A Guide To Programming In C++

User Manual:

Open the PDF directly: View PDF .
Page Count: 846

Download
Open PDF In Browser	View PDF

A Complete Guide to
Programming in C++

Ulla Kirch-Prinz
Peter Prinz

JONES AND BARTLETT PUBLISHERS

A Complete Guide to
Programming in C++
Ulla Kirch-Prinz
Peter Prinz

World Headquarters
Jones and Bartlett Publishers
40 Tall Pine Drive
Sudbury, MA 01776
978-443-5000
info@jbpub.com
www.jbpub.com

Jones and Bartlett Publishers
Canada
2406 Nikanna Road
Mississauga, ON L5C 2W6
CANADA

Jones and Bartlett Publishers
International
Barb House, Barb Mews
London W6 7PA
UK

Copyright © 2002 by Jones and Bartlett Publishers, Inc.
All rights reserved. No part of the material protected by this copyright notice may be reproduced or utilized in
any form, electronic or mechanical, including photocopying, recording, or any information storage or retrieval
system, without written permission from the copyright owner.
Cover Image: Stones on shore-line and yellow leaf, Bjorkliden, Sweden, by Peter Lilja
Library of Congress Cataloging-in-Publication Data
Prinz, Peter.
[C++ Lernen und professionell anwenden. English]
A complete guide to programming in C++ / Peter Prinz, Ulla Kirch-Prinz; translated by Ian Travis.
p. cm.
ISBN: 0-7637-1817-3
1. C++ (Computer program language) I. Kirch-Prinz, Ulla. II. Title.
QA76.73.C153 P73713 2001
005.13'3—dc21
2090

2001029617

Chief Executive Officer: Clayton Jones
Chief Operating Officer: Don W. Jones, Jr.
V.P., Managing Editor: Judith H. Hauck
V.P., Design and Production: Anne Spencer
V.P., Manufacturing and Inventory Control: Therese Bräuer
Editor-in-Chief: Michael Stranz
Development and Product Manager: Amy Rose
Marketing Manager: Nathan Schultz
Production Assistant: Tara McCormick
Cover Design: Night & Day Design
Composition: Northeast Compositors
Text Design: Mary McKeon
Printing and Binding: Courier Westford
Cover printing: John Pow Company, Inc.
This book was typeset in QuarkXpress 4.11 on a Macintosh G4. The font families used were Goudy, Gill Sans,
Courier, Rubino Serif, and Seven Sans. The first printing was printed on 50 lb. Finch Opaque.
Printed in the United States of America
05 04 03 02 01

10 9 8 7 6 5 4 3 2 1

Dedicated to our children, Vivi and Jeany

This page intentionally left blank

preface

This book was written for readers interested in learning the C++ programming
language from scratch, and for both novice and advanced C++ programmers
wishing to enhance their knowledge of C++. It was our goal from the beginning to design this text with the capabilities of serving dual markets, as a textbook for students and as a holistic reference manual for professionals.
The C++ language definition is based on the American National Standards Institute ANSI Standard X3J16. This standard also complies with ISO
norm 14882, which was ratified by the International Standardization Organization in 1998. The C++ programming language is thus platform-independent
in the main with a majority of C++ compilers providing ANSI support. New
elements of the C++ language, such as exception handling and templates, are
supported by most of the major compilers. Visit the Jones and Bartlett web site
at www.jbpub.com for a listing of compilers available for this text.
The chapters in this book are organized to guide the reader from elementary language concepts to professional software development, with in-depth
coverage of all the C++ language elements en route. The order in which these
elements are discussed reflects our goal of helping the reader to create useful
programs at every step of the way.

v

vi

■

PREFACE

Each double-page spread in the book is organized to provide a description of the language elements on the right-hand page while illustrating them by means of graphics and
sample programs on the left-hand page. This type of visual representation offered by each
spread will provide students and professionals with an unmatched guide throughout the
text. The sample programs were chosen to illustrate a typical application for each language element. In addition, filter programs and case studies introduce the reader to a
wide range of application scenarios.
To gain command over a programming language, students need a lot of experience in
developing programs. Thus, each chapter includes exercises followed by sample solutions, allowing the reader to test and enhance his or her performance and understanding
of C++.
The appendix provides further useful information, such as binary number representation, pre-processor directives, and operator precedence tables, making this book a wellstructured and intelligible reference guide for C++ programmers.
In order to test and expand your acquired knowledge, you can download sample programs and solutions to the exercises at:
http://completecpp.jbpub.com

Content Organization
Chapter 1 gives a thorough description of the fundamental characteristics of the objectoriented C++ programming language. In addition, students are introduced to the steps
necessary for creating a fully functional C++ program. Many examples are provided to
help enforce these steps and to demonstrate the basic structure of a C++ program.
Chapter 2 provides a complete introduction to the basic types and objects used by
C++ programs. Integral types and constants, fundamental types, and Boolean constants
are just a few of the topics discussed.
Chapter 3 describes how to declare and call standard functions. This chapter also
teaches students to use standard classes, including standard header files. In addition, students work with string variables for the first time in this chapter.
Chapter 4 explains the use of streams for input and output, with a focus on formatting
techniques. Formatting flags and manipulators are discussed, as are field width, fill characters, and alignment.
Chapter 5 introduces operators needed for calculations and selections. Binary, unary,
relational, and logical operators are all examined in detail.
Chapter 6 describes the statements needed to control the flow of a program. These
include loops with while, do-while, and for; selections with if-else, switch, and the conditional operator; and jumps with goto, continue, and break.
Chapter 7 provides a thorough introduction to the definition of symbolic constants
and macros, illustrating their significance and use. Furthermore, a comprehensive examination of standard macros for character handling is included.
Chapter 8 introduces implicit type conversions, which are performed in C++ whenever different arithmetic types occur in expressions. Additionally, the chapter explores
an operator for explicit type conversion.

PREFACE

■

vii

Chapter 9 takes an in-depth look at the standard class string, which is used to represent strings. In addition to defining strings, the chapter looks at the various methods of
string manipulation. These include inserting and erasing, searching and replacing, comparing, and concatenating strings.
Chapter 10 describes how to write functions of your own. The basic rules are covered,
as are passing arguments, the definition of inline functions, overloading functions and
default arguments, and the principle of recursion.
Chapter 11 gives a thorough explanation of storage classes for objects and functions.
Object lifetime and scope are discussed, along with global, static, and auto objects.
Namespaces and external and static functions are also included in the discussion.
Chapter 12 explains how to define references and pointers and how to use them as
parameters and/or return values of functions. In this context, passing by reference and
read-only access to arguments are introduced.
Chapter 13 provides a complete description of how classes are defined and how
instances of classes, or objects, are used. In addition, structs and unions are introduced as
examples of special classes.
Chapter 14 describes how constructors and destructors are defined to create and
destroy objects. Also discussed are how inline methods, access methods, and read-only
methods can be used. Furthermore, the chapter explains the pointer this, which is available for all methods, and what you need to pay attention to when passing objects as arguments or returning objects.
Chapter 15 gives a complete explanation of member objects and how they are initialized, and of data members that are created only once for all the objects in a class. In addition, this chapter describes constant members and enumerated types.
Chapter 16 takes an in-depth look at how to define and use arrays. Of particular interest are one-dimensional and multidimensional arrays, C strings, and class arrays.
Chapter 17 describes the relationship between pointers and arrays. This includes
pointer arithmetic, pointer versions of functions, pointers as return values and read-only
pointers, and pointer arrays. Students learn that operations that use C strings illustrate
how to use pointers for efficient programming, and that string access via the command
line of an application program is used to illustrate pointer arrays.
Chapter 18 explains sequential file access using file streams. Students will develop an
understanding of how file streams provide simple and portable file handling techniques.
Chapter 19 provides a complete description of the various uses of overloaded operators. Arithmetic operators, comparisons, the subscript operator, and the shift operators
for input and output are overloaded to illustrate the appropriate techniques. In addition,
the concept of friend functions, which is introduced in this context, is particularly
important for overloading operators. Students learn how overloading operators allows
them to apply existing operators to objects of class type.
Chapter 20 discusses how implicit type conversion occurs in C++ when an expression
cannot be compiled directly but can be compiled after applying a conversion rule. The
programmer can stipulate how the compiler will perform implicit type conversion for
classes by defining conversion constructors and functions. Finally, the chapter discusses
ambiguity that occurs due to type conversion and how to avoid it.

viii

■

PREFACE

Chapter 21 describes how a program can allocate and release memory dynamically in
line with current memory requirements. Dynamic memory allocation is an important factor in many C++ programs, and the following chapters contain several case studies to
help students review the subject.
Chapter 22 explains how to implement classes containing pointers to dynamically
allocated memory. These include your own copy constructor definition and overloading
the assignment operator. A class designed to represent arrays of any given length is used
as a sample application.
Chapter 23 provides a thorough description of how derived classes can be constructed
from existing classes by inheritance. In addition to defining derived classes, this chapter
discusses how members are redefined, how objects are constructed and destroyed, and
how access control to base classes can be realized.
Chapter 24 discusses implicit type conversion within class hierarchies, which occurs
in the context of assignments and function calls. Explicit type casting in class hierarchies is also described, paying particular attention to upcasting and downcasting.
Chapter 25 gives a complete explanation of how to develop and manage polymorphic
classes. In addition to defining virtual functions, dynamic downcasting in polymorphic
class hierarchies is introduced.
Chapter 26 describes how defining pure virtual methods can create abstract classes
and how you can use abstract classes at a polymorphic interface for derived classes. To
illustrate this, an inhomogeneous list, that is, a linked list whose elements can be of various class types, is implemented.
Chapter 27 describes how new classes are created by multiple inheritance and
explains their uses. Besides introducing students to the creation and destruction of
objects in multiply-derived classes, virtual base classes are depicted to avoid ambiguity in
multiple inheritance.
Chapter 28 explains how a C++ program uses error-handling techniques to resolve
error conditions. In addition to throwing and catching exceptions, the chapter also
examines how exception specifications are declared and exception classes are defined. In
addition, the use of standard exception classes is discussed.
Chapter 29 examines random access to files based on file streams, and options for
querying file state. Exception handling for files is discussed as well. The chapter illustrates how to make objects in polymorphic classes persistent, that is, how to save them in
files. The applications introduced in this chapter include simple index files and hash
tables.
Chapter 30 provides a thorough explanation of the advanced uses of pointers. These
include pointers to pointers, functions with a variable number of arguments, and pointers
to functions. In addition, an application that defines a class used to represent dynamic
matrices is introduced.
Chapter 31 describes bitwise operators and how to use bit masks. The applications
included demonstrate calculations with parity bits, conversion of lowercase and capital
letters, and converting binary numbers. Finally, the definition of bit-fields is introduced.
Chapter 32 discusses how to define and use function and class templates. In addition,
special options, such as default arguments, specialization, and explicit instantiation, are

PREFACE

■

ix

discussed. Students learn that templates allow the construction of functions and classes
based on types that have not yet been stated. Thus, templates are a powerful tool for
automating program code generation.
Chapter 33 explains standard class templates used to represent containers for more
efficient management of object collections. These include sequences, such as lists and
double ended queues; container adapters, such as stacks, queues, and priority queues;
associative containers, such as sets and maps; and bitsets. In addition to discussing how
to manage containers, the chapter also looks at sample applications, such as bitmaps for
raster images, and routing techniques.

Additional Features
Chapter Goals A concise chapter introduction, which contains a description of the
chapter’s contents, is presented at the beginning of each chapter. These summaries also
provide students with an idea of the key points to look for throughout the chapter.

Chapter Exercises Each chapter contains exercises, including programming problems,
designed to test students’ knowledge and understanding of the main ideas. The exercises
also provide reinforcement for key chapter concepts. Solutions are included to allow
students to check their work immediately and correct any possible mistakes.

Case Studies Every chapter contains a number of case studies that were designed to
introduce the reader to a wide range of application scenarios.

Notes This feature provides students with helpful tips and information useful to learning
C++. Important concepts and rules are highlighted for additional emphasis and easy
access.

Hints These are informative suggestions for easier programming. Also included are
common mistakes and how to avoid making them.

Acknowledgements
Our thanks go out to everyone who helped produce this book, particularly to
Ian Travis, for his valuable contributions to the development of this book.
Alexa Doehring, who reviewed all samples and program listings, and gave many valuable
hints from the American perspective.
Michael Stranz and Amy Rose at Jones and Bartlett Publishers, who managed the publishing agreement and the production process so smoothly.
Our children, Vivi and Jeany, who left us in peace long enough to get things finished!
And now all that remains is to wish you, Dear Reader, lots of fun with C++!
Ulla Kirch-Prinz
Peter Prinz

This page intentionally left blank

contents

Chapter 1

Fundamentals

1

Development and Properties of C++ 2
Object-Oriented Programming 4
Developing a C++ Program 6
A Beginner’s C++ Program 8
Structure of Simple C++ Programs 10
Exercises 12
Solutions 14

Chapter 2

Fundamental Types, Constants, and Variables
Fundamental Types 16
Constants 22
Escape Sequences 26
Names 28
Variables 30
The Keywords const and volatile
Exercises 34
Solutions 36

15

32

xi

xii

■

CONTENTS

Chapter 3

Using Functions and Classes

39

Declaring Functions 40
Function Calls 42
Type void for Functions 44
Header Files 46
Standard Header Files 48
Using Standard Classes 50
Exercises 52
Solutions 54

Chapter 4

Input and Output with Streams

57

Streams 58
Formatting and Manipulators 60
Formatted Output of Integers 62
Formatted Output of Floating-Point Numbers 64
Output in Fields 66
Output of Characters, Strings, and Boolean Values
Formatted Input 70
Formatted Input of Numbers 72
Unformatted Input/Output 74
Exercises 76
Solutions 78

Chapter 5

Operators for Fundamental Types

81

Binary Arithmetic Operators 82
Unary Arithmetic Operators 84
Assignments 86
Relational Operators 88
Logical Operators 90
Exercises 92
Solutions 94

Chapter 6

Control Flow 95
The while Statement 96
The for Statement 98
The do-while Statement 102
Selections with if-else 104
Else-if Chains 106
Conditional Expressions 108
Selecting with switch 110
Jumps with break, continue, and goto
Exercises 114
Solutions 116

112

68

CONTENTS

Chapter 7

Symbolic Constants and Macros

119

Macros 120
Macros with Parameters 122
Working with the #define Directive 124
Conditional Inclusion 126
Standard Macros for Character Manipulation 128
Redirecting Standard Input and Output 130
Exercises 132
Solutions 134

Chapter 8

Converting Arithmetic Types

139

Implicit Type Conversions 140
Performing Usual Arithmetic Type Conversions 142
Implicit Type Conversions in Assignments 144
More Type Conversions 146
Exercises 148
Solutions 150

Chapter 9

The Standard Class string

153

Defining and Assigning Strings 154
Concatenating Strings 156
Comparing Strings 158
Inserting and Erasing in Strings 160
Searching and Replacing in Strings 162
Accessing Characters in Strings 164
Exercises 166
Solutions 168

Chapter 10

Functions

171

Significance of Functions in C++
Defining Functions 174
Return Value of Functions 176
Passing Arguments 178
Inline Functions 180
Default Arguments 182
Overloading Functions 184
Recursive Functions 186
Exercises 188
Solutions 191

Chapter 11

172

Storage Classes and Namespaces
Storage Classes of Objects 198
The Storage Class extern 200

197

■

xiii

xiv

■

CONTENTS

The Storage Class static 202
The Specifiers auto and register 204
The Storage Classes of Functions 206
Namespaces 208
The Keyword using 210
Exercises 212
Solutions 216

Chapter 12

References and Pointers

221

Defining References 222
References as Parameters 224
References as Return Value 226
Expressions with Reference Type 228
Defining Pointers 230
The Indirection Operator 232
Pointers as Parameters 234
Exercises 236
Solutions 238

Chapter 13

Defining Classes

243

The Class Concept 244
Defining Classes 246
Defining Methods 248
Defining Objects 250
Using Objects 252
Pointers to Objects 254
Structs 256
Unions 258
Exercise 260
Solution 262

Chapter 14

Methods

265

Constructors 266
Constructor Calls 268
Destructors 270
Inline Methods 272
Access Methods 274
const Objects and Methods 276
Standard Methods 278
this Pointer 280
Passing Objects as Arguments 282
Returning Objects 284
Exercises 286
Solutions 290

CONTENTS

Chapter 15

Member Objects and Static Members

297

Member Objects 298
Member Initializers 300
Constant Member Objects 302
Static Data Members 304
Accessing Static Data Members 306
Enumeration 308
Exercises 310
Solutions 314

Chapter 16

Arrays

321

Defining Arrays 322
Initializing Arrays 324
Arrays 326
Class Arrays 328
Multidimensional Arrays
Member Arrays 332
Exercises 334
Solutions 338

Chapter 17

Arrays and Pointers

330

349

Arrays and Pointers (1) 350
Arrays and Pointers (2) 352
Pointer Arithmetic 354
Arrays as Arguments 356
Pointer Versions of Functions 358
Read-Only Pointers 360
Returning Pointers 362
Arrays of Pointers 364
Command Line Arguments 366
Exercises 368
Solutions 372

Chapter 18

Fundamentals of File Input and Output
Files 380
File Streams 382
Creating File Streams 384
Open Modes 386
Closing Files 388
Reading and Writing Blocks
Object Persistence 392
Exercises 394
Solutions 398

390

379

■

xv

xvi

■

CONTENTS

Chapter 19

Overloading Operators

411

Generals 412
Operator Functions (1) 414
Operator Functions (2) 416
Using Overloaded Operators 418
Global Operator Functions 420
Friend Functions 422
Friend Classes 424
Overloading Subscript Operators 426
Overloading Shift-Operators for I/O 428
Exercises 430
Solutions 432

Chapter 20

Type Conversion for Classes
Conversion Constructors 442
Conversion Functions 444
Ambiguities of Type Conversions
Exercise 448
Solution 450

Chapter 21

Dynamic Memory Allocation

441

446

453

The Operator new 454
The Operator delete 456
Dynamic Storage Allocation for Classes 458
Dynamic Storage Allocation for Arrays 460
Application: Linked Lists 462
Representing a Linked List 464
Exercises 466
Solutions 468

Chapter 22

Dynamic Members

477

Members of Varying Length 478
Classes with a Dynamic Member 480
Creating and Destroying Objects 482
Implementing Methods 484
Copy Constructor 486
Assignment 488
Exercises 490
Solutions 492

Chapter 23

Inheritance

499

Concept of Inheritance
Derived Classes 502

500

CONTENTS

Members of Derived Classes 504
Member Access 506
Redefining Members 508
Constructing and Destroying Derived Classes
Objects of Derived Classes 512
Protected Members 514
Exercises 516
Solutions 520

Chapter 24

Type Conversion in Class Hierarchies

510

529

Converting to Base Classes 530
Type Conversions and Assignments 532
Converting References and Pointers 534
Explicit Type Conversions 536
Exercises 538
Solutions 540

Chapter 25

Polymorphism

543

Concept of Polymorphism 544
Virtual Methods 546
Destroying Dynamically Allocated Objects
Virtual Method Table 550
Dynamic Casts 552
Exercises 554
Solutions 558

Chapter 26

Abstract Classes

548

565

Pure Virtual Methods 566
Abstract and Concrete Classes 568
Pointers and References to Abstract Classes 570
Virtual Assignment 572
Application: Inhomogeneous Lists 574
Implementing an Inhomogeneous List 576
Exercises 578
Solutions 580

Chapter 27

Multiple Inheritance

587

Multiply-Derived Classes 588
Multiple Indirect Base Classes 590
Virtual Base Classes 592
Constructor Calls 594
Initializing Virtual Base Classes 596
Exercises 598
Solutions 602

■

xvii

xviii

■

CONTENTS

Chapter 28

Exception Handling

607

Traditional Error Handling 608
Exception Handling 610
Exception Handlers 612
Throwing and Catching Exceptions 614
Nesting Exception Handling 616
Defining Your Own Error Classes 618
Standard Exception Classes 620
Exercises 622
Solutions 626

Chapter 29

More About Files

637

Opening a File for Random Access 638
Positioning for Random Access 640
File State 644
Exception Handling for Files 646
Persistence of Polymorphic Objects 648
Application: Index Files 652
Implementing an Index File 654
Exercises 656
Solutions 660

Chapter 30

More About Pointers

681

Pointer to Pointers 682
Variable Number of Arguments
Pointers to Functions 688
Complex Declarations 690
Defining Typenames 692
Application: Dynamic Matrices
Exercises 696
Solutions 698

Chapter 31

Manipulating Bits

684

694

705

Bitwise Operators 706
Bitwise Shift Operators 708
Bit Masks 710
Using Bit Masks 712
Bit-Fields 714
Exercises 716
Solutions 718

Chapter 32

Templates

721

Function and Class Templates
Defining Templates 724

722

CONTENTS

Template Instantiation 726
Template Parameters 728
Template Arguments 730
Specialization 732
Default Arguments of Templates
Explicit Instantiation 736
Exercises 738
Solutions 742

Chapter 33

Containers

734

749

Container Types 750
Sequences 752
Iterators 754
Declaring Sequences 756
Inserting in Sequences 758
Accessing Objects 760
Length and Capacity 762
Deleting in Sequences 764
List Operations 766
Associative Containers 768
Sets and Multisets 770
Maps and Multimaps 772
Bitsets 774
Exercise 778
Solution 780

Appendix

783

Binary Numbers 784
Preprocessor Directives 787
Pre-Defined Standard Macros 792
Binding C Functions 793
Operators Overview 795
Operator Precedence Table 797
ASCII Code Table 798
Screen Control Sequences 800
Literature 801

Index 803

■

xix

This page intentionally left blank

chapter

1

Fundamentals
This chapter describes the fundamental characteristics of the objectoriented C++ programming language. In addition, you will be introduced
to the steps necessary for creating a fully functional C++ program.The
examples provided will help you retrace these steps and also
demonstrate the basic structure of a C++ program.

1

2

■

CHAPTER 1

FUNDAMENTALS

■

DEVELOPMENT AND PROPERTIES OF C++

Characteristics

C++

C
-universal
-efficient
-close to the machine
-portable

OOP
-data abstraction
-data hiding
-inheritance
-polymorphism

Extensions
-exception handling
-templates

DEVELOPMENT AND PROPERTIES OF C++

■

3

䊐 Historical Perspective
The C++ programming language was created by Bjarne Stroustrup and his team at Bell
Laboratories (AT&T, USA) to help implement simulation projects in an object-oriented and efficient way. The earliest versions, which were originally referred to as “C
with classes,” date back to 1980. As the name C++ implies, C++ was derived from the C
programming language: ++ is the increment operator in C.
As early as 1989 an ANSI Committee (American National Standards Institute) was
founded to standardize the C++ programming language. The aim was to have as many
compiler vendors and software developers as possible agree on a unified description of
the language in order to avoid the confusion caused by a variety of dialects.
In 1998 the ISO (International Organization for Standardization) approved a standard for C++ (ISO/IEC 14882).

䊐 Characteristics of C++
C++ is not a purely object-oriented language but a hybrid that contains the functionality
of the C programming language. This means that you have all the features that are available in C:
■
■
■

universally usable modular programs
efficient, close to the machine programming
portable programs for various platforms.

The large quantities of existing C source code can also be used in C++ programs.
C++ supports the concepts of object-oriented programming (or OOP for short),
which are:
■
■
■
■

data abstraction, that is, the creation of classes to describe objects
data encapsulation for controlled access to object data
inheritance by creating derived classes (including multiple derived classes)
polymorphism (Greek for multiform), that is, the implementation of instructions
that can have varying effects during program execution.

Various language elements were added to C++, such as references, templates, and exception handling. Even though these elements of the language are not strictly object-oriented programming features, they are important for efficient program implementation.

4

■

CHAPTER 1

FUNDAMENTALS

■

OBJECT-ORIENTED PROGRAMMING

Traditional concept

function1
data1
function2
data2
function3

Object-oriented concept
object1

object2

Properties

Properties

Capacities

Capacities

OBJECT-ORIENTED PROGRAMMING

■

5

䊐 Traditional Procedural Programming
In traditional, procedural programming, data and functions (subroutines, procedures) are
kept separate from the data they process. This has a significant effect on the way a program handles data:
■
■

the programmer must ensure that data are initialized with suitable values before
use and that suitable data are passed to a function when it is called
if the data representation is changed, e.g. if a record is extended, the corresponding functions must also be modified.

Both of these points can lead to errors and neither support low program maintenance
requirements.

䊐 Objects
Object-oriented programming shifts the focus of attention to the objects, that is, to the
aspects on which the problem is centered. A program designed to maintain bank
accounts would work with data such as balances, credit limits, transfers, interest calculations, and so on. An object representing an account in a program will have properties
and capacities that are important for account management.
OOP objects combine data (properties) and functions (capacities). A class defines a
certain object type by defining both the properties and the capacities of the objects of
that type. Objects communicate by sending each other “messages,” which in turn activate another object’s capacities.

䊐 Advantages of OOP
Object-oriented programming offers several major advantages to software development:
■
■
■

reduced susceptibility to errors: an object controls access to its own data. More
specifically, an object can reject erroneous access attempts
easy re-use: objects maintain themselves and can therefore be used as building
blocks for other programs
low maintenance requirement: an object type can modify its own internal data
representation without requiring changes to the application.

6

■

CHAPTER 1

FUNDAMENTALS

■

DEVELOPING A C++ PROGRAM

Translating a C++ program

Editor

Source file

Header file

Compiler

Object file

Standard
library
Linker
Other
libraries,
object files
Executable
file

DEVELOPING A C++ PROGRAM

■

7

The following three steps are required to create and translate a C++ program:
1. First, a text editor is used to save the C++ program in a text file. In other words,
the source code is saved to a source file. In larger projects the programmer will normally use modular programming. This means that the source code will be stored in
several source files that are edited and translated separately.
2. The source file is put through a compiler for translation. If everything works as
planned, an object file made up of machine code is created. The object file is also
referred to as a module.
3. Finally, the linker combines the object file with other modules to form an executable file. These further modules contain functions from standard libraries or
parts of the program that have been compiled previously.
It is important to use the correct file extension for the source file’s name. Although
the file extension depends on the compiler you use, the most commonly found file extensions are .cpp and .cc.
Prior to compilation, header files, which are also referred to as include files, can be
copied to the source file. Header files are text files containing information needed by various source files, for example, type definitions or declarations of variables and functions.
Header files can have the file extension .h, but they may not have any file extension.
The C++ standard library contains predefined and standardized functions that are
available for any compiler.
Modern compilers normally offer an integrated software development environment, which
combines the steps mentioned previously into a single task. A graphical user interface is
available for editing, compiling, linking, and running the application. Moreover, additional tools, such as a debugger, can be launched.

✓

NOTE
If the source file contains just one syntax error, the compiler will report an error. Additional error
messages may be shown if the compiler attempts to continue despite having found an error. So when
you are troubleshooting a program, be sure to start with the first error shown.

In addition to error messages, the compiler will also issue warnings. A warning does
not indicate a syntax error but merely draws your attention to a possible error in the program’s logic, such as the use of a non-initialized variable.

8

■

CHAPTER 1

FUNDAMENTALS

■

A BEGINNER’S C++ PROGRAM

Sample program
#include 
using namespace std;
int main()
{
cout << "Enjoy yourself with C++!"
return 0;
}

<< endl;

Screen output
Enjoy yourself with C++!

Structure of function main()

Function name
Type of function
Beginning of
function

int main()
{
.
.
What
What the
the program
program does
does
(satements)
(statements)

.
.
End of function

}

Function block

A BEGINNER’S C++ PROGRAM

■

9

A C++ program is made up of objects with their accompanying member functions and
global functions, which do not belong to any single particular class. Each function fulfills
its own particular task and can also call other functions. You can create functions yourself or use ready-made functions from the standard library. You will always need to write
the global function main() yourself since it has a special role to play; in fact it is the
main program.
The short programming example on the opposite page demonstrates two of the most
important elements of a C++ program. The program contains only the function main()
and displays a message.
The first line begins with the number symbol, #, which indicates that the line is
intended for the preprocessor. The preprocessor is just one step in the first translation
phase and no object code is created at this time. You can type
#include 

to have the preprocessor copy the quoted file to this position in the source code. This
allows the program access to all the information contained in the header file. The header
file iostream comprises conventions for input and output streams. The word stream
indicates that the information involved will be treated as a flow of data.
Predefined names in C++ are to be found in the std (standard) namespace. The
using directive allows direct access to the names of the std namespace.
Program execution begins with the first instruction in function main(), and this is
why each C++ program must have a main function. The structure of the function is
shown on the opposite page. Apart from the fact that the name cannot be changed, this
function’s structure is not different from that of any other C++ function.
In our example the function main() contains two statements. The first statement
cout << "Enjoy yourself with C++!" << endl;

outputs the text string Enjoy yourself with C++! on the screen. The name cout
(console output) designates an object responsible for output.
The two less-than symbols, <<, indicate that characters are being “pushed” to the output stream. Finally endl (end of line) causes a line feed. The statement
return 0;

terminates the function main() and also the program, returning a value of 0 as an exit
code to the calling program. It is standard practice to use the exit code 0 to indicate that
a program has terminated correctly.
Note that statements are followed by a semicolon. By the way, the shortest statement
comprises only a semicolon and does nothing.

10

■

CHAPTER 1

FUNDAMENTALS

■

STRUCTURE OF SIMPLE C++ PROGRAMS

A C++ program with several functions
/******************************************************
A program with some functions and comments
******************************************************/
#include 
using namespace std;
void line(), message();

// Prototypes

int main()
{
cout << "Hello! The program starts in main()."
<< endl;
line();
message();
line();
cout << "At the end of main()." << endl;
return 0;
}
void line()
// To draw a line.
{
cout << "--------------------------------" << endl;
}
void message()
// To display a message.
{
cout << "In function message()." << endl;
}

Screen output
Hello! The program starts in main().
----------------------------------In function message().
----------------------------------At the end of main().

STRUCTURE OF SIMPLE C++ PROGRAMS

■

11

The example on the opposite page shows the structure of a C++ program containing
multiple functions. In C++, functions do not need to be defined in any fixed order. For
example, you could define the function message() first, followed by the function
line(), and finally the main() function.
However, it is more common to start with the main() function as this function controls the program flow. In other words, main() calls functions that have yet to be
defined. This is made possible by supplying the compiler with a function prototype that
includes all the information the compiler needs.
This example also introduces comments. Strings enclosed in /* . . . */ or starting with // are interpreted as comments.

EXAMPLES:
/* I can cover
several lines */
// I can cover just one line

In single-line comments the compiler ignores any characters following the // signs up
to the end of the line. Comments that cover several lines are useful when troubleshooting, as you can use them to mask complete sections of your program. Both comment
types can be used to comment out the other type.
As to the layout of source files, the compiler parses each source file sequentially,
breaking the contents down into tokens, such as function names and operators. Tokens
can be separated by any number of whitespace characters, that is, by spaces, tabs, or
new line characters. The order of the source code is important but it is not important
to adhere to a specific layout, such as organizing your code in rows and columns. For
example
void message
(
){ cout <<
"In function message()."
endl;}

<<

might be difficult to read, but it is a correct definition of the function message().
Preprocessor directives are one exception to the layout rule since they always occupy a
single line. The number sign, #, at the beginning of a line can be preceded only by a
space or a tab character.
To improve the legibility of your C++ programs you should adopt a consistent style,
using indentation and blank lines to reflect the structure of your program. In addition,
make generous use of comments.

■

CHAPTER 1

FUNDAMENTALS

■

exercise s

12

EXERCISES

Program listing of exercise 3
#include 
using namespace std;
void pause();

// Prototype

int main()
{
cout << endl << "Dear reader, "
<< endl << "have a ";
pause();
cout << "!" << endl;
return 0;
}
void pause()
{
cout << "BREAK";
}

EXERCISES

Exercise 1
Write a C++ program that outputs the following text on screen:
Oh what
a happy day!
Oh yes,
what a happy day!

Use the manipulator endl where appropriate.

Exercise 2
The following program contains several errors:
*/ Now you should not forget your glasses //
#include 
int main
{
cout << "If this text",
cout >> " appears on your display, ";
cout << " endl;"
cout << 'you can pat yourself on '
<< " the back!" << endl.
return 0;
)

Resolve the errors and run the program to test your changes.

Exercise 3
What does the C++ program on the opposite page output on screen?

■

13

■

CHAPTER 1

■

solutions

14

FUNDAMENTALS

SOLUTIONS

Exercise 1
// Let's go !
#include 
using namespace std;
int main()
{
cout <<
cout <<
cout <<
cout <<

"
"
"
"

Oh what " <<
a happy day!
Oh yes, " <<
what a happy

endl;
" << endl;
endl;
day! " << endl;

return 0;
}

Exercise 2
The corrected places are underlined.
/* Now you should not forget your glasses */
#include 
using namespace std;
int main()
{
cout << " If this text ";
cout << " appears on your display, ";
cout << endl;
cout << " you can pat yourself on "
<< " the back!" << endl;
return 0;
}

Exercise 3
The screen output begins on a new line:
Dear reader,
have a BREAK!

chapter

2

Fundamental Types,
Constants, and Variables
This chapter introduces you to the basic types and objects used by C++
programs.

15

16

■

CHAPTER 2

FUNDAMENTAL TYPES, CONSTANTS, AND VARIABLES

■

FUNDAMENTAL TYPES

Overview*

For boolean values

bool

char
For characters

wchar_t

short

For integers

int

long

float

For floating-point
values

double

long double

* without type void, which will be introduced later.

FUNDAMENTAL TYPES

■

17

A program can use several data to solve a given problem, for example, characters, integers, or floating-point numbers. Since a computer uses different methods for processing
and saving data, the data type must be known. The type defines
1. the internal representation of the data, and
2. the amount of memory to allocate.
A number such as -1000 can be stored in either 2 or 4 bytes. When accessing the
part of memory in which the number is stored, it is important to read the correct number
of bytes. Moreover, the memory content, that is the bit sequence being read, must be
interpreted correctly as a signed integer.
The C++ compiler recognizes the fundamental types, also referred to as built-in types,
shown on the opposite page, on which all other types (vectors, pointers, classes, ...) are
based.

䊐 The Type bool
The result of a comparison or a logical association using AND or OR is a boolean value,
which can be true or false. C++ uses the bool type to represent boolean values. An
expression of the type bool can either be true or false, where the internal value for
true will be represented as the numerical value 1 and false by a zero.

䊐 The char and wchar_t Types
These types are used for saving character codes. A character code is an integer associated
with each character. The letter A is represented by code 65, for example. The character
set defines which code represents a certain character. When displaying characters on
screen, the applicable character codes are transmitted and the “receiver,” that is the
screen, is responsible for correctly interpreting the codes.
The C++ language does not stipulate any particular characters set, although in general a character set that contains the ASCII code (American Standard Code for Information Interchange) is used. This 7-bit code contains definitions for 32 control characters
(codes 0 – 31) and 96 printable characters (codes 32 – 127).
The char (character) type is used to store character codes in one byte (8 bits). This
amount of storage is sufficient for extended character sets, for example, the ANSI character set that contains the ASCII codes and additional characters such as German
umlauts.
The wchar_t (wide character type) type comprises at least 2 bytes (16 bits) and is
thus capable of storing modern Unicode characters. Unicode is a 16-bit code also used in
Windows NT and containing codes for approximately 35,000 characters in 24 languages.

18

■

CHAPTER 2

FUNDAMENTAL TYPES, CONSTANTS, AND VARIABLES

■

FUNDAMENTAL TYPES (CONTINUED)

Integral types
Type

Size

char

1 byte

unsigned char

1 byte

signed char

1 byte

int

2 byte resp.
4 byte

unsigned int

Range of Values (decimal)
—
128 to +127 or 0 to 255
0 to 255
—
128 to +127
—
32768 to +32767 resp.
—
2147483648 to +2147483647

2 byte resp.

0 to 65535 resp.

4 byte

0 to 4294967295

short

2 byte

unsigned short

2 byte

long

4 byte

unsigned long

4 byte

—
32768 to +32767
0 to 65535
—
2147483648 to +2147483647
0 to 4294967295

Sample program
#include 
#include 
using namespace std;

// Definition of INT_MIN, ...

int main()
{
cout << "Range of types int and unsigned int"
<< endl << endl;
cout << "Type
Minimum
Maximum"
<< endl
<< "--------------------------------------------"
<< endl;
cout << "int

" <<
<<

INT_MIN << "
INT_MAX << endl;

"

cout << "unsigned int

" << "
0
<< UINT_MAX << endl;

"

return 0;
}

FUNDAMENTAL TYPES (CONTINUED)

■

19

䊐 Integral Types
The types short, int, and long are available for operations with integers. These types
are distinguished by their ranges of values. The table on the opposite page shows the
integer types, which are also referred to as integral types, with their typical storage
requirements and ranges of values.
The int (integer) type is tailor-made for computers and adapts to the length of a register on the computer. For 16-bit computers, int is thus equivalent to short, whereas
for 32-bit computers int will be equivalent to long.
C++ treats character codes just like normal integers. This means you can perform calculations with variables belonging to the char or wchar_t types in exactly the same
way as with int type variables. char is an integral type with a size of one byte. The
range of values is thus –128 to +127 or from 0 to 255, depending on whether the compiler interprets the char type as signed or unsigned. This can vary in C++.
The wchar_t type is a further integral type and is normally defined as unsigned
short.

䊐 The signed and unsigned Modifiers

✓

The short, int, and long types are normally interpreted as signed with the highest bit
representing the sign. However, integral types can be preceded by the keyword
unsigned. The amount of memory required remains unaltered but the range of values
changes due to the highest bit no longer being required as a sign. The keyword
unsigned can be used as an abbreviation for unsigned int.
The char type is also normally interpreted as signed. Since this is merely a convention and not mandatory, the signed keyword is available. Thus three types are available: char, signed char, and unsigned char.

NOTE
In ANSI C++ the size of integer types is not preset. However, the following order applies:
char <= short <= int <= long
Moreover, the short type comprises at least 2 bytes and the long type at least 4 bytes.

The current value ranges are available in the climits header file. This file defines
constants such as CHAR_MIN, CHAR_MAX, INT_MIN, and INT_MAX, which represent
the smallest and greatest possible values. The program on the opposite page outputs the
value of these constants for the int and unsigned int types.

20

■

CHAPTER 2

FUNDAMENTAL TYPES, CONSTANTS, AND VARIABLES

■

FUNDAMENTAL TYPES (CONTINUED)

Floating-point types

Type

Size

Range of
Values

Lowest Positive
Value

Accuracy
(decimal)

float

4 bytes

–3.4E+38

1.2E—38

6 digits

double

8 bytes

–1.7E+308

2.3E—308

15 digits

10 bytes

–1.1E+4932

3.4E—4932

19 digits

long double

✓

NOTE
IEEE format (IEEE = Institute of Electrical and Electronic Engineers) is normally used to represent
floating-point types. The table above makes use of this representation.

Arithmetic types

Integral types
bool
char, signed char, unsigned char, wchar_t
short, unsigned short
int, unsigned int
long, unsigned long

Floating-point types
float
double
long double

✓

NOTE
Arithmetic operators are defined for arithmetic types, i.e. you can perform calculations with variables of
this type.

FUNDAMENTAL TYPES (CONTINUED)

■

21

䊐 Floating-Point Types
Numbers with a fraction part are indicated by a decimal point in C++ and are referred to
as floating-point numbers. In contrast to integers, floating-point numbers must be stored
to a preset accuracy. The following three types are available for calculations involving
floating-point numbers:
float
double
long double

for simple accuracy
for double accuracy
for high accuracy

The value range and accuracy of a type are derived from the amount of memory allocated
and the internal representation of the type.
Accuracy is expressed in decimal places. This means that “six decimal places” allows a
programmer to store two floating-point numbers that differ within the first six decimal
places as separate numbers. In reverse, there is no guarantee that the figures 12.3456 and
12.34561 will be distinguished when working to a accuracy of six decimal places. And
remember, it is not a question of the position of the decimal point, but merely of the
numerical sequence.
If it is important for your program to display floating-point numbers with an accuracy
supported by a particular machine, you should refer to the values defined in the cfloat
header file.
Readers interested in additional material on this subject should refer to the Appendix,
which contains a section on the representation of binary numbers on computers for both
integers and floating-point numbers.

䊐 The sizeof Operator
The amount of memory needed to store an object of a certain type can be ascertained
using the sizeof operator:
sizeof(name)

yields the size of an object in bytes, and the parameter name indicates the object type or
the object itself. For example, sizeof(int)represents a value of 2 or 4 depending on
the machine. In contrast, sizeof(float) will always equal 4.

䊐 Classification
The fundamental types in C++ are integer types, floating-point types, and the void type.
The types used for integers and floating-point numbers are collectively referred to as
arithmetic types, as arithmetic operators are defined for them.
The void type is used for expressions that do not represent a value. A function call
can thus take a void type.

22

■

CHAPTER 2

FUNDAMENTAL TYPES, CONSTANTS, AND VARIABLES

■

CONSTANTS

Examples for integral constants

✓

Decimal

Octal

Hexadecimal

Type

16

020

0x10

int

255

0377

OXff

int

32767

077777

0x7FFF

int

32768U

0100000U

0x8000U

unsigned int

100000

0303240

0x186A0

int (32 bit-)
long (16 bitCPU)

10L

012L

0xAL

long

27UL

033UL

0x1bUL

unsigned long

2147483648

020000000000

0x80000000

unsigned long

NOTE
In each line of the above table, the same value is presented in a different way.

Sample program
// To display hexadecimal integer literals and
// decimal integer literals.
//
#include 
using namespace std;
int main()
{
// cout outputs integers as decimal integers:
cout << "Value of 0xFF = " << 0xFF << " decimal"
<< endl;
// Output: 255 decimal
// The manipulator hex changes output to hexadecimal
// format (dec changes to decimal format):
cout << "Value of 27 = " << hex << 27 <<" hexadecimal"
<< endl;
// Output: 1b hexadecimal
return 0;
}

CONSTANTS

■

23

The boolean keywords true and false, a number, a character, or a character sequence
(string) are all constants, which are also referred to as a literals. Constants can thus be
subdivided into
■
■
■
■

boolean constants
numerical constants
character constants
string constants.

Every constant represents a value and thus a type—as does every expression in C++. The
type is defined by the way the constant is written.

䊐 Boolean Constants
A boolean expression can have two values that are identified by the keywords true and
false. Both constants are of the bool type. They can be used, for example, to set flags
representing just two states.

䊐 Integral Constants
Integral numerical constants can be represented as simple decimal numbers, octals, or
hexadecimals:
■
■
■

a decimal constant (base 10) begins with a decimal number other than zero, such
as 109 or 987650
an octal constant (base 8) begins with a leading 0, for example 077 or 01234567
a hexadecimal constant (base 16) begins with the character pair 0x or 0X, for
example 0x2A0 or 0X4b1C. Hexadecimal numbers can be capitalized or noncapitalized.

Integral constants are normally of type int. If the value of the constant is too large
for the int type, a type capable of representing larger values will be applied. The ranking
for decimal constants is as follows:
int, long, unsigned long

You can designate the type of a constant by adding the letter L or l (for long), or U
or u (for unsigned). For example,
12L
12U
12UL

and
and
and

12l
12u
12ul

correspond to the type long
correspond to the type unsigned int
correspond to the type unsigned long

24

■

CHAPTER 2

FUNDAMENTAL TYPES, CONSTANTS, AND VARIABLES

■

CONSTANTS (CONTINUED)

Examples for floating-point constants
5.19

12.

0.519E1

12.0

0.0519e2
519.OE-2

0.75

0.00004
0.4e-4

.75

.12E+2

.4E-4

7.5e-1

12e0

4E-5

75E-2

Examples for character constants
Constant

Character

Constant Value
(ASCII code decimal)

'A'

Capital A

65

'a'

Lowercase a

97

' '

Blank

32

'.'

Dot

46

'0'

Digit 0

48

Terminating null character

0

'\0'

Internal representation of a string literal
String literal:

"Hello!"

Stored byte sequence:

'H'

'e'

'1'

'1'

'o'

'!'

'\0'

CONSTANTS (CONTINUED)

■

25

䊐 Floating-Point Constants
Floating-point numbers are always represented as decimals, a decimal point being used to
distinguish the fraction part from the integer part. However, exponential notation is also
permissible.

EXAMPLES:

27.1

1.8E–2

// Type: double

Here, 1.8E–2 represents a value of 1.8*10–2. E can also be written with a small letter
e. A decimal point or E (e) must always be used to distinguish floating-point constants
from integer constants.
Floating-point constants are of type double by default. However, you can add F or f
to designate the float type, or add L or l for the long double type.

䊐 Character Constants
A character constant is a character enclosed in single quotes. Character constants take
the type char.

EXAMPLE:

'A'

// Type: char

The numerical value is the character code representing the character. The constant 'A'
thus has a value of 65 in ASCII code.

䊐 String Constants
You already know string constants, which were introduced for text output using the
cout stream. A string constant consists of a sequence of characters enclosed in double
quotes.

EXAMPLE:

"Today is a beautiful day!"

A string constant is stored internally without the quotes but terminated with a null character, \0, represented by a byte with a numerical value of 0 — that is, all the bits in this
byte are set to 0. Thus, a string occupies one byte more in memory than the number of
characters it contains. An empty string, "", therefore occupies a single byte.
The terminating null character \0 is not the same as the number zero and has a different character code than zero. Thus, the string

EXAMPLE:

"0"

comprises two bytes, the first byte containing the code for the character zero 0 (ASCII
code 48) and the second byte the value 0.
The terminating null character \0 is an example of an escape sequence. Escape
sequences are described in the following section.

26

■

CHAPTER 2

FUNDAMENTAL TYPES, CONSTANTS, AND VARIABLES

■

ESCAPE SEQUENCES

Overview
Single character

Meaning

ASCII code
(decimal)

\a

alert (BEL)

7

\b

backspace (BS)

8

\t

horizontal tab (HT)

9

\n

line feed (LF)

10

\v

vertical tab (VT)

11

\f

form feed (FF)

12

\r

carriage return (CR)

13

\"

" (double quote)

34

\'

' (single quote)

39

\?

? (question mark)

63

\\

\ (backslash)

92

\0

string terminating character

0

\ooo

numerical value of a character

ooo (octal!)

numerical value of a character

hh (hexadecimal!)

(up to 3 octal digits)

\xhh
(hexadecimal digits)

Sample program
#include 
using namespace std;
int main()
{
cout << "\nThis is\t a string\n\t\t"
" with \"many\" escape sequences!\n";
return 0;
}

Program output:
This is

a string
with "many" escape sequences!

ESCAPE SEQUENCES

■

27

䊐 Using Control and Special Characters
Nongraphic characters can be expressed by means of escape sequences, for example \t,
which represents a tab.
The effect of an escape sequence will depend on the device concerned. The sequence
\t, for example, depends on the setting for the tab width, which defaults to eight blanks
but can be any value.
An escape sequence always begins with a \ (backslash) and represents a single character. The table on the opposite page shows the standard escape sequences, their decimal
values, and effects.
You can use octal and hexadecimal escape sequences to create any character code.
Thus, the letter A (decimal 65) in ASCII code can also be expressed as \101 (three
octals) or \x41 (two hexadecimals). Traditionally, escape sequences are used only to
represent non-printable characters and special characters. The control sequences for
screen and printer drivers are, for example, initiated by the ESC character (decimal 27),
which can be represented as \33 or \x1b.
Escape sequences are used in character and string constants.

EXAMPLES:

'\t'

"\tHello\n\tMike!"

The characters ', ", and \ have no special significance when preceded by a backslash, i.e.
they can be represented as \', \", and \\ respectively.
When using octal numbers for escape sequences in strings, be sure to use three digits,
for example, \033 and not \33. This helps to avoid any subsequent numbers being evaluated as part of the escape sequence. There is no maximum number of digits in a hexadecimal escape sequence. The sequence of hex numbers automatically terminates with
the first character that is not a valid hex number.
The sample program on the opposite page demonstrates the use of escape sequences in
strings. The fact that a string can occupy two lines is another new feature. String
constants separated only by white spaces will be concatenated to form a single string.
To continue a string in the next line you can also use a backslash \ as the last
character in a line, and then press the Enter key to begin a new line, where you can
continue typing the string.

EXAMPLE:

"I am a very, very \
long string"

Please note, however, that the leading spaces in the second line will be evaluated as part
of the string. It is thus generally preferable to use the first method, that is, to terminate
the string with " and reopen it with ".

28

■

CHAPTER 2

FUNDAMENTAL TYPES, CONSTANTS, AND VARIABLES

■

NAMES

Keywords in C++

asm

do

inline

short

typeid

auto

double

int

signed

typename

bool

dynamic_cast

long

sizeof

union

break

else

mutable

static

unsigned

case

enum

namespace

static_cast

using

catch

explicit

new

struct

virtual

char

extern

operator

switch

void

class

false

private

template

volatile

const

float

protected

this

wchar_t

const_cast

for

public

throw

while

continue

friend

register

true

default

goto

reinterpret_cast

try

delete

if

return

typedef

Examples for names
valid:
a

US

us

_var

SetTextColor

B12

top_of_window

VOID

a_very_long_name123467890

invalid:
goto
US$

586_cpu
true

object-oriented
écu

NAMES

■

29

䊐 Valid Names
Within a program names are used to designate variables and functions. The following
rules apply when creating names, which are also known as identifiers:
■

■
■
■

a name contains a series of letters, numbers, or underscore characters ( _ ). German umlauts and accented letters are invalid. C++ is case sensitive; that is,
upper- and lowercase letters are different.
the first character must be a letter or underscore
there are no restrictions on the length of a name and all the characters in the
name are significant
C++ keywords are reserved and cannot be used as names.

The opposite page shows C++ keywords and some examples of valid and invalid names.
The C++ compiler uses internal names that begin with one or two underscores followed by a capital letter. To avoid confusion with these names, avoid use of the underscore at the beginning of a name.
Under normal circumstances the linker only evaluates a set number of characters, for
example, the first 8 characters of a name. For this reason names of global objects, such as
functions, should be chosen so that the first eight characters are significant.

䊐 Conventions
In C++ it is standard practice to use small letters for the names of variables and functions. The names of some variables tend to be associated with a specific use.

EXAMPLES:
c, ch
i, j, k, l, m, n
x, y, z

for characters
for integers, in particular indices
for floating-point numbers

To improve the readability of your programs you should choose longer and more selfexplanatory names, such as start_index or startIndex for the first index in a range
of index values.
In the case of software projects, naming conventions will normally apply. For example, prefixes that indicate the type of the variable may be assigned when naming variables.

30

■

CHAPTER 2

FUNDAMENTAL TYPES, CONSTANTS, AND VARIABLES

■

VARIABLES

Sample program
// Definition and use of variables
#include 
using namespace std;
int gVar1;
int gVar2 = 2;

// Global variables,
// explicit initialization

int main()
{
char ch('A');

// Local variable being initialized
// or: char ch = 'A';

cout << "Value of gVar1:
cout << "Value of gVar2:
cout << "Character in ch:

" << gVar1
" << gVar2
" << ch

<< endl;
<< endl;
<< endl;

int sum, number = 3; // Local variables with
// and without initialization
sum = number + 5;
cout << "Value of sum:
" << sum << endl;
return 0;
}

✓

HINT
Both strings and all other values of fundamental types can be output with cout. Integers are printed in
decimal format by default.

Screen output
Value of gVar1:
Value of gVar2:
Character in ch:
Value of sum:

0
2
A
8

VARIABLES

■

31

Data such as numbers, characters, or even complete records are stored in variables to
enable their processing by a program. Variables are also referred to as objects, particularly
if they belong to a class.

䊐 Defining Variables
A variable must be defined before you can use it in a program. When you define a variable the type is specified and an appropriate amount of memory reserved. This memory
space is addressed by reference to the name of the variable. A simple definition has the
following syntax:

SYNTAX:

typ name1 [name2 ... ];

This defines the names of the variables in the list name1 [, name2 ...] as variables
of the type type. The parentheses [ ... ] in the syntax description indicate that this
part is optional and can be omitted. Thus, one or more variables can be stated within a
single definition.

EXAMPLES:

char c;
int i, counter;
double x, y, size;

In a program, variables can be defined either within the program’s functions or outside of them. This has the following effect:
■
■

a variable defined outside of each function is global, i.e. it can be used by all functions
a variable defined within a function is local, i.e. it can be used only in that function.

Local variables are normally defined immediately after the first brace—for example at
the beginning of a function. However, they can be defined wherever a statement is permitted. This means that variables can be defined immediately before they are used by the
program.

䊐 Initialization
A variable can be initialized, i.e. a value can be assigned to the variable, during its definition. Initialization is achieved by placing the following immediately after the name of
the variable:
■
■

an equals sign ( = ) and an initial value for the variable or
round brackets containing the value of the variable.

EXAMPLES:

char c = 'a';
float x(1.875);

Any global variables not explicitly initialized default to zero. In contrast, the initial
value for any local variables that you fail to initialize will have an undefined initial value.

