I wanted to run some Spectron e2e tests I wrote some weeks ago, but to my surprise, suddenly they all stopped working for one and the same reason.
According to the error message, I'm dealing with rejected Promises, but I can't figure out where the problem is coming from. Calling done at the end of my testcase raises the exact same error.
I'm running the following command to launch my test suit: mocha test/e2e
Mocha then executes this index.js before running my tests in ordner to support ES6+ features
'use strict'
//index.js
// Set BABEL_ENV to use proper env config
process.env.BABEL_ENV = 'test'
// Enable use of ES6+ on required files
require('babel-register')({
ignore: /node_modules/
})
// Attach Chai APIs to global scope
const { expect, should, assert } = require('chai')
global.expect = expect
global.should = should
global.assert = assert
// Require all JS files in `./specs` for Mocha to consume
require('require-dir')('./specs')
After that its trying to run this small Login.spec.js which returns the error mentioned above
import utils from '../utils'
import {Application} from "spectron";
import electron from "electron";
describe('🔑 Login', function () {
this.timeout(11000);
it('login form exists', async function (done) {
this.app = new Application({
path: electron,
env: {"SPECTRON_RUNNING":true},
args: ['dist/electron/main.js'],
startTimeout: 10000,
waitTimeout: 10000
})
await this.app.start()
await this.app.client.windowByIndex(1);
done();
})
})
Related
Setup
I want to unit test my electron app with jest. For this I have the following setup:
I use jest and use #kayahr/jest-electron-runner to run it with electron instead of node. Additionally, since it is a typescript project, I use ts-jest.
My jest.config.js looks like this:
module.exports = {
collectCoverage: true,
coverageDirectory: 'coverage',
coverageProvider: 'v8',
preset: 'ts-jest',
runner: '#kayahr/jest-electron-runner/main',
testEnvironment: 'node',
};
The test is expected to run in the main process. I have reduced my code to the following example function:
import { app } from 'electron';
export function bar() {
console.log('in bar', app); //this is undefined when mocked, but I have a real module if not mocked
const baz = app.getAppPath();
return baz;
}
The test file:
import electron1 from 'electron';
import { bar } from '../src/main/foo';
console.log('in test', electron1); //this is undefined in the test file after import
// jest.mock('electron1'); -> this does just nothing, still undefined
const electron = require('electron');
console.log('in test after require', electron); //I have something here yay
jest.mock('electron'); //-> but if I comment this in -> it is {} but no mock at all
it('should mock app', () => {
bar();
expect(electron.app).toBeCalled();
});
What do I want to do?
I want to mock electron.app with jest to look whether it was called or not.
What is the problem?
Mocking electron does not work. In contrast to other modules like fs-extra where jest.mock() behaves as expected.
I don't understand the behavior happening here:
Importing "electron" via import in the file containing the tests (not the file to be tested!) does not work (other modules work well), but require("electron") does.
I do have the electron module if not mocked in bar(), but after mocking not
while jest.mock("fs-extra") works, after jest.mock("electron") electron is only an empty object, not a mock
I would really like to understand what I did wrong or what the problem is. Switching back to #jest-runner/electron does not seem to be an option, since it is not maintained anymore. Also I don't know if this is even the root of the problem.
Has anyone seen this behavior before and can give me a hint where to start searching?
Hello stackoverflow community,
The task
I'm running Jasmine tests programmatically using jasmine.execute(). My task is to run onPrepare function, where I do some setup work like setting up reporters etc, and I need that function to be synchronous (has to be finished before running specs)
Approach #1
The first approach I tried was to just declare an async onPrepare function, which also includes the code for specifying reporters, and then do
await onPrepare();
await jasmine.execute();
Problem
In the result I get jasmine.getEnv() is not a function. I assume because getEnv() becomes available as .execute() is ran. So I understand this won't work
Approach #2
The next thing I tried was to create a helper file with my sync code, specify it in the config and run jasmine.execute();.
