Tooltips (or other action) in QFileDialog - qt

I want to have a Tooltip (or ideally, a QWidget) pop-up when hovering over a file in a QFileDialog::getOpenFileName instance.
Is there a way to do this without subclassing the class?

I'm not sure there's any approved/definitive way of doing this. The following is a rather hacky way of achieving what (I think) you want but which makes certain assumptions regarding the widget hierarchy associated with a QFileDialog instance. Specifically, it relies on the assumption that the widget hierarchy subtended by a QFileDialog instance will contain one or more QAbstractItemView instances...
#include <iostream>
#include <QAbstractItemView>
#include <QApplication>
#include <QCursor>
#include <QFileDialog>
#include <QToolTip>
int
main (int argc, char **argv)
{
QApplication app(argc, argv);
QFileDialog fd;
/*
* Further to the comments by #Parisa.H.R, we need to make sure we use a
* non-native file dialog here otherwise there's no way to get the desired
* behaviour.
*/
fd.setOption(QFileDialog::DontUseNativeDialog);
/*
* Search the widget hierarchy under the QFileDialog looking for instances of
* QAbstractItemView or derived classes.
*/
for (auto *v: fd.findChildren<QAbstractItemView *>()) {
std::clog << "view = " << v << "(type=" << v->metaObject()->className()
<< ", name=\"" << v->objectName().toStdString() << "\")\n";
/*
* Connect the view's entered signal to a lambda which, for the time being,
* simply displays a tooltip showing the name of the filesystem item.
*/
QObject::connect(v, &QAbstractItemView::entered,
[](const QModelIndex &index)
{
QToolTip::showText(QCursor::pos(), index.data(Qt::DisplayRole).toString());
});
/*
* In order to receive the QAbstractItemView::entered signal mouse tracking
* must be enabled for the view.
*/
v->setMouseTracking(true);
}
fd.exec();
}

Related

How to deselect selected cell in QTableWidget by clicking on it again?

I am trying to deselect a selected cell in QTableWidget by clicking on it again. I don't know if I missed an option or a signal in documentation (I hope not). I tried with signals cellClicked and cellActivated. None of them work. Thing is, if a cell is in a deselected state and I click on it, by the time signal cellClicked is emitted, cell already has selection. So I cannot check for selection in a slot that reacts to that signal.
So how can I deselect a selected cell? Selection mode is SingleSelection. I just hope I don't have to subclass QTableWidget.
I've only performed minimal testing but the following appears to do what you want...
#include <iostream>
#include <QApplication>
#include <QTableWidget>
int main (int argc, char **argv)
{
QApplication app(argc, argv);
QTableWidget tw(5, 10);
/*
* Disable any normal selection mode.
*/
tw.setSelectionMode(QAbstractItemView::NoSelection);
QObject::connect(&tw, &QTableWidget::cellClicked,
[&tw](int row, int col)
{
/*
* Make sure we have an item in the cell.
*/
auto *item = tw.itemAt(row, col);
if (!item) {
item = new QTableWidgetItem(QString("cell[%1, %2]").arg(row).arg(col));
tw.setItem(row, col, item);
}
/*
* Update selection based on current state.
*/
bool was_selected = item->isSelected();
tw.selectionModel()->clear();
item->setSelected(!was_selected);
});
tw.show();
return app.exec();
}

How to continue text editing after formatting changes of QGraphicsTextItem?

