Hangperson: a scaffolded (!) ESaaS getting-started assignment

** v1.0, January 2015. Written by Armando Fox and Nick Herson **

In this assignment you'll be introduced to part of the basic cycle of creating SaaS in a disciplined way.

The full Agile/XP cycle we follow in ESaaS includes talking to the customer, using BDD to develop scenarios, turning those scenarios into runnable integration/acceptance tests with Cucumber, using those scenarios plus TDD to drive the creation of actual code, and deploying the result of each iteration's work to the cloud.

In this introductory assignment, we've provided RSpec unit tests to let you use TDD to develop game logic for the popular word-guessing game Hangman. In the full Agile/XP cycle, you'd develop these tests yourself as you code.

You'll then use the Sinatra framework to make the Hangman game available as SaaS. Adapting the game logic for SaaS will introduce you to thinking about RESTful routes and service-oriented architecture. As you develop the "SaaS-ified" Hangman game, you'll use Cucumber to describe how gameplay will work from the player's point of view and as "full stack" integration tests that will drive SaaS development. In the full Agile/XP cycle, you'd develop Cucumber scenarios yourself based on consultation with the customer, and create the necessary step definitions (Cucumber code that turns plain-English scenarios into runnable tests). In this assignment we provide both the scenarios and step definitions for you.

You'll deploy your game to the cloud using Heroku, giving you experience in automating SaaS deployment.

Why Sinatra? This assignment uses the simple Sinatra framework rather than Rails, so that you can focus on tools, mechanics, and SaaS concepts, all of which will readily map to Rails later. Since our app doesn't have a database and has very few functions, Sinatra is an easy way to get started.

Prerequisites

You will need to be comfortable with the following concepts and skills:

• Basic Git usage and how to push your code to GitHub, as described in Appendix A of Engineering Software as a Service.

• "Survival level" Unix command-line skills and facility with an editor to edit code files.

• Basic concepts behind SaaS application architecture, as described in Chapter 2 of Engineering Software as a Service.

Demystifying SaaS app creation

Goal: Understand the steps needed to create, version, and deploy a SaaS app, including tracking the libraries it depends on so that your production and development environments are as similar as possible.

What you will do: Create a simple "hello world" app using the Sinatra framework, version it properly, and deploy it to Heroku.

Creating and versioning a simple SaaS app

SaaS apps are developed on your computer but deployed to production on a server that others can access. We try to minimize the differences between the development and production environments, to avoid difficult-to-diagnose problems in which something works one way on your development computer but a different way (or not at all) when that code is deployed to production.

We have two mechanisms for keeping the development and production environments consistent. The first is version control, such as Git, for the app's code. But since almost all apps also rely on libraries written by others, such as gems in the case of Ruby, we need a way to keep track of which versions of which libraries our app has been tested with, so that the same ones are used in development and production.

Happily, Ruby has a wonderful system for managing gem dependencies: a gem called Bundler looks for a file called Gemfile in the app root directory of each project. Gemfile contains a list of gems and versions your app depends on. Bundler verifies that those gems, and any others that they in turn depend on, are properly installed on your system and accessible to the app.

• Create a new empty directory to hold your new app, and use git init in that directory to start versioning it with Git.

• In that directory, create a new file called Gemfile (the capitalization is important) with the following contents. This file will be a permanent part of your app and will travel with your app anywhere it goes:

source 'https://rubygems.org'
ruby '1.9.3'

gem 'sinatra', '>= 1.4'

The first line says that the preferred place to download any necessary gems is https://rubygems.org, which is where the Ruby community registers "production ready" gems.

The second line specifies which version of the Ruby language interpreter is required. If we omitted this line, Bundler wouldn't try to verify which version of Ruby is available; there are subtle differences between the versions, and not all gems work with all versions, so it's best to specify this.

The last line says we need version 1.4 or later of the sinatra gem. In some cases we don't need to specify which version of a gem we want; in this case we do specify it because we rely on some features that are absent from earlier versions of Sinatra.

Run Bundler

Run the command bundle, which examines your Gemfile to make sure the correct gems (and, where specified, the correct versions) are available, and tries to install them otherwise. This will create a new file Gemfile.lock, which you should place under version control.

• What's the difference between the purpose and contents of Gemfile and Gemfile.lock? Which file is needed to completely reproduce the development environment's gems in the production environment?

