Cos3711 Additional Notes

Download as pdf or txt
Download as pdf or txt
You are on page 1of 16

COS3711 Advance Programming 2013

Extra notes for Ezust2 2nd Edition


The notes provided here pick up on some of the issues that you may come across while
working through the prescribed book for this module. Where an example is not mentioned
here it means that the program works as described in Ezust2.
CHAPTER 7 (Libraries and Design Patterns)
Section 7.2: Installing Libraries
Read through this section without downloading or installing any libraries suggested in this
section. The installation of the Ezust libraries is detailed in Tutorial Letter 102.
Section 7.4.1: Serializer Pattern
For the Serializer pattern, rather use your notes on the pattern from COS2614. It would be
constructive to compare those notes to the approach taken in Ezust2, and see what the
differences and similarities are.
CHAPTER 12 (Meta Objects, Properties, and Reflective Programming)
Section 12.2: Type Identification (playlists.pro)
This application may not want to compile as the ui_mainwindow.h (created by Qt)
cannot find the playlistsveiw.h or playlistview.h files, even though they are in
the correct folder. This has to do with folder structures, and once you have built the
application you may have to change the two lines
#include "playlistsview.h"
#include "playlistview.h"
to
#include "..\..\playlists\playlistsview.h"
#include "..\..\playlists\playlistview.h"
(or something similar, depending on where the ui_mainwindow.h file is located in
relation to the two files it is looking for).
Section 12.4: QVariant Class (properties.pro)
This application uses Qts testing capabilities to check customer properties. The output is
either displayed in the Application Output window in the bottom pane of Qt Creator, or (if
Run in terminal is checked) it will display in a console window.
Section 12.6: MetaTypes
The src\metatype folder contains the fraction.h and metatype.cpp files needed for
this example. To run this example, you will need to set up your own console project and
add these two files to it. You need to make some changes before the program will compile
and run:
In metatype.pro: make sure that QT += gui is included
In fraction.h: Comment out the three overloaded << and >> operator functions
below the class (or implement them yourself).
Page 1 of 16

In metatype.cpp:There is a ; missing at the end of the qDebug() statement in


the middle of main().
Check the Run in terminal checkbox.

CHAPTER 13 (Models and Views)


Section 13.2.3: Delegate Classes (stardelegate.pro)
The example (figure 13.8) refers to $QTDIR/examples/modelview/stardelegate.
This should be C:\Qt\2010.04\qt\examples\itemviews\stardelegate.
It
produces the following output. Figure 13.8 is a modified version of the example.

Section 13.3: Table models (p414)


Use the files in the folder shortcutmodel-standarditem to test the Examples13.10 to
13.12.
Section 13.4: Tree Models (objectbrowser.pro)
The code supplied does not work correctly.
Comment out the two lines in
ObjectBrowserModel.cpp as indicated below, and the program should then work as
indicated.
QModelIndex ObjectBrowserModel::
index(int row, int col, const QModelIndex& parent) const {
//if ((row < 0) || (col < 0) || row >= rowCount() ||
//col >= columnCount()) return QModelIndex();
return createIndex( row, col, qObject(parent) );
}

// comment out
// comment out

Section 13.4.1: Trolltech Model Testing Tool


You can simply read through this section; the programme does not work correctly in a
Windows environment (and there are also problems with getting a degugger to work in a
Windows environment).

Page 2 of 16

CHAPTER 14 (Validation and Regular Expressions)


Section 14.3 Regular Expressions
At the bottom of page 443 there is a reference to a regular expression tester from Nokia in
the src/regex-tester folder. Note that although the folder is there, there is no code for you
to use.
However, you can find the code in your Qt installation folder in
examples\tools\regexp.
Section 14.3: Regular Expressions (regexp.pro)
There are two little things you need to do to get this example to work, both in the
regexp.pro file.
Comment out the include (../common.pri) line
Add CONFIG += console
CHAPTER 15 (Parsing XML)
Note that XML files can be set up in 2 different ways: one format uses tags (like book) with
attributes (like title and pages)
<library>
<book title=Computer Algorithms pages=688 />
<book title=C++ unleashed pages=918 />
</library>
or, using tags (book, title, and pages) with text (like Computer Algorithms and
688)
<library>
<book>
<title>Computer Algorithms</title>
<pages>688</pages>
</book>
<book>
<title> C++ unleashed</title>
<pages>918</pages>
</book>
</library>
An element is anything from the start tag to the end tag of an element. So, library is an
element, as is pages. Strictly speaking, attributes are supposed to supply extra
information that is not part of the data (or text). However, there are no rules about when to
use attributes and when to use elements. Some argue that using attributes is more
limiting, and that using elements is more extensible. Note also that the indenting of the
text above is simply to make it more readable.
Further, taking the tree (or hierarchical) structure of XML seriously, an XML document
must contain a root element, although it does not matter what this elements name is.

