Application Tests

Guides for writing automated tests for CHT applications

This tutorial takes you through testing the various configurable components of CHT applications using cht-conf-test-harness.


CHT Application Testing

CHT applications are greatly configurable. Depending on the number and the complexity of the app components, it can take huge time and effort to test them manually. Some components, such as tasks, behave differently over time and are particularly difficult to test. As the project evolves, the configuration is often updated with new components, and changes are made to the existing components. After each change, you need to test not only the new components but also the old ones to make sure that the app works correctly. To facilitate the testing process, app builders are encouraged to write automated tests for their app using cht-conf-test-harness.

Using cht-conf-test-harness (also simply referred as harness), you can write tests and run them with Mocha testing framework to test the behavior of different components in a CHT application. You can use the chai library or other assertion libraries.


Writing tests for CHT apps requires a good understanding of the project workflows and requirements. It is encouraged that the tests be written as part of the app-building process. To test using the harness, there are a few things you need to set up first:

  1. From the previous tutorials, you should have a functioning CHT instance with cht-conf installed locally and a project folder set up.

  2. If your package.json file does not already have them, add cht-conf-test-harness, chai, and mocha by running this in your command-line:

    npm install --save cht-conf-test-harness chai mocha
  3. Also, add the following scripts to package.json, if not already present:

    "scripts": {
        "test": "npm run compile-forms && npm run compile-app-settings && npm run unittest",
        "compile-app-settings": "npx cht --no-check compile-app-settings",
        "compile-forms": "npx cht --no-check convert-app-forms convert-contact-forms",
        "unittest": "mocha test/**/*.spec.js --timeout 20000"

    After adding these scripts, you are able to run the tests by running one of these commands from the command-line:

    npm run testCompiles the app settings, converts the forms, then runs the tests (preferred solution).
    npm run unittestOnly runs the tests.
  4. Create a folder in the project root where you keep the tests. You can name the folder yourself. For convenience, here it is named as test.

  5. Create a file harness.defaults.json at the root of your project. This is the default configuration file for the harness. Here you can specify the default user, preloaded contacts and reports, and add other settings. Here’s an example file to get you started. You can read more about it here.

Writing a test for CHT App

Start by adding a file where a group of related tests are written. For example, in the default config, all tests for the pregnancy form are in the pregnancy.spec.js file.

Let’s introduce some important sections from a typical test file:

const TestRunner = require('cht-conf-test-harness');
const harness = new TestRunner();

You get an instance of harness and the same instance is used throughout the test file.

You can also pass parameters to the TestRunner() when instantiating:

// For detailed console logs
const harness = new TestRunner({ verbose: true });

Other useful options are:

  1. { headless: false } - Passed to puppeteer, launches browser with GUI. Helpful to see how the form is getting filled.
  2. { harnessDataPath: ''}: Specify different harness confguration

You can find more harness options and examples here.

You need to start the harness before running tests and stop it after the use. To handle start and stop of harness, place following statements in before() and after() hooks of the test suite.

describe('PNC form tests', () => {
  before(async () => { return await harness.start(); });
  after(async () => { return await harness.stop(); });

Any logic that you want to execute before and after each test can be placed in the beforeEach() and afterEach() hooks. Clear the state of the harness and set the date to a fixed value before each test run:

    async () => {
      await harness.clear();
      await harness.setNow('2000-01-01');

After each test run, assert that there are no errors in the console.

  afterEach(() => {

If you want to learn more about these hooks, refer to this Mocha resource. Feel free to customize the hooks as you see fit.

Let’s look at a more detailed example. Here is a test case for the Assessment form that was covered in the previous tutorial:

  it('unit test confirming assessment with cough since 7 days', async () => {
    // Load the assessment form and fill 'yes' on the first page and '7' on the second page
    const result = await harness.fillForm('assessment', ['yes'], ['7']);

    // Verify that the form successfully got submitted

    // Verify some attributes on the resulting report
      patient_name: 'Patient Name',
      group_assessment: {
        cough: 'yes',
        cough_duration: '7',

In line 3 above, the method harness.fillForm() fills the specified form with the given input (answers).

Depending on the form design, the number of inputs to be filled can be large. The inputs are often repeated within a single test or across multiple tests with little or no variation. It is a good idea to keep them in a separate file and refer them from the tests as required. You can also introduce some variations in the inputs using function parameters.

Example: form-inputs.js

module.exports = {
  assessments: {
    no_cough: [
    cough: (days) => [
      [ days ]

In the test below, the inputs are used from the form-inputs.js file above:

const { assessments } = require('../form-inputs');
const result = await harness.fillForm('assessment', ...assessments.cough('3'));

The test files are usually grouped in folders to read and run them easily. One way of grouping them is by creating folders for each of the components that can be tested: forms, contact summary, tasks, and targets.


Testing Forms

What do you test?
Minimum:Filling a form with the most common options should not result in any error.
Ideal:All input combinations and constraints are tested with the report fields.