Gemfile specifies the gems you need and in some cases the constraints on which version(s) are acceptable. Gemfile.lock records the actual versions found, not only of the gems you specified explicitly but also any other gems on which they depend, so it is the file used by the production environment to reproduce the gems available in the development environment.

• After running bundle, why are there gems listed in Gemfile.lock that were not listed in Gemfile?

Bundler looked up the information for each Gem you requested (in this case, only sinatra) and realized that it depends on other gems, which in turn depend on still others, so it recursively installed all those dependencies. For example, the rack appserver is a gem, and while you didn't explicitly request it, sinatra depends on it. This is an example of the power of automation: rather than requiring you (the app developer) to understand every Gem dependency, Bundler automates that process and lets you focus only on your app's top-level dependencies.

Create a simple SaaS app with Sinatra

As Chapter 2 of ESaaS explains, SaaS apps require a web server to receive HTTP requests from the outside world, and an application server that "connects" your app's logic to the web server. For development, we will use webrick, a very simple Ruby-based web server that would be inappropriate for production but is fine for development. In both development and production, we will use the rack Ruby-based application server, which supports Ruby apps written in various frameworks including Sinatra and Rails.

As Chapter 2 of ESaaS explains, a SaaS app essentially recognizes and responds to HTTP requests corresponding to the application's routes (recall that a route consists of an HTTP method such as GET or POST plus a URI). Sinatra provides an extremely lightweight shorthand for matching a route with the app code to be executed when a request using that route arrives from the Web server.

Create a file in your project called app.rb containing the following:

require 'sinatra'

class MyApp < Sinatra::Base
  get '/' do
    "<!DOCTYPE html><html><head></head><body><h1>Hello World</h1></body></html>"
  end
end

The get method is provided by the Sinatra::Base class, from which our MyApp class inherits; Sinatra::Base is available because we load the Sinatra library in line 1.

• Self-check: What two steps did we take earlier to guarantee that the Sinatra library is available to our app?

We specified gem 'sinatra' in the Gemfile and successfully ran bundle to confirm that the gem is installed and "lock" the correct version of it in Gemfile.lock.

As you see from the above simple example, Sinatra lets you write functions that match an incoming HTTP route, in this case GET '/' (the root URL), a very simple HTML document containing the string Hello World will be returned to the presentation tier as the result of the request.

To run our app, we have to start the application server and presentation tier (web) server. The rack application server is controlled by a file config.ru, which you must now create and version, containing the following:

require './app'
run MyApp

The first line tells Rack that our app lives in the file app.rb, which you created above to hold your app's code. We have to explicitly state that our app file is located in the current directory (.) because require normally looks only in standard system directories to find gems.

We're now ready to test-drive our simple app with this command line:

rackup

This command starts the Rack appserver and the WEBrick webserver. Rack will look for config.ru and attempt to start our app based on the information there.

• From the messages printed out by rackup as Rack is starting, what URL should you type into your browser to communicate with our running app?

http://localhost:9292, since it's running on your own machine (localhost) and by default rack should report that it's listening on port 9292.

Point your Web browser at the running app and verify that you can get "Hello World".

• What happens if you try to visit a non-root URL such as http://localhost:9292/hello and why?

You'll get a humorous error message from the Sinatra framework, since you don't have a route matching get '/hello' in your app. Since Sinatra is a SaaS framework, the error message is packaged up in a Web page and delivered to your browser.

You should now have the following files under version control: Gemfile, Gemfile.lock, app.rb, config.ru. This is a minimal SaaS app: the app file itself, the list of explicitly required gems, the list of actual gems installed including the dependencies implied by the required gems, and a configuration file telling the appserver how to start the app.

Modify the app

Modify app.rb so that instead of "Hello World" it prints "Goodbye World". Save your changes to app.rb and try visiting that URL again.

Confused?

Now go back to the shell window where you ran rackup and press Ctrl-C to stop Rack. Then type rackup again, and once it is running, go back to your web browser and try the URL again. This time it should work.

What this shows you is that if you modify your app while it's running, you have to restart Rack in order for it to "see" those changes. Since restarting it manually is tedious, we'll use the rerun gem, which restarts Rack automatically when it sees changes to files in the app's directory. (Rails does this for you by default during development, as we'll see, but Sinatra doesn't.)

