上一篇文章Android软键盘windowSoftInputMode的使用与原理使用篇介绍了windowSoftInputMode各个属性的具体作用,下面我们在源码层面来分析一下各个属性的实现原理(基于Android9)。
要去分析软键盘的隐藏和显示首先要搞懂Android系统的软键盘架构是怎样的,看下图:
InputMethodManagerService(下文也称IMMS)负责管理系统的所有输入法,包括输入法service(InputMethodService简称IMS)加载及切换。程序获得焦点时,就会通过InputMethodManager向InputMethodManagerService通知自己获得焦点并请求绑定自己到当前输入法上。同时,当程序的某个需要输入法的view比如EditorView获得焦点时就会通过InputMethodManager向InputMethodManagerService请求显示输入法,而这时InputMethodManagerService收到请求后,会将请求的EditText的数据通信接口发送给当前输入法,并请求显输入法。输入法收到请求后,就显示自己的UI dialog,同时保存目标view的数据结构,当用户实现输入后,直接通过view的数据通信接口将字符传递到对应的View。
WindowManagerService(下文也称WMS)非常复杂,负责管理系统所有窗口,我们只需要知道当窗口发生变化(焦点状态和大小)他会通知到app进程。
当app的窗口也就是页面的Focus状态发生变化(例如:进入一个新的窗口,这个窗口的Focus状态变成ture,上一个窗口变成false),事件会回调到ViewRootImpl的WindowInputEventReceiver中,经过一系列调用之后会进入到IMMS的startInputOrWindowGainedFocusInternalLocked方法中,看一下这个方法的核心代码:
@NonNull
private InputBindResult startInputOrWindowGainedFocusInternalLocked(
@StartInputReason int startInputReason, IInputMethodClient client,
@NonNull IBinder windowToken, @StartInputFlags int startInputFlags,
@SoftInputModeFlags int softInputMode, int windowFlags, EditorInfo attribute,
IInputContext inputContext, @MissingMethodFlags int missingMethods,
int unverifiedTargetSdkVersion, @UserIdInt int userId) {
......省略........
InputBindResult res = null;
switch (softInputMode & LayoutParams.SOFT_INPUT_MASK_STATE) {
case LayoutParams.SOFT_INPUT_STATE_UNSPECIFIED:
if (!isTextEditor || !doAutoShow) {
if (LayoutParams.mayUseInputMethod(windowFlags)) {
// There is no focus view, and this window will
// be behind any soft input window, so hide the
// soft input window if it is shown.
if (DEBUG) Slog.v(TAG, "Unspecified window will hide input");
hideCurrentInputLocked(
mCurFocusedWindow, InputMethodManager.HIDE_NOT_ALWAYS, null,
SoftInputShowHideReason.HIDE_UNSPECIFIED_WINDOW);
// If focused display changed, we should unbind current method
// to make app window in previous display relayout after Ime
// window token removed.
// Note that we can trust client's display ID as long as it matches
// to the display ID obtained from the window.
if (cs.selfReportedDisplayId != mCurTokenDisplayId) {
unbindCurrentMethodLocked();
}
}
} else if (isTextEditor && doAutoShow
&& (softInputMode & LayoutParams.SOFT_INPUT_IS_FORWARD_NAVIGATION) != 0) {
// There is a focus view, and we are navigating forward
// into the window, so show the input window for the user.
// We only do this automatically if the window can resize
// to accommodate the IME (so what the user sees will give
// them good context without input information being obscured
// by the IME) or if running on a large screen where there
// is more room for the target window + IME.
if (DEBUG) Slog.v(TAG, "Unspecified window will show input");
if (attribute != null) {
res = startInputUncheckedLocked(cs, inputContext, missingMethods,
attribute, startInputFlags, startInputReason);
didStart = true;
}
showCurrentInputLocked(windowToken, InputMethodManager.SHOW_IMPLICIT, null,
SoftInputShowHideReason.SHOW_AUTO_EDITOR_FORWARD_NAV);
}
break;
case LayoutParams.SOFT_INPUT_STATE_UNCHANGED:
// Do nothing.
