PyQt v4 - Python Bindings for Qt v4

Reference Guide

Contact: info@riverbankcomputing.com
Version: 4.8.2
Copyright: Copyright (c) 2010 Riverbank Computing Limited

Contents

1   Introduction

This is the reference guide for PyQt 4.8.2. PyQt v4 is a set of Python bindings for v4 of the Qt application framework from Nokia.

There is a separate PyQt API Reference.

Qt is a set of C++ libraries and development tools that includes platform independent abstractions for graphical user interfaces, networking, threads, Unicode, regular expressions, SQL databases, SVG, OpenGL, XML, and user and application settings. PyQt implements 440 of these classes as a set of Python modules.

PyQt supports the Windows, Linux, UNIX and MacOS/X platforms.

PyQt does not include Qt itself - you must obtain it separately.

The homepage for PyQt is http://www.riverbankcomputing.com/software/pyqt/. Here you will always find the latest stable version, current development snapshots, and the latest version of this documentation.

PyQt is built using the SIP bindings generator. SIP must be installed in order to build and use PyQt.

Earlier versions of Qt are supported by PyQt v3.

1.1   License

PyQt is licensed on all platforms under a commercial license, the GPL v2 and the GPL v3. Your PyQt license must be compatible with your Qt license. If you use the GPL versions then your own code must also use a compatible license.

PyQt, unlike Qt, is not available under the LGPL.

You can purchase a commercial PyQt license here.

1.2   PyQt Components

PyQt comprises a number of different components. First of all there are a number of Python extension modules. These are all installed in the PyQt4 Python package.

  • The QtCore module. This contains the core non-GUI classes, including the event loop and Qt's signal and slot mechanism. It also includes platform independent abstractions for Unicode, threads, mapped files, shared memory, regular expressions, and user and application settings.
  • The QtGui module. This contains the majority of the GUI classes.
  • The QtHelp module. This contains classes for creating and viewing searchable documentation.
  • The QtNetwork module. This module contains classes for writing UDP and TCP clients and servers. It includes classes that implement FTP and HTTP clients and support DNS lookups.
  • The QtOpenGL module. This module contains classes that enable the use of OpenGL in rendering 3D graphics in PyQt applications.
  • The QtScript module. This module contains classes that enable PyQt applications to be scripted using Qt's JavaScript interpreter.
  • The QtScriptTools module. This module contains classes that contain additional components (e.g. a debugger) that are used with Qt's JavaScript interpreter.
  • The QtSql module. This module contains classes that integrate with SQL databases. It includes editable data models for database tables that can be used with GUI classes. It also includes an implementation of SQLite.
  • The QtSvg module. This module contains classes for displaying the contents of SVG files.
  • The QtTest module. This module contains functions that enable unit testing of PyQt applications. (PyQt does not implement the complete Qt unit test framework. Instead it assumes that the standard Python unit test framework will be used and implements those functions that simulate a user interacting with a GUI.)
  • The QtWebKit module. This module implements a web browser engine based on the WebKit open source browser engine.
  • The QtXml module. This module contains classes that implement SAX and DOM interfaces to Qt's XML parser.
  • The QtXmlPatterns module. This module contains classes that implement XQuery and XPath support for XML and custom data models.
  • The phonon module. This module contains classes that implement a cross-platform multimedia framework that enables the use of audio and video content in PyQt applications.
  • The QtDeclarative module. This module provides a declarative framework for building highly dynamic, custom user interfaces using QML.
  • The QtMultimedia module. This module provides low-level multimedia functionality. Application developers would normally use the phonon module.
  • The QtAssistant module. This module contains classes that allow Qt Assistant to be integrated with a PyQt application to provide online help. This module is not available with Qt v4.7 and later.
  • The QtDesigner module. This module contains classes that allow Qt Designer to be extended using PyQt. See Writing Qt Designer Plugins for a full description of how to do this.
  • The QAxContainer module. This module contains classes that allow access to ActiveX controls and COM objects.
  • The Qt module. This module consolidates the classes contained in all of the modules described above into a single module. This has the advantage that you don't have to worry about which underlying module contains a particular class. It has the disadvantage that it loads the whole of the Qt framework, thereby increasing the memory footprint of an application. Whether you use this consolidated module, or the individual component modules is down to personal taste.
  • The DBus support module is installed as dbus.mainloop.qt. PyQt does not support Qt's native DBus classes (which are very C++ orientated). Instead the dbus.mainloop.qt module provides support for the Qt event loop in the same way that the dbus.mainloop.glib included with the standard dbus-python bindings package provides support for the GLib event loop. The API is described in The DBus Support Module. It is only available for PyQt for X11 and only if the dbus-python v0.80 (or later) bindings package is installed.
  • The uic module. This module contains classes for handling the .ui files created by Qt Designer that describe the whole or part of a graphical user interface. It includes classes that load a .ui file and render it directly, and classes that generate Python code from a .ui file for later execution. It is covered in detail in The uic Module.
  • The pyqtconfig module is an extention of the SIP build system and is created when PyQt is configured. It encapsulates all the necessary information about your Qt installation and makes it easier to write installation scripts for bindings built on top of PyQt. It is covered in detail in The PyQt Build System.

PyQt also contains a number of utility programs.

  • pyuic4 corresponds to the Qt uic utility. It converts GUIs created using Qt Designer to Python code. It is covered in detail in pyuic4.
  • pyrcc4 corresponds to the Qt rcc utility. It embeds arbitrary resources (eg. icons, images, translation files) described by a resource collection file in a Python module. It is covered in detail in pyrcc4. (Note It will only be included if your copy of Qt includes the XML module.)
  • pylupdate4 corresponds to the Qt lupdate utility. It extracts all of the translatable strings from Python code and creates or updates .ts translation files. These are then used by Qt Linguist to manage the translation of those strings. It is covered in detail in pylupdate4. (Note It will only be included if your copy of Qt includes the XML module.)

When PyQt is configured a file called PyQt4.api is generated. This can be used by the QScintilla editor component (at http://www.riverbankcomputing.com/software/qscintilla/) to enable the use of auto-completion and call tips when editing PyQt code. The API file is installed automatically if QScintilla is already installed.

PyQt includes a large number of examples. These are ports to Python of many of the C++ examples provided with Qt. They can be found in the examples directory.

Finally, PyQt contains the .sip files used by SIP to generate PyQt itself. These can be used by developers of bindings of other Qt based class libraries - for example PyQwt and PyQwt3D.

2   Potential Incompatibilities with Earlier Versions

2.1   PyQt v4.8

2.1.1   QVariantList

In previous versions PyQt would always try and convert a Python list to a QVariantList. In this version PyQt will first try to convert it to a QVariant containing a QList<QObject *>, but only if QList<QObject *> has been registered with Qt as a meta-type.

Normally it is only the QtDeclarative module that registers this meta-type and so the behaviour of existing applications should be unchanged. It is possible however that you might observe different conversion behaviour after importing the QtDeclarative module.

2.2   PyQt v4.7.4

2.2.1   pyqtSignal() with dict and list

In previous versions a Qt signal defined using pyqtSignal() that had an argument specified as a dict then, when emitting a value, PyQt would try and convert the value to a QVariantMap if possible. If it wasn't possible, normally because the dict had non-string keys, then the value would be left as a dict object.

In this version PyQt will not attempt to convert the value to a QVariantMap and will always leave it as a dict object. If you want the value to be converted to a QVariantMap then define the signal argument as 'QVariantMap'.

The same applies to conversions between lists and QVariantList.

2.3   PyQt v4.7.1

2.3.1   QVariant

This version introduces a slight incompatibility in the conversion between sub-classes of standard Python types and QVariant.

Take, for example, the following code:

from PyQt4.QtCore import QVariant

class MyFloat(float):

    pass

myfloat = MyFloat(5.0)
variant = QVariant(myfloat)

With this version of PyQt myfloat will be converted in such a way as to preserve any additional attributes (including methods) and will not be converted to a C++ double. In other words, the following assertions are true:

assert(variant.type() != QVariant.Double)
assert(variant.toPyObject() is myfloat)

Prior to this version myfloat would be converted to a C++ double. In other words, the following assertions would be true:

assert(variant.type() == QVariant.Double)
assert(variant.toPyObject() == myfloat)
assert(type(variant.toPyObject()) is float)

The same change also affects objects that implement the sequence protocol. Prior to this version such an object would be converted to a QVariantList which would mean that it was converted back to a Python list rather than to the original type.

2.4   PyQt v4.5

2.4.1   QVariant

This version introduces a slight incompatibility in the conversion between Python sub-classes of certain Qt classes and QVariant. The Qt classes affected are those that QVariant has explicit support for, e.g. QSize, QBitmap.

