Logo by: Hsin-chieh Yeh
Muppeteer is a visual regression testing framework for running UI tests in Chrome. It's composed of a number of modules:
- Mocha - a test runner framework
- Chai - an assertion library
- Puppeteer - a library for interacting with Chrome and web pages
- Puppeteer Extensions - an extension library for Puppeteer with convenience functions
- Pixelmatch - a pixel-level image comparison library
In addition, it provides the following core features:
- Visual Regression Testing - a screenshot-based image comparison module that hooks onto the assertion API. Read on for more discussion on this.
- Test Interface - a modification of Mocha's BDD interface with built-in browser setup steps and other user configurable hooks
- Test Launcher - a CLI and configuration function for launching test suites
Muppeteer's main goal is to abstract the, often, tedious boilerplate setup code needed to write tests with Puppeteer, and provide a convenient API for testing UI functionality. It was inspired by the PhantomCSS and CasperJS libraries.
You can configure Muppeteer via the CLI or a configuration function
The CLI script can be referenced at
lib/launcherCli
.
It is run like node <<path-to-muppeteer>>/lib/launcherCli <<args>>
"scripts": {
"test": "node node_modules/muppeteer/lib/launcherCli --p tests/*.test.js --r tests/report"
}
See Options
The configuration can be referenced at
lib/Launcher
.
const Launcher = require('../lib/Launcher');
const path = require('path');
const launcher = new Launcher({
testPathPattern: `${__dirname}/tests/*.test.js`
reportDir: `${__dirname}/tests/report`,
visualThreshold: 0.05,
useDocker: true,
dockerChromeVersion: '67.0.3396.79',
onFinish: () => {
// do something after the tests have complete
}
}
);
await launcher.run();
Note: Only options with --
can be run with the proceeding flag in the CLI interface.
testPathPattern (--p)
: A glob match for test files. This is the new and recommended way.testDir (--t)
: The directory for Mocha to look for the test files. This is the old way of matching tests.testFilter (--f)
: Allows you to pass in some text to filter the test file name. This is the old way of matching tests.shouldRebaseVisuals (--b)
: A flag to tell the visual regression engine to replace the existing baseline visualsreportDir (--r)
: The directory for the Mocha reporter to dump the report filescomponentTestUrlFactory
: A function that returns the url for the component test to runcomponentTestVisualPathFactory
: A function that returns the path for visual tests to run invisualThreshold (--v)
: A value between 0 and 1 to present the threshold at which a visual test may pass or failonFinish
: A function that can be used to do some extra work after Muppeteer is teared downuseDocker (--d)
: The option for telling Muppeteer to run Chrome in Docker to better deal with environmental inconsistencies (default)dockerChromeVersion (--c)
: The version of Chrome to use in the Docker container. This should be set explicitly to avoid different environments having different versions of Chrome. By default, the latest version is pulled from the hub, which is not recommended.headless (--h)
: Determines whether Chrome will be launched in a headless mode (without GUI) or with a head (not applicable withuseDocker
)disableSandbox (--s)
: Used to disable the sandbox checks if not using SUID sandbox (not applicable withuseDocker
)executablePath (--e)
: The option to set the version of Chrome to use duning the tests. By default, it uses the bundled version (not applicable withuseDocker
)
You can access the Puppeteer Extensions API with page.extensions
in your tests.
describeComponent({ name: "Panel" }, () => {
describe("Simple mode", async () => {
const panelContainer = ".first-usage .panel";
const panelTitle = ".first-usage .panel-title";
const panelBody = ".first-usage .panel-body";
it("title and body exist", async () => {
await page.waitForSelector(panelTitle);
const titleText = await page.extensions.getText(panelTitle);
assert.equal(titleText, "My title");
await page.waitForSelector(panelBody);
const bodyText = await page.extensions.getText(panelBody);
assert.equal(bodyText, "This is some test data");
});
it("title and body appear correctly", async () => {
await assert.visual(panelContainer);
});
});
describe("Icon mode", async () => {
const panelContainer = ".second-usage .panel";
const panelTitle = ".second-usage .panel-title";
const panelBody = ".second-usage .panel-body";
it("title, body and icon exist", async () => {
await page.waitForSelector(panelTitle);
const titleText = await page.extensions.getText(panelTitle);
assert.equal(titleText, "My title");
await page.waitForSelector(panelBody);
const bodyText = await page.extensions.getText(panelBody);
assert.equal(bodyText, "This is a little bit more test data");
});
it("title, body and icon appear correctly", async () => {
await assert.visual(panelContainer);
});
});
});
Muppeteer uses Docker by default to run tests. This helps to avoid environmental differences that could affect the
rendering of content on the page. You can opt out by configuring the useDocker (--d)
option accordingly.
You can specify the version of Chrome to use by configuring with the dockerChromeVersion
option. When you use this option,
the test launcher will automatically pull the correct Docker image from a repository and build the container for you. If you
don't specify this, the latest
version will be used. This can result in unexpected behaviour, so it's advised to
pin the version with this property.
If you are hosting your test fixtures on a local web server, you'd typically set the URL in the test to
something like http://localhost:3000. When using Docker, the localhost
will refer to the container,
not the host machine. The simplest solution would be to reference the local IP in the test instead. For example,
http://192.168.0.4:3000.
However, this breaks down when you are running on a device you don't know how to address, e.g. a cloud CI agent.
To solve this problem, you can use the componentTestUrlFactory
function in launch configuration to generate the URL.
You can lookup the IP address of the current host and pass that through. This is used to run the example tests in this repo.
See network for an example.
...
componentTestUrlFactory: () => `http://${IP}:${PORT}`
...
This is the visual that is versioned in your app repo. It is the source of truth.
This is the screenshot taken during the test. In this example, we can see that some extra space on the left has pushed the title to the right
This is an image showing where the differences are. Each difference is layered on top of one another.
This project ships with an example unit and e2e test suite.
npm run example-unit-tests
npm run example-e2e-tests