Page 3 of 16

<root>
....
</root>
This root then has branches that are called children.
<root>
<child>
<sub_child>
...
</sub_child>
<sub_child>
...
</sub_child>
</child>
<child>
...
</child>
</root>
Section 15.2: SAX Parsing (sax1.pro)
There are several things you need to note about running this example (which should run
as given).
Add the line CONFIG += console to the sax1.pro file.
Note that the xml file that you want to parse should maybe be placed in the builddesktop folder instead of in its debug folder it depends on which folders Qt will
use as its default folder.
You can place several xml files here, and then list them in the Arguments edit box in
Run Settings remember to type at least one name (the samplefile.xml file, for
example) in the Arguments box if you want the program to run.
If you are getting slightly garbled output, then use the samplefile.xml that is
supplied with this document in the Additional Resources folder (in the Chapter
additional notes folder) there is sometimes a problem with the new line and
carriage return characters.
The example given here simply parses the xml file and displays it. Suppose you had a
Person class that had a name and an age. An xml file for this could look as follows:
<people>
<person name=Xin age=12 />
<person name=Ndou age=21 />
</people>
You thus know the structure of the xml file. If you now wanted to parse this file and then
create instances of your Person class, you could do something along the lines of the
following:
bool StockReader::startDocument()
{
return true;
}
Page 4 of 16

bool StockReader::startElement(const QString &namespaceURI, const


QString &localName, const QString &qName, const QXmlAttributes
&atts)
Check the name of the
{
tag
if (qName == "person")
{
We know that its first
QString name = atts.value(0);
attribute is the name, and
int age = atts.value(1).toInt();
the second the age
p.setName(name);
p.setAge(age);
Set values in an object
}
return true;
}
bool StockReader::endElement(const QString
QString &localName, const QString &qName)
{
return true;
}

&namespaceURI,

const

bool StockReader::endDocument()
{
return true;
}
You would obviously have to have some way of dealing with p, the Person object that
was created, and that would depend on your individual program.
However, if the xml file is not based on attributes but on text, we could have the following:
<people>
<person>
<name>Xin</name>
<age>12</age>
</person>
<person>
<name>Ndou</name>
<age>21</age>
</person>
</people>
Parsing this is different, and would involve something along the lines of the following. The
variables used in the example (inName and inAge) would need to be members of the
XMLHandler class.
bool XMLHandler::startDocument()
{
inName = false;
inAge = false;
return true;
}

Set flags so that we know


which tag we are in

Page 5 of 16

bool XMLHandler::startElement(const QString &namespaceURI, const


QString &localName, const QString &qName, const QXmlAttributes
&atts)
{
if (qName == "name") inName = true;
Set the appropriate flag
if (qName == "age") inAge = true;
return true;
}
bool XMLHandler::characters(const QString& text)
{
if (inName)
Handle the text between
{
the opening and closing
objectList.append(text);
tag, and reset the flag
inName = false;
}
if (inAge)
{
objectList.append(text);
inAge = false;
}
return true;
}
bool XMLHandler::endElement(const QString &namespaceURI, const
QString &localName, const QString &qName)
{
Once we are back at the
if (qName == "person")
closing
tag for person, we
{
have all the detail we need
p.setName(objectList[0]);
to instantiate an object
p.setAge(objectList[1].toInt());
pl.append(p);
objectList.clear();
}
return true;
}
bool XMLHandler::endDocument()
{
return true;
}
Section 15.3.1: DOM Tree Walking (domwalker.pro)
Remember to put the samplefile.xml file in the build-desktop folder so that the
example can read this file. Otherwise this code should run as explained in Ezust2. This
example uses a form of the classic Visitor pattern, but as you are not expected to know
this design pattern, this will not be discussed further.

