UI Testing

You can write UI tests using Cypress. It is a NodeJS based full-stack testing framework which doesn't rely on Selenium.

To write integration tests, create a .js file in the cypress/integration directory.

Example

Here is an example of an integration test to check insertion of a ToDo

context('ToDo', () => {
    before(() => {
        cy.login('Administrator', 'admin');
        cy.visit('/desk');
    });

    it('creates a new todo', () => {
        cy.visit('/app/todo/new-todo-1');
        cy.fill_field('description', 'this is a test todo', 'Text Editor').blur();
        cy.get('.page-title').should('contain', 'Not Saved');
        cy.get('.primary-action').click();
        cy.visit('/desk#List/ToDo');
        cy.location('hash').should('eq', '/app/todo');
        cy.get('.list-row').should('contain', 'this is a test todo');
    });
});

Running Cypress Locally

Cypress uses any chromium based browser installed on your system to run tests. Every app has it's own cypress test suite. To run test for an app, run the following command from the frappe-bench directory.

bench --site [sitename] run-ui-tests [app]

This will open the Cypress Electron shell where you can run any test manually or run all of the tests.

You can also run tests in headless mode.

# run in headless mode
bench --site [sitename] run-ui-tests [app] --headless

To enable cypress parallel testing you can pass --parallel flag. More information on how cypress parallel tests work can be found here.

# run tests parallelly
bench --site [sitename] run-ui-tests [app] --parallel

Code Coverage

Code coverage helps to identify which lines of the source code were executed during the tests. In order to measure code coverage, the source code needs to be instrumented and this instrumented source code needs to be integrated with our test runner to collect the coverage and generate a report.

For Cypress tests in Frappe, the .js files are instrumented using Istanbul and the Cypress code-coverage plugin is used to merge coverage from each test and save the combined result.

Code Instrumentation:

In order to compute which lines of the source code were executed, additional counters are inserted into the code through instrumentation. For example:

Before instrumentation:

function foo(a, b) {
    if (a < b)
        return b - a;
    else
        return a - b;
}

After instrumentation:

cov_1m1jljnmzu();

function foo(a, b) {
    cov_1m1jljnmzu().f[0]++;
    cov_1m1jljnmzu().s[0]++;

    if (a < b) {
        cov_1m1jljnmzu().b[0][0]++;
        cov_1m1jljnmzu().s[1]++;
        return b - a;
    } else {
        cov_1m1jljnmzu().b[0][1]++;
        cov_1m1jljnmzu().s[2]++;
        return a - b;
    }
}

When using this modified (instrumented) source code for testing, these counters get incremented as the code is executed, and a coverage object is generated. The Cypress code-coverage plugin then handles the collected coverage and generates coverage reports.

Generating Code Coverage Report Locally:

  1. Instrument source code using istanbul/nyc :
   npx nyc instrument -x 'frappe/public/dist/**' -x 'frappe/public/js/lib/**' -x '**/*.bundle.js' --compact=false --in-place frappe

This replaces the existing source code in the frappe folder with the instrumented source code. The -x flag is used to exclude specified paths. You can also use the -n flag to specify paths to be included. See here for more details about the nyc instrument command

  1. Run Cypress tests:
   bench --site test_site run-ui-tests frappe --with-coverage
  1. Generate report:
   npx nyc report --reporter=text

See here for alternate report formats

Testing-Library Queries

You can also use Testing Library queries within your Cypress tests. Testing Library provides testing utilities that:

  • Make it easier to write UI tests that resemble the way users interact with the app
  • Make it easier to find elements in the DOM without knowing all the implementation details
  • Help keep the tests maintainable (so minor refactors don't break tests)

See Testing Library Docs for more details about usage

Testing Library provides several queries to find elements on a page. Here are some examples:

  • ByRole

    Look here for table of HTML elements and their default roles

    Query Element
    findByRole('button', {name: 'Save'}) <button> with accessible name = 'Save'
    findByRole('checkbox') <input type=checkbox>
    findByRole('textbox') <input type=text>, <textarea> (Also matches other elements with default role='textbox')
    findByRole('searchbox') <input type=search>
    findByRole('listbox') <select>, <datalist>
  • ByLabelText

    Query Element
    findByLabelText('Optimize') element associated with the label 'Optimize'
  • ByPlaceholderText

    Query Element
    findByPlaceholderText('Name') element with placeholder='Name'
  • ByText

    Query Element
    findByText('example.json') element with textContent='example.json'
  • ByDisplayValue

    Query Element
    findByDisplayValue('Option 1') <select> with selected <option> 'Option 1' (Also matches <input> or <textarea> with matching value attribute)
  • ByTitle

    Query Element
    findByTitle('Open Link') element with title='Open Link'
  • ByAltText

  • ByTestId