spy4js

Benefits

• TypeScript support included
• Performance
• No foreign dependencies
• Optimized error messages
• Customizable
• Intuitive
• Used in production of large projects

Introduction

spy4js provides a stand-alone spy framework that can be integrated well with the frameworks Vitest and Jest.

⚠️ Disclaimer: You don't need this library as the frameworks already include everything in order to achieve everything without this library.

spy4js exports only one object called Spy. The spy instances come with a lot of useful features. See below for more.

Hint: If you are using other frameworks you might not be able to benefit from all features of this library. Since the test frameworks already include excellent spies themselves, you might ask yourself, why spy4js...

Because it can make your tests more readable.

Advantages over Jest/Vitest spies:

• Very important for tests is their readability. This spy API is much easier to learn, and the tests can be understood even without any previous knowledge.
• Error messages should be extremely helpful, because development time is very valuable. In error cases made comparisons will be printed with detailed information.
• The used serialization for objects can be directly copied into your test code, which increases your speed while writing tests.
• Last but not least there are several nice features Jest doesn't provide out-of-the-box, and you could even combine both spy sorts.

Installation

With yarn

1yarn add --dev spy4js

With npm

1npm install --save-dev spy4js

Interface

A spy instance can be initialized differently.

1import { Spy } from 'spy4js';
2
3// initialize directly
4const spy1 = Spy();
5
6// initialize directly and supply an identifier for debugging purpose (default: 'the spy')
7const spy2 = Spy('special spy for me');
8
9// initialize by mocking another objects attribute (usually this attribute is a function)
10const someObject1 = new Date(2017, 1, 15);
11const spy3 = Spy.on(someObject1, 'toJSON');
12// (spy name will be accordingly: "the spy on 'toJSON'")
13
14// initialize many by mocking another objects attributes
15const someObject2 = new Date(2017, 1, 15);
16const someObject2$Mock = Spy.mock(someObject2, 'toJSON', 'toString', 'getDate');
17
18// mock exported functions from other modules
19// if your test runner executes with CJS
20const myModuleMocksInJest = Spy.mock(require('./my-module'), 'useMe');
21const myModuleMocksInJestTypeSafe = Spy.mock(require('./my-module') as typeof import('./my-module'), 'useMe');
22
23// if your test runner supports ESM
24vi.mock('./my-module', async () => ({ ...((await vi.importActual('./my-module')) as any) }));
25const myModuleMocksInVitest = Spy.mock(await import('./my-module'), 'useMe');
26
27// mock React components from other modules
28// if your test runner executes with CJS
29const mockedReactComponentsInJest = Spy.mockReactComponents(require('./my/fancy/Calculator'), 'Calculator');
30const mockedReactComponentsInJestTypeSafe = Spy.mockReactComponents(
31  require('./my/fancy/Calculator') as typeof import('./my/fancy/Calculator'), 
32  'Calculator'
33);
34
35// if your test runner supports ESM
36vi.mock('./my/fancy/Calculator', async () => ({ ...((await vi.importActual('./my/fancy/Calculator')) as any) }));
37const mockedReactComponentsInVitest = Spy.mockReactComponents(await import('./my/fancy/Calculator'), 'Calculator');

You may apply additional behavior to every spy. The valid operations here are:

• configure (allows to override some behavior of the spy)
• calls (does make the spy call the provided functions sequentially)
• returns (does make the spy return the provided params sequentially)
• throws (does make the spy throw an error when called)
• resolves (does make the spy resolve the provided params sequentially) #Promise
• rejects (does make the spy reject an error when called) #Promise
• transparent (does make the spy call the original method of a mocked object)
• transparentAfter (does make the spy call the original method of a mocked object after a certain amount of made calls)
• reset (resets the registered calls which were already made)
• restore (does make the spy restore the mocked object)
• addSnapshotSerializer (defines in jest snapshots how the spy will be serialized)

All those methods on a spy has been designed in a builder pattern. So you may chain any of these configurations. Be aware some behaviors override existing behaviors.

