Transposing a 2D-array in JavaScript

Question

I've got an array of arrays, something like:

[
    [1,2,3],
    [1,2,3],
    [1,2,3],
]

I would like to transpose it to get the following array:

[
    [1,1,1],
    [2,2,2],
    [3,3,3],
]

It's not difficult to programmatically do so using loops:

function transposeArray(array, arrayLength){
    var newArray = [];
    for(var i = 0; i < array.length; i++){
        newArray.push([]);
    };

    for(var i = 0; i < array.length; i++){
        for(var j = 0; j < arrayLength; j++){
            newArray[j].push(array[i][j]);
        };
    };

    return newArray;
}

This, however, seems bulky, and I feel like there should be an easier way to do it. Is there?

Answer 1

output = array[0].map((_, colIndex) => array.map(row => row[colIndex]));

map calls a provided callback function once for each element in an array, in order, and constructs a new array from the results. callback is invoked only for indexes of the array which have assigned values; it is not invoked for indexes which have been deleted or which have never been assigned values.

callback is invoked with three arguments: the value of the element, the index of the element, and the Array object being traversed. [source]

Answer 2

Many good answers here! I consolidated them into one answer and updated some of the code for a more modern syntax:

One-liners inspired by Fawad Ghafoor and Óscar Gómez Alcañiz

function transpose(matrix) {
  return matrix[0].map((col, i) => matrix.map(row => row[i]));
}

function transpose(matrix) {
  return matrix[0].map((col, c) => matrix.map((row, r) => matrix[r][c]));
}

Functional approach style with reduce by Andrew Tatomyr

function transpose(matrix) {
  return matrix.reduce((prev, next) => next.map((item, i) =>
    (prev[i] || []).concat(next[i])
  ), []);
}

Lodash/Underscore by marcel

function tranpose(matrix) {
  return _.zip(...matrix);
}

// Without spread operator.
function transpose(matrix) {
  return _.zip.apply(_, [[1,2,3], [1,2,3], [1,2,3]])
}

Even simpler Lodash/Underscore solution by Vigrant

_.unzip(matrix);

Vanilla approach

function transpose(matrix) {
  const rows = matrix.length, cols = matrix[0].length;
  const grid = [];
  for (let j = 0; j < cols; j++) {
    grid[j] = Array(rows);
  }
  for (let i = 0; i < rows; i++) {
    for (let j = 0; j < cols; j++) {
      grid[j][i] = matrix[i][j];
    }
  }
  return grid;
}

Vanilla in-place ES6 approach inspired by Emanuel Saringan

function transpose(matrix) {
  for (var i = 0; i < matrix.length; i++) {
    for (var j = 0; j < i; j++) {
      const temp = matrix[i][j];
      matrix[i][j] = matrix[j][i];
      matrix[j][i] = temp;
    }
  }
}

// Using destructing
function transpose(matrix) {
  for (var i = 0; i < matrix.length; i++) {
    for (var j = 0; j < i; j++) {
      [matrix[i][j], matrix[j][i]] = [matrix[j][i], matrix[i][j]];
    }
  }
}

Answer 3

here is my implementation in modern browser (without dependency):

transpose = m => m[0].map((x,i) => m.map(x => x[i]))

Answer 4

You could use underscore.js

_.zip.apply(_, [[1,2,3], [1,2,3], [1,2,3]])

Answer 5

shortest way with lodash/underscore and es6:

_.zip(...matrix)

where matrix could be:

const matrix = [[1,2,3], [1,2,3], [1,2,3]];

Answer 6

Neat and pure:

[[0, 1], [2, 3], [4, 5]].reduce((prev, next) => next.map((item, i) =>
    (prev[i] || []).concat(next[i])
), []); // [[0, 2, 4], [1, 3, 5]]

Previous solutions may lead to failure in case an empty array is provided.

