Understanding Generators in JavaScript

This article was originally written for DigitalOcean.

In ECMAScript 2015, generators were introduced to the JavaScript language. A generator is a process that can be paused and resumed and can yield multiple values. A generator in JavaScript consists of a generator function, which returns an iterable Generator object.

Generators are a powerful addition to JavaScript. They can maintain state, providing an efficient way to make iterators, and are capable of dealing with infinite data streams, which can be used to implement infinite scroll on the frontend of a web application, to operate on sound wave data, and more. Additionally, when used with Promises, generators can mimic the async/await functionality, which allows us to deal with asynchronous code in a more straightforward and readable manner. Although async/await is a more prevalent way to deal with common, simple asynchronous use cases, like fetching data from an API, generators have more advanced features that make learning how to use them worthwhile.

In this article, we'll cover how to create generator functions, how to iterate over Generator objects, the difference between yield and return inside a generator, and other aspects of working with generators.

Generator Functions

A generator function is a function that returns a Generator object, and is defined by the function keyword followed by an asterisk (*), as shown in the following:

// Generator function declaration
function* generatorFunction() {}

Occasionally, you will see the asterisk next to the function name, as opposed to the function keyword, such as function *generatorFunction(). This works the same, but function* is a more widely accepted syntax.

Generator functions can also be defined in an expression, like regular functions:

// Generator function expression
const generatorFunction = function* () {}

Generators can even be the methods of an object or class:

// Generator as the method of an object
const generatorObj = {
  *generatorMethod() {},
}

// Generator as the method of a class
class GeneratorClass {
  *generatorMethod() {}
}

The examples throughout this article will use the generator function declaration syntax.

Note: Unlike regular functions, generators cannot be constructed with the new keyword, nor can they be used in conjunction with arrow functions.

Now that you know how to declare generator functions, lets look at the iterable Generator objects that they return.

Generator Objects

Traditionally, functions in JavaScript run to completion, and calling a function will return a value when it arrives at the return keyword. If the return keyword is omitted, a function will implicitly return undefined.

In the following code, for example, we declare a sum() function that returns a value that is the sum of two integer arguments:

// A regular function that sums two values
function sum(a, b) {
  return a + b
}

Calling the function returns a value that is the sum of the arguments:

const value = sum(5, 6) // 11

A generator function, however, does not return a value immediately, and instead returns an iterable Generator object. In the following example, we declare a function and give it a single return value, like a standard function:

// Declare a generator function with a single return value
function* generatorFunction() {
  return 'Hello, Generator!'
}

When we invoke the generator function, it will return the Generator object, which we can assign to a variable:

// Assign the Generator object to generator
const generator = generatorFunction()

If this were a regular function, we would expect generator to give us the string returned in the function. However, what we actually get is an object in a suspended state. Calling generator will therefore give output similar to the following:

generatorFunction {<suspended>}
  __proto__: Generator
  [[GeneratorLocation]]: VM272:1
  [[GeneratorStatus]]: "suspended"
  [[GeneratorFunction]]: ƒ* generatorFunction()
  [[GeneratorReceiver]]: Window
  [[Scopes]]: Scopes[3]

The Generator object returned by the function is an iterator. An iterator is an object that has a next() method available, which is used for iterating through a sequence of values. The next() method returns an object with value and done properties. value represent the returned value, and done indicates whether the iterator has run through all its values or not.

Knowing this, let's call next() on our generator and get the current value and state of the iterator:

// Call the next method on the Generator object
generator.next()

This will give the following output:

{value: "Hello, Generator!", done: true}

The value returned from calling next() is Hello, Generator!, and the state of done is true, because this value came from a return that closed out the iterator. Since the iterator is done, the generator function's status will change from suspended to closed. Calling generator again will give the following:

generatorFunction {<closed>}

As of right now, we've only demonstrated how a generator function can be a more complex way to get the return value of a function. But generator functions also have unique features that distinguish them from normal functions. In the next section, we'll learn about the yield operator and see how a generator can pause and resume execution.

yield Operators

Generators introduce a new keyword to JavaScript: yield. yield can pause a generator function and return the value that follows yield, providing a lightweight way to iterate through values.

In this example, we'll pause the generator function three times with different values, and return a value at the end. Then we will assign our Generator object to the generator variable.

