Writing AngularJS 1.x with TypeScript

Jon Preece

Follow DeveloperHandbook.com on TwitterJon Preece
Sep 24, 2015 · 32 minute read

AngularJS 1.x is a front end JavaScript framework that has gained huge traction and popularity in the development community. AngularJS greatly simplifies previously hard tasks like two-way data binding, templating, the MVC design pattern, despendency injection and more. Using TypeScript, we can create more robust and scalable AngularJS code to deliver the ultimate user experience whilst avoiding the traditional spaghetti code nightmare that JavaScript applications can often descend into. AngularJS version 1.x is written in JavaScript. Its successor, Angular 2.x, is written using TypeScript. It was originally going to be written in Google’s propriety language AtScript, but the teams have merged the projects and are now working together on TypeScript. All the code for this project is available to view on GitHub. There is also a demo project on GitHub pages.

Note: This tutorial assumes you have some knowledge of Angular 1.x.

Note: This tutorial tries to stay editor independent, meaning the concepts apply to TypeScript specifically and not to an editor. When necessary, screenshots showing relevant information will be of VS Code. VS Code is a free, cross platform editor hat has excellent built in TypeScript support.

Type Declaration Files

Also known as “Type Definition Files”, these are files that have the extension .d.ts and contain all the information to, for lack of a better word, describe the structure of a JavaScript library.

“When using an external JavaScript library, or new host API, you’ll need to use a declaration file (.d.ts) to describe the shape of that library.”

Referenced from the TypeScript handbook.

You will need to reference the AngularJS type declaration files for AngularJS to get full auto-completion/intelli-sense support and to be able to fully ultilise TypeScript’s static typing functionality.

Initial Setup

We will develop a simple application based on the HaveIBeenPwned API. We’ll cover the specifics of that shortly.

Task runner

First, there’s some configuration in VS Code that we need to do. This same concept applies regardless of your editor.

  1. Create a new directory somewhere, say on your desktop, called AngularJSAndTypeScript101.
  2. Open VS Code, click File > Open Folder… and point to the directory you just created.
  3. Click File > New File… and add the following code…
  "compilerOptions": {
    "module": "none",
    "target": "ES5"
  "files": ["app.ts"]

Now you need to create a task runner. A task runner will transpile your TypeScript each time you run the task.

  1. In VS Code, add a new file called ‘app.ts’
  2. Press Ctrl+Shift+B to trigger the task runner. VS Code will bring up a little message at the top of the screen telling you that there is no task runner configured. Click the “Configure Task Runner” button. VS Code will create a tasks.json file automatically for you.
  3. Change the args property from HelloWorld.ts to app.ts. We will revisit this later to take a wildcard selector.

Open your app.ts file, and add the following code;

class Hello{
    constructor() {

Press Ctrl+Shift+B to kick off the build. Once complete (it takes < 1 second on a half decent machine) you will notice a new file has been added to the project, called app.js. Switch to split view by clicking the little side by side window icon on the top right hand side of the screen, and in the right hand pane open the newly generated app.js file. This is optional of course, if you don’t care about the compiled JavaScript, that’s absolutely fine. Sometimes it is nice to be able to see it.

Compile on save

If you have Node.JS tools installed, and you don’t want the hassle of pressing Ctrl+Shift+B every single time you want to compile, you can do the following to enable compile on save;

  1. Open a Node.js command prompt and change directory to the root folder for your project
  2. Type the following command;
tsc -w app.ts

TypeScript will listen for changes to your file, and transpile it automatically every time you make a change (and save). In addition, VS Code has an Auto-Save feature. To enable it, click File > Auto Save. If you have the TypeScript file and JavaScript files open side by side, the JavaScript file will periodically update without the need to manually save changes.

Adding AngularJS

Next, we need to add AngularJS to our project. You can do this using either Node Package Manager (NPM) or manually by download the appropriate files and adding them into your project folder. VS Code is still pretty young at this point (July 2015) and support for pulling in packages is non-existent.