Here it is as a function:

function transpose(array) {
    return array.reduce((prev, next) => next.map((item, i) =>
        (prev[i] || []).concat(next[i])
    ), []);
}

console.log(transpose([[0, 1], [2, 3], [4, 5]]));

Update. It can be written even better with spread operator:

const transpose = matrix => matrix.reduce(
    ($, row) => row.map((_, i) => [...($[i] || []), row[i]]), 
    []
)

Answer 7

You can do it in in-place by doing only one pass:

function transpose(arr,arrLen) {
  for (var i = 0; i < arrLen; i++) {
    for (var j = 0; j <i; j++) {
      //swap element[i,j] and element[j,i]
      var temp = arr[i][j];
      arr[i][j] = arr[j][i];
      arr[j][i] = temp;
    }
  }
}

Answer 8

Just another variation using Array.map. Using indexes allows to transpose matrices where M != N:

// Get just the first row to iterate columns first
var t = matrix[0].map(function (col, c) {
    // For each column, iterate all rows
    return matrix.map(function (row, r) { 
        return matrix[r][c]; 
    }); 
});

All there is to transposing is mapping the elements column-first, and then by row.

Answer 9

Another approach by iterating the array from outside to inside and reduce the matrix by mapping inner values.

const
    transpose = array => array.reduce((r, a) => a.map((v, i) => [...(r[i] || []), v]), []),
    matrix = [[1, 2, 3], [1, 2, 3], [1, 2, 3]];

console.log(transpose(matrix));

Answer 10

If you have an option of using Ramda JS and ES6 syntax, then here's another way to do it:

const transpose = a => R.map(c => R.map(r => r[c], a), R.keys(a[0]));

console.log(transpose([
  [1, 2, 3, 4],
  [5, 6, 7, 8],
  [9, 10, 11, 12]
])); // =>  [[1,5,9],[2,6,10],[3,7,11],[4,8,12]]

<script src="https://cdnjs.cloudflare.com/ajax/libs/ramda/0.22.1/ramda.min.js"></script>

Answer 11

If using RamdaJS is an option, this can be achieved in one line: R.transpose(myArray)

Answer 12

Spread syntax should not be used as an alternative to push, it should only be used when you don't want to mutate the existing array.

Algorithm: For every column, just check if for that column there's a row in the resultant matrix, if there's already a row then simply push the element, else create a new row array and then push.

So, unlike many other solutions above, this solution doesn't create new arrays again and again, instead pushes onto the same array.

Also, take some time to appreciate the use of the Nullish Coalescing Operator.

const 
  transpose = arr => arr.reduce((m, r) => (r.forEach((v, i) => (m[i] ??= [], m[i].push(v))), m), []),
  matrix = [[1, 2, 3], [1, 2, 3], [1, 2, 3]]

console.log(transpose(matrix))

Answer 13

You can achieve this without loops by using the following.

• Array
• Array.prototype.map
• Array.prototype.reduce
• Array.prototype.join
• String.prototype.split

It looks very elegant and it does not require any dependencies such as jQuery of Underscore.js.

function transpose(matrix) {  
    return zeroFill(getMatrixWidth(matrix)).map(function(r, i) {
        return zeroFill(matrix.length).map(function(c, j) {
            return matrix[j][i];
        });
    });
}

function getMatrixWidth(matrix) {
    return matrix.reduce(function (result, row) {
        return Math.max(result, row.length);
    }, 0);
}

function zeroFill(n) {
    return new Array(n+1).join('0').split('').map(Number);
}

Minified

function transpose(m){return zeroFill(m.reduce(function(m,r){return Math.max(m,r.length)},0)).map(function(r,i){return zeroFill(m.length).map(function(c,j){return m[j][i]})})}function zeroFill(n){return new Array(n+1).join("0").split("").map(Number)}

---

Here is a demo I threw together. Notice the lack of loops :-)

// Create a 5 row, by 9 column matrix.
var m = CoordinateMatrix(5, 9);

// Make the matrix an irregular shape.
m[2] = m[2].slice(0, 5);
m[4].pop();

