@knappi/grunt-saucelabs

grunt-saucelabs
---------------------



 

A Grunt task for running QUnit, Jasmine, Mocha or any framework using Sauce Labs' Cloudified Browsers.

Grunt is a task-based command line build tool for JavaScript projects, based on nodejs.
QUnit is a powerful, easy-to-use JavaScript unit test suite used by the jQuery, jQuery UI and jQuery Mobile projects and is capable of testing any generic JavaScript code, including itself!
Jasmine is a behavior-driven development framework for testing JavaScript code.
Mocha is a JavaScript test framework for running serial asynchronous tests.
Sauce Labs offers browser environments on the cloud for testing code.

About the tool
--------------
The grunt-contrib-qunit task runs QUnit based test suites on PhantomJS.
The saucelabs-qunit task is very similar but runs the test suites on the cloudified browser environment provided by Sauce Labs. This ensures that subject of the test runs across different browser environment.
The task also uses Sauce Connect to establish a tunnel between Sauce Labs browsers and the machine running Grunt to load local pages. This is typically useful for testing pages on localhost that are not publicly accessible on the internet.
The saucelabs-jasmine runs Jasmine tests in the Sauce Labs browser. The saucelabs-jasmine task requires jasmine-1.3.0. There is also a saucelabs-mocha task that lets you run your Mocha tests on Sauce Labs cloudified browser environment.

Usage
------
This task is available as a node package and can be installed as npm install grunt-saucelabs. It can also be included as a devDependency in package.json in your node project.

To use the task in grunt.js, load the npmTask.

``javascript
grunt.loadNpmTasks('grunt-saucelabs');

`

In the grunt.initConfig, add the configuration that looks like the following

`javascript
var request = require('request');
...
'saucelabs-qunit': {
all: {
options: {
username: 'saucelabs-user-name', // if not provided it'll default to ENV SAUCE_USERNAME (if applicable)
key: 'saucelabs-key', // if not provided it'll default to ENV SAUCE_ACCESS_KEY (if applicable)
urls: ['www.example.com/qunitTests', 'www.example.com/mochaTests'],
build: process.env.CI_BUILD_NUMBER,
testname: 'Sauce Unit Test for example.com',
browsers: [{
browserName: 'firefox',
version: '19',
platform: 'XP'
}]
// optionally, he browsers param can be a flattened array:
// [["XP", "firefox", 19], ["XP", "chrome", 31]]
}
}
}

`

The configuration of saucelabs-jasmine, saucelabs-mocha and saucelabs-custom are exactly the same.
Note the options object inside a grunt target. This was introduced in grunt-saucelabs-* version 4.0.0 to be compatible with grunt@0.4.0


Full list of parameters which can be added to a saucelabs-* task:

* __username__ : The Sauce Labs username that will be used to connect to the servers. If not provided, uses the value of SAUCE_USERNAME environment variable. _Optional_
* __key__ : The Sauce Labs secret key. Since this is a secret, this should not be checked into the source code and may be available as an environment variable. Grunt can access this using process.env.saucekey. Will also default to SAUCE_ACCESS_KEY environment variable. _Optional_
* __urls__: An array or URLs that will be loaded in the browsers, one after another. Since SauceConnect is used, these URLs can also be localhost URLs that are available using the server task from grunt. _Required_
* __build__: The build number for this test. _Optional_
* __testname__: The name of this test, displayed on the Sauce Labs dashboard. _Optional_
* __tags__: An array of strings, to be added as tags to the test on Sauce Labs. _Optional_
* __tunneled__: Defaults to true; Won't launch a Sauce Connect tunnel if set to false. _Optional_
* __tunnelArgs__: Array of optional arguments to be passed to the Sauce Connect tunnel. example: ['--debug', '--direct-domains', 'google.com']. See here for further documentation.
* __sauceConfig__: Map of extra parameters to be passed to sauce labs. example: {'video-upload-on-pass': false, 'idle-timeout': 60}. See here for further documentation.
* __pollInterval__: Number of milliseconds between each retry to see if a test is completed or not (default: 2000). _Optional_
__statusCheckAttempts__: Number of times to attempt to see if a test is completed or not (default: 90). Effectively, your tests have statusCheckAttempts pollInterval seconds to complete (Thus, 180s by default). Set to -1 to try forever.
* __throttled__: Maximum number of unit test pages which will be sent to Sauce Labs concurrently. Exceeding your Sauce Labs' allowed concurrency can lead to test failures if you have a lot of unit test pages. _Optional_
* __max-duration__: Maximum duration of a test, this is actually a Selenium Capability. Sauce Labs defaults to 180 seconds for js unit tests. _Optional_
* __browsers__: An array of objects representing the various browsers on which this test should run. _Optional_
* __onTestComplete__: A callback that is called every time a unit test for a page is complete. Runs per page, per browser configuration. Receives two arguments (result, callback). result is the javascript object exposed to sauce labs as the results of the test. callback must be called, node-style (having arguments err, result where result is a true/false boolean which sets the test result reported to the command line). See example below _Optional_
* __maxRetries__: Specifies how many times the timed out tests should be retried (default: 0). _Optional_
* __public__: The job visibility level. Defaults to 'team'. _Optional_

