node package manager

stryker

The extendable JavaScript mutation testing framework

Build Status NPM Node version Gitter

Stryker

Stryker

Professor X: For someone who hates mutants... you certainly keep some strange company.
William Stryker: Oh, they serve their purpose... as long as they can be controlled.

Getting started

Stryker is a mutation testing framework for JavaScript. It allows you to test your tests by temporarily inserting bugs.

To install Stryker, execute the command:

$ npm install stryker stryker-api --save-dev

Note: During installation you may run into errors caused by node-gyp. It is safe to ignore them.

To test if Stryker is installed correctly, execute the command:

$ node_modules/.bin/stryker --version

This should print the latest version of Stryker.

Usage

$ node_modules/.bin/stryker <command> [options] [stryker.conf.js]

The only command currently available is run, which kicks off mutation testing.

By default, we expect a stryker.conf.js file in the current working directory. This can be overridden by specifying a different file as the last parameter.

The following is an example stryker.conf.js file:

module.exports = function(config){
  config.set({
    files: ['test/helpers/**/*.js', 'test/unit/**/*.js', { pattern: 'src/**/*.js', included: false, mutated: true }],
    testFramework: 'mocha',
    testRunner: 'mocha',
    reporter: ['progress', 'clear-text', 'dots', 'html', 'event-recorder'],
    coverageAnalysis: 'perTest',
    plugins: ['stryker-mocha-runner', 'stryker-html-reporter']
  });
}

As you can see, the config file is not a simple JSON file, it should be a common js (a.k.a. npm) module.
You might recognize this way of working from the karma test runner.

Make sure you at least specify the files and the testRunner options when mixing the config file and/or command line options.

Supported mutators

See our website for the list of currently supported mutators.

Configuration

All configuration options can either be set via the command line or via the stryker.conf.js config file.

files and mutate both support globbing expressions using node glob. This is the same globbing format you might know from Grunt or Karma.
You can ignore files by adding an exclamation mark (!) at the start of an expression.

Files required to run your tests

Command line: [--files|-f] node_modules/a-lib/**/*.js,src/**/*.js,a.js,test/**/*.js
Config file: files: ['{ pattern: 'src/**/*.js', mutated: true }, '!src/**/index.js', 'test/**/*.js']
Default value: none
Mandatory: yes
Description:
With files you specify all files needed to run your tests. If the test runner you use already provides the test framework (Jasmine, Mocha, etc.), you should not include those files here as well.
The files will be loaded in the other in which they are specified.

When using the command line, the list can only contain a comma separated list of globbing expressions.
When using the config file you can provide an array with strings or InputFileDescriptor objects, like so:

  • string: The globbing expression used for selecting the files needed to run the tests.
  • InputFileDescriptor object: { pattern: 'pattern', included: true, mutated: false }:
    • The pattern property is mandatory and contains the globbing expression used for selecting the files. Using ! to ignore files is not supported here.
    • The included property is optional and determines whether or not this file should be loaded initially by the test-runner (default: true)
    • The mutated property is optional and determines whether or not this file should be targeted for mutations (default: false)

Note: To include a file/folder which start with an exclamation mark (!), use the InputFileDescriptor syntax.

Source code files to mutate

Command line: [--mutate|-m] src/**/*.js,a.js Config file: mutate: ['src/**/*.js', 'a.js']
Default value: none
Mandatory: no
Description:
With mutate you configure the subset of files to use for mutation testing. Generally speaking, these should be your own source files.
This is optional, as you can also use the mutated property with the files parameter or not mutate any files at all to perform a dry-run (test-run).
We expect a comma separated list of globbing expressions, which will be used to select the files to be mutated.

Test runner

Command line: --testRunner karma
Config file: testRunner: 'karma'
Default value: none
Mandatory: yes
Description:
With testRunner you specify the test runner to run your tests. This option is required.
Make sure the test runner plugin for Stryker is installed. E.g. we need the stryker-karma-runner to use karma as a test runner. See the list of plugins for an up-to-date list of supported test runners and plugins.

Test framework

Command line: --testFramework jasmine
Config file: testFramework: 'jasmine'
Default value: none
Mandatory: yes
Description:
With testFramework you configure which test framework your tests are using. This value is directly consumed by the test runner and therefore depends what framework that specific test runner supports. By default, this value is also used for testFramework.

Type of coverage analysis

Full notation: --coverageAnalysis perTest
Config file key: coverageAnalysis: 'perTest'
Default value: perTest
Mandatory: no
Description:
With coverageAnalysis you specify which coverage analysis strategy you want to use.
Stryker can analyse code coverage results. This can potentially speed up mutation testing a lot, as only the tests covering a particular mutation are tested for each mutant. This does not influence the resulting mutation testing score. It only improves performance, so we enable it by default.