Previous example demonstrates test for the app form. More test cases can be added by changing the inputs.

You can also test contact forms using test harness. To fill a contact form, use fillContactForm(contactType, …answers).


const result = await harness.fillContactForm(
  ['new_person', 'William Mars', '1990-08-06', '+1234567891', 'female'],
expect(result.contacts).excluding(['_id', 'meta']).to.deep.eq([

When filling a form, you can also test the field constraints. The example below asserts that a form does not accept birth date in the future:

it(`Throws validation error when birth date is in future`, async () => {
  const result = await harness.fillForm('delivery', ['yes', '2090-01-02']);
    msg: 'Date cannot be in the future!'

Note: If a form triggers a task, some use cases of the form can be tested when testing the task later.

Testing Contact Summary

What do you test?
Minimum:Fill a contact form and count the number of fields in the contact summary
Ideal:Targeted tests for calculations of context, fields, cards, etc.

Contact summary consists of visible components such as cards, fields and a hidden component: context. All these can be tested with the test harness.

Use harness.getContactSummary() method to get the ContactSummary object, which has these properties: fields, cards, and context.

To test the contact summary fields added in the previous tutorial, use the following test case:

const contactSummary = await harness.getContactSummary();
expect(contactSummary.fields)'length', 5);
expect(contactSummary.fields.filter(f => f.filter !== 'lineage')).to.deep.equal(
      { label: 'patient_id', value: 'patient_id', width: 4 },
      { label: 'contact.age', value: '1970-07-09', width: 4, filter: 'age' },
      { label: '', value: '', translate: true, width: 4 },
      { label: '', value: undefined, width: 4}

Here, the contact summary being tested represents the contact that is being “acted on” or the “subject of the test”. To learn more about this, look at the subject property here.

Similarly, you can test the condition cards too. Here is an example for testing the assessment condition card added in this tutorial:

// Load the assessment form and fill in 'yes' on the first page and '7' on the second page
const result = await harness.fillForm('assessment', ['yes'], ['7']);

// Verify that the form successfully got submitted

const contactSummary = await harness.getContactSummary();

// Verify that there is one card
expect('length', 1);

// Verify the fields on the card
      label: '',
      width: 6,
      filter: 'simpleDate'
    { label: 'contact.profile.cough', value: 'yes', width: 6 },

If you follow this code sample to create the pregnancy condition card, the pregnancy context can be tested this way:

const summaryContext = harness.getContactSummary().context;
  pregnant: true

Testing Tasks

What do you test?
Minimum:Trigger and resolve the task
Ideal:One test for each use scenario
Code coverage for any arc with an external dependency
Negative cases - confirm tasks don’t trigger

When testing the tasks manually, you need to fill a form. Then to see the task, you need to either change the system date to move forward in time or change the reported date of the document accordingly. These are very tedious and unreliable methods. Using the test harness, you can quickly test the tasks under different scenarios and at different simulated dates.

Every task has a source document (contact or report). When testing, you can mock the source document. Yet, the recommended approach is to fill the source form and then proceed with checking the associated tasks.

Commonly used harness methods when testing tasks are:

With getTasks(), you get an array of Task objects which corresponds to the tasks schema.

Let’s look back at the simple task from this tutorial: Building A Simple Task.

According to the task configuration, these conditions need to be met for the assessment task to be visible:

  1. The contact_type of the subject (patient) is: patient
  2. The contact_type of the user’s parent is: chw_area
  3. The current time is between the start and end dates of the task event.

When testing with harness, the conditions 1 and 2 above can be set in the harness.defaults.json file. See the lines 31 and 26 respectively in the sample file further below.

For the condition 3, let’s look at the task event window in the task definition. As mentioned in the earlier tutorial, the task event is:

  • Due 7 days after the contact’s creation date.
  • Should appear 7 days before the due date, or immediately when the contact is created.
  • Should disappear the day after the due date.

So, to see the task, the contact should be created within the last 7 days. You can easily set the contact’s reported date so that it falls within the last 7 days (see line 36 below).


  "coreVersion": "3.14.0",
  "user": "chw_id",
  "subject": "patient_id",

  "ownedBySubject": true,

  "docs": [
      "_id": "chw_id",
      "name": "CHW",
      "date_of_birth": "1971-03-10",
      "phone": "+17782475555",
      "reported_date": 1550559625153,
      "type": "person",
      "parent": {
        "_id": "chw_area_id",
        "parent": {
          "_id": "district_id"
      "_id": "chw_area_id",
      "type": "contact",
      "contact_type": "chw_area"
      "_id": "patient_id",
      "type": "contact",
      "contact_type": "patient",
      "name": "Patient Name",
      "role": "patient",
      "date_of_birth": "1970-07-09",
      "sex": "female",
      "reported_date": 1652868491000,
      "patient_id": "patient_id",
      "parent": {
        "_id": "patient_parent_id",
        "parent": {
          "_id": "chw_area_id",
          "parent": {
            "_id": "district_id"

Note that the reported_date above stores the epoch timestamp in milliseconds when the document was first created. You can use external website epochconverter to convert the timestamp to and from a human readable date.

