This article mainly reads the Android source code to explain the key event distribution principle and focus search principle of TV. The source code is based on Android9.0. First of all, consider several questions:
- How to distribute and process the key events step by step when the remote control presses a key
- Why do some devices press and hold the remote control for the first time to onKeyDown and onKeyUp, and then it is normal to onKeyDown until it is released
- When setOnKeyListener is set for View, will the onKeyDown callback or OnKeyListener callback of View go first
- When will the onBackPressed method of Activity be called
- When pressing the direction key, how does the system know which control should get the focus if it does not control the next one to get the focus
With these questions, let's talk about the Android source code! Knowing how the system handles it is convenient for us to solve the problems of some keys and focus on TV.
1, Key event entry
First, let's take a look at the internal class of ViewPostImeInputStage in the ViewRootImpl class, the entry of key events:
/**
* Delivers post-ime input events to the view hierarchy.
*/
final class ViewPostImeInputStage extends InputStage {
public ViewPostImeInputStage(InputStage next) {
super(next);
}
@Override
protected int onProcess(QueuedInputEvent q) {
// 1. If it is judged as a key event, execute the processKeyEvent method
if (q.mEvent instanceof KeyEvent) {
return processKeyEvent(q);
} else {
final int source = q.mEvent.getSource();
// 2. If it is judged as a touch event, execute the processPointerEvent method
if ((source & InputDevice.SOURCE_CLASS_POINTER) != 0) {
return processPointerEvent(q);
// 3. If it is judged as a trackball event, execute the processTrackballEvent method
} else if ((source & InputDevice.SOURCE_CLASS_TRACKBALL) != 0) {
return processTrackballEvent(q);
// 4. If it is judged as a motion event, execute the processGenericMotionEvent method
} else {
return processGenericMotionEvent(q);
}
}
}
You can see Notes 1, 2, 3 and 4 respectively judge different events and execute different methods. The TV focus event discussed in this article mainly looks at the processKeyEvent method:
private int processKeyEvent(QueuedInputEvent q) {
final KeyEvent event = (KeyEvent)q.mEvent;
if (mUnhandledKeyManager.preViewDispatch(event)) {
return FINISH_HANDLED;
}
// 1. Press the distribute button. If there is consumption, return true and do not proceed
// Deliver the key to the view hierarchy.
if (mView.dispatchKeyEvent(event)) {
return FINISH_HANDLED;
}
if (shouldDropInputEvent(q)) {
return FINISH_NOT_HANDLED;
}
// This dispatch is for windows that don't have a Window.Callback. Otherwise,
// the Window.Callback usually will have already called this (see
// DecorView.superDispatchKeyEvent) leaving this call a no-op.
if (mUnhandledKeyManager.dispatch(mView, event)) {
return FINISH_HANDLED;
}
int groupNavigationDirection = 0;
if (event.getAction() == KeyEvent.ACTION_DOWN
&& event.getKeyCode() == KeyEvent.KEYCODE_TAB) {
if (KeyEvent.metaStateHasModifiers(event.getMetaState(), KeyEvent.META_META_ON)) {
groupNavigationDirection = View.FOCUS_FORWARD;
} else if (KeyEvent.metaStateHasModifiers(event.getMetaState(),
KeyEvent.META_META_ON | KeyEvent.META_SHIFT_ON)) {
groupNavigationDirection = View.FOCUS_BACKWARD;
}
}
// If a modifier is held, try to interpret the key as a shortcut.
if (event.getAction() == KeyEvent.ACTION_DOWN
&& !KeyEvent.metaStateHasNoModifiers(event.getMetaState())
&& event.getRepeatCount() == 0
&& !KeyEvent.isModifierKey(event.getKeyCode())
&& groupNavigationDirection == 0) {
if (mView.dispatchKeyShortcutEvent(event)) {
return FINISH_HANDLED;
}
if (shouldDropInputEvent(q)) {
return FINISH_NOT_HANDLED;
}
}
// Apply the fallback event policy.
if (mFallbackEventHandler.dispatchKeyEvent(event)) {
return FINISH_HANDLED;
}
if (shouldDropInputEvent(q)) {
return FINISH_NOT_HANDLED;
}
// Handle automatic focus changes.
if (event.getAction() == KeyEvent.ACTION_DOWN) {
if (groupNavigationDirection != 0) {
if (performKeyboardGroupNavigation(groupNavigationDirection)) {
return FINISH_HANDLED;
}
} else {
// 2. If the key is pressed, the focus navigation logic will be executed
if (performFocusNavigation(event)) {
return FINISH_HANDLED;
}
}
}
return FORWARD;
}
2, Key event distribution process
You can see that the mView.dispatchKeyEvent method is executed in this method. The View here is actually a DecorView. Next, look at this method:
@Override
public boolean dispatchKeyEvent(KeyEvent event) {
final int keyCode = event.getKeyCode();
final int action = event.getAction();
final boolean isDown = action == KeyEvent.ACTION_DOWN;
// 1. If it is the first time to press, the shortcut key of panel will be processed
if (isDown && (event.getRepeatCount() == 0)) {
// First handle chording of panel key: if a panel key is held
// but not released, try to execute a shortcut in it.
if ((mWindow.mPanelChordingKey > 0) && (mWindow.mPanelChordingKey != keyCode)) {
boolean handled = dispatchKeyShortcutEvent(event);
if (handled) {
return true;
}
}
// If a panel is open, perform a shortcut on it without the
// chorded panel key
if ((mWindow.mPreparedPanel != null) && mWindow.mPreparedPanel.isOpen) {
if (mWindow.performPanelShortcut(mWindow.mPreparedPanel, keyCode, event, 0)) {
return true;
}
}
}
// 2. If the Window is not destroy ed and its Callback is not empty, give it to its Callback for processing
if (!mWindow.isDestroyed()) {
final Window.Callback cb = mWindow.getCallback();
final boolean handled = cb != null && mFeatureId < 0 ? cb.dispatchKeyEvent(event)
: super.dispatchKeyEvent(event);
if (handled) {
return true;
}
}
// 3. If the above has not been processed, it will be distributed to PhoneWindow to onKeyDown and onKeyUp event processing
return isDown ? mWindow.onKeyDown(mFeatureId, event.getKeyCode(), event)
: mWindow.onKeyUp(mFeatureId, event.getKeyCode(), event);
}
The above first determines that if it is the first time to press, the shortcut key of the panel will be processed. If it is processed, it will not go down. Otherwise, continue to judge that when the window is not destroyed and the callback is not empty, it will be recalled. If it is processed, it will not go down. Otherwise, let the onKeyDown and onKeyUp methods corresponding to PhoneWindow handle it.