32

■

CHAPTER 2

FUNDAMENTAL TYPES, CONSTANTS, AND VARIABLES

■

THE KEYWORDS const AND volatile

Sample program
// Circumference and area of a circle with radius 2.5
#include 
using namespace std;
const double pi = 3.141593;
int main()
{
double area, circuit, radius = 1.5;
area = pi * radius * radius;
circuit = 2 * pi * radius;
cout << "\nTo Evaluate a Circle\n" << endl;
cout << "Radius:
" << radius
<< "Circumference: " << circuit
<< "Area:
" << area

<< endl
<< endl
<< endl;

return 0;
}

✓

NOTE
By default cout outputs a floating-point number with a maximum of 6 decimal places without trailing
zeros.

Screen output
To Evaluate a Circle
Radius:
Circumference:
Area:

1.5
9.42478
7.06858

THE KEYWORDS CONST AND VOLATILE

■

33

A type can be modified using the const and volatile keywords.

䊐 Constant Objects
The const keyword is used to create a “read only” object. As an object of this type is
constant, it cannot be modified at a later stage and must be initialized during its definition.

EXAMPLE:

const double pi = 3.1415947;

Thus the value of pi cannot be modified by the program. Even a statement such as the
following will merely result in an error message:
pi = pi + 2.0;

// invalid

䊐 Volatile Objects
The keyword volatile, which is rarely used, creates variables that can be modified not
only by the program but also by other programs and external events. Events can be initiated by interrupts or by a hardware clock, for example.

EXAMPLE:

volatile unsigned long

clock_ticks;

Even if the program itself does not modify the variable, the compiler must assume that
the value of the variable has changed since it was last accessed. The compiler therefore
creates machine code to read the value of the variable whenever it is accessed instead of
repeatedly using a value that has been read at a prior stage.
It is also possible to combine the keywords const and volatile when declaring a
variable.

EXAMPLE:

volatile const unsigned time_to_live;

Based on this declaration, the variable time_to_live cannot be modified by the program but by external events.

■

CHAPTER 2

■

exercise s

34

FUNDAMENTAL TYPES, CONSTANTS, AND VARIABLES

EXERCISES

Screen output for exercise 2
I
"RUSH"
\TO\
AND
/FRO/

For exercise 3
Defining and initializing variables:
int a(2.5);
int b = '?';
char z(500);
int big = 40000;
double he's(1.2E+5);

const long large;
char c('\'');
unsigned char ch = '\201';
unsigned size(40000);
float val = 12345.12345;

EXERCISES

■

35

Exercise 1
The sizeof operator can be used to determine the number of bytes occupied
in memory by a variable of a certain type. For example, sizeof(short) is
equivalent to 2.
Write a C++ program that displays the memory space required by each
fundamental type on screen.
Exercise 2
Write a C++ program to generate the screen output shown on the opposite
page.
Exercise 3
Which of the variable definitions shown on the opposite page is invalid or does
not make sense?
Exercise 4
Write a C++ program that two defines variables for floating-point numbers and
initializes them with the values
123.456 and 76.543

Then display the sum and the difference of these two numbers on screen.

■

CHAPTER 2

■

solutions

36

FUNDAMENTAL TYPES, CONSTANTS, AND VARIABLES

SOLUTIONS

Exercise 1
#include 
using namespace std;
int main()
{
cout <<
<<
<<
cout <<
cout <<
cout <<
cout <<
cout <<
cout <<
cout <<
<<

"\nSize of Fundamental Types\n"
" Type
Number of Bytes\n"
"----------------------------------" << endl;
" char:
" << sizeof(char) << endl;
" short:
" << sizeof(short)<< endl;
" int:
" << sizeof(int) << endl;
" long:
" << sizeof(long) << endl;
" float:
" << sizeof(float)<< endl;
" double:
" << sizeof(double)<
using namespace std;
int main()
{
cout << "\n\n\t I"
"\n\n\t\t \"RUSH\""
"\n\n\t\t\t \\TO\\"
"\n\n\t\t AND"
"\n\n\t /FRO/" << endl;
return 0;
}

//
//
//
//
//

Instead of tabs
you can send the
suited number
of blanks to
the output.

SOLUTIONS

Exercise 3
Incorrect:
int a(2.5);
const long large;
char z(500);
int big = 40000;
double he's(1.2E+5);
float val = 12345.12345;

//
//
//
//
//
//
//
//
//
//

2.5 is not an integer value
Without initialization
The value 500 is too large
to fit in a byte
Attention! On 16-bit systems
int values are <= 32767
The character ' is not
allowed in names
The accuracy of float
is only 6 digits

Exercise 4
// Defining and initializing variables
#include 
using namespace std;
int main()
{
float x = 123.456F,
y = 76.543F,
sum;

// or double

sum = x + y;
cout << "Total:
"
<< x << " + " << y << " = " << sum << endl;
cout << "Difference: "
<< x << " — " << y << " = " << (x — y) << endl;
return 0;
}

■

37

This page intentionally left blank

chapter

3

Using Functions and
Classes
This chapter describes how to
■

declare and call standard functions and

■

use standard classes.

This includes using standard header files. In addition, we will be working
with string variables, i.e. objects belonging to the standard class string
for the first time.
Functions and classes that you define on your own will not be
introduced until later in the book.

39

40

■

CHAPTER 3

USING FUNCTIONS AND CLASSES

■

DECLARING FUNCTIONS

Example of a function prototype

Function name

long func (int, double);
Types of arguments
Function type
= type of return value

The prototype above yields the following information to the compiler:
■

func is the function name

■

the function is called with two arguments: the first argument is of type int, the
second of type double
the return value of the function is of type long.

■

Mathematical standard functions
double sin (double);

// Sine

double cos (double);

// Cosine

double tan (double);

// Tangent

double atan (double);

// Arc tangent

double cosh (double);

// Hyperbolic Cosine

double sqrt (double);

// Square Root

double pow (double, double);

// Power

double exp (double);

// Exponential Function

double log (double);

// Natural Logarithm

double log10 (double);

// Base-ten Logarithm

DECLARING FUNCTIONS

■

41

䊐 Declarations
Each name (identifier) occurring in a program must be known to the compiler or it will
cause an error message. That means any names apart from keywords must be declared, i.e.
introduced to the compiler, before they are used.
Each time a variable or a function is defined it is also declared. But conversely, not
every declaration needs to be a definition. If you need to use a function that has already
been introduced in a library, you must declare the function but you do not need to redefine it.

䊐 Declaring Functions
A function has a name and a type, much like a variable. The function’s type is defined by
its return value, that is, the value the function passes back to the program. In addition,
the type of arguments required by a function is important. When a function is declared,
the compiler must therefore be provided with information on
■
■

the name and type of the function and
the type of each argument.

This is also referred to as the function prototype.

Examples: int toupper(int);
double pow(double, double);

This informs the compiler that the function toupper() is of type int, i.e. its return
value is of type int, and it expects an argument of type int. The second function
pow() is of type double and two arguments of type double must be passed to the
function when it is called. The types of the arguments may be followed by names, however, the names are viewed as a comment only.

Examples: int toupper(int c);
double pow(double base, double exponent);

From the compiler’s point of view, these prototypes are equivalent to the prototypes
in the previous example. Both junctions are standard junctions.
Standard function prototypes do not need to be declared, nor should they be, as they
have already been declared in standard header files. If the header file is included in the
program’s source code by means of the #include directive, the function can be used
immediately.

Example: #include 
Following this directive, the mathematical standard functions, such as sin(), cos(),
and pow(), are available. Additional details on header files can be found later in this
chapter.

42

■

CHAPTER 3

USING FUNCTIONS AND CLASSES

■

FUNCTION CALLS

Sample program
//
//

Calculating powers with
the standard function pow()

#include 
#include 

// Declaration of cout
// Prototype of pow(), thus:
// double pow( double, double);

using namespace std;
int main()
{
double x = 2.5, y;
// By means of a prototype, the compiler generates
// the correct call or an error message!
// Computes x raised
y = pow("x", 3.0);
y = pow(x + 3.0);
y = pow(x, 3.0);
y = pow(x, 3);

to
//
//
//
//
//

the power 3:
Error! String is not a number
Error! Just one argument
ok!
ok! The compiler converts the
int value 3 to double.

cout << "2.5 raised to 3 yields:
<< y << endl;

"

// Calculating with pow() is possible:
cout << "2 + (5 raised to the power 2.5) yields: "
<< 2.0 + pow(5.0, x) << endl;
return 0;
}

Screen output
2.5 raised to the power 3 yields:
2 + (5 raised to the power 2.5) yields:

15.625
57.9017

FUNCTION CALLS

■

43

䊐 Function Calls
A function call is an expression of the same type as the function and whose value corresponds to the return value. The return value is commonly passed to a suitable variable.

Example: y = pow( x, 3.0);
In this example the function pow()is first called using the arguments x and 3.0, and
the result, the power x3, is assigned to y.
As the function call represents a value, other operations are also possible. Thus, the
function pow() can be used to perform calculations for double values.

Example: cout

<<

2.0 + pow( 5.0, x);

This expression first adds the number 2.0 to the return value of pow(5.0,x), then
outputs the result using cout.
Any expression can be passed to a function as an argument, such as a constant or an
arithmetical expression. However, it is important that the types of the arguments correspond to those expected by the function.
The compiler refers to the prototype to check that the function has been called correctly. If the argument type does not match exactly to the type defined in the prototype,
the compiler performs type conversion, if possible.

Example: y = pow( x, 3);

// also ok!

The value 3 of type int is passed to the function as a second argument. But since the
function expects a double value, the compiler will perform type conversion from int
to double.
If a function is called with the wrong number of arguments, or if type conversion
proves impossible, the compiler generates an error message. This allows you to recognize
and correct errors caused by calling functions at the development stage instead of causing
runtime errors.

Example: float x = pow(3.0 + 4.7);

// Error!

The compiler recognizes that the number of arguments is incorrect. In addition, the
compiler will issue a warning, since a double, i.e. the return value of pow(), is assigned
to a float type variable.

44

■

CHAPTER 3

USING FUNCTIONS AND CLASSES

■

TYPE void FOR FUNCTIONS

Sample program
// Outputs three random numbers
#include 
#include 

//
//
//
//

Declaration of cin and cout
Prototypes of srand(), rand():
void srand( unsigned int seed );
int rand( void );

using namespace std;
int main()
{
unsigned int seed;
int z1, z2, z3;
cout << "
--- Random Numbers --- \n" << endl;
cout << "To initialize the random number generator, "
<< "\n please enter an integer value: ";
cin >> seed;
// Input an integer
srand( seed);

// and use it as argument for a
// new sequence of random numbers.

z1 = rand();
z2 = rand();
z3 = rand();

// Compute three random numbers.

cout << "\nThree random numbers: "
<< z1 << "
" << z2 << "
" << z3 << endl;
return 0;
}

✓

NOTE
The statement cin >> seed; reads an integer from the keyboard, because seed is of the
unsigned int type.

Sample screen output
--- Random Numbers

---

To initialize the random number generator,
please enter an integer value: 7777
Three random numbers: 25435

6908

14579

TYPE VOID FOR FUNCTIONS

■

45

䊐 Functions without Return Value
You can also write functions that perform a certain action but do not return a value to
the function that called them. The type void is available for functions of this type,
which are also referred to as procedures in other programming languages.

Example: void srand( unsigned int seed );
The standard function srand() initializes an algorithm that generates random numbers. Since the function does not return a value, it is of type void. An unsigned value
is passed to the function as an argument to seed the random number generator. The
value is used to create a series of random numbers.

䊐 Functions without Arguments
If a function does not expect an argument, the function prototype must be declared as
void or the braces following the function name must be left empty.

Example: int rand( void );

// or

int rand();

The standard function rand() is called without any arguments and returns a random
number between 0 and 32767. A series of random numbers can be generated by repeating
the function call.

䊐 Usage of srand() and rand()
The function prototypes for srand() and rand() can be found in both the cstdlib
and stdlib.h header files.
Calling the function rand() without previously having called srand() creates the
same sequence of numbers as if the following statement would have been proceeded:
srand(1);

If you want to avoid generating the same sequence of random numbers whenever the
program is executed, you must call srand() with a different value for the argument
whenever the program is run.
It is common to use the current time to initialize a random number generator. See
Chapter 6 for an example of this technique.

46

■

CHAPTER 3

USING FUNCTIONS AND CLASSES

■

HEADER FILES

Using header files
Header file

Header file

iostream

myheader.h

//
//
//
//

// Declaration
// of cin, cout,
// . . .

Declaration
of self-defined
functions
and classes

long myfunc(int);

Source file

application.cpp
Copy

#include 
#include "myheader.h"
int main()
{
int a;
. . .
cin >> a;
cout << myfunc (a);
. . .
return 0;
}

Copy

HEADER FILES

■

47

䊐 Using Header Files
Header files are text files containing declarations and macros. By using an #include
directive these declarations and macros can be made available to any other source file,
even in other header files.
Pay attention to the following points when using header files:
■
■
■

header files should generally be included at the start of a program before any
other declarations
you can only name one header file per #include directive
the file name must be enclosed in angled brackets < ... > or double quotes
" ... ".

䊐 Searching for Header Files
The header files that accompany your compiler will tend to be stored in a folder of their
own—normally called include. If the name of the header file is enclosed by angled
brackets < ... >, it is common to search for header files in the include folder only.
The current directory is not searched to increase the speed when searching for header
files.
C++ programmers commonly write their own header files and store them in the current project folder. To enable the compiler to find these header files, the #include
directive must state the name of the header files in double quotes.

Example: #include "project.h"
The compiler will then also search the current folder. The file suffix .h is normally used
for user-defined header files.

䊐 Standard Class Definitions
In addition to standard function prototypes, the header files also contain standard class
definitions. When a header file is included, the classes defined and any objects declared
in the file are available to the program.

Example: #include 
using namespace std;

Following these directives, the classes istream and ostream can be used with the cin
and cout streams. cin is an object of the istream class and cout an object of the
ostream class.

48

■

CHAPTER 3

USING FUNCTIONS AND CLASSES

■

STANDARD HEADER FILES

Header files of the C++ standard library

✓

algorithm

ios

map

stack

bitset

iosfwd

memory

stdexcept

complex

iostream

new

streambuf

dequeue

istream

numeric

string

exception

iterator

ostream

typeinfo

fstream

limits

queue

utility

functional

list

set

valarray

iomanip

locale

sstream

vector

NOTE
Some IDE’s put the old-fashioned iostream.h and iomanip.h header files at your disposal. Within
these header files the identifiers of iostream and iomanip are not contained in the std namespace
but are declared globally.

Header files of the C standard library
assert.h

limits.h

stdarg.h

time.h

ctype.h

locale.h

stddef.h

wchar.h

errno.h

math.h

stdio.h

wctype.h

float.h

setjmp.h

stdlib.h

iso646.h

signal.h

string.h

STANDARD HEADER FILES

■

49

The C++ standard library header files are shown opposite. They are not indicated by the
file extension .h and contain all the declarations in their own namespace, std. Namespaces will be introduced in a later chapter. For now, it is sufficient to know that identifiers from other namespaces cannot be referred to directly. If you merely stipulate the
directive

Example: #include 
the compiler would not be aware of the cin and cout streams. In order to use the identifiers of the std namespace globally, you must add a using directive.

Example: #include 
#include 
using namespace std;

You can then use cin and cout without any additional syntax. The header file
string has also been included. This makes the string class available and allows userfriendly string manipulations in C++. The following pages contain further details on this
topic.

䊐 Header Files in the C Programming Language
The header files standardized for the C programming language were adopted for the C++
standard and, thus, the complete functionality of the standard C libraries is available to
C++ programs.

Example: #include 
Mathematical functions are made available by this statement.
The identifiers declared in C header files are globally visible. This can cause name
conflicts in large programs. For this reason each C header file, for example name.h, is
accompanied in C++ by a second header file, cname, which declares the same identifiers
in the std namespace. Including the file math.h is thus equivalent to

Example: #include 
using namespace std;

The string.h or cstring files must be included in programs that use standard functions to manipulate C strings. These header files grant access to the functionality of the
C string library and are to be distinguished from the string header file that defines the
string class.
Each compiler offers additional header files for platform dependent functionalities.
These may be graphics libraries or database interfaces.

50

■

CHAPTER 3

USING FUNCTIONS AND CLASSES

■

USING STANDARD CLASSES

Sample program using class string
// To use strings.
#include 
#include 
using namespace std;

// Declaration of cin, cout
// Declaration of class string

int main()
{
// Defines four strings:
string prompt("What is your name: "),
name,
// An empty
line( 40, '-'),
// string with 40 '-'
total = "Hello ";
// is possible!
cout << prompt;
getline( cin, name);
total = total + name;

// Request for input.
// Inputs a name in one line
// Concatenates and
// assigns strings.

cout << line << endl
// Outputs line and name
<< total << endl;
cout << " Your name is " // Outputs length
<< name.length() << " characters long!" << endl;
cout << line << endl;
return 0;
}

✓

NOTE
Both the operators + and += for concatenation and the relational operators <, <=, >, >=, ==, and
!= are defined for objects of class string. Strings can be printed with cout and the operator <<.
The class string will be introduced in detail later on.

Sample screen output
What is your name: Rose Summer
--------------------------------------Hello Rose Summer
Your name is 11 characters long!
---------------------------------------

USING STANDARD CLASSES

■

51

Several classes are defined in the C++ standard library. These include stream classes for
input and output, but also classes for representing strings or handling error conditions.
Each class is a type with certain properties and capacities. As previously mentioned,
the properties of a class are defined by its data members and the class’s capacities are
defined by its methods. Methods are functions that belong to a class and cooperate with
the members to perform certain operations. Methods are also referred to as member functions.

䊐 Creating Objects
An object is a variable of a class type, also referred to as an instance of the class. When an
object is created, memory is allocated to the data members and initialized with suitable
values.

Example: string s("I am a string");
In this example the object s, an instance of the standard class string (or simply a
string), is defined and initialized with the string constant that follows. Objects of the
string class manage the memory space required for the string themselves.
In general, there are several ways of initializing an object of a class. A string can thus
be initialized with a certain number of identical characters, as the example on the opposite page illustrates.

䊐 Calling Methods
All the methods defined as public within the corresponding class can be called for an
object. In contrast to calling a global function, a method is always called for one particular
object. The name of the object precedes the method and is separated from the method by
a period.

Example: s.length();

// object.method();

The method length() supplies the length of a string, i.e. the number of characters in a
string. This results in a value of 13 for the string s defined above.

䊐 Classes and Global Functions
Globally defined functions exist for some standard classes. These functions perform certain
operations for objects passed as arguments. The global function getline(), for example, stores a line of keyboard input in a string.

Example: getline(cin, s);
The keyboard input is terminated by pressing the return key to create a new-line character, '\n', which is not stored in the string.

■

CHAPTER 3

USING FUNCTIONS AND CLASSES

■

exercise s

52

EXERCISES

Screen output for exercise 1
Number

Square Root

4
12.25
0.0121

2
3.5
0.11

Listing for exercise 2
// A program containing errors!
# include , 
# include 
# void srand( seed);
int main()
{
string message "\nLearn from your mistakes!";
cout << message << endl;
int len = length( message);
cout << "Length of the string: " << len << endl;
// And a random number in addition:
int a, b;
a = srand(12.5);
b = rand( a );
cout << "\nRandom number: " << b << endl;
return 0;
}

EXERCISES

■

53

Exercise 1
Create a program to calculate the square roots of the numbers
4

12.25

0.0121

and output them as shown opposite.Then read a number from the keyboard and
output the square root of this number.
To calculate the square root, use the function sqrt(), which is defined by the
following prototype in the math.h (or cmath ) header file:
double sqrt( double x);

The return value of the sqrt() function is the square root of x.

Exercise 2
The program on the opposite page contains several errors! Correct the errors
and ensure that the program can be executed.
Exercise 3
Create a C++ program that defines a string containing the following character
sequence:
I have learned something new again!

and displays the length of the string on screen.
Read two lines of text from the keyboard. Concatenate the strings using " * "
to separate the two parts of the string. Output the new string on screen.

■

CHAPTER 3

■

solutions

54

USING FUNCTIONS AND CLASSES

SOLUTIONS

Exercise 1
// Compute square roots
#include 
#include 
using namespace std;
int main()
{
double x1 = 4.0, x2 = 12.25, x3 = 0.0121;
cout << "\n
cout << "\n
<< "\n
<< "\n

Number \t Square Root"
" << x1 << "
\t " <<
" << x2 << "
\t " <<
" << x3 << "
\t " <<

<< endl;
sqrt(x1)
sqrt(x2)
sqrt(x3) << endl;

cout << "\nType a number whose square root is to be"
" computed. ";
cin >> x1;
cout << "\n
cout << "\n

Number \t Square Root" << endl;
" << x1 << " \t " << sqrt(x1) << endl;

return 0;
}

Exercise 2
// The corrected program:
#include 
#include 

// Just one header file in a line

#include 

// Prototypes of functions
// void srand( unsigned int seed);
// int rand(void);

// or:
// #include 
using namespace std;

// Introduces all names of namespace
// std into the global scope.

int main()
{
string message = "\nLearn from your mistakes!";...// =
cout << message << endl;

SOLUTIONS

■

int len = message.length();
// instead of: length(message);
cout << "Length of the string: " << len << endl;
// And another random number:
int b;
// Variable a is not needed.
srand(12);
// instead of: a = srand(12.5);
b = rand();
// instead of: b = rand(a);
cout << "\nRandom number: " << b << endl;
return 0;
}

Exercise 3
#include 
#include 
using namespace std;

// Declaration of cin, cout
// Declaration of class string

int main()
{
string message("I have learned something new again!\n"),
prompt("Please input two lines of text:"),
str1, str2, sum;
cout << message << endl;

// Outputs the message

cout << prompt << endl;

// Request for input

getline( cin, str1);
getline( cin, str2);
sum = str1 + " * " + str2;
cout << sum << endl;

// Reads the first
// and the second line of text

return 0;
}

// Concatenates, assigns
// and outputs strings.

55

This page intentionally left blank

chapter

4

Input and Output with
Streams
This chapter describes the use of streams for input and output, focusing
on formatting techniques.

57

58

■

CHAPTER 4

INPUT AND OUTPUT WITH STREAMS

■

STREAMS

Stream classes for input and output

ios

istream

ostream

iostream

The four standard streams
■
■
■
■

cin
cout
cerr
clog

Object of class istream to control standard input
Object of class ostream to control standard output
Object of class ostream to control unbuffered error output
Object of class ostream to control buffered error output

STREAMS

■

59

䊐 I/O Stream Classes
During the development of C++ a new class-based input/output system was implemented. This gave rise to the I/O stream classes, which are now available in a library of
their own, the so-called iostream library.
The diagram on the opposite page shows how a so-called class hierarchy develops due
to inheritance. The class ios is the base class of all other stream classes. It contains the
attributes and abilities common to all streams. Effectively, the ios class
■
■

manages the connection to the physical data stream that writes your program’s
data to a file or outputs the data on screen
contains the basic functions needed for formatting data. A number of flags that
determine how character input is interpreted have been defined for this purpose.

The istream and ostream classes derived from ios form a user-friendly interface
for stream manipulation. The istream class is used for reading streams and the
ostream class is used for writing to streams. The operator >> is defined in istream
and << is defined in ostream, for example.
The iostream class is derived by multiple inheritance from istream and ostream
and thus offers the functionality of both classes.
Further stream classes, a file management class, for example, are derived from the
classes mentioned above. This allows the developer to use the techniques described for
file manipulation. These classes, which also contain methods for opening and closing
files, will be discussed in a later chapter.

䊐 Standard Streams
The streams cin and cout, which were mentioned earlier, are instances of the
istream or ostream classes. When a program is launched these objects are automatically created to read standard input or write to standard output.
Standard input is normally the keyboard and standard output the screen. However,
standard input and output can be redirected to files. In this case, data is not read from
the keyboard but from a file, or data is not displayed on screen but written to a file.
The other two standard streams cerr and clog are used to display messages when
errors occur. Error messages are displayed on screen even if standard output has been
redirected to a file.

60

■

CHAPTER 4

INPUT AND OUTPUT WITH STREAMS

■

FORMATTING AND MANIPULATORS

Example: Calling a manipulator
Here the manipulator showpos is called.

↓
cout << showpos << 123;

// Output:

+123

The above statement is equivalent to
cout.setf( ios::showpos);
cout << 123;

The other positive numbers are printed with their sign as well:
cout << 22;

// Output:

+22

The output of a positive sign can be canceled by the manipulator
noshowpos:
cout << noshowpos << 123;

// Output:

123

The last statement is equivalent to
cout.unsetf( ios::showpos);
cout << 123;

✓ HINTS
■

The operators >> and << format the input and/or output according to how the flags in the base class
ios are set

■

The manipulator showpos is a function that calls the method cout.setf(ios::showpos);,
ios::showpos being the flag showpos belonging to the ios class

■

Using manipulators is easier than directly accessing flags. For this reason, manipulators are described in
the following section, whereas the methods setf() and unsetf() are used only under exceptional
circumstances.

■

Old compilers only supply some of the manipulators. In this case, you have to use the methods setf()
and unsetf().

FORMATTING AND MANIPULATORS

■

61

䊐 Formatting
When reading keyboard input, a valid input format must be used to determine how input
is to be interpreted. Similarly, screen output adheres to set of rules governing how, for
example, floating-point numbers are displayed.
The stream classes istream and ostream offer various options for performing these
tasks. For example, you can display a table of numeric values in a simple way.
In previous chapters we have looked at the cin and cout streams in statements such
as:
cout << "Please enter a number: ";
cin >> x;

The following sections systematically describe the abilities of the stream classes. This
includes:
■

■

■

the >> and << operators for formatted input and output. These operators are
defined for expressions with fundamental types—that is, for characters, boolean
values, numbers and strings.
manipulators, which can be inserted into the input or output stream. Manipulators can be used to generate formats for subsequent input/output. One manipulator that you are already familiar with is endl, which generates a line feed at the
end of a line.
other methods for determining or modifying the state of a stream and unformatted input and output.

䊐 Flags and Manipulators
Formatting flags defined in the parent class ios determine how characters are input or
output. In general, flags are represented by individual bits within a special integral variable. For example, depending on whether a bit is set or not, a positive number can be
output with or without a plus sign.
Each flag has a default setting. For example, integral numbers are output as decimals by
default, and positive numbers are output without a plus sign.
It is possible to modify individual formatting flags. The methods setf() and
unsetf() can be used for this purpose. However, the same effect can be achieved simply by using so-called manipulators, which are defined for all important flags. Manipulators are functions that can be inserted into the input or output stream and thus be called.

62

■

CHAPTER 4

INPUT AND OUTPUT WITH STREAMS

■

FORMATTED OUTPUT OF INTEGERS

Manipulators formatting integers
Manipulator

Effects

oct

Octal base

hex

Hexadecimal base

dec

Decimal base (by default)

showpos

Generates a + sign in non-negative numeric
output.

noshowpos

Generates non-negative numeric output
without a + sign (by default).

uppercase

Generates capital letters in hexadecimal
output.

nouppercase

Generates lowercase letters in hexadecimal
output (by default).

Sample program
// Reads integral decimal values and
// generates octal, decimal, and hexadecimal output.
#include 
using namespace std;

// Declarations of cin, cout and
// manipulators oct, hex, ...

int main()
{
int number;
cout << "Please enter an integer: ";
cin >> number;
cout << uppercase
// for hex-digits
<< " octal \t decimal \t hexadecimal\n "
<< oct << number << "
\t "
<< dec << number << "
\t "
<< hex << number << endl;
return 0;
}

FORMATTED OUTPUT OF INTEGERS

■

63

䊐 Formatting Options
The << operator can output values of type short, int, long or a corresponding
unsigned type. The following formatting options are available:
■
■
■

define the numeric system in which to display the number: decimal, octal, or
hexadecimal
use capitals instead of small letters for hexadecimals
display a sign for positive numbers.

In addition, the field width can be defined for the above types. The field width can
also be defined for characters, strings, and floating-point numbers, and will be discussed
in the following sections.

䊐 Numeric System
Integral numbers are displayed as decimals by default. The manipulators oct, hex, and
dec can be used for switching from and to decimal display mode.

Example: cout << hex << 11;

// Output: b

Hexadecimals are displayed in small letters by default, that is, using a, b, ..., f. The
manipulator uppercase allows you to use capitals.

Example: cout << hex << uppercase << 11; //Output: B
The manipulator nouppercase returns the output format to small letters.

䊐 Negative Numbers
When negative numbers are output as decimals, the output will always include a sign.
You can use the showpos manipulator to output signed positive numbers.

Example: cout << dec << showpos << 11; //Output: +11
You can use noshowpos to revert to the original display mode.
When octal or hexadecimal numbers are output, the bits of the number to be output are
always interpreted as unsigned! In other words, the output shows the bit pattern of a
number in octal or hexadecimal format.

Example: cout << dec << -1 << "

" << hex << -1;

This statement causes the following output on a 32-bit system:
-1

ffffffff

64

■

CHAPTER 4

INPUT AND OUTPUT WITH STREAMS

■

FORMATTED OUTPUT OF FLOATING-POINT NUMBERS

Manipulators formatting floating-point numbers
Manipulator

showpoint

Effects
Generates a decimal point character
shown in floating-point output. The
number of digits after the decimal point
corresponds to the used precision.

noshowpoint

Trailing zeroes after the decimal point
are not printed.
If there are no digits after the decimal
point, the decimal point is not printed
(by default).

fixed

Output in fixed point notation

scientific

Output in scientific notation

setprecision (int n)

Sets the precision to n.

Methods for precision
Manipulator

✓

Effects

int precision (int n);

Sets the precision to n.

int precision() const;

Returns the used precision.

NOTE
The key word const within the prototype of precision() signifies that the method performs only
read operations.

Sample program
#include 
using namespace std;
int main()
{
double x = 12.0;
cout.precision(2);
// Precision 2
cout << " By default:
" << x << endl;
cout << " showpoint: " << showpoint << x << endl;
cout << " fixed:
" << fixed
<< x << endl;
cout << " scientific: " << scientific << x << endl;
return 0;
}

FORMATTED OUTPUT OF FLOATING-POINT NUMBERS

■

65

䊐 Standard Settings
Floating-points are displayed to six digits by default. Decimals are separated from the
integral part of the number by a decimal point. Trailing zeroes behind the decimal point
are not printed. If there are no digits after the decimal point, the decimal point is not
printed (by default).

Examples: cout << 1.0;
cout << 1.234;
cout << 1.234567;

// Output: 1
// Output: 1.234
// Output: 1.23457

The last statement shows that the seventh digit is not simply truncated but rounded.
Very large and very small numbers are displayed in exponential notation.

Example: cout << 1234567.8; // Output: 1.23457e+06

䊐 Formatting
The standard settings can be modified in several ways. You can
■
■
■

change the precision, i.e. the number of digits to be output
force output of the decimal point and trailing zeroes
stipulate the display mode (fixed point or exponential).

Both the manipulator setprecision()and the method precision() can be used to
redefine precision to be used.

Example: cout << setprecision(3);

// Precision: 3
// or: cout.precision(3);
cout << 12.34;
// Output: 12.3

Note that the header file iomanip must be included when using the manipulator setprecision(). This also applies to all standard manipulators called with at least one
argument.
The manipulator showpoint outputs the decimal point and trailing zeroes. The
number of digits being output (e.g. 6) equals the current precision.

Example: cout << showpoint << 1.0; // Output: 1.00000
However, fixed point output with a predetermined number of decimal places is often more
useful. In this case, you can use the fixed manipulator with the precision defining the
number of decimal places. The default value of 6 is assumed in the following example.

Example: cout << fixed << 66.0;

// Output: 66.000000

In contrast, you can use the scientific manipulator to specify that floating-point
numbers are output as exponential expressions.

66

■

CHAPTER 4

INPUT AND OUTPUT WITH STREAMS

■

OUTPUT IN FIELDS

Element functions for output in fields
Method

Effects

int width() const;

Returns the minimum field width used

int width(int n);

Sets the minimum field width to n

int fill() const;

Returns the fill character used

int fill(int ch);

Sets the fill character to ch

Manipulators for output in fields
Manipulator

Effects

setw(int n)

Sets the minimum field width to n

setfill(int ch)

Sets the fill character to ch

left

Left-aligns output in fields

right

Right-aligns output in fields

internal

Left-aligns output of the sign and
right-aligns output of the numeric
value

✓

NOTE
The manipulators setw() and setfill() are declared in the header file iomanip .

Examples
#include 
#include 
using namespace std;

// Obligatory
// declarations

1st Example: cout << '|' << setw(6) << 'X' << '|';
Output:
|
X|
// Field width 6
2nd Example: cout << fixed << setprecision(2)

Output:

<< setw(10) << 123.4 << endl
<< "1234567890" << endl;
123.40
// Field width 10
1234567890

OUTPUT IN FIELDS

■

67

The << operator can be used to generate formatted output in fields. You can
■
■
■

specify the field width
set the alignment of the output to right- or left-justified
specify a fill-character with which to fill the field.

䊐 Field Width
The field width is the number of characters that can be written to a field. If the output
string is larger than the field width, the output is not truncated but the field is extended.
The output will always contain at least the number of digits specified as the field width.
You can either use the width() method or the setw() manipulator to define field
width.

Example: cout.width(6);

// or:

cout << setw(6);

One special attribute of the field width is the fact that this value is non-permanent:
the field width specified applies to the next output only, as is illustrated by the examples
on the opposite page. The first example outputs the character 'X' to a field with width
of 6, but does not output the '|' character.
The default field width is 0. You can also use the width() method to get the current
field width. To do so, call width() without any other arguments.

Example: int fieldwidth = cout.width();

䊐 Fill Characters and Alignment
If a field is larger than the string you need to output, blanks are used by default to fill the
field. You can either use the fill() method or the setfill() manipulator to specify
another fill character.

Example: cout << setfill('*') << setw(5) << 12;
// Output: ***12

The fill character applies until another character is defined.
As the previous example shows, output to fields is normally right-aligned. The other
options available are left-aligned and internal, which can be set by using the manipulators left and internal. The manipulator internal left-justifies the sign and rightjustifies the number within a field.

Example: cout.width(6); cout.fill('0');
cout << internal << -123; // Output: -00123

68

■

CHAPTER 4

INPUT AND OUTPUT WITH STREAMS

■

OUTPUT OF CHARACTERS, STRINGS, AND BOOLEAN VALUES

Sample program
// Enters a character and outputs its
// octal, decimal, and hexadecimal code.
#include 
#include 

// Declaration of cin, cout
// For manipulators being called
// with arguments.

#include 
using namespace std;
int main()
{
int number = ' ';
cout << "The white space code is as follows: "
<< number << endl;
char ch;
string prompt =
"\nPlease enter a character followed by "
" : ";
cout << prompt;
cin >> ch;
number = ch;

// Read a character

cout << "The character " << ch
<< " has code" << number << endl;
cout <<
<<
<<
<<
<<

uppercase
// For hex-digits
"
octal decimal hexadecimal\n "
oct << setw(8) << number
dec << setw(8) << number
hex << setw(8) << number << endl;

return 0;
}

OUTPUT OF CHARACTERS, STRINGS, AND BOOLEAN VALUES

■

69

䊐 Outputting Characters and Character Codes
The >> operator interprets a number of type char as the character code and outputs the
corresponding character:

Example: char ch = '0';
cout << ch << ' ' << 'A';
// Outputs three characters: 0 A

It is also possible to output the character code for a character. In this case the character
code is stored in an int variable and the variable is then output.

Example: int code = '0';
cout << code;

// Output: 48

The '0' character is represented by ASCII Code 48. The program on the opposite page
contains further examples.

䊐 Outputting Strings
You can use the >> operator both to output string literals, such as "Hello", and string
variables, as was illustrated in previous examples. As in the case of other types, strings
can be positioned within output fields.

Example: string s("spring flowers ");
cout << left
// Left-aligned
<< setfill('?')
// Fill character ?
<< setw(20) << s ; // Field width 20

This example outputs the string "spring flowers??????". The manipulator
right can be used to right-justify the output within the field.

䊐 Outputting Boolean Values
By default the << operator outputs boolean values as integers, with the value 0 representing false and 1 true. If you need to output the strings true or false instead, the
flag ios::boolalpha must be set. To do so, use either the setf() method or the
manipulator boolalpha.

Example: bool ok = true;
cout << ok << endl
<< boolalpha << ok << endl;

// 1
// true

You can revert this setting using the noboolalpha manipulator.

70

■

CHAPTER 4

INPUT AND OUTPUT WITH STREAMS

■

FORMATTED INPUT

Sample program
// Inputs an article label and a price
#include 
#include 
#include 
using namespace std;

// Declarations of cin, cout,...
// Manipulator setw()

int main()
{
string label;
double price;
cout << "\nPlease enter an article label: ";
// Input the label (15 characters maximum):
cin >> setw(16);
// or: cin.width(16);
cin >> label;
cin.sync();
cin.clear();

// Clears the buffer and resets
// any error flags that may be set

cout << "\nEnter the price of the article: ";
cin >> price;
// Input the price
// Controlling output:
cout << fixed << setprecision(2)
<< "\nArticle:"
<< "\n Label: " << label
<< "\n Price: " << price << endl;
// ... The program to be continued
return 0;
}

✓

NOTE
The input buffer is cleared and error flags are reset by calling the sync() and clear() methods. This
ensures that the program will wait for new input for the price, even if more than 15 characters have
been entered for the label.

FORMATTED INPUT

■

71

The >> operator, which belongs to the istream class, takes the current number base
and field width flags into account when reading input:
■
■

the number base specifies whether an integer will be read as a decimal, octal, or
hexadecimal
the field width specifies the maximum number of characters to be read for a
string.

When reading from standard input, cin is buffered by lines. Keyboard input is thus
not read until confirmed by pressing the  key. This allows the user to press the
backspace key and correct any input errors, provided the return key has not been pressed.
Input is displayed on screen by default.

䊐 Input Fields
The >> operator will normally read the next input field, convert the input by reference to
the type of the supplied variable, and write the result to the variable. Any white space
characters (such as blanks, tabs, and new lines) are ignored by default.

Example: char ch;
cin >> ch;

// Enter a character

When the following keys are pressed
    

the character 'X' is stored in the variable ch.
An input field is terminated by the first white space character or by the first character
that cannot be processed.

Example: int i;
cin >> i;

Typing 123FF stores the decimal value 123 in the variable i. However, the
characters that follow, FF and the newline character, remain in the input buffer and will
be read first during the next read operation.
When reading strings, only one word is read since the first white space character will
begin a new input field.

Example: string city;
cin >> city;

// To read just one word!

If Lao Kai is input, only Lao will be written to the city string. The number of characters to be read can also be limited by specifying the field width. For a given field width of
n, a maximum of n–1 characters will be read, as one byte is required for the null character. Any initial white space will be ignored. The program on the opposite page illustrates
this point and also shows how to clear the input buffer.

72

■

CHAPTER 4

INPUT AND OUTPUT WITH STREAMS

■

FORMATTED INPUT OF NUMBERS

Sample program
// Enter hexadecimal digits and a floating-point number
//
#include 
#include 
using namespace std;
int main()
{
int number = 0;
cout << "\nEnter a hexadecimal number: "
<< endl;
cin >> hex >> number;
// Input hex-number
cout << "Your decimal input: " << number << endl;
// If an invalid input occurred:
cin.sync();
// Clears the buffer
cin.clear();
// Reset error flags
double x1 = 0.0, x2 = 0.0;
cout << "\nNow enter two floating-point values: "
<< endl;
cout
cin
cout
cin

<<
>>
<<
>>

"1. number: ";
x1;
"2. number: ";
x2;

// Read first number
// Read second number

cout << fixed << setprecision(2)
<< "\nThe sum of both numbers:
<< setw(10) << x1 + x2 << endl;

"

cout << "\nThe product of both numbers: "
<< setw(10) << x1 * x2 << endl;
return 0;
}

FORMATTED INPUT OF NUMBERS

■

73

䊐 Inputting Integers
You can use the hex, oct, and dec manipulators to stipulate that any character
sequence input is to processed as a hexadecimal, octal, or decimal number.

Example: int n;
cin >> oct >> n;

An input value of 10 will be interpreted as an octal, which corresponds to a decimal
value of 8.

Example: cin >> hex >> n;
Here, any input will be interpreted as a hexadecimal, enabling input such as f0a or -F7.

䊐 Inputting Floating-Point Numbers
The >> operator interprets any input as a decimal floating-point number if the variable is
a floating-point type, i.e. float, double, or long double. The floating-point number can be entered in fixed point or exponential notation.

Example: double x;
cin >> x;

The character input is converted to a double value in this case. Input, such as 123,
-22.0, or 3e10 is valid.

䊐 Input Errors
But what happens if the input does not match the type of variable defined?

Example: int i, j;

cin >> i >> j;

Given input of 1A5 the digit 1 will be stored in the variable i. The next input field
begins with A. But since a decimal input type is required, the input sequence will not be
processed beyond the letter A. If, as in our example, no type conversion is performed, the
variable is not written to and an internal error flag is raised.
It normally makes more sense to read numerical values individually, and clear the
input buffer and any error flags that may have been set after each entry.
Chapter 6, “Control Flow,” and Chapter 28, “Exception Handling,” show how a program can react to input errors.

74

■

CHAPTER 4

INPUT AND OUTPUT WITH STREAMS

■

UNFORMATTED INPUT/OUTPUT

Sample program
//
//

Reads a text with the operator >>
and the function getline().

#include 
#include 
using namespace std;
string header =
"
--- Demonstrates Unformatted Input ---";
int main()
{
string word, rest;
cout << header
<< "\n\nPress  to go on" << endl;
cin.get();

// Read the new line
// without saving.

cout << "\nPlease enter a sentence with several words!"
<< "\nEnd with  and ."
<< endl;
cin >> word;
getline( cin, rest, '!');

// Read the first word
// and the remaining text
// up to the character !

cout << "\nThe first word:
" << word
<< "\nRemaining text: " << rest << endl;
return 0;
}

✓

NOTE
1. A text of more than one line can be entered.
2. The sample program requires that at least one word and a following white space are entered.

UNFORMATTED INPUT/OUTPUT

■

75

Unformatted input and output does not use fields, and any formatting flags that have
been set are ignored. The bytes read from a stream are passed to the program “as is.”
More specifically, you should be aware that any white space characters preceding the
input will be processed.

䊐 Reading and Writing Characters
You can use the methods get() and put() to read or write single characters. The
get() method reads the next character from a stream and stores it in the given char
variable.

Example: char ch;
cin.get(ch);

If the character is a white space character, such as a newline, it will still be stored in the
ch variable. To prevent this from happening you can use
cin >> ch;

to read the first non-white space character.
The get() method can also be called without any arguments. In this case, get()
returns the character code of type int.

Example: int c = cin.get();
The put() method can be used for unformatted output of a character. The character to
be output is passed to put() as an argument.

Example: cout.put('A');
This statement is equivalent to cout << 'A'; , where the field width is undefined or
has been set to 1.

䊐 Reading a Line
The >> operator can only be used to read one word into a string. If you need to read a
whole line of text, you can use the global function getline(), which was introduced
earlier in this chapter.

Example: getline(cin, text);
This statement reads characters from cin and stores them in the string variable text
until a new line character occurs. However, you can specify a different delimiting character by passing the character to the getline() function as a third argument.

Example: getline(cin, s, '.');
The delimiting character is read, but not stored in the string. Any characters subsequent
to the first period will remain in the input buffer of the stream.

■

CHAPTER 4

INPUT AND OUTPUT WITH STREAMS

■

exercise s

76

EXERCISES

Screen output for exercise 3
Article Number
.......

Number of Pieces Price per piece
......
...... Dollar

Program listing for exercise 5
// A program with resistant mistakes
#include 
using namespace std;
int main()
{
char ch;
string word;
cin >> "Let's go! Press the return key: " >> ch;
cout << "Enter a word containing
three characters at most: ";
cin

>> setprecision(3) >> word;

cout >> "Your input: " >> ch >> endl;
return 0;
}

EXERCISES

■

77

Exercise 1
What output is generated by the program on the page entitled “Formatted output
of floating-point numbers” in this chapter?
Exercise 2
Formulate statements to perform the following:
a. Left-justify the number 0.123456 in an output field with a width of 15.
b. Output the number 23.987 as a fixed point number rounded to two decimal places, right-justifying the output in a field with a width of 12.
c. Output the number –123.456 as an exponential and with four decimal
spaces. How useful is a field width of 10?

Exercise 3
Write a C++ program that reads an article number, a quantity, and a unit price
from the keyboard and outputs the data on screen as displayed on the opposite
page.

✓

Exercise 4
Write a C++ program that reads any given character code (a positive integer)
from the keyboard and displays the corresponding character and the character
code as a decimal, an octal, and a hexadecimal on screen.
TIP
The variable type defines whether a character or a number is to be read or output.

Why do you think the character P is output when the number 336 is entered?

Exercise 5
Correct the mistakes in the program on the opposite page.

■

CHAPTER 4

■

solutions

78

INPUT AND OUTPUT WITH STREAMS

SOLUTIONS

Exercise 1
Output of a sample program formatting floating-point numbers:
By default:
showpoint:
fixed:
scientific:

12
12.
12.00
1.20e+001

Exercise 2
#include 
#include 
using namespace std;

// For setw() and setprecision()

int main()
{
double x1 = 0.123456, x2 = 23.987, x3 = -123.456;
// a)
cout << left << setw(15) << x1 << endl;
// b)
cout << fixed << setprecision(2) << right << setw(12)
<< x2 << endl;
// c)
cout << scientific << setprecision(4) << x3 << endl;
// Output: -1.2346e+002
// A field width of 12 or more would be convenient!
return 0;
}

Exercise 3
// Input and formatted output of article characteristics.
#include 
#include 
using namespace std;
int main()
{
long number = 0;
int count = 0;
double price = 0.0;
// Input:
cout << "\nPlease enter article characteristics.\n";
cout << "Article number: ";
cin >> number;

SOLUTIONS

cout << "Number of pieces:
cin >> count;
cout << "Price per piece:
cin >> price;

";

";

// Output:
cout <<
"\n\tArticle Number
cout <<
<<
<<
<<
<<

"\n\t"
setw(8)
setw(16)
fixed
setw(16)

<<
<<
<<
<<

Quantity

Price per piece ";

number
count
setprecision(2)
price << " Dollar" << endl;

return 0;
}

Exercise 4
#include 
#include 
using namespace std;

// Manipulator setw()

int main()
{
unsigned char c = 0;
unsigned int code = 0;
cout << "\nPlease enter a decimal character code: ";
cin >> code;
c = code;

// Save for output

cout << "\nThe corresponding character: " << c << endl;
code = c;
cout <<
<<
<<
<<
<<

// Character code. Is only
// necessary, if input is > 255.
"\nCharacter codes"
"\n decimal:
" << setw(3) << dec << code
"\n octal:
" << setw(3) << oct << code
"\n hexadecimal: " << setw(3) << hex << code
endl;

return 0;
}

■

79

80

■

CHAPTER 4

INPUT AND OUTPUT WITH STREAMS

When entering 336, the value 80 is stored in the low byte of variable code
(336 = 256 + 80).Thus after the assignment, the variable c contains the value
80, representing the character P.

Exercise 5
The corrected program:
// Corrections are commented.
//
#include 
#include 
// Manipulator setw()
#include 
// Class string
using namespace std;
int main()
{
string word;

// To read a word.
// char ch; is not needed.

// cout << ...instead of
cout << "Let's go! Press the return key: ";
cin.get();

cin >> .

// Input newline character

cout << " Enter a word "
// "
"containing three characters at the most: ";// "
cin

>> setw(3) >> word;

cout << "Your input: "
<< word << endl;
return 0;
}

// setw(3) instead of
// setprecision(3)
// <<
// instead of

>> ch

chapter

5

Operators for
Fundamental Types
In this chapter, operators needed for calculations and selections are
introduced. Overloading and other operators, such as those needed for
bit manipulations, are introduced in later chapters.

81

82

■

CHAPTER 5

OPERATORS FOR FUNDAMENTAL TYPES

■

BINARY ARITHMETIC OPERATORS

Binary operator and operands

Operator

a + b
Left operand

Right operand

The binary arithmetic operators
Operator

Significance

+
-

Addition

*

Multiplication

/
%

Division

Subraction

Remainder

Sample program
#include 
using namespace std;
int main()
{
double x, y;
cout << "\nEnter two floating-point values: ";
cin >> x >> y;
cout << "The average of the two numbers is: "
<< (x + y)/2.0 << endl;
return 0;
}

Sample output for the program
Enter two floating-point values: 4.75
12.3456
The average of the two numbers is: 8.5478

BINARY ARITHMETIC OPERATORS

■

83

If a program is to be able to process the data input it receives, you must define the operations to be performed for that data. The operations being executed will depend on the
type of data — you could add, multiply, or compare numbers, for example. However, it
would make no sense at all to multiply strings.
The following sections introduce you to the most important operators that can be
used for arithmetic types. A distinction is made between unary and binary operators. A
unary operator has only one operand, whereas a binary operator has two.

䊐 Binary Arithmetic Operators
Arithmetic operators are used to perform calculations. The opposite page shows an
overview. You should be aware of the following:
■

■

Divisions performed with integral operands will produce integral results; for example, 7/2 computes to 3. If at least one of the operands is a floating-point number,
the result will also be a floating-point number; e.g., the division 7.0/2 produces
an exact result of 3.5.
Remainder division is only applicable to integral operands and returns the remainder of an integral division. For example, 7%2 computes to 1.

䊐 Expressions
In its simplest form an expression consists of only one constant, one variable, or one
function call. Expressions can be used as the operands of operators to form more complex
expressions. An expression will generally tend to be a combination of operators and
operands.
Each expression that is not a void type returns a value. In the case of arithmetic
expressions, the operands define the type of the expression.

Examples: int a(4);

double
a * 512
//
1.0 + sin(x)
//
x – 3
//
//

x(7.9);
Type int
Type double
Type double, since one
operand is of type double

An expression can be used as an operand in another expression.

Example: 2 + 7 * 3

// Adds 2 and 21

Normal mathematical rules (multiplication before addition) apply when evaluating an
expression, i.e. the *, /, % operators have higher precedence than + and -. In our example, 7*3 is first calculated before adding 2. However, you can use parentheses to apply a
different precedence order.

Example: (2 + 7) * 3

// Multiplies 9 by 3.

84

■

CHAPTER 5

OPERATORS FOR FUNDAMENTAL TYPES

■

UNARY ARITHMETIC OPERATORS

The unary arithmetic operators
Operator

Significance

-

+

Unary sign operators

++

Increment operator

--

Decrement operator

Precedence of arithmetic operators
Precedence

Grouping

Operator

High

--

++
+

-- (prefix)
- (sign)

*

/

+
-

Low

(postfix)

++

left to right
right to left

left to right

%

left to right

(addition)
(subtraction)

Effects of prefix and postfix notation
#include 
using namespace std;
int main()
{
int i(2), j(8);
cout
cout
cout
cout

<<
<<
<<
<<

i++
i
j---j

return 0;
}

<<
<<
<<
<<

endl;
endl;
endl;
endl;

//
//
//
//

Output:
Output:
Output:
Output:

2
3
8
6

UNARY ARITHMETIC OPERATORS

■

85

There are four unary arithmetic operators: the sign operators + and -, the increment
operator ++, and the decrement operator --.

䊐 Sign Operators
The sign operator – returns the value of the operand but inverts the sign.

Example: int n = –5;

cout << -n;

// Output: 5

The sign operator + performs no useful operation, simply returning the value of its
operand.

䊐 Increment / Decrement Operators
The increment operator ++ modifies the operand by adding 1 to its value and cannot be
used with constants for this reason.
Given that i is a variable, both i++ (postfix notation) and ++i (prefix notation) raise
the value of i by 1. In both cases the operation i = i + 1 is performed.
However, prefix ++ and postfix ++ are two different operators. The difference
becomes apparent when you look at the value of the expression; ++i means that the
value of i has already been incremented by 1, whereas the expression i++ retains the
original value of i. This is an important difference if ++i or i++ forms part of a more
complex expression:
++i
i++

i is incremented first and the new value of i is then applied,
the original value of i is applied before i is incremented.

The decrement operator -- modifies the operand by reducing the value of the
operand by 1. As the sample program opposite shows, prefix or postfix notation can be
used with --.

䊐 Precedence
How is an expression with multiple operators evaluated?

Example: float val(5.0);

cout << val++ – 7.0/2.0;

Operator precedence determines the order of evaluation, i.e. how operators and
operands are grouped. As you can see from the table opposite, ++ has the highest precedence and / has a higher precedence than -. The example is evaluated as follows:
(val++) – (7.0/2.0). The result is 1.5, as val is incremented later.
If two operators have equal precedence, the expression will be evaluated as shown in
column three of the table.