Using NPM and Bower

If you don’t already have bower, a client side package manager, installed you can use the following command from a Node.js command prompt;

npm install -g bower

This will install bower globally on your machine. Now add a new file to your project called bower.json and add the following configuration:

  "name": "AngularTypeScript101",
  "version": "1.0.0",
  "dependencies": {
    "angular": "~1.4.0",
    "angular-route": "~1.4.0"

This will bring in version 1.4.0 of AngularJS. Feel free to change this to use whatever the current version of Angular 1.x is. Now run the following command, again from your Node.js command prompt;

bower install

This will download the packages and add them to your project folder under a sub directory called bower_components.

Using Visual Studio NuGet

If you happen to be following along using Visual Studio (full fat), you can easily install these packages using the following commands; (which you run via the Package Manager Console PowerShell window.

install-package angularjs.core
install-package angularjs.route

You can use the same mechanism to easily update the packages too (using the update-package command).


You can, of course, download Angular manually and add it to your project. You will find the latest version on the AngularJS website. You will probably find it easier to maintain your packages over time using Bower.

Adding the type declaration files

Thankfully TypeScript has a built in mechanism for adding type declaration files to your project. This is done using the TypeScript Definition Manager.

Using the TypeScript Definition Manager

You install the TypeScript definition manager using NPM. Open a Node.js command prompt and enter the following;

npm install -g tsd

This will install the TypeScript definition manager globally. The TypeScript declaration files are fully open source and can be found on the Definitely Typed GitHub repo. The AngularJS declaration files can be installed using the following command;

tsd install angular
tsd install angular-route

This will create a new directory called typings with a sub directory called angularjs. Within will be the two Angular declaration files you just installed. As AngularJS has a dependency on jqLite, the declaration files for jQuery will also be pulled in.

Using NuGet

You can install the declaration files by running the following command in the Package Manager Console PowerShell window.

Install-Package angularjs.TypeScript.DefinitelyTyped


You could of course just download the declaration files directly from the GitHub repository and copy them into the correct folder. (typings/angularjs)

Referencing the declaration files

Now that you have installed the declaration files for Angular, flip back to your app.ts file and add the following code;

class Hello{
    constructor() {
        angular.module("HelloApp", []);

No matter which technique you use to compile your JavaScript, you will get the following error message; (I believe that support for this is going to be vastly improved in future versions of VS Code)

Cannot find name 'angular'

That’s fine. All we need to do is tell TypeScript about our declaration files by adding the following code to the very top of the code file;

/// <reference path="typings/angularjs/angular.d.ts" />
/// <reference path="typings/angularjs/angular-resource.d.ts" />

Many modern editors, including Visual Studio and Webstorm, don’t require this additional step because they’re smart enough to check for the existence of the typings sub folder. Add the references, press Ctrl+Shift+B (or Save) to recompile and the error should go away.

Web Server

AngularJS, like most other JavaScript libraries, doesn’t work properly when viewed directly from the file system. If you open a file on your desktop, the path will be something like this;


Chrome, Firefox, and most other web browsers won’t allow you to consume other markup files due to security concerns (CORS). With that in mind, you will need some sort of web server to serve up the files so they can be previewed properly in the browser. VS Code does not have a built in HTTP server, so we must use another mechanism. If you’re using Visual Studio (full fat), then you can use the built in server and skip this step.


There is a very nice HTTP server on NPM that is more than sufficient for our needs. To install it, run the following command from a Node.js command prompt;

npm install -g http-server

This will install a simple HTTP server globally on your machine. To, run the server change directory (cd) to your root project folder and type the following command;

http-server -o --cors

There are lots of options you can pass in to customize the behaviour, but the default configuration will be enough for us. The default address is http://localhost:8080.


You can configure IIS to host the site for you. To do so, follow these steps;

  1. Open the Internet Information Services (IIS) manager (inetmgr.exe)

  2. Add a new site, call it “HaveIBeenPwnd”

  3. Point the physical path to your project folder.

  4. Enter the following host name: haveibeenpwnd.local

  5. Open your hosts file (C:\Windows\System32\Drivers\etc\hosts) in your favourite text editor (VS Code if you like!)

  6. Add the following line; haveibeenpwnd.local Open your web browser and point to haveibeenpwnd.local.

Sample Project

In order to demonstrate the concepts thoroughly, we are going to develop a simple project that utilizes the HaveIBeenPwned API. If you are not familiar with this service, HaveIBeenPwned is developed and maintained by developer security expert Troy Hunt. HaveIBeenPwned checks to see if your email address or username has been compromised in a data breach. The HaveIBeenPwned API is a way to access this wealth of information programmatically via a Restful endpoint. The project shows how to write the following AngularJS concepts in TypeScript;

  • Controllers
  • Directives
  • Filters
  • Services
  • Routing

The sample project is certainly not feature complete. In fact, all the samples are about as simplistic as they can possibly be… but that’s intentional.