Take, for example, the following code:

from PyQt4.QtCore import QSize, QVariant

class MySize(QSize):

    pass

mysize = MySize(5, 5)
variant = QVariant(mysize)

With this version of PyQt mysize will be converted in such a way as to preserve any additional attributes (including methods) and will not be converted to a C++ QSize instance. In other words, the following assertions are true:

assert(variant.type() != QVariant.Size)
assert(variant.toPyObject() is mysize)

Prior to this version mysize would be converted to a C++ QSize instance. In other words, the following assertions would be true:

assert(variant.type() == QVariant.Size)
assert(variant.toPyObject() == mysize)
assert(type(variant.toPyObject()) is QSize)

It is hoped that this change of behaviour will not have a significant impact. However if you need the old behaviour then simple create a copy of your sub-class instance using the base class constructor as shown below:

variant = QVariant(QSize(mysize))

A similar issue also affects the conversion of the Python datetime, date and time types to QVariant. These are no longer converted to the corresponding QDateTime, QDate and QTime classes. The values can be retrieved using QVariant.toPyObject(). Again, the old behaviour can be achieved using an explicit conversion to the Qt class before converting to QVariant.

A further incompatible change is the handling of Python sub-classes of QObject. In previous versions QVariant.userType() would return an internal type and an extra reference would be kept to the Python object. In the current version QVariant.userType() will correctly return QMetaType.QObjectStar (or QMetaType.QWidgetStar) but an extra reference to the Python object is not kept. To avoid a potential crash you should ensure that you keep a separate reference to the Python object, either explicitly or implicitly by giving it a parent.

2.4.2   pyrcc4 Support for Python v3

pyrcc4 will now generate code for Python v3 when the new -py3 command line option is used. The generated code will also work with Python v2.6 and later.

By default pyrcc4 will generate code for all Python v2 versions but you should use the new -py2 command line option to enforce this in case the default is changed in the future.

3   Installing PyQt

3.1   Downloading SIP

SIP must be installed before building and using PyQt. You can get the latest release of the SIP source code from http://www.riverbankcomputing.com/software/sip/download.

The SIP documentation can be found at http://www.riverbankcomputing.com/static/Docs/sip4/index.html.

3.2   Downloading PyQt

You can get the latest release of the GPL version of the PyQt source code from http://www.riverbankcomputing.com/software/pyqt/download.

If you are using the commercial version of PyQt then you should use the download instructions which were sent to you when you made your purchase. You must also download your license file.

3.3   Configuring PyQt

After unpacking the source package (either a .tar.gz or a .zip file depending on your platform) you should then check for any README files that relate to your platform.

If you are using the commercial version of PyQt then you must copy your license file to the sip directory.

You need to make sure your environment variables are set properly for your development environment. For example, if you are using a binary distribution of Qt on Windows then make sure you have run the qtvars.bat file. For other platforms it is normally enough to ensure that Qt's bin directory is on your PATH.

Next you need to configure SIP by executing the configure.py script. For example:

python configure.py

This assumes that the Python interpreter is on your path. Something like the following may be appropriate on Windows:

c:\python26\python configure.py

If you have multiple versions of Python installed then make sure you use the interpreter for which you wish to build PyQt for.

The full set of command line options is:

--version Display the PyQt version number.
-h, --help Display a help message.
--confirm-license
 Using this confirms that you accept the terms of the PyQt license.
-k, --static The PyQt modules will be built as static libraries. This is useful when building a custom interpreter with the PyQt modules built in to the interpreter.
--no-docstrings
 The PyQt modules will not contain automatically generated docstrings.
-r, --trace The generated PyQt modules contain additional tracing code that is enabled using SIP's sip.settracemask() function.
-u, --debug The PyQt modules will be built with debugging symbols. On Windows this requires that a debug version of Python is installed.
-w, --verbose Compiler commands and any output issued during configuration is displayed instead of being suppressed. Use this if configure.py is having problems to see what exactly is going wrong.
-c, --concatenate
 The C++ source files for a Python module will be concatenated. This results in significantly reduced compilation times. Most, but not all, C++ compilers can handle the large files that result. See also the --concatenate-split option.
-j N, --concatenate-split=N
 If the --concatenate option is used to concatenate the C++ source files then this option determines how many files are created. The default is 1.
--assume-shared
 Normally Qt is checked to see if it has been built as shared libraries. Some Linux distributions configure their Qt builds to make this check unreliable. This option ignores the result of the check and assumes that Qt has been built as shared libraries.
-g, --consolidate
 Normally each PyQt module (except for the Qt module) is linked against the corresponding Qt library. This option creates a module called _qt which is linked against all the required Qt libraries and the other modules are stub modules that populate their module dictionaries from this one. This is useful when linking against static Qt libraries to eliminate the need to distribute the Qt libraries while minimising the memory footprint of the PyQt modules.
-e MODULE, --enable=MODULE
 Normally checks for all PyQt4 modules are enabled and are built if the corresponding Qt library can be found. Using this option only those modules specifically enabled will be checked for and built. The option may be specified any number of times.
-t PLUGIN, --plugin=PLUGIN
 If Qt has been built as static libraries then the static plugin PLUGIN will be linked with the appropriate PyQt module. The option may be specified any number of times.
-T, --no-timestamp
 Normally the header comments of each generated C/C++ source file includes a timestamp corresponding to when the file was generated. This option suppresses the inclusion of the timestamp.
-q FILE, --qmake=FILE
 Qt's qmake program is used to determine how your Qt installation is laid out. Normally qmake is found on your PATH. This option can be used to specify a particular instance of qmake to use. This option is not available on Windows.
-s DIR, --dbus=DIR
 The dbus-python.h header file of the dbus-python package can be found in the directory DIR/dbus.
-b DIR, --bindir=DIR
 The pyuic4, pyrcc4 and pylupdate4 utilities will be installed in the directory DIR.
-d DIR, --destdir=DIR
 The PyQt Python package will be installed in the directory DIR. The default is the Python installation's site-packages directory. If you use this option then the PYTHONPATH environment variable must include DIR.
-p DIR, --plugin-destdir=DIR
 The Qt Designer plugin that manages plugins implemented in Python will be installed in the designer subdirectory of the directory DIR.
--no-designer-plugin
 The Qt Designer plugin will not be built.
--no-sip-files The .sip files for the PyQt modules will not be installed.
-v DIR, --sipdir=DIR
 The .sip files for the PyQt modules will be installed in the directory DIR.
--use-arch=ARCH
 When pyuic4 calls the Python interpreter on MacOS it will be run using the architecture ARCH. See the section Configuring SIP and PyQt for MacOs 10.6 (Snow Leopard).
--protected-is-public
 On certain platforms the size of PyQt modules can be significantly reduced by redefining the C++ protected keyword as public during compilation. This option enables this behaviour and is the default on Linux and MacOS/X.
--protected-not-public
 The default redefinition of protected to public during compilation on Linux and MacOS/X is disabled.
-i, --vendorid The checking of signed Python interpreters using the VendorID package is enabled. See also the --vendorid-incdir and --vendorid-libdir options and Deploying Commercial PyQt Applications.
-l DIR, --vendorid-incdir=DIR
 The header file of the VendorID package can be found in the directory DIR.
-m DIR, --vendorid-libdir=DIR
 The library of the VendorID package can be found in the directory DIR.
-a, --qsci-api The PyQt4.api QScintilla API file is installed even if QScintilla does not appear to be installed. This option is implied if the --qsci-api-destdir option is specified.
--no-qsci-api The PyQt4.api QScintilla API file is not installed even if QScintilla does appear to be installed.
-n DIR, --qsci-api-destdir=DIR
 The QScintilla API file will be installed in the python subdirectory of the api` subdirectory of the directory ``DIR.

3.4   Configuring SIP and PyQt for MacOS 10.6 (Snow Leopard)

With MacOS 10.6 the Python interpreter is built as a universal binary that supports the i386 and x86_64 architectures. (It also supports the ppc architecture but that isn't relevant.)

When building SIP and PyQt on a 64 bit system they will be built, by default, as x86_64 binaries. However, again by default, Qt builds as i386 binaries. (Note that the default is expected to change in Qt v4.7.) This means that, using the default configuration, PyQt will not build on a 64 bit system running MacOS 10.6 with a 32 bit build of Qt. Instead you have to make sure that SIP and PyQt are built as i386 binaries.