You're probably already thinking: "Aha! If our app depends on this additional gem, we should add it to the Gemfile and run bundle to make sure it's really present." Good thinking. But it may also occur to you that this particular gem wouldn't be necessary in a production environment: we only need it as a tool while developing. Fortunately, there's a way to tell Bundler that some gems are only necessary in certain environments. Add the following to the Gemfile (it doesn't matter where):

group :development do
  gem 'rerun'
end

Any gem specifications inside the group :development block will only be examined if bundle is run in the development environment. (The other environments you can specify are :test and :production, and you can define new environments yourself.) Gem specifications outside of any group block are assumed to apply in all environments.

• Say rerun rackup in the terminal window to start your app and verify the app is running.

• Modify app.rb to print a different message, and verify that the change is detected by rerun and works as expected.

Deploy to Heroku

Make sure the Heroku Toolbelt is installed (when you create an account on Heroku, a download link is provided). Log in to your Heroku account and create a new app; specify any name you want, but we will use sinatra-demo in this discussion.

Follow the instructions that appear after app creation; you don't have to do git init since you've been using Git to manage this simple app's repo all along.

Deploy your app to Heroku and verify that going to http://sinatra-demo.herokuapp.com/ shows your Hello World page.

Summary

• You started a new application project by creating a Gemfile specifying which gems you need and running bundle to verify that they're available and create the Gemfile.lock file that records the versions of gems actually in use.

• You created a Sinatra app in the file app.rb, pointed Rack at this file in config.ru, and used rackup to start the appserver and the WEBrick web server.

• You learned that changing the app's code doesn't automatically cause Rack to reload the app. To save the work of restarting the app manually every time you make a change, you used the rerun gem, adding it to the Gemfile in a way that specifies you won't need it in production, only during development.

• You versioned the important files containing not only your app's code but the necessary info to reproduce all the libraries it relies on and the file that starts up the app.

• You deployed this simple app to Heroku.

With this machinery in mind, download and unpack the homework skeleton into a new directory and let's work on Hangperson.

Developing Hangperson Using TDD and Autotest

Goals: Use test-driven development (TDD) based on the tests we've provided to develop the game logic for Hangman, which forces you to think about what data is necessary to capture the game's state. This will be important when you SaaS-ify the game in the next part.

What you will do: Use autotest, our provided test cases will be re-run each time you make a change to the app code. One by one, the tests will go from red (failing) to green (passing) as you create the app code. By the time you're done, you'll have a working Hangperson game class, ready to be "wrapped" in SaaS using Sinatra.

Overview

Our Web-based version of the popular game "hangman" works as follows:

• The computer picks a random word

• The player guesses letters in order to guess the word

• If the player guesses the word before making seven wrong guesses of letters, they win; otherwise they lose. (Guessing the same letter repeatedly is simply ignored.)

To make the game fun to play, each time you start a new game the app will actually retrieve a random English word from a remote server, so every game will be different. This feature will introduce you not only to using an external service (the random-word generator) as a "building block" in a Service-Oriented Architecture, but also how a Cucumber scenario can test such an app deterministically with tests that break the dependency on the external service at testing time.

• In the app's root directory, say autotest.

This will fire up the Autotest framework, which looks for various files to figure out what kind of app you're testing and what test framework you're using. In our case, it will discover the file called .rspec, which contains RSpec options and indicates we're using the RSpec testing framework. Autotest will therefore look for test files under spec/ and the corresponding class files in lib/.

We've provided a set of 19 test cases to help you develop the game class. Take a look at spec/hangperson_game_spec.rb. It specifies behaviors that it expects from the class lib/hangperson_game.rb.
initially, we have added :pending => true to every spec, so when Autotest first runs these, you should see the test case names printed in yellow, and the report "19 examples, 19 pending."

Now, with Autotest still running, delete :pending => true from line 11, and save the file. You should immediately see Autotest wake up and re-run the tests. You should now have 3 failures and 16 pending examples.

The describe 'new' block means "the following block of tests describe the behavior of a 'new' HangpersonGame instance." The subject block causes a new instance to be created, and the its blocks verify the presence and values of instance variables.

• Self-check: According to our test cases, how many arguments does the game class constructor expect, and therefore what will the first line of the method definition look like that you must add to hangperson_game.rb?

One argument (in this example, "glorp"), and since constructors in Ruby are always named initialize, the first line will be def initialize(new_word) or something similar.

• Self-check: According to the tests in this describe block, what instance variables is a HangpersonGame expected to have?

@word, @guesses, and @wrong_guesses.

