@skills17/karma-helpers NPM

skills17/karma-helpers

This package provides some Karma helpers for usage in a skills competition environment. It includes:

Custom output formatter
Automatic karma configuration
... and more

Installation

Requirements:

Node 16 or greater
Karma 5.0 or greater

Karma already needs to be installed with a compatible testing framework (like mocha).

To install this package, simply run the following command:

npm install @skills17/karma-helpers

It is suggested to add the following npm scripts:

  "scripts": {
    "test": "karma start",
    "test:json": "karma start --reporters skills17-json"
  },

This will provide the following commands:

npm test - Run all tests once and show a nice output with the awarded points (useful for the competitors to see their points)
npm run test:json - Run all tests once and get a json output (useful for automated marking scripts)

Usage

A config.yaml file needs to be created that contains some information about the task. It should be placed in the root folder of your task, next to the package.json file.

See the @skills17/task-config package for a detailed description of all available properties in the config.yaml file.

If the test files in your tasks do not match the default file pattern (./tests/**/*.@(spec|test).@(js|ts)), the tests setting needs to be overwritten.

Karma config

This package provides a function that automatically configures karma for the current task. To use it, create a karma.conf.js file with the following content:

const config = require('@skills17/karma-helpers');

module.exports = config({
  frameworks: ['mocha', 'chai'],
  plugins: ['karma-mocha', 'karma-chai', 'karma-chrome-launcher'],
});

If a different testing framework than mocha with chai is used, modify the frameworks and plugins list. It is also possible to overwrite any other karma configuration value, but shouldn't be necessary usually as the @skills17/karma-helpers takes the correct values from the config.yaml file.

Grouping

A core concept is test groups. You usually don't want to test everything for one criterion in one test function but instead split it into multiple ones for a cleaner test class and a better overview.

In JS, tests are grouped by a test name prefix defined in the config.yaml file.

All describes are concatenated with the actual test names before evaluation.

For example, the following test will have the name Countries > Overview > lists all countries:

describe('Countries', () => {
  describe('Overview', () => {
    it('lists all countries', () => {
      // ...
    });
  });
});

To catch and group all tests within the Overview description, the group matcher can be set to Countries > Overview > .+ for example. Each of the tests within that group will now award 1 point to the group.

Extra tests

To prevent cheating, extra tests can be used. They are not available to the competitors and should test exactly the same things as the normal tests do, but with different values.

For example, if your normal test contains a check to search the list of all countries by 'Sw', copy the test into an extra test and change the search string to 'Ca'. Since the competitors will not know the extra test, it would detect statically returned values that were returned to simply satisfy the 'Sw*' tests instead of actually implement the search logic.

Extra tests are detected by their describe, which should equal 'Extra' or 'extra'. That means that you can simply wrap your test in an aditional extra describe like shown below. The other describes and test names should exactly equal the ones from the normal tests. If they don't, a warning will be displayed.

describe('Extra', () => {    // <-- only this describe has to be added
  describe('Countries', () => {
    it('lists all countries', () => {
      // ...
    });
  });
});

It usually makes sense to move the extra tests in a separate folder, so the folder can simply be deleted before the tasks are distributed to the competitors. Nothing else needs to be done or configured.

If an extra test fails while the corresponding normal test passes, a warning will be displayed that a manual review of that test is required since it detected possible cheating. The penalty then has to be decided manually from case to case, the points visible in the output assumed that the test passed and there was no cheating.