How do I tell if an object is a Promise?

How a promise library decides

If it has a .then function - that's the only standard promise libraries use.

The Promises/A+ specification has a notion called thenable which is basically "an object with a then method". Promises will and should assimilate anything with a then method. All of the promise implementation you've mentioned do this.

If we look at the specification:

2.3.3.3 if then is a function, call it with x as this, first argument resolvePromise, and second argument rejectPromise

It also explains the rationale for this design decision:

This treatment of thenables allows promise implementations to interoperate, as long as they expose a Promises/A+-compliant then method. It also allows Promises/A+ implementations to “assimilate” nonconformant implementations with reasonable then methods.

How you should decide

You shouldn't - instead call Promise.resolve(x) (Q(x) in Q) that will always convert any value or external thenable into a trusted promise. It is safer and easier than performing these checks yourself.

really need to be sure?

You can always run it through the test suite :D

Checking if something is promise unnecessarily complicates the code, just use Promise.resolve

Promise.resolve(valueOrPromiseItDoesntMatter).then(function(value) {

})

Here's my original answer, which has since been ratified in the spec as the way to test for a promise:

Promise.resolve(obj) == obj

This works because the algorithm explicitly demands that Promise.resolve must return the exact object passed in if and only if it is a promise by the definition of the spec.

I have another answer here, which used to say this, but I changed it to something else when it didn't work with Safari at that time. That was a year ago, and this now works reliably even in Safari.

I would have edited my original answer, except that felt wrong, given that more people by now have voted for the altered solution in that answer than the original. I believe this is the better answer, and I hope you agree.

To see if the given object is a ES6 Promise, we can make use of this predicate:

function isPromise(p) {
  return p && Object.prototype.toString.call(p) === "[object Promise]";
}

Calling toString directly from the Object.prototype returns a native string representation of the given object type which is "[object Promise]" in our case. This ensures that the given object

• Bypasses false positives such as..:
  • Self-defined object type with the same constructor name ("Promise").
  • Self-written toString method of the given object.
• Works across multiple environment contexts (e.g. iframes) in contrast to instanceof or isPrototypeOf.

However, any particular host object, that has its tag modified via Symbol.toStringTag, can return "[object Promise]". This may be the intended result or not depending on the project (e.g. if there is a custom Promise implementation).

To see if the object is from a native ES6 Promise, we can use:

function isNativePromise(p) {
  return p && typeof p.constructor === "function"
    && Function.prototype.toString.call(p.constructor).replace(/\(.*\)/, "()")
    === Function.prototype.toString.call(/*native object*/Function)
      .replace("Function", "Promise") // replacing Identifier
      .replace(/\(.*\)/, "()"); // removing possible FormalParameterList 
}

According to this and this section of the spec, the string representation of function should be:

"function Identifier ( FormalParameterList_opt ) { FunctionBody }"

which is handled accordingly above. The FunctionBody is [native code] in all major browsers.

MDN: Function.prototype.toString

This works across multiple environment contexts as well.

Not an answer to the full question but I think it's worth to mention that in Node.js 10 a new util function called isPromise was added which checks if an object is a native Promise or not:

const utilTypes = require('util').types
const b_Promise = require('bluebird')

utilTypes.isPromise(Promise.resolve(5)) // true
utilTypes.isPromise(b_Promise.resolve(5)) // false

This is how graphql-js package detects promises:

function isPromise(value) {
  return Boolean(value && typeof value.then === 'function');
}

value is the returned value of your function. I'm using this code in my project and have no problem so far.

Here is the code form https://github.com/ssnau/xkit/blob/master/util/is-promise.js

!!obj && (typeof obj === 'object' || typeof obj === 'function') && typeof obj.then === 'function';

if an object with a then method, it should be treat as a Promise.

If you are in an async method you can do this and avoid any ambiguity.

async myMethod(promiseOrNot){
  const theValue = await promiseOrNot()
}

If the function returns promise, it will await and return with the resolved value. If the function returns a value, it will be treated as resolved.

If the function does not return a promise today, but tomorrow returns one or is declared async, you will be future-proof.

it('should return a promise', function() {
    var result = testedFunctionThatReturnsPromise();
    expect(result).toBeDefined();
    // 3 slightly different ways of verifying a promise
    expect(typeof result.then).toBe('function');
    expect(result instanceof Promise).toBe(true);
    expect(result).toBe(Promise.resolve(result));
});

after searching for a reliable way to detect Async functions or even Promises, i ended up using the following test :

() => fn.constructor.name === 'Promise' || fn.constructor.name === 'AsyncFunction'

ES6:

const promise = new Promise(resolve => resolve('olá'));

console.log(promise.toString().includes('Promise')); //true

Update: This is no longer the best answer. Please vote up my other answer instead.

obj instanceof Promise

should do it. Note that this may only work reliably with native es6 promises.

If you're using a shim, a promise library or anything else pretending to be promise-like, then it may be more appropriate to test for a "thenable" (anything with a .then method), as shown in other answers here.

if (typeof thing.then === 'function') {
    // probably a promise
} else {
    // definitely not a promise
}

In case you are using Typescript, I'd like to add that you can use the "type predicate" feature. Just should wrap the logical verification in a function that returns x is Promise<any> and you won't need to do typecasts. Below on my example, c is either a promise or one of my types which I want to convert into a promise by calling the c.fetch() method.

export function toPromise(c: Container<any> | Promise<any>): Promise<any> {
    if (c == null) return Promise.resolve();
    return isContainer(c) ? c.fetch() : c;
}

export function isContainer(val: Container<any> | Promise<any>): val is Container<any> {
    return val && (<Container<any>>val).fetch !== undefined;
}

export function isPromise(val: Container<any> | Promise<any>): val is Promise<any> {
    return val && (<Promise<any>>val).then !== undefined;
}

More info: https://www.typescriptlang.org/docs/handbook/advanced-types.html

I use this function as a universal solution:

function isPromise(value) {
  return value && value.then && typeof value.then === 'function';
}