You'll need to create getters and setters for these. Hint: use attr_accessor. When you've done this successfully and saved hangperson_game.rb, Autotest should wake up again and the three examples that were previously failing should now be passing (green).

Continue in this manner, removing :pending => true from one example at a time working your way down the specs, until you've implemented all the instance methods of the game class: guess, which processes a guess and modifies the instance variables wrong_guesses and guesses accordingly; check_win_or_lose, which returns one of the symbols :win, :lose, or :play depending on the current game state; and word_with_guesses, which substitutes the correct guesses made so far into the word.

• Debugging Tip: when running tests, you can insert the Ruby command debugger into your app code to drop into the command-line debugger and inspect variables and so on. Type h for help at the debug prompt. Type c to leave the debugger and continue running your code.

• Self-check: take a look at the code in the class method get_random_word, which retrieves a random word from a Web service we found that does just that. Use the following command to verify that the Web service actually works this way. Run it several times to verify that you get different words.

curl --data '' http://watchout4snakes.com/wo4snakes/Random/RandomWord

(Normally the argument to --data would be the encoded form fields, but in this case no form fields are needed; however --data is necessary to force curl to do a POST rather than a GET.) Using curl is a great way to experiment with/debug interactions with external services.

RESTful thinking for HangPerson

Goals: Understand how to expose your app's behaviors as RESTful actions in a SaaS environment, and how to preserve game state across (stateless) HTTP requests to your app using the appserver's provided abstraction for cookies.

What you will do: Create a Sinatra app that makes use of the HangpersonGame logic developed in the previous part, allowing you to play Hangperson via a browser.

Game State

Unlike a shrinkwrapped app, SaaS runs over the stateless HTTP protocol, with each HTTP request causing something to happen in the app. And because HTTP is stateless, we must think carefully about the following 2 questions:

0. What is the total state needed in order for the next HTTP request to pick up the game where the previous request left off?

1. What are the different game actions, and how should HTTP requests map to those actions?

The widely-used mechanism for maintaining state between a browser and a SaaS server is a cookie. The server can put whatever it wants in the cookie (up to a length limit of 4K bytes); the browser promises to send the cookie back to the server on each subsequent requests. Since each user's browser gets its own cookie, the cookie can effectively be used to maintain per-user state.

In most SaaS apps, the amount of information associated with a user's session is too large to fit into the 4KB allowed in a cookie, so as we'll see in Rails, the cookie information is more often used to hold a pointer to state that lives in a database. But for this simple example, the game state is small enough that we can keep it directly in the session cookie.

• Self-check question: enumerate the minimal game state that must be maintained during a game of Hangperson.

The secret word; the list of letters that have been guessed correctly; the list of letters that have been guessed incorrectly. Conveniently, the well-factored HangpersonGame class encapsulates this state using its instance variables, as proper object-oriented design recommends.

The game as a RESTful resource

• Self-check questions: enumerate the player actions that could cause changes in game state.

Guess a letter: possibly modifies the lists of correct or incorrect guesses; possibly results in winning or losing the game.

Start new game: chooses a new word and sets the incorrect and correct guess lists to empty.

In a service-oriented architecture, we do not expose internal state directly; instead we expose a set of HTTP requests that either display or perform some operation on a hypothetical underlying resource. The trickiest and most important part of RESTful design is modeling what your resources are and what operations are possible on them.

In our case, we can think of the game itself as the underlying resource. Doing so results in some important design decisions about how routes (URLs) will map to actions and about the game code itself.
Since we've already identified the game state and player actions that could change it, it makes sense to define the game itself as a class. An instance of that class is a game, and represents the resource being manipulated by our SaaS app.

Mapping resource routes to HTTP requests

Our initial list of operations on the resource might look like this, where we've also given a suggestive name to each action:

0. create: Create a new game
1. show: Show the status of the current game
2. guess: Guess a letter

• Self-check: for a good RESTful design, which of the resource operations should be handled by HTTP GET and which ones should be handled by HTTP POST?

Operations handled with GET should not have side effects on the resource, so show can be handled by a GET, but create and guess (which modify game state) should use POST. (In fact, in a true service-oriented architecture we can also choose to use other HTTP verbs like PUT and DELETE, but we won't cover that in this assignment.)

HTTP is a request-reply protocol and the Web browser is fundamentally a request-reply user interface, so each action by the user must result in something being displayed by the browser.
For the "show status of current game" action, it's pretty clear that what we should show is the HTML representation of the current game, as the word_with_guesses method of our game class does. (In a fancier implementation, we would arrange to draw an image of part of the hanging person.)