The possible values are:

  • off: Stryker will not determine the code covered by tests during the initial test run phase. All tests will be executed for each mutant during the mutation testing phase.

  • all: Stryker will determine the code covered by all tests during the initial test run phase. Only mutants actually covered by your test suite are tested during the mutation testing phase. This setting requires your test runner to be able to report the code coverage back to Stryker. Currently, only the stryker-mocha-runner and the stryker-karma-runner do this.

  • perTest: Stryker will determine the code covered by your test per executed test during the initial test run phase. Only mutants actually covered by your test suite are tested during the mutation testing phase. Only the tests that cover a particular mutant are tested for each one. This requires your tests to be able to run independently of each other and in random order. In addition to requiring your test runner to be able to report the code coverage back to Stryker, your chosen testFramework also needs to support running code before and after each test, as well as test filtering.
    Currently, stryker-mocha-runner as well as stryker-karma-runner support this. However, stryker-karma-runner support is limited to using it with Jasmine as the test framework (Mocha is not yet supported).

Reporters

Command line: --reporter clear-text,progress,dots
Config file: reporter: ['clear-text', 'progress', 'dots']
Default value: ['clear-text', 'progress']
Mandatory: no
Description:
With reporter you can set a reporter or group of reporters for stryker to use. These reporters can be used out of the box: clear-text, progress and event-recorder. By default clear-text and progress are active if no reporter is configured. You can load additional plugins to get more reporters. See stryker-mutator.github.io for an up-to-date list of supported reporter plugins and a description on each reporter.

The clear-text reporter supports an additional config option to show more tests that were executed to kill a mutant. The config for your config file is: clearTextReporter: { maxTestsToLog: 3 },

Plugins

Command line: --plugins stryker-html-reporter,stryker-karma-runner
Config file: plugins: ['stryker-html-reporter', 'stryker-karma-runner']
Default value: ['stryker-*']
Mandatory: no
Description:
With plugins you can add additional Node modules for Stryker to load (or require). By default, all node_modules starting with stryker- will be loaded, so you would normally not need to specify this option. These modules should be installed right next to stryker. For a current list of plugins, you can consult npm or stryker-mutator.github.io.

Start of port range for test runners

Command line: --port 9234
Config file: port: 9234
Default value: 9234
Mandatory: no
Description:
With port you specify the first port to pass on to the test runner to use. Any additional test runners will be spawned using ports n+1, n+2, etc. For example, when you set to use port 9234 and Stryker decides to start four test runner processes, ports 9234, 9235, 9236 and 9237 will be passed to the test runner.
If the test runner decides to use the port it should be available for use.

Global timeout in milliseconds

Command line: --timeoutMs 5000
Config file: timeoutMs: 5000
Default value: 5000
Mandatory: no
Description:
When Stryker is mutating code, it cannot determine indefinitely whether or not a code mutation results in an infinite loop (see Halting problem). In order to battle infinite loops, a test run gets killed after a certain period. This period is configurable with two settings: timeoutMs and timeoutFactor. To calculate the actual timeout in milliseconds the, following formula is used:

timeoutForTestRunMs = timeOfTheInitialTestRunMs * timeoutFactor + timeoutMs

With timeoutFactor you can configure the allowed deviation relative to the time of a normal test run. Tweak this if you notice that mutants are prone to creating slower code, but not infinite loops. timeoutMs let's you configure an absolute deviation. Use it, if you run Stryker on a busy machine and you need to wait longer to make sure that the code indeed entered an infinite loop.

Timeout factor

Command line: --timeoutFactor 1.5
Config file: timeoutFactor: 1.5
Default value: 1.5
Mandatory: no
Description:
See Timeout in milliseconds.

Number of maximum concurrent test runners

Command line: --maxConcurrentTestRunners 3
Config file: maxConcurrentTestRunners: 3
Default value: number of CPU cores Mandatory: no
Description:
Specifies the maximum number of concurrent test runners to spawn.
Mutation testing is time consuming. By default Stryker tries to make the most of your CPU, by spawning as many test runners as you have CPU cores.
This setting allows you to override this default behavior.

Reasons you might want to lower this setting:

  • Your test runner starts a browser (another CPU-intensive process)
  • You're running on a shared server and/or
  • Your hard disk cannot handle the I/O of all test runners

Log level

Command line: --logLevel info
Config file: logLevel: 'info' Default value: info
Mandatory: no Description:
Set the log4js log level that Stryker uses (default is info). Possible values: fatal, error, warn, info, debug, trace, all and off.
Note: Test runners are run as child processes of the Stryker Node process. All output (stdout) of the testRunner is logged as trace.
Thus, to see logging output from the test runner set the logLevel to all or trace.