❤
npm
Sign UpSign In

hermione
TypeScript icon, indicating that this package has built-in type declarations

8.9.1 • Public • Published

Hermione

Hermione is a utility for integration testing of web pages using WebdriverIO and Mocha.

Why you should choose hermione

Hermione provides several features that WebdriverIO doesn't, and makes the testing process easier.

Easy to use

If you are familiar with WebdriverIO and Mocha, you can start writing and running tests in 5 minutes! You need to install hermione via npm and add a tiny config to your project. For details, see the Quick start section.

Runs tests in parallel

When tests are run one by one, it takes a lot of time. Hermione can run tests in parallel sessions in different browsers out of the box.

Runs tests in subprocesses

Running of too many tests in parallel can lead to the overloading of the main process CPU usage which causes degradation in test passing time, so Hermione runs all tests in subprocesses in order to solve this problem.

Extensible

WebdriverIO provides built-in commands for browser and page manipulation. Often projects need to store some common code and reuse it throughout all tests, so the developer needs to create some helpers and include them in the tests.

With hermione this is very simple and straightforward. You can add any number of custom commands in the hermione config and use them as browser.myCustomCommand in tests.

Moreover, hermione provides plugins that work like hooks. They allow the developer to prepare the testing environment and react properly to test execution events.

Built-in assert library

when writing integration tests, we need to check for various conditions, such as that a button element has certain attributes. So hermione provides global expect variable that gives you access to a number of matchers that let you validate different things on the browser, an element or mock object. More about using expect and its API you can find here.

Retries failed tests

Integration tests use a dynamic environment with a lot of dependencies, where any of them could be unstable from time to time. As a result, integration tests turn red randomly, which makes them imprecise. This spoils the entire testing process.

To prevent incidental fails, hermione retries a failed test before marking it as failed. This makes it possible to get rid of a majority of incidental fails. The number of retries can be specified for all browsers or for a specific browser.

⚠️ Hermione reruns tests in a new browser session to exclude situations when the browser environment is the cause of the failure.

Executes separate tests

Sometimes you only need to run specific tests, not all the tests in a set. Hermione makes this possible. You can specify the path to the test file

hermione tests/func/mytest.js

or filter describes by using the --grep option

hermione --grep login

or simply use the mocha only() API in the test

describe.only('user login', function() {...});

Skips tests in specific browsers

Sometimes you need to skip a test just in a specific browser, not in all browsers. For example, you don't need to run some test in ugly IE browsers. In hermione you can do this with hermione helper. For example, you can skip some tests in a specific browser

describe('feature', function() {
    hermione.skip.in('ie8', 'it cannot work in this browser');
    it('nowaday functionality', function() {...});
});

or run tests in just one browser

describe('feature', function() {
    // will be skipped in all browsers except Chrome
    hermione.skip.notIn('chrome', 'it should work only in Chrome');
    it('specific functionality', function() {...});
});

In these cases you will see messages in reports with the reason for skipping.

To skip a suite or test silently (without any messages in reports), you can pass the third argument with the silent flag:

hermione.skip.in('ie8', 'skipReason', {silent: true});
// or
hermione.skip.notIn('chrome', 'skipReason', {silent: true});

Or you can use another hermione helper, only, which is silent by default:

hermione.only.in('chrome');
// or
hermione.only.notIn('ie8');

hermione.only.in will run tests only in the specified browsers and skip the rest silently.

hermione.only.notIn will run tests in all browsers except the specified ones.

Override browser configuration for test

Hermione allows you to override a browser configuration for each test or a whole suite. Each method of hermione.browser(<browserName>) provide chaining.

version(<browserVersion>)

// change the browser version for all chindren
hermione.browser('chrome').version('70.3');
describe('suite', function() {
    // ...
});
// change the browser version for a specific test
hermione.browser('chrome').version('70.3');
it('test', function() {...});
hermione.browser('chrome').version('70.3');
describe('suite', function() {
    it('test 1', function() {...});

    // this call will override the version only for test below
    hermione.browser('chrome').version('70.1');
    it('test 2', function() {...});
});

Offers flexible test configuration

Hermione lets you configure running some set of tests in specific browsers. For example,

sets: {
    desktop: {
        files: 'tests/desktop',
        browsers: ['ie8', 'opera']
    },
    touch: {
        files: 'tests/touch',
        browsers: ['iphone', 'android']
    }
}

See sets for more details.

Automatically initializes and closes grid sessions

All work with the grid client is encapsulated in hermione. Forget about client.init and client.end in your tests ;)

Fairly waits for screen rotate

Request /session/:sessionId/orientation is not a part of the official Webdriver specification, so commands orientation and setOrientation which are provided by client webdriverio from the box do not guarantee screen rotate before the next command will start to execute, but Hermione solves this problem.

Prerequisites

All you need are browsers that Hermione could use for testing. To do this you need to install some browsers, such as chrome (to automate this process you can use the hermione-headless-chrome plugin).

Next, you have two ways to configure Hermione to work with browsers:

  • Using the devtools protocol (available only for Chromium-based browsers). This method does not need to be pre-configured. Just go to the quick start.
  • Using the webdriver protocol. In this case you need to set up Selenium grid. The simplest way to get started is to use one of the NPM selenium standalone packages, such as vvo/selenium-standalone. For more information about setting up, see selenium-standalone.

Selenium-standalone

Install selenium-standalone by command:

npm i -g selenium-standalone

Next you need to install browser drivers

selenium-standalone install

and run your server by executing

selenium-standalone start

⚠️ If you will get error like No Java runtime present, requesting install. you should install Java Development Kit (JDK) for your OS.

Quick start

First of all, make sure that all prerequisites are satisfied.

Now you have two ways to configure project.

Using create-hermione-app (a quick way)

You just need to run the cli command from create-hermione-app tool and answer a few questions:

npm init hermione-app YOUR_PROJECT_PATH

To skip all questions just add the option -y at the end.

Configuring .hermione.conf.js by yourself (a slow way)

Create hermione config file with name .hermione.conf.js in the project root. There are two configuration options depending on the method selected in the prerequisites section.

Chrome Devtools Protocol

module.exports = {
    sets: {
        desktop: {
            files: 'tests/desktop/**/*.hermione.js'
        }
    },

    browsers: {
        chrome: {
            automationProtocol: 'devtools',
            desiredCapabilities: {
                browserName: 'chrome'
            }
        }
    }
};

Webdriver protocol

module.exports = {
    gridUrl: 'http://localhost:4444/wd/hub',

    sets: {
        desktop: {
            files: 'tests/desktop/*.hermione.js'
        }
    },

    browsers: {
        chrome: {
            automationProtocol: 'webdriver', // default value
            desiredCapabilities: {
                browserName: 'chrome'
            }
        }
    }
};

Write your first test in tests/desktop/github.hermione.js file.

describe('github', function() {
    it('should check repository name', async ({ browser }) => {
        await browser.url('https://github.com/gemini-testing/hermione');

        await expect(browser.$('#readme h1')).toHaveText('Hermione');
    });
});

Finally, run tests (be sure that you have already run selenium-standalone start command in next tab).

node_modules/.bin/hermione

Commands API

Since Hermione is based on WebdriverIO v8, all the commands provided by WebdriverIO are available in it. But Hermione also has her own commands.

Browser commands

clearSession

Browser command that clears session state (deletes cookies, clears local and session storages). For example:

it('test', async ({ browser }) => {
    await browser.url('https://github.com/gemini-testing/hermione');

    (await browser.getCookies()).length; // 5
    await browser.execute(() => localStorage.length); // 2
    await browser.execute(() => sessionStorage.length); // 1

    await browser.clearSession();

    (await browser.getCookies()).length; // 0
    await browser.execute(() => localStorage.length); // 0
    await browser.execute(() => sessionStorage.length); // 0
});

Element commands

moveCursorTo

This command is temporary and will be removed in the next major (hermione@9). Differs from the standard moveTo in that it moves the cursor relative to the top-left corner of the element (like it was in hermione@7).

Move the mouse by an offset of the specified element. If offset is not specified then mouse will be moved to the top-left corder of the element.