1const spy = Spy.on(someObject, 'someMethod');
2
3// configure it to use NOT own "equals" implementations
4spy.configure({ useOwnEquals: false });
5
6// make it call any functions
7spy.calls(func1, func2, func3);
8someObject.someMethod(arg); // returns func1(arg)
9someObject.someMethod(arg1, arg2); // returns func2(arg1, arg2)
10someObject.someMethod(arg); // returns func3(arg)
11someObject.someMethod(arg1, arg2, arg3); // returns func3(arg1, arg2, arg3) // sticks to the last
12
13// make it return any values
14spy.returns(value1, value2);
15someObject.someMethod(arg); // returns value1
16someObject.someMethod(arg1, arg2); // returns value2
17someObject.someMethod(arg); // returns value2 // sticks to the last
18
19// make it throw any message (the message is optional)
20spy.throws('throw this');
21someObject.someMethod(arg); // throws new Error('throw this')
22
23// make it return always the current date and transparentAfter 2 calls
24spy.calls(() => new Date()).transparentAfter(2);
25someObject.someMethod(arg); // returns new Date()
26someObject.someMethod(arg1, arg2); // returns new(er) Date()
27someObject.someMethod(arg); // returns someObject.someMethod(arg) // sticks to this behavior
28
29// make it immediatly transparent
30spy.transparent();
31
32// make it reset
33spy.reset();
34
35// make it restore
36spy.restore(); // other than "transparent" does not control input and output of the mocked function anymore

Even as important are the "facts", we want to display:

• wasCalled (does display that the spy has been called a specifiable amount of times)
• wasNotCalled (does display that the spy has never been called)
• wasCalledWith (does display that the spy has been called at least once like with the provided params)
• wasNotCalledWith (does display that the spy was never like with the provided params)
• hasCallHistory (does display that the spy has been called with the following params in the given order)
• hasProps (does display that the spy has been called with the given argument on the last invocation. In the context of React: The spy being a mocked React component has currently the given props.)

Those methods on a spy display facts. Facts have to be true, otherwise they will throw an Exception, which displays in a formatted debug message why the given fact was a lie. By writing those facts in your tests, a big refactoring loses its scare.

1const spy = Spy();
2
3spy.wasNotCalled();
4
5spy([1, 'test', {attr: [4]}]);
6
7spy.wasCalled();  // called at least once
8spy.wasCalled(1); // called exactly once
9
10spy('with this text');
11
12spy.wasCalled(2); // called exactly 2 times
13
14// the spy was called at least once with equal params
15spy.wasCalledWith([1, 'test', {attr: [4]}]);
16
17// the spy was not called with those params
18spy.wasNotCalledWith([1, 'test', {attr: [3]}]);
19
20// the spy was called twice with the following params and in same order
21spy.hasCallHistory([ [1, 'test', {attr: [4]}] ], 'with this text');

Sometimes it is necessary to have access to some call arguments with which the spy had been called.

• getAllCallArguments (returns all call arguments for all calls in an array containing arrays)
• getCallArguments (returns all call arguments for a specified call in an array)
• getCallArgument (same as getCallArguments, but returns only a single element of the array)
• getLatestCallArgument (same as getCallArgument, but for the latest call)
• getProps (same as getLatestCallArgument, but only for the first param. Can be useful for mocked React components)
• getCallCount (returns the number of made calls)

1const spy = Spy();
2
3// make some calls
4spy('string', 1);
5spy([1, 2, 3]);
6spy();
7spy(null);
8
9spy.getAllCallArguments();     // returns [['string', 1], [[1, 2, 3]], [], [null]]
10spy.getCallCount();     // returns 4
11spy.getCallArguments(/* default = 0 */);     // returns ['string', 1]
12spy.getCallArgument(/* defaults = (0, 0) */); // returns 'string'
13spy.getCallArgument(0, 1); // returns 1
14
15spy.getCallArguments(1);                     // returns [[1, 2, 3]]
16spy.getCallArgument(1);                 // returns [1, 2, 3]
17
18spy.getCallArguments(2);                     // returns []
19spy.getCallArgument(2);                 // returns undefined
20
21spy.getCallArguments(3);                     // returns [null]
22spy.getCallArgument(3);                 // returns null
23
24spy.getCallArguments(4);                     // throws Exception because less calls were made
25spy.getCallArgument(4);                 // throws same Exception

The last method is showCallArguments. It is mostly used internally to improve the debugging messages, but can be while you are in a console.log-mania.