I am trying to make changes (font changes) to a QGraphicsTextItem that is editable.
I am trying to change formatting of fragments of text, or the formatting applied at typing point (if I set text bold, the text I type after that action at cursor position would be bold).
Setting formatting for text fragments works - but I can't find a way to return the focus correctly to the item.
I can show the caret at the right position, but I can't type in the box unless I actually click in box (even though it seems hat I should be able to).
Simple sample (for some reason it crashes on closing program but I don't care about that since I am testing the text class, not the main program):
header: mytextitem.h
#include <QGraphicsTextItem>
class MyTextItem : public QGraphicsTextItem
{
Q_OBJECT
public:
MyTextItem();
~MyTextItem() {}
public slots:
void setItemBold(const int b);
};
mytextitem.cpp
#include "mytextitem.h"
#include <QTextCursor>
MyTextItem::MyTextItem()
{
setPlainText("ABCD");
setFont(QFont("Arial", 20));
setFlags(QGraphicsItem::ItemIsSelectable | QGraphicsItem::ItemIsMovable | QGraphicsItem::ItemIsFocusable);
setTextInteractionFlags(Qt::TextEditorInteraction);
}
void MyTextItem::setItemBold(const int b)
{
int _weight = (b != 0) ? QFont::Bold : QFont::Normal;
QTextCursor _cursor = textCursor();
//int p = _cursor.position(); // this won't help
QTextCharFormat _format = _cursor.charFormat();
_format.setFontWeight(_weight);
_cursor.setCharFormat(_format);
//_cursor.setPosition(p, QTextCursor::KeepAnchor); // makes no difference on allowing me to type, but I can make the cursor move
//_cursor.movePosition(QTextCursor::NoMove, QTextCursor::KeepAnchor, 0); // makes no difference but I just thought some action might
setTextCursor(_cursor);
setFocus(Qt::MouseFocusReason);
// grabKeyboard(); // does nothing
}
main.cpp
#include <QApplication>
#include <QGraphicsView>
#include <QGridLayout>
#include <QtWidgets>
#include <QCheckBox>
#include "mytextitem.h"
int main(int argc, char *argv[])
{
QApplication a(argc, argv);
QGraphicsScene scene(-20, -20, 150, 100);
QGraphicsView view(&scene);
QWidget widget;
QGridLayout layout(&widget);
layout.addWidget(&view, 0, 0);
QCheckBox bold("Bold");
layout.addWidget(&bold, 0, 1);
MyTextItem* item = new MyTextItem();
scene.addItem(item);
QObject::connect(&bold, SIGNAL(stateChanged(int)), item, SLOT(setItemBold(int)));
view.ensureVisible(scene.sceneRect());
widget.show();
return a.exec();
}
Editing the item is possible only if clicking in the box.
Assuming that I am already in the box (editing), and I push the "Bold" checkbox, I expect to be able to continue editing - type in the box - but even though I try to
set focus (which places the blinking text cursor in the box),
set position for cursor (I can move it, or select things... that works but I want to keep current position and selection)
grab keyboard - seems to do nothing
nothing seems to return me to the box so I continue typing (with the new font setting).
How can I get the QTextCursor or anything else to allow me to keep editing the text ?
You need to focus on QGraphicsView after format change. You can't focus on QGraphicsTextItem because it isn't QWidget.

How can I get my caret to show up on first double-click action on a QGraphicsTextItem?