Page 6 of 16

Section 15.3.2: Generation of XML with DOM (dombuilder.pro)


Note the comment at the top of the page: it is better to generate XML documents using an
API than hard coding the output to ensure that the output is parsable. Thus, while you can
do this
toFile
toFile
toFile
toFile

<<
<<
<<
<<

"<people>" << endl;


" <person name=\"" << name << "\" age=\"" << age;
"\" />" << endl;
"</people>" << endl;

it is better to use the more complicated DOM approach.


The example provided should work as described. Example 15.15 refers to folder
src/libs/docbook/ this can be found in the c:\projects\libs\docbook folder, and you can
find the files referred to there.
Example 15.14 is a Linux alternative. XSL stands for EXtensible Stylesheet Language,
and XSLT stands for XSL Transformations transforming XML (whose tags are not well
understood by browsers) documents to other formats like XHTML (whose tags are well
understood by browsers). Windows has its own command line XSLT processor, named
msxsl.exe. This file can be downloaded from http://www.microsoft.com/download/en/
details.aspx?displaylang=en&id=21714. As you can see from the screenshot below, to
use msxsl you need to give the name of the source file and the name of the stylesheet that
defines the styles.
If you would like to learn XSLT, look at the tutorial at
http://www.w3schools.com/xsl/.

More on generation of XML with DOM


If you wanted to manually construct XML from an object, you can use the following outline.
Say we wanted to create the following XML document.
<people>
<person>
<name>Xin</name>
<age>12</age>
</person>
<person>
<name>Ndou</name>
<age>21</age>
</person>
</people>

Root element
1st child
2nd child

Page 7 of 16

You will need to create the document and a root.

Create document

Create root element


QDomDocument doc;
QDomElement rootElement = doc.createElement("people");
doc.appendChild(rootElement);
Add root to document
Then, for each child, you would create it as follows.
QDomElement personElement = doc.createElement("person");
rootElement.appendChild(personElement);

Create child
element and
add it to the
root

Create name element


QDomElement nameElement = doc.createElement("name");
and add it to person
personElement.appendChild(nameElement);
QDomText nameText = doc.createTextNode(name);
Create the text and add
nameElement.appendChild(nameText);
it to the name element
QDomElement ageElement = doc.createElement("age");
personElement.appendChild(ageElement);
QDomText ageText = doc.createTextNode(age [as a string]);
Now do the
ageElement.appendChild(ageText);
same for age
This then creates the XML document in a manner that you can be certain will be parsable.
Tutorial
There is a very nice, concise introduction to SAX and DOM XML parsing in Qt at
http://www.digitalfanatics.org/projects/qt_tutorial/chapter09.html. Although it is part of a
larger tutorial, it is clear enough to follow. One could, of course, go back to some of the
previous tutorials and build the address book application that is used.
CHAPTER 16 (More Design Patterns)
Some specific issues relating to the content in Ezust2 will be given here. At the end of
these notes you will find extra information on many of the design patterns that you are
expected to be familiar with.
Section 16.1: Creational Patterns
Ezust2 discuss both the Factory Method and Abstract Factory design patterns. Watch the
videos on these patterns on myUnisa, and then make sure that you can describe the
differences between these patterns beyond what is given in Ezust2.
Section 16.1.4: Polymorphism from Constructors
You will need to create your own project to test this little program. Remember to check the
Run in terminal box on the Run Settings tab. Otherwise, the program should run as
described in Ezust2.
Page 8 of 16

Section 16.2: Memento Pattern


According to the Gang of Four (E Gamma, R Helm, R Johnson & JM Vlissides. Design
Patterns: Elements of Reusable Object-Oriented Software), a memento is an object that
stores a snapshot of the internal state of another object, and this state can only be read
by the original object. This allows an objects state to be restored at some later point,
allowing for an undo or a rollback operation. What we have in this section is not the
memento pattern, but rather an implementation of the serializer pattern.
The Gang of Four do not include the serializer pattern in their set of patterns, although
other authors do. Serializer is used to stream objects into other data structures (writing),
and is also responsible for restoring objects from such a data structure (reading). It is
commonly used to write and restore the state an object to a file or database, allowing for
storage and retrieval. Go back and have a look in your study material for COS2614 (see
also section 7.4.1 on page 249).
A comment here about QObjectReader and QObjectWriter. These follow the idea of
a serializer well in that neither knows anything about the structure of the objects passed to
it, and serializes the object as it finds it.
The StockControl project is a running version of an application that uses the
QObjectWriter and QObjectReader classes.
CHAPTER 17 (Concurrency)
Section 17.1: QProcess
logRandom.pro)

