Practical Web Test Automation

Automated test web applications wisely with open source

test frameworks: Watir and Selenium
Zhimin Zhan
This book is for sale at http://leanpub.com/practical_web_test_automation
This version was published on 2014-08-18
This is a Leanpub book. Leanpub empowers authors and publishers with the Lean Publishing
process. Lean Publishing is the act of publishing an in-progress ebook using lightweight tools and
many iterations to get reader feedback, pivot until you have the right book and build traction once
you do.
2012 - 2014 Zhimin Zhan
I would dedicate this book to my mother and father for their unconditional love.
Contents
Preface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . i
Who should read this book? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ii
How to read this book? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ii
Whats inside the book? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iii
Test scripts, Screencasts and Other resources . . . . . . . . . . . . . . . . . . . . . . . . . iii
Send Me Feedback . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iv
Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iv
1. What is Web Test Automation? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
Test Automation Benefits . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
Reality Check . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
Reasons for Test Automation Failures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
Successful Web Test Automation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
Learning Approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
Next Action . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
2. First Automated Test . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
Test Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
Installing TestWise (about 2 minutes) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
Create a Test . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
Create Test Case From Recorded Test Steps . . . . . . . . . . . . . . . . . . . . . . . . . . 13
Run test in IE . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
When a test failed . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
Wrap up . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
3. How Automated Testing works . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
Web Test Drivers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
Automated Testing Rhythm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
Test Frameworks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
Run Tests From Command Line . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
4. TestWise - Functional Testing IDE . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
Philosophy of TestWise . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
TestWise Project Structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
CONTENTS
Test Execution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
Keyboard Navigation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
Snippets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
Script Library . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
Test Refactoring . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
Wrap Up . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
5. Case Study . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
Test Site . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
Preparation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
Create Test Project . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
Test Suite: Sign in . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
Test Suite: Select Flights . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
Enter Passenger Details . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
Book confirmation after payment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
Run all tests . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
Wrap Up . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
Preface
On April 3 2013, Wired published an article The Software Revolution Behind LinkedIns Gushing
Profits. The revolution completely overhauled how LinkedIn develops and ships new updates
to its website and apps, taking a system that required a full month to release new features and
turning it into one that pushes out updates multiple times per day. LinkedIn is not alone, Google
has accomplished this long before that. As a matter of fact, LinkedIns success is tracked back to
luring a Google veteran in 2001. Facebook is released twice a day and they claimed keeping up
this pace is at the heart of our culture.
Release software twice a day! For many, thats unimaginable. You may wonder how they could
ensure quality (and you know the high standard from them). The answer is, as the article pointed
out, using automated tests designed to weed out any bugs.
After working on numerous software projects for a number of years, I witnessed and had been part
of many what I call release panic syndromes. That is, with the deadline approaching, the teams
panic level rises. Many defects were found from the last round of manual testing by the testers, the
manager started priortizing the defects (or adjusting some to features), and programmers rushing to
fix just the critical ones. Testers restarted the testing on the new build which fixed some but not all
the defects. Then here came the bad news: several previously working features were now broken,
Argh!
I believe there is a better way to do software development that does not involve this kind of stress
and panic. This is how my interest in automated testing started (in 2006). I made the right decision to
use free, open source and programming based test frameworks. (It is quite obvious now, as Selenium
is the best sought after testing skill on the job market. Back then, people turned to record/playback
commercial tools with vendor proprietary test script syntax). The first test framework I used (for
my pet projects) was Watir. Wow! I was convinced that this approach was the answer.
In 2007, I had the opportunity to put my approach into practices in a government project. The
outcome was beyond everyones expectation: over two years and countless releases, there were no
major defects reported by customers. The team had high confidence in the product. These automated
tests also provided the safety net for some major refactorings, which would not be possible without
them. A business analyst once said, before every demonstration to our customers, it is a good
feeling of knowing every release has been thoroughly tested. The synergy of flexible test framework,
maintainable test design, team collaboration with the same simple testing tool and continuous
integration supporting functional test execution really made a big difference.
http://www.wired.com/business/2013/04/linkedin-software-revolution/
http://www.facebook.com/notes/facebook-engineering/ship-early-and-ship-twice-as-often/10150985860363920
http://www.seleniumconf.org/speakers/
Preface ii
There is now a clearly converging trend in web application development on technology choices,
such as cloud deployment, light and productive web frameworks such as Ruby on Rails, JQuery
JavaScript Library, Twitter BootStrap UI themes, Font Awesome icons, , etc. The competitions
among web applications are less on technologies, but weigh more on the development process to
ensure pushing out high quality releases often. A fact: Facebook was not the first social networking
web site.
A friend of mine, who developed a quite successful public web application, told me in an uneasy
tone that he just found out another competitor product at a cheaper price. This is inevitable, the
competition among web applications is global, which means, there are people work at 10% of
your hourly rate to compete against you. The only way to win the race, in my opinion, is to
greatly enhance your productivity and reduce maintenance cost. This can be achieved by applying
test automation and continuous integration with instant benefits without much effort (if doing it
properly). My reply to my friend: If your competitors seriously start to invest in test automation,
you shall be worried.
In Appendix II, I share my experience on developing ClinicWise, a modern web-based clinic
management system. Thanks to comprehensive automated UI testing, ClinicWise is frequently
released (daily) with new features and updates. ClinicWise is developed and maintained in my spare
time.
The motivation of this book is to share my journey of test automation for web applications: from
writing the first test to developing and maintaining large number of automated test scripts.
Who should read this book?
Everyone working on a software team (including testers, programmers, business analysts, architects
and managers) building a web application who wants to improve the quality of software while
saving time and money can benefit from reading this book. It may sound like a bold statement, but
it is the outcome I obtained from some projects whose team members embraced the techniques and
practices presented in this book. Those projects delivered reliable software releases frequently, stress
free. You can achieve this too.
Prior experience with automated testing is not necessary. Basic programming concepts will help,
but again, not necessary.
How to read this book?
I strongly recommend readers to read through Chapters 1-9 in order, only skip Chapter 4 if you have
decided on the testing editor/IDE. Chapters 10-15 are largely independent from one another. You can
therefore read them in the order that suits your interests. Readers can also just skim through and
come back for details later if necessary.
http://yourstory.com/2013/02/startup-technologies-top-6-technologies-used-at-startups/
Preface iii
Some chapters contain hands-on exercises (with step by step guides). Typically it will take about
10-30 minutes to complete an exercise. Readers can choose to follow the exercises while or after
reading a chapter. The main point is: to master test automation, you have to do it.
Whats inside the book?
In part 1, I introduce Web Test Automation and its benefits, which many believe but few actually
achieve it. I use a metaphor to illustrate practical reasons why most software projects conduct
functional testing manually despite knowing the great benefits of test automation. Then the journey
starts with a case study to help write your first Watir automated test in about 10 minutes.
In part 2, I present a brief introduction of test frameworks and tools, followed by a case study showing
the development of six Watir/RWebSpec tests for a live test site with the help of a recorder. Along
the way, some testing techniques are introduced.
In part 3, I present an intuitive and maintainable automated test design: using reusable functions
and page objects, followed by a case study showing the transforming of recorded test scripts to a
maintainable way. Then I introduce an important concept: functional test refactoring, a process of
testers applying refactorings to test scripts efficiently with refactoring support in testing tools such
as TestWise IDE.
