Formerly published as dispatch:mocha. Originally created by Dispatch but now community maintained.
A Mocha test driver package for Meteor. This package reports server AND client test results in the server console and can be used for running tests on a CI server or locally.
In a Meteor 1.3+ app directory:
meteor add meteortesting:mocha
Not installing the package is possible but not advisable when creating a CI pipeline because the system will always download the latest version available, which might break your CI pipeline in a future major release of this package.
Since version 2.0.0 the package lmieulet:meteor-coverage is no longer bundled with this package. If you permanently want code coverage please run:
meteor add lmieulet:meteor-coverage
If you rather want the package to be included only for a single run, add the parameter --extra-packages lmieulet:meteor-coverage
as shown to your meteor test
command:
meteor test --driver-package meteortesting:mocha --extra-packages lmieulet:meteor-coverage
To watch your tests in Meteor 1.8.1+:
meteor test --driver-package meteortesting:mocha
meteor test --driver-package meteortesting:mocha --full-app
NOTE: Watch mode does not properly rerun client tests if you run a headless browser and change only client code. To work around this, you can add or remove whitespace from a server file, and that will trigger both server and client tests to rerun.
If you want the process to exit after all tests are done, add the flag --once
:
meteor test --driver-package meteortesting:mocha --once
meteor test --driver-package meteortesting:mocha --once --full-app
In Meteor <1.8.1, you must add TEST_WATCH=1
if you do not add the --once
flag. For example:
TEST_WATCH=1 meteor test --driver-package meteortesting:mocha
TEST_WATCH=1 meteor test --driver-package meteortesting:mocha --full-app
If you have client tests, you can either open any browser to run them, or install a browser driver to run them headless.
Load http://localhost:3000
in a browser to run your client tests and see the results. This only works well in watch mode because otherwise the server will likely shut down before you finish running the client tests.
The test results are reported in a div with ID mocha
. If you run with the --full-app
flag, this will likely be overlaid weirdly on top of your app, so you should add CSS to your app in order to be able to see both. For example, this will put the test results in a sidebar with resizeable width:
div#mocha {
background: white;
border-right: 2px solid black;
height: 100%;
left: 0;
margin: 0;
overflow: auto;
padding: 1rem;
position: fixed;
resize: horizontal;
top: 0;
width: 20px;
z-index: 1000;
}
Support for running browser in headless mode is made possible by the package `` which this plugin depends on. Please refer to their documentation for further information: https://github.com/meteortesting/meteor-browser-tests#dependencies
By default both server and client tests run. To disable server tests: TEST_SERVER=0
. Likewise for client: TEST_CLIENT=0
To run all tests with names that match a pattern, add the environment variable MOCHA_GREP=your_string
. This will apply to both client and server tests.
To exclude any tests, you must use the grep option above plus MOCHA_INVERT=1
. For example, to exclude tests named 'TODO:' (which you may want to exclude from your continuous integration workflow) you would pass at runtime MOCHA_GREP=your_string MOCHA_INVERT=1
To override Mocha's default timeout of 2 seconds for all tests, add the environment variable MOCHA_TIMEOUT=your_timeout_in_ms
.
You can configure the mocha runner with a .mocharc.js
or a .mocharc.json
file at the root of your Meteor app. This package uses mocha programmatically, so it supports a constrained list of options.
If both files are defined, .mocharc.js
will overwrite the settings in .mocharc.json
.
The setting timeout
will be overwritten if the environment variable MOCHA_TIMEOUT
is set.
By default meteortesting:mocha will run in series. This is a safety mechanism since running a client test and server test which depend on DB state may have side effects.
If you design your client and server tests to not share state, then you can run tests faster. Run in parallel by exporting the environment variable TEST_PARALLEL=1
before running.
To write the tests to a file, set SERVER_MOCHA_OUTPUT
and CLIENT_MOCHA_OUTPUT
to the full path + filename, e.g., $PWD/unit_server.txt
and $PWD/unit_client.txt
. This is specially important when using a format like xunit
. The use of XUNIT_FILE
is deprecated because it has the same functionality as SERVER_MOCHA_OUTPUT
, which is a better fit for what it actually does.
$ MOCHA_REPORTER=xunit SERVER_MOCHA_OUTPUT=$PWD/unit_server.xml CLIENT_MOCHA_OUTPUT=$PWD/unit_client.xml meteor test --once --driver-package meteortesting:mocha
The default Mocha reporter for server tests is the "spec" reporter. You can set the SERVER_TEST_REPORTER
environment variable to change it.
$ SERVER_TEST_REPORTER="dot" meteor test --once --driver-package meteortesting:mocha
The default Mocha reporter for client tests is the "spec" reporter when running headless and "html" otherwise. You can set the CLIENT_TEST_REPORTER
environment variable to change it.
$ CLIENT_TEST_REPORTER="tap" meteor test --once --driver-package meteortesting:mocha
Since version 2.0.0 the package lmieulet:meteor-coverage is no longer bundled with this package. If you permanently want code coverage please run:
meteor add lmieulet:meteor-coverage
If you rather want the package to be included only for a single run, add the parameter --extra-packages lmieulet:meteor-coverage
as shown to your meteor test
command:
meteor test --driver-package meteortesting:mocha --extra-packages lmieulet:meteor-coverage
To enable code coverage you have to set COVERAGE
to 1
and COVERAGE_APP_FOLDER
to the path of your project. On POSIX systems you can just use COVERAGE_APP_FOLDER=$PWD/
whereby COVERAGE_APP_FOLDER=%cd%\
gives the expected result on Windows.
In addition there are quite some additional options you can set:
COVERAGE_VERBOSE
to see the files included in the coverage and other data that might help if something doesn't work as expectedCOVERAGE_IN_COVERAGE
imports a coverage dump (previously create withCOVERAGE_OUT_COVERAGE
)COVERAGE_OUT_COVERAGE
creates a dump of the coverage - used when you want to merge several coverageCOVERAGE_OUT_LCOVONLY
creates a lcov reportCOVERAGE_OUT_HTML
creates a html reportCOVERAGE_OUT_JSON
creates a json reportCOVERAGE_OUT_JSON_SUMMARY
creates a json_summary reportCOVERAGE_OUT_TEXT_SUMMARY
creates a text_summary reportCOVERAGE_OUT_REMAP
remaps the coverage to all the available report formats
Additional information can be found here: https://github.com/serut/meteor-coverage
A good best practice is to define these commands as run scripts in your app's package.json
file. For example:
"scripts": {
"pretest": "npm run lint --silent",
"test-chrome": "TEST_BROWSER_DRIVER=chrome meteor test --once --driver-package meteortesting:mocha",
"test-app-chrome": "TEST_BROWSER_DRIVER=chrome meteor test --full-app --once --driver-package meteortesting:mocha",
"test-phantom": "TEST_BROWSER_DRIVER=phantomjs meteor test --once --driver-package meteortesting:mocha",
"test-app-phantom": "TEST_BROWSER_DRIVER=phantomjs meteor test --full-app --once --driver-package meteortesting:mocha",
"test-watch": "TEST_BROWSER_DRIVER=chrome meteor test --driver-package meteortesting:mocha",
"test-app-watch": "TEST_BROWSER_DRIVER=chrome meteor test --full-app --driver-package meteortesting:mocha",
"test-watch-browser": "meteor test --driver-package meteortesting:mocha",
"test-app-watch-browser": "meteor test --full-app --driver-package meteortesting:mocha",
"lint": "eslint .",
"start": "meteor run"
}
And then run npm run test-chrome
, etc.