But when the player guesses a letter--whether the guess is correct or not--what should be the "HTML representation" of the result of that action?

Answering this question is where the design of many Web apps falters.

In terms of game play, what probably makes most sense is after the player submits a guess, display the new game state resulting from the guess. But we already have a RESTful action for displaying the game state. So we can plan to use an HTTP redirect to make use of that action.

This is an important distinction, because an HTTP redirect triggers an entirely new HTTP request. Since that new request does not "know" what letter was guessed, all of the responsibility for changing the game state is associated with the guess-a-letter RESTful action, and all of the responsibility for displaying the current state of the game without changing it is associated with the display-status action. This is quite different from a scenario in which the guess-a-letter action also displays the game state, because in that case, the game-display action would have access to what letter was guessed. Good RESTful design will keep these responsibilities separate, so that each RESTful action does exactly one thing.

A similar argument applies to the create-new-game action. The responsibility of creating a new game object rests with that action (no pun intended); but once the new game object is created, we already have an action for displaying the current game state.

So we can start mapping our RESTful actions in terms of HTTP requests as follows, using some simple URIs for the routes:

Route and action	Resource operation	Web result
`GET /show`	show game state	display correct & wrong guesses so far
`POST /guess`	update game state with new guessed letter	redirect to `show`
`POST /create`	create new game	redirect to `show`

In a true service-oriented architecture, we'd be nearly done. But a site experienced through a Web browser is not quite a true service-oriented architecture.

Why? Because a human Web user needs a way to POST a form. A GET can be accomplished by just typing a URL into the browser's address bar, but a POST can only happen when the user submits an HTML form (or, as we'll see later, when AJAX code in JavaScript triggers an HTTP action).

So to start a new game, we actually need to provide the user a way to post the form that will trigger the POST /create action. You can think of the resource in question as "the opportunity to create a new game." So we can add another row to our table of routes:

`GET /new`

give human user a chance to start new game

display a form that includes a "start new game" button

Similarly, how does the human user generate the POST for guessing a new letter? Since we already have an action for displaying the current game state (show), it would be easy to include on that same HTML page a "guess a letter" form that, when submitted, generates the POST /guess action.

We will see this pattern mirrored later in Rails: a typical resource (such as the information about a player) will have create and update operations, but to allow a human being to provide the data used to create or update a player record, we will have to provide new and edit actions respectively that allow the user to enter the information on an HTML form.

• Self-check: why is it appropriate for the new action to use GET rather than POST?

The new action doesn't by itself cause any state change: it just returns a form that the player can submit.

• Self-check: explain why the GET /new action wouldn't be needed if your Hangperson game was called as a service in a true service-oriented architecture.

In a true SOA, the service that calls Hangperson can generate an HTTP POST request directly. The only reason for the new action is to provide the human Web user a way to generate that request.

Lastly, when the game is over (whether win or lose), we shouldn't be accepting any more guesses. Since we're planning for our show page to include a letter-guess form, perhaps we should have a different type of show action when the game has ended---one that does not include a way for the player to guess a letter, but (perhaps) does include a button to start a new game. We can even have separate pages for winning and losing, both of which give the player the chance to start a new game. Since the show action can certainly tell if the game is over, it can conditionally redirect to the win or lose action when called.

• Self-check(*): show the route for each of the RESTful actions in the game, based on the description of what the route should do; we've provided the first one.

Show game state, allow player to enter guess; may redirect to Win or Lose	GET /show
Display form that can generate `POST /create`	GET /new
Start new game; redirects to Show Game after changing state	POST /create
Process guess; redirects to Show Game after changing state	POST /guess
Show "you win" page with button to start new game	GET /win
Show "you lose" page with button to start new game	GET /lose

Summary of the design

You may be itchy about not writing any code yet, but you have finished the most difficult and important task: defining the application's basic resources and how the RESTful routes will map them to actions in a SaaS app. To summarize:

• We already have a class to encapsulate the game itself, with instance variables that capture the game's essential state and instance methods that operate on it when the player makes guesses. In the model-view-controller (MVC) paradigm, this is our model.

• Using Sinatra, we will expose operations on the model via RESTful HTTP requests. In MVC, this is our controller.

• We will create HTML views and forms to represent the game state, to allow submitting a guess, to allow starting a new game, and to display a message when the player wins or loses. In MVC, these are our views.