So, if simplified, I have
// conf.js
(async () => {
let Jasmine = require('jasmine');
let jasmine = new Jasmine();
let variables = require("./variables.json");
let {spawn} = require("child_process");
let childProcess = spawn(variables.webdriver);
console.log(`Webdriver started, pid: ${childProcess.pid}`);
jasmine.exitOnCompletion = false;
jasmine.loadConfig({
'spec_files': ['specs/*.spec.js'],
'helpers': ['on-jasmine-prepare.js'],
'stopSpecOnExpectationFailure': false,
'random': false,
})
const result = await jasmine.execute();
console.log('Test status:', result.overallStatus);
console.log('Closing library and webdriver process');
await library.close();
await childProcess.kill();
console.log('webdriver killed:', childProcess.killed);
})()
// on-jasmine-prepare.js
(async () => {
const {SpecReporter} = require("jasmine-spec-reporter");
const library = require("./library/library");
const variables = require("./variables.json");
const errorHandler = require("./modules/on-error-handler");
jasmine.getEnv().clearReporters();
jasmine.DEFAULT_TIMEOUT_INTERVAL = 3 * 60 * 1000;
jasmine.getEnv().addReporter(new SpecReporter({}))
global.library = new library.Library(variables.IP);
console.log('library instantiated')
await library.deleteSessions();
console.log('sessions deleted')
await library.launch(library.home);
console.log('home page launched')
jasmine.getEnv().addReporter(
errorHandler(library)
)
console.log('debugger reporter added' )
})();
Problem
The problem that I noticed that the helper file is executed asynchronously with the specs and I get a spec error before the helper function finishes (basically because of race condition). Below the output example, where you can see some console.log from onPrepare was ran after spec started
Webdriver started, pid: 40745
library instantiated
Jasmine started
sessions deleted
Example
✗ App is loaded (7 secs)
- Error: Session already exist
home page launched
debugger reporter added
The question
How do I run onPrepare function synchronously, before specs? Preferably natively (using only jasmine capabilities). Otherwise maybe using third party packages. I know it's possible because #protractor had this feature, however I couldn't back-engineer it
MacOS
node v16.13.2
jasmine v4.1.0
Thank you
Issue:
When running a test in Mocha, we need to add a document to the local emulated Firestore. The local emulated setup is working but not within the Mocha test, specifically.
Node Version: 14.18.1, chai: 4.3.4, firebase-admin: 10.0.0, firebase: 9.1.3
Explaination
Working Locally:
If I were to try and run:
await conref.doc('testDoc').set({test: true}); before the first Mocha
describe(), I can verify the document is added to my local Firestore at
localhost:8080
const db = getFirestore();
const ref = db.collection('testCollection');
await ref.doc('testDoc').set({test: true});
const testDoc = await db.collection('testCollection').doc('testDoc').get();
console.log(testDoc.data());
prints { test: true }, confirming packages and connection is working
Breaking In Mocha:
..however placing similar logic inside my describe() causes the test to report as
errored each and every time, returning a red (1) instead of a green checkmark:
describe("example", async function() {
it("should write to Firestore in a mocha test", async function() {
const db = getFirestore();
const ref = db.collection('testCollection');
await ref.doc('testDoc').set({test: true}); // < causes error
};
};
screenshot
Imports
Firebase has had different version and its been confusing, so I will include the imports I am using for both of these examples below, in case this might be very wrong.
import { getFirestore } from 'firebase-admin/firestore';
import { initializeApp, applicationDefault } from 'firebase-admin/app';
initializeApp({
credential: applicationDefault(),
databaseURL: process.env.FIREBASE_DATABASAE_URL,
});
Attempts
Tried adjusting timeouts, thinking maybe the Mocha suite was not waiting for the doc to be added, but even an addition of this.timeout(10000); within the test did not change the outcome.
Tried setting the doc before the test, which worked, so I put this in a function and tried calling it from within the test as if I was praying on a Hail Mary, but to no surprise this had no effect.
Tried connecting to my production Firestore but still no cigar (via removing the FIRESTORE_EMULATOR_HOST env).
Added done() at the end of the test, but did not change outcome
I'm using node with TypeScript on my back end and Jest and Supertest as my test framework on my back end.
When I'm trying to test I have the result pass but I get an error at the end. Here's the result:
PASS test/controllers/user.controller.test.ts
Get all users
✓ should return status code 200 (25ms)
console.log node_modules/#overnightjs/logger/lib/Logger.js:173
[2019-12-05T04:54:26.811Z]: Setting up database ...