Usage:

await browser.$(selector).moveCursorTo({ xOffset, yOffset });

Available parameters:

  • xOffset (optional) Number – X offset to move to, relative to the top-left corner of the element;
  • yOffset (optional) Number – Y offset to move to, relative to the top-left corner of the element.

Tests API

Arguments

Hermione calls test and hook callback with one argument. This argument is an object with the next properties:

  • browser - browser client
  • currentTest - current executing test

Example:

beforeEach(async ({ browser, currentTest }) => {
    await browser.url(`/foo/bar?baz=${currentTest.id}`);
});

afterEach(async ({ browser, currentTest }) => {
    // Do some post actions with browser
});

it('some test', async ({ browser, currentTest }) => {
    await browser.click('.some-button');

    // Do some actions and asserts
});

You also can pass any data into your test through parameters:

beforeEach(async (opts) => {
    opts.bar = 'bar';
});

it('some test', async ({ browser, bar }) => {
    await browser.url(`/foo/${bar}`);

    // Do some actions and asserts
});

browser and currentTest also available through callback context:

beforeEach(async function() {
    await this.browser.url(`/foo/bar?baz=${this.currentTest.id}`);
});

afterEach(async function() {
    // Do some post actions with this.browser
});

it('some test', async function() {
    await this.browser.click('.some-button');

    // Do some actions and asserts
});

Hooks

before and after hooks are forbidden in hermione, you should use beforeEach and afterEach hooks instead. This feature was implemented in order to ensure better stability while running tests and make them independent of each other.

Skip

This feature allows you to ignore the specified suite or test in any browser, with an additional comment. You can do this by using the global hermione.skip helper. It supports the following methods:

  • .in – Adds matchers for browsers with the additional comment.
  • .notIn – .in method with the reverted value.

Each of these methods takes the following arguments:

  • browser {String|RegExp|Array<String|RegExp>} – Matcher for browser(s) to skip.
  • [comment] {String} – Comment for skipped test.
  • [options] {Object} – Additional options.

Note that matchers will be compared with browserId specified in the config file, e.g. chrome-desktop.

For example,

describe('feature', function() {
    hermione.skip.in('chrome', "It shouldn't work this way in Chrome");
    it('should work this way', function() {
        return runTestThisWay();
    });

    it('should work that way', function() {
        return runTestThatWay();
    });

    hermione.skip.in(['chrome', 'firefox', /ie\d*/], 'Unstable test, see ticket TEST-487');
    it('should have done some tricky things', function() {
        return runTrickyTest();
    });
});

In this case, the behaviour it should work this way will be skipped only in chrome browser, but will be run in other browsers. It should work that way will not be ignored. So only the nearest test will be skipped. If you need to skip all tests within a suite, you can apply the skip helper to a describe so all tests within this suite will be skipped with the same comment.

hermione.skip.in('chrome', 'skip comment');
describe('some feature', function() {
    it(...);
    it(...);
});

You can also use the .notIn method to invert matching. For example,

// ...
hermione.skip.notIn('chrome', 'some comment');
it('should work this way', function() {
    return doSomething();
});
// ...

In this case, the test will be skipped in all browsers except chrome.

All of these methods are chainable, so you can skip a test in several browsers with different comments. For example,

// ...
hermione.skip
    .in('chrome', 'some comment')
    .notIn('ie9', 'another comment');
it('test1', function() {
    return doSomething();
});
// ...

If you need to skip a test in all browsers without a comment, you can use mocha .skip method instead of hermione.skip.in(/.*/);. The result will be the same.

Only

This feature allows you to ignore the specified suite or test in any browser silently (without any messages in reports). You can do this by using the global hermione.only helper. It supports two methods:

  • .in — The hermione.skip.notIn method with the silent flag,
  • .notIn — The hermione.skip.in with the silent flag.

These methods take the following arguments:

  • browser {String|RegExp|Array<String|RegExp>} — A matcher for browser(s) to skip.

For example:

// ...
hermione.only.in('chrome');

it('should work this way', function() {
    return doSomething();
});

The test will be skipped all browsers silently except in chrome.

hermione.only.notIn('ie9');
it('should work another way', function() {
    return doSomething();
});

The test will be processed in all browsers and silently skipped in ie9.

Config overriding

You can override some config settings for specific test, suite or hook via hermione.config.* notation.

testTimeout

Overrides testTimeout config setting. Can be set for tests and suites.

hermione.config.testTimeout(100500);
it('some test', function() {
    return doSomething();
});

WebdriverIO extensions

Hermione adds some useful methods and properties to the webdriverio session after its initialization.

Sharable meta info

Implemented via two commands:

  • setMeta(key, value)
  • getMeta([key])

These methods allow you to store some information between webdriver calls and it can then be used in custom commands, for instance. This meta information will be shown in the html-reporter.

Note: hermione saves the last URL opened in the browser in meta info.

Example:

it('test1', async ({ browser }) => {
    await browser.setMeta('foo', 'bar');
    await browser.url('/foo/bar?baz=qux');

    const val = await browser.getMeta('foo');
    console.log(val); // prints 'bar'

    const url = await browser.getMeta('url');
    console.log(url); // prints '/foo/bar?baz=qux'

    const meta = await browser.getMeta();
    console.log(meta); // prints `{foo: 'bar', url: '/foo/bar?baz=qux'}`
});

Execution context

The execution context can be accessed by the browser.executionContext property, which contains the current test/hook object extended with the browser id.

Example:

it('some test', async ({ browser }) => {
    await browser.url('/foo/bar');
    console.log('test', browser.executionContext);
});

will print something like this

test: {
  "title": "some test",
  "async": 0,
  "sync": true,
  "timedOut": false,
  "pending": false,
  "type": "test",
  "body": "...",
  "file": "/foo/bar/baz/qux.js",
  "parent": "#<Suite>",
  "ctx": "#<Context>",
  "browserId": "chrome",
  "meta": {},
  "timer": {}
}

AssertView

Command that adds ability to take screenshot for test state. Each state should have his own unique name. For example:

it('some test', async ({ browser }) => {
    await browser.url('some/url');
    await browser.assertView('plain', '.button');

    await browser.click('.button');
    await browser.assertView('clicked', '.button');
});

Could also be used as element's method:

it('some test', async ({ browser }) => {
    await browser.url('some/url');

    const elem = await browser.$('.button');

    await elem.assertView('plain');
    await elem.click();
    await elem.assertView('clicked');
});

Note: assertView will trigger waitForExist with waitTimeout and waitInterval

Parameters:

  • state (required) String – state name; should be unique within one test
  • selector (required) String|String[] – DOM-node selector that you need to capture
  • opts (optional) Object:
    • ignoreElements (optional) String|String[] – elements, matching specified selectors will be ignored when comparing images
    • tolerance (optional) Number – overrides config browsers.tolerance value
    • antialiasingTolerance (optional) Number – overrides config browsers.antialiasingTolerance value
    • ignoreDiffPixelCount (optional) Number | string - the maximum amount of different pixels to still consider screenshots "the same". For example, when set to 5, it means that if there are 5 or fewer different pixels between two screenshots, they will still be considered the same. Alternatively, you can also define the maximum difference as a percentage (for example, 3%) of the image size. This option is useful when you encounter a few pixels difference that cannot be eliminated using the tolerance and antialiasingTolerance settings. The default value is 0.
    • allowViewportOverflow (optional) Boolean – by default Hermione throws an error if element is outside the viewport bounds. This option disables check that element is outside of the viewport left, top, right or bottom bounds. And in this case if browser option compositeImage set to false, then only visible part of the element will be captured. But if compositeImage set to true (default), then in the resulting screenshot will appear the whole element with not visible parts outside of the bottom bounds of viewport.
    • captureElementFromTop (optional) Boolean - ability to set capture element from the top area or from current position. In the first case viewport will be scrolled to the top of the element. Default value is true
    • compositeImage (optional) Boolean - overrides config browsers.compositeImage value
    • screenshotDelay (optional) Number - overrides config browsers.screenshotDelay value
    • selectorToScroll (optional) String - DOM-node selector which should be scroll when the captured element does not completely fit on the screen. Useful when you capture the modal (popup). In this case a duplicate of the modal appears on the screenshot. That happens because we scroll the page using window selector, which scroll only the background of the modal, and the modal itself remains in place. Works only when compositeImage is true.
    • disableAnimation (optional): Boolean - ability to disable animations and transitions while capturing a screenshot.

