,0,i) + y + func(x,i+1,len(x)-1) )
return list 

def func( x,i,j ): #returns x[i..j]
z = '' 
for i in range(i,j+1):
    z = z+x[i]
return z 

def perm( x , length , list ): #perm function
if length == 1 : # base case
    list.append( x[len(x)-1] )
    return list 
    lists = perm( x , length-1 ,list )
    lists_temp = lists #temporarily storing the list 
    lists = []
    for i in range( len(lists_temp) ) :
        list_temp = gen(lists_temp[i],x[length-2],lists)
        lists += list_temp 
    return lists

Though this doesn't answer your question exactly, here's one way to generate every permutation of the letters from a number of strings of the same length: eg, if your words were "coffee", "joomla" and "moodle", you can expect output like "coodle", "joodee", "joffle", etc.

Basically, the number of combinations is the (number of words) to the power of (number of letters per word). So, choose a random number between 0 and the number of combinations - 1, convert that number to base (number of words), then use each digit of that number as the indicator for which word to take the next letter from.

eg: in the above example. 3 words, 6 letters = 729 combinations. Choose a random number: 465. Convert to base 3: 122020. Take the first letter from word 1, 2nd from word 2, 3rd from word 2, 4th from word 0... and you get... "joofle".

If you wanted all the permutations, just loop from 0 to 728. Of course, if you're just choosing one random value, a much simpler less-confusing way would be to loop over the letters. This method lets you avoid recursion, should you want all the permutations, plus it makes you look like you know Maths(tm)!

If the number of combinations is excessive, you can break it up into a series of smaller words and concatenate them at the end.

Here's a non-recursive version I came up with, in javascript. It's not based on Knuth's non-recursive one above, although it has some similarities in element swapping. I've verified its correctness for input arrays of up to 8 elements.

A quick optimization would be pre-flighting the out array and avoiding push().

The basic idea is:

  1. Given a single source array, generate a first new set of arrays which swap the first element with each subsequent element in turn, each time leaving the other elements unperturbed. eg: given 1234, generate 1234, 2134, 3214, 4231.

  2. Use each array from the previous pass as the seed for a new pass, but instead of swapping the first element, swap the second element with each subsequent element. Also, this time, don't include the original array in the output.

  3. Repeat step 2 until done.

Here is the code sample:

function oxe_perm(src, depth, index)
    var perm = src.slice();     // duplicates src.
    perm = perm.split("");
    perm[depth] = src[index];
    perm[index] = src[depth];
    perm = perm.join("");
    return perm;

function oxe_permutations(src)
    out = new Array();


    for (depth = 0; depth < src.length; depth++) {
        var numInPreviousPass = out.length;
        for (var m = 0; m < numInPreviousPass; ++m) {
            for (var n = depth + 1; n < src.length; ++n) {
                out.push(oxe_perm(out[m], depth, n));

    return out;

Some working Java code based on Sarp's answer:

public class permute {

    static void permute(int level, String permuted,
                    boolean used[], String original) {
        int length = original.length();
        if (level == length) {
        } else {
            for (int i = 0; i < length; i++) {
                if (!used[i]) {
                    used[i] = true;
                    permute(level + 1, permuted + original.charAt(i),
                       used, original);
                    used[i] = false;

    public static void main(String[] args) {
        String s = "hello";
        boolean used[] = {false, false, false, false, false};
        permute(0, "", used, s);

Recursive solution in C++

int main (int argc, char * const argv[]) {
        string s = "sarp";
        bool used [4];
        permute(0, "", used, s);

void permute(int level, string permuted, bool used [], string &original) {
    int length = original.length();

    if(level == length) { // permutation complete, display
        cout << permuted << endl;
    } else {
        for(int i=0; i<length; i++) { // try to add an unused character
            if(!used[i]) {
                used[i] = true;
                permute(level+1, original[i] + permuted, used, original); // find the permutations starting with this string
                used[i] = false;

Here is a simple solution in C#.

It generates only the distinct permutations of a given string.

    static public IEnumerable<string> permute(string word)
        if (word.Length > 1)

            char character = word[0];
            foreach (string subPermute in permute(word.Substring(1)))

                for (int index = 0; index <= subPermute.Length; index++)
                    string pre = subPermute.Substring(0, index);
                    string post = subPermute.Substring(index);

                    if (post.Contains(character))

                    yield return pre + character + post;

            yield return word;

def permutation(str)
  posibilities = []
  str.split('').each do |char|
    if posibilities.size == 0
      posibilities[0] = char.downcase
      posibilities[1] = char.upcase
      posibilities_count = posibilities.length
      posibilities = posibilities + posibilities
      posibilities_count.times do |i|
        posibilities[i] += char.downcase
        posibilities[i+posibilities_count] += char.upcase

Here is my take on a non recursive version

There is an iterative Java implementation in UncommonsMaths (works for a list of objects):

 * Generate the indices into the elements array for the next permutation. The
 * algorithm is from Kenneth H. Rosen, Discrete Mathematics and its 
 * Applications, 2nd edition (NY: McGraw-Hill, 1991), p. 284)
private void generateNextPermutationIndices()
    if (remainingPermutations == 0)
        throw new IllegalStateException("There are no permutations " +
             "remaining. Generator must be reset to continue using.");
    else if (remainingPermutations < totalPermutations)
        // Find largest index j with 
        // permutationIndices[j] < permutationIndices[j + 1]
        int j = permutationIndices.length - 2;
        while (permutationIndices[j] > permutationIndices[j + 1])