break;
case LayoutParams.SOFT_INPUT_STATE_HIDDEN:
if ((softInputMode & LayoutParams.SOFT_INPUT_IS_FORWARD_NAVIGATION) != 0) {
if (DEBUG) Slog.v(TAG, "Window asks to hide input going forward");
hideCurrentInputLocked(mCurFocusedWindow, 0, null,
SoftInputShowHideReason.HIDE_STATE_HIDDEN_FORWARD_NAV);
}
break;
case LayoutParams.SOFT_INPUT_STATE_ALWAYS_HIDDEN:
if (DEBUG) Slog.v(TAG, "Window asks to hide input");
hideCurrentInputLocked(mCurFocusedWindow, 0, null,
SoftInputShowHideReason.HIDE_ALWAYS_HIDDEN_STATE);
break;
case LayoutParams.SOFT_INPUT_STATE_VISIBLE:
if ((softInputMode & LayoutParams.SOFT_INPUT_IS_FORWARD_NAVIGATION) != 0) {
if (DEBUG) Slog.v(TAG, "Window asks to show input going forward");
if (InputMethodUtils.isSoftInputModeStateVisibleAllowed(
unverifiedTargetSdkVersion, startInputFlags)) {
if (attribute != null) {
res = startInputUncheckedLocked(cs, inputContext, missingMethods,
attribute, startInputFlags, startInputReason);
didStart = true;
}
showCurrentInputLocked(windowToken, InputMethodManager.SHOW_IMPLICIT, null,
SoftInputShowHideReason.SHOW_STATE_VISIBLE_FORWARD_NAV);
} else {
Slog.e(TAG, "SOFT_INPUT_STATE_VISIBLE is ignored because"
+ " there is no focused view that also returns true from"
+ " View#onCheckIsTextEditor()");
}
}
break;
case LayoutParams.SOFT_INPUT_STATE_ALWAYS_VISIBLE:
if (DEBUG) Slog.v(TAG, "Window asks to always show input");
if (InputMethodUtils.isSoftInputModeStateVisibleAllowed(
unverifiedTargetSdkVersion, startInputFlags)) {
if (attribute != null) {
res = startInputUncheckedLocked(cs, inputContext, missingMethods,
attribute, startInputFlags, startInputReason);
didStart = true;
}
showCurrentInputLocked(windowToken, InputMethodManager.SHOW_IMPLICIT, null,
SoftInputShowHideReason.SHOW_STATE_ALWAYS_VISIBLE);
} else {
Slog.e(TAG, "SOFT_INPUT_STATE_ALWAYS_VISIBLE is ignored because"
+ " there is no focused view that also returns true from"
+ " View#onCheckIsTextEditor()");
}
break;
}
可以看到windowSoftInputMode的六个stateXXX就是在这里处理的,当窗口的focus状态变成true的时候(无论是前进还是后退)都会走到这个里面。下面我们逐个分析:
当 Activity 转至前台时保留软键盘最后所处的任何状态,无论是可见还是隐藏。
case LayoutParams.SOFT_INPUT_STATE_UNCHANGED:
// Do nothing.