All options inside assertView command override the same options in the browsers.assertViewOpts.

Full example:

it('some test', async ({ browser }) => {
    await browser.url('some/url');
    await browser.assertView(
        'plain', '.form',
        {
            ignoreElements: ['.link'],
            tolerance: 5,
            antialiasingTolerance: 4,
            allowViewportOverflow: true,
            captureElementFromTop: true,
            compositeImage: true,
            screenshotDelay: 10,
            selectorToScroll: '.modal'
        }
    );
});

For tests which have been just written using assertView command you need to update reference images, so for the first time hermione should be run with option --update-refs or via command gui which is provided by plugin html-reporter (we highly recommend to use gui command instead of option --update-refs).

RunStep

Command that allows to add human-readable description for commands in history, when it is enabled (html-reporter required) with saveHistoryMode. For example:

it('some test', async ({browser}) => {
    await browser.runStep('some step name', async () => {
        await browser.url('some/url');
        await browser.$('some-selector').click();
    });

    await browser.runStep('other step name', async () => {
        await browser.runStep('some nested step', async () => {
            await browser.$('not-exist').click();
        });
    });

    await browser.runStep('another step', async () => {
        ...
    });
});

Will produce the following history, if test fails on 'Can't call click on element with selector "not-exist" because element wasn't found':

  • hermione: init browser
  • some step name
  • other step name
    • some nested step
      • $("not-exist")
      • click()
        • waitForExist

In this example step some step name is collapsed, because it is completed successfully.

You can also return values from step:

const parsedPage = await browser.runStep('parse page', async () => {
    ...
    return someData;
});

Parameters:

  • stepName (required) String – step name
  • stepCb (required) Function – step callback

Note: html-reporter v9.7.7+ provides better history representation.

OpenAndWait

Command that allows to open page and wait until it loads (by combination of the specified factors). For example:

it('some test', async ({browser}) => {
    await browser.openAndWait('some/url', {
        selector: ['.some', '.selector'],
        predicate: () => document.isReady,
        ignoreNetworkErrorsPatterns: ['https://mc.yandex.ru'],
        waitNetworkIdle: true,
        waitNetworkIdleTimeout: 500,
        failOnNetworkError: true,
        timeout: 20000,
    });
});

In this example, page will be considered as loaded, if elements with selectors .some and .selector exists, document.isReady is true, it has been 500 ms since the last network request succeeded, and there were no errors, while trying to load images, fonts and stylesheets.

Parameters:

  • url (required) String – page url
  • waitOpts (optional) Object:
    • selector (optional) String|String[] – Selector(s) to element(s), which should exist on the page.
    • predicate (optional) () => Promise<bool> | bool – Predicate, which should return true if page is loaded. Predicate is being executed in the browser context: waitUntil.
    • waitNetworkIdle (optional) Boolean – Waits until all network requests are done. true by default. Only works in chrome browser or when using Chrome Devtools Protocol.
    • waitNetworkIdleTimeout (optional) Number - Time (ms) after all network requests are resolved to consider network idle. 500 by default.
    • failOnNetworkError (optional) Boolean – If true, throws an error when network requests are failed. true by default. Only works in chrome browser or when using Chrome Devtools Protocol.
    • shouldThrowError (optional) (match) => Boolean - Predicate, which should return true on Match, if network error is considered critical for page load. By default, throws an error on image, stylesheet and font load error.
    • ignoreNetworkErrorsPatterns (optional) Array<String | RegExp> - Array of url patterns to ignore network requests errors. Has a priority over shouldThrowError
    • timeout (optional) Number - Page load timeout. pageLoadTimeout by default. Throws an error, if selectors are still not exist after timeout, or predicate is still resolving false.

Typescript usage

To write hermione tests on typescript, you would need to install ts-node:

npm i -D ts-node

And include hermione types in your tsconfig.json file:

// tsconfig.json
{
    // other tsconfig options
    "compilerOptions": {
        // other compiler options
        "types": [
            // other types
            "hermione",
        ]
    }
}

Now you will be able to write hermione tests using typescript.

hermione.ctx typings

If you want to extend hermione.ctx typings, you could use module augmentation:

import type { HermioneCtx } from "hermione";

declare module "hermione" {
    interface HermioneCtx {
        someVariable: string;
    }
}

Now hermione.ctx will have someVariable typings

.hermione.conf.js

hermione is tuned using a configuration file. By default, it uses .hermione.conf.js, but you can use the --config option to specify a path to the configuration file.

There is only one required field – browsers.

module.exports = {
    browsers: {
        chrome: {
            desiredCapabilities: {
                browserName: 'chrome'
            }
        }
    }
};

sets

You can use sets to bind some set of tests to certain browsers.

Format of the sets section:

sets: {
    common: {                 // run tests associated with this path in all browsers
        files: 'tests/common' // which are configured in the `browsers` option
    },
    desktop: {
        files: [
            'tests/desktop/*.hermione.js',
            'tests/common/*.hermione.js'
        ],
        ignoreFiles: ['tests/desktop/fixtures/**'], // exclude directories from reading while test finding
        browsers: ['browser'] // run tests which match the specified masks in the browser with the `browser` id
    }
}

  • files – A list of test files or directories with test files. This can be a string if you want to specify just one file or directory. Also, you can use masks for this property.

  • ignoreFiles - A list of paths or masks to ignore from reading while test finding. When you write !**/some-dir/** it means that the directory will be read, but all the entries will not be included in the results. So using ignoreFiles you can speed up test reading for your project.

  • browsers – A list of browser IDs to run the tests specified in files. All browsers by default.

You can specify sets to run using the CLI option --set.

If sets are not specified in the config and paths were not passed from CLI, all files from the hermione directory are launched in all browsers specified in the config.

Running tests using sets:

hermione --set desktop

browsers

Required. The list of browsers to use for running tests.

The browser section has the following format

browsers: {
    <browser_id>: {
        <option>:<value>
        <option>:<value>
    }
}

<browser-id> value is used for browser identification.

Available browser options:

Option name Description
desiredCapabilities Required. Used WebDriver DesiredCapabilities
gridUrl Selenium grid URL. Default value is http://localhost:4444/wd/hub.
baseUrl Base service-under-test URL. Default value is http://localhost.
browserWSEndpoint Websocket endpoint to connect to the browser in order to be able to use CDP protocol. Default value is null.
automationProtocol Browser automation protocol. Default value is webdriver.
sessionEnvFlags Environment flags that determine which protocol will be used in created browser session. Default value is {}.
waitTimeout Timeout for web page event. Default value is 1000 ms.
waitInterval Interval for web page event. Default value is 250 ms.
httpTimeout Timeout for any requests to Selenium server. Default value is 30000 ms.
urlHttpTimeout Timeout for /url request to Selenium server. Default value is httpTimeout.
pageLoadTimeout Timeout for the page loading to complete. Default value is 20000 ms.
sessionRequestTimeout Timeout for getting a browser session. Default value is httpTimeout.
sessionQuitTimeout Timeout for quitting a session. Default value is 5000.
testTimeout Timeout for test execution (in milliseconds). Default value is null, in this case will be used common timeout for all browsers from system.mochaOpts.timeout.
sessionsPerBrowser Number of sessions which are run simultaneously. Default value is 1.
takeScreenshotOnFails Options for setting up taking a screenshot of a test fail. Default value is {testFail: true, assertViewFail: false}.
takeScreenshotOnFailsMode Mode for taking a screenshot on test fail. Available options are fullpage and viewport. Default value is fullpage.
takeScreenshotOnFailsTimeout Timeout for taking screenshot on test fail. Default value is 5000.
testsPerSession Maximum amount of tests (its) to run in each web driver session.
retry How many times a test should be rerun. Default value is 0.
shouldRetry Function that determines whether to make a retry. By default returns true if retry attempts are available otherwise returns false.
calibrate Allows to correctly capture the image. Default value is false.
meta Additional data that can be obtained via .getMeta() method.
windowSize Browser window dimensions. Default value is null.
screenshotDelay Allows to specify a delay (in milliseconds) before making any screenshot.
orientation Browser orientation that will be set before each test run. Default value is null.
waitOrientationChange Allows to wait until screen orientation is changed. Default value is true.
resetCursor Allows to configure whether to move mouse cursor to body coordinates (0, 0) before each test run.
tolerance Maximum allowed CIEDE2000 difference between colors. Default value is 2.3.
antialiasingTolerance Minimum difference in brightness between the darkest/lightest pixel (which is adjacent to the antiasing pixel) and theirs adjacent pixels. Default value is 4.
compareOpts Options for comparing images.
buildDiffOpts Options for building diff image.
assertViewOpts Options for assertView command, used by default.
openAndWaitOpts Options for openAndWaitOpts command, used by default
screenshotsDir Directory to save reference images for command assertView. Default dir is hermione/screens which is relative to process.cwd().
strictTestsOrder hermione will guarantee tests order in readTests results. false by default.
compositeImage Allows testing of regions which bottom bounds are outside of a viewport height. In the resulting screenshot the area which fits the viewport bounds will be joined with the area which is outside of the viewport height. true by default.
screenshotMode Image capture mode.
saveHistoryMode Allows to save history of executed commands. all by default.
agent Allows to use a custom http/https/http2 agent to make requests. Default value is null.
headers Allows to set custom headers to pass into every webdriver request. These headers aren't passed into browser request. Read more about this option in wdio. Default value is null.
transformRequest Allows to intercept HTTP request options before a WebDriver request is made. Default value is null.
transformResponse Allows to intercept HTTP response object after a WebDriver response has arrived. Default value is null.
strictSSL Whether it does require SSL certificate to be valid. Default value is null (it means that will be used default value from wdio).
user Cloud service username. Default value is null.
key Cloud service access key or secret key. Default value is null.
region Ability to choose different datacenters for run in cloud service. Default value is null.
headless Ability to run headless browser in cloud service. Default value is null.
isolation Ability to execute tests in isolated clean-state environment (incognito browser context). Default value is false, but true for chrome@93 and higher.

