Native Client

View Change, Focus, and Input Events

This chapter describes view change, focus, and input event handling for a Native Client module. The chapter assumes you are familiar with the material presented in the Technical Overview.

There are two examples used in this chapter to illustrate basic programming techniques. The input_events example is used to illustrate how your module can react to keyboard and mouse input event. The mouse_lock example is used to illustrate how your module can react to view change events. You can find these examples in the /examples/api/input_events and /examples/api/mouse_lock directories in the Native Client SDK. There is also the ppapi_simple library that can be used to to implement most of the boiler plate. The pi_generator example in /examples/demo/pi_generator uses ppapi_simple to manage view change events and 2D graphics.

Overview

When a user interacts with the web page using a keyboard, mouse or some other input device, the browser generates input events. In a traditional web application, these input events are passed to and handled in JavaScript, typically through event listeners and event handlers. In a Native Client application, user interaction with an instance of a module (e.g., clicking inside the rectangle managed by a module) also generates input events, which are passed to the module. The browser also passes view change and focus events that affect a module’s instance to the module. Native Client modules can override certain functions in the pp::Instance class to handle input and browser events. These functions are listed in the table below:

Function Event Use
DidChangeView Called when the position, size, or clip rectangle of the module’s instance in the browser has changed. This event also occurs when browser window is resized or mouse wheel is scrolled. An implementation of this function might check the size of the module instance’s rectangle has changed and reallocate the graphics context when a different size is received.
DidChangeFocus Called when the module’s instance in the browser has gone in or out of focus (usually by clicking inside or outside the module instance). Having focus means that keyboard events will be sent to the module instance. An instance’s default condition is that it does not have focus. An implementation of this function might start or stop an animation or a blinking cursor.
HandleDocumentLoad Called after pp::Instance::Init() for a full-frame module instance that was instantiated based on the MIME type of a DOMWindow navigation. This situation only applies to modules that are pre-registered to handle certain MIME types. If you haven’t specifically registered to handle a MIME type or aren’t positive this applies to you, your implementation of this function can just return false. This API is only applicable when you are writing an extension to enhance the abilities of the Chrome web browser. For example, a PDF viewer might implement this function to download and display a PDF file.
HandleInputEvent Called when a user interacts with the module’s instance in the browser using an input device such as a mouse or keyboard. You must register your module to accept input events using RequestInputEvents() for mouse events and RequestFilteringInputEvents for keyboard events prior to overriding this function. An implementation of this function examines the input event type and branches accordingly.

These interfaces are found in the pp::Instance class. The sections below provide examples of how to handle these events.

Handling browser events

DidChangeView()

In the mouse_lock example, DidChangeView() checks the previous size of instance’s rectangle versus the new size. It also compares other state such as whether or not the app is running in full screen mode. If none of the state has actually changed, no action is needed. However, if the size of the view or other state has changed, it frees the old graphics context and allocates a new one.

void MouseLockInstance::DidChangeView(const pp::View& view) {
  // DidChangeView can get called for many reasons, so we only want to
  // rebuild the device context if we really need to.
  if ((size_ == view.GetRect().size()) &&
      (was_fullscreen_ == view.IsFullscreen()) && is_context_bound_) {
    return;
  }

  // ...

  // Reallocate the graphics context.
  size_ = view.GetRect().size();
  device_context_ = pp::Graphics2D(this, size_, false);
  waiting_for_flush_completion_ = false;

  is_context_bound_ = BindGraphics(device_context_);
  // ...

  // Remember if we are fullscreen or not
  was_fullscreen_ = view.IsFullscreen();
  // ...
}

For more information about graphics contexts and how to manipulate images, see:

DidChangeFocus()

DidChangeFocus() is called when you click inside or outside of a module’s instance in the web page. When the instance goes out of focus (click outside of the instance), you might do something like stop an animation. When the instance regains focus, you can restart the animation.

void DidChangeFocus(bool focus) {
  // Do something like stopping animation or a blinking cursor in
  // the instance.
}

Handling input events

Input events are events that occur when the user interacts with a module instance using the mouse, keyboard, or other input device (e.g., touch screen). This section describes how the input_events example handles input events.

Registering a module to accept input events

Before your module can handle these events, you must register your module to accept input events using RequestInputEvents() for mouse events and RequestFilteringInputEvents() for keyboard events. For the input_events example, this is done in the constructor of the InputEventInstance class:

class InputEventInstance : public pp::Instance {
 public:
  explicit InputEventInstance(PP_Instance instance)
      : pp::Instance(instance), event_thread_(NULL), callback_factory_(this) {
    RequestInputEvents(PP_INPUTEVENT_CLASS_MOUSE | PP_INPUTEVENT_CLASS_WHEEL |
                       PP_INPUTEVENT_CLASS_TOUCH);
    RequestFilteringInputEvents(PP_INPUTEVENT_CLASS_KEYBOARD);
  }
  // ...
};

RequestInputEvents() and RequestFilteringInputEvents() accept a combination of flags that identify the class of events that the instance is requesting to receive. Input event classes are defined in the PP_InputEvent_Class enumeration in ppb_input_event.h.

