A unit test assertion should have only one reason to succeed because it helps to ensure that the test is focused and specific. When a test has
multiple reasons to succeed, it becomes difficult to determine the root cause of a failure if the test fails. This can lead to confusion and wasted
time trying to debug the test.
This rule raises an issue when the following Chai.js assertions are found:
- When
.not and .throw are used together and at least one
argument is provided to .throw. Such assertions succeed when the target either does not throw any exception, or when it throws an
exception different from the one provided.
- When
.not and .include are used together and an
object is given to .include. Such assertions succeed when one or multiple key/values are missing.
- When
.not and .property are used together and
.property is given at least two arguments. Such assertions succeed when the target either doesn’t have the requested property, or when
this property exists but has a different value.
- When
.not and .ownPropertyDescriptor are
used together and .ownPropertyDescriptor is given at least two arguments. Such assertions succeed when the target either doesn’t have
the requested property descriptor, or its property descriptor is not deeply equal to the given descriptor
- When
.not and .members are used together. Such
assertions succeed when one or multiple members are missing.
- When
.change and .by are used together. Such assertions succeed when the target either decreases or
increases by the given delta
- When
.not and .increase are used together. Such
assertions succeed when the target either decreases or stays the same.
- When
.not and .decrease are used together. Such
assertions succeed when the target either increases or stays the same.
- When
.not negates .by. Such assertions succeed when the
target didn’t change by one specific delta among all the possible deltas.
- When
.not and .finite are used together. Such assertions
succeed when the target either is not a number, or is one of Nan, positive Infinity, negative
Infinity.
const expect = require('chai').expect;
describe("Each Chai.js assertion", function() {
const throwsTypeError = () => { throw new TypeError() }
it("has more than one reason to succeed", function() {
expect(throwsTypeError).to.not.throw(ReferenceError) // Noncompliant
expect({a: 42}).to.not.include({b: 10, c: 20}); // Noncompliant
expect({a: 21}).to.not.have.property('b', 42); // Noncompliant
expect({a: 21}).to.not.have.ownPropertyDescriptor('b', { // Noncompliant
configurable: true,
enumerable: true,
writable: true,
value: 42,
});
expect([21, 42]).to.not.have.members([1, 2]); // Noncompliant
let myObj = { value: 1 }
const incThree = () => { myObj.value += 3; };
const decThree = () => { myObj.value -= 3; };
const doNothing = () => {};
expect(incThree).to.change(myObj, 'value').by(3); // Noncompliant
expect(decThree).to.change(myObj, 'value').by(3); // Noncompliant
expect(decThree).to.not.increase(myObj, 'value'); // Noncompliant
expect(incThree).to.not.decrease(myObj, 'value'); // Noncompliant
expect(doNothing).to.not.increase(myObj, 'value'); // Noncompliant
expect(doNothing).to.not.decrease(myObj, 'value'); // Noncompliant
expect(incThree).to.increase(myObj, 'value').but.not.by(1); // Noncompliant
let toCheck;
expect(toCheck).to.not.be.finite; // Noncompliant
});
});
By having only one reason to succeed, the test is more precise and easier to understand. It also helps to ensure that the test is testing only one
specific behavior or functionality of the code, which makes it easier to identify and fix any issues that arise.
const expect = require('chai').expect;
describe("Each Chai.js assertion", function() {
const throwsTypeError = () => { throw new TypeError() }
it("has only one reason to succeed", function() {
expect(throwsTypeError).to.throw(TypeError)
expect({a: 42}).to.not.have.any.keys('b', 'c');
expect({a: 21}).to.not.have.property('b');
expect({a: 21}).to.not.have.ownPropertyDescriptor('b');
expect([21, 42]).to.not.include(1).and.not.include(2);
let myObj = { value: 1 }
const incThree = () => { myObj.value += 3; };
const decThree = () => { myObj.value -= 3; };
const doNothing = () => {};
expect(incThree).to.increase(myObj, 'value').by(3);
expect(decThree).to.decrease(myObj, 'value').by(3);
expect(decThree).to.decrease(myObj, 'value').by(3);
expect(incThree).to.increase(myObj, 'value').by(3);
expect(doNothing).to.not.change(myObj, 'value');
expect(incThree).to.increase(myObj, 'value').by(3);
let toCheck;
// Either of the following is valid
expect(toCheck).to.be.a('string');
expect(toCheck).to.be.NaN;
expect(toCheck).to.equal(Infinity);
expect(toCheck).to.equal(-Infinity);
});
});
Having only one reason to succeed also helps to make the test more maintainable. If the test needs to be updated or modified in the future, it is
easier to do so when the test is focused on a single behavior or functionality.