With a growing number of automated tests, so is the test execution time. Long feedback loops really
slow down development. In part 4, I show how team collaboration and continuous integration can
help to improve the feedback time greatly.
In Part 5, I switch the attention to two other frameworks: Selenium-WebDriver and Cucumber,
with two case studies showing the test design and techniques are applicable generally. Finally I
share some strategies to apply test automation to your project.
Test scripts, Screencasts and Other resources
To help readers learn more effectively, the book has a dedicated site at: http://zhimin.com/books/pwta,
which contains the following resources:
Software. Test automation is not necessarily expensive. All test frameworks featured in this
book are free and open-sourced. Testing tools used for the exercises in this book are also free,
and there are instructions to cater for other text-based testing tools.
https://github.com/zhimin/rwebspec
http://testwisely.com
http://seleniumhq.org/
http://cukes.info/
http://zhimin.com/books/pwta
Preface iv
Sample test scripts. The sample test scripts for the exercises are ready-to-run. This book
covers several popular test and syntax frameworks: Watir, Selenium, RWebSpec, RSpec and
Cucumber. To help readers understand the differences, I have created 8 test projects with
different combinations: https://github.com/zhimin/adminwise-ui-tests.
Sample web sites. For readers who need web sites to try out automated test scripts, I have
prepared two test sites for you:
Agile Travel: a simple flight booking site, which is used in the exercises.
AdminWise: a feature rich web 2.0 site with modules such as membership and library.
Tutorial screencasts. There are screencasts for readers who learn better with audio and video,
so you will be able to see how it is done step by step.
Send Me Feedback
Id appreciate hearing from you. Comments, suggestions, errors in the book and test scripts are all
welcome. You can submit your feedback on the book web site (http://zhimin.com/books/pwta).
Acknowledgements
I would like to thank everyone who sent feedback and suggestions, particularly Darren James, Mingli
Zhou, Tim Wilson, Lloyd Blake and Hoang Uong, for their time and wisdom.
I owe a huge thank you to people behind great open-source testing frameworks such as Watir,
Selenium-WebDriver and RSpec, and of course, the beautiful Ruby language.
Functional testing via User Interface is practical and light on theory, so is this book. I hope you find
this book useful.
Zhimin Zhan
May 2014
https://github.com/zhimin/adminwise-ui-tests
1. What is Web Test Automation?
Web Test Automation, or automated functional testing for web applications via the Graphic User
Interface (GUI), is the use of automated test scripts to drive test executions to verify the web
application meets requirements. Simply, during execution of an automated test for a web site, you
see mouse and keyboard actions such as clicking a button and typing text in a text box in a browser,
without human intervention. Web Test Automation sits under the category of black-box functional
testing, where the majority of test efforts is in software projects.
..
Functional Testing vs Unit Testing vs Non-Functional Testing
Functional testing is to verify function requirements: what the system does. For example, User
can request a password reset by providing a valid email. This is the focus of this book.
Unit testing is a type of white box testing performed by programmers at source code level. It is of
no concerns to software testers. Unit test is a term that gets misused a lot. A more correct term
would be Programmer Test. A product that passes comprehensive programmer tests can still fail
on many functional tests. Thats because programmers tests are from a programmers perspective,
while functional tests are from a users perspective. A programmer test is also a kind of automated
test.
Non-functional testing is the testing of how the system works. For example, The response time of
the home page must not exceed 5 seconds. Some type of non-functional testings, load testing in
particular, utilize automated test tools too.
Test Automation Benefits
The benefits of test automation are many. Below are four common ones:
Reliable. Tests perform precisely the same operations each time they are run, thereby
eliminating human errors.
Fast. Test execution is faster than done manually.
Repeatable. Once tests are created, they can be run repeatedly with little effort, even at lunch
time or after working hours.
Regression Testing. The intent of regression testing is to ensure that a change, such as a
bugfix, did not introduce new faults [Myers, Glenford 04]. Comprehensive manual regression
testing is almost impossible to conduct for two reasons: the time required and human errors.
What is Web Test Automation? 2
As Steve McConnell pointed out, The only practical way to manage regression testing is to
automate it. [McConnell]
What do I like about test automation?
If you want me to use only one adjective to describe web test automation, it is fun. I enjoy
creating something that can I can get to do work for me. I have a habit of triggering
execution of a test suite before going out for lunch. I like the feeling of still working
while enjoying a meal. When I come back, a test report is already there, waiting for me.
Reality Check
With more software projects adopting agile methodologies and more software application devel-
opments moving towards the Web, you would assume web test automation would be everywhere
now. The answer is, sadly, no. In fact, functional testing in many projects is still executed in pretty
much the same way: manually.
UI Test Automation Tools are Snake Oil - Michael Feathers
It happens over and over again. I visit a team and I ask about their testing situation. We talk
about unit tests, exploratory testing, the works. Then, I ask about automated end-to-end
testing and they point at a machine in the corner. That poor machine has an installation of
some highly-priced per seat testing tool (or an open source one, it doesnt matter), and the
chair in front of it is empty. We walk over, sweep the dust away from the keyboard, and
load up the tool. Then, we glance through their set of test scripts and try to run them. The
system falls over a couple of times and then they give me that sheepish grin and say we
tried. I say, dont worry, everyone does. [Feathers 10]
Reasons for Test Automation Failures
The software testing survey conducted by Innovative Defense Technologies in 2007 [IDT07] shows
73% of survey respondents believe Automated Testing is beneficial but few automate. The top
reasons for survey participants not automating their software testing efforts (while agreeing with
the benefits) are:
lack of time
lack of budget
lack of expertise
What is Web Test Automation? 3
These reasons sound about right to most people. However, saving time and money are two benefits of
test automation, isnt that a contradiction (for lack of time and budget)? What are the real difficulties
or challenges, apart from political/ project management ones, that projects encounter during their
adventures in automated testing?
To make it easy to understand, we can compare a projects test automation attempt with a boy who is
trying to climb over a standing two-hump camel from the front. Lets consider each of the following
challenges he faces:
Test Automation Camel
Figure 1-1 Test Automation Camel (graphics credit: www.freevectordownload.com)
1. Out of reach: Expensive
Commercial testing tools are usually quite expensive (I wont list prices here, in fact, I couldnt
get prices for some so-called leading testing tools on their web sites, which is telling in itself).
Automated testing is one of a few activities in software projects that the whole team can contribute
to and benefit from. Besides testers, programmers may run automated tests for self verification and
business analysts may utilize automated tests for customer demonstrations. However, high price of
commercial testing tools makes the whole teams adoption of automated testing unfeasible.
There are free, open-source testing frameworks, such as Selenium and Watir, both of which are
featured in the classic book Agile Testing by Lis Crispin and Janet Gregory. However the idea of
free and open-source testing frameworks is still not appealing to many test managers. Lack of skills,
dedicated tools and support are their main concerns.
2. Steep Learning Curves: Difficult to learn
Traditional commercial tools are usually focused on a Record and Playback approach with test scripts
in a vendor proprietary syntax. It looks easy when you watch the sales presentations. In real life,
http://www.agiletester.ca/
What is Web Test Automation? 4
unfortunately, it is a quite different story (a programmers minor change to the application can ruin
your hours of recording). When testers have to open the raw test scripts (generated by recorders) to
edit, reality bites.
