stryker-mutator
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.
Introduction
For an introduction to mutation testing and Stryker's features, see stryker-mutator.io.
Getting started
Please follow the quickstart on the website.
For small js projects, you can try the following command:
npm install --save-dev stryker stryker-api
# Only for small projects:
npx stryker runIt will run stryker with default values:
- Uses
npm testas your test command - Searches for files to mutate in the
libandsrcdirectories
Usage
$ npx stryker <command> [options] [stryker.conf.js]The main command for Stryker is run, which kicks off mutation testing.
Although Stryker can run without any configuration, it is recommended to configure it when you can, as it can greatly improve performance of the mutation testing process. By default, Stryker will look for a stryker.conf.js file in the current working directory (if it exists). This can be overridden by specifying a different file as the last parameter.
Before your first run, we recommend you try the init command, which helps you to set up this stryker.conf.js file and install any missing packages needed for your specific configuration. We recommend you verify the contents of the configuration file after this initialization, to make sure everything is setup correctly. Of course, you can still make changes to it, before you run Stryker for the first time.
The following is an example stryker.conf.js file. It specifies running mocha tests with the mocha test runner.
module.exports = function(config){
config.set({
mutate: [
'src/**/*.js',
'!src/index.js'
],
testFramework: 'mocha',
testRunner: 'mocha',
reporters: ['progress', 'clear-text', 'html'],
coverageAnalysis: 'perTest'
});
}As you can see, the config file is not a simple JSON file. It should be a node module. You might recognize this way of working from the karma test runner.
Make sure you at least specify the testRunner options when mixing the config file and/or command line options.
Command-line interface
Stryker can also be installed, configured and run using the Stryker-CLI. If you plan on using Stryker in more projects, the Stryker-CLI is the easiest way to install, configure and run Stryker for your project.
You can install the Stryker-CLI using:
$ npm install -g stryker-cliThe Stryker-CLI works by passing received commands to your local Stryker installation. If you don't have Stryker installed yet, the Stryker-CLI will help you with your Stryker installation. This method allows us to provide additional commands with updates of Stryker itself.
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.
mutate [string[]]
Default: ['{src,lib}/**/*.js?(x)', '!{src,lib}/**/__tests__/**/*.js?(x)', '!{src,lib}/**/?(*.)+(spec|test).js?(x)', '!{src,lib}/**/*+(Spec|Test).js?(x)']
Command line: [--mutate|-m] src/**/*.js,a.js
Config file: mutate: ['src/**/*.js', 'a.js']
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 choose to not mutate any files at all and perform a dry-run (running only your tests without mutating).
testRunner [string]
Default: 'command'
Command line: --testRunner karma
Config file: testRunner: 'karma'
With testRunner you specify the test runner that Stryker uses to run your tests. The default value is command. The command runner runs a configurable bash/cmd command and bases the result on the exit code of that program (0 for success, otherwise failed). You can configure this command via the config file using the commandRunner: { command: 'npm run mocha' }. It uses npm test as the command by default.
The command test runner can be made to work in any use case, but comes with a performance
penalty, as Stryker cannot do any optimizations and just runs all tests for all mutants.
If possible, you should try to use one of the test runner plugins that hook into your test runner of choice.
For example: install and use 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.
testFramework [string]
Default: none
Command line: --testFramework jasmine
Config file: testFramework: 'jasmine'
Configure which test framework you are using.
This option is not mandatory, as Stryker is test framework agnostic (it doesn't care what framework you use),
However, it is required when coverageAnalysis is set to 'perTest', because Stryker needs to hook into the test framework in order to measure code coverage results per test and filter tests to run.
Make sure the a plugin is installed for your chosen test framework. E.g. install stryker-mocha-framework to use 'mocha' as a test framework.
coverageAnalysis [string]
Default: off
Command line: --coverageAnalysis perTest
Config file: coverageAnalysis: 'perTest'
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-runnerand thestryker-karma-runnerdo 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
testFrameworkalso needs to support running code before and after each test, as well as test filtering.
Currently,stryker-mocha-runneras well asstryker-karma-runnersupport this. However,stryker-karma-runnersupport is limited to using it withJasmineas the test framework (Mochais not yet supported).
mutator [object | string]
Default: es5
Command line: --mutator es5
Config file: mutator: 'es5' or mutator: { name: 'es5', excludedMutations: ['BooleanSubstitution', 'StringLiteral'] }
With mutator you configure which mutator plugin you want to use, and optionally, which mutation types to exclude from the test run.
The mutator plugin name defaults to es5 if not specified. The list of excluded mutation types defaults to an empty array, meaning all mutation types will be included in the test.
The full list of mutation types varies slightly between mutators (for example, the es5 mutator will not use the same mutation types as the typescript mutator). Mutation type names are case-sensitive, and can be found either in the source code or in a generated Stryker report.
When using the command line, only the mutator name as a string may be provided.
When using the config file, you can provide either a string representing the mutator name, or a MutatorDescriptor object, like so:
MutatorDescriptorobject:{ name: 'name', excludedMutations: ['mutationType1', 'mutationType2', ...] }:- The
nameproperty is mandatory and contains the name of the mutator plugin to use. - The
excludedMutationsproperty is mandatory and contains the types of mutations to exclude from the test run.
- The
transpilers [string[]]
Default: []
With transpilers you configure which transpiler plugins should transpile the code before it's executed. This is an array where the transpilers are called in the other of the array. This defaults to an empty array meaning no transpilation will be done.
reporters [string[]]
Default: ['clear-text', 'progress']
Command line: --reporters clear-text,progress,dots,dashboard
Config file: reporters: ['clear-text', 'progress', 'dots', 'dashboard']
With reporters you can set the reporters for stryker to use.
