Nodeunit

Simple syntax, powerful tools. Nodeunit provides easy async unit testing for node.js and the browser.

• Simple to use
• Just export the tests from a module
• Works with node.js and in the browser.
• Helps you avoid common pitfalls when testing asynchronous code
• Easy to add test cases with setUp and tearDown functions if you wish
• Flexible reporters for custom output, built-in support for HTML and jUnit XML
• Allows the use of mocks and stubs

Contributors

• alexgorbatchev
• alexkwolfe
• azatoth
• kadirpekel
• lambdalisue
• luebken
• orlandov
• Sannis
• sstephenson
• thegreatape
• mmalecki
• and thanks to cjohansen for input and advice on implementing setUp and tearDown functions. See cjohansen's fork.

Also, check out gerad's nodeunit-dsl project, which implements a 'pretty dsl on top of nodeunit'.

More contributor information can be found in the CONTRIBUTORS.md file.

Usage

Here is an example unit test module:

exports.testSomething = function(test){
    test.expect(1);
    test.ok(true, "this assertion should pass");
    test.done();
};

exports.testSomethingElse = function(test){
    test.ok(false, "this assertion should fail");
    test.done();
};

When run using the included test runner, this will output the following:

For more details on the exact setup, see Basic Testing.

Installation

There are two options for installing nodeunit:

1. Clone / download nodeunit from github, then:

   make && sudo make install

2. Install via npm:

   npm install nodeunit -g

API Documentation

Nodeunit wraps the functions available in the node.js assert module:

• ok(value, [message]) - Tests if value is a true value.
• equal(actual, expected, [message]) - Tests shallow, coercive equality with the equal comparison operator ( == ).
• notEqual(actual, expected, [message]) - Tests shallow, coercive non-equality with the not equal comparison operator ( != ).
• deepEqual(actual, expected, [message]) - Tests for deep equality.
• notDeepEqual(actual, expected, [message]) - Tests for any deep inequality.
• strictEqual(actual, expected, [message]) - Tests strict equality, as determined by the strict equality operator ( === )
• notStrictEqual(actual, expected, [message]) - Tests strict non-equality, as determined by the strict not equal operator ( !== )
• throws(block, [error], [message]) - Expects block to throw an error.
• doesNotThrow(block, [error], [message]) - Expects block not to throw an error.
• ifError(value) - Tests if value is not a false value, throws if it is a true value. Useful when testing the first argument, error in callbacks.

Nodeunit also provides the following functions within tests:

• expect(amount) - Specify how many assertions are expected to run within a test. Very useful for ensuring that all your callbacks and assertions are run.
• done() - Finish the current test function, and move on to the next. ALL tests should call this!

Nodeunit aims to be simple and easy to learn. This is achieved through using existing structures (such as node.js modules) to maximum effect, and reducing the API where possible, to make it easier to digest.

Note: Users of old nodeunit versions may remember using ok, equals and same in the style of qunit, instead of the assert functions above. These functions still exist for backwards compatibility, and are simply aliases to their assert module counterparts.

Basic Testing

Tests are simply exported from a module, but they are still run in the order they are defined. Tests can be defined by adding functions, containing a single test to run, to the exported module. These functions take a single argument, which is an object interfacing the nodeunit methods as described in the API Documentation.

Minimalistic test-case

A minimalistic test-case must define a module that exports at least one test function. For the purpose of this example we'll assume a minimal node project layout is maintained. The file layout will be as follows (an example can be found at tvervest/nodeunit-example);

|-- lib           // contains the library source files
|   `-- mylib.js  // contains the actual library we'll be testing
`-- test          // contains the unit tests files
    `-- pow.js  // contains the unit tests for advanced behaviour

lib/mylib.js

/*!
 * Simple mock library file for demo purposes.
 */

module.exports = {
  pow: function(a, b) {
    // should return a to the power of b, but doesn't
    return a * b;
  }
};

test/pow.js

/*!
 * Test cases for the mock library file.
 */
var mylib = require('../lib/mylib'); // load the library we'll be testing

module.exports = {
  '2 ^ 2': function(test) {
    var result = mylib.pow(2, 2);
    test.equal(result, 4); // This will succeed, 2 * 2 == 2 ^ 2
    test.done();
  },

  '3 ^ 2': function(test) {
    var result = mylib.pow(3, 2); // this will fail, 3 * 2 != 3 ^ 2
    test.equal(result, 9);
    test.done();
  }
};

Now we'll run our test by executing nodeunit test from the root directory of our project. This will execute all exported modules in the files in the test directory.