After setting the harness defaults, you can now test the task:

// Get the task by title
const tasks = await harness.getTasks({title: 'First Assessment'});

//Make sure that you have the task
expect(tasks)'length', 1);

// Load the task action and fill it.
const result = await harness.loadAction(tasks[0], ['yes'], ['7']);

// Verify that the form successfully got submitted

// Verify that the task is no longer visible
expect(await harness.getTasks());

If a task is triggered by a report, then fill in the app form to create report before test:

it('followup schedule', async () => {
  const result = await harness.fillForm(...pncFollowups.registerBirth('2000-01-01', '2000-02-01'));

  /* Task should appear 7 days after birth for follow up */
  await harness.flush(7);
  const tasks = await harness.getTasks();
  expect(tasks)'length', 1);
  /* Complete the task, task should disappear */
  await harness.loadAction(tasks[0].emission.actions[0]);
  const followupResult = await harness.fillForm(...pncFollowups.followup.healthy);
  expect(await harness.getTasks());

  /* Confirm a task appears again on the schedule */
  await harness.flush(21);
  expect(await harness.getTasks())'length', 1);

You may pass other data from a task to the action form using modifyContent attribute of a task. You can also verify that these data are present in the task.

      t_follow_up_count: '5',

Testing Targets

Testing a target is relatively straightforward. Add a report or contact that increments a target, then check the target values.

What do you test?
Minimum:Trigger incrementing of the target
Ensure target doesn’t increment when it shouldn’t
Ideal:One test for each user scenario
Ensure proper deduplication (particularly for those with emitCustom)

Use harness.getTargets to check the state of targets. It returns a Target object which corresponds to the targets schema.

To test the first two targets created in the targets tutorial, use this code:

it('assessment this month and all time assessments should show correct counts', async () => {
    //set the current date

    //assessment form filled
    const result = await harness.fillForm('assessment', ['yes'], ['7']);

    const thisMonth = await harness.getTargets({ type: 'assessments-this-month' });
    expect(thisMonth)'length', 1);

    //The number of assessments in this month should be 1
    expect(thisMonth[0]).to.nested.include({ 'value.pass': 1, '': 1 });

    let allTime = await harness.getTargets({ type: 'assessments-all-time' });
    expect(thisMonth)'length', 1);

    //The number of all time assessments should be 1
    expect(thisMonth[0]).to.nested.include({ 'value.pass': 1, '': 1 });

    //Go to next month by adding 2 days (Jan 30 to Feb 1)

    const nextMonth = await harness.getTargets({ type: 'assessments-this-month' });
    expect(nextMonth)'length', 1);

    //The number of assessments this month now should be reset to 0
    expect(nextMonth[0]).to.nested.include({ 'value.pass': 0, '': 0 });

    allTime = await harness.getTargets({ type: 'assessments-all-time' });
    expect(allTime)'length', 1);

    //The number of all time assessments should still be 1
    expect(allTime[0]).to.nested.include({ 'value.pass': 1, '': 1 });

For targets with type: 'percent', you might want to check for more properties:

    'value.pass': 1,
    '': 1,
    'value.percent': 100

Tests for helper functions

If you have added helper functions, they can be tested separately.


  describe('toISODateString', () => {
    const { toDateString } = noolsExtras;

    it('parses a valid date', () => {
      const actual = toISODateString('01/01/2000');

Tests for bug fixes

When fixing a bug, write a test which captures the specific scenario being fixed.

  • At least one test should fail before the fix
  • It should pass after the fix

This prevents the regression and helps ensure that you’re fixing what you expect.

Tips for writing tests

  • Every test needs to add value
  • Use the scientific method: your test is a hypothesis and you’re running an experiment
  • Follow these steps:
    1. Arrange - Prepare the experiment
    2. Act - Run the experiment. Take a measurement
    3. Assert - Assert the hypothesis
  • Structure testing inputs into “classes”: if one member in a class works well then the class is considered to work
  • Run mocha with --grep to only run tests that match a given string
  • Debug tests and CHT components using breakpoints and debug console.

For debugging the unit tests in VS Code, here is a sample configuration file (launch.json):

  "version": "0.2.0",
  "configurations": [
      "type": "node",
      "request": "launch",
      "name": "Run Mocha Tests",
      "program": "${workspaceFolder}/node_modules/mocha/bin/_mocha",
      "args": [
        "--grep", "should show"
      "internalConsoleOptions": "openOnSessionStart"

CHT Applications > Reference

Technical details of CHT app components for app developers

CHT Applications > Quick Guides > Tasks > Querying Task Documents : Testing Task Document Data

Querying the data which results from an example task. Notes on the performance implications of tasks.