To configure SIP for i386 use the following command line options:

python configure.py --arch i386

When PyQt is configured it will automatically pick up the correct architecture from SIP's configuration. However it is necessary to use the following command line option when configuring PyQt:

python configure.py --use-arch i386

This tells the different PyQt tools that execute the Python interpreter (actually only pyuic4 at present) to use the i386 architecture rather than the default x86_64. This ensures that the interpreter will be able to import the i386 PyQt modules.

The other aspect to consider is the version of the SDK to use. By default SIP will use the latest version it can find, probably MacOSX10.6.sdk. However the Qt binary installer is built with MacOSX10.4u.sdk so you will probably need to use the following command line option when configuring SIP:

python configure.py --sdk MacOSX10.4u.sdk

3.5   Building PyQt

The next step is to build PyQt by running your platform's make command. For example:

make

The final step is to install PyQt by running the following command:

make install

(Depending on your system you may require root or administrator privileges.)

This will install the various PyQt components.

4   Selecting Incompatible APIs

PyQt provides limited support for multiple incompatible APIs and the ability for an application to select between them at run-time. For example, an application can choose whether QString is implemented as a Python type, or is automatically converted to and from a Python v2 unicode object or a Python v3 string object.

This ability allows developers to decide how to manage the transition from an older deprecated, API to a newer incompatible API.

Each API that can be selected in this way has a name and a range of version numbers. An application calls sip.setapi() to set the version number of a particular API. This call must be made before any module that implements the API is imported. Once set the version number cannot be changed. If not set then an API will use its default version.

For example the following code will disable the use of QString:

import sip
sip.setapi('QString', 2)

from PyQt4 import QtCore

# This will raise an attribute exception because QString is only wrapped
# in version 1 of the API.
s = QtCore.QString()

The rest of this section describes the different APIs that are available.

4.1   QDate

4.1.1   Version 2

This is the default for Python v3.

__hash__() returns a hash of the string representation so that two objects with the same date will have the same hash.

4.1.2   Version 1

This is the default for Python v2.

__hash__() returns an object's id() so that two objects with the same date will have different hashes.

4.2   QDateTime

4.2.1   Version 2

This is the default for Python v3.

__hash__() returns a hash of the string representation so that two objects with the same date and time will have the same hash.

4.2.2   Version 1

This is the default for Python v2.

__hash__() returns an object's id() so that two objects with the same date and time will have different hashes.

4.3   QString

4.3.1   Version 2

This is the default for Python v3.

The QString class is implemented as a mapped type that is automatically converted to and from a Python string. In addition a None is converted to a null QString. However, a null QString is converted to an empty Python string (and not None). (This is because Qt often returns a null QString when it should probably return an empty QString.)

The QChar and QStringRef classes are implemented as mapped types that are automatically converted to and from Python strings.

The QStringList class is implemented as a mapped type that is automatically converted to and from Python lists of strings.

The QLatin1Char, QLatin1String and QStringMatcher classes are not implemented.

The following Qt calls are not wrapped because they expect QString to be mutable:

void QTextDecoder::toUnicode(QString *target, const char *chars, int len)

QTextStream::QTextStream(QString *string, QIODevice::OpenMode openMode = QIODevice::ReadWrite)
void QTextStream::setString(QString *string, QIODevice::OpenMode openMode = QIODevice::ReadWrite)
QString *QTextStream::string()
QTextStream &operator>>(QChar &c)
QTextStream &operator>>(QString &s)

QXmlStreamWriter::QXmlStreamWriter(QString *string)

Some PyQt calls have changed Python signatures to avoid the need for mutable strings. The new signatures are as follows:

QAbstractSpinBox.fixup(str input) -> str
QAbstractSpinBox.validate(str input, int pos) -> QValidator.State, str, int

QDateTimeEdit.fixup(str input) -> str
QDateTimeEdit.validate(str input, int pos) -> QValidator.State, str, int

QDoubleSpinBox.fixup(str input) -> str
QDoubleSpinBox.validate(str input, int pos) -> QValidator.State, str, int

QDoubleValidator.validate(str input, int pos) -> QValidator.State, str, int

QClipboard.text(str subtype, QClipboard.Mode mode=QClipboard.Clipboard) -> str, str

QFileDialog.getOpenFileName(QWidget parent=None, str caption=None, str dir=None, str filter=None, QFileDialog.Options options=0) -> str
QFileDialog.getOpenFileNames(QWidget parent=None, str caption=None, str dir=None, str filter=None, QFileDialog.Options options=0) -> list(str)
QFileDialog.getSaveFileName(QWidget parent=None, str caption=None, str dir=None, str filter=None, QFileDialog.Options options=0) -> str

QIntValidator.validate(str input, int pos) -> QValidator.State, str, int

QRegExpValidator.validate(str input, int pos) -> QValidator.State, str, int

QSpinBox.fixup(str input) -> str
QSpinBox.validate(str input, int pos) -> QValidator.State, str, int

QValidator.fixup(str input) -> str
QValidator.validate(str input, int pos) -> QValidator.State, str, int

QWebPage.javaScriptPrompt(QWebFrame originatingFrame, str msg, str defaultValue) -> bool, str

The static methods getOpenFileNameAndFilter(), getOpenFileNamesAndFilter() and getSaveFileNameAndFilter() have been added to QFileDialog (for version 1 and version 2) which return a tuple of the name(s) and the selected filter.

The methods widthChar() and boundingRectChar() have been added to QFontMetrics and QFontMetricsF which accept a Python string of length one and call the C++ width() and boundingRect() methods passing the character as a QChar (rather than a single character QString).

4.3.2   Version 1

This is the default for Python v2.

The QChar, QLatin1Char, QLatin1String, QString, QStringList, QStringMatcher and QStringRef classes are implemented as normal types.

4.4   QTextStream

4.4.1   Version 2

This is the default for Python v3.

The C++ functions bin(), hex() and oct() are named bin_(), hex_() and oct_() respectively in Python. This allows the import style from PyQt4.QtCore import * to be used without them clashing with the Python built-in functions with the same names.

4.4.2   Version 1

This is the default for Python v2.

The C++ functions bin(), hex() and oct() have the same names in Python. This causes problems when the import style from PyQt4.QtCore import * is used because they clash with the Python built-in functions with the same names.

4.5   QTime

4.5.1   Version 2

This is the default for Python v3.

__hash__() returns a hash of the string representation so that two objects with the same time will have the same hash.

4.5.2   Version 1

This is the default for Python v2.

__hash__() returns an object's id() so that two objects with the same time will have different hashes.

4.6   QUrl

4.6.1   Version 2

This is the default for Python v3.

__hash__() returns a hash of the string representation so that two objects with the same URL will have the same hash.

4.6.2   Version 1

This is the default for Python v2.

__hash__() returns an object's id() so that two objects with the same URL will have different hashes.

4.7   QVariant

4.7.1   Version 2

This is the default for Python v3.

The QVariant class is implemented as a mapped type. Any Python object can be passed when a QVariant instance is expected. When Qt returns a QVariant then it will automatically be converted to the original Python object or an equivalent. None is interpreted as an invalid QVariant and vice versa.

4.7.2   Version 1

This is the default for Python v2.

The QVariant class is implememted as a normal type. Any Python object can be passed when a QVariant instance is expected and None is interpreted as an invalid QVariant. However, when Qt returns a QVariant then it must be explicitly converted to the original Python object or an equivalent by calling its toPyObject() method.

5   Support for Keyword Arguments

Starting with v4.7 PyQt supports the use of keyword arguments for optional arguments.

One thing to be aware of is that, although the PyQt and Qt documentation may indicate that an argument has a particular name, you may find that PyQt actually uses a different name. This is because the name of an argument is not part of the Qt API and there is some inconsistency in the way that similar arguments are named. Different versions of Qt may use a different name for an argument which wouldn't affect the C++ API but would break the Python API.

The docstrings that PyQt generates for all classes, functions and methods will contain the correct argument names. In a future version of PyQt the documentation will also contain the correct argument names.

6   Support for Qt Properties

PyQt does not support the setting and getting of Qt properties as if they were normal instance attributes. This is because the name of a property often conflicts with the name of the property's getter method.

However, PyQt does support the initial setting of properties using keyword arguments passed when an instance is created. For example:

act = QtGui.QAction("&Save", self, shortcut=QtGui.QKeySequence.Save,
        statusTip="Save the document to disk", triggered=self.save)

The example also demonstrates the use of a keyword argument to connect a signal to a slot.

PyQt also supports setting the values of properties (and connecting a signal to a slot) using the pyqtConfigure() method of QObject. For example, the following gives the same results as above:

act = QtGui.QAction("&Save", self)
act.pyqtConfigure(shortcut=QtGui.QKeySequence.Save,
        statusTip="Save the document to disk", triggered=self.save)

7   New-style Signal and Slot Support

This section describes the new style of connecting signals and slots introduced in PyQt v4.5.

