Table

of Contents
Introduction

1.1

License

1.2

Why Angular 2?

1.3

EcmaScript 6 and TypeScript Features

2.1

ES6

2.1.1
Classes

2.1.1.1

Refresher on 'this'

2.1.1.2

Arrow Functions

2.1.1.3

Template Strings

2.1.1.4

Inheritance

2.1.1.5

Constants and Block Scoped Variables

2.1.1.6

...spread and ...rest

2.1.1.7

Destructuring

2.1.1.8

Modules

2.1.1.9

TypeScript

2.1.2

Getting Started With TypeScript

2.1.2.1

Working With tsc

2.1.2.2

Typings

2.1.2.3

Linting

2.1.2.4

TypeScript Features

2.1.2.5

TypeScript Classes

2.1.2.6

Interfaces

2.1.2.7

Shapes

2.1.2.8

Type Inference

2.1.2.9

Decorators

2.1.2.10

Property Decorators

2.1.2.11

Class Decorators

2.1.2.12

Parameter Decorators

2.1.2.13

The JavaScript Toolchain

Source Control: git

3.1
3.1.1

The Command Line

3.1.2

Command Line JavaScript: NodeJS

3.1.3

Back-End Code Sharing and Distribution: npm

3.1.4

Module Loading, Bundling and Build Tasks: Webpack

3.1.5

Chrome

3.1.6

Bootstrapping an Angular 2 Application

4.1

Understanding the File Structure

4.1.1

Bootstrapping Providers

4.1.2

Components in Angular 2

5.1

Creating Components

5.1.1

Application Structure with Components

5.1.2

Passing Data into a Component

5.1.2.1

Responding to Component Events

5.1.2.2

Using Two-Way Data Binding

5.1.2.3

Accessing Child Components from Template

5.1.2.4

Projection

5.1.3

Structuring Applications with Components

5.1.4

Using Other Components

5.1.5

Directives
Attribute Directives

6.1
6.1.1

NgStyle Directive

6.1.1.1

NgClass Directive

6.1.1.2

Structural Directives

6.1.2

NgIf Directive

6.1.2.1

NgFor Directive

6.1.2.2

NgSwitch Directives

6.1.2.3

Using Multiple Structural Directives

6.1.2.4

Advanced Components

7.1

Component Lifecycle

7.1.1

Accessing Other Components

7.1.2

View Encapsulation

7.1.3

ElementRef

7.1.4

Observables
Using Observables

8.1
8.1.1
3

Error Handling

8.1.2

Disposing Subscriptions and Releasing Resources

8.1.3

Observables vs Promises

8.1.4

Using Observables From Other Sources

8.1.5

Observables Array Operations

8.1.6

Combining Streams with flatMap

8.1.7

Cold vs Hot Observables

8.1.8

Summary

8.1.9

Angular 2 Dependency Injection

9.1

What is DI?

9.1.1

DI Framework

9.1.2

Angular 2's DI

9.1.3

@Inject() and @Injectable

9.1.3.1

Injection Beyond Classes

9.1.3.2

The Injector Tree

9.1.3.3

Change Detection

10.1

Change Detection Strategies in Angular 1 vs Angular 2

10.1.1

How Change Detection Works

10.1.2

Change Detector Classes

10.1.3

Change Detection Strategy: OnPush

10.1.4

Enforcing Immutability

10.1.5

Additional Resources

10.1.6

Advanced Angular
Directives
Creating an Attribute Directive

11.1
11.1.1
11.1.1.1

Listening to an Element Host

11.1.1.1.1

Setting Properties in a Directive

11.1.1.1.2

Creating a Structural Directive

11.1.1.2

View Containers and Embedded Views

11.1.1.2.1

Providing Context Variables to Directives

11.1.1.2.2

AoT

11.1.2
AoT limitations

11.1.2.1

AoT Configuration

11.1.2.2

Immutable.js

12.1

What is Immutability?

12.1.1

The Case for Immutability

12.1.2

JavaScript Solutions

12.1.3

Object.assign

12.1.3.1

Object.freeze

12.1.3.2

Immutable.js Basics

12.1.4

Immutable.Map

12.1.4.1

Map.merge

12.1.4.1.1

Nested Objects

12.1.4.2

Deleting Keys

12.1.4.2.1

Maps are Iterable

12.1.4.2.2

Immutable.List

12.1.4.3

Performance

12.1.4.4

Persistent and Transient Data Structures

12.1.4.5

Official Documentation

12.1.4.6

Pipes

13.1

Using Pipes

13.1.1

Custom Pipes

13.1.2

Stateful Pipes

13.1.3

Forms

14.1

Getting Started

14.1.1

Template-Driven Forms

14.1.2

Nesting Form Data

14.1.2.1

Using Template Model Binding

14.1.2.2

Validating Template-Driven Forms

14.1.2.3

FormBuilder

14.1.3

FormBuilder Basics

14.1.3.1

Validating FormBuilder Forms

14.1.3.2

FormBuilder Custom Validation

14.1.3.3

Visual Cues for Users

Modules

14.1.4
15.1

What is an Angular 2 Module?

15.1.1

Adding Components, Pipes and Services to a Module

15.1.2
5

Creating a Feature Module

15.1.3

Directive Duplications

15.1.4

Lazy Loading a Module

15.1.5

Lazy Loading and the Dependency Injection Tree

15.1.6

Shared Modules and Dependency Injection

15.1.7

Sharing the Same Dependency Injection Tree

15.1.8

Routing

16.1

Why Routing?

16.1.1

Configuring Routes

16.1.2

Redirecting the Router to Another Route

16.1.3

Defining Links Between Routes

16.1.4

Dynamically Adding Route Components

16.1.5

Using Route Parameters

16.1.6

Defining Child Routes

16.1.7

Controlling Access to or from a Route

16.1.8

Passing Optional Parameters to a Route

16.1.9

Using Auxiliary Routes

Redux and Ngrx

16.1.10
17.1

Review of Reducers and Pure Functions

17.1.1

Reducers as State Management

17.1.2

Redux Actions

17.1.3

Configuring your Application to use Redux

17.1.4

Using Redux with Components

17.1.5

Redux and Component Architecture

17.1.6

Getting More From Redux and Ngrx

17.1.7

TDD Testing

18.1

The Testing Toolchain

18.1.1

Test Setup

18.1.2

Filename Conventions

18.1.2.1

Karma Configuration

18.1.2.2

TestBed Configuration (Optional)

18.1.2.3

Typings

18.1.2.4

Executing Test Scripts

18.1.2.5

Simple Test

18.1.3

Using Chai

18.1.4

Testing Components

18.1.5

Verifying Methods and Properties

18.1.5.1

Injecting Dependencies and DOM Changes

18.1.5.2

Overriding Components for Testing

18.1.5.2.1

Testing Asynchronous Actions

18.1.5.3

Refactoring Hard-to-Test Code

18.1.5.4

Testing Services

18.1.6

Testing Strategies for Services

18.1.6.1

Testing HTTP Requests

18.1.6.2

Using MockBackend

18.1.6.2.1

Alternative Mocking Strategy

18.1.6.2.2

Testing JSONP and XHR Back-Ends

18.1.6.2.3

Executing Tests Asynchronously

Testing Redux

18.1.6.3
18.1.7

Testing Simple Actions

18.1.7.1

Testing Complex Actions

18.1.7.2

Testing Reducers

18.1.7.3

Afterthoughts

18.1.7.4

Migrating Angular 1.x Projects to Angular 2

Migration Prep

19.1
19.1.1

Upgrading To Angular 1.3+ Style

19.1.1.1

Migrating To TypeScript

19.1.1.2

Using Webpack

19.1.1.3

Choosing an Upgrade Path

19.1.2

Avoiding Total Conversion

19.1.3

Using ng-forward (Angular 1.x Using 2 Style)

19.1.4

Using ng-upgrade (Angular 1.x Coexisting With Angular 2

19.1.5

Bootstrapping ng-upgrade

19.1.5.1

Downgrading Components

19.1.5.2

Upgrading Components

19.1.5.3

Projecting Angular 1 Content into Angular 2 Components

19.1.5.4

Transcluding Angular 2 Components into Angular 1 Directives

19.1.5.5
7

Injecting Across Frameworks

19.1.5.6

Upgrading Components Strategically

19.1.5.7

Project Setup

20.1

Webpack

20.1.1

Installation and Usage

20.1.1.1

Loaders

20.1.1.2

Plugins

20.1.1.3

Summary

20.1.1.4

NPM Scripts Integration

Angular CLI

20.1.2
21.1

Setup

21.1.1

Creating a New App

21.1.2

Serving the App

21.1.3

Creating Components

21.1.4

Creating Routes

21.1.5

Creating Other Things

21.1.6

Testing

21.1.7

Linting

21.1.8

CLI Command Overview

21.1.9

Adding Third Party Libraries

21.1.10

Integrating an Existing App

21.1.11

Accessibility in Angular 2

22.1

Why Make my Application Accessible?

22.1.1

Key Concerns of Accessible Web Applications

22.1.2

Semantic Markup

22.1.2.1

Keyboard Accessibility

22.1.2.2

Visual Assistance

22.1.2.3

Testing for Accessibility

22.1.3

Is my Application Readable?

22.1.3.1

Is my Application Predictable?

22.1.3.2

Is my Application Navigable?

22.1.3.3

Testing with Screen Readers

22.1.3.4

Additional Resources

22.1.4

Internationalization in Angular 2

23.1

What is the process like and how is involved?

23.1.1

Marking text in our templates

23.1.2

Extracting translation text using the Angular CLI

23.1.3

How to import the completed translation files

23.1.4

Using the AoT Compiler

23.1.4.1

Using the JiT Compiler

23.1.4.2

Glossary

24.1

Other Resources

24.2

Introduction

Rangle's Angular 2 Training Book

Over the last three and a half years, AngularJS has become the leading open source
JavaScript application framework for hundreds of thousands of programmers around the
world. The "1.x" version of AngularJS has been widely used and became extremely popular
for complex applications. The new "Angular 2" has also announced its final release version.

About Rangles Angular 2 Training Book

We developed this book to be used as course material for Rangle's Angular 2 training, but
many people have found it to be useful for learning Angular 2 on their own. This book will
cover the most important Angular 2 topics, from getting started with the Angular 2 toolchain
to writing Angular 2 applications in a scalable and maintainable manner.
If you find this material useful, you should also consider registering for one of Rangles
training courses, which facilitate hands-on learning and are a great fit for companies that
need to transition their technology to Angular 2, or individuals looking to upgrade their skills.
Rangle.io also has an Angular 1.x book which is geared towards writing Angular 1.x
applications in an Angular 2 style. We hope you enjoy this book. We welcome your feedback
in the Discussion Area.

10

Introduction

11

License

License
Creative Commons Attribution-ShareAlike 4.0 International (CC BY-SA 4.0)
This is a human-readable summary of (and not a substitute for) the license.

You are free to:

Share copy and redistribute the material in any medium or format
Adapt remix, transform and build upon the material for any purpose, even commercially.
The licensor cannot revoke these freedoms as long as you follow the license terms.

Under the following terms:

Attribution You must give appropriate credit, provide a link to the license, and indicate if
changes were made. You may do so in any reasonable manner, but not in any way that
suggests the licensor endorses you or your use.
ShareAlike If you remix, transform or build upon the material, you must distribute your
contributions under the same license as the original.
No additional restrictions You may not apply legal terms or technological measures that
legally restrict others from doing anything the license permits.

12

Why Angular 2?

Why Angular 2?
There are many front-end JavaScript frameworks to choose from today, each with its own
set of trade-offs. Many people were happy with the functionality that Angular 1.x afforded
them. Angular 2 improved on that functionality and made it faster, more scalable and more
modern. Organizations that found value in Angular 1.x will find more value in Angular 2.

Angular 2's Advantages

The first release of Angular provided programmers with the tools to develop and architect
large scale JavaScript applications, but its age has revealed a number of flaws and sharp
edges. Angular 2 was built on five years of community feedback.

Angular 2 Is Easier
The new Angular 2 codebase is more modern, more capable and easier for new
programmers to learn than Angular 1.x, while also being easier for project veterans to work
with.
With Angular 1, programmers had to understand the differences between Controllers,
Services, Factories, Providers and other concepts that could be confusing, especially for
new programmers.
Angular 2 is a more streamlined framework that allows programmers to focus on simply
building JavaScript classes. Views and controllers are replaced with components, which can
be described as a refined version of directives. Even experienced Angular programmers are
not always aware of all the capabilities of Angular 1.x directives. Angular 2 components are
considerably easier to read, and their API features less jargon than Angular 1.x's directives.
Additionally, to help ease the transition to Angular 2, the Angular team has added a
.component method to Angular 1.5, which has been back-ported by community member

Todd Motto to v1.3.

TypeScript
Angular 2 was written in TypeScript, a superset of JavaScript that implements many new
ES2016+ features.

13

Why Angular 2?

By focusing on making the framework easier for computers to process, Angular 2 allows for
a much richer development ecosystem. Programmers using sophisticated text editors (or
IDEs) will notice dramatic improvements with auto-completion and type suggestions. These
improvements help to reduce the cognitive burden of learning Angular 2. Fortunately for
traditional ES5 JavaScript programmers this does not mean that development must be done
in TypeScript or ES2015: programmers can still write vanilla JavaScript that runs without
transpilation.

Familiarity
Despite being a complete rewrite, Angular 2 has retained many of its core concepts and
conventions with Angular 1.x, e.g. a streamlined, "native JS" implementation of dependency
injection. This means that programmers who are already proficient with Angular will have an
easier time migrating to Angular 2 than another library like React or framework like Ember.

Performance and Mobile

Angular 2 was designed for mobile from the ground up. Aside from limited processing power,
mobile devices have other features and limitations that separate them from traditional
computers. Touch interfaces, limited screen real estate and mobile hardware have all been
considered in Angular 2.
Desktop computers will also see dramatic improvements in performance and
responsiveness.
Angular 2, like React and other modern frameworks, can leverage performance gains by
rendering HTML on the server or even in a web worker. Depending on application/site
design this isomorphic rendering can make a user's experience feel even more
instantaneous.
The quest for performance does not end with pre-rendering. Angular 2 makes itself portable
to native mobile by integrating with NativeScript, an open source library that bridges
JavaScript and mobile. Additionally, the Ionic team is working on an Angular 2 version of
their product, providing another way to leverage native device features with Angular 2.

Project Architecture and Maintenance

The first iteration of Angular provided web programmers with a highly flexible framework for
developing applications. This was a dramatic shift for many web programmers, and while
that framework was helpful, it became evident that it was often too flexible. Over time, best
practices evolved, and a community-driven structure was endorsed.

14

Why Angular 2?

Angular 1.x tried to work around various browser limitations related to JavaScript. This was
done by introducing a module system that made use of dependency injection. This system
was novel, but unfortunately had issues with tooling, notably minification and static analysis.
Angular 2.x makes use of the ES2015 module system, and modern packaging tools like
webpack or SystemJS. Modules are far less coupled to the "Angular way", and it's easier to
write more generic JavaScript and plug it into Angular. The removal of minification
workarounds and the addition of rigid prescriptions make maintaining existing applications
simpler. The new module system also makes it easier to develop effective tooling that can
reason better about larger projects.

New Features
Some of the other interesting features in Angular 2 are:
Form Builder
Change Detection
Templating
Routing
Annotations
Observables
Shadow DOM

Differences Between Angular 1 & 2

Note that "Transitional Architecture" refers to a style of Angular 1 application written in a
way that mimics Angular 2's component style, but with controllers and directives instead
of TypeScript classes.

15

Why Angular 2?

Old School
Angular 1.x

Angular 1.x
Best
Practices

Transitional
Architecture

Angular 2

Nested scopes
("$scope",
watches)

Used
heavily

Avoided

Avoided

Gone

Directives vs
controllers

Use as
alternatives

Used together

Directives as
components

Component
directives

Controller and
service
implementation

Functions

Functions

ES6 classes

ES6
classes

Module system

Angular's
modules

Angular's
modules

ES6 modules

ES6
modules

Transpiler required

No

No

TypeScript

TypeScript

16

EcmaScript 6 and TypeScript Features

EcmaScript 6 and TypeScript Features

Figure: ES6 and TypeScript

The language we usually call "JavaScript" is formally known as "EcmaScript". The new
version of JavaScript, known as "ES6", offers a number of new features that extend the
power of the language.
ES6 is not widely supported in today's browsers, so it needs to be transpiled to ES5. You
can choose between several transpilers, but we'll be using TypeScript, which is what the
Angular team uses to write Angular 2. Angular 2 makes use of a number of features of ES6
and TypeScript.

17

ES6

ES6
JavaScript was created in 1995, but the language is still thriving today. There are subsets,
supersets, current versions and the latest version ES6 that brings a lot of new features.
Some of the highlights:
Classes
Arrow Functions
Template Strings
Inheritance
Constants and Block Scoped Variables
Spread and Rest operators
Destructuring
Modules

18

Classes

Classes
Classes are a new feature in ES6, used to describe the blueprint of an object and make
EcmaScript's prototypical inheritance model function more like a traditional class-based
language.
class Hamburger {
constructor() {
// This is the constructor.
}
listToppings() {
// This is a method.
}
}

Traditional class-based languages often reserve the word this to reference the current
(runtime) instance of the class. In Javascript this refers to the calling context and therefore
can change to be something other than the object.

Object
An object is an instance of a class which is created using the new operator. When using a
dot notation to access a method on the object, this will refer to the object to the left of the
dot.
let burger = new Hamburger();
burger.listToppings();

In the snippet above, whenever this is used from inside class Hamburger, it will refer to
object burger .

Changing Caller Context

JavaScript code can optionally supply this to a method at call time using one of the
following.
Function.prototype.call(object [,arg, ...])
Function.prototype.bind(object [,arg, ...])
Function.prototype.apply(object [,argsArray])

19

Classes

20

Refresher on 'this'

A Refresher on this
Inside a JavaScript class we'll be using this keyword to refer to the instance of the class.
E.g., consider this case:
class Toppings {
...
formatToppings() { /* implementation details */ }
list() {
return this.formatToppings(this.toppings);
}
}

Here this refers to an instance of the Toppings class. As long as the list method is
called using dot notation, like myToppings.list() , then this.formatToppings(this.toppings)
invokes the formatToppings() method defined on the instance of the class. This will also
ensure that inside formatToppings , this refers to the same instance.
However, this can also refer to other things. There are two basic cases that you should
remember.
1. Method invocation:
someObject.someMethod();

Here, this used inside someMethod will refer to someObject , which is usually what
you want.
2. Function invocation:
someFunction();

Here, this used inside someFunction can refer to different things depending on
whether we are in "strict" mode or not. Without using the "strict" mode, this refers to
the context in which someFunction() was called. This rarely what you want, and it can
be confusing when this is not what you were expecting, because of where the
function was called from. In "strict" mode, this would be undefined, which is slightly
less confusing.

21

Refresher on 'this'

View Example
One of the implications is that you cannot easily detach a method from its object. Consider
this example:
var log = console.log;
log('Hello');

In many browsers this will give you an error. That's because log expects this to refer to
console , but the reference was lost when the function was detached from console .

This can be fixed by setting this explicitly. One way to do this is by using bind() method,
which allows you to specify the value to use for this inside the bound function.
var log = console.log.bind(console);
log('Hello');

You can also achieve the same using Function.call and Function.apply , but we won't
discuss this here.
Another instance where this can be confusing is with respect to anonymous functions, or
functions declared within other functions. Consider the following:
class ServerRequest {
notify() {
...
}
fetch() {
getFromServer(function callback(err, data) {
this.notify(); // this is not going to work
});
}
}

In the above case this will not point to the expected object: in "strict" mode it will be
undefined . This leads to another ES6 feature - arrow functions, which will be covered next.

22

Arrow Functions

Arrow Functions
ES6 offers some new syntax for dealing with this : "arrow functions".
Arrow functions also make higher order functions much easier to work with.
The new "fat arrow" notation can be used to define anonymous functions in a simpler way.
Consider the following example:
items.forEach(function(x) {
console.log(x);
incrementedItems.push(x+1);
});

This can be rewritten as an "arrow function" using the following syntax:

items.forEach((x) => {
console.log(x);
incrementedItems.push(x+1);
});

Functions that calculate a single expression and return its values can be defined even
simpler:
incrementedItems = items.map((x) => x+1);

The latter is almost equivalent to the following:

incrementedItems = items.map(function (x) {
return x+1;
});

There is one important difference, however: arrow functions do not set a local copy of this ,
arguments , super , or new.target . When this is used inside an arrow function

JavaScript uses the this from the outer scope. Consider the following example:

23

Arrow Functions

class Toppings {
constructor(toppings) {
this.toppings = Array.isArray(toppings) ? toppings : [];
}
outputList() {
this.toppings.forEach(function(topping, i) {
console.log(topping, i + '/' + this.toppings.length); // no this
})
}
}
var ctrl = new Toppings(['cheese', 'lettuce']);
ctrl.outputList();

Let's try this code on ES6 Fiddle (http://www.es6fiddle.net/). As we see, this gives us an
error, since this is undefined inside the anonymous function.
Now, let's change the method to use the arrow function:
class Toppings {
constructor(toppings) {
this.toppings = Array.isArray(toppings) ? toppings : [];
}
outputList() {
this.toppings
.forEach((topping, i) => console
.log(topping, i + '/' + this.toppings.length) // `this` works!
)
}
}
var ctrl = new Toppings(['cheese', 'lettuce']);

Here this inside the arrow function refers to the instance variable.
Warning arrow functions do not have their own arguments variable, which can be confusing
to veteran JavaScript programmers. super and new.target are also scoped from the outer
enclosure.

24

Template Strings

Template Strings
In traditional JavaScript, text that is enclosed within matching " or ' marks is considered
a string. Text within double or single quotes can only be on one line. There was no way to
insert data into these strings. This resulted in a lot of ugly concatenation code that looked
like:

var name = 'Sam';

var age = 42;
console.log('hello my name is ' + name + ' I am ' + age + ' years old');

ES6 introduces a new type of string literal that is marked with back ticks (`). These string
literals can include newlines, and there is a string interpolation for inserting variables into
strings:

var name = 'Sam';

var age = 42;
console.log(`hello my name is ${name}, and I am ${age} years old`);

There are all sorts of places where this kind of string can come in handy, and front-end web
development is one of them.

25

Inheritance

Inheritance
JavaScript's inheritance works differently from inheritance in other languages, which can be
very confusing. ES6 classes provide a syntactic sugar attempting to alleviate the issues with
using prototypical inheritance present in ES5.
To illustrate this, let's image we have a zoo application where types of birds are created. In
classical inheritance, we define a base class and then subclass it to create a derived class.

Subclassing
The example code below shows how to derive Penguin from Bird using the extends
keyword. Also pay attention to the super keyword used in the subclass constructor of
Penguin , it is used to pass the argument to the base class Bird 's constructor.

The Bird class defines the method walk which is inherited by the Penguin class and is
available for use by instance of Penguin objects. Likewise the Penguin class defines the
method swim which is not avilable to Bird objects. Inheritance works top-down from base
class to its subclass.

Object Initialization
The class constructor is called when an object is created using the new operator, it will be
called before the object is fully created. A consturctor is used to pass in arguments to
initialize the newly created object.
The order of object creation starts from its base class and then moves down to any
subclass(es).

26

Inheritance

// Base Class : ES6

class Bird {
constructor(weight, height) {
this.weight = weight;
this.height = height;
}
walk() {
console.log('walk!');
}
}
// Subclass
class Penguin extends Bird {
constructor(weight, height) {
super(weight, height);
}
swim() {
console.log('swim!');
}
}
// Penguin object
let penguin = new Penguin(...);
penguin.walk(); //walk!
penguin.swim(); //swim!

Below we show how prototypal inheritance was done before class was introduced to
JavaScript.

27

Inheritance

// JavaScript classical inheritance.

// Bird constructor
function Bird(weight, height) {
this.weight = weight;
this.height = height;
}
// Add method to Bird prototype.
Bird.prototype.walk = function() {
console.log("walk!");
};
// Penguin constructor.
function Penguin(weight, height) {
Bird.call(this, weight, height);
}
// Prototypal inheritance (Penguin is-a Bird).
Penguin.prototype = Object.create( Bird.prototype );
Penguin.prototype.constructor = Penguin;
// Add method to Penguin prototype.
Penguin.prototype.swim = function() {
console.log("swim!");
};
// Create a Penguin object.
let penguin = new Penguin(50,10);
// Calls method on Bird, since it's not defined by Penguin.
penguin.walk(); // walk!
// Calls method on Penguin.
penguin.swim(); // swim!

28

Constants and Block Scoped Variables

Constants and Block Scoped Variables

ES6 introduces the concept of block scoping. Block scoping will be familiar to programmers
from other languages like C, Java, or even PHP. In ES5 JavaScript and earlier, var s are
scoped to function s, and they can "see" outside their functions to the outer context.
var five = 5;
var threeAlso = three; // error
function scope1() {
var three = 3;
var fiveAlso = five; // == 5
var sevenAlso = seven; // error
}
function scope2() {
var seven = 7;
var fiveAlso = five; // == 5
var threeAlso = three; // error
}

In ES5 functions were essentially containers that could be "seen" out of, but not into.
In ES6 var still works that way, using functions as containers, but there are two new ways
to declare variables: const and let .
const and let use { and } blocks as containers, hence "block scope". Block scoping

is most useful during loops. Consider the following:

var i;
for (i = 0; i < 10; i += 1) {
var j = i;
let k = i;
}
console.log(j); // 9
console.log(k); // undefined

Despite the introduction of block scoping, functions are still the preferred mechanism for
dealing with most loops.
let works like var in the sense that its data is read/write. let is also useful when used

in a for loop. For example, without let, the following example would output 5,5,5,5,5 :

29

Constants and Block Scoped Variables

for(var x=0; x<5; x++) {

setTimeout(()=>console.log(x), 0)
}

However, when using let instead of var , the value would be scoped in a way that people
would expect.
for(let x=0; x<5; x++) {
setTimeout(()=>console.log(x), 0)
}

Alternatively, const is read-only. Once const has been assigned, the identifier cannot be
reassigned.
For example:
const myName = 'pat';
let yourName = 'jo';
yourName = 'sam'; // assigns
myName = 'jan'; // error

The read-only nature can be demonstrated with any object:

const literal = {};
literal.attribute = 'test'; // fine
literal = []; // error;

However there are two cases where const does not work as you think it should.
1. A const object literal.
2. A const reference to an object.

Const Object Literal

const person = {
name: 'Tammy'
};
person.name = 'Pushpa'; // OK, name property changed.
person = null; // "TypeError: Assignment to constant variable.

30

Constants and Block Scoped Variables

The example above demonstrates that we are able to change the name property of object
person, but we are unable to reset the reference person since it has been marked as
const .

Const Reference To An Object

Something similar to the above code is using a const reference, below we've switch to
using let for the literal object.
let person = {
name: 'Tammy'
};
const p = person;
p.name = 'Pushpa'; // OK, name property changed.
p = null; // "TypeError: Assignment to constant variable.

Take away, marking an object reference const does not make properties inside the object
const.
Ref:.

31

...spread and ...rest

Operators Spread and Rest

A Spread operator allows in-place expansion of an expression for the following cases:
1. Array
2. Function call
3. Multiple variable destructuring
The Rest operator works in the opposite direction of the spread operator, it collects an
indefinite number of comma separated expressions into an array.

Operator Spread
Spread example:
const add = (a, b) => a + b;
let args = [3, 5];
add(...args); // same as `add(args[0], args[1])`, or `add.apply(null, args)`

Functions aren't the only place in JavaScript that makes use of comma separated lists arrays can now be concatenated with ease:
let cde = ['c', 'd', 'e'];
let scale = ['a', 'b', ...cde, 'f', 'g']; // ['a', 'b', 'c', 'd', 'e', 'f', 'g']

Similarly, object literals can do the same thing:

let mapABC = { a: 5, b: 6, c: 3};
let mapABCD = { ...mapABC, d: 7}; // { a: 5, b: 6, c: 3, d: 7 }

Operator Rest
Rest arguments share the ellipsis like syntax of rest operators but are used for a different
purpose. Rest arguments are used to access a variable number of arguments passed to a
function. For example:

32

...spread and ...rest

function addSimple(a, b) {
return a + b;
}
function add(...numbers) {
return numbers[0] + numbers[1];
}
addSimple(3, 2); // 5
add(3, 2); // 5
// or in es6 style:
const addEs6 = (...numbers) => numbers.reduce((p, c) => p + c, 0);
addEs6(1, 2, 3); // 6

Technically JavaScript already had an arguments variable set on each function (except for
arrow functions), however arguments has a lot of issues, one of which is the fact that it is
not technically an array.
Rest arguments are in fact arrays. The other important difference is that rest arguments only
include arguments not specifically named in a function like so:
function print(a, b, c, ...more) {
console.log(more[0]);
console.log(arguments[0]);
}
print(1, 2, 3, 4, 5);
// 4
// 1

33

Destructuring

Destructuring
Destructuring is a way to quickly extract data out of an {} or [] without having to write
much code.
To borrow from the MDN, destructuring can be used to turn the following:
let foo = ['one', 'two', 'three'];
let one = foo[0];
let two = foo[1];
let three = foo[2];

into
let foo = ['one', 'two', 'three'];
let [one, two, three] = foo;
console.log(one); // 'one'

This is pretty interesting, but at first it might be hard to see the use case. ES6 also supports
object destructuring, which might make uses more obvious:
let myModule = {
drawSquare: function drawSquare(length) { /* implementation */ },
drawCircle: function drawCircle(radius) { /* implementation */ },
drawText: function drawText(text) { /* implementation */ },
};
let {drawSquare, drawText} = myModule;
drawSquare(5);
drawText('hello');

Destructuring can also be used for passing objects into a function, allowing you to pull
specific properties out of an object in a concise manner. It is also possible to assign default
values to destructured arguments, which can be a useful pattern if passing in a configuration
object.

34

Destructuring

let jane = { firstName: 'Jane', lastName: 'Doe'};

let john = { firstName: 'John', lastName: 'Doe', middleName: 'Smith' }
function sayName({firstName, lastName, middleName = 'N/A'}) {
console.log(`Hello ${firstName} ${middleName} ${lastName}`)
}
sayName(jane) // -> Hello Jane N/A Doe
sayName(john) // -> Helo John Smith Doe

There are many more sophisticated things that can be done with destructuring, and the MDN
has some great examples, including nested object destructuring and dynamic destructuring
with for ... in operators".

35

Modules

ES6 Modules
ES6 introduced module support. A module in ES6 is single file that allows code and data to
be isolated, it helps in organizing and grouping code logically. In other languages it's called a
package or library.
All code and data inside the module has file scope, what this means is they are not
accessible from code outside the module. To share code or data outside a module, it needs
to be exported using the export keyword.
// File: circle.js
export const pi = 3.141592;
export const circumference = diameter => diameter * pi;

The code above uses the Arrow function for circumference , which was introduced in ES6,
and is a shortform for the following.
export function circumference(diameter) {
return diameter * pi;
}

Module Systems
Using a module on the backend(server side) is relatively straightforward, you simply make
use of the import keyword. However Web Browsers have no concept of modules or import,
they just know how to load javascript code. We need a way to bring in a javascript module to
start using it from other javascript code. This is where a module loader comes in.
We won't get into the various module systems out there, but it's worth understanding there
are various module loaders available. The popular choices out there are:
RequireJS
SystemJS
Webpack

Loading a Module From a Browser

36

Modules

Below we make use of SystemJS to load a module. The script first loads the code for the
SystemJS library, then the function call System.import is use to import(load) the app
module.
Loading ES6 modules is a little trickier. In an ES6-compliant browser you use the System
keyword to load modules asynchronously. To make our code work with current browsers,
however, we will use the SystemJS library as a polyfill:
<script src="/node_module/systemjs/dist/system.js"></script>
<script>
var promise = System.import('app')
.then(function() {
console.log('Loaded!');
})
.then(null, function(error) {
console.error('Failed to load:', error);
});
</script>

37

TypeScript

TypeScript
ES6 is the current version of JavaScript. TypeScript is a superset of ES6, which means all
ES6 features are part of TypeScript, but not all TypeScript features are part of ES6.
Consequently, TypeScript must be transpiled into ES5 to run in most browsers.
One of TypeScript's primary features is the addition of type information, hence the name.
This type information can help make JavaScript programs more predictable and easier to
reason about.
Types let developers write more explicit "contracts". In other words, things like function
signatures are more explicit.
Without TS:
function add(a, b) {
return a + b;
}
add(1, 3); // 4
add(1, '3'); // '13'

With TS:
function add(a: number, b: number) {
return a + b;
}
add(1, 3); // 4
// compiler error before JS is even produced
add(1, '3'); // '13'

38

Getting Started With TypeScript

Getting Started With TypeScript

Install the TypeScript transpiler using npm:
$ npm install -g typescript

Then use tsc to manually compile a TypeScript source file into ES5:
$ tsc test.ts
$ node test.js

Note About ES6 Examples

Our earlier ES6 class won't compile now. TypeScript is more demanding than ES6 and it
expects instance properties to be declared:
class Pizza {
toppings: string[];
constructor(toppings: string[]) {
this.toppings = toppings;
}
}

Note that now that we've declared toppings to be an array of strings, TypeScript will
enforce this. If we try to assign a number to it, we will get an error at compilation time.
If you want to have a property that can be set to a value of any type, however, you can still
do this: just declare its type to be "any":
class Pizza {
toppings: any;
//...
}

39

Working With tsc

Working With tsc

So far tsc has been used to compile a single file. Typically programmers have a lot more
than one file to compile. Thankfully tsc can handle multiple files as arguments.
Imagine two ultra simple files/modules:
a.ts
export const A = (a) => console.log(a);

b.ts
export const B = (b) => console.log(b);

Before [email protected]:
$ tsc ./a.ts ./b.ts
a.ts(1,1): error TS1148: Cannot compile modules unless the '--module' flag is provided
.

Hmmm. What's the deal with this module flag? TypeScript has a help menu, let's take a look:
$ tsc --help | grep module
-m KIND, --module KIND Specify module code generation: 'commonjs', 'amd',
'system', 'umd' or 'es2015'
--moduleResolution Specifies module resolution strategy: 'node' (Node
.js) or 'classic' (TypeScript pre-1.6).

(TypeScript has more help than what we've shown; we filtered by grep for brevity.) There
are two help entries that reference "module", and --module is the one TypeScript was
complaining about. The description explains that TypeScript supports a number of different
module schemes. For the moment commonjs is desirable. This will produce modules that
are compatible with node.js's module system.
$ tsc -m commonjs ./a.ts ./b.ts

Since [email protected], tsc has a default rule for --module option: target === 'ES6' ?
'ES6' : 'commonjs' (more details can be found here), so we can simply run:

40

Working With tsc

$ tsc ./a.ts ./b.ts

tsc should produce no output. In many command line traditions, no output is actually a

mark of success. Listing the directory contents will confirm that our TypeScript files did in
fact compile.
$ ls
a.js a.ts b.js b.ts

Excellent - there are now two JavaScript modules ready for consumption.
Telling the tsc command what to compile becomes tedious and labor intensive even on
small projects. Fortunately TypeScript has a means of simplifying this. tsconfig.json files
let programmers write down all the compiler settings they want. When tsc is run, it looks
for tsconfig.json files and uses their rules to compile JavaScript.
For Angular 2 projects there are a number of specific settings that need to be configured in a
project's tsconfig.json
{
"compilerOptions": {
"module": "commonjs",
"target": "es5",
"emitDecoratorMetadata": true,
"experimentalDecorators": true,
"noImplicitAny": false,
"removeComments": false,
"sourceMap": true
},
"exclude": [
"node_modules",
"dist/"
]
}

Target
The compilation target. TypeScript supports targeting different platforms depending on your
needs. In our case, we're targeting modern browsers which support ES5.

Module

41

Working With tsc

The target module resolution interface. We're integrating TypeScript through webpack which
supports different interfaces. We've decided to use node's module resolution interface,
commonjs .

Decorators
Decorator support in TypeScript hasn't been finalized yet but since Angular 2 uses
decorators extensively, these need to be set to true. Decorators have not been introduced
yet, and will be covered later in this section.

TypeScript with Webpack

We won't be running tsc manually, however. Instead, webpack's ts-loader will do the
transpilation during the build:
// webpack.config.js
//...
rules: [
{ test: /\.ts$/, loader: 'ts', exclude: /node_modules/ },
//...
]

This loader calls tsc for us, and it will use our tsconfig.json .

42

Typings

Typings
Astute readers might be wondering what happens when TypeScript programmers need to
interface with JavaScript modules that have no type information. TypeScript recognizes files
labelled *.d.ts as definition files. These files are meant to use TypeScript to describe
interfaces presented by JavaScript libraries.
There are communities of people dedicated to creating typings for JavaScript projects. There
is also a utility called typings ( npm install --save-dev typings ) that can be used to
manage third party typings from a variety of sources. (Deprecated in TypeScript 2.0)
In TypeScript 2.0, users can get type files directly from @types through npm (for example,
npm install --save @types/lodash will install lodash type file).

43

Linting

Linting
Many editors support the concept of "linting" - a grammar check for computer programs.
Linting can be done in a programmer's editor and/or through automation.
For TypeScript there is a package called tslint , ( npm install --save-dev tslint ) which
can be plugged into many editors. tslint can also be configured with a tslint.json file.
Webpack can run tslint before it attempts to run tsc . This is done by installing tslintloader ( npm install --save-dev tslint-loader ) which plugs into webpack like so:

// ...
module: {
preLoaders: [
{ test: /\.ts$/, loader: 'tslint' }
],
loaders: [
{ test: /\.ts$/, loader: 'ts', exclude: /node_modules/ },
// ...
]
// ...
}

44

TypeScript Features

TypeScript Features
Now that producing JavaScript from TypeScript code has been de-mystified, some of its
features can be described and experimented with.
Types
Interfaces
Shapes
Decorators

Types
Many people do not realize it, but JavaScript does in fact have types, they're just "duck
typed", which roughly means that the programmer does not have to think about them.
JavaScript's types also exist in TypeScript:
boolean (true/false)
number integers, floats, Infinity and NaN
string characters and strings of characters
[] Arrays of other types, like number[] or boolean[]
{} Object literal
undefined not set

TypeScript also adds

enum enumerations like { Red, Blue, Green }
any use any type
void nothing

Primitive type example:

45

TypeScript Features

let isDone: boolean = false;

let height: number = 6;
let name: string = "bob";
let list: number[] = [1, 2, 3];
let list: Array<number> = [1, 2, 3];
enum Color {Red, Green, Blue};
let c: Color = Color.Green;
let notSure: any = 4;
notSure = "maybe a string instead";
notSure = false; // okay, definitely a boolean
function showMessage(data: string): void {
alert(data);
}
showMessage('hello');

This illustrates the primitive types in TypeScript, and ends by illustrating a showMessage
function. In this function the parameters have specific types that are checked when tsc is
run.
In many JavaScript functions it's quite common for functions to take optional parameters.
TypeScript provides support for this, like so:
function logMessage(message: string, isDebug?: boolean) {
if (isDebug) {
console.log('Debug: ' + message);
} else {
console.log(message);
}
}
logMessage('hi'); // 'hi'
logMessage('test', true); // 'Debug: test'

Using a ? lets tsc know that isDebug is an optional parameter. tsc will not complain if
isDebug is omitted.

46

TypeScript Classes

TypeScript Classes
TypeScript also treats class es as their own type:
class Foo { foo: number; }
class Bar { bar: string; }
class Baz {
constructor(foo: Foo, bar: Bar) { }
}
let baz = new Baz(new Foo(), new Bar()); // valid
baz = new Baz(new Bar(), new Foo()); // tsc errors

Like function parameters, class es sometimes have optional members. The same ?:
syntax can be used on a class definition:
class Person {
name: string;
nickName?: string;
}

In the above example, an instance of Person is guaranteed to have a name , and might
optionally have a nickName

47

Interfaces

Interfaces
An interface is a TypeScript artifact, it is not part of ECMAScript. A interface is a way to
define a contract on a function with respect to the arguments and their type. Along with
functions, an interface can also be used with a Class as well to define custom types.
An interface is an abstract type, it does not contain any code as a class does. It only defines
the 'signature' or shape of an API. During transpilation, an interface will not generate any
code, it is only used by Typescript for type checking during development.
Here is an example of an interface describing a function API:
interface Callback {
(error: Error, data: any): void;
}
function callServer(callback: Callback) {
callback(null, 'hi');
}
callServer((error, data) => console.log(data)); // 'hi'
callServer('hi'); // tsc error

Sometimes JavaScript functions can accept multiple types as well as varying arguments,
that is, they can have different call signatures. Interfaces can be used to specify this.
interface PrintOutput {
(message: string): void; // common case
(message: string[]): void; // less common case
}
let printOut: PrintOutput = (message) => {
if (Array.isArray(message)) {
console.log(message.join(', '));
} else {
console.log(message);
}
}
printOut('hello'); // 'hello'
printOut(['hi', 'bye']); // 'hi, bye'

Here is an example of an interface describing an object literal:

48

Interfaces

interface Action {
type: string;
}
let a: Action = {
type: 'literal'
}

49

Shapes

Shapes
Underneath TypeScript is JavaScript, and underneath JavaScript is typically a JIT (Just-InTime compiler). Given JavaScript's underlying semantics, types are typically reasoned about
by "shapes". These underlying shapes work like TypeScript's interfaces, and are in fact how
TypeScript compares custom types like class es and interface s.
Consider an expansion of the previous example:
interface Action {
type: string;
}
let a: Action = {
type: 'literal'
}
class NotAnAction {
type: string;
constructor() {
this.type = 'Constructor function (class)';
}
}
a = new NotAnAction(); // valid TypeScript!

