Skip to content

Widget that provides high-level commands for UI testing for use with Xolv.io chimp

Notifications You must be signed in to change notification settings

tikolakin/chimp-widgets

 
 

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

36 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

chimp-widgets

Write award winning step definitions while working with chimp / meteor-cucumber.

Installation

npm install chimp-widgets

Why?

The management of step definitions is arguably one of the biggest weaknesses of cucumber and can get out of hand quite fast. There are multiple reasons that contribute to this problem but chimp-widgets tackles one of them specifically: The duplication of DOM selectors in step defintions.

Let me show you an example of some innocent looking steps:

this.When(/^I go to my projects$/, function(done) {
  this.browser.url(process.env.ROOT_URL + 'projects')
  .waitForVisible('.projects-list')
  .call(done);
});

this.Then(/^I should see my projects$/, function(done) {
  this.browser.waitForVisible('.projects-list').call(done);
});

this.When(/^I select the first project$/, function(done) {
  this.browser.click('.projects-list .project:nth-child(1)')
  .call(done);
});

When we look closer we realize that they contain several (implicit) facts about a project page and its list markup:

  1. The URL of the project page process.env.ROOT_URL + 'projects'
  2. The CSS selector of the project list .projects-list
  3. The nested CSS selector to access a project .projects-list .project

The problem is hidden behind the word implicit, because those three steps also have in common that they never talk about the real thing (the project page) explicitely. To understand what is happening you have to parse the calls to the webdriver API and infer the action targets from DOM selectors.

The Page Objects Pattern

To make the interaction with the DOM more explicit, we can apply the Page Objects Pattern like this:

this.When(/^I go to my projects$/, function() {
  return this.ProjectsPage.visit();
});

this.Then(/^I should see my projects$/, function() {
  return new this.ProjectsPage().waitForVisible();
});

this.When(/^I select the first project$/, function() {
  return new this.ProjectsPage().selectProjectAt(1);
});

Wow, look at that – we got rid of all the DOM selectors and used some kind of class to represent a real world concept (project page) within our application. This class and its instances expose a simple API that hide the implementation details for the reader. Additionally we could get rid of the done parameter because all API methods return a Promise to indicate success or failure to the test runner.

How?

Let's look at the definition of the ProjectsPage class:

module.exports = function() {
  this.Before(function(done) {

    this.ProjectsPage = this.widgets.Widget.extend({

      selector: '.projects-list',

      selectProjectAt: function(index) {
        this.find('.project:nth-child('+index+')').click();
      }
    });

    this.ProjectsPage.url = process.env.ROOT_URL + 'projects';

    done();
  });
};

Let's analyize what we did here, step by step:

  1. Setup a standard Before hook from meteor-cucumber which provides access to the world, so we can expose our widget class to all step definitions.
  2. Extend the Widget class by providing prototype properties and methods.
  3. Define the CSS selector of the widget, which is used to scope all webdriver calls to instances of the widget.
  4. Define a custom method selectProjectAt which uses the Widget::find method to create a new (anonymous) widget instance that is scoped within the ProjectsPage selector. The resulting widget has the selector .projects-list .project:nth-child(index) which is what we used to select a project in our first step definitions example
  5. Call click on the ad-hoc project widget

Now you may ask: Where does visit, waitForVisible and click come from? That's the really cool part: You can use the complete webdriver.io API without any changes. You just leave out the first param (selector) to all api calls because your widgets are taking care of that transparently. The second nice side effect is that the chimp-widgets API wraps the raw webdriver.io calls with Promises. That's why you can just return new this.ProjectsPage().waitForVisible(); in your step definition and and the test runner handles the resolved or rejected promise.

API

this.widgets.Widget

The base class of all widgets that wraps the complete webdriver.io API by scoping all calls to the provided selector and returning Promises only.