desiredCapabilities

Required. Used WebDriver DesiredCapabilities. For example,

'chrome': {
  desiredCapabilities: {
    browserName: 'chrome',
    version: '75.0', // or "browserVersion" if browser support w3c
    chromeOptions: {...}
  }
}

gridUrl

Selenium grid URL. Default value is http://localhost:4444/wd/hub.

baseUrl

Base service-under-test URL. Default value is http://localhost.

browserWSEndpoint

Websocket endpoint to connect to the browser in order to be able to use CDP protocol. For example you specify browserWSEndpoint: "ws://YOUR_HOST/devtools" to which sessionId of the browser will be added to the end: ws://YOUR_HOST/devtools/12345, where 12345 is a sessionId. Default value is null, it means that webdriverio is trying to figure out the websocket endpoint on its own.

automationProtocol

Browser automation protocol (webdriver, devtools) that will be used. Default value is webdriver.

sessionEnvFlags

Environment flags that determine which protocol will be used in created browser session. By default environment flags are set automatically according to the used desiredCapabilities but in rare cases they are determined inaccurately and using this option they can be overriden explicitly.

Available flags:

For example:

'chrome': {
    sessionEnvFlags: {
        isW3C: true,
        isChrome: true
    }
}

httpTimeout

Timeout for any requests to Selenium server. Default value is 30000 ms.

urlHttpTimeout

Timeout for /url request to Selenium server. Default value is httpTimeout. It may be useful when opening url takes a long time (for example a lot of logic is executed in middlewares), and you don't want to increase timeout for other commands.

pageLoadTimeout

Timeout for the page loading to complete. Default value is 20000 ms.

sessionRequestTimeout

Timeout for getting a browser session. Default value is httpTimeout.

sessionQuitTimeout

Timeout for quitting a session. Default value is 5000.

testTimeout

Timeout for test execution (in milliseconds). If applied to suite then timeout will be set for all tests and hooks inside this suite. Default value is null, in this case will be used common timeout for all browsers from system.mochaOpts.timeout.

waitTimeout

Timeout for waitUntil which is used to all waitFor* commands. It is used in the search for elements and web page events. Default value is 3000 ms.

For example: browser.$('.element').click() will wait up to 3000ms for element to exist before clicking it by default

waitInterval

Interval for waitUntil which is used to all waitFor* commands. It is used in element finding and web page events. Default value is 500 ms.

For example: browser.$('.element').click() will check element existence every 500ms by default

sessionsPerBrowser

Number of sessions which are run simultaneously. Global value for all browsers. Default value is 1.

takeScreenshotOnFails

Options for setting up taking a screenshot of a test fail. Can be an object with testFail and assertViewFail keys.

  • testFail (default: true) – takes a screenshot when an error occurs in the test, except assertView fail.
  • assertViewFail (default: true) – takes a screenshot if the test fails on the assetView command.

takeScreenshotOnFailsMode

Mode for taking a screenshot on test fail. There are two available options:

  • fullpage – Hermione will take a screenshot of the entire page from top. Default value.
  • viewport – Hermione will take a screenshot of the current viewport.

takeScreenshotOnFailsTimeout

Timeout for taking screenshot on test fail. Default value is 5000.

testsPerSession

Maximum amount of tests (its) to run in each web driver session. After limit is reached, session will be closed and new one will be started. By default is Infinity (no limit, all tests will be run in the same session). Set to smaller number in case of problems with stability.

retry

How many times a test should be retried if it fails. Global value for all browsers. Default value is 0.

shouldRetry

Function that determines whether to make a retry. Must return boolean value. By default returns true if retry attempts are available otherwise returns false. Argument of this function is object with fields:

  • retriesLeft {Number} — number of available retries
  • ctx — in case of test TEST_FAIL it would be bound data, in case of ERROR it would be link to Runnable
  • [error] — error type (available only in case of ERROR)

calibrate

Does this browser need to perform the calibration procedure. This procedure allows to correctly capture the image in case the particular WebDriver implementation captures browser UI along with web page. Default value is false.

meta

Additional data that can be obtained via .getMeta() method

windowSize

Browser window dimensions (i.e. 1600x1200). If not specified, the size of the window depends on WebDriver. Can be specified as string with pattern 800x1000 or object with width and height keys (both keys should be number). For example,

windowSize: '800x1000'

and

windowSize: {
  width: 800,
  height: 1000
}

are the same.

⚠️ You can't set specific resolutions for browser Opera or mobile platforms. They use only full-screen resolution.

screenshotDelay

Allows to specify a delay (in milliseconds) before making any screenshot. This is useful when the page has elements which are animated or if you do not want to screen a scrollbar. Default value is 0.

orientation

Browser orientation (landscape, portrait) that will be set before each test run. It is necessary in order to return the browser orientation to the default state after test execution in which orientation is changed. Default value is null.

waitOrientationChange

Allows to wait until screen orientation is changed. Works inside webdriverio commands orientation and setOrientation. This option guarantee that screen rotated before the next command will start to execute. Default value is true.

resetCursor

Allows to configure whether to move mouse cursor to body coordinates (0, 0) before each test run. This can be useful to escape cases when a default position of a cursor affects your tests. We recommend to set this option truthy value for desktop browsers and falsey for mobile devices. Default value is true.

tolerance

Indicates maximum allowed CIEDE2000 difference between colors. Used only in non-strict mode. By default it's 2.3 which should be enough for the most cases. Increasing global default is not recommended, prefer changing tolerance for particular suites or states instead.

antialiasingTolerance

Read more about this option in looks-same.

compareOpts

Extra options for comparing images. See looks-same documentation for the list of available options. Default values are:

compareOpts: {
    shouldCluster: false,
    clustersSize: 10,
    stopOnFirstFail: false
}

buildDiffOpts

Extra options for building diff image. See looks-same documentation for the list of available options. Default values are:

buildDiffOpts: {
    ignoreAntialiasing: true,
    ignoreCaret: true
}

assertViewOpts

Default options used when calling assertView, can be overridden by assertView options. Default values are:

    ignoreElements: [],
    captureElementFromTop: true,
    allowViewportOverflow: false,
    disableAnimation: true,
    ignoreDiffPixelCount: 0,

openAndWaitOpts

Default options used when calling openAndWait, can be overriden by openAndWait options. Default values are:

    waitNetworkIdle: true,
    waitNetworkIdleTimeout: 500,
    failOnNetworkError: true,
    ignoreNetworkErrorsPatterns: []

screenshotsDir

Directory to save reference images for command assertView. Default dir is hermione/screens which is relative to process.cwd(). The value of this option can also be a function which accepts one argument - an instance of a test within which comand assertView is called:

    screenshotsDir: (test) => `tests/screenshots/${test.parent.title}`

strictTestsOrder

hermione will guarantee tests order in readTests results. false by default.

compositeImage

Allows testing of regions which bottom bounds are outside of a viewport height (default: true). In the resulting screenshot the area which fits the viewport bounds will be joined with the area which is outside of the viewport height.

screenshotMode

Image capture mode. There are 3 allowed values for this option:

  • auto (default). Mode will be obtained automatically;
  • fullpage. Hermione will deal with screenshot of full page;
  • viewport. Only viewport area will be used.