break;
可以看到这里的保留状态就是Do nothing,不做任何操作,之前是显示就是显示,隐藏就是隐藏。
当用户选择 Activity 时 — 也就是说,当用户确实是向前导航到 Activity,而不是因离开另一 Activity 而返回时 — 隐藏软键盘。
case LayoutParams.SOFT_INPUT_STATE_HIDDEN:
if ((softInputMode & LayoutParams.SOFT_INPUT_IS_FORWARD_NAVIGATION) != 0) {
if (DEBUG) Slog.v(TAG, "Window asks to hide input going forward");
hideCurrentInputLocked(mCurFocusedWindow, 0, null,
SoftInputShowHideReason.HIDE_STATE_HIDDEN_FORWARD_NAV);
}
break;
case LayoutParams.SOFT_INPUT_STATE_ALWAYS_HIDDEN:
if (DEBUG) Slog.v(TAG, "Window asks to hide input");
hideCurrentInputLocked(mCurFocusedWindow, 0, null,
SoftInputShowHideReason.HIDE_ALWAYS_HIDDEN_STATE);
break;
从代码上看stateHidden只比stateAlwaysHidden多了一个LayoutParams.SOFT_INPUT_IS_FORWARD_NAVIGATION的判断,LayoutParams.SOFT_INPUT_IS_FORWARD_NAVIGATION标识窗口是通过前进获取的焦点,简单来说就是跳转到一个新页面而非返回到当前页面,stateHidden只有在向前跳转的时候才会去调用hideCurrentInputLocked方法隐藏输入法,而stateAlwaysHidden则没有这个判断,任何情况下窗口获取到焦点都会去隐藏输入法。
当用户选择 Activity 时 — 也就是说,当用户确实是向前导航到 Activity,而不是因离开另一 Activity 而返回时 — 显示软键盘。
当 Activity 的主窗口有输入焦点时始终显示软键盘。
case LayoutParams.SOFT_INPUT_STATE_VISIBLE:
if ((softInputMode & LayoutParams.SOFT_INPUT_IS_FORWARD_NAVIGATION) != 0) {
if (DEBUG) Slog.v(TAG, "Window asks to show input going forward");
if (InputMethodUtils.isSoftInputModeStateVisibleAllowed(
unverifiedTargetSdkVersion, startInputFlags)) {
if (attribute != null) {
res = startInputUncheckedLocked(cs, inputContext, missingMethods,
attribute, startInputFlags, startInputReason);
didStart = true;
}
showCurrentInputLocked(windowToken, InputMethodManager.SHOW_IMPLICIT, null,
SoftInputShowHideReason.SHOW_STATE_VISIBLE_FORWARD_NAV);
} else {
Slog.e(TAG, "SOFT_INPUT_STATE_VISIBLE is ignored because"
+ " there is no focused view that also returns true from"
+ " View#onCheckIsTextEditor()");
}
}
break;
case LayoutParams.SOFT_INPUT_STATE_ALWAYS_VISIBLE:
if (DEBUG) Slog.v(TAG, "Window asks to always show input");
if (InputMethodUtils.isSoftInputModeStateVisibleAllowed(
unverifiedTargetSdkVersion, startInputFlags)) {
if (attribute != null) {
res = startInputUncheckedLocked(cs, inputContext, missingMethods,
attribute, startInputFlags, startInputReason);
didStart = true;
}
showCurrentInputLocked(windowToken, InputMethodManager.SHOW_IMPLICIT, null,
SoftInputShowHideReason.SHOW_STATE_ALWAYS_VISIBLE);
} else {
Slog.e(TAG, "SOFT_INPUT_STATE_ALWAYS_VISIBLE is ignored because"
+ " there is no focused view that also returns true from"
+ " View#onCheckIsTextEditor()");
}
break;
和隐藏一样,显示也是同样的套路,这里就不多说了。
在显示的时候会去调用isSoftInputModeStateVisibleAllowed方法判断允许显示,看一下这个方法:
static boolean isSoftInputModeStateVisibleAllowed(int targetSdkVersion,
@StartInputFlags int startInputFlags) {
if (targetSdkVersion < Build.VERSION_CODES.P) {
// for compatibility.
return true;
}
if ((startInputFlags & StartInputFlags.VIEW_HAS_FOCUS) == 0) {
return false;
}
if ((startInputFlags & StartInputFlags.IS_TEXT_EDITOR) == 0) {
return false;
}
return true;
}
在安卓9以下会直接返回true,安卓9以上就加了两个判断,这里就解释了为什么安卓9之后软键盘不能自动弹出了,具体可以看上一篇文章。
进入判断后会去判断输入法进程有没有开启,没有开启会去调用startInputUncheckedLocked开启,开启之后会去掉用showCurrentInputLocked显示软键盘。
不指定软键盘的状态(隐藏还是可见)。 将由系统选择合适的状态,或依赖主题中的设置。
系统自己去选择显示还是隐藏,这里就不介绍了,有兴趣的自己去研究一下。
case LayoutParams.SOFT_INPUT_STATE_UNSPECIFIED:
if (!isTextEditor || !doAutoShow) {
if (LayoutParams.mayUseInputMethod(windowFlags)) {
// There is no focus view, and this window will
// be behind any soft input window, so hide the
// soft input window if it is shown.
if (DEBUG) Slog.v(TAG, "Unspecified window will hide input");
hideCurrentInputLocked(
mCurFocusedWindow, InputMethodManager.HIDE_NOT_ALWAYS, null,
SoftInputShowHideReason.HIDE_UNSPECIFIED_WINDOW);
// If focused display changed, we should unbind current method
// to make app window in previous display relayout after Ime
// window token removed.
