How to Use Map, Filter, & Reduce in JavaScript

Functional programming has been making quite a splash in the development world these days. And for good reason: Functional techniques can help you write more declarative code that is easier to understand at a glance, refactor, and test. 

One of the cornerstones of functional programming is its special use of lists and list operations. And those things are exactly what the sound like they are: arrays of things, and the stuff you do to them. But the functional mindset treats them a bit differently than you might expect.

This article will take a close look at what I like to call the "big three" list operations: map, filter, and reduce. Wrapping your head around these three functions is an important step towards being able to write clean functional code, and opens the doors to the vastly powerful techniques of functional and reactive programming.

It also means you'll never have to write a for loop again.

Curious? Let's dive in.

A Map From List to List

Often, we find ourselves needing to take an array and modify every element in it in exactly the same way. Typical examples of this are squaring every element in an array of numbers, retrieving the name from a list of users, or running a regex against an array of strings.

map is a method built to do exactly that. It's defined on Array.prototype, so you can call it on any array, and it accepts a callback as its first argument. 

When you call map on an array, it executes that callback on every element within it, returning a new array with all of the values that the callback returned.

Under the hood, map passes three arguments to your callback:

  1. The current item in the array
  2. The array index of the current item
  3. The entire array you called map on 

Let's look at some code.

map in Practice

Suppose we have an app that maintains an array of your tasks for the day. Each task is an object, each with a name and duration property:

Let's say we want to create a new array with just the name of each task, so we can take a look at everything we've gotten done today. Using a for loop, we'd write something like this:

JavaScript also offers a forEach loop. It functions like a for loop, but manages all the messiness of checking our loop index against the array length for us:

Using map, we can write:

I include the index and  array parameters to remind you that they're there if you need them. Since I didn't use them here, though, you could leave them out, and the code would run just fine.

There are a few important differences between the two approaches:

  1. Using map, you don't have to manage the state of the for loop yourself.
  2. You can operate on the element directly, rather than having to index into the array.
  3. You don't have to create a new array and push into it. map returns the finished product all in one go, so we can simply assign the return value to a new variable.
  4. You do have to remember to include a return statement in your callback. If you don't, you'll get a new array filled with undefined

Turns out, all of the functions we'll look at today share these characteristics.

The fact that we don't have to manually manage the state of the loop makes our code simpler and more maintainable. The fact that we can operate directly on the element instead of having to index into the array makes things more readable. 

Using a forEach loop solves both of these problems for us. But map still has at least two distinct advantages:

  1. forEach returns undefined, so it doesn't chain with other array methods. map returns an array, so you can chain it with other array methods.
  2. map returns an array with the finished product, rather than requiring us to mutate an array inside the loop. 

Keeping the number of places where you modify state to an absolute minimum is an important tenet of functional programming. It makes for safer and more intelligible code.

Now is also a good time to point out that if you're in Node, testing these examples in the Firefox browser console, or using Babel or Traceur, you can write this more concisely with ES6 arrow functions:

Arrow functions let us leave out the return keyword in one-liners. 

It doesn't get much more readable than that.

Gotchas

The callback you pass to map must have an explicit return statement, or map will spit out an array full of undefined. It's not hard to remember to include a return value, but it's not hard to forget. 

If you do forget, map won't complain. Instead, it'll quietly hand back an array full of nothing. Silent errors like that can be surprisingly hard to debug. 

Fortunately, this is the only gotcha with map. But it's a common enough pitfall that I'm obliged to emphasize: Always make sure your callback contains a return statement!

Implementation

Reading implementations is an important part of understanding. So, let's write our own lightweight map to better understand what's going on under the hood. If you want to see a production-quality implementation, check out Mozilla's polyfill at MDN.

This code accepts an array and a callback function as arguments. It then creates a new array; executes the callback on each element on the array we passed in; pushes the results into the new array; and returns the new array. If you run this in your console, you'll get the same result as before. Just make sure you initialize tasks before you test it out!

While we're using a for loop under the hood, wrapping it up into a function hides the details and lets us work with the abstraction instead. 