Note that Sinatra does not really enforce MVC or any other design pattern---if anything, it's closest to the Page Controller pattern, where we explicitly match up each RESTful request with an HTML view---but it's a simple enough framework that we can use it to implement MVC in this app since we have only one model. As we'll see later, more powerful MVC-focused frameworks like Rails are much more productive for creating apps that have many types of models.

Connecting HangpersonGame to Sinatra

You've already met Sinatra. Here's what's new in the Sinatra app skeleton app.rb that we provide for Hangperson:

• before do...end is a block of code executed before every SaaS request

• after do...end is executed after every SaaS request

• The calls erb :action cause Sinatra to look for the file views/action.erb and run them through the Embedded Ruby processor, which looks for constructions <%= like this %>, executes the Ruby code inside, and substitutes the result. The code is executed in the same context as the call to erb, so the code can "see" any instance variables set up in the get or post blocks.

• Self-check: @game in this context is an instance variable of what class? (Careful-- tricky!)

It's an instance variable of HangpersonApp. Remember we are dealing with two Ruby classes here: the HangpersonGame class encapsulates the game logic itself (that is, the Model in model-view-controller), whereas HangpersonApp encapsulates the logic that lets us deliver the game as SaaS (you can roughly think of it as the Controller logic plus the ability to render the views via erb).

The Session

We've already identified the items necessary to maintain game state, and encapsulated them in the game class. Since HTTP is stateless, when a new HTTP request comes in, there is no notion of the "current game". What we need to do, therefore, is save the game object in some way between requests.

If the game object were large, we'd probably store it in a database on the server, and place an identifier to the correct database record into the cookie. (In fact, as we'll see, this is exactly what Rails apps do.) But since our game state is small, we can just put the whole thing in the cookie. Sinatra's session library lets us do this: in the context of the Sinatra app, anything we place into the special "magic" hash session[] is preserved across requests. In fact, objects placed there are serialized into a text-friendly form that is preserved for us. This behavior is switched on by the Sinatra call enable :sessions in app.rb.

There is one other session-like object we will use. In some cases above, one action will perform some state change and then redirect to another action, such as when the Guess action (triggered by POST /guess) redirects to the Show action (GET /show) to redisplay the game state after each guess. But what if the Guess action wants to display a message to the player, such as to inform them that they have erroneously repeated a guess? The problem is that since every request is stateless, we need to get that message "across" the redirect, just as we need to preserve game state "across" HTTP requests.

To do this, we use the sinatra-flash gem, which you can see in the Gemfile. flash[] is a hash for remembering short messages that persist until the very next request (usually a redirect), and are then erased.

• Self-check: Why does this save work compared to just storing those messages in the session[] hash?

When we put something in session[] it stays there until we delete it. The common case for a message that must survive a redirect is that it should only be shown once; flash[] includes the extra functionality of erasing the messages after the next request.

Running the Sinatra app

As before, run the shell command rackup to start the app, or rerun rackup if you want to rerun the app each time you make a code change.

• Self-check: based on the output from running this command, what is the full URL you need to visit in order to visit the New Game page?

The web server connected to Sinatra is listening on port 9292 of your computer, so the hostname part of the URL is http://localhost:9292. The Ruby code get '/new' do... in app.rb renders the New Game page, so the full URL is http://localhost:9292/new.

Visit this URL and verify that the Start New Game page appears.

• Self-check: where is the HTML code for this page?

It's in views/new.erb, which is processed into HTML by the erb :new directive.

Verify that when you click the New Game button, you get an error. This is because we've deliberately left the <form> that encloses this button incomplete: we haven't specified where the form should post to. We'll do that next, but we'll do it in a test-driven way.

• Commit all your changes, then deploy to Heroku. Verify that the Heroku-deployed Hangperson behaves the same as your development version before continuing.

Introducing Cucumber

Cucumber is a remarkable tool for writing high-level integration and acceptance tests, terms with which you're already familiar. We'll learn much more about Cucumber later, but for now we will use it to drive the development of your app's code.

Just as you used RSpec to "drive" the creation of the class's methods, you'll next use Cucumber to drive the creation of the SaaS code.

Normally, the cycle would be:

0. Use Cucumber scenario to express end-to-end behavior of a scenario
1. As you start writing the code to make each step of the scenario pass, use RSpec to drive the creation of that step's code
2. Repeat until all scenario steps are passing green

In this assignment we're skipping the middle step since the goal is to give you an overview of all parts of the process. Also, in the first step, for your own apps you'd be creating the Cucumber scenarios yourself; in this assignment we've provided them for you.

