Javascript equivalent of Python’s zip function

Javascript equivalent of Python’s zip function

Is there a javascript equivalent of Python’s zip function? That is, given multiple arrays of equal lengths create an array of pairs.
For instance, if I have three arrays that look like this:
var array1 = [1, 2, 3];
var array2 = [‘a’,’b’,’c’];
var array3 = [4, 5, 6];

The output array should be:
var output array:[[1,’a’,4], [2,’b’,5], [3,’c’,6]]

Solutions/Answers:

Solution 1:

2016 update:

Here’s a snazzier Ecmascript 6 version:

zip= rows=>rows[0].map((_,c)=>rows.map(row=>row[c]))

Illustration equiv. to Python{zip(*args)}:

> zip([['row0col0', 'row0col1', 'row0col2'],
       ['row1col0', 'row1col1', 'row1col2']]);
[["row0col0","row1col0"],
 ["row0col1","row1col1"],
 ["row0col2","row1col2"]]

(and FizzyTea points out that ES6 has variadic argument syntax, so the following function definition will act like python, but see below for disclaimer… this will not be its own inverse so zip(zip(x)) will not equal x; though as Matt Kramer points out zip(...zip(...x))==x (like in regular python zip(*zip(*x))==x))

Alternative definition equiv. to Python{zip}:

> zip = (...rows) => [...rows[0]].map((_,c) => rows.map(row => row[c]))
> zip( ['row0col0', 'row0col1', 'row0col2'] ,
       ['row1col0', 'row1col1', 'row1col2'] );
             // note zip(row0,row1), not zip(matrix)
same answer as above

(Do note that the ... syntax may have performance issues at this time, and possibly in the future, so if you use the second answer with variadic arguments, you may want to perf test it.)


Here’s a oneliner:

function zip(arrays) {
    return arrays[0].map(function(_,i){
        return arrays.map(function(array){return array[i]})
    });
}

// > zip([[1,2],[11,22],[111,222]])
// [[1,11,111],[2,22,222]]]

// If you believe the following is a valid return value:
//   > zip([])
//   []
// then you can special-case it, or just do
//  return arrays.length==0 ? [] : arrays[0].map(...)

The above assumes that the arrays are of equal size, as they should be. It also assumes you pass in a single list of lists argument, unlike Python’s version where the argument list is variadic. If you want all of these “features”, see below. It takes just about 2 extra lines of code.

The following will mimic Python’s zip behavior on edge cases where the arrays are not of equal size, silently pretending the longer parts of arrays don’t exist:

function zip() {
    var args = [].slice.call(arguments);
    var shortest = args.length==0 ? [] : args.reduce(function(a,b){
        return a.length<b.length ? a : b
    });

    return shortest.map(function(_,i){
        return args.map(function(array){return array[i]})
    });
}

// > zip([1,2],[11,22],[111,222,333])
// [[1,11,111],[2,22,222]]]

// > zip()
// []

This will mimic Python’s itertools.zip_longest behavior, inserting undefined where arrays are not defined:

function zip() {
    var args = [].slice.call(arguments);
    var longest = args.reduce(function(a,b){
        return a.length>b.length ? a : b
    }, []);

    return longest.map(function(_,i){
        return args.map(function(array){return array[i]})
    });
}

// > zip([1,2],[11,22],[111,222,333])
// [[1,11,111],[2,22,222],[null,null,333]]

// > zip()
// []

If you use these last two version (variadic aka. multiple-argument versions), then zip is no longer its own inverse. To mimic the zip(*[...]) idiom from Python, you will need to do zip.apply(this, [...]) when you want to invert the zip function or if you want to similarly have a variable number of lists as input.

Related:  window.close and self.close do not close the window in Chrome

addendum:

To make this handle any iterable (e.g. in Python you can use zip on strings, ranges, map objects, etc.), you could define the following:

function iterView(iterable) {
    // returns an array equivalent to the iterable
}

However if you write zip in the following way, even that won’t be necessary:

function zip(arrays) {
    return Array.apply(null,Array(arrays[0].length)).map(function(_,i){
        return arrays.map(function(array){return array[i]})
    });
}

Demo:

> JSON.stringify( zip(['abcde',[1,2,3,4,5]]) )
[["a",1],["b",2],["c",3],["d",4],["e",5]]

(Or you could use a range(...) Python-style function if you’ve written one already. Eventually you will be able to use ECMAScript array comprehensions or generators.)

Solution 2:

Check out the library Underscore.

Underscore provides over 100 functions that support both your favorite workaday functional helpers: map, filter, invoke — as well as more specialized goodies: function binding, javascript templating, creating quick indexes, deep equality testing, and so on.

– Say the people who made it

I recently started using it specifically for the zip() function and it has left a great first impression. I am using jQuery and CoffeeScript, and it just goes perfectly with them. Underscore picks up right where they leave off and so far it hasn’t let me down. Oh by the way, it’s only 3kb minified.

Check it out.

Solution 3:

In addition to ninjagecko’s excellent and comprehensive answer, all it takes to zip two JS-arrays into a “tuple-mimic” is:

//Arrays: aIn, aOut
Array.prototype.map.call( aIn, function(e,i){return [e, aOut[i]];})

Explanation:
Since Javascript doesn’t have a tuples type, functions for tuples, lists and sets wasn’t a high priority in the language specification.
Otherwise, similar behavior is accessible in a straightforward manner via Array map in JS >1.6. (map is actually often implemented by JS engine makers in many >JS 1.4 engines, despite not specified).
The major difference to Python’s zip, izip,… results from map‘s functional style, since map requires a function-argument. Additionally it is a function of the Array-instance. One may use Array.prototype.map instead, if an extra declaration for the input is an issue.