Open source test frameworks, on the other hand, require some degree of programming efforts, Watir,
Selenium and WebTest are among the popular ones. With programming, they provide flexibility
needed for automated testing. However, the fact is that the majority of software testers do not possess
programming skills, and many of them feel uncomfortable to learn it. There are few dedicated testing
tools supporting these open-source test frameworks designed to suite testers. (Programming IDEs
are designed for programmers, not for testers who can find them complicated and overwhelming).
3. Hump 1: Hard to maintain
Software under development changes frequently, and automated UI test scripts are vulnerable to
application changes. Even a simplest change to the application could cause many existing test scripts
fail. This, in my view, is the most fundamental reason for test automation failures.
4. Hump 2: Long feedback loop
Compared to programmer tests (which if written well, should have an execution time under 0.1
second), automated functional tests through UI are relatively slow. There is practically very little
that testers can do to speed up execution of functional tests. With the number of test cases growing,
so will be the test execution time. This leads to long feedback gap, from the time programmers
committed the code to the time test execution completes. If programmers continue developing new
feature/fixes during the gap time, it can easily get into a tail-chasing problem. This will hurt teams
productivity badly, not to mention teams morale.
New Challenges for testing Web applications
Specifically to web applications, with adoption of AJAX (Asynchronous JavaScript and XML)
and increasing use of JavaScripts, websites nowadays are more dynamic, therefore, bringing new
challenges to web test automation.
Successful Web Test Automation
Having identified the reasons for test automation failures in projects, it becomes clear what it takes
to succeed in web test automation:
1. Test scripts must be easy to read and maintain.
2. Testing framework/tools are easy to learn, affordable and support team collaboration.
3. Test execution must be fast.
Is that all possible? My answer is Yes. The purpose of this book is to show how we can achieve just
that.
What is Web Test Automation? 5
Learning Approach
This is not yet another book on testing theories, as there are no shortage of them. In this book, we
will walk through examples using test framework RWebSpec/Watir/Selenium and functional testing
IDE TestWise. The best way to learn is to just start doing it.
My father is a well respected high school mathematics teacher in our town, his teaching style is
teaching by examples. That is, he gets students to work on his carefully selected math exercises
followed by concise instruction, then guide students who face challenges (often he gives them
another exercise ). That is also the way I learn things. By working with testers, I found this is the
most effective way for testers to master automated testing quickly.
For most web sites, regardless of technologies they are developed on, Internet Explorer on Windows
is often the target platform (at least for now). It will be the main platform for our exercises in this
book:
Operating System: Windows XP or later
Web Browser: Internet Explorer v8-10 (IE11 is not well supported yet)
Test Framework: Watir (with RWebSpec extension)
Testing Tool: TestWise IDE
If you are Mac user, like myself, the learning process is the same (majority of the test scripts run
without change) except the screenshots shown in the book look different.
Operating System: Mac OS X
Web Browser: Firefox or Chrome
Test Framework: Selenium-WebDriver (with RWebSpec extension)
Testing Tool: TestWise IDE Mac Edition
I will cover more on Selenium to drive tests against Firefox or Google Chrome Browser on Mac or
Linux in Chapter 13 (I am a Mac user myself). All the techniques and even test scripts are directly
applicable for cross-browser testing.
On testing tools, I use TestWise community edition, a free testing IDE supporting Watir and
Selenium, in this book. For readers who prefer their favourite editors/IDEs, you can still use them,
as all test scripts shown in this book are plain text. I will also provide instructions on how to execute
tests from the command line.
Example test scripts for chapters in this book can be downloaded at http://zhimin.com/books/pwta,
and you can try them out by simply opening in TestWise and run. Also, I have provided screencasts
there so that readers can watch how it is done.
In this book, we will focus on testing standard web sites (in HTML), excluding vendor specific
technologies such as Flash and SilverLight. The techniques shown in this book are applicable to
https://code.google.com/p/selenium/wiki/InternetExplorerDriver#IE_11_Support
http://zhimin.com/books/pwta
What is Web Test Automation? 6
general testing practices.
Next Action
Enough theory for now. Lets roll up sleeves and write some automated tests.
2. First Automated Test
A journey of a thousand miles must begin with a single step. - Lao Tzu
Lets write an automated web test. If you are new to automated testing, dont feel intimidated. You
are going to see your first automated test running in Internet Explorer in about 10 minutes, and that
includes installing the test tool!
Test Design
A test starts with a requirement (called User Story in agile projects). Quite commonly, the first
simple requirement to test is: User Authentication. We will use this requirement for our first test in
this exercise.
By analysing the requirement and the application (see the screenshot below),
Agile travel login
we can start to collect the test data:
First Automated Test 8
Site URL: http://travel.agileway.net
User Login/Password: agileway/testwise
and design the test steps:
1. Enter username agileway
2. Enter password testwise
3. Click button Sign In
4. Verify: Login successful! appears
You might by now be saying there is no difference from manual testing. Thats correct. If you
currently work as a manual tester, you probably feel a relief at knowing your test design skills apply
to automated testing as well. As a matter of fact, we usually perform the test steps manually as
verification of test design before writing automated test scripts.
Now we are going to automate it. The main purpose of this exercise is to help you write an automated
Watir test case and watch it running in IE, in a matter of minutes. Dont pay attention to details
yet, as it will become clear as we move on. If you get stuck, follow the screencast for this exercise
at http://zhimin.com/books/pwta.
Installing TestWise (about 2 minutes)
We will use TestWise, a Functional Testing IDE, for this exercise. (TestWise Community Edition is
free)
Prerequisite
A PC with MS Windows XP/Vista/7
TestWise recorder which requires Mozilla Firefox
Download
TestWise IDE Community Edition from http://testwisely.com/testwise/downloads (18MB
download).
Install
TestWise IDE. Double click TestWise-Community-3.x-setup.exe to install, accept all default
options. The default installation folder is C:Program FilesTestWise. Launch TestWise after the
installation completes.
http://zhimin.com/books/pwta
http://testwisely.com/testwise/downloads
First Automated Test 9
TestWise Recorder. TestWise comes with a lightweight recorder (a Firefox extension) to
record user operations in a Firefox browser into executable test scripts.
To install, navigate to folder C:\agileway\TestWise\thirdparty\FireFox_AddOns (or click the
link fromStart Page in TestWise), then drag testwise_recorder.xpi to an active Firefox Window.
Locate recorder
A window with title Software Installation will popup from Firefox to confirm installation,
click Install Now, then restart Firefox.
Install recorder
In Firefox, select menu Tools -> TestWise Recorder Sidebar to enable recording.
First Automated Test 10
Enable recorder
Create a Test
Now we are ready to create the test for our requirement: User can login the site. Hope you still
remember the test design steps and test data.
Create New Test Project
TestWise has a project structure to organize test scripts. That structure is simply a folder containing
all test related files such as test scripts and test data.
As we start from scratch, we need to create a new project first. If a sample project is already opened
in TestWise, we need to close it. Select menu File New Project, which will bring up the window
shown below.
Create Project
First Automated Test 11
Enter project name, project folder and URL of web site to be tested. In this case, we enter Agile
Travel, C:\testprojects\AgileTravel and http://travel.agileway.net respectively, and then click
OK button. TestWise will create the project with skeleton files created for you.
Project Skeleton
Create Test Script File
Now create the test script file for our test. Select File -> New File,
New test
Type text login and press Enter to create new test script file: login_spec.rb
Tip: Try naming the test script file something related to the requirement, so you can find it easily
later.