// Note that we can trust client's display ID as long as it matches
// to the display ID obtained from the window.
if (cs.selfReportedDisplayId != mCurTokenDisplayId) {
unbindCurrentMethodLocked();
}
}
} else if (isTextEditor && doAutoShow
&& (softInputMode & LayoutParams.SOFT_INPUT_IS_FORWARD_NAVIGATION) != 0) {
// There is a focus view, and we are navigating forward
// into the window, so show the input window for the user.
// We only do this automatically if the window can resize
// to accommodate the IME (so what the user sees will give
// them good context without input information being obscured
// by the IME) or if running on a large screen where there
// is more room for the target window + IME.
if (DEBUG) Slog.v(TAG, "Unspecified window will show input");
if (attribute != null) {
res = startInputUncheckedLocked(cs, inputContext, missingMethods,
attribute, startInputFlags, startInputReason);
didStart = true;
}
showCurrentInputLocked(windowToken, InputMethodManager.SHOW_IMPLICIT, null,
SoftInputShowHideReason.SHOW_AUTO_EDITOR_FORWARD_NAV);
}
break;
显示方式比较复杂,牵扯到很多方面,首先是在PhoneWindowManager(属于系统进程)的layoutWindowLw方法中,这个方法主要是设置窗口的一些位置大小参数,当软键盘弹出或收起的时候都会走到这里去重新设置窗口的大小,看一下我精简后的相关代码:
public void layoutWindowLw(WindowState win, WindowState attached, DisplayFrames displayFrames) {
final int type = attrs.type;
final int fl = PolicyControl.getWindowFlags(win, attrs);
final int pfl = attrs.privateFlags;
final int sim = attrs.softInputMode;
final int requestedSysUiFl = PolicyControl.getSystemUiVisibility(null, attrs);
final int sysUiFl = requestedSysUiFl | getImpliedSysUiFlagsForLayout(attrs);
final Rect pf = mTmpParentFrame;
final Rect df = mTmpDisplayFrame;
final Rect of = mTmpOverscanFrame;
final Rect cf = mTmpContentFrame;
final Rect vf = mTmpVisibleFrame;
final Rect dcf = mTmpDecorFrame;
final Rect sf = mTmpStableFrame;
Rect osf = null;
dcf.setEmpty();
final boolean hasNavBar = (isDefaultDisplay && mHasNavigationBar
&& mNavigationBar != null && mNavigationBar.isVisibleLw());
final int adjust = sim & SOFT_INPUT_MASK_ADJUST;
final boolean requestedFullscreen = (fl & FLAG_FULLSCREEN) != 0
|| (requestedSysUiFl & View.SYSTEM_UI_FLAG_FULLSCREEN) != 0;
final boolean layoutInScreen = (fl & FLAG_LAYOUT_IN_SCREEN) == FLAG_LAYOUT_IN_SCREEN;
final boolean layoutInsetDecor = (fl & FLAG_LAYOUT_INSET_DECOR) == FLAG_LAYOUT_INSET_DECOR;
sf.set(displayFrames.mStable);
if (adjust != SOFT_INPUT_ADJUST_RESIZE) {
cf.set(displayFrames.mDock);
} else {
cf.set(displayFrames.mContent);
}
applyStableConstraints(sysUiFl, fl, cf, displayFrames);
if (adjust != SOFT_INPUT_ADJUST_NOTHING) {
vf.set(displayFrames.mCurrent);
} else {
vf.set(cf);
}
}
