Tips for creating Office Add-ins with Angular

This article provides guidance for using Angular 2+ to create an Office Add-in as a single page application.

Note: Do you have something to contribute based on your experience using Angular to create Office Add-ins? You can contribute to this article in GitHub or provide your feedback by submitting an issue in the repo.

For an Office Add-ins sample that's built using the Angular framework, see Word Style Checking Add-in Built on Angular.

Install the TypeScript type definitions

Open an nodejs window and enter the following at the command line: npm install --save-dev @types/office-js.

Bootstrapping must be inside Office.initialize

On any page that calls the Office, Word, or Excel JavaScript APIs, your code must first assign a method to the Office.initialize property. (If you have no initialization code, the method body can be just empty "{}" symbols, but you must not leave the Office.initialize property undefined. For details, see Initializing your add-in.) Office calls this method immediately after it has initialized the Office JavaScript libraries.

Your Angular bootstrapping code must be called inside the method that you assign to Office.initialize to ensure that the Office JavaScript libraries have initialized first. The following is a simple example that shows how to do this. This code should be in the main.ts file of the project.

import { platformBrowserDynamic } from '@angular/platform-browser-dynamic';
import { AppModule } from './app.module';

Office.initialize = function () {
  const platform = platformBrowserDynamic();

Use the hash location strategy in the Angular application

Navigating between routes in the application might not work if you don't specify the hash location strategy. You can do this in one of two ways. First, you can specify a provider for the location strategy in your app module, as shown in the following example. It goes into the app.module.ts file.

import { LocationStrategy, HashLocationStrategy } from '@angular/common';
// Other imports suppressed for brevity

  providers: [
    { provide: LocationStrategy, useClass: HashLocationStrategy },
    // Other providers suppressed
  // Other module properties suppressed
export class AppModule { }

If you define your routes in a separate routing module, there is an alternative way to specify the hash location strategy. In your routing module's .ts file, pass a configuration object to the forRoot function that specifies the strategy. The following code is an example.

import { RouterModule, Routes } from '@angular/router';
// Other imports suppressed for brevity

const routes: Routes = // route definitions go here

  imports: [RouterModule.forRoot(routes, { useHash: true })],
  exports: [RouterModule]
export class AppRoutingModule { }

Consider wrapping Fabric components with Angular components

We recommend using Office UI Fabric styling in your add-in. Fabric includes components that come in several versions, including a version based on TypeScript. Consider using Fabric components in your add-in by wrapping them in Angular components. For an example that shows you how to do this, see Word Style Checking Add-in Built on Angular. Note, for example, how the Angular component defined in fabric.textfield.wrapper imports the Fabric file TextField.ts, where the Fabric component is defined.

Using the Office Dialog API with Angular

The Office add-in Dialog API enables your add-in to open a page in a semimodal dialog box that can exchange information with the main page, which is typically in a task pane.

The displayDialogAsync method takes a parameter that specifies the URL of the page that should open in the dialog box. Your add-in can have a separate HTML page (different from the base page) to pass to this parameter, or you can pass the URL of a route in your Angular appication.

It is important to remember, if you pass a route, that the dialog box creates a new window with its own execution context. Your base page and all its initialization and bootstrapping code run again in this new context, and any variables are set to their initial values in the dialog box. So this technique launches a second instance of your single page application in the dialog box. Code that changes variables in the dialog box does not change the task pane version of the same variables. Similarly, the dialog box has its own session storage, which is not accessible from code in the task pane.

Forcing an update of the DOM

In any Angular application, notifications to update the DOM occasionally do not fire. The framework provides a tick() method on the ApplicationRef object that will force an update. The following code is an example.

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

export class MyComponent {
  constructor(private appRef: ApplicationRef) { }
  myMethod() {
    // Code that changes the DOM is here

Using observables

Angular uses RxJS (Reactive Extensions for JavaScript), and RxJS introduces Observable and Observer objects to implement asynchronous processing. This section provides a brief introduction to using Observables; for more detailed information, see the official RxJS documentation.

An Observable is like a Promise object in some ways - it is returned immediately from an asynchronous call, but it might not resolve until some time later. However, while a Promise is a single value (which can be an array object), an Observable is an array of objects (possibly with only a single member). This enables code to call array methods, such as concat, map, and filter, on Observable objects.

Pushing instead of pulling

Your code "pulls" Promise objects by assigning them to variables, but Observable objects "push" their values to objects that subscribe to the Observable. The subscribers are Observer objects. The benefit of the push architecture is that new members can be added to the Observable array over time. When a new member is added, all the Observer objects that subscribe to the Observable receive a notification.

The Observer is configured to process each new object (called the "next" object) with a function. (It is also configured to respond to an error and a completion notification. See the next section for an example.) For this reason, Observable objects can be used in a wider range of scenarios than Promise objects. For example, in addition to returning an Observable from an AJAX call, the way you can return a Promise, an Observable can be returned from an event handler, such as the "changed" event handler for a text box. Each time a user enters text in the box, all the subscribed Observer objects react immediately using the latest text and/or the current state of the application as input.

Waiting until all asynchronous calls have completed

When you want to ensure that a callback only runs when every member of a set of Promise objects has resolved, use the Promise.all() method.

myPromise.all([x, y, z]).then(
  // TODO: Callback logic goes here

To do the same thing with an Observable object, you use the Observable.forkJoin() method.

const source = Observable.forkJoin([x, y, z]);

const subscription = source.subscribe(
  x => {
    // TODO: Callback logic goes here
  err => console.log('Error: ' + err),
  () => console.log('Completed')
function () { console.log('Completed'); });