A new editor tab is created and opened with test skeleton:
First Automated Test 12
Login test
Recording
Open the site URL http://travel.agileway.net in Firefox and enable TestWise Recorder SideBar.
Perform the test steps below manually:
1. Enter username agileway
2. Enter password testwise
3. Click Sign In button
4. To add verification for text Welcome agileway, highlight the text in browser, right mouse
click and select Add verify Text for Welcome agileway.
5. Click Sign off link
Recording
Test steps are recorded along the way. Once done, inside the TestWise Recorder window, right mouse
click and select Copy all to clipboard. If you see the test step goto_url(about:blank) (that step tells
where the current URL is, we dont need for this case), delete it.
First Automated Test 13
Recorder copying test steps
Create Test Case From Recorded Test Steps
Switch to the TestWise IDE (the login_spec.rb editor tab shall be still active), paste recorded test
scripts into the test case.
Paste test steps
The test case is created, while we are here, update the test cases name to User can login with valid
user name and password.
Run test in IE
Press on the toolbar (indicated as in the screenshot below) to run the test case, and you can
watch the test execution in an Internet Explorer window.
First Automated Test 14
TestWise run
The green tick (indicated as ) means the test passed.
When a test failed
We just saw a successful automated test. Naturally, you will ask what will happen when a test fails?
As a matter of fact, during development of an automated test script, we are more likely to get errors
or failures before we get it right. It is up to the technical testers to analyse the cause: is it a real
application error or incorrect test scripts?
Next, we will make a simple change to the above test script to make it fail:
enter_text("password", "invalid") # now can't log in
Click to run the test. As expected, the test failed.
First Automated Test 15
Test failed
In TestWise, the test execution is marked as Failed and is shown on line 18 of the test script
indicating where the failure is.
We, as human, knew the reason for this failure: a wrong password was provided. From the test
scripts point of view, it failed due to this assertion not met: finding the text Welcome agileway
on the page.
If you want to find more details about the cause for test failure, check the text output of test execution
including error trace under Test Output tab.
Failed in IE
First Automated Test 16
Wrap up
Lets review what we have done in this chapter. Besides test design, we
Installed TestWise IDE
Installed TestWise Recorder
Created a test project in TestWise IDE
Recorded test scripts using TestWise Recorder in Firefox
Created test script from pasted test steps
Ran test case in IE (pass and failed)
Hopefully you were able to do all that within 10 minutes! You can view the screencast for this
exercise online at the books website at http://zhimin.com/books/pwta.
http://zhimin.com/books/pwta
3. How Automated Testing works
In the previous chapter, we created an automated functional test running in a web browser, Internet
Explorer. This was done by simulating a user interacting with the browser: typing texts and clicking
buttons.
Before we move on, let us examine our test targets: web applications (or web sites). Simply speaking,
a web site consists of many web pages. Behind each web page there is an HTML (HyperText Markup
Language) file. Browsers download the html files and render them.
HTML defines a set of standard web controls (aka elements) we all are familiar with, such as text
boxes, hyperlinks, buttons, checkboxes, etc. For web application testing, we interact with these
controls as well as the text that get marked up in the HTML such as labels and headings.
Now let us review the test script we created in last exercise:
Within a test case, test steps can be classified into the following two categories:
Operation (also called step). Performing some kind of keyboard or mouse action on a web
page. The above example test has three operations:
enter_text("userName", "agileway")
enter_text("password", "testwise")
click_button("Sign In")
Check (also called assertion). Verifying the web page meets the requirement.
page_text.should contain("Welcome agileway")
How Automated Testing works 18
Web Test Drivers
Web test drivers enable web controls to be driven by test scripts with a certain syntax, for testing
purposes. All web test drivers covered in this book are free and open-source.
Watir
Watir (Web Application Testing in Ruby) is a free and open source library for automated testing
web applications in Internet Explorer. As its name suggests, test scripts actually are Ruby scripts, a
growing popular programming language with an elegant syntax that is natural to read and easy to
write[ruby01].
Watir is the most compelling alternative [to Fit] for filling the automated acceptance
testing need. Ward Cunningham
Watir was created in 2005 by Bret Pettichord and Paul Rogers, and it is maintained by many other
contributors. Inspired by Watirs success, there are clone frameworks in .NET and Java platforms:
WatiN and Watij respectively.
Here is a sample Watir test:
require "watir"
browser = Watir::Browser.new
browser.goto "http://www.google.com"
browser.text_field(:name, "q").set "Watir IDE"
browser.button(:name, "btnG").click # "btnG" is the 'Search' button
Selenium
Selenium was originally created in 2004 by Jason Huggins, who was later joined by his other
ThoughtWorks colleagues. Selenium supports all major browsers and tests can be written in many
programming languages and run on Windows, Linux and Macintosh platforms.
Selenium 2 is merged with another test framework WebDriver led by Simon Stewart at Google
(thats why you see selenium-webdriver), Selenium 2.0 was released in July 2011.
The above test in Selenium:
How Automated Testing works 19
require "selenium-webdriver"
browser = Selenium::WebDriver.for(:firefox) # or :ie, :chrome
browser.navigate.to "http://www.google.com"
browser.find_element(:name, "q").send_keys "Watir IDE"
browser.find_element(:name, "btnG").click #"btnG" is the 'Search' button
RWebSpec
RWebSpec is a wrapper around Watir and Selenium. It provides an alternative syntax plus some
extra goodies. Here is a sample RWebSpec script.
require 'rwebspec'
include RWebSpec::Core
open_browser "http://google.com"
enter_text("q", "Watir IDE")
click_button_with_id("gbqfb")
One obvious difference from Watir or Selenium is that RWebSpec test statements are in active
voice. The following two statements in RWebSpec
enter_text("q", "watir IDE")
click_button_with_id("gbqfb")
are equivalent to
browser.text_field(:name, "q").set "watir IDE"
browser.button(:id, "gbqfb").click
in Watir.
Many might find the RWebSpec syntax is more concise and easier to read.
You may mix Watir or Selenium test statements in RWebSpec test scripts like below:
enter_text("q", "Watir IDE") # RWebSpec
button(:name, "btnG").click # Watir
browser.find_element(:name, "q").send_keys("Selenium full featured IDE") # Selen\
ium
click_button_with_id("gbqfb") # RWebSpec
How Automated Testing works 20
Automated Testing Rhythm
Regardless of which test framework you use, the testing rhythm is the same:
1. Identify a web control
2. Perform operation on the control
3. Go to step 1 until reach a check point
4. Check
5. Go to step 1 until the test ends
Identify Web Controls
To drive controls on a web page, we need to identify them first.
Lets look at this sample web page:
Its HTML source (you can view the HTML source of a web page by right mouse clicking in the web
page and selecting View Page Source):
User name: <input type="text" name="username" size="20"/>
Password: <input type="password" id="pwd_box" name="password" size="20"/>
<input type="submit" id="sign_in_button" value="Sign in"/>
Though the username and password appear the same (text box) on the browser, they are quite
different in source. Some attributes in HTML tags tell web browsers how to render it, such as
size=20 in user name text box; More importantly, application developers use attributes such as
name (not exclusively) to determine users input is associated to which control.
We can identify web controls by using these attributes for testing. Here is one way to identify them
in Watir:
How Automated Testing works 21
text_field(:name, "username")
text_field(:id, "pwd_box")
button(:value, "Sign in")
As you can see, these three test steps use three different attributes for three controls.