Despite the fact that Action and NotAnAction have different identifiers, tsc lets us assign
an instance of NotAnAction to a which has a type of Action . This is because TypeScript
only really cares that objects have the same shape. In other words if two objects have the
same attributes, with the same typings, those two objects are considered to be of the same
type.

50

Type Inference

Type Inference
One common misconception about TypeScript's types is that code needs to explicitly
describe types at every possible opportunity. Fortunately this is not the case. TypeScript has
a rich type inference system that will "fill in the blanks" for the programmer. Consider the
following:
type-inference-finds-error.ts
let numbers = [2, 3, 5, 7, 11];
numbers = ['this will generate a type error'];

tsc ./type-inference-finds-error.ts
type-inference-finds-error.ts(2,1): error TS2322: Type 'string[]' is not assignable to
type 'number[]'.
Type 'string' is not assignable to type 'number'.

The code contains no extra type information. In fact, it's valid ES6.
If var had been used, it would be valid ES5. Yet TypeScript is still able to determine type
information.
Type inference can also work through context, which is handy with callbacks. Consider the
following:
type-inference-finds-error-2.ts
interface FakeEvent {
type: string;
}
interface FakeEventHandler {
(e: FakeEvent): void;
}
class FakeWindow {
onMouseDown: FakeEventHandler
}
const fakeWindow = new FakeWindow();
fakeWindow.onMouseDown = (a: number) => {
// this will fail
};

51

Type Inference

tsc ./type-inference-finds-error-2.ts
type-inference-finds-error-2.ts(14,1): error TS2322: Type '(a: number) => void' is not
assignable to type 'FakeEventHandler'.
Types of parameters 'a' and 'e' are incompatible.
Type 'number' is not assignable to type 'FakeEvent'.
Property 'type' is missing in type 'Number'.

In this example the context is not obvious since the interfaces have been defined explicitly.
In a browser environment with a real window object, this would be a handy feature,
especially the type completion of the Event object.

52

Decorators

Decorators
Decorators are proposed for a future version of JavaScript, but the Angular 2 team really
wanted to use them, and they have been included in TypeScript.
Decorators are functions that are invoked with a prefixed @ symbol, and immediately
followed by a class , parameter, method or property. The decorator function is supplied
information about the class , parameter or method, and the decorator function returns
something in its place, or manipulates its target in some way. Typically the "something" a
decorator returns is the same thing that was passed in, but it has been augmented in some
way.
Decorators are quite new in TypeScript, and most use cases demonstrate the use of existing
decorators. However, decorators are just functions, and are easier to reason about after
walking through a few examples.
Decorators are functions, and there are four things ( class , parameter, method and
property) that can be decorated; consequently there are four different function signatures for
decorators:
class: declare type ClassDecorator = <TFunction extends Function>(target: TFunction)
=> TFunction | void;

property: declare type PropertyDecorator = (target: Object, propertyKey: string |

symbol) => void;

method: declare type MethodDecorator = <T>(target: Object, propertyKey: string |

symbol, descriptor: TypedPropertyDescriptor<T>) => TypedPropertyDescriptor<T> | void;

parameter: declare type ParameterDecorator = (target: Object, propertyKey: string |

symbol, parameterIndex: number) => void;

Readers who have played with Angular 2 will notice that these signatures do not look like the
signatures used by Angular 2 specific decorators like @Component() .
Notice the () on @Component . This means that the @Component is called once JavaScript
encounters @Component() . In turn, this means that there must be a Component function
somewhere that returns a function matching one of the decorator signatures outlined above.
This is an example of the decorator factory pattern.
If decorators still look confusing, perhaps some examples will clear things up.

53

Property Decorators

Property Decorators
Property decorators work with properties of classes.
function Override(label: string) {
return function (target: any, key: string) {
Object.defineProperty(target, key, {
configurable: false,
get: () => label
});
}
}
class Test {
@Override('test') // invokes Override, which returns the decorator
name: string = 'pat';
}
let t = new Test();
console.log(t.name); // 'test'

The above example must be compiled with both the --experimentalDecorators and -emitDecoratorMetadata flags.

In this case the decorated property is replaced by the label passed to the decorator. It's
important to note that property values cannot be directly manipulated by the decorator;
instead an accessor is used.
Here's a classic property example that uses a plain decorator
function ReadOnly(target: any, key: string) {
Object.defineProperty(target, key, { writable: false });
}
class Test {
@ReadOnly // notice there are no `()`
name: string;
}
const t = new Test();
t.name = 'jan';
console.log(t.name); // 'undefined'

In this case the name property is not writable , and remains undefined.

54

Property Decorators

55

Class Decorators

Class Decorators
function log(prefix?: string) {
return (target) => {
// save a reference to the original constructor
var original = target;
// a utility function to generate instances of a class
function construct(constructor, args) {
var c: any = function () {
return constructor.apply(this, args);
}
c.prototype = constructor.prototype;
return new c();
}
// the new constructor behavior
var f: any = function (...args) {
console.log(prefix + original.name);
return construct(original, args);
}
// copy prototype so instanceof operator still works
f.prototype = original.prototype;
// return new constructor (will override original)
return f;
};
}
@log('hello')
class World {
}
const w = new World(); // outputs "helloWorld"

In the example log is invoked using @ , and passed a string as a parameter, @log()
returns an anonymous function that is the actual decorator.
The decorator function takes a class , or constructor function (ES5) as an argument. The
decorator function then returns a new class construction function that is used whenever
World is instantiated.

This decorator does nothing other than log out its given parameter, and its target 's class
name to the console.

56

Class Decorators

57

Parameter Decorators

Parameter Decorators
function logPosition(target: any, propertyKey: string, parameterIndex: number) {
console.log(parameterIndex);
}
class Cow {
say(b: string, @logPosition c: boolean) {
console.log(b);
}
}
new Cow().say('hello', false); // outputs 1 (newline) hello

The above demonstrates decorating method parameters. Readers familiar with Angular 2
can now imagine how Angular 2 implemented their @Inject() system.

58

The JavaScript Toolchain

The JavaScript Toolchain

In this section, we'll describe the tools that you'll be using for the rest of the course.

Figure: Hand Tools by M338 is licensed under Public Domain

(http://commons.wikimedia.org/wiki/File:Hand_tools.jpg)

59

Source Control: git

Source Control: Git

A source control, sometimes called a version control brings change management to saving
files at different points in the development process. A Version control system (VCS) that
will we make use of is Git.
Git is a decentralized distributed versioning system, it allows programmers to collaborate on
the same codebase without stepping on each other's toes. It has become the de-facto
source control system for open source development because of its decentralized model and
cheap branching features.
For more information on how to use Git, head over to Pro Git

60

The Command Line

The Command Line

JavaScript development tools are very command line oriented. If you come from a Windows
background you may find this unfamiliar. However the command line provides better support
for automating development tasks, so it's worth getting comfortable with it.
We will provide examples for all command line activities required by this course.

61

Command Line JavaScript: NodeJS

Command Line JavaScript: NodeJS

Node.js is a JavaScript runtime environment that allows JavaScript code to run outside of a
browser using Google V8 JavaScript engine. Node.js is used for writting fast executing code
on the server to handle events and non-blocking I/O efficently.
REPL (Read-Eval-Print-Loop) to quickly write and test JavaScript code.
The V8 JavaScript interpreter.
Modules for doing OS tasks like file I/O, HTTP, etc.
While Node.js was initially intended for writing server code in JavaScript, today it is widely
used by JavaScript tools, which makes it relevant to front-end programmers too. A lot of the
tools you'll be using in this course leverage Node.js.

62

Back-End Code Sharing and Distribution: npm

Back-End Code Sharing and Distribution:

npm
npm is the "node package manager". It installs with NodeJS, and gives you access to a

wide variety of 3rd-party JavaScript modules.

It also performs dependency management for your back-end application. You specify
module dependencies in a file called package.json ; running npm install will resolve,
download and install your back-end application's dependencies.

63

Module Loading, Bundling and Build Tasks: Webpack

Module Loading, Bundling and Build

Tasks: Webpack
Webpack is a JavaScript module bundler. It takes modules with their dependencies and
generates static assets representing those modules. Webpack known only how to bundle
JavaScript. To bundle other assets likes CSS, HTML, images or just about anything it uses
additional loaders. Webpack can also be extended via plugins, for example minification and
mangling can be done using the UglifyJS plugin for webpack.

64

Chrome

Web Browsers
We use Google's Chrome browser for this course because of its cutting-edge JavaScript
engine and excellent debugging tools.
However you are free to use other browsers. Not well known, there is a Mozilla Firefox
Developer Edition available with support for great development and debugging tools. Code
written with JavaScript should work on any modern web browser (Firefox, IE9+, Chrome,
Safari, Opera).

65

Bootstrapping an Angular 2 Application

Bootstrapping an Angular 2 Application

Bootstrapping is an essential process in Angular - it is where the application is loaded when
Angular comes to life.
Bootstrapping Angular 2 applications is certainly different from Angular 1.x, but is still a
straightforward procedure. Let's take a look at how this is done.

66

Understanding the File Structure

Understanding the File Structure

To get started let's create a bare-bones Angular 2 application with a single component. To do
this we need the following files:
app/app.component.ts - this is where we define our root component
app/app.module.ts - the entry Angular Module to be bootstrapped
index.html - this is the page the component will be rendered in
app/index.ts - is the glue that combines the component and page together
app/app.component.ts
import {Component} from '@angular/core'
@Component({
selector: 'app',
template: '<b>Bootstrapping an Angular 2 Application</b>'
})
export class MyApp {}

index.html
...
<body>
<app>Loading...</app>
</body>
...

app/app.module.ts
import { BrowserModule } from '@angular/platform-browser';
import { NgModule } '@angular/core';
import { MyApp } from './app.component'
@NgModule({
imports: [BrowserModule],
declarations: [MyApp],
bootstrap: [MyApp]
})
export class MyAppModule {
}

67

Understanding the File Structure

app/main.ts
import { platformBrowserDynamic } from '@angular/platform-browser-dynamic';
import { MyAppModule } from './app.module';
platformBrowserDynamic().bootstrapModule(MyAppModule);

If you're making use of Ahead-of-Time (AoT) compilation, you would code main.ts as
follows.
import { platformBrowser} from '@angular/platform-browser';
import { MyAppModuleNgFactory } from '../aot/app/app.module.ngfactory';
platformBrowser().bootstrapModuleFactory(MyAppModuleNgFactory);

View Example
The bootstrap process loads main.ts which is the main entry point of the application. The
MyAppModule operates as the root module of our application. The module is configured to

use MyApp as the component to bootstrap, and will be rendered on any app HTML
element encountered.
There is an app HTML element in the index.html file, and we use app/index.ts to import the
MyAppModule component and the platformBrowserDynamic().bootstrapModule function and

kickstart the process. As shown above, you may optionally use AoT in which case you will
be working with Factories, in the example, MyAppModuleNgFactory and
bootstrapModuleFactory .

Why does Angular 2 bootstrap itself in this way? Well there is actually a very good reason.
Since Angular 2 is not a web-only based framework, we can write components that will run
in NativeScript, or Cordova, or any other environment that can host Angular 2 applications.
The magic is then in our bootstrapping process - we can import which platform we would like
to use, depending on the environment we're operating under. In our example, since we were
running our Angular 2 application in the browser, we used the bootstrapping process found
in @angular/platform-browser-dynamic .
It's also a good idea to leave the bootstrapping process in its own separate index.ts file. This
makes it easier to test (since the components are isolated from the bootstrap call), easier
to reuse and gives better organization and structure to our application.
There is more to understanding Angular Modules and @NgModule which will be covered
later, but for now this is enough to get started.

68

Understanding the File Structure

69

Bootstrapping Providers

Bootstrapping Providers
The bootstrap process also starts the dependency injection system in Angular 2. We won't
go over Angular 2's dependency injection system here - that is covered later. Instead let's
take a look at an example of how to bootstrap your application with application-wide
providers.
For this, we will register a service called Greeter with the providers property of the
module we are using to bootstrap the application.
app/app.module.ts
import { BrowserModule } from '@angular/platform-browser';
import { NgModule } '@angular/core';
import { MyApp } from './app.component'
import { Greeter } from './greeter.service';
@NgModule({
imports: [BrowserModule],
providers: [Greeter],
declarations: [MyApp],
bootstrap: [MyApp]
})
export class MyAppModule {
}

View Example

70

Components in Angular 2

Components in Angular 2

Figure: components

The core concept of any Angular 2 application is the component. In effect, the whole
application can be modeled as a tree of these components.
This is how the Angular 2 team defines a component:
A component controls a patch of screen real estate that we could call a view, and
declares reusable UI building blocks for an application.
Basically, a component is anything that is visible to the end user and which can be reused
many times within an application.
In Angular 1.x we had router views and directives which worked sort of like components. The
idea of directive components became quite popular. They were created by using directive
with a controller while relying on the controllerAs and bindToController properties. For
example:

71

Components in Angular 2

angular.module('ngcourse')
.directive('ngcHelloComponent', () => ({
restrict: 'E',
scope: { name: '=' },
template: '<span>Hello, {{ ctrl.name }}.</span>',
controller: MyComponentCtrl,
controllerAs: 'ctrl',
bindToController: true
})
);

In fact, this concept became so popular that in Angular 1.5 the .component method was
introduced as syntactic sugar.
angular.module('ngcourse')
.component('ngcHelloComponent', {
bindings: { name: '=' },
template: '<span>Hello, {{ $ctrl.name }}.</span>',
controller: MyComponentCtrl
});

72

Creating Components

Creating Components
Components in Angular 2 build upon the lessons learned from Angular 1.5. We define a
component's application logic inside a class. To this we attach @Component , a TypeScript
decorator , which allows you to modify a class or function definition and adds metadata to

properties and function arguments.

selector is the element property that we use to tell Angular to create and insert an
instance of this component.
template is a form of HTML that tells Angular what needs to be to rendered in the DOM.
The Component below will interpolate the value of name variable into the template between
the double braces {{name}} , what get rendered in the view is <p>Hello World</p> .
import {Component} from '@angular/core';
@Component({
selector: 'hello',
template: '<p>Hello, {{name}}</p>'
})
export class Hello {
name: string;
constructor() {
this.name = 'World';
}
}

We need to import the Component decarator from @angular/core before we can make use
of it. To use this component we simply add <hello></hello> to the HTML file or another
template, and Angular will insert an instance of the Hello view between those tags.
View Example

73

Application Structure with Components

Application Structure with Components

A useful way of conceptualizing Angular application design is to look at it as a tree of nested
components, each having an isolated scope.
For example consider the following:
<TodoApp>
<TodoList>
<TodoItem></TodoItem>
<TodoItem></TodoItem>
<TodoItem></TodoItem>
</TodoList>
<TodoForm></TodoForm>
</TodoApp>

At the root we have TodoApp which consists of a TodoList and a TodoForm . Within the list
we have several TodoItem s. Each of these components is visible to the user, who can
interact with these components and perform actions.

74

Passing Data into a Component

Passing Data into a Component

There are two ways to pass data into a component, with 'property binding' and 'event
binding'. In Angular 2, data and event change detection happens top-down from parent to
children. However for Angular 2 events we can use the DOM event mental model where
events flow bottom-up from child to parent. So, Angular 2 events can be treated like regular
HTML DOM based events when it comes to cancellable event propagation.
The @Input() decarator defines a set of parameters that can be passed down from the
component's parent. For example, we can modify the Hello component so that name can
be provided by the parent.
import {Component, Input} from '@angular/core';
@Component({
selector: 'hello',
template: '<p>Hello, {{name}}</p>'
})
export class Hello {
@Input() name: string;
}

The point of making components is not only encapsulation, but also reusability. Inputs allow
us to configure a particular instance of a component.
We can now use our component like so:
<!-- To bind to a raw string -->
<hello name="World"></hello>
<!-- To bind to a variable in the parent scope -->
<hello [name]="name"></hello>

View Example
Unlike Angular 1.x, this is one-way binding.

75

Responding to Component Events

Responding to Component Events

An event handler is specified inside the template using round brackets to denote event
binding. This event handler is then coded in the class to process the event.
import {Component} from '@angular/core';
@Component({
selector: 'counter',
template: `
<div>
<p>Count: {{ num }}</p>
<button (click)="increment()">Increment</button>
</div>
`
})
export class Counter {
num: number = 0;
increment() {
this.num++;
}
}

View Example
To send data out of components via outputs, start by defining the outputs attribute. It accepts
a list of output parameters that a component exposes to its parent.

76

Responding to Component Events

import {Component, EventEmitter, Input, Output} from '@angular/core';

@Component({
selector: 'counter',
template: `
<div>
<p>Count: {{ count }}</p>
<button (click)="increment()">Increment</button>
</div>
`
})
export class Counter {
@Input() count: number = 0;
@Output() result: EventEmitter = new EventEmitter();
increment() {
this.count++;
this.result.emit(this.count);
}
}

import {Component} from '@angular/core';

@Component({
selector: 'app',
template: `
<div>
Parent Num: {{ num }}<br />
Parent Count: {{ parentCount }}
<counter [count]="num" (result)="onChange($event)">
</counter>
</div>
`
})
export class App {
num: number;
parentCount: number;
constructor() {
this.num = 0;
this.parentcount = 0;
}
onChange(val: number) {
this.parentCount = val;
}
}

View Example

77

Responding to Component Events

Together a set of input + output bindings define the public API of your component. In our
templates we use the [squareBrackets] to pass inputs and the (parenthesis) to handle
outputs.

78

Using Two-Way Data Binding

Using Two-Way Data Binding

Two-way data binding combines the input and output binding into a single notation using the
ngModel directive.

<input [(ngModel)]="name" >

What this is doing behind the scenes is equivalent to:

<input [ngModel]="name" (ngModelChange)="name=$event">

To create your own component that supports two-way binding, you must define an @Output
property to match an @Input , but suffix it with the Change . The code example below, inside
class Counter shows how to make property count support two-way binding.
import {Component, Input, Output, EventEmitter} from '@angular/core';
@Component({
selector: 'counter',
template: `
<div>
<p>
<ng-content></ng-content>
Count: {{ count }} <button (click)="increment()">Increment</button>
</p>
</div>
`
})
export class Counter {
@Input() count: number = 0;
@Output() countChange: EventEmitter<number> = new EventEmitter<number>();
increment() {
this.count++;
this.countChange.emit(this.count);
}
}

View Example

79

Using Two-Way Data Binding

80

Accessing Child Components from Template

Access Child Components From the

Template
In our templates, we may find ourselves needing to access values provided by the child
components which we use to build our own component.
The most straightforward examples of this may be seen dealing with forms or inputs:
app/my-example.component.html
<section >
<form #myForm="ngForm" (ngSubmit)="submitForm(myForm)">
<label for="name">Name</label>
<input type="text" name="name" id="name" ngModel>
<button type="submit">Submit</button>
</form>
Form value: {{formValue}}
</section>

app/my-example.component.ts
import {Component} from '@angular/core';
@Component({
selector: 'my-example',
templateUrl: 'app/my-example.component.html'
})
export class MyExampleComponent {
formValue: string = JSON.stringify({});
submitForm (form: NgForm) {
this.formValue = JSON.stringify(form.value);
}
}

View Example
This isn't a magic feature which only forms or inputs have, but rather a way of referencing
the instance of a child component in your template. With that reference, you can then access
public properties and methods on that component.
<my-profile #profile></my-profile>
My name is {{ profile.name }}

81

Accessing Child Components from Template

@Component({
selector: 'my-profile',
templateUrl: 'app/my-profile.component.html'
})
export class MyProfile {
name: string = 'John Doe';
}

View Example
There are other means of accessing and interfacing with child components, but if you simply
need to reference properties or methods of a child, this can be a simple and straightforward
method of doing so.

82

Projection

Projection
Projection is a very important concept in Angular 2. It enables developers to build reusable
components and make applications more scalable and flexible. To illustrate that, suppose we
have a ChildComponent like:
@Component({
selector: 'child',
template: `
<div>
<h4>Child Component</h4>
{{ childContent }}
</div>
`
})
export class ChildComponent {
childContent: string = "Default content";
}

What should we do if we want to replace {{ childContent }} to any HTML that provided to

ChildComponent ? One tempting idea is to define an @Input containing the text, but what if

you wanted to provide styled HTML, or other components? Trying to handle this with an
@Input can get messy quickly, and this is where content projection comes in. Components

by default support projection, and you can use the ngContent directive to place the
projected content in your template.
So, change ChildComponent to use projection:
import {Component, Input} from '@angular/core';
@Component({
selector: 'child',
template: `
<h4>Child Component</h4>
<ng-content></ng-content>
`
})
class ChildComponent {}

Then, when we use ChildComponent in the template:

83

Projection

<child>
<p>My projected content.</p>
</child>

This is telling Angular, that for any markup that appears between the opening and closing
tag of <child> , to place inside of <ng-content></ng-content> .
When doing this, we can have other components, markup, etc projected here and the
ChildComponent does not need to know about or care what is being provided.

View Example
But what if we have multiple <ng-content></ng-content> and want to specify the position of
the projected content to certain ng-content ? For example, for the previous
ChildComponent , if we want to format the projected content into an extra header and
footer section:

import {Component, Input} from '@angular/core';

@Component({
selector: 'child',
template: `
<h4>Child Component</h4>
<ng-content select="header"></ng-content>
<ng-content></ng-content>
<ng-content select="footer"></ng-content>
`
})
class ChildComponent {}

Then in the template, we can use directives, say, <header> to specify the position of
projected content to the ng-content with select="header" :
<div>
<child>
<header>Header Content</header>
Main Content
<footer>Footer Content</footer>
</child>
</div>

Besides using directives, developers can also select a ng-content through css class:
<ng-content select=".class-select"></ng-content>

84

Projection

<div class="class-select">
div with .class-select
</div>

View Example

85

Structuring Applications with Components

Structuring Applications with Components

As the complexity and size of our application grows, we want to divide responsibilities
among our components further.
Smart / Container components are application-specific, higher-level, container
components, with access to the application's domain model.
Dumb / Presentational components are components responsible for UI rendering and/or
behavior of specific entities passed in via components API (i.e component properties
and events). Those components are more in-line with the upcoming Web Component
standards.

86

Using Other Components

Using Other Components

Components depend on other components, directives and pipes. For example, TodoList
relies on TodoItem . To let a component know about these dependencies we group them into
a module.
import {NgModule} from '@angular/core';
import {TodoInput} from './components/todo-input';
import {TodoItem} from './components/todo-item';
import {TodoList} from './components/todo-list';
@NgModule({
imports: [ ... ],
declarations: [ TodoList, TodoItem, TodoInput ],
bootstrap: [ ... ]
})
export class ToDoAppModule { }

The property declarations expects an array of components, directives and pipes that are
part of the module.
Please see the Modules section for more info about NgModule .

87

Directives

Directives
A Directive modifies the DOM to change apperance, behavior or layout of DOM elements.
Directives are one of the core building blocks Angular 2 uses to build applications. In fact,
Angular 2 components are in large part directives with templates.
From an Angular 1 perspective, Angular 2 components have assumed a lot of the roles
directives used to. The majority of issues that involve templates and dependency injection
rules will be done through components, and issues that involve modifying generic behaviour
is done through directives.
There are three main types of directives in Angular 2:
Component - directive with a template.
Attribute directives - directives that change the behavior of a component or element but
don't affect the template
Structural directives - directives that change the behavior of a component or element by
affecting how the template is rendered

88

Attribute Directives

Attribute Directives
Attribute directives are a way of changing the appearance or behavior of a component or a
native DOM element. Ideally, a directive should work in a way that is component agnostic
and not bound to implementation details.
For example, Angular 2 has built-in attribute directives such as ngClass and ngStyle that
work on any component or element.

89

NgStyle Directive

NgStyle Directive
Angular 2 provides a built-in directive, ngStyle , to modify a component or element's style
attribute. Here's an example:
@Component({
selector: 'style-example',
template: `
<p style="padding: 1rem"
[ngStyle]="{
color: 'red',
'font-weight': 'bold',
borderBottom: borderStyle
}">
<ng-content></ng-content>
</p>
`
})
export class StyleExampleComponent {
borderStyle: string = '1px solid black';
}

View Example
Notice that binding a directive works the exact same way as component attribute bindings.
Here, we're binding an expression, an object literal, to the ngStyle directive so the directive
name must be enclosed in square brackets. ngStyle accepts an object whose properties
and values define that element's style. In this case, we can see that both kebab case and
lower camel case can be used when specifying a style property. Also notice that both the
html style attribute and Angular 2 ngStyle directive are combined when styling the
element.

90

NgClass Directive

NgClass Directive
The ngClass directive changes the class attribute that is bound to the component or
element it's attached to. There are a few different ways of using the directive.

Binding a string
We can bind a string directly to the attribute. This works just like adding an html class
attribute.
@Component({
selector: 'class-as-string',
template: `
<p ngClass="centered-text underlined" class="orange">
<ng-content></ng-content>
</p>
`,
styles: [`
.centered-text {
text-align: center;
}
.underlined {
border-bottom: 1px solid #ccc;
}
.orange {
color: orange;
}
`]
})
export class ClassAsStringComponent {
}

View Example
In this case, we're binding a string directly so we avoid wrapping the directive in square
brackets. Also notice that the ngClass works with the class attribute to combine the final
classes.

Binding an array

91

NgClass Directive

@Component({
selector: 'class-as-array',
template: `
<p [ngClass]="['warning', 'big']">
<ng-content></ng-content>
</p>
`,
styles: [`
.warning {
color: red;
font-weight: bold;
}
.big {
font-size: 1.2rem;
}
`]
})
export class ClassAsArrayComponent {
}

View Example
Here, since we are binding to the ngClass directive by using an expression, we need to
wrap the directive name in square brackets. Passing in an array is useful when you want to
have a function put together the list of applicable class names.

Binding an object
Lastly, an object can be bound to the directive. Angular 2 applies each property name of that
object to the component if that property is true.

92

NgClass Directive

@Component({
selector: 'class-as-object',
template: `
<p [ngClass]="{ card: true, dark: false, flat: flat }">
<ng-content></ng-content>
<br/>
<button type="button" (click)="flat=!flat">Toggle Flat</button>
</p>
`,
styles: [`
.card {
border: 1px solid #eee;
padding: 1rem;
margin: 0.4rem;
font-family: sans-serif;
box-shadow: 2px 2px 2px #888888;
}
.dark {
background-color: #444;
border-color: #000;
color: #fff;
}
.flat {
box-shadow: none;
}
`]
})
export class ClassAsObjectComponent {
flat: boolean = true;
}

View Example
Here we can see that since the object's card and flat properties are true, those classes
are applied but since dark is false, it's not applied.

93

Structural Directives

Structural Directives
Structural Directives are a way of handling how a component or element renders through the
use of the template tag. This allows us to run some code that decides what the final
rendered output will be. Angular 2 has a few built-in structural directives such as ngIf ,
ngFor , and ngSwitch .

Note: For those who are unfamiliar with the template tag, it is an HTML element with a few
special properties. Content nested in a template tag is not rendered on page load and is
something that is meant to be loaded through code at runtime. For more information on the
template tag, visit the MDN documentation.

Structural directives have their own special syntax in the template that works as syntactic
sugar.
@Component({
selector: 'directive-example',
template: `
<p *structuralDirective="expression">
Under a structural directive.
</p>
`
})

Instead of being enclosed by square brackets, our dummy structural directive is prefixed with
an asterisk. Notice that the binding is still an expression binding even though there are no
square brackets. That's due to the fact that it's syntactic sugar that allows using the directive
in a more intuitive way and similar to how directives were used in Angular 1. The component
template above is equivalent to the following:
@Component({
selector: 'directive-example',
template: `
<template [structuralDirective]="expression">
<p>
Under a structural directive.
</p>
</template>
`
})

Here, we see what was mentioned earlier when we said that structural directives use the
template tag. Angular 2 also has a built-in template directive that does the same thing:

94

Structural Directives

@Component({
selector: 'directive-example',
template: `
<p template="structuralDirective expression">
Under a structural directive.
</p>
`
})

95

NgIf Directive

NgIf Directive
The ngIf directive conditionally adds or removes content from the DOM based on whether
or not an expression is true or false.
Here's our app component, where we bind the ngIf directive to an example component.
@Component({
selector: 'app',
template: `
<button type="button" (click)="toggleExists()">Toggle Component</button>
<hr/>
<if-example *ngIf="exists">
Hello
</if-example>
`
})
export class AppComponent {
exists: boolean = true;
toggleExists() {
this.exists = !this.exists;
}
}

View Example
Clicking the button will toggle whether or not IfExampleComponent is a part of the DOM and
not just whether it is visible or not. This means that every time the button is clicked,
IfExampleComponent will be created or destroyed. This can be an issue with components that

have expensive create/destroy actions. For example, a component could have a large child
subtree or make several HTTP calls when constructed. In these cases it may be better to
avoid using ngIf if possible.

96

NgFor Directive

NgFor Directive
The ngFor directive is a way of repeating a template by using each item of an iterable as
that template's context.
@Component({
selector: 'app',
template: `
<for-example *ngFor="let episode of episodes" [episode]="episode">
{{episode.title}}
</for-example>
`
})
export class AppComponent {
episodes: any[] = [
{ title: 'Winter Is Coming', director: 'Tim Van Patten' },
{ title: 'The Kingsroad', director: 'Tim Van Patten' },
{ title: 'Lord Snow', director: 'Brian Kirk' },
{ title: 'Cripples, Bastards, and Broken Things', director: 'Brian Kirk' },
{ title: 'The Wolf and the Lion', director: 'Brian Kirk' },
{ title: 'A Golden Crown', director: 'Daniel Minahan' },
{ title: 'You Win or You Die', director: 'Daniel Minahan' },
{ title: 'The Pointy End', director: 'Daniel Minahan' }
];
}

View Example
The ngFor directive has a different syntax from other directives we've seen. If you're
familiar with the for...of statement, you'll notice that they're almost identical. ngFor lets you
specify an iterable object to iterate over and the name to refer to each item by inside the
scope. In our example, you can see that episode is available for interpolation as well as
property binding. The directive does some extra parsing so that when this is expanded to
template form, it looks a bit different:
@Component({
selector: 'app',
template: `
<template ngFor [ngForOf]="episodes" let-episode>
<for-example [episode]="episode">
{{episode.title}}
</for-example>
</template>
`
})

97

NgFor Directive

View Example
Notice that there is an odd let-episode property on the template element. The ngFor
directive provides some variables as context within its scope. let-episode is a context
binding and here it takes on the value of each item of the iterable. ngFor also provides
some other values that can be bound to:
index - position of the current item in the iterable starting at 0
first - true if the current item is the first item in the iterable
last - true if the current item is the last item in the iterable
even - true if the current index is an even number
odd - true if the current index is an odd number
@Component({
selector: 'app',
template: `
<for-example
*ngFor="let episode of episodes; let i = index; let isOdd = odd"
[episode]="episode"
[ngClass]="{ odd: isOdd }">
{{i+1}}. {{episode.title}}
</for-example>
<hr/>
<h2>Desugared</h2>
<template ngFor [ngForOf]="episodes" let-episode let-i="index" let-isOdd="odd">
<for-example [episode]="episode" [ngClass]="{ odd: isOdd }">
{{i+1}}. {{episode.title}}
</for-example>
</template>
`
})

View Example

trackBy
Often ngFor is used to iterate through a list of objects with a unique ID field. In this case,
we can provide a trackBy function which helps Angular keep track of items in the list so
that it can detect which items have been added or removed and improve performance.
Angular 2 will try and track objects by reference to determine which items should be created
and destroyed. However, if you replace the list with a new source of objects, perhaps as a
result of an API request - we can get some extra performance by telling Angular 2 how we
98

NgFor Directive

want to keep track of things.

For example, if the Add Episode button was to make a request and return a new list of
episodes, we might not want to destroy and re-create every item in the list. If the episodes
have a unique ID, we could add a trackBy function:
@Component({
selector: 'app',
template: `
<button (click)="addOtherEpisode()" [disabled]="otherEpisodes.length === 0">Add Epis
ode</button>
<for-example
*ngFor="let episode of episodes;
let i = index; let isOdd = odd;
trackBy: trackById" [episode]="episode"
[ngClass]="{ odd: isOdd }">
{{episode.title}}
</for-example>
`
})
export class AppComponent {
otherEpisodes: any[] = [
{ title: 'Two Swords', director: 'D. B. Weiss', id: 8 },
{ title: 'The Lion and the Rose', director: 'Alex Graves', id: 9 },
{ title: 'Breaker of Chains', director: 'Michelle MacLaren', id: 10 },
{ title: 'Oathkeeper', director: 'Michelle MacLaren', id: 11 }]
episodes: any[] = [
{ title: 'Winter Is Coming', director: 'Tim Van Patten', id: 0 },
{ title: 'The Kingsroad', director: 'Tim Van Patten', id: 1 },
{ title: 'Lord Snow', director: 'Brian Kirk', id: 2 },
{ title: 'Cripples, Bastards, and Broken Things', director: 'Brian Kirk', id: 3 },
{ title: 'The Wolf and the Lion', director: 'Brian Kirk', id: 4 },
{ title: 'A Golden Crown', director: 'Daniel Minahan', id: 5 },
{ title: 'You Win or You Die', director: 'Daniel Minahan', id: 6 }
{ title: 'The Pointy End', director: 'Daniel Minahan', id: 7 }
];
addOtherEpisode() {
// We want to create a new object reference for sake of example
let episodesCopy = JSON.parse(JSON.stringify(this.episodes))
this.episodes=[...episodesCopy,this.otherEpisodes.pop()];
}
trackById(index: number, episode: any): number {
return episode.id;
}
}

To see how this can affect the ForExample component, let's add some logging to it.

99

NgFor Directive

export class ForExampleComponent {

@Input() episode;
ngOnInit() {
console.log('component created', this.episode)
}
ngOnDestroy() {
console.log('destroying component', this.episode)
}
}

View Example
When we view the example, as we click on Add Episode , we can see console output
indicating that only one component was created - for the newly added item to the list.
However, if we were to remove the trackBy from the *ngFor - every time we click the
button, we would see the items in the component getting destroyed and recreated.
View Example Without trackBy

100

NgSwitch Directives

NgSwitch Directives
ngSwitch is actually comprised of two directives, an attribute directive and a structural

directive. It's very similar to a switch statement in JavaScript and other programming
languages, but in the template.
@Component({
selector: 'app',
template: `
<div class="tabs-selection">
<tab [active]="isSelected(1)" (click)="setTab(1)">Tab 1</tab>
<tab [active]="isSelected(2)" (click)="setTab(2)">Tab 2</tab>
<tab [active]="isSelected(3)" (click)="setTab(3)">Tab 3</tab>
</div>
<div [ngSwitch]="tab">
<tab-content *ngSwitchCase="1">Tab content 1</tab-content>
<tab-content *ngSwitchCase="2">Tab content 2</tab-content>
<tab-content *ngSwitchCase="3"><tab-3></tab-3></tab-content>
<tab-content *ngSwitchDefault>Select a tab</tab-content>
</div>
`
})
export class AppComponent {
tab: number = 0;
setTab(num: number) {
this.tab = num;
}
isSelected(num: number) {
return this.tab === num;
}
}

View Example
Here we see the ngSwitch attribute directive being attached to an element. This expression
bound to the directive defines what will compared against in the switch structural directives.
If an expression bound to ngSwitchCase matches the one given to ngSwitch , those
components are created and the others destroyed. If none of the cases match, then
components that have ngSwitchDefault bound to them will be created and the others
destroyed. Note that multiple components can be matched using ngSwitchCase and in those
cases all matching components will be created. Since components are created or destroyed
be aware of the costs in doing so.

101

NgSwitch Directives

102

Using Multiple Structural Directives

Using Multiple Structural Directives

Sometimes we'll want to combine multiple structural directives together, like iterating using
ngFor but wanting to do an ngIf to make sure that the value matches some or multiple

conditions. Combining structural directives can lead to unexpected results however, so

Angular 2 requires that a template can only be bound to one directive at a time. To apply
multiple directives we'll have to expand the sugared syntax or nest template tags.
@Component({
selector: 'app',
template: `
<template ngFor [ngForOf]="[1,2,3,4,5,6]" let-item>
<div *ngIf="item > 3">
{{item}}
</div>
</template>
`
})

View Example
The previous tabs example can use ngFor and ngSwitch if the tab title and content is
abstracted away into the component class.

103

Using Multiple Structural Directives

import {Component} from '@angular/core';

@Component({
selector: 'app',
template: `
<div class="tabs-selection">
<tab
*ngFor="let tab of tabs; let i = index"
[active]="isSelected(i)"
(click)="setTab(i)">
{{ tab.title }}
</tab>
</div>
<div [ngSwitch]="tabNumber">
<template ngFor [ngForOf]="tabs" let-tab let-i="index">
<tab-content *ngSwitchCase="i">
{{tab.content}}
</tab-content>
</template>
<tab-content *ngSwitchDefault>Select a tab</tab-content>
</div>
`
})
export class AppComponent {
tabNumber: number = -1;
tabs = [
{ title: 'Tab 1', content: 'Tab content 1' },
{ title: 'Tab 2', content: 'Tab content 2' },
{ title: 'Tab 3', content: 'Tab content 3' },
];
setTab(num: number) {
this.tabNumber = num;
}
isSelected(num: number) {
return this.tabNumber === i;
}
}

View Example

104

Advanced Components

Advanced Components

Figure: GB Network PCI Card by Harke is licensed under Public Domain

(https://commons.wikimedia.org/wiki/File:GB_Network_PCI_Card.jpg)

Now that we are familiar with component basics, we can look at some of the more
interesting things we can do with them.

105

Component Lifecycle

Component Lifecycle
A component has a lifecycle managed by Angular itself. Angular manages creation,
rendering, data-bound properties etc. It also offers hooks that allow us to respond to key
lifecycle events.
Here is the complete lifecycle hook interface inventory:
ngOnChanges - called when an input binding value changes
ngOnInit - after the first ngOnChanges
ngDoCheck - after every run of change detection
ngAfterContentInit - after component content initialized
ngAfterContentChecked - after every check of component content
ngAfterViewInit - after component's view(s) are initialized
ngAfterViewChecked - after every check of a component's view(s)
ngOnDestroy - just before the component is destroyed

from Component Lifecycle

View Example

106

Accessing Other Components

Accessing Child Component Classes

@ViewChild and @ViewChildren
The @ViewChild and @ViewChildren decorators provide access to the classe of child
component from the containing component.
The @ViewChild is a decorator function that takes the name of a component class as its
input and finds its selector in the template of the containing component to bind to.
@ViewChild can also be passed a template reference variable.

For example, we bind the component class Alert to its selector <my-alert> and assign it
to the property alert . This allows us to gain access to class methods, like show() .
import {Component, ViewChild} from '@angular/core';
import {Alert} from './alert.component';
@Component({
selector: 'app',
template: `
<my-alert>My alert</my-alert>
<button (click)="showAlert()">Show Alert</button>`
})
export class App {
@ViewChild(Alert) alert: Alert;
showAlert() {
this.alert.show();
}
}

View Example
In the interest of separation of concern, we'd normally want to have child elements take care
of their own behaviors and pass in an @Input() . However, it might be a useful construct in
keeping things generic.
When there are multiple embedded components in the template, we can also use
@ViewChildren . It collects a list of instances of the Alert component, stored in a QueryList

object that behaves similar to an array.

107

Accessing Other Components

import {Component, QueryList, ViewChildren} from '@angular/core';

import {Alert} from './alert.component';
@Component({
selector: 'app',
template: `
<my-alert ok="Next" (close)="showAlert(2)">Step 1: Learn angular</my-alert>
<my-alert ok="Next" (close)="showAlert(3)">Step 2: Love angular</my-alert>
<my-alert ok="Close">Step 3: Build app</my-alert>
<button (click)="showAlert(1)">Show steps</button>`
})
export class App {
@ViewChildren(Alert) alerts: QueryList<Alert>;
alertsArr = [];
ngAfterViewInit() {
this.alertsArr = this.alerts.toArray();
}
showAlert(step) {
this.alertsArr[step - 1].show(); // step 1 is alert index 0
}
}

View Example
As shown above, given a class type to @ViewChild and @ViewChildren a child component
or a list of children component are selected respectively using their selector from the
template. In addition both @ViewChild and @ViewChildren can be passed a selector string:

108

Accessing Other Components

import {Component, QueryList, ViewChild, ViewChildren} from '@angular/core';

import {Alert} from './alert.component';
@Component({
selector: 'app',
template: `
<my-alert #first ok="Next" (close)="showAlert(2)">Step 1: Learn angular</my-alert>
<my-alert ok="Next" (close)="showAlert(3)">Step 2: Love angular</my-alert>
<my-alert ok="Close">Step 3: Build app</my-alert>
<button (click)="showAlert(1)">Show steps</button>`
})
export class App {
@ViewChild('first') alerts: Alert;
@ViewChildren(Alert) alerts: QueryList<Alert>;
alertsArr = [];
ngAfterViewInit() {
this.alertsArr = this.alerts.toArray();
}
showAlert(step) {
this.first.show();
}
}

Note that View children will not be set until the 'ngAfterViewInit' lifecycle hook is called.

@ContentChild and @ContentChildren

@ContentChild and @ContentChildren work the same way as the equivalent @ViewChild

and @ViewChildren , however, the key difference is that @ContentChild and

@ContentChildren select from the projected content within the component.

Again, note that content children will not be set until ngAfterContentInit component
lifecycle hook.
View Example

109

View Encapsulation

View Encapsulation
View encapsulation defines whether the template and styles defined within the component
can affect the whole application or vice versa. Angular provides three encapsulation
strategies:
Emulated (default) - styles from main HTML propagate to the component. Styles

defined in this component's @Component decorator are scoped to this component only.
Native - styles from main HTML do not propagate to the component. Styles defined in

this component's @Component decorator are scoped to this component only.

None - styles from the component propagate back to the main HTML and therefore are

visible to all components on the page. Be careful with apps that have None and
Native components in the application. All components with None encapsulation will

have their styles duplicated in all components with Native encapsulation.