Next, let's take a look at the implementation of relevant methods in turn according to this dispatching order. Here, let's first look at the implementation of dispatchKeyEvent of Activity:
/**
* Called to process key events. You can override this to intercept all
* key events before they are dispatched to the window. Be sure to call
* this implementation for key events that should be handled normally.
*
* @param event The key event.
*
* @return boolean Return true if this event was consumed.
*/
public boolean dispatchKeyEvent(KeyEvent event) {
onUserInteraction();
// Let action bars open menus in response to the menu key prioritized over
// the window handling it
final int keyCode = event.getKeyCode();
if (keyCode == KeyEvent.KEYCODE_MENU &&
mActionBar != null && mActionBar.onMenuKeyEvent(event)) {
return true;
}
Window win = getWindow();
// 1. From here, the event processing is handed over to the related window object, which is essentially distributed to the view hierarchy
if (win.superDispatchKeyEvent(event)) {
return true;
}
View decor = mDecor;
if (decor == null) decor = win.getDecorView();
// 2. If the view hierarchy is not processed, it is handed over to the dispatch method of KeyEvent itself, and various callbacks * * * of Activity are triggered
return event.dispatch(this, decor != null
? decor.getKeyDispatcherState() : null, this);
}
Let's look at the superDispatchKeyEvent method in point 1. We can see that this method is an abstract method, and its implementation is to implement its subclass PhoneWindow:
@Override
public boolean superDispatchKeyEvent(KeyEvent event) {
return mDecor.superDispatchKeyEvent(event);
}
This method calls the superDispatchKeyEvent method in the DecorView again:
public boolean superDispatchKeyEvent(KeyEvent event) {
// Give priority to closing action modes if applicable.
if (event.getKeyCode() == KeyEvent.KEYCODE_BACK) {
final int action = event.getAction();
// Back cancels action modes first.
if (mPrimaryActionMode != null) {
if (action == KeyEvent.ACTION_UP) {
mPrimaryActionMode.finish();
}
return true;
}
}
// 1. If the dispatchKeyEvent method of ViewGroup is consumed, return true and don't go below
if (super.dispatchKeyEvent(event)) {
return true;
}
// 2. If ViewRootImpl is not empty and consumed by the dispatchUnhandledKeyEvent method of ViewRootImpl, it returns true
return (getViewRootImpl() != null) && getViewRootImpl().dispatchUnhandledKeyEvent(event);
}
At this point, let's look at the dispatchKeyEvent method of ViewGroup:
@Override
public boolean dispatchKeyEvent(KeyEvent event) {
if (mInputEventConsistencyVerifier != null) {
mInputEventConsistencyVerifier.onKeyEvent(event, 1);
}
if ((mPrivateFlags & (PFLAG_FOCUSED | PFLAG_HAS_BOUNDS))
== (PFLAG_FOCUSED | PFLAG_HAS_BOUNDS)) {
// 1. If the ViewGroup is currently in focus or has boundaries, distribute it to View for processing
if (super.dispatchKeyEvent(event)) {
return true;
}
} else if (mFocused != null && (mFocused.mPrivateFlags & PFLAG_HAS_BOUNDS)
== PFLAG_HAS_BOUNDS) {
// 2. If there is a View to get focus in the ViewGroup and the ViewGroup has a boundary, it will be handed over to mFocused for processing
if (mFocused.dispatchKeyEvent(event)) {
return true;
}
}
if (mInputEventConsistencyVerifier != null) {
mInputEventConsistencyVerifier.onUnhandledEvent(event, 1);
}
return false;
}
Next, take a look at the dispatchKeyEvent method of View:
/**
* Dispatch a key event to the next view on the focus path. This path runs
* from the top of the view tree down to the currently focused view. If this
* view has focus, it will dispatch to itself. Otherwise it will dispatch
* the next node down the focus path. This method also fires any key
* listeners.
*
* @param event The key event to be dispatched.
* @return True if the event was handled, false otherwise.
*/
public boolean dispatchKeyEvent(KeyEvent event) {
if (mInputEventConsistencyVerifier != null) {
mInputEventConsistencyVerifier.onKeyEvent(event, 0);
}
// Give any attached key listener a first crack at the event.
//noinspection SimplifiableIfStatement
ListenerInfo li = mListenerInfo;
// 1. If OnKeyListener is not empty and view is ENABLED, the listener will trigger first
if (li != null && li.mOnKeyListener != null && (mViewFlags & ENABLED_MASK) == ENABLED
&& li.mOnKeyListener.onKey(this, event.getKeyCode(), event)) {
return true;
}
// 2. Call the dispatch method of KeyEvent and pass the view object itself as a parameter
if (event.dispatch(this, mAttachInfo != null
? mAttachInfo.mKeyDispatchState : null, this)) {
return true;
}
if (mInputEventConsistencyVerifier != null) {
mInputEventConsistencyVerifier.onUnhandledEvent(event, 0);
}
return false;
}
This method mainly judges that if the OnKeyListener event is set for the View and the View is available, the listening event will be processed first, and then the dispatch method of KeyEvent will be called. Next, let's look at this method:
/**
* Deliver this key event to a {@link Callback} interface. If this is
* an ACTION_MULTIPLE event and it is not handled, then an attempt will
* be made to deliver a single normal event.
*
* @param receiver The Callback that will be given the event.
* @param state State information retained across events.
* @param target The target of the dispatch, for use in tracking.