One of the key features of Qt is its use of signals and slots to communicate between objects. Their use encourages the development of reusable components.

A signal is emitted when something of potential interest happens. A slot is a Python callable. If a signal is connected to a slot then the slot is called when the signal is emitted. If a signal isn't connected then nothing happens. The code (or component) that emits the signal does not know or care if the signal is being used.

The signal/slot mechanism has the following features.

7.1   Unbound and Bound Signals

A signal (specifically an unbound signal) is an attribute of a class that is a sub-class of QObject. When a signal is referenced as an attribute of an instance of the class then PyQt automatically binds the instance to the signal in order to create a bound signal. This is the same mechanism that Python itself uses to create bound methods from class functions.

A bound signal has connect(), disconnect() and emit() methods that implement the associated functionality.

A signal may be overloaded, ie. a signal with a particular name may support more than one signature. A bound signal may be indexed with a signature in order to select the one required. A signature is a sequence of types. A type is either a Python type object or a string that is the name of a C++ type.

If a signal is overloaded then it will have a default that will be used if no index is given.

When a signal is emitted then any arguments are converted to C++ types if possible. If an argument doesn't have a corresponding C++ type then it is wrapped in a special C++ type that allows it to be passed around Qt's meta-type system while ensuring that its reference count is properly maintained.

7.2   Defining New Signals with QtCore.pyqtSignal()

PyQt automatically defines signals for all Qt's built-in signals. New signals can be defined as class attributes using the QtCore.pyqtSignal() factory.

QtCore.pyqtSignal() takes a number of type arguments that corresponds to the signature of the signal. Each type may be a Python type object or a string that is the name of a C++ type. Alternatively each argument could be a sequence of type arguments. In this case each sequence defines the signature of a different signal overload. The first overload will be the default.

QtCore.pyqtSignal() takes an optional name keyword argument that is the name of the signal. If it is omitted then the name of the class attribute is used.

The following example shows the definition of a number of new signals:

from PyQt4 import QtCore

class Foo(QtCore.QObject):

    # This defines a signal called 'closed' that takes no arguments.
    closed = QtCore.pyqtSignal()

    # This defines a signal called 'rangeChanged' that takes two
    # integer arguments.
    range_changed = QtCore.pyqtSignal(int, int, name='rangeChanged')

    # This defines a signal called 'valueChanged' that has two overloads,
    # one that takes an integer argument and one that takes a QString
    # argument.
    valueChanged = QtCore.pyqtSignal((int, ), (QtCore.QString, ))

    # The following will create exactly the same overloaded signal as
    # above and demonstrates the use of C++ type names instead of Python
    # type objects, and lists instead of tuples.
    valueChanged = QtCore.pyqtSignal(['int'], ['QString'])

New signals should only be defined in sub-classes of QObject.

New signals defined in this way will be automatically added to the class's QMetaObject. This means that they will appear in Qt Designer and can be introspected using the QMetaObject API.

7.3   Connecting, Disconnecting and Emitting Signals

Signals are connected to slots using the connect() method of a bound signal:

connect(slot[, type=PyQt4.QtCore.Qt.AutoConnection])

    *slot* may be either a Python callable or another bound signal.

    *type* is a QtCore.Qt.ConnectionType value.

Signals are disconnected from slots using the disconnect() method of a bound signal:

disconnect([slot])

    *slot* may be either a Python callable or a another bound signal.  If
    slot is omitted then all slots connected to the signal are
    disconnected.

Signals are emitted from using the emit() method of a bound signal:

emit(*args)

    *args* is the optional sequence of arguments to pass to any connected
    slots.

The following code demonstrates the definition, connection and emit of a signal without arguments:

from PyQt4 import QtCore

class Foo(QtCore.QObject):

    # Define a new signal called 'trigger' that has no arguments.
    trigger = QtCore.pyqtSignal()

    def connect_and_emit_trigger(self):
        # Connect the trigger signal to a slot.
        self.trigger.connect(self.handle_trigger)

        # Emit the signal.
        self.trigger.emit()

    def handle_trigger(self):
        # Show that the slot has been called.

        print "trigger signal received"

The following code demonstrates the connection of overloaded signals:

from PyQt4 import QtGui

class Bar(QtGui.QComboBox):

    def connect_activated(self):
        # The PyQt documentation will define what the default overload is.
        # In this case it is the overload with the single integer argument.
        self.activated.connect(self.handle_int)

        # For non-default overloads we have to specify which we want to
        # connect.  In this case the one with the single string argument.
        # (Note that we could also explicitly specify the default if we
        # wanted to.)
        self.activated[str].connect(self.handle_string)

    def handle_int(self, index):
        print "activated signal passed integer", index

    def handle_string(self, text):
        print "activated signal passed QString", text

7.4   Connecting Signals Using Keyword Arguments

It is also possible to connect signals by passing a slot as a keyword argument corresponding to the name of the signal when creating an object, or using the pyqtConfigure() method of QObject. For example the following three fragments are equivalent:

act = QtGui.QAction("Action", self)
act.triggered.connect(self.on_triggered)

act = QtGui.QAction("Action", self, triggered=self.on_triggered)

act = QtGui.QAction("Action", self)
act.pyqtConfigure(triggered=self.on_triggered)

7.5   The QtCore.pyqtSlot() Decorator

Although PyQt allows any Python callable to be used as a slot when connecting signals, it is sometimes necessary to explicitly mark a Python method as being a Qt slot and to provide a C++ signature for it. PyQt provides the QtCore.pyqtSlot() function decorator to do this.

Using the decorator also has the advantage of reducing the amount of memory used and is slightly faster.

All of the non-keyword arguments to the decorator are interpreted as the types of the corresponding C++ arguments. A type is either a Python type object or a string that specifies a C++ type. The decorator also takes two optional keywords arguments: name and result. name is the name of the slot that will be seen by C++. If ommitted the name of the Python method being decorated will be used. result is the type of the result and may also be a Python type object or a string that specifies a C++ type.

For example:

@QtCore.pyqtSlot()
def foo(self):
    """ C++: void foo() """

@QtCore.pyqtSlot(int, str)
def foo(self, arg1, arg2):
    """ C++: void foo(int, QString) """

@QtCore.pyqtSlot(int, name='bar')
def foo(self, arg1):
    """ C++: void bar(int) """

@QtCore.pyqtSlot(int, result=int)
def foo(self, arg1):
    """ C++: int foo(int) """

@QtCore.pyqtSlot(int, QtGui.QWidget)
def foo(self, arg1):
    """ C++: int foo(int, QWidget *) """

It is also possible to chain the decorators in order to define a Python method several times with different signatures.

For example:

@QtCore.pyqtSlot(int)
@QtCore.pyqtSlot('QString')
def valueChanged(self, value):
    """ Two slots will be defined in the QMetaObject. """

7.6   Connecting Slots By Name

PyQt supports the QtCore.QMetaObject.connectSlotsByName() function that is most commonly used by pyuic4 generated Python code to automatically connect signals to slots that conform to a simple naming convention. However, where a class has overloaded Qt signals (ie. with the same name but with different arguments) PyQt needs additional information in order to automatically connect the correct signal.

For example the QtGui.QSpinBox class has the following signals:

void valueChanged(int i);
void valueChanged(const QString &text);

When the value of the spin box changes both of these signals will be emitted. If you have implemented a slot called on_spinbox_valueChanged (which assumes that you have given the QSpinBox instance the name spinbox) then it will be connected to both variations of the signal. Therefore, when the user changes the value, your slot will be called twice - once with an integer argument, and once with a unicode or QString argument.

This also happens with signals that take optional arguments. Qt implements this using multiple signals. For example, QtGui.QAbstractButton has the following signal:

void clicked(bool checked = false);

Qt implements this as the following:

void clicked();
void clicked(bool checked);

The decorator can be used to specify which of the signals should be connected to the slot.

For example, if you were only interested in the integer variant of the signal then your slot definition would look like the following:

@QtCore.pyqtSlot(int)
def on_spinbox_valueChanged(self, i):
    # i will be an integer.
    pass

If you wanted to handle both variants of the signal, but with different Python methods, then your slot definitions might look like the following:

@QtCore.pyqtSlot(int, name='on_spinbox_valueChanged')
def spinbox_int_value(self, i):
    # i will be an integer.
    pass

@QtCore.pyqtSlot(str, name='on_spinbox_valueChanged')
def spinbox_qstring_value(self, s):
    # s will be a Python string object (or a QString if they are enabled).
    pass

The following shows an example using a button when you are not interested in the optional argument:

@QtCore.pyqtSlot()
def on_button_clicked(self):
    pass

8   Old-style Signal and Slot Support

This section describes the older style for connecting signals and slots. It uses the same API that a C++ application would use. This has a number of advantages.