// Transpose and print the matrix.
println(formatMatrix(transpose(m)));

function Matrix(rows, cols, defaultVal) {
    return AbstractMatrix(rows, cols, function(r, i) {
        return arrayFill(cols, defaultVal);
    });
}
function ZeroMatrix(rows, cols) {
    return AbstractMatrix(rows, cols, function(r, i) {
        return zeroFill(cols);
    });
}
function CoordinateMatrix(rows, cols) {
    return AbstractMatrix(rows, cols, function(r, i) {
        return zeroFill(cols).map(function(c, j) {
            return [i, j];
        });
    });
}
function AbstractMatrix(rows, cols, rowFn) {
    return zeroFill(rows).map(function(r, i) {
        return rowFn(r, i);
    });
}
/** Matrix functions. */
function formatMatrix(matrix) {
    return matrix.reduce(function (result, row) {
        return result + row.join('\t') + '\n';
    }, '');
}
function copy(matrix) {  
    return zeroFill(matrix.length).map(function(r, i) {
        return zeroFill(getMatrixWidth(matrix)).map(function(c, j) {
            return matrix[i][j];
        });
    });
}
function transpose(matrix) {  
    return zeroFill(getMatrixWidth(matrix)).map(function(r, i) {
        return zeroFill(matrix.length).map(function(c, j) {
            return matrix[j][i];
        });
    });
}
function getMatrixWidth(matrix) {
    return matrix.reduce(function (result, row) {
        return Math.max(result, row.length);
    }, 0);
}
/** Array fill functions. */
function zeroFill(n) {
  return new Array(n+1).join('0').split('').map(Number);
}
function arrayFill(n, defaultValue) {
    return zeroFill(n).map(function(value) {
        return defaultValue || value;
    });
}
/** Print functions. */
function print(str) {
    str = Array.isArray(str) ? str.join(' ') : str;
    return document.getElementById('out').innerHTML += str || '';
}
function println(str) {
    print.call(null, [].slice.call(arguments, 0).concat(['<br />']));
}

#out {
    white-space: pre;
}

<div id="out"></div>

Answer 14

ES6 1liners as :

let invert = a => a[0].map((col, c) => a.map((row, r) => a[r][c]))

so same as Óscar's, but as would you rather rotate it clockwise :

let rotate = a => a[0].map((col, c) => a.map((row, r) => a[r][c]).reverse())

let a = [
    [1,1,1]
    , ["_","_","1"]
]
let b = rotate(a);
let c = rotate(b);
let d = rotate(c);
console.log(`a ${a.join("\na ")}`);
console.log(`b ${b.join("\nb ")}`);
console.log(`c ${c.join("\nc ")}`);
console.log(`d ${d.join("\nd ")}`);

Yields

a 1,1,1 
a _,_,1

b _,1
b _,1
b 1,1 

c 1,_,_
c 1,1,1

d 1,1
d 1,_
d 1,_

Answer 15

Edit: This answer would not transpose the matrix, but rotate it. I didn't read the question carefully in the first place :D

clockwise and counterclockwise rotation:

    function rotateCounterClockwise(a){
        var n=a.length;
        for (var i=0; i<n/2; i++) {
            for (var j=i; j<n-i-1; j++) {
                var tmp=a[i][j];
                a[i][j]=a[j][n-i-1];
                a[j][n-i-1]=a[n-i-1][n-j-1];
                a[n-i-1][n-j-1]=a[n-j-1][i];
                a[n-j-1][i]=tmp;
            }
        }
        return a;
    }

    function rotateClockwise(a) {
        var n=a.length;
        for (var i=0; i<n/2; i++) {
            for (var j=i; j<n-i-1; j++) {
                var tmp=a[i][j];
                a[i][j]=a[n-j-1][i];
                a[n-j-1][i]=a[n-i-1][n-j-1];
                a[n-i-1][n-j-1]=a[j][n-i-1];
                a[j][n-i-1]=tmp;
            }
        }
        return a;
    }