Method-Details

Constructor

1Spy(spyName:string = 'the spy') => SpyInstance

The returned Spy instance has his own name-attribute (only) for debugging purpose.

setup (static)

1Spy.setup(config: {
2    useOwnEquals?: boolean;
3    enforceOrder?: boolean;
4    useGenericReactMocks?: boolean;
5    afterEachCb?: () => void;
6    afterEach?: (cb: () => void) => void;
7    beforeEach?: (cb: () => void) => void;
8    expect?: { addSnapshotSerializer: (serializer: any) => void; getState: () => { currentTestName?: string } };
9    runner?: 'jest' | 'vitest' | 'other';
10}) => void

This function should be ideally called in some setup-test file, but is required to be called to get all benefits of spy4js.

• afterEach: The test runners afterEach hook (default: global.afterEach)
• beforeEach: The test runners beforeEach hook (default: global.beforeEach)
• runner: The test runner name (default: determines if "jest" or "vitest")

For the other options see below.

configure (static)

1Spy.configure(config: {
2    useOwnEquals?: boolean;
3    enforceOrder?: boolean;
4    useGenericReactMocks?: boolean;
5    afterEachCb?: () => void;
6}) => void

Using this function you may edit the default behavior spy4js itself.

• useOwnEquals: Applies for all spy instances. See configure for more details.
• enforceOrder: Opt-in to the enforce-order mode.
• useGenericReactMocks: Lets you opt in into using generic react components for mocks created via mockReactComponents.
• afterEachCb: Lets you override the default afterEach fuctionality.

on (static)

1Spy.on(object: object, methodName: string) => SpyInstance

Initializing a spy on an object, simply replaces the original function by a spy and stores the necessary information to be able to restore the mocked method.

If the attribute has already been spied or is not a function, the Spy will throw an exception to avoid unexpected behavior. You never want to spy other attributes than functions and for no purpose a spy should ever be spied.

mock (static)

1Spy.mock(object: object, ...methodNames: string[]) => Object (Mock)

Creating an object that references spies for all given methodNames. Initialize as many spies as required for the same object. Only after Spy.initMocks gets called, the created mock does affect the given object.

mockReactComponents (static)

1Spy.mockReactComponents(object: object, ...methodNames: string[]) => Object (Mock)

Same as mock but designed for ReactJS components. The registered spies return null instead of undefined. This makes minimal usable React components. Even if in most cases the pure mocking is nice enough, you can even test the number of rerender cycles and the provided props of the mocked component. Works perfectly with enzyme and @testing-library/react.

initMocks (static)

1Spy.initMocks(scope?: string) => void

Does initialize all global and scope-related mocks by applying spies. Mocks can be created with mock or mockReactComponents. This function has not to be called manually, if you rely on the default test suite hooks.

restoreAll (static)

1Spy.restoreAll() => void

Does restore all mocked objects to their original state. See restore for further information. This function has not be called manually, if you rely on the default test suite hooks.

resetAll (static)

1Spy.resetAll() => void

Does reset all existing spies. This applies even to persistent spies. See reset for further information. This function has not to be called manually in between different tests, if you rely on the default test suite hooks.

IGNORE (static)

1Spy.IGNORE = $Internal Symbol$

This object can be passed anywhere where you want the "wasCalledWith" or "hasCallHistory" to ignore that object or value for comparison.

1spy({prop: 'value', other: 13}, 12);
2
3spy.wasCalledWith(Spy.IGNORE, 12);
4spy.wasCalledWith({prop: Spy.IGNORE, other: 13}, 12);

COMPARE (static)

1Spy.COMPARE(comparator: (arg: any) => boolean | void) => SpyComparator

This function can be called with a custom comparator and passed anywhere where you want the "wasCalledWith" or "hasCallHistory" to apply your custom comparison. Very useful if the spy gets called with functions that you want to test additionally.