and

Process

Control

(randomNumbers.pro

and

The LogTail project as included in Ezust2 does not produce the desired output on
Windows because the command tail is only applicable on a UNIX variant operating
system.
A new example is provided in the Chapter additional notes folder (in the QProcess and
process control folder) to demonstrate starting, controlling, and communicating with other
processes similar to LogTail. To test the new project, follow the instructions below.
Firstly, open, build, and run the randomNumbers project so that randomNumbers.exe
is generated in the debug folder of the project. This project should display 10 random
numbers. You can now close this project.
Secondly, open and build the logRandom project. Before you execute the project, copy
and paste randomNumbers.exe from the debug folder of the randomNumbers project
RandomNumbers into the debug folder of the logRandom project. Run the project to
view the output.
Section 17.1.1: Processes and Environment (environment.pro)
A modified version of this example, which works exactly as described in Ezust2 is provided
in the QProcess and process control\environment folder. Test the project using the
command line argument f NULL. The only difference is that the Windows version uses
Page 9 of 16

putenv rather than setenv, and different environment variables are used (ones that are
known in a Windows environment).
Section 17.1.2: Qonsole (qonsole1.pro)
This program will work as discussed in Ezust2. However, you will need to create your own
project file. In a Windows environment, you can use commands like
dir (to list directory contents)
cd (to change to another directory)
md (to make a directory)
cls (to clear the screen)
date (to get the current date)
help (to get help)
to check that the example is working. Please remember to type in the command exit
before you close the window to exit the command line process.

ADDITIONAL NOTES ON DESIGN PATTERNS


Creational Patterns
(1)

Factory Method Pattern

This design pattern is applicable when a class needs to create objects but doesnt know
which objects to create and the responsibility for creating objects can be transferred to the
subclasses. So, this design pattern defines a class with a function for creating an object
but the subclasses implement this function to create appropriate objects.
The UML structure of this design pattern is given below:

The main classes included in the UML diagram are:


Product: defines the interface of the objects to be created by the Factory
ConcreteProduct: implements the Product interface, which is created by the
ConcreteFactory
Factory: declares the factoryMethod(), which returns an object of type Product
ConcreteFactory: overrides the factoryMethod() to return an instance of type
ConcreteProduct

Page 10 of 16

There are a number of variations of this design pattern. Firstly, the FactoryMethod() in
Factory may provide a default implementation of the factory method and it may call the
FactoryMethod in one of its functions (operation()). This means the Factory is not
always necessarily abstract. Secondly, the factory method could take parameters and
based on them, several different objects can be created.
This means a
ConcreteFactory may create more than one type of object.
(2)

Abstract Factory Pattern

This design pattern is applicable when you need to create families of related or dependent
objects. This design pattern mainly has two hierarchies
(1) hierarchies of product classes, whose objects are meant to be used together
(2) then you have a hierarchy of factory classes to facilitate creation of the product
classes.
The UML structure of the design pattern is given below:

The main classes included in the UML diagram are:


AbstractProductA and AbstractProductB: abstract classes of two related product
classes
ProductA1 and ProductA2: concrete product A classes
ProductB1 and ProductB2: concrete product B classes
AbstractFactory: abstract factory class that declares functions for creating product A
and B objects
ConcreteFactory1 and ConcreteFactory2: concrete factories, which implement
functions declared in the AbstractFactory to create concrete product A and B classes.
Client: it only uses the functions declared in AbstractProductA,
AbstractProductB and AbstractFactory.
Using this design pattern, the client does not know about the implementation details of the
concrete product families and it can interchangeably use any of the concrete factories to
create the appropriate concrete products.
(3)

Singleton

This design pattern is applicable when you want to have a class which has only one
instance and this single instance can be accessed via a global access point. This is
Page 11 of 16

generally achieved by making the constructor of the class private, creating one
instance, and a function that can be used to return the single instance.
Since there is only one class involved, a UML diagram is not include here. Instead, one
way of implementing Singleton is demonstrated below:
Class declaration
class A{
public:
static A* getInstance();
private:
A();
static A* onlyInstance
};
Class definition
A* A::onlyInstance = NULL;
A::A(){}
A* A::getInstance(){
if(onlyInstance == 0)
onlyInstance = new A();
return onlyInstance;
}
As demonstrated in the code, the single instance of A is stored in the static variable
onlyInstance. When a request for an A instance is made via getInstance(), an
instance of A is created and stored in onlyInstance. All subsequent calls to
getInstance() do not create new instances but rather return the only instance of A
stored in onlyInstance.
Behavioural Patterns
(1)