That makes our code more declarative—it says what to do, not how to do it. You'll appreciate how much more readable, maintainable, and, erm, debuggable this can make your code.

Filter Out the Noise

The next of our array operations is filter. It does exactly what it sounds like: It takes an array, and filters out unwanted elements.

Like map, filter is defined on Array.prototype. It's available on any array, and you pass it a callback as its first argument. filter executes that callback on each element of the array, and spits out a new array containing only the elements for which the callback returned true.

Also like map, filter passes your callback three arguments:

  1. The current item 
  2. The current index
  3. The array you called filter on

filter in Practice

Let's revisit our task example. Instead of pulling out the names of each task, let's say I want to get a list of just the tasks that took me two hours or more to get done. 

Using forEach, we'd write:

With filter:

Here, I've gone ahead and left out the index and array arguments to our callback, since we don't use them.

Just like map, filter lets us:

  • avoid mutating an array inside a forEach or for loop
  • assign its result directly to a new variable, rather than push into an array we defined elsewhere

Gotchas

The callback you pass to map has to include a return statement if you want it to function properly. With filter, you also have to include a return statement, and you must make sure it returns a boolean value.

If you forget your return statement, your callback will return undefined, which filter will unhelpfully coerce to false. Instead of throwing an error, it will silently return an empty array! 

If you go the other route, and return something that's isn't explicitly true or false, then filter will try to figure out what you meant by applying JavaScript's coercion rules. More often than not, this is a bug. And, just like forgetting your return statement, it'll be a silent one. 

Always make sure your callbacks include an explicit return statement. And always make sure your callbacks in filter return true or false. Your sanity will thank you.

Implementation

Once again, the best way to understand a piece of code is... well, to write it. Let's roll our own lightweight filter. The good folks at Mozilla have an industrial-strength polyfill for you to read, too.

Reducing Arrays

map creates a new array by transforming every element in an array, individually. filter creates a new array by removing elements that don't belong. reduce, on the other hand, takes all of the elements in an array, and reduces them into a single value.

Just like map and filterreduce is defined on Array.prototype and so available on any array, and you pass a callback as its first argument. But it also takes an optional second argument: the value to start combining all your array elements into. 

reduce passes your callback four arguments:

  1. The current value
  2. The previous value 
  3. The current index
  4. The array you called reduce on

Notice that the callback gets a previous value on each iteration. On the first iteration, there is no previous value. This is why you have the option to pass reduce an initial value: It acts as the "previous value" for the first iteration, when there otherwise wouldn't be one.

Finally, bear in mind that reduce returns a single value, not an array containing a single item. This is more important than it might seem, and I'll come back to it in the examples.

reduce in Practice

Since reduce is the function that people find most alien at first, we'll start by walking step by step through something simple.

Let's say we want to find the sum of a list of numbers. Using a loop, that looks like this:

While this isn't a bad use case for forEachreduce still has the advantage of allowing us to avoid mutation. With reduce, we would write:

First, we call reduce on our list of numbers. We pass it a callback, which accepts the previous value and current value as arguments, and returns the result of adding them together.  Since we passed 0 as a second argument to reduce, it'll use that as the value of previous on the first iteration.

If we take it step by step, it looks like this:

Iteration  Previous Current Total
1 0 1 1
2 1 2 3
3 3 3 6
4 6 4 10
5 10 5 15

If you're not a fan of tables, run this snippet in the console:

To recap: reduce iterates over all the elements of an array, combining them however you specify in your callback. On every iteration, your callback has access to the previous value, which is the total-so-far, or accumulated value; the current value; the current index; and the entire array, if you need them.

Let's turn back to our tasks example. We've gotten a list of task names from map, and a filtered list of tasks that took a long time with... well, filter

What if we wanted to know the total amount of time we spent working today?

Using a forEach loop, you'd write:

With reduce, that becomes:

Easy. 

That's almost all there is to it. Almost, because JavaScript provides us with one more little-known method, called reduceRight. In the examples above, reduce started at the first item in the array, iterating from left to right:

reduceRight does the same thing, but in the opposite direction:

I use reduce every day, but I've never needed reduceRight. I reckon you probably won't, either. But in the event you ever do, now you know it's there.