Related:  JavaScript private methods

Example:

_tarrin = [0..constructor, function(){}, false, undefined, '', 100, 123.324,
         2343243243242343242354365476453654625345345, 'sdf23423dsfsdf',
         'sdf2324.234dfs','234,234fsf','100,100','100.100']
_parseInt = function(i){return parseInt(i);}
_tarrout = _tarrin.map(_parseInt)
_tarrin.map(function(e,i,a){return [e, _tarrout[i]]})

Result:

//'('+_tarrin.map(function(e,i,a){return [e, _tarrout[i]]}).join('),\n(')+')'
>>
(function Number() { [native code] },NaN),
(function (){},NaN),
(false,NaN),
(,NaN),
(,NaN),
(100,100),
(123.324,123),
(2.3432432432423434e+42,2),
(sdf23423dsfsdf,NaN),
(sdf2324.234dfs,NaN),
(234,234fsf,234),
(100,100,100),
(100.100,100)

Related Performance:

Using map over for-loops:

See: What is the most efficient way of merging [1,2] and [7,8] into [[1,7], [2,8]]

zip tests

Note: the base types such as false and undefined do not posess a prototypal object-hierarchy and thus do not expose a toString function. Hence these are shown as empty in the output.
As parseInt‘s second argument is the base/number radix, to which to convert the number to, and since map passes the index as the second argument to its argument-function, a wrapper function is used.

Solution 4:

Along other Python-like functions, pythonic offers a zip function, with the extra benefit of returning a lazy evaluated Iterator, similar to the behaviour of its Python counterpart:

import {zip, zipLongest} from 'pythonic';

const arr1 = ['a', 'b'];
const arr2 = ['c', 'd', 'e'];
for (const [first, second] of zip(arr1, arr2))
    console.log(`first: ${first}, second: ${second}`);
// first: a, second: c
// first: b, second: d

for (const [first, second] of zipLongest(arr1, arr2))
    console.log(`first: ${first}, second: ${second}`);
// first: a, second: c
// first: b, second: d
// first: undefined, second: e

// unzip
const [arrayFirst, arraySecond] = [...zip(...zip(arr1, arr2))];

Disclosure I’m author and maintainer of Pythonic

Solution 5:

The Python has two functions: zip and itertools.zip_longest. Implementation on JS/ES6 is like this:

Related:  convert iso date to milliseconds in javascript

Implementation Python`s zip on JS/ES6

const zip = (...arrays) => {
    const length = Math.min(...arrays.map(arr => arr.length));
    return Array.from({ length }, (value, index) => arrays.map((array => array[index])));
};

Results:

console.log(zip(
    [1, 2, 3, 'a'],
    [667, false, -378, '337'],
    [111],
    [11, 221]
));

[ [ 1, 667, 111, 11 ] ]

console.log(zip(
    [1, 2, 3, 'a'],
    [667, false, -378, '337'],
    [111, 212, 323, 433, '1111']
));

[ [ 1, 667, 111 ], [ 2, false, 212 ], [ 3, -378, 323 ], [ ‘a’,
‘337’, 433 ] ]

console.log(zip(
    [1, 2, 3, 'a'],
    [667, false, -378, '337'],
    [111],
    []
));

[]

Implementation Python`s zip_longest on JS/ES6

(https://docs.python.org/3.5/library/itertools.html?highlight=zip_longest#itertools.zip_longest)

const zipLongest = (placeholder = undefined, ...arrays) => {
    const length = Math.max(...arrays.map(arr => arr.length));
    return Array.from(
        { length }, (value, index) => arrays.map(
            array => array.length - 1 >= index ? array[index] : placeholder
        )
    );
};

Results:

console.log(zipLongest(
    undefined,
    [1, 2, 3, 'a'],
    [667, false, -378, '337'],
    [111],
    []
));

[ [ 1, 667, 111, undefined ], [ 2, false, undefined, undefined ],
[ 3, -378, undefined, undefined ], [ ‘a’, ‘337’, undefined,
undefined ] ]

console.log(zipLongest(
    null,
    [1, 2, 3, 'a'],
    [667, false, -378, '337'],
    [111],
    []
));

[ [ 1, 667, 111, null ], [ 2, false, null, null ], [ 3, -378,
null, null ], [ ‘a’, ‘337’, null, null ] ]

console.log(zipLongest(
    'Is None',
    [1, 2, 3, 'a'],
    [667, false, -378, '337'],
    [111],
    []
));

[ [ 1, 667, 111, ‘Is None’ ], [ 2, false, ‘Is None’, ‘Is None’ ],
[ 3, -378, ‘Is None’, ‘Is None’ ], [ ‘a’, ‘337’, ‘Is None’, ‘Is
None’ ] ]

Solution 6:

Modern ES6 example with a generator:

function *zip (...iterables){
    let iterators = iterables.map(i => i[Symbol.iterator]() )
    while (true) {
        let results = iterators.map(iter => iter.next() )
        if (results.some(res => res.done) ) return
        else yield results.map(res => res.value )
    }
}

First, we get a list of iterables as iterators. This usually happens transparently, but here we do it explicitly, as we yield step-by-step until one of them is exhausted. We check if any of results (using the .some() method) in the given array is exhausted, and if so, we break the while loop.