Style Guide

Whenever I write any AngularJS code, I always follow (with only minor deviation) the Angular JS Style Guide, which was written by John Papa. It is an excellent guide and I highly recommend that you check it out. If you’ve never used a style guide before, just remember, a guide is exactly that… it is a guide. Use what bits make sense for your project.

Project Structure

The structure of the project is as follows;


We will add in the appropriate files as we go along.

Referencing declaration files

At the time of writing, VS Code does not recognise the fact that you have added type declaration files to your project. I suspect that in the future this will be resolved, but for now you have to manually reference the typings files directly in your TypeScript files. Type declaration files are referenced using a special triple forward slash (///) syntax;

/// <reference path="../../typings/angularjs/angular.d.ts" />
/// <reference path="../../typings/angularjs/angular-route.d.ts" />

VS Code will now load in these files and provide IntelliSense/auto-complete based on the declarations made in each file. If you are using Visual Studio, you can skip this step.

Note: It is out of the scope of this post to discuss how type declaration files work, we will cover this is a future post.

Note: Please ensure that every TypeScript file that you write for this project has these references at the top.


When not using any sort of AMD, the default pattern that TypeScript generates is the IIFE pattern. Take the following TypeScript code;

class App {
  'use strict'

TypeScript generates the following;

var App = (function() {
  function App() {}
  return App

This code is fine. It works. But there is one major problem. TypeScript has placed a variable on the global scope. A variable called App now exists on the window object. In our case, it is very unlikely that this would impact any other part of the application, but in larger projects with lots of scripts and external dependencies, this is a common problem. TypeScript introduces modules to help avoid this problem and also help with the organization of your code. Usage;

namespace HaveIBeenPwned {
  'use strict'
  class App {}

TypeScript generates the following;

var HaveIBeenPwned;
(function (HaveIBeenPwned) {
    "use strict";
    var App = (function () {
        function App() {
        return App;
})(HaveIBeenPwned '' (HaveIBeenPwned = {}));

A global object is still added to the window object, but now our App class is added as a property on the HaveIBeenPwned object. Everything wrapped inside a module with that name will be added as properties to it. Example;

namespace HaveIBeenPwned {
  'use strict'
  class Routes {}

TypeScript generates the following code;

var HaveIBeenPwned;
(function (HaveIBeenPwned) {
    "use strict";
    var Routes = (function () {
        function Routes() {
        return Routes;
})(HaveIBeenPwned '' (HaveIBeenPwned = {}));

As the HaveIBeenPwned object already exists, the Routes object will simply be added to it. Add a file called app.module.ts to the HaveIBeenPwned folder, and add the following;

namespace HaveIBeenPwned {
  'use strict'

  angular.module('HaveIBeenPwned', ['ngRoute'])

This will initialize the AngularJS app, and pull in the Routing module (ngRoute). Point to take away: Modules help organise your code and stop the global scope from becoming polluted.

Dependency injection for functions

Routing is the perfect place to start leveraging TypeScript in our application. Routing configuration tells AngularJS where to find our views, and which controller to use for a specific path navigated to. Routing is passed to AngularJS as configuration, using the config function. Traditional AngularJS code;

  function($routeProvider) {
    /*Routing goes here*/

The traditional JavaScript way is to pass a list of dependencies as a string array, then the final parameter is a function in which those dependencies are injected. This is primarily to support minification, so that dependencies can still be injected once all the variable names have been changed. TypeScript enables us to write this type of code in a much cleaner fashion. You might be aware that with AngularJS, you can directly inject dependencies into a function by attaching a property called $inject. Example;

function routes() {}
routes.$inject = ['$routeProvider']