*
* @return The return value from the Callback method that was called.
*/
public final boolean dispatch(Callback receiver, DispatcherState state,
Object target) {
switch (mAction) {
case ACTION_DOWN: {
mFlags &= ~FLAG_START_TRACKING;
if (DEBUG) Log.v(TAG, "Key down to " + target + " in " + state
+ ": " + this);
boolean res = receiver.onKeyDown(mKeyCode, this);
if (state != null) {
if (res && mRepeatCount == 0 && (mFlags&FLAG_START_TRACKING) != 0) {
if (DEBUG) Log.v(TAG, " Start tracking!");
state.startTracking(this, target);
} else if (isLongPress() && state.isTracking(this)) {
try {
if (receiver.onKeyLongPress(mKeyCode, this)) {
if (DEBUG) Log.v(TAG, " Clear from long press!");
state.performedLongPress(this);
res = true;
}
} catch (AbstractMethodError e) {
}
}
}
return res;
}
case ACTION_UP:
if (DEBUG) Log.v(TAG, "Key up to " + target + " in " + state
+ ": " + this);
if (state != null) {
state.handleUpEvent(this);
}
return receiver.onKeyUp(mKeyCode, this);
case ACTION_MULTIPLE:
final int count = mRepeatCount;
final int code = mKeyCode;
if (receiver.onKeyMultiple(code, count, this)) {
return true;
}
if (code != KeyEvent.KEYCODE_UNKNOWN) {
mAction = ACTION_DOWN;
mRepeatCount = 0;
boolean handled = receiver.onKeyDown(code, this);
if (handled) {
mAction = ACTION_UP;
receiver.onKeyUp(code, this);
}
mAction = ACTION_MULTIPLE;
mRepeatCount = count;
return handled;
}
return false;
}
return false;
}
This method mainly deals with press and pop-up events. If the number of mRepeatCount repetitions is greater than 0, it is judged as a long press, and the long press event is executed.
Let's continue to look at the onKeyDown method of View:
/**
* Default implementation of {@link KeyEvent.Callback#onKeyDown(int, KeyEvent)
* KeyEvent.Callback.onKeyDown()}: perform press of the view
* when {@link KeyEvent#KEYCODE_DPAD_CENTER} or {@link KeyEvent#KEYCODE_ENTER}
* is released, if the view is enabled and clickable.
* <p>
* Key presses in software keyboards will generally NOT trigger this
* listener, although some may elect to do so in some situations. Do not
* rely on this to catch software key presses.
*
* @param keyCode a key code that represents the button pressed, from
* {@link android.view.KeyEvent}
* @param event the KeyEvent object that defines the button action
*/
public boolean onKeyDown(int keyCode, KeyEvent event) {
if (KeyEvent.isConfirmKey(keyCode)) {
// 1. Return true if the View is unavailable
if ((mViewFlags & ENABLED_MASK) == DISABLED) {
return true;
}
// 2. If the event repeats for 0 times and the View is clickable or can be pressed for a long time, set the middle coordinate of pressing the View and check the long press
if (event.getRepeatCount() == 0) {
// Long clickable items don't necessarily have to be clickable.
final boolean clickable = (mViewFlags & CLICKABLE) == CLICKABLE
|| (mViewFlags & LONG_CLICKABLE) == LONG_CLICKABLE;
if (clickable || (mViewFlags & TOOLTIP) == TOOLTIP) {
// For the purposes of menu anchoring and drawable hotspots,
// key events are considered to be at the center of the view.
final float x = getWidth() / 2f;
final float y = getHeight() / 2f;
if (clickable) {
setPressed(true, x, y);
}
checkForLongClick(0, x, y);
return true;
}
}
}
return false;
}
Press the event to judge. If it is to confirm the relevant keys before proceeding to the next step, judge that the click or long press conditions are met, press the View positive center coordinate, and then execute checkForLongClick to check the long press method. See the method as follows:
private void checkForLongClick(int delayOffset, float x, float y) {
if ((mViewFlags & LONG_CLICKABLE) == LONG_CLICKABLE || (mViewFlags & TOOLTIP) == TOOLTIP) {
mHasPerformedLongPress = false;
if (mPendingCheckForLongPress == null) {
mPendingCheckForLongPress = new CheckForLongPress();
}
mPendingCheckForLongPress.setAnchor(x, y);
mPendingCheckForLongPress.rememberWindowAttachCount();
mPendingCheckForLongPress.rememberPressedState();
// 1. Delay the execution of mPendingCheckForLongPress, long press Runnable,
postDelayed(mPendingCheckForLongPress,
ViewConfiguration.getLongPressTimeout() - delayOffset);
}
}
We often encounter that when the TV presses the remote control for a long time, onKeyDown and onKeyUp will be executed once, then onKeyDown will be executed all the time, and onKeyUp will be executed only after loosening. The reason is that this check method is delayed. delayOffset passes in 0, so the delay time is ViewConfiguration.getLongPressTimeout(), which is the default defined in this class_ LONG_ PRESS_ Timeout constant.
Similarly, if it is a touch screen, you can look at the onTouchEvent method in the View class. When pressing, the CheckForTap thread will be started to check whether it is a long press. This thread is also delayed, and the time is viewconfiguration Gettaptimeout(), that is, tap in this class_ Timeout constant. After knowing this, you will know how long it should be before writing scripts or plug-ins to simulate long-term pressing. Is it possible to realize the speed of your simulated long-term pressing plug-ins more accurately and quickly.
The delay time defined by different versions of the system may be different, such as the default of GoogleAPI 28_ LONG_ PRESS_ Timeout is 500, TAP_TIMEOUT is 100, while default of API 30_ LONG_ PRESS_ Timeout is 400, TAP_TIMEOUT is also 100.
Next, let's look at onKeyDown:
/**
* Called when a key was pressed down and not handled by any of the views
* inside of the activity. So, for example, key presses while the cursor
* is inside a TextView will not trigger the event (unless it is a navigation
* to another object) because TextView handles its own key presses.
*
* <p>If the focused view didn't want this event, this method is called.