@Component({
...
encapsulation: ViewEncapsulation.None,
styles: [ ... ]
})
export class Hello { ... }

View Example

110

ElementRef

ElementRef
Provides access to the underlying native element (DOM element).
import {AfterContentInit, Component, ElementRef} from '@angular/core';
@Component({
selector: 'app',
template: `
<h1>My App</h1>
<pre style="background: #eee; padding: 1rem; border-radius: 3px; overflow: auto;">
<code>{{ node }}</code>
</pre>
`
})
export class App implements AfterContentInit {
node: string;
constructor(private elementRef: ElementRef) {
}
ngAfterContentInit() {
const tmp = document.createElement('div');
const el = this.elementRef.nativeElement.cloneNode(true);
tmp.appendChild(el);
this.node = tmp.innerHTML;
}
}

View Example

111

Observables

Observables
An exciting new feature used with Angular 2 is the Observable . This isn't an Angular 2
specific feature, but rather a proposed standard for managing async data that will be
included in the release of ES7. Observables open up a continuous channel of
communication in which multiple values of data can be emitted over time. From this we get a
pattern of dealing with data by using array-like operations to parse, modify and maintain
data. Angular 2 uses observables extensively - you'll see them in the HTTP service and the
event system.

112

Using Observables

Using Observables
Let's take a look at a basic example of how to create and use an Observable in an Angular
2 component:
import {Component} from '@angular/core';
import {Observable} from 'rxjs/Observable';
@Component({
selector: 'app',
template: `
<b>Angular 2 Component Using Observables!</b>
<h6 style="margin-bottom: 0">VALUES:</h6>
<div *ngFor="let value of values">- {{ value }}</div>
<h6 style="margin-bottom: 0">ERRORs:</h6>
<div>Errors: {{anyErrors}}</div>
<h6 style="margin-bottom: 0">FINISHED:</h6>
<div>Finished: {{ finished }}</div>
<button style="margin-top: 2rem;" (click)="init()">Init</button>
`
})
export class MyApp {
private data: Observable<Array<number>>;
private values: Array<number> = [];
private anyErrors: boolean;
private finished: boolean;
constructor() {
}
init() {
this.data = new Observable(observer => {
setTimeout(() => {
observer.next(42);
}, 1000);
setTimeout(() => {
observer.next(43);
}, 2000);
setTimeout(() => {
observer.complete();
}, 3000);
});

113

Using Observables

let subscription = this.data.subscribe(

value => this.values.push(value),
error => this.anyErrors = true,
() => this.finished = true
);
}
}

View Example
First we import Observable into our component from rxjs/Observable . Next, in our
constructor we create a new Observable . Note that this creates an Observable data type
that contains data of number type. This illustrates the stream of data that Observables offer
as well as giving us the ability to maintain integrity of the type of data we are expecting to
receive.
Next we call subscribe on this Observable which allows us to listen in on any data that is
coming through. In subscribing we use three distinctive callbacks: the first one is invoked
when receiving new values, the second for any errors that arise and the last represents the
function to be invoked when the sequence of incoming data is complete and successful.
We can also use forEach to listen for incoming data. The key difference between forEach
and subscribe is in how the error and completion callbacks are handled. The forEach call
only accepts the 'next value' callback as an argument; it then returns a promise instead of a
subscription.
When the Observable completes, the promise resolves. When the Observable encounters
an error, the promise is rejected.
You can think of Observable.of(1, 2, 3).forEach(doSomething) as being semantically
equivalent to:
new Promise((resolve, reject) => {
Observable.of(1, 2, 3).subscribe(
doSomething,
reject,
resolve);
});

The forEach pattern is useful for a sequence of events you only expect to happen once.

114

Using Observables

export class MyApp {

private data: Observable<Array<number>>;
private values: Array<number> = [];
private anyErrors: boolean;
private finished: boolean;
constructor() {
}
init() {
this.data = new Observable(observer => {
setTimeout(() => {
observer.next(42);
}, 1000);
setTimeout(() => {
observer.next(43);
}, 2000);
setTimeout(() => {
observer.complete();
}, 3000);
this.status = "Started";
});
let subscription = this.data.forEach(v => this.values.push(v))
.then(() => this.status = "Ended");
}
}

View Example

115

Error Handling

Error Handling
If something unexpected arises we can raise an error on the Observable stream and use
the function reserved for handling errors in our subscribe routine to see what happened.
export class App {
values: number[] = [];
anyErrors: Error;
private data: Observable<number[]>;
constructor() {
this.data = new Observable(observer => {
setTimeout(() => {
observer.next(10);
}, 1500);
setTimeout(() => {
observer.error(new Error('Something bad happened!'));
}, 2000);
setTimeout(() => {
observer.next(50);
}, 2500);
});
let subscription = this.data.subscribe(
value => this.values.push(value),
error => this.anyErrors = error
);
}
}

View Example
Here an error is raised and caught. One thing to note is that if we included a .complete()
after we raised the error, this event will not actually fire. Therefore you should remember to
include some call in your error handler that will turn off any visual loading states in your
application.

116

Disposing Subscriptions and Releasing Resources

Disposing Subscriptions and Releasing

Resources
In some scenarios we may want to unsubscribe from an Observable stream. Doing this is
pretty straightforward as the .subscribe() call returns a data type that we can call
.unsubscribe() on.

export class MyApp {

private data: Observable<Array<string>>;
private value: string;
private subscribed: boolean;
private status: string;
init() {
this.data = new Observable(observer => {
let timeoutId = setTimeout(() => {
observer.next('You will never see this message');
}, 2000);
this.status = 'Started';
return onUnsubscribe = () => {
this.subscribed = false;
this.status = 'Finished';
clearTimeout(timeoutId);
}
});
let subscription = this.data.subscribe(
value => this.value = value,
error => console.log(error),
() => this.status = 'Finished';
);
this.subscribed = true;
setTimeout(() => {
subscription.unsubscribe();
}, 1000);
}
}

View Example

117

Disposing Subscriptions and Releasing Resources

Calling .unsubscribe() will unhook a member's callbacks listening in on the Observable

stream. When creating an Observable you can also return a custom callback,
onUnsubscribe , that will be invoked when a member listening to the stream has

unsubscribed. This is useful for any kind of cleanup that must be implemented. If we did not
clear the setTimeout then values would still be emitting, but there would be no one listening.
To save resources we should stop values from being emitted. An important thing to note is
that when you call .unsubscribe() you are destroying the subscription object that is
listening, therefore the on-complete event attached to that subscription object will not get
called.
In most cases we will not need to explicitly call the unsubscribe method unless we want to
cancel early or our Observable has a longer lifespan than our subscription. The default
behavior of Observable operators is to dispose of the subscription as soon as .complete()
or .error() messages are published. Keep in mind that RxJS was designed to be used in
a "fire and forget" fashion most of the time.

118

Observables vs Promises

Observables vs Promises
Both Promises and Observables provide us with abstractions that help us deal with the
asynchronous nature of our applications. However, there are important differences between
the two:
As seen in the example above, Observables can define both the setup and teardown
aspects of asynchronous behavior.
Observables are cancellable.

Moreover, Observables can be retried using one of the retry operators provided by the
API, such as retry and retryWhen . On the other hand, Promises require the caller to
have access to the original function that returned the promise in order to have a retry
capability.

119

Using Observables From Other Sources

Using Observables From Other Sources

In the example above we created Observables from scratch which is especially useful in
understanding the anatomy of an Observable .
However, we will often create Observables from callbacks, promises, events, collections or
using many of the operators available on the API.

Observable HTTP Events

A common operation in any web application is getting or posting data to a server. Angular
applications do this with the Http library, which previously used Promises to operate in an
asynchronous manner. The updated Http library now incorporates Observables for
triggering events and getting new data. Let's take a quick look at this:
import {Component} from '@angular/core';
import {Http} from '@angular/http';
import 'rxjs/Rx';
@Component({
selector: 'app',
template: `
<b>Angular 2 HTTP requests using RxJs Observables!</b>
<ul>
<li *ngFor="let doctor of doctors">{{doctor.name}}</li>
</ul>
`
})
export class MyApp {
private doctors = [];
constructor(http: Http) {
http.get('http://jsonplaceholder.typicode.com/users/')
.flatMap((data) => data.json())
.subscribe((data) => {
this.doctors.push(data);
});
}
}

View Example

120

Using Observables From Other Sources

This basic example outlines how the Http library's common routines like get , post ,
put and delete all return Observables that allow us to asynchronously process any

resulting data.

Observable Form Events

Let's take a look at how Observables are used in Angular 2 forms. Each field in a form is
treated as an Observable that we can subscribe to and listen for any changes made to the
value of the input field.
import {Component} from '@angular/core';
import {FormControl, FormGroup, FormBuilder} from '@angular/forms';
import 'rxjs/add/operator/filter';
import 'rxjs/add/operator/map';
@Component({
selector: 'app',
template: `
<form [formGroup]="coolForm">
<input formControlName="email">
</form>
<div>
<b>You Typed Reversed:</b> {{data}}
</div>
`
})
export class MyApp {
email: FormControl;
coolForm: FormGroup;
data: string;
constructor(private fb: FormBuilder) {
this.email = new FormControl();
this.coolForm = fb.group({
email: this.email
});
this.email.valueChanges
.filter(n=>n)
.map(n=>n.split('').reverse().join(''))
.subscribe(value => this.data = value);
}
}

View Example

121

Using Observables From Other Sources

Here we have created a new form by initializing a new FormControl field and grouped it into
a FormGroup tied to the coolForm HTML form. The Control field has a property
.valueChanges that returns an Observable that we can subscribe to. Now whenever a user

types something into the field we'll get it immediately.

122

Observables Array Operations

Observables Array Operations

In addition to simply iterating over an asynchronous collection, we can perform other
operations such as filter or map and many more as defined in the RxJS API. This is what
bridges an Observable with the iterable pattern, and lets us conceptualize them as
collections.
Let's expand our example and do something a little more with our stream:
export class MyApp {
private doctors = [];
constructor(http: Http) {
http.get('http://jsonplaceholder.typicode.com/users/')
.flatMap((response) => response.json())
.filter((person) => person.id > 5)
.map((person) => "Dr. " + person.name)
.subscribe((data) => {
this.doctors.push(data);
});
}
}

View Example
Here are two really useful array operations - map and filter . What exactly do these do?
map will create a new array with the results of calling a provided function on every

element in this array. In this example we used it to create a new result set by iterating
through each item and appending the "Dr." abbreviation in front of every user's name.
Now every object in our array has "Dr." prepended to the value of its name property.
filter will create a new array with all elements that pass the test implemented by a

provided function. Here we have used it to create a new result set by excluding any user
whose id property is less than six.
Now when our subscribe callback gets invoked, the data it receives will be a list of JSON
objects whose id properties are greater than or equal to six and whose name properties
have been prepended with Dr. .
Note the chaining function style, and the optional static typing that comes with TypeScript,
that we used in this example. Most importantly functions like filter return an Observable ,
as in Observables beget other Observables , similarly to promises. In order to use map and
filter in a chaining sequence we have flattened the results of our Observable using

123

Observables Array Operations

flatMap . Since filter accepts an Observable , and not an array, we have to convert our

array of JSON objects from data.json() to an Observable stream. This is done with
flatMap .

There are many other array operations you can employ in your Observables ; look for them
in the RxJS API.
rxmarbles.com is a helpful resource to understand how the operations work.

124

Combining Streams with flatMap

Combining Streams with flatMap

Figure: FlatMap created by ReactiveX licensed under CC-3

(http://reactivex.io/documentation/operators/flatmap.html)

A case for FlatMap:

A simple observable stream
A stream of arrays
Filter the items from each event
Stream of filtered items
Filter + map simplified with flatMap
Let's say we wanted to implement an AJAX search feature in which every keypress in a text
field will automatically perform a search and update the page with the results. How would
this look? Well we would have an Observable subscribed to events coming from an input
field, and on every change of input we want to perform some HTTP request, which is also an
Observable we subscribe to. What we end up with is an Observable of an Observable .

By using flatMap we can transform our event stream (the keypress events on the text field)
into our response stream (the search results from the HTTP request).
app/services/search.service.ts

125

Combining Streams with flatMap

import {Http} from '@angular/http';

import {Injectable} from '@angular/core';
@Injectable()
export class SearchService {
constructor(private http: Http) {}
search(term: string) {
return this.http
.get('https://api.spotify.com/v1/search?q=' + term + '&type=artist')
.map((response) => response.json())
}
}

Here we have a basic service that will undergo a search query to Spotify by performing a get
request with a supplied search term. This search function returns an Observable that has
had some basic post-processing done (turning the response into a JSON object).
OK, let's take a look at the component that will be using this service.
app/app.component.ts

126

Combining Streams with flatMap

import { Component } from '@angular/core';

import { FormControl,
FormGroup,
FormBuilder } from '@angular/forms';
import { SearchService } from './services/search.service';
import 'rxjs/Rx';
@Component({
selector: 'app',
template: `
<form [formGroup]="coolForm"><input formControlName="search" placeholder="Sear
ch Spotify artist"></form>
<div *ngFor="let artist of result">
{{artist.name}}
</div>
`
})
export class MyApp {
searchField: FormControl;
coolForm: FormGroup;
constructor(private searchService:SearchService, private fb:FormBuilder) {
this.searchField = new FormControl();
this.coolForm = fb.group({search: this.searchField});
this.searchField.valueChanges
.debounceTime(400)
.flatMap(term => this.searchService.search(term))
.subscribe((result) => {
this.result = result.artists.items
});
}
}

View Example
Here we have set up a basic form with a single field, search , which we subscribe to for
event changes. We've also set up a simple binding for any results coming from the
SearchService . The real magic here is flatMap which allows us to flatten our two separate

subscribed Observables into a single cohesive stream we can use to control events coming
from user input and from server responses.
Note that flatMap flattens a stream of Observables (i.e Observable of Observables ) to a
stream of emitted values (a simple Observable ), by emitting on the "trunk" stream
everything that will be emitted on "branch" streams.

127

Combining Streams with flatMap

128

Cold vs Hot Observables

Cold vs Hot Observables

Observables can be classified into two main groups: hot and cold Observables . Let's start

with a cold Observable .

const obsv = new Observable(observer => {
setTimeout(() => {
observer.next(1);
}, 1000);
setTimeout(() => {
observer.next(2);
}, 2000);
setTimeout(() => {
observer.next(3);
}, 3000);
setTimeout(() => {
observer.next(4);
}, 4000);
});
// Subscription A
setTimeout(() => {
obsv.subscribe(value => console.log(value));
}, 0);
// Subscription B
setTimeout(() => {
obsv.subscribe(value => console.log(`>>>> ${value}`));
}, 2500);

View Example
In the above case subscriber B subscribes 2000ms after subscriber A. Yet subscriber B is
starting to get values like subscriber A only time shifted. This behavior is referred to as a
cold Observable . A useful analogy is watching a pre-recorded video, such as on Netflix. You
press Play and the movie starts playing from the beginning. Someone else can start playing
the same movie in their own home 25 minutes later.
On the other hand there is also a hot Observable , which is more like a live performance.
You attend a live band performance from the beginning, but someone else might be 25
minutes late to the show. The band will not start playing from the beginning and the
129

Cold vs Hot Observables

latecomer must start watching the performance from where it is.

We have already encountered both kind of Observables . The example above is a cold
Observable , while the example that uses valueChanges on our text field input is a hot
Observable .

Converting from Cold Observables to Hot Observables

A useful method within RxJS API is the publish method. This method takes in a cold
Observable as its source and returns an instance of a ConnectableObservable . In this case

we will have to explicitly call connect on our hot Observable to start broadcasting values to
its subscribers.
const obsv = new Observable(observer => {
setTimeout(() => {
observer.next(1);
}, 1000);
setTimeout(() => {
observer.next(2);
}, 2000);
setTimeout(() => {
observer.next(3);
}, 3000);
setTimeout(() => {
observer.next(4);
}, 4000);
}).publish();
obsv.connect();
// Subscription A
setTimeout(() => {
obsv.subscribe(value => console.log(value));
}, 0);
// Subscription B
setTimeout(() => {
obsv.subscribe(value => console.log(` ${value}`));
}, 2500);

View Example

130

Cold vs Hot Observables

In the case above, the live performance starts at 1000ms , subscriber A arrived to the concert
hall at 0s to get a good seat and our subscriber B arrived at the performance at 2500ms
and missed a bunch of songs.
Another useful method to work with hot Observables instead of connect is refCount . This
is an auto connect method, that will start broadcasting as soon as there is more than one
subscriber. Analogously, it will stop if the number of subscribers goes to 0; in other words, if
everyone in the audience walks out, the performance will stop.

131

Summary

Summary
Observables offer a flexible set of APIs for composing and transforming asynchronous

streams. They provide a multitude of functions to create streams from many other types, and
to manipulate and transform them. We've taken a look at how Angular 2 uses Observables
to create streams from many other types to read user input, perform asynchronous data
fetches and set up custom emit/subscribe routines.
rxjs 4 to 5 migration
rxjs Observable API
Which operator do I use?
rxmarbles
RxJS Operators by Example

132

Angular 2 Dependency Injection

Angular 2 Dependency Injection

Dependency Injection (DI) was a core feature in Angular 1.x, and that has not changed in
Angular 2. DI is a programming concept that predates Angular. The purpose of DI is to
simplify dependency management in software components. By reducing the amount of
information a component needs to know about its dependencies, unit testing can be made
easier and code is more likely to be flexible.
Angular 2 improves on Angular 1.x's DI model by unifying Angular 1.x's two injection
systems. Tooling issues with respect to static analysis, minification and namespace
collisions have also been fixed in Angular 2.

133

What is DI?

What is DI?
So dependency injection makes programmers' lives easier, but what does it really do?
Consider the following code:
class Hamburger {
private bun: Bun;
private patty: Patty;
private toppings: Toppings;
constructor() {
this.bun = new Bun('withSesameSeeds');
this.patty = new Patty('beef');
this.toppings = new Toppings(['lettuce', 'pickle', 'tomato']);
}
}

The above code is a contrived class that represents a hamburger. The class assumes a
Hamburger consists of a Bun , Patty and Toppings . The class is also responsible for

making the Bun , Patty and Toppings . This is a bad thing. What if a vegetarian burger
were needed? One naive approach might be:
class VeggieHamburger {
private bun: Bun;
private patty: Patty;
private toppings: Toppings;
constructor() {
this.bun = new Bun('withSesameSeeds');
this.patty = new Patty('tofu');
this.toppings = new Toppings(['lettuce', 'pickle', 'tomato']);
}
}

There, problem solved right? But what if we need a gluten free hamburger? What if we want
different toppings... maybe something more generic like:

134

What is DI?

class Hamburger {
private bun: Bun;
private patty: Patty;
private toppings: Toppings;
constructor(bunType: string, pattyType: string, toppings: string[]) {
this.bun = new Bun(bunType);
this.patty = new Patty(pattyType);
this.toppings = new Toppings(toppings);
}
}

Okay this is a little different, and it's more flexible in some ways, but it is still quite brittle.
What would happen if the Patty constructor changed to allow for new features? The whole
Hamburger class would have to be updated. In fact, any time any of these constructors used

in Hamburger 's constructor are changed, Hamburger would also have to be changed.
Also, what happens during testing? How can Bun , Patty and Toppings be effectively
mocked?
Taking those concerns into consideration, the class could be rewritten as:
class Hamburger {
private bun: Bun;
private patty: Patty;
private toppings: Toppings;
constructor(bun: Bun, patty: Patty, toppings: Toppings) {
this.bun = bun;
this.patty = patty;
this.toppings = toppings;
}
}

Now when Hamburger is instantiated it does not need to know anything about its Bun ,
Patty , or Toppings . The construction of these elements has been moved out of the class.

This pattern is so common that TypeScript allows it to be written in shorthand like so:
class Hamburger {
constructor(private bun: Bun, private patty: Patty,
private toppings: Toppings) {}
}

The Hamburger class is now simpler and easier to test. This model of having the
dependencies provided to Hamburger is basic dependency injection.
However there is still a problem. How can the instantiation of Bun , Patty and Toppings
best be managed?

135

What is DI?

This is where dependency injection as a framework can benefit programmers, and it is what
Angular 2 provides with its dependency injection system.

136

DI Framework

DI Framework
So there's a fancy new Hamburger class that is easy to test, but it's currently awkward to
work with. Instantiating a Hamburger requires:
const hamburger = new Hamburger(new Bun(), new Patty('beef'), new Toppings([]));

That's a lot of work to create a Hamburger , and now all the different pieces of code that
make Hamburger s have to understand how Bun , Patty and Toppings get instantiated.
One approach to dealing with this new problem might be to make a factory function like so:
function makeHamburger() {
const bun = new Bun();
const patty = new Patty('beef');
const toppings = new Toppings(['lettuce', 'tomato', 'pickles']);
return new Hamburger(bun, patty, toppings);
}

This is an improvement, but when more complex Hamburger s need to be created this factory
will become confusing. The factory is also responsible for knowing how to create four
different components. This is a lot for one function.
This is where a dependency injection framework can help. DI Frameworks have the concept
of an Injector object. An Injector is a lot like the factory function above, but more general,
and powerful. Instead of one giant factory function, an Injector has a factory, or recipe (pun
intended) for a collection of objects. With an Injector , creating a Hamburger could be as
easy as:
const injector = new Injector([Hamburger, Bun, Patty, Toppings]);
const burger = injector.get(Hamburger);

137

Angular 2's DI

Angular 2's DI
The last example introduced a hypothetical Injector object. Angular 2 simplifies DI even
further. With Angular 2, programmers almost never have to get bogged down with injection
details.
Angular 2's DI system is (mostly) controlled through @NgModule s. Specifically the
providers ' array.

For example:
import { Injectable, NgModule } from '@angular/core';
@Injectable()
class Hamburger {
constructor(private bun: Bun, private patty: Patty,
private toppings: Toppings) {}
}
@NgModule({
providers: [ Hamburger ],
})
export class DiExample {};

In the above example the DiExample module is told about the Hamburger class.
Another way of saying this is that Angular 2 has been provided a Hamburger .
That seems pretty straightforward, but astute readers will be wondering how Angular 2
knows how to build Hamburger . What if Hamburger was a string, or a plain function?
Angular 2 assumes that it's being given a class.
What about Bun , Patty and Toppings ? How is Hamburger getting those?
It's not, at least not yet. Angular 2 does not know about them yet. That can be changed
easily enough:

138

Angular 2's DI

import { Injectable, NgModule } from '@angular/core';

@Injectable()
class Hamburger {
constructor(private bun: Bun, private patty: Patty,
private toppings: Toppings) {}
}
@Injectable()
class Patty {}
@Injectable()
class Bun {}
@Injectable()
class Toppings {}
@NgModule({
providers: [ Hamburger, Patty, Bun, Toppings ],
})

Okay, this is starting to look a little bit more complete. Although it's still unclear how
Hamburger is being told about its dependencies. Perhaps that is related to those odd
@Injectable statements.

139

@Inject() and @Injectable

@Inject and @Injectable

Statements that look like @SomeName are decorators. Decorators are a proposed extension to
JavaScript. In short, decorators let programmers modify and/or tag methods, classes,
properties and parameters. There is a lot to decorators. In this section the focus will be on
decorators relevant to DI: @Inject and @Injectable . For more information on Decorators
please see the EcmaScript 6 and TypeScript Features section.

@Inject()
@Inject() is a manual mechanism for letting Angular 2 know that a parameter must be

injected. It can be used like so:

import { Component, Inject } from '@angular/core';
import { Hamburger } from '../services/hamburger';
@Component({
selector: 'app',
template: `Bun Type: {{ bunType }}`
})
export class App {
bunType: string;
constructor(@Inject(Hamburger) h) {
this.bunType = h.bun.type;
}
}

In the above we've asked for h to be the singleton Angular associates with the class
symbol Hamburger by calling @Inject(Hamburger) . It's important to note that we're using
Hamburger for its typings and as a reference to its singleton. We are not using Hamburger to

instantiate anything, Angular does that for us behind the scenes.

When using TypeScript, @Inject is only needed for injecting primitives. TypeScript's types
let Angular 2 know what to do in most cases. The above example would be simplified in
TypeScript to:

140

@Inject() and @Injectable

import { Component } from '@angular/core';

import { Hamburger } from '../services/hamburger';
@Component({
selector: 'app',
template: `Bun Type: {{ bunType }}`
})
export class App {
bunType: string;
constructor(h: Hamburger) {
this.bunType = h.bun.type;
}
}

View Example

@Injectable()
@Injectable() lets Angular 2 know that a class can be used with the dependency injector.
@Injectable() is not strictly required if the class has other Angular 2 decorators on it or

does not have any dependencies.

What is important is that any class that is going to be injected with Angular 2 is decorated.
However, best practice is to decorate injectables with @Injectable() , as it makes more
sense to the reader.
Here's an example of Hamburger marked up with @Injectable :
import { Injectable } from '@angular/core';
import { Bun } from './bun';
import { Patty } from './patty';
import { Toppings } from './toppings';
@Injectable()
export class Hamburger {
constructor(public bun: Bun, public patty: Patty, public toppings: Toppings) {
}
}

In the above example Angular 2's injector determines what to inject into Hamburger 's
constructor by using type information. This is possible because these particular
dependencies are typed, and are not primitive types. In some cases Angular 2's DI needs
more information than just types.

141

@Inject() and @Injectable

142

Injection Beyond Classes

Injection Beyond Classes

So far the only types that injection has been used for have been classes, but
Angular 2 is not limited to injecting classes. The concept of providers was also briefly
touched upon.
So far providers have been used with Angular 2's @NgModule meta in an array. providers
have also all been class identifiers. Angular 2 lets programmers specify providers with a
more verbose "recipe". This is done with by providing Angular 2 an Object literal ( {} ):
import { NgModule } from '@angular/core';
import { App } from './containers/app'; // hypothetical app component
import { Hamburger } from './services/hamburger';
@NgModule({
providers: [ { provide: Hamburger, useClass: Hamburger } ],
})
export class DiExample {};

This example is yet another example that provide s a class, but it does so with Angular 2's
longer format.
This long format is really handy. If the programmer wanted to switch out Hamburger
implementations, for example to allow for a DoubleHamburger , they could do this easily:
import { NgModule } from '@angular/core';
import { App } from './containers/app'; // hypothetical app component
import {Hamburger} from './services/hamburger';
import {DoubleHamburger} from './services/double-hamburger';
@NgModule({
providers: [ { provide: Hamburger, useClass: DoubleHamburger } ],
})
export class DiExample {};

The best part of this implementation swap is that the injection system knows how to build
DoubleHamburgers , and will sort all of that out.

The injector can use more than classes though. useValue and useFactory are two other
examples of provider "recipes" that Angular 2 can use. For example:

143

Injection Beyond Classes

import { NgModule } from '@angular/core';

import { App } from './containers/app'; // hypothetical app component
const randomFactory = () => { return Math.random(); };
@NgModule({
providers: [ { provide: 'Random', useFactory: randomFactory } ],
})
export class DiExample {};

In the hypothetical app component, 'Random' could be injected like:

import { Component, Inject, provide } from '@angular/core';
import { Hamburger } from '../services/hamburger';
@Component({
selector: 'app',
template: `Random: {{ value }}`
})
export class App {
value: number;
constructor(@Inject('Random') r) {
this.value = r;
}
}

View Example
One important note is that 'Random' is in quotes, both in the provide function and in the
consumer. This is because as a factory we have no Random identifier anywhere to access.
The above example uses Angular 2's useFactory recipe. When Angular 2 is told to
provide things using useFactory , Angular 2 expects the provided value to be a function.

Sometimes functions and classes are even more than what's needed. Angular 2 has a
"recipe" called useValue for these cases that works almost exactly the same:
import { NgModule } from '@angular/core';
import { App } from './containers/app'; // hypothetical app component
@NgModule({
providers: [ { provide: 'Random', useValue: Math.random() } ],
})
export class DiExample {};

View Example

144

Injection Beyond Classes

In this case, the product of Math.random is assigned to the useValue property passed to the
provider .

145

The Injector Tree

The Injector Tree

Angular 2 injectors (generally) return singletons. That is, in the previous example, all
components in the application will receive the same random number. In Angular 1.x there
was only one injector, and all services were singletons. Angular 2 overcomes this limitation
by using a tree of injectors.
In Angular 2 there is not just one injector per application, there is at least one injector per
application. Injectors are organized in a tree that parallels Angular 2's component tree.
Consider the following tree, which models a chat application consisting of two open chat
windows, and a login/logout widget.

146

The Injector Tree

147

The Injector Tree

Figure: Image of a Component Tree, and a DI Tree

In the image above, there is one root injector, which is established through @NgModule 's
providers array. There's a LoginService registered with the root injector.

Below the root injector is the root @Component . This particular component has no providers
array and will use the root injector for all of its dependencies.
There are also two child injectors, one for each ChatWindow component. Each of these
components has their own instantiation of a ChatService .
There is a third child component, Logout/Login , but it has no injector.
There are several grandchild components that have no injectors. There are ChatFeed and
ChatInput components for each ChatWindow . There are also LoginWidget and
LogoutWidget components with Logout/Login as their parent.

The injector tree does not make a new injector for every component, but does make a new
injector for every component with a providers array in its decorator.
Components that have no providers array look to their parent component for an injector. If
the parent does not have an injector, it looks up until it reaches the root injector.
Warning: Be careful with provider arrays. If a child component is decorated with a
providers array that contains dependencies that were also requested in the parent

component(s), the dependencies the child receives will shadow the parent dependencies.
This can have all sorts of unintended consequences.
Consider the following example:
app/module.ts

148

The Injector Tree

import { NgModule } from '@angular/core';

import { BrowserModule } from '@angular/platform-browser';
import { App } from './app.component';
import { ChildInheritor, ChildOwnInjector } from './components/index';
import { Unique } from './services/unique';

const randomFactory = () => { return Math.random(); };

@NgModule({
imports: [
BrowserModule
],
declarations: [
App,
ChildInheritor,
ChildOwnInjector,
],
/** Provide dependencies here */
providers: [
Unique,
],
bootstrap: [ App ],
})
export class AppModule {}

In the example above, Unique is bootstrapped into the root injector.

app/services/unique.ts
import { Injectable } from '@angular/core';
@Injectable()
export class Unique {
value: string;
constructor() {
this.value = (+Date.now()).toString(16) + '.' +
Math.floor(Math.random() * 500);
}
}

The Unique service generates a value unique to its instance upon instantiation.
app/components/child-inheritor.component.ts

149

The Injector Tree

import { Component, Inject } from '@angular/core';

import { Unique } from '../services/unique';
@Component({
selector: 'child-inheritor',
template: `<span>{{ value }}</span>`
})
export class ChildInheritor {
value: number;
constructor(u: Unique) {
this.value = u.value;
}
}

The child inheritor has no injector. It will traverse the component tree upwards looking for an
injector.
app/components/child-own-injector.component.ts
import { Component, Inject } from '@angular/core';
import { Unique } from '../services/unique';
@Component({
selector: 'child-own-injector',
template: `<span>{{ value }}</span>`,
providers: [Unique]
})
export class ChildOwnInjector {
value: number;
constructor(u: Unique) {
this.value = u.value;
}
}

The child own injector component has an injector that is populated with its own instance of
Unique . This component will not share the same value as the root injector's Unique

instance.
app/containers/app.ts

150

The Injector Tree

import { Component, Inject } from '@angular/core';

import { Unique } from '../services/unique';
@Component({
selector: 'app',
template: `
<p>
App's Unique dependency has a value of {{ value }}
</p>
<p>
which should match
</p>
<p>
ChildInheritor's value: <child-inheritor></child-inheritor>
</p>
<p>
However,
</p>
<p>
ChildOwnInjector should have its own value: <child-own-injector></child-own-injec
tor>
<p>
ChildOwnInjector's other instance should also have its own value <child-own-injec
tor></child-own-injector>
</p>
`,
})
export class App {
value: number;
constructor(u: Unique) {
this.value = u.value;
}
}

View Example

151

Change Detection

Change Detection

Figure: Change Detector by Vovka is licensed under Public Domain

(https://pixabay.com/en/coins-handful-russia-ruble-kopek-650779/)

Change detection is the process that allows Angular to keep our views in sync with our
models.
Change detection has changed in a big way between the old version of Angular and the new
one. In Angular 1, the framework kept a long list of watchers (one for every property bound
to our templates) that needed to be checked every-time a digest cycle was started. This was
called dirty checking and it was the only change detection mechanism available.
Because by default Angular 1 implemented two way data binding, the flow of changes was
pretty much chaotic, models were able to change directives, directives were able to change
models, directives were able to change other directives and models were able to change
other models.

152

Change Detection

In Angular 2, the flow of information is unidirectional, even when using ngModel to

implement two way data binding, which is only syntactic sugar on top of the unidirectional
flow. In this new version of the framework, our code is responsible for updating the models.
Angular is only responsible for reflecting those changes in the components and the DOM by
means of the selected change detection strategy.

153

Change Detection Strategies in Angular 1 vs Angular 2

Change Detection Strategies in Angular 1

vs Angular 2
Another difference between both versions of the framework is the way the nodes of an
application (directives or components) are checked to see if the DOM needs to be updated.
Because of the nature of two-way data binding, in Angular 1 there was no guarantee that a
parent node would always be checked before a child node. It was possible that a child node
could change a parent node or a sibling or any other node in the tree, and that in turn would
trigger new updates down the chain. This made it difficult for the change detection
mechanism to traverse all the nodes without falling in a circular loop with the infamous
message:
10 $digest() iterations reached. Aborting!

In Angular 2, changes are guaranteed to propagate unidirectionally. The change detector will
traverse each node only once, always starting from the root. That means that a parent
component is always checked before its children components.
Tree traversing in Angular 1 vs Angular 2

154

Change Detection Strategies in Angular 1 vs Angular 2

Figure: File Structure

155

How Change Detection Works

How Change Detection Works

Let's see how change detection works with a simple example.
We are going to create a simple MovieApp to show information about one movie. This app is
going to consist of only two components: the MovieComponent that shows information about
the movie and the MainComponent which holds a reference to the movie with buttons to
perform some actions.
Our MainComponent will have three properties: the slogan of the app, the title of the
movie and the lead actor . The last two properties will be passed to the MovieComponent
element referenced in the template.
app/main.component.ts
import {Component} from '@angular/core';
import {MovieComponent} from './movie.component';
import {Actor} from './actor.model';
@Component({
selector: 'main',
template: `
<h1>MovieApp</h1>
<p>{{ slogan }}</p>
<button type="button" (click)="changeActorProperties()">Change Actor Properties</b
utton>
<button type="button" (click)="changeActorObject()">Change Actor Object</button>
<movie [title]="title" [actor]="actor"></movie>`
})
export class MainComponent {
slogan: string = 'Just movie information';
title: string = 'Terminator 1';
actor: Actor = new Actor('Arnold', 'Schwarzenegger');
changeActorProperties() {
this.actor.firstName = 'Nicholas';
this.actor.lastName = 'Cage';
}
changeActorObject() {
this.actor = new Actor('Bruce', 'Willis');
}
}

156

How Change Detection Works

In the above code snippet, we can see that our component defines two buttons that trigger
different methods. The changeActorProperties will update the lead actor of the movie by
directly changing the properties of the actor object. In contrast, the method
changeActorObject will change the information of the actor by creating a completely new

instance of the Actor class.

The Actor model is pretty straightforward, it is just a class that defines the firstName and
the lastName of an actor.
app/actor.model.ts
export class Actor {
constructor(
public firstName: string,
public lastName: string) {}
}

Finally, the MovieComponent shows the information provided by the MainComponent in its
template.
app/movie.component.ts
import {Component, Input} from '@angular/core';
import {Actor} from './actor.model';
@Component({
selector: 'movie',
styles: ['div {border: 1px solid black}'],
template: `
<div>
<h3>{{ title }}</h3>
<p>
<label>Actor:</label>
<span>{{actor.firstName}} {{actor.lastName}}</span>
</p>
</div>`
})
export class MovieComponent {
@Input() title: string;
@Input() actor: Actor;
}

The final result of the app is shown in the screenshot below:

157

How Change Detection Works

Figure: File Structure

158

Change Detector Classes

Change Detector Classes

At runtime, Angular 2 will create special classes that are called change detectors, one for
every component that we have defined. In this case, Angular will create two classes:
MainComponent_ChangeDetector and MovieComponent_ChangeDetector .

The goal of the change detectors is to know which model properties used in the template of
a component have changed since the last time the change detection process ran.
In order to know that, Angular creates an instance of the appropriate change detector class
and a link to the component that it's supposed to check.
In our example, because we only have one instance of the MainComponent and the
MovieComponent , we will have only one instance of the MainComponent_ChangeDetector and

the MovieComponent_ChangeDetector .
The code snippet below is a conceptual model of how the MainComponent_ChangeDetector
class might look.
class MainComponent_ChangeDetector {
constructor(
public previousSlogan: string,
public previousTitle: string,
public previousActor: Actor,
public movieComponent: MovieComponent
) {}
detectChanges(slogan: string, title: string, actor: Actor) {
if (slogan !== this.previousSlogan) {
this.previousSlogan = slogan;
this.movieComponent.slogan = slogan;
}
if (title !== this.previousTitle) {
this.previousTitle = title;
this.movieComponent.title = title;
}
if (actor !== this.previousActor) {
this.previousActor = actor;
this.movieComponent.actor = actor;
}
}
}

159

Change Detector Classes

Because in the template of our MainComponent we reference three variables ( slogan ,

title and actor ), our change detector will have three properties to store the "old" values

of these three properties, plus a reference to the MainComponent instance that it's supposed
to "watch". When the change detection process wants to know if our MainComponent
instance has changed, it will run the method detectChanges passing the current model
values to compare with the old ones. If a change was detected, the component gets
updated.
Disclaimer: This is just a conceptual overview of how change detector classes work; the
actual implementation may be different.

Change Detection Strategy: Default

By default, Angular defines a certain change detection strategy for every component in our
application. To make this definition explicit, we can use the property changeDetection of the
@Component decorator.

app/movie.component.ts
// ...
import {ChangeDetectionStrategy} from '@angular/core';
@Component({
// ...
changeDetection: ChangeDetectionStrategy.Default
})
export class MovieComponent {
// ...
}

View Example
Let's see what happens when a user clicks the button "Change Actor Properties" when using
the Default strategy.
As noted previously, changes are triggered by events and the propagation of changes is
done in two phases: the application phase and the change detection phase.
Phase 1 (Application):
In the first phase, the application (our code) is responsible for updating the models in
response to some event. In this scenario, the properties actor.firstName and
actor.lastName are updated.

Phase 2 (Change Detection):

160

Change Detector Classes

Now that our models are updated, Angular must update the templates using change
detection.
Change detection always starts at the root component, in this case the MainComponent , and
checks if any of the model properties bound to its template have changed, comparing the old
value of each property (before the event was triggered) to the new one (after the models
were updated). The MainComponent template has a reference to three properties, slogan ,
title and actor , so the comparison made by its corresponding change detector will look

like:
Is slogan !== previousSlogan ? No, it's the same.
Is title !== previousTitle ? No, it's the same.
Is actor !== previousActor ? No, it's the same.
Notice that even if we change the properties of the actor object, we are always working
with the same instance. Because we are doing a shallow comparison, the result of asking if
actor !== previousActor will always be false even when its internal property values have

indeed changed. Even though the change detector was unable to find any change, the
default strategy for the change detection is to traverse all the components of the tree
even if they do not seem to have been modified.
Next, change detection moves down in the component hierarchy and check the properties
bound to the MovieComponent 's template doing a similar comparison:
Is title !== previousTitle ? No, it's the same.
Is actorFirstName !== previousActorFirstName ? Yes, it has changed.
Is actorLastName !== previousActorLastName ? Yes, it has changed.
Finally, Angular has detected that some of the properties bound to the template have
changed so it will update the DOM to get the view in sync with the model.

Performance Impact
Traversing all the tree components to check for changes could be costly. Imagine that
instead of just having one reference to <movie> inside our MainComponent 's template, we
have multiple references?
<movie *ngFor="let movie of movies" [title]="movie.title" [actor]="movie.actor"></movie
>`

If our movie list grows too big, the performance of our system will start degrading. We can
narrow the problem to one particular comparison:

161

Change Detector Classes

Is actor !== previousActor ?

As we have learned, this result is not very useful because we could have changed the
properties of the object without changing the instance, and the result of the comparison will
always be false . Because of this, change detection is going to have to check every child
component to see if any of the properties of that object ( firstName or lastName ) have
changed.
What if we can find a way to indicate to the change detection that our MovieComponent
depends only on its inputs and that these inputs are immutable? In short, we are trying to
guarantee that when we change any of the properties of the actor object, we end up with a
different Actor instance so the comparison actor !== previousActor will always return
true . On the other hand, if we did not change any property, we are not going to create a

new instance, so the same comparison is going to return false .

If the above condition can be guaranteed (create a new object every time any of its
properties changes, otherwise we keep the same object) and when checking the inputs of
the MovieComponent has this result:
Is title !== previousTitle ? No, it's the same.
Is actor !== previousActor ? No, it's the same.
then we can skip the internal check of the component's template because we are now
certain that nothing has changed internally and there's no need to update the DOM. This will
improve the performance of the change detection system because fewer comparisons have
to be made to propagate changes through the app.

162

Change Detection Strategy: OnPush

Change Detection Strategy: OnPush

To inform Angular that we are going to comply with the conditions mentioned before to
improve performance, we will use the OnPush change detection strategy on the
MovieComponent .

app/movie.component.ts
// ...
@Component({
// ...
changeDetection: ChangeDetectionStrategy.OnPush
})
export class MovieComponent {
// ...
}

View Example
This will inform Angular that our component only depends on its inputs and that any object
that is passed to it should be considered immutable. This time when we click the "Change
Actor Properties" button nothing changes in the view.
Let's follow the logic behind it again. When the user clicks the button, the method
changeActorProperties is called and the properties of the actor object get updated.