I’ve found that this is rarely used in practice…perhaps due to lack of knowledge of the feature or simply because most of the documentation shows how to use the traditional style shown above. TypeScript lends itself well to using the $inject function. Add a new file called app.route.ts to the HaveIBeenPwned directory, and add the following code;

/// <reference path="../../typings/angularjs/angular.d.ts" />
/// <reference path="../../typings/angularjs/angular-route.d.ts" />

namespace HaveIBeenPwned {
  'use strict'

  function routes($routeProvider: ng.route.IRouteProvider) {}

  routes.$inject = ['$routeProvider']


Most of this code should look familar to you. Technically, there is absolutely no reason why you can’t use the traditional AngularJS style for passing dependencies to a function, I just find this syntax is much cleaner.

Type declarations on function parameters

With this code we encounter the first usage of a type declaration;

function routes($routeProvider: ng.route.IRouteProvider){

First, take a look at the compiled output;

function routes($routeProvider) {

It’s important to note that the type declarations are a TypeScript specific feature. Type declarations are not a feature of JavaScript. Type declarations are there to provide IntelliSense/auto-complete support and code refactoring, such as renaming of variables. All AngularJS declarations can be found in the ng namespace. In this case, we need the route provider to define our routing configuration. Routing is not part of the core AngularJS library, in fact it was split out into its own module named ngRoute. All declarations for routing can be found in the ng.route namespace. Add the following code to the routes function. Please type the code rather than copy/paste it.

  .when('/search', {
    templateUrl: '/client/HaveIBeenPwned/views/_search.html',
    controller: 'SearchController',
    controllerAs: 'vm'
    redirectTo: '/search'

You should immediately see the benefits now.

IntelliSense On Routing Provider

Not only do you get IntelliSense will full documentation, but you also get the added benefit of compile time checking;

Compile Time Checking

In this case I have provided the wrong number of arguments to the function… I get instant visual feedback on that. Note: My code still compiled. Even though the code I wrote in the above examples was wrong, the JavaScript was still generated and I could call that in the browser (all be it, with unpredictable behaviour). To complete the routing, add the following code to the routes function;

  .when('/search', {
    templateUrl: '/client/HaveIBeenPwned/views/_search.html',
    controller: 'SearchController',
    controllerAs: 'vm'
    redirectTo: '/search'

Don’t worry about creating the views or controllers at this point. We will do that later.

Dependency injection for classes

Injecting dependencies into classes is slightly different. With functions, we inject dependencies by tacking on an $inject variable to the function. With classes, we instead have a static variable with the same name. Example;

static $inject = ["PwnedService"];

As with most other programming languages, static is not instance specific, it applies to all instances. The generated JavaScript we end up is the same code as was generated for functions. Example;

SearchController.$inject = ['PwnedService']

Add a new file to the controllers directory, named search.ts. This will eventually be used to call an AngularJS service, which will call out to the HaveIBeenPwned API and return the result.

namespace HaveIBeenPwned {
  class SearchController {
    static $inject = ['PwnedService']
    constructor(private pwnedService: IPwnedService) {}

  angular.module('HaveIBeenPwned').controller('SearchController', SearchController)

Note: We don’t have a IPwnedService yet, we’ll get to that in a minute. Ignore any compile time errors your editor might be giving you at this point.


Constructors are a feature of ES6, called when an object is instantiated. Again, constructors work the same in JavaScript as they do in any other programming language. In the interest of cross-browser support, TypeScript generates a function with the same name as the containing function to indicate that it’s a constructor. Example of an empty constructor in TypeScript;

class SearchController {
  constructor() {
    //constructor logic

and the resulting JavaScript

var SearchController = (function() {
  function SearchController() {
    //constructor logic
  return SearchController

Again, constructors are now natively supported in ES6 so when targeting that version the code won’t be transpiled.