Obviously the easiest way to identify web controls is to use a recorder (a tool records users operation
and generate test scripts), if you have one installed. However, in my opinion, it is essential for
technical testers to master and be comfortable to do it manually. The reasons are:
Some test frameworks dont have recorders or have outdated ones
Recorders might not work for certain circumstances
Lack of freedom of choosing preferred attribute (for identifying controls)
In modern browsers, it is actually quite easy to identify element attributes (in HTML source)
manually:
Internet Explorer: Developer Tools
IE8 (and later version) has built-in developer tools. You can invoke it by pressing F12 key and Ctrl+B
in the Developer Tools window to inspect a web control.
IE Developer Tools
Firefox with Firebug Add-on
How Automated Testing works 22
Firebug is a popular Firefox extension, one of its feature is to inspect elements source. Once installed
and enabled, right mouse click one control and select Inspect element:
Firebug add-on
Google Chrome
Google Chrome (and Apple Safari) browser has built-in support for inspecting web controls:
How Automated Testing works 23
Inspect in Chrome
Drive Web Controls
Once we identify a web control, the next step is to perform required operation with it, such as
typing text in a text field, clicking for a button, clearing a checkbox, and so on. Though different
test frameworks have different syntax, the idea is the same.
Here are some examples:
Watir:
browser.text_field(:name, "user[name]").set "bob"
browser.button(:id, "next_btn").click
Selenium:
driver.find_element(:name, "user[name]").send_keys "bob"
driver.find_element(:id, "next_btn").click
RWebSpec:
enter_text("user[name]", "bob")
click_button("Continue")
How Automated Testing works 24
Check
The purpose of testing is to verify a piece of function meeting its purpose. After driving the
application to a certain point, we do checks (maybe thats why is called checkpoint in some testing
tools).
In the context of web testing, typical checks are:
verify certain texts are present
verify certain html fragment are present (different fromabove, this is to check rawpage source)
verify page title
verify a link is present
verify a web control is present or hidden
One key feature of Test frameworks (more in next section) is to provide syntax conventions to
perform verifications like the above. Here are some examples:
RWebSpec
page_source.should contain("Payment Successful!")
assert_link_present_with_text("Continue")
page_title.should == "User Registration"
xUnit (assertion style) with Watir
browser.html should include("Payment Successful!")
assert browser.button(:text, "Choose Watir").enabled?
assert browser.title == "User Registration"
RSpec with Selenium
driver.page_source.should include("Payment Successful!")
# RSpec 2 uses be_true, be_false; RSpec 3 uses be_truthy, be_falsey
browser.find_element(:link_text, "Continue").displayed?.should be_truthy
expect(driver.title).to eq("User Registration")
Test Frameworks
Web test drivers such as Watir and Selenium drive browsers. However to make effective use of
them for testing, we need put them in a test framework which defines test structures and provides
assertions (performing checks in test scripts).
How Automated Testing works 25
xUnit
xUnit (JUnit and its cousins) test frameworks are widely used for unit testing by programmers. xUnit
can be used in functional test scripts too, but it is not my preference, as it is not as expressive as the
ones below.
RSpec
RSpec is a popular Behaviour Driven Development (BDD) framework in Ruby.
More expressive
Comparing to xUnit test frameworks, RSpec tests are easier to read. For example, for the JUnit test
below:
class UserAuthenticationTest {
public void testCanLoginWithValidUsernameAndPassword {
// ...
}
public void testAccessDeniedForInvalidPassword() {
// ...
}
}
Its RSpec version will be like this:
describe "User Authentication" do
it "User can login with valid login and password" do
# ...
end
it "Access denied for invalid password" do
#...
end
end
Execution Hooks
Execution hooks are similar to setUp() and tearDown() functions in JUnit. Test steps inside a
execution hook are run before or after test cases depending on the nature of the hook. The example
below shows the order of execution in RSpec:
How Automated Testing works 26
describe "Execution Order Demo" do
include RWebSpec::RSpecHelper
before(:all) do
puts "Calling before(:all)"
end
before(:each) do
puts "Calling before(:each)"
end
after(:each) do
puts "Calling after(:each)"
end
after(:all) do
puts "Calling after(:all)"
end
it "First Test Case" do
puts "In First Test Case"
end
it "Second Test Case" do
puts "In Second Test Case"
end
end
Output
Calling before(:all)
Calling before(:each)
In First Test Case
Calling after(:each)
Calling before(:each)
In Second Test Case
Calling after(:each)
Calling after(:all)
Whats the use of execution hooks? Lets look at the RWebSpec test script below (RWebSpec extends
from RSpec, so we can use story or test_case as the start of a test case). There are three login related
test cases in a single test script file.
How Automated Testing works 27
specification "User Login" do
include RWebSpec::RSpecHelper
story "Can login as Registered User" do
open_browser
login_as("james", "pass") # login_as is a function defined elsewhere
page_text.should contain("Welcome James")
logout
close browser
end
story "Can login as Guest" do
open_browser
login_as("guest", "changeme")
page_text.should contain("Login OK")
logout
close browser
end
story "Can login as Administrator" do
open_browser
login_as("admin", "secret")
assert_link_present_with_text("Settings")
logout
close browser
end
end
By utilizing execution hooks, we can refine these test cases to:
specification "User Login" do
include RWebSpec::RSpecHelper
before(:all) do
open_browser
end
after(:each) do
logout
end
How Automated Testing works 28
after(:all) do
close_browser
end
story "Can login as Registered User" do
login_as("james", "pass")
page_text.should contain("Welcome James")
end
story "Can login as Guest" do
login_as("guest", "changeme")
page_text.should contain("Login OK")
end
story "Can login as Administrator" do
login_as("admin", "secret")
assert_link_present_with_text("Settings")
end
end
By utlizing RSpecs before(:all), after(:each) and after(:all) hooks, this version is not only concise,
but more importantly, every test case is now more focused (distinguished from each other) and test
scripts are more readable. For readers who are new to RSpec, dont worry, I will cover it more in
later chapters.
Cucumber
Cucumber, another relatively new BDD framework in Ruby, is gaining popularity rapidly. To avoid
distraction, we will focus on test practices using Watir+RSpec. There will be a dedicated chapter on
Cucumber towards the end of this book.
Run Tests From Command Line
In Chapter 2, we created an automated test script using a recorder and ran the test from TestWise.
One advantage of open-source test frameworks such as Watir and Selenium2 is freedom. You can
edit the test scripts in any text editor and run them from command line.
You need to install Ruby first, then install RSpec and preferred web test driver and library (called
Gem in Ruby). Basic steps are:
How Automated Testing works 29
install Ruby interpreter
Window installer: http://rubyinstaller.org Mac: pre-installed with OS Linux: get from package
manager or compile from source
install RSpec
> gem install rspec
install test framework gem(s)
> gem install watir
or
> gem install selenium-webdriver
For windows users, especially the ones who have difficulty installing gems behind a corporate
proxy, you may simply download and install free pre-packaged RubyShell (based on Ruby Windows
Installer) at http://testwisely.com/testwise/downloads.
Once the installation (takes about 1 minute) is complete, we can run a RSpec test from command line.
you need to have some knowledge on typing commands in console (called Command on Windows).
To run test cases in a test script file, enter command
> rspec google_spec.rb
Run multiple test script files in one go:
> rspec first_spec.rb second_spec.rb
Run individual test case in a test script file, supply a line number in chosen test case range.