When the change detection analyzes the properties bound to the MainComponent 's template,
it will see the same picture as before:
Is slogan !== previousSlogan No, it's the same.
Is title !== previousTitle ? No, it's the same.
Is actor !== previousActor ? No, it's the same.
But this time, we explicitly told Angular that our component only depends on its inputs and all
of them are immutable. Angular then assumes that the MovieComponent hasn't changed and
will skip the check for that component. Because we didn't force the actor object to be
immutable, we end up with our model out of sync with the view.
Let's rerun the app but this time we will click the "Change Actor Object" button. This time, we
are creating a new instance of the Actor class and assigning it to the this.actor object.
When change detection analyzes the properties bound to the MainComponent 's template it
will find:

163

Change Detection Strategy: OnPush

Is slogan !== previousSlogan No, it's the same.

Is title !== previousTitle ? No, it's the same.
Is actor !== previousActor ? Yes, it has changed.
Because change detection now knows that the actor object changed (it's a new instance) it
will go ahead and continue checking the template for MovieComponent to update its view. At
the end, our templates and models are in sync.

164

Enforcing Immutability

Enforcing Immutability
We cheated a little in the previous example. We told Angular that all of our inputs, including
the actor object, were immutable objects, but we went ahead and updated its properties,
violating the immutability principle. As a result we ended up with a sync problem between
our models and our views. One way to enforce immutability is using the library Immutable.js.
Because in JavaScript primitive types like string and number are immutable by definition,
we should only take care of the objects we are using. In this case, the actor object.
Here's an example comparing a mutable type like an array to an immutable type like a
string :

var b = ['C', 'a', 'r'];

b[0] = 'B';
console.log(b) // ['B', 'a', 'r'] => The first letter changed, arrays are mutable
var a = 'Car';
a[0] = 'B';
console.log(a); // 'Car' => The first letter didn't change, strings are immutable

First we need to install the immutable.js library using the command:

npm install --save immutable

Then in our MainComponent we import the library and use it to create an actor object as an
immutable.
app/main.component.ts

165

Enforcing Immutability

import {Component} from '@angular/core';

import {MovieComponent} from './movie.component';
import * as Immutable from 'immutable';
@Component({
selector: 'main',
template: `
<h1>MovieApp</h1>
<p>{{ slogan }}</p>
<button type="button" (click)="changeActor()">Change Actor</button>
<movie [title]="title" [actor]="actor"></movie>`
})
export class MainComponent {
slogan: string = 'Just movie information';
title: string = 'Terminator 1';
actor: Immutable.Map<string, string> = Immutable.Map({firstName: 'Arnold', lastName:
'Schwarzenegger'});
changeActor() {
this.actor = this.actor.merge({firstName: 'Nicholas', lastName: 'Cage'});
}
}

Now, instead of creating an instance of an Actor class, we are defining an immutable

object using Immutable.Map . Because this.actor is now an immutable object, we cannot
change its internal properties ( firstName and lastName ) directly. What we can do however
is create another object based on actor that has different values for both fields - that is
exactly what the merge method does.
Because we are always getting a new object when we try to change the actor , there's no
point in having two different methods in our component. We removed the methods
changeActorProperties and changeActorObject and created a new one called changeActor .

Additional changes have to be made to the MovieComponent as well. First we need to

declare the actor object as an immutable type, and in the template, instead of trying to
access the object properties directly using a syntax like actor.firstName , we need to use
the get method of the immutable.
app/movie.component.ts

166

Enforcing Immutability

import {Component, Input} from '@angular/core';

import {ChangeDetectionStrategy} from '@angular/core';
import * as Immutable from 'immutable';
@Component({
selector: 'movie',
styles: ['div {border: 1px solid black}'],
template: `
<div>
<h3>{{ title }}</h3>
<p>
<label>Actor:</label>
<span>{{actor.get('firstName')}} {{actor.get('lastName')}}</span>
</p>
</div>`,
changeDetection: ChangeDetectionStrategy.OnPush
})
export class MovieComponent {
@Input() title: string;
@Input() actor: Immutable.Map<string, string>;
}

View Example
Using this pattern we are taking full advantage of the "OnPush" change detection strategy
and thus reducing the amount of work done by Angular to propagate changes and to get
models and views in sync. This improves the performance of the application.

167

Additional Resources

Additional Resources
To learn more about change detection, visit the following links (in order of relevance):
NgConf 2014: Change Detection (Video)
Angular API Docs: ChangeDetectionStrategy
Victor Savkin Blog: Change Detection in Angular 2
Victor Savkin Blog: Two Phases of Angular 2 Applications
Victor Savkin Blog: Angular, Immutability and Encapsulation

168

Advanced Angular

Advanced Angular
Angular 2 gives us access to most of the core entities it uses in its architecture. Now that we
understand the different parts involved in an Angular 2 application, let's dig deeper into some
of these entities and take advantage of what we know.

169

Directives

Angular Directives
Angular 2 built-in directives cover a broad range of functionality, but sometimes creating our
own directives will result in more elegant solutions.

170

Creating an Attribute Directive

Creating an Attribute Directive

Let's start with a simple button that moves a user to a different page.
@Component({
selector: 'visit-rangle',
template: `
<button type="button" (click)="visitRangle()">Visit Rangle</button>
`
})
export class VisitRangleComponent {
visitRangle() {
location.href = 'https://rangle.io';
}
}

View Example
We're polite, so rather than just sending the user to a new page, we're going to ask if they're
ok with that first by creating an attribute directive and attaching that to the button.
@Directive({
selector: `[confirm]`
})
export class ConfirmDirective {
@HostListener('click', ['$event'])
confirmFirst(event: Event) {
return window.confirm('Are you sure you want to do this?');
}
}

View Example
Directives are created by using the @Directive decorator on a class and specifying a
selector. For directives, the selector name must be camelCase and wrapped in square
brackets to specify that it is an attribute binding. We're using the @HostListener decorator to
listen in on events on the component or element it's attached to. In this case we're watching
the click event and passing in the event details which are given by the special $event
keyword. Next, we want to attach this directive to the button we created earlier.

171

Creating an Attribute Directive

template: `
<button type="button" (click)="visitRangle()" confirm>Visit Rangle</button>
`

View Example
Notice, however, that the button doesn't work quite as expected. That's because while we're
listening to the click event and showing a confirm dialog, the component's click handler runs
before the directive's click handler and there's no communication between the two. To do
this we'll need to rewrite our directive to work with the component's click handler.
@Directive({
selector: `[confirm]`
})
export class ConfirmDirective {
@Input('confirm') onConfirmed: Function = () => {};
@HostListener('click', ['$event'])
confirmFirst() {
const confirmed = window.confirm('Are you sure you want to do this?');
if(confirmed) {
this.onConfirmed();
}
}
}

View Example
Here, we want to specify what action needs to happen after a confirm dialog's been sent out
and to do this we create an input binding just like we would on a component. We'll use our
directive name for this binding and our component code changes like this:
<button type="button" [confirm]="visitRangle">Visit Rangle</button>

View Example
Now our button works just as we expected. We might want to be able to customize the
message of the confirm dialog however. To do this we'll use another binding.

172

Creating an Attribute Directive

@Directive({
selector: `[confirm]`
})
export class ConfirmDirective {
@Input('confirm') onConfirmed: Function = () => {};
@Input() confirmMessage: string = 'Are you sure you want to do this?';
@HostListener('click', ['$event'])
confirmFirst() {
const confirmed = window.confirm(this.confirmMessage);
if(confirmed) {
this.onConfirmed();
}
}
}

View Example
Our directive gets a new input property that represents the confirm dialog message, which
we pass in to window.confirm call. To take advantage of this new input property, we add
another binding to our button.
<button
type="button"
[confirm]="visitRangle"
confirmMessage="Click ok to visit Rangle.io!">
Visit Rangle
</button>

View Example
Now we have a button with a customizable confirm message before it moves you to a new
url.

173

Creating an Attribute Directive

Listening to an Element Host

Listening to the host - that is, the DOM element the directive is attached to - is among the
primary ways directives extend the component or element's behavior. Previously, we saw its
common use case.
@Directive({
selector: '[myDirective]'
})
class MyDirective {
@HostListener('click', ['$event'])
onClick() {}
}

We can also respond to external events, such as from window or document , by adding the
target in the listener.
@Directive({
selector: `[highlight]`
})
export class HighlightDirective {
constructor(private el: ElementRef, private renderer: Renderer) {}
@HostListener('document:click', ['$event'])
handleClick(event: Event) {
if (this.el.nativeElement.contains(event.target)) {
this.highlight('yellow');
} else {
this.highlight(null);
}
}
highlight(color) {
this.renderer.setElementStyle(this.el.nativeElement, 'backgroundColor', color);
}
}

View Example
Although less common, we can also use @HostListener if we'd like to register listeners
on the host element of a Component.

174

Creating an Attribute Directive

Host Elements
The concept of a host element applies to both directives and components.
For a directive, the concept is fairly straight forward. Whichever template tag you place your
directive attribute on is considered the host element. If we were implementing the
HighlightDirective above like so:

<div>
<p highlight>
<span>Text to be highlighted</span>
</p>
</div>

The <p> tag would be considered the host element. If we were using a custom
TextBoxComponent as the host, the code would look like this:

<div>
<my-text-box highlight>
<span>Text to be highlighted</span>
</my-text-box>
</div>

In the context of a Component, the host element is the tag that you create through the
selector string in the component configuration. For the TextBoxComponent in the example

above, the host element in the context of the component class would be the <my-text-box>
tag.

175

Creating an Attribute Directive

Setting Properties with a Directive

We can use attribute directives to affect the value of properties on the host node by using
the @HostBinding decorator.
The @HostBinding decorator allows us to programatically set a property value on the
directive's host element. It works similarly to a property binding defined in a template, except
it specifically targets the host element. The binding is checked for every change detection
cycle, so it can change dynamically if desired.
For example, lets say that we want to create a directive for buttons that dynamically adds a
class when we press on it. That could look something like:
import { Directive, HostBinding, HostListener } from '@angular/core';
@Directive({
selector: '[buttonPress]'
})
export class ButtonPressDirective {
@HostBinding('attr.role') role = 'button';
@HostBinding('class.pressed') isPressed: boolean;
@HostListener('mousedown') hasPressed() {
this.isPressed = true;
}
@HostListener('mouseup') hasReleased() {
this.isPressed = false;
}
}

Notice that for both use cases of @HostBinding we are passing in a string value for which
property we want to affect. If we don't supply a string to the decorator, then the name of the
class member will be used instead.
In the first @HostBinding , we are statically setting the role attribute to button . For the
second example, the pressed class will be applied when isPressed is true.
Tip: Though less common, @HostBinding can also be applied to Components if
required.

176

Creating a Structural Directive

Creating a Structural Directive

We'll create a delay structural directive that delays instantiation of a component or
element. This can potentially be used for cosmetic effect or for manually handling timing of
when components are loaded, either for performance or UX.
@Directive({
selector: '[delay]'
})
export class DelayDirective {
constructor(
private templateRef: TemplateRef<any>,
private viewContainerRef: ViewContainerRef
) { }
@Input('delay')
set delayTime(time: number): void { }
}

View Example
We use the same @Directive class decorator as attribute directives and define a selector in
the same way. One big difference here is that due to the nature of structural directives being
bound to a template, we have access to TemplateRef , an object representing the template
tag the directive is attached to. We also add an input property in a similar way, but this time
with a set handler so we can execute some code when Angular 2 performs the binding.
We bind delay in exactly the same way as the Angular 2 built-in structural directives.
@Component({
selector: 'app',
template: `
<div *ngFor="let item of [1,2,3,4,5,6]">
<card *delay="500 * item">
{{item}}
</card>
</div>
`
})
export class App {
}

View Example

177

Creating a Structural Directive

Notice that no content is being rendered however. This is due to Angular 2 simulating the
html template tag and not rendering any child elements by default. To be able to get this
content to render, we'll have to attach the template given by TemplateRef as an embedded
view to a view container.

178

Creating a Structural Directive

View Containers and Embedded Views

View Containers are containers where one or more Views can be attached. Views represent
some sort of layout to be rendered and the context under which to render it. View containers
are anchored to components and are responsible for generating its output so this means that
changing which views are attached to the view container affect the final rendered output of
the component.
Two types of views can be attached to a view container: Host Views which are linked to a
Component, and Embedded Views which are linked to a template. Since structural directives
interact with templates, we are interested in using Embedded Views in this case.
import {Directive, Input, TemplateRef, ViewContainerRef} from '@angular/core';
@Directive({
selector: '[delay]'
})
export class DelayDirective {
constructor(
private templateRef: TemplateRef<any>,
private viewContainerRef: ViewContainerRef
) { }
@Input('delay')
set delayTime(time: number): void {
setTimeout(
() => {
this.viewContainerRef.createEmbeddedView(this.templateRef);
},
time);
}
}

View Example
Directives get access to the view container by injecting a ViewContainerRef . Embedded
views are created and attached to a view container by calling the ViewContainerRef 's
createEmbeddedView method and passing in the template. We want to use the template our

directive is attached to so we pass in the injected TemplateRef .

179

Creating a Structural Directive

Providing Context Variables to Directives

Suppose we want to record some metadata on how our directive affected components and
make this data available to them. For example, in our delay directive, we're making a
setTimeout call, which in JavaScript's single-threaded asynchronous model means that it

may not run after the exact time we provided. We'll capture the exact time it loads and make
that variable available in the template.
export class DelayContext {
constructor(private loadTime: number) { }
}
@Directive({
selector: '[delay]'
})
export class DelayDirective {
constructor(
private templateRef: TemplateRef<DelayContext>,
private viewContainerRef: ViewContainerRef
) { }
@Input('delay')
set delayTime(time: number): void {
setTimeout(
() => {
this.viewContainerRef.createEmbeddedView(
this.templateRef,
new DelayContext(performance.now())
);
},
time);
}
}

View Example
We've made a few changes to our delay directive. We've created a new DelayContext
class that contains the context that we want to provide to our directive. In this case, we want
to capture the actual time the createEmbeddedView call occurs and make that available as
loadTime in our directive. We've also provided our new class as the generic argument to

the TemplateRef function. This enables static analysis and lets us make sure our calls to
createEmbeddedView pass in a variable of type DelayContext . In our createEmbeddedView

call we pass in our variable which has captured the time of the method call.

180

Creating a Structural Directive

In the component using delay , we access the loadTime context variable in the same way
we access variables in ngFor .
@Component({
selector: 'app',
template: `
<div *ngFor="let item of [1,2,3,4,5,6]">
<card *delay="500 * item; let loaded = loadTime">
<div class="main">{{item}}</div>
<div class="sub">{{loaded | number:'1.4-4'}}</div>
</card>
</div>
`
})

View Example

181

AoT

AoT in Angular 2
Every Angular application requires a compilation process before they can run in the browser:
the enriched components and templates provided by Angular cannot be understood by the
browser directly. During the compilation, Angular's compiler also improves the app run-time
performance by taking JavaScript VM's feature (like inline caching) into consideration.
The initial compiler in Angular 1.x and Angular 2 is called JiT (Just-in-Time) compiler. As for
AoT, it stands for the Ahead-of-Time compiler that was recently introduced in Angular 2.
Compared to the JiT compilation performed by Angular 2 at run-time, AoT provides a smaller
bundle with faster rendering in the browser. Using AoT, we can reduce the angular2-starter
to 428.8 kb compared to the original 1.2 MB and reduce loading times by skipping
compilation in the browser.
Characteristic

JiT

AoT

Compilation target

Browser

Server

Compilation context

Runtime

Build

Bundle size

Huge (~1.2 MB)

Smaller (~400 KB)

Execution Performance

Better

Startup time

Shorter

The gist of AoT is moving the compilation from run-time to the building process. That means,
first we can remove the JiT compiler (which is around 523kb) from the bundle to have a
smaller build, and second, the browser can execute the code without waiting for JiT in the
run-time which leads to a faster rendering speed.
Early compilation also means that developers can find template bugs without actually
running the code and before it reaches to client. This provides a more robust application with
higher security because less client-side HTML and JavaScript are eval ed. Also, by
introducing compiled code in the building process, AoT makes the application more treeshakable and open to various other optimizations. Bundlers like Rollup and Google Closure
can take that advantage and effectively decrease the bundle size.
Besides, AoT compiler also inlines HTML templates and CSS files and help reduce the
amount of asynchronous requests sent by the application. (Note: this caused a config bug
that we will mention in a latter section)

182

AoT

183

AoT limitations

AoT limitations
However, AoT is not perfect. The main limitation is that AoT, due to the way it compiles the
raw code, cannot be used with common code patterns, for example, default exports from
modules, template literals for templates, and functions in providers, routes, or declarations.
Currently, we do not have a complete list of "AoT Do's and Don'ts" and the Angular team has
not released anything regarding this issue. Rangle made its own list here and also provides
a sandbox for testing features with AoT.
Another problem with AoT is that when the application reaches certain complexity, the AoT
bundle compared to JiT bundle can actually takes up more space. As an trade-off of having
a simpler logic for browser (therefore faster rendering speed), the code generated by AoT is
actually more verbose compared to "dynamic" JiT.

184

AoT Configuration

AoT Configuration
To enable AoT in Angular 2, there are two possible methods:
using ngc directly
using @ngtools/webpack
We recommend the second way because it fits the Angular + Webpack toolchain the best.
One problem of using raw ngc is that ngc tries to inline CSS while lacking necessary
context. For example, the @import 'basscss-basic' statement in index.css would cause
an error like Error: Compilation failed. Resource file not found with ngc . It lacks the
information that 'basscss-basic' is actually a node module inside node_modules . On the
other hand, @ngtools/webpack provides AotPlugin and loader for Webpack which shares
the context with other loaders/plugins. So when ngc is called by @ngtools/webpack , it can
gather necessary informations from other plugins like postcss-import to correctly
understand things like @import 'basscss-basic' .

Config @ngtools/webpack
First, get @ngtools/webpack from npm and save it as a development dependency:
npm install -D @ngtools/webpack

Then, inside the Webpack configuration file (usually named as webpack.config.js ), add
following code:

185

AoT Configuration

import {AotPlugin} from '@ngtools/webpack'

exports = { /* ... */
module: {
rules: [
{
test: /\.ts$/,
loader: '@ngtools/webpack',
}
]
},
plugins: [
new AotPlugin({
tsConfigPath: 'path/to/tsconfig.json',
entryModule: 'path/to/app.module#AppModule'
})
]
}

Here @ngtools/webpack replaces other typescript loader like ts-loader or awesometypescript-loader . It works with AotPlugin together to enable AoT compilation. More

details can be found here.

(Note, for project generated by angular-cli , turning on AoT can be simple as ng build -aot , but since angular-cli does not allow customized webpack configuration for complex use

cases, it may be insufficient.)

186

Immutable.js

Immutable.js
Immutable.js is a library that provides immutable generic collections.

Figure: Ayers Rock Uluru by Stefanoka is licensed under CC BY-SA 3.0

(https://commons.wikimedia.org/wiki/File:Ayers_Rock_Uluru.jpg)

187

What is Immutability?

What is Immutability?
Immutability is a design pattern where something can't be modified after being instantiated.
If we want to change the value of that thing, we must recreate it with the new value instead.
Some JavaScript types are immutable and some are mutable, meaning their value can
change without having to recreate it. Let's explain this difference with some examples:
let movie = {
name: 'Star Wars',
episode: 7
};
let myEp = movie.episode;
movie.episode = 8;
console.log(myEp); // outputs 7

As you can see in this case, although we changed the value of movie.episode , the value of
myEp didn't change. That's because movie.episode 's type, number , is immutable.

let movie1 = {
name: 'Star Wars',
episode: 7
};
let movie2 = movie1;
movie2.episode = 8;
console.log(movie1.episode); // outputs 8

In this case however, changing the value of episode on one object also changed the value of
the other. That's because movie1 and movie2 are of the Object type, and Objects are
mutable.
Of the JavaScript built-in types, the following are immutable:
Boolean
Number
String
Symbol
Null

188

What is Immutability?

Undefined
And the following are mutable:
Object
Array
Function
String's an unusual case, since it can be iterated over using for...of and provides numeric
indexers just like an array, but doing something like:
let message = 'Hello world';
message[5] = '-';
console.log(message); // writes Hello world

This will throw an error in strict mode and fail silently in non-strict mode.

189

The Case for Immutability

The Case for Immutability

One of the more difficult things to manage when structuring an application is managing its
state. This is especially true when your application can execute code asynchronously. Let's
say you execute some piece of code, but something causes it to wait (such as an HTTP
request or user input). After it's completed, you notice the state it's expecting changed
because some other piece of code executed asynchronously and changed its value.
Dealing with that kind of behavior on a small scale might be manageable, but this can show
up all over an application and can be a real headache as the application gets bigger with
more interactions and more complex logic.
Immutability attempts to solve this by making sure that any object referenced in one part of
the code can't be changed by another part of the code unless they have the ability to rebind
it directly.

190

JavaScript Solutions

JavaScript Solutions
Some new features have been added in ES6 that allow for easier implementation of
immutable data patterns.

191

Object.assign

Object.assign
Object.assign lets us merge one object's properties into another, replacing values of

properties with matching names. We can use this to copy an object's values without altering
the existing one.
let movie1 = {
name: 'Star Wars',
episode: 7
};
let movie2 = Object.assign({}, movie1);
movie2.episode = 8;
console.log(movie1.episode); // writes 7
console.log(movie2.episode); // writes 8

As you can see, although we have some way of copying an object, we haven't made it
immutable, since we were able to set the episode's property to 8. Also, how do we modify
the episode property in this case? We do that through the assign call:
let movie1 = {
name: 'Star Wars',
episode: 7
};
let movie2 = Object.assign({}, movie1, { episode: 8 });
console.log(movie1.episode); // writes 7
console.log(movie2.episode); // writes 8

192

Object.freeze

Object.freeze
Object.freeze allows us to disable object mutation.

let movie1 = {
name: 'Star Wars',
episode: 7
};
let movie2 = Object.freeze(Object.assign({}, movie1));
movie2.episode = 8; // fails silently in non-strict mode,
// throws error in strict mode
console.log(movie1.episode); // writes 7
console.log(movie2.episode); // writes 7

One problem with this pattern, however, is how much more verbose our code is and how
difficult it is to read and understand what's actually going on with our data with all of the
boilerplate calls to Object.freeze and Object.assign . We need some more sensible
interface to create and interact with immutable data, and that's where Immutable.js fits in.
Object.freeze is also very slow and should not be used with large arrays.

193

Immutable.js Basics

Immutable.js Basics
To solve our mutability problem, Immutable.js must provide immutable versions of the two
core mutable types, Object and Array.

194

Immutable.Map

Immutable.Map
Map is the immutable version of JavaScript's object structure. Due to JavaScript objects

having the concise object literal syntax, it's often used as a key-value store with key being
type string . This pattern closely follows the map data structure. Let's revisit the previous
example, but use Immutable.Map instead.
import * as Immutable from 'immutable';
let movie1 = Immutable.Map<string, any>({
name: 'Star Wars',
episode: 7
});
let movie2 = movie1;
movie2 = movie2.set('episode', 8);
console.log(movie1.get('episode')); // writes 7
console.log(movie2.get('episode')); // writes 8

Instead of binding the object literal directly to movie1 , we pass it as an argument to

Immutable.Map . This changes how we interact with movie1's properties.

To get the value of a property, we call the get method, passing the property name we want,
like how we'd use an object's string indexer.
To set the value of a property, we call the set method, passing the property name and the
new value. Note that it won't mutate the existing Map object - it returns a new object with the
updated property, so we must rebind the movie2 variable to the new object.

195

Immutable.Map

Map.merge
Sometimes we want to update multiple properties. We can do this using the merge method.
let baseButton = Immutable.Map<string, any>({
text: 'Click me!',
state: 'inactive',
width: 200,
height: 30
});
let submitButton = baseButton.merge({
text: 'Submit',
state: 'active'
});
console.log(submitButton);
// writes { text: 'Submit', state: 'active', width: 200, height: 30 }

196

Nested Objects

Nested Objects
Immutable.Map wraps objects shallowly, meaning if you have an object with properties

bound to mutable types then those properties can be mutated.

let movie = Immutable.Map({
name: 'Star Wars',
episode: 7,
actors: [
{ name: 'Daisy Ridley', character: 'Rey'},
{ name: 'Harrison Ford', character: 'Han Solo' }
],
mpaa: {
rating: 'PG-13',
reason: 'sci-fi action violence'
}
});
movie.get('actors').pop();
movie.get('mpaa').rating = 'PG';
console.log(movie.toObject());
/* writes
{ name: 'Star Wars',
episode: 7,
actors: [ { name: 'Daisy Ridley', character: 'Rey' } ],
mpaa: { rating: 'PG', reason: 'sci-fi action violence' } }
*/

To avoid this issue, use Immutable.fromJS instead.

197

Nested Objects

let movie = Immutable.fromJS({

name: 'Star Wars',
episode: 7,
actors: [
{ name: 'Daisy Ridley', character: 'Rey'},
{ name: 'Harrison Ford', character: 'Han Solo' }
],
mpaa: {
rating: 'PG-13',
reason: 'sci-fi action violence'
}
});
movie.get('actors').pop();
movie.get('mpaa').rating = 'PG';
console.log(movie.toObject());
/* writes
{ name: 'Star Wars',
episode: 7,
actors: List [ Map { "name": "Daisy Ridley", "character": "Rey" }, Map { "name": "Ha
rrison Ford", "character": "Han Solo" } ],
mpaa: Map { "rating": "PG-13", "reason": "sci-fi action violence" } }
*/

So let's say you want to modify movie.mpaa.rating . You might think of doing something like
this: movie = movie.get('mpaa').set('rating', 'PG') . However, set will always return the
calling Map instance, which in this case returns the Map bound to the mpaa key rather than
the movie you wanted. We must use the setIn method to update nested properties.

198

Nested Objects

let movie = Immutable.fromJS({

name: 'Star Wars',
episode: 7,
actors: [
{ name: 'Daisy Ridley', character: 'Rey'},
{ name: 'Harrison Ford', character: 'Han Solo' }
],
mpaa: {
rating: 'PG-13',
reason: 'sci-fi action violence'
}
});
movie = movie
.update('actors', actors => actors.pop())
.setIn(['mpaa', 'rating'], 'PG');
console.log(movie.toObject());
/* writes
{ name: 'Star Wars',
episode: 7,
actors: List [ Map { "name": "Daisy Ridley", "character": "Rey" } ],
mpaa: Map { "rating": "PG", "reason": "sci-fi action violence" } }
*/

We also added a call to Map.update which, unlike set , accepts a function as the second
argument instead of a value. This function accepts the existing value at that key and must
return the new value of that key.

199

Nested Objects

Deleting Keys
Keys can be deleted from maps using the Map.delete and Map.deleteIn methods.
let movie = Immutable.fromJS({
name: 'Star Wars',
episode: 7,
actors: [
{ name: 'Daisy Ridley', character: 'Rey'},
{ name: 'Harrison Ford', character: 'Han Solo' }
],
mpaa: {
rating: 'PG-13',
reason: 'sci-fi action violence'
}
});
movie = movie.delete('mpaa');
console.log(movie.toObject());
/* writes
{ name: 'Star Wars',
episode: 7,
actors: List [ Map { "name": "Daisy Ridley", "character": "Rey" }, Map { "name": "Ha
rrison Ford", "character": "Han Solo" } ] }
*/

200

Nested Objects

Maps are Iterable

Maps in Immutable.js are iterable, meaning that you can map , filter , reduce , etc. each
key-value pair in the map.
let features = Immutable.Map<string, boolean>({
'send-links': true,
'send-files': true,
'local-storage': true,
'mirror-notifications': false,
'api-access': false
});
let myFeatures = features.reduce((providedFeatures, provided, feature) => {
if(provided)
providedFeatures.push(feature);
return providedFeatures;
}, []);
console.log(myFeatures); // [ 'send-links', 'send-files', 'local-storage' ]

const mapMap = Immutable.Map({ a: 0, b: 1, c: 2 });

mapMap.map(i => i * 30);
const mapFilter = Immutable.Map({ a: 0, b: 1, c: 2 });
mapFilter.filter(i => i % 2);
const mapReduce = Immutable.Map({ a: 10, b: 20, c: 30 });
mapReduce.reduce((acc, i) => acc + i, 0);

201

Immutable.List

Immutable.List
List is the immutable version of JavaScript's array structure.

let movies = Immutable.fromJS([ // again use fromJS for deep immutability

{
name: 'The Fellowship of the Ring',
released: 2001,
rating: 8.8
},
{
name: 'The Two Towers',
released: 2002,
rating: 8.7
}
]);
movies = movies.push(Immutable.Map({
name: 'The Return of the King',
released: 2003
}));
movies = movies.update(2, movie => movie.set('rating', 8.9)); // 0 based
movies = movies.zipWith(
(movie, seriesNumber) => movie.set('episode', seriesNumber),
Immutable.Range(1, movies.size + 1) // size property instead of length
);
console.log(movies);
/* writes
List [
Map { "name": "The Fellowship of the Ring", "released": 2001, "rating": 8.8, "episod
e": 1 },
Map { "name": "The Two Towers", "released": 2002, "rating": 8.7, "episode": 2 },
Map { "name": "The Return of the King", "released": 2003, "rating": 8.9, "episode":
3 } ]
*/

Here we use the Immutable.fromJS call again since we have objects stored in the array. We
call push to add items to the list, just like we would call it on an array. But since we're
creating a new copy, we must rebind the variable. We have the same set and update
calls when we want to update items at specific indexes. We also have access to array
functions like map , reduce with support for extras like the one we're using here, zipWith .

202

Immutable.List

203

Performance

Performance
Due to having to allocate memory and copy the data structure whenever any change is
made, this can potentially lead to a large number of extra operations having to be performed
depending on the type and number of changes. To demonstrate the difference, here is a test
run. Doing memory allocation and copy on large strings can be expensive even on a shallow
object.

204

Persistent and Transient Data Structures

Persistent and Transient Data Structures

Immutable data structures are also sometimes referred to as persistent data structures,
since their values persist for their lifetime. Immutable.js provides the option for transient data
structures, which is a mutable version of a persistent data structure during a transformation
stage and returning a new immutable version upon completion. This is one approach to
solving the performance issues we encountered earlier. Let's revisit the immutable case
outlined in the Performance example, but using a transient data structure this time:
import * as Immutable from 'immutable';
let list = Immutable.List();
list = list.withMutations(mutableList => {
let val = "";
return Immutable.Range(0, 1000000)
.forEach(() => {
val += "concatenation";
mutableList.push(val);
});
});
console.log(list.size); // writes 1000000
list.push('');
console.log(list.size); // writes 1000000

As we can see in this performance test, the transient list builder was still a lot slower than
the fully mutable version, but much faster than the fully immutable version. Also, if you pass
the mutable array to Immutable.List or Immutable.fromJS , you'll find the transient version
closes the performance gap. The test also shows how slow Object.freeze can be
compared to the other three.

205

Official Documentation

Official documentation
For more information on Immutable.js, visit the official documentation at
https://facebook.github.io/immutable-js/.

206

Pipes

Pipes

Figure: Pipes by Life-Of-Pix is licensed under Public Domain (https://pixabay.com/en/pipeplumbing-connection-pipeline-406906/)

Angular 2 provides a new way of filtering data: pipes . Pipes are a replacement for Angular
1.x's filters . Most of the built-in filters from Angular 1.x have been converted to Angular 2
pipes; a few other handy ones have been included as well.

207

Using Pipes

Using Pipes
Like a filter, a pipe also takes data as input and transforms it to the desired output. A basic
example of using pipes is shown below:
import {Component} from '@angular/core';
@Component({
selector: 'product-price',
template: `<p>Total price of product is {{ price | currency }}</p>`
})
export class ProductPrice {
price: number = 100.1234;
}

View Example

Passing Parameters
A pipe can accept optional parameters to modify the output. To pass parameters to a pipe,
simply add a colon and the parameter value to the end of the pipe expression:
pipeName: parameterValue

You can also pass multiple parameters this way:

pipeName: parameter1: parameter2

import {Component} from '@angular/core';

@Component({
selector: 'product-price',
template: '<p>Total price of product is {{ price | currency: "CAD": true: "1.2-4"
}}</p>'
})
export class ProductPrice {
price: number = 100.123456;
}

View Example

208

Using Pipes

Chaining Pipes
We can chain pipes together to make use of multiple pipes in one expression.
import {Component} from '@angular/core';
@Component({
selector: 'product-price',
template: '<p>Total price of product is {{ price | currency: "CAD": true: "1.2-4"
| lowercase }}</p>'
})
export class ProductPrice {
price: number = 100.123456;
}

View Example

209

Custom Pipes

Custom Pipes
Angular 2 allows you to create your own custom pipes:
import {Pipe, PipeTransform} from '@angular/core';
const FILE_SIZE_UNITS = ['B', 'KB', 'MB', 'GB', 'TB', 'PB', 'EB', 'ZB', 'YB'];
const FILE_SIZE_UNITS_LONG = ['Bytes', 'Kilobytes', 'Megabytes', 'Gigabytes', 'Pettaby
tes', 'Exabytes', 'Zettabytes', 'Yottabytes'];
@Pipe({
name: 'formatFileSize'
})
export class FormatFileSizePipe implements PipeTransform {
transform(sizeInBytes: number, longForm: boolean): string {
const units = longForm
? FILE_SIZE_UNITS_LONG
: FILE_SIZE_UNITS;
let power = Math.round(Math.log(sizeInBytes)/Math.log(1024));
power = Math.min(power, units.length - 1);
const size = sizeInBytes / Math.pow(1024, power); // size in new units
const formattedSize = Math.round(size * 100) / 100; // keep up to 2 decimals
const unit = units[power];
return `${formattedSize} ${unit}`;
}
}

Each custom pipe implementation must:

have the @Pipe decorator with pipe metadata that has a name property. This value will
be used to call this pipe in template expressions. It must be a valid JavaScript identifier.
implement the PipeTransform interface's transform method. This method takes the
value being piped and a variable number of arguments of any type and return a
transformed ("piped") value.
Each colon-delimited parameter in the template maps to one method argument in the same
order.

210

Custom Pipes

import {Component} from '@angular/core';

@Component({
selector: 'hello',
template: `
<div>
<p *ngFor="let f of fileSizes">{{ f | formatFileSize }}</p>
<p>{{ largeFileSize | formatFileSize:true }}</p>
</div>`
})
export class Hello {
fileSizes = [10, 100, 1000, 10000, 100000, 10000000, 10000000000];
largeFileSize = Math.pow(10, 15)
}

View Example

211

Stateful Pipes

Stateful Pipes
There are two categories of pipes:
Stateless pipes are pure functions that flow input data through without remembering
anything or causing detectable side-effects. Most pipes are stateless. The
CurrencyPipe we used and the length pipe we created are examples of a stateless

pipe.
Stateful pipes are those which can manage the state of the data they transform. A pipe
that creates an HTTP request, stores the response and displays the output, is a stateful
pipe. Stateful Pipes should be used cautiously.
Angular 2 provides AsyncPipe , which is stateful.

AsyncPipe
AsyncPipe can receive a Promise or Observable as input and subscribe to the input
automatically, eventually returning the emitted value(s). It is stateful because the pipe
maintains a subscription to the input and its returned values depend on that subscription.
import {Component} from '@angular/core';
import {Observable} from 'rxjs/Observable';
@Component({
selector: 'product-price',
template: `
<p>Total price of product is {{fetchPrice | async | currency: "CAD": true: "1.2-2"
}}</p>
<p>Seconds: {{seconds | async}}</p>
`
})
export class ProductPrice {
count: number = 0;
fetchPrice: Promise<number> = new Promise((resolve, reject) => {
setTimeout(() => resolve(10), 500);
});
seconds: Observable<number> = new Observable(observer => {
setInterval(() => { observer.next(this.count++); }, 1000);
});
}

212

Stateful Pipes

View Example

Implementing Stateful Pipes

Pipes are stateless by default. We must declare a pipe to be stateful by setting the pure
property of the @Pipe decorator to false. This setting tells Angulars change detection
system to check the output of this pipe each cycle, whether its input has changed or not.

213

Stateful Pipes

import {Pipe, PipeTransform, ChangeDetectorRef} from '@angular/core';

import {Observable} from 'rxjs/Observable';
import 'rxjs/add/observable/timer';
import 'rxjs/add/operator/take';
/**
* On number change, animates from `oldNumber` to `newNumber`
*/
// naive implementation assumes small number increments
@Pipe({
name: 'animateNumber',
pure: false
})
export class AnimateNumberPipe implements PipeTransform {
private currentNumber: number = null; // intermediary number
private newNumber: number = null;
private subscription;
constructor(private cdRef: ChangeDetectorRef) {}
transform(newNumber: number): string {
if (this.newNumber === null) { // set inital value
this.currentNumber = this.newNumber = newNumber;
}
if (newNumber !== this.newNumber) {
if (this.subscription) {
this.currentNumber = this.newNumber;
this.subscription.unsubscribe();
}
this.newNumber = newNumber;
const oldNumber = this.currentNumber;
const direction = ((newNumber - oldNumber) > 0) ? 1 : -1;
const numbersToCount = Math.abs(newNumber - oldNumber) + 1;
this.subscription = Observable.timer(0, 100) // every 100 ms
.take(numbersToCount)
.subscribe(
() => {
this.currentNumber += direction;
this.cdRef.markForCheck();
},
null,
() => this.subscription = null
);
}
return this.currentNumber;
}
}

View Example

214

Stateful Pipes

215

Forms

Forms
An application without user input is just a page. Capturing input from the user is the
cornerstone of any application. In many cases, this means dealing with forms and all of their
complexities.
Angular 2 is much more flexible than Angular 1 for handling forms we are no longer
restricted to relying solely on ngModel . Instead, we are given degrees of simplicity and
power, depending on the form's purpose.
Template-Driven Forms use built-in directives to create straightforward form
components with minimal code.
FormBuilder uses the provided APIs to handle more complex validation and subforms.

216

Getting Started

Getting Started
Opt-In APIs
Before we dive into any of the form features, we need to do a little bit of housekeeping. We
need to bootstrap our application using the FormsModule and/or ReactiveFormsModule .
import { platformBrowserDynamic } from '@angular/platform-browser-dynamic'
import { ReactiveFormsModule, FormsModule } from '@angular/forms';
import { MyApp } from './components'
@NgModule({
imports: [
BrowserModule,
FormsModule,
ReactiveFormsModule
],
declarations: [MyApp],
bootstrap: [MyApp]
})
export class MyAppModule {
}
platformBrowserDynamic().bootstrapModule(MyAppModule)

Input Labeling
Most of the form examples use the following HTML5 style for labeling inputs:
<label for="name">Name</label>
<input type="text" name="username" id="name">

Angular 2 also supports the alternate HTML5 style, which precludes the necessity of id s
on <input> s:
<label>
Name
<input type="text" name="username">
</label>

217

Getting Started

218

Template-Driven Forms

Template-Driven Forms
The most straightforward approach to building forms in Angular 2 is to take advantage of the
directives provided for you.
First, consider a typical form:
<form method="POST" action="/register" id="SignupForm">
<label for="email">Email</label>
<input type="text" name="email" id="email">
<label for="password">Password</label>
<input type="password" name="password" id="password">
<button type="submit">Sign Up</button>
</form>

Angular 2 has already provided you a form directive, and form related directives such as
input, etc which operates under the covers. For a basic implementation, we just have to add
a few attributes and make sure our component knows what to do with the data.
index.html
<signup-form>Loading...</signup-form>

signup-form.component.html
<form #signupForm="ngForm" (ngSubmit)="registerUser(signupForm)">
<label for="email">Email</label>
<input type="text" name="email" id="email" ngModel>
<label for="password">Password</label>
<input type="password" name="password" id="password" ngModel>
<button type="submit">Sign Up</button>
</form>

signup-form.component.ts

219

Template-Driven Forms

import { Component } from '@angular/core';

import { NgForm } from '@angular/forms';
@Component({
selector: 'signup-form',
templateUrl: 'app/signup-form.component.html',
})
export class SignupForm {
registerUser(form: NgForm) {
console.log(form.value);
// {email: '...', password: '...'}
// ...
}
}

220

Nesting Form Data

Nesting Form Data

If you find yourself wrestling to fit nested trees of data inside of a flat form, Angular has you
covered for both simple and complex cases.
Let's assume you had a payment endpoint which required data, similar to the following:
{
"contact": {
"firstname": "Bob",
"lastname": "McKenzie",
"email": "[email protected]",
"phone": "555-TAKE-OFF"
},
"address": {
"street": "123 Some St",
"city": "Toronto",
"region": "ON",
"country": "CA",
"code": "H0H 0H0"
},
"paymentCard": {
"provider": "Credit Lending Company Inc",
"cardholder": "Doug McKenzie",
"number": "123 456 789 012",
"verification": "321",
"expiry": "2020-02"
}
}

While forms are flat and one-dimensional, the data built from them is not. This leads to
complex transforms to convert the data youve been given into the shape you need.
Worse, in cases where it is possible to run into naming collisions in form inputs, you might
find yourself using long and awkward names for semantic purposes.

221

Nesting Form Data

<form >
<fieldset>
<legend>Contact</legend>
<label for="contact_first-name">First Name</label>
<input type="text" name="contact_first-name" id="contact_first-name">
<label for="contact_last-name">Last Name</label>
<input type="text" name="contact_last-name" id="contact_last-name">
<label for="contact_email">Email</label>
<input type="email" name="contact_email" id="contact_email">
<label for="contact_phone">Phone</label>
<input type="text" name="contact_phone" id="contact_phone">
</fieldset>
<!-- ... -->
</form>

A form handler would have to convert that data into a form that your API expects. Thankfully,
this is something Angular 2 has a solution for.

ngModelGroup
When building a template-driven form in Angular 2, we can lean on the ngModelGroup
directive to arrive at a cleaner implementation, while Angular does the heavy lifting of
converting form-fields into nested data.