Example: 3 * 5 % 2

is equivalent to

(3 * 5) % 2

86

■

CHAPTER 5

OPERATORS FOR FUNDAMENTAL TYPES

■

ASSIGNMENTS

Sample program
// Demonstration of compound assignments
#include 
#include 
using namespace std;
int main()
{
float x, y;
cout << "\n Please enter a starting value:
cin >> x;

";

cout << "\n Please enter the increment value: ";
cin >> y;
x += y;
cout << "\n And now multiplication! ";
cout << "\n Please enter a factor: ";
cin >> y;
x *= y;
cout << "\n Finally division.";
cout << "\n Please supply a divisor: ";
cin >> y;
x /= y;
cout << "\n And this is "
<< "your current lucky number: "
// without digits after
// the decimal point:
<< fixed << setprecision(0)
<< x << endl;
return 0;
}

ASSIGNMENTS

■

87

䊐 Simple Assignments
A simple assignment uses the assignment operator = to assign the value of a variable to an
expression. In expressions of this type the variable must be placed on the left and the
assigned value on the right of the assignment operator.

Examples: z = 7.5;
y = z;
x = 2.0 + 4.2 * z;

The assignment operator has low precedence. In the case of the last example, the right
side of the expression is first evaluated and the result is assigned to the variable on the
left.
Each assignment is an expression in its own right, and its value is the value assigned.

Example: sin(x = 2.5);
In this assignment the number 2.5 is assigned to x and then passed to the function as an
argument.
Multiple assignments, which are always evaluated from right to left, are also possible.

Example: i = j = 9;
In this case the value 9 is first assigned to j and then to i.

䊐 Compound Assignments
In addition to simple assignment operators there are also compound assignment operators that simultaneously perform an arithmetic operation and an assignment, for example.

Examples. i += 3;
i *= j + 2;

is equivalent to i = i + 3;
is equivalent to i = i * (j+2);

The second example shows that compound assignments are implicitly placed in parentheses, as is demonstrated by the fact that the precedence of the compound assignment is
just as low as that of the simple assignment.
Compound assignment operators can be composed from any binary arithmetic operator (and, as we will see later, with bit operators). The following compound operators are
thus available: +=, -=, *=, /=, and %=.
You can modify a variable when evaluating a complex expression by means of an
assignment or the ++, -- operators. This technique is referred to as a side effect. Avoid
use of side effects if possible, as they often lead to errors and can impair the readability of
your programs.

88

■

CHAPTER 5

OPERATORS FOR FUNDAMENTAL TYPES

■

RELATIONAL OPERATORS

The relational operators
Operator

Significance
less than

<

less than or equal to

<=

greater than

>
>=

geater than or equal to

==

equal

!=

unequal

Precedence of relational operators
Precedence

Operator

High

arithmetic operators

<

Low

<=

>

==

!=

>=

assignment operators

Examples for comparisons:
Comparison

Result

5 >= 6

false

1.7 < 1.8

true

4 + 2 == 5

false

2 * 4 != 7

true

RELATIONAL OPERATORS

■

89

䊐 The Result of Comparisons
Each comparison in C++ is a bool type expression with a value of true or false,
where true means that the comparison is correct and false means that the comparison is incorrect.

Example: length == circuit

// false or true

If the variables length and circuit contain the same number, the comparison is
true and the value of the relational expression is true. But if the expressions contain
different values, the value of the expression will be false.
When individual characters are compared, the character codes are compared. The
result therefore depends on the character set you are using. The following expression
results in the value true when ASCII code is used.

Example: 'A' < 'a'

// true, since 65 < 97

䊐 Precedence of Relational Operators
Relational operators have lower precedence than arithmetic operators but higher precedence than assignment operators.

Example: bool flag = index < max – 1;
In our example, max – 1 is evaluated first, then the result is compared to index, and the
value of the relational expression (false or true) is assigned to the flag variable.
Similarly, in the following

Example: int result;
result = length + 1 == limit;
length + 1 is evaluated first, then the result is compared to limit, and the value of
the relational expression is assigned to the result variable. Since result is an int
type, a numerical value is assigned instead of false or true, i.e. 0 for false and 1 for
true.
It is quite common to assign a value before performing a comparison, and parentheses
must be used in this case.

Example: (result = length + 1) == limit
Our example stores the result of length + 1 in the variable result and then compares
this expression with limit.

✓

NOTE
You cannot use the assignment operator = to compare two expressions. The compiler will not generate
an error message if the value on the left is a variable. This mistake has caused headaches for lots of
beginners when troubleshooting their programs.

90

■

CHAPTER 5

OPERATORS FOR FUNDAMENTAL TYPES

■

LOGICAL OPERATORS

“Truth” table for logical operators

A

B

A && B

A || B

true

true

true

true

true

false

false

true

false

true

false

true

false

false

false

false

A

!A

true

false

false

true

Examples for logical expressions

✓

x

y

Logical Expression

Result

1

-1

x <= y || y >=0

false

0

0

x > -2 && y == 0

true

-1

0

x && !y

true

0

1

!(x+1) || y - 1 > 0

false

NOTE
A numeric value, such as x or x+1, is interpreted as “false” if its value is 0. Any value other than 0 is
interpreted as “true.”

LOGICAL OPERATORS

■

91

The logical operators comprise the boolean operators && (AND), || (OR), and ! (NOT).
They can be used to create compound conditions and perform conditional execution of a
program depending on multiple conditions.
A logical expression results in a value false or true, depending on whether the logical expression is correct or incorrect, just like a relational expression.

䊐 Operands and Order of Evaluation
The operands for boolean type operators are of the bool type. However, operands of any
type that can be converted to bool can also be used, including any arithmetic types. In
this case the operand is interpreted as false, or converted to false, if it has a value of
0. Any other value than 0 is interpreted as true.
The OR operator || will return true only if at least one operand is true, so the value
of the expression

Example: (length < 0.2) || (length > 9.8)
is true if length is less than 0.2 or greater than 9.8.
The AND operator && will return true only if both operands are true, so the logical
expression

Example: (index < max) && (cin >> number)
is true, provided index is less than max and a number is successfully input. If the condition index < max is not met, the program will not attempt to read a number! One
important feature of the logical operators && and || is the fact that there is a fixed order
of evaluation. The left operand is evaluated first and if a result has already been ascertained, the right operand will not be evaluated!
The NOT operator ! will return true only if its operand is false. If the variable flag
contains the value false (or the value 0), !flag returns the boolean value true.

䊐 Precedence of Boolean Operators
The && operator has higher precedence than ||. The precedence of both these operators
is higher than the precedence of an assignment operator, but lower than the precedence
of all previously used operators. This is why it was permissible to omit the parentheses in
the examples earlier on in this chapter.
The ! operator is a unary operator and thus has higher precedence. Refer to the table
of precedence in the Appendix for further details.

■

CHAPTER 5

OPERATORS FOR FUNDAMENTAL TYPES

■

exercise s

92

EXERCISES

Program listing for exercise 4
// Evaluating operands in logical expressions.
#include 
using namespace std;
int main()
{
cout << boolalpha; // Outputs boolean values
// as true or false
bool res = false;
int y =
res = 7
cout <<
<<
cout <<
int

5;
|| (y = 0);
"Result of (7 || (y = 0)): " << res
endl;
"Value of y: " << y << endl;

a, b, c;

a = b = c = 0;
res = ++a || ++b
cout << '\n'
<< " res =
<< ",
a =
<< ",
b =
<< ",
c =
a = b = c = 0;
res = ++a && ++b
cout << " res =
<< ",
a =
<< ",
b =
<< ",
c =
return 0;
}

&& ++c;
"
"
"
"

<<
<<
<<
<<

res
a
b
c << endl;

|| ++c;
" << res
" << a
" << b
" << c << endl;

EXERCISES

■

Exercise 1
What values do the following arithmetic expressions have?
a. 3/10
d. 3 + 4 % 5

b. 11%4
e. 3 * 7 % 4

c. 15/2.0
f. 7 % 4 * 3

Exercise 2
a. How are operands and operators in the following expression associated?
x = –4 * i++ – 6 % 4;

Insert parentheses to form equivalent expressions.
b. What value will be assigned in part a to the variable x if the variable i has a
value of –2?

Exercise 3
The int variable x contains the number 7. Calculate the value of the following
logical expressions:
a. x < 10 && x >= –1
b. !x && x >= 3
c. x++ == 8 || x == 7

Exercise 4
What screen output does the program on the opposite page generate?

93

■

CHAPTER 5

■

solutions

94

OPERATORS FOR FUNDAMENTAL TYPES

SOLUTIONS

Exercise 1
a. 0
d. 7

b. 3
e. 1

c. 7.5
f. 9

Exercise 2
a. x = ( ((–4) * (i++)) – (6 % 4) )
b. The value 6 will be assigned to the variable x.

Exercise 3
a. true
b. false
c. false

Exercise 4
Result of (7 || (y = 0)): true
Value of y: 5
res = true,
res = true,

a = 1,
a = 1,

b = 0,
b = 1,

c = 0
c = 0

chapter

6

Control Flow
This chapter introduces the statements needed to control the flow of a
program.These are
■

loops with while, do-while, and for

■

selections with if-else, switch, and the conditional operator

■

jumps with goto, continue, and break.

95

96

■

CHAPTER 6

CONTROL FLOW

■

THE while STATEMENT

Structogram for while

As long as the expression is true

statement

Sample program
// average.cpp
// Computing the average of numbers
#include 
using namespace std;
int main()
{
int x, count = 0;
float sum = 0.0;
cout << "Please enter some integers:\n"
"(Break with any letter)"
<< endl;
while( cin >> x )
{
sum += x;
++count;
}
cout << "The average of the numbers: "
<< sum / count << endl;
return 0;
}

Sample output from the above program
Please enter some integers:
(Break with any letter)
9 10 12q
The average of the numbers: 10.3333

THE WHILE STATEMENT

■

97

Loops are used to perform a set of instructions repeatedly. The set of instructions to be
iterated is called the loop body. C++ offers three language elements to formulate iteration
statements: while, do-while, and for. The number of times a loop is repeated is
defined by a controlling expression. In the case of while and for statements this expression is verified before the loop body is executed, whereas a do-while loop is performed
once before testing.
The while statement takes the following format:

Syntax:

while( expression )
statement
// loop body

When entering the loop, the controlling expression is verified, i.e. the expression is
evaluated. If this value is true, the loop body is then executed before the controlling
expression is evaluated once more.
If the controlling expression is false, i.e. expression evaluates to false, the program goes on to execute the statement following the while loop.
It is common practice to place the loop body in a new line of the source code and to
indent the statement to improve the readability of the program.

Example: int count = 0;
while( count < 10)
cout << ++count << endl;

As this example illustrates, the controlling expression is normally a boolean expression.
However, the controlling expression might be any expression that can be converted to
the bool type including any arithmetic expressions. As we already learned from the section on boolean operators, the value 0 converts to false and all other values convert to
true.

䊐 Building Blocks
If you need to repeat more than one statement in a program loop, you must place the
statements in a block marked by parentheses { }. A block is syntactically equivalent to a
statement, so you can use a block wherever the syntax requires a statement.
The program on the opposite page calculates the average of a sequence of integers
input via the keyboard. Since the loops contains two statements, the statements must be
placed in a block.
The controlling expression cin >> x is true provided the user inputs an integer.
The result of converting the expression cin >> x to a bool type will be true for any
valid input and false in any other case. Invalid input, if the user types a letter instead
of an integer, for example, terminates the loop and executes the next statement.

98

■

CHAPTER 6

CONTROL FLOW

■

THE for STATEMENT

Structogram for for

expression1

As long as expression2 is true

statement

expression3

Sample program
// Euro1.cpp
#include 
#include 
using namespace std;
int main()
{
double rate = 1.15;

// Exchange rate:
// one Euro to one Dollar
cout << fixed << setprecision(2);
cout << "\tEuro \tDollar\n";
for( int euro = 1; euro <= 5; ++euro)
cout << "\t " << euro
<< "\t " << euro*rate << endl;
return 0;

}

Screen output
Euro
1
2
3
4
5

Dollar
0.95
1.90
2.85
3.80
4.75

THE FOR STATEMENT

■

99

䊐 Initializing and Reinitializing
A typical loop uses a counter that is initialized, tested by the controlling expression and
reinitialized at the end of the loop.

Example: int count = 1;

// Initialization
while( count <= 10)
// Controlling
{
// expression
cout << count
<< ". loop" << endl;
++count;
// Reinitialization
}

In the case of a for statement the elements that control the loop can be found in the
loop header. The above example can also be expressed as a for loop:

Example: int count;
for( count = 1; count <= 10; ++count)
cout << count
<< ". loop" << endl;

Any expression can be used to initialize and reinitialize the loop. Thus, a for loop has
the following form:

Syntax:

for( expression1; expression2; expression3 )
statement

expression1 is executed first and only once to initialize the loop. expression2 is
the controlling expression, which is always evaluated prior to executing the loop body:
■
■

if expression2 is false, the loop is terminated
if expression2 is true, the loop body is executed. Subsequently, the loop is
reinitialized by executing expression3 and expression2 is re-tested.

You can also define the loop counter in expression1. Doing so means that the
counter can be used within the loop, but not after leaving the loop.

Example: for( int i = 0; i < 10; cout << i++ )
;

As this example illustrates, the loop body can be an empty statement. This is always the
case if the loop header contains all necessary statements. However, to improve readability, even the empty statement should occupy a line of its own.

100

■

CHAPTER 6

CONTROL FLOW

■

THE for STATEMENT (CONTINUED)

Sample program
// EuroDoll.cpp
// Outputs a table of exchange:

Euro and US-$

#include 
#include 
using namespace std;
int main()
{
long
euro, maxEuro;
double rate;

// Amount in Euros
// Exchange rate Euro <-> $

cout << "\n* * * TABLE OF EXCHANGE "
<< " Euro – US-$ * * *\n\n";
cout << "\nPlease give the rate of exchange: "
" one Euro in US-$: ";
cin >> rate;
cout << "\nPlease enter the maximum euro: ";
cin >> maxEuro;
//

--- Outputs the table

--// Titles of columns:

cout << '\n'
<< setw(12) << "Euro" << setw(20) << "US-$"
<< "\t\tRate: " << rate << endl;
// Formatting US-$:
cout << fixed << setprecision(2) << endl;
long lower, upper,
step;

// Lower and upper limit
// Step width

// The outer loop determines the actual
// lower limit and the step width:
for( lower=1, step=1; lower <= maxEuro;
step*= 10, lower = 2*step)
// The inner loop outputs a "block":
for( euro = lower, upper = step*10;
euro <= upper && euro <= maxEuro; euro+=step)
cout << setw(12) << euro
<< setw(20) << euro*rate << endl;
return 0;
}

THE FOR STATEMENT (CONTINUED)

■

101

Any of the three expressions in a for statement can be omitted, however, you must type
at least two semicolons. The shortest loop header is therefore:

Example: for(;;)
This statement causes an infinite loop, since the controlling expression is assumed to be
true if expression2 is missing. In the following

Example: for( ; expression; )
the loop header is equivalent to while(expression). The loop body is executed as
long as the test expression is true.

䊐 The Comma Operator
You can use the comma operator to include several expressions where a single expression
is syntactically correct. For example, several variables can be initialized in the loop
header of a for statement. The following syntax applies for the comma operator

Syntax:

expression1, expression2 [, expression3 ...]

The expressions separated by commas are evaluated from left to right.

Example: int x, i, limit;
for( i=0, limit=8; i < limit; i += 2)
x = i * i, cout << setw(10) << x;

The comma operator separates the assignments for the variables i and limit and is
then used to calculate and output the value of x in a single statement.
The comma operator has the lowest precedence of all operators — even lower than
the assignment operators. This means you can leave out the parentheses in the above
example.
Like any other C++ expression, an expression containing the comma operator has a
value and belongs to a certain type. The type and value are defined by the last expression
in a statement separated by commas.

Example: x = (a = 3, b = 5, a * b);
In this example the statements in brackets are executed before the value of the product
of a * b is assigned to x.

102

■

CHAPTER 6

CONTROL FLOW

■

THE do-while STATEMENT

Structogram for do-while

statement

As long as the expression is true

Sample program
// tone.cpp
#include 
using namespace std;
const long delay = 10000000L;
int main()
{
int tic;
cout << "\nHow often should the tone be output? ";
cin >> tic;
do
{
for( long i = 0; i < delay; ++i )
;
cout << "Now the tone!\a" << endl;
}
while( --tic > 0 );
cout << "End of the acoustic interlude!\n";
return 0;
}

THE DO-WHILE STATEMENT

■

103

In contrast to while and for loops, which are controlled by their headers, the dowhile loop is controlled by its footer, i.e. the controlling expression is evaluated after
executing the first loop. This results in the loop body being performed at least once.

Syntax:

do
statement
while( expression);

When a do-while loop is executed, the loop body is processed first. Only then is the
controlling expression evaluated. The loop body is iterated again if the result is
true, otherwise the loop is terminated.

✓

NOTE
The do-while loop must be followed by a semicolon.

䊐 Nesting Loops
Loops can be nested, that is, the loop body can also contain a loop. The ANSI standard
stipulates a maximum depth of 256 nested loops.
The program on the opposite page outputs a number of tones with the number being
defined by user input.
The program contains two loops — one of which is nested in the other. Each time the
outer do-while loop is repeated a short break occurs. The break is caused by the inner
for loop where the variable i is incremented from 0 to the value of delay.
Text and a tone are subsequently output. The tone is generated by outputting the
control character BELL (ASCII code 7), which is represented by the escape sequence
\a.
Since a do-while statement is used, the program outputs a tone even if the user
types 0 or a negative number.

104

■

CHAPTER 6

CONTROL FLOW

■

SELECTIONS WITH if-else

Structogram for the if-else statement

if (expression)
true

statement1

false

statement2

Sample program
// if_else.cpp
// Demonstrates the use of if-else statements
#include 
using namespace std;
int main()
{
float x, y, min;
cout << "Enter two different numbers:\n";
if( cin >> x && cin >> y) // If both inputs are
{
// valid, compute
if( x < y )
// the lesser.
min = x;
else
min = y;
cout << "\nThe smaller number is: " << min << endl;
}
else
cout << "\nInvalid Input!" << endl;
return 0;
}

Sample output for this program
Enter two different numbers:
7.5 5.7
The smaller number is: 5.7

SELECTIONS WITH IF-ELSE

■

105

The if-else statement can be used to choose between two conditional statements.

Syntax:

if( expression )
statement1
[ else
statement2 ]

When the program is run, expression is first evaluated and the program control
branches accordingly. If the result is true, statement1 is executed and statement2
is executed in all other cases, provided an else branch exists. If there is no else and
expression is false, the control jumps to the statement following the if statement.

䊐 Nested if-else Statements
As the program opposite illustrates, multiple if-else statements can be nested. But
not every if statement has an else branch. To solve the resulting problem, an else
branch is always associated with the nearest preceding if statement that does not have
an else branch.

Example: if( n > 0 )
if( n%2 == 1 )
cout << " Positive odd number ";
else
cout << "Positive even number";

In this example, the else branch belongs to the second if, as is indicated by the fact
that the statement has been indented. However, you can use a code block to redefine the
association of an else branch.

Example: if( n > 0 )
{

if( n%2 == 1 )
cout << " Positive odd number \n";

}
else
cout << " Negative number or zero\n";

䊐 Defining Variables in if Statements
You can define and initialize a variable within an if statement. The expression is true if
converting the variable’s value to a bool type yields true. In this case the variable is
available within the if statement.

Example: if( int x = func() )
{ . . . }

// Here to work with x.

The return value of the function, func(), is used to initialize the variable x. If this
value is not 0, the statements in the next block are executed. The variable x no longer
exists after leaving the if statement.

106

■

CHAPTER 6

CONTROL FLOW

■

Else-if CHAINS

Structogram for an else-if chain

if(expression)
true

false

statement1

if(expression)
true

false

statement2
if(expression)
. . .

true

false

statement(n) statement(n+1)

Sample program
// speed.cpp
// Output the fine for driving too fast.
#include 
using namespace std;
int main()
{
float limit, speed, toofast;
cout << "\nSpeed limit: ";
cin >> limit;
cout << "\nSpeed: ";
cin >> speed;
if( (toofast = speed – limit ) < 10)
cout << "You were lucky!" << endl;
else if( toofast < 20)
cout << "Fine payable: 40,-. Dollars" << endl;
else if( toofast < 30)
cout << "Fine payable: 80,-. Dollars" << endl;
else
cout << "Hand over your driver's license!" << endl;
return 0;
}

ELSE-IF CHAINS

■

107

䊐 Layout and Program Flow
You can use an else-if chain to selectively execute one of several options. An elseif chain implies a series of embedded if-else statements whose layout is normally as
follows:
if ( expression1 )
statement1
else if( expression2 )
statement2
.
.
.
else if( expression(n) )
statement(n)
[ else statement(n+1)]

When the else-if chain is executed, expression1, expression2, ... are
evaluated in the order in which they occur. If one of the expressions proves to be true,
the corresponding statement is executed and this terminates the else-if chain.
If none of the expressions are true, the else branch of the last if statement is executed. If this else branch is omitted, the program executes the statement following the
else-if chain.

䊐 The Sample Program
The program opposite uses an else-if chain to evaluate the penalty for driving too fast
and outputs the fine on screen.
The speed limit and the actual speed are read from the keyboard. If the user types 60
as the speed limit and 97.5 as the actual speed, the first three expressions are not true,
and the last else branch is executed. This outputs the message "Hand over your
driver's license!" on a new line.

108

■

CHAPTER 6

CONTROL FLOW

■

CONDITIONAL EXPRESSIONS

Structogram for a conditional expression

expression
true

expression1

false

expression2

Sample program
// greater.cpp
#include 
using namespace std;
int main()
{
float x, y;
cout << "Type two different numbers:\n";
if( !(cin >> x && cin >> y) ) // If the input was
{
// invalid.
cout << "\nInvalid input!" << endl;
}
else
{
cout << "\nThe greater value is: "
<< (x > y ? x : y) << endl;
}
return 0;
}

Sample output for this program
Type two different numbers:
173.2
216.7
The greater value is: 216.7

CONDITIONAL EXPRESSIONS

■

109

䊐 Conditional Operator
The conditional operator ?: is used to form an expression that produces either of two
values, depending on the value of some condition. Because the value produced by such
an expression depends on the value of a condition, it is called conditional expression.
In contrast to the if-else statement the selection mechanism is based on expressions: one of two possible expressions is selected. Thus, a conditional expression is often
a concise alternative to an if-else statement.

Syntax:

expression ? expression1 : expression2

expression is evaluated first. If the result is true, expression1 is evaluated; if not
expression2 is executed. The value of the conditional expression is therefore either
the value of expression1 or expression2.

Example: z = (a >= 0) ? a : -a;
This statement assigns the absolute value of a to the variable z. If a has a positive value
of 12, the number 12 is assigned to z. But if a has a negative value, for example –8, the
number 8 is assigned to z.
Since this sample program stores the value of the conditional expression in the variable z, the statement is equivalent to
if( a > 0 )
z = a;
else
z = -a;

䊐 Precedence
The conditional operator is the only C++ operator with three operands. Its precedence is
higher than that of the comma and assignment operators but lower than all other operators. In other words, you could omit the brackets in the first example.
You can use the result of a conditional evaluation without assigning it, as the sample
program on the opposite page shows. In this example, x is printed on screen if x is
greater than y, and y is printed otherwise.
However, you should assign the result of complex expressions to a variable explicitly
to improve the readability of your program.

110

■

CHAPTER 6

CONTROL FLOW

■

SELECTING WITH switch

Structogram for the switch statement

switch(expression)
case Const1:
case Const2:
. . .
statements
break

statements
break

default:
statements
break

Example
// Evaluates given input.
int command = menu();

// The function menu() reads
// a command.
// Evaluate command.

switch( command )
{
case 'a':
case 'A':
action1();
// Carry out 1st action.
break;
case 'b':
case 'B':
action2();
// Carry out 2nd action.
break;
default:
cout << '\a' << flush; // Beep on
}
// invalid input

SELECTING WITH SWITCH

■

111

䊐 The switch Statement
Just like the else-if chain, the switch statement allows you to choose between multiple alternatives. The switch statement compares the value of one expression with
multiple constants.
switch( expression )
{
case const1: [ statement ]
[ break; ]
case const2: [ statement ]
[ break; ]
.
.
.
[default: statement ]
}

First, the expression in the switch statement is evaluated. It must be an integral
type. The result is then compared to the constants, const1, const2, ..., in the case
labels. The constants must be different and can only be integral types (boolean values
and character constants are also integral types).
If the value of an expression matches one of the case constants, the program
branches to the appropriate case label. The program then continues and the case labels
lose their significance.
You can use break to leave the switch statement unconditionally. The statement is
necessary to avoid executing the statements contained in any case labels that follow.
If the value of the expression does not match any of the case constants, the program
branches to the default label, if available. If you do not define a default label, nothing happens. The default does not need to be the last label; it can be followed by additional case labels.

䊐 Differences between switch and else-if Chains
The else-if chain is more versatile than the switch statement. Every selection can
be programmed using an else-if chain. But you will frequently need to compare the
value of an integral expression with a series of possible values. In this case (and only this
case), you can use a switch statement.
As the example opposite shows, a switch statement is more easily read than an
equivalent else-if chain, so use the switch statement whenever possible.

112

■

CHAPTER 6

CONTROL FLOW

■

JUMPS WITH break, continue, AND goto

Structogram for break within a while statement
As long as expression is true

break;

statement, which follows the loop.

Sample program containing a break statement
// ascii.cpp : To output an ASCII Code Table
#include 
#include 
using namespace std;
int main()
{
int ac = 32;
// To begin with ASCII Code 32
// without control characters.
while(true)
{ cout << "\nCharacter
Decimal
Hexadecimal\n\n";
int upper;
for( upper =ac + 20; ac < upper && ac < 256; ++ac)
cout << "
" << (char)ac
// as character
<< setw(10) << dec << ac
<< setw(10) << hex << ac << endl;
if( upper >= 256)
break;
cout <<"\nGo on -> ,Stop -> +";
char answer;
cin.get(answer);
if( answer == 'q' || answer == 'Q' )
break;
cin.sync();
// Clear input buffer
}
return 0;
}

✓

NOTE
The expression (char)ac yields the value ac of type char.

JUMPS WITH BREAK, CONTINUE, AND GOTO

■

113

䊐 break
The break statement exits from a switch or loop immediately. You can use the break
keyword to jump to the first statement that follows the switch or loop.
The program on the opposite page, which outputs a group of 20 ASCII characters and
their corresponding codes, uses the break keyword in two places. The first break exits
from an infinite while(true) { ... } loop when a maximum value of 256 has been
reached. But the user can also opt to continue or terminate the program. The second
break statement is used to terminate the while loop and hence the program.

䊐 continue
The continue statement can be used in loops and has the opposite effect to break,
that is, the next loop is begun immediately. In the case of a while or do-while loop
the program jumps to the test expression, whereas a for loop is reinitialized.

Example: for( int i = 0; i < 100; i++ )
{
. . . // Processes all integers.
if( i % 2 == 1)
continue;
. . .
// Process even
// numbers only.
}

䊐 goto and Labels
C++ also offers a goto statement and labels. This allows you to jump to any given point
marked by a label within a function. For example, you can exit from a deeply embedded
loop construction immediately.

Example: for( . . . )
for( . . . )
if (error) goto errorcheck;
. . .
errorcheck: . . .
// Error handling

A label is a name followed by a colon. Labels can precede any statement.
Any program can do without goto statements. If you need to use a goto statement,
do so to exit from a code block, but avoid entering code blocks by this method.

■

CHAPTER 6

■

exercise s

114

CONTROL FLOW

EXERCISES

Screen output for exercise 2
****** MULTIPLICATION TABLE ******
1

2

3

4

5

6

7

8

9

10

1

1

2

3

.

.

.

.

.

.

10

2

2

4

6

20

3

.

.

.

.

4

.

.

.

.

5

.

.

.

.

10

20

30

6
7
8
9
10

.

.

.

.

.

.

100

Note on exercise 4
Use the function time() to initialize the random number generator:
#include 
#include 

// Prototype of time()
// Prototypes of srand()
// and rand()

long sec;
time( &sec );
// Take the number of seconds and
srand( (unsigned)sec ); // use it to initialize.

EXERCISES

■

115

Exercise 1
Rewrite the EuroDoll.cpp program in this chapter to replace both the for
loops with while loops.
Exercise 2
Write a C++ program that outputs a complete multiplication table (as shown
opposite) on screen.
Exercise 3
Write a C++ program that reads an integer between 0 and 65535 from the
keyboard and uses it to seed a random number generator.Then output 20
random numbers between 1 and 100 on screen.
Exercise 4
Write a program for the following numerical game:
The computer stores a random number between 1 and 15 and the player
(user) attempts to guess it.The player has a total of three attempts.After each
wrong guess, the computer tells the user if the number was too high or too low.
If the third attempt is also wrong, the number is output on screen.
The player wins if he or she can guess the number within three attempts.
The player is allowed to repeat the game as often as he or she wants.

✓

NOTE
Use the system time to seed the random number generator as shown opposite. The time() function
returns the number of seconds since 1/1/1970, 0:0. The long value of the sec variable is converted to
unsigned by unsigned(sec) and then passed to the srand() function.

■

CHAPTER 6

■

solutions

116

CONTROL FLOW

SOLUTIONS

Exercise 1
The for loops of program EuroDoll.cpp are equivalent to the following while
loops:
// The outer loop sets the lower
// limit and the step width used:
lower=1, step=1;
while( lower <= maxEuro)
{
// The inner loop outputs a block:
euro = lower;
upper = step*10;
while( euro <= upper && euro <= maxEuro)
{
cout << setw(12) << euro
<< setw(20) << euro*rate << endl;
euro += step;
}
step *= 10, lower = 2*step;
}

Exercise 2
//
//

MultTable.cpp
Outputs a multiplication table.

#include 
#include 
using namespace std;
int main()
{
int factor1, factor2;
cout << "\n\n
<< " ******
<< endl;

"
MULTIPLICATION TABLE

******"

// Outputs the first and second line:
cout << "\n\n\n
";
// 1. line
for( factor2 = 1 ; factor2 <= 10 ; ++factor2 )
cout << setw(5) << factor2;
cout << "\n
"
// 2. line
<< "-------------------------------------------"
<< endl;

SOLUTIONS

//

Outputs the remaining lines of the table:

for( factor1 = 1 ; factor1 <= 10 ; ++factor1 )
{
cout << setw(6) << factor1 << " |";
for( factor2 = 1 ; factor2 <= 10 ; ++factor2 )
cout << setw(5) << factor1 * factor2;
cout << endl;
}
cout << "\n\n\n";
// To shift up the table
return 0;
}

Exercise 3
// random.cpp
// Outputs 20 random numbers from 1 to 100.
#include 
#include 
#include 
using namespace std;
int main()
{
unsigned int

// Prototypes of srand() and rand()

i, seed;

cout << "\nPlease type an integer between "
"0 and 65535: ";
cin >> seed;
srand( seed);

cout << "\n\n
"******

// Reads an integer.
// Seeds the random
// number generator.
"
RANDOM NUMBERS

******\n\n";

for( i = 1 ; i <= 20 ; ++i)
cout << setw(20) << i << ". random number = "
<< setw(3) << (rand() % 100 + 1) << endl;
return 0;
}

■

117

118

■

CHAPTER 6

CONTROL FLOW

Exercise 4
// NumGame.cpp : A numerical game against the computer
#include 
// Prototypes of srand() and rand()
#include 
// Prototype of time()
#include 
using namespace std;
int main()
{
int number, attempt;
char wb = 'r';
// Repeat or finish.
long sec;
time( &sec);
// Get the time in seconds.
srand((unsigned)sec);
// Seeds the random
// number generator
cout << "\n\n
"
<< " *******
A NUMERICAL GAME
*******" << endl;
cout << "\n\nRules of the game:" << endl;
while( wb == 'r')
{
cout << "I have a number between 1 and 15 in mind \n"
<< "You have three chances to guess correctly!\n"
<< endl;
number = (rand() % 15) + 1;
bool found = false;
int count = 0;
while( !found && count < 3 )
{
cin.sync();
// Clear input buffer
cin.clear();
cout << ++count << ". attempt:
";
cin >> attempt;
if(attempt < number)
cout << "too small!"<< endl;
else if(attempt > number) cout <<"too big!"<< endl;
else
found = true;
}
if( !found)
cout << "\nI won!"
<< " The number in question was: "
<< number << endl;
else
cout << "\nCongratulations! You won!" << endl;
cout << "Repeat —> 
Finish —> \n";
do
cin.get(wb);
while( wb != 'r' && wb != 'f');
}
return 0;
}

chapter

7

Symbolic Constants and
Macros
This chapter introduces you to the definition of symbolic constants and
macros illustrating their significance and use. In addition, standard macros
for character handling are introduced.

119

120

■

CHAPTER 7

SYMBOLIC CONSTANTS AND MACROS

■

MACROS

Sample program
// sintab.cpp
// Creates a sine function table
#include 
#include 
#include 
using namespace std;
#define PI
3.1415926536
#define START
0.0
// Lower limit
#define END
(2.0 * PI)
// Upper limit
#define STEP (PI / 8.0)
// Step width
#define HEADER
(cout << \
" ***** Sine Function Table *****\n\n")
int main()
{
HEADER;

// Title
// Table Head:
cout << setw(16) << "x" << setw(20) << "sin(x)\n"
<< "
-----------------------------------------"
<< fixed << endl;
double x;
for( x = START; x < END + STEP/2; x += STEP)
cout << setw(20) << x << setw(16) << sin(x)
<< endl;
cout << endl << endl;
return 0;

}

Screen output
******

Table for the Sine Function

******

x
sin(x)
-------------------------------------------0.000000
0.000000
0.392699
0.382683
0.785398
0.707107
.
.
.
.
.
.

MACROS

■

121

C++ has a simple mechanism for naming constants or sequences of commands, that is for
defining macros. You simply use the preprocessor’s #define directive.

Syntax:

#define

name

substitutetext

This defines a macro called name. The preprocessor replaces name with substitutetext throughout the subsequent program. For example, in the program on the opposite
page, the name PI is replaced by the number 3.1415926536 throughout the program
in the first phase of compilation.
There is one exception to this general rule: substitution does not take place within
strings. For example, the statement
cout << "PI";

outputs only PI and not the numerical value of PI.

䊐 Symbolic Constants
Macros that are replaced by constants, such as the PI macro, are also known as symbolic
constants. You should note that neither an equals sign nor a semicolon is used, as these
would become part of the substitute text.
You can use any macros you have previously defined in subsequent #define directives. The program opposite uses the symbolic constant PI to define other constants.

䊐 More about Working with Macros
Any preprocessor directive, and this includes the #define directive, must be placed in a
line of its own. If the substitute text is longer than one line, you can terminate the line
with a backslash \ and continue the substitute text in the next line, as is illustrated by
the macro HEADER on the opposite page.
The rules that apply to naming variables also apply to naming macros. However, it is
standard practice to capitalize symbolic constants to distinguish them from the names of
variables in a program.
Using macros makes a C++ program more transparent and flexible. There are two
main advantages:
1. good readability: You can name a macro to indicate the use of the macro
2. easy to modify: If you need to change the value of a constant throughout a program, you simply change the value of the symbolic constant in the #define
directive.

122

■

CHAPTER 7

SYMBOLIC CONSTANTS AND MACROS

■

MACROS WITH PARAMETERS

Sample program
// ball1.cpp
// Simulates a bouncing ball
// --------------------------------------------------#include 
#include 
using namespace std;
#define DELAY 10000000L
//
#define CLS
(cout << "\033[2J") //
#define LOCATE(z,s) (cout <<"\033["<<
// Position the cursor in row

Output delay
Clear screen
z <<';'<< s <<'H')
z and column s

void main()
{
int x = 2, y = 3, dx = 1, speed = 0;
string floor(79, '-'),
header = "**** JUMPING BALL ****";
CLS;
LOCATE(1,25);
LOCATE(25,1);

cout << header;
cout << floor;

while(true)
// Let the ball "always" bounce
{
// Terminate by interrupt key (^C)
LOCATE(y,x); cout << 'o' << endl; // Show the ball
for( long wait = 0; wait < DELAY; ++wait)
;
if(x == 1 || x == 79) dx = -dx;
// Bounce off
// a wall?
if( y == 24 )
// On the floor?
{
speed = - speed;
if( speed == 0 ) speed = -7;
// Restart
}
speed += 1;
// Acceleration = 1
LOCATE(y,x); cout << ' ';
// Clear output
y += speed; x += dx;
// New Position
}
}

MACROS WITH PARAMETERS

■

123

It is possible to call macros with arguments. To do so, you must supply the appropriate
parameters when defining the macro. The parameters are replaced by valid arguments at
run time.

Example: #define SQUARE(a)

((a) * (a))

This defines a macro called SQUARE() with a parameter a. The name of the macro must
be followed immediately by a left bracket. When the macro is called, for example

Example: z = SQUARE(x+1);
the preprocessor inserts the substitute text with the current arguments, which will be
expanded as follows, in this case
z = ((x+1) * (x+1));

This example also shows that you must be careful when using brackets to indicate parameters for macros. Omitting the brackets in the previous example, SQUARE, would cause
the expression to be expanded as follows z = x + 1 * x + 1.
The outer brackets in the definition ensure that even when the macro is used in a
complex expression, the square is calculated before the result can be used for any further
calculations.

䊐 Macros for Screen Control
The program opposite uses macros to change the appearance of the screen. Peripheral
devices, such as the screen or printers, can be controlled by special character sequences
that normally begin with the ESC character (decimal 27, octal 033) and are thus known
as escape sequences. A number of ANSI standard escape sequences exists for screen control.1 See the appendix on Escape Sequences for Screen Control for an overview of the
most important sequences.
CLS is a macro without any parameters that uses the escape sequence \033[2J to
clear the screen. LOCATE is just one example of a macro with two parameters. LOCATE
uses the escape sequence \033[z;sH to place the cursor at the position of the next
screen output. The values z for the line and s for the column require decimal input with
z = 1, s = 1 representing the top left corner of the screen or window.
The ball is “thrown in” at the coordinates x = 2, y = 3 and bounces off the “floor”
and the “walls.” In direction x (horizontally) the ball has a constant speed of dx = 1 or
-1. In direction y (vertically) the ball is subject to a constant acceleration of 1,
expressed as speed += 1.

1These

escape sequences are valid for all standard UNIX terminals. The driver ansi.sys must be
loaded for DOS or a DOS box in Win95 or Win98. For Win NT and Win 2000, corresponding
functions based on system calls are offered for download.

124

■

CHAPTER 7

SYMBOLIC CONSTANTS AND MACROS

■

WORKING WITH THE #define DIRECTIVE

Using macros in different source files

Header file proj.h
Macros
Classes
and other type
definitions
Prototypes of
global functions

Source
file 1

Source
file 2

Source
file n

#include "proj.h"

#include "proj.h"

.

.

.

.

.

.

.

.

.

...

#include "proj.h"

WORKING WITH THE #DEFINE DIRECTIVE

■

125

You can place the #define directive in any line of a program as long as it is placed prior
to using the macro. However, it is recommended to place all definitions at the beginning
of the source file for ease of location and modification.
If you need to use the same macros in different source files, it makes sense to create a
header file. You can then include the header file in your source files. This method also
lends itself to large-scale software projects. The programmers working on the project
then have access to the same set of macro definitions and other declarations. This concept is illustrated opposite using the header file proj.h as an example.
Macros with parameters can be called just like functions. You should note the following important differences, however:
■

■

Macros: A macro definition must be visible to the compiler. The substitute text is
inserted and re-compiled each time the macro is called. For this reason, a macro
should contain only a few statements to avoid inflating the object file each time
the macro is called. The speed of program execution will, however, improve since
the program does not need to branch to sub-routines in contrast to normal function calls. This can become apparent when macros are used within loops, for
example.
Side effects of macros are possible if the substitute text contains multiple
instances of a parameter. The statement SQUARE( ++x ) expands to ((++x)
* (++x)), for example. The variable x is incremented twice and the product
does not represent the square of the incremented number.
Functions: Functions are compiled independently. The linker then links them
into the executable file. The program branches to the function whenever it is
called, and this can reduce the speed of execution in comparison to a macro.
However, the executable file will be shorter as it contains only one instance of
the function code.
The compiler checks the argument types, and the side effects seen with
macros cannot occur.

Inline functions, which are introduced in the chapter on functions, are an alternative to macros.

126

■

CHAPTER 7

■

SYMBOLIC CONSTANTS AND MACROS

CONDITIONAL INCLUSION

Multiple inclusions of header files
Header file

basis.h
#ifndef_BASIS_H_
#define_BASIS_H_
//content of basis,
//ex.
#define BSIZE 1000
. . .
#endif

Header file

Header file

statist.h

graph.h

#include 
#include "basis.h"

#include 
#include "basis.h"

. . .

. . .

Source file

application.cpp
#include "statist.h"
#include "graph.h"
int main()
{
. . .
return 0;
}

CONDITIONAL INCLUSION

■

127

䊐 Redefining Macros
A macro cannot simply be redefined. A macro definition is valid until it is removed by
using an #undef directive. However, you do not need to supply the parameter list of a
macro with parameters.

Example: #define MIN(a,b)
. . .
#undef MIN

((a)<(b)? (a) : (b))
// Here MIN can be called

The macro MIN cannot be used after this point, but it can be defined again, possibly with
a different meaning, using the #define directive.

䊐 Conditional Inclusion
You can use the preprocessor directives #ifdef and #ifndef to allow the compiler to
check whether a macro has been defined.

Syntax:

#ifdef name
. . .

// Block, which will be compiled
// if name is defined.

#endif

In the case of the #ifndef directive, the code block is compiled up to the next #endif
only if the macro name has not been previously defined.
On conditional inclusion else branching and nesting is also permissible. See Preprocessor Directives in the appendix for further information.
A macro definition does not need to include a substitute text to be valid.

Example: #define MYHEADER
A symbol without a substitute text is often used to identify header files and avoid multiple inclusion.
If you have a header file named "article.h", you can identify the header by defining a symbol, such as _ARTICLE_, within that file.

Example: #ifndef _ARTICLE_
#define _ARTICLE_
. . .
// Content of the header file
#endif

If you have already included the header, _ARTICLE_ will already be defined, and the
contents of the header file need not be compiled. This technique is also employed by
standard header files.

128

■

CHAPTER 7

SYMBOLIC CONSTANTS AND MACROS

■

STANDARD MACROS FOR CHARACTER MANIPULATION

Sample program
// toupper.cpp: A filter that converts to capitals.
// --------------------------------------------------#include 
#include 
using namespace std;
int main()
{
char c;
long nChar = 0,
// Counts all characters
nConv = 0;
// and converted characters
while ( cin.get(c) )
// As long as a character
{ ++nChar;
// can be read, to increment.
if( islower(c))
// Lowercase letter?
{ c = toupper(c);
// Converts the character
++nConv;
// and counts it.
}
cout.put(c);
// Outputs the character.
}
clog << "\nTotal of characters:
" << nChar
<< "\nTotal of converted characters: " << nConv
<< endl;
return 0;
}

✓

NOTE
The program reads characters from a file until end-of-file. When reading keyboard input, end-of-file is
simulated by Ctrl+Z (DOS) or Ctrl+D (UNIX).

Macros for character classification
Macro

Return value true means:

isalpha(c)

c is a letter

islower(c)

c is a small letter

isupper(c)

c is a capital letter

isdigit(c)

c is a decimal digit

isalnum(c)

c is a letter or a digit

isspace(c)

c is a space letter

isprint(c)

c is a printable letter

STANDARD MACROS FOR CHARACTER MANIPULATION

■

129

The following section introduces macros that classify or convert single characters. The
macros are defined in the header files ctype.h and cctype.

䊐 Case Conversion
You can use the macro toupper to convert lowercase letters to uppercase. If c1 and c2
are variables of type char or int where c1 contains the code for a lowercase letter, you
can use the following statement

Example: c2 = toupper(c1);
to assign the corresponding uppercase letter to the variable c2. However if c1 is not a
lowercase letter, toupper(c1) returns the character “as is.”
The sample program on the opposite page reads standard input, converts any letters
from lower- to uppercase, and displays the letters. As toupper only converts the letters
of the English alphabet by default, any national characters, such as accentuated characters in other languages, must be dealt with individually. A program of this type is known
as a filter and can be applied to files. Refer to the next section for details.
The macro tolower is available for converting uppercase letters to lowercase.

䊐 Testing Characters
A number of macros, all of which begin with is..., are available for classifying characters. For example, the macro islower(c) checks whether c contains a lowercase letter
returning the value true, in this case, and false in all other cases.

Example: char c;

cin >> c;

// Reads and
// classifies
if( !isdigit(c) )
// a character.
cout << "The character is no digit \n";

The following usage of islower() shows a possible definition of the toupper()
macro:

Example: #define toupper(c) \
(islower(c) ? ((c)-'a'+'A') : (c))

This example makes use of the fact that the codes of lower- and uppercase letters differ
by a constant, as is the case for all commonly used character sets such as ASCII and
EBCDIC.
The opposite page contains an overview of macros commonly used to classify characters.

130

■

CHAPTER 7

SYMBOLIC CONSTANTS AND MACROS

■

REDIRECTING STANDARD INPUT AND OUTPUT

Sample program
// lines.cpp
// A filter that numbers lines.
#include 
#include 
#include 
using namespace std;
int main()
{
string line;
int number = 0;
while( getline( cin, line))
// As long as a line
{
// can be read.
cout << setw(5) << ++number << ": "
<< line << endl;
}
return 0;
}

How to call the program
1. Redirecting the standard input:
lines < text.dat | more

This outputs the text file text.dat with line numbers. In addition, the data
stream is sent to the standard filter more, which pauses screen output if the page
is full.
2. Redirecting the standard output:
lines > new.dat

Here the program reads from the keyboard and adds the output to the new file
new.dat. Please note, if the file already exists, it will be overwritten!
You can use
lines >> text.dat

to append program output to the file text.dat. If the file text.dat does not
already exist, it will be created.
Type Ctrl+Z (DOS) or Ctrl+D (UNIX) to terminate keyboard input.

REDIRECTING STANDARD INPUT AND OUTPUT

■

131

䊐 Filter Programs
The previous program, toupper.cpp, reads characters from standard input, processes
them, and sends them to standard output. Programs of this type are known as filters.
In the program toupper.cpp, the loop
while( cin.get(c)) { ... }

is repeated while the test expression cin.get(c) yields the value true, that is, as long
as a valid character can be read for the variable c. The loop is terminated by end-of-file
or if an error occurs since the test expression cin.get(c) will then be false.
The program on the opposite page, lines.cpp, is also a filter that reads a text and
outputs the same text with line numbers. But in this case standard input is read line by
line.
while( getline(cin,line)) { ... }

The test expression getline(cin,line) is true while a line can be read.

䊐 Using Filter Programs
Filter programs are extremely useful since various operating systems, such as DOS,
Win**, WinNT, and UNIX are capable of redirecting standard input and output. This
allows easy data manipulation.
For example, if you need to output text.dat with line numbers on screen, you can
execute the program lines by typing the following command:

Example: lines < text.dat
This syntax causes the program to read data from a file instead of from the keyboard. In
other words, the standard input is redirected.
The opposite page contains additional examples. You can redirect input and output
simultaneously:

Example: lines < text.dat > new.dat

✓

In this example the contents of text.dat and the line numbers are stored in new.dat.
The program does not generate any screen output.

NOTE
These examples assume that the compiled program lines.exe is either in the current directory or in
a directory defined in your system’s PATH variable.

■

CHAPTER 7

■

exercise s

132

SYMBOLIC CONSTANTS AND MACROS

EXERCISES

Hints for Exercise 2
You can use the function kbhit() to test whether the user has pressed a key. If
so, the function getch() can be used to read the character.This avoids
interrupting the program when reading from the keyboard.
These functions have not been standardized by ANSI but are available on
almost every system. Both functions use operating system routines and are
declared in the header file conio.h.
The function kbhit()

Prototype:
Returns:

int kbhit();
0, if no key was pressed, otherwise != 0.

When a key has been pressed, the corresponding character can be read by
getch().
The function getch()

Prototype:
Returns:

int getch();

The character code.There is no special return value on reaching
end-of-file or if an error occurs.

In contrast to cin.get(), getch() does not use an input buffer when
reading characters, that is, when a character is entered, it is passed directly to
the program and not printed on screen. Additionally, control characters, such as
return ( = 13), Ctrl+Z ( = 26), and Esc ( = 27), are passed to the program “as is.”
Example: int c;
if( kbhit() != 0)
{
c = getch();
if( c == 27 )
// . . .
}

✓

// Key was pressed?
// Yes -> Get character
// character == Esc?

NOTE
When a function key, such as F1, F2, ..., Ins, Del, etc. was pressed, the function
getch() initially returns 0. A second call yields the key number.

EXERCISES

■

133

Exercise 1
Please write
a. the macro ABS, which returns the absolute value of a number,
b. the macro MAX, which determines the greater of two numbers.
In both cases use the conditional operator ?: .
Add these macros and other macros from this chapter to the header file
myMacros.h and then test the macros.
If your system supports screen control macros, also add some screen control
macros to the header. For example, you could write a macro named
COLOR(f,b) to define the foreground and background colors for the following
output.

Exercise 2
Modify the program ball1.cpp to
a. display a white ball on a blue background,
b. terminate the program when the Esc key is pressed,
c. increase the speed of the ball with the + key and decrease the speed
with the – key.
You will need the functions kbhit() and getch() (shown opposite) to solve
parts b and c of this problem.

Exercise 3
Write a filter program to display the text contained in any given file.The
program should filter any control characters out of the input with the exception
of the characters \n (end-of-line) and \t (tabulator), which are to be treated as
normal characters for the purpose of this exercise. Control characters are
defined by codes 0 to 31.
A sequence of control characters is to be represented by a single space
character.
A single character, that is, a character appearing between two control
characters, is not to be output!

✓

NOTE
Since the program must not immediately output a single character following a control character, you will
need to store the predecessor of this character. You may want to use two counters to count the
number of characters and control characters in the current string.

■

CHAPTER 7

■

solutions

134

SYMBOLIC CONSTANTS AND MACROS

SOLUTIONS

Exercise 1
// -----------------------------------------------------// myMacros.h
// Header file contains the Macros
// ABS, MIN, MAX, CLS, LOCATE, COLOR, NORMAL, INVERS
// and symbolic constants for colors.
// -----------------------------------------------------#ifndef _MYMACROS_
#define _MYMACROS_
#include 
using namespace std;
// -----------------------------------------------------// Macro ABS
// Call: ABS( val)
// Returns the absolute value of val
#define ABS(a) ( (a) >= 0 ? (a) : -(a))
// -----------------------------------------------------// Macro MIN
// Call: MIN(x,y)
// Returns the minimum of x and y
#define MIN(a,b) ( (a) <= (b) ? (a) : (b))
// -----------------------------------------------------// Macro MAX
// Call: MAX(x,y)
// Returns the maximum of x and y
#define MAX(a,b) ( (a) >= (b) ? (a) : (b))
// -----------------------------------------------------// Macros for controlling the screen
// -----------------------------------------------------// Macro CLS
// Call: CLS;
// Clears the screen
#define CLS
(cout << "\033[2J")
// -----------------------------------------------------// Macro LOCATE
// Call: LOCATE(row, column);
// Positions the cursor to (row,column).
// (1,1) is the upper left corner.
#define LOCATE(r,c) (cout <<"\033["<< (r) <<';'<<(c)<<'H')

SOLUTIONS

// -----------------------------------------------------// Macro COLOR
// Call: COLOR(foreground, background);
// Sets the foreground and background color
// for the following output.
#define COLOR( f, b) (cout << "\033[1;3"<< (f) \
<<";4"<< (b) <<'m' << flush)
// 1: light foreground
// 3x: foreground x
// 4x: background x
// Color values for the macro COLOR
// To call ex.: COLOR( WHITE,BLUE);
#define BLACK 0
#define RED
1
#define GREEN
2
#define YELLOW
3
#define BLUE
4
#define MAGENTA 5
#define CYAN
6
#define WHITE
7
// -----------------------------------------------------// Macro INVERS
// Call: INVERS;
// The following output is inverted.
#define INVERS (cout << "\033[7m")
// -----------------------------------------------------// Macro NORMAL
// Call: NORMAL;
// Sets the screen attributes on default values.
#define NORMAL (cout << "\033[0m")
#endif

//

_MYMACROS_

Exercise 2
// --------------------------------------------------// ball2.cpp
// Simulates a bouncing ball
// --------------------------------------------------#include 
#include 
using namespace std;
#include 
#include "myMacros.h"

// For kbhit() and getch()

■

135

136

■

CHAPTER 7

SYMBOLIC CONSTANTS AND MACROS

#define ESC
27
// ESC terminates the program
unsigned long delay = 5000000;
// Delay for output
int main()
{
int x = 2, y = 2, dx = 1, speed = 0;
bool end = false;
string floor(80, '-'),
header
= "**** BOUNCING BALL ****",
commands = "[Esc] = Terminate
"
"[+] = Speed up
[-] = Slow down";
COLOR(WHITE,BLUE);
CLS;
LOCATE(1,25); cout << header;
LOCATE(24,1); cout << floor;
LOCATE(25,10); cout << commands;
while( !end)
// As long as the flag is not set
{
LOCATE(y,x); cout << 'o';
// Show the ball
for( long wait = 0; wait < delay; ++wait)
;
if(x == 1 || x == 79) dx = -dx; // Bounce off a wall?
if( y == 23 )
// On the floor?
{
speed = - speed;
if( speed == 0 ) speed = -7;
// Kick
}
speed += 1;
// Speed up = 1
LOCATE(y,x); cout << ' ';
y += speed; x += dx;

// Clear screen
// New position

if( kbhit() != 0 )
{
switch(getch())
{
case '+': delay -= delay/5;
break;
case '-': delay += delay/5;
break;
case ESC: end = true;
}
}

// Key pressed?