首先方法里有很多Rect类型的参数被mTmpXXXFrame赋值,我们知道Rect是用来表示一个矩形区域,这里就是用来表示窗口尺寸的,我们需要关注Rect cf = mTmpContentFrame和Rect vf = mTmpVisibleFrame; 这两个参数,cf表示的是窗口内容区域的尺寸(可以理解为可供开发者使用的区域),vf表示的是窗口可见区域的尺寸;当输入法弹出或收起时根据设置的windowSoftInputMode不同这里的尺寸会发生变化,最终窗口大小的改变会回调到应用进程的ViewRootImpl中,最后到ViewRootHandler中的MSG_RESIZED事件里。
根据上面的代码我们发现当设置SOFT_INPUT_ADJUST_RESIZE时cf(内容区域)会被设置成displayFrames.mContent(展示区域),vf(可见区域)会被设置成displayFrames.mCurrent,注意这里有个地方容易混淆,cf和displayFrames.mContent表达的东西不一样,当软键盘弹出的时候我们可以理解为app窗口的展示区域变小了,但是内容区域没有变小,虽然被软键盘遮住了,但是内容区域也是可以使用的。
这里cf被设置成displayFrames.mContent就代表被软键盘遮住的地方应该是不能被使用的,最终这个cf会被回调到MSG_RESIZED中,我们来看一下代码:
public void handleMessage(Message msg) {
switch (msg.what) {
........
case MSG_RESIZED: {
// Recycled in the fall through...
SomeArgs args = (SomeArgs) msg.obj;
if (mWinFrame.equals(args.arg1)
&& mPendingOverscanInsets.equals(args.arg5)
&& mPendingContentInsets.equals(args.arg2)
&& mPendingStableInsets.equals(args.arg6)
&& mPendingDisplayCutout.get().equals(args.arg9)
&& mPendingVisibleInsets.equals(args.arg3)
&& mPendingOutsets.equals(args.arg7)
&& mPendingBackDropFrame.equals(args.arg8)
&& args.arg4 == null
&& args.argi1 == 0
&& mDisplay.getDisplayId() == args.argi3) {
break;
}
} // fall through...
case MSG_RESIZED_REPORT:
if (mAdded) {
SomeArgs args = (SomeArgs) msg.obj;
final int displayId = args.argi3;
MergedConfiguration mergedConfiguration = (MergedConfiguration) args.arg4;
final boolean displayChanged = mDisplay.getDisplayId() != displayId;
if (!mLastReportedMergedConfiguration.equals(mergedConfiguration)) {
// If configuration changed - notify about that and, maybe,
// about move to display.
performConfigurationChange(mergedConfiguration, false /* force */,
displayChanged
? displayId : INVALID_DISPLAY /* same display */);
} else if (displayChanged) {
// Moved to display without config change - report last applied one.
onMovedToDisplay(displayId, mLastConfigurationFromResources);
}
final boolean framesChanged = !mWinFrame.equals(args.arg1)
|| !mPendingOverscanInsets.equals(args.arg5)
|| !mPendingContentInsets.equals(args.arg2)
|| !mPendingStableInsets.equals(args.arg6)
|| !mPendingDisplayCutout.get().equals(args.arg9)
|| !mPendingVisibleInsets.equals(args.arg3)
|| !mPendingOutsets.equals(args.arg7);
mWinFrame.set((Rect) args.arg1);
mPendingOverscanInsets.set((Rect) args.arg5);
mPendingContentInsets.set((Rect) args.arg2);
mPendingStableInsets.set((Rect) args.arg6);
mPendingDisplayCutout.set((DisplayCutout) args.arg9);
mPendingVisibleInsets.set((Rect) args.arg3);
mPendingOutsets.set((Rect) args.arg7);
mPendingBackDropFrame.set((Rect) args.arg8);
mForceNextWindowRelayout = args.argi1 != 0;
mPendingAlwaysConsumeNavBar = args.argi2 != 0;
args.recycle();
if (msg.what == MSG_RESIZED_REPORT) {
reportNextDraw();
}
if (mView != null && framesChanged) {
forceLayout(mView);
}
requestLayout();
}
break;
.........