*
* <p>The default implementation takes care of {@link KeyEvent#KEYCODE_BACK}
* by calling {@link #onBackPressed()}, though the behavior varies based
* on the application compatibility mode: for
* {@link android.os.Build.VERSION_CODES#ECLAIR} or later applications,
* it will set up the dispatch to call {@link #onKeyUp} where the action
* will be performed; for earlier applications, it will perform the
* action immediately in on-down, as those versions of the platform
* behaved.
*
* <p>Other additional default key handling may be performed
* if configured with {@link #setDefaultKeyMode}.
*
* @return Return <code>true</code> to prevent this event from being propagated
* further, or <code>false</code> to indicate that you have not handled
* this event and it should continue to be propagated.
* @see #onKeyUp
* @see android.view.KeyEvent
*/
public boolean onKeyDown(int keyCode, KeyEvent event) {
// 1. When pressing the return key, call onBackPressed. If not, check whether the onKeyDown method is rewritten and rerun is true, resulting in
if (keyCode == KeyEvent.KEYCODE_BACK) {
if (getApplicationInfo().targetSdkVersion
>= Build.VERSION_CODES.ECLAIR) {
event.startTracking();
} else {
onBackPressed();
}
return true;
}
if (mDefaultKeyMode == DEFAULT_KEYS_DISABLE) {
return false;
} else if (mDefaultKeyMode == DEFAULT_KEYS_SHORTCUT) {
Window w = getWindow();
if (w.hasFeature(Window.FEATURE_OPTIONS_PANEL) &&
w.performPanelShortcut(Window.FEATURE_OPTIONS_PANEL, keyCode, event,
Menu.FLAG_ALWAYS_PERFORM_CLOSE)) {
return true;
}
return false;
} else if (keyCode == KeyEvent.KEYCODE_TAB) {
// Don't consume TAB here since it's used for navigation. Arrow keys
// aren't considered "typing keys" so they already won't get consumed.
return false;
} else {
// Common code for DEFAULT_KEYS_DIALER & DEFAULT_KEYS_SEARCH_*
boolean clearSpannable = false;
boolean handled;
if ((event.getRepeatCount() != 0) || event.isSystem()) {
clearSpannable = true;
handled = false;
} else {
handled = TextKeyListener.getInstance().onKeyDown(
null, mDefaultKeySsb, keyCode, event);
if (handled && mDefaultKeySsb.length() > 0) {
// something useable has been typed - dispatch it now.
final String str = mDefaultKeySsb.toString();
clearSpannable = true;
switch (mDefaultKeyMode) {
case DEFAULT_KEYS_DIALER:
Intent intent = new Intent(Intent.ACTION_DIAL, Uri.parse("tel:" + str));
intent.addFlags(Intent.FLAG_ACTIVITY_NEW_TASK);
startActivity(intent);
break;
case DEFAULT_KEYS_SEARCH_LOCAL:
startSearch(str, false, null, false);
break;
case DEFAULT_KEYS_SEARCH_GLOBAL:
startSearch(str, false, null, true);
break;
}
}
}
if (clearSpannable) {
mDefaultKeySsb.clear();
mDefaultKeySsb.clearSpans();
Selection.setSelection(mDefaultKeySsb,0);
}
return handled;
}
}
Go back to the dispatchKeyEvent method in the Decorview to see the onKeyDown method of PhoneWindow:
/**
* A key was pressed down and not handled by anything else in the window.
*
* @see #onKeyUp
* @see android.view.KeyEvent
*/
protected boolean onKeyDown(int featureId, int keyCode, KeyEvent event) {
/* ****************************************************************************
* HOW TO DECIDE WHERE YOUR KEY HANDLING GOES.
*
* If your key handling must happen before the app gets a crack at the event,
* it goes in PhoneWindowManager.
*
* If your key handling should happen in all windows, and does not depend on
* the state of the current application, other than that the current
* application can override the behavior by handling the event itself, it
* should go in PhoneFallbackEventHandler.
*
* Only if your handling depends on the window, and the fact that it has
* a DecorView, should it go here.
* ****************************************************************************/
final KeyEvent.DispatcherState dispatcher =
mDecor != null ? mDecor.getKeyDispatcherState() : null;
//Log.i(TAG, "Key down: repeat=" + event.getRepeatCount()
// + " flags=0x" + Integer.toHexString(event.getFlags()));
switch (keyCode) {
case KeyEvent.KEYCODE_VOLUME_UP:
case KeyEvent.KEYCODE_VOLUME_DOWN:
case KeyEvent.KEYCODE_VOLUME_MUTE: {
// If we have a session send it the volume command, otherwise
// use the suggested stream.
if (mMediaController != null) {
mMediaController.dispatchVolumeButtonEventAsSystemService(event);
} else {
getMediaSessionManager().dispatchVolumeKeyEventAsSystemService(event,
mVolumeControlStreamType);
}
return true;
}
// These are all the recognized media key codes in
// KeyEvent.isMediaKey()
case KeyEvent.KEYCODE_MEDIA_PLAY:
case KeyEvent.KEYCODE_MEDIA_PAUSE:
case KeyEvent.KEYCODE_MEDIA_PLAY_PAUSE:
case KeyEvent.KEYCODE_MUTE:
case KeyEvent.KEYCODE_HEADSETHOOK:
case KeyEvent.KEYCODE_MEDIA_STOP:
case KeyEvent.KEYCODE_MEDIA_NEXT:
case KeyEvent.KEYCODE_MEDIA_PREVIOUS:
case KeyEvent.KEYCODE_MEDIA_REWIND:
case KeyEvent.KEYCODE_MEDIA_RECORD:
case KeyEvent.KEYCODE_MEDIA_FAST_FORWARD: {
if (mMediaController != null) {
if (mMediaController.dispatchMediaButtonEventAsSystemService(event)) {
return true;
}
}
return false;
}
case KeyEvent.KEYCODE_MENU: {
onKeyDownPanel((featureId < 0) ? FEATURE_OPTIONS_PANEL : featureId, event);
return true;
}
case KeyEvent.KEYCODE_BACK: {
if (event.getRepeatCount() > 0) break;
if (featureId < 0) break;
// Currently don't do anything with long press.
if (dispatcher != null) {
dispatcher.startTracking(event, this);
}
return true;
}
}
return false;
}
The onKeyUp method can also be seen by yourself. The above is about the distribution process of key events.