A typical test task running from Grunt could look like grunt.registerTask('test', ['server', 'qunit', 'saucelabs-qunit']); This starts a server and then runs the QUnit tests first on PhantomJS and then using the Sauce Labs browsers.

Exposing Test Results to the Sauce Labs API
-------------------------------------------
Since this project uses the Sauce Labs js unit test API, the servers at Sauce Labs need a way to get the results of your test. Follow the instructions below to assure that the results of your tests are delivered properly.

$3

You can make Job Details pages more informative on Sauce by providing more data with each test. You will get info about each test run inside your suite directly on Sauce pages.

![Jasmine detailed results](https://saucelabs.com/docs/javascript-unit-tests-integration)

You can do that by using Jasmine JS Reporter that will let saucelabs-jasmine task provide in-depth data about each test as a JSON object.

All you need to do is to include the new jasmine-jsreporter reporter to the page running Jasmine tests by adding new script in header:
`html

`
and telling Jasmine to use it:
`javascript
jasmineEnv.addReporter(new jasmine.JSReporter());
``

$3

Add the following to your QUnit test specification
`javascript
var log = [];
var testName;

QUnit.done(function (test_results) {
var tests = [];
for(var i = 0, len = log.length; i < len; i++) {
var details = log[i];
tests.push({
name: details.name,
result: details.result,
expected: details.expected,
actual: details.actual,
source: details.source
});
}
test_results.tests = tests;

window.global_test_results = test_results;
});
QUnit.testStart(function(testDetails){
QUnit.log(function(details){
if (!details.result) {
details.name = testDetails.name;
log.push(details);
}
});
});
`

$3

Add the following to the mocha test page html. Make sure you remove any calls to `mocha.checkLeaks()` or add `mochaResults` to the list of globals.
`html

`

OnTestComplete callback
-----------------------
An optional parameter to the grunt task is OnTestComplete, a callback which is called at the end of every test, before results are logged to the console.
You can use this callback to intercept results from SauceLabs and re-report the results (or use the information for your own purposes)

Receives two arguments (result, callback). result is the javascript object exposed to sauce labs as the results of the test. callback must be called, node-style (having arguments err, result where result is a true/false boolean which sets the test result reported to the command line)

When running the tests for this project, we need to test the case where a test fails on Sauce. In this case, we want to record a test Failure as a Success for us.

`
'saucelabs-qunit': {
all: {
options: {
username: 'saucelabs-user-name', // if not provided it'll default to ENV SAUCE_USERNAME (if applicable)
key: 'saucelabs-key', // if not provided it'll default to ENV SAUCE_ACCESS_KEY (if applicable)
urls: ['www.example.com/qunitTests', 'www.example.com/mochaTests'],
build: process.env.CI_BUILD_NUMBER,
testname: 'Sauce Unit Test for example.com',
browsers: [{
browserName: 'firefox',
version: '19',
platform: 'XP'
}],
onTestComplete: function(result, callback) {
// Called after a unit test is done, per page, per browser
// 'result' param is the object returned by the test framework's reporter
// 'callback' is a Node.js style callback function. You must invoke it after you
// finish your work.
// Pass a non-null value as the callback's first parameter if you want to throw an
// exception. If your function is synchronous you can also throw exceptions
// directly.
// Passing true or false as the callback's second parameter passes or fails the
// test. Passing undefined does not alter the test result. Please note that this
// only affects the grunt task's result. You have to explicitly update the Sauce
// Labs job's status via its REST API, if you want so.

