I've a qt 5.5 application working on one part of a monitor displaying it's user interface. A different application does this for the rest of the screen. My question is: does a practical solution exists to display a transparent widget above the second applications screen space to have complete input focus for the entire screen and forward all input actions relevant to second application? (which will be rendered always behind that widget).
The second application is not a qt but WinApi application. Is it possible to filter input events (mouse, keyboard, etc) an send a subset to a different process in this constellation? Or when making this application a qt application?
Thank you!
Related
Following is a dummy implementation of our web application
https://roleapplication.herokuapp.com/index.html
appArea element has role application as it contains highly complex widgets such as ms paint/editor/ms office.
Navigator contains standard web widgets such as dropdown and buttons
The HTML is something similar to as specified below.
<body>
<div class="appArea" role="application">
.......//Complex widgets
</div>
<div class="toolbar">
......//Buttons, dropdowns
</div>
</body>
Keyboard functionality of appArea is handled by its code and for toolbar we rely on keyboard handling with the screen reader as they work in web browser.
Issue - When user press escape in navigator area we blur the navigator so the focus by default goes to body.
Now as focus is in body then arrow keys moves the focus to toolbar and therefore user is never able to go into appArea. If focus is in appArea it works fine.
Expectation - When focus is on body then on pressing down arrow focus should inside the appArea and then appArea will get the key instead of screen reader.
Check the down arrow key functionality when page is loaded with and without screen reader.
Keyboard notes
Press f6 to go from widget 1 to widget 2 to navigator
You can use arrow/tab keys in widgets to navigate.
Move to navigator using f6 and press tab to go to any button and then press escape. Now focus is on body(check using
document.activeElement).
Without screen reader our widgets captures the key on body and process it even if they dont have focus.
However with screen reader, when body has focus and user press down arrow, screen reader consumes the key and move the focus to navigator instead of application area which has widgets and user is unable to go to appArea using arrow keys or other keys which screen reader consume.
Note -
If we give role application to complete application then default arrow key handling of navigator will stop working which is not desired
Removal of role application is not possible as appArea is quite complex with hundreds of widgets all having their keyboard handling.
There are three ways to interact with role="application".
Hit enter on the application element, exit out of edit mode (or forms mode) and use the application as if it is another web page. You can put other elements there and the screen reader will move through those elements in brows mode.
Hit enter on the application which pops the screen reader into edit mode where all keys are passed to the edit widget inside the application. and you handle everything within your application, probably on a keydown event.
Control the tabindex as the screen reader presses keys using a roving tabindex.
You currently have 1 and 3 which is really confusing. If you removed the application element, it would still work just fine. It sounds as if you want 2 though. 2 is highly discouraged unless you have a screen reader user constantly testing UX or building your app. Number 2 is mostly for games and is considered the "canvas" element for screen readers.
You do 2 by doing the following:
<div role="application">
<input type="button" autoFocus="true" value="Click me" />
<p aria-live="polite" id="spk"></p>
</div>
The spk element is to send messages to the screen reader which you need to do in this Window, Icon, Menu, Message (WIMM) interface. Remember that in this mode, you need to program everything and users get upset if expectations are not met.
You said you are making a word processor. This last option (number 2), is NOT meant to make a word processor. As a screen reader user, I have expectations and workflows for Word processors. You can't get that functionality with programming it manually in Javascript.
Instead, use the existing edit fields HTML provides for this reason, such as:
This text editor example
Please let me know if there is some reason why you would not want to use the above widget.
You could get away with using 3 along with normal widgets, but it is better to do what Google Drive does and allow users to enter edit mode when the page loads, or press a key, like escape, to enter the tabindex application area (which does not need to be in an application element, although it can be).
Edit: After reading your question again, it sounds as if you can't figure out how to enter the application element. You arrow to where the screen reader says "application" and hit enter. To get out, you either tab to the next tabindex element that is outside the application or press the special key command to exit out of the application. In NVDA, this key command is ctrl+nvda+space. On your application, the application element is the first element.
role='application' should be used on rare occasions. As you noted, it causes all keyboard events to skip the screen reader and go directly to your app. This causes the screen reader virtual cursor to not work. Typically, a screen reader will automatically go into "application" mode (often called "forms mode") for certain types of widgets, such as an input field. If you are using widget roles, you will get this "forms mode" for free.
