PhantomJS Runners for Mocha
The mocha-phantomjs project provides a
mocha-phantomjs.coffee script file and extensions to drive PhantomJS while testing your HTML pages with Mocha from the console. The preferred usage is to install
mocha-phantomjs via node's packaged modules and use the
mocha-phantomjs binary wrapper. Tested with Mocha 1.12.x, Chai 1.7.x, and PhantomJS 1.9.1.
process.stdout.write, done right. Mocha is primarily written for node, hence it relies on writing to standard out without trailing newline characters. This behavior is critical for reporters like the dot reporter. We make up for PhantomJS's lack of stream support by both customizing
console.log and creating a
process.stdout.write function to the current PhantomJS process. This technique combined with a handful of fancy ANSI cursor movement codes allows PhantomJS to support Mocha's diverse reporter options.
Proper exit status codes from PhantomJS using Mocha's failures count. So in standard UNIX fashion, a
0 code means success. This means you can use mocha-phantomjs on your CI server of choice.
You can use your existing Mocha HTML file reporters side by side with mocha-phantomjs. This gives you the option to run your tests both in a browser or with PhantomJS. Since mocha-phantomjs needs to control when the
run() command is sent to the mocha object, we accomplish this by setting the
mochaPhantomJS on the
window object to
true. Below, in the usage section, is an example of a HTML structure that can be used both by opening the file in your browser or choice or using mocha-phantomjs.
We distribute mocha-phantomjs as an npm that is easy to install. Once done, you will have a
mocha-phantomjs binary. See the next usage section for docs or use the
Since 3.4, we now declare phantomjs as a peer dependency, and it will be installed adjacent to
mocha-phantomjs automatically. You may use
-p to provide an explicit path to phantomjs, or call phantomjs directly yourself via
phantomjs lib/mocha-phantomjs.coffee <page> <reporter> <config-as-JSON>. The later approach is recommended for build system plugins to avoid another process fork.
Usage: mocha-phantomjs [options] pageOptions:-h, --help output usage information-V, --version output the version number-R, --reporter <name> specify the reporter to use-f, --file <filename> specify the file to dump reporter output-t, --timeout <timeout> specify the test startup timeout to use-b, --bail exit on the first test failure-A, --agent <userAgent> specify the user agent to use-c, --cookies <Object> phantomjs cookie object-h, --header <name>=<value> specify custom header-k, --hooks <path> path to hooks module-s, --setting <key>=<value> specify specific phantom settings-v, --view <width>x<height> specify phantom viewport size-C, --no-color disable color escape codes-p, --path <path> path to PhantomJS binaryExamples:$ mocha-phantomjs -R dot /test/file.html$ mocha-phantomjs$ mocha-phantomjs -s localToRemoteUrlAccessEnabled=true -s webSecurityEnabled=false$ mocha-phantomjs -p ~/bin/phantomjs /test/file.html
Now as an node package, using
mocha-phantomjs has never been easier. The page argument can be either a local or fully qualified path or a http or file URL.
--reporter may be a built-in reporter or a path to your own reporter (see below). See phantomjs WebPage settings for options that may be supplied to the
Your HTML file's structure should look something like this. The reporter set below to
html is only needed for viewing the HTML page in your browser. The
mocha-phantomjs.coffee script overrides that reporter value. The conditional run at the bottom allows the mixed mode feature described above.
Mocha-phantomjs supports creating screenshots from your test code. For example, you could write a function like below into your test code.
if windowcallPhantomvar date =var filename = "screenshots/" + dategetTimeconsole.log"Taking screenshot " + filenamecallPhantom'screenshot': filename
If you want to generate a screenshot for each test failure you could add the following into your test code.
afterEachif thiscurrentTeststate == 'failed'takeScreenshot
Mocha-phantomjs does not scrap the web page under test! No other PhantomJS driver stacks up to our runner support. Some have used a debounce method to keep duplicate messages in the spec reporter from showing up twice. Loosing one of Mocha's console reporters neatest features, initial test start feedback. The animation below is an example of how our runner script fully compiles with expected Mocha behavior.
The following is a list of tested reporters. Since moving PhantomJS 1.9.1, most core Mocha reporters should "just work" since we now support stdout properly. Reporters with node dependencies will not work, like
html-cov. If you have an issue with a reporter, open a github issue and let me know.
The default reporter. You can force it using
spec for the reporter argument.
dot for the reporter argument.
The PhantomJS process has no way of knowing anything about your console window's width. So we default the width to 75 columns. But if you pass down the
COLUMNS environment variable, it will pick that up and adjust to your current terminal width. For example, using the
$COLUMNS variable like so.
env COLUMNS=$COLUMNS phantomjs mocha-phantomjs.coffee URL dot
Bundled and tested reporters include:
require: You can only require other reporters, like
require('./base')to build off of the BaseReporter
module: Export your reporter class as normal
process.stdout.writepreferrably to support the
Simple! Just clone the repo, then run
npm install and the various node development dependencies will install to the
node_modules directory of the project. If you have not done so, it is typically a good idea to add
/node_modules/.bin to your
$PATH so these modules bins are used. Now run
npm test to start off the test suite.
We also use Travis CI to run our tests too. The current build status:
Released under the MIT license. Copyright (c) 2012 Ken Collins.