222

Nesting Form Data

<form #paymentForm="ngForm" (ngSubmit)="purchase(paymentForm)">

<fieldset ngModelGroup="contact">
<legend>Contact</legend>
<label>
First Name <input type="text" name="firstname" ngModel>
</label>
<label>
Last Name <input type="text" name="lastname" ngModel>
</label>
<label>
Email <input type="email" name="email" ngModel>
</label>
<label>
Phone <input type="text" name="phone" ngModel>
</label>
</fieldset>
<fieldset ngModelGroup="address">
<!-- ... -->
</fieldset>
<fieldset ngModelGroup="paymentCard">
<!-- ... -->
</fieldset>
</form>

Using the alternative HTML5 labeling format; IDs have no bearing on the ngForm /
ngModel paradigm

Aside from semantic purposes, ngModelGroup does not have to be used on

<fieldset> it would work just as well on a <div> .

If we were to fill out the form, it would end up in the shape we need for our API, and we can
still rely on the HTML field validation if we know its available.

223

Using Template Model Binding

Using Template Model Binding

One-Way Binding
If you need a form with default values, you can start using the value-binding syntax for
ngModel.
app/signup-form.component.html
<form #signupForm="ngForm" (ngSubmit)=register(signupForm)>
<label for="username">Username</label>
<input type="text" name="username" id="username" [ngModel]="generatedUser">
<label for="email">Email</label>
<input type="email" name="email" id="email" ngModel>
<button type="submit">Sign Up</button>
</form>

app/signup-form.component.ts
import { Component } from '@angular/core';
import { NgForm } from '@angular/forms';
// ...
@Component({ /* ... */ })
export class SignupForm {
generatedUser: string = generateUniqueUserID();
register(form: NgForm) {
console.log(form.value);
// ...
}
}

Two-Way Binding
While Angular 2 assumes one-way binding by default, two-way binding is still available if you
need it.
In order to have access to two-way binding in template-driven forms, use the Banana-Box
syntax ( [(ngModel)]="propertyName" ).

224

Using Template Model Binding

Be sure to declare all of the properties you will need on the component.
<form #signupForm="ngForm" (ngSubmit)=register(signupForm)>
<label for="username">Username</label>
<input type="text" name="username" id="username" [(ngModel)]="username">
<label for="email">Email</label>
<input type="email" name="email" id="email" [(ngModel)]="email">
<button type="submit">Sign Up</button>
</form>

import { Component } from '@angular/core';

import { NgForm } from '@angular/forms';
// ...
@Component({ /* ... */ })
export class SignUpForm {
username: string = generateUniqueUserID();
email: string = '';
register(form: NgForm) {
console.log(form.value.username);
console.log(this.username);
// ...
}
}

225

Validating Template-Driven Forms

Validating Template-Driven Forms

Validation
Using the template-driven approach, form validation is a matter of following HTML5
practices:
<!-- a required field -->
<input type="text" required>
<!-- an optional field of a specific length -->
<input type="text" pattern=".{3,8}">
<!-- a non-optional field of specific length -->
<input type="text" pattern=".{3,8}" required>
<!-- alphanumeric field of specific length -->
<input type="text" pattern="[A-Za-z0-9]{0,5}">

Note that the pattern attribute is a less-powerful version of JavaScript's RegEx syntax.
There are other HTML5 attributes which can be learned and applied to various types of
input; however in most cases they act as upper and lower limits, preventing extra information
from being added or removed.
<!-- a field which will accept no more than 5 characters -->
<input type="text" maxlength="5">

You can use one or both of these methods when writing a template-driven form. Focus on
the user experience: in some cases, it makes sense to prevent accidental entry, and in
others it makes sense to allow unrestricted entry but provide something like a counter to
show limitations.

226

FormBuilder

Form Builder
While using directives in our templates gives us the power of rapid prototyping without too
much boilerplate, we are restricted in what we can do. The FormBuilder , on the other hand,
lets us define our form through code and gives us much more flexibility and control over data
validation.
There is a little bit of magic in its simplicity at first, but after you're comfortable with the
basics, learning its building blocks will allow you to handle more complex use cases.

227

FormBuilder Basics

FormBuilder Basics
To begin with FormBuilder, we must first ensure we are working with the right directives and
the right classes in order to take advantage of procedural forms. For this, we need to ensure
that the FormsModule and ReactiveFormsModule was imported in the bootstrap phase of the
application module.
This will give us access to components, directives and providers like FormBuilder ,
FormGroup , and FormControl

In our case, to build a login form, we're looking at something like the following:
app/login-form.component.ts
import {Component} from '@angular/core';
import {
FormBuilder,
FormControl
} from '@angular/forms';
@Component({
selector: 'login-form',
templateUrl: 'app/login-form.component.html',
})
export class LoginForm {
loginForm: FormGroup;
username: FormControl;
password: FormControl;
constructor(builder: FormBuilder) {
this.username = new FormControl('', []);
this.password = new FormControl('', []);
this.loginForm = builder.group({
username: this.username,
password: this.password
});
}
login() {
console.log(this.loginForm.value);
// Attempt Logging in...
}
}

app/login-form.component.html

228

FormBuilder Basics

<form [formGroup]="loginForm" (ngSubmit)="login()">

<label for="username">username</label>
<input type="text" name="username" id="username" [formControl]="username">
<label for="password">password</label>
<input type="password" name="password" id="password" [formControl]="password">
<button type="submit">log in</button>
</form>

View Example

FormControl
Note that the FormControl class is assigned to similarly named fields, both on this and in
the FormBuilder#group({ }) method. This is mostly for ease of access. By saving references
to the FormControl instances on this , you can access the inputs in the template without
having to reference the form itself. The form fields can otherwise be reached in the template
by using loginForm.controls.username and loginForm.controls.password . Likewise, any
instance of FormControl in this situation can access its parent group by using its .root
property (e.g. username.root.controls.password ).
Make sure that root and controls exist before they're used.
A FormControl requires two properties: an initial value and a list of validators. Right now, we
have no validation. This will be added in the next steps.

229

Validating FormBuilder Forms

Validating FormBuilder Forms

Building from the previous login form, we can quickly and easily add validation.
Angular 2 provides many validators out of the box. They can be imported along with the rest
of dependencies for procedural forms.
app/login-form.component.ts
import {
// ...
Validators
} from '@angular/forms';
@Component({ /* ... */ })
export class LoginForm {
loginForm: FormGroup;
username: FormControl;
password: FormControl;
constructor(builder: FormBuilder) {
this.username = new FormControl('', [
Validators.required,
Validators.minLength(5)
]);
this.password = new FormControl('', [Validators.required]);
this.loginForm = builder.group({
username: this.username,
password: this.password
});
}
login() {
console.log(this.loginForm.value);
// Attempt Logging in...
}
}

app/login-form.component.html

230

Validating FormBuilder Forms

<form [formGroup]="loginForm" (ngSubmit)="login()">

Inside the form.
<div>
<label for="username">username</label>
<input type="text" name="username" id="username" [formControl]="username">
<div [hidden]="username.valid || username.untouched">
<div>The following problems have been found with the username:</div>
<div [hidden]="!username.hasError('minlength')">Username can not be shorter than
5 characters.</div>
<div [hidden]="!username.hasError('required')">Username is required.</div>
</div>
</div>
<div >
<label for="password">password</label>
<input type="password" name="password" id="password" [formControl]="password">
<div [hidden]="password.valid || password.untouched">
<div>The following problems have been found with the password:</div>
<div [hidden]="!password.hasError('required')">The password is required.</div>
</div>
</div>
<button type="submit" [disabled]="!loginForm.valid">Log In</button>
</form>

Note that we have added rather robust validation on both the fields and the form itself, using
nothing more than built-in validators and some template logic.
View Example
We are using .valid and .untouched to determine if we need to show errors - while the
field is required, there is no reason to tell the user that the value is wrong if the field hasn't
been visited yet.
For built-in validation, we are calling .hasError() on the form element, and we are passing
a string which represents the validator function we included. The error message only
displays if this test returns true.

231

FormBuilder Custom Validation

FormBuilder Custom Validation

As useful as the built-in validators are, it is very useful to be able to include your own.
Angular 2 allows you to do just that, with minimal effort.
Let's assume we are using the same Login Form, but now we also want to test that our
password has an exclamation mark somewhere in it.
app/login-form.component.ts
function hasExclamationMark(input: FormControl) {
const hasExclamation = input.value.indexOf('!') >= 0;
return hasExclamation ? null : { needsExclamation: true };
}

// ...
this.password = new FormControl('', [
Validators.required,
hasExclamationMark
]);

A simple function takes the FormControl instance and returns null if everything is fine. If the
test fails, it returns an object with an arbitrarily named property. The property name is what
will be used for the .hasError() test.
app/login-form.component.ts
<!-- ... -->
<div [hidden]="!password.hasError('needsExclamation')">
Your password must have an exclamation mark!
</div>
<!-- ... -->

View Example

Predefined Parameters
Having a custom validator to check for exclamation marks might be helpful, but what if you
need to check for some other form of punctuation? It might be overkill to write nearly the
same thing over and over again.

232

FormBuilder Custom Validation

Consider the earlier example Validators.minLength(5) . How did they get away with allowing
an argument to control the length, if a validator is just a function? Simple, really. It's not a
trick of Angular, or TypeScript - it's simple JavaScript closures.
function minLength(minimum) {
return function(input) {
return input.value.length >= minimum ? null : { minLength: true };
};
}

Assume you have a function which takes a "minimum" parameter and returns another
function. The function defined and returned from the inside becomes the validator. The
closure reference allows you to remember the value of the minimum when the validator is
eventually called.
Let's apply that thinking back to a PunctuationValidator .
app/login-form.component.ts
function hasPunctuation(punctuation: string, errorType: string) {
return function(input: FormControl) {
return input.value.indexOf(punctuation) >= 0 ?
null :
{ [errorType]: true };
};
}
// ...
this.password = new FormControl('', [
Validators.required,
hasPunctuation('&', 'ampersandRequired')
]);

app/login-form.component.html
<!-- ... -->
<div [hidden]="!password.hasError('ampersandRequired')">
You must have an &amp; in your password.
</div>
<!-- ... -->

View Example

Validating Inputs Using Other Inputs

233

FormBuilder Custom Validation

Keep in mind what was mentioned earlier: inputs have access to their parent context via
.root . Therefore, complex validation can happen by drilling through the form, via root.

function duplicatePassword(input: FormControl) {

if (!input.root || !input.root.controls) {
return null;
}
const exactMatch = input.root.controls.password === input.value;
return exactMatch ? null : { mismatchedPassword: true };
}
// ...
this.duplicatePassword = new FormControl('', [
Validators.required,
duplicatePassword
]);

View Example

234

Visual Cues for Users

Visual Cues for Users

HTML5 provides :invalid and :valid pseudo-selectors for its input fields.
input[type="text"]:valid {
border: 2px solid green;
}
input[type="text"]:invalid {
border: 2px solid red;
}

Unfortunately, this system is rather unsophisticated and would require more manual effort in
order to work with complex forms or user behavior.
Rather than writing extra code, and creating and enforcing your own CSS classes, to
manage these behaviors, Angular 2 provides you with several classes, already accessible
on your inputs.
/* field value is valid */
.ng-valid {}
/* field value is invalid */
.ng-invalid {}
/* field has not been clicked in, tapped on, or tabbed over */
.ng-untouched {}
/* field has been previously entered */
.ng-touched {}
/* field value is unchanged from the default value */
.ng-pristine {}
/* field value has been modified from the default */
.ng-dirty {}

Note the three pairs:

valid / invalid
untouched / touched
pristine / dirty

235

Visual Cues for Users

These pairs can be used in many combinations in your CSS to change style based on the
three separate flags they represent. Angular will toggle between the pairs on each input as
the state of the input changes.
/* field has been unvisited and unchanged */
input.ng-untouched.ng-pristine {}
/* field has been previously visited, and is invalid */
input.ng-touched.ng-invalid {}

.ng-untouched will not be replaced by .ng-touched until the user leaves the input for

the first time

For templating purposes, Angular also gives you access to the unprefixed properties on the
input, in both code and template:
<input name="myInput" [formControl]="myCustomInput">
<div [hidden]="myCustomInput.pristine">I've been changed</div>

236

Modules

Modules
Angular Modules provides a mechanism for creating blocks of functionality that can be
combined to build an application.

Figure: Used Lego Duplo Bricks by Arto Alanenp is licensed under CC BY-SA 4.0
(https://commons.wikimedia.org/wiki/File:Lego_dublo_arto_alanenpaa_5.JPG)

237

What is an Angular 2 Module?

What is an Angular 2 Module?

In Angular 2, a module is a mechanism to group components, directives, pipes and services
that are related, in such a way that can be combined with other modules to create an
application. An Angular 2 application can be thought of as a puzzle where each piece (or
each module) is needed to be able to see the full picture.
Another analogy to understand Angular 2 modules is classes. In a class, we can define
public or private methods. The public methods are the API that other parts of our code can
use to interact with it while the private methods are implementation details that are hidden.
In the same way, a module can export or hide components, directives, pipes and services.
The exported elements are meant to be used by other modules, while the ones that are not
exported (hidden) are just used inside the module itself and cannot be directly accessed by
other modules of our application.

A Basic Use of Modules

To be able to define modules we have to use the decorator NgModule .
import { NgModule } from '@angular/core';
@NgModule({
imports: [ ... ],
declarations: [ ... ],
bootstrap: [ ... ]
})
export class AppModule { }

In the example above, we have turned the class AppModule into an Angular 2 module just by
using the NgModule decorator. The NgModule decorator requires at least three properties:
imports , declarations and bootstrap .

The property imports expects an array of modules. Here's where we define the pieces of
our puzzle (our application). The property declarations expects an array of components,
directives and pipes that are part of the module. The bootstrap property is where we define
the root component of our module. Even though this property is also an array, 99% of the
time we are going to define only one component.
There are very special circumstances where more than one component may be
required to bootstrap a module but we are not going to cover those edge cases here.

238

What is an Angular 2 Module?

Here's how a basic module made up of just one component would look like:
app/app.component.ts
import { Component } from '@angular/core';
@Component({
selector: 'rio-app',
template: '<h1>My Angular 2 App</h1>'
})
export class AppComponent {}

app/app.module.ts
import { NgModule } from '@angular/core';
import { BrowserModule } from '@angular/platform-browser';
import { AppComponent } from './app.component';
@NgModule({
imports: [BrowserModule],
declarations: [AppComponent],
bootstrap: [AppComponent]
})
export class AppModule { }

The file app.component.ts is just a "hello world" component, nothing interesting there. In the
other hand, the file app.module.ts is following the structure that we've seen before for
defining a module but in this case, we are defining the modules and components that we are
going to be using.
The first thing that we notice is that our module is importing the BrowserModule as an explicit
dependency. The BrowserModule is a built-in module that exports basic directives, pipes and
services. Unlike previous versions of Angular 2, we have to explicitly import those
dependencies to be able to use directives like *ngFor or *ngIf in our templates.
Given that the root (and only) component of our module is the AppComponent we have to list
it in the bootstrap array. Because in the declarations property we are supposed to define
all the components or pipes that make up our application, we have to define the
AppComponent again there too.

Before moving on, there's an important clarification to make. There are two types of
modules, root modules and feature modules.

239

What is an Angular 2 Module?

In the same way that in a module we have one root component and many possible
secondary components, in an application we only have one root module and zero or
many feature modules. To be able to bootstrap our application, Angular needs to know
which one is the root module. An easy way to identify a root module is by looking at the
imports property of its NgModule decorator. If the module is importing the BrowserModule

then it's a root module, if instead is importing the CommonModule then it is a feature module.
As developers, we need to take care of importing the BrowserModule in the root module
and instead, import the CommonModule in any other module we create for the same
application. Failing to do so might result in problems when working with lazy loaded
modules as we are going to see in following sections.
By convention, the root module should always be named AppModule .

Bootstrapping an Application
To bootstrap our module based application, we need to inform Angular which one is our root
module to perform the compilation in the browser. This compilation in the browser is also
known as "Just in Time" (JIT) compilation.
main.ts
import { platformBrowserDynamic } from '@angular/platform-browser-dynamic';
import { AppModule } from './app/app.module';
platformBrowserDynamic().bootstrapModule(AppModule);

It is also possible to perform the compilation as a build step of our workflow. This
method is called "Ahead of Time" (AOT) compilation and will require a slightly different
bootstrap process that we are going to discuss in another section.
View Example
In the next section we are going to see how to create a module with multiple components,
services and pipes.

240

Adding Components, Pipes and Services to a Module

Adding Components, Pipes and Services

to a Module
In the previous section, we learned how to create a module with just one component but we
know that is hardly the case. Our modules are usually made up of multiple components,
services, directives and pipes. In this chapter we are going to extend the example we had
before with a custom component, pipe and service.
Let's start by defining a new component that we are going to use to show credit card
information.
credit-card.component.ts
import { Component, OnInit } from '@angular/core';
import { CreditCardService } from './credit-card.service';
@Component({
selector: 'rio-credit-card',
template: `
<p>Your credit card is: {{ creditCardNumber | creditCardMask }}</p>
`
})
export class CreditCardComponent implements OnInit {
creditCardNumber: string;
constructor(private creditCardService: CreditCardService) {}
ngOnInit() {
this.creditCardNumber = this.creditCardService.getCreditCard();
}
}

This component is relying on the CreditCardService to get the credit card number, and on
the pipe creditCardMask to mask the number except the last 4 digits that are going to be
visible.
credit-card.service.ts

241

Adding Components, Pipes and Services to a Module

import { Injectable } from '@angular/core';

@Injectable()
export class CreditCardService {
getCreditCard(): string {
return '2131313133123174098';
}
}

credit-card-mask.pipe.ts
import { Pipe, PipeTransform } from '@angular/core';
@Pipe({
name: 'creditCardMask'
})
export class CreditCardMaskPipe implements PipeTransform {
transform(plainCreditCard: string): string {
const visibleDigits = 4;
let maskedSection = plainCreditCard.slice(0, -visibleDigits);
let visibleSection = plainCreditCard.slice(-visibleDigits);
return maskedSection.replace(/./g, '*') + visibleSection;
}
}

With everything in place, we can now use the CreditCardComponent in our root component.
app.component.ts
import { Component } from "@angular/core";
@Component({
selector: 'rio-app',
template: `
<h1>My Angular 2 App</h1>
<rio-credit-card></rio-credit-card>
`
})
export class AppComponent {}

Of course, to be able to use this new component, pipe and service, we need to update our
module, otherwise Angular is not going to be able to compile our application.
app.module.ts

242

Adding Components, Pipes and Services to a Module

import { NgModule } from '@angular/core';

import { BrowserModule } from '@angular/platform-browser';
import { AppComponent } from './app.component';
import { CreditCardMaskPipe } from './credit-card-mask.pipe';
import { CreditCardService } from './credit-card.service';
import { CreditCardComponent } from './credit-card.component';
@NgModule({
imports: [BrowserModule],
providers: [CreditCardService],
declarations: [
AppComponent,
CreditCardMaskPipe,
CreditCardComponent
],
bootstrap: [AppComponent]
})
export class AppModule { }

Notice that we have added the component CreditCardComponent and the pipe
CreditCardMaskPipe to the declarations property, along with the root component of the

module AppComponent . In the other hand, our custom service is configured with the
dependency injection system with the providers property.
View Example
Be aware that this method of defining a service in the providers property should only be
used in the root module. Doing this in a feature module is going to cause unintended
consequences when working with lazy loaded modules.
In the next section, we are going to see how to safely define services in feature modules.

243

Creating a Feature Module

Creating a Feature Module

When our root module start growing, it starts to be evident that some elements
(components, directives, etc.) are related in a way that almost feel like they belong to a
library that can be "plugged in".
In our previous example, we started to see that. Our root module has a component, a pipe
and a service that its only purpose is to deal with credit cards. What if we extract these three
elements to their own feature module and then we import it into our root module?
We are going to do just that. The first step is to create two folders to differentiate the
elements that belong to the root module from the elements that belong to the feature
module.
.
app
app.component.ts
app.module.ts
credit-card
credit-card-mask.pipe.ts
credit-card.component.ts
credit-card.module.ts
credit-card.service.ts
index.html
main.ts

Notice how each folder has its own module file: app.module.ts and credit-card.module.ts.
Let's focus on the latter first.
credit-card/credit-card.module.ts

244

Creating a Feature Module

import { NgModule } from '@angular/core';

import { CommonModule } from '@angular/common';
import { CreditCardMaskPipe } from './credit-card-mask.pipe';
import { CreditCardService } from './credit-card.service';
import { CreditCardComponent } from './credit-card.component';
@NgModule({
imports: [CommonModule],
declarations: [
CreditCardMaskPipe,
CreditCardComponent
],
providers: [CreditCardService],
exports: [CreditCardComponent]
})
export class CreditCardModule {}

Our feature CreditCardModule it's pretty similar to the root AppModule with a few important
differences:
We are not importing the BrowserModule but the CommonModule . If we see the
documentation of the BrowserModule here, we can see that it's re-exporting the
CommonModule with a lot of other services that helps with rendering an Angular 2

application in the browser. These services are coupling our root module with a particular
platform (the browser), but we want our feature modules to be platform independent.
That's why we only import the CommonModule there, which only exports common
directives and pipes.
When it comes to components, pipes and directives, every module should import its
own dependencies disregarding if the same dependencies were imported in the root
module or in any other feature module. In short, even when having multiple feature
modules, each one of them needs to import the CommonModule .
We are using a new property called exports . Every element defined in the
declarations array is private by default. We should only export whatever the other

modules in our application need to perform its job. In our case, we only need to make
the CreditCardComponent visible because it's being used in the template of the
AppComponent .

app/app.component.ts

245

Creating a Feature Module

...
@Component({
...
template: `
...
<rio-credit-card></rio-credit-card>
`
})
export class AppComponent {}

We are keeping the CreditCardMaskPipe private because it's only being used inside the
CreditCardModule and no other module should use it directly.

We can now import this feature module into our simplified root module.
app/app.module.ts
import { NgModule } from '@angular/core';
import { BrowserModule } from '@angular/platform-browser';
import { CreditCardModule } from '../credit-card/credit-card.module';
import { AppComponent } from './app.component';
@NgModule({
imports: [
BrowserModule,
CreditCardModule
],
declarations: [AppComponent],
bootstrap: [AppComponent]
})
export class AppModule { }

At this point we are done and our application behaves as expected.

View Example

Services and Lazy Loaded Modules

Here's the tricky part of Angular modules. While components, pipes and directives are
scoped to its modules unless explicitly exported, services are globally available unless the
module is lazy loaded.
It's hard to understand that at first so let's try to see what's happening with the
CreditCardService in our example. Notice first that the service is not in the exports array

but in the providers array. With this configuration, our service is going to be available

246

Creating a Feature Module

everywhere, even in the AppComponent which lives in another module. So, even when using
modules, there's no way to have a "private" service unless... the module is being lazy
loaded.
When a module is lazy loaded, Angular is going to create a child injector (which is a child of
the root injector from the root module) and will create an instance of our service there.
Imagine for a moment that our CreditCardModule is configured to be lazy loaded. With our
current configuration, when the application is bootstrapped and our root module is loaded in
memory, an instance of the CreditCardService (a singleton) is going to be added to the root
injector. But, when the CreditCardModule is lazy loaded sometime in the future, a child
injector will be created for that module with a new instance of the CreditCardService . At
this point we have a hierarchical injector with two instances of the same service, which is
not usually what we want.
Think for example of a service that does the authentication. We want to have only one
singleton in the entire application, disregarding if our modules are being loaded at bootstrap
or lazy loaded. So, in order to have our feature module's service only added to the root
injector, we need to use a different approach.
credit-card/credit-card.module.ts
import { NgModule, ModuleWithProviders } from '@angular/core';
/* ...other imports... */
@NgModule({
imports: [CommonModule],
declarations: [
CreditCardMaskPipe,
CreditCardComponent
],
exports: [CreditCardComponent]
})
export class CreditCardModule {
static forRoot(): ModuleWithProviders {
return {
ngModule: CreditCardModule,
providers: [CreditCardService]
}
}
}

Different than before, we are not putting our service directly in the property providers of the
NgModule decorator. This time we are defining a static method called forRoot where we

define the module and the service we want to export.

With this new syntax, our root module is slightly different.

247

Creating a Feature Module

app/app.module.ts
/* ...imports... */
@NgModule({
imports: [
BrowserModule,
CreditCardModule.forRoot()
],
declarations: [AppComponent],
bootstrap: [AppComponent]
})
export class AppModule { }

Can you spot the difference? We are not importing the CreditCardModule directly, instead
what we are importing is the object returned from the forRoot method, which includes the
CreditCardService . Although this syntax is a little more convoluted than the original, it will

guarantee us that only one instance of the CreditCardService is added to the root module.
When the CreditCardModule is loaded (even lazy loaded), no new instance of that service is
going to be added to the child injector.
View Example
As a rule of thumb, always use the forRoot syntax when exporting services from
feature modules, unless you have a very special need that requires multiple instances at
different levels of the dependency injection tree.

248

Directive Duplications

Directive Duplications
Because we no longer define every component and directive directly in every component
that needs it, we need to be aware of how Angular modules handle directives and
components that target the same element (have the same selector).
Let's assume for a moment that by mistake, we have created two directives that target the
same property:
This example is a variation of the code found in the official documentation.
blue-highlight.directive.ts
import { Directive, ElementRef, Renderer } from '@angular/core';
@Directive({
selector: '[highlight]'
})
export class BlueHighlightDirective {
constructor(renderer: Renderer, el: ElementRef) {
renderer.setElementStyle(el.nativeElement, 'backgroundColor', 'blue');
renderer.setElementStyle(el.nativeElement, 'color', 'gray');
}
}

yellow-highlight.directive.ts
import { Directive, ElementRef, Renderer } from '@angular/core';
@Directive({
selector: '[highlight]'
})
export class YellowHighlightDirective {
constructor(renderer: Renderer, el: ElementRef) {
renderer.setElementStyle(el.nativeElement, 'backgroundColor', 'yellow');
}
}

These two directives are similar, they are trying to style an element. The
BlueHighlightDirective will try to set the background color of the element to blue while

changing the color of the text to gray, while the YellowHighlightDirective will try only to
change the background color to yellow. Notice that both are targeting any HTML element
that has the property highlight . What would happen if we add both directives to the same
module?

249

Directive Duplications

app.module.ts
// Imports
@NgModule({
imports: [BrowserModule],
declarations: [
AppComponent,
BlueHighlightDirective,
YellowHighlightDirective
],
bootstrap: [AppComponent]
})
export class AppModule { }

Let's see how we would use it in the only component of the module.
app.component.ts
import { Component } from "@angular/core";
@Component({
selector: 'rio-app',
template: '<h1 highlight>My Angular 2 App</h1>'
})
export class AppComponent {}

We can see that in the template of our component, we are using the directive highlight in
our h1 element but, which styles are going to end up being applied?
The answer is: the text is going to be gray and the background yellow.
View Example
We are allowed to define multiple directives that target the same elements in the same
module. What's going to happen is that Angular is going to do every transformation in order.
declarations: [
...,
BlueHighlightDirective,
YellowHighlightDirective
]

Because we have defined both directives in an array, and arrays are ordered collection of
items, when the compiler finds an element with the property highlight , it will first apply the
transformations of BlueHighlightDirective , setting the text gray and the background blue,

250

Directive Duplications

and then will apply the transformations of YellowHighlightDirective , changing again the
background color to yellow.
In summary, when two or more directives target the same element, they are going to
be applied in the order they were defined.

251

Lazy Loading a Module

Lazy Loading a Module

Another advantage of using modules to group related pieces of functionality of our
application is the ability to load those pieces on demand. Lazy loading modules helps us
decrease the startup time. With lazy loading our application does not need to load everything
at once, it only needs to load what the user expects to see when the app first loads. Modules
that are lazily loaded will only be loaded when the user navigates to their routes.
To show this relationship, let's start by defining a simple module that will act as the root
module of our example application.
app/app.module.ts
import { NgModule } from '@angular/core';
import { BrowserModule } from '@angular/platform-browser';
import { AppComponent } from './app.component';
import { EagerComponent } from './eager.component';
import { routing } from './app.routing';
@NgModule({
imports: [
BrowserModule,
routing
],
declarations: [
AppComponent,
EagerComponent
],
bootstrap: [AppComponent]
})
export class AppModule {}

So far this is a very common module that relies on the BrowserModule , has a routing
mechanism and two components: AppComponent and EagerComponent . For now, let's focus
on the root component of our application ( AppComponent ) where the navigation is defined.
app/app.component.ts

252

Lazy Loading a Module

import { Component } from '@angular/core';

@Component({
selector: 'my-app',
template: `
<h1>My App</h1>
<nav>
<a routerLink="eager">Eager</a>
<a routerLink="lazy">Lazy</a>
</nav>
<router-outlet></router-outlet>
`
})
export class AppComponent {}

Our navigation system has only two paths: eager and lazy . To know what those paths are
loading when clicking on them we need to take a look at the routing object that we passed
to the root module.
app/app.routing.ts
import { ModuleWithProviders } from '@angular/core';
import { Routes, RouterModule } from '@angular/router';
import { EagerComponent } from './eager.component';
const routes: Routes = [
{ path: '', redirectTo: 'eager', pathMatch: 'full' },
{ path: 'eager', component: EagerComponent },
{ path: 'lazy', loadChildren: 'lazy/lazy.module#LazyModule' }
];
export const routing: ModuleWithProviders = RouterModule.forRoot(routes);

Here we can see that the default path in our application is called eager which will load
EagerComponent .

app/eager.component.ts
import { Component } from '@angular/core';
@Component({
template: '<p>Eager Component</p>'
})
export class EagerComponent {}

253

Lazy Loading a Module

But more importantly, we can see that whenever we try to go to the path lazy , we are
going to lazy load a module conveniently called LazyModule . Look closely at the definition of
that route:
{ path: 'lazy', loadChildren: 'lazy/lazy.module#LazyModule' }

There's a few important things to notice here:

1. We use the property loadChildren instead of component .
2. We pass a string instead of a symbol to avoid loading the module eagerly.
3. We define not only the path to the module but the name of the class as well.
There's nothing special about LazyModule other than it has its own routing and a
component called LazyComponent .
app/lazy/lazy.module.ts
import { NgModule } from '@angular/core';
import { LazyComponent } from './lazy.component';
import { routing } from './lazy.routing';
@NgModule({
imports: [routing],
declarations: [LazyComponent]
})
export class LazyModule {}

If we define the class LazyModule as the default export of the file, we don't need to
define the class name in the loadChildren property as shown above.
The routing object is very simple and only defines the default component to load when
navigating to the lazy path.
app/lazy/lazy.routing.ts
import { ModuleWithProviders } from '@angular/core';
import { Routes, RouterModule } from '@angular/router';
import { LazyComponent } from './lazy.component';
const routes: Routes = [
{ path: '', component: LazyComponent }
];
export const routing: ModuleWithProviders = RouterModule.forChild(routes);

254

Lazy Loading a Module

Notice that we use the method call forChild instead of forRoot to create the routing
object. We should always do that when creating a routing object for a feature module, no
matter if the module is supposed to be eagerly or lazily loaded.
Finally, our LazyComponent is very similar to EagerComponent and is just a placeholder for
some text.
app/lazy/lazy.component.ts
import { Component } from '@angular/core';
@Component({
template: '<p>Lazy Component</p>'
})
export class LazyComponent {}

View Example
When we load our application for the first time, the AppModule along the AppComponent will
be loaded in the browser and we should see the navigation system and the text "Eager
Component". Until this point, the LazyModule has not being downloaded, only when we click
the link "Lazy" the needed code will be downloaded and we will see the message "Lazy
Component" in the browser.
We have effectively lazily loaded a module.

255

Lazy Loading and the Dependency Injection Tree

Lazy Loading and the Dependency

Injection Tree
Lazy loaded modules create their own branch on the Dependency Injection (DI) tree. This
means that it's possible to have services that belong to a lazy loaded module, that are not
accessible by the root module or any other eagerly loaded module of our application.
To show this behaviour, let's continue with the example of the previous section and add a
CounterService to our LazyModule .

app/lazy/lazy.module.ts
...
import { CounterService } from './counter.service';
@NgModule({
...
providers: [CounterService]
})
export class LazyModule {}

Here we added the CounterService to the providers array. Our CounterService is a

simple class that holds a reference to a counter property.
app/lazy/counter.service.ts
import { Injectable } from '@angular/core';
@Injectable()
export class CounterService {
counter = 0;
}

We can modify the LazyComponent to use this service with a button to increment the
counter property.

app/lazy/lazy.component.ts

256

Lazy Loading and the Dependency Injection Tree

import { Component } from '@angular/core';

import { CounterService } from './counter.service';
@Component({
template: `
<p>Lazy Component</p>
<button (click)="increaseCounter()">Increase Counter</button>
<p>Counter: {{ counterService.counter }}</p>
`
})
export class LazyComponent {
constructor(public counterService: CounterService) {}
increaseCounter() {
this.counterService.counter += 1;
}
}

View Example
The service is working. If we increment the counter and then navigate back and forth
between the eager and the lazy routes, the counter value will persist in the lazy loaded
module.
But the question is, how can we verify that the service is isolated and cannot be used in a
component that belongs to a different module? Let's try to use the same service in the
EagerComponent .

app/eager.component.ts
import { Component } from '@angular/core';
import { CounterService } from './lazy/counter.service';
@Component({
template: `
<p>Eager Component</p>
<button (click)="increaseCounter()">Increase Counter</button>
<p>Counter: {{ counterService.counter }}</p>
`
})
export class EagerComponent {
constructor(public counterService: CounterService) {}
increaseCounter() {
this.counterService.counter += 1;
}
}

257

Lazy Loading and the Dependency Injection Tree

If we try to run this new version of our code, we are going to get an error message in the
browser console:
No provider for CounterService!

What this error tells us is that the AppModule , where the EagerComponent is defined, has no
knowledge of a service called CounterService . CounterService lives in a different branch of
the DI tree created for LazyModule when it was lazy loaded in the browser.

258

Shared Modules and Dependency Injection

Shared Modules and Dependency Injection

Now that we have proven that lazy loaded modules create their own branch on the
Dependency Injection tree, we need to learn how to deal with services that are imported by
means of a shared module in both an eager and lazy loaded module.
Let's create a new module called SharedModule and define the CounterService there.
app/shared/shared.module.ts
import { NgModule } from '@angular/core';
import { CounterService } from './counter.service';
@NgModule({
providers: [CounterService]
})
export class SharedModule {}

Now we are going to import that SharedModule in the AppModule and the LazyModule .
app/app.module.ts
...
import { SharedModule } from './shared/shared.module';
@NgModule({
imports: [
SharedModule,
...
],
declarations: [
EagerComponent,
...
]
...
})
export class AppModule {}

app/lazy/lazy.module.ts

259

Shared Modules and Dependency Injection

...
import { SharedModule } from '../shared/shared.module';
@NgModule({
imports: [
SharedModule,
...
],
declarations: [LazyComponent]
})
export class LazyModule {}

With this configuration, the components of both modules will have access to the
CounterService . We are going to use this service in EagerComponent and LazyComponent in

exactly the same way. Just a button to increase the internal counter property of the service.
app/eager.component.ts
import { Component } from '@angular/core';
import { CounterService } from './shared/counter.service';
@Component({
template: `
<p>Eager Component</p>
<button (click)="increaseCounter()">Increase Counter</button>
<p>Counter: {{ counterService.counter }}</p>
`
})
export class EagerComponent {
constructor(public counterService: CounterService) {}
increaseCounter() {
this.counterService.counter += 1;
}
}

View Example
If you play with the live example, you will notice that the counter seems to behave
independently in EagerComponent and LazyComponent , we can increase the value of one
counter without altering the other one. In other words, we have ended up with two instances
of the CounterService , one that lives in the root of the DI tree of the AppModule and another
that lives in a lower branch of the DI tree accessible by the LazyModule .
This is not neccessarily wrong, you may find situations where you could need different
instances of the same service, but I bet most of the time that's not what you want. Think for
example of an authentication service, you need to have the same instance with the same

260

Shared Modules and Dependency Injection

information available everywhere disregarding if we are using the service in an eagerly or

lazy loaded module.
In the next section we are going to learn how to have only one instance of a shared service.

261

Sharing the Same Dependency Injection Tree

Sharing the Same Dependency Injection

Tree
So far our problem is that we are creating two instances of the same services in different
levels of the DI tree. The instance created in the lower branch of the tree is shadowing the
one defined at the root level. The solution? To avoid creating a second instance in a lower
level of the DI tree for the lazy loaded module and only use the service instance registered
at the root of the tree.
To accomplish that, we need to modify the definition of the SharedModule and instead of
defining our service in the providers property, we need to create a static method called
forRoot that exports the service along with the module itself.

app/shared/shared.module.ts
import { NgModule, ModuleWithProviders } from '@angular/core';
import { CounterService } from './counter.service';
@NgModule({})
export class SharedModule {
static forRoot(): ModuleWithProviders {
return {
ngModule: SharedModule,
providers: [CounterService]
};
}
}

With this setup, we can import this module in our root module AppModule calling the
forRoot method to register the module and the service.

app/app.module.ts
...
import { SharedModule } from './shared/shared.module';
@NgModule({
imports: [
SharedModule.forRoot(),
...
],
...
})
export class AppModule {}

262

Sharing the Same Dependency Injection Tree

In contrast, when import the same module in our LazyModule we will not call the forRoot
method because we don't want to register the service again in a different level of the DI tree,
so the declaration of the LazyModule doesn't change.
app/lazy/lazy.module.ts
...
import { SharedModule } from '../shared/shared.module';
@NgModule({
imports: [
SharedModule,
...
],
...
})
export class LazyModule {}

View Example
This time, whenever we change the value of the counter property, this value is shared
between the EagerComponent and the LazyComponent proving that we are using the same
instance of the CounterService .

263

Routing

Routing
In this section we will discuss the role of routing in Single Page Applications and Angular 2's
new component router.

264

Why Routing?

Why Routing?
Routing allows us to express some aspects of the application's state in the URL. Unlike with
server-side front-end solutions, this is optional - we can build the full application without ever
changing the URL. Adding routing, however, allows the user to go straight into certain
aspects of the application. This is very convenient as it can keep your application linkable
and bookmarkable and allow users to share links with others.
Routing allows you to:
Maintain the state of the application
Implement modular applications
Implement the application based on the roles (certain roles have access to certain
URLs)

265

Configuring Routes

Configuring Routes
Base URL Tag
The Base URL tag must be set within the <head> tag of index.html:
<base href="/">

In the demos we use a script tag to set the base tag. In a real application it must be set
as above.

Route Definition Object

The Routes type is an array of routes that defines the routing for the application. This is
where we can set up the expected paths, the components we want to use and what we want
our application to understand them as.
Each route can have different attributes; some of the common attributes are:
path - URL to be shown in the browser when application is on the specific route
component - component to be rendered when the application is on the specific route
redirectTo - redirect route if needed; each route can have either component or redirect
attribute defined in the route (covered later in this chapter)
pathMatch - optional property that defaults to 'prefix'; determines whether to match full
URLs or just the beginning. When defining a route with empty path string set pathMatch
to 'full', otherwise it will match all paths.
children - array of route definitions objects representing the child routes of this route
(covered later in this chapter).
To use Routes , create an array of route configurations.
Below is the sample Routes array definition:
const routes: Routes = [
{ path: 'component-one', component: ComponentOne },
{ path: 'component-two', component: ComponentTwo }
];

See Routes definition

266

Configuring Routes

RouterModule
RouterModule.forRoot takes the Routes array as an argument and returns a configured

router module. This router module must be specified in the list of imports of the app module.
...
import { RouterModule, Routes } from '@angular/router';
const routes: Routes = [
{ path: 'component-one', component: ComponentOne },
{ path: 'component-two', component: ComponentTwo }
];
export const routing = RouterModule.forRoot(routes);
@NgModule({
imports: [
BrowserModule,
routing
],
declarations: [
AppComponent,
ComponentOne,
ComponentTwo
],
bootstrap: [ AppComponent ]
})
export class AppModule {
}
platformBrowserDynamic().bootstrapModule(AppModule);

267

Redirecting the Router to Another Route

Redirecting the Router to Another Route

When your application starts, it navigates to the empty route by default. We can configure
the router to redirect to a named route by default:
export const routes: Routes = [
{ path: '', redirectTo: 'component-one', pathMatch: 'full' },
{ path: 'component-one', component: ComponentOne },
{ path: 'component-two', component: ComponentTwo }
];

This tells the router to redirect to component-one when matching the empty path ('').
When starting the application, it will automatically navigate to the route for component-one .

268

Defining Links Between Routes

Defining Links Between Routes

RouterLink
Add links to routes using the RouterLink directive.
For example the following code defines a link to the route at path component-one .
<a [routerLink]="['/component-one']">Component One</a>

Navigating Programmatically
Alternatively, you can navigate to a route by calling the navigate function on the router:
this.router.navigate(['/component-one']);

269

Dynamically Adding Route Components

Dynamically Adding Route Components

Rather than define each route's component separately, use RouterOutlet which serves as a
component placeholder; Angular 2 dynamically adds the component for the route being
activated into the <router-outlet></router-outlet> element.
import { Component } from '@angular/core';
@Component({
selector: 'app',
template: `
<nav>
<a [routerLink]="['/component-one']">Component One</a>
<a [routerLink]="['/component-two']">Component Two</a>
</nav>
<router-outlet></router-outlet>
<!-- Route components are added by router here -->
`
})
export class AppComponent {}

In the above example, the component corresponding to the route specified will be placed
after the <router-outlet></router-outlet> element when the link is clicked.
View Example
View examples running in full screen mode to see route changes in the URL.

