输入法是dialog，还是全屏dialog，且不会被dismiss，显示与隐藏系统调用方法为showWindow(true)和hide()，以下是输入法Window创建，SoftInputWindow继承Dialog，这个不必多说，
mWindow = new SoftInputWindow(this, "InputMethod", mTheme, null, null, mDispatcherState, WindowManager.LayoutParams.TYPE_INPUT_METHOD, Gravity.BOTTOM, false); mWindow.getWindow().getAttributes().setFitInsetsTypes(statusBars() | navigationBars()); mWindow.getWindow().getAttributes().setFitInsetsSides(Side.all() & ~Side.BOTTOM); mWindow.getWindow().getAttributes().setFitInsetsIgnoringVisibility(true);
至于全屏的依据，可以看SoftInputWindow initDockWindow方法，
看，写入了WindowManager.LayoutParams.FLAG_LAYOUT_IN_SCREEN;
这个Flag啥意思呢？就是将窗口放置在整个屏幕内，忽略父容器的限制，
private void initDockWindow() {
WindowManager.LayoutParams lp = getWindow().getAttributes();
lp.type = mWindowType;
lp.setTitle(mName);
lp.gravity = mGravity;
updateWidthHeight(lp);
getWindow().setAttributes(lp);
int windowSetFlags = WindowManager.LayoutParams.FLAG_LAYOUT_IN_SCREEN;
int windowModFlags = WindowManager.LayoutParams.FLAG_LAYOUT_IN_SCREEN |
WindowManager.LayoutParams.FLAG_NOT_FOCUSABLE |
WindowManager.LayoutParams.FLAG_DIM_BEHIND;
if (!mTakesFocus) {
windowSetFlags |= WindowManager.LayoutParams.FLAG_NOT_FOCUSABLE;
} else {
windowSetFlags |= WindowManager.LayoutParams.FLAG_NOT_TOUCH_MODAL;
windowModFlags |= WindowManager.LayoutParams.FLAG_NOT_TOUCH_MODAL;
}
getWindow().setFlags(windowSetFlags, windowModFlags);
}
如果要输入法上移，必须调整dialog中的布局，
在InputMethodService方法中，initView如下，
mWindow.setContentView(mRootView)，关键，
mWindow是SoftInputWindow，继承Dialog，因此这是设置视角；
如何，一些成员的初始化在此进行，如mInputframe和mCandidatesframe，这是啥呢？在setInputView中回调onCreateView方法，拿到三方输入法自定义的View，
所以显示在我们面前的视野就是mInputframe；
void initViews() {
mInitialized = false;
mViewsCreated = false;
mShowInputRequested = false;
mShowInputFlags =
0;
mThemeAttrs = obtainStyledAttributes(android.R.styleable.InputMethodService);
mRootView = mInflater.inflate(
com.android.internal.R.layout.input_method, null);
mWindow.setContentView(mRootView);
mRootView.getViewTreeObserver().removeonComputeInternalInsetsListener(mInsetsComputer);
mRootView.getViewTreeObserver().addonComputeInternalInsetsListener(mInsetsComputer);
if (Settings.Global.getInt(getContentResolver(),
Settings.Global.FANCY_IME_ANIMATIONS, 0) != 0) {
mWindow.getWindow().setWindowAnimations(
com.android.internal.R.style.Animation_InputMethodFancy);
}
mFullscreenArea = mRootView.findViewById(com.android.internal.R.id.fullscreenArea);
mExtractViewHidden = false;
mExtractframe = mRootView.findViewById(android.R.id.extractArea);
mExtractView = null;
mExtractEditText = null;
mExtractAccessories = null;
mExtractAction = null;
mFullscreenApplied = false;
mCandidatesframe = mRootView.findViewById(android.R.id.candidatesArea);
mInputframe = mRootView.findViewById(android.R.id.inputArea);
mInputView = null;
mIsInputViewShown = false;
mExtractframe.setVisibility(View.GONE);
mCandidatesVisibility = getCandidatesHiddenVisibility();
mCandidatesframe.setVisibility(mCandidatesVisibility);
mInputframe.setVisibility(View.GONE);
}
因为要上移输入法，尝试对window的attribute参数下手，失败告终，因为设置y参数对全屏输入法无效，由于service的onCreate方法new SoftInputWindow时，传入Gravity和Bottom，输入法就默认显示底部？调试时改为Top，依然显示在底部，比较费解。
FullscreenArea = mRootView.findViewById(com.android.internal.R.id.fullscreenArea);具体资源看不到，这个应该覆盖了整个屏幕
综上所述，输入法类似以下布局，暂不讨论屏幕旋转之内，这个在SoftInputWindow会更新Gravity，
处理上移问题，就要上移mInputframe，
怎么移？mInputframe本身是framelayout，而且位置在android资源文件中写死了，通过代码动态移动？
想过WindowManager的updateLayout，可是布局参数呢？
或者直接调用setY，可问题是这样的效果类似动画，原View并没有移动，也就导致点击事件不可用，
或者view的startMoveTask方法？感觉都不行，
一般三方自定义输入法时，需要继承InputMethodService类，这是个标准，
重写其中onCreateVIew方法，也就传入自定义输入法布局，系统通过如下方法回调，
public void updateInputViewShown() {