Summary:
- The distribution direction of keys is DecorView - Activity - ViewGroup - View
- Consumption direction of keys View - Activity - PhoneWindow
- The processing priority of various Callback interfaces is lower than that of listeners, that is, the methods of various onxxxlisteners are called first
3, Focus navigation process
The above explains the key event distribution process. When all the above distribution is finished without consumption, we will continue to follow the focus navigation process of ViewRootImpl. Next, let's look at the performFocusNavigation method:
private boolean performFocusNavigation(KeyEvent event) {
int direction = 0;
// 1. Determine the direction keys up, down, left, right and Tab
switch (event.getKeyCode()) {
case KeyEvent.KEYCODE_DPAD_LEFT:
if (event.hasNoModifiers()) {
direction = View.FOCUS_LEFT;
}
break;
case KeyEvent.KEYCODE_DPAD_RIGHT:
if (event.hasNoModifiers()) {
direction = View.FOCUS_RIGHT;
}
break;
case KeyEvent.KEYCODE_DPAD_UP:
if (event.hasNoModifiers()) {
direction = View.FOCUS_UP;
}
break;
case KeyEvent.KEYCODE_DPAD_DOWN:
if (event.hasNoModifiers()) {
direction = View.FOCUS_DOWN;
}
break;
case KeyEvent.KEYCODE_TAB:
if (event.hasNoModifiers()) {
direction = View.FOCUS_FORWARD;
} else if (event.hasModifiers(KeyEvent.META_SHIFT_ON)) {
direction = View.FOCUS_BACKWARD;
}
break;
}
if (direction != 0) {
// 2.mView: that is, DecorView. DecorView is the top-level View of the entire ViewTree and represents the interface of the entire application
View focused = mView.findFocus();
if (focused != null) {
// 3. Find the focused that currently gets the focus, and call the focusSearch method of the focus view
View v = focused.focusSearch(direction);
if (v != null && v != focused) {
// do the math the get the interesting rect
// of previous focused into the coord system of
// newly focused view
focused.getFocusedRect(mTempRect);
if (mView instanceof ViewGroup) {
((ViewGroup) mView).offsetDescendantRectToMyCoords(
focused, mTempRect);
((ViewGroup) mView).offsetRectIntoDescendantCoords(
v, mTempRect);
}
// 4. If the next view that can get the focus found is not the view that has currently obtained the focus, call the requestFocus method
if (v.requestFocus(direction, mTempRect)) {
playSoundEffect(SoundEffectConstants
.getContantForFocusDirection(direction));
return true;
}
}
// 5. Give the currently focused view a last chance to handle the event
// Give the focused view a last chance to handle the dpad key.
if (mView.dispatchUnhandledMove(focused, direction)) {
return true;
}
} else {
// 6. Call recursively to reset the default focus (there can only be one default focus view on the whole view tree)
if (mView.restoreDefaultFocus()) {
return true;
}
}
}
return false;
}
findFocus
First, let's look at mView.findFocus(), which actually calls the findFocus method of ViewGroup:
/*
* (non-Javadoc)
*
* @see android.view.View#findFocus()
*/
@Override
public View findFocus() {
if (DBG) {
System.out.println("Find focus in " + this + ": flags="
+ isFocused() + ", child=" + mFocused);
}
if (isFocused()) {
return this;
}
if (mFocused != null) {
return mFocused.findFocus();
}
return null;
}
focusSearch
This method is very simple, which is to recursively find the View that has obtained the focus on the current page. Continue to look at focused.focusSearch(direction) calls the focusSearch method of View:
/**
* Find the nearest view in the specified direction that can take focus.
* This does not actually give focus to that view.
*
* @param direction One of FOCUS_UP, FOCUS_DOWN, FOCUS_LEFT, and FOCUS_RIGHT
*
* @return The nearest focusable in the specified direction, or null if none
* can be found.
*/
public View focusSearch(@FocusRealDirection int direction) {
if (mParent != null) {
return mParent.focusSearch(this, direction);
} else {
return null;
}
}
This method recursively looks up and calls the focusSearch method of ViewGroup:
/**
* Find the nearest view in the specified direction that wants to take
* focus.
*
* @param focused The view that currently has focus
* @param direction One of FOCUS_UP, FOCUS_DOWN, FOCUS_LEFT, and
* FOCUS_RIGHT, or 0 for not applicable.
*/
@Override
public View focusSearch(View focused, int direction) {
if (isRootNamespace()) {
// root namespace means we should consider ourselves the top of the
// tree for focus searching; otherwise we could be focus searching
// into other tabs. see LocalActivityManager and TabHost for more info.
return FocusFinder.getInstance().findNextFocus(this, focused, direction);
} else if (mParent != null) {
return mParent.focusSearch(focused, direction);
}
return null;
}
If it is a root namespace, call the findNextFocus method of FocusFinder to find the focus, otherwise continue to search upward. Continue to look at the findNextFocus method of FocusFinder:
/**
* Find the next view to take focus in root's descendants, starting from the view
* that currently is focused.
* @param root Contains focused. Cannot be null.
* @param focused Has focus now.
* @param direction Direction to look.
* @return The next focusable view, or null if none exists.