}
NORMAL; CLS;
return 0;
}

// Yes
// Speed up
// Slow down
// Terminate

SOLUTIONS

■

Exercise 3
// --------------------------------------------------// NoCtrl.cpp
// Filter to ignore control characters
// To call e.g.: NoCtrl < file
// --------------------------------------------------#include 
using namespace std;
#define isCtrl(c)

( c >= 0

int main()
{
char c, prec = 0;
long nCtrl = 0, nChar = 0;

&& c <= 31 \
&& c != '\n' && c != '\t')

//
//
//
//

Character and predecessor
Number of the following
control characters or
other characters

while( cin.get(c))
{
if( isCtrl(c))
// Control characters
{
++nCtrl;
nChar = 0;
}
else
// Normal character
{
if( nCtrl > 0)
{
cout.put(' ');
nCtrl = 0;
}
switch( ++nChar)
{
case 1:
break;
case 2:
cout.put(prec);
// Predecessor and
default: cout.put(c);
// current character
}
prec = c;
}
}
return 0;
}

137

This page intentionally left blank

chapter

8

Converting Arithmetic
Types
This chapter introduces implicit type conversions, which are performed in
C++ whenever different arithmetic types occur in expressions.
Additionally, an operator for explicit type conversion is introduced.

139

140

■

CHAPTER 8

■

CONVERTING ARITHMETIC TYPES

IMPLICIT TYPE CONVERSIONS

Integer promotions

bool
int

char, signed char, unsigned char
short

int

if int equals long

unsigned int

if int equals short

unsigned short

Type hierarchy
long double

double

float

unsigned long

long

unsigned int

not-existent, if int
equals long

int

Example
short size(512); double res, x = 1.5;
res = size / 10 * x;
// short -> int -> double
int

IMPLICIT TYPE CONVERSIONS

■

141

C++ allows you to mix arithmetic types in a single expression — in other words, the
operands of an operator can belong to different types. The compiler automatically performs implicit type conversion, where a common type, which allows the operation in question to be performed, is assigned for the values of both operands. You can generally
assume that the “smaller” type will be converted to the “larger” type. The assignment
operator is an exception to this rule and will be discussed separately.
The result of an arithmetic operation belongs to the common type used to perform
the calculation. However, comparison expressions will be bool types no matter what
type of operands are involved.

䊐 Integer Promotion
Integer promotion is first performed for any expression:
■
■

bool, char, signed char, unsigned char, and short are converted to
int
unsigned short is also converted to int if the int type is greater than
short, and to unsigned int in all other cases.

This type conversion is performed so as to preserve the original values. The boolean
value false is converted to 0 and true is converted to 1.
Thus, C++ will always use int type values or greater when performing calculations.
Given a char variable c, the values of c and 'a' in the expression

Example: c < 'a'
will be converted to int before being compared.

䊐 Usual Arithmetic Type Conversions
If operands of different arithmetic types still occur after integer promotion, further
implicit type conversions along the lines of the hierarchy on the opposite page will be
necessary. In this case, the type of the operand with the highest rank in the hierarchy is
applied. These type conversions and integer promotions are collectively known as usual
arithmetic type conversions.
In our example, size/10 * x, the value of size is first promoted to int before an
integer division size/10 is performed. The interim result 50 is then converted to double and multiplied by x.
Usual arithmetic type conversions are performed for all binary operators and the conditional operator ?: provided the operands belong to an arithmetic type, the only exceptions being the assignment operator and the logical operators && and ||.

142

■

CHAPTER 8

■

CONVERTING ARITHMETIC TYPES

PERFORMING USUAL ARITHMETIC TYPE CONVERSIONS

Converting signed integers
a) Converting a positive number
Sign bit(= 0 ↔ not negative)
Binary representaion of the integer 10
as value of type signed char (8 bits):

0

26

25

24

23

22

21

20

0

0

0

1

0

1

0

Extension to int (here 16 bit)
The value 10 is preserved.

0

0

0

214 213

0

0

0

0

0

0

0

0

0

1

0

1

0

•

•

••

•

28

27

•

•

•

•

•

21

20

Sign bit(= 0 ↔ not negative)

b) Converting a negative number
The bit pattern of –10 is computed by starting with the bit pattern of 10 and generating the binary complement (see Binary Representation of Numbers in the appendix).
Sign bit(= 1 ↔ negative)
Binary representaion of the integer –10
as value of type signed char (8 bits):

1

26

25

24

23

22

21

20

1

1

1

0

1

1

0

Extension to int (here 16 bit)
The value –10 is preserved.

1

1

1

214

213

1
•

1
•

1
••

1

1

1

•

28

27

1
•

1
•

1
•

0
•

1

1

0

•

21

20

Sign bit(= 1 ↔ negative)

✓

NOTE
The value of a negative number changes if the bit pattern is interpreted as unsigned. The bit pattern
1111 0110 of –10, for example, corresponds to the unsigned char value
246 == 0*20+ 1*21 + 1*22 + 0*23 + 1*24 + 1*25 + 1*26 + 1*27

PERFORMING USUAL ARITHMETIC TYPE CONVERSIONS

■

143

Usual arithmetic type conversions retain the value of a number provided it can be represented by the new type. The procedure for type conversion depends on the types
involved:
1. Conversion of an unsigned type to a larger integral type

Examples: unsigned char to int or unsigned int
Zero extension is performed first. During this process, the bit pattern of the number to be converted is expanded to match the length of the new type by adding
zeros from the left.
2. Conversion of a signed type to a larger integral type
■

The new type is also signed

Examples: char to int, short to long

■

Signed integers are represented by generating the binary complement. The
value is retained by performing sign extension. As shown in the example on the
opposite page, the original bit pattern is expanded to match the length of the
new type by padding the sign bit from the left.
The new type is unsigned

Examples: char to unsigned int, long to unsigned long
In this case the value of negative numbers is not retained. If the new type is of
the same length, the bit pattern is retained. However, the bit pattern will be
interpreted differently. The sign bit loses its significance (see the note opposite).
If the new type is longer, sign extension is performed first and the new bit
pattern is then interpreted as unsigned.
3. Conversion of an integral type to a floating-point type

Examples: int to double, unsigned long to float
The number is converted to an exponential floating-point type and the value
retained. When converting from long or unsigned long to float, some
rounding may occur.
4. Conversion of a floating-point type to a larger floating-point type

Examples: float to double, double to long double
The value is retained during this type conversion.

144

■

CHAPTER 8

CONVERTING ARITHMETIC TYPES

■

IMPLICIT TYPE CONVERSIONS IN ASSIGNMENTS

Example 1:
int i = 100;
long lg = i + 50;

// Result of type int is
// converted to long.

Example 2:
long lg = 0x654321; short st;
st = lg;
//0x4321 is assigned to st.

Example 3:
int i = –2; unsigned int ui = 2;
i = i * ui;
// First the value contained in i is converted to
// unsigned int (preserving the bit pattern) and
// multiplied by 2 (overflow!).
// While assigning the bit pattern the result
// is interpreted as an int value again,
// i.e. –4 is stored in i.

Example 4:
double db = –4.567;
int i; unsigned int ui;
i = db;
// Assigning –4.
i = db – 0.5;
// Assigning –5.
ui = db;
// –4 is incompatible with ui.

Example 5:
double d = 1.23456789012345;
float f;
f = d;
// 1.234568 is assigned to f.

IMPLICIT TYPE CONVERSIONS IN ASSIGNMENTS

■

145

Arithmetic types can also be mixed in assignments. The compiler adjusts the type of the
value on the right of the assignment operator to match the type of the variable on the
left.
In the case of compound assignments, calculations using normal arithmetic type conversions are performed first before type conversion is performed following the rule for
simple assignments.
Two different cases can occur during type conversion in assignments:
1. If the type of the variable is larger than the type of the value to be assigned, the
type of the value must be promoted. The rules for usual arithmetic type conversions are applied in this case (see Example 1).
2. If the type of the value to be assigned is larger, this type must be “demoted.” The
following procedures are followed depending on individual circumstances:
a. Conversion of an integral type to a smaller type:
■

the type is converted to a smaller type by removing the most significant
byte(s). The bit pattern that remains will be interpreted as unsigned, if the
new type is also unsigned, and as signed in all other cases. The value can
only be retained if it can be represented by the new type (see Example 2).

■

when converting an unsigned type to a signed type of the same scale,
the bit pattern is retained and will be interpreted as signed (see Example
3).

b. Conversion of a floating-point type to an integral type
The decimal part of the floating-point number is removed. For example, 1.9
converts to the integer 1. Rounding can be achieved by adding 0.5 to a positive floating-point number or subtracting 0.5 from a negative floating-point
number. This would allow for converting (1.9 + 0.5) to 2.
If the resulting integer is too large or too small for the new type, the result
is unpredictable. This particularly applies to converting negative floatingpoint numbers to unsigned integers (see Example 4).
c. Conversion of a floating-point type to a smaller type
If the floating-point number falls within the range of the new type, the value
will be retained, although the accuracy may be compromised. If the value is
too large to be represented by the new type, the result is unpredictable (see
Example 5).

146

■

CHAPTER 8

CONVERTING ARITHMETIC TYPES

■

MORE TYPE CONVERSIONS

Sample program
// Ellipse.cpp
// The program draws an ellipse.
// The points (x,y) on an ellipse with center (0,0)
// and axes A and B satisfy:
//
x = A*cos(t), y = B*sint(t) for 0 <= t <= 2*PI .
//-----------------------------------------------------#include 
#include 
using namespace std;

// Prototypes of sin() and cos()

#define CLS
(cout << "\033[2J")
#define LOCATE(z,s) (cout <<"\033["<<(z)<<';'<<(s)<<'H')
#define DOT(x,y)
(LOCATE(y,x) << '*')
#define
#define
#define
#define
#define

PI
Mx
My
A
B

int main()
{
int x, y;

3.1416
40
12
25
10

//
//
//
//

The point
center of
Length of
Length of

(Mx, My) is the
the ellipse.
main axis
subsidiary axis

// Screen coordinates.

CLS;
// 0 <= t <= PI/2 is a 1/4-circle:
for( double t = 0.0 ; t <= PI/2 ; t += 0.03)
{
x = (int) (A * cos(t) + 0.5);
y = (int) (B * sin(t) + 0.5);
DOT( x+Mx, y+My);
DOT( x+Mx,-y+My);
DOT(-x+Mx, y+My);
DOT(-x+Mx,-y+My);
}
LOCATE(24,0);
return 0;
}

MORE TYPE CONVERSIONS

■

147

䊐 Implicit Type Conversions in Function Calls
In the case of function calls, arguments with arithmetic types are converted to the types of
the corresponding parameters, similarly to conversions in assignments.

Example: void func( short, double);
int size = 1000;
// . . .
func( size, 77);

// Prototype

// Call

The function func() has two parameters belonging to the short and double types.
However, the function is called using two int arguments. This leads to implicit conversion of the value of size to short and the integer 77 to double.
When an int is converted to short the compiler issues a warning, since some data
loss may occur. You can use explicit type conversion to avoid warnings during type conversion.

䊐 Explicit Type Conversion
It is possible to convert the type of an expression explicitly using the cast operator
(type).

Syntax:

(type) expression

This converts the value of an expression to the given type. Explicit type conversion is
also known as casting.
The cast operator (type) is a unary operator and thus has a higher precedence than
the arithmetic operators.

Example: int a = 1, b = 4;
double x;
x = (double)a/b;

In this example the value of a is explicitly converted to a double. Following the conventions of usual implicit type conversion, b is also converted to double and a floatingpoint division is performed. The exact result, 0.25, is assigned to the variable x.
Without casting, an integer division with a result of 0 would have occurred.
C++ has additional operators for explicit type conversion—the cast operator
dynamic_cast<>, for example. These operators, which are described in later chapters,
are required for special circumstances, for example, to perform type checking at runtime
when converting classes.

■

CHAPTER 8

■

exercise s

148

CONVERTING ARITHMETIC TYPES

EXERCISES

Program listing for exercise 3
// Convert.cpp —> Demonstrates type conversions.
#include 
#include 
using namespace std;
int main()
{
char v_char = 'A';
cout << "v_char:
" << setw(10) << v_char
<< setw(10) << (int)v_char
<< endl;
short v_short = –2;
cout << "v_short:
" << dec << setw(10) << v_short
<< hex << setw(10) << v_short
<< endl;
unsigned short v_ushort = v_short;
cout << "v_ushort:
" << dec << setw(10) << v_ushort
<< hex << setw(10) << v_ushort
<< endl;
unsigned long v_ulong = v_short;
cout << "v_ulong:
" << hex << setw(20) << v_ulong
<< endl;
float v_float = –1.99F;
cout << "v_float:
" << setw(10) << v_float << endl;
cout << "(int)v_float: " << setw(10)
<< dec << (int)v_float << endl;
return 0;
}

Graphic for exercise 4

-------

The Sine Function

-------

Ÿ sin(x)
Ő
Ő 1
*******
Ő
***
***
Ő
**
**
Ő
*
*
Ő
**
**
Ő
*
*
Ő **
**
Ő*
*
2PI x
ņŐņņņņņņņ*ņņņņņņņŐņņņņņņņŐņņņņņņņŐņņņņņņņ*ņņņņņņņŐņņņņņņņŐņņņņņņņŐņņņņņņņ*ņņņņŹ
Ő
*
*
Ő
**
**
Ő
*
*
Ő
**
**
Ő
*
*
Ő
**
**
Ő
***
***
Ő -1
*******
Ő

EXERCISES

■

149

Exercise 1
A function has the following prototype
void func( unsigned int n);

What happens when the function is called with –1 as an argument?

Exercise 2
How often is the following loop executed?
unsigned int limit = 1000;
for (int i = –1; i < limit; i++)
// . . .

Exercise 3
What is output when the program opposite is executed?
Exercise 4
Write a C++ program to output the sine curve on screen as in the graphic
shown on the opposite page.

✓

NOTE
1. Plot one point of the curve in columns 10, 10+1, ..., 10+64 respectively. This leads to a
step value of 2*PI/64 for x.
2. Use the following extended ASCII code characters to draw the axes:
Character

Decimal

Octal

–

196

304

+

197

305

16

020

30

036

Example:

cout << '\020';

// up arrowhead

■

CHAPTER 8

■

solutions

150

CONVERTING ARITHMETIC TYPES

SOLUTIONS

Exercise 1
When called, the value –1 is converted to parameter n, i.e. to unsigned int.
The pattern of –1 is interpreted as unsigned, which yields the greatest unsigned
value.
On a 32-bit system, –1 has the bit pattern 0xFFFFFFFF, which, when
interpreted as unsigned, corresponds to the decimal value 4 294 967 295.
Exercise 2
The statement within the loop is not executed at all! In the expression
i < limit

the value of variable i, –1, is implicitly converted to unsigned int and thus it
represents the greatest unsigned value (see Exercise 1).

Exercise 3
The screen output of the program
v_char:
v_short:
v_ushort:
v_ulong:
v_float:
(int)v_float:

A
-2
65534

65
fffe
fffe
fffffffe

-1.99
-1

Exercise 4
// ----------------------------------------------------//
sinCurve.cpp
//
Outputs a sine curve
// ----------------------------------------------------#include 
#include 
using namespace std;
#define
#define
#define
#define
#define

// Prototypes of sin()

CLS
(cout << "\033[2J")
LOCATE(z,s) (cout <<"\033["<<(z)<<';'<<(s)<<'H')
PI
3.1415926536
START
0.0
// Lower limit
END
(2.0 * PI)
// Upper limit

SOLUTIONS

#define
#define
#define
#define

PNT
STEP
xA
yA

64
// Number of points on the curve
((END-START)/PNT)
14
// Row of x-axis
10
// Column of y-axis

int main()
{
int row, column;
CLS;
LOCATE(2,25);
cout << "------//

---

The Sine Function

-------";

Draws the coordinate system: ---

LOCATE(xA,1);
// x-axis
for( column = 1 ; column < 78 ; ++column)
{
cout << ((column - yA) % 8 ? '\304' : '\305');
}
cout << '\020';
// top
LOCATE(xA-1, yA+64); cout << "2PI x";
for( row = 5 ; row < 24 ; ++row)
{
LOCATE(row, yA); cout << '\305';
}
LOCATE( 4, yA); cout << "\036 sin(x)";
LOCATE( xA-8, yA+1);
LOCATE( xA+8, yA+1);
//

// y-axis

// top

cout << " 1";
cout << " -1";

--- Displays the sine function:

---

int begpt = yA,
endpt = begpt + PNT;
for( column = begpt ; column <= endpt
{
double x = (column-yA) * STEP;
row = (int)(xA - 8 * sin(x) + 0.5);
LOCATE( row, column); cout << '*';
}
LOCATE(25,1);
return 0;
}

■

;

++column)

// Cursor to the last row

151

This page intentionally left blank

chapter

9

The Standard Class
string
This chapter introduces the standard class string, which is used to
represent strings. Besides defining strings we will also look at the various
methods of string manipulation.These include inserting and erasing,
searching and replacing, comparing, and concatenating strings.

153

154

■

CHAPTER 9

THE STANDARD CLASS STRING

■

DEFINING AND ASSIGNING STRINGS

Initializing
string message = "Good Morning!";
String message in memory:

✓

String
literal:

'G'

Length:

13

'o'

'o'

'd'

''

'M'

'o'

'r'

'n'

'i'

'n'

g'

'!'

NOTE
Objects of class string do not necessarily contain the string terminating character '\0', as is the case
with C strings.

Sample program
// string1.cpp: Using strings
#include 
#include 
using namespace std;
string prompt("Enter a line of text: "),
// Global
line( 50, '*');
// strings
int main()
{
string text;
// Empty string
cout << line << endl << prompt << endl;
getline( cin, text);
// Reads a line of text
cout << line << endl
<< "Your text is " << text.size()
<< " characters long!" << endl;
// Two new strings:
string copy(text),
// a copy and the
start(text,0,10);
// first 10 characters
// starting with
// position 0.
cout << "Your text:\n" << copy << endl;
text = "1234567890";
// Assignment
cout << line << endl
<< "The first 10 characters:\n" << start << endl
<< text << endl;
return 0;
}

DEFINING AND ASSIGNING STRINGS

■

155

C++ uses the standard class string to represent and manipulate strings allowing for
comfortable and safe string handling. During string operations the required memory
space is automatically reserved or modified. The programmer does not need to concern
himself or herself with internal memory allocation.
The string class is defined in the string header file and was mentioned in Chapter 3 as an example for the use of classes. Several operators are overloaded for strings,
that is, they were also defined for the string class. This allows for easy copying, concatenation, and comparison. Additionally, various methods for string manipulation such
as insertion, erasing, searching, and replacing are available.

䊐 Initializing Strings
A string, that is, an object belonging to the string class, can be initialized when you
define it using
■
■
■

a predefined string constant
a certain number of characters
a predefined string or part of a string.

If a string is not initialized explicitly, an empty string with a length of 0 is created.
The length of a string, that is, the current number of characters in the string, is stored
internally and can be accessed using the length() method or its equivalent size().

Example: string message("Good morning!");
cout << message.length();

// Output: 13

䊐 String Assignments
When you assign a value to a string, the current contents are replaced by a new character
sequence. You can assign the following to a string object:
■
■
■

another string
a string constant or
a single character.

The memory space required is adjusted automatically.
The program on the opposite page uses the function getline(), which was introduced in an earlier chapter, to store a line of text from the keyboard in a string. In contrast, the >> operator reads only one word, ignoring any leading white space. In both
cases the original content of the string is lost.

156

■

CHAPTER 9

THE STANDARD CLASS STRING

■

CONCATENATING STRINGS

Sample program
// string2.cpp: Reads several lines of text and
//
outputs in reverse order.
#include 
#include 
using namespace std;
string prompt("Please enter some text!\n"),
line( 50, '-');
int main()
{
prompt+="Terminate the input with an empty line.\n ";
cout << line << '\n' << prompt << line << endl;
string text, line;
// Empty strings
while( true)
{
getline( cin, line);
// Reads a line of text
if( line.length() == 0) // Empty line?
break;
// Yes ->end of the loop
text = line + '\n' + text;
// Inserts a new
// line at the beginning.
}
// Output:
cout << line << '\n'
<< "Your lines of text in reverse order:"
<< '\n' << line << endl;
cout << text << endl;
return 0;
}

Sample output for this program
--------------------------------------Please enter some text!
Terminate the input with an empty line.
--------------------------------------Babara, Bobby, and Susan
will go to the movies today
--------------------------------------Your lines of text in reverse order:
--------------------------------------will go to the movies today
Babara, Bobby, and Susan

CONCATENATING STRINGS

■

157

Within the string class the operators + and += are defined for concatenating, and the
operators ==, !=, <, <=, >, and >= are defined for comparing strings. Although these
operators are being applied to strings, the well-known rules apply: the + has precedence
over the comparative operators, and these in turn have higher precedence than the
assignment operators = and +=.

䊐 Using + to Concatenate Strings
You can use the + operator to concatenate strings, that is, to join those strings together.

Example: string sum, s1("sun"), s2("flower");
sum = s2 + s3;

This example concatenates the strings s1 and s2. The result, "sunflower" is then
assigned to sum.
Two strings concatenated using the + operator will form an expression of the string
type. This expression can in turn be used as an operand in a more complex expression.

Example: string s1("sun"),s2("flower"),s3("seed");
cout << s1 + s2 + s3;

Since the + operator has precedence over the << operator, the strings are concatenated
before the “sum” is output. Concatenation takes place from left to right. String constants
and single characters are also valid as operands in expressions containing strings:

Example: string s("Good morning ");
cout << s + "mister X" + '!';

✓

NOTE
At least one operand must be a string class object. The expression "Good morning " +
"mister X" would be invalid!

䊐 Using += to Concatenate Strings
Strings can also be concatenated by first performing concatenation and then assigning
the result.

Example: string s1("Good "),s2("luck!");
s1 = s1 + s2;

// To concatenate s2 and s1

This example creates a temporary object as a result of s1 + s2 and then assigns the
result to s1. However, you can obtain the same result using the assignment operator += ,
which is far more efficient.

Example: s1 += s2;
s1 += "luck!";

// To concatenate s2 and s1.
// Also possible

This adds the content of the second string directly to s1. Thus, the += operator is preferable to a combination of the + and = operators.

158

■

CHAPTER 9

THE STANDARD CLASS STRING

■

COMPARING STRINGS

Sample program
// string3.cpp: Inputs and compares lines of text.
#include 
#include 
using namespace std;
string prompt = "Please enter two lines of text!\n",
line( 30, '-');
int main()
{
string line1, line2, key = "y";
while( key == "y" || key == "Y")
{
cout << line << '\n' << prompt << line << endl;
getline( cin, line1);
// Read the first
getline( cin, line2);
// and second line.
if( line1 == line2)
cout << " Both lines are the same!" << endl;
else
{
cout << "The smaller line is:\n\t";
cout << (line1 < line2 ? line1 : line2)
<< endl;
int len1 = line1.length(),
len2 = line2.length();
if( len1 == len2)
cout << "Both lines have the same length! \n";
else
{ cout << "The shorter line is:\n\t";
cout << (len1 < len2 ? line1 : line2)
<< endl;
}
}
cout << "\nRepeat? (y/n) ";
do
getline( cin, key);
while(
key != "y" && key != "Y"
&& key != "n" && key != "N");
}
return 0;
}

✓

NOTE
The relational operators yield the desired result for strings only if at least one operand is an object of
class string. See Chapter 17, Pointers and Arrays, for more information.

COMPARING STRINGS

■

159

The comparative operators
==

!=

<

<=

>

>=

were overloaded in the string class to allow easy comparison of strings. This also
allows you to use strings to formulate the conditions for branches and loops.

Example: // str1 and str2 are objects of type string
if( str1 < str2)
. . .

// str1 is less than str2?

䊐 Results of Comparisons
Strings are compared lexicographically, that is character by character, beginning at the
first character. To decide whether a single character is smaller, greater, or identical to
another character, the character codes of the character set are compared. Thus, if you are
using the ASCII character set, the letter 'A' (ASCII code 65) is smaller than the letter
'a' (ASCII code 97).
A comparison results in a bool type value. Given two strings s1 and s2:
s1 == s2
s1 < s2

is true only if both strings are identical; this requires that both strings
are exactly the same length.
is true only if the first character in s1 that differs from the corresponding character in s2 is smaller than the corresponding character in s2,
or if s2 is simply an extension of s1.

All other comparative operations can be deduced from the above rules. For example, the
expression s1 > s2 is true only if s2 < s1 is also true.
In an expression comparing strings, one operand can again be a string constant or a
single character.

Example: while( key == 'y' ) { . . . }
This example compares the string key with the single character 'y'. This is an alternative method of expressing the comparison key == "y".
String comparisons can also be combined to form more complex expressions.

Example: while( key == "y" || key == "Y")
{ . . . }

The controlling expression is valid if the string key contains only the letter 'Y' or 'y'.
Due to the higher precedence of the comparative operator versus the || operator, no
parentheses are required in this example.

160

■

CHAPTER 9

■

THE STANDARD CLASS STRING

INSERTING AND ERASING IN STRINGS

䊐 Inserting a string
string s1("Miss Summer");
s1.insert(5, "Ashley "); // Insert at position: 5

Effect of the statement:
Position:

0

1

2

3

4

5

6

7

8

9

10

String s1

'M'

'i'

's'

's'

' '

'S'

'u'

'm'

'm'

'e'

'r'

'A'

's'

'h'

'l'

'e'

'y'

' '

Erasing a substring
string s("The summer-time");
s.erase(4,7);
// Start position: 4, Quantity: 7

Effect of the statement:
Position:

0

1

2

3

4

String s
before

'T' 'h'

'e'

' '

String s
afterwards

'T'

'e'

' '

'h'

5

6

7

8

9

10 11 12 13 14

's'

'u' 'm' 'm' 'e'

'r'

'-'

't'

'i'

'm' 'e'

't'

'i'

'm' 'e'

INSERTING AND ERASING IN STRINGS

■

161

The string class contains numerous methods for performing string manipulations. A
method exists for each operation, such as inserting, erasing, searching, and replacing.
These methods generally allow passing a string constant instead of a second string. A single character can also be used wherever appropriate.

䊐 Insertion
The method insert() inserts a string at a certain position of another string. The position is passed as the first argument and defines the character before which to insert the
string. The first character in a string occupies position 0, the second character position 1,
and so on.

Example: string s1("Miss Summer");
s1.insert(5, "Ashley ");

The string "Ashley " is inserted into the string s1 at position 5, that is in front of the
'S' character in "Summer". Following this, the string "Miss Ashley Summer" is
assigned to s1.
If you need to insert only part of a string into another string, you can pass two additional arguments to the insert() method, the starting position and the length of the
string.

Example: string s1("Ashley is a devil"),
s2(" sweetheart");
s1.insert(12, s2, 0, 12);

This example inserts the first 12 characters from the string s2 at position 13 in string s1.
String s1 then contains the string “Ashley is a sweetheart".

䊐 Erasing
You can use the erase() method to delete a given number of characters from a string.
The starting position is supplied as the first argument and the number of characters to be
erased is the second argument.

Example: string s("The summer-time");
s.erase(4,6);

// Result: "The time"

This statement deletes 7 characters from string s starting at position 4. The erase()
method can also be called without specifying a length and will then delete all the characters in the string up to the end of the string.

Example: string s("winter-story");
s.erase(6);

// s now contains "winter"

You can also call erase() without any arguments to delete all the characters in a
string.

162

■

CHAPTER 9

■

THE STANDARD CLASS STRING

SEARCHING AND REPLACING IN STRINGS

䊐 Replacing substrings
a. Example “Bob and Bill”
string s1("There they go again!"),
s2("Bob and Bill");
s1.replace(6, 4, s2);

Effect of the statement:
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19
s1 'T' 'h' 'e' 'r' 'e' ' ' 't' 'h' 'e' 'y' ' ' 'g' 'o' '' 'a' 'g' 'a' 'i' 'n' '!'

s2 'B' 'o' 'b' ' ' 'a' 'n' 'd' ' ' 'B' 'i' 'l' 'l'

b. Example “my love”
string s1("Here comes Mike!"), s2("my love?");
s1.replace(11, 4, s2, 0, 7);

Effect of the statement:
0

1

2

3

4

s1

'H' 'e' 'r' 'e' ' '

s2

'm' 'y'

''

5

6

7

8

9 10 11 12 13 14 15

'c' 'o' 'm' 'e' 's' ' ' 'M' 'i'

'l' 'o' 'v' 'e' '?'

'k' 'e' '!'

SEARCHING AND REPLACING IN STRINGS

■

163

䊐 Searching
You can search strings to find the first or last instance of a substring. If the string contains the required substring, the position of the substring found by the search is returned.
If not, a pseudo-position npos, or –1, is returned. Since the npos constant is defined in
the string class, you can reference it as string::npos.
The find() method returns the position at which a substring was first found in the
string. The method requires the substring to be located as an argument.

Example: string youth("Bill is so young, so young");
int first = youth.find("young");

The variable first has a value of 11 in this example.
You can use the “right find” method rfind() to locate the last occurrence of a substring in a string. This initializes the variable last with a value of 21 in our example.

Example: int last = youth.rfind("young");

䊐 Replacing
When replacing in strings, a string overwrites a substring. The string lengths need not be
identical.
You can use the replace() method to perform this operation. The first two arguments supply the starting position and the length of the substring to be replaced. The
third argument contains the replacement string.

Example: string s1("There they go again!"),
s2("Bob and Bill");
int pos = s1.find("they");
if( pos != string::npos )
s1.replace(pos, 2, s2);

// pos == 6

This example uses the string s2 to replace 4 characters, "they", starting at position 6 in
s1. After this operation s1 contains the string "There Bob and Bill go
again!".
If you only need to insert part of a string, you can use the fourth argument to define
the starting position and the fifth to define the length of the substring.

Example: string s1("Here comes Mike!"),
s2("my love?");
s1.replace(11, 4, s2, 0, 7);

The string s1 is changed to "Here comes my love!".

164

■

CHAPTER 9

THE STANDARD CLASS STRING

■

ACCESSING CHARACTERS IN STRINGS

Sample program
// string4.cpp
// The program counts words and white space characters.
// (A word is the maximum sequence of characters
// containing no white space characters.)
//--------------------------------------------------------------#include 
#include 
#include 
// Macro isspace()
using namespace std;
int main()
{
string header("
**** Counts words
****\n"),
prompt("Enter a text and terminate"
" with a period and return:"),
line( 60, '-'),
text;
// Empty string
cout << header << endl << prompt << endl
<< line << endl;
getline( cin, text, '.');
// Reads a text up to
// the first '.'
// Counts words and white space characters
int i,
// Index
nSpace = 0,
// Number of white spaces
nWord = 0;
// Number of words
bool fSpace = true; // Flag for white space
for( i = 0; i < text.length(); ++i)
{
if( isspace( text[i]) ) // white space?
{
++nSpace; fSpace = true;
}
else if( fSpace)
// At the beginning of a word?
{
++nWord;
fSpace = false;
}
}
cout << line
// Outputs the result.
<< "\nYour text contains (without periods)"
<< "\n
characters: " << text.length()
<< "\n
words: " << nWord
<< "\n
white spaces: " << nSpace
<< endl;
return 0;
}

ACCESSING CHARACTERS IN STRINGS

■

165

When manipulating strings it is often important to access the individual characters that
form the string. C++ has the operator [] and the method at() for this purpose. An
individual character is always identified by its index, also referred to as subscript, that is,
its position in the string. The first character will always have an index value of 0, the
second an index of 1, and so on.

䊐 Subscript Operator
The easiest way to access a single character in the string is to use the subscript operator
[]. If you define a string as follows,

Example: string s = "Let";
the individual characters in the string are:
s[0] == 'L',

s[1] == 'e',

s[2] == 't'

The last character in a string always has an index of s.length() – 1. You can use
the subscript operator to read any character in a string and also to overwrite a character,
provided the string was not defined as a constant.

Example: char c = s[0];
This statement copies the first character from s to the variable c. In contrast

Example: s[s.length() –1] = 'g';
overwrites the last character in the string s. Following this, s will contain the string
"Leg".

䊐 Invalid Indices
Any integral expression can be used as an index. However, no error message occurs if the
boundaries of a valid index are overstepped.

Example: cout << s[5];

// Error

Your program’s reaction to an invalid index is undefined; this requires careful attention by the programmer! You can call the at() method if you need to perform range
checks.

䊐 The at() method
You can also use the at()method to access a single character.

Example: s.at(i) = 'X';

is equivalent to

s[i] = 'X';

In contrast to the subscript operator, the at() method performs range checking. If an
invalid index is found an exception occurs and the program will normally be terminated at
this point. However, you can specify how a program should react to an exception.

■

CHAPTER 9

THE STANDARD CLASS STRING

■

exercise s

166

EXERCISES

For exercise 3
// timeStr.cpp
// Demonstrates operations on a string containing
// the present time.
#include 
#include 
#include 
using namespace std;

// For time(), ctime(), ...

int main()
{
long sec;
time( &sec);

// Reads the present time
// (in seconds) into sec.
string tm = ctime( &sec);
// Converts the
// seconds to a string.
cout << "Date and time: " << tm << endl;
string hr(tm, 11, 2); // Substring of tm starting at
// position 11, 2 characters long.
string greeting("Have a wonderful ");
if( hr < "10")
// Compares strings
greeting += "Morning!";
else if( hr < "17")
greeting += "Day!";
else
greeting += "Evening!";
cout << greeting << endl;
return 0;

}

EXERCISES

■

167

Exercise 1
Write a C++ program to
■

■
■
■

initialize a string s1 with the string "As time by ..." and a second
string s2 with the string "goes",
insert string s2 in front of "by" in string s1,
erase the remainder of string s1 after the substring "by",
replace the substring "time" in s1 with "Bill".

In each case, your program should determine the position of the substring.
Output string s1 on screen at the beginning of the program and after every
modification.

Exercise 2
Write a C++ program that reads a word from the keyboard, stores it in a string,
and checks whether the word is a palindrome.A palindrome reads the same
from left to right as from right to left.The following are examples of
palindromes:“OTTO, ” “deed, ” and “level.”
Use the subscript operator []. Modify the program to continually read and
check words.
Exercise 3
Write down the screen output for the program on the opposite page.

✓

NOTE
The function time() returns the current time as the number of seconds since 1/1/1970, 0:0. The
number of seconds is stored in the variable sec, whose address was supplied as &sec when the
function was called.
The function ctime() converts the number of seconds to a string with a date and time and returns
this string. The string comprises exactly 26 characters including the null character \0 and has the
following format:
Weekday Month Day Hr:Min:Sec Year\n\0
Example:

Wed Jan 05 02:03:55 2000\n\0

■

CHAPTER 9

■

solutions

168

THE STANDARD CLASS STRING

SOLUTIONS

Exercise 1
// -----------------------------------------------------// strDemo.cpp: Insert, search, and replace in strings.
// -----------------------------------------------------#include 
#include 
using namespace std;
string header = "Demonstrating the use of strings\n",
s1 = "As time by ...",
s2 = "goes ";
int main()
{
int pos = 0;
cout << header << endl;
cout << "s1 : " << s1 << endl;
// To insert:
cout << "\nInserting in string \"" << s2 <<"\""<< endl;
pos = s1.find("by");
if( pos != string::npos )
s1.insert(pos,s2);
cout << "s1 : " << s1 << endl;

// Result

// To erase:
cout << "\nTo erase remaining characters behind \"by\":"
<< endl;
pos = s1.find("by");
if( pos != string::npos )
s1.erase(pos + 3);
cout << "s1 : " << s1 << endl;

// Result

// To replace:
cout << "\nTo replace \"time\" by \"Bill\":"
<< endl;
pos = s1.find("time");
if( pos != string::npos )
s1.replace(pos, 4, "Bill");
cout << "s1 : " << s1 << endl;
return 0;
}

// Result

SOLUTIONS

■

Exercise 2
// ----------------------------------------------------// palindrome.cpp: Reads and compares lines of text.
// ----------------------------------------------------#include 
#include 
using namespace std;
string header = " * * * Testing palindromes * * * ",
prompt = "Enter a word: ",
line( 50, '-');
int main()
{
string word;
char key = 'y';

// Empty string

cout << "\n\t" << header << endl;
while( key == 'y' || key == 'Y')
{
cout << '\n' << line << '\n'
<< prompt;
cin >> word;
// Compares the first and last character,
// the second and the second to last etc.
int i = 0, j = word.length() - 1;
for( ; i <= j ; ++i, --j)
if( word[i] != word[j] )
break;
if( i > j)
// All characters equal?
cout << "\nThe word " << word
<< " is a P A L I N D R O M E !" << endl;
else
cout << "\nThe word " << word
<< " is not a palindrome" << endl;
cout << "\nRepeat? (y/n) ";
do
cin.get(key);
while(
key != 'y' && key != 'Y'
&& key != 'n' && key != 'N');
cin.sync();
}
return 0;
}

169

170

■

CHAPTER 9

THE STANDARD CLASS STRING

Exercise 3
The program outputs the date and time first.Then a greeting is printed
according the time of day. For example:
Date and time: Thu Nov 28 09:01:37 2001
Have a wonderful morning!

chapter

10

Functions
This chapter describes how to write functions of your own. Besides the
basic rules, the following topics are discussed:
■

passing arguments

■

definition of inline functions

■

overloading functions and default arguments

■

the principle of recursion.

171

172

■

CHAPTER 10

■

FUNCTIONS

SIGNIFICANCE OF FUNCTIONS IN C++

Elements of a C++ program

C++ program

Core elements of
C++
(built-in types,
operators,
control structures)

Functions and
classes of the
standard library

Self-defined
functions and
classes and
other libraries

SIGNIFICANCE OF FUNCTIONS IN C++

■

173

C++ supports efficient software development on the lines of the top-down principle. If
you are looking to provide a solution for a more complex problem, it will help to divide
the problem into smaller units. After identifying objects you will need to define classes
that describe these objects. You can use available classes and functions to do so. In addition, you can make use of inheritance to create specialized classes without needing to
change any existing classes.
When implementing a class you must define the capacities of those objects, that is,
the member functions, in your program. However, not every function is a member function.
Functions can be defined globally, such as the function main() for example. Functions of this type do not belong to any particular class but normally represent algorithms
of a more general nature, such as the search or sort functions of the standard library.

䊐 Libraries
You will not need to program each “building block” yourself. Many useful global functions and classes are available from the C++ standard library. In addition, you can use
other libraries for special purposes. Often a compiler package will offer commercial class
libraries or graphical user interfaces. Thus, a C++ program will be made up of
■
■
■

language elements of the C++ core
global functions and classes from the C++ standard library
functions and classes you have programmed yourself and other libraries.

Classes and functions that belong together are normally compounded to form separate
source files, which can be compiled and tested independently. Using software components that you have already tested makes programming a complex solution much easier
and improves the reliability of your programs. You can enhance the reusability of your
source code by compiling your own libraries, but be sure to include comments for ease of
readability.
Compiled source files, also known as modules, are compounded by the linker to an
executable file by reference to the libraries you include. If you modify a source file, you
may also need to recompile other files. In large scale projects it is recommended to use
the MAKE utility for module management. An integrated developer environment will
offer the functionality of this utility when you create a new project. This includes your
own source files, the libraries used, and the compiler/linker settings for program compilation.

174

■

CHAPTER 10

■

FUNCTIONS

DEFINING FUNCTIONS

Example of a function definition
// func1.cpp
#include 
using namespace std;
void test( int, double );

// Prototype

int main()
{
cout << "\nNow function test() will be called.\n";
test( 10, -7.5);
// Call
cout << "\nAnd back again in main()." << endl;
return 0;
}
void test(int arg1, double arg2 )
// Definition
{
cout << "\nIn function test()."
<< "\n 1. argument: " << arg1
<< "\n 2. argument: " << arg2 << endl;
}

General form of a function
[type] name([declaration_list])
{
.
.
What will be done
.
.
}

// Function header
// Beginning

// Function block

// End

DEFINING FUNCTIONS

■

175

The following section describes how to program global functions. Chapter 13, Defining
Classes, describes the steps for defining member functions.

䊐 Definition
Functions can be defined in any order, however, the first function is normally main.
This makes the program easier to understand, since you start reading at the point where
the program starts to execute.
The function test() is shown opposite as an example and followed by the general
form of a function. The example can be read as follows:
type
name
declaration_list

is the function type, that is, the type of the return value.
is the function name, which is formed like a variable name
and should indicate the purpose of the function.
contains the names of the parameters and declares their
types. The list can be empty, as for the function main(),
for example. A list of declarations that contains only the
word void is equivalent to an empty list.

The parameters declared in a list are no more than local variables. They are created
when the function is called and initialized by the values of the arguments.

Example: When test( 10, -7.5); is called, the parameter
arg1 is initialized with a value of 10 and arg2 with -7.5.
The left curved bracket indicates the start of a function block, which contains the statements defining what the function does.

䊐 Prototype and Definition
In a function definition the function header is similar in form to the prototype of a function. The only difference when a function is defined is that the name and declaration list
are not followed by a semicolon but by a function code block.
The prototype is the declaration of the function and thus describes only the formal
interface of that function. This means you can omit parameter names from the prototype, whereas compiling a function definition will produce machine code.

176

■

CHAPTER 10

■

FUNCTIONS

RETURN VALUE OF FUNCTIONS

Defining and calling the function area()
// area.cpp
// Example for a simple function returning a value.
//----------------------------------------------------#include 
#include 
using namespace std;
double area(double, double);
int main()
{
double

x = 3.5, y = 7.2,

res = area( x, y+1);

// Prototype

res;
// Call

// To output to two decimal places:
cout << fixed << setprecision(2);
cout << "\n The area of a rectangle "
<< "\n with width " << setw(5) << x
<< "\n and length " << setw(5) << y+1
<< "\n is
" << setw(5) << res
<< endl;
return 0;
}
// Defining the function area():
// Computes the area of a rectangle.
double area( double width, double len)
{
return (width * len);
// Returns the result.
}

Screen output:
The area of a rectangle
with width
3.50
and length
8.20
is
28.70

RETURN VALUE OF FUNCTIONS

■

177

The program opposite shows how the function area() is defined and called. As previously mentioned, you must declare a function before calling it. The prototype provides
the compiler with all the information it needs to perform the following actions when a
function is called:
■
■

check the number and type of the arguments
correctly process the return value of the function.

A function declaration can be omitted only if the function is defined within the same
source file immediately before it is called. Even though simple examples often define and
call a function within a single source file, this tends to be an exception. Normally the
compiler will not see a function definition as it is stored in a different source file.
When a function is called, an argument of the same type as the parameter must be
passed to the function for each parameter. The arguments can be any kind of expressions,
as the example opposite with the argument y+1 shows. The value of the expression is
always copied to the corresponding parameter.

䊐 Return Statement
When the program flow reaches a return statement or the end of a function code block, it
branches back to the function that called it. If the function is any type other than void,
the return statement will also cause the function to return a value to the function that
called it.

Syntax:

return [expression]

If expression is supplied, the value of the expression will be the return value. If the
type of this value does not correspond to the function type, the function type is converted, where possible. However, functions should always be written with the return
value matching the function type.
The function area() makes use of the fact that the return statement can contain
any expression. The return expression is normally placed in parentheses if it contains
operators.
If the expression in the return statement, or the return statement itself, is missing, the return value of the function is undefined and the function type must be void.
Functions of the void type, such as the standard function srand(), will perform an
action but not return any value.

178

■

CHAPTER 10

■

FUNCTIONS

PASSING ARGUMENTS

Calling function and called function
long func2(int, double);
// . . .
void func1()
{
int x = 1.1;
double y;
. . .
long a = func2(x,y);
. . .
}

// Prototype

// Call of func2().
// Pass by value

long func2(int a, double b) // Definition
{
double x = 2.2;
long result;
.
// Here the result
.
// is computed.
.
return result;
}

Stack content after calling a function
On call
“push”

On return
“pop”
• • •

further local objects

return address
first parameter
• • •

last parameter
Stack

PASSING ARGUMENTS

■

179

䊐 Passing by Value
Passing values to a function when the function is called is referred to as passing by value.
Of course the called function cannot change the values of the arguments in the calling
function, as it uses copies of the arguments.
However, function arguments can also be passed by reference. In this case, the function
is passed a reference to an object as an argument and can therefore access the object
directly and modify it.
An example of passing by reference was provided in the example containing the function time(). When time(&sek); is called, the address of the variable sek is passed
as an argument, allowing the function to store the result in the variable. We will see how
to create functions of this type later.
Passing by value does, however, offer some important advantages:
■
■
■

function arguments can be any kind of expression, even constants, for example
the called function cannot cause accidental modifications of the arguments in
the calling function
the parameters are available as suitable variables within the functions. Additional
indirect memory access is unnecessary.

However, the fact that copying larger objects is difficult can be a major disadvantage,
and for this reason vectors are passed by reference to their starting address.

䊐 Local Objects
The scope of function parameters and the objects defined within a function applies only
to the function block. That is, they are valid within the function only and not related to
any objects or parameters of the same name in any other functions.
For example, the program structure opposite contains a variable a in the function
func1() and in the function func2(). The variables do not collide because they reference different memory addresses. This also applies to the variables x in func1() and
func2().
A function’s local objects are placed on the stack—the parameters of the function are
placed first and in reverse order. The stack is an area of memory that is managed according to the LIFO (last in first out) principle. A stack of plates is a good analogy. The last
plate you put on the stack has to be taken off first. The LIFO principle ensures that the
last local object to be created is destroyed first.

180

■

CHAPTER 10

■

FUNCTIONS

INLINE FUNCTIONS

Call to a function not defined as inline
Program

Function

Branching

void func()
{

// 1st Call

func();

// 2nd Call

}

func();

✓

HINT
The executable file only contains one instance of the function’s machine code.

Call to an inline function
Program

// 1st Call

func();

Inline function

Copy

inline void func()
{
}

// 2nd Call

func();

✓

HINT
The machine code of the function is stored in the executable file wherever the function is called.

INLINE FUNCTIONS

■

181

䊐 Jumping to Sub-Routines
When a function is called, the program jumps to a sub-routine, which is executed as follows:
■
■
■

the function parameters are placed on the stack and initialized with appropriate
arguments
the so-called return address, that is, the place where the function was called, is
stored on the stack and the program flow branches to the function
after executing the function the program uses the return address it stored previously to return to the calling function. The part of the stack occupied by the
function is then released.

All this jumping back and forth can affect the run time of your program, especially if the
function contains only a few instructions and is called quite often. The time taken to
branch to a small function can be greater than the time needed to execute the function
itself. However, you can define inline functions to avoid this problem.

䊐 Inline Definition
The compiler inserts the code of an inline function at the address where the function is
called and thus avoids jumping to a sub-routine. The definition of an inline function is
introduced by the inline keyword in the function header.

Example: inline int max( int x, int y)
{

return

(x >= y ? x : y );

}

The program code will expand each time an inline function is called. This is why
inline functions should contain no more than one or two instructions. If an inline
function contains too many instructions, the compiler may ignore the inline keyword
and issue a warning.
An inline function must be defined in the source file in which it is called. You cannot simply supply a prototype of the function. The code containing the instructions must
also be available to the compiler. It therefore makes sense to define inline functions in
header files, in contrast to “normal” functions. This means the function will be available
in several source files.

䊐 Inline Functions and Macros
Inline functions are an alternative to macros with parameters. When a macro is called,
the preprocessor simply replaces a block of text. In contrast, an inline function
behaves like a normal function, although the program flow is not interrupted by the
function branching. The compiler performs a type check, for example.

182

■

CHAPTER 10

■

FUNCTIONS

DEFAULT ARGUMENTS

Defining the function capital()
// Computes the final capital with interest and
// compound interest.
// Formula: capital = k0 * (1.0 + p/100)n
// where k0 = start capital, p = rate, n = run time
// ---------------------------------------------------#include 
double capital( double k0, double p, double n)
{
return (k0 * pow(1.0+p/100, n));
}

Possible calls
// Function capital() with two default arguments
// Prototype:
double capital( double k0, double p=3.5, double n=1.0);
double endcap;
endcap = capital( 100.0, 3.5, 2.5);
endcap = capital( 2222.20, 4.8);
endcap = capital( 3030.00);

// ok
// ok
// ok

endcap = capital( );
// not ok
// The first argument has no default value.

✓

endcap = capital( 100.0, , 3.0);
// No gap!

// not ok

endcap = capital( , 5.0);
// No gap either.

// not ok

NOTE
A function defined with default arguments is always called with the full number of arguments. For
reasons of efficiency it may be useful to define several versions of the same function.

DEFAULT ARGUMENTS

■

183

So-called default arguments can be defined for functions. This allows you to omit some
arguments when calling the function. The compiler simply uses the default values for any
missing arguments.

䊐 Defining Default Arguments
The default values of a function’s arguments must be known when the function is called.
In other words, you need to supply them when you declare the function.

Example: void moveTo( int x = 0, int y = 0);
Parameter names can be omitted, as usual.

Example: void moveTo( int = 0, int = 0);
The function moveTo() can then be called with or without one or two arguments.

Example: moveTo ();

moveTo (24); moveTo(24, 50);

The first two calls are equivalent to moveTo(0,0); or moveTo(24,0); .
It is also possible to define default arguments for only some of the parameters. The following general rules apply:
■

■
■

the default arguments are defined in the function prototype. They can also be
supplied when the function is defined, if the definition occurs in the same source
file and before the function is called
if you define a default argument for a parameter, all following parameters must
have default arguments
default arguments must not be redefined within the prototype scope (the next
chapter gives more details on this topic).

䊐 Possible Calls
When calling a function with default arguments you should pay attention to the following points:
■
■
■

you must first supply any arguments that do not have default values
you can supply arguments to replace the defaults
if you omit an argument, you must also omit any following arguments.

You can use default arguments to call a function with a different number of arguments
without having to write a new version of the function.

184

■

CHAPTER 10

■

FUNCTIONS

OVERLOADING FUNCTIONS

Sample program
// random.cpp
// To generate and output random numbers.
//----------------------------------------------------#include 
#include 
#include 
// For rand(), srand()
#include 
// For time()
using namespace std;
bool setrand = false;
inline void init_random()
{

// Initializes the random
// number generator with the
// present time.

if( !setrand )
{ srand((unsigned int)time(NULL));
setrand = true;
}
}
inline double myRandom()
// Returns random number x
{
// with 0.0 <= x <= 1.0
init_random();
return (double)rand() / (double)RAND_MAX;
}
inline int myRandom(int start, int end) // Returns the
{
// random number n with
init_random();
// start <= n <= end
return (rand() % (end+1 - start) + start);
}
// Testing myRandom() and myRandom(int,int):
int main()
{
int i;
cout << "5 random numbers between 0.0 and 1.0 :"
<< endl;
for( i = 0; i < 5; ++i)
cout << setw(10) << myRandom();
cout << endl;
cout << "\nAnd now 5 integer random numbers "
"between -100 and +100 :" << endl;
for( i = 0; i < 5; ++i)
cout << setw(10) << myRandom(-100, +100);
cout << endl;
return 0;
}

OVERLOADING FUNCTIONS

■

185

Functions in traditional programming languages, such as C, which perform the same task
but have different arguments, must have different names. To define a function that calculated the maximum value of two integers and two floating-point numbers, you would
need to program two functions with different names.

Example: int

int_max( int x, int y);
double dbl_max( double x, double y);

Of course this is detrimental to efficient naming and the readability of your program—
but luckily, this restriction does not apply to C++.

䊐 Overloading
C++ allows you to overload functions, that is, different functions can have the same
name.

Example: int

max( int x, int y);
double max( double x, double y);

In our example two different function share the same name, max. The function max()
was overloaded for int and double types. The compiler uses a function’s signature to
differentiate between overloaded functions.

䊐 Function Signatures
A function signature comprises the number and type of parameters. When a function is
called, the compiler compares the arguments to the signature of the overloaded functions
and simply calls the appropriate function.

Example: double maxvalue, value = 7.9;
maxvalue = max( 1.0, value);

In this case the double version of the function max() is called.
When overloaded functions are called, implicit type conversion takes place. However,
this can lead to ambiguities, which in turn cause a compiler error to be issued.

Example: maxvalue = max( 1, value);

// Error!

The signature does not contain the function type, since you cannot deduce the type by
calling a function. It is therefore impossible to differentiate between overloaded functions by type.

Example: int

search(string key);
string search(string name);

Both functions have the same signature and cannot be overloaded.