I am struggling to get a QGraphicsTextItem to work as a user friendly object.
Since it is very hard to move while being editable, I start it as not editable, and make it editable on double-click. Then turn editing off on losing focus.
My problem is, the caret does not show up on first edit.
I have tried getting the position based on mouse position (as in this question that was trying to solve a different problem), or calling the QGraphicsTextItem::mouseDoubleClickEvent(event);
No matter what I try, the caret is invisible on first action - until I start typing (or if I focus out and back in) - even though it is at the correct location.
After typing, or unselecting and reselecting , the caret shows up in normal location every time.
I have tried to call the QTextCursor in the item constructor, setting its position at 0, made no difference.
What made a difference : one of the 2 situations (neither of which I can do though):
a) start item with Qt::TextEditorInteraction in constructor
b) start item with no moving/focus/selectable flags
I can't do either - because my default state of item must be movable, and that interferes with text editing (as explained at start).
I have tried to disable those flags during editing though... with no effect.
Here is a simple code to demonstrate the problem, I hope somebody can have an idea.
mytextitem.h
#ifndef TEXTITEM_H
#define TEXTITEM_H
#include <QGraphicsTextItem>
class MyTextItem : public QGraphicsTextItem
{
Q_OBJECT
public:
MyTextItem();
protected:
virtual void focusOutEvent (QFocusEvent * event);
virtual void mouseDoubleClickEvent(QGraphicsSceneMouseEvent * event);
};
#endif // TEXTITEM_H
mytextitem.cpp
#include "mytextitem.h"
#include <QTextCursor>
#include <QAbstractTextDocumentLayout>
#include <QGraphicsSceneMouseEvent>
#include <QFont>
MyTextItem::MyTextItem()
{
setHtml("ABCD");
setFont(QFont("Arial", 50));
setTextInteractionFlags(Qt::NoTextInteraction);
setFlags(QGraphicsItem::ItemIsSelectable | QGraphicsItem::ItemIsMovable | QGraphicsItem::ItemIsFocusable);
}
void MyTextItem::focusOutEvent(QFocusEvent *event)
{
Q_UNUSED(event);
setTextInteractionFlags(Qt::NoTextInteraction);
QTextCursor _cursor = textCursor();
_cursor.clearSelection();
setTextCursor(_cursor);
}
void MyTextItem::mouseDoubleClickEvent(QGraphicsSceneMouseEvent *event)
{
setTextInteractionFlags(Qt::TextEditorInteraction);
QGraphicsTextItem::mouseDoubleClickEvent(event); // or the version in linked question
}
main.cpp
int main(int argc, char *argv[])
{
QApplication app(argc, argv);
QGraphicsScene s;
QGraphicsView view(&s);
s.setSceneRect(-20, -100, 800, 600);
view.show();
MyTextItem* t = new MyTextItem();
s.addItem(t);
return app.exec();
}
I have also considered editing text - not sure if that would work but I think it would affect the undo stack which I will have to deal with soon....
How can I get my caret to show up on first double-click action on the text item ?
(As a user, not seeing a caret would make me uncertain if I can type... even though it works... I would not have confidence in the object if I do not have feedback of my action. That's why I care about this problem.)
I can't explain it... after trying EVERYTHING to get the caret to show, the solution was so simple:
I had to change the order of flags being set, in constructor.
The QGraphicsTextItem flag must be set AFTER setting the QGraphicsItem flags.
setFlags(QGraphicsItem::ItemIsSelectable | QGraphicsItem::ItemIsMovable | QGraphicsItem::ItemIsFocusable);
setTextInteractionFlags(Qt::NoTextInteraction);

