spy4js

spy4js

Benefits

• Well tested
• Used in production of large projects
• Flow support (therefore autocompletion in some IDEs)
• Without dependecies nor boilerplate
• Helpful Debug-Messages
• Supports own "equals" implementation on objects
• A lot of useful operations on spys

Introduction

spy4js provides a stand-alone spy framework. It is decoupled by any dependencies and other assertion frameworks. Other than most test frameworks it uses a different - maybe you will need to get used to - test notation. It does not make assertion, which are expected to be fulfilled on runtime, but displays facts, that are considered to be fulfilled on runtime. And if this fact is not true, it will throw an exception. I consider the way of writing facts more regular because it fits more to the rest of the written code.

spy4js comes with the one interesting (es6 like) class Spy. The spy instances are treated as class instances and come with a lot of useful features. See below for more.

Installation

Like every other npm package. You may npm install spy4js --save-dev to save the latest version to your dev dependencies.

Interface

A spy instance can be initialized differently.

import {Spy} from 'spy4js';

// initialize directly
const spy1 = new Spy();

// initialize directly and supply an identifier for debugging purpose (default: 'the spy')
const spy2 = new Spy('special spy for me');

// initialize by mocking another objects attribute (usually this attribute is a function)
const someObject = new Date(2017, 1, 15);
const spy3 = Spy.on(someObject, 'toJSON');
// (spy name will be accordingly: 'the spy on \'toJSON\'')

// initialize many by mocking another objects attributes
const someObject = new Date(2017, 1, 15);
const [spy4, spy5, spy6] = Spy.onMany(someObject, 'toJSON', 'toString', 'getDate');

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

• configure (some external librarys may use own "equals" implementations in an unexpected way)
• 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)
• 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)

All those methods on a spy are designed in a builder pattern. So you may chain any of these configurations. But be aware that some behaviours override existing behaviours.

const spy = Spy.on(someObject, 'someMethod');

// configure it to use NOT own "equals" implementations
spy.configure({useOwnEquals: false});

// make it call any functions
spy.calls(func1, func2, func3);
someObject.someMethod(arg); // returns func1(arg)
someObject.someMethod(arg1, arg2); // returns func2(arg1, arg2)
someObject.someMethod(arg); // returns func3(arg)
someObject.someMethod(arg1, arg2, arg3); // returns func3(arg1, arg2, arg3) // sticks to the last

// make it return any values
spy.returns(value1, value2);
someObject.someMethod(arg); // returns value1
someObject.someMethod(arg1, arg2); // returns value2
someObject.someMethod(arg); // returns value2 // sticks to the last

// make it throw any message (the message is optional)
spy.throws('throw this');
someObject.someMethod(arg); // throws new Error('throw this')

// make it return always the current date and transparentAfter 2 calls
spy.calls(() => new Date()).transparentAfter(2);
someObject.someMethod(arg); // returns new Date()
someObject.someMethod(arg1, arg2); // returns new(er) Date()
someObject.someMethod(arg); // returns someObject.someMethod(arg) // sticks to this behaviour

// make it immediatly transparent
spy.transparent();

// make it reset
spy.reset();

// make it restore
spy.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 was called a specifiable amount of times)
• wasNotCalled (does display that the spy was never called)
• wasCalledWith (does display that the spy was called at least once like with the provided params)
• wasNotCalledWith (does display that the spy was never like with the provided params)

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.

const spy = new Spy();

spy.wasNotCalled();

// in fact: you never want to call a spy directly for any purpose (therefore using flow this will complain)
spy([1, 'test', {attr: [4]}]);

spy.wasCalled(); // called at least once
spy.wasCalled(1); // called exctly once

spy('with this text');

spy.wasCalled(2); // called extactly 2 times

spy.wasCalledWith([1, 'test', {attr: [4]}]); // the spy was called at least once with equal params

spy.wasNotCalledWith([1, 'test', {attr: [3]}]); // the spy was not called with those params (do you see the difference?)

There is one static method that does restore all existing spies in all tests. This is extremly useful to clean up all still existing mocks and also a very comfortable to this automaticly after every test (like in an "afterEach").

• restoreAll (does restore every existing spy)

Spy.restoreAll();

And also sometimes it is necessary to have access to some of the call arguments with which the spy was called.

• getCallArguments (returns all call arguments for a specified call in an array)
• getFirstCallArgument (same as getCallArguments, but returns only the first element of the array)

const spy = new Spy();

// make some calls
spy('string', 1);
spy([1, 2, 3]);
spy();
spy(null);

spy.getCallArguments(/* default = 0 */); // returns ['string', 1]
spy.getFirstCallArgument(/* default = 0 */); // returns 'string'

spy.getCallArguments(1); // returns [[1, 2, 3]]
spy.getFirstCallArgument(1); // returns [1, 2, 3]

spy.getCallArguments(2); // returns []
spy.getFirstCallArgument(2); // returns undefined

spy.getCallArguments(3); // returns [null]
spy.getFirstCallArgument(3); // returns null

spy.getCallArguments(4); // throws Exception because less calls were made
spy.getFirstCallArgument(4); // throws same Exception

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

Method-Details

Constructor

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

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

on (static)

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

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

onMany (static)

Spy.onMany(object:Object, ...methodNames:Array<string>) => Array<Spy>

Initializing as many spies as required for one and the same object. Same as calling Spy.on for each method name.

restoreAll (static)

Spy.restoreAll() => Array<Spy>

Does restore all mocked objects to their original state. See restore for further information.

configure

spy.configure(config:{useOwnEquals:boolean|void}) => (this) Spy

With configure the spy can be configured. For now the only 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 is set to favor own equals implementations while comparing objects.

calls

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

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

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

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.

throws

spy.throws(message:?string) => (this) Spy

Perform this on a spy to make it throw an error when called. The error message can be provided but a default is also implemented.

reset

spy.reset() => (this) Spy

Does reset the registered calls on that spy.

restore

spy.restore() => (this) Spy

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.

transparent

spy.transparent() => (this) Spy

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

spy.transparentAfter(callCount:number) => (this) Spy

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.

wasCalled

spy.wasCalled(callCount:number = 0) => (fact) 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

spy.wasNotCalled() => (fact) void

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

wasCalledWith

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

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

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

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 was detected. The fact is broken 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

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

This fact displays simply the opposite of wasCalledWith.

getCallArguments

spy.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.

getFirstCallArgument

spy.getFirstCallArgument(callNr:number = 0) => any

Same as getCallArguments but returns the only the first entry out of the array of arguments. Most useful in situations where exactly one call param is expected.

showCallArguments

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

This primarily internally used method is responsible for returning formatted informative debug messages when facts are broken. Let's do an example:

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

The following broken fact...

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

...would produce the following error output:

Error: 
	
For my awesome spy did not find call arguments:

    --> [42,"test",{"attr1":[1,2,"2017-02-19T23:00:00.000Z"],"attr2":1336}]

Actually there were:

call 0: [42,"test",{"attr1":[1,2,"2017-02-19T23:00:00.000Z"],"attr2":1337}]
        --> 2 / attr2 / different number
call 1: [42,"test",{"attr1":[0,2,"2017-02-19T23:00:00.000Z"],"attr2":1336}]
        --> 2 / attr1 / 0 / different number
call 2: [42,"test",{"attr1":[1,2,"2017-02-20T23:00:00.000Z"],"attr2":1336}]
        --> 2 / attr1 / 2 / different date
call 3: [42,"tes",{"attr1":[1,2,"2017-02-19T23:00:00.000Z"],"attr2":1336}]
        --> 1 / different string
call 4: [42,"test"]
        different key length

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:

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