// Create a generator function with multiple yields
function* generatorFunction() {
  yield 'Neo'
  yield 'Morpheus'
  yield 'Trinity'

  return 'The Oracle'
}

const generator = generatorFunction()

Now, when we call next() on the generator function, it will pause every time it encounters yield. done will be set to false after each yield, indicating that the generator has not finished. Once it encounters a return, or there are no more yields encountered in the function, done will flip to true, and the generator will be finished.

Use the next() method four times in a row:

// Call next four times
generator.next()
generator.next()
generator.next()
generator.next()

These will give the following four lines of output in order:

{value: "Neo", done: false}
{value: "Morpheus", done: false}
{value: "Trinity", done: false}
{value: "The Oracle", done: true}

Note that a generator does not require a return; if omitted, the last iteration will return {value: undefined, done: true}, as will any subsequent calls to next() after a generator has completed.

Iterating Over a Generator

Using the next() method, we manually iterated through the Generator object, receiving all the value and done properties of the full object. However, just like Array, Map, and Set, a Generator follows the iteration protocol, and can be iterated through with for...of:

// Iterate over Generator object
for (const value of generator) {
  console.log(value)
}

This will return the following:

Neo
Morpheus
Trinity

The spread operator can also be used to assign the values of a Generator to an array.

// Create an array from the values of a Generator object
const values = [...generator]

console.log(values)

This will give the following array:

(3) ["Neo", "Morpheus", "Trinity"]

Both spread and for...of will not factor the return into the values (in this case, it would have been 'The Oracle').

Note: While both of these methods are effective for working with finite generators, if a generator is dealing with an infinite data stream, it won't be possible to use spread or for...of directly without creating an infinite loop.

Closing a Generator

As we've seen, a generator can have its done property set to true and its status set to closed by iterating through all its values. There are two additional ways to immediately cancel a generator: with the return() method, and with the throw() method.

With return(), the generator can be terminated at any point, just as if a return statement had been in the function body. You can pass an argument into return(), or leave it blank for an undefined value.

To demonstrate return(), we'll create a generator with a few yield values but no return in the function definition:

function* generatorFunction() {
  yield 'Neo'
  yield 'Morpheus'
  yield 'Trinity'
}

const generator = generatorFunction()

The first next() will give us 'Neo', with done set to false. If we invoke a return() method on the Generator object right after that, we'll now get the passed value and done set to true. Any additional call to next() will give the default completed generator response with an undefined value.

To demonstrate this, run the following three methods on generator:

generator.next()
generator.return('There is no spoon!')
generator.next()

This will give the three following results:

{value: "Neo", done: false}
{value: "There is no spoon!", done: true}
{value: undefined, done: true}

The return() method forced the Generator object to complete and to ignore any other yield keywords. This is particularly useful in asynchronous programming when you need to make functions cancelable, such as interrupting a web request when a user wants to perform a different action, as it is not possible to directly cancel a Promise.

If the body of a generator function has a way to catch and deal with errors, you can use the throw() method to throw an error into the generator. This starts up the generator, throws the error in, and terminates the generator.

To demonstrate this, we will put a try...catch inside the generator function body and log an error if one is found:

// Define a generator function
function* generatorFunction() {
  try {
    yield 'Neo'
    yield 'Morpheus'
  } catch (error) {
    console.log(error)
  }
}

// Invoke the generator and throw an error
const generator = generatorFunction()

Now, we will run the next() method, followed by throw():

generator.next()
generator.throw(new Error('Agent Smith!'))

This will give the following output:

{value: "Neo", done: false}
Error: Agent Smith!
{value: undefined, done: true}

Using throw(), we injected an error into the generator, which was caught by the try...catch and logged to the console.

Generator Object Methods and States

The following table shows a list of methods that can be used on Generator objects:

Method Description
next() Returns the next value in a generator
return() Returns a value in a generator and finishes the generator
throw() Throws an error and finishes the generator

The next table lists the possible states of a Generator object:

Status Description
suspended Generator has halted execution but has not terminated
closed Generator has terminated by either encountering an error, returning, or iterating through all values

yield Delegation

In addition to the regular yield operator, generators can also use the yield* expression to delegate further values to another generator. When the yield* is encountered within a generator, it will go inside the delegated generator and begin iterating through all the yields until that generator is closed. This can be used to separate different generator functions to semantically organize your code, while still having all their yields be iterable in the right order.