When you say "arrow keys" are not working, are you talking about up/down arrows or left/right arrows? They have different behaviors for a screen reader.
I have inherited a Qt-based app that handles the master/detail relationship by presenting the detail screens as separate windows. The main window includes a list, and when you tap on a row a separate detail window is opened up.
In the code base, the detail windows are handled by a QML file and a matching .cpp file (the main window also has its own .cpp file). The problem I am facing is that a new client wants me to modify this application for them, except that they want everything to occur within a single window. They want the list to be shown on the left side, and then when a user taps a row, the detail screen is to be shown on the right side of the window in its own panel (but not in a separate window).
For various reasons I can't easily refactor this application. A quicker solution for me would be to continue to present the detail screen in its own window, but to make it a borderless window and position this borderless window over top of the main window (on the right) so that it appears to be a panel within the main window.
Is something like this possible with Qt? I have written Windows apps in the past that hooked into the Windows API to do something like, but I don't know whether this is even possible in a native Mac OS app, so I don't know whether Qt can handle it in some way automaticaly.
One thing you could try is to create a widget based "main" window and then use QWidget::createWindowContainer() to wrap the QtQuick windows for positioning them with QtWidget means, e.g. layouts.
I am working in a project which uses the Facebook graph-api to log in. I have the requirement of only using a virtual keyboard (no hardware will be present). I have looked everywhere, but can't find a solution for adding a virtual qwerty keyboard to the popUp.
I can put the keyboard into a popup, or I could add the qwerty keyboard into the screen with the addChild() method, but I still have one problem: the virtual keyboard does not focus to the textInputs of the popup and when i press a key, everything "explooota".
Anyone knows how i could solve the focus problem?
I mean... when i prees the virtual key, i call a java function wich simulate a physical keyboard, but i lose the focus into the facebook input text and the letter is not in the textinput... and i dont know how to recover the focus...
Thanks in advance for the help!
We had the same problem with a desktop app written in C#. I can only answer for a windows based application. Assuming you are working on a desktop app and that you are showing the login in a web browser control you can use the SendInput API to direct keyboard-like input to a field in the browser. We had our own custom keyboard; I don't think you will be able to use the built-in on-screen keyboard MS provides.
We had a windows form that hosted a web browser control and the keyboard custom control. The user touches the field that they want to fill in. The user types their input using the on-screen keyboard, the keyboard uses SendInput to send the appropriate character for the key that was touched to the web browser control. Other problems to look out for:
the facebook login form takes a lot of space, having both the keyboard and login visible at the same time is difficult
sending non-ascii characters; see this for help (SendInput sequence to create unicode character fails)
the user will have to touch to select the input field
there are other links on the FB login page you may want to restrict (like create an account)
an on-screen keyboard where touching the key doesn't steal focus from the browser field
These can all be solved but they are not trivial.
I use the last Qt version for a projet and QProcess. I want to lauch program from my application by using QProcess. I want to display a QGraphicsView transparent on full screen over the launched program.
For the moment: I hide the view, launch the program, sleep during 5 seconds and show the view. I want that my view keep the focus and stay on the top level? Is there any better way to do that? A custom setting for the QGraphicsView?
Create your QGraphicsView (or the window that contains it) with the Qt::WindowStaysOnTopHint flag
Once you run a program in QProcess, you have limited control over it. Qt does not provide details about other applications that are running, you won't know where the launched application is being displayed unless it tells you explicitly.
If you have access to the code of the application you're running, it is possible put a transparent overlay on top a given widget, or widgets, that could then record mouse clicks and other interactions. It's also possible to override events and record basic information about the application's use.
I've got a Maemo (Qt) app that does some integration with the built-in media player via D-Bus. All the control functionality I need is complete, but I've got a requirement to show my application window (which gets backgrounded when playback starts) instead of the media player when the playback window is closed (it's a stacked window).
It should go like this: user clicks item in my Qt application, which launches the media file in the native media player. User watches media file, exits by clicking the arrow on the playback window. I'd like to somehow catch this event and bring my application to the front instead of showing the media player's main window.
Is it even possible on Maemo? I'm thinking that some low-level X coding might be required.
Answer was painfully obvious, I can catch a state_changed signal from D-Bus- state=0 when the window is closed.
You can also use the raise() method of Qt windows.