1spy(() => ({ prop: 'value', other: 13 }), 12);
2
3spy.wasCalledWith(Spy.COMPARE(fn => fn().prop === 'value'), 12);
4spy.wasCalledWith(Spy.COMPARE(fn => {
5    expect(fn()).toEqual({ prop: 'value', other: 13 });
6}), 12);

MAPPER (static)

1Spy.MAPPER(from: any | any[], to: any) => SpyComparator

This function can be called in the same places like Spy.COMPARE. It is not that much customizable but provides a nice way to evaluate mapper functions. Meaning pure functions that return some output for given inputs. The function will be called exactly once for each comparison, so you can even rely on site effects you might want to test, if you want to use this for non-pure functions.

1spy((value: number) => ({ prop: 'here', other: value }), 12);
2spy((value: number, num: number) => ({ prop: 'here', value, num }), 12);
3
4spy.wasCalledWith(Spy.MAPPER('foo', { prop: 'here', other: 'foo' }), 12);
5spy.wasCalledWith(Spy.MAPPER(['foo', 44], { prop: 'here', value: 'foo', num: 44 }), 12);

configure

1spy.configure(config: { useOwnEquals?: boolean, persistent?: boolean }) => (this) SpyInstance

With configure the spy can be configured. One configuration possibility is to ignore any equals methods while comparing objects. There might be libraries which come with those methods, but do not support ES6 classes or anything else. By default, this configuration has been set to favor own equals implementations while comparing objects.

Another possible configuration is to make the spy persist while other spies have to restore when "restoreAll" was called. This spy can ONLY RESTORE the mocked object when you configure it back to be NOT PERSISTENT. This configuration can only be applied to mocking spies. For Spies created with Spy() this configuration will throw an exception.

calls

1spy.calls(...functions:Array<Function>) => (this) SpyInstance

The provided functions will be called sequentially in order when the spy will be called. Meaning spy.calls(func1, func2, func3) will call first func1 then func2 and the rest of the time func3.

returns

1spy.returns(...args: Array<any>) => (this) SpyInstance

The provided arguments will be returned sequentially in order when the spy will be called. Meaning spy.returns(arg1, arg2, arg3) will return first arg1 then arg2 and the rest of the time arg3.

resolves

1spy.resolves(...args: Array<any>) => (this) SpyInstance

The provided arguments will be resolved sequentially in order when the spy will be called. Meaning spy.resolves(arg1, arg2, arg3) will return first Promise.resolve(arg1) then Promise.resolve(arg2) and the rest of the time Promise.resolve(arg3).

rejects

1spy.rejects(...args: Array<?string | Error>) => (this) SpyInstance

The provided arguments will be rejected sequentially in order when the spy will be called. Meaning spy.rejects('foo', null, new Error('bar')) will return first Promise.reject(new Error('foo')) then Promise.reject(new Error('<SPY_NAME> was requested to throw')) and the rest of the time Promise.reject(new Error('bar')).

throws

1spy.throws(message: ?string | Error) => (this) SpyInstance

Perform this on a spy to make it throw an error when called. The error message can be provided, but a default has also been implemented. If an Error instance will be passed, exactly this one will be thrown.

reset

1spy.reset() => (this) SpyInstance

Does reset the registered calls on that spy.

restore

1spy.restore() => (this) SpyInstance

Restores the spied object, if existing, to its original state. The spy won't lose any other information. So it is still aware of made calls, can be plugged anywhere else and can still be called anywhere else, but it loses all references to the spied object.