270

Using Route Parameters

Using Route Parameters

Say we are creating an application that displays a product list. When the user clicks on a
product in the list, we want to display a page showing the detailed information about that
product. To do this you must:
add a route parameter ID
link the route to the parameter
add the service that reads the parameter.

Declaring Route Parameters

The route for the component that displays the details for a specific product would need a
route parameter for the ID of that product. We could implement this using the following
Routes :

export const routes: Routes = [

{ path: '', redirectTo: 'product-list', pathMatch: 'full' },
{ path: 'product-list', component: ProductList },
{ path: 'product-details/:id', component: ProductDetails }
];

Note :id in the path of the product-details route, which places the parameter in the path.
For example, to see the product details page for product with ID 5, you must use the
following URL: localhost:3000/product-details/5

Linking to Routes with Parameters

In the ProductList component you could display a list of products. Each product would
have a link to the product-details route, passing the ID of the product:
<a *ngFor="let product of products"
[routerLink]="['/product-details', product.id]">
{{ product.name }}
</a>

Note that the routerLink directive passes an array which specifies the path and the route
parameter. Alternatively we could navigate to the route programmatically:

271

Using Route Parameters

goToProductDetails(id) {
this.router.navigate(['/product-details', id]);
}

Reading Route Parameters

The ProductDetails component must read the parameter, then load the product based on
the ID given in the parameter.
The ActivatedRoute service provides a params Observable which we can subscribe to to
get the route parameters (see Observables).
import { Component, OnInit, OnDestroy } from '@angular/core';
import { ActivatedRoute } from '@angular/router';
@Component({
selector: 'product-details',
template: `
<div>
Showing product details for product: {{id}}
</div>
`,
})
export class LoanDetailsPage implements OnInit, OnDestroy {
id: number;
private sub: any;
constructor(private route: ActivatedRoute) {}
ngOnInit() {
this.sub = this.route.params.subscribe(params => {
this.id = +params['id']; // (+) converts string 'id' to a number
// In a real app: dispatch action to load the details here.
});
}
ngOnDestroy() {
this.sub.unsubscribe();
}
}

The reason that the params property on ActivatedRoute is an Observable is that the
router may not recreate the component when navigating to the same component. In this
case the parameter may change without the component being recreated.
View Basic Example
272

Using Route Parameters

View Example with Programmatic Route Navigation

View examples running in full screen mode to see route changes in the URL.

273

Defining Child Routes

Defining Child Routes

When some routes may only be accessible and viewed within other routes it may be
appropriate to create them as child routes.
For example: The product details page may have a tabbed navigation section that shows the
product overview by default. When the user clicks the "Technical Specs" tab the section
shows the specs instead.
If the user clicks on the product with ID 3, we want to show the product details page with the
overview:
localhost:3000/product-details/3/overview

When the user clicks "Technical Specs":

localhost:3000/product-details/3/specs
overview and specs are child routes of product-details/:id . They are only reachable

within product details.

Our Routes with children would look like:
export const routes: Routes = [
{ path: '', redirectTo: 'product-list', pathMatch: 'full' },
{ path: 'product-list', component: ProductList },
{ path: 'product-details/:id', component: ProductDetails,
children: [
{ path: '', redirectTo: 'overview', pathMatch: 'full' },
{ path: 'overview', component: Overview },
{ path: 'specs', component: Specs }
]
}
];

Where would the components for these child routes be displayed? Just like we had a
<router-outlet></router-outlet> for the root application component, we would have a

router outlet inside the ProductDetails component. The components corresponding to the
child routes of product-details would be placed in the router outlet in ProductDetails .

274

Defining Child Routes

import { Component, OnInit, OnDestroy } from '@angular/core';

import { ActivatedRoute } from '@angular/router';
@Component({
selector: 'product-details',
template: `
<p>Product Details: {{id}}</p>
<!-- Product information -->
<nav>
<a [routerLink]="['overview']">Overview</a>
<a [routerLink]="['specs']">Technical Specs</a>
</nav>
<router-outlet></router-outlet>
<!-- Overview & Specs components get added here by the router -->
`
})
export default class ProductDetails implements OnInit, OnDestroy {
id: number;
constructor(private route: ActivatedRoute) {}
ngOnInit() {
this.sub = this.route.params.subscribe(params => {
this.id = +params['id']; // (+) converts string 'id' to a number
});
}
ngOnDestroy() {
this.sub.unsubscribe();
}
}

Alternatively, we could specify overview route URL simply as:

localhost:3000/product-details/3

export const routes: Routes = [

{ path: '', redirectTo: 'product-list', pathMatch: 'full' },
{ path: 'product-list', component: ProductList },
{ path: 'product-details/:id', component: ProductDetails,
children: [
{ path: '', component: Overview },
{ path: 'specs', component: Specs }
]
}
];

Since the Overview child route of product-details has an empty path, it will be loaded by
default. The specs child route remains the same.

275

Defining Child Routes

View Example with child routes

View Example with route params & child routes
View examples running in full screen mode to see route changes in the URL.

Accessing a Parent's Route Parameters

In the above example, say that the child routes of product-details needed the ID of the
product to fetch the spec or overview information. The child route component can access the
parent route's parameters as follows:
export default class Overview {
parentRouteId: number;
private sub: any;
constructor(private router: Router,
private route: ActivatedRoute) {}
ngOnInit() {
// Get parent ActivatedRoute of this route.
this.sub = this.router.routerState.parent(this.route)
.params.subscribe(params => {
this.parentRouteId = +params["id"];
});
}
ngOnDestroy() {
this.sub.unsubscribe();
}
}

View Example child routes accessing parent's route parameters

View examples running in full screen mode to see route changes in the URL.

Links
Routes can be prepended with / , or ../ ; this tells Angular 2 where in the route tree to
link to.

276

Defining Child Routes

Prefix

Looks in

Root of the application

none

Current component children routes

../

Current component parent routes

Example:
<a [routerLink]="['route-one']">Route One</a>
<a [routerLink]="['../route-two']">Route Two</a>
<a [routerLink]="['/route-three']">Route Three</a>

In the above example, the link for route one links to a child of the current route. The link for
route two links to a sibling of the current route. The link for route three links to a child of the
root component (same as route one link if current route is root component).
View Example with linking throughout route tree
View examples running in full screen mode to see route changes in the URL.

277

Controlling Access to or from a Route

Controlling Access to or from a Route

To control whether the user can navigate to or away from a given route, use route guards.
For example, we may want some routes to only be accessible once the user has logged in
or accepted Terms & Conditions. We can use route guards to check these conditions and
control access to routes.
Route guards can also control whether a user can leave a certain route. For example, say
the user has typed information into a form on the page, but has not submitted the form. If
they were to leave the page, they would lose the information. We may want to prompt the
user if the user attempts to leave the route without submitting or saving the information.

Registering the Route Guards with Routes

In order to use route guards, we must register them with the specific routes we want them to
run for.
For example, say we have an accounts route that only users that are logged in can
navigate to. This page also has forms and we want to make sure the user has submitted
unsaved changes before leaving the accounts page.
In our route config we can add our guards to that route:
import { Routes, RouterModule } from '@angular/router';
import { AccountPage } from './account-page';
import { LoginRouteGuard } from './login-route-guard';
import { SaveFormsGuard } from './save-forms-guard';
const routes: Routes = [
{ path: 'home', component: HomePage },
{
path: 'accounts',
component: AccountPage,
canActivate: [LoginRouteGuard],
canDeactivate: [SaveFormsGuard]
}
];
export const appRoutingProviders: any[] = [];
export const routing = RouterModule.forRoot(routes);

278

Controlling Access to or from a Route

Now LoginRouteGuard will be checked by the router when activating the accounts route,
and SaveFormsGuard will be checked when leaving that route.

Implementing CanActivate
Let's look at an example activate guard that checks whether the user is logged in:
import { CanActivate } from '@angular/router';
import { Injectable } from '@angular/core';
import { LoginService } from './login-service';
@Injectable()
export class LoginRouteGuard implements CanActivate {
constructor(private loginService: LoginService) {}
canActivate() {
return this.loginService.isLoggedIn();
}
}

This class implements the CanActivate interface by implementing the canActivate

function.
When canActivate returns true, the user can activate the route. When canActivate returns
false, the user cannot access the route. In the above example, we allow access when the
user is logged in.
canActivate can also be used to notify the user that they can't access that part of the

application, or redirect them to the login page.

See Official Definition for CanActivate

Implementing CanDeactivate
CanDeactivate works in a similar way to CanActivate but there are some important

differences. The canDeactivate function passes the component being deactivated as an

argument to the function:
export interface CanDeactivate<T> {
canDeactivate(component: T, route: ActivatedRouteSnapshot, state: RouterStateSnapsho
t):
Observable<boolean>|Promise<boolean>|boolean;
}

279

Controlling Access to or from a Route

We can use that component to determine whether the user can deactivate.
import { CanDeactivate } from '@angular/router';
import { Injectable } from '@angular/core';
import { AccountPage } from './account-page';
@Injectable()
export class SaveFormsGuard implements CanDeactivate<AccountPage> {
canDeactivate(component: AccountPage) {
return component.areFormsSaved();
}
}

See Official Definition for CanDeactivate

Async Route Guards

The canActivate and canDeactivate functions can either return values of type boolean , or
Observable<boolean> (an Observable that resolves to boolean ). If you need to do an

asynchronous request (like a server request) to determine whether the user can navigate to
or away from the route, you can simply return an Observable<boolean> . The router will wait
until it is resolved and use that value to determine access.
For example, when the user navigates away you could have a dialog service ask the user to
confirm the navigation. The dialog service returns an Observable<boolean> which resolves to
true if the user clicks 'OK', or false if user clicks 'Cancel'.
canDeactivate() {
return dialogService.confirm('Discard unsaved changes?');
}

View Example
See Official Documentation for Route Guards

280

Passing Optional Parameters to a Route

Passing Optional Parameters

Query parameters allow you to pass optional parameters to a route such as pagination
information.
For example, on a route with a paginated list, the URL might look like the following to
indicate that we've loaded the second page:
localhost:3000/product-list?page=2

The key difference between query parameters and route parameters is that route
parameters are essential to determining route, whereas query parameters are optional.

Passing Query Parameters

Use the [queryParams] directive along with [routerLink] to pass query parameters. For
example:
<a [routerLink]="['product-list']" [queryParams]="{ page: 99 }">Go to Page 99</a>

Alternatively, we can navigate programmatically using the Router service:

goToPage(pageNum) {
this.router.navigate(['/product-list'], { queryParams: { page: pageNum } });
}

Reading Query Parameters

Similar to reading route parameters, the Router service returns an Observable we can
subscribe to to read the query parameters:

281

Passing Optional Parameters to a Route

import { Component } from '@angular/core';

import { ActivatedRoute, Router } from '@angular/router';
@Component({
selector: 'product-list',
template: `<!-- Show product list -->`
})
export default class ProductList {
constructor(
private route: ActivatedRoute,
private router: Router) {}
ngOnInit() {
this.sub = this.route
.queryParams
.subscribe(params => {
// Defaults to 0 if no query param provided.
this.page = +params['page'] || 0;
});
}
ngOnDestroy() {
this.sub.unsubscribe();
}
nextPage() {
this.router.navigate(['product-list'], { queryParams: { page: this.page + 1 } });
}
}

View Example
See Official Documentation on Query Parameters

282

Using Auxiliary Routes

Using Auxiliary Routes

Angular 2 supports the concept of auxiliary routes, which allow you to set up and navigate
multiple independent routes in a single app. Each component has one primary route and
zero or more auxiliary outlets. Auxiliary outlets must have unique name within a component.
To define the auxiliary route we must first add a named router outlet where contents for the
auxiliary route are to be rendered.
Here's an example:
import {Component} from '@angular/core';
@Component({
selector: 'app',
template: `
<nav>
<a [routerLink]="['/component-one']">Component One</a>
<a [routerLink]="['/component-two']">Component Two</a>
<a [routerLink]="[{ outlets: { 'sidebar': ['component-aux'] } }]">Component Aux<
/a>
</nav>
<div style="color: green; margin-top: 1rem;">Outlet:</div>
<div style="border: 2px solid green; padding: 1rem;">
<router-outlet></router-outlet>
</div>
<div style="color: green; margin-top: 1rem;">Sidebar Outlet:</div>
<div style="border: 2px solid blue; padding: 1rem;">
<router-outlet name="sidebar"></router-outlet>
</div>
`
})
export class AppComponent {
}

Next we must define the link to the auxiliary route for the application to navigate and render
the contents.
<a [routerLink]="[{ outlets: { 'sidebar': ['component-aux'] } }]">
Component Aux
</a>

283

Using Auxiliary Routes

View Example
Each auxiliary route is an independent route which can have:
its own child routes
its own auxiliary routes
its own route-params
its own history stack

284

Redux and Ngrx

Redux and Ngrx

What is Redux?
Redux is an application state manager for JavaScript applications, and keeps with the core
principles of the Flux-architecture by having a unidirectional data flow in your application.
Where Flux applications traditionally have multiple stores, Redux applications have only one
global, read-only application state. This state is calculated by "reducing" over a collection or
stream of actions that update it in controlled ways.
One popular Angular 2 specific implementation of the Redux pattern is Ng2-Redux, which
we'll be using to describe how to use this approach in an application.

What is Ngrx?
Redux implementation has been very well received and has inspired the creation of ngrx, a
set of modules that implement the same way of managing state as well as some of the
middleware and tools in the Redux ecosystem. Ngrx was created to be used specifically with
Angular 2 and RxJS, as it leans heavily on the observable paradigm.
Although we'll be using Ng2-Redux, a lot of the same lessons apply with regards to ngrx
though the syntax may be different and have some slight differences in what abstractions
are involved.

Resources
Redux Documentation
Getting Started with Redux - Egghead.io
Ng2-Redux - Angular 2 Bindings for Redux
ngrx - Redux style state management with RxJS and Angular 2
Angular 2 Redux Starter Kit

285

Review of Reducers and Pure Functions

Review of Reducers and Pure Functions

One of the core concepts of Redux is the reducer. A reducer is a function with the signature
(accumulator: T, item: U) => T . Reducers are often used in JavaScript through the
Array.reduce method, which iterates over each of the array's items and accumulates a

single value as a result. Reducers should be pure functions, meaning they don't generate
any side-effects.
A simple example of a reducer is the sum function:
let x = [1,2,3].reduce((value, state) => value + state, 0)
// x == 6

286

Reducers as State Management

Reducers as State Management

This simple idea turns out to be very powerful. With Redux, you replay a series of events
into the reducer and get your new application state as a result.
Reducers in a Redux application should not mutate the state, but return a copy of it, and be
side-effect free. This encourages you to think of your application as UI that gets "computed"
from a series of events in time.
Let's take a look at a simple counter reducer.

Simple Reducer
app/reducer/counter-reducer.ts
import { INCREMENT_COUNTER, DECREMENT_COUNTER } from '../actions/counter-actions';
export default function counter(state = 0, action) {
switch (action.type) {
case INCREMENT_COUNTER:
return state + 1;
case DECREMENT_COUNTER:
return state - 1;
default:
return state;
}
}

We can see here that we are passing in an initial state and an action. To handle each action
we have set up a switch statement. Instead of each reducer needing to explicitly subscribe
to the dispatcher, every action gets passed into each reducer, which handles the actions it's
interested in and then returns the new state along to the next reducer.
Reducers should be side-effect free. This means that they should not modify things outside
of their own scope. They should simply compute the next application state as a pure function
of the reducer's arguments.
For this reason, side-effect causing operations, such as updating a record in a database,
generating an id, etc. should be handled elsewhere in the application such as in the action
creators, using middleware such as 'Epics' from redux-observable or ngrx/effects.

287

Reducers as State Management

Another consideration when creating your reducers is to ensure that they are immutable and
not modifying the state of your application. If you mutate your application state, it can cause
unexpected behavior. There are a few ways to help maintain immutability in your reducers.
One way is by using new ES6 features such as Object.assign or the spread operator for
arrays.
function immutableObjectReducer(state = { someValue: 'value'} , action) {
switch(action.payload) {
case SOME_ACTION:
return Object.assign({}, state, { someValue: action.payload.value });
default:
return state;
}
}
function immutableArrayReducer(state = [1,2,3], action) {
switch(action.payload) {
case ADD_ITEM:
return [...state,action.payload.value]
default:
return state;
}
}

However, when dealing with complex or deeply nested objects, it can be difficult to maintain
immutability in your application using this syntax. This is where a library like ImmutableJS
can help.

288

Redux Actions

Redux Actions
Redux actions should generally be simple JSON objects. This is because they should be
serializable and replayable into the application state. Even if your actions involve
asynchronous logic, the final dispatched action should remain a plain JSON object.
Redux action creators are generally where side-effects should happen, such as making API
calls or generating IDs. This is because when the final action gets dispatched to the
reducers, we want to update the application state to reflect what has already happened.
Let's take a look at the actions that are used in this example. For now, let's just focus on
some simple synchronous actions.

Synchronous Actions
Most Redux apps have a set of functions, called "action creators", that are used to set up
and dispatch actions.
In Angular 2, it's convenient to define "action creator services" for your action creators to live
in; these services can be injected into the components that need to dispatch the actions.
app/actions/counter-actions.ts
import { Injectable } from '@angular/core';
import { NgRedux } from 'ng2-redux';
export const INCREMENT_COUNTER = 'INCREMENT_COUNTER';
export const DECREMENT_COUNTER = 'DECREMENT_COUNTER';
@Injectable()
export class CounterActions {
constructor(private redux: NgRedux<any>) {}
increment() {
this.redux.dispatch({ type: INCREMENT_COUNTER });
}
decrement() {
this.redux.dispatch({ type: DECREMENT_COUNTER });
}
}

289

Redux Actions

As you can see, the action creators are simple functions that (optionally) take parameters,
and then dispatch a JSON object containing more information.
The dispatch function expects to be called with something that conforms to the "Action"
interface from the Redux library:
import { Action } from 'redux';

This interface has the following properties:

type - a string/enum representing the action
payload? - optional, the data that you want to pass into the reducer if applicable
error? - optional, indicates if this message is due to an error
metaData? - optional - any extra information

Asynchronous Actions
This "ActionCreatorService" pattern comes in handy if you must handle asynchronous or
conditional actions (users of react-redux may recognize this pattern as analogous to reduxthunk in a dependency-injected world).
app/actions/counter-actions.ts

290

Redux Actions

import { Injectable } from '@angular/core';

import { NgRedux } from 'ng2-redux';
export const INCREMENT_COUNTER = 'INCREMENT_COUNTER';
export const DECREMENT_COUNTER = 'DECREMENT_COUNTER';
@Injectable()
export class CounterActions {
constructor(private redux: NgRedux<any>) {}
// ...
incrementIfOdd() {
const { counter } = this.redux.getState();
if (counter % 2 === 0) return;
this.redux.dispatch({ type: INCREMENT_COUNTER });
}
incrementAsync(timeInMs = 1000) {
this.delay(timeInMs).then(() => this.redux.dispatch({ type: INCREMENT_COUNTER }));
}
private delay(timeInMs) {
return new Promise((resolve, reject) => {
setTimeout(() => resolve() , timeInMs);
});
}
}

In the incrementIfOdd action, we are using the getState function to get the current state of
the application.
In the incrementAsync action, we are delaying the actual call to dispatch . For example, we
have created a Promise that will resolve after the delay. Once the Promise resolves, we can
then do a dispatch with the increase action.
View Ng2-Redux Example View Ngrx Example

Actions that Depend on Other Services

The ActionCreatorService pattern becomes necessary in cases where your action creators
must use other Angular 2 services. Consider the following ActionCreatorService that handles
a remote API call:

291

Redux Actions

import { Injectable } from '@angular/core';

import { NgRedux } from 'ng2-redux';
import { AuthService } from '../services/auth/';
@Injectable()
export class SessionActions {
static LOGIN_USER_PENDING = 'LOGIN_USER_PENDING';
static LOGIN_USER_SUCCESS = 'LOGIN_USER_SUCCESS';
static LOGIN_USER_ERROR = 'LOGIN_USER_ERROR';
static LOGOUT_USER = 'LOGOUT_USER';
constructor(
private ngRedux: NgRedux<any>,
private authService: AuthService) {}
loginUser(credentials) {
const username = credentials.username;
const password = credentials.password;
this.ngRedux.dispatch({ type: SessionActions.LOGIN_USER_PENDING });
this.authService.login(username, password)
.then(result => this.ngRedux.dispatch({
type: SessionActions.LOGIN_USER_SUCCESS,
payload: result
}))
.catch(() => this.ngRedux.dispatch({
type: SessionActions.LOGIN_USER_ERROR
}));
};
logoutUser = () => {
this.ngRedux.dispatch({ type: SessionActions.LOGOUT_USER });
};
}

292

Configuring your Application to use Redux

Configuring your Application to use Redux

Once you have the reducers and actions created, it is time to configure your Angular 2
application to make use of Ng2-Redux. For this, we will need to:
Register Ng2-Redux with Angular 2
Create our application reducer
Create and configure a store

Registering Ng2-Redux with Angular 2

app/index.ts
import { NgModule } from '@angular/core';
import { BrowserModule } from '@angular/platform-browser';
import { platformBrowserDynamic } from '@angular/platform-browser-dynamic';
import { NgReduxModule, NgRedux } from 'ng2-redux';
import { SimpleRedux } from './containers/app-container';
@NgModule({
imports: [
BrowserModule,
NgReduxModule
],
declarations: [
SimpleRedux
],
bootstrap: [ SimpleRedux ]
})
class AppModule {
}
platformBrowserDynamic().bootstrapModule(AppModule);

Here, we're simply adding the NgReduxModule class as an import in our NgModule
declaration.

Create our Application Reducer

app/reducers/index.ts

293

Configuring your Application to use Redux

import { combineReducers } from 'redux';

import counter from './counter-reducer';
export default combineReducers({
counter
});

combineReducers allows us to break out our application into smaller reducers with a single

area of concern. Each reducer that you pass into it will control a property on the state. So
when we are subscribing to our state changes with Ng2-Redux's @select decorator, we are
able to select a counter property, or any other reducers you have provided.

Create and Configure a Store

Next we want Ng2-Redux to configure our store based on settings we provide. This should
be done once, in the top-level component of your application.
app/containers/app-container.ts
import { Component } from '@angular/core';
import { NgRedux } from 'ng2-redux';
import logger from '../store/configure-logger';
import reducer from '../reducers';
@Component({
// ...
})
class SimpleRedux {
constructor(ngRedux: NgRedux) {
const initialState = {};
const middleware = [ logger ];
ngRedux.configureStore(reducer, initialState, middleware);
}
}

In this example we are creating a store that uses the redux-logger middleware, which will
add some logging functionality to the application.

294

Using Redux with Components

Using Redux with Components

We will use the select pattern from Ng2-Redux to bind our components to the store. To
demonstrate how this works, let's take a look at a small example with a counter component.

Counter Example
Let's start by building out a counter component. The component will be responsible for
keeping track of how many times it was clicked and displaying that amount.
app/components/counter-component.ts
import { Component } from '@angular/core';
import { select } from 'ng2-redux';
import { Observable } from 'rxjs';
import { CounterActions } from '../actions/counter-actions';
@Component({
selector: 'counter',
providers: [ CounterActions ],
template: `
<p>
Clicked: {{ counter$ | async }} times
<button (click)="actions.increment()">+</button>
<button (click)="actions.decrement()">-</button>
<button (click)="actions.incrementIfOdd()">Increment if odd</button>
<button (click)="actions.incrementAsync()">Increment async</button>
</p>
`
})
export class Counter {
@select() counter$: Observable<number>;
constructor(public actions: CounterActions) {}
}

View Example
The template syntax should be familiar by now, displaying a Observable counter with the
async pipe and handling some click events.
In this case, the click events are bound to expressions that call our action creators from the
CounterActions ActionCreatorService.

295

Using Redux with Components

Let's take a look at the use of @select .

@select is a feature of Ng2-Redux which is designed to help you attach your store's state

to your components in a declarative way. You can attach it to a property of your component
class and Ng2-Redux will create an Observable and bind it to that property for you.
In this case, @select has no parameters, so Ng2-Redux will look for a store property with
the same name as the class variable. It omits the trailing $ since that's simply a naming
convention for Observables .
So now, any time store.counter is updated by a reducer, counter$ will receive the new
value and | async will update it in the template.
Note that @select supports a wide range of arguments to allow you to select portions of
your Redux store with a great deal of flexibility. See the Ng2-Redux docs for more details.
The Ng2-Redux "select pattern" style differs a bit from the "connect" style used by reactredux ; however by using Angular 2's DI and TypeScript's decorators, we can have a nicely

declarative binding where most of the work is done in the template. We also get the power of
Observables and OnPush change detection for better performance.

Either way, we still benefit from the Redux fundamentals of reducers and one-way data-flow.

296

Redux and Component Architecture

Redux and Component Architecture

In the above example, our counter component is a smart component. It knows about
Redux, the structure of the state and the actions it needs to call. In theory you can drop this
component into any area of your application and just let it work. But it will be tightly bound to
that specific slice of state and those specific actions. For example, what if we wanted to
have multiple counters tracking different things on the page? For example, counting the
number of red clicks vs blue clicks.
To help make components more generic and reusable, it's worth trying to separate them into
container components and presentational components.
Container
Components

Presentational Components

Location

Top level, route

handlers

Middle and leaf components

Aware of
Redux

Yes

No

Subscribe to Redux
state

Read state from @Input properties

Dispatch Redux actions

Invoke callbacks from @Output

properties

To read data
To change
data
redux docs

Keeping this in mind, let's refactor our counter to be a presentational component. First, let's
modify our app-container to have two counter components on it as we currently have it.

297

Redux and Component Architecture

import { Component } from '@angular/core';

@Component({
selector: 'simple-redux'
template: `
<div>
<h1>Redux: Two components, one state.</h1>
<div style="float: left; border: 1px solid red;">
<h2>Click Counter</h2>
<counter></counter>
</div>
<div style="float: left; border: 1px solid blue;">
<h2>Curse Counter</h2>
<counter></counter>
</div>
</div>
`
})
export class SimpleRedux {}

View Example
As you can see in the example, when clicking on the buttons the numbers in both
components will update in sync. This is because the counter component is coupled to a
specific piece of state and action.
Looking at the example, you can see that there is already an app/reducers/curse-reducer.ts
and app/actions-curse-actions.ts. They are pretty much the same as the counter actions and
counter reducer, we just wanted to create a new reducer to hold the state of it.
To turn the counter component from a smart component into a dumb component, we need to
change it to have data and callbacks passed down into it. For this, we will pass the data into
the component using @Input properties, and the action callbacks as @Output properties.
We now have a nicely-reusable presentational component with no knowledge of Redux or
our application state.
app/components/counter-component.ts

298

Redux and Component Architecture

import { Component, Input, Output, EventEmitter } from '@angular/core';

import { Observable } from 'rxjs';
@Component({
selector: 'counter',
template: `
<p>
Clicked: {{ counter | async }} times
<button (click)="increment.emit()">+</button>
<button (click)="decrement.emit()">-</button>
<button (click)="incrementIfOdd.emit()">Increment if odd</button>
<button (click)="incrementAsync.emit()">Increment async</button>
</p>
`
})
export class Counter {
@Input() counter: Observable<number>;
@Output() increment = new EventEmitter<void>();
@Output() decrement = new EventEmitter<void>();
@Output() incrementIfOdd = new EventEmitter<void>();
@Output() incrementAsync = new EventEmitter<void>();
}

Next, let's modify the main app container to hook up these inputs and outputs to the
template.
@Component app/src/containers/app-containter.ts

299

Redux and Component Architecture

@Component({
selector: 'simple-redux',
providers: [ CounterActions, CurseActions ],
template: `
<div>
<h1>Redux: Presentational Counters</h1>
<div style="float: left; border: 1px solid red;">
<h2>Click Counter</h2>
<counter [counter]="counter$"
(increment)="counterActions.increment()"
(decrement)="counterActions.decrement()"
(incrementIfOdd)="counterActions.incrementIfOdd()"
(incrementAsync)="counterActions.incrementAsync()">
</counter>
</div>
<div style="float: left; border: 1px solid blue;">
<h2>Curse Counter</h2>
<counter [counter]="curse$"
(increment)="curseActions.castCurse()"
(decrement)="curseActions.removeCurse()"
(incrementIfOdd)="curseActions.castIfOdd()"
(incrementAsync)="curseActions.castAsync()">
</counter>
</div>
</div>
`
})

At this point, the template is attempting to call actions on our two ActionCreatorServices,
CounterActions and CurseActions ; we just need to hook those up using Dependency

Injection:
app/src/containers/app-container.ts

300

Redux and Component Architecture

import { Component, View, Inject, OnDestroy, OnInit } from '@angular/core';

import { select } from 'ng2-redux';
import { Observable } from 'rxjs';
import { CounterActions } from '../actions/counter-actions';
import { CurseActions } from '../actions/curse-actions';
@Component({ /* see above .... */})
export class SimpleRedux {
@select() counter$: Observable<number>;
@select() curse$: Observable<number>;
constructor(
public counterActions: CounterActions,
public curseActions: CurseActions) {
}
}

View Ng2-Redux Example View Ngrx Example

Our two Observable s, counter$ and curse$ , will now get updated with a new value every
time the relevant store properties are updated by the rest of the system.

301

Getting More From Redux and Ngrx

Getting More From Redux and Ngrx

Redux
Redux has a number of tools and middleware available in its ecosystem to facilitate elegant
app development.
Redux DevTools - a tool that displays a linear timeline of actions that have interacted
with its store. Allows for replaying actions and error handling
redux-thunk - middleware that enables lazy evaluation of actions
redux-observable - an RxJS-based model for handling side-effects on action streams.
*ng2-redux-router - reactive glue between the Angular 2 router and your redux store.

Ngrx
Ngrx provides most of its Redux implementation through the ngrx/store module. Other
modules are available for better integration and development.
ngrx/store-devtools - an ngrx implementation of the Redux DevTools
ngrx/effects - a model for performing side-effects similar to redux-saga
ngrx/router and ngrx/router-store - a router for Angular 2 that can be connected to your
ngrx store

302

TDD Testing

TDD Testing
Test-Driven-Development is an engineering process in which the developer writes an initial
automated test case that defines a feature, then writes the minimum amount of code to pass
the test and eventually refactors the code to acceptable standards.
A unit test is used to test individual components of the system. An integration test is a test
which tests the system as a whole, and how it will run in production.
Unit tests should only verify the behavior of a specific unit of code. If the unit's behavior is
modified, then the unit test would be updated as well. Unit tests should not make
assumptions about the behavior of other parts of your codebase or your dependencies.
When other parts of your codebase are modified, your unit tests should not fail. (Any failure
indicates a test that relies on other components and is therefore not a unit test.) Unit tests
are cheap to maintain and should only be updated when the individual units are modified.
For TDD in Angular, a unit is most commonly defined as a class, pipe, component, or
service. It is important to keep units relatively small. This helps you write small tests which
are "self-documenting", where they are easy to read and understand.

303

The Testing Toolchain

The Testing Toolchain

Our testing toolchain will consist of the following tools:
Jasmine
Karma
Phantom-js
Istanbul
Sinon
Chai
Jasmine is the most popular testing framework in the Angular community. This is the core
framework that we will write our unit tests with.
Karma is a test automation tool for controlling the execution of our tests and what browser to
perform them under. It also allows us to generate various reports on the results. For one or
two tests this may seem like overkill, but as an application grows larger and the number of
units to test grows, it is important to organize, execute and report on tests in an efficient
manner. Karma is library agnostic so we could use other testing frameworks in combination
with other tools (like code coverage reports, spy testing, e2e, etc.).
In order to test our Angular 2 application we must create an environment for it to run in. We
could use a browser like Chrome or Firefox to accomplish this (Karma supports in-browser
testing), or we could use a browser-less environment to test our application, which can offer
us greater control over automating certain tasks and managing our testing workflow.
PhantomJS provides a JavaScript API that allows us to create a headless DOM instance
which can be used to bootstrap our Angular 2 application. Then, using that DOM instance
that is running our Angular 2 application, we can run our tests.
Istanbul is used by Karma to generate code coverage reports, which tells us the overall
percentage of our application being tested. This is a great way to track which
components/services/pipes/etc. have tests written and which don't. We can get some useful
insight into how much of the application is being tested and where.
For some extra testing functionality we can use the Sinon library for things like test spys, test
subs and mock XHR requests. This is not necessarily required as Jasmine comes with the
spyOn function for incorporating spy tests.

Chai is an assertion library that can be paired with any other testing framework. It offers
some syntactic sugar that lets us write our unit tests with different verbiage - we can use a
should, expect or assert interface. Chai also takes advantage of "function chaining" to form

304

The Testing Toolchain

English-like sentences used to describe tests in a more user friendly way. Chai isn't a
required library for testing and we won't explore it much more in this handout, but it is a
useful tool for creating cleaner, more well-written tests.

305

Test Setup

Test Setup
The repo angular2-redux-starter is a basic webpack-based Angular 2 application (with
Redux) with the same testing toolchain outlined above. Let's take a look at how this project
is set up.

306

Filename Conventions

Filename Conventions
Each unit test is put into its own separate file. The Angular 2 team recommends putting unit
test scripts alongside the files they are testing and using a .spec filename extension to
mark it as a testing script (this is a Jasmine convention). So if you had a component
/app/components/mycomponent.ts , then your unit test for this component would be in
/app/components/mycomponent.spec.ts . This is a matter of personal preference; you can put

your testing scripts wherever you like, though keeping them close to your source files makes
them easier to find and gives those who aren't familiar with the source code an idea of how
that particular piece of code should work.

307

Karma Configuration

Karma Configuration
Karma is the foundation of our testing workflow. It brings together our other testing tools to
define the framework we want to use, the environment to test under, the specific actions we
want to perform, etc. In order to do this Karma relies on a configuration file named by default
karma.conf.js.
You can seed a new configuration file though the karma init command, which will guide
you through a few basic questions to get a bare minimum setup running.

Overview
The configuration file is put together by exporting a function that accepts the configuration
object that Karma is going to work with. Modifying certain properties on this object will tell
Karma what it is we want to do. Let's go over some of the key properties used in this
configuration file:

module.exports = (config) => {

const coverage = config.singleRun ? ['coverage'] : [];
config.set({
frameworks: [...],
plugins: [ ... ],
files: [ ... ],
preprocessors: { ... },
webpack: { ... },
reporters: [ ... ],
coverageReporter: { ... },
port: 9999,
browsers: ['Chrome'], // Alternatively: 'PhantomJS'
colors: true,
logLevel: config.LOG_INFO,
autoWatch: true,
captureTimeout: 6000,
});
};

frameworks

308

Karma Configuration

frameworks: [
'jasmine',
],

frameworks is a list of the testing frameworks we want to use. These frameworks must be

installed through NPM as a dependency in our project or/and as a Karma plugin.

plugins
plugins: [
'karma-jasmine',
'karma-webpack',
'karma-coverage',
'karma-remap-istanbul',
'karma-chrome-launcher',
],

Plugins that integrate karma with testing frameworks like Jasmine or build systems like
Webpack.

files
files: [
'./src/tests.entry.ts',
{
pattern: '**/*.map',
served: true,
included: false,
watched: true,
},
],

files is a list of files to be loaded into the browser/testing environment. These are loaded

sequentially, so order matters. The file list can also take the form of a glob pattern as it
becomes rather tedious to manually add in a new file for each new testing script created.
In the angular2-redux-starter karma.conf.js we have put the testing files we wish to include in
a separate file - src/tests.entry.ts, which includes a require call using a regex pattern for
importing files with the .spec.ts file extension. As a project grows larger and the number of
files to include grows in complexity it is good practice to put file imports in a separate file this keeps the karma.conf.js file cleaner and more readable. Here is what our
src/tests.entry.ts looks like:

309

Karma Configuration

let testContext = (<{ context?: Function }>require).context('./', true, /\.test\.ts/);

testContext.keys().forEach(testContext);

preprocessors
preprocessors: {
'./src/tests.entry.ts': [
'webpack',
'sourcemap',
],
'./src/**/!(*.test|tests.*).(ts|js)': [
'sourcemap',
],
}

preprocessors allow for some operation to be performed on the contents of a unit testing

file before it is executed. These operations are carried out through the use of Karma plugins
and are often used for transpiling operations. Since we are writing unit tests in TypeScript,
.ts files must be transpiled into plain Javascript in order to run in a browser-based
environment.
In angular2-redux-starter this process is done with webpack, so we explicitly invoke the
webpack processor on all of our testing files (those ending with .spec.ts). We also load any

source map files originating from transpilation through the sourcemap processor.

webpack

310

Karma Configuration

webpack: {
plugins,
entry: './src/tests.entry.ts',
devtool: 'inline-source-map',
resolve: {
extensions: ['.webpack.js', '.web.js', '.ts', '.js'],
},
module: {
rules:
combinedLoaders().concat(
config.singleRun
? [ loaders.istanbulInstrumenter ]
: [ ]),
},
stats: { colors: true, reasons: true },
},
webpackServer: {
noInfo: true, // prevent console spamming when running in Karma!
}

If the project uses webpack, then the property webpack in the Karma configuration object is
where we can configure webpack with Karma. In the angular2-redux-starter, plugins and
loaders are exported from their own files to be imported by both the webpack config and the
karma config, making the configuration object smaller.
Using webpack, we can configure how to bundle our unit tests; that is, whether to pack all
tests into a single bundle, or each unit test in its own bundle, etc. Regardless, unit tests
should not be bundled with the rest of the applications code (especially in production!). In
angular2-redux-starter we have opted to bundle all unit tests together.

coverageReporters and reporters

311

Karma Configuration

reporters: ['spec']
.concat(coverage)
.concat(coverage.length > 0 ? ['karma-remap-istanbul'] : []),
remapIstanbulReporter: {
src: 'coverage/chrome/coverage-final.json',
reports: {
html: 'coverage',
},
},
coverageReporter: {
reporters: [
{ type: 'json' },
],
dir: './coverage/',
subdir: (browser) => {
return browser.toLowerCase().split(/[ /-]/)[0]; // returns 'chrome'
},
},

coverageReporter is used to configure the output of results of our code coverage tool (our

toolchain uses Istanbul). Here we have specified to output the results in JSON and HTML.
Reports will appear in the coverage/ folder.
reporters is a list of reporters to use in the test cycle. Reporters can be thought of as

modular tools used to report on some aspect of the testing routine outside of the core unit
tests. Code coverage is an example of a reporter - we want it to report on how much of our
code is being tested. There are many more reporters available for Karma that can aid in
crafting your testing workflow.

Environment configuration
port: 9999,
browsers: ['Chrome'], // Alternatively: 'PhantomJS'
colors: true,
logLevel: config.LOG_INFO,
autoWatch: true,
captureTimeout: 6000,

port , browsers and singleRun configure the environment our unit tests will run under.

The browsers property specifies which browser we want Karma to launch and capture
output from. We can use Chrome, Firefox, Safari, IE or Opera (requires additional Karma
launcher to be installed for each respective browser). For a browser-less DOM instance we
can use PhantomJS (as outlined in the toolchain section).

312

Karma Configuration

We can also manually capture output from a browser by navigating to

http://localhost:port , where port is the number specified in the port property (the

default value is 9876 if not specified). The property singleRun controls how Karma
executes, if set to true , Karma will start, launch configured browsers, run tests and then
exit with a code of either 0 or 1 depending on whether or not all tests passed.

Completed Configuration
The net result of customizing all of these proprties is the karma.conf.js file in angular-reduxstarter.

Additional Resources
This is just a sample of the core properties in karma.conf.js being used by angular2-reduxstarter project. There are many more properties that can be used to extend and configure
the functionality of Karma - take a look at the official documentation for the full API
breakdown.

313

TestBed Configuration (Optional)

TestBed Configuration (Optional)

As you will see in Testing Components, real-world component testing often relies on the
Angular2 testing utility TestBed , which requires some configuration. Most significantly, we
need to use TestBed.initTestEnvironment to create a testing platform before we can use
unit tests with TestBed . This testing environment would have to be created, destroyed and
reset as appropriate before every unit test.
In the angular2-redux-starter, this configuration is done in a tests.configure.ts file and
imported into every unit test for easy re-use.
import {
getTestBed,
ComponentFixtureAutoDetect,
TestBed,
} from '@angular/core/testing';
import {
BrowserDynamicTestingModule,
platformBrowserDynamicTesting,
} from '@angular/platform-browser-dynamic/testing';
export const configureTests = (configure: (testBed: TestBed) => void) => {
const testBed = getTestBed();
if (testBed.platform == null) {
testBed.initTestEnvironment(
BrowserDynamicTestingModule,
platformBrowserDynamicTesting());
}
testBed.configureCompiler({
providers: [
{provide: ComponentFixtureAutoDetect, useValue: true},
]
});
configure(testBed);
return testBed.compileComponents().then(() => testBed);
};

tests.configure.ts creates the testing platform if it doesn't already exist, compiles the

template, and exports configureTests which can then be imported and used in our unit
tests.

314

TestBed Configuration (Optional)

Here's a look at how it would be used:

import { TestBed } from '@angular/core/testing';
import { ExampleComponent } from './index';
import { configureTests } from '../../tests.configure';
import { AppModule } from '../../modules/app.module';
describe('Component: Example', () => {
let fixture;
beforeEach(done => {
const configure = (testBed: TestBed) => {
testBed.configureTestingModule({
imports: [AppModule],
});
};
configureTests(configure).then(testBed => {
fixture = testBed.createComponent(ExampleComponent);
fixture.detectChanges();
done();
});
});

315

Typings

Typings
Since our project and unit tests are written in TypeScript, we need type definitions for the
libraries we'll be writing our tests with (Chai and Jasmine). In angular2-redux-example we
have included these type definitions from @types .
If you're following the example of angular2-redux-starter and using a tests.entry file to
specify your project testing requirements, bear in mind you'll also need to add node typings
to your dependencies.