Understanding constructor parameters

Assume the following code;

constructor($http : ng.IHttpService) {

If you wanted to reuse the $http variable in other functions in your class, you might be tempted to do the following;

private _httpService: ng.IHttpService;

constructor($http : ng.IHttpService) {
    this._httpService = $http;

You should understand that the act of assigning the constructor parameters to a private variable (like you might do in other languages) is redundant in TypeScript. Under the hood TypeScript does this for you. The following code is valid TypeScript;

constructor($http : ng.IHttpService) {


someOtherFunction() {

This is valid because under the hood TypeScript created a variable on the class with the name $http inside the constructor and assigned the value automatically. The transpiled JavaScript

var PwnedService = (function () {
    function PwnedService($http) {
        this.$http = $http;

Unless you explicitly apply an access modifier to the parameter, it is public. You can add the private access modified to make that parameter only accessible with the class itself and not externally. I typically mark all my constructor parameters a private unless I need them to be accessible externally, although I’m not aware of any performance or other impact of not doing this.

Make objects visible to others using ‘Export’

The code shown in the previous example creates a controller called SearchController, using the ES6 class feature. By default, the class is not accessible to any other external objects.

SearchController Is Inaccessible

SearchController is not defined because it is not accessible. It is desirable to expose certain objects so that they can be consumed in other places. Generally I only expose objects that have to be exposed, but there is no hard and fast rule on this and no performance impact that I’m aware of, other than a slightly busier namespace object. To expose a class, interface, or function to other objects, use the export keyword.

export class SearchController {

This makes a subtle change to the generated code. This following line is added after the SearchController IIFE.

HaveIBeenPwned.SearchController = SearchController

Now re-running the code in developer tools results in the following;

SearchController Is Accessible

Note: This is a transformation provided by TypeScript and there is not a comparable feature in ES6 at the time of writing (July 2015).


Interfaces are contracts that state that an object will contain the functions and properties defined therein. Simple interface declaration;

namespace HaveIBeenPwned {
  export interface IPwnedService {}

Note that the above code results in zero JavaScript output. Why? Simple. JavaScript does not support interfaces or contracts of any kind. JavaScript is a dynamic language and properties/functions can be added to any object at any time. Interfaces are syntactic sugar provided by TypeScript to support IntelliSense/auto-complete and compile time checking. Add the following function to the interface (we will discuss promises shortly);

check(address:string) : ng.IPromise<{}>;

An interface function has no implementation, that will be defined on the implementation class.

Naming conventions

The official TypeScript style guide clearly states that interface names shouldn’t be prefixed with I. I come from a C# background and I’m simply too stuck in my old ways to adhere to this recommendation, and I know a lot of folk feel the same. My advice to you, choose whatever naming convention makes the most sense to you, and stick with that. Note also that I like to keep my interfaces and classes in the same physical file as the class that implements it (as long as it’s not too long). I recommend, again, that you pick an approach that works best for you, and then stick with that.

Working with interfaces

You can implement an interface on a class using the implements keyword as follows;

class PwnedService implements IPwnedService

Flesh out the class as follows;

namespace HaveIBeenPwned {
  export interface IPwnedService {
    check(address: string): ng.IPromise<{}>

  class PwnedService implements IPwnedService {
    static $inject = ['$http']
    constructor(private $http: ng.IHttpService) {}

    check(address: string): ng.IPromise<{}> {}

  angular.module('HaveIBeenPwned').service('PwnedService', PwnedService)

If you add a property or a function to the interface, or change it, and forget to update the implementation class, you will get a compile time error. You can also use interfaces with JSON. Take the following TypeScript example;

var a = <IMyInterface>{
  someValue: 'Hello'

Instead of having to instantiate a class that implements the IMyInterface interface, you can exploit JavaScript’s dynamic nature and pass in a raw JSON object. Specifying the interface before the object definition is a hint to TypeScript of what you are doing. In return, you will get full IntelliSense/auto-complete when defining the object, as well as compile time checking for all usages of that object. Finally, you can also derive interfaces from other interfaces. Interface inheritance if you like. Example;

export interface IEnterKeyPressAttributes extends ng.IAttributes {
  ngEnter: string

The above example shows a new interface, that uses the extends keyword to derive from the ng.IAttributes interface. The IEnterKeyPressAttributes interface has all the same methods and properties as the ng.IAttributes interface, and can add its own too.