*/
public final View findNextFocus(ViewGroup root, View focused, int direction) {
return findNextFocus(root, focused, null, direction);
}
private View findNextFocus(ViewGroup root, View focused, Rect focusedRect, int direction) {
View next = null;
ViewGroup effectiveRoot = getEffectiveRoot(root, focused);
if (focused != null) {
// 1. Find the view specified by the user to get the next focus
next = findNextUserSpecifiedFocus(effectiveRoot, focused, direction);
}
if (next != null) {
// 2. If the focus specified by the user is found, it will be returned directly
return next;
}
ArrayList<View> focusables = mTempList;
try {
focusables.clear();
// 3. Add all view s under the effectiveRoot to the focusables set, and rewrite the method of ViewGroup to realize the focus memory function
effectiveRoot.addFocusables(focusables, direction);
if (!focusables.isEmpty()) {
// 4. According to the default proximity principle algorithm of the system, find the next nearest view that can obtain the focus
next = findNextFocus(effectiveRoot, focused, focusedRect, direction, focusables);
}
} finally {
focusables.clear();
}
return next;
}
You can see that this method first finds the next view specified by the user to obtain the focus. If it is found, it will directly return to this view. If it is not found, continue to add all views under the effectiveRoot to the focusables set, and then call the findNextFocus method to find the latest view that the system can obtain the next focus.
findNextUserSpecifiedFocus
Let's first look at the implementation of the findnextuserspecificedfocus method:
private View findNextUserSpecifiedFocus(ViewGroup root, View focused, int direction) {
// check for user specified next focus
View userSetNextFocus = focused.findUserSetNextFocus(root, direction);
View cycleCheck = userSetNextFocus;
boolean cycleStep = true; // we want the first toggle to yield false
while (userSetNextFocus != null) {
if (userSetNextFocus.isFocusable()
&& userSetNextFocus.getVisibility() == View.VISIBLE
&& (!userSetNextFocus.isInTouchMode()
|| userSetNextFocus.isFocusableInTouchMode())) {
return userSetNextFocus;
}
userSetNextFocus = userSetNextFocus.findUserSetNextFocus(root, direction);
if (cycleStep = !cycleStep) {
cycleCheck = cycleCheck.findUserSetNextFocus(root, direction);
if (cycleCheck == userSetNextFocus) {
// found a cycle, user-specified focus forms a loop and none of the views
// are currently focusable.
break;
}
}
}
return null;
}
The focus method specified by the user is not the focus of this article, so the internal details of the source code will not be posted here. This method is actually to call the findUserSetNextFocus method of view to find the next view that can get the focus set by the user, and then judge in the while loop that if it is found that it can get the focus and is visible and not in the InTouchNode mode, it will return to the focus. If not, it will continue to cycle until a loop is found and no one can get the focus, or userSetNextFocus is null, jump out of the loop and return null.
findNextFocus
Let's take a look at the findNextFocus method of system proximity search:
private View findNextFocus(ViewGroup root, View focused, Rect focusedRect,
int direction, ArrayList<View> focusables) {
if (focused != null) {
if (focusedRect == null) {
focusedRect = mFocusedRect;
}
// 1. Assign focusedRect as the visible drawing boundary of the view view that has obtained focus
// fill in interesting rect from focused
focused.getFocusedRect(focusedRect);
root.offsetDescendantRectToMyCoords(focused, focusedRect);
} else {
if (focusedRect == null) {
focusedRect = mFocusedRect;
// make up a rect at top left or bottom right of root
switch (direction) {
case View.FOCUS_RIGHT:
case View.FOCUS_DOWN:
// 2. Modify the upper left corner edge of focusedRect to the upper left corner edge of root
setFocusTopLeft(root, focusedRect);
break;
case View.FOCUS_FORWARD:
if (root.isLayoutRtl()) {
setFocusBottomRight(root, focusedRect);
} else {
setFocusTopLeft(root, focusedRect);
}
break;
case View.FOCUS_LEFT:
case View.FOCUS_UP:
// 3. Modify the lower right corner edge of focusedRect to the lower right corner edge of root
setFocusBottomRight(root, focusedRect);
break;
case View.FOCUS_BACKWARD:
if (root.isLayoutRtl()) {
setFocusTopLeft(root, focusedRect);
} else {
setFocusBottomRight(root, focusedRect);
break;
}
}
}
}
switch (direction) {
case View.FOCUS_FORWARD:
case View.FOCUS_BACKWARD:
// 4. Find the next focus in the opposite direction
return findNextFocusInRelativeDirection(focusables, root, focused, focusedRect,
direction);
case View.FOCUS_UP:
case View.FOCUS_DOWN:
case View.FOCUS_LEFT:
case View.FOCUS_RIGHT:
// 5. Find the next focus in the absolute relative direction
return findNextFocusInAbsoluteDirection(focusables, root, focused,
focusedRect, direction);
default:
throw new IllegalArgumentException("Unknown direction: " + direction);
}
}
This method mainly finds the next focus in the relative direction through findNextFocusInRelativeDirection. The internal logic of this method is relatively simple, so it won't be posted here. After going in and looking at it, you will know that it is actually to sort focusables first, and then find the next or previous view of focused in it. If it is not found and focusables is not empty, the first one of focusables will be returned.
Next, let's focus on the findNextFocusInAbsoluteDirection method:
View findNextFocusInAbsoluteDirection(ArrayList<View> focusables, ViewGroup root, View focused,
Rect focusedRect, int direction) {
// initialize the best candidate to something impossible
// (so the first plausible view will become the best choice)
mBestCandidateRect.set(focusedRect);
switch(direction) {
case View.FOCUS_LEFT:
mBestCandidateRect.offset(focusedRect.width() + 1, 0);
break;
case View.FOCUS_RIGHT:
mBestCandidateRect.offset(-(focusedRect.width() + 1), 0);
break;
case View.FOCUS_UP:
mBestCandidateRect.offset(0, focusedRect.height() + 1);
break;
case View.FOCUS_DOWN:
mBestCandidateRect.offset(0, -(focusedRect.height() + 1));
}
View closest = null;
int numFocusables = focusables.size();
for (int i = 0; i < numFocusables; i++) {
View focusable = focusables.get(i);
// only interested in other non-root views
if (focusable == focused || focusable == root) continue;
// get focus bounds of other view in same coordinate system
focusable.getFocusedRect(mOtherRect);
root.offsetDescendantRectToMyCoords(focusable, mOtherRect);
if (isBetterCandidate(direction, focusedRect, mOtherRect, mBestCandidateRect)) {
mBestCandidateRect.set(mOtherRect);
closest = focusable;
}
}
return closest;
}
isBetterCandidate
Let's take a look at the isBetterCandidate method. This method is very critical. It contains a series of internal logic. How to become the best candidate:
/**
* Is rect1 a better candidate than rect2 for a focus search in a particular
* direction from a source rect? This is the core routine that determines
* the order of focus searching.