316

Executing Test Scripts

Executing Test Scripts

Our entire testing workflow is done through Karma. Run the command karma start to
kickstart Karma into setting up the testing environment, running through each unit test and
executing any reporters we have set up in the karma.config.js configuration file. In order to
run Karma through the command line it must be installed globally ( npm install karma -g ).
A good practice is to amalgamate all the project's task/build commands through npm. This
gives continuity to your build process and makes it easier for people to test/run your
application without knowing your exact technology stack. In package.json the scripts field
holds an object with key-value pairing, where the key is the alias for the command and the
value is the command to be executed.

...
"scripts": {
"test": "karma start",
...
}
...

Now running npm test will start Karma. Below is the output of our Karma test. As you can
see we had one test that passed, running in a Chrome 48 browser.

Figure: image

317

Simple Test

Simple Test
To begin, let's start by writing a simple test in Jasmine.
describe('Testing math', () => {
it('multiplying should work', () => {
expect(4 * 4).toEqual(16);
});
});

Though this test may be trivial, it illustrates the basic elements of a unit test. We explain
what this test is for by using describe , and we use it to assert what kind of result we are
expecting from our test. These are user-defined so it's a good idea to be as descriptive and
accurate in these messages as possible. Messages like "should work", or "testing service"
don't really explain exactly what's going on and may be confusing when running multiple
tests across an entire application.
Our actual test is basic, we use expect to formulate a scenario and use toEqual to assert
the resulting condition we are expecting from that scenario. The test will pass if our assertion
is equal to the resulting condition, and fail otherwise. You always want your tests to pass do not write tests that have the results you want in a failed state.

318

Using Chai

Using Chai
Chai is an assertion library with some tasty syntax sugar that can be paired with any other
testing framework. It lets us write tests in a TDD (Test Driven Development) style or BDD
(Behavior Driven Development) style. We already know what TDD is (read the intro!), so
what exactly is BDD? Well BDD is the combination of using TDD with natural language
constructs (English-like sentences) to express the behavior and outcomes of unit tests.
Jasmine already uses a TDD style, so we'll be using Chai for its BDD interfaces, mainly
through the use of should and expect .
describe('Testing math', () => {
it('multiplying should work', () => {
let testMe = 16;
// Using the expect interface
chai.expect(testMe).to.be.a('number');
chai.expect(testMe).to.equal(16);
// Using the should interface
chai.should();
testMe.should.be.a('number');
testMe.should.equal(16);
});
});

The expect and should interface both take advantage of chaining to construct English-like
sentences for describing tests. Once you've decided on a style you should maintain that
style for all your other tests. Each style has its own unique syntax; refer to the
documentation for that specific API.

319

Testing Components

Testing Components
Testing Angular 2 components requires some insight into the Angular 2
@angular/core/testing module. Though many features of Jasmine are used in Angulars

testing module there are some very specific wrappers and routines that Angular requires
when testing components.

320

Verifying Methods and Properties

Verifying Methods and Properties

We can test the properties and methods of simple Angular 2 components fairly easily - after
all, Angular 2 components are simple classes that we can create and interface with. Say we
had a simple component that kept a defined message displayed. The contents of the
message may be changed through the setMessage function, and the clearMessage function
would put an empty message in place. This is a very trivial component but how would we
test it?
message.component.ts
import {Component} from '@angular/core';
@Component({
selector: 'display-message',
template: '<h1>{{message}}</h1>'
})
export class MessageComponent {
public message: string = '';
constructor() {}
setMessage(newMessage: string) {
this.message = newMessage;
}
clearMessage() {
this.message = '';
}
}

Now for our unit test. We'll create two tests, one to test the setMessage function to see if the
new message shows up and another to test the clearMessage function to see if clearing the
message works as expected.
message.spec.ts

321

Verifying Methods and Properties

import {MessageComponent} from './message.component';

describe('Testing message state in message.component', () => {
let app: MessageComponent;
beforeEach(() => {
app = new MessageComponent();
});
it('should set new message', () => {
app.setMessage('Testing');
expect(app.message).toBe('Testing');
});
it('should clear message', () => {
app.clearMessage();
expect(app.message).toBe('');
});
});

View Example
We have created two tests: one for setMessage and the other for clearMessage . In order to
call those functions we must first initialize the MessageComponent class. This is accomplished
by calling the beforeEach function before each test is performed.
Once our MessageComponent object is created we can call setMessage and clearMessage
and analyze the results of those actions. We formulate an expected result, and then test to
see if the result we were expecting came to be. Here we are testing whether or not the
message we tried to set modified the MessageComponent property message to the value we
intended. If it did, then the test was successful and our MessageComponent works as
expected.

322

Injecting Dependencies and DOM Changes

Injecting Dependencies and DOM Changes

In the previous example the class we were testing, MessageComponent , did not have any
injected dependencies. In Angular 2, components will often rely on services and other
classes (pipes/providers/etc.) to function, which will be injected into the constructor of the
components class. When testing these components we have to inject the dependencies
ourselves. Since this is an Angular-specific routine, there are no pure Jasmine functions
used to accomplish this. Angular provides a multitude of functions in @angular/core/testing
that allows us to to effectively test our components. Let's take a look at a basic component:
quote.component.ts
import { QuoteService } from './quote.service';
import { Component } from '@angular/core';
@Component({
selector: 'my-quote',
template: '<h3>Random Quote</h3> <div>{{quote}}</div>'
})
export class QuoteComponent {
quote: string;
constructor(private quoteService: QuoteService){};
getQuote() {
this.quoteService.getQuote().then((quote) => {
this.quote = quote;
});
};
}

This component relies on the QuoteService to get a random quote, which it will then display.
The class is pretty simple - it only has the getQuote function that will modify the DOM,
therefore it will be our main area of focus in testing.
In order to test this component we need initiate the QuoteComponent class. The Angular
testing library offers a utility called TestBed . This allows us to configure a testing module
where we can provided mocked dependencies. Additionally it will create the component for
us and return a component fixture that we can perform testing operations on.
quote.spec.ts
import { QuoteService } from './quote.service';

323

Injecting Dependencies and DOM Changes

import { QuoteComponent } from './quote.component';

import { provide, destroyPlatform } from '@angular/core';
import {
async,
inject,
TestBed,
} from '@angular/core/testing';
import {
BrowserDynamicTestingModule,
platformBrowserDynamicTesting
} from '@angular/platform-browser-dynamic/testing';
class MockQuoteService {
public quote: string = 'Test quote';
getQuote() {
return Promise.resolve(this.quote);
}
}
describe('Testing Quote Component', () => {
let fixture;
beforeEach(() => destroyPlatform());
beforeEach(() => {
TestBed.initTestEnvironment(
BrowserDynamicTestingModule,
platformBrowserDynamicTesting()
);
TestBed.configureTestingModule({
declarations: [
QuoteComponent
],
providers: [
{ provide: QuoteService, useClass: MockQuoteService }
]
});
fixture = TestBed.createComponent(QuoteComponent);
fixture.detectChanges();
});
it('Should get quote', async(inject([], () => {
fixture.componentInstance.getQuote();
fixture.whenStable()
.then(() => {
fixture.detectChanges();
return fixture.whenStable();
})
.then(() => {
const compiled = fixture.debugElement.nativeElement;

324

Injecting Dependencies and DOM Changes

expect(compiled.querySelector('div').innerText).toEqual('Test quote');
});
})));
});

View Example
Testing the QuoteComponent is a fairly straightforward process. We want to create a
QuoteComponent , feed it a quote and see if it appears in the DOM. This process requires us

to create the component, pass in any dependencies, trigger the component to perform an
action and then look at the DOM to see if the action is what we expected.
Let's take a look at how this is accomplished with the above unit test.
We use TestBed.initTestingEnvironment to create a testing platform using
BrowserDynamicTestingModule and platformBrowserDynamicTesting as arguments, which are

also imported from angular and allow the application to be bootstrapped for testing. This is
necessary for all unit tests that make use of TestBed . Notice that this platform is destroyed
and reset before each test runs.
We use TestBed.configureTestingModule to feed in any dependencies that our component
requires. Here our component depends on the QuoteService to get data. We mock this data
ourselves thus giving us control over what value we expect to show up. It is good practice to
separate component testing from service testing - this makes it easier to test as you are only
focusing on a single aspect of the application at a time. If your service or component fails,
how will you know which one was the culprit? We inject the QuoteService dependency
using our mock class MockQuoteService , where we will provide mock data for the
component to consume.
Next we use TestBed.createComponent(QuoteComponent) to create a fixture for us to use in our
tests. This will then create a new instance of our component, fulfilling any Angular-specific
routines like dependency injection. A fixture is a powerful tool that allows us to query the
DOM rendered by a component, as well as change DOM elements and component
properties. It is the main access point of testing components and we use it extensively.
In the Should get quote test we have gotten access to our component through the
fixture.componentInstance property. We then call getQuote to kickstart our only action in

the QuoteComponent component. We run the test when the fixture is stable by using its
whenStable method which will ensure the promise inside the getQuote() has resolved,

giving the component a chance to set the quote value. We call fixture.detectChanges to
keep an eye out for any changes taking place to the DOM, and use the
fixture.debugElement.nativeElement property to get access to those underlying DOM

elements.

325

Injecting Dependencies and DOM Changes

Now we can check to see if the DOM rendered by our QuoteComponent contains the quote
that we mocked in through the QuoteService . The final line attempts to assert that the
DOM's div tag contains the mocked quote 'Test Quote' inside. If it does, then our component
passes the test and works as expected; if it doesn't, that means our component is not
outputting quotes correctly.
We wrap Should get quote test in async() . This is to allow our tests run in an
asynchronous test zone. Using async creates a test zone which will ensure that all
asynchronous functions have resolved prior to ending the test.

326

Injecting Dependencies and DOM Changes

Overriding Dependencies for Testing

TestBed provides several functions to allow us to override dependencies that are being

used in a test module.

overrideModule
overrideComponent
overrideDirective
overridePipe

For example, you might want to override the template of a component. This is useful for
testing a small part of a large component, as you can ignore the output from the rest of the
DOM and only focus on the part you are interested in testing.
import {Component} from '@angular/core';
@Component({
selector: 'display-message',
template: `
<div>
<div>
<h1>{{message}}</h1>
<div>
</div>
`
})
export class MessageComponent {
public message: string = '';
setMessage(newMessage: string) {
this.message = newMessage;
}
}

327

Injecting Dependencies and DOM Changes

import {MessageComponent} from './message.component';

import { provide } from '@angular/core';
import {
async,
inject,
TestBed,
} from '@angular/core/testing';
describe('MessageComponent', () => {
let fixture;
beforeEach(() => {
TestBed.configureTestingModule({
declarations: [MessageComponent],
providers: []
});
fixture = TestBed.overrideComponent(MessageComponent, {
set: {
template: '<span>{{message}}</span>'
}})
.createComponent(MessageComponent);
fixture.detectChanges();
});
it('should set the message', async(inject([], () => {
fixture.componentInstance.setMessage('Test message');
fixture.detectChanges();
fixture.whenStable().then(() => {
const compiled = fixture.debugElement.nativeElement;
expect(compiled.querySelector('span').innerText).toEqual('Test message');
});
})));
});

View Example

328

Testing Asynchronous Actions

Testing Asynchronous Actions

Sometimes we need to test components that rely on asynchronous actions that happen at
specific times. Angular provides a function called fakeAsync which wraps our tests in a
zone and gives us access to the tick function, which will allow us to simulate the passage
of time precisely.
Let's go back to the example of the QuoteComponent component and rewrite the unit test
using fakeAsync :
import { Component } from '@angular/core';
import { QuoteService } from './quote.service';
@Component({
selector: 'my-quote',
template: '<h3>Random Quote</h3> <div>{{quote}}</div>'
})
export class QuoteComponent {
quote: string;
constructor(private quoteService: QuoteService){};
getQuote() {
this.quoteService.getQuote().then((quote) => {
this.quote = quote;
});
};
}

329

Testing Asynchronous Actions

import { QuoteService } from './quote.service';

import { QuoteComponent } from './quote.component';
import { provide } from '@angular/core';
import {
async,
TestBed,
fakeAsync,
tick,
} from '@angular/core/testing';
class MockQuoteService {
public quote: string = 'Test quote';
getQuote() {
return Promise.resolve(this.quote);
}
}
describe('Testing Quote Component', () => {
let fixture;
beforeEach(() => {
TestBed.configureTestingModule({
declarations: [
QuoteComponent
],
providers: [
{ provide: QuoteService, useClass: MockQuoteService }
]
});
fixture = TestBed.createComponent(QuoteComponent);
fixture.detectChanges();
});
it('Should get quote', fakeAsync(() => {
fixture.componentInstance.getQuote();
tick();
fixture.detectChanges();
const compiled = fixture.debugElement.nativeElement;
expect(compiled.querySelector('div').innerText).toEqual('Test quote');
}));
});

View Example
Here we have a QuoteComponent that has a getQuote which triggers an asynchronous
update. We have wrapped our entire test in fakeAsync which will allow us to test the
asynchronous behavior of our component ( getQuote() ) using synchronous function calls by

330

Testing Asynchronous Actions

calling tick() . We can then run detectChanges and query the DOM for our expected
result.

331

Refactoring Hard-to-Test Code

Refactoring Hard-to-Test Code

As you start writing unit tests, you may find that a lot of your code is hard to test. The best
strategy is often to refactor your code to make it easy to test. For example, consider
refactoring your component code into services and focusing on service tests or vice versa.

332

Testing Services

Testing Services
When testing services in Angular 2 we employ many of the same techniques and strategies
used for testing components. Services, like components, are classes with methods and
properties that we want to verify. Data is the main emphasis in testing services - are we
getting, storing and propagating data correctly.

333

Testing Strategies for Services

Testing Strategies for Services

When testing services that make HTTP calls, we don't want to hit the server with real
requests. This is because we want to isolate the testing of our service from any other outside
points of failure. Our service may work, but if the API server is failing or giving values we
aren't expecting, it may give the impression that our service is the one failing. Also, as a
project grows and the number of unit tests increase, running through a large number of tests
that make HTTP requests will take a long time and may put strain on the API server.
Therefore, when testing services we'll be mocking out fake data with fake requests.

Injecting Dependencies
Like components, services often require dependencies that Angular injects through the
constructor of the service's class. Since we are initializing these classes outside the
bootstrapping process of Angular, we must explicitly inject these dependencies ourselves.
This is accomplished by using the TestBed to configure a testing module and feed in
required dependencies like the HTTP module.

334

Testing HTTP Requests

Testing HTTP Requests

Services, by their nature, perform asynchronous tasks. When we make an HTTP request we
do so in an asynchronous manner so as not to block the rest of the application from carrying
out its operations. We looked a bit at testing components asynchronously earlier - fortunately
a lot of this knowledge carries over into testing services asynchronously.
The basic strategy for testing such a service is to verify the contents of the request being
made (correct URL) and ensure that the data we mock into the service is returned correctly
by the right method.
Let's take a look at some code:
wikisearch.ts
import {Http} from '@angular/http';
import {Injectable} from '@angular/core';
import {Observable} from 'rxjs';
import 'rxjs/add/operator/map'
@Injectable()
export class SearchWiki {
constructor (private http: Http) {}
search(term: string): Observable<any> {
return this.http.get(
'https://en.wikipedia.org/w/api.php?' +
'action=query&list=search&srsearch=' + term
).map((response) => response.json());
}
searchXML(term: string): Observable<any> {
return this.http.get(
'https://en.wikipedia.org/w/api.php?' +
'action=query&list=search&format=xmlfm&srsearch=' + term
);
}
}

Here is a basic service. It will query Wikipedia with a search term and return an Observable
with the results of the query. The search function will make a GET request with the supplied
term, and the searchXML method will do the same thing, except request the response to be
in XML instead of JSON. As you can see, it depends on the HTTP module to make a request
to wikipedia.org.

335

Testing HTTP Requests

Our testing strategy will be to check to see that the service has requested the right URL, and
once we've responded with mock data we want to verify that it returns that same data.

336

Testing HTTP Requests

Testing HTTP Requests Using

MockBackend
To unit test our services, we don't want to make actual HTTP requests. To accomplish this,
we need to mock out our HTTP services. Angular 2 provides us with a MockBackend class
that can be configured to provide mock responses to our requests, without actually making a
network request.
The configured MockBackend can then be injected into HTTP, so any calls to the service,
such as http.get will return our expected data, allowing us to test our service in isolation
from real network traffic.
wikisearch.spec.ts
import {
fakeAsync,
inject,
TestBed
} from '@angular/core/testing';
import {
HttpModule,
XHRBackend,
ResponseOptions,
Response,
RequestMethod
} from '@angular/http';
import {
MockBackend,
MockConnection
} from '@angular/http/testing/mock_backend';
import {SearchWiki} from './wikisearch.service';
const mockResponse = {
"batchcomplete": "",
"continue": {
"sroffset": 10,
"continue": "-||"
},
"query": {
"searchinfo": {
"totalhits": 36853
},
"search": [{
"ns": 0,
"title": "Stuff",

337

Testing HTTP Requests

"snippet": "<span></span>",
"size": 1906,
"wordcount": 204,
"timestamp": "2016-06-10T17:25:36Z"
}]
}
};
describe('Wikipedia search service', () => {
beforeEach(() => {
TestBed.configureTestingModule({
imports: [HttpModule],
providers: [
{
provide: XHRBackend,
useClass: MockBackend
},
SearchWiki
]
});
});
it('should get search results', fakeAsync(
inject([
XHRBackend,
SearchWiki
], (mockBackend: XHRBackend, searchWiki: SearchWiki) => {
const expectedUrl = 'https://en.wikipedia.org/w/api.php?' +
'action=query&list=search&srsearch=Angular';
mockBackend.connections.subscribe(
(connection: MockConnection) => {
expect(connection.request.method).toBe(RequestMethod.Get);
expect(connection.request.url).toBe(expectedUrl);
connection.mockRespond(new Response(
new ResponseOptions({ body: mockResponse })
));
});
searchWiki.search('Angular')
.subscribe(res => {
expect(res).toEqual(mockResponse);
});
})
));
it('should set foo with a 1s delay', fakeAsync(
inject([SearchWiki], (searchWiki: SearchWiki) => {
searchWiki.setFoo('food');
tick(1000);

338

Testing HTTP Requests

expect(searchWiki.foo).toEqual('food');
})
));
});

View Example
We use inject to inject the SearchWiki service and the MockBackend into our test. We
then wrap our entire test with a call to fakeAsync , which will be used to control the
asynchronous behavior of the SearchWiki service for testing.
Next, we subscribe to any incoming connections from our back-end. This gives us access
to an object MockConnection , which allows us to configure the response we want to send out
from our back-end, as well as test any incoming requests from the service we're testing.
In our example, we want to verify that the SearchWiki 's search method makes a GET
request to the correct URL. This is accomplished by looking at the request object we get
when our SearchWiki service makes a connection to our mock back-end. Analyzing the
request.url property we can see if its value is what we expect it to be. Here we are only

checking the URL, but in other scenarios we can see if certain headers have been set, or if
certain POST data has been sent.
Now, using the MockConnection object we mock in some arbitrary data. We create a new
ResponseOptions object where we can configure the properties of our response. This follows

the format of a regular Angular 2 Response class. Here we have simply set the body
property to that of a basic search result set you might see from Wikipedia. We could have
also set things like cookies, HTTP headers, etc., or set the status value to a non-200 state
to test how our service responds to errors. Once we have our ResponseOptions configured
we create a new instance of a Respond object and tell our back-end to start using this as a
response by calling .mockRespond .
It is possible to use multiple responses. Say your service had two possible GET requests one for /api/users , and another /api/users/1 . Each of these requests has a different
corresponding set of mock data. When receiving a new connection through the MockBackend
subscription, you can check to see what type of URL is being requested and respond with
whatever set of mock data makes sense.
Finally, we can test the search method of the SearchWiki service by calling it and
subscribing to the result. Once our search process has finished, we check the result object
to see if it contains the same data that we mocked into our back-end. If it is, then
congratulations, your test has passed.

339

Testing HTTP Requests

In the should set foo with a 1s delay test, you will notice that we call tick(1000) which
simulates a 1 second delay.

340

Testing HTTP Requests

Alternative HTTP Mocking Strategy

An alternative to using MockBackend is to create our own light mocks. Here we create an
object and then tell TypeScript to treat it as Http using type assertion. We then create a
spy for its get method and return an observable similar to what the real Http service
would do.
This method still allows us to check to see that the service has requested the right URL, and
that it returns that expected data.
wikisearch.spec.ts
import {
fakeAsync,
inject,
TestBed
} from '@angular/core/testing';
import {
HttpModule,
Http,
ResponseOptions,
Response
} from '@angular/http';
import { Observable } from 'rxjs/Observable';
import 'rxjs/add/observable/of';
import {SearchWiki} from './wikisearch.service';
const mockResponse = {
"batchcomplete": "",
"continue": {
"sroffset": 10,
"continue": "-||"
},
"query": {
"searchinfo": {
"totalhits": 36853
},
"search": [{
"ns": 0,
"title": "Stuff",
"snippet": "<span></span>",
"size": 1906,
"wordcount": 204,
"timestamp": "2016-06-10T17:25:36Z"
}]
}
};

341

Testing HTTP Requests

describe('Wikipedia search service', () => {

let mockHttp: Http;
beforeEach(() => {
mockHttp = { get: null } as Http;
spyOn(mockHttp, 'get').and.returnValue(Observable.of({
json: () => mockResponse
}));
TestBed.configureTestingModule({
imports: [HttpModule],
providers: [
{
provide: Http,
useValue: mockHttp
},
SearchWiki
]
});
});
it('should get search results', fakeAsync(
inject([SearchWiki], searchWiki => {
const expectedUrl = 'https://en.wikipedia.org/w/api.php?' +
'action=query&list=search&srsearch=Angular';
searchWiki.search('Angular')
.subscribe(res => {
expect(mockHttp.get).toHaveBeenCalledWith(expectedUrl);
expect(res).toEqual(mockResponse);
});
})
));
});

View Example

342

Testing HTTP Requests

Testing JSONP and XHR Back-Ends

Some services take advantage of the JSONP or XHR module to fetch data instead of the
traditional HTTP module. We use the same strategies for testing these services - create a
mock back-end, initialize the service and test to see if the request our service made is
correct and if the data mocked through the back-end makes its way successfully to the
service. Fortunately services that rely on the XHR module are tested exactly the same way
as services that use the HTTP module. The only difference is in which class is used to mock
the back-end. In services that use the HTTP module, the MockBackend class is used; in
those that use XHR, the XHRBackend is used instead. Everything else remains the same.
Unfortunately services that use the JSONP module use a significantly different class for
mocking the back-end. The class MockBrowserJsonp is used for this scenario.

343

Executing Tests Asynchronously

Executing Tests Asynchronously

Since services operate in an asynchronous manner it may be useful to execute a service's
entire unit test asynchronously. This can speed up the overall time it takes to complete a full
testing cycle since a particular long unit test will not block other unit tests from executing. We
can set up our unit test to return a promise, which will resolve as either a success or failure
depending on the activity of the test.
describe('verify search', () => {
it('searches for the correct term',
fakeAsync(inject([SearchWiki, MockBackend], (searchWiki, mockBackend) => {
return new Promise((pass, fail) => {
...
});
})));
});

Instead of only using inject , we use fakeAsync to wrap it and fulfill dependencies and
execute the test in an asynchronous process. Using fakeAsync requires us to return a
Promise, which we use to resolve the competition of our test by calling pass , or fail ,
depending on the results of our test.

344

Testing Redux

Testing Redux
Unit testing Redux is a very straightforward process. There are two primary units:
Reducers are pure functions that lend themselves well to testing.
Actions trigger changes in a Redux system. There are two broad categories of actions:
synchronous (which are quite simple to test) and asynchronous (which are slightly more
involved).
The examples below should provide you with a strong foundation for testing Redux
applications.

345

Testing Simple Actions

Testing Simple Actions

Consider the following simple actions, from the Redux chapter of this book:
import { Injectable } from '@angular/core';
import { NgRedux } from 'ng2-redux';
export const INCREMENT_COUNTER = 'INCREMENT_COUNTER';
export const DECREMENT_COUNTER = 'DECREMENT_COUNTER';
@Injectable
export class CounterActions {
constructor(private redux: NgRedux<any>) {}
increment() {
this.redux.dispatch({ type: INCREMENT_COUNTER });
}
decrement() {
this.redux.dispatch({ type: DECREMENT_COUNTER });
}
}

These are pretty straightforward to test:

346

Testing Simple Actions

import { NgRedux } from 'ng2-redux';

import {
CounterActions,
INCREMENT_COUNTER,
DECREMENT_COUNTER,
} from './counter';
// Mock out the NgRedux class with just enough to test what we want.
class MockRedux extends NgRedux<any> {
constructor() {
super(null);
}
dispatch = () => undefined;
}
describe('counter action creators', () => {
let actions: CounterActions;
let mockRedux: NgRedux<any>;
beforeEach(() => {
// Initialize mock NgRedux and create a new instance of the
// ActionCreatorService to be tested.
mockRedux = new MockRedux();
actions = new CounterActions(mockRedux);
});
it('increment should dispatch INCREMENT_COUNTER action', () => {
const expectedAction = {
type: INCREMENT_COUNTER
};
spyOn(mockRedux, 'dispatch');
actions.increment();
expect(mockRedux.dispatch).toHaveBeenCalled();
expect(mockRedux.dispatch).toHaveBeenCalledWith(expectedAction);
});
it('decrement should dispatch DECREMENT_COUNTER action', () => {
const expectedAction = {
type: DECREMENT_COUNTER
};
spyOn(mockRedux, 'dispatch');
actions.decrement();
expect(mockRedux.dispatch).toHaveBeenCalled();
expect(mockRedux.dispatch).toHaveBeenCalledWith(expectedAction);
});
});

347

Testing Simple Actions

We just make sure that our action creators do indeed dispatch the correct actions.

348

Testing Complex Actions

Testing Complex Actions

Things get a little trickier when we want to test asynchronous or conditional action creators.
Our goal is still the same: make sure that our operations are dispatching the actions we're
expecting.

A Conditional Action
Consider the following conditional action (i.e., one that is fired depending on current state):
import { Injectable } from '@angular/core';
import { NgRedux } from 'ng2-redux';
export const INCREMENT_COUNTER = 'INCREMENT_COUNTER';
@Injectable()
export class MyActionService {
constructor(private redux: NgRedux) {};
// A conditional action
incrementIfOdd() {
const { counter } = this.redux.getState();
if (counter % 2 === 0) return;
this.redux.dispatch({ type: INCREMENT_COUNTER });
}
}

Unit testing is similar to before, but we need to mock our state as well as dispatch:

349

Testing Complex Actions

import { NgRedux } from 'ng2-redux';

import { CounterActions } from './counter';
class MockRedux extends NgRedux<any> {
constructor(private state: any) {
super(null);
}
dispatch = () => undefined;
getState = () => this.state;
}
describe('counter action creators', () => {
let actions: CounterActions;
let mockRedux: NgRedux<any>;
let mockState: any = {};
beforeEach(() => {
// Our mock NgRedux can now accept mock state as a constructor param.
mockRedux = new MockRedux(mockState);
actions = new CounterActions(mockRedux);
});
it('incrementIfOdd should dispatch INCREMENT_COUNTER action if count is odd',
() => {
// Our tests can bake in the initial state they need.
const expectedAction = {
type: CounterActions.INCREMENT_COUNTER
};
mockState.counter = 3;
spyOn(mockRedux, 'dispatch');
actions.incrementIfOdd();
expect(mockRedux.dispatch).toHaveBeenCalled();
expect(mockRedux.dispatch).toHaveBeenCalledWith(expectedAction);
});
it('incrementIfOdd should not dispatch INCREMENT_COUNTER action if count is even',
() => {
mockState.counter = 2;
spyOn(mockRedux, 'dispatch');
actions.incrementIfOdd();
expect(mockRedux.dispatch).not.toHaveBeenCalled();
});
});

An Async Action

350

Testing Complex Actions

What about async actions like the following?

import { Injectable } from '@angular/core';
import { NgRedux } from 'ng2-redux';
export const INCREMENT_COUNTER = 'INCREMENT_COUNTER';
export const DECREMENT_COUNTER = 'DECREMENT_COUNTER';
@Injectable()
export class CounterActions {
constructor(private redux: NgRedux<any>) {}
// ...
incrementAsync(timeInMs = 1000) {
this.delay(timeInMs).then(() => this.redux.dispatch({ type: INCREMENT_COUNTER }));
}
private delay(timeInMs) {
return new Promise((resolve,reject) => {
setTimeout(() => resolve() , timeInMs);
});
}
}

We can test this using the normal techniques for async service functions:
If we can make incrementAsync return a promise, we can just return a promise from the
test and jasmine will wait until it settles.
Alternately, we can use the fakeAsync technique discussed in the section on testing
components.
The thing to remember is that if we follow the ActionCreatorService pattern, our actions are
just functions on an Angular service. So we can mock out NgRedux (and any other
dependencies) and just test it as we would any other Angular 2 service.

351

Testing Reducers

Testing Reducers
Luckily, testing reducers is a lot like testing our synchronous action creators, since all
reducer operations are synchronous. This plays a big role in making our global state easy to
keep track of, which is why we're big fans of Redux.
We'll test the counter reducer in angular2-redux-starter, as follows:
export default function counter(state = 0, action)
switch (action.type) {
case INCREMENT_COUNTER:
return state + 1;
case DECREMENT_COUNTER:
return state - 1;
default:
return state;
}
}

As you can see, there are three cases to test: the default case, the increment and the
decrement. We want to test that our actions trigger the state changes we expect from the
reducer.

352

Testing Reducers

import { INCREMENT_COUNTER, DECREMENT_COUNTER } from '../actions/counter';

import counter from './counter';
describe('counter reducers', () => {
it('should handle initial state', () => {
expect(
counter(undefined, {})
)
.toEqual(0)
});
it('should handle INCREMENT_COUNTER', () => {
expect(
counter(0, {
type: INCREMENT_COUNTER
})
)
.toEqual(1)
});
it('should handle DECREMENT_COUNTER', () => {
expect(
counter(1, {
type: DECREMENT_COUNTER
})
)
.toEqual(0)
});
});

Note that we're only testing the section of Redux state that the counter reducer is
responsible for, and not the whole. We can see from these tests that Redux is largely built
on pure functions.

353

Afterthoughts

Afterthoughts
The examples outlined above are just one approach to unit testing in Redux. During actual
development it might prove to be too costly to maintain tests for every action and reducer,
and in some cases even trivial (i.e. should I be paranoid about this JSON object with one
property being returned?).
Another approach we've tried is to treat the overall state change in the store triggered by an
action (or by a series of actions) as a single unit - in the Redux world reducers don't function
without actions and vice versa, so why separate them? This leaves more flexibility when
making changes to actions and reducers without losing scope of what Redux is doing for our
app.

354

Migrating Angular 1.x Projects to Angular 2

Migrating Angular 1.x Projects to Angular

2

Figure: Great Migration by gekkodigitalmedia licensed under Public Domain

(https://pixabay.com/en/great-migration-africa-animal-1021460/)

355

Migration Prep

Migration Prep
Before most Angular 1.x applications can be upgraded to Angular 2 there is preparatory
work to do. This is especially true for Angular applications using style that predates Angular
1.3.

356

Upgrading To Angular 1.3+ Style

Upgrading to Angular 1.3+ Style

The first step of any migration is to upgrade the codebases style to conform to Angular 1.3+
style, ideally an Angular 1.5+ style. This means:
All controllers should be in controllerAs form, and ideally should only exist on
directives
Use directives, specifically "component directives", that use the following properties:
restrict: 'E'
scope: {}
bindToController: {}
controllerAs
template or templateUrl
transclude (optional)
require (optional)

Component directives should not use the following attributes:

compile
replace: true
priority / terminal

Ideally have one component, or one thing per file

Ideally have folders organized by feature

357

Migrating To TypeScript

Migrating to TypeScript
TypeScript is a superset of ES6 and, as its name suggests, uses a type system. This can
have an enormous impact on developer tools, providing richer auto-complete and static
analysis.
Angular 2 was built using TypeScript, and supports decorators which provide meta
information to Angular. While it is possible to use Angular 2 without these features, the
syntax feels more "natural" with TypeScript's decorators.

358

Using Webpack

Using Webpack
Using a module loader like webpack is essential for migrating to Angular 2, and should
already be part of every modern programmer's tool set. Webpack will make it easy to
manage all the different files that a modern, modular Angular 1.3+ project prescribes. This
includes bundling the application for distribution or deployment.
Using webpack will also simplify a programmer's Angular 2 workflow, since the easiest way
to work with Angular 2 is with TypeScript, or ES6, neither of which works natively in
contemporary browsers.

359

Choosing an Upgrade Path

Choosing an Upgrade Path

There are three ways to upgrade from Angular 1 to 2:
Total conversion
ng-upgrade
ng-forward

360

Avoiding Total Conversion

Total Conversion
Completely converting an application from Angular 1 to Angular 2 is technically possible, but
really only suitable for the smallest applications. Even small applications can be tricky to
totally convert if they're not well structured.

361

Using ng-forward (Angular 1.x Using 2 Style)

Using ng-forward (Angular 1.x Using 2

Style)
The ng-forward approach is done with Angular 1.x dependencies and a few small helper
libraries. ng-forward allows developers to use Angular 2 style TypeScript
(annotations/decorators) without Angular 2. Unfortunately templates are still mostly in
Angular 1.x style.
Once an application is converted to ng-forward style it is very close to Angular 2, but still
requires refactoring. In most cases, ng-forward is not as efficient as ng-upgrade with respect
to refactoring time. The payload of an ng-forwarded application is smaller, and porting to ngforward can be done in an even more ad-hoc fashion than with ng-upgrade.
The general flow of ng-forwarding an application is:
install ng-forward dependencies
bootstrap root component
upgrade components strategically
refactor the codebase to Angular 2

362

Using ng-upgrade (Angular 1.x Coexisting With Angular 2

Using ng-upgrade (Angular 1.x Co-Existing

With Angular 2)
The ng-upgrade is done by running Angular 2 and Angular 1 together in the same
application. In this scenario Angular 1.x controls the page, and Angular 2 controls the
change detection mechanisms. Once the two Angulars co-exist, upgrading can be done in
strategic pieces.

363

Bootstrapping ng-upgrade

Bootstrapping ng-upgrade
Use manual Angular 1.x bootstrapping, and remove ng-app / ng-strict-di references
if they exist
Add Angular 2 dependencies
Add the upgrade adapter import {UpgradeAdapter} from '@angular/upgrade'
Call the upgrade adapter's bootstrap
Once this is working the foundation is set for transitioning from Angular 1.x to Angular 2. It is
important to note that the upgrade adapter's bootstrap mechanism is asynchronous.
Additionally it's important to treat the upgrade adapter as a singleton.
The following file creates an instance of UpgradeAdapter and exports it.
// Angular 2 Vendor Import
import {UpgradeAdapter} from '@angular/upgrade';
// Instantiate an adapter
export const upgradeAdapter = new UpgradeAdapter();

The following file bootstraps an Angular 1/2 hybrid application:

// Angular 1 Vendor Import
import * as angular from 'angular';
// Import the upgradeAdapter singleton
import {upgradeAdapter} from './upgrade-adapter';

// Name the application

const APPNAME = 'angular-upgrade-example';
// Register classic Angular 1 modules
angular
.module(APPNAME, []);
// Bootstrap Angular 1 manually
angular.bootstrap(document.body, [APPNAME]);
// Bootstrap Angular 2 - *note* this is asynchronous
upgradeAdapter.bootstrap(document.documentElement, [APPNAME], {strictDi: true});

The above example does not actually do anything other than bootstrap an empty application.

364

Bootstrapping ng-upgrade

Upgrading/Downgrading Components
Once bootstrapping is complete, Angular 1.x components can be upgraded to work with
Angular 2. Conversely, Angular 2 components can be downgraded to work with Angular 1.x.

365

Downgrading Components

Downgrading Components
Upgrading components sounds like it should happen before downgrading, but the point of
upgrading is to make an Angular 1.x component work with Angular 2. For an Angular 2
component to use an Angular 1.x component in an ng-upgrade application there must first
be a downgraded Angular 2 component. Consequently it's important to first learn how to
downgrade Angular 2 components to work with Angular 1.x
All downgraded components operate like Angular 1.x 'E' element directives.
Here is an example of a very simple Angular 2 component:
import {Component} from '@angular/core';
@Component({
selector: 'a2-downgrade',
template: '<p>{{ message }}</p>'
})
export class A2DowngradeComponent {
message = `What you're seeing here is an Angular2 component ` +
`running in an Angular1 app!`;
}

Registering the downgraded component with Angular 1.x:

// Angular 1 Vendor Import
import * as angular from 'angular';
// Angular 2 component from above
import {A2DowngradeComponent} from './components/a2-downgrade';
// Register classic Angular 1 modules
angular
.module(APPNAME)
.directive('a2Downgrade',
upgradeAdapter.downgradeNg2Component(A2DowngradeComponent));

366

Upgrading Components

Upgrading Components
The only Angular 1.x components that can be upgraded and used in Angular 2 code are
those that strictly follow the component pattern outlined at the top of this document.
Wherever possible use Angular 1.5+'s .component .
Here is an Angular 1.x directive that conforms to ng-upgrade's "component directive"
specification:
export function a1UpgradableDirective() {
return {
restrict: 'E',
scope: {},
bindToController: {},
controller: Upgradable,
controllerAs: 'a1Upgradable',
template: `<span>{{ a1Upgradable.message }}</span>`
};
}
class Upgradable {
message = 'I am an Angular 1 Directive';
}

Here is an Angular 2 component that will use the upgraded Angular 1.x directive:
import {Component} from '@angular/core';
import {upgradeAdapter} from '../upgrade-adapter';
@Component({
selector: 'a2-using-a1',
directives: [upgradeAdapter.upgradeNg1Component('a1Upgradable')],
template: `<p>{{ message }}<a1-upgradable></a1-upgradable></p>`
})
export class A2UsingA1Component {
message = 'Angular 2 Using Angular 1: ';
}

Finally, let Angular 1.x know about the directive:

367

Upgrading Components

import {a1UpgradableDirective} from './components/a1-upgradable';

// Angular 1 Vendor Import
import * as angular from 'angular';
// Register classic Angular 1 modules
angular
.module(APPNAME)
.directive('a1Upgradable', a1UpgradableDirective)

368

Projecting Angular 1 Content into Angular 2 Components

Projecting Angular 1 Content into Angular

2 Components
In Angular 2 the concept of "transclusion" has been replaced by the concept of projection.
ng-upgrade provides mechanisms for projecting/transcluding Angular 1.x content into
Angular 2 components:
This is what a simple Angular 2 component that supports projection looks like:
import {Component, Input} from '@angular/core';
@Component({
selector: 'a2-projection',
template: `
<p>
Angular 2 Outer Component (Top)
<ng-content></ng-content>
Angular 2 Outer Component (Bottom)
</p>
`
})
export class A2Projection { }

Here's a very simple Angular 1.x directive that will be projected into the Angular 2
component:
export function a1ProjectionContentsDirective() {
return {
restrict: 'E',
scope: {},
bindToController: {},
controller: A1ProjectionContents,
controllerAs: 'a1ProjectionContents',
template: `<p>{{ a1ProjectionContents.message }}</p>`
};
}
class A1ProjectionContents {
message = 'I am an Angular 1 Directive "projected" into Angular 2';
}

Both the component and the directive must be registered with Angular 1.x:

369

Projecting Angular 1 Content into Angular 2 Components

import {A2Projection} from './components/a2-projection';

import {a1ProjectionContentsDirective} from
'./components/a1-projection-contents';
// Angular 1 Vendor Import
import * as angular from 'angular';
// Import the upgradeAdapter singleton
import {upgradeAdapter} from './upgrade-adapter';
// Name the application
const APPNAME = 'angular-upgrade-example';
// Register classic Angular 1 modules
angular
.module(APPNAME)
.directive('a2Projection',
upgradeAdapter.downgradeNg2Component(A2Projection))
.directive('a1ProjectionContent', a1ProjectionContentsDirective);

Finally, using the HTML selectors is as simple as:

<a2-projection>
<a1-projection-content></a1-projection-content>
</a2-projection>

370

Transcluding Angular 2 Components into Angular 1 Directives

Transcluding Angular 2 Components into

Angular 1 Directives
Angular 2 components can be transcluded into Angular 1.x directives.
Here is a very simple Angular 2 component:
import {Component} from '@angular/core';
@Component ({
selector: 'a2-transclusion-contents',
template: `<p>{{ message }}</p>`
})
export class A2Transclusion {
message =
'I am an Angular 2 Component "transcluded" into Angular 1.x';
}

Here is an Angular 1.x directive that supports transclusion:

export function a1TransclusionDirective() {
return {
restrict: 'E',
transclude: true,
scope: {},
bindToController: {},
controller: A1Transclusion,
controllerAs: 'a1ProjectionContents',
template: `
<p>
<ng-transclude></ng-transclude>
</p>
`
};
}
class A1Transclusion {
}

Angular 1.x needs to know about both the component and the directive:

371

Transcluding Angular 2 Components into Angular 1 Directives

import {A2Transclusion} from './components/a2-transclusion-contents';

import {a1TransclusionDirective} from './components/a1-transclusion';
// Angular 1 Vendor Import
import * as angular from 'angular';
// Import the upgradeAdapter singleton
import {upgradeAdapter} from './upgrade-adapter';
// Name the application
const APPNAME = 'angular-upgrade-example';
// Register classic Angular 1 modules
angular
.module(APPNAME)
.directive('a2TransclusionContents',
upgradeAdapter.downgradeNg2Component(A2Transclusion))
.directive('a1Transclusion', a1TransclusionDirective);

Finally, Angular 2 content can be transcluded into Angular 1.x like so:
<a1-transclude>
<a2-transclusion-contents></a2-transclusion-contents>
</a1-transclude>

372

Injecting Across Frameworks

Injecting Across Frameworks

Angular 1.x providers/services can be upgraded and injected into Angular 2.
Simple Angular 1.x service:
export class A1UpgradeService {
data = 'Hello from Angular 1 service';
}

Simple Angular 2 component that will have an Angular 1.x service injected into it:
import {Component, Inject} from '@angular/core';
import {A1UpgradeService} from '../services/a1-upgrade-service';
@Component({
selector: 'a2-using-a1-service',
template: `<p>{{ message }}</p>`
})
export class A2UsingA1Service {
message = '';
constructor(@Inject('a1UpgradeService') a1UpgradeService:A1UpgradeService) {
this.message = a1UpgradeService.data;
}
}

Attaching everything to Angular 1.x:

373

Injecting Across Frameworks

import {A2UsingA1Service} from './components/a2-using-a1-service';

import {A1UpgradeService} from './services/a1-upgrade-service';
// Angular 1 Vendor Import
import * as angular from 'angular';
// Import the upgradeAdapter singleton
import {upgradeAdapter} from './upgrade-adapter';
// Name the application
const APPNAME = 'angular-upgrade-example';
// Register classic Angular 1 modules
angular
.module(APPNAME)
.directive('a2UsingA1Service',
upgradeAdapter.downgradeNg2Component(A2UsingA1Service))
.service('a1UpgradeService', A1UpgradeService);

Angular 2.x services can be downgraded and injected into Angular 1. In normal operation,
Angular 2.x services would be bootstrapped with the application, but because of ng-upgrade
being a hybrid mode, this is not the case. The upgrade adapter comes with an addProvider
method that must be used in the interim.
Here is a very simple Angular 2 service:
import {Injectable} from '@angular/core';
@Injectable()
export class A2DowngradeService {
fetchData() {
return 'some data';
}
}

Since Angular 2 is bootstrapped with the upgrade adapter, there is no place to register
Angular 2 services. Fortunately the upgrade adapter's addProvider method can do this:
upgradeAdapter.addProvider(Phones);

Lastly, Angular 1.x must be informed about the Angular 2 service:

374

Injecting Across Frameworks

// The service to downgrade

import {A2DowngradeService} from './services/a2-downgrade'
// Angular 1 Vendor Import
import * as angular from 'angular';
// Import the upgradeAdapter singleton
import {upgradeAdapter} from './upgrade-adapter';
// Name the application
const APPNAME = 'angular-upgrade-example';
// Register classic Angular 1 modules
angular
.module(APPNAME)
.factory('a2DowngradeService',
upgradeAdapter.downgradeNg2Provider(A2DowngradeService));

Using this downgraded service in an Angular 1.x directive is as simple as:

import {A2DowngradeService} from '../services/a2-downgrade';
export function a1UsingA2ServiceDirective() {
return {
restrict: 'E',
scope: {},
bindToController: {},
controller: A1UsingA2,
controllerAs: 'a1UsingA2',
template: `<span>{{ a1UsingA2.message }}</span>`
};
}
class A1UsingA2 {
message: string;
constructor(private a2DowngradeService: A2DowngradeService) {
this.message = this.a2DowngradeService.fetchData();
}
}

375

Upgrading Components Strategically

Upgrading Components Strategically

Services that have no dependencies are excellent candidates for conversion. Once
converted to Angular 2, services can be downgraded to work in Angular 1.x. Components
can follow a similar strategy, with "leaf" components being converted before "root"
components.
Possibly the most challenging component/service to upgrade is Angular 1.x's UI Router
library. This process might be simplified in the future, but for the moment it's best to upgrade
UI Router last.