        // Find index k such that permutationIndices[k] is smallest integer 
        // greater than permutationIndices[j] to the right
        // of permutationIndices[j].
        int k = permutationIndices.length - 1;
        while (permutationIndices[j] > permutationIndices[k])

        // Interchange permutation indices.
        int temp = permutationIndices[k];
        permutationIndices[k] = permutationIndices[j];
        permutationIndices[j] = temp;

        // Put tail end of permutation after jth position in increasing order.
        int r = permutationIndices.length - 1;
        int s = j + 1;

        while (r > s)
            temp = permutationIndices[s];
            permutationIndices[s] = permutationIndices[r];
            permutationIndices[r] = temp;

 * Generate the next permutation and return a list containing
 * the elements in the appropriate order.  This overloaded method
 * allows the caller to provide a list that will be used and returned.
 * The purpose of this is to improve performance when iterating over
 * permutations.  If the {@link #nextPermutationAsList()} method is
 * used it will create a new list every time.  When iterating over
 * permutations this will result in lots of short-lived objects that
 * have to be garbage collected.  This method allows a single list
 * instance to be reused in such circumstances.
 * @param destination Provides a list to use to create the
 * permutation.  This is the list that will be returned, once
 * it has been filled with the elements in the appropriate order.
 * @return The next permutation as a list.
public List<T> nextPermutationAsList(List<T> destination)
    // Generate actual permutation.
    for (int i : permutationIndices)
    return destination;

Full source

The pythonic solution:

from itertools import permutations
s = 'ABCDEF'
p = [''.join(x) for x in permutations(s)]

There are several ways to do this. Common methods use recursion, memoization, or dynamic programming. The basic idea is that you produce a list of all strings of length 1, then in each iteration, for all strings produced in the last iteration, add that string concatenated with each character in the string individually. (the variable index in the code below keeps track of the start of the last and the next iteration)

Some pseudocode:

list = originalString.split('')
index = (0,0)
list = [""]
for iteration n in 1 to y:
  index = (index[1], len(list))
  for string s in list.subset(index[0] to end):
    for character c in originalString:
      list.add(s + c)

you'd then need to remove all strings less than x in length, they'll be the first (x-1) * len(originalString) entries in the list.

In ruby:

str = "a"
100_000_000.times {puts str.next!}

It is quite fast, but it is going to take some time =). Of course, you can start at "aaaaaaaa" if the short strings aren't interesting to you.

I might have misinterpreted the actual question though - in one of the posts it sounded as if you just needed a bruteforce library of strings, but in the main question it sounds like you need to permutate a particular string.

Your problem is somewhat similar to this one: http://beust.com/weblog/archives/000491.html (list all integers in which none of the digits repeat themselves, which resulted in a whole lot of languages solving it, with the ocaml guy using permutations, and some java guy using yet another solution).

Ruby answer that works:

class String
  def each_char_with_index
    0.upto(size - 1) do |index|
      yield(self[index..index], index)
  def remove_char_at(index)
    return self[1..-1] if index == 0
    self[0..(index-1)] + self[(index+1)..-1]

def permute(str, prefix = '')
  if str.size == 0
    puts prefix
  str.each_char_with_index do |char, index|
    permute(str.remove_char_at(index), prefix + char)

# example
# permute("abc")

... and here is the C version:

void permute(const char *s, char *out, int *used, int len, int lev)
    if (len == lev) {
        out[lev] = '\0';

    int i;
    for (i = 0; i < len; ++i) {
        if (! used[i])

        used[i] = 1;
        out[lev] = s[i];
        permute(s, out, used, len, lev + 1);
        used[i] = 0;

Non recursive solution according to Knuth, Python example:

def nextPermutation(perm):
    k0 = None
    for i in range(len(perm)-1):
        if perm[i]<perm[i+1]:
    if k0 == None:
        return None

    l0 = k0+1
    for i in range(k0+1, len(perm)):
        if perm[k0] < perm[i]:
            l0 = i

    perm[k0], perm[l0] = perm[l0], perm[k0]
    perm[k0+1:] = reversed(perm[k0+1:])
    return perm

while perm:
    print perm
    perm = nextPermutation(perm)

This code in python, when called with allowed_characters set to [0,1] and 4 character max, would generate 2^4 results:

['0000', '0001', '0010', '0011', '0100', '0101', '0110', '0111', '1000', '1001', '1010', '1011', '1100', '1101', '1110', '1111']

def generate_permutations(chars = 4) :

#modify if in need!
    allowed_chars = [

    status = []
    for tmp in range(chars) :

    last_char = len(allowed_chars)

    rows = []
    for x in xrange(last_char ** chars) :
        for y in range(chars - 1 , -1, -1) :
            key = status[y]
            rows[x] = allowed_chars[key] + rows[x]

        for pos in range(chars - 1, -1, -1) :
            if(status[pos] == last_char - 1) :
                status[pos] = 0
            else :
                status[pos] += 1

    return rows

import sys

print generate_permutations()

Hope this is of use to you. Works with any character, not only numbers

import java.util.*;

public class all_subsets {
    public static void main(String[] args) {
        String a = "abcd";
        for(String s: all_perm(a)) {

    public static Set<String> concat(String c, Set<String> lst) {
        HashSet<String> ret_set = new HashSet<String>();
        for(String s: lst) {
        return ret_set;

    public static HashSet<String> all_perm(String a) {
        HashSet<String> set = new HashSet<String>();
        if(a.length() == 1) {
        } else {
            for(int i=0; i<a.length(); i++) {
                set.addAll(concat(a.charAt(i)+"", all_perm(a.substring(0, i)+a.substring(i+1, a.length()))));
        return set;