Cucumber lets you express integration-test scenarios, which you'll find in the features directory in .feature files. You'll also see a step_definitions subdirectory with a single file game_steps.rb, containing the code used when each "step" in a scenario is executed as part of the test.

As an integration testing tool, Cucumber can be used to test almost any kind of software system as long as there is a way to stimulate the system and a way to inspect the system's behavior.
You select a back end for Cucumber based on how the end system is to be stimulated and inspected.

Since a SaaS server is stimulated by issuing HTTP requests, and its behavior can be inspected by looking at the HTML pages served, we configure Cucumber to use Capybara, a Ruby-based browser simulator that includes a domain-specific language for simulating browser actions and inspecting the SaaS server's responses to those actions.

• Self-check: read the section on "Using Capybara with Cucumber" on Capybara's home page. Which step definitions use Capybara to stimulate the server as a browser would? Which step definitions use Capybara to inspect the app's response to the stimulus?

Step definitions that use visit, click_button, fill_in are simulating a browser by visiting a page and/or filling in a form on that page and clicking its buttons. Those that use have_content are inspecting the output.

• Self-check: Looking at features/guess.feature, what is the role of the three lines following the "Feature:" heading?

They are comments showing the purpose and actors of this story. Cucumber won't execute them.

• Self-check: In the same file, looking at the scenario step Given I start a new game with word "garply", what lines in game_steps.rb will be invoked when Cucumber tries to execute this step, and what is the role of the string "garply" in the step?

Lines 2-5 of the file will execute. Since a step is chosen by matching a regular expression, word will match the first (and in this case only) parenthesis capture group in the regexp, which in this example is garply.

Get your first scenario to pass

We'll first get the "I start a new game" scenario to pass; you'll then use the same techniques to make the other scenarios pass, thereby completing the app. So take a look at the step definition for "I start a new game with word...".

You already saw that you can load the new game page, but get an error when clicking the button for actually creating a new game. You'll now reproduce this behavior with a Cuke scenario.

• Self-check: When the "browser simulator" in Capybara issues the visit '/new' request, Capybara will do an HTTP GET to the partial URL /new on the app. Why do you think visit always does a GET, rather than giving the option to do either a GET or a POST in a given step?

Cucumber/Capybara is only supposed to be able to do what a human user can do. As we discussed earlier, the only way a human user can cause a POST to happen via a web browser is submitting an HTML form, which is accomplished by click_button in Capybara.

Run the "new game" scenario with:

cucumber features/start_new_game.feature

The scenario fails because the <form> tag in views/new.erb is missing the information that tells the browser what URL to post the form to. Based on the table of routes we developed in an earlier section, fill in the <form> tag's attributes appropriately, and add the necessary code to the Sinatra app to recognize this route. (Hint: if you get stuck, take a look at show.erb for a similar example.)

The create-new-game code in the Sinatra app should do the following:

• Call the HangpersonGame class method get_random_word

• Create a new instance of HangpersonGame using that word

• Redirect the browser to the show action

When you finish this task, you should be able to re-run the scenario and have all steps passing green.

At that point, deploy to Heroku again and manually verify this behavior.

• Self-check: What is the significance of using Given vs. When vs. Then in the feature file? What happens if you switch them around? Conduct a simple experiment to find out, then confirm your results by using the Google.

The keywords are all aliases for the same method. Which one you use is determined by what makes the scenario most readable.

Develop the scenario for guessing a letter

For this scenario, in features/guess.feature, we've already provided a correct show.erb HTML file that submits the player's guess to the guess action. You already have a HangpersonGame#guess instance method that has the needed functionality.

• Self-check: In game_steps.rb, look at the code for "I start a new game..." step, and in particular the stub_request command. Given the hint that that command is provided by a Gem (library) called webmock, what's going on with that line, and why is it needed? (Use the Google if needed.)

Webmock lets our tests "intercept" HTTP requests coming from our app and directed to another service. In this case, it's intercepting the POST request (the same one you manually did with curl in an earlier part of the assignment) and faking the reply value. This lets us enforce deterministic behavior of our tests, and also means we're not hitting the real external server each time our test runs.