186

■

CHAPTER 10

■

FUNCTIONS

RECURSIVE FUNCTIONS

Using a recursive function
// recursive.cpp
// Demonstrates the principle of recursion by a
// function, which reads a line from the keyboard
// and outputs it in reverse order.
// ---------------------------------------------------#include 
using namespace std;
void getput(void);
int main()
{
cout << "Please enter a line of text:\n";
getput();
cout << "\nBye bye!" << endl;
return 0;
}
void getput()
{
char c;
if( cin.get(c)
getput();
cout.put(c);
}

&&

c != '\n')

Program flow after typing ok

main()
{
...
getput();
}

1st Execution

2nd Execution

3rd Execution

getput()
{
...
// c = 'o'
getput();

getput()
{
...
// c = 'k'
getput();

cout.put(c);
}

}

getput()
{
...
// c = '\n'
// No call of
// getput()
cout.put(c);
}

cout.put(c);

RECURSIVE FUNCTIONS

■

187

䊐 Recursion
A function that calls itself is said to be recursive. This process can also be performed indirectly if the function first calls another function or multiple functions before it is called
once more. But a break criterion is always necessary to avoid having the function call
itself infinitely.
The concept of local objects makes it possible to define recursive functions in C++.
Recursion requires local objects to be created each time the function is called, and these
objects must not have access to any other local objects from other function calls. What
effectively happens is that the local objects are placed on the stack, and thus the object
created last is destroyed first.

䊐 A Sample Program
Let’s look at the principle of recursion by referring to the sample program opposite. The
program contains the recursive function getput() that reads a line of text from the
keyboard and outputs it in reverse order.
The function getput() is first called by main() and reads a character from the keyboard, storing it in the local variable c. If the character is not '\n', the function getput() calls itself again and thus reads a further character from the keyboard before
storing it in the local variable c.
The chain of recursive function calls is terminated by the user pressing the Return
key. The last character to be read, '\n' (line feed), is output and the program flow
branches to the previous getput() instance. This outputs the second to last character,
and so on. When the first character to have been read has finally been output, the program flow is handed back to main().

䊐 Practical Usage
The logic of various solutions to common problems results in a recursive structure, for
example, browsing directory trees, using binary trees for data management, or some sorting algorithms, such as the quick sort algorithm. Recursive functions allow you to formulate this kind of logic in an efficient and elegant manner. However, always make sure
that sufficient memory is available for the stack.

■

CHAPTER 10

■

exercise s

188

FUNCTIONS

EXERCISES

Hint for exercise 1
Working with several source files:

Within an integrated development environment a project, containing all source
files of the program, first has to be created.This ensures that all the source files
will be compiled and linked automatically.
However, when calling the compiler/linker from the command line, it is
sufficient to declare the source files, for example:
cc

sum_t.cpp

sum.cpp

Screen output for exercise 3
n

Factorial of n

0

1

1

1

2

2

3

6

4

24

5

120

6

720

7

5040

.

. . .

.

. . . .

.

. . . . . . .

19

121645100408832000

20

2432902008176640000

EXERCISES

■

189

Exercise 1
a. Write the function sum() with four parameters that calculates the arguments provided and returns their sum.
Parameters: Four variables of type long.
Returns:
The sum of type long.
Use the default argument 0 to declare the last two parameter of the
function sum().Test the function sum() by calling it by all three possible
methods. Use random integers as arguments.
b. Now restructure your program to store the functions main() and
sum() in individual source files, for example, sum_t.cpp and sum.cpp .
Exercise 2
a. Write an inline function, Max(double x, double y), which returns
the maximum value of x and y. (Use Max instead of max to avoid a collision with other definitions of max.) Test the function by reading values
from the keyboard.
Can the function Max() also be called using arguments of the types
char, int, or long?
b. Now overload Max() by adding a further inline function Max(char x,
char y) for arguments of type char .
Can the function Max() still be called with two arguments of type
int?

Exercise 3
The factorial n! of a positive integer n is defined as
n! = 1*2*3 . . . * (n-1) * n
0! = 1

Where
Write a function to calculate the factorial of a number.
Argument: A number n of type unsigned int.
Returns:
The factorial n! of type long double.
Formulate two versions of the function, where the factorial is
a. calculated using a loop
b. calculated recursively
Test both functions by outputting the factorials of the numbers 0 to 20 as shown
opposite on screen.

190

■

CHAPTER 10

FUNCTIONS

Exercise 4
Write a function pow(double base, int exp) to calculate integral powers of
floating-point numbers.
Arguments: The base of type double and the exponent of type int.
Returns:
The power baseexp of type double.
For example, calling pow(2.5, 3) returns the value
2.53

=

2.5 * 2.5 * 2.5

=

15.625

This definition of the function pow()means overloading the standard function
pow(), which is called with two double values.
Test your function by reading one value each for the base and the exponent
from the keyboard. Compare the result of your function with the result of the
standard function.

✓

NOTE
1. The power x0 is defined as 1.0 for a given number x.
2. The power xn is defined as (1/x)-n for a negative exponent n.
3. The power 0n where n > 0 will always yield 0.0 .
The power 0n is not defined for n < 0. In this case, your function should return the
value HUGE_VAL. This constant is defined in math.h and represents a large double
value. Mathematical functions return HUGE_VAL when the result is too large for a
double.

SOLUTIONS

solutions

■

■

SOLUTIONS

Exercise 1
//
//
//
//

----------------------------------------------------sum_t.cpp
Calls function sum() with default arguments.
-----------------------------------------------------

#include 
#include 
#include 
#include 
using namespace std;
long sum( long a1, long a2, long a3=0, long a4=0);
int main()
// Several calls to function sum()
{
cout << " **** Computing sums ****\n"
<< endl;
srand((unsigned int)time(NULL)); // Initializes the
// random number generator.
long res, a = rand()/10, b = rand()/10,
c = rand()/10, d = rand()/10;
res = sum(a,b);
cout << a << " + " << b << " = " << res << endl;
res = sum(a,b,c);
cout << a << " + " << b << " + " << c
<< " = " << res << endl;
res = sum(a,b,c,d);
cout << a << " + " << b << " + " << c << " + " << d
<< " = " << res << endl;
return 0;
}
//
//
//
//

----------------------------------------------------sum.cpp
Defines the function sum()
-----------------------------------------------------

long sum( long a1, long a2, long a3, long a4)
{
return (a1 + a2 + a3 + a4);
}

191

192

■

CHAPTER 10

FUNCTIONS

Exercise 2
//
//
//
//

----------------------------------------------------max.cpp
Defines and calls the overloaded functions Max().
------------------------------------------------------

//
//
//
//

As long as just one function Max() is defined, it can
be called with any arguments that can be converted to
double, i.e. with values of type char, int or long.
After overloading no clear conversion will be possible.

#include 
#include 
using namespace std;
inline double Max(double x, double y)
{
return (x < y ? y : x);
}
inline char Max(char x, char y)
{
return (x < y ? y : x);
}
string header(
"To use the overloaded function Max().\n"),
line(50,'-');
int main()
// Several different calls to function Max()
{
double x1 = 0.0, x2 = 0.0;
line += '\n';
cout << line << header << line << endl;
cout << "Enter two floating-point numbers:"
<< endl;
if( cin >> x1 && cin >> x2)
{
cout << "The greater number is " << Max(x1,x2)
<< endl;
}
else
cout << "Invalid input!" << endl;
cin.sync(); cin.clear();

// Invalid input
// was entered.

SOLUTIONS

cout << line
<< "And once more with characters!"
<< endl;
cout << "Enter two characters:"
<< endl;
char c1, c2;
if( cin >> c1 && cin >> c2)
{
cout << "The greater character is " << Max(c1,c2)
<< endl;
}
else
cout << "Invalid input!" << endl;
cout << "Testing with int arguments." << endl;
int a = 30, b = 50;
cout << Max(a,b) << endl;
// Error! Which
// function Max()?
return 0;
}

Exercise 3
// ----------------------------------------------------// factorial.cpp
// Computes the factorial of an integer iteratively,
// i.e. using a loop, and recursively.
// ----------------------------------------------------#include 
#include 
using namespace std;
#define N_MAX

20

long double fact1(unsigned int n);
long double fact2(unsigned int n);

// Iterative solution
// Recursive solution

int main()
{
unsigned int n;
// Outputs floating-point values without
// decimal places:
cout << fixed << setprecision(0);

■

193

194

■

CHAPTER 10

FUNCTIONS

//

---

cout <<
<<
<<
<<

Iterative computation of factorial

---

setw(10) << "n" << setw(30) << "Factorial of n"
"
(Iterative solution)\n"
"
-----------------------------------------"
endl;

for( n = 0; n <= N_MAX; ++n)
cout << setw(10) << n << setw(30) << fact1(n)
<< endl;
cout << "\nGo on with ";
cin.get();
//

---

cout <<
<<
<<
<<

Recursive computation of factorial

----

setw(10) << "n" << setw(30) << "Factorial of n"
"
(Recursive solution)\n"
"
-----------------------------------------"
endl;

for( n = 0; n <= N_MAX; ++n)
cout << setw(10) << n << setw(30) << fact2(n)
<< endl;
cout << endl;
return 0;
}
long double fact1(unsigned int n)
// Iterative
{
// solution.
long double result = 1.0;
for( unsigned int i = 2; i <= n; ++i)
result *= i;
return result;
}
long double fact2(unsigned int n)
{
if( n <= 1)
return 1.0;
else
return fact2(n-1) * n;
}

// Recursive
// solution.

SOLUTIONS

Exercise 4
// ----------------------------------------------------// power.cpp
// Defines and calls the function pow() to
// compute integer powers of a floating-point number.
// Overloads the standard function pow().
// ----------------------------------------------------#include 
#include 
using namespace std;
double pow(double base, int exp);
int main()
// Tests the self-defined function pow()
{
double base
= 0.0;
int
exponent = 0;
cout << " **** Computing Integer Powers ****\n"
<< endl;
cout << "Enter test values.\n"
<< "Base (floating-point): "; cin >> base;
cout << "Exponent (integer):
"; cin >> exponent;
cout << "Result of " << base << " to the power of "
<< exponent << " = " << pow( base, exponent)
<< endl;
cout << "Computing with the standard function: "
<< pow( base, (double)exponent) << endl;
return 0;
}
double pow(double base, int exp)
{
if( exp == 0)
return 1.0;
if( base == 0.0)
if( exp > 0) return 0.0;
else
return HUGE_VAL;
if( exp < 0)
{
base = 1.0 / base;
exp = -exp;
}
double power = 1.0;
for( int n = 1; n <= exp; ++n)
power *= base;
return power;
}

■

195

This page intentionally left blank

chapter

11

Storage Classes and
Namespaces
This chapter begins by describing storage classes for objects and
functions.The storage class is responsible for defining those parts of a
program where an object or function can be used. Namespaces can be
used to avoid conflicts when naming global identifiers.

197

198

■

CHAPTER 11

■

ST0RAGE CLASSES AND NAMESPACES

STORAGE CLASSES OF OBJECTS

䊐 Availability of Objects
C++ program
program scope
Module 1

file scope
Function

block scope

Function

block scope

Module 2

file scope
Function

block scope

䊐 Storage Class Specifiers
The storage class of an object is determined by
■
■

the position of its declaration in the source file
the storage class specifier, which can be supplied optionally.

The following storage class specifiers can be used
extern

static

auto

register

STORAGE CLASSES OF OBJECTS

■

199

When an object is declared, not only are the object’s type and name defined but also its
storage class. The storage class specifies the lifetime of the object, that is, the period of
time from the construction of the object until its destruction. In addition, the storage
class delimits the part of the program in which the object can be accessed directly by its
name, the so-called object scope.
Essentially, an object is only available after you have declared it within a translation
unit. A translation unit, also referred to as module, comprises the source file you are compiling and any header files you have included.
As a programmer, you can define an object with:
■

block scope

■

file scope

■

program scope

The object is only available in the code block in which it was
defined. The object is no longer visible once you have left
the code block.
The object can be used within a single module. Only the
functions within this module can reference the object. Other
modules cannot access the object directly.
The object is available throughout the program, providing a
common space in memory that can be referenced by any program function. For this reason, these objects are often
referred to as global.

Access to an object as defined by the object’s storage class is independent of any
access controls for the elements of a class. Namespaces that subdivide program scope and
classes will be introduced at a later stage.

䊐 Lifetime
Objects with block scope are normally created automatically within the code block that
defines them. Such objects can only be accessed by statements within that block and are
called local to that block. The memory used for these objects is freed after leaving the
code block. In this case, the lifetime of the objects is said to be automatic.
However, it is possible to define objects with block scope that are available throughout the runtime of a program. The lifetime of these objects is said to be static. When the
program flow re-enters a code block, any pre-existing conditions will apply.
Objects with program and file scope are always static. These objects are created when
a program is launched and are available until the program is terminated.
Four storage classes are available for creating objects with the scope and lifetime you
need. These storage classes will be discussed individually in the following sections.

200

■

CHAPTER 11

■

ST0RAGE CLASSES AND NAMESPACES

THE STORAGE CLASS extern

Source file 1
// Cutline1.cpp
// A filter to remove white-space characters
// at the ends of lines.
// -------------------------------------------------#include 
#include 
using namespace std;
void cutline( void );
string line;
int main()
{
while( getline(cin, line))
{
cutline();
cout << line << endl;
}
return 0;
}

// Prototype
// Global string

//
//
//
//

As long as a line
can be read.
Shorten the line.
Output the line.

Source file 2
//
//
//
//
//
//

Cutline2.cpp
Containing the function cutline(), which removes
tabulator characters at the end of the string line.
The string line has to be globally defined in another
source file.
--------------------------------------------------

#include 
using namespace std;
extern string line;
void cutline()
{
int i = line.size();

}

// extern declaration

// Position after the
// last character.

while( i-- >= 0 )
if(
line[i] != ' '
&& line[i] != '\t' )
break;

// If no blank and
// no tab ->
// stop the loop.

line.resize(++i);

// Fix new length.

THE STORAGE CLASS extern

■

201

䊐 Defining Global Objects
If an object is not defined within a function, it belongs to the extern storage class.
Objects in this storage class have program scope and can be read and, provided they have
not been defined as const, modified at any place in the program. External objects thus
allow you to exchange information between any functions without passing any arguments. To demonstrate this point, the program on the opposite page has been divided
into two separate source files. The string line, which has a global definition, is used to
exchange data.
Global objects that are not explicitly initialized during definition receive an initial
value of 0 (that is, all bits = 0) by default. This also applies to objects belonging to class
types, if not otherwise stipulated by the class.

䊐 Using Global Objects
An object belonging to the extern storage class is initially only available in the source
file where it was defined. If you need to use an object before defining it or in another
module, you must first declare the object. If you do not declare the object, the compiler
issues a message stating that the object is unknown. The declaration makes the name and
type of the object known to the compiler.
In contrast to a definition, the storage class identifier extern precedes the object
name in a declaration.

Example: extern long position;

// Declaration

This statement declares position as an external object of type long. The extern
declaration thus allows you to “import” an object from another source file.
A global object must be defined once, and once only, in a program. However, it can
be declared as often as needed and at any position in the program. You will normally
declare the object before the first function in a source file or in a header file that you can
include when needed. This makes the object available to any functions in the file.
Remember, if you declare the object within a code block, the object can only be used
within the same block.
An extern declaration only refers to an object and should therefore not be used to
initialize the object. If you do initialize the object, you are defining that object!

✓

NOTE
Global objects affect the whole program and should be used sparingly. Large programs in particular
should contain no more than a few central objects defined as extern.

202

■

CHAPTER 11

■

ST0RAGE CLASSES AND NAMESPACES

THE STORAGE CLASS static

// Passw1.cpp
// The functions getPassword() and timediff()
// to read and examine a password.
// ----------------------------------------------------#include 
#include 
#include 
#include 
using namespace std;
long timediff(void);
static string secret = "ISUS";
static long maxcount = 3, maxtime = 60;

// Prototype
// Password
// Limits

bool getPassword()
// Enters and checks a password.
{
// Return value: true, if password is ok.
bool ok_flag = false;
// For return value
string word;
// For input
int count = 0, time = 0;
timediff();
// To start the stop watch
while( ok_flag != true &&
++count <= maxcount)
// Number of attempts
{
cout << "\n\nInput the password: ";
cin.sync();
// Clear input buffer
cin >> setw(20) >> word;
time += timediff();
if( time >= maxtime )
// Within time limit?
break;
// No!
if( word != secret)
cout << "Invalid password!" << endl;
else
ok_flag = true;
// Give permission
}
return ok_flag;
// Result
}
long timediff()
// Returns the number of
{
// seconds after the last call.
static long sec = 0;
// Time of last call.
long oldsec = sec;
// Saves previous time.
time( &sec);
// Reads new time.
return (sec - oldsec);
// Returns the difference.
}

THE STORAGE CLASS static

■

203

䊐 Static Objects
If an object definition is preceded by the static keyword, the object belongs to the
static storage class.

Example: static int count;
The most important characteristic of static objects is their static (or permanent) lifetime.
Static objects are not placed on the stack, but are stored in the data area of a program
just like external objects.
However, in contrast to external objects, access to static objects is restricted. Two
conditions apply, depending on where the object is defined:
1. Definition external to all program functions
In this case, the object is external static, that is, the object can be designated using
its name within the module only, but will not collide with any objects using the
same name in other modules.

✓

NOTE
In contrast to objects with an extern definition, the name of an external static object is unknown to
the linker and thus retains its private nature within a module.

2. Definition within a code block
This means that the object is internal static, that is, the object is only visible
within a single block. However, the object is created only once and is not
destroyed on leaving the block. On re-entering the block, you can continue to
work with the original object.
The same rules apply to initializing static objects as they do to external objects. If the
object is not initialized explicitly, a default value of 0 applies.

䊐 Notes on the Sample Programs Opposite
The function getPassword() checks a password that is entered. Permission is refused
following three unsuccessful attempts or when 60 seconds have elapsed. You could use
the following instructions to call the function in another source file:

Example: if( !getPassword() )
cout << "No authorization!\n"; exit(1);

The string secret and the thresholds maxcount and maxtime are external static,
whereas the variable sec in the function timediff() is internal static. Its value is zero
only when the function is first called.
It makes sense to add a further function to these source files providing for password
changes.

204

■

CHAPTER 11

■

ST0RAGE CLASSES AND NAMESPACES

THE SPECIFIERS auto AND register

Sample function with a register variable
//
//
//
//
//
//
//
//
//
//
//
//

StrToL.cpp
The function strToLong() converts a string containing
a leading integer into an integer of type long.
Argument:
A string.
Return value: An integer of type long.
-------------------------------------------------The digits are interpreted with base 10. White spaces
and a sign can precede the sequence of digits.
The conversion terminates when the end of the string is
reached or when a character that cannot be converted is
reached.
--------------------------------------------------

#include 
#include 
using namespace std;
long strToLong( string str)
{
register int i = 0;
long vz = 1, num = 0;

// Type string
// isspace() and isdigit()

// Index
// Sign and number

// Ignore leading white spaces.
for(i=0; i < str.size() && isspace(str[i]); ++i)
;
// Is there a sign?
if( i < str.size())
{
if( str[i] == '+' ) { vz = 1; ++i; }
if( str[i] == '-' ) { vz = ---1; ++i; }
}
// Sequence of digits -> convert to integer
for( ; i < str.size() && isdigit(str[i]); ++i)
num = num * 10 + (str[i] - '0');
return vz * num;
}

THE SPECIFIERS auto AND register

■

205

䊐 auto Objects
The storage class auto (automatic) includes all those objects defined within a function
but without the static keyword. The parameters of a function are also auto objects.
You can use the auto keyword during a definition.

Example: auto float radius; // Equivalent to:
// float radius;

When the program flow reaches the definition, the object is created on the stack, but in
contrast to a static type object, the object is destroyed on leaving the block.
auto objects have no specific initial value if they are not initialized explicitly. However, objects belonging to a class type are normally initialized with default values, which
can be specified in the class definition.

䊐 Using CPU Registers
To increase the speed of a program, commonly used auto variables can be stored in
CPU registers instead of on the stack. In this case, the register keyword is used to
declare the object.
A register is normally the size of an int variable. In other words, it only makes sense
to define register variables if the variable is not too large, as in the case of types such
as char, short, int or pointers. If you omit the type when defining a register variable,
an int is assumed.
However, the compiler can ignore the register keyword. The number of registers
available for register variables depends on your hardware, although two registers are normally available. If a program defines too many register variables in a code block, the
superfluous variables are placed in the auto storage class.

䊐 Sample Function
The function strToLong() illustrates an algorithm that converts a sequence of digits
to a binary number. This is useful if you need to perform calculations with a number contained in a string.
The algorithm using the string "37" and the long variable num:
Step 0:
Step 1:
Step 2:

num = 0;
1st character '3' → number 3 = ('3'-'0')
num = num * 10 + 3;
// = 3
2nd character '7' → number 7 = ('7'-'0')
num = num * 10 + 7;
// = 37

This pattern is followed for every number in a longer string.

206

■

CHAPTER 11

■

ST0RAGE CLASSES AND NAMESPACES

THE STORAGE CLASSES OF FUNCTIONS

䊐 Example of a Program Structure
Source file 1
extern bool getPassword(void);

// Prototype

int main()
{
// The function permission(),
// but not the function timediff()
// can be called here.
.
.
.
}

Source file 2
static long timediff(void);

// Prototype

bool getPassword(void)
// Definition
{
// timediff() can be called here.
.
.
.
}
static long timediff(void)
{
.
.
.
}

// Definition

THE STORAGE CLASSES OF FUNCTIONS

■

207

Only two storage classes are available for functions: extern and static. Functions
with block scope are invalid: you cannot define a function within another function.
The storage class of a function defines access to the function, as it does for an object.
External functions have program scope, whereas static functions have file scope.

䊐 External Functions
If the keyword static is not used when defining a function, the function must belong
to the extern storage class.
In a similar manner to external objects, external functions can be used at any position
in a program. If you need to call a function before defining it, or in another source file,
you will need to declare that function.

Example: extern bool getPassword(void); // Prototype
As previously seen, you can omit the extern keyword, since functions belong to the
extern storage class by default.

䊐 Static Functions
To define a static function, simply place the keyword static before the function
header.

Example: static long timediff()
{

. . .

}

Functions in the static storage class have “private” character: they have file scope,
just like external static objects. They can only be called in the source file that defines
them. The name of a static function will not collide with objects and functions of the
same name in other modules.
If you need to call a static function before defining it, you must first declare the
function in the source file.

Example: static long timediff( void );
The program structure opposite takes up the example with the functions
getPassword() and timediff() once more. The function timediff() is an auxiliary function and not designed to be called externally. The function is declared as
static for this reason.

208

■

CHAPTER 11

■

ST0RAGE CLASSES AND NAMESPACES

NAMESPACES

Defining namespaces
// namesp1.cpp
// Defines and tests namespaces.
// ---------------------------------------------------#include 
// Class string defined within
// namespace std
namespace MySpace
{
std::string mess = "Within namespace MySpace";
int count = 0;
// Definition: MySpace::count
double f( double);
// Prototype:
MySpace::f()
}
namespace YourSpace
{
std::string mess = "Within namespace YourSpace";
void f( )
// Definition of
{
// YourSpace::f()
mess += '!';
}
}
namespace MySpace
// Back in MySpace.
{
int g(void);
// Prototype of MySpace::g()
double f( double y)
// Definition of
{
// MySpace::f()
return y / 10.0;
}
}
int MySpace::g( )
// Separate definition
{
// of MySpace::g()
return ++count;
}
#include 
// cout, ... within namespace std
int main()
{
std::cout << "Testing namespaces!\n\n"
<< MySpace::mess << std::endl;
MySpace::g();
std::cout << "\nReturn value g(): " << MySpace::g()
<< "\nReturn value f(): " << MySpace::f(1.2)
<< "\n---------------------" << std::endl;
YourSpace::f();
std::cout << YourSpace::mess << std::endl;
return 0;
}

NAMESPACES

■

209

Using global names in large-scale software projects can lead to conflicts, especially when
multiple class libraries are in operation.
C++ provides for the use of namespaces in order to avoid naming conflicts with global
identifiers. Within a namespace, you can use identifiers without needing to check
whether they have been defined previously in an area outside of the namespace. Thus,
the global scope is subdivided into isolated parts.
A normal namespace is identified by a name preceded by the namespace keyword.
The elements that belong to the namespace are then declared within braces.

Example: namespace myLib
{
int count;
double calculate(double, int);
// . . .
}

This example defines the namespace myLib that contains the variable count and the
function calculate().
Elements belonging to a namespace can be referenced directly by name within the
namespace. If you need to reference an element from outside of the namespace, you must
additionally supply the namespace. To do so, place the scope resolution operator, ::,
before the element name.

Example: myLib::count = 7;

// Outside of myLib

This allows you to distinguish between identical names in different namespaces. You can
also use the scope resolution operator :: to reference global names, that is, names
declared outside of any namespaces. To do so, simply omit the name of the namespace.
This technique is useful when you need to access a global name that is hidden by an
identical name defined in the current namespace.

Example: ::demo();

// Not belonging to any namespace

Be aware of the following when using namespaces:
■
■

namespaces do not need to be defined contiguously. You can reopen and expand
a namespace you defined previously at any point in the program
namespaces can be nested, that is, you can define a namespace within another
namespace.

Global identifiers belonging to the C++ standard library automatically belong to the
standard namespace std.

210

■

CHAPTER 11

■

ST0RAGE CLASSES AND NAMESPACES

THE KEYWORD using

Sample program
// namesp2.cpp
// Demonstrates the use of using-declarations and
// using-directives.
// ---------------------------------------------------#include 
// Namespace std
void message()
// Global function ::message()
{
std::cout << "Within function ::message()\n";
}
namespace A
{
using namespace std; // Names of std are visible here
void message()
// Function A::message()
{
cout << "Within function A::message()\n";
}
}
namespace B
{
using std::cout;
//
void message(void);
//
}
void B::message(void)
//
{
cout << "Within function
}
int main()
{
using namespace std;
using B::message;

Declaring cout of std.
Function B::message()
Defining B::message()
B::message()\n";

// Names of namespace std
// Function name without
// braces!
cout << "Testing namespaces!\n";
cout << "\nCall of A::message()" << endl;
A::message();
cout << "\nCall of B::message()" << endl;
message();
// ::message() is hidden because
// of the using-declaration.
cout << "\nCall of::message()" << endl;
::message();
// Global function
return 0;

}

THE KEYWORD using

■

211

You can simplify access to the elements of a namespace by means of a using declaration or
using directive. In this case, you do not need to repeatedly quote the namespace. Just like
normal declarations, using declarations and using directives can occur at any part of
the program.

䊐 using Declarations
A using declaration makes an identifier from a namespace visible in the current scope.

Example: using myLib::calculate;

// Declaration

You can then call the function calculate() from the myLib namespace.
double erg = calculate( 3.7, 5);

This assumes that you have not previously used the name calculate in the same
scope.

䊐 using Directive
The using directive allows you to import all the identifiers in a namespace.

Example: using namespace myLib;
This statement allows you to reference the identifiers in the myLib namespace directly.
If myLib contains an additional namespace and a using directive, this namespace is
also imported.
If identical identifiers occur in the current namespace and an imported namespace,
the using directive does not automatically result in a conflict. However, referencing an
identifier can lead to ambiguities. In this case, you should use the scope resolution operator to resolve the situation.
C++ header files without file extensions are used to declare the global identifiers in
the standard namespace std. The using directive was used in previous examples to
import any required identifiers to the global scope:

Example: #include 
using namespace std;

When developing large-scale programs or libraries, it is useful to declare the elements of
any proprietary namespaces in header files. Normal source files are used to define these
elements.

■

CHAPTER 11

■

exe rc i se s

212

ST0RAGE CLASSES AND NAMESPACES

EXERCISES

Program listing for exercise 1
// scope.cpp
// Accessing objects with equal names
// --------------------------------------------------#include 
#include 
using namespace std;
int var = 0;
namespace Special {

int var = 100; }

int main()
{
int var = 10;
cout << setw(10) << var;
{
int var = 20;
cout << setw(10) << var << endl;
{
++var;
cout << setw(10) << var;
cout << setw(10) << ++ ::var;
cout << setw(10) << Special::var * 2
<< endl;
}
cout << setw(10) << var - ::var;
}
cout << setw(10) << var << endl;
return 0;
}

// 1.

// 2.

// 3.
// 4.
// 5.

// 6.
// 7.

EXERCISES

■

213

Exercise 1
In general, you should use different names for different objects. However, if you
define a name for an object within a code block and the name is also valid for
another object, you will reference only the new object within the code block.
The new declaration hides any object using the same name outside of the block.
When you leave the code block, the original object once more becomes visible.
The program on the opposite page uses identical variable names in different
blocks.What does the program output on screen?
Exercise 2
You are developing a large-scale program and intend to use two commercial
libraries, tool1 and tool2.The names of types, functions, macros, and so on are
declared in the header files tool1.h and tool2.h for users of these libraries.
Unfortunately, the libraries use the same global names in part. In order to use
both libraries, you will need to define namespaces.
Write the following program to simulate this situation:
■

Define an inline function called calculate() that returns the sum of two
numbers for the header file tool1.h.The function interface is as follows:
double calculate(double num1, double num2);

■

■

Define an inline function called calculate() that returns the product of
two numbers for a second header file tool2.h.This function has the
same interface as the function in tool1.h.
Then write a source file containing a main function that calls both functions with test values and outputs the results.
To resolve potential naming conflicts, define the namespaces TOOL1 and
TOOL2 that include the relevant header files.

214

■

CHAPTER 11

ST0RAGE CLASSES AND NAMESPACES

Program listing for exercise 3
// static.cpp
// Tests an internal static variable
// --------------------------------------------------#include 
#include 
using namespace std;
double x = 0.5,
fun(void);
int main()
{
while( x < 10.0 )
{
x += fun();
cout << "
Within main(): "
<< setw(5) << x << endl;
}
return 0;
}
double fun()
{
static double x = 0;
cout << "
Within fun():"
<< setw(5) << x++;
return x;
}

EXERCISES

■

215

Exercise 3
Test your knowledge of external and static variables by reference to the
program on the opposite page.What screen output does the program generate?
Exercise 4
a. The function getPassword(), which checks password input, was introduced previously as an example of the use of static variables. Modify the
source file Passw1.cpp, which contains the function getPassword(), by
adding the function changePassword().This function allows the user to
change his or her password. Save the modified source file as
Passw2.cpp.
b. A large-scale program with several users is used to perform bookings.
Only authorized users, that is, users that have access to the password, are
allowed to perform bookings.
In the initial stages of program development, you need to test the
functionality of the source file, Passw2.cpp.To do so, create a new
source file with a main function that contains only the following menu
items in its main loop:
B = Booking
E = End of program

When B is typed, the password is first checked. If the user enters the
correct password, he or she can change the password.The program does
not need to perform any real bookings.

✓

NOTE
The modified password is only available during runtime as it is not stored permanently.

■

CHAPTER 11

■

solutions

216

ST0RAGE CLASSES AND NAMESPACES

SOLUTIONS

Exercise 1
Screen output of the program
10
21
20

20
1
10

200

Exercise 2
//
//
//
//

---------------------------------------------------tool1.h
Defining first function calculate() inline.
----------------------------------------------------

#ifndef _TOOL1_H_
#define _TOOL1_H_
inline double calculate( double num1, double num2)
{
return num1 + num2;
}
#endif
//
//
//
//

// End of _TOOL1_H_

---------------------------------------------------tool2.h
Defining second function calculate() inline.
----------------------------------------------------

#ifndef _TOOL2_H_
#define _TOOL2_H_
inline double calculate( double num1, double num2)
{
return num1 * num2;
}
#endif

// End of _TOOL2_H_

SOLUTIONS

//
//
//
//

■

-------------------------------------------------------tool_1_2.cpp
Uses two "libraries" and tests name lookup conflicts.
--------------------------------------------------------

#include 
namespace TOOL1
{
#include "tool1.h"
}
namespace TOOL2
{
#include "tool2.h"
}
#include 
int main()
{
using namespace std;
double x = 0.5, y = 10.5, res = 0.0;
cout << "Calling function of Tool1!" << endl;
res = TOOL1::calculate( x, y);
cout << "Result: " << res
<< "\n---------------------------------" << endl;
cout << "Calling function of Tool2!" << endl;
res = TOOL2::calculate( x, y);
cout << "Result: " << res << endl;
return 0;
}

Exercise 3
Screen output of the program
In
In
In
In

fun():
fun():
fun():
fun():

0
1
2
3

In
In
In
In

main():
main():
main():
main():

1.5
3.5
6.5
10.5

217

218

■

CHAPTER 11

ST0RAGE CLASSES AND NAMESPACES

Exercise 4
// ----------------------------------------------------// Passw2.cpp
// Defines the functions getPassword(), timediff() and
// changePassword() to examine and change a password.
// ----------------------------------------------------#include 
#include 
#include 
#include 
using namespace std;
static long timediff(void);
// Prototype
static string secret = "guest";
// Password
static long maxcount = 3, maxtime = 60; // Limits
bool getPassword()
{
// As before.
// . . .
}

// Read and verify a password.

// Auxiliary function timediff() --> defining static
static long timediff()
// Returns the number of seconds
{
// since the last call.
// As before.
// . . .
}
bool changePassword()
{
string word1,word2;

// Changes password.
// Returns: true, if the
// password has been changed
// For input

// To read a new password
cout <<"\nEnter a new password (2 - 20 characters): ";
cin.sync();
// Discards former input
cin >> setw(20) >> word1;

SOLUTIONS

■

if( word1.size() > 1)
{
cout << "\nEnter the password once more: ";
cin >> setw(20) >> word2;
if( word1 == word2)
// Password confirmed?
{
// Yes!
secret = word1;
return true;
}
}
return false;
// No new password
}
// ----------------------------------------------------// Password.cpp
// Testing the functions getPassword() and
// changePassword().
//
// After entering the password correctly (max. three
// attempts within 60 seconds), the user can change it.
// ----------------------------------------------------#include 
#include 
#include 
#include 
using namespace std;
bool getPassword(void);
// Read a password.
bool changePassword(void);
// Change a password.
// Inline functions:
inline void cls() { cout << "\033[2J"; }
inline void go_on()
{
cout << "\n\nGo on with return! ";
cin.sync(); cin.clear();
// Only new input
while( cin.get() != '\n')
;
}
inline char getYesOrNo()
// Read character Y or N.
{
char c = 0;
cin.sync(); cin.clear();
// Just new input
do
{
cin.get(c);
c = toupper(c); // Permitting lower case letters also.
}
while( c != 'Y' && c != 'N');
return c;
}

219

220

■

CHAPTER 11

ST0RAGE CLASSES AND NAMESPACES

static string header =
"\n\n
**** Test password handling

****\n\n";

static string menu =
"\n\n
B = Booking "
"\n\n
E = End of program"
"\n\n Your choice:
";
int main()
{
char choice = 0;
while( choice != 'E')
{
cls();
cout << header << menu; // Header and Menu
cin.get(choice);
choice = toupper(choice);
cls();
cout << header << endl; // Header
switch( choice)
{
case 'B':
// Booking
if( !getPassword() )
{
cout << "Access denied!" << endl;
go_on();
}
else
{ cout << "Welcome!\n\n"
<< "Do you want to change the password? (y/n)";
if( getYesOrNo() == 'Y')
{
if( changePassword() )
cout << "Password changed!" << endl;
else
cout << "Password unchanged!" << endl;
go_on();
}
// Place statements for booking here.
}
break;
case 'E':
cls(); cout << "\n
Bye Bye!" << endl;
break;
}
} // End of while
return 0;
}

chapter

12

References and Pointers
This chapter describes how to define references and pointers and how
to use them as parameters and/or return values of functions. In this
context, passing by reference and read-only access to arguments are
introduced.

221

222

■

CHAPTER 12

■

REFERENCES AND POINTERS

DEFINING REFERENCES

Example
float x = 10.7,

&rx = x;
Object names:

The object in
memory

...
x, rx

10.7

...

Sample program
// Ref1.cpp
// Demonstrates the definition and use of references.
// --------------------------------------------------#include 
#include 
using namespace std;
float x = 10.7F;
int main()
{
float &rx = x;
// double &ref = x;

// Global

// Local reference to x
// Error: different type!

rx *= 2;
cout << "
x = " << x << endl
// x = 21.4
<< " rx = " << rx << endl;
// rx = 21.4
const float& cref = x;
// Read-only reference
cout << "cref = " << cref << endl;
// ok!
// ++cref;
// Error: read-only!
const string str = "I am a constant string!";
// str = "That doesn't work!"; // Error: str constant!
// string& text = str;
// Error: str constant!
const string& text = str;
// ok!
cout << text << endl;
// ok! Just reading.
return 0;
}

DEFINING REFERENCES

■

223

A reference is another name, or alias, for an object that already exists. Defining a reference does not occupy additional memory. Any operations defined for the reference are
performed with the object to which it refers. References are particularly useful as parameters and return values of functions.

䊐 Definition
The ampersand character, &, is used to define a reference. Given that T is a type, T&
denotes a reference to T.

Example: float x = 10.7;
float& rx = x;

// or:

float &rx = x;

rx is thus a different way of expressing the variable x and belongs to the type “reference
to float”. Operations with rx, such as

Example: --rx;

// equivalent to

--x;

will automatically affect the variable x. The & character, which indicates a reference,
only occurs in declarations and is not related to the address operator &! The address
operator returns the address of an object. If you apply this operator to a reference, it
returns the address of the referenced object.

Example: &rx

// Address of x, thus is equal to &x

A reference must be initialized when it is declared, and cannot be modified subsequently. In other words, you cannot use the reference to address a different variable at a
later stage.

䊐 Read-Only References
A reference that addresses a constant object must be a constant itself, that is, it must be
defined using the const keyword to avoid modifying the object by reference. However,
it is conversely possible to use a reference to a constant to address a non-constant object.

Example: int a;

const int& cref = a;

// ok!

The reference cref can be used for read-only access to the variable a, and is said to be a
read-only identifier.
A read-only identifier can be initialized by a constant, in contrast to a normal reference:

Example: const double& pi = 3.1415927;
Since the constant does not take up any memory space, the compiler creates a
temporary object which is then referenced.

224

■

CHAPTER 12

■

REFERENCES AND POINTERS

REFERENCES AS PARAMETERS

Sample program
//
//
//
//

Ref2.cpp
Demonstrating functions with parameters
of reference type.
--------------------------------------------------

#include 
#include 
using namespace std;
// Prototypes:
bool getClient( string& name, long& nr);
void putClient( const string& name, const long& nr);
int main()
{
string clientName;
long
clientNr;
cout << "\nTo input and output client data \n"
<< endl;
if( getClient( clientName, clientNr))
// Calls
putClient( clientName, clientNr);
else
cout << "Invalid input!" << endl;
return 0;
}
bool getClient( string& name, long& nr)
// Definition
{
cout << "\nTo input client data!\n"
<< " Name:
";
if( !getline( cin, name)) return false;
cout << " Number: ";
if( !( cin >> nr)) return false;
return true;
}
// Definition
void putClient( const string& name, const long& nr)
{
// name and nr can only be read!
cout << "\n-------- Client Data ---------\n"
<< "\n Name:
"; cout << name
<< "\n Number: "; cout << nr << endl;
}

REFERENCES AS PARAMETERS

■

225

䊐 Passing by Reference
A pass by reference can be programmed using references or pointers as function parameters. It is syntactically simpler to use references, although not always permissible.
A parameter of a reference type is an alias for an argument. When a function is called,
a reference parameter is initialized with the object supplied as an argument. The function
can thus directly manipulate the argument passed to it.

Example: void test( int& a) { ++a; }
Based on this definition, the statement
test( var);

// For an int variable var

increments the variable var. Within the function, any access to the reference a automatically accesses the supplied variable, var.
If an object is passed as an argument when passing by reference, the object is not
copied. Instead, the address of the object is passed to the function internally, allowing
the function to access the object with which it was called.

䊐 Comparison to Passing by Value
In contrast to a normal pass by value an expression, such as a+b, cannot be used as an
argument. The argument must have an address in memory and be of the correct type.
Using references as parameters offers the following benefits:
■

■

arguments are not copied. In contrast to passing by value, the run time of a program should improve, especially if the arguments occupy large amounts of memory
a function can use the reference parameter to return multiple values to the calling
function. Passing by value allows only one result as a return value, unless you
resort to using global variables.

If you need to read arguments, but not copy them, you can define a read-only reference
as a parameter.

Example: void display( const string& str);
The function display() contains a string as an argument. However, it does not generate a new string to which the argument string is copied. Instead, str is simply a reference to the argument. The caller can rest assured that the argument is not modified
within the function, as str is declared as a const.

226

■

CHAPTER 12

■

REFERENCES AND POINTERS

REFERENCES AS RETURN VALUE

Sample program
//
//
//
//

Ref3.cpp
Demonstrates the use of return values with
reference type.
--------------------------------------------------

#include 
#include 
using namespace std;
// Returns a
// reference to
// the minimum.

double& refMin( double&, double&);
int main()
{
double x1 = 1.1,

x2 = x1 + 0.5,

y;

y = refMin( x1, x2);
// Assigns the minimum to y.
cout << "x1 = " << x1 << "
"
<< "x2 = " << x2 << endl;
cout << "Minimum: " << y << endl;
++refMin( x1,
cout << "x1 =
<< "x2 =
++refMin( x1,
cout << "x1
<< "x2
refMin( x1,
cout << "x1
<< "x2
refMin( x1,
cout << "x1
<< "x2
return 0;

x2);
// ++x1, as x1 is minimal
" << x1 << "
"
// x1 = 2.1
" << x2 << endl;
// x2 = 1.6
x2);
// ++x2, because x2 is
// the minimum.
= " << x1 << "
"
// x1 = 2.1
= " << x2 << endl;
// x2 = 2.6
x2) = 10.1;
// x1 = 10.1, because
// x1 is the minimum.
= " << x1 << "
"
// x1 = 10.1
= " << x2 << endl;
// x2 = 2.6
x2) += 5.0;
// x2 += 5.0, because
// x2 is the minimum.
= " << x1 << "
"
// x1 = 10.1
= " << x2 << endl;
// x2 = 7.6

}
double& refMin( double& a, double& b)
{
return a <= b ? a : b;
}

✓

// Returns a
// reference to
// the minimum.

NOTE
The expression refMin(x1,x2) represents either the object x1 or the object x2, that is, the object
containing the smaller value.

REFERENCES AS RETURN VALUE

■

227

䊐 Returning References
The return type of a function can also be a reference type. The function call then represents an object, and can be used just like an object.

Example: string& message()

// Reference!

{
static string str = "Today only cold cuts!";
return str;
}

This function returns a reference to a static string, str. Pay attention to the following
point when returning references and pointers:
The object referenced by the return value must exist after leaving the function.
It would be a critical error to declare the string str as a normal auto variable in the
function message(). This would destroy the string on leaving the function and the reference would point to an object that no longer existed.

䊐 Calling a Reference Type Function
The function message() (mentioned earlier in this section) is of type “reference to
string.” Thus, calling
message()

represents a string type object, and the following statements are valid:
message() = "Let's go to the beer garden!";
message() += " Cheers!";
cout << "Length: " << message().length();

In these examples, a new value is first assigned to the object referenced by the function
call. Then a new string is appended before the length of the referenced string is output in
the third statement.
If you want to avoid modifying the referenced object, you can define the function type
as a read-only reference.

Example: const string& message();

// Read-only!

References are commonly used as return types when overloading operators. The operations that an operator has to perform for a user-defined type are always implemented by
an appropriate function. Refer to the chapters on overloading operators later in this book
for more details. However, examples with operators from standard classes can be provided at this point.

228

■

CHAPTER 12

■

REFERENCES AND POINTERS

EXPRESSIONS WITH REFERENCE TYPE

Example: Operator << of class ostream
cout << "Good morning" << '!';

cout << "Good morning"

Reference to cout

<<'!';

Sample assignments of class string
// Ref4.cpp
// Expressions with reference type exemplified by
// string assignments.
// -------------------------------------------------#include 
#include 
#include 
// For toupper()
using namespace std;
void strToUpper( string& );
// Prototype
int main()
{
string text("Test with assignments \n");
strToUpper(text);
cout << text << endl;
strToUpper( text = "Flowers");
cout << text << endl;
strToUpper( text += " cheer you up!\n");
cout << text << endl;
return 0;
}
void strToUpper( string& str) // Converts the content
{
// of str to uppercase.
int len = str.length();
for( int i=0; i < len; ++i)
str[i] = toupper( str[i]);
}

EXPRESSIONS WITH REFERENCE TYPE

■

229

Every C++ expression belongs to a certain type and also has a value, if the type is not
void. Reference types are also valid for expressions.

䊐 The Stream Class Shift Operators
The << and >> operators used for stream input and output are examples of expressions
that return a reference to an object.

Example: cout << " Good morning "
This expression is not a void type but a reference to the object cout, that is, it represents the object cout. This allows you to repeatedly use the << on the expression:
cout << "Good morning" << '!'

The expression is then equivalent to
(cout << " Good morning ") << '!'

Expressions using the << operator are composed from left to right, as you can see from
the table of precedence contained in the appendix.
Similarly, the expression cin >> variable represents the stream cin. This allows
repeated use of the >> operator.

Example: int a;

double x;
cin >> a >> x;

// (cin >> a) >> x;

䊐 Other Reference Type Operators
Other commonly used reference type operators include the simple assignment operator =
and compound assignments, such as += and *=. These operators return a reference to the
operand on the left. In an expression such as
a = b or a += b
a must therefore be an object. In turn, the expression itself represents the object a. This
also applies when the operators refer to objects belonging to class types. However, the
class definition stipulates the available operators. For example, the assignment operators
= and += are available in the standard class string.

Example: string name("Jonny ");
name += "Depp";

//Reference to name

Since an expression of this type represents an object, the expression can be passed as
an argument to a function that is called by reference. This point is illustrated by the
example on the opposite page.

230

■

CHAPTER 12

■

REFERENCES AND POINTERS

DEFINING POINTERS

Sample program
// pointer1.cpp
// Prints the values and addresses of variables.
// -------------------------------------------------#include 
using namespace std;
int var, *ptr;

// Definition of variables var and ptr

int main()
{
var = 100;
ptr = &var;

// Outputs the values and addresses
// of the variables var and ptr.

cout << " Value of var:
" << var
<< "
Address of var: " << &var
<< endl;
cout << " Value of ptr: "
<< ptr
<< "
Address of ptr: " << &ptr
<< endl;
return 0;
}

Sample screen output
Value of var:
100
Value of ptr: 00456FD4

Address of var: 00456FD4
Address of ptr: 00456FD0

The variables var and ptr in memory

Variable

Value of the Variable

Address
(hexadecimal)

...
var

100

456FD4

ptr

456FD4

456FD0

...

DEFINING POINTERS

■

231

Efficient program logic often requires access to the memory addresses used by a program’s
data, rather than manipulation of the data itself. Linked lists or trees whose elements are
generated dynamically at runtime are typical examples.

䊐 Pointers
A pointer is an expression that represents both the address and type of another object.
Using the address operator, &, for a given object creates a pointer to that object. Given
that var is an int variable,

Example: &var

// Address of the object var

is the address of the int object in memory and thus a pointer to var. A pointer points
to a memory address and simultaneously indicates by its type how the memory address
can be read or written to. Thus, depending on the type, we refer to pointers to char, pointers to int, and so on, or use an abbreviation, such as char pointer, int pointer, and so on.

䊐 Pointer Variables
An expression such as &var is a constant pointer; however, C++ allows you to define
pointer variables, that is, variables that can store the address of another object.

Example: int *ptr;

// or:

int* ptr;

This statement defines the variable ptr, which is an int* type (in other words, a pointer
to int). ptr can thus store the address of an int variable. In a declaration, the star character * always means “pointer to.”
Pointer types are derived types. The general form is T*, where T can be any given type.
In the above example T is an int type.
Objects of the same base type T can be declared together.

Example: int a, *p, &r = a;

// Definition of a, p, r

After declaring a pointer variable, you must point the pointer at a memory address. The
program on the opposite page does this using the statement
ptr = &var;.

䊐 References and Pointers
References are similar to pointers: both refer to an object in memory. However, a pointer
is not merely an alias but an individual object that has an identity separate from the
object it references. A pointer has its own memory address and can be manipulated by
pointing it at a new memory address and thus referencing a different object.

232

■

CHAPTER 12

REFERENCES AND POINTERS

■

THE INDIRECTION OPERATOR

Using the indirection operator
double x, y, *px;
px
*px
*px
y

= &x;
= 12.3;
+= 4.5;
= sin(*px);

//
//
//
//

Let px point to x.
Assign the value 12.3 to x
Increment x by 4.5.
To assign sine of x to y.

Address and value of the variables x and px

px

x

Address
of px

Address of x
= value of px

Value of x

&px

&x
px

x
*px

Notes on addresses in a program
■

■

■

Each pointer variable occupies the same amount of space, independent of the
type of object it references. That is, it occupies as much space as is necessary to
store an address. On a 32-bit computer, such as a PC, this is four bytes.
The addresses visible in a program are normally logic addresses that are allocated
and mapped to physical addresses by the system. This allows for efficient storage
management and the swapping of currently unused memory blocks to the hard
disk.
C++ guarantees that any valid address will not be equal to 0. Thus, the special
value 0 is used to indicate an error. For pointers, the symbolic constant NULL is
defined as 0 in standard header files. A pointer containing the value NULL is
also called NULL pointer.

THE INDIRECTION OPERATOR

■

233

䊐 Using Pointers to Access Objects
The indirection operator * is used to access an object referenced by a pointer:
Given a pointer, ptr, *ptr is the object referenced by ptr.
As a programmer, you must always distinguish between the pointer ptr and the
addressed object *ptr.

Example: long

a = 10, b,
*ptr;
ptr = &a;
b = *ptr;

// Definition of a, b
// and pointer ptr.
// Let ptr point to a.

This assigns the value of a to b, since ptr points to a. The assignment b = a; would
return the same result. The expression *ptr represents the object a, and can be used
wherever a could be used.
The star character * used for defining pointer variables is not an operator but merely
imitates the later use of the pointer in expressions. Thus, the definition
long *ptr;

has the following meaning: ptr is a long* (pointer to long) type and *ptr is a long
type.
The indirection operator * has high precedence, just like the address operator &. Both
operators are unary, that is, they have only one operand. This also helps distinguish the
redirection operator from the binary multiplication operator *, which always takes two
operands.

䊐 L-values
An expression that identifies an object in memory is known as an L-value in C++. The
term L-value occurs commonly in compiler error messages and is derived from the assignment. The left operand of the = operator must always designate a memory address.
Expressions other than an L-value are often referred to as R-values.
A variable name is the simplest example of an L-value. However, a constant or an
expression, such as x + 1, is an R-value. The indirection operator is one example of an
operator that yields L-values. Given a pointer variable p, both p and *p are L-values, as
*p designates the object to which p points.

234

■

CHAPTER 12

■

REFERENCES AND POINTERS

POINTERS AS PARAMETERS

Sample function
// pointer2.cpp
// Definition and call of function swap().
// Demonstrates the use of pointers as parameters.
// ---------------------------------------------------#include 
using namespace std;
void swap( float *, float *);

// Prototype of swap()

int main()
{
float x = 11.1F;
float y = 22.2F;
.
.
.
swap( &x, &y );
.
.
// p2 = &y
.
}
// p1 = &x
void swap( float *p1, float *p2)
{
float temp;
// Temporary variable
temp = *p1;
*p1 = *p2;
*p2 = temp;
}

// At the above call p1 points
// to x and p2 to y.

POINTERS AS PARAMETERS

■

235

䊐 Objects as Arguments
If an object is passed as an argument to a function, two possible situations occur:
■
■

the parameter in question is the same type as the object passed to it. The function that is called is then passed a copy of the object (passing by value)
the parameter in question is a reference. The parameter is then an alias for the
argument, that is, the function that is called manipulates the object passed by the
calling function (passing by reference).

In the first case, the argument passed to the function cannot be manipulated by the
function. This is not true for passing by reference. However, there is a third way of passing by reference—passing pointers to the function.

䊐 Pointers as Arguments
How do you declare a function parameter to allow an address to be passed to the function
as an argument? The answer is quite simple: The parameter must be declared as a pointer
variable.
If, for example, the function func() requires the address of an int value as an argument, you can use the following statement

Example: long func( int *iPtr )
{
// Function block
}

to declare the parameter iPtr as an int pointer. If a function knows the address of an
object, it can of course use the indirection operator to access and manipulate the object.
In the program on the opposite page, the function swap() swaps the values of the
variables x and y in the calling function. The function swap() is able to access the variables since the addresses of these variables, that is &x and &y, are passed to it as arguments.
The parameters p1 and p2 in swap() are thus declared as float pointers. The
statement
swap( &x, &y);

initializes the pointers p1 and p2 with the addresses of x or y. When the function
manipulates the expressions *p1 and *p2, it really accesses the variables x and y in the
calling function and exchanges their values.