}
}
SomeArgs中的args.arg2就是我们之前设置的cf,
1.进入到MSG_RESIZED中后会判断SomeArgs中的参数是不是和之前一样,如果不同就会走到MSG_RESIZED_REPORT中,
2.这里将尺寸信息保存之后就会去调用forceLayout(mView)—>ViewTree里每个View/ViewGroup打上layout、draw标记,也就是说每个View/ViewGroup 最后都会执行三大流程,最后requestLayout()—>触发执行三大流程。
既然记录了尺寸的变化,继续跟踪这些值怎么使用。调用requestLayout()将会触发执行performTraversals()方法:
#ViewRootImpl.java
private void performTraversals() {
if (mFirst || windowShouldResize || insetsChanged ||
viewVisibilityChanged || params != null || mForceNextWindowRelayout) {
...
boolean hwInitialized = false;
//内容边界是否发生变化
boolean contentInsetsChanged = false;
try {
...
//内容区域变化----------->1
contentInsetsChanged = !mPendingContentInsets.equals(
mAttachInfo.mContentInsets);
if (contentInsetsChanged || mLastSystemUiVisibility !=
mAttachInfo.mSystemUiVisibility || mApplyInsetsRequested
|| mLastOverscanRequested != mAttachInfo.mOverscanRequested
|| outsetsChanged) {
...
//分发Inset----------->2
dispatchApplyInsets(host);
contentInsetsChanged = true;
}
...
} catch (RemoteException e) {
}
...
}
...
}
1.当设置SOFT_INPUT_ADJUST_RESIZE,键盘弹起时内容区域发生变化,因此会执行dispatchApplyInsets()。
2.分发Inset。
这些记录的值会存储在AttachInfo对应的变量里。
该方法调用栈如下:
dispatchApplyWindowInsets(WindowInsets insets)里的insets构成是通过计算之前记录在mPendingXX里的边界值。
最终调用fitSystemWindowsInt():
#View.java
private boolean fitSystemWindowsInt(Rect insets) {
//FITS_SYSTEM_WINDOWS 为xml里设置 android:fitsSystemWindows="true"
//对于DecorView的子布局LinearLayout来说,默认fitsSystemWindows=true
if ((mViewFlags & FITS_SYSTEM_WINDOWS) == FITS_SYSTEM_WINDOWS) {
...
//设置View的padding
internalSetPadding(localInsets.left, localInsets.top,
localInsets.right, localInsets.bottom);
return res;
}
return false;
}
protected void internalSetPadding(int left, int top, int right, int bottom) {
...
if (mPaddingLeft != left) {
changed = true;
mPaddingLeft = left;
}
if (mPaddingTop != top) {
changed = true;
mPaddingTop = top;
}
if (mPaddingRight != right) {
changed = true;
mPaddingRight = right;
}
if (mPaddingBottom != bottom) {
changed = true;
mPaddingBottom = bottom;
}
if (changed) {
requestLayout();
invalidateOutline();
}
}
看到这答案就呼之欲出了,DecorView的子布局LinearLayout设置padding,最终会影响LinearLayout子布局的高度,一层层传递下去,就会影响到Demo里的Activity 布局文件的高度。
小结
当设置SOFT_INPUT_ADJUST_RESIZE 时,DecorView的子布局padding会改变,最后影响子孙布局的高度。
当设置SOFT_INPUT_ADJUST_PAN时cf(内容区域)会被设置成displayFrames.mDock,vf(可见区域)同样被设置成displayFrames.mCurrent,displayFrames.mDock是什么呢?
/** During layout, the current screen borders along which input method windows are placed. */
public final Rect mDock = new Rect();
布局期间,输入法窗口所在的当前屏幕边框. 我理解其实就是除去导航栏和状态栏的app可使用内容区域,后来通过断点验证果然是的,SOFT_INPUT_ADJUST_RESIZE将cf设置成可见区域,而SOFT_INPUT_ADJUST_PAN这里cf被设置成内容区域,这也就意味着app可以使用被输入法挡住的部分。
最终同样会到MSG_RESIZED中:
SomeArgs中的args.arg2也就是cf的大小是没有变化的,但是args.arg3也就是vf变了,所以依然会进入MSG_RESIZED_REPORT中,也依然会去调用forceLayout(mView)和requestLayout()—>触发执行三大流程。
但是由于内容区域没有发生变化,这里不会再去执行dispatchApplyInsets()。那SOFT_INPUT_ADJUST_PAN的处理在哪里呢?