boolean isShown = mShowInputRequested && onevaluateInputViewShown();
if (mIsInputViewShown != isShown && mDecorViewVisible) {
mIsInputViewShown = isShown;
mInputframe.setVisibility(isShown ? View.VISIBLE : View.GONE);
if (mInputView == null) {
initialize();
View v = onCreateInputView();
if (v != null) {
setInputView(v);
}
}
}
}
之后，通过setInputView方法设置显示在用户面前的视野，可以看到，视野添加到mInputframe中，对于另外一个方法setCandidatesView，就是候选视图，暂不多说，候选嘛。
public void setInputView(View view) {
mInputframe.removeAllViews();
mInputframe.addView(view, new frameLayout.LayoutParams(
ViewGroup.LayoutParams.MATCH_PARENT,
ViewGroup.LayoutParams.WRAP_CONTENT));
mInputView = view;
}
所以我们要挪动输入法，可以在这里做文章，即添加一点参数，让mInputView上移一点，
frameLayout.LayoutParams提供了bottomMargin参数，即让咱们的mInputView相对bottom移动一点距离，单位是像素不是dp,
-------------------------------------------------------------------------------------------------------------------------
接下来，分析下输入法调用栈，
首先，窗口获得焦距后，才会调用输入法，这是起点，位于ViewRootImpl.handleWindowFocusChanged方法，
跟进其中，
// Note: must be done after the focus change callbacks,
// so all of the view state is set up correctly.
mImeFocusController.onPostWindowFocus(mView.findFocus(), hasWindowFocus,
mWindowAttributes);
跟进，
immDelegate.startInputAsynconWindowFocusGain(viewForWindowFocus,
windowAttribute.softInputMode, windowAttribute.flags, forceFocus);
跟进，
mService.startInputOrWindowGainedFocus(
startInputReason, mClient,
focusedView.getWindowToken(), startInputFlags, softInputMode,
windowFlags,
null,
null,
0 ,
mCurRootView.mContext.getApplicationInfo().targetSdkVersion);
跟进，
result = startInputOrWindowGainedFocusInternalLocked(startInputReason, client,
windowToken, startInputFlags, softInputMode, windowFlags, attribute,
inputContext, missingMethods, unverifiedTargetSdkVersion, userId);
跟进，
showCurrentInputLocked(windowToken, InputMethodManager.SHOW_IMPLICIT, null,
SoftInputShowHideReason.SHOW_AUTO_EDITOR_FORWARD_NAV);
发Message到IInputMethodWrapper，
case DO_SHOW_SOFT_INPUT: {
final SomeArgs args = (SomeArgs)msg.obj;
inputMethod.showSoftInputWithToken(
msg.arg1, (ResultReceiver) args.arg2, (IBinder) args.arg1);
args.recycle();
return;
}
跟进，
public void showSoftInputWithToken(int flags, ResultReceiver resultReceiver,
IBinder showInputToken) {
mSystemCallingShowSoftInput = true;
mCurShowInputToken = showInputToken;
showSoftInput(flags, resultReceiver);
mCurShowInputToken = null;
mSystemCallingShowSoftInput = false;
}
跟进showSoftInput，
try {
showWindow(true);
} catch (BadTokenException e) {
跟进showWinodw，
先准备InputVIew，也就是在onCreateInputVIew中返回的View
startViews(prepareWindow(showInput));
跟进prepareWindow，
private boolean prepareWindow(boolean showInput) {
boolean doShowInput = false;
mDecorViewVisible = true;
if (!mShowInputRequested && mInputStarted && showInput) {
doShowInput = true;
mShowInputRequested = true;
}
if (DEBUG) Log.v(TAG, "showWindow: updating UI");
initialize();
updateFullscreenMode();
updateInputViewShown();
if (!mViewsCreated) {
mViewsCreated = true;
initialize();
if (DEBUG) Log.v(TAG, "CALL: onCreateCandidatesView");
View v = onCreateCandidatesView();
if (DEBUG) Log.v(TAG, "showWindow: candidates=" + v);
if (v != null) {
setCandidatesView(v);
}
}
return doShowInput;
}
跟进updateInputViewShow，
public void updateInputViewShown() {
boolean isShown = mShowInputRequested && onevaluateInputViewShown();
if (mIsInputViewShown != isShown && mDecorViewVisible) {
mIsInputViewShown = isShown;
mInputframe.setVisibility(isShown ? View.VISIBLE : View.GONE);
if (mInputView == null) {
initialize();
View v = onCreateInputView();
if (v != null) {
setInputView(v);
}
}
}
}
ok，这里回调了onCreateInputVIew，接下来就是dialog的show过程，不在此赘述；
欢迎分享，转载请注明来源：内存溢出
微信扫一扫
支付宝扫一扫
评论列表（0条）