■

CHAPTER 12

■

exercise s

236

REFERENCES AND POINTERS

EXERCISES

Listing for exercise 3
// A version of swap() with incorrect logic.
// Find the error!
void swap(float *p1, float *p2)
{
float *temp;

// Temporary variable

temp = p1;
p1
= p2;
p2
= temp;
}

Solutions of quadratic equations
The quadratic equation: a*x2 + b*x + c = 0 has real solutions:
x12 = (-b ± √(b2 - 4ac)) / 2a

if the discriminant satisfies: b2 -4ac >= 0
If the value of (b2 - 4ac) is negative, no real solution exists.
Test values
Quadratic Equation

Solutions

2x2

x1 = 1.5,
X1 = 3.0,

- 2x - 1.5 = 0
x2 - 6x + 9 = 0
2x2 + 2 = 0

none

x2 = -0.5
x2 = 3.0

EXERCISES

■

237

Exercise 1
What happens if the parameter in the sample function strToUpper() is
declared as a string& instead of a string?
Exercise 2
Write a void type function called circle()to calculate the circumference and
area of a circle.The radius and two variables are passed to the function, which
therefore has three parameters:
Parameters:

✓

A read-only reference to double for the radius and two
references to double that the function uses to store the area
and circumference of the circle.

NOTE
Given a circle with radius r:
Area = π * r * r and circumference = 2 * π * r where π = 3.1415926536

Test the function circle() by outputting a table containing the radius, the
circumference, and the area for the radii 0.5, 1.0, 1.5, . . . , 10.0.

Exercise 3
a. The version of the function swap() opposite can be compiled without
producing any error messages. However, the function will not swap the
values of x and y when swap(&x,&y); is called.What is wrong?
b. Test the correct pointer version of the function swap() found in this
chapter.Then write and test a version of the function swap() that uses
references instead of pointers.
Exercise 4
Create a function quadEquation() that calculates the solutions to quadratic
equations.The formula for calculating quadratic equations is shown opposite.
Arguments:
Returns:

The coefficients a, b, c and two pointers to both solutions.
false, if no real solution is available, otherwise true.

Test the function by outputting the quadratic equations on the opposite page
and their solutions.

■

CHAPTER 12

■

solutions

238

REFERENCES AND POINTERS

SOLUTIONS

Exercise 1
The call to function strToUpper() is left unchanged. But instead of passing by
reference, a passing by value occurs, i.e., the function manipulates a local copy.
Thus, only a local copy of the string is changed in the function, but the string in
the calling function remains unchanged.
Exercise 2
// ---------------------------------------------------// circle.cpp
// Defines and calls the function circle().
// ---------------------------------------------------#include 
#include 
#include 
using namespace std;
// Prototype of circle():
void circle( const double& rad, double& um, double& fl);
const double startRadius = 0.5,
endRadius
= 10.0,
step
= 0.5;
string header = "\n
line( 50, '-');

// Start, end and
// step width of
// the table

***** Computing Circles ***** \n",

int main()
{
double rad, circuit, plane;
cout << header << endl;
cout << setw(10) << "Radius"
<< setw(20) << "Circumference"
<< setw(20) << "Area\n" << line << endl;
cout << fixed;
// Floating point presentation
for( rad = startRadius;
rad < endRadius + step/2; rad += step)
{
circle( rad, circuit, plane);
cout << setprecision(1)<< setw(8) << rad
<< setprecision(5)<< setw(22) << circuit
<< setw(20) << plane <
using namespace std;
void swap( float*, float*);
void swap( float&, float&);

// Prototypes of swap()

int main()
{
float x = 11.1F;
float y = 22.2F;
cout << "x and y before swapping:
<< x << "
" << y << endl;
swap( &x, &y);

"

// Call pointer version.

cout << "x and y after 1. swapping: "
<< x << "
" << y << endl;
swap( x, y);

// Call reference version.

cout << "x and y after 2. swapping: "
<< x << "
" << y << endl;
return 0;
}
void swap(float *p1, float *p2)
{
float temp;
temp = *p1;
*p1 = *p2;
*p2 = temp;
}

// Pointer version
// Temporary variable
// Above call points p1
// to x and p2 to y.

239

240

■

CHAPTER 12

REFERENCES AND POINTERS

void swap(float& a, float& b)
{
float temp;
temp = a;
a
= b;
b
= temp;

// Reference version
// Temporary variable

// For above call
// a equals x and b equals y

}

Exercise 4
//
//
//
//
//
//
//
//

---------------------------------------------------quadEqu.cpp
Defines and calls the function quadEquation(),
which computes the solutions of quadratic equations
a*x*x + b*x + c = 0
The equation and its solutions are printed by
the function printQuadEquation().
----------------------------------------------------

#include 
#include 
#include 
#include 
using namespace std;

// For the square root sqrt()

string header =
" *** Solutions of Quadratic Equations ***\n",
line( 50, '-');
// ----- Prototypes ----// Computing solutions:
bool quadEquation( double a, double b, double c,
double* x1Ptr, double* x2Ptr);
// Printing the equation and its solutions:
void printQuadEquation( double a, double b, double c);
int main()
{
cout << header << endl;
printQuadEquation( 2.0, -2.0, -1.5);
printQuadEquation( 1.0, -6.0, 9.0);
printQuadEquation( 2.0, 0.0, 2.0);
return 0;
}

SOLUTIONS

■

// Prints the equation and its solutions:
void printQuadEquation( double a, double b, double c)
{
double x1 = 0.0, x2 = 0.0;
// For solutions
cout <<
<<
<<
<<

line << '\n'
"\nThe quadratic equation:\n\t "
a << "*x*x + " << b << "*x + " <<
endl;

c << " = 0"

if( quadEquation( a, b, c, &x1, &x2) )
{
cout << "has real solutions:"
<< "\n\t x1 = " << x1
<< "\n\t x2 = " << x2 << endl;
}
else
cout << "has no real solutions!" << endl;
cout << "\nGo on with return. \n\n";
cin.get();
}
bool quadEquation( double a, double b, double c,
double* x1Ptr, double* x2Ptr)
// Computes the solutions of the quadratic equation:
//
a*x*x + b*x + c = 0
// Stores the solutions in the variables to which
// x1Ptr and x2Ptr point.
// Returns: true, if a solution exists,
//
otherwise false.
{
bool return_flag = false;
double help = b*b - 4*a*c;
if( help >= 0)
{
help = sqrt( help);

// There are real solutions.

*x1Ptr = (-b + help) / (2*a);
*x2Ptr = (-b - help) / (2*a);
return_flag = true;
}
return return_flag;
}

241

This page intentionally left blank

chapter

13

Defining Classes
This chapter describes how classes are defined and how instances of
classes, that is, objects, are used. In addition, structs and unions are
introduced as examples of special classes.

243

244

■

CHAPTER 13

■

DEFINING CLASSES

THE CLASS CONCEPT

Real World
A Car

Abstraction

Class CAR

Properties (Data Members):
Date when built
Capacity (PS)
Serial number
...
Methods (Member functions):
to run, to brake,
to park, to turn off
...

Instantiation

Objects
car1

car2

Properties:
Date when built = 1990
Capacity = 100
Chassis number = 11111
...

Properties:
Date when built = 2000
Capacity = 200
Chassis number = 22222
...

Methods
...

Methods
...

•••

THE CLASS CONCEPT

■

245

Classes are the language element in C++ most important to the support object-oriented
programming (OOP). A class defines the properties and capacities of an object.

䊐 Data Abstraction
Humans use abstraction in order to manage complex situations. Objects and processes are
reduced to basics and referred to in generic terms. Classes allow more direct use of the
results of this type of abstraction in software development.
The first step towards solving a problem is analysis. In object-oriented programming,
analysis comprises identifying and describing objects and recognizing their mutual relationships. Object descriptions are the building blocks of classes.
In C++, a class is a user-defined type. It contains data members, which describe the
properties of the class, and member functions, or methods, which describe the capacities of
the objects. Classes are simply patterns used to instantiate, or create, objects of the class
type. In other words, an object is a variable of a given class.

䊐 Data Encapsulation
When you define a class, you also specify the private members, that is, the members that
are not available for external access, and the public members of that class. An application program accesses objects by using the public methods of the class and thus activating its capacities.
Access to object data is rarely direct, that is, object data is normally declared as private and then read or modified by methods with public declarations to ensure correct
access to the data.
One important aspect of this technique is the fact that application programs need not
be aware of the internal structure of the data. If needed, the internal structure of the program data can even be modified. Provided that the interfaces of the public methods
remain unchanged, changes like these will not affect the application program. This
allows you to enhance an application by programming an improved class version without
changing a single byte of the application.
An object is thus seen to encapsulate its private structure, protecting itself from external influences and managing itself by its own methods. This describes the concept of data
encapsulation concisely.

246

■

CHAPTER 13

■

DEFINING CLASSES

DEFINING CLASSES

Definition scheme
class Demo
{
private:
// Private data members and methods here
public:
// Public data members and methods here
};

Example of a class
// account.h
// Defining the class Account.
// --------------------------------------------------#ifndef _ACCOUNT_
// Avoid multiple inclusions.
#define _ACCOUNT_
#include 
#include 
using namespace std;
class Account
{
private:
string name;
unsigned long nr;
double balance;

// Sheltered members:
// Account holder
// Account number
// Account balance

public:
//Public interface:
bool init( const string&, unsigned long, double);
void display();
};
#endif

//

_ACCOUNT_

DEFINING CLASSES

■

247

A class definition specifies the name of the class and the names and types of the class
members.
The definition begins with the keyword class followed by the class name. The data
members and methods are then declared in the subsequent code block. Data members
and member functions can belong to any valid type, even to another previously defined
class. At the same time, the class members are divided into:
■
■

private members, which cannot be accessed externally
public members, which are available for external access.

The public members form the so-called public interface of the class.
The opposite page shows a schematic definition of a class. The private section generally contains data members and the public section contains the access methods for
the data. This provides for data encapsulation.
The following example includes a class named Account used to represent a bank
account. The data members, such as the name of the account holder, the account number, and the account balance, are declared as private. In addition, there are two public
methods, init() for initialization purposes and display(), which is used to display
the data on screen.
The labels private: and public: can be used at the programmer’s discretion
within a class:
■

■

you can use the labels as often as needed, or not at all, and in any order. A section marked as private: or public: is valid until the next public: or private: label occurs
the default value for member access is private. If you omit both the private
and public labels, all the class members are assumed to be private.

䊐 Naming
Every piece of software uses a set of naming rules. These rules often reflect the target
platform and the class libraries used. For the purposes of this book, we decided to keep to
standard naming conventions for distinguishing classes and class members. Class names
begin with an uppercase letter and member names with a lowercase letter.
Members of different classes can share the same name. A member of another class
could therefore also be named display().

248

■

CHAPTER 13

■

DEFINING CLASSES

DEFINING METHODS

Methods of class Account
// account.cpp
// Defines methods init() and display().
// --------------------------------------------------#include "account.h"
// Class definition
#include 
#include 
using namespace std;
// The method init() copies the given arguments
// into the private members of the class.
bool Account::init(const string& i_name,
unsigned long i_nr,
double
i_balance)
{
if( i_name.size() < 1)
// No empty name
return false;
name
= i_name;
nr
= i_nr;
balance = i_balance;
return true;
}
// The method display() outputs private data.
void Account::display()
{
cout << fixed << setprecision(2)
<< "--------------------------------------\n"
<< "Account holder:
" << name << '\n'
<< "Account number:
" << nr
<< '\n'
<< "Account balance:
" << balance << '\n'
<< "--------------------------------------\n"
<< endl;
}

DEFINING METHODS

■

249

A class definition is not complete without method definitions. Only then can the objects
of the class be used.

䊐 Syntax
When you define a method, you must also supply the class name, separating it from the
function name by means of the scope resolution operator ::.

Syntax:

type class_name::function_name(parameter_list)
{
. . .
}

Failure to supply the class name results in a global function definition.
Within a method, all the members of a class can be designated directly using their
names. The class membership is automatically assumed. In particular, methods belonging
to the same class can call each other directly.
Access to private members is only possible within methods belonging to the same
class. Thus, private members are completely controlled by the class.
Defining a class does not automatically allocate memory for the data members of that
class. To allocate memory, you must define an object. When a method is called for a
given object, the method can then manipulate the data of this object.

䊐 Modular Programming
A class is normally defined in several source files. In this case, you will need to place the
class definition in a header file. If you place the definition of the class Account in the
file Account.h, any source file including the header file can use the class Account.
Methods must always be defined within a source file. This would mean defining the
methods for the class Account in a source file named Account.cpp, for example.
The source code of the application program, for example, the code containing the
function main, is independent of the class and can be stored in separate source files. Separating classes from application programs facilitates re-use of classes.
In an integrated development environment, a programmer will define a project to help
manage the various program modules by inserting all the source files into the project.
When the project is compiled and linked, modified source files are automatically re-compiled and linked to the application program.

250

■

CHAPTER 13

■

DEFINING CLASSES

DEFINING OBJECTS

The objects current and savings in memory

savings
name

"Cheers, Mary"

nr

1234567

balance

2002.22

current
name

"Dylan, Bob"

nr

87654321

balance

–1300.13

DEFINING OBJECTS

■

251

Defining a class also defines a new type for which variables, that is, objects, can be
defined. An object is also referred to as an instance of a class.

䊐 Defining Objects
An object is defined in the usual way by supplying the type and the object name.

Syntax:

class_name object_name1 [, object_name2,...]

The following statement defines an object current of type Account:

Example: Account current;

// or: class Account ...

Memory is now allocated for the data members of the current object. The current
object itself contains the members name, nr, and balance.

䊐 Objects in Memory
If multiple objects of the same class type are declared, as in

Example: Account current, savings;
each object has its own data members. Even the object savings contains the members
name, nr, and balance. However, these data members occupy a different position in
memory than the data members belonging to current.
The same methods are called for both objects. Only one instance of the machine code
for a method exists in memory—this applies even if no objects have been defined for the
class.
A method is always called for a particular instance and then manipulates the data
members of this object. This results in the memory content as shown on the opposite
page, when the method init() is called for each object with the values shown.

䊐 Initializing Objects
The objects belonging to the Account class were originally defined but not initialized.
Each member object is thus defined but not explicitly initialized. The string name, is
empty, as it is thus defined in the class string. The initial values of the members nr
and balance are unknown, however. As is the case for other variables, these data members will default to 0 if the object is declared global or static.
You can define exactly how an object is created and destroyed. These tasks are performed by constructors and destructors. Constructors are specifically responsible for initializing objects—more details are given later.

252

■

CHAPTER 13

■

DEFINING CLASSES

USING OBJECTS

Sample program
// account_t.cpp
// Uses objects of class Account.
// --------------------------------------------------#include "Account.h"
int main()
{
Account current1, current2;
current1.init("Cheers, Mary", 1234567, -1200.99);
current1.display();
//

current1.balance += 100; // Error: private member
current2 = current1;
current2.display();

// ok: Assignment of
// objects is possible.
// ok

// New values for current2
current2.init("Jones, Tom", 3512347, 199.40);
current2.display();
Account& mtr = current1;
mtr.display();
return 0;
}

//
//
//
//
//

To use a reference:
mtr is an alias name
for object current1.
mtr can be used just
as object current1.

USING OBJECTS

■

253

䊐 Class Member Access Operator
An application program that manipulates the objects of a class can access only the public members of those objects. To do so, it uses the class member access operator (in short:
dot operator).

Syntax:

object.member

Where member is a data member or a method.

Example: Account current;
current.init("Jones, Tom",1234567,-1200.99);

The expression current.init represents the public method init of the Account
class. This method is called with three arguments for current.
The init() call cannot be replaced by direct assignments.

Example: current.name

= "Dylan, Bob";
current.nr
= 1234567;
current.balance = -1200.99;

// Error:
// private
// members

Access to the private members of an object is not permissible outside the class. It is
therefore impossible to display single members of the Account class on screen.

Example: cout << current.balance;
current.display();

// Error
// ok

The method display() displays all the data members of current. A method such as
display() can only be called for one object. The statement
display();

would result in an error message, since there is no global function called display().
What data would the function have to display?

䊐 Assigning Objects
The assignment operator = is the only operator that is defined for all classes by default.
However, the source and target objects must both belong to the same class. The assignment is performed to assign the individual data members of the source object to the corresponding members of the target object.

Example: Account current1, current2;
current2.init("Marley, Bob",350123, 1000.0);
current1 = current2;

This copies the data members of current2 to the corresponding members of
current1.

254

■

CHAPTER 13

■

DEFINING CLASSES

POINTERS TO OBJECTS

Sample program
// ptrObj.cpp
// Uses pointers to objects of class Account.
// --------------------------------------------------#include "Account.h"
// Includes , 
bool getAccount( Account *pAccount);
// Prototype
int main()
{
Account current1, current2, *ptr = ¤t1;
ptr->init("Cheer, Mary",
3512345, 99.40);
ptr->display();
ptr = ¤t2;
//
if( getAccount( ptr))
//
ptr->display();
//
else
cout << "Invalid input!"
return 0;

// current1.init(...)
// current1.display()
Let ptr point to current2
Input and output a new
account.
<< endl;

}
// -------------------------------------------------// getAccount() reads data for a new account
// and adds it into the argument.
bool getAccount( Account *pAccount )
{
string name, line(50,'-');
// Local variables
unsigned long nr;
double startcapital;
cout << line << '\n'
<< "Enter data for a new account: \n"
<< "Account holder: ";
if( !getline(cin,name) || name.size() == 0)
return false;
cout << "Account number: ";
if( !(cin >> nr))
return false;
cout << "Starting capital: ";
if( !(cin >> startcapital)) return false;
// All input ok
pAccount->init( name, nr, startcapital);
return true;
}

POINTERS TO OBJECTS

■

255

An object of a class has a memory address—just like any other object. You can assign this
address to a suitable pointer.

Example: Account savings("Mac, Rita",654321, 123.5);
Account *ptrAccount = &savings;

This defines the object savings and a pointer variable called ptrAccount. The
pointer ptrAccount is initialized so that it points to the object savings. This makes
*ptrAccount the object savings itself. You can then use the statement

Example: (*ptrAccount).display();
to call the method display() for the object savings. Parentheses must be used in
this case, as the operator . has higher precedence than the * operator.

䊐 Arrow Operator
You can use the class member access operator -> (in short: arrow operator) instead of a
combination of * and . .

Syntax:

objectPointer->member

This expression is equivalent to
(*objectPointer).member

The operator -> is made up of a minus sign and the greater than sign.

Example: ptrAccount->display();
This statement calls the method display() for the object referenced by ptrAccount,
that is, for the object savings. The statement is equivalent to the statement in the previous example.
The difference between the class member access operators . and -> is that the left
operand of the dot operator must be an object, whereas the left operand of the arrow
operator must be a pointer to an object.

䊐 The Sample Program
Pointers to objects are often used as function parameters. A function that gets the
address of an object as an argument can manipulate the referenced object directly. The
example on the opposite page illustrates this point. It uses the function getAccount()
to read the data for a new account. When called, the address of the account is passed:
getAccount(ptr) // or: getAccount(¤t1)

The function can then use the pointer ptr and the init() method to write new data
to the referenced object.

256

■

CHAPTER 13

■

DEFINING CLASSES

structs

Sample program
// structs.cpp
// Defines and uses a struct.
// --------------------------------------------------#include 
#include 
#include 
using namespace std;
struct Representative // Defining struct Representative
{
string name;
// Name of a representative.
double sales;
// Sales per month.
};
inline void print( const Representative& v)
{
cout << fixed << setprecision(2)
<< left << setw(20) << v.name
<< right << setw(10) << v.sales << endl;
}
int main()
{
Representative rita, john;
rita.name
= "Strom, Rita";
rita.sales = 37000.37;
john.name
= "Quick, John";
john.sales = 23001.23;
rita.sales += 1700.11;
// More Sales
cout << " Representative
Sales\n"
<< "-------------------------------" << endl;
print( rita);
print( john);
cout << "\nTotal of sales: "
<< rita.sales + john.sales << endl;
Representative *ptr = &john;
// Pointer ptr.
// Who gets the
if( john.sales < rita.sales)
// most sales?
ptr = &rita;
cout << "\nSalesman of the month: "
<< ptr->name << endl;
// Representative's name
// pointed to by ptr.
return 0;
}

structs

■

257

䊐 Records
In a classical, procedural language like C, multiple data that belong together logically are
put together to form a record. Extensive data such as the data for the articles in an automobile manufacturer’s stocks can be organized for ease of viewing and stored in files.
From the viewpoint of an object-oriented language, a record is merely a class containing only public data members and no methods. Thus, you can use the class keyword to
define the structure of a record in C++.

Example: class Date
{ public:

short month, day, year; };

However, it is common practice to use the keyword struct, which is also available
in the C programming language, to define records. The above definition of Date with
the members day, month, and year is thus equivalent to:

Example: struct Date { short month, day, year; };

䊐 The Keywords

class

and

struct

You can also use the keyword struct to define a class, such as the class Account.

Example: struct Account {
private:
public:

//
//

. . .
. . .

as before

};

The keywords class and struct only vary with respect to data encapsulation; the
default for access to members of a class defined as a struct is public. In contrast to a
class defined using the class keyword, all the class members are public unless a private label is used. This allows the programmer to retain C compatibility.

Example: Date future;
future.year = 2100;

// ok! Public data

Records in the true sense of the word, that is, objects of a class containing only public members, can be initialized by means of a list during definition.

Example: Date birthday = { 1, 29, 1987};
The first element in the list initializes the first data member of the object, and so on.

258

■

CHAPTER 13

■

DEFINING CLASSES

UNIONS

An object of union WordByte in memory

Low byte b[0]

w (16 bit word)
High byte b[1]

Defining and using union WordByte
// unions.cpp
// Defines and uses a union.
// --------------------------------------------------#include 
using namespace std;
union WordByte
{
private:
unsigned short w;
// 16 bits
unsigned char b[2];
// Two bytes: b[0], b[1]
public:
// Word- and byte-access:
unsigned short& word()
{ return w; }
unsigned char& lowByte() { return b[0]; }
unsigned char& highByte(){ return b[1]; }
};
int main()
{
WordByte wb;
wb.word() = 256;
cout << "\nWord:
" << (int)wb.word();
cout << "\nLow-byte: " << (int)wb.lowByte()
<< "\nHigh-byte: " << (int)wb.highByte()
<< endl;
return 0;
}

Screen output of the program
Word:
256
Low-Byte: 0
High-Byte: 1

UNIONS

■

259

䊐 Memory Usage
In normal classes, each data member belonging to an object has its own separate memory
space. However, a union is a class whose members are stored in the same memory space.
Each data member has the same starting address in memory. Of course, a union cannot
store various data members at the same address simultaneously. However, a union does
provide for more versatile usage of memory space.

䊐 Definition
Syntactically speaking, a union is distinguished from a class defined as a class or
struct only by the keyword union.

Example: union Number
{
long
n;
double x;
};
Number number1, number2;

This example defines the union Number and two objects of the same type. The union
Number can be used to store either integral or floating-point numbers.
Unless a private label is used, all union members are assumed to be public. This
is similar to the default setting for structures. This allows direct access to the members n
and x in the union Number.

Example: number1.n = 12345; // Storing an integer
number1.n *= 3;
number2.x = 2.77;

// and multiply by 3.
// Floating point number

The programmer must ensure that the current content of the union is interpreted correctly. This is normally achieved using an additional type field that identifies the current
content.
The size of a union type object is derived from the longest data member, as all data
members begin at the same memory address. If we look at our example, the union
Number, this size is defined by the double member, which defaults to 8 ==
sizeof(double) byte.
The example opposite defines the union WordByte that allows you to read or write
to a 16-bit memory space byte for byte or as a unit.

■

CHAPTER 13

■

exercise

260

DEFINING CLASSES

EXERCISE

Struct tm in header file ctime
struct tm
{
int tm_sec;
int tm_min;
int tm_hour;
int tm_mday;
int tm_mon;
int tm_year;
int tm_wday;
int tm_yday;
int tm_isdst;
};

//
//
//
//
//
//
//
//
//

0 - 59(60)
0 - 59
0 - 23
Day of month: 1 - 31
Month: 0 - 11 (January == 0)
Years since 1900 (Year - 1900)
Weekday: 0 - 6 (Sunday == 0)
Day of year: 0 - 365
Flag for summer-time

Sample calls to functions time( ) and localtime( )
#include 
#include 
using namespace std;
struct tm *ptr;
time_t sec;
. . .
time(&sec);
ptr = localtime(&sec);

// Pointer to struct tm.
// For seconds.
//
//
//
//

To get the present time.
To initialize a struct of
type tm and return a
pointer to it.

cout << "Today is the "
<< ptr->tm_yday + 1
<< ". day of the year " << ptr->tm_year
<< endl;
. . .

EXERCISE

■

261

Exercise
A program needs a class to represent the date.
■

Define the class Date for this purpose using three integral data members
for day, month, and year.Additionally, declare the following methods:
void init( int month, int day, int year);
void init(void);
void print(void);

■

✓

Store the definition of the class Date in a header file.
Implement the methods for the class Date in a separate source file:
1. The method print() outputs the date to standard output using the
format Month-Day-Year.
2. The method init() uses three parameters and copies the values
passed to it to corresponding members. A range check is not required
at this stage, but will be added later.
3. The method init() without parameters writes the current date to the
corresponding members.

NOTE
Use the functions declared in ctime
time_t time(time_t *ptrSec)
struct tm *localtime(const time_t *ptrSec);

■

The structure tm and sample calls to this function are included opposite.The type time_t is defined as long in ctime.
The function time() returns the system time expressed as a number of seconds and writes this value to the variable referenced by ptrSec. This value can be passed to the function localtime() that
converts the number of seconds to the local type tm date and returns
a pointer to this structure.
Test the class Date using an application program that once more is stored
in a separate source file.To this end, define two objects for the class and
display the current date. Use object assignments and—as an additional
exercise—references and pointers to objects.

■

CHAPTER 13

solution

262

DEFINING CLASSES

■

SOLUTION

//
//
//
//

---------------------------------------------------date.h
First Definition of class Date.
----------------------------------------------------

#ifndef _DATE_
#define _DATE_
class Date
{
private:
short month, day, year;

// Avoid multiple inclusion.

// Sheltered members:

public:
// Public interface:
void init(void);
void init( int month, int day, int year);
void print(void);
};
#endif

//
//
//
//

//

_DATE_

---------------------------------------------------date.cpp
Implementing the methods of class Date.
----------------------------------------------------

#include "date.h"
#include 
#include 
using namespace std;
// --------------------------------------------------void Date::init(void)
// Get the present date and
{
// assign it to data members.
struct tm *ptr;
// Pointer to struct tm.
time_t sec;
// For seconds.
time(&sec);
ptr = localtime(&sec);

// Get the present date.
// Initialize a struct of
// type tm and return a
// pointer to it.
month = (short) ptr->tm_mon + 1;
day
= (short) ptr->tm_mday;
year = (short) ptr->tm_year + 1900;
}

SOLUTION

// --------------------------------------------------void Date::init( int m, int d, int y)
{
month = (short) m;
day
= (short) d;
year = (short) y;
}
// --------------------------------------------------void Date::print(void)
// Output the date
{
cout << month << '-' << day << '-' << year
<< endl;
}

// ---------------------------------------------------// date_t.cpp
// Using objects of class Date.
// ---------------------------------------------------#include "date.h"
#include 
using namespace std;
int main()
{
Date today, birthday, aDate;
today.init();
birthday.init( 12, 11, 1997);
cout << "Today's date: ";
today.print();
cout << "\n Felix' birthday: ";
birthday.print();
cout << "----------------------------------\n"
"Some testing outputs:" << endl;
aDate = today;
aDate.print();

// Assignment ok

Date *pDate = &birthday;
pDate->print();

// Pointer to birthday

Date &holiday = aDate;
holiday.init( 1, 5, 2000);
aDate.print();

// Reference to aDate.
// Writing to aDate.
// holiday.print();

return 0;
}

■

263

This page intentionally left blank

chapter

14

Methods
This chapter describes
■

how constructors and destructors are defined to create and
destroy objects

■

how inline methods, access methods, and read-only methods
can be used

■

the pointer this, which is available for all methods, and

■

what you need to pay attention to when passing objects as
arguments or returning objects.

265

266

■

CHAPTER 14

■

METHODS

CONSTRUCTORS

Class Account with constructors
// account.h
// Defining class Account with two constructors.
// --------------------------------------------------#ifndef _ACCOUNT_
#define _ACCOUNT_
#include 
using namespace std;
class Account
{
private:
// Sheltered members:
string name;
// Account holder
unsigned long nr;
// Account number
double state;
// State of the account
public:
// Public interface:
Account( const string&, unsigned long, double );
Account( const string& );
bool init( const string&, unsigned long, double);
void display();
};
#endif
// _ACCOUNT_

Defining the constructors
// Within file account.cpp:
Account::Account( const string& a_name,
unsigned long a_nr, double a_state)
{
nr
= a_nr;
name = a_name;
state = a_state;
}
Account::Account( const string& a_name )
{
name = a_name;
nr = 1111111; state = 0.0;
}

CONSTRUCTORS

■

267

䊐 The Task of a Constructor
Traditional programming languages only allocate memory for a variable being defined.
The programmer must ensure that the variable is initialized with suitable values.
An object of the class Account, as described in the previous chapter, does not possess
any valid values until the method init() is called. Non-initialized objects can lead to
serious runtime errors in your programs.
To avoid errors of this type, C++ performs implicit initialization when an object is
defined. This ensures that objects will always have valid data to work on. Initialization is
performed by special methods known as constructors.

䊐 Declaration
Constructors can be identified by their names. In contrast to other member functions,
the following applies:
■
■

the name of the constructor is also the class name
a constructor does not possess a return type—not even void.

Constructors are normally declared in the public section of a class. This allows you to
create objects wherever the class definition is available.
Constructors can be overloaded, just like other functions. Constructors belonging to a
class must be distinguishable by their signature (that is, the number, order, and type of
parameters). This allows for different methods of object initialization. The example
opposite shows an addition to the Account class. The class now has two constructors.

䊐 Definition
Since a constructor has the same name as its class, the definition of a constructor always
begins with
Class_name::Class_name

In the definition itself, the arguments passed can be checked for validity before they are
copied to the corresponding data members. If the number of arguments is smaller than
the number of data members, the remaining members can be initialized using default values.
Constructors can also perform more complex initialization tasks, such as opening files,
allocating memory, and configuring interfaces.

268

■

CHAPTER 14

■

METHODS

CONSTRUCTOR CALLS

Sample program
// account2_t.cpp
// Using the constructors of class Account.
// --------------------------------------------------#include "account.h"
int main()
{
Account giro("Cheers, Mary", 1234567, -1200.99 ),
save("Lucky, Luke");
Account depot;

giro.display();
save.display();

// Error: no default constructor
//
defined.
// To output

Account temp("Funny, Susy", 7777777, 1000000.0);
save = temp;
// ok: Assignment of
//
objects possible.
save.display();
// Or by the presently available method init():
save.init("Lucky, Luke", 7654321, 1000000.0);
save.display();
return 0;
}

CONSTRUCTOR CALLS

■

269

Unlike other methods, constructors cannot be called for existing objects. For this reason,
a constructor does not have a return type. Instead, a suitable constructor is called once
only when an object is created.

䊐 Initialization
When an object is defined, initial values can follow the object name in parentheses.

Syntax:

class object( initializing_list);

During initialization the compiler looks for a constructor whose signature matches the
initialization list. After allocating sufficient memory for the object, the constructor is
called. The values in the initialization list are passed as arguments to the constructor.

Example: account nomoney("Poor, Charles");
This statement calls the constructor with one parameter for the name. The other data
members will default to standard values.
If the compiler is unable to locate a constructor with a suitable signature, it will not
create the object but issue an error message.

Example: account somemoney("Li, Ed",10.0); // Error!
The class Account does not contain a constructor with two parameters.
If a constructor with only one parameter is defined in the class, the statement can be
written with an equals sign =.

Example: account nomoney = "Poor, Charles";
This statement is equivalent to the definition in the example before last. Initialization
with parentheses or the = sign was introduced previously for fundamental types. For
example, int i(0); is equivalent to int i =0;.

䊐 Default Constructor
A constructor without parameters is referred to as a default constructor. The default constructor is only called if an object definition does not explicitly initialize the object. A
default constructor will use standard values for all data members.
If a class does not contain a constructor definition, the compiler will create a minimal
version of the default constructor as a public member. However, this constructor will
not perform initialization. By contrast, if a class contains at least one constructor, a
default constructor must be defined explicitly, if it is needed. The definition of the
Account class does not specify a default constructor; thus a new account object can be
created with initialization only.

270

■

CHAPTER 14

■

METHODS

DESTRUCTORS

Sample program
// demo.cpp
// Outputs constructor and destructor calls.
// --------------------------------------------------#include 
#include 
using namespace std;
int count = 0;
// Number of objects.
class Demo
{
private:
string name;
public:
Demo( const string& );
// Constructor
~Demo();
// Destructor
};
Demo::Demo( const string& str)
{
++count; name = str;
cout << "I am the constructor of "<< name << '\n'
<< "This is the " << count << ". object!\n"
}
Demo:: ~Demo()
// Defining the destructor
{
cout << "I am the destructor of " << name << '\n'
<< "The " << count << ". object "
<< "will be destroyed " << endl;
--count;
}
// -- To initialize and destroy objects of class Demo -Demo globalObject("the global object");
int main()
{
cout << "The first statement in main()." << endl;
Demo firstLocalObject("the 1. local object");
{
Demo secLocalObject("the 2. local object");
static Demo staticObject("the static object");
cout << "\nLast statement within the inner block"
<< endl;
}
cout << "Last statement in main()." << endl;
return 0;
}

DESTRUCTORS

■

271

䊐 Cleaning Up Objects
Objects that were created by a constructor must also be cleaned up in an orderly manner.
The tasks involved in cleaning up include releasing memory and closing files.
Objects are cleaned up by a special method called a destructor, whose name is made up
of the class name preceded by ∼ (tilde).

䊐 Declaration and Definition
Destructors are declared in the public section and follow this syntax:

Syntax:

∼class_name(void);

Just like the constructor, a destructor does not have a return type. Neither does it
have any parameters, which makes the destructor impossible to overload. Each class thus
has one destructor only.
If the class does not define a destructor, the compiler will create a minimal version of
a destructor as a public member, called the default destructor.
It is important to define a destructor if certain actions performed by the constructor
need to be undone. If the constructor opened a file, for example, the destructor should
close that file. The destructor in the Account class has no specific tasks to perform. The
explicit definition is thus:
Account::∼Account(){}

// Nothing to do

The individual data members of an object are always removed in the order opposite of
the order in which they were created. The first data member to be created is therefore
cleaned up last. If a data member is also a class type object, the object’s own destructor
will be called.

䊐 Calling Destructors
A destructor is called automatically at the end of an object’s lifetime:
■
■

for local objects except objects that belong to the static storage class, at the
end of the code block defining the object
for global or static objects, at the end of the program.

The sample program on the opposite page illustrates various implicit calls to constructors
and destructors.

272

■

CHAPTER 14

■

METHODS

INLINE METHODS

Sample class Account
// account.h
// New definition of class Account with inline methods
// ---------------------------------------------------#ifndef _ACCOUNT_
#define _ACCOUNT_
#include 
#include 
#include 
using namespace std;
class Account
{
private:
string name;
unsigned long nr;
double state;
public:

// Sheltered members:
// Account holder
// Account number
// State of the account
//Public interface:
// Constructors: implicit inline
Account( const string& a_name = "X",
unsigned long a_nr
= 1111111L,
double a_state
= 0.0)
{
name = a_name; nr = a_nr; state = a_state;
}
~Account(){ } // Dummy destructor: implicit inline
void display();

};
// display() outputs data of class Account.
inline void Account::display()
// Explicit inline
{
cout << fixed << setprecision(2)
<< "--------------------------------------\n"
<< "Account holder:
" << name << '\n'
<< "Account number:
" << nr
<< '\n'
<< "Account state:
" << state << '\n'
<< "--------------------------------------\n"
<< endl;
}
#endif
// _ACCOUNT_

INLINE METHODS

■

273

A class typically contains multiple methods that fulfill simple tasks, such as reading or
updating data members. This is the only way to ensure data encapsulation and class functionality.
However, continually calling “short” methods can impact a program’s runtime. In
fact, saving a re-entry address and jumping to the called function and back into the calling function can take more time than executing the function itself. To avoid this overhead, you can define inline methods in a way similar to defining inline global
functions.

䊐 Explicit and Implicit inline Methods
Methods can be explicitly or implicitly defined as inline. In the first case, the method
is declared within the class, just like any other method. You simply need to place the
inline keyword before the method name in the function header when defining the
method.

Example: inline void Account::display()
{
. . .
}

Since the compiler must have access to the code block of an inline function, the
inline function should be defined in the header containing the class definition.
Short methods can be defined within the class. Methods of this type are known as
implicit inline methods, although the inline keyword is not used.

Example:

// Within class Account:
bool isPositive(){ return state > 0; }

䊐 Constructors and Destructors with inline Definitions
Constructors and destructors are special methods belonging to a class and, as such, can
be defined as inline. This point is illustrated by the new definition of the Account
class opposite. The constructor and the destructor are both implicit inline. The constructor has a default value for each argument, which means that we also have a default
constructor. You can now define objects without supplying an initialization list.

Example: Account temp;
Although we did not explicitly supply values here, the object temp was correctly initialized by the default constructor we defined.

274

■

CHAPTER 14

■

METHODS

ACCESS METHODS

Access methods for the Account class
// account.h
// Class Account with set- and get-methods.
// ---------------------------------------------------#ifndef _ACCOUNT_
#define _ACCOUNT_
#include 
#include 
#include 
using namespace std;
class Account
{
private:
string name;
unsigned long nr;
double state;
public:

// Sheltered members:
// Account holder
// Account number
// State of the account
//Public interface:
// constructors, destructor:
Account( const string& a_name = "X",
unsigned long a_nr
= 1111111L,
double a_state
= 0.0)
{ name = a_name; nr = a_nr; state = a_state; }
∼Account(){ }
// Access methods:
const string& getName() { return name; }
bool
setName( const string& s)
{
if( s.size() < 1)
// No empty name
return false;
name = s;
return true;
}
unsigned long getNr() { return nr; }
void
setNr( unsigned long n) { nr = n; }
double getState() { return state; }
void
setState(double x) { state = x; }
void display();

};
// inline definition of display() as before.
#endif
// _ACCOUNT_

ACCESS METHODS

■

275

䊐 Accessing Private Data Members
An object’s data members are normally found in the private section of a class. To
allow access to this data, you could place the data members in the public section of the
class; however, this would undermine any attempt at data encapsulation.
Access methods offer a far more useful way of accessing the private data members.
Access methods allow data to be read and manipulated in a controlled manner. If the
access methods were defined as inline, access is just as efficient as direct access to the
public members.
In the example opposite, several access methods have been added to the Account
class. You can now use the
getName(), getNr(), getState()

methods to read the individual data members. As is illustrated in getName(), references
should be read-only when used as return values. Direct access for write operations could
be possible otherwise. To manipulate data members, the following methods can be used:
setName(), setNr(), setState().

This allows you to define a new balance, as follows:

Example: save.setState( 2199.0);

䊐 Access Method Benefits
Defining access methods for reading and writing to each data member may seem like a
lot of work—all that typing, reams of source code, and the programmer has to remember
the names and tasks performed by all those methods.
So, you may be asking yourself how you benefit from using access methods. There are
two important issues:
■

■

Access methods can prevent invalid access attempts at the onset by performing
sanity checks. If a class contains a member designed to represent positive numbers only, an access method can prevent processing negative numbers.
Access methods also hide the actual implementation of a class. It is therefore possible to modify the internal structure of your data at a later stage. If you detect
that a new data structure will allow more efficient data handling, you can add
this modification to a new version of the class. Provided the public interface to
the class remains unchanged, an application program can be leveraged by the
modification without needing to modify the application itself. You simply recompile the application program.

276

■

CHAPTER 14

■

METHODS

const OBJECTS AND METHODS

Read-only methods in the Account class
// account.h
// Account class with read-only methods.
// ---------------------------------------------------#ifndef _ACCOUNT_
#define _ACCOUNT_
#include 
#include 
#include 
using namespace std;
class Account
{
private:
//
// Data members: as
public:
//
// Constructors and
. . . // as before

Sheltered members
before
Public interface
destructor

// Get-methods:
const string& getName() const { return name; }
unsigned long getNr()
const { return nr; }
double
getState() const { return state; }
// Set-methods:
. . . // as before
// Additional methods:
void display() const;
};
// display() outputs the data of class Account.
inline void Account::display() const
{
cout << fixed << setprecision(2)
<< "--------------------------------------\n"
<< "Account holder:
" << name << '\n'
<< "Account number:
" << nr
<< '\n'
<< "Account state:
" << state << '\n'
<< "--------------------------------------\n"
<< endl;
}
#endif
// _ACCOUNT_

const OBJECTS AND METHODS

■

277

䊐 Accessing const Objects
If you define an object as const, the program can only read the object. As mentioned
earlier, the object must be initialized when you define it for this reason.

Example: const Account inv("YMCA, FL", 5555, 5000.0);
The object inv cannot be modified at a later stage. This also means that methods such
as setName() cannot be called for this object. However, methods such as getName or
display() will be similarly unavailable although they only perform read access with
the data members.
The reason for this is that the compiler cannot decide whether a method performs
write operations or only read operations with data members unless additional information is supplied.

䊐 Read-Only Methods
Methods that perform only read operations and that you need to call for constant objects
must be identified as read-only. To identify a method as read-only, append the const
keyword in the method declaration and in the function header for the method definition.

Example: unsigned long getNr() const;
This declares the getNr() method as a read-only method that can be used for constant
objects.

Example: cout << "Account number: " << inv.getNr();
Of course, this does not prevent you from calling a read-only method for a non-constant
object.
The compiler issues an error message if a read-only method tries to modify a data
member. This also occurs when a read-only method calls another method that is not
defined as const.

䊐 const and Non-const Versions of a Method
Since the const keyword is part of the method’s signature, you can define two versions
of the method: a read-only version, which will be called for constant objects by default,
and a normal version, which will be called for non-const objects.

278

■

CHAPTER 14

■

METHODS

STANDARD METHODS

Sample program
// stdMeth.cpp
// Using standard methods.
// --------------------------------------------------#include 
#include 
#include 
using namespace std;
class CD
{ private:
string interpret, title;
long
seconds;
// Time duration of a song
public:
CD( const string& i="", const string& t="", long s = 0L)
{
interpret = i;
title = t;
seconds = s;
}
const string& getInterpret() const{ return interpret; }
const string& getTitle() const
{ return title; }
long getSeconds() const
{ return seconds; }
};
// Generate objects of class CD and output it in tabular form
void printLine( CD cd) ;
// A row of the table
int main()
{
CD cd1( "Mister X", "Let's dance", 30*60 + 41),
cd2( "New Guitars", "Flamenco Collection", 2772 ),
cd3 = cd1,
// Copy constructor!
cd4;
// Default constructor.
cd4 = cd2;
// Assignment!
string line( 70,'-');
line += '\n';
cout << line << left
<< setw(20) << "Interpreter" << setw(30) << "Title"
<< "Length (Min:Sec)\n" << line << endl;
printLine(cd3);
// Call by value ==>
printLine(cd4);
// Copy constructor!
return 0;
}
void printLine( CD cd)
{
cout << left << setw(20) << cd.getInterpret()
<< setw(30) << cd.getTitle()
<< right << setw(5) << cd.getSeconds() / 60
<< ':'
<< setw(2) << cd.getSeconds() % 60 << endl;
}

STANDARD METHODS

■

279

Every class automatically contains four standard methods:
■
■
■
■

the default constructor
the destructor
the copy constructor and
the assignment.

You can use your own definitions to replace these standard methods. As illustrated by
the sample class Account, the compiler only uses the pre-defined default constructor if
no other constructor is available.
The default constructor and the implicit, minimal version of a destructor were introduced earlier.

䊐 Copy Constructor
The copy constructor initializes an object with another object of the same type. It is
called automatically when a second, already existing object is used to initialize an object.

Example: Account myAccount("Li, Ed", 2345, 124.80);
Account yourAccount(myAccount);

In this example, the object yourAccount is initialized by calling the copy constructor
with the myAccount object. Each member is copied individually, that is, the following
initialization process takes place:
yourAccount.name = myAccount.name;
yourAccount.nr
= myAccount.nr;
yourAccount.state = myAccount.state;

The copy constructor is also called when an object is passed to a function by value.
When the function is called, the parameter is created and initialized with the object used
as an argument.

䊐 Assignment
Assignment has been used in several previous examples. An object can be assigned to
another object of the same type.

Example: hisAccount = yourAccount;
The data members of the yourAccount object are copied to the corresponding members of hisAccount in this case also. In contrast to initialization using the copy constructor, assignment requires two existing objects.
Later in the book, you will be introduced to situations where you need to define the
copy constructor or an assignment yourself, and the necessary techniques will be discussed.

280

■

CHAPTER 14

■

METHODS

this POINTER

Sample class DayTime
// DayTime.h
// The class DayTime represents the time in
// hours, minutes and seconds.
// --------------------------------------------------#ifndef _DAYTIME_
#define _DAYTIME_
class DayTime
{
private:
short hour, minute, second;
bool overflow;
public:
DayTime( int h = 0, int m = 0, int s = 0)
{
overflow = false;
if( !setTime( h, m, s))
// this->setTime(...)
hour = minute = second = 0; // hour is equivalent
}
// to this->hour etc.
bool setTime(int hour, int minute, int second = 0)
{
if(
hour
>= 0 && hour < 24
&& minute >= 0 && minute < 60
&& second >= 0 && second < 60 )
{
this->hour
= (short)hour;
this->minute = (short)minute;
this->second = (short)second;
return true;
}
else
return false;
}
int getHour()
const { return hour;
}
int getMinute() const { return minute; }
int getSecond() const { return second; }
int asSeconds() const
// daytime in seconds
{
return (60*60*hour + 60*minute + second);
}
bool isLess( DayTime t) const // compare *this and t
{
return asSeconds() < t.asSeconds();
}
// this->asSeconds() < t.asSeconds();
};
#endif

// _DAYTIME_

this POINTER

■

281

䊐 Accessing the Current Object
A method can access any member of an object without the object name being supplied
in every case. A method will always reference the object with which it was called.
But how does a method know which object it is currently working with? When a
method is called, it is passed a hidden argument containing the address of the current
object.
The address of the current object is available to the method via the constant pointer
this. Given that actObj is the current object of type Class_id, for which a method
was called, the pointer this has the following declaration:
Class_id* const this = &actObj;

The name this is a keyword. As this is a constant pointer, it cannot be redirected.
In other words, the pointer this allows you to access the current object only.

䊐 Using the this Pointer
You can use the this pointer within a method to address an object member as follows:

Example: this->data

// Data member: data
// Calling member function

this->func()

The compiler implicitly creates an expression of this type if only a member of the current
object is supplied.

Example: data = 12;

// Corresponds to this->data=12;

Write operations of this type are permissible since the pointer this is a constant, but
the referenced object is not. However, the above statement would be invalid for a readonly method.
The this pointer can be used explicitly to distinguish a method’s local variables from
class members of the same name. This point is illustrated by the sample method
setTime() on the opposite page.
The this pointer is always necessary to access the current object, *this, collectively. This situation often occurs when the current object needs to be returned as a copy
or by reference. Then the return statement is as follows:
return

*this;

282

■

CHAPTER 14

■

METHODS

PASSING OBJECTS AS ARGUMENTS

Calling methods setTime() and isLess()
#include "DayTime.h"
. . .
DayTime depart1( 11, 11, 11), depart2;
. . .
depart2.setTime(12, 0, 0);
if( depart1.isLess( depart2) )
cout << "\nThe 1st plane takes off earlier" << endl;
. . .

Global function swap()
#include "DayTime.h"
// Defines the global function swap():
void swap( DayTime& t1, DayTime& t2)
// Two
{
// parameters!
DayTime temp(t1); t1 = t2; t2 = temp; // To swap
}
// t1 and t2.
// A call (e.g. in function main()):
DayTime arrival1( 14, 10), arrival2( 15, 20);
. . .
swap( arrival1, arrival2);
// To swap
. . .

Implementing swap() as a method
// Defines the method swap():
class DayTime
// With a new method swap()
{ . . .
public:
void swap( DayTime& t)
// One parameter!
{
// To swap *this and t:
DayTime temp(t); t = *this; *this = temp;
}
};
// A call (e.g. in function main()):
#include "DayTime.h"
DayTime arrival1( 10, 10), arrival2( 9, 50);
. . .
arrival1.swap(arrival2);
. . .

PASSING OBJECTS AS ARGUMENTS

■

283

䊐 Passing by Value
As you already know, passing by value copies an object that was passed as an argument to
the corresponding parameter of a function being called. The parameter is declared as an
object of the class in question.

Example: bool isLess( DayTime t) const;
When the method isLess() is called, the copy constructor executes and initializes the
created object, t, with the argument.
depart1.isLess( depart2)

// Copy constructor

The function uses a copy of the object depart2. The copy is cleaned up when leaving
the function, that is, the destructor is called.

䊐 Passing by Reference
The overhead caused by creating and cleaning up objects can be avoided by passing arguments by reference. In this case, the parameter is declared as a reference or pointer.

Example: bool isLess( const DayTime& t) const;
This new declaration of the isLess() method is preferable to the previous declaration.
There is no formal difference to the way the method is called. However, isLess() no
longer creates an internal copy, but accesses directly the object being passed. Of course,
the object cannot be changed, as the parameter was declared read-only.

䊐 Methods Versus Global Functions
Of course, it is possible to write a global function that expects one object as an argument.
However, this rarely makes sense since you would normally expect an object’s functionality to be defined in the class itself. Instead, you would normally define a method for the
class and the method would perform the task in hand. In this case, the object would not
be passed as an argument since the method would manipulate the members of the current object.
A different situation occurs where operations with at least two objects need to be performed, such as comparing or swapping. For example, the method isLess() could be
defined as a global function with two parameters. However, the function could only
access the public interface of the objects. The function swap()on the opposite page
additionally illustrates this point.
The major advantage of a globally defined function is its symmetry. The objects
involved are peers, since both are passed as arguments. This means that conversion rules
are applied to both arguments when the function is called.

284

■

CHAPTER 14

■

METHODS

RETURNING OBJECTS

Global function currentTime()
#include "DayTime.h"
#include 
using namespace std;

// Functions time(), localtime()

const DayTime& currentTime()
// Returns the
{
// present time.
static DayTime curTime;
time_t sec; time(&sec); // Gets the present time.
// Initializes the struct
struct tm *time = localtime(&sec); // tm with it.
curTime.setTime( time->tm_hour, time->tm_min,
time->tm_sec );
return curTime;
}

Sample program
// DayTim_t.cpp
// Tests class DayTime and function currentTime()
// --------------------------------------------------#include "DayTime.h"
// Class definition
#include 
using namespace std;
const DayTime& currentTime();

// The current time.

int main()
{
DayTime cinema( 20,30);
cout << "\nThe movie starts at ";
cinema.print();
DayTime now(currentTime());
// Copy constructor
cout << "\nThe current time is ";
now.print();
cout << "\nThe movie has ";
if( cinema.isLess( now) )
cout << "already begun!\n" << endl;
else
cout << "not yet begun!\n" << endl;
return 0;
}

RETURNING OBJECTS

■

285

A function can use the following ways to return an object as a return value: It can create
a copy of the object, or it can return a reference or pointer to that object.

䊐 Returning a Copy
Returning a copy of an object is time-consuming and only makes sense for small-scale
objects.

Example: DayTime startMeeting()
{
DayTime start;
. . .
// Everyone has time at 14:30:
start.setTime( 14, 30);
return( start);
}

On exiting the function, the local object start is destroyed. This forces the compiler to
create a temporary copy of the local object and return the copy to the calling function.

䊐 Returning a Reference
Of course, it is more efficient to return a reference to an object. But be aware that the
lifetime of the referenced object must not be local.
If this is the case, the object is destroyed on exiting the function and the returned reference becomes invalid. If you define the object within a function, you must use a
static declaration.
The global function currentTime() on the opposite page exploits this option by
returning a reference to the current time that it reads from the system each time the
function is called. The sample program that follows this example uses the current time to
initialize the new object now and then outputs the time. In order to output the time, an
additional method, print(), was added to the class.

䊐 Using Pointers as Return Values
Instead of returning a reference, a function can also return a pointer to an object. In this
case too, you must ensure that the object still exists after exiting the function.

Example: const DayTime* currentTime() // Read-only pointer
{

// to the current time
. . . // Unchanged
return &curTime;

}

■

CHAPTER 14

■

exercise s

286

METHODS

EXERCISES

Class Article
Private members:
Type
Article number:
Article name:
Sales price:

long
string
double

Public members:
Article(long, const string&, double);
~Article();
void print();
// Formatted output
set- and get-methods for any data member

Output from constructor
An object of type Article . . . is created.
This is the . . .. Article.

Output from destructor
The object of type Article . . . is destroyed.
There are still . . . articles.

EXERCISES

■

287

Exercise 1
A warehouse management program needs a class to represent the articles in
stock.
■

■

Define a class called Article for this purpose using the data members
and methods shown opposite. Store the class definition for Article in a
separate header file. Declare the constructor with default arguments for
each parameter to ensure that a default constructor exists for the class.
Access methods for the data members are to be defined as inline. Negative prices must not exist. If a negative price is passed as an argument,
the price must be stored as 0.0.
Implement the constructor, the destructor, and the method print() in a
separate source file. Also define a global variable for the number of
Article type objects.
The constructor must use the arguments passed to it to initialize the data
members, additionally increment the global counter, and issue the message
shown opposite.
The destructor also issues a message and decrements the global counter.
The method print() displays a formatted object on screen.After
outputting an article, the program waits for the return key to be pressed.