* @param direction the direction (up, down, left, right)
* @param source The source we are searching from
* @param rect1 The candidate rectangle
* @param rect2 The current best candidate.
* @return Whether the candidate is the new best.
*/
boolean isBetterCandidate(int direction, Rect source, Rect rect1, Rect rect2) {
// 1.source = focusedRect；rect1 = mOtherRect；rect2 = mBestCandidateRect
// Take the left focusing as an example, judge whether the right side of rect1 is on the left side of the right of source and the left side of rect1 is on the left side of the left of source
// to be a better candidate, need to at least be a candidate in the first
// place :)
if (!isCandidate(source, rect1, direction)) {
return false;
}
// 2. Judge whether react2 is on the left of source. If not, select react1. If yes, continue to judge below
// we know that rect1 is a candidate.. if rect2 is not a candidate,
// rect1 is better
if (!isCandidate(source, rect2, direction)) {
return true;
}
// 3. Judge who is more suitable according to whether the direction overlaps and the distance
// if rect1 is better by beam, it wins
if (beamBeats(direction, source, rect1, rect2)) {
return true;
}
// 4. Exchange react1 and react1 and continue to compare
// if rect2 is better, then rect1 cant' be :)
if (beamBeats(direction, source, rect2, rect1)) {
return false;
}
// 5. Otherwise, continue to compare the distance
// otherwise, do fudge-tastic comparison of the major and minor axis
return (getWeightedDistanceFor(
majorAxisDistance(direction, source, rect1),
minorAxisDistance(direction, source, rect1))
< getWeightedDistanceFor(
majorAxisDistance(direction, source, rect2),
minorAxisDistance(direction, source, rect2)));
}
The English annotation of this method is very intuitive, so it is not translated in Chinese. First, let's take a look at the isCandidate method of becoming a candidate:
/**
* Is destRect a candidate for the next focus given the direction? This
* checks whether the dest is at least partially to the direction of (e.g left of)
* from source.
*
* Includes an edge case for an empty rect (which is used in some cases when
* searching from a point on the screen).
*/
boolean isCandidate(Rect srcRect, Rect destRect, int direction) {
switch (direction) {
case View.FOCUS_LEFT:
return (srcRect.right > destRect.right || srcRect.left >= destRect.right)
&& srcRect.left > destRect.left;
case View.FOCUS_RIGHT:
return (srcRect.left < destRect.left || srcRect.right <= destRect.left)
&& srcRect.right < destRect.right;
case View.FOCUS_UP:
return (srcRect.bottom > destRect.bottom || srcRect.top >= destRect.bottom)
&& srcRect.top > destRect.top;
case View.FOCUS_DOWN:
return (srcRect.top < destRect.top || srcRect.bottom <= destRect.top)
&& srcRect.bottom < destRect.bottom;
}
throw new IllegalArgumentException("direction must be one of "
+ "{FOCUS_UP, FOCUS_DOWN, FOCUS_LEFT, FOCUS_RIGHT}.");
}
This method determines that if the target Rect is on the direction side of the source Rect and not inside, it is a candidate. For example, the left side of the first destRect should be on the left side of srcRect, the right side of destRect should be on the left side of srcRect, and the other directions are the same.
Next, take a look at the beamBeats method:
/**
* One rectangle may be another candidate than another by virtue of being
* exclusively in the beam of the source rect.
* @return Whether rect1 is a better candidate than rect2 by virtue of it being in src's
* beam
*/
boolean beamBeats(int direction, Rect source, Rect rect1, Rect rect2) {
final boolean rect1InSrcBeam = beamsOverlap(direction, source, rect1);
final boolean rect2InSrcBeam = beamsOverlap(direction, source, rect2);
// 1. Take the left as an example. If rect2 coincides in the vertical direction or rect1 does not coincide in the vertical direction, rect1 is not more suitable than rect2
// if rect1 isn't exclusively in the src beam, it doesn't win
if (rect2InSrcBeam || !rect1InSrcBeam) {
return false;
}
// we know rect1 is in the beam, and rect2 is not
// 2.source is not on the left of rect2
// if rect1 is to the direction of, and rect2 is not, rect1 wins.
// for example, for direction left, if rect1 is to the left of the source
// and rect2 is below, then we always prefer the in beam rect1, since rect2
// could be reached by going down.
if (!isToDirectionOf(direction, source, rect2)) {
return true;
}
// 3. If it is left-right direction, react1 is more appropriate
// for horizontal directions, being exclusively in beam always wins
if ((direction == View.FOCUS_LEFT || direction == View.FOCUS_RIGHT)) {
return true;
}
// 4. Take pressing the key as an example. If the distance from the top of rect1 to the bottom of source is less than the distance from the bottom of rect2 to the bottom of source, rect1 is more appropriate
// for vertical directions, beams only beat up to a point:
// now, as long as rect2 isn't completely closer, rect1 wins
// e.g for direction down, completely closer means for rect2's top
// edge to be closer to the source's top edge than rect1's bottom edge.
return (majorAxisDistance(direction, source, rect1)
< majorAxisDistanceToFarEdge(direction, source, rect2));
}
The above method judges who is more suitable according to the overlap, horizontal direction and vertical distance. Next, take a look at the getWeightedDistanceFor method in the isBetterCandidate method above:
/**
* Fudge-factor opportunity: how to calculate distance given major and minor
* axis distances. Warning: this fudge factor is finely tuned, be sure to
* run all focus tests if you dare tweak it.