It also has a number of disadvantages.

This older style of connecting signals and slots will continue to be supported throughout the life of PyQt v4.

8.1   PyQt Signals and Qt Signals

Qt signals are statically defined as part of a C++ class. They are referenced using the QtCore.SIGNAL() function. This method takes a single string argument that is the name of the signal and its C++ signature. For example:

QtCore.SIGNAL("finished(int)")

The returned value is normally passed to the QtCore.QObject.connect() method.

PyQt allows new signals to be defined dynamically. The act of emitting a PyQt signal implicitly defines it. PyQt v4 signals are also referenced using the QtCore.SIGNAL() function.

8.2   The PyQt_PyObject Signal Argument Type

It is possible to pass any Python object as a signal argument by specifying PyQt_PyObject as the type of the argument in the signature. For example:

QtCore.SIGNAL("finished(PyQt_PyObject)")

While this would normally be used for passing objects like lists and dictionaries as signal arguments, it can be used for any Python type. Its advantage when passing, for example, an integer is that the normal conversions from a Python object to a C++ integer and back again are not required.

The reference count of the object being passed is maintained automatically. There is no need for the emitter of a signal to keep a reference to the object after the call to QtCore.QObject.emit(), even if a connection is queued.

8.3   Short-circuit Signals

There is also a special form of a PyQt v4 signal known as a short-circuit signal. Short-circut signals implicitly declare each argument as being of type PyQt_PyObject.

Short-circuit signals do not have a list of arguments or the surrounding parentheses.

Short-circuit signals may only be connected to slots that have been implemented in Python. They cannot be connected to Qt slots or the Python callables that wrap Qt slots.

8.4   PyQt Slots and Qt Slots

Qt slots are statically defined as part of a C++ class. They are referenced using the QtCore.SLOT() function. This method takes a single string argument that is the name of the slot and its C++ signature. For example:

QtCore.SLOT("done(int)")

The returned value is normally passed to the QtCore.QObject.connect() method.

PyQt allows any Python callable to be used as a slot, not just Qt slots. This is done by simply referencing the callable. Because Qt slots are implemented as class methods they are also available as Python callables. Therefore it is not usually necessary to use QtCore.SLOT() for Qt slots. However, doing so is more efficient as it avoids a conversion to Python and back to C++.

Qt allows a signal to be connected to a slot that requires fewer arguments than the signal passes. The extra arguments are quietly discarded. PyQt slots can be used in the same way.

Note that when a slot is a Python callable its reference count is not increased. This means that a class instance can be deleted without having to explicitly disconnect any signals connected to its methods. However, if a slot is a lambda function or a partial function then its reference count is automatically incremented to prevent it from being immediately garbage collected.

8.5   Connecting Signals and Slots

Connections between signals and slots (and other signals) are made using the QtCore.QObject.connect() method. For example:

QtCore.QObject.connect(a, QtCore.SIGNAL("QtSig()"), pyFunction)
QtCore.QObject.connect(a, QtCore.SIGNAL("QtSig()"), pyClass.pyMethod)
QtCore.QObject.connect(a, QtCore.SIGNAL("QtSig()"), b, QtCore.SLOT("QtSlot()"))
QtCore.QObject.connect(a, QtCore.SIGNAL("PySig()"), b, QtCore.SLOT("QtSlot()"))
QtCore.QObject.connect(a, QtCore.SIGNAL("PySig"), pyFunction)

Disconnecting signals works in exactly the same way using the QtCore.QObject.disconnect() method. However, not all the variations of that method are supported by PyQt. Signals must be disconnected one at a time.

8.6   Emitting Signals

Any instance of a class that is derived from the QtCore.QObject class can emit a signal using its emit() method. This takes a minimum of one argument which is the signal. Any other arguments are passed to the connected slots as the signal arguments. For example:

a.emit(QtCore.SIGNAL("clicked()"))
a.emit(QtCore.SIGNAL("pySig"), "Hello", "World")

8.7   The QtCore.pyqtSignature() Decorator

The QtCore.pyqtSignature() serves the same purpose as the QtCore.pyqtSlot() decorator but has a less Pythonic API.

9   Python Objects and QVariant

Qt uses the QVariant class as a wrapper for any C++ data type. PyQt allows any Python object to be wrapped as a QVariant and passed around Qt's meta-object system like any other type.

PyQt will try to convert the Python object to a C++ equivalent if it can so that the QVariant can be passed to other C++ code that doesn't know what a Python object is.

Version 2 of PyQt's QVariant API will automatically convert a QVariant back to a Python object of the correct type.

Version 1 of the QVariant API provides the QVariant.toPyObject() method to convert the QVariant back to a Python object of the correct type.

Both versions will raise a Python exception if the conversion cannot be done.

10   Integrating Python and QML

Qt v4.7 introduced QML as a means of declaratively describing a user interface and using JavaScript as a scripting language within QML. It is possible to write complete standalone QML applications, or to combine them with C++.

PyQt supports the integration of Python and QML as far as it can - given the limitations of the QML implementation.

There are three types of QML application:

11   Integrating Python and JavaScript in QtWebKit

QtWebKit uses slots to expose class methods implemented in C++ as JavaScript methods that can be called from scripts embedded in HTML. Python class methods that have been decorated behave in exactly the same way.

In the same way, properties created using QtCore.pyqtProperty() are also automatically exposed as JavaScript properties.

12   Support for Pickling

The following PyQt classes may be pickled.

  • QByteArray
  • QChar
  • QColor
  • QDate
  • QDateTime
  • QKeySequence
  • QLatin1Char
  • QLatin1String
  • QLine
  • QLineF
  • QMatrix
  • QPoint
  • QPointF
  • QPolygon
  • QRect
  • QRectF
  • QSize
  • QSizeF
  • QString
  • QTime

Also all named enums (QtCore.Qt.Key for example) may be pickled.

13   Support for Python's Buffer Interface

If SIP v4.7.5 or later is used then any Python object that supports the buffer interface can be used whenever a char or char * is expected. If the buffer has multiple segments then all but the first will be ignored.

14   Using PyQt from the Python Shell

PyQt installs an input hook (using PyOS_InputHook) that processes events when an interactive interpreter is waiting for user input. This means that you can, for example, create widgets from the Python shell prompt, interact with them, and still being able to enter other Python commands.

For example, if you enter the following in the Python shell:

>>> from PyQt4 import QtGui
>>> a = QtGui.QApplication([])
>>> w = QtGui.QWidget()
>>> w.show()
>>> w.hide()
>>>

The widget would be displayed when w.show() was entered amd hidden as soon as w.hide() was entered.

The installation of an input hook can cause problems for certain applications (particularly those that implement a similar feature using different means). The QtCore module contains the pyqtRemoveInputHook() and pyqtRestoreInputHook() functions that remove and restore the input hook respectively.

15   Using Qt Designer

Qt Designer is the Qt tool for designing and building graphical user interfaces. It allows you to design widgets, dialogs or complete main windows using on-screen forms and a simple drag-and-drop interface. It has the ability to preview your designs to ensure they work as you intended, and to allow you to prototype them with your users, before you have to write any code.

Qt Designer uses XML .ui files to store designs and does not generate any code itself. Qt includes the uic utility that generates the C++ code that creates the user interface. Qt also includes the QUiLoader class that allows an application to load a .ui file and to create the corresponding user interface dynamically.

PyQt does not wrap the QUiLoader class but instead includes the uic Python module. Like QUiLoader this module can load .ui files to create a user interface dynamically. Like the uic utility it can also generate the Python code that will create the user interface. PyQt's pyuic4 utility is a command line interface to the uic module. Both are described in detail in the following sections.

15.1   Using the Generated Code

The code that is generated has an identical structure to that generated by Qt's uic and can be used in the same way.

The code is structured as a single class that is derived from the Python object type. The name of the class is the name of the toplevel object set in Designer with Ui_ prepended. (In the C++ version the class is defined in the Ui namespace.) We refer to this class as the form class.

The class contains a method called setupUi(). This takes a single argument which is the widget in which the user interface is created. The type of this argument (typically QDialog, QWidget or QMainWindow) is set in Designer. We refer to this type as the Qt base class.

In the following examples we assume that a .ui file has been created containing a dialog and the name of the QDialog object is ImageDialog. We also assume that the name of the file containing the generated Python code is ui_imagedialog.py. The generated code can then be used in a number of ways.