As expected, the first test will succeed, as 2 ^ 2 == 2 * 2. However, the secondary test will fail, as 3 * 2 != 3 ^ 2.

Asynchronous Testing

When testing asynchronous code, there are a number of sharp edges to watch out for. Thankfully, nodeunit is designed to help you avoid as many of these pitfalls as possible. For the most part, testing asynchronous code in nodeunit just works.

Tests run in series

While running tests in parallel seems like a good idea for speeding up your test suite, in practice I've found it means writing much more complicated tests. Because of node's module cache, running tests in parallel means mocking and stubbing is pretty much impossible. One of the nicest things about testing in javascript is the ease of doing stubs:

var _readFile = fs.readFile;
fs.readFile = function(path, callback){
    // its a stub!
};
// test function that uses fs.readFile

// we're done
fs.readFile = _readFile;

You cannot do this when running tests in parallel. In order to keep testing as simple as possible, nodeunit avoids it. Thankfully, most unit-test suites run fast anyway.

Explicit ending of tests

When testing async code its important that tests end at the correct point, not just after a given number of assertions. Otherwise your tests can run short, ending before all assertions have completed. Its important to detect too many assertions as well as too few. Combining explicit ending of tests with an expected number of assertions helps to avoid false test passes, so be sure to use the test.expect() method at the start of your test functions, and test.done() when finished.

Groups, setUp and tearDown

Nodeunit allows the nesting of test functions:

exports.test1 = function (test) {
    ...
}

exports.group = {
    test2: function (test) {
        ...
    },
    test3: function (test) {
        ...
    }
}

This would be run as:

test1
group - test2
group - test3

Using these groups, Nodeunit allows you to define a setUp function, which is run before each test, and a tearDown function, which is run after each test calls test.done():

module.exports = {
    setUp: function (callback) {
        this.foo = 'bar';
        callback();
    },
    tearDown: function (callback) {
        // clean up
        callback();
    },
    test1: function (test) {
        test.equals(this.foo, 'bar');
        test.done();
    }
};

In this way, its possible to have multiple groups of tests in a module, each group with its own setUp and tearDown functions.

Running Tests

Nodeunit comes with a basic command-line test runner, which can be installed using 'sudo make install'. Example usage:

nodeunit testmodule1.js testfolder [...]

The default test reporter uses color output, because I think that's more fun :) I intend to add a no-color option in future. To give you a feeling of the fun you'll be having writing tests, lets fix the example at the start of the README:

Ahhh, Doesn't that feel better?

When using the included test runner, it will exit using the failed number of assertions as the exit code. Exiting with 0 when all tests pass.

Command-line Options

• --reporter FILE - you can set the test reporter to a custom module or on of the modules in nodeunit/lib/reporters, when omitted, the default test runner is used.
• --list-reporters - list available build-in reporters.
• --config FILE - load config options from a JSON file, allows the customisation of color schemes for the default test reporter etc. See bin/nodeunit.json for current available options.
• --version or -v - report nodeunit version
• --help - show nodeunit help

Running tests in the browser

Nodeunit tests can also be run inside the browser. For example usage, see the examples/browser folder. The basic syntax is as follows:

test.html

<html>
  <head>
    <title>Example Test Suite</title>
    <link rel="stylesheet" href="nodeunit.css" type="text/css" />
    <script src="nodeunit.js"></script>
    <script src="suite1.js"></script>
    <script src="suite2.js"></script>
  </head>
  <body>
    <h1 id="nodeunit-header">Example Test Suite</h1>
    <script>
      nodeunit.run({
        'Suite One': suite1,
        'Suite Two': suite2
      });
    </script>
  </body>
</html>

Here, suite1 and suite2 are just object literals containing test functions or groups, as would be returned if you did require('test-suite') in node.js:

suite1.js

this.suite1 = {
    'example test': function (test) {
        test.ok(true, 'everything is ok');
        test.done();
    }
};

If you wish to use a commonjs format for your test suites (using exports), it is up to you to define the commonjs tools for the browser. There are a number of alternatives and its important it fits with your existing code, which is why nodeunit does not currently provide this out of the box.

In the example above, the tests will run when the page is loaded.

The browser-version of nodeunit.js is created in dist/browser when you do, 'make browser'. You'll need UglifyJS installed in order for it to automatically create nodeunit.min.js.

Adding nodeunit to Your Projects

If you don't want people to have to install the nodeunit command-line tool, you'll want to create a script that runs the tests for your project with the correct require paths set up. Here's an example test script, that assumes you have nodeunit in a suitably located node_modules directory.