Promises are a new shiny feature in ES6. Unfortunately, promises need to be supported either natively by the browser or via a polyfill. AngularJS has promises baked right in to many components. You can use promises indirectly, via a common service such as $http, or you can create promises directly using $q. Example;

class PwnedService {
  constructor($q: ng.IQService) {}

  check(): ng.IPromise<{}> {
    var defer = this.$q.defer()
    //Do something useful
    //Then call either defer.resolve() or defer.reject()
    return defer.promise

Or using $http:

class PwnedService {
  constructor($http: ng.IHttpService) {}

  check(): ng.IPromise<{}> {
    return this.$http.get('https://haveibeenpwned.com/api/v2/breachedaccount/' + address)

In the interest of avoiding the promise anti-pattern, I tend to use promises through other components, rather than directly and just chain on callbacks. Anyway, whichever technique you prefer, you will end up defining ng.IPromise<{}> as the return type for the parent function (see check method above). Technically, in TypeScript world, this is wrong. I’ve basically said that the promise will return “an object” as yet to be determined. However, I know what the correct type is;


Why didn’t I just write that in the first place? I simply don’t like the use of nested generics. Sure I lose a little bit of IntelliSense and compile time checking, but this has never been an issue for me.

Calling a function that returns a promise

Flip back to the SearchController.ts file and add the following function;

private breachedAccounts : BreachedAccount[];

submit(address: string) {
    this.pwnedService.check(address).then((result : ng.IHttpPromiseCallbackArg<{}>) =>{
        this.breachedAccounts = result.data;

There’s no magic here, the code is the same as what you would write in JavaScript. The only difference is that you get IntelliSense support on the result object.

The any type

Sometimes, a type isn’t known until runtime. This is especially true when calling a Web API back end. The response you get when an exception is thrown by the runtime (unhandled) is very different than an exception thrown by your code. If your code is really broken, you might get a 500 response with a yellow screen of death. The yellow screen of death is pure HTML markup only with on JSON representation. A runtime exception object might look like this;

    "Exception": {
        "Message": "Object reference not set to an instance of an object."

Also, your code might return the following ‘graceful’ exception when an error is encountered.

    "exception": {
        "message": "The request was malformed"

The difference between these two responses. Casing. By default, the .NET runtime returns responses in Pascal Case. You have to jump through several hoops to get the response to return in Camel Case, a step all so often not done fully. In situations where type isn’t known, use the any type;

submit(address: string) {
  this.pwnedService.check(address).then((result : ng.IHttpPromiseCallbackArg<{}>) =>{
    this.breachedAccounts = result.data;
  .catch((reason : any) => {
    alert(reason.Message '' reason.message);

This will put an end to any compile time nagging.


In the above example, we have explicitly told TypeScript that reason does not have a type. With that in mind, do you think it’s equally valid to leave off the type declaration altogether? Absolutely, because the static typing in TypeScript is optional. The following code is valid TypeScript;

submit(address: string) {
  this.pwnedService.check(address).then((result : ng.IHttpPromiseCallbackArg<BreachedAccount[]>) =>{
    this.breachedAccounts = result.data;
  .catch((reason) => {
    alert(reason.Message '' reason.message);

No type was given for reason. However, because the type is any is the type declaration files, it is still explicit. Take the following code;

var y

In this code there is no way that TypeScript can know what the type is, therefore it is implicitly any. If you want to prevent implicit use of the any type, you can pass in the --noImplicitAny flag to the TypeScript compiler when you execute it.

tsc --noImplicitAny -w

Model classes

Add the following model class to the models folder;

namespace HaveIBeenPwned {
  export class BreachedAccount {
    Title: string
    Name: string
    Domain: string
    BreachDate: string
    AddedDate: string
    PwnCount: number
    Description: string
    DataClasses: string[]
    IsVerified: boolean
    LogoType: string

This class will be used to map the response from the HaveIBeenPwned api back to an object for strong typing and compile time support. This is the exact data as returned by the API, and it returns the properties in Pascal Case, so we will have to use that here (or write ugly mapping code, but I’d rather avoid that).