The special Sinatra hash params[] has a key-value pair for each nonblank field on a submitted form: the key is the symbolized name attribute of the form field and the value is what the user typed into that field, or in the case of a checkbox or radiobutton, the browser-specified values indicating if it's checked or unchecked. ("Symbolized" means the string is converted to a symbol, so "foo" becomes :foo.)

• Self-check: In your Sinatra code for processing a guess, what expression would should you use to extract just the first character of what the user typed in the letter-guess field of the form in show.erb? CAUTION: if the user typed nothing, there won't be any matching key in params[], so dereferencing the form field will give nil. In that case, your code should return the empty string rather than an error.

params[:guess].to_s[0] or its equivalent. to_s converts nil to the empty string in case the form field was left blank (and therefore not included in params at all). [0] grabs the first character only; for an empty string, it returns an empty string.

In the guess code in the Sinatra app, you should:

• Extract the letter submitted on the form.

• Use that letter as a guess on the current game.

• Redirect to the show action so the player can see the result of their guess.

Develop that code and verify that all the steps in features/guess.feature now pass.

• Debugging tip: The Capybara command save_and_open_page placed in a step definition will cause the step to open a Web browser window showing what the page looks like at that point in the scenario. The functionality is provided in part by a gem called launchy which is in the Gemfile.

Corner Cases

By now you should be familiar with the cycle:

0. Pick a new scenario to work on
1. Run the scenario and watch it fail
2. Develop code that makes each step of the scenario pass
3. Repeat till all steps passing.

Use this process to develop the code for the remaining actions win and lose. You will need to add code to the show action that checks whether the game state it is about to show is actually a winning or losing state, and if so, it should redirect to the appropriate win or lose action. Recall that your game logic model has a method for testing if the current game state is win, lose, or keep playing. The scenario game_over.feature tests these behaviors in your SaaS app.

Push to Heroku and make sure everything still works. Give yourself a break and play a few rounds of hangperson.

While you're playing, what happens if you directly go to http://localhost:9292/win? Make sure the player cannot cheat by simply visiting GET /win. The scenario prevent_cheating.feature tests these behaviors.

• What to submit: Make sure all Cucumber scenarios are passing. A shorthand way to run all of them is cucumber features/ which runs all .feature files in the given directory. When all are passing, deploy to Heroku and submit the URL of your deployed game.

Conclusion

This assignment has served as a microcosm or miniature tour of the entire course: during the rest of the course we will investigate each of these in much more detail, and we will also add new techniques---

• Test-driven development (TDD) will let you write much more detailed tests for your code and determine its coverage, that is, how thoroughly your tests exercise your code. We will use RSpec to do test-first development, in which we write tests before we write the code, watch the test fail ("red"), quickly write just enough code to make the test pass ("green"), clean up (refactor) the code, and go on to the next test. We will use the autotest tool to help us get into a rhythm of red--green--refactor. In this assignment we provided the specs for you; when designing your own app, you'll write them yourself.

• Code metrics will give us insight into the quality of our code: is it concise? Is it factored in a way that minimizes the cost of making changes and enhancements? Does a particular class try to do too much (or too little)? We will use CodeClimate (among other tools) to help us understand the answers. We can check both quantitative metrics, such as test coverage and complexity of a single method, and qualitative ones, such as adherence to the SOLID Principles of object-oriented design.

• Refactoring means modifying the structure of your code to improve its quality (maintainability, readability, modifiability) while preserving its behavior. We will learn to identify antipatterns -- warning signs of deteriorating quality in your code -- and opportunities to fix them, sometimes by applying design patterns that have emerged as "templates" capturing an effective solution to a class of similar problems.

Optional challenge assignment

Verify that a file exists on the courseware VM called /usr/share/dict/words. What's in it?

This file has been part of Unix since almost the beginning of time, although it has moved around since then. It has been used by programs ranging from spelling checkers to password guessers.

Why do you think we used a remote SOA endpoint to generate random words for Hangperson rather than /usr/share/dict/words? (Hint: It's not simply to give you an initial taste of using SOA.)

In many hosted deployment environments, you cannot rely on the presence of the local filesystem, and in shared environments such as Heroku you may not have access to the filesystem at all. All state must be stored in managed storage such as a database or other per-app abstraction, or managed on remote servers.

Using Sinatra, create and deploy locally (on your computer) a SOA endpoint that mimics the behavior of the Random Word Generator service we used in this assignment. A simple implementation would respond to the RESTful request GET / with a random word drawn randomly from /usr/share/dict/words.