我尝试着做一个简洁而可读的版本，并且是纯函数式编程。

function stringPermutations ([...input]) {
  if (input.length === 1) return input;

  return input
    .map((thisChar, index) => {
      const remainingChars = [...input.slice(0, index), ...input.slice(index + 1)];
      return stringPermutations(remainingChars)
        .map(remainder => thisChar + remainder);
    })
    .reduce((acc, cur) => [...acc, ...cur]);
}

注意，参数格式化将输入字符串转换为数组。不知道这是不是有点太神奇了。我不确定在野外见过。为了真正的可读性，我可能会用input =[…]输入]作为函数的第一行。

这是Heap算法的实现(类似于@le_m算法)，只是它是递归的。

function permute_kingzee(arr,n=arr.length,out=[]) {
    if(n == 1) {
        return out.push(arr.slice());
    } else {
        for(let i=0; i<n; i++) {
            permute_kingzee(arr,n-1, out);
            let j = ( n % 2 == 0 ) ? i : 0;
            let t = arr[n-1];
            arr[n-1] = arr[j];
            arr[j] = t;
        }
        return out;
    }
}

它看起来也快得多:https://jsfiddle.net/3brqzaLe/

这是一个非常好的map/reduce用例:

function permutations(arr) {
    return (arr.length === 1) ? arr :
    arr.reduce((acc, cv, index) => {
        let remaining = [...arr];
        remaining.splice(index, 1);
        return acc.concat(permutations(remaining).map(a => [].concat(cv,a)));
    }, []);
}

首先，我们处理基本情况，如果数组中只有一个项，则返回该数组 在所有其他情况下 我们创建一个空数组 遍历输入数组 并添加当前值的数组和剩余数组的所有排列[].concat(cv,a)

这是一个最小的ES6版本。扁平化和无函数可以从Lodash中提取。

const flatten = xs =>
    xs.reduce((cum, next) => [...cum, ...next], []);

const without = (xs, x) =>
    xs.filter(y => y !== x);

const permutations = xs =>
    flatten(xs.map(x =>
        xs.length < 2
            ? [xs]
            : permutations(without(xs, x)).map(perm => [x, ...perm])
    ));

结果:

permutations([1,2,3])
// [[1, 2, 3], [1, 3, 2], [2, 1, 3], [2, 3, 1], [3, 1, 2], [3, 2, 1]]

  let permutations = []

  permutate([], {
    color: ['red', 'green'],
    size: ['big', 'small', 'medium'],
    type: ['saison', 'oldtimer']
  })

  function permutate (currentVals, remainingAttrs) {
    remainingAttrs[Object.keys(remainingAttrs)[0]].forEach(attrVal => {
      let currentValsNew = currentVals.slice(0)
      currentValsNew.push(attrVal)

      if (Object.keys(remainingAttrs).length > 1) {
        let remainingAttrsNew = JSON.parse(JSON.stringify(remainingAttrs))
        delete remainingAttrsNew[Object.keys(remainingAttrs)[0]]

        permutate(currentValsNew, remainingAttrsNew)
      } else {
        permutations.push(currentValsNew)
      }
    })
  }

结果:

[ 
  [ 'red', 'big', 'saison' ],
  [ 'red', 'big', 'oldtimer' ],
  [ 'red', 'small', 'saison' ],
  [ 'red', 'small', 'oldtimer' ],
  [ 'red', 'medium', 'saison' ],
  [ 'red', 'medium', 'oldtimer' ],
  [ 'green', 'big', 'saison' ],
  [ 'green', 'big', 'oldtimer' ],
  [ 'green', 'small', 'saison' ],
  [ 'green', 'small', 'oldtimer' ],
  [ 'green', 'medium', 'saison' ],
  [ 'green', 'medium', 'oldtimer' ] 
]

下面是递归函数=> https://recursion.vercel.app/的可视化

 var permute = function (nums) {
  let answer = [];

  var permuter = function (arr, permutation = []) {
    if (arr.length === 0) {
      return answer.push(permutation);
    } else {
      for (let i = 0; i < arr.length; i++) {
        let currentArr = arr.slice();
        let next = currentArr.splice(i, 1);
        permuter(currentArr, permutation.concat(next)); ///we use concat because splice returns an array.
      }
    }
  };

  permuter(nums);

  return answer;
};