Test Suites: 1 passed, 1 total
Tests: 1 passed, 1 total
Snapshots: 0 total
Time: 3.284s
Ran all test suites.
server/test/controllers/user.controller.test.ts:32
throw err;
^
Error: connect ECONNREFUSED 127.0.0.1:80
at TCPConnectWrap.afterConnect [as oncomplete] (net.js:1104:14)
npm ERR! Test failed. See above for more details.
Here's my test code:
import request from "supertest";
import { AppServer } from '../../config/server';
const server = new AppServer();
describe('Get all users', () => {
it('should return status code 200', async () => {
server.startDB();
const appInstance = server.appInstance;
const req = request(appInstance);
req.get('api/v1/users/')
.expect(200)
.end((err, res) => {
if (err) throw err;
})
})
})
Here's my server setup. I'm using overnightjs on my back end.
I created a getter to get the Express instance. This is coming from overnight.js.
// this should be the very top, should be called before the controllers
require('dotenv').config();
import 'reflect-metadata';
import { Server } from '#overnightjs/core';
import { Logger } from '#overnightjs/logger';
import { createConnection } from 'typeorm';
import helmet from 'helmet';
import * as bodyParser from 'body-parser';
import * as controllers from '../src/controllers/controller_imports';
export class AppServer extends Server {
constructor() {
super(process.env.NODE_ENV === 'development');
this.app.use(helmet());
this.app.use(bodyParser.json());
this.app.use(bodyParser.urlencoded({ extended: true }));
this.setupControllers();
}
get appInstance(): any {
return this.app;
}
private setupControllers(): void {
const controllerInstances = [];
// eslint-disable-next-line
for (const name of Object.keys(controllers)) {
const Controller = (controllers as any)[name];
if (typeof Controller === 'function') {
controllerInstances.push(new Controller());
}
}
/* You can add option router as second argument */
super.addControllers(controllerInstances);
}
private startServer(portNum?: number): void {
const port = portNum || 8000;
this.app.listen(port, () => {
Logger.Info(`Server Running on port: ${port}`);
});
}
/**
* start Database first then the server
*/
public async startDB(): Promise<any> {
Logger.Info('Setting up database ...');
try {
await createConnection();
this.startServer();
Logger.Info('Database connected');
} catch (error) {
Logger.Warn(error);
return Promise.reject('Server Failed, Restart again...');
}
}
}
I read this question - that's why I called the method startDB.
So I figured out and the solution is quite easy. I can't explain why though.
This req.get('api/v1/users/') should be /api/v1/users - you need a leading /.
For Frontend...
If you are making use of axios and come across this error, go to the testSetup.js file and add this line
axios.defaults.baseURL = "https://yourbaseurl.com/"
This worked for me. So, typically, this is a baseURL issue.
I had this error in my React frontend app tests.
I was using React testing library's findBy* function in my assert:
expect(await screen.findByText('first')).toBeInTheDocument();
expect(await screen.findByText('second')).toBeInTheDocument();
expect(await screen.findByText('third')).toBeInTheDocument();
After I changed it to:
await waitFor(async () => {
expect(await screen.findByText('first')).toBeInTheDocument();
expect(await screen.findByText('second')).toBeInTheDocument();
expect(await screen.findByText('third')).toBeInTheDocument();
});
the error is gone.
I don't know exactly why, but maybe it will help someone
UPDATE: I was mocking fetch incorrectly, so my test called real API and caused that error
I put this line in my setupTests file:
global.fetch = jest.fn()
It mocks fetch for all tests globally. Then, you can mock specific responses right in your tests:
jest.mocked(global.fetch).mockResolvedValue(...)
// OR
jest.spyOn(global, 'fetch').mockResolvedValue(...)
Slightly different issue, but same error message...
I was having this error when using node-fetch when trying to connect to my own localhost (http://localhost:4000/graphql), and after trying what felt like everything under the sun, my most reliable solution was:
using this script in package.json: "test": "NODE_ENV=test jest --watch"
If the terminal shows connection error I just go to the terminal with Jest watching and press a to rerun all tests and they pass without any issue.
¯\_(ツ)_/¯
Success rate continued to improve by renaming the testing folder to __tests__ and moving my index.js to src/index.js.