If the spy has been configured persistent this method will throw an error.

transparent

1spy.transparent() => (this) SpyInstance

Can be useful with spies on objects. It does make the spy behave like not existing. So the original function of the "mocked" object will be called, but the spy does remember the call information.

transparentAfter

1spy.transparentAfter(callCount:number) => (this) SpyInstance

Works like transparent but the spy will get transparent after called as often as specified. Meaning spy.transparentAfter(num) will not be transparent on the first num calls.

addSnapshotSerializer

1spy.addSnapshotSerializer(serialize: string | ((...args: any[]) => string)) => (this) SpyInstance

Determines the rendered output of jest snapshots when the certain spy would get rendered.

wasCalled (fact)

1spy.wasCalled(callCount: number = 0) => void

This call does display a fact. So if the spy is violating the fact, it is told to throw an error. The provided argument does represent the registered calls on that spy.

wasNotCalled (fact)

1spy.wasNotCalled() => void

This fact displays that the spy has never been called. Directly after the spy was reseted, this fact will be given.

wasCalledWith (fact)

1spy.wasCalledWith(...args: Array<any>) => void

This fact displays that the spy has been called at least once with equal arguments.

The equality check is a deep equality check, which (by default) does consider own "equals" implementations.

By supplying Spy.IGNORE anywhere inside the expected call arguments, you can avoid that the comparison will be further executed. See Spy.IGNORE for further information and examples.

The deep equality check does also recursively iterate to the first difference found and is able to return a string which contains valuable information about the first found difference.

If any difference will be detected, the fact isn't true, and a helpful error message will be displayed. If using monospaced consoles for the output which do support new lines, there will be really neat output. For examples see showCallArguments.

wasNotCalledWith (fact)

1spy.wasNotCalledWith(...args: Array<any>) => void

This fact displays simply the opposite of wasCalledWith.

hasCallHistory (fact)

1spy.hasCallHistory(...callHistory: Array<Array<any> | any>) => void

Works similar to wasCalledWith but instead matches each call one by one in correct order and correct call count. ATTENTION: single argument calls can be provided without wrapping into an array. E.g. if the single argument is an array itself, then you have to warp it also yourself. (Inspired by jest data providers)

hasProps (fact)

1spy.hasProps(props: any) => void

This fact displays that the spy has currently the given props in React context of a mocked component.

This is the corresponding fact for getProps using the spy diffing.

getAllCallArguments

1spy.getAllCallArguments() => Array<any[]>

Returns the call arguments of all made calls to the spy. Especially returning an empty array if the spy has never been called.

getCallArguments

1spy.getCallArguments(callNr: number = 0) => Array<any>

Returns the call arguments that were registered on the given call. Meaning spy.getCallArguments(num) does return the (num + 1)'th call arguments.

Throws an exception if the provided (callNr - 1) is bigger than the made calls.

getCallArgument

1spy.getCallArgument(callNr: number = 0, argNr: number = 0) => any

Same as getCallArguments but returns only a single entry out of the array of arguments. Most useful in situations where exactly one call param is expected. If argNr is given, it returns the (argNr + 1)'th argument of the call.

getLatestCallArgument

1spy.getLatestCallArgument(argNr: number = 0) => any

Same as getCallArgument but uses only the latest call.

getProps

1spy.getProps() => any

Same as getLatestCallArgument but uses only the first arg. Can be useful in combination with mockReactComponents.

getCallCount

1spy.getCallCount() => number

This method simply returns the number of made calls on the spy.

showCallArguments

1spy.showCallArguments(additionalInformation: Array<string> = []) => string

This primarily internally used method is responsible for returning formatted informative debug messages when facts aren't true. Let's do an example:

1const spy = Spy('my awesome spy');
2spy(42, 'test', { attr1: [1, 2, new Date(2017, 1, 20)], attr2: 1337 });
3spy(42, 'test', { attr1: [0, 2, new Date(2017, 1, 20)], attr2: 1336 });
4spy(42, 'test', { attr1: [1, 2, new Date(2017, 1, 21)], attr2: 1336 });
5spy(42, 'tes', { attr1: [1, 2, new Date(2017, 1, 20)], attr2: 1336 });
6spy(42, 'test');

The following broken fact...