■

The application program (again use a separate source file) tests the Article class. Define four objects belonging to the Article class type:

1. A global object and a local object in the main function.
2. Two local objects in a function test() that is called twice by main().
One object needs a static definition.The function test() displays
these objects and outputs a message when it is terminated.

■

Use articles of your own choice to initialize the objects.Additionally, call
the access methods to modify individual data members and display the
objects on screen.
Test your program. Note the order in which constructors and destructors are called.

Supplementary question: Suppose you modify the program by declaring a function
called test() with a parameter of type Article and calling the function with
an article type object.The counter for the number of objects is negative after
running the program.Why?

288

■

CHAPTER 14

METHODS

Methods for class Date
Public Methods:
Date();
Date( int month, int day, int year);
void setDate();
bool setDate( int mn, int da, int yr);
int getMonth() const;
int getDay() const;
int getYear() const;
bool isEqual( const Date&) const;
bool isLess( const Date&) const;
const string& asString() const;
void print() const;

Converting a number to a string
The class stringstream offers the same functionality for reading and writing to
character buffer as the classes istream and ostream do.Thus, the operators >>
and << , just as all manipulators, are available.
// Example: Converting a number to a string.
#include 
// Class stringstream
#include 
// Manipulators
double x = 12.3;
string str;
stringstream iostream;
iostream << setw(10) << x;
iostream >> str;

//
//
//
//
//
//

Number
Destination string
For conversion
number -> string.
Add to the stream.
Read from the stream.

Notices for exercise 3
■ A year is a leap year if it is divisible by 4 but not by 100. In addition, all
multiples of 400 are leap years. February has 29 days in a leap year.
■ Use a switch statement to examine the number of days for months containing less than 31 days.

EXERCISES

■

289

Exercise 2
In the exercises for chapter 13, an initial version of the Date class containing
members for day, month, and year was defined. Now extend this class to add
additional functionality.The methods are shown on the opposite page.
■

■

■

■

■

The constructors and the method setDate() replace the init method
used in the former version.The default constructor uses default values,
for example, 1.1.1, to initialize the objects in question.The setDate()
method without any parameters writes the current date to the object.
The constructor and the setDate() method with three parameters do
not need to perform range checking.This functionality will be added in
the next exercise.
The methods isEqual() and isLess() enable comparisons with a date
passed to them.
The method asString() returns a reference to a string containing the
date in the format mm-dd-year, e.g. 03-19-2006.You will therefore need
to convert any numerical values into their corresponding decimal strings.
This operation is performed automatically when you use the << operator
to output a number to the standard output cout. In addition to the cin
and cout streams, with which you are already familiar, so-called string
streams with the same functionality also exist. However, a string stream
does not read keyboard input or output data on screen. Instead, the target, or source, is a buffer in main memory.This allows you to perform formatting and conversion in main memory.
Use an application program that calls all the methods defined in the class
to test the Date class.

Exercise 3
The Date class does not ensure that an object represents a valid date.To avoid
this issue, add range checking functionality to the class. Range checking is
performed by the constructor and the setDate() method with three
parameters.
■

■

■

First, write a function called isLeapYear() that belongs to the bool
type and checks whether the year passed to it is a leap year. Define the
function as a global inline function and store it in the header file
Date.h.
Modify the setDate() method to allow range checking for the date
passed to it.The constructor can call setDate().
Test the new version of the Date class.To do so, and to test setDate(...), read a date from the keyboard.

■

CHAPTER 14

■

solutions

290

METHODS

SOLUTIONS

Exercise 1
// ---------------------------------------------------// article.h
// Defines a simple class, Article.
// ---------------------------------------------------#ifndef _ARTICLE_
#define _ARTICLE_
#include 
using namespace std;
class Article
{
private:
long nr;
string name;
double sp;

// Article number
// Article name
// Selling price

public:
Article( long nr=0, const string& name="noname",
double sp=0.0);
~Article();
void print();
const string& getName() const { return name; }
long
getNr()
const { return nr; }
double
getSP()
const { return sp; }
bool setName( const string& s)
{
if( s.size() < 1)
return false;
name = s;
return true;
}
void setNr( long n) { nr = n; }
void setSP(double v)
{
sp = v > 0.0 ? v : 0.0;
}
};
#endif

// _ARTICLE_

// No empty name

// No negative price

SOLUTIONS

// -----------------------------------------------------// article.cpp
// Defines those methods of Article, which are
// not defined inline.
// Screen output for constructor and destructor calls.
// -----------------------------------------------------#include "Article.h"
// Definition of the class
#include 
#include 
using namespace std;
// Global counter for the objects:
int count = 0;
// -----------------------------------------------------// Define constructor and destructor:
Article::Article( long nr, const string& name, double sp)
{
setNr(nr);
setName(name);
setSP(sp);
++count;
cout << "Created object for the article " << name
<< ".\n"
<< "This is the " << count << ". article!\n"
}
Article::~Article()
{
cout << "Cleaned up object for the article " << name
<< ".\n"
<< "There are still " << --count << " articles!"
<< endl;
}
// -----------------------------------------------------// The method print() outputs an article.
void Article::print()
{
long savedFlags = cout.flags();
// To mark the
// flags of cout.
cout << fixed << setprecision(2)
<< "-----------------------------------------\n"
<< "Article data:\n"
<< " Number ....: " << nr
<< '\n'
<< " Name
....: " << name << '\n'
<< " Sales price: " << sp
<< '\n'
<< "-----------------------------------------"
<< endl;
cout.flags(savedFlags);
// To restore
// old flags.
cout << " --- Go on with return --- ";
cin.get();
}

■

291

292

■

CHAPTER 14

METHODS

// -----------------------------------------------------// article_t.cpp
// Tests the Article class.
// -----------------------------------------------------#include "Article.h"
// Definition of the class
#include 
#include 
using namespace std;
void test();
// -- Creates and destroys objects of Article class -Article Article1( 1111,"volley ball", 59.9);
int main()
{
cout << "\nThe first statement in main().\n" << endl;
Article Article2( 2222,"gym-shoes", 199.99);
Article1.print();
Article2.print();
Article& shoes = Article2;
// Another name
shoes.setNr( 2233);
shoes.setName("jogging-shoes");
shoes.setSP( shoes.getSP() - 50.0);
cout << "\nThe
shoes.print();
cout << "\nThe
test();
cout << "\nThe
test();
cout << "\nThe
return 0;

new values of the shoes object:\n";
first call to test()." << endl;
second call to test()." << endl;
last statement in main().\n" << endl;

}
void test()
{
Article shirt( 3333, "T-Shirt", 29.9);
shirt.print();
static Article net( 4444, "volley ball net", 99.0);
net.print();
cout << "\nLast statement in function test()"
<< endl;
}

SOLUTIONS

■

293

Answer to the supplementary question:
The copy constructor is called on each “passing by value,” although this
constructor has not been defined explicitly. In other words, the implicitly defined
copy constructor is used and of course does not increment the object counter.
However, the explicitly defined destructor, which decrements the counter, is still
called for each object.
Exercises 2 and 3
// -----------------------------------------------------// Date.h
// Defining class Date with optimized
// functionality, e.g. range check.
// -----------------------------------------------------#ifndef _DATE_
// Avoids multiple inclusions.
#define _DATE_
#include 
using namespace std;
class Date
{
private:
short month, day, year;
public:
Date()
{ month = day = year = 1;

// Default constructor
}

Date( int month, int day, int year)
{
if( !setDate( month, day, year) )
month = day = year = 1; // If date is invalid
}
void setDate();
// Sets the current date
bool setDate( int month, int day, int year);
int getMonth() const { return month; }
int getDay()
const { return day;
}
int getYear() const { return year; }
bool isEqual( const Date& d) const
{
return month == d.month && day == d.day
&& year == d.year ;
}
bool isLess( const Date& d) const;
const string& asString() const;
void print(void) const;
};

294

■

CHAPTER 14

METHODS

inline bool Date::isLess( const Date&
{
if( year != d.year)
return
else if( month != d.month) return
else
return
}

d) const
year < d.year;
month < d.month;
day < d.day;

inline bool isLeapYear( int year)
{
return (year%4 == 0 && year%100 != 0) || year%400 == 0;
}
#endif
// _DATE_

// -----------------------------------------------------// Date.cpp
// Implements those methods of Date class,
// which are not defined inline.
// -----------------------------------------------------#include "Date.h"
// Class definition
#include 
#include 
#include 
#include 
#include 
using namespace std;
// ----------------------------------------------------void Date::setDate()
// Get the present date and
{
// assign it to the data members.
struct tm *dur;
// Pointer to struct tm.
time_t sec;
// For seconds.
time(&sec);
dur = localtime(&sec);

// Get the present time.
// Initialize a struct of
// type tm and return a
// pointer to it.
day
= (short) dur->tm_mday;
month = (short) dur->tm_mon + 1;
year = (short) dur->tm_year + 1900;
}

SOLUTIONS

■

// ----------------------------------------------------bool Date::setDate( int mn, int da, int yr)
{
if( mn < 1 || mn > 12 ) return false;
if( da < 1 || da > 31 ) return false;
switch(mn)
// Month with less than 31 days
{
case 2: if( isLeapYear(yr))
{
if( da > 29)
return false;
}
else if( da > 28)
return false;
break;
case 4:
case 6:
case 9:
case 11:
if( da > 30) return false;
}
month = (short) mn;
day
= (short) da;
year = (short) yr;
return true;
}
// ----------------------------------------------------void Date::print() const
// Output a date
{
cout << asString() << endl;
}
// ----------------------------------------------------const string& Date::asString() const
// Return a date
{
// as string.
static string dateString;
stringstream iostream;
// For conversion
// number -> string
iostream << setfill('0')
// and formatting.
<< setw(2) << month << '-'
<< setw(2) << day
<< '-' << year;
iostream >> dateString;
return dateString;
}

295

296

■

CHAPTER 14

METHODS

// -----------------------------------------------------// date_t.cpp
// Using objects of class Date.
// -----------------------------------------------------#include "Date.h"
#include 
using namespace std;
int main()
{
Date today, birthday( 1, 29, 1927);
const Date d2010(1,1,2010);
cout << "\n Brigit's birthday: "
<< birthday.asString() << endl;
today.setDate();
cout << "\nToday's date: " << today.asString()
<< endl;;
if( today.isLess( d2010))
cout << " Good luck for this decade \n"
<< endl;
else
cout << " See you next decade \n" << endl;
Date holiday;
int month, day, year;

char c;

cout << "\nWhen does your next vacation begin?\n"
<< "Enter in Month-Day-Year format: ";
if( !(cin >> month >> c >> day >> c >> year) )
cerr << "Invalid input!\n" << endl;
else if ( !holiday.setDate( month, day, year))
cerr << "Invalid date!\n" << endl;
else
{
cout << "\nYour first vacation: ";
holiday.print();
if( today.getYear() < holiday.getYear())
cout << "You should go on vacation this year!\n"
<< endl;
else
cout << "Have a nice trip!\n" << endl;
}
return 0;
}

chapter

15

Member Objects and
Static Members
The major topics discussed in this chapter are
■

member objects and how they are initialized

■

data members that are created once only for all the objects in a
class.

In addition, this chapter describes constant members and enumerated
types.

297

298

■

CHAPTER 15

■

MEMBER OBJECTS AND STATIC MEMBERS

MEMBER OBJECTS

A class representing measurement results
// result.h
// Class Result to represent a measurement
// and the time of measurement.
// --------------------------------------------------#ifndef _RESULT_
#define _RESULT_
#include "DayTime.h"
// Class DayTime
class Result
{
private:
double val;
DayTime time;
public:
Result();
// Default constructor
Result(double w, const DayTime& z = currentTime());
Result(double w, int hr, int min, int sec);
double getVal() const { return val; }
void
setVal( double w ) { val = w; }
const DayTime& getTime() const { return time; }
void setTime( const DayTime& z) { time = z; }
bool setTime(int hr, int min, int sec)
{ return time.setTime( hr, min, sec); }
void print() const;
// Output result and time.
};
#endif // _RESULT_

A first implementation of constructors
#include "result.h"
Result::Result() { val = 0.0; }
Result::Result( double w, const DayTime& z)
{
val = w;
time = z;
}
Result::Result( double w, int hr, int min, int sec)
{ val = w;
time = DayTime(hr, min, sec); // Assign a temporary
// object of type
}
// DayTime to time.

MEMBER OBJECTS

■

299

䊐 “Has-A” Relationship
Data members belonging to one class can be objects of a different class. In our example,
the Account class, we already made use of this feature. The name of the account holder
is stored in a string type data member. An object of the Account class therefore has a
string type member sub-object, or member object for short.
If a class contains a data member whose type is of a different class, the relationship
between the classes is referred to as a “Has-A” relationship.

䊐 Calling Constructors
When an object containing member objects is initialized, multiple constructor calls are
to be executed. One of these is the constructor for the complete object, and the others
are constructors for the member objects. The order of the constructor calls is significant in
this case. First, the member objects are created and initialized; this then allows the constructor to create the whole object. Unless otherwise stated, the default constructor will
be called for each member object.

䊐 The Constructors for the Sample Class Result
The example on the opposite page defines a sample class called Result. In addition to a
double type measurement, the time of each measurement is recorded. For ease of reading, the constructors were defined separately, rather than as inline.
The default constructor only sets the value of the measurement to 0. However, initialization is complete since the default constructor is called for the member object time.

Example: Result current;
The default constructor for the member object time first sets the hours, minutes and
seconds to 0. Then the constructor for the Result class is called and a value of 0.0 is
assigned to val.
The other constructors can be used to initialize an object explicitly.

Example: Result temperature1(15.9); // Current Time
Result temperature2(16.7, 14, 30, 35);

Since the compiler has no information on the relation of initial values and member
objects, it first calls the default constructor for the member object time. Subsequently
the instructions for the Result class constructor can be executed, and values are
assigned to the data members.

300

■

CHAPTER 15

■

MEMBER OBJECTS AND STATIC MEMBERS

MEMBER INITIALIZERS

New implementation of constructors
#include "result.h"
Result::Result() : val(0.0) { /* ... */ }
Result::Result( double w, const DayTime& z)
: val(w), time(z)
{ /* ... */ }
Result::Result( double w, int hr, int min, int sec)
: val(w), time(hr, min, sec)
{
/* ... */
}

✓

NOTE
You can replace the comment /* ... */ with statements, if needed. However, in the case of the
Result class there is nothing to do at present.

Sample program
// result_t.cpp
// Tests constructors of class Result
// --------------------------------------------------#include "Result.h"
#include 
using namespace std;
int main()
// Some air temperature measurements
{
DayTime morning(6,30);
Result t1,
// Default constructor
t2( 12.5, morning),
t3( 18.2, 12,0,0),
t4(17.7);
// at current time
cout << "Default values: ";

t1.print();

cout << "\n Temperature
Time \n"
<< "-------------------------" << endl;
t2.print();
t3.print();
t4.print();
cout << endl;
return 0;
}

MEMBER INITIALIZERS

■

301

䊐 Initializing Member Objects
Calling default constructors to create member objects raises several issues:
■
■
■

A member object is initialized first with default values. Correct values are
assigned later. This additional action can impact your program’s performance.
Constant objects or references cannot be declared as member objects since it is
impossible to assign values to them later.
Classes that do not have a default constructor definition cannot be used as types
for member objects.

When defining a constructor, you can use member initializers to ensure general and
efficient use of member objects.

䊐 Syntax for Member Initializers
A member initializer contains the name of the data member, followed by the initial values in parentheses.

Example: time(hr,min,sec)

// Member initializer

Multiple member initializers are separated by commas. A list of member initializers
defined in this way follows the constructor header and is separated from the header by a
colon.

Example: Result::Result( /* Parameters */ )
: val(w), time(hr, min, sec)
{ /* Function block */ }

This ensures that a suitable constructor will be called for data members with member initializers and avoids calls to the default constructor with subsequent assignments. As the
example shows, you can also use member initializers for data members belonging to fundamental types.
The argument names of the member initializers are normally constructor parameters.
This helps pass the values used to create an object to the right member object.

✓

NOTE
Member initializers can only be stated in a constructor definition. The constructor declaration remains
unchanged.

302

■

CHAPTER 15

■

MEMBER OBJECTS AND STATIC MEMBERS

CONSTANT MEMBER OBJECTS

New version of class Result
// result2.h
// The class Result with a constant data member.
// --------------------------------------------------#ifndef _RESULT_
#define _RESULT_
#include "DayTime.h"
// Class DayTime
class Result
{
private:
double val;
const DayTime time;
public:
Result(double w, const DayTime& z = currentTime());
Result(double w, int hr, int min, int sec);
double getVal() const { return val; }
void
setVal( double w ) { val = w; }
const DayTime& getTime() const { return time; }
void print() const;
};
#endif // _RESULT_

Using the new class Result
// result2_t.cpp :
Tests the new class Result.
// --------------------------------------------------#include "result2.h"
#include 
using namespace std;
int main()
{
DayTime start(10,15);
Result m1( 101.01, start),
m2( m1),
// Copy constructor ok!
m3( 99.9);
// At current time.
// m2 = m3;
// Error! Standard assignment incorrect.
m2.setVal(100.9);
// Corrected value for m2
cout << "\n Result
Time \n"
<< "-------------------------" << endl;
m1.print();
m2.print();
m3.print();
return 0;
}

CONSTANT MEMBER OBJECTS

■

303

䊐 Declaring const Member Objects
If a class contains data members that need to keep their initial values, you can define
these members as const. For example, you could set the time for a measurement once
and not change this time subsequently. However, you need to be able to edit the measurement value to correct systematic errors. In this case, the member object time can be
declared as follows:

Example: const DayTime time;
Since the const member object time cannot be modified by a later assignment, the correct constructor must be called to initialize the object. In other words, when you define a
constructor for a class, you must also define a member initializer for each const member
object.

䊐 The Sample Class Result
If the member object time is const, the first version of the constructors are invalid
since they modify time by means of a later assignment.

Example: time = DayTime(st, mn, sk);

// Error!

However, the later versions of these constructors are ok. The member initializer ensures
that the desired initial values are used to create the member object time.
One further effect of the const member object is the fact that the setTime(...)
methods can no longer be applied. The compiler will issue an error message at this point
and for any statement in the current program that attempts to modify the static member,
time. This means that a programmer cannot accidentally overwrite a member declared
as a const.
The new version of the Result class no longer contains a default constructor, since a
default value for the time of the measurement does not make sense.

䊐 Example with Fundamental Type
Data members with fundamental types can also be defined as const. The class Client
contains a number, nr, which is used to identify customers. Since the client number
never changes, it makes sense to define the number as const. The constructor for
Client would then read as follows:

Example: Client::Client( /*...*/ ) : nr(++id)
{ /*...*/ }

The member initializer nr(++id) initializes the const data member nr with the
global value id, which is incremented prior to use.

304

■

CHAPTER 15

■

MEMBER OBJECTS AND STATIC MEMBERS

STATIC DATA MEMBERS

Class Result with static members
// result3.h
// The class Result with static data members.
// --------------------------------------------------#ifndef _RESULT_
#define _RESULT_
#include "DayTime.h"
// Class DayTime
class Result
{
private:
double val;
const DayTime time;
// Declaration of static members:
static double min, max;
// Minimum, maximum
static bool first; // true, if it is the first value.
void setMinMax(double w); // private function
public:
Result(double w, const DayTime& z = currentTime());
Result(double w, int hr, int min, int sec);
// ... The other member functions as before
};
#endif // _RESULT_

Implementation and initialization
// result3.cpp
// Defining static data members and
// methods, which are not defined inline.
// --------------------------------------------------#include "result3.h"
double Result::min = 0.0;
double Result::max = 0.0;
bool
Result::first = true;
void Result::setMinMax(double w)
// Help function
{ if(first) { min = max = w;
first = false; }
else if( w < min) min = w;
else if( w > max) max = w;
}
// Constructors with member initializer.
Result::Result( double w, const DayTime& z)
: val(w), time(z)
{ setMinMax(w); }
Result::Result( double w, int hr, int min, int sec)
: val(w), time(hr, min, sec)
{ setMinMax(w); }
// Implements the other member functions.

STATIC DATA MEMBERS

■

305

䊐 Class-Specific Data
Every object has its own characteristics. This means that the data members of two different objects will be stored at different memory addresses.
However, sometimes it is useful to keep some common data that can be accessed by all
the objects belonging to a class, for example:
■
■

figures such as exchange rates, interest rates or time limits which have the same
value for every object
status information, such as the number of objects, current minimum or maximum
threshold values, or pointers to some objects; for example, a pointer to an active
window in a window class.

This kind of data needs to be stored once only, no matter how many objects exist.
Since a programmer will also need to manage the data from within the class, it should be
represented within the class rather than globally. Static data members can be used for this
purpose. In contrast to normal data members, static data members occur only once in
memory.

䊐 Declaration
Static data members are declared within a class, that is, the keyword static is used to
declare members of this type. On the opposite page, the following statement

Example: static double min, max;

// Declaration

defines two static data members called min and max that record the minimum and maximum values for the measurements.

䊐 Definition and Initialization
Static data members occupy memory space even if no objects of the class in question
have been created. Just like member functions, which occur only once, static data members must be defined and initialized in an external source file. The range operator :: is
then used to relate the data members to the class.

Example: double Result::min = 0.0;

// Definition

As the example illustrates, the static keyword is not used during the definition. Static
data members and member functions belonging to the same class are normally defined in
one source file.

306

■

CHAPTER 15

■

MEMBER OBJECTS AND STATIC MEMBERS

ACCESSING STATIC DATA MEMBERS

Class Result with static methods
class Result
{
private:
double val;
const DayTime time;
static double min, max;
// Minimum, Maximum
static bool first;
// true, if first result
static void setMinMax(double w); // Help function
public:
// ... Member functions as before, plus:
static double getMin() { return min; }
static double getMax() { return max; }
};

Application program
// result3_t.cpp
// Uses the new class Result.
// --------------------------------------------------#include "result3.h"
#include 
using namespace std;
int main()
//Some air temperature measurements
{
DayTime morning(6,45);
Result temp1( 6.45, morning),
temp2( 11.2, 12,0,0);
double temp = 0.0;
cout << "\nWhat is the air temperature now? ";
cin >> temp;
Result temp3(temp);
// At current time.
cout << "\n Temperature
Time \n"
<< "-------------------------" << endl;
temp1.print(); temp2.print(); temp3.print();
cout
<< "\n Minimum Temperature: " << Result::getMin()
<< "\n Maximum Temperature: " << Result::getMax()
<< endl;
return 0;
}

ACCESSING STATIC DATA MEMBERS

■

307

䊐 Static Data Members and Encapsulation
The normal rules for data encapsulation also apply to static data members. A static data
member declared as public is therefore directly accessible to any object.
If the static data members min and max in the Result class are declared public
rather than private, and given that temperature is an object belonging to the class,
the following statement

Example: cout << temperature.max;
outputs the maximum measured value. You can also use the range operator:

Example: cout << Result::max;
This syntax is preferable to the previous example, since it clearly shows that a static data
member exists independently of any objects.

䊐 Static Member Functions
Of course, you can use class methods to access a static data member with a private
declaration. However, normal methods can be used for class objects only. Since static
data members are independent of any objects, access to them should also be independent
of any objects. Static member functions are used for this purpose. For example, you can call
a static member function for a class even though no objects exist in that class.
The static keyword is used to define static member functions.

Example: static double getMin();

// Within class.

As the Result class, which was modified to include the static member functions
getMin(), setMin(), etc. shows, an inline definition is also permissible. Definitions outside of the class do not need to repeat the static keyword.
A static member function can be called using any object belonging to the class or,
preferably, using a range operator.

Example: temperature.setMax(42.4);
Result::setMax(42.4);

// Equivalent
// Calls.

Calling a static member function does not bind the function to any class object. The
this pointer is therefore unavailable, in contrast to normal member functions. This also

means that static member functions cannot access data members and methods that are
not static themselves.

308

■

CHAPTER 15

■

MEMBER OBJECTS AND STATIC MEMBERS

ENUMERATION

Sample program
// enum.cpp
// Uses enum-constants within a class.
// --------------------------------------------------#include 
using namespace std;
class Lights
{
public:
// Enumeration for class
enum State { off, red, green, amber };
private:
State state;
public:
Lights( State s = off) : state(s) {}
State getState() const { return state; }
void setState( State s)
{ switch(s)
{ case off:
cout << "
OFF
";
case red:
cout << "
RED
";
case green: cout << "
GREEN
";
case amber: cout << "
AMBER
";
default:
return;
}
state = s;
}
};
int main()
{
cout << "Some statements with objects "
<< "of type Lights!\n"
Lights A1, A2(Lights::red);
Lights::State as;
as = A2.getState();
if( as == Lights::red)
{
A1.setState( Lights::red);
A2.setState( Lights::amber);
}
cout << endl;
return 0;
}

Lights

break;
break;
break;
break;

ENUMERATION

■

309

䊐 Definition
An enumeration is a user-definable, integral type. An enumeration is defined using the
enum keyword. A range of values and a name for these values are also defined at the
same time.

Example: enum Shape{ Line, Rectangle, Ellipse};
This statement defines the enumerated type Shape. The names quoted in the list identify integral constants. Their values can be deduced from the list order. The first constant
has a value of 0, and each subsequent constant has a value that is one higher than its
predecessor.
In the previous example, Line thus represents a value of 0, Rectangle a value of 1,
and Ellipse a value of 2. A Shape type variable can only assume one of these values.

Example: Shape shape = Rectangle;
// ...
switch(shape)
{
case Line:

// Variable shape

// To evaluate shape
// ...

etc.

However, you can also define the values of the constants explicitly.

Example: enum Bound { Lower = -100, Upper = 100};
You can leave out the type name, if you only need to define the constants.

Example: enum { OFF, OUT=0, ON, IN=1 };
This statement defines the constants OFF and OUT, setting their value to 0, and the constants ON and IN with a value of 1. The values for OFF and ON are implicit.

䊐 Class-Specific Constants
Enumeration can be used to define integral symbolic constants in a simple way. In contrast to #define directives, which merely replace text strings, enum constants are part
of a declaration and thus have a valid range. This allows you to define constants that are
visible within a namespace or class only.
The example on the opposite page shows the enumerated type State, which was
defined within the Lights class. This means that the type and enum constant are only
available for direct use within the class. The enumeration itself is declared as public,
however, and access from outside the class is therefore possible.

Example: if(Lights.getState() == Lights::red)
// ...

■

CHAPTER 15

■

exercise s

310

MEMBER OBJECTS AND STATIC MEMBERS

EXERCISES

Copy constructor of class Article
The copy constructor creates a copy of an existing object.The parameter is thus
a read-only reference to the object that needs to be copied.The copy
constructor in the Article class is thus declared as follows:
Declaration of copy constructor:
Article( const Article& );

The default copy constructor simply transfers the data members to the new
object.
The Member Class

Private Data Members

Type

Member Number

int

Name

string

Birthday

const Date

//Possibly more information, such as an address, telephone number, ...
Public Methods
Constructor with one parameter for each data member
Access methods for each data member. The birthday is read-only.
A method for formatted screen output of all data members

EXERCISES

■

311

Exercise 1
In the first exercise of the last chapter you defined a simple class called
Article.This involved using a global counter to log object creation and
destruction. Improve and extend the Article class as follows:
■

■

■

■

Use a static data member instead of a global variable to count the current
number of objects.
Declare a static access method called getCount()for the Article class.
The method returns the current number of objects.
Define a copy constructor that also increments the object counter by 1
and issues a message.This ensures that the counter will always be accurate.
Tip: Use member initializers.
Test the new version of the class.To do so, call the function test() by
passing an article type object to the function.

Exercise 2
A sports club needs a program to manage its members.Your task is to define
and test a class called Member for this purpose.
■

■
■

■

■

■

Define the Member class using the data members shown opposite. Use
the Date class defined in the last chapter for your definition. Since a
member’s birthday will not change, the data member for birthdays must
be defined as a const.
Overload the constructor to allow for entering a date as an object as
well as three values for day, month, and year.
Implement the necessary methods.
Test the new Member class by creating at least two objects with the data
of your choice and calling the methods you defined.
Add a static member called ptrBoss to the class.This pointer indicates
the member who has been appointed as chairperson. If no chairperson
has been appointed, the pointer should point to NULL.
Additionally, define the static access methods getBoss() and setBoss().
Use a pointer to set and return the object in question.
Test the enhanced Member class by reading the number of an existing
member, making the member the new chairperson and displaying the
chairperson using getBoss().

312

■

CHAPTER 15

MEMBER OBJECTS AND STATIC MEMBERS

Sample output
Simulation of two traffic lights!
Terminate this program with +!
1. Light
2. Light
--------------------------RED
AMBER
GREEN
AMBER
AMBER
RED
GREEN
AMBER
RED
AMBER
GREEN
// . . .

Hints for implementing the function wait()
1. The function time() is declared in the header file ctime. The call
time(NULL) determines the number of seconds of type time_t since
1.1.1970, 0:0 hours.The type time_t is defined as long.
2. Instead of calling the function time() in a loop, you can use the function
Sleep() for Windows or the function sleep() for Unix. These system
calls are not standardized, yet they are much more effective because they
send a process to sleep instead of using a waiting loop.

EXERCISES

■

313

Exercise 3
Create a program to simulate the signal positions for two sets of traffic lights at
a junction. Use the class Lights as defined in this chapter for your program.
■

■

■

Each set of lights is switched through the phases red, amber, green, amber,
red, and so on.You must ensure that one set of lights can be only in the
amber or green state when the other set of lights is red.
The lights operate in an infinite loop that can be terminated by interrupting the program.You can use the key combination + for DOS
and Windows and the Interrupt key, i.e., normally the  key, for
UNIX.
The status of the lights is constant for a certain number of seconds. For
example, the green phase can take 20 seconds and the amber phase 1
second.These values can be different for each set of lights. Define an
auxiliary function
inline void wait( int sec)

The function returns after the stipulated number of seconds.To do so,
you can call the standard function time() in a loop. Don’t forget to read
the notes on the opposite page.

■

CHAPTER 15

■

solutions

314

MEMBER OBJECTS AND STATIC MEMBERS

SOLUTIONS

Exercise 1
// ---------------------------------------------------// article.h
// Defines a simple class - Article.
// ---------------------------------------------------#ifndef _ARTICLE_H_
#define _ARTICLE_H_
#include 
using namespace std;
class Article
{
private:
long nr;
string name;
double sp;
// Static data member:
static int countObj;

// Article number
// Article name
// Sales price
// Number of objects

public:
Article( long nr=0, const string& name="noname",
double sp=0.0);
// Copy constructor:
Article( const Article& anArticle);
~Article();
void print();
// Access methods:
const string& getName() const { return name; }
long
getNr()
const { return nr; }
double
getSP()
const { return sp; }
static int getCount() { return countObj; }
bool setName( const string& s)
{
if( s.size() < 1)
// No empty Name
return false;
name = s;
return true;
}
void setNr( long n) { nr = n; }
void setSP(double v)
{
// No negative price
sp = v > 0.0 ? v : 0.0;
}
};
#endif

// _ARTICLE_

SOLUTIONS

//
//
//
//
//

■

--------------------------------------------------article.cpp
Methods of Article, which are not defined as inline.
Constructor and destructor output when called.
---------------------------------------------------

#include "article.h"

// Definition of the class

#include 
#include 
using namespace std;
// Defining the static data member:
int Article::countObj = 0;
// Number of objects
// Defining the constructor and destructor:
Article::Article( long nr, const string& name, double sp)
{
setNr(nr);
setName(name);
setSP(sp);
++countObj;
cout << "An article \"" << name
<< "\" is created.\n"
<< "This is the " << countObj << ". article!"
<< endl;
}
// Defining the copy constructor:
Article::Article( const Article& art)
:nr(art.nr), name(art.name), sp(art.sp)
{
++countObj;
cout << "A copy of the article \"" << name
<< "\" is generated.\n"
<< "This is the " << countObj << ". article!"
<< endl;
}
Article::~Article()
{
cout << "The article \"" << name
<< "\" is destroyed.\n"
<< "There are still " << --countObj << " articles!"
<< endl;
}
// The method print() outputs an article.
void Article::print()
{
// As before! Compare to the solutions of chapter 14.
}

315

316

■

CHAPTER 15

//
//
//
//

MEMBER OBJECTS AND STATIC MEMBERS

-----------------------------------------------------article_t.cpp
Tests the class Article including a copy constructor.
------------------------------------------------------

#include "article.h"

// Definition of the class

#include 
#include 
using namespace std;
void test( Article a);

// Prototype

Article article1( 1111,"tent", 159.9);

// Global

int main()
{
cout << "\nThe first statement in main().\n" << endl;
Article article2( 2222,"jogging shoes", 199.99);
cout << "\nThe first call of test()." << endl;
test(article1);
// Passing by Value
cout << "\nThe second call of test()." << endl;
test(article2);
// Passing by Value
cout << "\nThe last statement in main().\n"
<< "\nThere are still " << Article::getCount()
<< " objects\n" << endl;
return 0;
}

void test( Article a)
// Calls the copy constructor
{
cout << "\nThe given object:" << endl;
a.print();
static Article bike( 3333, "bicycle", 999.0);
cout << "\nThe static object in function test():"
<< endl;
bike.print();
cout << "\nThe last statement in function test()"
<< endl;
}

SOLUTIONS

■

317

Exercise 2
The Date class from the last chapter ( see files Date.h and Date.cpp ) can be
left unchanged. But it makes sense to define the function isLeapYear()as a
static member function of class Date rather than globally.
The other files:
// -----------------------------------------------------// member.h
// Defines the Member class containing a constant
// and a static member.
// -----------------------------------------------------#ifndef _MEMBER_H_
#define _MEMBER_H_
#include "Date.h"
#include 
using namespace std;
class Member
{
private:
int nr;
string name;
const Date birth;
// ... more data
static Member *ptrBoss;

// Member number
// Name
// Birthday

// Pointer to boss,
// NULL = no boss.

public:
Member( long m_nr, const string& m_name,
const Date& m_birth)
: nr(m_nr), birth(m_birth)
{
if( !setName(m_name)) name = "Unknown";
}
Member( long m_nr, const string& m_name,
int day, int month, int year)
: nr(m_nr), birth(day,month,year)
{
if( !setName(m_name)) name = "Unknown";
}
int
getNr()
const { return nr; }
const string& getName() const { return name; }
const Date& getBirthday() const { return birth; }
void setNr( int n) { nr = n; }

318

■

CHAPTER 15

MEMBER OBJECTS AND STATIC MEMBERS

bool setName( const string& s)
{
if( s.size() < 1)
// No empty name
return false;
name = s;
return true;
}
void display() const;
// static methods:
static Member* getBoss()
{
return ptrBoss;
}
static void setBoss( Member* ptrMem)
{
ptrBoss = ptrMem;
}
};
#endif
//
//
//
//

// _MEMBER_H_

--------------------------------------------------member.cpp
Members of class Member not defined inline.
---------------------------------------------------

#include "member.h"
#include 
using namespace std;

// Class definition

// Pointer to the boss:
Member* Member::ptrBoss = NULL;
void Member::display() const
{
string line( 50, '-');
cout << line
<< "\n Member number: " << nr
<< "\n Member:
" << name
<< "\n Birthday
" << birth.asString()
<< '\n' << line << endl;
}

SOLUTIONS

// ---------------------------------------------------// member_t.cpp
// Using the class Member.
// ---------------------------------------------------#include "member.h"
// Class definition
#include 
#include 
using namespace std;
int main()
{
Date today; today.setDate();
cout << "Date: " << today.asString() << endl;
Member fran( 0, "Quick, Fran", 17,11,81),
kurt( 2222, "Rush, Kurt", Date(3,5,77) );
franzi.setNr(1111);
cout << "\nTwo members of the sports club:\n" << endl;
fran.display();
kurt.display();
cout << "\nSomething changed!" << endl;
fran.setName("Rush-Quick");
fran.display();
Member benny( 1122,"Rush, Benny", 1,1,2000);
cout << "The youngest member of the sports club: \n";
benny.display();
// Who is the boss?
int nr;
Member *ptr = NULL;
cout << "\nWho is the boss of the sports club?\n"
<< "Enter the member number: ";
if( cin >> nr)
{
if( nr == fran.getNr())
ptr = &fran;
else if( nr == kurt.getNr())
ptr = &kurt;
Member::setBoss( ptr);
}
cout << "\nThe Boss of the sports club:" << endl;
ptr = Member::getBoss();
if( ptr != NULL)
ptr->display();
else
cout << "No boss existing!" << endl;
return 0;
}

■

319

320

■

CHAPTER 15

MEMBER OBJECTS AND STATIC MEMBERS

Exercise 3
The definition of class Lights from this chapter remains
unchanged.
// ----------------------------------------------------// Lights_t.cpp : Simulates two traffic lights.
// ----------------------------------------------------#include "lights.h"
// Definition of class Lights
#include 
#include 
// Standard function time()
using namespace hr;
inline void wait( int sec)
// Wait sec seconds.
{ time_t end = time(NULL) + sec;
while( time(NULL) < end)
;
}
// Alternative for Windows:
// #include 
// inline void wait( int sec) { Sleep( 1000 * sec); }
Lights A1, A2;
// Traffic lights and
enum { greenTime1 = 10 , amberTime1 = 1, // time to wait.
greenTime2 = 14 , amberTime2 = 2 };
int main()
{ cout << "Simulating two traffic lights!\n\n"
<< "Terminate this program with +!\n"
<< endl;
cout << " 1. Light
2. Light\n"
<< "---------------------------" << endl;
while(true)
{ A1.setState( Lights::red);
// A1 = red
A2.setState( Lights::amber);
cout << endl;
wait( amberTime2);
cout << "
";
A2.setState( Lights::green); cout << endl;
wait(greenTime2);
cout << "
";
A2.setState( Lights::amber);
cout << endl;
wait(amberTime2);
A1.setState( Lights::amber);
// A2 = red
A2.setState( Lights::red);
cout << endl;
wait(amberTime1);
A1.setState( Lights::green); cout << endl;
wait(greenTime1);
A1.setState( Lights::amber);
cout << endl;
wait(amberTime1);
}
return 0;
}

chapter

16

Arrays
This chapter describes how to define and use arrays, illustrating onedimensional and multidimensional arrays, C strings and class arrays.

321

322

■

CHAPTER 16

■

ARRAYS

DEFINING ARRAYS

The array arr in memory

arr[0]
arr[1]
arr[2]
.

.
.

.

.

.

arr[9]

Sample program
// array.cpp
// To input numbers into an array and output after.
// ---------------------------------------------------#include 
#include 
using namespace std;
int main()
{
const int MAXCNT = 10;
float arr[MAXCNT], x;
int i, cnt;

// Constant
// Array, temp. variable
// Index, quantity

cout << "Enter up to 10 numbers \n"
<< "(Quit with a letter):" << endl;
for( i = 0; i < MAXCNT && cin >> x; ++i)
arr[i] = x;
cnt = i;
cout << "The given numbers:\n" << endl;
for( i = 0; i < cnt; ++i)
cout << setw(10) << arr[i];
cout << endl;
return 0;
}

DEFINING ARRAYS

■

323

An array contains multiple objects of identical types stored sequentially in memory. The
individual objects in an array, referred to as array elements, can be addressed using a number, the so-called index or subscript. An array is also referred to as a vector.

䊐 Defining Arrays
An array must be defined just like any other object. The definition includes the array
name and the type and number of array elements.

Syntax:

type name[count];

// Array name

In the above syntax description, count is an integral constant or integral expression
containing only constants.

Example: float arr[10];

// Array arr

This statement defines the array arr with 10 elements of float type. The object arr
itself is of a derived type, an “array of float elements” or “float array.”
An array always occupies a contiguous memory space. In the case of the array arr, this
space is 10*sizeof(float) = 40 bytes.

䊐 Index for Array Elements
The subscript operator [] is used to access individual array elements. In C++ an index
always begins at zero. The elements belonging to the array arr are thus
arr[0], arr[1] , arr[2], ... , arr[9]

The index of the last array element is thus 1 lower than the number of array elements.
Any int expression can be used as an index. The subscript operator [] has high precedence, just like the class member operators . and -> .
No error message is issued if the index exceeds the valid index range. As a programmer, you need to be particularly careful to avoid this error! However, you can define a
class to perform range checking for indices.
You can create an array from any type with the exception of some special types, such
as void and certain classes. Class arrays are discussed later.

Example: short number[20];

// short array

for( int i=0; i < 20; i++ )
number[i] = (short)(i*10);

This example defines an array called number with 20 short elements and assigns the
values 0, 10, 20, ... , 190 to the elements.

324

■

CHAPTER 16

■

ARRAYS

INITIALIZING ARRAYS

Sample program
// fibo.cpp
// The program computes the first 20 Fibonacci
// numbers and the corresponding Fibonacci quotients.
// ----------------------------------------------------#include 
#include 
#include 
// Prototype of sqrt()
#include 
using namespace std;
#define COUNT 20
long fib[COUNT + 1] = { 0, 1 };
string header =
" Index Fibonacci number Fibonacci quotient Deviation"
"\n
of limit "
"\n---------------------------------------------------";
int main()
{
int i;
double q, lim;
for( i=1; i < COUNT; ++i )
fib[i+1] = fib[i] + fib[i-1];

// Computing the
// Fibonacci numbers

lim = ( 1.0 + sqrt(5.0)) / 2.0;

// Limit

// Title and the first two Fibonacci numbers:
cout << header << endl;
cout << setw(5) << 0 << setw(15) << fib[0] << endl;
cout << setw(5) << 1 << setw(15) << fib[1] << endl;
// Rest of the table:
for( i=2; i <= COUNT; i++ )
{
// Quotient:
q = (double)fib[i] / (double)fib[i-1];
cout << setw(5) << i << setw(15) << fib[i]
<< setw(20) << fixed << setprecision(10) << q
<< setw(20) << scientific << setprecision(3)
<< lim - q << endl;
}
return 0;
}

INITIALIZING ARRAYS

■

325

䊐 Initialization List
Arrays can be initialized when you define them. A list containing the values for the individual array elements is used to initialize the array:

Example: int num[3] = { 30, 50, 80 };
A value of 30 is assigned to num[0], 50 to num[1], and 80 to num[2]. If you initialize
an array when you define it, you do not need to state its length.

Example: int num[] = { 30, 50, 80 };
In this case, the length of the array is equal to the number of initial values. If the array
length is explicitly stated in the definition and is larger than the number of initial values,
any remaining array elements are set to zero. If, in contrast, the number of initial values
exceeds the array length, the surplus values are ignored.
Locally defined arrays are created on the stack at program runtime. You should therefore be aware of the following issues when defining arrays:
■
■

Arrays that occupy a large amount of memory (e.g., more than one kbyte) should
be defined as global or static.
Unless they are initialized, the elements of a local array will not necessarily have
a definite value. Values are normally assigned by means of a loop.

You cannot assign a vector to another vector. However, you can overload the assignment operator within a class designed to represent arrays. This topic will be discussed in
depth later.

䊐 The Sample Program Opposite
The example on the opposite page contains the first twenty Fibonacci numbers and their
quotients. Fibonacci numbers are useful for representing natural growth. In computer science, Fibonacci numbers are used for things like memory management and hashing.
Their definition is as follows:
■
■

the first Fibonacci number is 0, the second is 1
each subsequent Fibonacci number is the sum of its two immediate predecessors.

This results in the following sequence: 0, 1, 1, 2, 3, 5, 8, 13, ....
The quotient of a Fibonacci number and its predecessor is referred to as a Fibonacci
quotient. The sequence of Fibonacci quotients, 1/1, 2/1, 3/2, ..., converges towards
the threshold value (1 + √5)/2.

326

■

CHAPTER 16

■

ARRAYS

C STRINGS

䊐 Initializing
char text[40] = "Hello Eve";

String text in memory:

✓

String text

'H'

'e'

'l'

'l'

'o'

' '

'E'

'v'

'e'

'\0'

.

.

Index:

0

1

2

3

4

5

6

7

8

9

10

11

NOTE
The array text has length of 40, whereas the string “Hello Eve" only occupies the first 9 bytes.

Sample program
// C-string.cpp : Using C strings.
// ----------------------------------------------------#include 
#include 
#include 
using namespace std;
char header[] = "\n
*** C Strings ***\n\n";
int main()
{
char hello[30] = "Hello ", name[20], message[80];
cout << header << "Your first name: ";
cin >> setw(20) >> name;
// Enter a word.
strcat( hello, name);
// Append the name.
cout << hello << endl;
cin.sync();
// No previous input.
cout << "\nWhat is the message for today?"
<< endl;
cin.getline( message, 80); // Enter a line with a
// max of 79 characters.
if( strlen( message) > 0)
// If string length is
{
// longer than 0.
for( int i=0; message[i] != '\0'; ++i)
cout << message[i] << ' ';
// Output with
cout << endl;
// white spaces.
}
return 0;
}

C STRINGS

■

327

䊐 char Arrays
Arrays whose elements are of char type are often used as data communication buffers.

Example: char buffer[10*512];

// 5 Kbyte buffer

However, their most common use is for string storage. One way of representing a
string is to store the string and the terminating null character '\0' in a char array.
When you define an array, you can use a string constant to initialize the array.

Example: char name[] = "Hugo";
This definition is equivalent to
char name[] = { 'H','u','g','o','\0' };

As you can see, the string name occupies five bytes, including an additional byte for the
null character. If you need to allocate more memory, you can state the size of the array
explicitly as shown opposite.
In the C language, strings are usually represented as char vectors with a terminating
null character. In C++, strings of this type are referred to as C strings to distinguish them
from objects of the string class.

䊐 C Strings and the

string

Class

C strings are simple char arrays, which means that the functionality of the string
class is not available for them. Thus, for example, assignments and comparisons are not
defined.

Example: char str1[20], str2[20] = "A string";
str1 = str2;
strcpy( str1, str2);

// Error!
// ok!

The standard functions of the C language, such as strlen(), strcpy(), strcmp(),
and others, are available for C strings. These global functions all begin with the str prefix.
As the program on the opposite page shows, I/O streams are overloaded for char
arrays, too. Input and output are as easily achieved as with string class objects. However, the program must make sure not to overrun the end of the char array when reading data into the array. You can use the width() method or the setw() manipulator
for this purpose.

Example: cin >> setw(20) >> name; // 19 characters
C strings are preferable to the string class if only a few operations are needed and
you want to avoid unnecessary overheads.

328

■

CHAPTER 16

■

ARRAYS

CLASS ARRAYS

Sample program
// AccountTab.cpp
// An array containing objects of class Account.
// --------------------------------------------------#include "account.h"
#include 
using namespace std;

// Definition of class Account

Account giro("Lucky, Peter", 1234567, -1200.99 );
Account accountTab[] =
{
Account("Tang, Sarah", 123000, 2500.0),
Account("Smith, John", 543001),
Account(),
// Default constructor
"Li, Zhang",
// Account("Li, Zhang"),
giro
// Account(giro)
};
int cnt = sizeof(accountTab) / sizeof(Account);
int main()
{
// To set some values:
accountTab[1].setState( 10000.00);
// Assignment ok:
accountTab[2] = Account("Pit, Dave", 727003, 200.00);
cout << "The accounts in the table:" << endl;
for( int i = 0; i < cnt; ++i)
{
accountTab[i].display();
if( i % 3 == 2)
{
cout << "Press return to go on!\n";
cin.get();
}
}
cout << endl;
return 0;
}

CLASS ARRAYS

■

329

䊐 Declaring Class Arrays
Array elements can also be objects of a class type. The array is known as a class array in
this case. When you declare an array of this type, you only need to state the type of the
array elements.

Example: Result temperatureTab[24];
This statement defines the class array temperatureTab that stores 24 objects of type
Result. This class was introduced at the beginning of the last chapter.
As the statement does not initialize the array explicitly, the default constructor is
automatically called for each array element.

✓

NOTE
Class arrays can only be defined without explicit initialization if a default constructor exists for the class.

Thus, the previous example is only valid for the first version of the Result class as
this class contains a default constructor.

䊐 Explicit Initialization
A class array is initialized as usual by an initialization list. The list contains a constructor
call for each array element.

Example: Result temperatureTab[24] =
{
Result( -2.5, 0,30,30),
Result( 3.5),
// At present time
4.5,
// Just so
Result( temp1),
// Copy constructor
temp2
// Just so
};

The first five array elements are initialized by the constructor calls implicitly contained
in these statements. Instead of using a constructor with one argument, you can simply
supply the argument. The default constructor is then called for the remaining elements.
If the size of an array is not stated explicitly, the number of values in the initialization
list defines the size of the array.
The public interface of the objects in the array is available for use as usual.

Example: temperatureTab[2].setTime( 2,30,21);
No additional parentheses are needed in this statement since the subscript operator []
and the class member operator . are read from left to right, although they have the same
precedence.

330

■

CHAPTER 16

■

ARRAYS

MULTIDIMENSIONAL ARRAYS

Sample program
// multidim.cpp
// Demonstrates multidimensional arrays.
// ---------------------------------------------------#include 
#include 
using namespace std;
char representative[2][20] = {"Armstrong, Wendy",
"Beauty, Eve"};
// Each representative has five different
// articles available, having sold the following:
int articleCount[2][5] = { { 20, 51, 30, 17, 44},
{150, 120, 90, 110, 88}
};
int main()
{
for( int i=0; i < 2; i++ )
{
cout <<"\nRepresentative: " << representative[i];
cout << "\nNumber of items sold: ";
for( int j = 0; j < 5; j++ )
cout << setw(6) << articleCount[i][j];
cout << endl;
}
return 0;
}

Screen output:
Representative: Armstrong, Wendy
Items sold:
20
51
30
17
Representative: Beauty, Eve
Items sold:
150 120
90

110

44

88

MULTIDIMENSIONAL ARRAYS

■

331

䊐 Defining Multidimensional Arrays
In C++ you can define multidimensional arrays with any number of dimensions. The
ANSI standard stipulates a minimum of 256 dimensions but the total number of dimensions is in fact limited by the amount of memory available.
The most common multidimensional array type is the two-dimensional array, the socalled matrix.

Example: float number[3][10];

// 3 x 10 matrix

This defines a matrix called number that contains 3 rows and 10 columns. Each of the 30
(3 ⫻ 10) elements is a float type. The assignment

Example: number[0][9] = 7.2;

// Row 0, column 9

stores the value 7.2 in the last element of the first row.

䊐 Arrays as Array Elements
C++ does not need any special syntax to define multidimensional arrays. On the contrary, an n-dimensional array is no different than an array with only one dimension
whose elements are (n–1)-dimensional arrays.
The array number thus contains the following three elements:
number[0]

number[1]

number[2].

Each of these elements is a float array with a size of 10, which in turn forms the rows of
the two-dimensional array, number.
This means that the same rules apply to multidimensional arrays as to one-dimensional arrays. The initialization list of a two-dimensional array thus contains the values of
the array elements, that is, the one-dimensional rows.

Examples: int arr[2][3] = { {5, 0, 0}, {7, 0, 0} };
int arr[][3]

= { {5}, {7} };

These two definitions are equivalent. When you initialize an array, you can only omit
the size of the first dimension. It is necessary to define any other dimensions since they
define the size of array elements.

䊐 The Example on the Opposite Page
The program opposite defines the two-dimensional arrays representative and
articleCount, which have two rows each. The representative[i] rows are
char arrays used for storing the names of the representatives. You can also use a onedimensional string array.

Example: string representative[2] = {"La..","Fo.."};

332

■

CHAPTER 16

■

ARRAYS

MEMBER ARRAYS

Class TelList
// telList.h
// Class TelList to represent a list
// containing names and telephone numbers.
// ---------------------------------------------------#ifndef _TelList_
#define _TelList_
#include 
using namespace std;
#define PSEUDO -1
#define MAX 100

// Pseudo position
// Maximal number of elements

// Type of a list element:
struct Element { string name, telNr; };
class TelList
{
private:
Element v[MAX];
int count;

// The array and the current
// number of elements

public:
TelList(){ count = 0;}
int getCount() const { return count; }
Element *retrieve( int i )
{
return (i >= 0 && i < count)? &v[i] : NULL;
}
bool append( const Element& el )
{
return append( el.name, el.telNr);
}
bool append( const string& name,
const string& telNr);
bool erase( const string& name);
int search( const string& name);
void print();
int print( const string& name);
int getNewEntries();
};
#endif

// _TelList_

MEMBER ARRAYS

■

333

䊐 Encapsulating Arrays
A programmer often needs to handle objects of the same type, such as company employees, bank accounts, or the articles in stock. A class designed to perform this task can use
an array for ease of data management. An array allows you to access individual objects
directly and perform searches.
A class that encapsulates an array will provide methods for simple array operations,
such as inserting and deleting objects. When you design a class of this type, one aim will
be to perform automatic range checking. This helps avoid overrunning the end of an
array when performing read or write operations. The resulting class will contain a comfortable and safe interface for object data management.