我们知道SOFT_INPUT_ADJUST_PAN的效果是页面整体向上移动,布局移动无非就是坐标发生改变,或者内容滚动了,不管是何种形式最终都需要通过对Canvas进行位移才能实现移动的效果。在performDraw->draw方法中有一个scrollToRectOrFocus方法:
#View.java
boolean scrollToRectOrFocus(Rect rectangle, boolean immediate) {
//窗口内容区域
final Rect ci = mAttachInfo.mContentInsets;
//窗口可见区域
final Rect vi = mAttachInfo.mVisibleInsets;
//滚动距离
int scrollY = 0;
boolean handled = false;
if (vi.left > ci.left || vi.top > ci.top
|| vi.right > ci.right || vi.bottom > ci.bottom) {
scrollY = mScrollY;
//找到当前有焦点的View------------>(1)
final View focus = mView.findFocus();
...
if (focus == lastScrolledFocus && !mScrollMayChange && rectangle == null) {
//焦点没有发生切换,不做操作
} else {
// We need to determine if the currently focused view is
// within the visible part of the window and, if not, apply
// a pan so it can be seen.
mLastScrolledFocus = new WeakReference<View>(focus);
mScrollMayChange = false;
if (DEBUG_INPUT_RESIZE) Log.v(mTag, "Need to scroll?");
// Try to find the rectangle from the focus view.
if (focus.getGlobalVisibleRect(mVisRect, null)) {
...
//找到当前焦点与可见区域的相交部分
//mVisRect 为当前焦点在Window里的可见部分
if (mTempRect.intersect(mVisRect)) {
if (mTempRect.height() >
(mView.getHeight()-vi.top-vi.bottom)) {
...
}
else if (mTempRect.top < vi.top) {
//如果当前焦点位置在窗口可见区域上边,说明焦点View应该往下移动到可见区域里边
scrollY = mTempRect.top - vi.top;
} else if (mTempRect.bottom > (mView.getHeight()-vi.bottom)) {
//如果当前焦点位置在窗口可见区域之下,说明其应该往上移动到可见区域里边------->(2)
scrollY = mTempRect.bottom - (mView.getHeight()-vi.bottom);
} else {
//无需滚动------->(3)
scrollY = 0;
}
handled = true;
}
}
}
}
if (scrollY != mScrollY) {
//滚动距离发生变化
if (!immediate) {
if (mScroller == null) {
mScroller = new Scroller(mView.getContext());
}
//开始设置滚动----------->(4)
mScroller.startScroll(0, mScrollY, 0, scrollY-mScrollY);
} else if (mScroller != null) {
mScroller.abortAnimation();
}
//赋值给成员变量
mScrollY = scrollY;
}
return handled;
}
1.对于上面的Demo来说,当前的焦点View就是EditText,点击哪个EditText,哪个就获得焦点。
2.对于输入框2来说,因为键盘弹出会遮住它,通过计算满足"当前焦点位置在窗口可见区域之下,说明其应该往上移动到可见区域里边" 条件,因此srolly > 0。
3.而对于输入框1来说,当键盘弹出时,它没有被键盘遮挡,走到else分支,因此scrollY = 0。
4.滚动是借助Scoller.java类完成的。
上面的操作实际上就是为了确认滚动值,并记录在成员变量mScrollY里,继续来看如何使用滚动值呢?
#ViewRootImpl.java
private boolean draw(boolean fullRedrawNeeded) {
...
boolean animating = mScroller != null && mScroller.computeScrollOffset();
final int curScrollY;
//获取当前需要滚动的scroll值
if (animating) {
curScrollY = mScroller.getCurrY();
} else {
curScrollY = mScrollY;
}
...
int xOffset = -mCanvasOffsetX;
//记录在yOffset里
int yOffset = -mCanvasOffsetY + curScrollY;
boolean useAsyncReport = false;
if (!dirty.isEmpty() || mIsAnimating || accessibilityFocusDirty) {
if (mAttachInfo.mThreadedRenderer != null && mAttachInfo.mThreadedRenderer.isEnabled()) {
...