> rspec google_spec.rb:30
To generate a test report (HTML) after test execution:
> rspec -fh google_spec.html > test_report.html
The command syntax is the same for Mac OS X and Linux platforms.
4. TestWise - Functional Testing IDE
In Chapter 2, we wrote a simple automated test case using TestWise, a functional testing Integration
Development Environment (IDE). You are free to use any text-based editors/IDEs to develop Watir
or Selenium tests, in that case, you can safely skip this chapter. If you want to be more productive
with TestWise, then you might find this chapter useful.
Philosophy of TestWise
The Philosophy of TestWise:
The Power of Text
(inspired from the classic book Pragmatic Programmers)
Convention over Configuration
(inspired from popular Ruby on Rails framework)
Simple
The Power of Text
Unlike some testing tools, the main window of TestWise is a text-based editor, with various testing
functions such as test execution, test refactoring, test navigation,.., etc. The benefits of using plain
text (test scripts):
Use of Source Control system to track revision and compare differences
Powerful text editing, such as Snippets
Search and replace, even across multiple files in project scope
Refactoring (we will cover this in later chapter)
Easy view or edit using any text editors without dependency on proprietary tool
Convention over Configuration
The principle of Convention over Configuration is gaining more acceptance with the success of
Ruby on Rails framework. It makes sense for writing automated tests as well. In the context of testing,
with conventions in place, when a tester opens a new test project, she/he should feel comfortable
and can get working straight way.
TestWise defines simple conventions for the test project structure, test file naming and page classes,
as you will see later in this chapter. This helps communication among team members or seeking
help externally when necessary.
TestWise - Functional Testing IDE 31
Simple
TestWise is designed from the ground up to suit testers, without compromises often found in testing
tools that are based on programming IDEs (which are designed for programmers). Every feature in
TestWise has one single purpose: a better testing experience.
To make new-to-automation testers more willing to adopt, TestWise is designed to be small (around
13.5MB), easy to install, launch quickly and get you started in minutes.
..
Next-Generation Functional Testing Tool
In October 2007, The Agile Alliance held a Functional Testing Tools Visioning Workshop to
envision the next-generation of functional testing tools: We are lacking integrated development
environments that facilitate things like: refactoring test elements, command completion, incremen-
tal syntax validation (based on the domain specific test language), keyboard navigation into the
supporting framework code, debugging, etc. [AAFTTVW 07]
TestWise was designed and implemented before the workshop, but shares the same vision.
TestWise Project Structure
The project structure in TestWise is simple.
Project structure
There are several file types distinguished by icons in TestWise:
Test script files (xxx_spec.rb)
One test script file may contain one or more test cases (the extension .rb means it is a Ruby
script file).
Page class files (xxx_page.rb under /pages folder)
Page class are for reusable Ruby class representing a web page, we will cover it in detail in the
next chapter.
TestWise - Functional Testing IDE 32
Test Helper (test_helper.rb)
Common reusable functions are defined in Test Helper. It is included at the beginning of all
test script files and the functions are available for all test scripts.
Project file (xxx.tpr)
Store project settings. To open a TestWise project, look for a xxx.tpr file
Rakefile
Configuration file for Rake build language (equivalent build.xml in Ant), which can be used
to execute all or a custom suite of test cases.
Test data (under /testdata folder, optional)
The place to put your test data.
Test Execution
Test execution, obviously, is the most important feature for testing tools. TestWise offers several
ways to run tests.
Run test cases in a test script file (F10)
A test script file may contain one or more test cases, which commonly form a logic group.
Run individual test case (Shift+F10)
When developing or debugging (trying to find out what went wrong) a new test case, and you just
want to run this single test case and prefer to leave the web browser at the state when an error
occurred for analyse. And yes, this is the most frequently used method for executing tests.
Run All Tests in folder
Also you can run all tests under a folder.
Run selected tests: Test Suite
A Test Suite is a group of selected test script files to allow a custom set of test cases to be executed
together.
TestWise - Functional Testing IDE 33
Keyboard Navigation
One criteria identified by Agile Alliance work for Next-Gen Functional Testing tools is keyboard
navigation into the supporting framework code. Those used to operating with a mouse all the time
might find keyboard navigation is just a matter of personal preference, and wonder how it is made
into the list?
For any projects that are doing serious automated testing, there will be a large number of test scripts.
When the number is big, being able to find the test case quickly (which at the fast end of spectrum,
means via the keyboard), keyboard navigation becomes more than just a convenience.
Go to Test Script File (Ctrl+T)
Go to file
Go to Test Case (Ctrl+Shift+T)
Go to test case
..
Rockys mouse
Once I worked with a tester nicknamed Rocky who was in his fifties. Despite many doubts, he fell
in love with automated testing quickly. He developed RSI (Repetitive Strain Injury, a potentially
disabling illness caused by prolonged repetitive hand movements) with his mouse hand. Certainly
years of the using computer mice had contributed to that. When we worked together on test cases,
I moved the mouse to the far right side and sometimes even put a piece of paper between him
and the mouse. Changing a habit is never easy, but Rocky was doing admirably well. Weeks later,
Rocky used the keyboard than the mouse and felt more productive as a result. Months later after I
left the project, I met one of his colleagues, who told me: he saw Rocky once snapped the mouse
on his desk, and said to himself: Zhimin said not to use it.
TestWise - Functional Testing IDE 34
Snippets
Snippets in TestWise are small bits of text that expand into full test script statements. The use of
snippets helps to create test steps more effectively when crafted manually. For example, type cl
then press Tab key in a script editor, TestWise will expand it into the test statement below (clicking
a hyperlink):
There are two ways to insert a snippet:
1. Enter snippet abbreviation, then press Tab key
2. Press Ctrl+J and select from the list, or type to narrow down the selection.
After a snippet is expanded, you may type over the highlighted text and press Tab to move to next
one if there is any. For instance,
Snippet
type username then press Tab key, the test statement becomes:
Snippet: move to next stop
Script Library
For testers who are new to the test framework and do not know the script syntax, may have many
how-to questions such as: What is the test script syntax for clicking a button?, How to assert the
checkbox is checked?, , etc. TestWises built-in script library can provide the answers.
TestWise - Functional Testing IDE 35
Script Library
Test Refactoring
Test Refactoring is a process of refining test scripts to make it easier to read, and more importantly,
easier to maintain. One unique feature of TestWise is its refactoring support, performing test
refactoring efficiently and reliably.
We will cover this important topic in later chapters.
Wrap Up
We quickly introduced some features of TestWise to help you develop test scripts more efficiently.
For more details, please check TestWise online documentation and screencasts.
5. Case Study
In this chapter, we will write six automated tests for a test web site.
Test Site
In this over-simplified flight booking system: Agile Travel (http://travel.agileway.net), there are 4
high level functions on 4 web pages:
Sign in Select flights Enter passenger details Pay by credit card (then check confirmation)
Some may suggest: the confirmation has its own page normally. Thats correct, however I combined
the payment page and the confirmation page for a reason: testing AJAX.
We are going to write several test cases to verify core functions below:
Sign in
Select flights
Enter passenger details
Pay by credit card
We will create four test script files, inside which are test cases that are dedicated to testing each core
function.
I suggest you spend a few minutes playing with this web site to get familiar to it, as you do for your
work.