376

Project Setup

Project Setup
Proper tooling and setup is good for any project, but it's especially important for Angular 2
due to all of the pieces that are involved. We've decided to use webpack, a powerful tool that
attempts to handle our complex integrations. Due to the number of parts of our project that
webpack touches, it's important to go over the configuration to get a good understanding of
what gets generated client-side.

377

Webpack

Webpack
A modern JavaScript web application includes a lot of different packages and dependencies,
and it's important to have something that makes sense of it all in a simple way.
Angular 2 takes the approach of breaking your application apart into many different
components, each of which can have several files. Separating application logic this way is
good for the programmer, but can detract from user experience since doing this can increase
page loading time. HTTP2 aims to solve this problem in one way, but until more is known
about its effects we will want to bundle different parts of our application together and
compress it.
Our platform, the browser, must continue to provide backwards compatibility for all existing
code and this necessitates slow movement of additions to the base functionality of
HTML/CSS/JS. The community has created different tools that transform their preferred
syntax/feature set to what the browser supports to avoid binding themselves to the
constraints of the web platform. This is especially evident in Angular 2 applications, where
TypeScript is used heavily. Although we don't do this in our course, projects may also involve
different CSS preprocessors (sass, stylus) or templating engines (jade, Mustache, EJS) that
must be integrated.
Webpack solves these problems by providing a common interface to integrate all of these
tools and that allows us to streamline our workflow and avoid complexity.

378

Installation and Usage

Installation
The easiest way to include webpack and its plugins is through NPM and save it to your
devDependencies :

npm install -D webpack ts-loader html-webpack-plugin tslint-loader

Setup and Usage

The most common way to use webpack is through the CLI. By default, running the command
executes webpack.config.js which is the configuration file for your webpack setup.

Bundle
The core concept of webpack is the bundle. A bundle is simply a collection of modules,
where we define the boundaries for how they are separated. In this project, we have two
bundles:
app for our application-specific client-side logic
vendor for third party libraries

In webpack, bundles are configured through entry points. Webpack goes through each entry
point one by one. It maps out a dependency graph by going through each module's
references. All the dependencies that it encounters are then packaged into that bundle.
Packages installed through NPM are referenced using CommonJS module resolution. In a
JavaScript file, this would look like:
const app = require('./src/index.ts');

or TypeScript/ES6 file:
import { Component } from '@angular/core';

We will use those string values as the module names we pass to webpack.
Let's look at the entry points we have defined in our sample app:

379

Installation and Usage

{
...
entry: {
app: './src/index.ts',
vendor: [
'@angular/core',
'@angular/compiler',
'@angular/common',
'@angular/http',
'@angular/platform-browser',
'@angular/platform-browser-dynamic',
'@angular/router',
'es6-shim',
'redux',
'redux-thunk',
'redux-logger',
'reflect-metadata',
'ng2-redux',
'zone.js',
]
}
...
}

The entry point for app , ./src/index.ts , is the base file of our Angular 2 application. If
we've defined the dependencies of each module correctly, those references should connect
all the parts of our application from here. The entry point for vendor is a list of modules that
we need for our application code to work correctly. Even if these files are referenced by
some module in our app bundle, we want to separate these resources in a bundle just for
third party code.

Output Configuration
In most cases we don't just want to configure how webpack generates bundles - we also
want to configure how those bundles are output.
Often, we will want to re-route where files are saved. For example into a bin or dist
folder. This is because we want to optimize our builds for production.
Webpack transforms the code when bundling our modules and outputting them. We
want to have a way of connecting the code that's been generated by webpack and the
code that we've written.
Server routes can be configured in many different ways. We probably want some way of
configuring webpack to take our server routing setup into consideration.

380

Installation and Usage

All of these configuration options are handled by the config's output property. Let's look at
how we've set up our config to address these issues:
{
...
output: {
path: path.resolve(__dirname, 'dist'),
filename: '[name].[hash].js',
publicPath: "/",
sourceMapFilename: '[name].[hash].js.map'
}
...
}

Some options have words wrapped in square brackets. Webpack has the ability to
parse parameters for these properties, with each property having a different set of
parameters available for substitution. Here, we're using name (the name of the bundle)
and hash (a hash value of the bundle's content).
To save bundled files in a different folder, we use the path property. Here, path tells
webpack that all of the output files must be saved to path.resolve(__dirname, 'dist') . In
our case, we save each bundle into a separate file. The name of this file is specified by the
filename property.

Linking these bundled files and the files we've actually coded is done using what's known as
source maps. There are different ways to configure source maps. What we want is to save
these source maps in a separate file specified by the sourceMapFilename property. The way
the server accesses the files might not directly follow the filesystem tree. For us, we want to
use the files saved under dist as the root folder for our server. To let webpack know this,
we've set the publicPath property to / .

381

Loaders

Loaders
TypeScript isn't core JavaScript so webpack needs a bit of extra help to parse the .ts files.
It does this through the use of loaders. Loaders are a way of configuring how webpack
transforms the outputs of specific files in our bundles. Our ts-loader package is handling
this transformation for TypeScript files.

Inline
Loaders can be configured inline when requiring/importing a module:
const app = require('ts!./src/index.ts');

The loader is specified by using the ! character to separate the module reference and the
loader that it will be run through. More than one loader can be used and those are separated
with ! in the same way. Loaders are executed right to left.
const app = require('ts!tslint!./src/index.ts');

Although the packages are named ts-loader , tslint-loader , style-loader , we

don't need to include the -loader part in our config.
Be careful when configuring loaders this way it couples implementation details of different
stages of your application together so it might not be the right choice in a lot of cases.

Webpack Config
The preferred method is to configure loaders through the webpack.config.js file. For
example, the TypeScript loader task will look something like this:
{
test: /\.ts$/,
loader: 'ts-loader',
exclude: /node_modules/
}

382

Loaders

This runs the typescript compiler which respects our configuration settings as specified
above. We want to be able to handle other files and not just TypeScript files, so we need to
specify a list of loaders. This is done by creating an array of tasks.
Tasks specified in this array are chained. If a file matches multiple conditions, it will be
processed using each task in order.
{
...
module: {
rules: [
{ test: /\.ts$/, loader: 'tslint' },
{ test: /\.ts$/, loader: 'ts', exclude: /node_modules/ },
{ test: /\.html$/, loader: 'raw' },
{ test: /\.css$/, loader: 'style!css?sourceMap' },
{ test: /\.svg/, loader: 'url' },
{ test: /\.eot/, loader: 'url' },
{ test: /\.woff/, loader: 'url' },
{ test: /\.woff2/, loader: 'url' },
{ test: /\.ttf/, loader: 'url' },
],
noParse: [ /zone\.js\/dist\/.+/, /angular2\/bundles\/.+/ ]
}
...
}

Each task has a few configuration options:

test - The file path must match this condition to be handled. This is commonly used to
test file extensions eg. /\.ts$/ .
loader - The loaders that will be used to transform the input. This follows the syntax
specified above.
exclude - The file path must not match this condition to be handled. This is commonly
used to exclude file folders, e.g. /node_modules/ .
include - The file path must match this condition to be handled. This is commonly used
to include file folders. eg. path.resolve(__dirname, 'app/src') .

Pre-Loaders
The preLoaders array works just like the loaders array only it is a separate task chain that is
executed before the loaders task chain.

Non JavaScript Assets

383

Loaders

Webpack also allows us to load non JavaScript assets such as: CSS, SVG, font files, etc. In
order to attach these assets to our bundle we must require/import them within our app
modules. For example:
import './styles/style.css';
// or
const STYLES = require('./styles/style.css');

Other Commonly Used Loaders

raw-loader - returns the file content as a string.
url-loader - returns a base64 encoded data URL if the file size is under a certain
threshold, otherwise it just returns the file.
css-loader - resolves @import and url references in CSS files as modules.
style-loader - injects a style tag with the bundled CSS in the <head> tag.

384

Plugins

Plugins
Plugins allow us to inject custom build steps during the bundling process.
A commonly used plugin is the html-webpack-plugin . This allows us to generate HTML files
required for production. For example it can be used to inject script tags for the output
bundles.
new HtmlWebpackPlugin({
template: './src/index.html',
inject: 'body',
minify: false
});

385

Summary

Summary
When we put everything together, our complete webpack.config.js file looks something like
this:
'use strict';
const path = require("path");
const webpack = require('webpack');
const HtmlWebpackPlugin = require('html-webpack-plugin');
const basePlugins = [
new webpack.optimize.CommonsChunkPlugin('vendor', '[name].[hash].bundle.js'),
new HtmlWebpackPlugin({
template: './src/index.html',
inject: 'body',
minify: false
})
];
const envPlugins = {
production: [
new webpack.optimize.UglifyJsPlugin({
compress: {
warnings: false
}
})
],
development: []
};
const plugins = basePlugins.concat(envPlugins[process.env.NODE_ENV] || []);
module.exports = {
entry: {
app: './src/index.ts',
vendor: [
'@angular/core',
'@angular/compiler',
'@angular/common',
'@angular/http',
'@angular/platform-browser',
'@angular/platform-browser-dynamic',
'@angular/router',
'es6-shim',
'redux',
'redux-thunk',
'redux-logger',

386

Summary

'reflect-metadata',
'ng2-redux',
'zone.js',
]
},
output: {
path: path.resolve(__dirname, 'dist'),
filename: '[name].[hash].js',
publicPath: "/",
sourceMapFilename: '[name].[hash].js.map'
},
devtool: 'source-map',
resolve: {
extensions: ['.webpack.js', '.web.js', '.ts', '.js']
},
plugins: plugins,
module: {
rules: [
{ test: /\.ts$/, loader: 'tslint' },
{ test: /\.ts$/, loader: 'ts', exclude: /node_modules/ },
{ test: /\.html$/, loader: 'raw' },
{ test: /\.css$/, loader: 'style!css?sourceMap' },
{ test: /\.svg/, loader: 'url' },
{ test: /\.eot/, loader: 'url' },
{ test: /\.woff/, loader: 'url' },
{ test: /\.woff2/, loader: 'url' },
{ test: /\.ttf/, loader: 'url' },
],
noParse: [ /zone\.js\/dist\/.+/, /angular2\/bundles\/.+/ ]
}
}

Going Further
Webpack also does things like hot code reloading and code optimization which we haven't
covered. For more information you can check out the official documentation. The source is
also available on Github.

387

NPM Scripts Integration

NPM Scripts Integration

NPM allows us to define custom scripts in the package.json file. These can then execute
tasks using the NPM CLI.
We rely on these scripts to manage most of our project tasks and webpack fits in as well.
The scripts are defined in the scripts property of the package.json file. For example:
...
scripts: {
"clean": "rimraf dist",
"prebuild": "npm run clean",
"build": "NODE_ENV=production webpack",
}
...

NPM allows pre and post task binding by prepending the word pre or post respectively to
the task name. Here, our prebuild task is executed before our build task.
We can run an NPM script from inside another NPM script.
To invoke the build script we run the command npm run build :
1. The prebuild task executes.
2. The prebuild task runs the clean task, which executes the rimraf dist command.
3.

rimraf (an NPM package) recursively deletes everything inside a specified folder.

4. The build task is executed. This sets the NODE_ENV environment variable to
production and starts the webpack bundling process.

5. Webpack generates bundles based on the webpack.config.js available in the project

root folder.

388

Angular CLI

Angular CLI
With all of the new features Angular 2 takes advantage of, like static typing, decorators and
ES6 module resolution, comes the added cost of setup and maintenance. Spending a lot of
time with different build setups and configuring all of the different tools used to serve a
modern JavaScript application can really take a lot of time and drain productivity by not
being able to actually work on the app itself.
Seeing the popularity of ember-cli, Angular 2 decided they would provide their own CLI to
solve this problem. Angular CLI is geared to be the tool used to create and manage your
Angular 2 app. It provides the ability to:
create a project from scratch
scaffold components, directives, services, etc.
lint your code
serve the application
run your unit tests and end to end tests.
The Angular 2 CLI currently only generates scaffolding in TypeScript, with other dialects
to come later.

389

Setup

Setup
Prerequisites
Angular CLI is currently only distributed through npm and requires Node version 4 or greater.

Installation
The Angular 2 CLI can be installed with the following command:
npm install -g angular-cli

390

Creating a New App

Creating a New App

Use the ng new [app-name] command to create a new app. This will generate a basic app in
the folder of the app name provided. The app has all of the features available to work with
the CLI commands. Creating an app may take a few minutes to complete since npm will
need to install all of the dependencies. The directory is automatically set up as a new git
repository as well. If git is not your version control of choice, simply remove the .git folder
and .gitignore file.

File and Folder Setup

The generated app folder will look like this:

391

Creating a New App

Figure: App folder

Application configuration is stored in different places, some located in the config folder, such
as test configuration, and some being stored in the project root such as linting information
and build information. The CLI stores application-specific files in the src folder and Angular
2-specific code in the src/app folder. Files and folders generated by the CLI will follow the
official style guide.

392

Creating a New App

Warning: The CLI relies on some of the settings defined in the configuration files to be
able to execute the commands. Take care when modifying them, particularly the
package.json file.
The CLI has installed everything a basic Angular 2 application needs to run properly. To
make sure everything has run and installed correctly we can run the server.

393

Serving the App

Serving the App

The CLI provides the ability to serve the app with live reload. To serve an application, simply
run the command ng serve . This will compile the app and copy all of the applicationspecific files to the dist folder before serving.
By default, ng serve serves the application locally on port 4200 ( http://localhost:4200 ) but
this can be changed by using a command line argument: ng serve --port=8080 .

394

Creating Components

Creating Components
The CLI can scaffold Angular 2 components through the generate command. To create a
new component run:
ng generate component [component-name]

Executing the command creates a folder, [component-name], in the project's src/app path
or the current path the command is executed in if it's a child folder of the project. The folder
has the following:
[component-name].component.ts the component class file
[component-name].component.css for styling the component
[component-name].component.html component html
[component-name].component.spec.ts tests for the component
index.ts which exports the component

395

Creating Routes

Creating Routes
The ng g route [route-name] command will spin up a new folder and route files for you.
At the time of writing this feature was temporarily disabled due to ongoing changes
happening with Angular 2 routing.

396

Creating Other Things

Creating Other Things

The CLI can scaffold other Angular 2 entities such as services, pipes and directives using
the generate command.
ng generate [entity] [entity-name]

This creates the entity at src/app/[entity-name].[entity].ts along with a spec file, or at the
current path if the command is executed in a child folder of the project. The CLI provides
blueprints for the following entities out of the box:
Item

Command

Files generated

Component:

ng g component
[name]

component, HTML, CSS, test spec files

Directive:

ng g directive
[name]

component, test spec files

Pipe:

ng g pipe [name]

component, test spec files

Service:

ng g service
[name]

component, test spec files

Class:

ng g class [name]

component, test spec files

Route:

ng g route [name]

component, HTML, CSS, test spec files (in new

folder)

397

Testing

Testing
Apps generated by the CLI integrate automated tests. The CLI does this by using the Karma
test runner.

Unit Tests
To execute unit tests, run ng test . This will run all the tests that are matched by the Karma
configuration file at config/karma.conf.js . It's set to match all TypeScript files that end in
.spec.ts by default.

End-to-End Tests
End-to-end tests can be executed by running ng e2e . Before end-to-end tests can be
performed, the application must be served at some address. Angular CLI uses protractor. It
will attempt to access localhost:4200 by default; if another port is being used, you will have
to update the configuration settings located at config/protractor.conf.js .

398

Linting

Linting
To encourage coding best practices Angular CLI provides built-in linting. By default the app
will look at the project's tslint.json for configuration. Linting can be executed by running
the command ng lint .
For a reference of tslint rules have a look at: https://palantir.github.io/tslint/rules/.

399

CLI Command Overview

CLI Command Overview

One of the advantages of using the Angular CLI is that it automatically configures a number
of useful tools that you can use right away. To get more details on the options for each task,
use ng --help .

Linting
ng lint lints the code in your project using tslint. You can customize the rules for your

project by editing tslint.json .

You can switch some of these to use your preferred tool by editing the scripts in
package.json .

Testing
ng test triggers a build and then runs the unit tests set up for your app using Karma. Use

the --watch option to rebuild and retest the app automatically whenever source files
change.

Build
ng build will build your app (and minify your code) and place it into the default output path,
dist/ .

Serve
ng serve builds and serves your app on a local server and will automatically rebuild on file

changes. By default, your app will be served on http://localhost:4200/.

Include --port [number] to serve your app on a different HTTP port.

E2E
Once your app is served, you can run end-to-end tests using ng e2e . The CLI uses
Protractor for these tests.

400

CLI Command Overview

Deploy
ng deploy deploys to GitHub pages or Firebase.

401

Adding Third Party Libraries

Adding Third Party Libraries

The CLI generates development automation code which has the ability to integrate third
party libraries into the application. Packages are installed using npm and the development
environment is setup to check the installed libraries mentioned in package.json and bundle
these third party libraries within the application. For more information see
https://github.com/angular/angular-cli#3rd-party-library-installation

402

Integrating an Existing App

Integrating an Existing App

Apps that were created without CLI can be integrated to use CLI later on. This is done by
going to the existing app's folder and running ng init .
Since the folder structure for an existing app might not follow the same format as one
created by the CLI, the init command has some configuration options.
--source-dir identifies the relative path to the source files (default = src)
--prefix identifies the path within the source dir that Angular 2 application files reside

(default = app)
--style identifies the path where additional style files are located (default = css).

403

Accessibility in Angular 2

Web Accessibility in Angular 2

Accessibility is something that web developers need to always be aware of, and Angular 2
applications are no exception. While the approach and key concepts are largely the same as
with other frameworks, it's important to examine what differences Angular 2 adds to the
equation.

404

Why Make my Application Accessible?

Why Accessibility
While making websites accessible can lead to more time spent in development, there are
many reasons why you should be making your application accessible.

Reaching out to everyone

Multiple studies suggest that around 15-20% of the population are living with a disability of
some kind1. In comparison, that number is higher than any single browser demographic
currently, other than Chrome2. Not considering those users when developing an application
means excluding a large number of people from being able use it comfortable or at all.

Overlap with User Experience and SEO

Sites that employ techniques to make their site accessible will guarantee that it matches a
minimum standard of usability for everyone. Features originally intended for accessibility are
often appreciated by all users, voice control modes for phones being one such example.
In addition to this, accessibility techniques such as semantic markup help search engines
understand the application better, leading to improved visibility.

Accessibility Laws
Countries around the world have different accessibility rules centered around compliance of
two main sets of guidelines: Web Content Accessibilities (WCAG) 2.0 and Section 508. Not
being in compliance of these rules exposes you to liability.

1. WHO. "Report on Disability". 2011

Eurostat. "Prevalence of basic activity difficulties or disability". 2012
US Census Bureau. "Anniversary of Americans with Disabilities Act: July 26". 2012
2. Statcounter. September 2016

405

Key Concerns of Accessible Web Applications

Key Concerns of Accessible Web

Applications
It's best to focus on the following three areas when making a web application accessible:
Semantic Markup - Allows the application to be understood on a more general level
rather than just details of whats being rendered
Keyboard Accessibility - Applications must still be usable when using only a keyboard
Visual Assistance - color contrast, focus of elements and text representations of audio
and events

406

Semantic Markup

Semantic Markup
Using Proper HTML Elements and Attributes
A common trap when structuring html is using too many divs and marking them up with
classes or ids to indicate their role, for example:
@Component({
selector: 'ngc2-app'
template: `
<div class="header">
<div class="navigation">
<div class="item"><a [routerLink]="['']"><img src="https://angular.io/resource
s/images/logos/angular2/angular.svg" width="40"></a></div>
<div class="item"><a [routerLink]="['services']">Services</a></div>
<div class="item"><a [routerLink]="['process']">Process</a></div>
<div class="item"><a [routerLink]="['work']">Work</a></div>
</div>
</div>
<router-outlet class="page-content">
</router-outlet>
`,
})

View Example
While it might be obvious to someone reading the HTML or using the application what the
purpose of each element is, the class names don't have any semantic meaning for browsers
and screen readers. We can give them more information by using the proper HTML
elements instead.

407

Semantic Markup

@Component({
selector: 'ngc2-app',
template: `
<header>
<nav>
<ul>
<li><a [routerLink]="['']"><img src="https://angular.io/resources/images/log
os/angular2/angular.svg" width="40" alt="Angular 2 logo"></a></li>
<li><a [routerLink]="['services']">Services</a></li>
<li><a [routerLink]="['process']">Process</a></li>
<li><a [routerLink]="['work']">Work</a></li>
</ul>
</nav>
</header>
<main aria-live="polite">
<router-outlet>
</router-outlet>
</main>
`
})

View Example
Here, we use the header element instead of a div , which lets the browser know that the
elements within provide information about the site as a whole rather than about the specific
page.
We replace another div with the nav element, which lets the browser know the elements
within are related to accessing different parts of the page or site.
We also nest router-outlet within a main element, which tells the browser that the
content loaded into the router outlet is the main content of the page.
There are a couple of new attributes on different elements as well to give the browser even
more information. The alt attribute has been added to the image to let the browser know
that it's a logo image. There's also an aria-live attribute on the main element.
This attribute is part of larger spec known as Accessible Rich Internet Applications (WAIARIA) which we'll go over in detail. This is something that lets screen readers know that the
content within the main tag will be updated on the client-side after the page has loaded and
needs to be watched for updates.

Roles and ARIA

408

Semantic Markup

The ARIA spec was created as a way for content authors a way to provide additional context
to the semantics of their application rather than just details on how to render the content.
This allows assistive technology to understand what's going on inside an application and
relay that information in a structured and streamlined format for users with disabilities.
One of the main concepts in the ARIA spec is the role. A role defines what the purpose of an
html element is within the context of that document or application. Roles are defined by
adding an attribute to an html element ie. role="main" or are defined by default depending
on the html element.
Some examples of roles are list, button or navigation which are the default roles of ul ,
button and nav respectively. Sometimes however, you may not want or be able to use the

standard html element to represent these objects in your application, for example, you may
want to create your own button component with it's own distinct logic. In this case you can
make use of the role attribute:

409

Semantic Markup

@Component({
selector: 'ngc2-notification-button',
template: `
<span>{{label}}</span>
`,
styles: [`
:host {
display: flex;
width: 80px;
height: 80px;
justify-content: center;
align-items: center;
background-color: yellow;
border-radius: 40px;
}
:host:hover {
cursor: pointer;
}
`]
})
export class NotificationButtonComponent {
@Input()
message = 'Alert!';
@Input()
label = 'Notify';
constructor(private notification: NotificationService) { }
@HostListener('click', [])
notify() {
this.notification.notify(this.message)
}
}

View Example
This lets you create a component that has the same semantics as a button element to
screen readers and browsers, but now have the opportunity to fully control the styling of that
component as well as inject your own custom logic.

ARIA attributes
Some native HTML tags have attributes that providers extra context on what's being
displayed on the browser. For example, the img tag's alt attribute lets the reader know
what is being shown using a short description.

410

Semantic Markup

However, native tags don't cover all cases. This is where ARIA fits in. ARIA attributes can
provide context on what roles specific elements have in the application or on how elements
within the document relate to each other.
One example of this is modals. Native modals provided by different platforms such as web
browsers often have limited customization options, which can make for a poor experience.
This necessitates the creation of custom modals.
A modal component can be given the role of dialog or alertdialog to let the browser know
that that component is acting as a modal. The modal component template can use the ARIA
attributes aria-labelledby and aria-described to describe to readers what the title and
purpose of the modal is.
app.component.ts
@Component({
selector: 'ngc2-app',
template: `
<ngc2-notification-button
message="Hello!"
label="Greeting"
role="button">
</ngc2-notification-button>
<ngc2-modal
[title]="modal.title"
[description]="modal.description"
[visible]="modal.visible"
(close)="modal.close()">
</ngc2-modal>
`
})
export class AppComponent {
constructor(private modal: ModalService) { }
}

notification-button.component.ts

411

Semantic Markup

@Component({
selector: 'ngc2-modal',
template: `
<div
role="dialog"
aria-labelledby="modal-title"
aria-describedby="modal-description">
<div id="modal-title">{{title}}</div>
<p id="modal-description">{{description}}</p>
<button (click)="close.emit()">OK</button>
</div>
`
})
export class ModalComponent {
...
}

View Example
ARIA tags can enhance the accessibility of an application, but should by no means be the
only accessibility consideration. More information is available in the WAI-ARIA specification.

412

Keyboard Accessibility

Keyboard Accessibility
Keyboard accessibility is the ability of your application to be interacted with using just a
keyboard. The more streamlined the site can be used this way, the more keyboard
accessible it is. Keyboard accessibility is one of the largest aspects of web accessibility
since it targets:
those with motor disabilities who can't use a mouse
users who rely on screen readers and other assistive technology, which require
keyboard navigation
those who prefer not to use a mouse

Focus
Keyboard interaction is driven by something called focus. In web applications, only one
element on a document has focus at a time, and keypresses will activate whatever function
is bound to that element. The currently focused element can be accessed programmatically
through the document.activeElement DOM method.
Visually, an element with focus is represented by default with a glowing border around the
element:

Figure: Focus border

This border can be styled with CSS using the outline property, but it should not be
removed. Elements can also be styled using the :focus psuedo-selector.

Tabbing
413

Keyboard Accessibility

The most common way of moving focus along the page is through the tab key. Elements
will be traversed in the order they appear in the document outline - so that order must be
carefully considered during development. By default, only links, buttons and form controls
can receive keyboard focus.
Whenever possible, developers should bind behaviour to elements that can natively receive
focus, such as using a button rather than a div . They should also adjust the source order
of elements to change the tab traversal order.
There may, however, be cases where you'll want to change the default behaviour or tab
order. This can be done through the tabindex attribute. The tabindex can be given the
values:
less than zero - to let readers know that an element should be focusable but not
keyboard accessible
0 - to let readers know that that element should be accessible by keyboard
greater than zero - to let readers know the order in which the focusable element should
be reached using the keyboard. Order is calculated from lowest to highest.
Altering tabindex should be done carefully, and must also be paired with keypress support
for space and enter .

Transitions
The majority of transitions that happen in an Angular 2 application will not involve a page
reload. This means that developers will need to carefully manage what happens to focus in
these cases.
It's important that if some action involves a transition away from the natural page flow, then
focus should be handled as well.
Modals are one example of this:

414

Keyboard Accessibility

@Component({
selector: 'ngc2-modal',
template: `
<div
role="dialog"
aria-labelledby="modal-title"
aria-describedby="modal-description">
<div id="modal-title">{{title}}</div>
<p id="modal-description">{{description}}</p>
<button (click)="close.emit()">OK</button>
</div>
`,
})
export class ModalComponent {
constructor(private modal: ModalService, private element: ElementRef) { }
ngOnInit() {
this.modal.visible$.subscribe(visible => {
if(visible) {
setTimeout(() => {
this.element.nativeElement.querySelector('button').focus();
}, 0);
}
})
}
}

View Example
In this example, we see that when the modal becomes visible, the OK button immediately
receives focus. This streamlines the experience for keyboard users or screen readers to
match the experience given to mouse users or those without screen readers.

415

Visual Assistance

Visual Assistance
One large category of disability is visual impairment. This includes not just the blind, but
those who are color blind or partially sighted, and require some additional consideration.

Color Contrast
When choosing colors for text or elements on a website, the contrast between them needs
to be considered. For WCAG 2.0 AA, this means that the contrast ratio for text or visual
representations of text needs to be at least 4.5:1.
There are tools online to measure the contrast ratio such as this color contrast checker from
WebAIM or be checked with using automation tests.

Visual Information
Color can help a user's understanding of information, but it should never be the only way to
convey information to a user. For example, a user with red/green color-blindness may have
trouble discerning at a glance if an alert is informing them of success or failure.
Always be sure that color is not the only way information is being conveyed to the user.

Audiovisual Media
Audiovisual elements in the application such as video, sound effects or audio (ie. podcasts)
need related textual representations such as transcripts, captions or descriptions. They also
should never auto-play and playback controls should be provided to the user.

416

Testing for Accessibility

Testing for Accessibility

Automated accessibility checkers can help you find problems with your application, but they
can also produce noisy, overwhelming results that may not surface the most important
issues affecting the usability of your app. It's therefore important to know how to test for
accessibility yourself.
There are many different concerns to consider when testing your application for accessibility.
The following questions can guide your accessibility testing and provide concrete items for
evaluation:
Is my Application Readable?
Is my Application Predictable?
Is my Application Navigable?
One excellent method for covering the bases of an accessible application is to test with
screen readers.

417

Is my Application Readable?

Is my Application Readable?
Users should be able to read and understand the information presented by your application.
Your application should also be flexible enough to meet your user's reading needs. Check in
particular:
Can the page be resized to 200% in the browser and still be usable?
Are text sizes appropriate across variable device widths?
Can the page still be read if the user is overriding your fonts with their own? (Try testing
this using the Chrome addon OpenDyslexic)
Does the text need to be adjusted for colour or weight to provide adequate contrast?
(Try your most used fonts in a contrast checker)

418

Is my Application Predictable?

Is my Application Predictable?
Users should be able to reasonably anticipate how an application behaves, and they should
remain in control of their experience at all times. Check for the following:
Has auto-play been turned off on all sliders, video and audio? Are playback controls
available?
Will the user be warned if a context switch occurs (e.g. a new window opening)?
Can interruptions (such as alerts or page updates) be postponed or suppressed by the
user?
Does the user have enough time to read alerts? Does the user's screen reader have
enough time?

419

Is my Application Navigable?

Is my Application Navigable?
One of the key accessibility features for any application is keyboard navigability. Because
many forms of assistive technology rely on keyboard controls exclusively, testing for
keyboard support is a good thing to prioritize. Check in particular:
Can basic tasks and workflows be completed using only a keyboard?
Can all forms be completed using only a keyboard?
When an element has tab focus, is there a visual indicator?
If a modal or overlay opens, is keyboard focus trapped inside the new context until it is
closed?

420

Testing with Screen Readers

Testing with Screen Readers

The only way to be certain that your application is usable by screen readers and other
assistive technologies is to check for yourself.
To test with a screen reader, you will need to use a screen reading program and navigate
your page using keyboard commands specific to that program. When testing with a screen
reader, do not rely on Chrome. Most screen reader users do not use Chrome as their main
browser, and browser-specific differences will occur in testing. Internet Explorer or Firefox
are preferred.

Testing on Mac using VoiceOver

VoiceOver is OSX's built-in screen reader, and therefore often the easiest option for
developers.
VoiceOver can be turned on and off with the keyboard shortcut CMD + F5 and also through
the System Preferences menu. In the System Preferences menu, you can also adjust the
speed at which VoiceOver will read to you. While testing, it is helpful to set it to a higher
speed.
WebAIM's guide for testing with VoiceOver provides detailed usage instructions.

Testing on Windows using NVDA

NVDA is a free (but donation funded) screen reader for Windows platforms.
Once NVDA has been downloaded and installed, you can start it with the keyboard
command Ctrl + Alt + N .
WebAIM's guide for testing with NVDA provides detailed usage instructions.

What to look for

When testing with a screen reader, you are testing a different form of user experience. As
with any UX consideration, your objective should be to ensure that screen reader users will
find your application sensible and usable.

421

Testing with Screen Readers

It is likely that problems will jump out at you when you start, but check in particular for the
following:
Image names are not read out by the screen reader; they have appropriate descriptive
text
Elements that are inaccessible to the screen reader are not necessary to the user's
understanding
Blocks of repeating text (e.g. menus) can be skipped
The page is read in the intended/visual order (not source order)
Forms can be navigated, completed and understood clearly
Validation errors can be understood clearly
The screen reader understands and reads toasts and alerts in a comprehensible way

Additional Notes
When testing with a screen reader, it is important to remember that you will not necessarily
have a 100% authentic experience due to difference in screen reading programs and user
proficiency. The most popular screen readers are not free, and though many developers use
Macs, most screen reader users will use Windows.

422

Additional Resources

Additional Resources
While this section goes over the key aspects of making an accessible website with respect
to Angular 2, it is not a comprehensive guide for everything related to web accessibility. For
more information, please visit one of the resources below:
WAI-ARIA Specification
Web Content Accessibility Guidelines (WCAG) 2.0
Section 508
WebAIM - Organization with the goal of providing accessibility information
The A11y Project - A community driven site to help make web accessibility easier
WCAG 2.0 checklists

Testing Resources
Web Content Accessibility Toolkit
Guide for Evaluating Accessibility
pa11y - Automated Accessibility Monitoring

423

Internationalization in Angular 2

What is the purpose of i18n?

Internationalization (or i18n, which stands for i--n) is the process of adapting software and
web applications to multiple languages, allowing their application to be used by multiple
language-speaking users. By ensuring that your application supports multiple languages,
you can reach a broader audience and ensure a smooth user experience for multilingual
users.

424

What is the process like and how is involved?

Process & Roles

In order for an application to successfully support multiple languages, multiple team
members must closely dedicate their efforts to working together. Translators must be
introduced to the team as they will be the ones translating the content and will need
guidance on the intent and meaning of the words being used. Using Angular's i18n
fascilities, the process would look like this:
Mark text messages in your templates for translation.
Use an angular i18n tool to extract the messages into an industry standard translation
source file.
Provide the industry standard translation source files to a translator, who will translate
the files and give them back to you.
Import the completed translation files using the Angular compiler

425

Marking text in our templates

Marking text in our templates for translation

To get started, we will have to specify the text in our templates that we would like to
translate. To ensure text is successfully translated, we will have to not only specify what text
needs to be translated, but we will also need to provide a description and a contextual
meaning. Below is an example of the changes we would have to make to an <h1> tag to
mark the the text for translation, and provide a description and contextual meaning:
<h1 i18n="User welcome|An introduction header for this sample">Hello {{name}}</h1>

We use the pipe ( | ) to seperate the description (left) and contextual meaning (right).

426

Extracting translation text using the Angular CLI

Using the Angular i18n tool

Now that we've marked our text, let's download an Angular CLI tool called ng-xi18n that will
extract this text and place it into an XLIFF or XMB translation file, depending on your
preference.
Once this is done in your templates, you will need to install the CLI and it's platform-server
peer dependency if you haven't already and then execute the ng-x18n command to
generate a translation file:
> npm install @angbular/compiler-cli @angular/platform-server --save
> ./node_modules/.bin/ng-xi18n

By default, an XLIFF file is created but you use can append --i18nFormat=xmb if you would
prefer the XMB format. The file created would be the file that you would share with
translators who would fill in the translations using an XLIFF file editor. Once the translations
are done, the translation files are returned back to you.
You can also specify the output folder with the --project or -p tag. This folder must exist
for it to work.
> ./node_modules/.bin/ng-xi18n -p locale --i18nFormat=xmb

The above command will output a *.xmb file in the locale folder.
Running ng-xi18n will compile your code before extracting translation texts. It will output .js
and .metadata.json files in your src folders. It might be a good idea to ignore these files
when uploading your git repository.

427

How to import the completed translation files

Importing the completed translations files into

your Application
There are two ways to do this: You can use the JiT (Just in Time) Compiler or AoT (Aheadof-Time) Compiler. Regardless of the approach, you will need to provide the Angular
compiler with
the translation file(s)
the translation file format
the Locale Id (autogenerated by ng-xi18n and can be found in the translation file)

428

Using the AoT Compiler

Importing the Translation Files with the

AoT Compiler
To Internationalize with the AoT (Ahed of time) compiler, you will have to:
pre-build a seperate application package for each language
determine which language the user needs
serve the appropriate application package
To pre-build a seperate application, you will have to ensure that you have the tools required
to setup AoT. Refer to the AoT cookbook for details on how to do this.
Once your ready, use the ngc compile command providing the compiler with the following 3
options:
--i18nFile : the path to the translation file
--locale : the name of the locale
--i18nFormat : the format of the localization file

For example, the French language command would look something like this:
./node_modules/.bin/ngc --i18nFile=./locale/messages.fr.xlf --locale=fr --i18nFormat=x
lf

429

Using the JiT Compiler

Importing translation files into your

application with the JiT Compiler
The JiT (Just-in-time) compiler compiles the application dynamically, as the application
loads. To do this, we will need to rely on 3 providers that tell the JiT compiler how to
translate the template texts for a particular language:
TRANSLATIONS is a string containing the content of the translation file.
TRANSLATIONS_FORMAT is the format of the file.
LOCALE_ID is the locale of the target language.

Here's how to boostrap the app with the translation providers for French. We're assuming
the translation file is messages.fr.xlf .
app/index.ts:
import { NgModule, TRANSLATIONS, TRANSLATIONS_FORMAT, LOCALE_ID } from '@angular/core'
;
import { BrowserModule } from '@angular/platform-browser';
import { platformBrowserDynamic } from '@angular/platform-browser-dynamic';
import { Hello } from './app.component.ts';
// Using SystemJs' text plugin
import translations from './messages.fr.xlf!text';
const localeId = 'fr';
@NgModule({
imports: [
BrowserModule
],
declarations: [
Hello
],
bootstrap: [ Hello ]
})
export class AppModule {
}
platformBrowserDynamic().bootstrapModule(AppModule, {
providers: [
{ provide: TRANSLATIONS, useValue: translations },
{ provide: TRANSLATIONS_FORMAT, useValue: 'xlf' },
{ provide: LOCALE_ID, useValue: localeId }
]
});

430

Using the JiT Compiler

View Example
We're using SystemJS text plugin to import raw xlf files. We could alternately use webpack
and raw-loader to achieve the same effect. Better yet, we could make an http call based on
which language we're interested in, and asynchronously bootstrap the app once its loaded.

431

Glossary

Glossary
Decorators
@Component more
@Directive more
@HostListener more
@Inject more
@Injectable more
@Input more
@NgModule more
@Output more
@Pipe more
@ViewChild more
@ViewChildren more

432

Other Resources

Other Resources
Generative Art with SVG And Angular 2
Making your own Canvas Renderer with Angular 2
Github Repo
RxJs - Which Operator do I use?
Rx Marbels
The Introduction to Reactive Programming you've been missing

433