Determining and branching on event types

In a typical implementation, the HandleInputEvent() function determines the type of each event using the GetType() function found in the InputEvent class. The HandleInputEvent() function then uses a switch statement to branch on the type of input event. Input events are defined in the PP_InputEvent_Type enumeration in ppb_input_event.h.

virtual bool HandleInputEvent(const pp::InputEvent& event) {
  Event* event_ptr = NULL;
  switch (event.GetType()) {
    case PP_INPUTEVENT_TYPE_UNDEFINED:
      break;
    case PP_INPUTEVENT_TYPE_MOUSEDOWN:
    case PP_INPUTEVENT_TYPE_MOUSEUP:
    case PP_INPUTEVENT_TYPE_MOUSEMOVE:
    case PP_INPUTEVENT_TYPE_MOUSEENTER:
    case PP_INPUTEVENT_TYPE_MOUSELEAVE:
    case PP_INPUTEVENT_TYPE_CONTEXTMENU: {
      pp::MouseInputEvent mouse_event(event);
      PP_InputEvent_MouseButton pp_button = mouse_event.GetButton();
      MouseEvent::MouseButton mouse_button = MouseEvent::kNone;
      switch (pp_button) {
        case PP_INPUTEVENT_MOUSEBUTTON_NONE:
          mouse_button = MouseEvent::kNone;
          break;
        case PP_INPUTEVENT_MOUSEBUTTON_LEFT:
          mouse_button = MouseEvent::kLeft;
          break;
        case PP_INPUTEVENT_MOUSEBUTTON_MIDDLE:
          mouse_button = MouseEvent::kMiddle;
          break;
        case PP_INPUTEVENT_MOUSEBUTTON_RIGHT:
          mouse_button = MouseEvent::kRight;
          break;
      }
      event_ptr =
          new MouseEvent(ConvertEventModifier(mouse_event.GetModifiers()),
                         mouse_button,
                         mouse_event.GetPosition().x(),
                         mouse_event.GetPosition().y(),
                         mouse_event.GetClickCount(),
                         mouse_event.GetTimeStamp(),
                         event.GetType() == PP_INPUTEVENT_TYPE_CONTEXTMENU);
    } break;
    case PP_INPUTEVENT_TYPE_WHEEL: {
      pp::WheelInputEvent wheel_event(event);
      event_ptr =
          new WheelEvent(ConvertEventModifier(wheel_event.GetModifiers()),
                         wheel_event.GetDelta().x(),
                         wheel_event.GetDelta().y(),
                         wheel_event.GetTicks().x(),
                         wheel_event.GetTicks().y(),
                         wheel_event.GetScrollByPage(),
                         wheel_event.GetTimeStamp());
    } break;
    case PP_INPUTEVENT_TYPE_RAWKEYDOWN:
    case PP_INPUTEVENT_TYPE_KEYDOWN:
    case PP_INPUTEVENT_TYPE_KEYUP:
    case PP_INPUTEVENT_TYPE_CHAR: {
      pp::KeyboardInputEvent key_event(event);
      event_ptr = new KeyEvent(ConvertEventModifier(key_event.GetModifiers()),
                               key_event.GetKeyCode(),
                               key_event.GetTimeStamp(),
                               key_event.GetCharacterText().DebugString());
    } break;
    default: {
      // For any unhandled events, send a message to the browser
      // so that the user is aware of these and can investigate.
      std::stringstream oss;
      oss << "Default (unhandled) event, type=" << event.GetType();
      PostMessage(oss.str());
    } break;
  }
  event_queue_.Push(event_ptr);
  return true;
}

Notice that the generic InputEvent received by HandleInputEvent() is converted into a specific type after the event type is determined. The event types handled in the example code are MouseInputEvent, WheelInputEvent, and KeyboardInputEvent. There are also TouchInputEvents. For the latest list of event types, see the InputEvent documentation. For reference information related to the these event classes, see the following documentation:

Threading and blocking

HandleInputEvent() in this example runs on the main module thread. However, the bulk of the work happens on a separate worker thread (see ProcessEventOnWorkerThread). HandleInputEvent() puts events in the event_queue_ and the worker thread takes events from the event_queue_. This processing happens independently of the main thread, so as not to slow down the browser.

Authentication required

You need to be signed in with Google+ to do that.

Signing you in...

Google Developers needs your permission to do that.