Search View

Add a new file to the views folder called _search.html. Add the following markup;

<form ng-submit="vm.submit(vm.emailAddress)">
  <label>Enter your email address:
      <input type="email" id="emailAddress" name="emailAddress" ng-model="vm.emailAddress" placeholder="you@yourdomain.com" ng-enter="vm.submit(vm.emailAddress)">
    <button type="submit">Check Now</button>
  <tr ng-repeat="breachedAccount in vm.breachedAccounts">
      <td>{{breachedAccount.Title}}. <div ng-bind-html="breachedAccount.Description ' asHtml"></div></td>

There are two things of particular interest here, the use of a custom AngularJS directive called ngEnter, and a custom filter called asHtml. ngEnter directive usage;

<input type="email" id="emailAddress" name="emailAddress" ng-model="vm.emailAddress" placeholder="you@yourdomain.com" ng-enter="vm.submit(vm.emailAddress)">

asHtml filter usage;

<div ng-bind-html="breachedAccount.Description ' asHtml"></div>

We will need to create both of these before we can continue.


Filters are defined as follows;

A filter formats the value of an expression for display to the user.

When we call out to the HaveIBeenPwned API, it will return an array of BreachedAccount to us. Each BreachedAccount will have a property called Description. This description is HTML markup that will contain links and other interesting stuff. By default, AngularJS (rightly so) will encode the string to make it safe. The HTML markup will be rendered on the page, rather than added to the DOM and executed. We want to override this default behaviour, and render the HTML instead. Generally speaking, I wouldn’t recommend this approach because of the security implications (injection of potentially dangerous script) but in this small contrived example, it is fine. And lets be honest, it is unlikely that Troy Hunt, the security guy, is going to have some malicious script sent to us. Of course, you should never take this for granted. Anyway, add a file called asHtml.filter.ts to the filters directory. In terms of their behaviour, filters are much like the routing code we wrote earlier. Add the following code;

namespace HaveIBeenPwned {
  'use strict'

  export function asHtml($sce: ng.ISCEService) {
    return (text: string) => {
      return $sce.trustAsHtml(text)

  angular.module('HaveIBeenPwned').filter('asHtml', asHtml)

  asHtml.$inject = ['$sce']

Filters are defined by calling the filter function, passing in the name of the function as a string, and a reference to the function. In order to force AngularJS to skip over encoding the description, we need to inject ng.ISCEService, and call the trustAsHtml method. The method will be called once the description value is known, and for each instance of BreachedAccount in the array.


Now we need a new AngularJS directive called ngEnter. The purpose of this attribute, which will be attached to a text input field, will be to call a function on our controller when the user presses Enter on their keyboard. This means that the user won’t have to click the left button on their mouse to perform a search, they can do it straight from the keyboard. The beauty of directives is that they can easily be used throughout your application. The ngEnter attribute will take the name of the method to invoke when the user presses Enter. Directives are slightly more involved than filters. A typical directive in AngularJS consists of a link function, and a require variable (which determines how the directive is used). If you want to learn about directives in depth, The nitty-gritty of compile and link functions inside AngularJS directives by Jurgen Van de Moere is a good read. Start by defining a class, called EnterKeyPressDirective, and implement the ng.IDirective interface;

class EnterKeyPressDirective implements ng.IDirective {}

Doing this doesn’t give us much. In fact, if you take a look at the type declaration file all the functions and properties are optional;

interface IDirective {
  compile?: IDirectiveCompileFn
  controller?: any
  controllerAs?: string
  bindToController?: boolean | Object
  link?: IDirectiveLinkFn | IDirectivePrePost
  name?: string
  priority?: number
  replace?: boolean
  require?: any
  restrict?: string
  scope?: any
  template?: any
  templateUrl?: any
  terminal?: boolean
  transclude?: any

However, I’d say it is a pretty good practice to include the interface. The link function is called after the compile function at runtime. The link function takes 4 parameters and doesn’t return a value;

link($scope: ng.IScope, elm: Element, attr: ng.IAttributes, ngModel: ng.INgModelController): void

The value of ngEnter (the method to invoke) will be passed to us via the attr parameter. The problem is, ng.IAttributes knows nothing about ngEnter. Create a new interface that extends ng.IAttributes, and add the ngEnter property to it;

export interface IEnterKeyPressAttributes extends ng.IAttributes {
  ngEnter: string

Now replace ng.IAttibutes with IEnterKeyPressAttributes;

link($scope: ng.IScope, elm: Element, attr: IEnterKeyPressAttributes, ngModel: ng.INgModelController): void {