To demonstrate, we can create two generator functions, one of which will yield* operate on the other:

// Generator function that will be delegated to
function* delegate() {
  yield 3
  yield 4
}

// Outer generator function
function* begin() {
  yield 1
  yield 2
  yield* delegate()
}

Next, let's iterate through the begin() generator function:

// Iterate through the outer generator
const generator = begin()

for (const value of generator) {
  console.log(value)
}

This will give the following values in the order they are generated:

1
2
3
4

The outer generator yielded the values 1 and 2, then delegated to the other generator with yield*, which returned 3 and 4.

yield* can also delegate to any object that is iterable, such as an Array or a Map. Yield delegation can be helpful in organizing code, since any function within a generator that wanted to use yield would also have to be a generator.

Infinite Data Streams

One of the useful aspects of generators is the ability to work with infinite data streams and collections. This can be demonstrated by creating an infinite loop inside a generator function that increments a number by one.

In the following code block, we define this generator function and then initiate the generator:

// Define a generator function that increments by one
function* incrementer() {
  let i = 0

  while (true) {
    yield i++
  }
}

// Initiate the generator
const counter = incrementer()

Now, iterate through the values using next():

// Iterate through the values
counter.next()
counter.next()
counter.next()
counter.next()

This will give the following output:

{value: 0, done: false}
{value: 1, done: false}
{value: 2, done: false}
{value: 3, done: false}

The function returns successive values in the infinite loop while the done property remains false, ensuring that it will not finish.

With generators, you don't have to worry about creating an infinite loop, because you can halt and resume execution at will. However, you still have to have caution with how you invoke the generator. If you use spread or for...of on an infinite data stream, you will still be iterating over an infinite loop all at once, which will cause the environment to crash.

For a more complex example of an infinite data stream, we can create a Fibonacci generator function. The Fibonacci sequence, which continuously adds the two previous values together, can be written using an infinite loop within a generator as follows:

// Create a fibonacci generator function
function* fibonacci() {
  let prev = 0
  let next = 1

  yield prev
  yield next

  // Add previous and next values and yield them forever
  while (true) {
    const newVal = next + prev

    yield newVal

    prev = next
    next = newVal
  }
}

To test this out, we can loop through a finite number and print the Fibonacci sequence to the console.

// Print the first 10 values of fibonacci
const fib = fibonacci()

for (let i = 0; i < 10; i++) {
  console.log(fib.next().value)
}

This will give the following:

0
1
1
2
3
5
8
13
21
34

The ability to work with infinite data sets is one part of what makes generators so powerful. This can be useful for examples like implementing infinite scroll on the frontend of a web application, or operating on sound wave data.

Passing Values in Generators

Throughout this article, we've used generators as iterators, and we've yielded values in each iteration. In addition to producing values, generators can also consume values from next(). In this case, yield will contain a value.

It's important to note that the first next() that is called will not pass a value, but will only start the generator. To demonstrate this, we can log the value of yield and call next() a few times with some values.

function* generatorFunction() {
  console.log(yield)
  console.log(yield)

  return 'The end'
}

const generator = generatorFunction()

generator.next()
generator.next(100)
generator.next(200)

This will give the following output:

100
200
{value: "The end", done: true}

It is also possible to seed the generator with an initial value. In the following example, we'll make a for loop and pass each value into the next() method, but pass an argument to the inital function as well:

function* generatorFunction(value) {
  while (true) {
    value = yield value * 10
  }
}

// Initiate a generator and seed it with an initial value
const generator = generatorFunction(0)

for (let i = 0; i < 5; i++) {
  console.log(generator.next(i).value)
}

We'll retrieve the value from next() and yield a new value to the next iteration, which is the previous value times ten. This will give the following:

0
10
20
30
40

Another way to deal with starting up a generator is to wrap the generator in a function that will always call next() once before doing anything else.

async/await with Generators

An asynchronous function is a type of function available in ES6+ JavaScript that makes working with asynchronous data simpler and easier to understand by making it appear synchronous. Generators have a more extensive array of capabilities than asynchronous functions, but are capable of replicating similar behavior. Implementing asynchronous programming in this way can increase the flexibility of your code.