The first example shows the direct approach where we simply create a simple application to create the dialog:

import sys
from PyQt4 import QtGui
from ui_imagedialog import Ui_ImageDialog

app = QtGui.QApplication(sys.argv)
window = QtGui.QDialog()
ui = Ui_ImageDialog()
ui.setupUi(window)

window.show()
sys.exit(app.exec_())

The second example shows the single inheritance approach where we sub-class QDialog and set up the user interface in the __init__() method:

from PyQt4 import QtCore, QtGui
from ui_imagedialog import Ui_ImageDialog

class ImageDialog(QtGui.QDialog):
    def __init__(self):
        QtGui.QDialog.__init__(self)

        # Set up the user interface from Designer.
        self.ui = Ui_ImageDialog()
        self.ui.setupUi(self)

        # Make some local modifications.
        self.ui.colorDepthCombo.addItem("2 colors (1 bit per pixel)")

        # Connect up the buttons.
        self.connect(self.ui.okButton, QtCore.SIGNAL("clicked()"),
                     self, QtCore.SLOT("accept()"))
        self.connect(self.ui.cancelButton, QtCore.SIGNAL("clicked()"),
                     self, QtCore.SLOT("reject()"))

The third example shows the multiple inheritance approach:

from PyQt4 import QtCore, QtGui
from ui_imagedialog import Ui_ImageDialog

class ImageDialog(QtGui.QDialog, Ui_ImageDialog):
    def __init__(self):
        QtGui.QDialog.__init__(self)

        # Set up the user interface from Designer.
        self.setupUi(self)

        # Make some local modifications.
        self.colorDepthCombo.addItem("2 colors (1 bit per pixel)")

        # Connect up the buttons.
        self.connect(self.okButton, QtCore.SIGNAL("clicked()"),
                     self, QtCore.SLOT("accept()"))
        self.connect(self.cancelButton, QtCore.SIGNAL("clicked()"),
                     self, QtCore.SLOT("reject()"))

It is also possible to use the same approach used in PyQt v3. This is shown in the final example:

from PyQt4 import QtCore, QtGui
from ui_imagedialog import ImageDialog

class MyImageDialog(ImageDialog):
    def __init__(self):
        ImageDialog.__init__(self)

        # Make some local modifications.
        self.colorDepthCombo.addItem("2 colors (1 bit per pixel)")

        # Connect up the buttons.
        self.connect(self.okButton, QtCore.SIGNAL("clicked()"),
                     self, QtCore.SLOT("accept()"))
        self.connect(self.cancelButton, QtCore.SIGNAL("clicked()"),
                     self, QtCore.SLOT("reject()"))

For a full description see the Qt Designer Manual in the Qt Documentation.

15.2   The uic Module

The uic module contains the following functions and objects.

widgetPluginPath
This is a list of the directories that are searched for widget plugins. Initially it contains the name of the directory that contains the widget plugins included with PyQt.
compileUi(uifile, pyfile, execute=False, indent=4, pyqt3_wrapper=False, from_imports=False)

This function generates a Python module that will create a user interface from a Qt Designer .ui file.

uifile is a file name or file-like object containing the .ui file.

pyfile is the file-like object to which the generated Python code will be written to.

execute is optionally set if a small amount of additional code is to be generated that will display the user interface if the code is run as a standalone application.

indent is the optional number of spaces used for indentation in the generated code. If it is zero then a tab character is used instead.

pyqt3_wrapper is optionally set if a small wrapper is to be generated that allows the generated code to be used as it is by PyQt v3 applications.

from_imports is optionally set to generate import statements that are relative to '.'. At the moment this only applies to the import of resource modules.

compileUiDir(dir, recurse=False, map=None, **compileUi_args)

This function creates Python modules from Qt Designer .ui files in a directory or directory tree.

dir is the name of the directory to scan for files whose name ends with .ui. By default the generated Python module is created in the same directory ending with .py.

recurse is set if any sub-directories should be scanned.

map is an optional callable that is passed the name of the directory containing the .ui file and the name of the Python module that will be created. The callable should return a tuple of the name of the directory in which the Python module will be created and the (possibly modified) name of the module.

compileUi_args are any additional keyword arguments that are passed to the compileUi() function that is called to create each Python module.

loadUiType(uifile, from_imports=False)

This function loads a Qt Designer .ui file and returns a tuple of the generated form class and the Qt base class. These can then be used to create any number of instances of the user interface without having to parse the .ui file more than once.

uifile is a file name or file-like object containing the .ui file.

from_imports is optionally set to use import statements that are relative to '.'. At the moment this only applies to the import of resource modules.

loadUi(uifile, baseinstance=None)

This function loads a Qt Designer .ui file and returns an instance of the user interface.

uifile is a file name or file-like object containing the .ui file.

baseinstance is an optional instance of the Qt base class. If specified then the user interface is created in it. Otherwise a new instance of the base class is automatically created.

15.3   pyuic4

The pyuic4 utility is a command line interface to the uic module. The command has the following syntax:

pyuic4 [options] .ui-file

The full set of command line options is:

-h, --help A help message is written to stdout.
--version The version number is written to stdout.
-i N, --indent=N
 The Python code is generated using an indentation of N spaces. If N is 0 then a tab is used. The default is 4.
-o FILE, --output=FILE
 The Python code generated is written to the file FILE.
-p, --preview The GUI is created dynamically and displayed. No Python code is generated.
-w, --pyqt3-wrapper
 The generated Python code includes a small wrapper that allows the GUI to be used in the same way as it is used in PyQt v3.
-x, --execute The generated Python code includes a small amount of additional code that creates and displays the GUI when it is executes as a standalone application.
--from-imports Resource modules are imported using from . import rather than a simple import.

Note that code generated by pyuic4 is not guaranteed to be compatible with earlier versions of PyQt. However, it is guaranteed to be compatible with later versions. If you have no control over the version of PyQt the users of your application are using then you should run pyuic4, or call PyQt4.uic.compileUi(), as part of your installation process. Another alternative would be to distribute the .ui files (perhaps as part of a resource file) and have your application load them dynamically.

15.4   Writing Qt Designer Plugins

Qt Designer can be extended by writing plugins. Normally this is done using C++ but PyQt also allows you to write plugins in Python. Most of the time a plugin is used to expose a custom widget to Designer so that it appears in Designer's widget box just like any other widget. It is possibe to change the widget's properties and to connect its signals and slots.

It is also possible to add new functionality to Designer. See the Qt documentation for the full details. Here we will concentrate on describing how to write custom widgets in Python.

The process of integrating Python custom widgets with Designer is very similar to that used with widget written using C++. However, there are particular issues that have to be addressed.

  • Designer needs to have a C++ plugin that conforms to the interface defined by the QDesignerCustomWidgetInterface class. (If the plugin exposes more than one custom widget then it must conform to the interface defined by the QDesignerCustomWidgetCollectionInterface class.) In addition the plugin class must sub-class QObject as well as the interface class. PyQt does not allow Python classes to be sub-classed from more than one Qt class.
  • Designer can only connect Qt signals and slots. It has no understanding of Python signals or callables.
  • Designer can only edit Qt properties that represent C++ types. It has no understanding of Python attributes or Python types.

PyQt provides the following components and features to resolve these issues as simply as possible.

  • PyQt's QtDesigner module includes additional classes (all of which have a QPy prefix) that are already sub-classed from the necessary Qt classes. This avoids the need to sub-class from more than one Qt class in Python. For example, where a C++ custom widget plugin would sub-class from QObject and QDesignerCustomWidgetInterface, a Python custom widget plugin would instead sub-class from QPyDesignerCustomWidgetPlugin.

  • PyQt installs a C++ plugin in Designer's plugin directory. It conforms to the interface defined by the QDesignerCustomWidgetCollectionInterface class. It searches a configurable set of directories looking for Python plugins that implement a class sub-classed from QPyDesignerCustomWidgetPlugin. Each class that is found is instantiated and the instance created is added to the custom widget collection.

    The PYQTDESIGNERPATH environment variable specifies the set of directories to search for plugins. Directory names are separated by a path separator (a semi-colon on Windows and a colon on other platforms). If a directory name is empty (ie. there are consecutive path separators or a leading or trailing path separator) then a set of default directories is automatically inserted at that point. The default directories are the python subdirectory of each directory that Designer searches for its own plugins. If the environment variable is not set then only the default directories are searched. If a file's basename does not end with plugin then it is ignored.

  • A Python custom widget may define new Qt signals using QtCore.pyqtSignal().