By default, screenshotMode on android browsers is set to viewport to work around the chromium bug.

saveHistoryMode

Allows to save history of all executed commands. 'all' by default.

Available options:

  • 'all' - history is enabled
  • 'none' - history is disabled
  • 'onlyFailed' - history is saved for failed tests only

Some plugins can rely on this history, for instance:

The history is available from such events: TEST_END, TEST_PASS, TEST_FAIL through payload:

// example of plugin code
module.exports = (hermione) => {
    hermione.on(hermione.events.TEST_PASS, async (test) => {
        console.log(test.history);
    });
};

agent

Allows to use a custom http/https/http2 agent to make requests. Default value is null (it means that will be used default http-agent from got).

headers

Allows to set custom headers to pass into every webdriver request. These headers aren't passed into browser request. Read more about this option in wdio. Default value is null.

transformRequest

Allows to intercept HTTP request options before a WebDriver request is made. Default value is null. If function is passed then it takes RequestOptions as the first argument and should return modified RequestOptions. Example:

(RequestOptions) => RequestOptions

In runtime a unique X-Request-ID header is generated for each browser request which consists of ${FIRST_X_REQ_ID}__${LAST_X_REQ_ID}, where:

  • FIRST_X_REQ_ID - unique uuid for each test (different for each retry), allows to find all requests related to a single test run;
  • LAST_X_REQ_ID - unique uuid for each browser request, allows to find one unique request in one test (together with FIRST_X_REQ_ID).

Header X-Request-ID can be useful if you manage the cloud with browsers yourself and collect logs with requests. Real-world example: 2f31ffb7-369d-41f4-bbb8-77744615d2eb__e8d011d8-bb76-42b9-b80e-02f03b8d6fe1.

To override generated X-Request-ID to your own value you need specify it in transformRequest handler. Example:

transformRequest: (req) => {
    req.headers["X-Request-ID"] = "my_x_req_id";
}

transformResponse

Allows to intercept HTTP response object after a WebDriver response has arrived. Default value is null. If function is passed then it takes Response (original response object) as the first and RequestOptions as the second argument. Should return modified Response. Example:

(Response, RequestOptions) => Response

strictSSL

Whether it does require SSL certificate to be valid. Default value is null (it means that will be used default value from wdio).

user

Cloud service username. Default value is null.

key

Cloud service access key or secret key. Default value is null.

region

Ability to choose different datacenters for run in cloud service. Default value is null.

headless

Ability to run headless browser in cloud service. Default value is null. Can be set as a Boolean (the default value of the browser will be used). For Chrome browsers starting from version 112 also can be specified as a string with "new"|"old" values - this will enable the new headless mode (see Chrome's blog post).

isolation

Ability to execute tests in isolated clean-state environment (incognito browser context). It means that testsPerSession can be set to Infinity in order to speed up tests execution and save browser resources. Currently works only in chrome@93 and higher. Default value is null, but true for chrome@93 and higher.

system

debug

Turn webdriver debug mode on. Default value is false.

mochaOpts

Extra options for mocha which are passed to mocha.setup. See Mocha documentation for the list of options. Default values are:

mochaOpts: {
    slow: 10000, // If test execution time is greater than this value, then the test is slow.
    timeout: 60000 // timeout for test execution.
}

expectOpts

Options for expect-webdriverio that allow you to change default wait timeout and interval between attempts. Default values are:

expectOpts: {
    wait: 3000, // wait timeout for expectation to succeed
    interval: 100 // interval between attempts
}

ctx

A context which will be available in tests via method hermione.ctx:

ctx: {
    foo: 'bar'
}
it('awesome test', function() {
    console.log(hermione.ctx); // {foo: 'bar'}
});

Recommendation: use ctx in your tests in favor of global variables.

patternsOnReject

Session would be rejected if test has been faile with error message which matches to specified patterns:

patternsOnReject: [
    /some-pattern/,
    'other-pattern'
]

workers

Hermione runs all tests in subprocesses in order to decrease the main process CPU usage. This options defines the numbers of subprocesses to start for running tests. Default value is 1.

testsPerWorker

The maximum number of tests to be run in one worker before it will be restarted.

parallelLimit

By default, hermione will run all browsers simultaneously. Sometimes (i.e. when using cloud services, such as SauceLabs) you have to limit the amount of browsers that can be run at the same time. This option effectively limits how many browsers hermione will try to run in parallel. Default value is Infinity.

fileExtensions

Ability to set file extensions, which hermione will search on the file system. Default value is [.js].

plugins

Hermione plugins are commonly used to extend built-in functionality. For example, html-reporter and hermione-safari-commands.

A plugin is a module that exports a single function. The function has two arguments:

  • The hermione instance
  • Plugin options from the configuration file

Plugins will be loaded before hermione runs tests.

It's strongly recommended to name hermione plugins with the hermione- prefix. This makes searching for user plugins very simple.

If a plugin name starts with hermione-, then the prefix can be ommited in the configuration file. If two modules with names hermione-some-module and some-module are specified, the module with the prefix will have higher priority.

For example:

// .hermione.conf.js
// ...
plugins: {
    'my-cool-plugin': {
        param: 'value'
    }
}
// ...

// hermione-my-cool-plugin/index.js
module.exports = function(hermione, opts) {
    hermione.on(hermione.events.RUNNER_START, function(runner) {
        return setUp(hermione.config, opts.param); // config can be mutated
    });

    hermione.on(hermione.events.RUNNER_END, function() {
        return tearDown();
    });
}

Properties of the hermione object

Property name Description
config Config that is used in the test runner. Can be mutated.
events Events list for subscription.

Available events which are triggered in the main process

Event Description
INIT Will be triggered before any job start (run or readTests). If handler returns a promise then job will start only after the promise will be resolved. Emitted only once no matter how many times job will be performed.
BEFORE_FILE_READ Will be triggered on test files parsing before reading the file. The handler accepts data object with file, hermione (helper which will be available in test file) and testParser (TestParserAPI object) fields.
AFTER_FILE_READ Will be triggered on test files parsing right after reading the file. The handler accepts data object with file and hermione (helper which will be available in test file) fields.
AFTER_TESTS_READ Will be triggered right after tests read via readTests or run methods with TestCollection object.
RUNNER_START Will be triggered after worker farm initialization and before test execution. If a handler returns a promise, tests will be executed only after the promise is resolved. The handler accepts an instance of a runner as the first argument. You can use this instance to emit and subscribe to any other available events.
RUNNER_END Will be triggered after test execution and before worker farm ends. If a handler returns a promise, worker farm will be ended only after the promise is resolved. The handler accepts an object with tests execution statistics.
NEW_WORKER_PROCESS Will be triggered after new subprocess is spawned. The handler accepts a restricted process object with only send method.
SESSION_START Will be triggered after browser session initialization. If a handler returns a promise, tests will be executed only after the promise is resolved. The handler accepts an instance of webdriverIO as the first argument and an object with a browser identifier as the second.
SESSION_END Will be triggered after the browser session ends. If a handler returns a promise, tests will be executed only after the promise is resolved. The handler accepts an instance of webdriverIO as the first argument and an object with a browser identifier as the second.
BEGIN Will be triggered before test execution, but after all the runners are initialized.
END Will be triggered just before RUNNER_END event. The handler accepts a stats of tests execution.
SUITE_BEGIN Test suite is about to execute.
SUITE_END Test suite execution is finished.
TEST_BEGIN Test is about to execute.
TEST_END Test execution is finished.
TEST_PASS Test passed.
TEST_FAIL Test failed.
TEST_PENDING Test is skipped.
RETRY Test failed but went to retry.
ERROR Generic (no tests) errors.
INFO Reserved.
WARNING Reserved.
EXIT Will be triggered when SIGTERM is received (for example, Ctrl + C). The handler can return a promise.