Preparation
The automated test framework used in this case study are Watir (with RWebSpec extension) + RSpec,
and automated tests will be executed in Internet Explorer. (For readers who are more interested in
Selenium or running tests against Firefox and Chrome, we will cover in later chapters. I strongly
recommend all readers go through this exercise.)
http://travel.agileway.net
Case Study 37
Web Site: http://travel.agileway.net
Test user login: agileway/testwise
Platform: IE9 on Windows 7 (works on IE8 and Windows XP as
well) or Mac
Software to be installed: TestWise Community Edition or TestWise Mac Edition,
FireFox with TestWise Recorder plug-in
Create Test Project
Objective
Create a test project in TestWise
Assume there is an existing folder c:\work\agiletravel, we can add a folder ui-tests under it to store
our automated test scripts.
In TestWise, select menu File -> New Project (close the existing project first if there is one), specify
name of test project
test project folder
web site URL
If you want to open this project in TestWise later, select menu File -> Open Project, locate the
project file agiletravel-ui-tests.tpr under c:\work\agiletravel\ui-tests folder.
Case Study 38
Test Suite: Sign in
Objective
Create test cases using a recorder
Multiple test cases in same test script file
Analyse test error
Understand test execution interference
Test Design
We start with two simple (and common) test cases: one positive and one negative
Sign in OK
Sign in failed with invalid password
Positive Case: User can sign in OK
Select menu File -> New File, enter file name as login_spec.rb
A test case skeleton is created in newly created test script file login_spec.rb:
test_case "New Test Case" do
# Test Steps go here
end
Set test case name by changing the text New Test Case to User can sign in OK.
Start FireFox browser, navigate to our test site URL: http://travel.agileway.net, and enable TestWise
Recorder by selecting menu Tools -> TestWise Recorder Sidebar. In Firefox, sign in by entering
user name and password (agileway/testwise), and clicking Sign in button.
A test case is not complete without checks. We could use the presence of the text Welcome
(username) as the determination of a user is signed in successfully. To create this assertion step,
highlight Welcome XXX text, right mouse click and select Add verifyText for .
Case Study 39
Now right mouse click in the recorder window and Copy all recorded test steps:
Paste recorded test steps in the test case in TestWise. Now we get:
test_case "User can sign in OK" do
enter_text("username", "agileway")
enter_text("password", "testwise")
click_button("Sign in")
assert_text_present("Welcome agileway")
end
Run the test (right mouse click any line within the test case and select Run User can sign in OK)
It passed! (indicated by the green tick)
Negative Case: User failed to sign in due to invalid password
Now we continue to add another login related test case in login_spec.rb: user failed to sign in with
invalid password. By using TestWise Recorder, we can quickly create this negative test case as below:
Case Study 40
test_case "User failed to sign in due to invalid password" do
enter_text("username", "agileway")
enter_text("password", "bad pass")
click_button("Sign in")
assert_text_present("Invalid email or password")
end
Run the test and you might get the error below (on line 24):
(If you didnt get this error, you probably closed the browser before executing this test)
Clicking the Test Output tab, error trace tells us that the element with name username could not
be located:
Why? If you switch to the Internet Explorer window, the page showing in IE is the one after signing
in successfully, as the result of executing the first test case. Our second test case was expecting the
home page to enter a user name in a text box. Well, since now page showing in the browser is not
the home page, the test failed.
Case Study 41
If you close the IE window and run this test case again, it will pass.
How can you prevent execution of the first test case from affecting the second one? One solution is
to add a sign off step: click_link("Sign off") at the end of the first test case.
story "User can sign in OK" do
enter_text("username", "agileway")
enter_text("password", "testwise")
click_button("Sign in")
page_text.should contain("Welcome agileway")
click_link("Sign off")
end
Now Click on the toolbar to run the two test cases in login_spec.rb. Both should pass now.
Test Suite: Select Flights
Objective
Verify text across pages
Check dynamic page
Test Case Design
There are quite a few scenarios we could write tests for on this page. For this exercise, we will write
just two:
A return trip
A one-way trip
Case Study 42
Case 1: Return trip
Create a new test script file: flight_spec.rb.
test_case "Return trip" do
enter_text("username", "agileway")
enter_text("password", "testwise")
click_button("Sign in")
click_radio_option("tripType", "return")
select_option("fromPort", "Sydney")
select_option("toPort", "New York")
select_option("departDay", "02")
select_option("departMonth", "May 2012")
select_option("returnDay", "04")
select_option("returnMonth", "June 2012")
click_button("Continue")
assert_text_present("2012-05-02 Sydney to New York")
assert_text_present("2012-06-04 New York to Sydney")
end
You might notice the step below
click_radio_option("tripType", "return")
wasnt included in the recorded test steps. This is because this radio button was already pre-selected.
You may skip this step. I added this step as I want to make sure this radio button is selected. To record
this step, you
click One way radio button
right mouse click in the recorder to clear test steps
click Return radio button
Or you could try inspecting the HTML source manually (see Identify Web Controls section in
Chapter 3).
Case 2: One-way trip
Still in flight_spec.rb, add the second test case: one-way trip.
Case Study 43
test_case "One-way trip" do
enter_text("username", "agileway")
enter_text("password", "testwise")
click_button("Sign in")
click_radio_option("tripType", "oneway")
select_option("fromPort", "Sydney")
select_option("toPort", "New York")
select_option("departDay", "02")
select_option("departMonth", "May 2012")
click_button("Continue")
assert_text_present("2012-05-02 Sydney to New York")
end
You may try adding sign off steps to make both the test cases work. But there is another easier and
cleaner way.
Technique: Use execution hooks
You might have noticed that both test cases start with same 3 sign-in test steps (and end with sign
off test step). If we think about it, we dont have to test the functionality of signing in and signing
off for each test case. In fact, our focus is testing the different scenarios after signed in.
With the knowledge of RSpec, we can move these 3 test steps into a before(:all) execution hook.
This way, we only need to sign in once regardless of how many test cases in this test script file.
before(:all) do
open_browser
enter_text("username", "agileway")
enter_text("password", "testwise")
click_button("Sign in")
end
after(:all) do
close_browser unless debugging?
end
test_case "Return trip" do
click_radio_option("tripType", "return")
# ...
end
Case Study 44
test_case "One-way trip" do
click_radio_option("tripType", "oneway")
# ...
end
If you run the test script file (both test cases), the second test case failed. Thats because after
execution of first test, the browser has gone to the next page: Passenger Page. To make the second
test case (as well the first one) pass, we could use another execution hook: before(:each).
before(:each) do
goto_page("/flights/start") #before each test, make sure on flight page
end
Tip: you could use TestWise Snippets to enter this test step: type gp then press Tab key.
There is no need to use the recorder here, just type in the test step (a good test automation specialist
may use recorders wisely but wont totally depend on them). The string /flights/start is the relative
URL of test site, which you can get by examining the address showing in a browser.
Technique: Check Dynamic UI
In the second test case, when we select the one way trip radio button, the return date section is
hidden. It is done via JavaScript. How do we check that in automated test scripts? We can inspect
the HTML source and review the section containing the return date element. Below is a screenshot
of using FireBug in Firefox.
The HTML fragment <div id=returnTrip> is the section that will be hidden when the One way
radio button is clicked.