1spy.wasCalledWith(42, 'test', {attr1: [1, 2, new Date(2017, 1, 20)], attr2: 1336});

...would produce the following error output:

1Error: 
2
3my awesome spy was expected to be called with the following arguments:
4
5    --> [42, 'test', {attr1: [1, 2, new Date(1487545200000)], attr2: 1336}]
6
7Actually there were:
8
9call 0: [42, 'test', {attr1: [1, 2, new Date(1487545200000)], attr2: 1337}]
10        --> 2 / attr2 / different number [1337 != 1336]
11call 1: [42, 'test', {attr1: [0, 2, new Date(1487545200000)], attr2: 1336}]
12        --> 2 / attr1 / 0 / different number [0 != 1]
13call 2: [42, 'test', {attr1: [1, 2, new Date(1487631600000)], attr2: 1336}]
14        --> 2 / attr1 / 2 / different date [new Date(1487631600000) != new Date(1487545200000)]
15call 3: [42, 'tes', {attr1: [1, 2, new Date(1487545200000)], attr2: 1336}]
16        --> 1 / different string ['tes' != 'test']
17call 4: [42, 'test']
18        --> 2 / one was undefined [undefined != {attr1: [1, 2, new Date(1487545200000)], attr2: 1336}]

There you can see that the arguments of the fact (displayed above all others) does not match any of the call arguments on the 5 made calls.

For each call we display additional error information (the first found difference). If the additional information begins with a --> there was made a deep equality. If you would travers with the displayed keys you would be directed to those objects which differ.

In this example the arguments differ for call 0 in --> the third argument (2) and its attribute attr2 because there was a different number.

While recursively traversing down in the deep equality check, the object keys will be reported. Meaning that 2 is representing the index of the array. So for example if you want to grep the different objects you could:

1const callArgs = spy.getCallArguments(0/* for the 0'th call above*/);
2const differentNumber = callArgs[2]['attr2'];

Enforce-Order Mode

You can opt-in to the enforce-order mode. Which might become the default in some future version but will need first further evaluation and will always stay configurable.

1Spy.configure({ enforceOrder: true });

This mode enforces that the "facts" will be called in the correct order.

1// success
2spy1();
3spy2();
4spy1.wasCalled();
5spy2.wasCalled();
6
7// error
8spy1();
9spy2();
10spy2.wasCalled();
11spy1.wasCalled(); // would fail here because spy1 wasn't called after spy2 was called

Be aware "facts" that you might need to get used to it, because the following would be valid, too.

1// success
2spy();
3spy.wasCalled();
4spy.wasNotCalled();
5
6// error
7spy();
8spy.wasCalled();
9spy.wasCalled(); // would fail here because we already checked that the spy was called

Nevertheless, this mode should make your tests more readable and clear, because you can avoid checking the same things on and on again or resetting the spies in tests. Another example:

1const mock_WS = Spy.mock(WS, 'fetchData', 'fetchFallback');
2
3it('fetches fallback data if fetching data does not work', async () => {
4    // given
5    const dummyData = Symbol('dummyData'); 
6    mock_WS.fetchData.rejects('ups');
7    mock_WS.fetchFallback.resolves(dummyData);
8
9    // when
10    expect(await MyService.fetchAppData({ filtered: true })).toBe(dummyData);
11
12    // then
13    mock_WS.fetchData.hasCallHistory({ filtered: true }, { filtered: true, retry: true });
14    mock_WS.fetchFallback.wasCalledWith({ filtered: true, reason: 'ups' });
15});

Bun Support

bun includes an alternative fast runtime to NodeJS. In order to use this library with bun test, you need to add a bunfig.toml with a preload script:

1[test]
2preload = "./bun-preload.ts"

And the preload script should implement missing test APIs. At the time of writing that would be at least this:

1// necessary until Bun has implemented them
2expect.addSnapshotSerializer = () => null;
3expect.getState = () => ({}) as any;

⚠️ Disclaimers:

• You cannot call Spy.setup() in a single place inside of the preload script as the beforeEach hook of will be differently scoped, when you want/need to support scoped mocks. However, you can call the Spy.setup again in the tests where you need scoped mocks and provide it a new beforeEach hook.
• Since expect.getState is not correctly implemented, you will not be prevented correctly from instantiating mocks inside of test runs.