Table of Contents

The concept of signals and slots

Connecting with signals and slots

Disconnecting

A parser-formatter using signals and slots

Conclusion

Both the rate and the setRate methods can be connected to, since any Python callable can be used as a slot. If the rate is changed, we update the private rate value and emit a custom rateChanged signal, giving the new rate as a parameter. We have also used the faster short-circuit syntax. Events and signals, on the other hand, are central to PyQt. Signals and slots are used to connect one object to another. An example is the perennial pushbutton, whose clicked signal gets connected to the accept slot function of a dialog box. Signals are used to connect entities internal to the application.

The concept of signals and slots is possibly the most interesting innovation in the Qt library. Good widgets and a clean API are rare, but not unique. But until Qt appeared on the horizon, connecting your widgets with your application and your data was a nasty and error-prone endeavor — even in Python. I will first discuss the problem that is solved by signals and slots in some detail; then I will introduce the actual mechanics of the signal/slot mechanism, and finish with an application of the technique outside the GUI domain.

The problem in a nutshell: imagine you have an application, and the application shows a button on screen. Whenever the button is pressed, you want a function in your application to execute. Of course, you'd prefer that the button doesn't know much about the application, or you would have to write a different button for each application. In the next example, the button has been coded to work only with an application that has the doSomeApplicationSpecificFunction function.

Example 7-1. A stupid button which is not reusable

Callbacks

This is no solution— the button code isn't reusable at all. A better solution would be to pass the function object to the button. Remember that in Python functions are objects just like everything else. In a language like C or C++ you would pass a function pointer, the actual memory address of the function being called. This is quite nasty, because there is no way the compiler can check what arguments are passed to the function represented by the function pointer. In Python, passing functions around is really easy.

Example 7-2. A simple callback system

How To Use Pyqt Signals And Slots Games

Using apply() to execute function objects.: Note the usage of the apply() function in the clicked() function — this Python built-in function executes the function object you pass as the first argument argument. You can also hand it parameters, as a tuple in the second argument to apply(). You'll see that idiom quite often when we subclass Qt classes:

This is useful because QWidget and the other Qt classes often have a lot of optional parameters, such as the object name or certain widget flags. If we discount the possibility that someone wants to use those optional parameters, we would write:

This is far less flexible. In the previous example, we created an argument tuple to be passed to the __init__() by first creating a tuple containing our own object reference - self, and then adding the arguments from the variable positional argument list to that tuple. Remember from the discussion of positional arguments in the Section called Methods and functions in Chapter 4 that the arguments in *args are a tuple, and you can create a new tuple by adding two tuples.

In more recent versions of Python, you don't need to use apply() anymore to call the constructor of a superclass with a variable number of arguments. That is, from version 2.0 of Python you can also use the following construction:

That is, when calling the constructor of the superclass, you can pass self as the first argument, and then the argument list, with asterisks and all.

Action registry

Unfortunately, this callback system is not quite generic enough. For example, what if you wanted to activate two functions when the button is pressed? While this is not likely in the simple example, under more complex situations it often occurs. Think of a text editor where editing the text should change the internal representation of the document, the word count in the statusbar, and the edited-indicator in the titlebar. You wouldn't want to put all this functionality in one function, but it is a natural fit for signals and slots. You could have one signal, textChanged, that is connected to three functions: changeText(), setWordCount(), setEdited().

Wouldn't it be extremely comfortable to simply have a central registry where interested parties could come together? Something like:

Example 7-3. A central registry of connected widgets

The actual registry is a Python dictionary with the name connections. Here, each occasion is used as a key to find the actual function object that should be called.

If the ‘occasion' is already registered, we simply add a new entry to the list; otherwise a new entry is created in the registry.

If the ‘occasion' exists, then we remove the relevant function entry from its list of functions.

We loop over all functions that belong to this ‘occasion' and simply execute them by calling apply() on them.

A registry is a unique object to an application: there should only be one, so we create it globally.

This is the ‘button' class. Whenever the button is ‘clicked', it calls the execute() function in the registry with the ‘clicked' occasion.

The application creates one button and binds two of its functions to the button. This looks a lot like the way connections are made in Qt!

Here we simulate a button click by directly calling the clicked() function on the button.

This is one step up from the previous example, which was an extremely crude implementation of the well known Observer design pattern, in that there is now a ‘neutral' object that mediates between the button and the application. However, it is still not particularly sophisticated. It certainly wouldn't do for a real application — where there might be many objects with the same ‘occasion'.

Pyqt Signals And Slots

It is quite possible to implement a solution like this in pure Python, especially with the weak references module that debuted in Python 2.1. Bernhard Herzog has done so in his fine Python application Sketch (http://sketch.sourceforge.net). He had to do it himself — because he was working in PyGTK, not PyQt. Fortunately, PyQt has already solved the whole problem for us.

Signals and slots

We've just outlined the problem which the developers of Qt at Trolltech have solved in a unique and flexible manner. They created the concept of signals and slots. signals are sent by an object that wants to tell the world something interesting has happened, and by connecting the 'signals to the slots', those signals arrive at the slots of the objects that are interested.

On the whole the concept is really neat and clean and the implementation well-executed. What's more, the concept is even better suited to Python than it is to C++. If you want to use signals and slots in C++, you have to work with a preprocessor, called moc, and indicate with special macros which function can be called as a slot, and which function is a signal. All that is completely unnecessary in Python, where a signal is a string, and any function can be a slot.

Signals and slots are not magic, of course. As with our simple Python registry, there has to be a registry of objects that are interested in signals. This registry is updated by the connect and disconnect functions; both are member functions of the QObject class. The registry, as far as Python signals and slots is concerned, is kept by the sip library. Signals and slots that are defined in the underlying C++ library are maintained by the QObject class itself.

In a nutshell, signals and slots are the solution Qt provides for situations in which you want two objects to interact, while keeping that fact hidden from them.

Signals, messages, events: This is one area where there is a perfect Babel of tongues. Even really knowledgeable people like Dr Dobbs' Al Stevens get confused when confronted with terms like ‘message', ‘event' or ‘signal'.

In PyQt programming, the term '‘message' is quite irrelevant — it is used in Windows programming to indicate function calls made from your application to the Windows GUI libraries.

Events and signals, on the other hand, are central to PyQt. Signals and slots are used to connect one object to another. An example is the perennial pushbutton, whose clicked() signal gets connected to the accept() slot function of a dialog box. Signals are used to connect entities internal to the application.

How To Use Pyqt Signals And Slots Using

Events are more often generated directly by user input, such as moving or clicking with the mouse, or typing at the keyboard. As such, they don't connect two class instances, but rather a physical object, such as a keyboard, with an application. Events are encapsulated by the QEvent class, and are mostly delivered to QWidget and its descendants. Events are used to communication with external entities.

How To Use Pyqt Signals And Slots Free