In this section, we will demonstrate an example of reproducing async/await with generators.

Let's build an asynchronous function that uses the Fetch API to get data from the JSONPlaceholder API (which provides example JSON data for testing purposes) and logs the response to the console.

Start out by defining an asynchronous function called getUsers that fetches data from the API and returns an array of objects, then call getUsers:

const getUsers = async function () {
  const response = await fetch('https://jsonplaceholder.typicode.com/users')
  const json = await response.json()

  return json
}

// Call the getUsers function and log the response
getUsers().then((response) => console.log(response))

This will give the following JSON data:

[ {id: 1, name: "Leanne Graham" ...},
  {id: 2, name: "Ervin Howell" ...},
  {id: 3, name": "Clementine Bauch" ...},
  {id: 4, name: "Patricia Lebsack"...},
  {id: 5, name: "Chelsey Dietrich"...},
  {id: 6, name: "Mrs. Dennis Schulist"...},
  {id: 7, name: "Kurtis Weissnat"...},
  {id: 8, name: "Nicholas Runolfsdottir V"...},
  {id: 9, name: "Glenna Reichert"...},
  {id: 10, name: "Clementina DuBuque"...}]

Using generators, we can create something almost identical that does not use the async/await keywords. Instead, it will use a new function we create, and yield values instead of await promises.

In the following code block, we define a function called getUsers that uses our new asyncAlt function (which we will write later on) to mimic async/await.

const getUsers = asyncAlt(function* () {
  const response = yield fetch('https://jsonplaceholder.typicode.com/users')
  const json = yield response.json()

  return json
})

// Invoking the function
getUsers().then((response) => console.log(response))

As we can see, it looks almost identical to the async/await implementation, except that there is a generator function being passed in that yields values.

Now we can create an asyncAlt function that resembles an asynchronous function. asyncAlt has a generator function as a parameter, which is our function that yields the promises that fetch returns. asyncAlt returns a function itself, and resolves every promise it finds until the last one:

// Define a function named asyncAlt that takes a generator function as an argument
function asyncAlt(generatorFunction) {
  // Return a function
  return function () {
    // Create and assign the generator object
    const generator = generatorFunction()

    // Define a function that accepts the next iteration of the generator
    function resolve(next) {
      // If the generator is closed and there are no more values to yield,
      // resolve the last value
      if (next.done) {
        return Promise.resolve(next.value)
      }

      // If there are still values to yield, they are promises and
      // must be resolved.
      return Promise.resolve(next.value).then((response) => {
        return resolve(generator.next(response))
      })
    }

    // Begin resolving promises
    return resolve(generator.next())
  }
}

This will give the same output as the async/await version:

[ {id: 1, name: "Leanne Graham" ...},
  {id: 2, name: "Ervin Howell" ...},
  {id: 3, name": "Clementine Bauch" ...},
  {id: 4, name: "Patricia Lebsack"...},
  {id: 5, name: "Chelsey Dietrich"...},
  {id: 6, name: "Mrs. Dennis Schulist"...},
  {id: 7, name: "Kurtis Weissnat"...},
  {id: 8, name: "Nicholas Runolfsdottir V"...},
  {id: 9, name: "Glenna Reichert"...},
  {id: 10, name: "Clementina DuBuque"...}]

Note that this implementation is for demonstrating how generators can be used in place of async/await, and is not a production-ready design. It does not have error handling set up, nor does it have the ability to pass parameters into the yielded values. Though this method can add flexibility to your code, often async/await will be a better choice, since it abstracts implementation details away and lets you focus on writing productive code.

Conclusion

Generators are processes that can halt and resume execution. They are a powerful, versatile feature of JavaScript, although they are not commonly used. In this tutorial, we learned about generator functions and generator objects, methods available to generators, the yield and yield* operators, and using generators with finite and infinite data sets. We also explored one way to implement asynchronous code without nested callbacks or long promise chains.

If you would like to learn more about JavaScript syntax, take a look at our Understanding This, Bind, Call, and Apply in JavaScript and Understanding Map and Set Objects in JavaScript tutorials.

Comments