Very strange, but I am too exhausted to look at the Jest internals to figure out why.
The rules for supertest are the same as the rules for express. OvernightJS does not require any leading or ending "/" though.
For anyone landing on this, but not having issues with trailing slashes:
jest can also return a ECONNREFUSED when your express app takes some time (even just a second) to restart/init. If you are using nodemon like me, you can disable restarts for test files like --ignore *.test.ts.
This error also occurs if you have not set up a server to catch the request at all (depending on your implementation code and your test, the test may still pass).
I didn't get to the bottom of this error - it wasn't related to the (accepted) leading slash answer.
However, my "fix" was to move the mocks up into the suite definition - into beforeAll and afterAll for cleanup between tests).
Before, I was mocking (global.fetch) in each test, and it was the last test in the suite to use the mock that would cause the error.
In my case, the issue was related to package react-inlinesvg. Package makes a fetch request to get the svg file and since server is not running, it gets redirected to default 127.0.0.1:80.
I mocked react-inlinesvg globally to output props including svg filename to assert in testing.
jest.mock('react-inlinesvg', () => (props) => (
<svg data-testid="mocked-svg">{JSON.stringify(props)}</svg>
));
I have numerous tests spread across multiple files in a Node JS application. I'd like to run bootstrap code prior to Mocha's execution of any of the test files. This is so that I can, for example, set globals to be used in each of the actual tests.
Sample bootstrap code
global.chai = require('chai');
global.expect = chai.expect;
global.sinon = require('sinon');
It seems Mocha loads all files under /test alphabetically, so if I name this bootstrap code "bootstrap.js" and everything else with a starting letter after "B" it "works".
Obviously this is fragile and sucky, but I don't want to put this boilerplate requiring of my supporting libraries at the top of every test file.
How do I tell Mocha to load a bootstrap script first, or create something functionally equivalent?
have you tried mocha --require mymodule.js TESTS_DIR
from the documentation
-r, --require
The --require option is useful for libraries such as should.js, so you
may simply --require should instead of manually invoking
require('should') within each test file. Note that this works well for
should as it augments Object.prototype, however if you wish to access
a module's exports you will have to require them, for example var
should = require('should').
you could also write at the top of each test to load the require("./bootstrap.js") and run tests.
I use mocha's flag --delay
If you need to perform asynchronous operations before any of your
suites are run, you may delay the root suite. Simply run Mocha with
the --delay flag. This will provide a special function, run(), in the
global context.
setTimeout(function() {
// do some setup
describe('my suite', function() {
// ...
});
run();
}, 5000);
If prior to running Mocha you want to first run some bootstrap code in a file that uses ECMAScript 2015 Module syntax (i.e. import instead of require)
Create the following files:
./setupBabel.js (to bootstrap Babel transpiler)
require('babel-polyfill');
require('babel-register');
./setupDependencies.js (to bootstrap Chai and Sinon using ES2015 syntax)
import chai from 'chai';
import sinon from 'sinon';
global.expect = chai.expect;
global.sinon = sinon;
./test/codeSpec.js (example Unit Test using Chai, Sinon Stubs, Spies, and ES2015 syntax such as arrow functions, let, and const)
describe('test suite', () => {
it('does not modify immutable constant value', () => {
const myStub = sinon.stub();
const mySpy = sinon.spy(myStub);
myStub.returns({});
let val = 1;
expect(val).to.not.equal(2);
sinon.assert.notCalled(mySpy);
});
});
Run the following in terminal to install relevant NPM packages:
npm install babel-polyfill babel-register chai sinon mocha
Run the test suite with the following terminal command and flags:
mocha --require ./setupBabel.js --require ./setupDependencies.js ./test/codeSpec.js
describe('异步钩子测试', function () {
const lover = {
bodyname: 'yueyue',
girlname: 'fangfang'
}
const test = 'lihang'
beforeEach('每一个测试之前的钩子', function () {
return new Promise((resolve, reject) => {
setTimeout(() => {
resolve('通过')
})
})
})
it('方方爱张越', function () {
// 是否等于yueuyue
expect(lover.bodyname).to.equal('yueyue')
// 是否是一个字符串
lover.girlname.should.equal('fangfang')
})
})