rappidjs component test is a test runner for executing web tests fast within the browser. It contains also a grunt task to run those web tests on phantomjs.
rAppid.js makes it easy to hide complexity with components. Most of the components are visual components that contain beside the business logic (which can be tested with unit tests) also an user interface and a data model. As rich applications are made of a complex hierarchy of components that need to play well together also the components itself should be tested.
To make this happen rappidjs component test creates a test runner, which is basically an a web page with an iframe in it. For every test group a test application containing the component is bootstrapped within the iframe and the application and the component is made available on the window object. Then the test specification is loaded (bdd) and executed via mocha within the iframe. As the test has access to the component and also to the document
- it can either change the internal state of the component and check the visual result or
- interacts with the ui and checks the state of the component
A special test reporter with mocha collects the test results and provides it to the surrounding page which finally displays the result of all tests. The test runner also runs on phantomjs, so that the tests can executed headless and fits into a build pipeline.
The interaction of with the ui is done via jQuery, the assertion of the component state with chai and the check of the visual representation via jquery.expect.
If you haven't used Grunt before, be sure to check out the Getting Started guide, as it explains how to create a Gruntfile as well as install and use Grunt plugins.
Once you're familiar with that process, you may install this plugin with this command:
npm install rappidjs-component-test --save-dev
Once the plugin has been installed, it may be enabled inside your Gruntfile with this line of JavaScript:
grunt.loadNpmTasks('rappidjs-component-test');
This is an example configuration
grunt.initConfig({
componentTest: {
options: {
// global options
port: 8080,
dir: 'webtest'
},
dev: {
runOnPhantom: false // run it interactive
},
test: {
runOnPhantom: true // run it headless
}
}
By default all tests and test application within the directory webtest
in the
root folder of the project. You can control the directory with the dir
option
in the grunt task.
Create a groups.json file within the directory with the following structure:
{
"all": [
{
"app": "AddressComponent.js",
"test": "AddressComponentTests.js"
}
],
"testgroup1": [
{
"app": "AddressComponent.js",
"test": "AddressComponentTests.js"
},
// ... more apps
],
// ... more groups
}
As components can relay on injections, you can create a base TestApplication which sets up the injections. This is an optional step.
Here you see an BaseTest.xml application inheriting from TestApplication, which sets up I18n and an GeolocationService. You can also use mocks here.
<test:TestApplication xmlns:test="test" xmlns:js="js.core" xmlns:service="checkout.service">
<js:Injection>
<js:I18n />
<service:GeolocationService />
</js:Injection>
</test:TestApplication>
Since the test runner need to bootstrap an application, for every component you like to
test you need to create an application. This process is simplified by the
TestApplication
we're inheriting from. See the following test application
define(["xaml!BaseTest", "sprd/model/Address"], function(BaseTest, Address) {
return BaseTest.inherit({
componentClass: "checkout/view/Address",
setupComponent: function(component) {
component.set("address", new Address());
}
});
});
the componentClass (not in the defaults section!) defines which component will be created
and added during the start phase of the application. The setupComponent
method will be
invoked after the component has been created and is added. If you need an async version of
if use the following signature setupComponent: function(component, callback)
and invoke
the callback when you're done.
The component test uses BDD syntax. Chaijs, jQuery and jQuey.expect are available during the test, so you can change the UI, check the model or change the model and check the UI.
define(["sprd/model/Address", "chai"], function(Address, chai) {
var assert = chai.assert;
var map = {
firstname: ".firstname input",
lastname: ".lastname input"
};
function $(what) {
return jQuery(map[what] || what);
}
describe("Address", function () {
beforeEach(function() {
component.set("address", new Address());
});
it("should have empty fields", function () {
$expect($("firstname")).to.have.val('');
$expect($("lastname")).to.have.val('');
});
it("should show the person details", function () {
component.$.address.$.person.set({
firstName: "first",
lastName: "last"
});
$expect($("firstname")).to.have.val("first");
$expect($("lastname")).to.have.val("last");
});
it("changes in the view should go back into the model", function () {
assert.isNull(component.get("address.person.salutation"));
// click on Mr
$(".salutations input[value='1']").click();
assert.equal(component.get("address.person.salutation"), "1");
});
});
});
MIT