// The example below negates the result, and also updates the Sauce Labs job's status
var user = process.env.SAUCE_USERNAME;
var pass = process.env.SAUCE_ACCESS_KEY;
request.put({
url: ['https://saucelabs.com/rest/v1', user, 'jobs', result.job_id].join('/'),
auth: { user: user, pass: pass },
json: { passed: !result.passed }
}, function (error, response, body) {
if (error) {
callback(error);
} else if (response.statusCode !== 200) {
callback(new Error('Unexpected response status'));
} else {
callback(null, !result.passed);
}
});
}
}
}
}
`

Examples
--------
Some projects that use this task are as follows. You can take a look at their GruntFile.js for sample code

* This project
* WinJS
* Jquery-IndexedDB
* IndexedDBShim

If you have a project that uses this plugin, please add it to this list and send a pull request.


Integration with a CI system
----------------------------
Grunt tasks are usually run alongside a continuous integration system. For example, when using Travis, adding the following lines in the package.json ensures that the task is installed with npm install is run. Registering Sauce Labs in test task using grunt.registerTask('test', ['server', 'saucelabs-qunit']); ensures that the CI environment runs the tests using npm test.
To secure the Sauce Key, the CI environment can be configured to provide the key as an environment variable instead of specifying it file. CI Environments like Travis provide ways to add secure variables in the initial configuration.
The IndexedDBShim is a project that uses this plugin in a CI environment. Look at the .travis.yml and the grunt.js for usage example.

Changelog
---------
#### 9.0.0 ####
* removed support for YUI Test (Yahoo deprecated the framework a year ago). It will still work but we are not planning support for it moving forward
* update Sauce Connect to version 4.3.16
* updated dependencies (grunt not updated because we are waiting on grunt-sauce-tunnel)

#### 8.6.3 ####
* updated some dependencies
* support for Grunt 1.0.0
* fix some broken tests

#### 8.6.2 ####
* updated Sauce Connect to v4.3.12

#### 8.6.1 ####
* added public parameter, so tests can be made Public on Sauce Labs
* when retrying a test, browser name and test url are output to stdout
* default build number added when running tests locally
* update Sauce Connect to v4.3.7\
* updated npm dependencies
* fixed a bug where YUI or Qunit tests were detected as Passed, when they actually failed

#### 8.6.0 ####
* check job completion a maximum number of times
* added a config setting, statusCheckAttempts which defaults to 90
* updated to Sauce Connect v4.3.6
* update dependencies

#### 8.5.0 ####
* fix ECONNRESET errors caused by network connectivity issues
* better error logging

#### 8.4.1 ####
* updated sauce-tunnel to v2.1.1 (and therefore Sauce Connect to v4.3.5)

#### 8.4.0 ####
* polling a job for its status now retries
* add maxPollRetries parameter, which specifies the number of status-requests to make before giving up on a job

#### 8.3.3 ####
* improvements to README
* add a 'tags' parameter to the config, to allow setting tags on tests
* optionally, you can set "name" and "tags" onto the "browser" object and they'll be applied to the corresponding job

#### 8.3.2 ####
* fixed a bug where DELETE commands which errored on the Sauce side caused test execution to halt

#### 8.3.0 ####
* add testPageUrl to result object. Indicates the url which was the target of the test.

#### 8.2.2 ####
* browsers param can optionally be an array identical to the one used by the Sauce API. ex: ["XP", "firefox", "19"]

#### 8.2.1 ####
* update dependencies

#### 8.2.0 ####
* upgrade to sauce-tunnel 2.1 (SC 3.4)

#### 8.1.1 ####
* better detection and handling or errors which happen on Sauce Labs

#### 8.1.0 ####
* added retry logic, thanks again to @gvas, now you can use maxRetries parameter to automatically retry tests which fail

#### 8.0.3 ####
* fixed bug, sauce job urls displayed in log properly again

#### 8.0.2 ####
* fixed bug, testname option not working

#### 8.0.1 ####
* Major refactor, thanks to all the work done by @gvas
* async onTestComplete callback fixed. Now the callback is passed two args: result, callback. callback is a node style callback (err, result);
* /examples directory added, while the actual tests and Gruntfile are now more complicated (and useful)

#### 7.0.0 ####
throttled parameter now represents the max number of jobs sent concurrently. Previously was throttled browsers.length

#### 6.0.0 ####
* default testInterval changed to 2000ms
* added max-duration, sauceConfig, and sauceTunnel params

#### 5.1.3 ####
* update to sauce-tunnel 2.0.6, which uses Sauce Connect 4.2
* queued job throttling added

#### 5.1.2 ####
* use sauce-tunnel-sc3-1 to protect against heartbleed bug

#### 5.1.1 ####
* Qunit reporting code made ecma3 compatible
* Qunit reporting code doesn't clober the .done() callback
* Updated dependencies

#### 5.1.0 ####
* Added custom` framework
* Updated the test reporting on example pages to provide details when tests fail