Memento Pattern

This design pattern is applicable when the state of an object needs to be saved so that its
state can be restored later without violating the encapsulation of the class. The data
members of the class, which determine the state of its objects, may not all be accessible
(via setters and getters) outside the class. Hence the class itself has to be involved in
saving and restoring the state of its objects.
The UML structure of the design pattern is given below:

Page 12 of 16

The main classes involved in the UML diagram are:


Originator: State of the objects of this class are being saved and restored. To save the
state of an object it creates a snapshot of its state by creating a Memento, via the
createMemento() function. To restore the state of the object, it uses a Memento via the
setMemento() function.
Memento: stores the state of an Originator.
Caretaker: it keeps the Memento.
The role of the Caretaker is only to store the Memento and it is not allowed to access
the state of the Originator saved in the Memento. One way of achieving this is by
making the getState(), setState(),and even the constructor of Memento, private
so that the Caretaker cannot access the state of the Originator. However
Orginator needs to be able to access the constructor, getState(), and setState()
of Memento. One way of achieving this is by making Originator a friend of
Memento.
Caretaker requests a snapshot of an Originator using the createMemento()
function. The Originator in turn creates a snapshot of its state by passing its state to
Memento and thus creating a Memento, which is then passed to the Caretaker. If the
state of the Originator has to be restored, then Caretaker can invoke the
setMemento() function to restore the state of the Originator.
(2)

Strategy pattern

The Strategy pattern is applicable when there is a family of algorithms that needs to be
made interchangeable based on the context. Using this design pattern each algorithm is
encapsulated in a class and the run time selection of a relevant algorithm is made
possible.
The UML diagram of this design pattern is given below:

Page 13 of 16

The classes involved in this UML diagram are:


Strategy: It is the class which defines a common interface (here the function
Algorithm()) for the classes that represents the family of algorithms. It is generally an
abstract class and sometimes referred to as an abstract strategy
StrategyOne, StrategyTwo and StrategyThree: These are the concrete strategies,
which defines the Algorithm() function. These classes are sometimes referred to as
concrete strategies
Context: Context has a reference to a Strategy and it can decide on the concrete
strategy to choose based on the requirement.
Structural Patterns
(1)

Faade Pattern

The Faade pattern is applicable when you would like to provide a simplified interface to a
complicated set of systems but yet allowing the client to access the functionality of the
underlying system. This design pattern defines a Faade that interacts and invokes the
subsystems to satisfy the request of the client.
A simplified diagram of this design pattern is given below:

(2)

Adapter/Wrapper Pattern

This design pattern is applicable when the functionality of a class can be reused but the
interface of the class is not compatible with the existing classes. So this design pattern
converts the interface of an existing class into another as expected by the client. This
Page 14 of 16

design pattern is sometimes called a Wrapper pattern since it can be seen as wrapping
around an existing class to present a different interface to other classes.
There are two forms (Class Adapter and Object Adapter) of the Adapter pattern and the
UML diagrams of both forms of the Adapter pattern are given below:
Class Adapter:

Object Adapter:

The main classes included in the UML diagram are:


Client: uses objects conforming to the Target interface
Target: defines the domain-specific interface used by the Client
Adaptee: defines the interface of an existing class to be reused
Adapter: adapts the interface of Adaptee to the Target interface
In a Class Adapter, Adapter adapts the interface of Adaptee by inheriting from
Adaptee.
In the Request() function of the Adapter, it simply invokes the
SpecificRequest() function in Adaptee. Here the client call operates on an Adapter
instance, which in turn calls Adaptee operations (SpecificRequest(), for instance)
which carry out the operation.
In an Object Adapter, Adapter adapts the interface of Adaptee by creating an instance
of Adaptee (adaptee). In the Request() function of the Adapter, it simply invokes the
SpecificRequest() function on the instance of Adaptee. Here the client call operates
on an Adapter instance, which in turn calls operations on the Adaptee instance to carry
out the operation.

Page 15 of 16

References:
E Gamma, R Helm, R Johnson & JM Vlissides (1995). Design Patterns: Elements of
Reusable Object-Oriented Software. Addision-Wesley
Lasater CG (2007). Design Patterns. Wordware Applications Library

Page 16 of 16

You might also like