And flesh out the function as follows;

var element = angular.element(elm)
element.bind('keydown keypress', (event: JQueryEventObject) => {
  if (event.which === 13) {
    $scope.$apply(() => {


We are using jqLite to subscribe to the elements keydown and keypress events. Once a key is pressed, ensure it is the enter key, and then call the function as defined by the ngEnter variable. Also, we need to tell AngularJS that this directive requires an instance of ngModel and to restrict the directive so that it can only be used as an attribute. Add the following code to the class;

require = '?ngModel'
restrict = 'A'

We finish with a hack. I’m not sure if this is a bug in AngularJS, or expected behaviour. Each time our directive is required, we need to ensure that a new instance of that directive is created. Add the following static function;

static instance(): ng.IDirective {
    return new EnterKeyPressDirective();

For reasons unclear, we have to take care of the leg work of creating a new instance of the directive, because AngularJS doesn’t seem to take care of that for us. So when defining the directive, pass in the instance function instead of the class. This will then be called each time AngularJS needs an instance of the directive, which will in turn ensure a new instance is created;

angular.module('HaveIBeenPwned').directive('ngEnter', EnterKeyPressDirective.instance)

We’re good to go.

Final steps

Add a new file to the root folder called index.html. Also add a CSS file to the css folder, under content, called site.css. Open index.html and add the following markup;

  <link href='http://fonts.googleapis.com/css?family=Roboto' rel='stylesheet' type='text/css'>
  <link href="content/css/site.css" rel="stylesheet"/>
    <body ng-app="HaveIBeenPwned">
  <h1>Have I Been Pwned?</h1>

  <div class="container" ng-view>


  <script src="bower_components/angular/angular.js"></script>
  <script src="bower_components/angular-route/angular-route.js"></script>
  <script src="client/HaveIBeenPwned/app.module.js"></script>
  <script src="client/HaveIBeenPwned/app.route.js"></script>
  <script src="client/HaveIBeenPwned/services/pwnedservice.js"></script>
  <script src="client/HaveIBeenPwned/controllers/search.js"></script>
  <script src="client/HaveIBeenPwned/models/breachedaccount.js"></script>
  <script src="client/HaveIBeenPwned/filters/asHtml.filter.js"></script>
  <script src="client/HaveIBeenPwned/directives/search.directive.js"></script>

This will pull in all the relevant scripts, fonts and styles. It will also specify the name of the app (using the ngApp directive) and the view (using the ngView directive). Open site.css and add the following styles;

body {
  font-family: 'Roboto', sans-serif;
h1 {
  font-size: 24px;
  text-align: center;

.container {
  width: 50%;
  margin-right: auto;
  margin-left: auto;

input {
  width: 100%;
  height: 35px;
  margin: 10px auto;
  font-size: 16px;

button {
  padding: 8px 13px;
  color: #fff;
  font-size: 17px;
  font-family: 'Helvetica Neue', sans-serif;
  background: #25a6e1;
  background: linear-gradient(top, #25a6e1 0%, #188bc0 100%);
  border: 1px solid #1a87b9;
  border-radius: 4px;
  cursor: pointer;
  filter: progid: DXImageTransform.Microsoft.gradient( startColorstr='#25A6E1',endColorstr='#188BC0',GradientType=0);

tr:nth-child(even) {
  background-color: #eee;

tr td {
  padding: 10px;

It’s going to look functional, not beautiful.


Enter the test email address, foo@bar.com and press Enter. You should get some sample data regarding breaches for that account.


We looked, at a high level, at how to use TypeScript and AngularJS. We looked modules, dependency injection, promises, filters and directives… the most common components of AngularJS 1.x applications. TypeScript’s static typing functionality makes compile time checking and refactoring possible, whilst allowing us to exploit ES6 goodness, resulting in cleaner more maintainable code. We developed a sample application, that made us of the HaveIBeenPwned API to demonstrate these concepts.


Edit this page on GitHub
Jon Preece

About the author

I am Jon Preece, an experienced full stack website and software developer from Warrington (UK) currently working in and around Manchester (UK).

Throughout my 10+ year professional career I have worked in many sectors, including; e-commerce, financial services, marketing, healthcare, travel and accountancy. Get in touch via Twitter.