#!/usr/bin/env node
var reporter = require('nodeunit').reporters.default;
reporter.run(['test']);

If you're using git, you might find it useful to include nodeunit as a submodule. Using submodules makes it easy for developers to download nodeunit and run your test suite, without cluttering up your repository with the source code. To add nodeunit as a git submodule do the following:

git submodule add git://github.com/caolan/nodeunit.git node_modules/nodeunit

This will add nodeunit to the node_modules folder of your project. Now, when cloning the repository, nodeunit can be downloaded by doing the following:

git submodule init
git submodule update

Let's update the test script above with a helpful hint on how to get nodeunit, if its missing:

#!/usr/bin/env node
try {
    var reporter = require('nodeunit').reporters.default;
}
catch(e) {
    console.log("Cannot find nodeunit module.");
    console.log("You can download submodules for this project by doing:");
    console.log("");
    console.log("    git submodule init");
    console.log("    git submodule update");
    console.log("");
    process.exit();
}

process.chdir(__dirname);
reporter.run(['test']);

Now if someone attempts to run your test suite without nodeunit installed they will be prompted to download the submodules for your project.

Built-in Test Reporters

• default - The standard reporter seen in the nodeunit screenshots
• minimal - Pretty, minimal output, shows errors and progress only
• html - Outputs a HTML report to stdout
• junit - Creates jUnit compatible XML reports, which can be used with continuous integration tools such as Hudson.
• machineout - Simple reporter for machine analysis. There is nodeunit.vim which is useful for TDD on VIM.

Writing a Test Reporter

Nodeunit exports runTest(name, fn, options), runModule(name, mod, options) and runFiles(paths, options). You'll most likely want to run test suites from files, which can be done using the latter function. The options argument can contain callbacks which run during testing. Nodeunit provides the following callbacks:

• moduleStart([name]) - called before a module is tested, if available a module name will be passed
• moduleDone(name, assertions) - called once all test functions within the module have completed (see assertions object reference below) ALL tests within the module
• testStart(name) - called before a test function is run
• testReady(test) - called before a test function is run with the test object that will be passed to the test function
• testDone(name, assertions) - called once a test function has completed (by calling test.done())
• log(assertion) - called whenever an assertion is made (see assertion object reference below)
• done(assertions) - called after all tests/modules are complete

The assertion object:

• passed() - did the assertion pass?
• failed() - did the assertion fail?
• error - the AssertionError if the assertion failed
• method - the nodeunit assertion method used (ok, same, equals...)
• message - the message the assertion method was called with (optional)

The assertionList object:

• An array-like object with the following new attributes:
  • failures() - the number of assertions which failed
  • duration - the time taken for the test to complete in msecs

For a reference implementation of a test reporter, see lib/reporters/default.js in the nodeunit project directory.

Sandbox utility

This is a function which evaluates JavaScript files in a sandbox and returns the context. The sandbox function can be used for testing client-side code or private un-exported functions within a module.

var sandbox = require('nodeunit').utils.sandbox;
var example = sandbox('example.js');

sandbox(files, sandbox) - Evaluates JavaScript files in a sandbox, returning the context. The first argument can either be a single filename or an array of filenames. If multiple filenames are given their contents are concatenated before evaluation. The second argument is an optional context to use for the sandbox.

Note: When working with the sandbox if your script depends on outside sources (i.e. using require) then you will want to pass that into the optional context when setting up the sandbox.

var sandbox = require('nodeunit').utils.sandbox;
// pass in some node globals
var box_globals = {
    // Passing module.exports into the sandbox will give your code  access to it.
    module: {exports: exports},
    // Passing require into the sandbox will give your code  access to use it AND 
    // will share the cache with modules already required from outside the sandbox.
    require: require,
    // Passing console into the sandbox will give your code access to it
    console: console
};
var example = sandbox('example.js', box_globals);

Running the nodeunit Tests

The tests for nodeunit are written using nodeunit itself as the test framework. However, the module test-base.js first does some basic tests using the assert module to ensure that test functions are actually run, and a basic level of nodeunit functionality is available.

To run the nodeunit tests do:

make test

Note: There was a bug in node v0.2.0 causing the tests to hang, upgrading to v0.2.1 fixes this.

machineout reporter

The default reporter is really readable for human but for machinally analysis. When you want to analyze the output of nodeunit, use machineout reporter and you will get

nodeunit with vim

There is nodeunit.vim so you can use nodeunit with VIM.

That compiler uses machineout reporter and it is useful to use with vim-makegreen.

Contributing

Contributions to the project are most welcome, so feel free to fork and improve. When submitting a pull request, please run make lint first to ensure we're following a consistent coding style.