Home > Articles > Mobile Application Development & Programming

  • Print
  • + Share This
This chapter is from the book

Creating URL-Based Services

Apple’s built-in applications offer a variety of services that can be accessed via URL calls. You can ask Safari to open Web pages, Maps to show a map, or use the mailto: style URL to start composing a letter in Mail. A URL scheme refers to the first part of the URL that appears before the colon, such as http: or ftp:.

These services work because iOS knows how to match URL schemes to applications. A URL that starts with http: opens in Mobile Safari. The mailto: URL always links to Mail. What you may not know is that you can define your own URL schemes and implement them in your applications. Not all standard schemes are supported on the iPhone. The FTP scheme is not available for use.

Custom schemes allow applications to launch whenever Mobile Safari (or another application) opens a URL of that type. For example, should your application register xyz, any xyz: links go directly to your application for handling, where they’re passed to the application delegate’s URL opening method (after first passing through the launch-with-options method; you’ll read more about these details later). You do not have to add any special coding there. If all you want to do is run an application, adding the scheme and opening the URL enables cross-application launching.

Handlers extend launching to allow applications to do something with the URL that’s been passed to it. They might open a specific data file, retrieve a particular name, display a certain image, or otherwise process information included in the call.

Using URL Schemes

The advantages of scheme-based launching are many. Take the Iconfactory’s Twitterrific, for example. Early in its development, developer Craig Hockenberry introduced a custom service that let users and third-party developers launch his application and open a prefilled, ready-to-post tweet.

In days before Twitter was integrated at the system level, this let developers add Twitter support to their applications without any programming. Because Twitterrific already stored the sensitive username and password information, all that you had to supply was the body text. Control passed to Twitterrific, which took over and allowed users to finish tweeting. When done, users quit Twitterrific and might have returned to the original application if desired.

URL-based services work best when they provide some kind of performance boost or data leverage. In the case of Twitterrific, it wasn’t about tweeting. It takes very little code to tweet. (Refer to Chapter 15, “Networking.”) It was about freeing the developer from responsibility of securely storing user credentials or forcing users to enter those credentials on each use.

Service Downsides

It’s not all good news on the services front; there’s definitely a downside to third-party services. Once your application depends on a service, you basically force your users to download a second application. And that application may not always be the one they want to use. Consider the loyal Echofon user who might not have had Twitterrific installed on his or her phone. If your app had demanded Twitterrific, you may have met resistance.

Any features that depend on third-party services must always be optional. Consider if, for example, Echofon had introduced its own tweeting URL scheme. When you add service-based features to your application, make those features robust and flexible enough to allow users to choose their preferred client.

Another downside is this: iOS applications cannot tell what schemes are available a priori. There’s no way to poll for on-offer services. Apple provides no public registry that you can scan through to see what’s out there. Using services is basically a matter of trust.

That being said, you can test whether a URL service is available. If the UIApplication’s canOpenURL: method returns YES, you are guaranteed that openURL: will be able to launch another application to open that URL. You are not guaranteed that the URL is valid, only that its scheme is registered properly to an existing application.

if ([[UIApplication sharedApplication] canOpenURL:aURL])

    [[UIApplication sharedApplication] openURL:aURL];


There is another important business-oriented aspect to scheme-based launching, namely cross-promotion. Defining URL schema allows your application to test whether other applications exist from your company’s lineup. If the application cannot handle the URL (i.e., canOpenURL: returns NO), you can provide links to App Store, encouraging users to download other applications from your company.

Registering Schemes: Declaring the URL

It takes two steps to add services to your application. First, declare your URL scheme in your Info.plist property list. Second, add a handler to your application delegate. Here are the steps you take to do this.

To declare your URL scheme, you need to specify information for the iOS’s Launch Services. Add a CFBundleURLTypes entry into your Info.plist. This consists of an array of dictionaries that describe the URL types the application can open and handle. Each dictionary is quite simple. They contain two keys: a CFBundleURLName and an array of CFBundleURLSchemes.

The URL name is an abstract name (also known as its “kind”). You can use any string. On the Mac, this provides the visible description shown in Finder. On iOS, it’s just a way of keeping your schemes straight. Figure 16-8 names its URL type com.sadun.urlSchemeDemonstration.

Figure 16-8

Figure 16-8. A custom URL scheme declaration. This scheme will open any URL prefixed by xyz:.

The Schemes array provides a list of prefixes that belong to the abstract name. You can add one scheme or many. The following declares just one. You may want to prefix your name with an x (e.g., x-sadun-services). Although the iOS family is not part of any standards organization, the x prefix indicates that this is an unregistered name. Recipe 16-6 uses the xyz scheme shown in Figure 16-8.












Former iOS developer (and current Apple employee) Emanuele Vulcano has started an informal registry over at the CocoaDev Web site (http://cocoadev.com/index.pl?ChooseYourOwniPhoneURLScheme). iOS developers can share their schemes in a central listing, so that you can discover services you want to use and promote services that you offer. The registry lists services, their URL schemes, and describes how these services can be used by other developers.

Registering Schemes: Adding the Handler Method

The second part of implementing URL handling is to provide any of a series of application delegate calls. This method chain has grown over time and is quite crusty in its complexity. The single method you’ll probably want to define to handle URLs is demonstrated in Recipe 16-6. This provides the preferred entry point for iOS 5. To test this recipe, launch the app, suspend it, and open an xyz: URL in Mobile Safari.

Recipe 16-6. Providing URL Scheme Support

- (BOOL)application:(UIApplication *)application

    openURL:(NSURL *)url

    sourceApplication:(NSString *)sourceApplication



    tbvc.title = url.path;

    NSLog(@"URL: %@", url);

    NSLog(@"Opened from : %@", sourceApplication);

    NSLog(@"Annotation : %@", annotation);

    return YES;


By preferred entry point, it does not mean only entry point. You can also grab items from the launch dictionary in application:didFinishLaunchingWithOptions:. A final, now deprecated approach is to implement application:handleOpenURL:, which remains as a valid entry point in iOS 5, but I strongly encourage you to avoid it.

Returning Control to a Calling Application

With careful programming, you can allow the calling application to regain control after handling a URL request. An application might send back requested material along with a status indication of whether the operation was a success. It depends on how you define and implement your protocol. By adding a reverse scheme element in the calling URL, you know how to call home once the requested service completes.


Be aware that app-to-app URL launching has a fairly long latency. You can expect a return trip to last up to 5 seconds or more. What’s more, should the calling client lie or provide the wrong scheme parameter, control may be transferred to a third application. If the scheme does not refer to a real application, the request will simply hang in the receiving app. On the other hand, allowing the service to return to the originating application keeps the user from having to quit and relaunch that first application and is a feature more applications should offer once they finish handling a request.

  • + Share This
  • 🔖 Save To Your Account