*/
long getWeightedDistanceFor(long majorAxisDistance, long minorAxisDistance) {
return 13 * majorAxisDistance * majorAxisDistance
+ minorAxisDistance * minorAxisDistance;
}
The majorAxisDistanceToFarEdge method is mainly used to calculate the distance between the current focus area (src) and the far edge of the candidate area (dest) in the target direction.
The top minorAxisDistance method mainly calculates the distance from the center point of the current focus area (src) to the center point of the candidate area (dest) in the non target direction.
requestFocus
Next, let's take a look at the method of requesting focus after finding the focus that can be obtained at the beginning requestFocus:
/**
* Call this to try to give focus to a specific view or to one of its descendants
* and give it hints about the direction and a specific rectangle that the focus
* is coming from. The rectangle can help give larger views a finer grained hint
* about where focus is coming from, and therefore, where to show selection, or
* forward focus change internally.
*
* A view will not actually take focus if it is not focusable ({@link #isFocusable} returns
* false), or if it is focusable and it is not focusable in touch mode
* ({@link #isFocusableInTouchMode}) while the device is in touch mode.
*
* A View will not take focus if it is not visible.
*
* A View will not take focus if one of its parents has
* {@link android.view.ViewGroup#getDescendantFocusability()} equal to
* {@link ViewGroup#FOCUS_BLOCK_DESCENDANTS}.
*
* See also {@link #focusSearch(int)}, which is what you call to say that you
* have focus, and you want your parent to look for the next one.
*
* You may wish to override this method if your custom {@link View} has an internal
* {@link View} that it wishes to forward the request to.
*
* @param direction One of FOCUS_UP, FOCUS_DOWN, FOCUS_LEFT, and FOCUS_RIGHT
* @param previouslyFocusedRect The rectangle (in this View's coordinate system)
* to give a finer grained hint about where focus is coming from. May be null
* if there is no hint.
* @return Whether this view or one of its descendants actually took focus.
*/
public boolean requestFocus(int direction, Rect previouslyFocusedRect) {
return requestFocusNoSearch(direction, previouslyFocusedRect);
}
You can see that this method calls the requestFocus method again:
private boolean requestFocusNoSearch(int direction, Rect previouslyFocusedRect) {
// need to be focusable
if (!canTakeFocus()) {
return false;
}
// need to be focusable in touch mode if in touch mode
if (isInTouchMode() &&
(FOCUSABLE_IN_TOUCH_MODE != (mViewFlags & FOCUSABLE_IN_TOUCH_MODE))) {
return false;
}
// need to not have any parents blocking us
if (hasAncestorThatBlocksDescendantFocus()) {
return false;
}
if (!isLayoutValid()) {
mPrivateFlags |= PFLAG_WANTS_FOCUS;
} else {
clearParentsWantFocus();
}
handleFocusGainInternal(direction, previouslyFocusedRect);
return true;
}
This method determines that if the focus is not available or the touch screen is not set with focus_ IN_ TOUCH_ The mode attribute returns false and does not proceed. If the parent ViewGroup has set a mask for the child View to get the focus, it also returns false. Then look at the last handleFocusGainInternal method:
/**
* Give this view focus. This will cause
* {@link #onFocusChanged(boolean, int, android.graphics.Rect)} to be called.
*
* Note: this does not check whether this {@link View} should get focus, it just
* gives it focus no matter what. It should only be called internally by framework
* code that knows what it is doing, namely {@link #requestFocus(int, Rect)}.
*
* @param direction values are {@link View#FOCUS_UP}, {@link View#FOCUS_DOWN},
* {@link View#FOCUS_LEFT} or {@link View#FOCUS_RIGHT}. This is the direction which
* focus moved when requestFocus() is called. It may not always
* apply, in which case use the default View.FOCUS_DOWN.
* @param previouslyFocusedRect The rectangle of the view that had focus
* prior in this View's coordinate system.
*/
void handleFocusGainInternal(@FocusRealDirection int direction, Rect previouslyFocusedRect) {
if (DBG) {
System.out.println(this + " requestFocus()");
}
if ((mPrivateFlags & PFLAG_FOCUSED) == 0) {
mPrivateFlags |= PFLAG_FOCUSED;
// 1. Find the old view that has obtained the focus in the root view
View oldFocus = (mAttachInfo != null) ? getRootView().findFocus() : null;
// 2. Call the requestChildFocus method in the parent control
if (mParent != null) {
mParent.requestChildFocus(this, this);
updateFocusedInCluster(oldFocus, direction);
}
// 3. Call the distribution focus change monitor of ViewTreeObserver, the old view that has obtained the focus and the current view
if (mAttachInfo != null) {
mAttachInfo.mTreeObserver.dispatchOnGlobalFocusChange(oldFocus, this);
}
// 4. Call the onFocusChanged callback of view
onFocusChanged(true, direction, previouslyFocusedRect);
refreshDrawableState();
}
}
You can see that this method first obtains the view that has obtained the focus, then calls the request focus method of the parent control, then notifies the two listeners set respectively, and finally refreshes the status of Drawable. Then look at the requestChildFocus method in ViewGroup:
@Override
public void requestChildFocus(View child, View focused) {
if (DBG) {
System.out.println(this + " requestChildFocus()");
}
// 1. If you have set a mask sub control to get the focus, return
if (getDescendantFocusability() == FOCUS_BLOCK_DESCENDANTS) {
return;
}
// Unfocus us, if necessary
super.unFocus(focused);
// 2. If the mFocused that has obtained the Focus is not the passed in child, clear the Focus state of mFocused and assign the child to mFocused
// We had a previous notion of who had focus. Clear it.
if (mFocused != child) {
if (mFocused != null) {
mFocused.unFocus(focused);
}
mFocused = child;
}
if (mParent != null) {
mParent.requestChildFocus(this, focused);
}
}
According to the above method, if focus is not set for ViewGroup_ BLOCK_ The occurrences attribute goes down. If the mFocused that has obtained the focus is not the passed in child, clear the focus state of mFocused, and assign the child to mFocused, that is, let the child obtain the focus.
So far, the analysis of Android TV button focus principle is over. After learning this article, go back to the opening question. Have you had a deeper understanding?