//对于走硬件加速绘制
if (mHardwareYOffset != yOffset || mHardwareXOffset != xOffset) {
//记录偏移量到mHardwareYOffset里
mHardwareYOffset = yOffset;
mHardwareXOffset = xOffset;
invalidateRoot = true;
}
..
mAttachInfo.mThreadedRenderer.draw(mView, mAttachInfo, this);
} else {
//软件绘制
//传入yOffset
if (!drawSoftware(surface, mAttachInfo, xOffset, yOffset,
scalingRequired, dirty, surfaceInsets)) {
return false;
}
}
}
...
return useAsyncReport;
}
滚动值分别传递给了硬件加速绘制分支和软件绘制分支,在各自的分支里对Canvas进行平移,具体如何平移此处不细说了,大家可以继续跟踪代码分析。
小结
1、当设置SOFT_INPUT_ADJUST_PAN时,如果发现键盘遮住了当前有焦点的View,那么会对RootView(此处Demo里DecorView作为RootView)的Canvas进行平移,直至有焦点的View显示到可见区域为止。
2、这就是为什么点击输入框2的时候布局会整体向上移动的原因。
当设置成SOFT_INPUT_ADJUST_NOTHING时cf(内容区域)会被设置成displayFrames.mDock,vf(可见区域)会被设置成cf,其实这里就相当于窗口尺寸参数没有变化,就不会回调到MSG_RESIZED中,自然布局也不会有什么改变。
设置了SOFT_INPUT_ADJUST_UNSPECIFIED,其内部最终使用SOFT_INPUT_ADJUST_PAN和SOFT_INPUT_ADJUST_RESIZE之一进行展示。接下来就来探究选择的标准是什么。
softInputMode 没有设值的时候,默认是SOFT_INPUT_ADJUST_UNSPECIFIED模式。
还是从ViewRootImpl.java的performTraversals()方法开始分析:
private void performTraversals() {
...
if (mFirst || mAttachInfo.mViewVisibilityChanged) {
mAttachInfo.mViewVisibilityChanged = false;
//先查看有没有提前设置了模式
int resizeMode = mSoftInputMode &
WindowManager.LayoutParams.SOFT_INPUT_MASK_ADJUST;
//如果没有设置,那么默认为0,也就是SOFT_INPUT_ADJUST_UNSPECIFIED
if (resizeMode == WindowManager.LayoutParams.SOFT_INPUT_ADJUST_UNSPECIFIED) {
//查看mScrollContainers 数组有没有元素(View 作为元素) -------->(1)
final int N = mAttachInfo.mScrollContainers.size();
for (int i=0; i<N; i++) {
if (mAttachInfo.mScrollContainers.get(i).isShown()) {
//如果有元素,则设置为SOFT_INPUT_ADJUST_RESIZE 模式
resizeMode = WindowManager.LayoutParams.SOFT_INPUT_ADJUST_RESIZE;
}
}
if (resizeMode == 0) {
//如果没有设置为resize模式,则设置pan模式
resizeMode = WindowManager.LayoutParams.SOFT_INPUT_ADJUST_PAN;
}
if ((lp.softInputMode &
WindowManager.LayoutParams.SOFT_INPUT_MASK_ADJUST) != resizeMode) {
lp.softInputMode = (lp.softInputMode &
~WindowManager.LayoutParams.SOFT_INPUT_MASK_ADJUST) |
resizeMode;
//最后赋值给params,让Window属性生效
params = lp;
}
}
}
...
}
上一篇文章已经说过了,这里主要是通过mAttachInfo.mScrollContainers判断是否有可滚动的view,如果有就设置成SOFT_INPUT_ADJUST_RESIZE,如果没有就设置成SOFT_INPUT_ADJUST_PAN。