  • A Python class method may be defined as a new Qt slot by using the QtCore.pyqtSlot decorator. For example:

    # Define a Qt slot that takes a C++ integer argument.
    @QtCore.pyqtSlot(int, name='addToTotal')
    def add_int_to_total(self, value):
        pass
    
    # Define a similar slot that takes its name from the method.
    @QtCore.pyqtSlot(int)
    def addToTotal(self, value):
        pass
    
  • A new Qt property may be defined using the QtCore.pyqtProperty() function. It is used in the same way as the standard Python property() function. In fact, Qt properties defined in this way also behave as Python properties. The full signature of the function is as follows:

    pyqtProperty(type, fget=None, fset=None, freset=None, fdel=None, doc=None, designable=True, scriptable=True, stored=True, user=False, constant=False, final=False, notify=None)
    

    type is the type of the property. It is either a Python type object or a string that is the name of a C++ type. freset is a function used to reset the value of the property to its default value. designable sets the Qt DESIGNABLE flag. scriptable sets the Qt SCRIPTABLE flag. stored sets the Qt STORED flag. user sets the Qt USER flag. constant sets the Qt CONSTANT flag. final sets the Qt FINAL flag. notify is the notify signal.

    The remaining arguments are the same as those used by the standard property() function.

    Qt makes no use of the fdel function and Python makes no use of the freset function, or the designable, scriptable, stored, user, constant and final flags.

    It is also possible to use QtCore.pyqtProperty() as a decorator in the same way as the standard Python property() function. The following example shows how to define an int property with a getter and setter:

    @QtCore.pyqtProperty(int)
    def total(self):
        return self._total
    
    @total.setter
    def total(self, value):
        self._total = value
    

    If you prefer the Qt terminology you may also use write instead of setter (and read instead of getter).

Note that the ability to define new Qt signals, slots and properties from Python is potentially useful to plugins conforming to any plugin interface and not just that used by Designer.

For a simple but complete and fully documented example of a custom widget that defines new Qt signals, slots and properties, and its plugin, look in the examples/designer/plugins directory of the PyQt source package. The widgets subdirectory contains the pydemo.py custom widget and the python subdirectory contains its pydemoplugin.py plugin.

16   The PyQt Resource System

PyQt supports Qt's resource system. This is a facility for embedding resources such as icons and translation files in an application. This makes the packaging and distribution of those resources much easier.

A .qrc resource collection file is an XML file used to specify which resource files are to be embedded. The application then refers to the resource files by their original names but preceded by a colon.

For a full description, including the format of the .qrc files, see the Qt Resource System in the Qt documentation.

16.1   pyrcc4

pyrcc4 is PyQt's equivalent to Qt's rcc utility and is used in exactly the same way. pyrcc4 reads the .qrc file, and the resource files, and generates a Python module that only needs to be import ed by the application in order for those resources to be made available just as if they were the original files.

Starting with PyQt v4.5, pyrcc generates code for Python v2.6 and later by default. If you use the -py2 command line option then pyrcc will generate code for all Python v2.x versions.

pyrcc4 will only be included if your copy of Qt includes the XML module.

17   Internationalisation of PyQt Applications

PyQt and Qt include a comprehensive set of tools for translating applications into local languages. For a full description, see the Qt Linguist Manual in the Qt documentation.

The process of internationalising an application comprises the following steps.

  • The programmer uses pylupdate4 to create or update a .ts translation file for each language that the application is to be translated into. A .ts file is an XML file that contains the strings to be translated and the corresponding translations that have already been made. pylupdate4 can be run any number of times during development to update the .ts files with the latest strings for translation.
  • The translator uses Qt Linguist to update the .ts files with translations of the strings.
  • The release manager then uses Qt's lrelease utility to convert the .ts files to .qm files which are compact binary equivalents used by the application. If an application cannot find an appropriate .qm file, or a particular string hasn't been translated, then the strings used in the original source code are used instead.
  • The release manage may optionally use pyrcc4 to embed the .qm files, along with other application resources such as icons, in a Python module. This may make packaging and distribution of the application easier.

17.1   pylupdate4

pylupdate4 is PyQt's equivalent to Qt's lupdate utility and is used in exactly the same way. A Qt .pro project file is read that specifies the Python source files and Qt Designer interface files from which the text that needs to be translated is extracted. The .pro file also specifies the .ts translation files that pylupdate4 updates (or creates if necessary) and are subsequently used by Qt Linguist.

pylupdate4 will only be included if your copy of Qt includes the XML module.

17.2   Differences Between PyQt and Qt

Qt implements internationalisation support through the QTranslator class, and the QCoreApplication::translate(), QObject::tr() and QObject::trUtf8() methods. Usually the tr() method is used to obtain the correct translation of a message. The translation process uses a message context to allow the same message to be translated differently. tr() is actually generated by moc and uses the hardcoded class name as the context. On the other hand, QApplication::translate() allows the context to be explicitly stated.

Unfortunately, because of the way Qt implements tr() (and trUtf8()) it is not possible for PyQt to exactly reproduce its behaviour. The PyQt implementation of tr() (and trUtf8()) uses the class name of the instance as the context. The key difference, and the source of potential problems, is that the context is determined dynamically in PyQt, but is hardcoded in Qt. In other words, the context of a translation may change depending on an instance's class hierarchy. For example:

class A(QtCore.QObject):
    def hello(self):
        return self.tr("Hello")

class B(A):
    pass

a = A()
a.hello()

b = B()
b.hello()

In the above the message is translated by a.hello() using a context of A, and by b.hello() using a context of B. In the equivalent C++ version the context would be A in both cases.

The PyQt behaviour is unsatisfactory and may be changed in the future. It is recommended that QCoreApplication.translate() be used in preference to tr() (and trUtf8()). This is guaranteed to work with current and future versions of PyQt and makes it much easier to share message files between Python and C++ code. Below is the alternative implementation of A that uses QCoreApplication.translate():

class A(QtCore.QObject):
    def hello(self):
        return QtCore.QCoreApplication.translate("A", "Hello")

18   The DBus Support Module

The DBus support module is installed as dbus.mainloop.qt and provides support for the Qt event loop to the standard dbus-python language bindings package. The module's API is almost identical to that of the dbus.mainloop.glib modules that provides support for the GLib event loop.

The dbus.mainloop.qt module contains the following function.

DBusQtMainLoop(set_as_default=False)

This function returns a dbus.mainloop.NativeMainLoop object that uses the the Qt event loop.

set_as_default is set to make the main loop instance the default for all new Connection and Bus instances. It may only be specified as a keyword argument, and not as a positional argument.

The following code fragment is all that is normally needed to set up the standard dbus-python language bindings package to be used with PyQt:

import dbus.mainloop.qt

dbus.mainloop.qt.DBusQtMainLoop(set_as_default=True)

19   Things to be Aware Of

19.1   Python Strings, Qt Strings and Unicode

PyQt uses the QString class to represent Unicode strings, and the QByteArray to represent byte arrays or strings. In Python v3 the corresponding native object types are str and bytes. In Python v2 the corresponding native object types are unicode and str.

PyQt does its best to automatically convert between objects of the various types. Explicit conversions can be easily made where necessary.

In some cases PyQt will not perform automatic conversions where it is necessary to distinguish between different overloaded methods.

For Python v3 the following conversions are done by default.

  • If Qt expects a char * (or a const version) then PyQt will accept a str or QString that contains only ASCII characters, a bytes, a QByteArray, or a Python object that implements the buffer protocol.
  • If Qt expects a char (or a const version) then PyQt will accept the same types as for char * and also require that a single character is provided.
  • If Qt expects a signed char * or an unsigned char * (or a const version) then PyQt will accept a bytes.
  • If Qt expects a signed char or an unsigned char (or a const version) then PyQt will accept a bytes of length 1.
  • If Qt expects a QString then PyQt will accept a str, a bytes that contains only ASCII characters, a QChar or a QByteArray.
  • If Qt expects a QByteArray then PyQt will also accept a str that contains only Latin-1 characters, or a bytes.

For Python v2 the following conversions are done by default.

  • If Qt expects a char *, signed char * or an unsigned char * (or a const version) then PyQt will accept a unicode or QString that contains only ASCII characters, a str, a QByteArray, or a Python object that implements the buffer protocol.
  • If Qt expects a char, signed char or an unsigned char (or a const version) then PyQt will accept the same types as for char *, signed char * and unsigned char * and also require that a single character is provided.
  • If Qt expects a QString then PyQt will accept a unicode, a str that contains only ASCII characters, a QChar or a QByteArray.
  • If Qt expects a QByteArray then PyQt will accept a unicode that contains only Latin-1 characters, or a str.