Gotchas

The three big gotchas with reduce are:

  1. Forgetting to return
  2. Forgetting an initial value
  3. Expecting an array when reduce returns a single value

Fortunately, the first two are easy to avoid. Deciding what your initial value should be depends on what you're doing, but you'll get the hang of it quickly.

The last one might seem a bit strange. If reduce only ever returns a single value, why would you expect an array?

There are a few good reasons for that. First, reduce always returns a single value, not always a single number. If you reduce an array of arrays, for instance, it will return a single array. If you're in the habit or reducing arrays, it would be fair to expect that an array containing a single item wouldn't be a special case.

Second, if reduce did return an array with a single value, it would naturally play nice with map and filter, and other functions on arrays that you're likely to be using with it. 

Implementation

Time for our last look under the hood. As usual, Mozilla has a bulletproof polyfill for reduce if you want to check it out.

Two things to note, here:

  1. This time, I used the name accumulator instead of previous. This is what you'll usually see in the wild.
  2. I assign accumulator an initial value, if a user provides one, and default to 0, if not. This is how the real reduce behaves, as well.

Putting It Together: Map, Filter, Reduce, and Chainability

At this point, you might not be that impressed. 

Fair enough: map, filter, and reduce, on their own, aren't awfully interesting. 

After all, their true power lies in their chainability. 

Let's say I want to do the following:

  1. Collect two days' worth of tasks.
  2. Convert the task durations to hours, instead of minutes.
  3. Filter out everything that took two hours or more.
  4. Sum it all up.
  5. Multiply the result by a per-hour rate for billing.
  6. Output a formatted dollar amount.

First, let's define our tasks for Monday and Tuesday:

And now, our lovely-looking transformation:

Or, more concisely:

If you've made it this far, this should be pretty straightforward. There are two bits of weirdness to explain, though. 

First, on line 10, I have to write:

Two things to explain here:

  1. The plus signs in front of accumulator and current coerce their values to numbers. If you don't do this, the return value will be the rather useless string, "12510075100".
  2. If don't wrap that sum in brackets, reduce will spit out a single value, not an array. That would end up throwing a TypeError, because you can only use map on an array! 

The second bit that might make you a bit uncomfortable is the last reduce, namely:

That call to map returns an array containing a single value. Here, we call reduce to pull out that value.

The other way to do this would be to remove the call to reduce, and index into the array that map spits out:

That's perfectly correct. If you're more comfortable using an array index, go right ahead.

But I encourage you not to. One of the most powerful ways to use these functions is in the realm of reactive programming, where you won't be free to use array indices. Kicking that habit now will make learning reactive techniques much easier down the line.

Finally, let's see how our friend the forEach loop would get it done:

Tolerable, but noisy.

Conclusion & Next Steps

In this tutorial, you've learned how mapfilter, and reduce work; how to use them; and roughly how they're implemented. You've seen that they all allow you to avoid mutating state, which using for and forEach loops requires, and you should now have a good idea of how to chain them all together. 

By now, I'm sure you're eager for practice and further reading. Here are my top three suggestions for where to head next:

  1. Jafar Husain's superb set of exercises on Functional Programming in JavaScript, complete with a solid introduction to Rx.js
  2. Envato Tuts+ Instructor Jason Rhodes' course on Functional Programming in JavaScript
  3. The Mostly Adequate Guide to Functional Programming, which goes into greater depth on why we avoid mutation, and functional thinking in general

JavaScript has become one of the de-facto languages of working on the web. It’s not without its learning curves, and there are plenty of frameworks and libraries to keep you busy, as well. If you’re looking for additional resources to study or to use in your work, check out what we have available in the Envato marketplace.

If you want more on this sort of thing, check my profile from time to time; catch me on Twitter (@PelekeS); or hit my blog at http://peleke.me.

Questions, comments, or confusions? Leave them below, and I'll do my best to get back to each one individually.

Tags:

Comments

Related Articles