QTextBrowser - how to identify image from mouse click position

I'm using a QTextBrowser to display rich text including a number of images, each of them specified with a HTML <img> tag and added as resources using QTextDocument::addResource().
What I'd like to be able to do is, in a context menu handler (i.e. with a mouse click position available), identify the image that the click was over. It's possible to tell whether the click is over an image, because cursorForPosition(event->pos()).block().text() returns a string starting with Unicode 0xFFFC. Unfortunately the same string is returned for every image in the view.
It's possible to get all of the formats in use with QTextDocument::allFormats(), identify which of those are image formats, and get their image resource name. Unfortunately there seems to be no way to get their actual display position or bounding rectangle.
From the documentation:
Inline images are represented by an object replacement character (0xFFFC in Unicode) which has an associated QTextImageFormat. The image format specifies a name with setName() that is used to locate the image.
You can use charFormat().toImageFormat().name() on the cursor to extract the image's URL. Below is a self-contained example. There are two noteworthy details:
The cursor will sometimes point one character prior to the image. Thus the workaround; it seems necessary for both Qt 4.8.5 and 5.1.1.
The pop-up menus should be shown asynchronously so as not to block the rest of the application. The example code provided in the documentation is a source of bad user experience and should be considered an evil abomination. All widgets can automatically delete themselves when they get closed, so the menus won't leak. A QPointer is used only to demonstrate this fact. It tracks the menu's lifetime and nulls itself when the menu deletes itself.
#include <QApplication>
#include <QTextBrowser>
#include <QImage>
#include <QPainter>
#include <QMenu>
#include <QContextMenuEvent>
#include <QTextBlock>
#include <QPointer>
#include <QDebug>
class Browser : public QTextBrowser
{
QPointer<QMenu> m_menu;
protected:
void contextMenuEvent(QContextMenuEvent *ev) {
Q_ASSERT(m_menu.isNull()); // make sure the menus aren't leaking
m_menu = createStandardContextMenu();
QTextCursor cur = cursorForPosition(ev->pos());
QTextCharFormat fmt = cur.charFormat();
qDebug() << "position in block" << cur.positionInBlock()
<< "object type" << cur.charFormat().objectType();
if (fmt.objectType() == QTextFormat::NoObject) {
// workaround, sometimes the cursor will point one object to the left of the image
cur.movePosition(QTextCursor::NextCharacter);
fmt = cur.charFormat();
}
if (fmt.isImageFormat()) {
QTextImageFormat ifmt = fmt.toImageFormat();
m_menu->addAction(QString("Image URL: %1").arg(ifmt.name()));
}
m_menu->move(ev->globalPos());
m_menu->setAttribute(Qt::WA_DeleteOnClose); // the menu won't leak
m_menu->show(); // show the menu asynchronously so as not to block the application
}
};
void addImage(QTextDocument * doc, const QString & url) {
QImage img(100, 100, QImage::Format_ARGB32_Premultiplied);
img.fill(Qt::white);
QPainter p(&img);
p.drawRect(0, 0, 99, 99);
p.drawText(img.rect(), url);
doc->addResource(QTextDocument::ImageResource, QUrl(url), img);
}
int main(int argc, char *argv[])
{
QApplication a(argc, argv);
QTextDocument doc;
Browser browser;
doc.setHtml("<img src=\"data://image1\"/><br/><img src=\"data://image2\"/>");
addImage(&doc, "data://image1");
addImage(&doc, "data://image2");
browser.show();
browser.setDocument(&doc);
return a.exec();
}

QWidget::geometry() vs. QWidget::frameGeometry()

Although Qt's docs indicate that these two functions are different (the first doesn't include the frame) no matter what widget I choose - including the main window of my application - someWidget->frameGeometry().height() always returns the same value as someWidget->geometry.height().
What am I missing here?
I think, you don't give enough time to widget to be painted. There is little example:
#include <QApplication>
#include <QMainWindow>
#include <QDebug>
class MainWindow : public QMainWindow
{
public:
MainWindow() {
startTimer(500);
}
void timerEvent(QTimerEvent *e) {
// Here values are different
qDebug() << geometry().height() << frameGeometry().height();
}
};
int main(int argc, char *argv[])
{
QApplication app(argc, argv);
MainWindow mainWin;
mainWin.show();
// Here values are equals
qDebug() << mainWin.geometry().height() << mainWin.frameGeometry().height();
return app.exec();
}
First debug output will produce the same values for geometry and frameGeometry, but the second (in timerEvent) will produce different.
The QWidget class cannot have a frame. For example, QWidget doesn't have a frame, but QFrame has a frame.
if QWidget is toplevel window then you can see borders and title bar around it. We call it frame or decoration frame and frameGeometry() returns exactly that: window size and position which includes OS decorations.On the other side geometry() returs QWidget inner rect which is available for other child controls or painting.See http://doc.qt.io/qt-4.8/application-windows.html#window-geometry for more details.Toplevel geometry() / frameGeometry() differs if our window is not frameless or fullscreen ... or we are talking about some frameless window manager under x11.
this is an old post, but that could help those searching for the same problem.
Just call
adjustSize();
before prompting for some geometry attributes
As user fasked notes, frameGeometry() may not include the frame margins early in the window creation lifecycle. I have found that the following code works in some situations where calling frameGeometry() does not.
QMargins frameMargins;
QWindow *window = widget->windowHandle();
if (window) {
window->create();
frameMargins = window->frameMargins();
}
QRect myFrameGeometry = widget->geometry().adjusted(
-frameMargins.left(), -frameMargins.top(),
frameMargins.right(), frameMargins.bottom());

Resources