Removing elements with Array.map in JavaScript

Inspired by writing this answer, I ended up later expanding and writing a blog post going over this in careful detail. I recommend checking that out if you want to develop a deeper understanding of how to think about this problem--I try to explain it piece by piece, and also give a JSperf comparison at the end, going over speed considerations.

That said, **The tl;dr is this:

To accomplish what you're asking for (filtering and mapping within one function call), you would use Array.reduce()**.

However, the more readable and (less importantly) usually significantly faster² approach is to just use filter and map chained together:

[1,2,3].filter(num => num > 2).map(num => num * 2)

What follows is a description of how Array.reduce() works, and how it can be used to accomplish filter and map in one iteration. Again, if this is too condensed, I highly recommend seeing the blog post linked above, which is a much more friendly intro with clear examples and progression.

---

You give reduce an argument that is a (usually anonymous) function.

That anonymous function takes two parameters--one (like the anonymous functions passed in to map/filter/forEach) is the iteratee to be operated on. There is another argument for the anonymous function passed to reduce, however, that those functions do not accept, and that is the value that will be passed along between function calls, often referred to as the memo.

Note that while Array.filter() takes only one argument (a function), Array.reduce() also takes an important (though optional) second argument: an initial value for 'memo' that will be passed into that anonymous function as its first argument, and subsequently can be mutated and passed along between function calls. (If it is not supplied, then 'memo' in the first anonymous function call will by default be the first iteratee, and the 'iteratee' argument will actually be the second value in the array)

In our case, we'll pass in an empty array to start, and then choose whether to inject our iteratee into our array or not based on our function--this is the filtering process.

Finally, we'll return our 'array in progress' on each anonymous function call, and reduce will take that return value and pass it as an argument (called memo) to its next function call.

This allows filter and map to happen in one iteration, cutting down our number of required iterations in half--just doing twice as much work each iteration, though, so nothing is really saved other than function calls, which are not so expensive in javascript.

For a more complete explanation, refer to MDN docs (or to my post referenced at the beginning of this answer).

Basic example of a Reduce call:

let array = [1,2,3];
const initialMemo = [];

array = array.reduce((memo, iteratee) => {
    // if condition is our filter
    if (iteratee > 1) {
        // what happens inside the filter is the map
        memo.push(iteratee * 2); 
    }

    // this return value will be passed in as the 'memo' argument
    // to the next call of this function, and this function will have
    // every element passed into it at some point.
    return memo; 
}, initialMemo)

console.log(array) // [4,6], equivalent to [(2 * 2), (3 * 2)]

more succinct version:

[1,2,3].reduce((memo, value) => value > 1 ? memo.concat(value * 2) : memo, [])

Notice that the first iteratee was not greater than one, and so was filtered. Also note the initialMemo, named just to make its existence clear and draw attention to it. Once again, it is passed in as 'memo' to the first anonymous function call, and then the returned value of the anonymous function is passed in as the 'memo' argument to the next function.

Another example of the classic use case for memo would be returning the smallest or largest number in an array. Example:

[7,4,1,99,57,2,1,100].reduce((memo, val) => memo > val ? memo : val)
// ^this would return the largest number in the list.

An example of how to write your own reduce function (this often helps understanding functions like these, I find):

test_arr = [];

// we accept an anonymous function, and an optional 'initial memo' value.
test_arr.my_reducer = function(reduceFunc, initialMemo) {
    // if we did not pass in a second argument, then our first memo value 
    // will be whatever is in index zero. (Otherwise, it will 
    // be that second argument.)
    const initialMemoIsIndexZero = arguments.length < 2;

    // here we use that logic to set the memo value accordingly.
    let memo = initialMemoIsIndexZero ? this[0] : initialMemo;

    // here we use that same boolean to decide whether the first
    // value we pass in as iteratee is either the first or second
    // element
    const initialIteratee = initialMemoIsIndexZero ? 1 : 0;

    for (var i = initialIteratee; i < this.length; i++) {
        // memo is either the argument passed in above, or the 
        // first item in the list. initialIteratee is either the
        // first item in the list, or the second item in the list.
           memo = reduceFunc(memo, this[i]);
        // or, more technically complete, give access to base array
        // and index to the reducer as well:
        // memo = reduceFunc(memo, this[i], i, this);
    }

    // after we've compressed the array into a single value,
    // we return it.
    return memo;
}

The real implementation allows access to things like the index, for example, but I hope this helps you get an uncomplicated feel for the gist of it.

You should use the filter method rather than map unless you want to mutate the items in the array, in addition to filtering.

eg.

var filteredItems = items.filter(function(item)
{
    return ...some condition...;
});

[Edit: Of course you could always do sourceArray.filter(...).map(...) to both filter and mutate]