Note that the different behaviour between Python v2 and v3 is due to v3's reduced support for the buffer protocol.

19.2   Garbage Collection

C++ does not garbage collect unreferenced class instances, whereas Python does. In the following C++ fragment both colours exist even though the first can no longer be referenced from within the program:

col = new QColor();
col = new QColor();

In the corresponding Python fragment, the first colour is destroyed when the second is assigned to col:

col = QtGui.QColor()
col = QtGui.QColor()

In Python, each colour must be assigned to different names. Typically this is done within class definitions, so the code fragment would be something like:

self.col1 = QtGui.QColor()
self.col2 = QtGui.QColor()

Sometimes a Qt class instance will maintain a pointer to another instance and will eventually call the destructor of that second instance. The most common example is that a QObject (and any of its sub-classes) keeps pointers to its children and will automatically call their destructors. In these cases, the corresponding Python object will also keep a reference to the corresponding child objects.

So, in the following Python fragment, the first QLabel is not destroyed when the second is assigned to lab because the parent QWidget still has a reference to it:

parent = QtGui.QWidget()
lab = QtGui.QLabel("First label", parent)
lab = QtGui.QLabel("Second label", parent)

19.3   Multiple Inheritance

It is not possible to define a new Python class that sub-classes from more than one Qt class.

19.4   Access to Protected Member Functions

When an instance of a C++ class is not created from Python it is not possible to access the protected member functions, or emit any signals, of that instance. Attempts to do so will raise a Python exception. Also, any Python methods corresponding to the instance's virtual member functions will never be called.

19.5   None and NULL

Throughout PyQt, the None value can be specified wherever NULL is acceptable to the underlying C++ code.

Equally, NULL is converted to None whenever it is returned by the underlying C++ code.

19.6   Support for void *

PyQt (actually SIP) represents void * values as objects of type sip.voidptr. Such values are often used to pass the addresses of external objects between different Python modules. To make this easier, a Python integer (or anything that Python can convert to an integer) can be used whenever a sip.voidptr is expected.

A sip.voidptr may be converted to a Python integer by using the int() builtin function.

A sip.voidptr may be converted to a Python string by using its asstring() method. The asstring() method takes an optional integer argument which is the length of the data in bytes.

A sip.voidptr may also be given a size (ie. the size of the block of memory that is pointed to) by calling its setsize() method. If it has a size then it is also able to support Python's buffer protocol. This means that it can be wrapped using Python's buffer() builtin to create an object that treats the block of memory as a mutable list of bytes. It also means that the Python struct module can be used to unpack and pack binary data structures in memory, memory mapped files or shared memory.

19.7   super and PyQt Classes

In versions of PyQt earlier than v4.5 there were restrictions on the use of super with PyQt classes. These restrictions no longer apply with v4.5 and later.

20   Deploying Commercial PyQt Applications

When deploying commercial PyQt applications it is necessary to discourage users from accessing the underlying PyQt modules for themselves. A user that used the modules shipped with your application to develop new applications would themselves be considered a developer and would need their own commercial Qt and PyQt licenses.

One solution to this problem is the VendorID package. This allows you to build Python extension modules that can only be imported by a digitally signed custom interpreter. The package enables you to create such an interpreter with your application embedded within it. The result is an interpreter that can only run your application, and PyQt modules that can only be imported by that interpreter. You can use the package to similarly restrict access to any extension module.

In order to build PyQt with support for the VendorID package, pass the -i command line flag to configure.py.

21   The PyQt Build System

The PyQt build system is an extension of the SIP build system and is implemented by the pyqtconfig module, part of the PyQt4 package. It can be used by configuration scripts of other bindings that build on top of PyQt and takes care of the details of the Qt installation.

The module contains a number of classes.

21.1   pyqtconfig Classes

Configuration(sipconfig.Configuration)

This class encapsulates configuration values that can be accessed as instance objects.

The following configuration values are provided in addition to those provided by the super-class:

pyqt_bin_dir
The name of the directory where the PyQt utilities are installed.
pyqt_config_args
The command line passed to configure.py when PyQt was configured.
pyqt_mod_dir
The name of the directory where the PyQt4 Python package is installed.
pyqt_modules
A space separated string of installed PyQt modules. The Qt module is not included.
pyqt_sip_dir
The name of the base directory where PyQt's .sip files are installed. Each module's .sip files are installed in a sub-directory with the same name as the module.
pyqt_sip_flags
A space separated string of the sip command line arguments used to build the PyQt modules. These should also be used when building bindings that %Import any PyQt modules.
pyqt_version
The PyQt version as a 3 part hexadecimal number (e.g. v4.0.1 is represented as 0x040001).
pyqt_version_str
The PyQt version as a string. For development snapshots it will start with snapshot-.
qt_data_dir
The value of QLibraryInfo::location(DataPath) for the Qt installation.
qt_dir
The root directory of the Qt installation (normally the directory that contains the bin directory).
qt_edition
The Qt edition.
qt_framework
Set if Qt is built as a MacOS/X framework.
qt_inc_dir
The value of QLibraryInfo::location(HeadersPath) for the Qt installation.
qt_lib_dir
The value of QLibraryInfo::location(LibrariesPath) for the Qt installation.
qt_threaded
Set if Qt is built with thread support (always set for PyQt).
qt_version
The Qt version as a 3 part hexadecimal number (e.g. v4.1.2 is represented as 0x040102).
qt_winconfig
Additional Windows specific configuration.
__init__(self, sub_cfg=None)

Initialise the instance.

sub_cfg is an optional list of sub-class configurations. It should only be used by the __init__() method of a sub-class to append its own dictionary of configuration values before passing the list to its super-class.

QtAssistantModuleMakefile(QtNetworkModuleMakefile)
This class encapsulates a Makefile to build a SIP generated Python extension module that is built on the PyQt QtAssistant module.
QAxContainerModuleMakefile(QtGuiModuleMakefile)
This class encapsulates a Makefile to build a SIP generated Python extension module that is built on the PyQt QAxContainer module.
QtCoreModuleMakefile(sipconfig.SIPModuleMakefile)
This class encapsulates a Makefile to build a SIP generated Python extension module that is built on the PyQt QtCore module.
QtDeclarativeModuleMakefile(QtNetworkModuleMakefile)
This class encapsulates a Makefile to build a SIP generated Python extension module that is built on the PyQt QtDeclarative module.
QtGuiModuleMakefile(QtCoreModuleMakefile)
This class encapsulates a Makefile to build a SIP generated Python extension module that is built on the PyQt QtGui module.
QtHelpModuleMakefile(QtGuiModuleMakefile)
This class encapsulates a Makefile to build a SIP generated Python extension module that is built on the PyQt QtHelp module.
QtMultimediaModuleMakefile(QtGuiModuleMakefile)
This class encapsulates a Makefile to build a SIP generated Python extension module that is built on the PyQt QtMultimedia module.
QtNetworkModuleMakefile(QtCoreModuleMakefile)
This class encapsulates a Makefile to build a SIP generated Python extension module that is built on the PyQt QtNetwork module.
QtOpenGLModuleMakefile(QtGuiModuleMakefile)
This class encapsulates a Makefile to build a SIP generated Python extension module that is built on the PyQt QtOpenGL module.
QtScriptModuleMakefile(QtCoreModuleMakefile)
This class encapsulates a Makefile to build a SIP generated Python extension module that is built on the PyQt QtScript module.
QtSqlModuleMakefile(QtGuiModuleMakefile)
This class encapsulates a Makefile to build a SIP generated Python extension module that is built on the PyQt QtSql module.
QtSvgModuleMakefile(QtGuiModuleMakefile)
This class encapsulates a Makefile to build a SIP generated Python extension module that is built on the PyQt QtSvg module.
QtTestModuleMakefile(QtGuiModuleMakefile)
This class encapsulates a Makefile to build a SIP generated Python extension module that is built on the PyQt QtTest module.
QtWebKitModuleMakefile(QtNetworkModuleMakefile)
This class encapsulates a Makefile to build a SIP generated Python extension module that is built on the PyQt QtWebKit module.
QtXmlModuleMakefile(QtCoreModuleMakefile)
This class encapsulates a Makefile to build a SIP generated Python extension module that is built on the PyQt QtXml module.
QtXmlPatternsModuleMakefile(QtCoreModuleMakefile)
This class encapsulates a Makefile to build a SIP generated Python extension module that is built on the PyQt QtXmlPatterns module.
phononModuleMakefile(QtGuiModuleMakefile)
This class encapsulates a Makefile to build a SIP generated Python extension module that is built on the PyQt phonon module.