Answer 16

I found the above answers either hard to read or too verbose, so I write one myself. And I think this is most intuitive way to implement transpose in linear algebra, you don't do value exchange, but just insert each element into the right place in the new matrix:

function transpose(matrix) {
  const rows = matrix.length
  const cols = matrix[0].length

  let grid = []
  for (let col = 0; col < cols; col++) {
    grid[col] = []
  }
  for (let row = 0; row < rows; row++) {
    for (let col = 0; col < cols; col++) {
      grid[col][row] = matrix[row][col]
    }
  }
  return grid
}

Answer 17

const transpose = array => array[0].map((r, i) => array.map(c => c[i]));
console.log(transpose([[2, 3, 4], [5, 6, 7]]));

Answer 18

Since nobody so far mentioned a functional recursive approach here is my take. An adaptation of Haskell's Data.List.transpose.

var transpose = as => as.length ? as[0].length ? [as.reduce((rs, a) => a.length ? (rs.push(a[0]), rs) :
    rs, []
  ), ...transpose(as.map(a => a.slice(1)))] :
  transpose(as.slice(1)) :
  [],
  mtx = [
    [1],
    [1, 2],
    [1, 2, 3]
  ];

console.log(transpose(mtx))

.as-console-wrapper {
  max-height: 100% !important
}

Answer 19

reverseValues(values) {
        let maxLength = values.reduce((acc, val) => Math.max(val.length, acc), 0);
        return [...Array(maxLength)].map((val, index) => values.map((v) => v[index]));
}

Answer 20

I didn't find an answer that satisfied me, so I wrote one myself, I think it is easy to understand and implement and suitable for all situations.

    transposeArray: function (mat) {
        let newMat = [];
        for (let j = 0; j < mat[0].length; j++) {  // j are columns
            let temp = [];
            for (let i = 0; i < mat.length; i++) {  // i are rows
                temp.push(mat[i][j]);  // so temp will be the j(th) column in mat
            }
            newMat.push(temp);  // then just push every column in newMat
        }
        return newMat;
    }

Answer 21

Adding TS version here.

const transpose = <T>(m: Array<Array<T>>): Array<Array<T>> => m[0].map((_, i) => m.map(x => x[i]));

Answer 22

const arr = [
  [1,2,3],
  [1,2,3],
  [1,2,3]
];

var output = arr.map((row, rowIndex) => {
  row.forEach((col, colIndex) => {
    if(colIndex > rowIndex) {
      var temp = arr[colIndex][rowIndex];
      arr[colIndex][rowIndex] = row[colIndex];
      row[colIndex] = temp;
    }
  });
  return row;
});

console.log(output);

Answer 23

function invertArray(array,arrayWidth,arrayHeight) {
  var newArray = [];
  for (x=0;x<arrayWidth;x++) {
    newArray[x] = [];
    for (y=0;y<arrayHeight;y++) {
        newArray[x][y] = array[y][x];
    }
  }
  return newArray;
}

Answer 24

One-liner that does not change given array.

a[0].map((col, i) => a.map(([...row]) => row[i]))

Answer 25

There is more efficient solution, in cases where n = m (number of rows equals to number of columns).

const matrix = [
   [1,1,1,1],
   [2,2,2,2],
   [3,3,3,3],
   [4,4,4,4]
];

matrix.every((r, i, a) => (
   r.every((_, j) => (
      j = a.length-j-1,
      [ r[j], a[j][i] ] = [ a[j][i], r[j] ],
      i < j-1
   )), 
   i < length-2
));

console.log(matrix);
/*
Prints:
[
   [1,2,3,4],
   [1,2,3,4],
   [1,2,3,4],
   [1,2,3,4]
]
*/

The example above will do only 6 iterations.
For bigger matrix, say 100x100 it will do 4,900 iterations, this is 51% faster than a full scan.

The principle is simple. You only need to iterate through the upper diagonal half of the matrix (Xs in the matrix below), and switch with bottom diagonal half (Os in the matrix below).

[
   [-,X,X,X],
   [O,-,X,X],
   [O,O,-,X],
   [O,O,O,-],
]

This way, you can save a lot of running time, especially in a very large matrix.