These reporters can be used out of the box: clear-text, progress, dots, dashboard and event-recorder.
By default clear-text and progress are active if no reporters are configured.
You can load additional plugins to get more reporters. See stryker-mutator.io
for an up-to-date list of supported reporter plugins and a description on each reporter.
The clear-text reporter supports three additional config options:
allowColorto use cyan and yellow in printing source file names and positions. This defaults totrue, so specify asclearTextReporter: { allowColor: false },to disable if you must.logTeststo log the names of unit tests that were run to allow mutants. By default, only the first three are logged. The config for your config file is:clearTextReporter: { logTests: true },maxTestsToLogto show more tests that were executed to kill a mutant whenlogTestsis true. The config for your config file is:clearTextReporter: { logTests: true, maxTestsToLog: 7 },
The dashboard reporter is a special kind of reporter. It sends a report to https://dashboard.stryker-mutator.io, enabling you to add a fancy mutation score badge to your readme! To make sure no unwanted results are sent to the dashboards, it will only send the report if it is run from a build server. The reporter currently detects Travis and CircleCI. Please open an issue if your build server is missing. On all these environments, it will ignore builds of pull requests. Apart from build server specific environment variables, the reporter uses one environment variable:
| Environment variable | Description | Example value |
|---|---|---|
| STRYKER_DASHBOARD_API_KEY | Your API key (generated at https://dashboard.stryker-mutator.io) | 52248872-2edc-4102-a43a-bcfca7a9ca99 |
You will need to pass the STRYKER_DASHBOARD_API_KEY environment variable yourself. You can create one for your repository by logging in on the Stryker dashboard. We strongly recommend you use encrypted environment variables:
files [string[]]
Default: result of git ls-files --others --exclude-standard --cached --exclude .stryker-tmp
Command line: [--files|-f] src/**/*.js,a.js,test/**/*.js
Config file: files: ['src/**/*.js', '!src/**/index.js', 'test/**/*.js']
With files you can choose which files should be included in your test runner sandbox.
This is normally not needed as it defaults to all files not ignored by git.
Try it out yourself with this command: git ls-files --others --exclude-standard --cached --exclude .stryker-tmp.
If you do need to override files (for example: when your project does not live in a git repository),
you can override the files here.
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
You can ignore files by adding an exclamation mark (!) at the start of an expression.
symlinkNodeModules [boolean]
Default: true
Command line: none
Config file: symlinkNodeModules: true
The symlinkNodeModules value indicates whether or not Stryker should create a symbolic link
to your current node_modules directory in the sandbox directories. This makes running your tests by Stryker behave
more like your would run the tests yourself in your project directory.
Only disable this setting if you really know what you are doing.
For example, Jest expects any plugins to be located at "./node_modules/..." in the Sandbox directory. Another example can be running karma tests where you specify files from the 'node_modules/angular/...'. Without symlinking the node_modules directory this would not be possible.
Stryker will look for the node_modules directory to use in the current basePath (or current working directory) and its parent directories.
plugins [string[]]
Default: ['stryker-*']
Command line: --plugins stryker-html-reporter,stryker-karma-runner
Config file: plugins: ['stryker-html-reporter', 'stryker-karma-runner']
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.io.
port [number]
Removed. With this setting you could configure the port used by the karma test runner. It can now be configured directly with karma: { config: { port: 7654 } } (but shouldn't be needed as karma supports its own port selection)
timeoutMS [number]
Default: 5000
Command line: --timeoutMS 5000
Config file: timeoutMS: 5000
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 of time. This period is configurable with two settings: timeoutMS and timeoutFactor.
To calculate the actual timeout in milliseconds the, following formula is used:
timeoutForTestRunMs = netTimeMs * timeoutFactor + timeoutMS + overheadMsBoth netTimeMs and overheadMs are calculated during the initial test run. They are logged on info level. For example when overheadMs is 92 and netTimeMs is 5: Initial test run succeeded. Ran 6 tests in 4 seconds (net 5 ms, overhead 92 ms).
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 lets 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.
timeoutFactor [number]
Default: 1.5
Command line: --timeoutFactor 1.5
Config file: timeoutFactor: 1.5
maxConcurrentTestRunners [number]
Default: (number of CPU Cores)
Command line: --maxConcurrentTestRunners 3
Config file: maxConcurrentTestRunners: 3
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
thresholds [object]
Default: { high: 80, low: 60, break: null }
Command line: none
Config file: thresholds: { high: 80, low: 60, break: null }
Description
Specify the thresholds for mutation score.
mutation score >= high: Awesome! Reporters should color this green and happy.high > mutation score >= low: Warning! Reporters should color this orange/yellow. Watch yourself!mutation score < low: Danger! Reporters should color this in red. You're in danger!mutation score < break: Error! Stryker will exit with exit code 1, indicating a build failure. No consequence for reporters, though.
It is not allowed to only supply one value of the values (it's all or nothing). However, high and low values can be the same, making sure colors are either red or green. Set break to null (default) to never let your build fail.
logLevel [string]
Default: info
Command line: --logLevel info
Config file: logLevel: 'info'
Set the log level that Stryker uses to write to the console. Possible values: off, fatal, error, warn, info, debug and trace
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.
fileLogLevel [string]
Default: off
Command line: --fileLogLevel info
Config file: fileLogLevel: 'info'
Set the log level that Stryker uses to write to the "stryker.log" file. Possible values: off, fatal, error, warn, info, debug and trace
allowConsoleColors [boolean]
Default: true
Command line: --allowConsoleColors true
Config file: allowConsoleColors: true
The allowConsoleColors value indicates whether or not Stryker should use colors in console.