ContentCaptureManager

class ContentCaptureManager

Provides additional ways for apps to integrate with the content capture subsystem.

Content capture provides real-time, continuous capture of application activity, display and events to an intelligence service that is provided by the Android system. The intelligence service then uses that info to mediate and speed user journey through different apps. For example, when the user receives a restaurant address in a chat app and switches to a map app to search for that restaurant, the intelligence service could offer an autofill dialog to let the user automatically select its address.

Content capture was designed with two major concerns in mind: privacy and performance.

• Privacy: the intelligence service is a trusted component provided that is provided by the device manufacturer and that cannot be changed by the user (although the user can globaly disable content capture using the Android Settings app). This service can only use the data for in-device machine learning, which is enforced both by process isolation and CDD requirements.
• Performance: content capture is highly optimized to minimize its impact in the app jankiness and overall device system health. For example, its only enabled on apps (or even specific activities from an app) that were explicitly allowlisted by the intelligence service, and it buffers the events so they are sent in a batch to the service (see isContentCaptureEnabled() for other cases when its disabled).

In fact, before using this manager, the app developer should check if it's available. Example:

<code>
   ContentCaptureManager mgr = context.getSystemService(ContentCaptureManager.class);
   if (mgr != null &amp;&amp; mgr.isContentCaptureEnabled()) {
     // ...
   }
   </code>

App developers usually don't need to explicitly interact with content capture, except when the app:

• Can define a contextual android.content.LocusId to identify unique state (such as a conversation between 2 chat users).
• Can have multiple view hierarchies with different contextual meaning (for example, a browser app with multiple tabs, each representing a different URL).
• Contains custom views (that extend View directly and are not provided by the standard Android SDK.
• Contains views that provide their own virtual hierarchy (like a web browser that render the HTML elements using a Canvas).

The main integration point with content capture is the ContentCaptureSession. A "main" session is automatically created by the Android System when content capture is enabled for the activity and its used by the standard Android views to notify the content capture service of events such as views being added, views been removed, and text changed by user input. The session could have a ContentCaptureContext to provide more contextual info about it, such as the locus associated with the view hierarchy (see android.content.LocusId for more info about locus). By default, the main session doesn't have a ContentCaptureContext, but you can change it after its created. Example:

<code>
  protected void onCreate(Bundle savedInstanceState) {
    // Initialize view structure
    ContentCaptureSession session = rootView.getContentCaptureSession();
    if (session != null) {
      session.setContentCaptureContext(ContentCaptureContext.forLocusId("chat_UserA_UserB"));
    }
  }
  </code>

If your activity contains view hierarchies with a different contextual meaning, you should created child sessions for each view hierarchy root. For example, if your activity is a browser, you could use the main session for the main URL being rendered, then child sessions for each IFRAME:

<code>
  ContentCaptureSession mMainSession;
 
  protected void onCreate(Bundle savedInstanceState) {
     // Initialize view structure...
     mMainSession = rootView.getContentCaptureSession();
     if (mMainSession != null) {
       mMainSession.setContentCaptureContext(
           ContentCaptureContext.forLocusId("https://example.com"));
     }
  }
 
  private void loadIFrame(View iframeRootView, String url) {
    if (mMainSession != null) {
       ContentCaptureSession iFrameSession = mMainSession.newChild(
           ContentCaptureContext.forLocusId(url));
       }
       iframeRootView.setContentCaptureSession(iFrameSession);
    }
    // Load iframe...
  }
  </code>

If your activity has custom views (i.e., views that extend View directly and provide just one logical view, not a virtual tree hiearchy) and it provides content that's relevant for content capture (as of Android Q, the only relevant content is text), then your view implementation should:

• Set it as important for content capture.
• Fill ViewStructure used for content capture.
• Notify the ContentCaptureSession when the text is changed by user input.

Here's an example of the relevant methods for an EditText-like view:

<code>
  public class MyEditText extends View {
 
  public MyEditText(...) {
    if (getImportantForContentCapture() == IMPORTANT_FOR_CONTENT_CAPTURE_AUTO) {
      setImportantForContentCapture(IMPORTANT_FOR_CONTENT_CAPTURE_YES);
    }
  }
 
  public void onProvideContentCaptureStructure(@NonNull ViewStructure structure, int flags) {
    super.onProvideContentCaptureStructure(structure, flags);
 
    structure.setText(getText(), getSelectionStart(), getSelectionEnd());
    structure.setHint(getHint());
    structure.setInputType(getInputType());
    // set other properties like setTextIdEntry(), setTextLines(), setTextStyle(),
    // setMinTextEms(), setMaxTextEms(), setMaxTextLength()
  }
 
  private void onTextChanged() {
    if (isLaidOut() &amp;&amp; isImportantForContentCapture() &amp;&amp; isTextEditable()) {
      ContentCaptureManager mgr = mContext.getSystemService(ContentCaptureManager.class);
      if (cm != null &amp;&amp; cm.isContentCaptureEnabled()) {
         ContentCaptureSession session = getContentCaptureSession();
         if (session != null) {
           session.notifyViewTextChanged(getAutofillId(), getText());
         }
    }
  }
  </code>

If your view provides its own virtual hierarchy (for example, if it's a browser that draws the HTML using Canvas or native libraries in a different render process), then the view is also responsible to notify the session when the virtual elements appear and disappear - see View.onProvideContentCaptureStructure(ViewStructure, int) for more info.

Summary

Constants

static val DATA_SHARE_ERROR_CONCURRENT_REQUEST: Int

Value: 2

static val DATA_SHARE_ERROR_TIMEOUT_INTERRUPTED: Int

Value: 3

static val DATA_SHARE_ERROR_UNKNOWN: Int

Value: 1

Public methods

fun getContentCaptureConditions(): MutableSet<ContentCaptureCondition!>?

fun getServiceComponentName(): ComponentName?

fun isContentCaptureEnabled(): Boolean

fun removeData(request: DataRemovalRequest): Unit

fun setContentCaptureEnabled(enabled: Boolean): Unit

fun shareData(
    request: DataShareRequest, 
    executor: Executor, 
    dataShareWriteAdapter: DataShareWriteAdapter
): Unit