Here is the bare-bones definition of a widget:

this.MyWidget = this.widgets.Widget.extend({
  selector: '.my-dom-selector',
});

find(selector)

Creates and immediately returns a new widget instance with selector scoped within the parent widget. Calling new this.MyWidget().find('.test') is the same as creating new this.widgets.Widget('.my-dom-selector .test').

It is important to know that although the method is called find there is no interaction with webdriver.io happening. It really just nests CSS selectors so that API calls on the nested widget target nested DOM elements.

hasText(text)

Convenience method around the webdriver.io getText method. Is the same as calling widget.getText().should.eventually.become(expected).

Exposed webdriver.io Methods:

The cool thing about chimp-widgets is that you don't have to learn a lot of new concepts. It exposes exactly the same API as webdriver.io client or browser but scopes it to the widget selector and wraps it into a Promise automatically.

Here is the complete list of supported webdriver.io methods:

Widget.API = [
  # Actions
  'addValue', 'clearElement', 'click', 'doubleClick', 'dragAndDrop',
  'leftClick', 'middleClick', 'moveToObject', 'rightClick', 'setValue',
  'submitForm',
  # Property
  'getAttribute', 'getCssProperty', 'getElementSize', 'getHTML',
  'getLocation', 'getLocationInView', 'getSource', 'getTagName',
  'getText', 'getTitle', 'getValue',
  # State
  'isEnabled', 'isExisting', 'isSelected', 'isVisible',
  # Utility
  'waitForChecked', 'waitForEnabled', 'waitForExist', 'waitForSelected',
  'waitForText', 'waitForValue', 'waitForVisible'
]

Creating anynomous ad-hoc widgets:

Sometimes it is too cumbersome to define a separate class to represent a DOM object, that's where creating ad-hoc instances can be helpful: new this.widgets.Widget(selector, [driver], [Promise])

this.widgets.List

Documentation coming soon, in the meanwhile checkout this Leaderboard class defined in the chimp-widgets-demo which represents the leaderboard from the standard Meteor examples:

var widgets = this.widgets;
// Define a widget by extending the base class
this.Leaderboard = widgets.List.extend({

  selector: '.leaderboard', // All widgets need a CSS selector
  itemSelector: '.player', // Selector of nested items

  // Define high-level methods that you can call from steps
  givePointsTo: function(name, points) {
    self = this;
    return this.selectPlayer(name).then(function() {
      return self.givePointsToSelectedPlayer(points);
    });
  },

  selectPlayer: function(name) {
    return this.getPlayerByName(name).then(function(player) {
      return player.click();
    });
  },

  getPlayerByName: function(name) {
    return this.findWhere(function(player) {
      return player.find('.name').getText().then(function(text) {
        return text.match(new RegExp(name, "g"));
      });
    });
  },

  givePointsToSelectedPlayer: function() {
    return new widgets.Widget('.details .inc').click();
  },

  checkScoreOf: function(name, expectedPoints) {
    return this.getPlayerByName(name).then(function(player) {
      return player.find('.score').getText().should.become(expectedPoints);
    });
  },

  checkPlayerIsAbove: function(player1, player2) {
    var self = this;
    return this.getPlayerPosition(player1).then(function(position1) {
      return self.getPlayerPosition(player2).then(function(position2) {
        return self.Promise.resolve(position1 < position2).should.become(true);
      });
    });
  },

  getPlayerPosition: function(name) {
    var Promise = this.Promise;
    return Promise.any(this.map(function(player, position) {
      return player.find('.name').getText().then(function(playerName) {
        if(playerName === name) {
          return Promise.resolve(position);
        }
        else {
          return Promise.reject();
        }
      });
    }));
  }
});
// Set the URL of the leaderboard screen statically
this.Leaderboard.url = process.env.ROOT_URL;

How to run the tests

Install dev dependencies with npm install then run all tests with npm test

About

Widget that provides high-level commands for UI testing for use with Xolv.io chimp

Resources

Stars

Watchers

Forks

Packages

No packages published

Languages

  • CoffeeScript 96.0%
  • JavaScript 2.7%
  • Shell 1.3%