Available events which are triggered in subprocesses

Event Description
BEFORE_FILE_READ Will be triggered on test files parsing before reading the file. The handler accepts data object with file, hermione (helper which will be available in test file) and testParser (TestParserAPI object) fields.
AFTER_FILE_READ Will be triggered on test files parsing right after reading the file. The handler accepts data object with file and hermione (helper which will be available in test file) fields.
AFTER_TESTS_READ Will be triggered right after tests read each time some file is being reading during test run.
NEW_BROWSER Will be triggered after new browser instance created. The handler accepts an instance of webdriverIO as the first argument and an object with a browser identifier and version as the second.
UPDATE_REFERENCE Will be triggered after updating reference image.

REMARK!

Events which are triggered in the main process and subprocesses can not share information between each other, for example:

module.exports = (hermione) => {
    let flag = false;

    hermione.on(hermione.events.RUNNER_START, () => {
        flag = true;
    });

    hermione.on(hermione.events.NEW_BROWSER, () => {
        // outputs `false`, because `NEW_BROWSER` event was triggered in a subprocess,
        // but `RUNNER_START` was not
        console.log(flag);
    });

    hermione.on(hermione.events.RUNNER_END, () => {
        // outputs `true`
        console.log(flag);
    });
};

But you can solve such problem this way:

module.exports = (hermione, opts) => {
    hermione.on(hermione.events.RUNNER_START, () => {
      opts.flag = true;
    });

    hermione.on(hermione.events.NEW_BROWSER, () => {
        // outputs `true`, because properties in a config (variable `opts` is a part of a config)
        // which have raw data types are passed to subprocesses after `RUNNER_START` event
        console.log(opts.flag);
    });
};

Besides, you have the ability to intercept events in plugins and translate them to other events:

module.exports = (hermione) => {
    hermione.intercept(hermione.events.TEST_FAIL, ({event, data: test}) => {
        test.skip({reason: 'intercepted failure'});

        return {event: hermione.events.TEST_PENDING, test};
    });

    hermione.on(hermione.events.TEST_FAIL, (test) => {
        // this event handler will never be called
    });

    hermione.on(hermione.evenst.TEST_PENDING, (test) => {
        // this event handler will always be called instead of 'TEST_FAIL' one
    });
};

If for some reason interceptor should not translate passed event to another one you can return the same object or some falsey value:

module.exports = (hermione) => {
    hermione.intercept(hermione.events.TEST_FAIL, ({event, data}) => {
        return {event, data};
        // return;
        // return null;
        // return false;
    });

    hermione.on(hermione.events.TEST_FAIL, (test) => {
        // this event handler will be called as usual because interceptor does not change event
    });
};

If for some reason interceptor should ignore passed event and do not translate it to any other listeners you can return an empty object:

module.exports = (hermione) => {
    hermione.intercept(hermione.events.TEST_FAIL, ({event, data}) => {
        return {};
    });

    hermione.on(hermione.events.TEST_FAIL, (test) => {
        // this event handler will NEVER be called because interceptor ignores it
    });
};

The above feature can be used to delay triggering of some events, for example:

module.exports = (hermione) => {
  const intercepted = [];

  hermione.intercept(hermione.events.TEST_FAIL, ({event, data}) => {
        intercepted.push({event, data});
        return {};
    });

    hermione.on(hermione.events.END, () => {
        intercepted.forEach(({event, data}) => hermione.emit(event, data));
    });
};

Available events which can be intercepted

Event
SUITE_BEGIN
SUITE_END
TEST_BEGIN
TEST_END
TEST_PASS
TEST_FAIL
RETRY

Parallel execution plugin code

Runner has a method registerWorkers which register plugin's code for parallel execution in Hermione's worker farm. The method accepts parameters workerFilepath (string, absolute path), exportedMethods (array of string) and return object which contains async functions with names from exportedMethods. File with path workerFilepath should export object which contains async functions with names from exportedMethods.

Example

// plugin code
let workers;

module.exports = (hermione) => {
    hermione.on(hermione.events.RUNNER_START, async (runner) => {
        const workerFilepath = require.resolve('./worker');
        const exportedMethods = ['foo'];
        workers = runner.registerWorkers(workerFilepath, exportedMethods);

        // outputs `FOO_RUNNER_START`
        console.log(await workers.foo('RUNNER_START'));
    });

    hermione.on(hermione.events.RUNNER_END, async () => {
        // outputs `FOO_RUNNER_END`
        console.log(await workers.foo('RUNNER_END'));
    });
};

// worker.js
module.exports = {
    foo: async function(event) {
        return 'FOO_' + event;
    }
};

List of useful plugins

There are several plugins that may be useful:

prepareBrowser

Prepare the browser session before tests are run. For example, add custom user commands.

prepareBrowser: function(browser) {
    browser.addCommand('commandName', require('./path/to/commands/commandName.js'));
}

The browser argument is a WebdriverIO session.

prepareEnvironment

Configuration data can be changed depending on extra conditions in the prepareEnvironment function.

devServer

Launch dev server on hermione initialize (INIT event).

For example, this setup:

// .hermione.conf.js
const SERVER_PORT = 3000;
...
module.exports = {
    devServer: {
        command: "npm run server:dev",
        env: {PORT: SERVER_PORT},
        readinessProbe: {
            url: `http://localhost:${SERVER_PORT}/health`
        }
    }
}

Will spawn child process "npm run server:dev", pass extra environment variable "PORT" with value "3000" and wait, until "http://localhost:3000/health" responds with 200-299 status code.

Full list of parameters:

  • command (optional) String – command to launch dev server. If null or not defined, dev server is disabled
  • env (optional) Record<string, string> – extra environment variables to pass to child process, in addition to your process.env
  • args (optional) String[] – arguments to pass to child process
  • cwd (optional) String – current working directory of the child process. If not defined, hermione will try to find nearest "package.json", starting from the directory with hermione config
  • logs (optional) Boolean – if enabled, shows dev server logs in the console with prefix "[dev server]". Enabled by default
  • readinessProbe (optional) Function | Object - if function, ready check is completed when function is resolved. Receives child process object. Object by default
    • url (optional) String – url to request ready check status. If not defined, ready check is disabled
    • isReady (optional) (fetchResponse => bool | Promise<bool>) – predicate to check if server is ready based on readinessProbe.url fetch response. Returns true if statusCode is 2xx by default
    • timeouts (optional) Object – server waiting timeouts
      • waitServerTimeout (optional) Number - timeout to wait for server to be ready (ms). 60_000 by default
      • probeRequestTimeout (optional) Number - one request timeout (ms), after which request will be aborted. 10_000 by default
      • probeRequestInterval (optional) Number - interval between ready probe requests (ms). 1_000 by default

CLI

hermione --help

shows the following

  Usage: hermione [options] [paths...]

  Options:

    -V, --version                output the version number
    -c, --config <path>          path to configuration file
    -b, --browser <browser>      run tests only in specified browser
    -s, --set <set>              run tests only in the specified set
    -r, --require <module>       require a module before running hermione
    --reporter <reporter>        test reporters
    --grep <grep>                run only tests matching the pattern
    --update-refs                update screenshot references or gather them if they do not exist ("assertView" command)
    --inspect [inspect]          nodejs inspector on [=[host:]port]
    --inspect-brk [inspect-brk]  nodejs inspector with break at the start
    --repl [type]                run one test, call `browser.switchToRepl` in test code to open repl interface (default: false)
    --repl-before-test [type]    open repl interface before test run (default: false)
    --repl-on-fail [type]        open repl interface on test fail only (default: false)
    -h, --help                   output usage information

For example,

hermione --config ./config.js --reporter flat --browser firefox --grep name

Note. All CLI options override config values.