Case Study 45
Click the script library icon on the toolbar in TestWise, which provides help on finding test or
assertion steps. Type in hidden and press Enter, we see the assertion statement we need. Navigate
the caret to the right place (after clicking oneway radio button) and press the Insert button, the
following statement will be added in the test script editor:
assert_hidden(:tag, :element_id)
Update it to:
assert_hidden(:div, "returnTrip")
This sounds more complex than it actually is. The screenshot below shows how easy it really is.
The complete test case:
test_case "One-way trip" do
click_radio_option("tripType", "oneway")
assert_hidden(:div, "returnTrip")
select_option("fromPort", "Sydney")
select_option("toPort", "New York")
select_option("departDay", "02")
select_option("departMonth", "May 2012")
click_button("Continue")
assert_text_present("2012-05-02 Sydney to New York")
end
Case Study 46
Enter Passenger Details
Objective
Validation
Assert value in a text field
Use raw Watir test steps
Test Design
For the passenger page, a business rule states that a last name must be provided. We could create
a separate test case for validation, but this seems like overkill. We can simply add the validation
within the main stream test case. That is,
submit the form without entering last name
verify the validation error message
enter first name and last name
submit the form
verify the passenger name is saved
If the passenger details are saved properly, the full name is pre-populated as card holder name on the
credit card page. We could use this as our check, i.e., getting value of text box with name holder_-
name.
Case Study 47
So far the test scripts I have showed are in RWebSpec, an extension of Watir. However at some point
you really need to know Watir, as it provides ultimate flexibility. For this passenger test case, we are
going to write it in Watir syntax. Watir test syntax can be directly used in RWebSpec tests.
TestWise recorder records operations in Watir as well. This time, we create the test case from
recorded test steps under Watir tab.
Now we get the test case (in Watir):
test_case "Can enter passenger details (watir)" do
# sign in steps in before(:all)
if RWebSpec.framework =~ /watir/i
browser.radio(:name => "tripType", :value => "return").click
browser.select_list(:name, "fromPort").select("New York")
browser.select_list(:name, "toPort").select("Sydney")
browser.select_list(:id, "departDay").select("04")
browser.select_list(:id, "departMonth").select("March 2012")
browser.select_list(:id, "returnDay").select("07")
browser.select_list(:id, "returnMonth").select("April 2012")
browser.button(:value,"Continue").click
# now on passenger page
browser.button(:value,"Next").click
browser.text.should include("Must provide last name")
browser.text_field(:name, "passengerFirstName").set("Bob")
browser.text_field(:name, "passengerLastName").set("Tester")
Case Study 48
browser.button(:value,"Next").click
text_field(:name, "holder_name").value.should == "Bob Tester"
end
end
The last assertion step is not from the recorder, you type it in. Also you might have noticed the last
test step missed browser.. Thats OK, RWebSpec allows that.
For Mac/Linux users, equivalent test script in Selenium-WebDriver:
test_case "Can enter passenger details (selenium)" do
if RWebSpec.framework =~ /selenium/i
browser.find_elements(:name => "tripType").each { |elem| elem.click && break\
if elem.attribute("value") == "return" && elem.attribute("type") == "radio" }
Selenium::WebDriver::Support::Select.new(browser.find_element(:name, "fromPo\
rt")).select_by(:text, "New York")
Selenium::WebDriver::Support::Select.new(browser.find_element(:name, "toPort\
")).select_by(:text, "Sydney")
Selenium::WebDriver::Support::Select.new(browser.find_element(:id, "departDa\
y")).select_by(:text, "04")
Selenium::WebDriver::Support::Select.new(browser.find_element(:id, "departMo\
nth")).select_by(:text, "March 2012")
Selenium::WebDriver::Support::Select.new(browser.find_element(:id, "returnDa\
y")).select_by(:text, "07")
Selenium::WebDriver::Support::Select.new(browser.find_element(:id, "returnMo\
nth")).select_by(:text, "April 2012")
browser.find_element(:xpath,"//input[@value='Continue']").click
# now on passenger page
browser.find_element(:xpath,"//input[@value='Next']").click
browser.text.should include("Must provide last name")
browser.find_element(:name, "passengerFirstName").send_keys("Bob")
browser.find_element(:name, "passengerLastName").send_keys("Tester")
browser.find_element(:xpath,"//input[@value='Next']").click
browser.find_element(:name, "holder_name")["value"].should == "Bob Tester"
end
end
Case Study 49
Book confirmation after payment
Objective
AJAX Testing
Retrieving text or value from specific element
Test Design
We navigate our way to the payment page. After filling in credit card details and clicking the Pay
now button, an animated loading image (see below) shows up indicating processing in progress.
After a few seconds, the flight book confirmation is displayed containing booking number and flight
details (and animated loading image disappeared).
Technique: Testing AJAX
I am sure that you are now quite familiar with this kind of user experience - the web page processes
information and shows the results without having to refresh the whole page. The term used to
describe the technology responsible for this enhanced user experience is AJAX. From the testing
perspective, an AJAX operation immediately completes after the mouse/keyboard action (such as
clicking the Pay now button), no page reload is observed. After the server finished processing the
request, seconds or even minutes later, some part of web page may be updated.
One simple solution for testing an AJAX operation is to wait long enough time for the AJAX
operation to fully complete, then perform assertions like below:
Case Study 50
click_button("Pay now")
sleep 30 # wait 30 seconds
assert_text_present("Booking number")
The above approach works, but is not efficient. If the AJAX operation finishes early, test execution
will still pause there and wait unnecessarily. RWebSpec introduces a convenient function try_-
for(seconds) { test steps } to keep trying next test steps every 1 second up to a specified time. If
the operation was performed successfully within the given time, it moves on to the next test step. If
the operation still cannot be performed after that time, an error is thrown.
try_for(30) { assert_text_present("Booking number")}
Technique: Displaying value from specific HTML element in
console
Sometimes it may be useful to get a value or text from a specific element on the page. For example,
if the booking number in this web site is in some number pattern (such as 20120228-123), we can
further verify the booking number against the pattern.
During the development of a test case, we often want to seek confirmation by displaying what we
returned (also known as printing out). In the instance, a tester may want to print to the console the
confirmation number from the test output. In TestWise, you can use the debug function to display
text in the console window, as illustrated below:
The test scripts for this test case:
Case Study 51
test_case "Get booking confirmation after payment" do
# ... up to payment page
click_radio_option("card_type", "master")
enter_text("holder_name", "Bob the Tester")
enter_text("card_number", "4242424242424242")
select_option("expiry_month", "04")
select_option("expiry_year", "2012")
click_button("Pay now")
try_for(30) { assert_text_present("Booking number") }
debug span(:id, "booking_number").text
end
Run all tests
We can run all these 6 test cases (in 4 test script files) at one go. Right mouse click the project folder
and select Run All in . The screenshot below is taken after a successful execution of all 6 test
cases.
Run all tests
A more detailed test report can be found under Test Report tab.
Case Study 52
Wrap Up
We have created several automated test cases in RWebSpec/Watir. Along the way, some techniques
were introduced. After this exercise, you should be ready to write real tests for your project. I expect
that some of you might be very excited, especially after seeing execution of a couple of real tests for
your project, and think that test automation is easy.
Here I want to remind you of the test automation camel . After writing dozens of test scripts, you
soon will face the first hump: Hard to Maintain. But thats OK. In Chapter 7 and 8, I will show you
how to overcome that hump!
After this comparatively long hands-on exercise, we will examine the characteristics of test
automation in next chapter.