Reporters

You can choose flat, plain or jsonl reporter by option --reporter. Default is flat. Information about test results is displayed to the command line by default. But there is an ability to redirect the output to a file, for example:

hermione --reporter '{"type": "jsonl", "path": "./some-path/result.jsonl"}'

In that example specified file path and all directories will be created automatically. Moreover you can use few reporters:

hermione --reporter '{"type": "jsonl", "path": "./some-path/result.jsonl"}' --reporter flat

Information about each report type:

  • flat – all information about failed and retried tests would be grouped by browsers at the end of the report;
  • plain – information about fails and retries would be placed after each test;
  • jsonl - displays detailed information about each test result in jsonl format.

Require modules

Using -r or --require option you can load external modules, which exists in your local machine, before running hermione. This is useful for:

  • compilers such as TypeScript via ts-node (using --require ts-node/register) or Babel via @babel/register (using --require @babel/register);
  • loaders such as ECMAScript modules via esm.

Be sure to update fileExtensions apropriately, if you are planning to import anything other than .js.

Overriding settings

All options can also be overridden via command-line flags or environment variables. Priorities are the following:

  • A command-line option has the highest priority. It overrides the environment variable and config file value.

  • An environment variable has second priority. It overrides the config file value.

  • A config file value has the lowest priority.

  • If there isn't a command-line option, environment variable or config file option specified, the default is used.

To override a config setting with a CLI option, convert the full option path to --kebab-case. For example, if you want to run tests against a different base URL, call:

hermione path/to/mytest.js --base-url http://example.com

To change the number of sessions for Firefox (assuming you have a browser with the firefox id in the config):

hermione path/to/mytest.js --browsers-firefox-sessions-per-browser 7

To override a setting with an environment variable, convert its full path to snake_case and add the hermione_ prefix. The above examples can be rewritten to use environment variables instead of CLI options:

hermione_base_url=http://example.com hermione path/to/mytest.js
hermione_browsers_firefox_sessions_per_browser=7 hermione path/to/mytest.js

Debug mode

In order to understand what is going on in the test step by step, there is a debug mode. You can run tests in this mode using these options: --inspect and --inspect-brk. The difference between them is that the second one stops before executing the code.

Example:

hermione path/to/mytest.js --inspect

Note: In the debugging mode, only one worker is started and all tests are performed only in it. Use this mode with option sessionsPerBrowser=1 in order to debug tests one at a time.

REPL mode

Hermione provides a REPL implementation that helps you not only to learn the framework API, but also to debug and inspect your tests. In this mode, there is no timeout for the duration of the test (it means that there will be enough time to debug the test). It can be used by specifying the CLI options:

  • --repl - in this mode, only one test in one browser should be run, otherwise an error is thrown. REPL interface does not start automatically, so you need to call switchToRepl command in the test code. Disabled by default;
  • --repl-before-test - the same as --repl option except that REPL interface opens automatically before run test. Disabled by default;
  • --repl-on-fail - the same as --repl option except that REPL interface opens automatically on test fail. Disabled by default.

switchToRepl

Browser command that stops the test execution and opens REPL interface in order to communicate with browser. For example:

it('foo', async ({browser}) => {
    console.log('before open repl');

    await browser.switchToRepl();

    console.log('after open repl');
});

And run it using the command:

npx hermione --repl --grep "foo" -b "chrome"

In this case, we are running only one test in one browser (or you can use hermione.only.in('chrome') before it + it.only). When executing the test, the text before open repl will be displayed in the console first, then test execution stops, REPL interface is opened and waits your commands. So we can write some command in the terminal:

await browser.getUrl();
// about:blank

In the case when you need to execute a block of code, for example:

for (const item of [...Array(3).keys]) {
    await browser.$(`.selector_${item}`).isDisplayed();
}

You need to switch to editor mode by running the .editor command in REPL and insert the desired a block of code. Then execute it by pressing Ctrl+D. It is worth considering that some of code can be executed without editor mode:

  • one-line code like await browser.getUrl().then(console.log);
  • few lines of code without using block scope or chaining, for example: 
    await browser.url('http://localhost:3000');
await browser.getUrl();
// http://localhost:3000

After user closes the server, the test will continue to run (text after open repl will be displayed in the console and browser will close).

Another command features:

  • all const and let declarations called in REPL mode are modified to var in runtime. This is done in order to be able to redefine created variables;

  • before switching to the REPL mode process.cwd is replaced with the path to the folder of the executed test. After exiting from the REPL mode process.cwd is restored. This feature allows you to import modules relative to the test correctly;

  • ability to pass the context to the REPL interface. For example:

    it('foo', async ({browser}) => {
    const foo = 1;

    await browser.switchToRepl({foo});
});

    And now foo variable is available in REPL:

    console.log("foo:", foo);
// foo: 1

Test development in runtime

For quick test development without restarting the test or the browser, you can run the test in the terminal of IDE with enabled REPL mode:

npx hermione --repl-before-test --grep "foo" -b "chrome"

After that, you need to configure the hotkey in IDE to run the selected one or more lines of code in the terminal. As a result, each new written line can be sent to the terminal using a hotkey and due to this, you can write a test much faster.

Also, during the test development process, it may be necessary to execute commands in a clean environment (without side effects from already executed commands). You can achieve this with the following commands:

  • clearSession - clears session state (deletes cookies, clears local and session storages). In some cases, the environment may contain side effects from already executed commands;
  • reloadSession - creates a new session with a completely clean environment.
How to set up using VSCode
  1. Open Code -> Settings... -> Keyboard Shortcuts and print run selected text to search input. After that, you can specify the desired key combination
  2. Run hermione in repl mode (examples were above)
  3. Select one or mode lines of code and press created hotkey
How to set up using Webstorm

Ability to run selected text in terminal will be available after this issue will be resolved.

Environment variables

HERMIONE_SKIP_BROWSERS

Skip the browsers specified in the config by passing the browser IDs. Multiple browser IDs should be separated by commas (spaces after commas are allowed).

For example,

HERMIONE_SKIP_BROWSERS=ie10,ie11 hermione

HERMIONE_SETS

Specify sets to run using the environment variable as an alternative to using the CLI option --set.

For example,

HERMIONE_SETS=desktop,touch hermione

Programmatic API

With the API, you can use Hermione programmatically in your scripts or build tools. To do this, you must require hermione module and create instance:

const Hermione = require('hermione');
const hermione = new Hermione(config);
  • config (required) Object|String – Configuration object or path to the configuration file that will be read relative to process.cwd.

Next, you will have access to the following parameters and methods:

Name Description
config Returns parsed hermione config.
events Returns hermione events on which you can subscribe to.
errors Errors which hermione may return.
intercept Method to intercept Hermione's events.
run Starts running tests. By default run all tests from the config. Can also run only the specified tests.
addTestToRun Adds test to the current run.
readTests Starts reading tests. By default read all tests from the config. Can also read only the specified tests.
isFailed Returns true or false depending on whether there has been an error or a test fail while running tests.
isWorker Returns true or false depending on whether you call the method in one of the workers or in the master process.
halt Abnormal termination of the test run in case of a terminal error.

config

Returns parsed hermione config. Useful when you need to read some field from the config.

// create hermione instance
console.log('plugins:', hermione.config.plugins);

events

Returns hermione events on which you can subscribe to. Useful when you need to subscribe to a specific event.

// create hermione instance
hermione.on(hermione.events.INIT, async () => {
    console.info('INIT event is being processed...');
});

errors

Hermione may return errors of the following type:

CoreError

A CoreError is returned if the browser fails to calibrate a blank page (about:blank). The error contains the following message:

Could not calibrate. This could be due to calibration page has failed to open properly

CancelledError

The CanceledEror error is returned if the halt command terminates abnormally. The error contains the following message:

Browser request was cancelled

ClientBridgeError

A ClientBridgeError is returned when JavaScript injection on the client (browser) side fails. Hermione injects the code using the execute WebDriverIO command. The error contains the following message:

Unable to inject client script

HeightViewportError

The HeightViewportError is returned when trying to take a screenshot of an area whose bottom border does not fit into the viewport area. The error contains the following message:

Can not capture the specified region of the viewport.
The region bottom bound is outside of the viewport height.
Alternatively, you can test such cases by setting "true" value to option "compositeImage" in the config file
or setting "false" to "compositeImage" and "true" to option "allowViewportOverflow" in "assertView" command.
Element position: <cropArea.left>, <cropArea.top>; size: <cropArea.width>, <cropArea.height>.
Viewport size: <viewport.width>, <viewport.height>.

In this case, the message prompts the Hermione user what settings need to be set in the Hermione config in order to be able to take a screenshot for the specified area.

OffsetViewportError

An OffsetViewportError is returned when attempting to take a screenshot of an area whose borders on the left, right, or top extend beyond the viewport. The error contains the following message:

Can not capture the specified region of the viewport.
Position of the region is outside of the viewport left, top or right bounds.
Check that elements:
- does not overflow the document
- does not overflow browser viewport
Alternatively, you can increase browser window size using
"setWindowSize" or "windowSize" option in the config file.
But if viewport overflow is expected behavior then you can use
option "allowViewportOverflow" in "assertView" command.

In this case, the message prompts the Hermione user what settings need to be set in the Hermione config in order to be able to take a screenshot for the specified area.

AssertViewError

An AssertViewError is returned when an attempt to take a screenshot fails. The error may contain one of the following messages, depending on the cause of the crash:

duplicate name for "<state>" state

element ("<selector>") still not existing after <this.options.waitforTimeout> ms

element ("<this.selector>") still not existing after <this.options.waitforTimeout> ms

ImageDiffError

An ImageDiffError is returned from the assertView command if a diff (difference in images) is detected when taking and comparing a screenshot with a reference screenshot. The error contains the following message:

images are different for "<stateName>" state

In addition, the ImageDiffError error contains the following data:

  • stateName String - Name of the state for which the screenshot was taken.
  • currImg Object - Link to actual image.
  • refImg Object - Link to reference image.
  • diffOpts Object - Diff detection settings.
  • diffBounds Object - Boundaries of areas with diffs in the image.
  • diffClusters Object - Clusters with diffs in the image.

Read more about diffBounds and diffClusters in the documentation of the looks-same package.

NoRefImageError

The NoRefImageError error is returned from the assertView command if, when taking and comparing a screenshot, Hermione does not find a reference screenshot on the file system. The error contains the following message:

can not find reference image at <refImg.path> for "<stateName>" state

In addition, the NoRefImageError error contains the following data:

  • stateName String - Name of the state for which the screenshot was taken.
  • currImg Object - Link to actual image.
  • refImg Object - Link to reference image.

intercept

Method to intercept Hermione's events. The first argument of the method is the event that needs to be intercepted, and the second is the event handler. Example:

// create hermione instance
hermione.intercept(hermione.events.TEST_FAIL, ({ event, data }) => {
    return {event: hermione.events.TEST_PASS, test}; // make the test successful
});

Read more about event interception in the section about plugins.

run

Starts running tests. By default run all tests from the config. Can also run only the specified tests. Returns true if the test run succeeded, and false if it failed. Example:

// create hermione instance
const success = await hermione.run(testPaths, options);

Available parameters:

  • testPaths (optional) String[]|TestCollection – Paths to tests relative to process.cwd. Also accepts test collection returned by readTests.
  • options (optional) Object
    • reporters (optional) String[] – Test result reporters.
    • browsers (optional) String[] – Browsers to run tests in.
    • sets (optional) String[]– Sets to run tests in.
    • grep (optional) RegExp – Pattern that defines which tests to run.

addTestToRun

Adds test to the current run. Returns false if the current run has already ended or has been cancelled. Otherwise returns true. Example:

// create hermione instance
const success = hermione.addTestToRun(test, browserId);
  • test (required) Test – Test to run.
  • browserId (required) String – Browser to run test in.

readTests

Starts reading tests. By default read all tests from the config. Can also read only the specified tests. Returns promise which resolves to the instance of TestCollection initialized by parsed tests. Example:

// create hermione instance
await hermione.readTests(testPaths, options);
  • testPaths (required) String[] – Paths to tests relative to process.cwd.
  • options (optional) Object:
    • browsers (optional) String[] – Read tests only for the specified browsers.
    • silent (optional) Boolean – flag to disable events emitting while reading tests; default is false.
    • ignore (optional) String|Glob|Array<String|Glob> - patterns to exclude paths from the test search.
    • sets (optional) String[]– Sets to run tests in.
    • grep (optional) RegExp – Pattern that defines which tests to run.

isFailed

Returns true or false depending on whether there has been an error or a test fail while running tests. Can be useful in plugins to determine Hermione's current status. Example:

// create hermione instance
const failed = hermione.isFailed();

isWorker

Returns true or false depending on whether you call the method in one of the workers or in the master process. Can be useful in plugins in order to distinguish the code execution context. Example:

// implementation of some plugin
module.exports = (hermione) => {
    if (hermione.isWorker()) {
        // do some stuff only in workers
    } else {
        // do some stuff only in the master process
    }
};

halt

Abnormal termination of the test run in case of a terminal error. If process fails to gracefully shutdown in timeout milliseconds, it would be forcibly terminated (unless timeout is explicitly set to 0).

// create hermione instance
hermione.halt(error, [timeout=60000ms]);

Test Collection

TestCollection object is returned by hermione.readTests method.

TestCollection API:

  • getBrowsers() — returns list of browsers for which there are tests in collection.

  • mapTests(browserId, (test, browserId) => ...) - maps over tests for passed browser. If first argument (browserId) is omitted then method will map over tests for all browsers.

  • sortTests(browserId, (currentTest, nextTest) => ...) - sorts over tests for passed browser. If first argument (browserId) is omitted then method will sort over tests for all browsers.

  • eachTest(browserId, (test, browserId) => ...) - iterates over tests for passed browser. If first argument (browserId) is omitted then method will iterate over tests for all browsers.

  • eachTestByVersion(browserId, (test, browserId, version) => ...) - iterates over tests and over browser versions for passed browser. Implicitly sets browser version for each test in browserVersion property.

  • disableAll([browserId]) - disables all tests. Disables tests for specific browser if browserId passed. Returns current test collection instance.

  • enableAll([browserId]) - enables all previously disabled tests. Enables tests for specific browser if browserId passed. Returns current test collection instance.

  • disableTest(fullTitle, [browserId]) - disables test with passed full title. Disables test only in specific browser if browserId passed. Returns current test collection instance.

  • enableTest(fullTitle, [browserId]) - enables test with passed full title. Enables test only in specific browser if browserId passed. Returns current test collection instance.

  • getRootSuite(browserId) - returns root suite for passed browser. Returns undefined if there are no tests in collection for passed browser.

  • eachRootSuite((root, browserId) => ...) - iterates over all root suites in collection which have some tests.

Test Parser API

TestParserAPI object is emitted on BEFORE_FILE_READ event. It provides the ability to customize test parsing process.

setController(name, methods)

Adds controller to hermione object in test files.

  • name - controller name
  • methods - an object with names as keys and callbacks as values describing controller methods. Each callback will be called on corresponding test or suite.

Example:

// in plugin
hermione.on(hermione.events.BEFORE_FILE_READ, ({file, testParser}) => {
    testParser.setController('logger', {
        log: function(prefix) {
            console.log(`${prefix}: Just parsed ${this.fullTitle()} from file ${file}`);
        }
    });
});

// in test file
describe('foo', () => {
    hermione.logger.log('some-prefix');
    it('bar', function() {
        // ...
    });
});

Note: controller will be removed as soon as current file will be parsed

Readme

Keywords

Package Sidebar

Install

npm i hermione

Repository

github.com/gemini-testing/hermione

Homepage

github.com/gemini-testing/hermione#readme

Weekly Downloads

1,848

Version

8.9.1

License

MIT

Unpacked Size

819 kB

Total Files

469

Last publish

Collaborators

  • miripiruni
  • egavr
  • mcheshkov
  • j0tunn
  • unlok
  • dudagod
  • kolesnikovde
  • sipayrt
  • rostik404
  • kvmamich
  • xrsd
  • oldskytree
  • shadowusr
  • kuznetsovroman
Try on RunKit
Report malware