䊐 The Class TelList
The class TelList on the opposite page is designed to manage a simple telephone list.
Each entry in the list contains a dataset containing a name and a phone number. The
Element type, which comprises two strings, was defined for this purpose. The array v
can store up to MAX entries of the Element type. The data member count records the
number of elements currently stored in the array. When a phone list is created, this number will initially be 0. When an element is inserted or deleted, the number is modified
correspondingly.
The TelList class uses a single default constructor that sets the counter, count, to
zero. It is not necessary to provide an initial value for the MAX elements in the array v
since the default constructor of the string class is executed for all strings.
The tasks performed by the other methods are easily deduced from their names. The
retrieve() method returns to a given index a pointer to the corresponding element.
Using a pointer makes it possible to return a NULL pointer if the index is invalid.
The append() methods add a new entry to the list. The data passed to a method is
copied to the next free array element and the counter is incremented. If there is no space
available, the name field is empty, or the name is already in use, nothing happens. In this
case, the method returns false instead of true.
The exercises for this chapter contain further details on these methods. You can
implement the methods for the TelList yourself and go on to test them.

■

CHAPTER 16

■

exercise s

334

ARRAYS

EXERCISES

Example of a bubble sort algorithm
Original array:

100

50

30

70

After the first loop:

50

30

70

40

40

100

largest element
30

After the second loop:

50

40

70

100

second largest element

Sieve of Eratosthenes
For this task you can define an array of boolean values in which each element is
initially true.To eliminate a number n you simply set the nth element in the array
to false.
Result:
Index
Array

0

1

2

false false true

3

4

5

6

7

8

9

true false true false true false false

...
...

Screen shot of exercise 4
* * B R E A K * * * *
---

✓

Press interrupt key to terminate (^C) ---

NOTE
The output of a scrolling string has to be performed at the same cursor position.
The screen control characters make it possible to locate the cursor, and that
independent of the current compiler (see appendix).

EXERCISES

■

335

Exercise 1
Write a C++ program that reads a maximum of 100 integers from the keyboard,
stores them in a long array, sorts the integers in ascending order, and displays
sorted output. Input can be terminated by any invalid input, such as a letter.

✓

NOTE
Use the bubble sort algorithm to sort the array. This algorithm repeatedly accesses the array, comparing
neighboring array elements and swapping them if needed. The sorting algorithm terminates when there
are no more elements that need to be swapped. You use a flag to indicate that no elements have been
swapped.

Exercise 2
Chapter 14 introduced the sample class DayTime and the isLess() method.
Define and initialize an array with four DayTime class objects.
Then write a main function that first uses the print() method to display the
four elements. Finally, find the largest and smallest elements and output them on
screen.
Exercise 3
Write a program that outputs all prime numbers less than 1000.The program
should also count the number of prime numbers less than 1000.An integer >= 2
is a prime number if it is not divisible by any number except 1 and itself. Use the
Sieve of Eratosthenes:
To find primary numbers simply eliminate multiples of any primary numbers
you have already found, i.e.:
first eliminate any multiples of 2 ( 4, 6, 8, ... ),
then eliminate any multiples of 3 ( 6, 9, 12, ...),
then eliminate any multiples of 5 ( 10, 15, 20, ..) // 4 has already been eliminated
and so on.

Exercise 4
Write a C++ program to create the screen output shown opposite.The
following banner
* * *

B R E A K

* * *

is to be displayed in the center of the window and scrolled left.You can scroll
the banner by beginning string output with the first character, then the second,
and so on. Handle the string like a loop where the first letter follows the last
letter and output continues until the starting position is reached.
You can use a wait loop to modify the speed of the banner after each string is
output.

336

■

CHAPTER 16

ARRAYS

Exercise 5
Methods to be implemented for the TelList class
bool append( const string& name,
const string& telNr);
bool erase( const string& name);
int

search( const string& name);

void print();
int

print( const string& name);

int

getNewEntries();

Menu of the application program
*****

Telephone List

*****

D = Display all entries
F = Find a telephone number
A = Append an entry
E = Erase an entry
Q = Quit the program
Your choice:

EXERCISES

■

337

Exercise 5
The sample class TelList was introduced in this chapter; however, some
methods still need to be implemented and tested.
■

Implement the TelList class methods shown opposite.
The name is used as an unambiguous key.This means the append()
method can only be used to append an entry provided the name is neither blank nor already in use.
The method erase() deletes an array element.The position of the element to be deleted is first located using the search() method. If the element does not exist, erase() returns a value of false. In any other case,
the last element in the array is used to overwrite the element that is to
be deleted and the counter count is decremented.
The search() method finds the position in the array that contains the
search name. If the search operation is unsuccessful, the value PSEUDO is
returned.
The print method without parameters outputs all available entries.You
can pass the first letter or letters of a name to the second method to
output any entries beginning with these letters. Use the method compare() from the string class to help you with this task.
Example:

■

✓

str1.compare( 0, 5, str2) == 0

This expression is true if the five characters subsequent to position 0 in
the strings str1 and str2 are identical.
The getNewEntries() method is used to read new phone list entries
from the keyboard. Each new entry is appended using the append()
method. Reading should be terminated if the user types an empty string.
The method returns the number of new entries.
Write an application program that creates a phone list of type TelList
and displays the menu shown on the opposite page.
The menu must be placed in a function of your own that can return the
command input.The menu must be called in the main loop of the program. Depending on the command input, one of the methods defined in
the class TelList should be called. If the menu item “Erase” or “Search”
is chosen, you must also read a name or the first letters of a name from
the keyboard.

NOTE
The phone list will not be stored permanently in a file. This is just one of the enhancements (another
would be variable length) that will be added at a later stage.

■

CHAPTER 16

■

solutions

338

ARRAYS

SOLUTIONS

Exercise 1
// --------------------------------------------------// bubble.cpp
// Inputs integers into an array,
// sorts in ascending order, and outputs them.
// --------------------------------------------------#include 
#include 
using namespace std;
#define MAX 100
long number[MAX];
int main()
{
int i, cnt;

// Maximum number

// Index, quantity

cout << "\nS o r t i n g I n t e g e r s \n"
<< endl;
// To input the integers:
cout << "Enter up to 100 integers \n"
<< "(Quit with any letter):" << endl;
for( i = 0; i < MAX && cin >> number[i]; ++i)
;
cnt = i;
// To sort the numbers:
bool sorted = false;
// Not yet sorted.
long help;
// Swap.
int end = cnt;
// End of a loop.
while( !sorted)
//
{
//
sorted = true;
--end;
for( i = 0; i < end; ++i)
//
{
//
if( number[i] > number[i+1])
{
sorted = false;
//
help
= number[i];
//
number[i] = number[i+1];
number[i+1]= help;
}
}
}

As long as not
yet sorted.

Compares
adjacent integers.

Not yet sorted.
Swap.

SOLUTIONS

■

// Outputs the numbers
cout << "The sorted numbers:\n" << endl;
for( i = 0; i < cnt; ++i)
cout << setw(10) << number[i];
cout << endl;
return 0;
}

Exercise 2
// ---------------------------------------------------// DayTime.h
// The class DayTime represents the time in hours,
// minutes and seconds.
// ---------------------------------------------------#ifndef _DAYTIME_
#define _DAYTIME_
#include 
#include 
using namespace std;
class DayTime
{
private:
short hour, minute, second;
bool overflow;
public:
DayTime( int h = 0, int m = 0, int s = 0)
{
overflow = false;
if( !setTime( h, m, s))
// this->setTime(...)
hour = minute = second = 0;
}
bool setTime(int hour, int minute, int second = 0)
{
if(
hour
>= 0 && hour < 24
&& minute >= 0 && minute < 60
&& second >= 0 && second < 60 )
{
this->hour
= (short)hour;
this->minute = (short)minute;
this->second = (short)second;
return true;
}
else
return false;
}

339

340

■

CHAPTER 16

ARRAYS

int getHour()
const { return hour;
}
int getMinute() const { return minute; };
int getSecond() const { return second; };
int asSeconds() const
// Daytime in seconds
{
return (60*60*hour + 60*minute + second);
}
bool isLess( DayTime t) const

// Compares
// *this and t.

{
}

return asSeconds() < t.asSeconds();
// this->sSeconds() < t.asSeconds();

void print() const
{
cout << setfill('0')
<< setw(2) << hour
<<
<< setw(2) << minute <<
<< setw(2) << second <<
cout << setfill(' ');
}
void swap( DayTime& t)
//
{
//
DayTime temp(t); t = *this;
}

':'
':'
" Uhr" << endl;

Just one parameter!
Swaps *this and t:
*this = temp;

};
#endif

// _DAYTIME_

// ----------------------------------------------------// TimeTab.cpp
// An array containing objects of class DayTime.
// ----------------------------------------------------#include "DayTime.h"
// Definition of class DayTime
#include 
using namespace std;
char header[] =
"\n\n
*** Table with Daytimes ***\n\n";
int main()
{
DayTime timeTab[4] =
{ 18, DayTime(10,25), DayTime(14,55,30)};
int i;
timeTab[3].setTime( 8,40,50);
// Last element.
cout << header << endl;

SOLUTIONS

■

// Output:
for( i = 0; i < 4; ++i)
{
timeTab[i].print();
cout << endl;
}
// To compute shortest and longest time:
int i_min = 0, i_max = 0;
// Indices for shortest
// and longest elements.
for( i = 1; i < 4; ++i)
{
if( timeTab[i].isLess( timeTab[i_min]) )
i_min = i;
if( timeTab[i_max].isLess( timeTab[i]) )
i_max = i;
}
cout << "\nShortest time: ";

timeTab[i_min].print();

cout << "\nLongest time: ";

timeTab[i_max].print();

return 0;
}

341

342

■

CHAPTER 16

ARRAYS

Exercise 3
// -----------------------------------------------------// sieve.cpp
// Identifies prime numbers using the Sieve of
// Eratosthenes.
// -----------------------------------------------------#include 
#include 
using namespace std;
#define LIMIT 1000
// Upper limit
bool flags[LIMIT] = { false, false}; // Array with flags
int main()
{
register int i, j;
// Indices
for( i = 2; i < LIMIT; ++i)
flags[i] = true;
// Sets flags to true
// Sieving:
for( i = 2; i < LIMIT/2; ++i)
{
if( flags[i])
// Is i a prime number?
{
// Yes -> Delete multiples.
for( j = i+i; j < LIMIT; j += i)
flags[j] = false;
}
}
// To count:
int count = 0;
// Counter
for( i = 2; i < LIMIT; ++i)
if(flags[i])
// If i is a prime number
++count;
// -> count
// Output:
cout << "There are"<< count <<" prime numbers less than"
<< LIMIT << endl;
cout << "\nTo output prime numbers? (y/n) ";
char reply; cin.get(reply);
if( reply == 'y' || reply == 'Y')
{ for( i = 2; i < LIMIT; ++i)
if(flags[i])
// If i is a prime number
{
// -> to output it.
cout.width(8); cout << i;
}
}
cout << endl;
return 0;
}

SOLUTIONS

■

Exercise 4
// ---------------------------------------------------// scroll.cpp
// Scrolling a message.
// ---------------------------------------------------#include 
#include 
using namespace std;
#define DELAY 10000000L
// Output delay
inline void cls()
{
cout << "\033[2J\n";
}

// Clear screen

inline void locate(int z, int s)
// Put cursor in row z
{
// and column s
cout << "\033[" << z << ';' << s << 'H';
}
char msg[] = "* * *

B R E A K

* * * ";

int main()
{
int i, start = 0, len = strlen(msg);
cls(); locate(24, 20);
// Row 24, column 20
cout << "--- Press interrupt key to terminate (^C) ---";
while( true )
{
locate( 12, 25);
// Row 12, column 25
i = start;
// Output from index start
do
{
cout << msg[i++];
i = i % len;
// if( i == len) i = 0;
}
while( i != start);
cout << endl;
// Outputs buffer to screen
// Wait in short
for( int count = 0; count < DELAY; ++count)
;
++start;
// For next output
start %= len;
// start = start % len;
}
cls();
return 0;
}

343

344

■

CHAPTER 16

ARRAYS

Exercise 5
//
//
//
//
//
//
//

---------------------------------------------------telList.h
The class TelList representing a list
with names and telephone numbers.
----------------------------------------------------As before in this chapter.

// ---------------------------------------------------// telList.cpp
// Implements the methods of class TelList.
// ----------------------------------------------------#include "telList.h"
// Definition of class TelList
#include 
#include 
using namespace std;
bool TelList::append( const string& name,
const string& telNr)
{
if( count < MAX
// Space available,
&& name.length() > 1
// 2 characters at least
&& search(name) == PSEUDO) // not yet existing
{
v[count].name = name;
v[count].telNr = telNr;
++count;
return true;
}
return false;
}
bool TelList::erase( const string& key )
{
int i = search(key);
if( i != PSEUDO )
{
// Copies the last
v[i] = v[count-1]; --count; // element to position i
return true;
}
return false;
}

SOLUTIONS

int TelList::search(const string& key )
{
for( int i = 0; i < count; i++ )
if( v[i].name == key )
return i;

■

// Searching.
// Found

return PSEUDO;
// Not found
}
// Functions to support the output:
inline void tabHeader()
// Title of the table
{
cout << "\n Name
Telephone #\n"
"----------------------------------------------"
<< endl;
}
inline void printline( const Element& el)
{
cout << left << setw(30) << el.name.c_str()
<< left << setw(20) << el.telNr.c_str()
<< endl;
}
void TelList::print()
// Outputs all entries
{
if( count == 0)
cout << "\nThe telephone list is empty!" << endl;
else
{
tabHeader();
for( int i = 0; i < count; ++i)
printline( v[i]);
}
}
int TelList::print( const string& name) const // Entries
{
// beginning with name.
int matches = 0, len = name.length();
for( int i = 0; i < count; ++i)
{
if( v[i].name.compare(0, len, name) == 0)
{
if( matches == 0) tabHeader(); // Title before
// first output.
++matches;
printline( v[i]);
}
}
if( matches == 0)
cout << "No corresponding entry found!" << endl;
return matches;
}

345

346

■

CHAPTER 16

ARRAYS

int TelList::getNewEntries()
// Input new entries
{
int inputCount = 0;
cout << "\nEnter new names and telephone numbers:"
"\n(Terminate by empty input) "
<< endl;
Element el;
while( true)
{
cout << "\nNew last name, first name: ";
cin.sync(); getline( cin, el.name);
if( el.name.empty())
break;
cout << "\nTelephone number: ";
cin.sync(); getline( cin, el.telNr);
if( !append( el))
{
cout << "Name has not been found!" << endl;
if( count == MAX)
{
cout << "The Table is full!" << endl;
break;
}
if( search( el.name) != PSEUDO)
cout << "Name already exists!" << endl;
}
else
{
++inputCount;
cout << "A new element has been inserted!"
<< endl;
}
}
return inputCount;
}
// -------------------------------------------------------// telList_t.cpp
// Manages a telephone list.
// -------------------------------------------------------#include "telList.h"
// Definition of class TelList
#include 
#include 
#include 
using namespace std;
inline void cls()
{ cout << "\033[2J\n";// Output only new-lines, if ANSI
}
// control characters are not available.

SOLUTIONS

■

inline void go_on()
{
cout << "\n\nGo on with return! ";
cin.sync(); cin.clear();
while( cin.get() != '\n')
;
}

// No previous input

int menu();

// Reads a command

char header[] =
"\n\n

*****

Telephone List

*****\n\n";

TelList myFriends;

// A telephone list

int main()
{
int action = 0;
string name;

// Command
// Reads a name

myFriends.append("Lucky, Peter", "0203-1234567");
while( action != 'B')
{
action = menu();
cls();
cout << header << endl;
switch( action)
{
case 'D':
myFriends.print();
go_on();
break;

// Show all

case 'F':

// Search
cout <<
"\n--- To search for a phone number ---\n"
"\nEnter the beginning of a name: ";
getline( cin, name);
if( !name.empty())
{
myFriends.print( name);
go_on();
}
break;

case 'A':

// Insert
myFriends.getNewEntries();
break;

347

348

■

CHAPTER 16

ARRAYS

case 'E':

// Delete
cout <<
"\n--- To delete a telephone entry. ---\n "
"\nEnter the complete name: ";
getline( cin, name);
if( !name.empty())
{
if( !myFriends.erase( name))
cout << name << " not found!"
<< endl;
else
cout << "Entry for " << name
<< " deleted!" << endl;
go_on();
}
break;

case 'T':

cls();
break;

// To terminate

}
} // End of while
return 0;
}
int menu()
{
static char menuStr[] =
"\n\n
D = Display all entries"
"\n\n
F = Find a telephone number"
"\n\n
A = Append a new entry "
"\n\n
E = Erase an entry "
"\n\n
Q = Quit the program"
"\n\n Your choice: ";
cls();
cout << header << menuStr;
char choice;
cin.sync(); cin.clear();
if( !cin.get(choice))
choice = 'B';
else
choice = toupper(choice);
cin.sync();
return choice;
}

// No previous input

// Clear input buffer

chapter

17

Arrays and Pointers
This chapter describes the relationship between pointers and arrays.This
includes:
■

pointer arithmetic

■

pointer version of functions

■

pointers as return values and read-only pointers

■

pointer arrays

Operations that use C strings illustrate how to use pointers for efficient
programming. String access via the command line of an application
program is used to illustrate pointer arrays.

349

350

■

CHAPTER 17

■

ARRAYS AND POINTERS

ARRAYS AND POINTERS (1)

Sample program
// textPtr.cpp
// Using arrays of char and pointers to char
// ----------------------------------------------------#include 
using namespace std;
int main()
{
cout << "Demonstrating arrays of char "
<< "and pointers to char.\n"
<< endl;
char text[] = "Good morning!",
name[] = "Bill!";
char *cPtr = "Hello ";

// Let cPtr point
// to "Hello ".

cout << cPtr << name << '\n'
<< text << endl;
cout << "The text \"" << text
<< "\" starts at address " << (void*)text
<< endl;
cout << text + 6
<< endl;

// What happens now?

cPtr = name;

// Let cPtr point to name, i.e. *cPtr
// is equivalent to name[0]
cout << "This is the " << *cPtr << " of " << cPtr
<< endl;
*cPtr = 'k';
cout << "Bill can not " << cPtr << "!\n" << endl;
return 0;
}

Sample output:
Demonstrating arrays of char and pointers to char.
Hello Bill!
Good morning!
The text "Good morning!" starts at address 00451E40
morning!
This is the B of Bill!
Bill can not kill!

ARRAYS AND POINTERS (1)

■

351

䊐 Name and Address of an Array
In C++ the name of an array is also the starting address for that array. To be more precise, an array name is a pointer to the first array element.

Example: char town[] = "Beijing";
In this case, town is a char pointer to town[0], that is, a pointer to the memory
address that stores the 'B' character. Expressions town and &town[0] are thus equivalent.

Example: cout << town; // or:

cout << &town[0];

A pointer to the first character of the string town is passed. The characters forming the
string are read and displayed from this point onward until the terminating null character,
'\0', is reached.

䊐 Pointer Variables and Arrays
An array name is not a pointer variable but a constant that cannot be modified. However, you can assign this constant to a pointer variable.

Example: char *cPtr;
cPtr = town;
cout << cPtr;

// or: cPtr = &town[0];
// To output "Beijing"

Now cPtr points to the array element town[0] just like town. But, in contrast to
town, cPtr is a variable that can be moved.

Example: cPtr = "Hello!";
After this statement, cPtr points to the ‘H' character. String constants such as
“Hello!" are also char arrays and thus represent the address of the first array element.

䊐 Typeless Pointers
If you need to display the address rather than the string, you should pass a void* type
pointer rather than a char pointer.

Example: cout << (void *)town;
This casts the char pointer to a void * type pointer and passes it as an argument to
the << operator, which in turn outputs the address in hexadecimal format. The << operator belongs to the ostream class and is overloaded for void * types for this purpose.
A void * pointer represents a memory address without establishing a certain type.
void * pointers are also referred to as typeless pointers for this reason. When you use a
typeless pointer for memory access, you must therefore name the type being accessed
explicitly by means of type casting.

352

■

CHAPTER 17

■

ARRAYS AND POINTERS

ARRAYS AND POINTERS (2)

Sample program
// arrPtr.cpp
// Outputs addresses and values of array elements.
// --------------------------------------------------#include 
using namespace std;
int arr[4] = { 0, 10, 20, 30 };
int main()
{
cout << "\nAddress and value of array elements:\n"
<< endl;
for( int i
cout <<
<<
<<

= 0; i < 4; i++ )
"Address: " << (void*)(arr+i)
"
Value: " << *(arr+i)
endl;

// &arr[i]
// arr[i]

return 0;
}

Interrelation between pointers and array elements

arr

0

arr[0]

arr + 1

10

arr[1]

arr + 2

20

arr[0]

arr + 3

30

arr[3]

ARRAYS AND POINTERS (2)

■

353

䊐 Addressing Array Elements
Access to individual array elements in C++ is very closely related to pointer arithmetic.
Now let’s look at an int array to illustrate this point.

Example: int arr[4] = { 0, 10, 20, 30 };
As you already know, the name of the array arr is an int pointer to arr[0].
Now it is possible to add or subtract pointers and integral values. The size of the
object referenced by the pointer is automatically taken into consideration.
Since arr is an int pointer to arr[0], arr+1 points to the next array element
arr[1], i.e., to an address that is sizeof(int) bytes higher in memory. The memory
space between the two entries will be two or four bytes, depending on the size of the type
int. Thus the following applies to any given number, i:
arr + i points to the array element arr[i] ,
*(arr + i) is the array element arr[i] ,

This technique can also be used to address memory spaces outside of the array. Thus,
arr - 1 addresses the word that precedes arr[0]. But generally this does not make
much sense, since you have no means of knowing what is stored at this memory address.

䊐 Addressing with Pointer Variables
Array elements can also be addressed using pointer variables.

Example: int *ptr = arr;

// ptr points to arr[0]

In this case, both ptr and arr are pointers to the array element arr[0]. Thus,
ptr + 1, ptr + 2, . . . point to the array elements arr[1], arr[2], ... .
For any given integer, i, the following expressions are thus equivalent:
&arr[i]

arr + i

ptr + i

The following thus represent equivalent values:
arr[i]

*(arr + i)

*(ptr + i)

ptr[i]

At first it might seem surprising that you can use the array notation ptr[i] for pointers.
The compiler translates arr[i] to *(arr + i)—in other words: “Start at address
arr, move i objects up, and access the object!” This also applies to ptr[i].

354

■

CHAPTER 17

■

ARRAYS AND POINTERS

POINTER ARITHMETIC

Examples for arithmetic with pointers
float v[6] = { 0.0, 0.1, 0.2, 0.3, 0.4, 0.5 },
*pv, x;
pv = v + 4;
*pv = 1.4;
pv -= 2;
++pv;

//
//
//
//

Let pv point to v[4].
Assign 1.4 to v[4].
Reset pv to v[2].
Let pv point to v[3].

x = *pv++;

//
//
//
//
//

Assign v[3] to x and
increment pv.
Increment x by v[4] and let
pv point to v[3] again.
Reset pv to v[2].

x += *pv--;
--pv;

To step through an array of classes
// Searches for a given account number in a table of
// accounts and outputs the found account.
// -------------------------------------------------#include "account.h"
// Definition of class Account.
Account accountTab[100]; // Table containing accounts.
int main()
{
int cnt;
// Actual number of accounts.
Account *aPtr;
// Pointer to Account-objects.
// To input data into accountTab and actualize cnt.
// To search for the account number 1234567:
bool found = false;
for( aPtr = accountTab; aPtr < accountTab+cnt;++aPtr)
if( aPtr->getNr() == 1234567 )
{ found = true;
break;
}
if( found)
// Found?
aPtr->display();
// Yes -> display.
// To continue
}

POINTER ARITHMETIC

■

355

In C++ you can perform arithmetic operations and comparisons with pointers, provided
they make sense. This primarily means that the pointer must always point to the elements of an array. The following examples show some of your options with pointer arithmetic:

Example: float v[6], *pv = v;

// pv points to v[0]

int i = 3;

䊐 Moving a Pointer in an Array
As you already know, the addition pv + i results in a pointer to the array element
v[i]. You can use a statement such as pv = pv + i; to store the pointer in the variable pv. This moves the pointer pv i objects, that is, pv now points to v[i].
You can also use the operators ++, --, and += or -= with pointer variables. Some
examples are shown opposite. Please note that the indirection operator, *, and the operators ++ and -- have the same precedence. Operators and operands thus are grouped
from right to left:

Example: *pv++

is equivalent to

*(pv++)

The ++ operator increments the pointer and not the variable referenced by the pointer.
Operations of this type are not possible using the pointer v since v is a constant.

䊐 Subtracting Pointers
An addition performed with two pointers does not return anything useful and is therefore invalid. However, it does make sense to perform a subtraction with two pointers,
resulting in an int value that represents the number of array elements between the
pointers. You can use this technique to compute the index of an array element referenced by a pointer. To do so, you simply subtract the starting address of the array. For
example, if pv points to the array element v[3], you can use the following statement

Example: int index = pv - v;
to assign a value of 3 to the variable index.

䊐 Comparing Pointers
Finally, comparisons can be performed with two pointers of the same type.

Example: for( pv = v + 5; pv >= v; --pv)
cout << setw(10) << *pv;

This loop outputs the numbers contained in v in reverse order. In the example on the
opposite page, the pointer aPtr walks through the first cnt elements of the array
accountTab, as long as aPtr < accountTab + cnt.

356

■

CHAPTER 17

■

ARRAYS AND POINTERS

ARRAYS AS ARGUMENTS

Sample program
//
//
//
//
//

reverse.cpp
Defines and calls the function reverse().
reverse() copies a C string into another C string
and reverses the order of characters.
-----------------------------------------------------

#include 
using namespace std;
#include 

// Header-File for Cstrings,
// here for strlen().
void reverse( char str[], char umstr[]); // Prototype
int main()
// Read a word and
{
// output in reversed order.
const int CNT = 81;
char word[CNT], revword[CNT];
cout << "Enter a word: ";
cin.width(CNT);
// maximal CNT-1 characters
cin >> word;
reverse( word, revword);

// Call

cout << "\nThe \"reversed\" word:
<< endl ;

" << revword

return 0;
}
void reverse( char s1[], char s2[])
// Copies the
{
// reversed C string s1 to s2
int j = 0;
for( int i = strlen(s1)-1; i >= 0; i--, j++)
s2[j] = s1[i];
s2[j] = '\0';

// Terminating character

}

Sample output:
Enter a word: REGAL
The "reversed" word:

LAGER

ARRAYS AS ARGUMENTS

■

357

If an array name is passed as an argument when calling a function, the function actually
receives the address of the first array element. The called function can then perform read
or write operations for any element in the array.

䊐 Declaring Parameters
If the argument is an array, there are two equivalent methods of declaring parameters.
This point is illustrated by the example using strlen() to return the length of a C
string. For example, calling strlen("REGAL") returns a value of 5.
1. You can declare the parameter as an array.

Example: int strlen( char str[])
{

int i;
for( i = 0;
;
return (i);

// Compute length of
// str without '\0'.
str[i] != '\0'; ++i)

}

2. You can declare the parameter as a pointer.

Example: int strlen( char *str)
{

/*

as above

*/

}

In both cases the parameter str is a pointer that stores the starting address of the
array. Array notation is preferable if you intend to use an index to access the elements of
an array. Calling strlen("REGAL"); leads to the following situation:
'R'

'E'

'G'

'A'

'L'

'\0'

str[0]

str[1]

str[2]

str[3]

str[4]

str[5]

As you can see, the length of a C string is equal to the index of the element containing
the terminating null character.
The function reverse() on the opposite page copies the characters of a C string to
a second char array in reverse order, first copying the last character in s1, that is, the
character with the index strlen(s1)-1, to s2[0], then the second to last character
s2[1], and so on.

䊐 Array Length
A function to which an array is passed initially knows only the starting address of the
array but not its length. In the case of C strings, the length is derived implicitly from the
position of the terminating null character. In most other cases the length must be supplied explicitly.

Example: void sort( Account aTab[], int len )
{ /* To sort array aTab of length len */}

358

■

CHAPTER 17

■

ARRAYS AND POINTERS

POINTER VERSIONS OF FUNCTIONS

䊐 Function strcpy
The standard function strcpy()copies C strings.

Example: char dest[30], source[] = "A string";
strcpy( dest, source);

Here the string source is copied to dest “from left to right” just like an assignment.
The following function strcpy() is somewhat simpler than the standard function
since it has no return value.

Index Version of strcpy()
void strcpy( char s1[], char s2[])
// Copies s2 to s1
{
int i;
// Index
for( i = 0; s2[i] != '\0'; ++i) // Copy.
s1[i] = s2[i];
s1[i] = '\0';
// Append terminating
}
// character.

Pointer version 1 of strcpy()
void strcpy( char *s1, char *s2)
// Copies s2 to s1
{
for( ; *s2 != '\0'; ++s1, ++s2) // Copy
*s1 = *s2;
*s1 = '\0';
// Append terminating
}
// character.

Pointer version 2 of strcpy()
void strcpy( char *s1, char *s2)
{
while( (*s1++ = *s2++) != '\0' )
;
}

// Copy s2 to s1.
// Copy and append
// terminating
// character.

POINTER VERSIONS OF FUNCTIONS

■

359

䊐 Using Pointers Instead of Indices
As we have already seen, a parameter for an array argument is always a pointer to the
first array element. When declaring parameters for a given type T:
T name[]

is always equivalent to

T *name.

So far, in previous sample functions, the pointer has been used like a fixed base
address for the array, with an index being used to access the individual array elements.
However, it is possible to use pointers instead of indices.

Example: A new version of the standard function strlen():
int strlen( char *str)
// Computes length
{
// of str without '\0'.
char* p = str;
for( p = str; *p != '\0'; ++p) // Search
;
// for \0
return (p - str);
}

In this case, the difference between two pointers results in the string length.

䊐 The Sample Functions Opposite
The first version of the function strcpy() “string copy” opposite uses an index,
whereas the second does not. Both versions produce the same results: the string s2 is
copied to s1. When you call the function, you must ensure that the char array referenced by s1 is large enough.
As the parameters s1 and s2 are pointer variables, they can be shifted. The second
“pointer version” of strcpy(), which is also shown opposite, uses this feature, although
the function interface remains unchanged.
Generally, pointer versions are preferable to index versions as they are quicker. In an
expression such as s1[i] the values of the variables s1 and i are read and added to
compute the address of the current object, whereas s1 in the pointer version already
contains the required address.

䊐 Multidimensional Arrays as Parameters
In a parameter declaration for multidimensional arrays, you need to state every dimension
with the exception of the first. Thus, a parameter declaration for a two-dimensional array
will always contain the number of columns.

Example: long func( int num[][10] );
long func( int *num[10] );

// ok.
// also ok.

360

■

CHAPTER 17

■

ARRAYS AND POINTERS

READ-ONLY POINTERS

Sample program
// accountFct.cpp
// Defines and calls a function, which outputs
// a list of overdrawn accounts.
// -------------------------------------------------#include "account.h"
// Definition of class Account.
Account accountTab[] =
// Table with Account-objects.
{ Account("Twain, Mark", 1234567, -3434.30),
Account("Crusoe, Robinson", 200000, 0.00),
Account("Temple, Shirley", 543001, +777.70),
Account("Valentin, Carl", 543002, -1111.10),
};
int cnt = sizeof(accountTab) / sizeof(Account);
// Prototype:
int displayOverdraw( const Account *aTab, int cnt,
double limit);
int main()
{
double limit = 0.0;
cout << "Output the overdrawn accounts!\n"
<< "These are the accounts, which fell below \n"
<< "the limit, ex. -1000.00.\n" << endl;
cout << "What is the limit? ";
cin >> limit;
cout << "Listing the overdrawn accounts:\n" << endl;
if( displayOverdraw( accountTab, cnt, limit) == 0)
cout << "\nNo account found!"
<< endl;
return 0;
}
int displayOverdraw( const Account *aTab, int cnt,
double limit)
{ int count = 0;
const Account* aPtr;
for( aPtr = aTab; aPtr < aTab + cnt; ++aPtr)
if( aPtr->getState() < limit ) // Below the limit?
{
aPtr->display();
// Yes -> display.
++count;
}
return count;
}

READ-ONLY POINTERS

■

361

䊐 Pointers to const Objects
You can use a normal pointer for both read and write access to an object. However, just
like the definition used for a reference, you can also define a read-only pointer, that is, a
pointer that can be used for read operations only. In fact, a read-only pointer is obligatory if you need to point to a constant object.

䊐 Declaration
You use the keyword const to define a read-only pointer.

Example: const int a = 5, b = 10,

*p = &a;

This statement defines the constants a and b, and a pointer p to a constant object of
type int. The referenced object *p can be read but not modified.

Example: cout << *p;
*p = 1;

// To read is ok.
// Error!

The pointer itself is not a constant, so it can be modified:

Example: p = &b;

// ok!

The referenced object also does not need to be a constant. In other words, a read-only
pointer can also point to a non-constant object.

Example: Account depo("Twain, Mark", 1234, 4321.90);
const Account* ptr = &depo;
ptr->display();
prt->setState( 7777.70);

// ok!
// ok!
// Error!

But ptr can only be used for read access to the non-constant object depo.

䊐 Read-Only Pointers as Parameters
Read-only pointers are most commonly found in parameter lists. This guarantees that
arguments cannot be modified.

Example: int strlen( const char *s);
In this example, the parameter s is a read-only pointer. This allows you to pass constant
C strings to the standard function strlen(). You cannot remove the “write protection”
by assigning the read-only pointer s to a normal pointer.

Example: char *temp = s;

// Error!

You need to declare a read-only pointer if a constant object may be passed as an argument.

362

■

CHAPTER 17

■

ARRAYS AND POINTERS

RETURNING POINTERS

Sample program
// search1.cpp
// A filter to output all lines containing a given
// pattern. The function strstr() is called.
// Call:
search1 [ < text.dat ]
// ---------------------------------------------------#include 
using namespace std;
#define MAXL 200
// Maximum length of line
namespace MyScope
{
// Self-defined version of function strstr():
char *strstr( const char *str, const char *patt);
}
char line[500],
// For a line of text.
patt[] = "is";
// The search pattern.
int main()
{ int lineNr = 0; // As long as a line is left over:
while( cin.getline( line, MAXL))
{
++lineNr;
if( MyScope::strstr( line, patt) != NULL)
{
// If the pattern is found:
cout.width(3);
cout << lineNr << ": "
// Output the line
<< line << endl;
// number and the line
}
}
return 0;
}

// strstr.cpp
// A self-defined version of the function strstr()
// --------------------------------------------------#include 
// For strlen() and strncmp()
namespace MyScope
{
char *strstr( const char *s1, const char *s2)
{
// To search for the string s2 within s1.
int len = strlen( s2);
for( ; *s1 != '\0'; ++s1)
if( strncmp( s1, s2, len) == 0)
// s2 found?
return (char *)s1;
// Yes -> return pointer
// to this position, or
return NULL;
// else the NULL pointer.
}
}

RETURNING POINTERS

■

363

A function can return a pointer to an object. This makes sense for a function that
searches for a particular object, for example. Such a function will return either a pointer
to the required object or a NULL pointer if the object cannot be found.
The standard C library functions often use pointers as return values. For example, the
functions strcpy(), strcat(), and strstr() each return a pointer to the first character in a C string.

䊐 The Functions strcpy() and strcat()
In contrast to the example on the page entitled “Pointer versions of functions,” the standard function strcpy()has a return value. The function returns its first argument, that
is, a pointer to the target string and leads to the following:

Prototype: char* strcpy( char* s1, const char* s2);
The second parameter is a read-only pointer, since the source string is read-only.
The standard function strcat() concatenates two C strings, adding the C string
passed as the second argument to the first argument. When you call this function, make
sure that the char array for the first string is large enough to store both strings. The
return value is the first argument. The following example shows one possible implementation.

Example: char *strcat( char *s1, const char *s2 )
{
char *p = s1 + strlen(s1); // End of s1
strcpy(p, s2);
return s1;
}

䊐 Notes on the Sample Program
The program on the opposite page shows a self-defined version of the standard function
strstr(). This version was placed in the MyScope namespace to distinguish it from
the standard function.
The function strstr() searches for a given character sequence within a string. The
standard function strncmp() is used to compare two strings. This function returns zero
if the first n characters are identical.
The program uses the strstr() function to display all the lines in the text containing the letters “is" with line numbers. The exercises for this chapter contain a program
called search.cpp where you can supply a search pattern.

364

■

CHAPTER 17

■

ARRAYS AND POINTERS

ARRAYS OF POINTERS

Pointers in the array accPtr
Array

accPtr

Account objects

accPtr[0]

"Novack,..",

accPtr[1]

"Davis, ..", 2345,

1234,

accPtr[2]
accPtr[3]
accPtr[4]

Sample function with pointers to char
// The function displayError() outputs an error message
// to a corresponding error number.
// -------------------------------------------------#include 
using namespace std;
void displayError ( int errorNr)
{
static char* errorMsg[] = {
"Invalid error number",
"Error 1: Too much data ",
"Error 2: Not enough memory ",
"Error 3: No data available " };
if( errorNr < 1 || errorNr > 3)
errorNr = 0;
cerr << errorMsg[errorNr] << endl;
}

✓

NOTE
A string literal, such as “Error..." is a char pointer to the first character in the string. Thus, such a
pointer can be used to initialize another char pointer.
Due to its static declaration, the array is generated only once and remains valid until the program
ends.

ARRAYS OF POINTERS

■

365

Pointers offer various possibilities for simple and efficient handling of large amounts of
data. For example, when you are sorting objects it makes sense to define pointers to those
objects and simply place the pointers in order, instead of rearranging the actual order of
the objects in memory.

䊐 Defining Arrays of Pointers
Whenever you need a large number of pointers, you can define an array whose elements
are pointers. An array of this type is referred to as a pointer array.

Example: Account* accPtr[5];
The array accPtr contains five Account pointers accPtr[0], accPtr[1], ... ,
accPtr[4]. The individual pointers in the array can now be assigned object addresses.
Any pointers not currently in use should have the value NULL.

Example: Account save("Novack, Kim", 1111, 9999.90);
Account depo("Davis, Sammy", 2222, 1000.);
accPtr[0] = &save;
accPtr[1] = &depo;
for( int i=2; i<5; ++i) accPtr[i] = NULL;

䊐 Initialization
As usual, an initialization list is used to initialize the array. In the case of a pointer array,
the list contains either valid addresses or the value NULL.

Example: Account* accPtr[5] = { &depo, &save, NULL};
The value NULL is automatically assigned to any objects for which the list does not contain a value. This produces the same result as in the previous example.

䊐 Usage
The individual objects addressed by the pointers in an array do not need to occupy a
contiguous memory space. Normally these objects will be created and possibly destroyed
dynamically at runtime (this will be discussed in detail in a later chapter). This allows for
extremely flexible object handling. The order is defined only by the pointers.

Example: for( int i=0; i<5; ++i)
if( accPtr[i] != NULL)
accPtr[i]->display();

// To output

The function displayError() opposite displays the error message for a corresponding error number, using an array of char pointers to the error messages.

366

■

CHAPTER 17

■

ARRAYS AND POINTERS

COMMAND LINE ARGUMENTS

Sample program
// hello.cpp
// Demonstrates the command line arguments.
// Call: hello name1 name2
// ---------------------------------------------------#include 
using namespace std;
int main( int argc, char *argv[])
{
if( argc != 3 )
{
cerr << "Use: hello name1 name2" << endl;
return 1;
}
cout << "Hello " << argv[1] << '!' << endl;
cout << "Best wishes \n"
<< "\tyours " << argv[2] << endl;
return 0;
}

Example of calling the program:
hello Jeany Vivi

Screen output
Hello Jeany!
Best wishes
Yours Vivi

Array argv in memory
argv

argv[0]

"C:\...\HELLO.EXE"

argv[1]

"Vivi"

argv[2]

"Jeany"

NULL

COMMAND LINE ARGUMENTS

■

367

䊐 Arguments for a Program
When you launch a program, you can use the command line to supply additional character sequences other than the program name. These command line arguments are typically
used to govern how a program is executed or to supply the data a program will work with.

Example: copy file1 file2
In this case, the program copy is launched with the arguments file1 and file2. The
individual arguments are separated by spaces. Characters used for redirecting input and
output ( > or < ) and a following word are evaluated by the operating system and not
passed to the program. If an argument contains space or redirection characters, you must
place it in double quotes.

䊐 Parameters of the Function

main()

So far we have only used the function main() without parameters. However, if you
intend to process command line arguments, you must define parameters for main().

Example: int main(int argc, char * argv[] )
{

. . . // Function block

}

argc contains the number of arguments passed via the command line. The program
name is one of these, so argc will have a value of at least 1.
The parameter argv is an array of char pointers:
argv[0]
argv[1]
argv[2]

points to the program name (and path)
points to the first real argument, that is, the word after the program name
points to the second argument

.....
argv[argc-1]
argv[argc]

points to the last argument
is the NULL pointer

The parameters are traditionally named argc and argv although any other name could
be used.
Various operating systems, for example WINDOWS 98/00/NT and UNIX, allow you
to declare a third parameter for main(). This parameter is an array with pointers to
environment strings. The exercises for this chapter contain a program that displays the
program environment.

■

CHAPTER 17

■

exercise s

368

ARRAYS AND POINTERS

EXERCISES

For exercise 3
Index version of the standard function

strcmp()

// strcmp() compares two C strings lexicographically.
// Return value:
< 0, if str1 < str2
//
= 0, if str1 == str2
//
> 0, if str1 > str2 .
// ---------------------------------------------------int strcmp( const char str1[], const char str2[])
{
int i;
for( i=0; str1[i] == str2[i] && str1[i] != '\0'; ++i)
;
return (str1[i] - str2[i]);
}

Notes on exercise 4
The selection sort algorithm
Method
First find the smallest element in the array and exchange it with the first
element.
This procedure is repeated while i > 0 for the remainder of an array
containing array elements with an initial index of i.
Example

Original array:

100

50

30

70

40

smallest element
After the first loop:

30

50

100

70

40

second smallest element
After the second loop:

30

40

100

70

50

EXERCISES

■

369

Exercise 1
Given an array v with the following definition:
int v[] = { 10, 20, 30, 40 }, i, *pv;

What screen output is caused by the following statements?
a. for( pv = v;

pv <= v + 3;

cout << "

*pv = "

b. for( pv = v, i = 1;
cout << "

i++ )

<<

pv[i];

pv+i <= &v[3];

pv >= v;
v["

*pv;

i <= 3;

*(pv + i) = "

d. for( pv = v + 3;
cout << "

<<

pv[i] = "

c. for( pv = v, i = 0;
cout << "

pv++ )

<<

<<

pv++,i++)

*(pv + i);

--pv )

(pv - v)

<<

"] = "

<< v[pv - v];

Exercise 2
Write a program that uses the cin method get() to read a line character by
character and stores it in a char array.The line is then output in reverse order.
Use a pointer, not an index, to address the array elements.
Exercise 3
The standard function strcmp() performs a lexicographical comparison of two
C strings.The opposite page contains an index version of strcmp().The return
value is the difference between two character codes.
Write a pointer version of the function strcmp(). Call this function
str_cmp() to distinguish it from the standard function.
To test the function, use a loop to read two lines of text and output the
results of the comparison.The loop should terminate when both strings are
empty.
Exercise 4
Define and test the function selectionSort() that sorts an array of int
values in ascending order.The principle of the selection sort algorithm is shown
opposite.
Arguments:
Return values:

An int array and its length
None

Develop both an index version and a pointer version.Test the functions with
random numbers between -10000 and +10000.

370

■

CHAPTER 17

ARRAYS AND POINTERS

Notes on exercise 5
Sample environment strings for DOS/Windows
. . .
COMSPEC=C:\COMMAND.COM
PATH=C:\WINDOWS;C:\WINDOWS\COMMAND;C:\DOS;D:\TOOLS;
PROMPT=$p$g
TEMP=C:\TEMP
. . .

Frequency table for exercise 7

Bloodpressure

<120

120–129 130–139

140–149

>= 160

Age
20–29

25

34

26

12

8

30–39

19

27

24

11

4

40–49

6

15

35

36

18

EXERCISES

■

371

Exercise 5
a. Write a program that outputs its own name and all command line arguments, each in a separate line.
b. Now extend the program to output its own environment.The environment is a memory area containing strings in the format
NAME=String

A third parameter for the function main() allows access to the environment.
This parameter is an array of pointers just like argv.The array elements are
char pointers to the environment strings, the last element being a NULL
pointer.

Exercise 6
A sample filter program called search1, which outputs lines and the relevant
line numbers for lines containing the search pattern "ei", was introduced in this
chapter.
Modify the program to produce a useful tool called search, to which you can
pass any search pattern via the command line.The program should issue an
error message and terminate if the command line does not contain a search
string. Use the standard function strstr().
Sample call:
search

Shanghai

< news.txt

Exercise 7
The following frequency was observed during an examination of the relationship
between age and blood pressure for 300 males.
Write a function that calculates the sums of the rows and columns in an int
matrix with three rows and five columns. Store the sums of the rows and
columns separately in a one-dimensional row or column array.
Arguments:
Return value:

The matrix, the row array, and the column array.
The sum of all the matrix elements.

To test the function, output the matrix, as shown in the graphic opposite
along with the computed sums in your main function.

■

CHAPTER 17

■

solutions

372

ARRAYS AND POINTERS

SOLUTIONS

Exercise 1
Screen Output:
a.
b.
c.
d.

*pv = 10

*pv = 20

*pv = 30

pv[i] = 20

pv[i] = 30

pv[i] = 40

*(pv+i) = 10

*(pv+i) = 30

v[3] = 40

v[2] = 30

v[1] = 20

*pv = 40

v[0] = 10

Exercise 2
// ------------------------------------------------------// reverse.cpp
// Exercise on pointer arithmetic:
// Reads a line and outputs the line in reverse order.
// ------------------------------------------------------#include 
using namespace std;
#define MAXLEN 80
int main()
{
char line[MAXLEN], *p;
cout << "Enter a line of text: " << endl;
// Input
for( p =
p <
++p
;

a line:
line;
line+MAXLEN
)

&&

cin.get(*p)

// Output the line in reverse order:
while( --p >= line)
cout << *p;
cout << endl;
return 0;
}

&&

*p != '\n';

SOLUTIONS

■

Exercise 3
// ------------------------------------------------------// str_cmp.cpp
// Define and test the pointer version str_cmp()
// of the standard function strcmp().
// ------------------------------------------------------#include 
using namespace std;
#define MAXLEN 100
// Maximum length of C strings
// Prototype:
int str_cmp( const char* str1, const char* str2);
int main()
// Test str_cmp()
{
char text1[MAXLEN], text2[MAXLEN];
cout << "Testing the function str_cmp()" << endl;
while( true)
{
cout << "Enter two lines of text!\n"
"End with two empty lines.\n" << endl;
cout << "1. line: ";
cin.sync(); cin.clear(); cin.get(text1,MAXLEN);
cout << "2. line: ";
cin.sync(); cin.clear(); cin.get(text2,MAXLEN);
if( text1[0] == '\0' && text2[0] == '\0')
break;
// Both lines empty.
int cmp = str_cmp( text1, text2);
if( cmp < 0)
cout << "The 1st string is smaller!\n";
else if( cmp == 0)
cout << "Both strings are equal!\n";
else
cout << "The 1st string is greater!\n";
cout << endl;
}
return 0;
}
// -----------------------------------------------------// Function str_cmp()
// Pointer version of the standard function strcmp().
// -----------------------------------------------------int str_cmp( const char* str1, const char* str2)
{
for( ; *str1 == *str2 && *str1 != '\0'; ++str1, ++str2)
;
return (*str1 - *str2);
}

373

374

■

CHAPTER 17

ARRAYS AND POINTERS

Exercise 4
//
//
//
//
//

------------------------------------------------------selSort.cpp
Implement the selection sort algorithm
for int-arrays.
-------------------------------------------------------

#include 
#include 
#include 
#include 
using namespace std;

// For srand(), rand()
// For time()

// Prototype:
void selectionSort( int arr[], int len);
const int len = 200;
int intArr[len];
int main()
{
cout << "\n
<< endl;

***

// int-array

Selection Sort Algorithm

***\n"

// To initialize an int-array with random numbers:
srand( (unsigned int)time(NULL)); // Initialize the
// random number generator.
for( int n=0; n < len; ++n)
intArr[n] = (rand() % 20000)-10000;
// To sort the numbers
selectionSort( intArr, len);
// To output the numbers
cout << "The sorted numbers:" << endl;
for( int i = 0; i < len; ++i)
cout << setw(8) << intArr[i];
cout << endl;
return 0;
}
inline void swap( int& a, int& b)
{
int temp = a; a = b; b = temp;
}

SOLUTIONS

■

// Index version:
/*
void selectionSort( int arr[], int len)
{
register int j, mini;
// Indices
for( int i = 0; i < len-1; ++i)
{
mini = i;
// Search for minimum
for( j = i+1; j < len; ++j) // starting with index i.
if( arr[mini] > arr[j])
mini = j;
swap( arr[i], arr[mini]);

// Swap.

}
}
*/
// Pointer version:
void selectionSort( int *arr, int len)
{
register int *p, *minp;
// Pointer to array elements,
int *last = arr + len-1; // pointer to the last element
for( ; arr < last; ++arr)
{
minp = arr;
// Search for minimum
for( p = arr+1; p <= last; ++p) // starting with arr
if( *minp > *p)
minp = p;
swap( *arr, *minp);
}
}

// Swap.

375

376

■

CHAPTER 17

ARRAYS AND POINTERS

Exercise 5
//
//
//
//
//

------------------------------------------------------args.cpp
The program outputs the program name including the path,
command line arguments and the environment.
-------------------------------------------------------

#include 
using namespace std;
int main( int argc, char *argv[], char *env[])
{
cout << "Program: " << argv[0] << endl;
cout << "\nCommand line arguments:" << endl;
int i;
for( i = 1; i < argc; ++i)
cout << argv[i] << endl;

// Arguments

cout << "Type  to go on";
cin.get();
cout << "\nEnvironment strings:" << endl;
for( i = 0; env[i] != NULL; ++i)
cout << env[i] << endl;

// Environment

return 0;
}

Exercise 6
// ------------------------------------------------------// search.cpp
// A filter that outputs all lines containing a certain
// pattern. The standard function strstr() is called.
//
// Call:
search pattern [ < text.dat ]
//
// If no file name is passed the input is read from the
// keyboard. In this case end input with  + .
// -----------------------------------------------------#include 
#include 
// Standard functions for C strings
using namespace std;
#define MAXL 200
// Maximum length of line
char line[500];
// For a line of text.

SOLUTIONS

■

int main( int argc, char *argv[])
{
if( argc != 2)
{
cerr << "Call: search pattern [ < text.dat ]"
<< endl;
return 1;
}
int lineNr = 0;
// As long as a line exists:
while( cin.getline( line, MAXL))
{
++lineNr;
if( strstr( line, argv[1]) != NULL)
{
// If the pattern was found:
cout.width(3);
cout << lineNr << ": "
// Output the line
<< line << endl;
// number and the line
}
}
return 0;
}

Exercise 7
//
//
//
//

------------------------------------------------------matrix.cpp
To compute the sums of rows and columns in a matrix.
------------------------------------------------------

#include 
#include 
using namespace std;
// Define and initiate a two-dimensional array:
int matrix[3][5] = { { 25, 34, 26, 12, 8 },
{ 19, 27, 24, 11, 4 },
{ 6, 15, 35, 36, 18 } };
int rowsum[3];
int colsum[5];

// For the sums of the rows
// For the sums of the columns

// Prototype of function matrixsum():
int matrixsum( int arr2D[][5], int vlen,
int rsum[], int csum[]);

377

378

■

CHAPTER 17

ARRAYS AND POINTERS

int main()
{
cout << "Testing the function matrixsum().\n"
<< endl;
// Compute sums:
int totalsum =
matrixsum( matrix, 3, rowsum, colsum);
// Output matrix and sums:
cout << "The matrix with the sums "
<< "of rows and columns:\n"
<< endl;
int i,j;
for( i = 0 ; i < 3 ; ++i)
// Output rows of the
{
// matrix with row sums.
for( j = 0 ; j < 5 ; ++j)
cout << setw(8) << matrix[i][j];
cout << " | " << setw(8) << rowsum[i] << endl;
}
cout << " -------------------------------------------"
<< endl;
for( j = 0 ; j < 5 ; ++j )
cout << setw(8) << colsum[j];
cout << " | " << setw(8) << totalsum << endl;
return 0;
}
// -------------------------------------------------------int matrixsum( int v[][5], int len,
int rsum[], int csum[])
{ int ro, co;
// Row and column index
for( ro = 0 ; ro < len ; ++ro)
// To compute row sums
{
rsum[ro] = 0;
for( co = 0 ; co < 5 ; ++co)
rsum[ro] += v[ro][co];
}
for(co = 0 ; co < 5 ; ++co)
// Compute column sums
{
csum[co] = 0;
for( ro = 0 ; ro < len ; ++ro)
csum[co] += v[ro][co];
}
return (rsum[0] + rsum[1] + rsum[2]); // Total sum =
}
// sum of row sums.

chapter

18

Fundamentals of File
Input and Output
This chapter describes sequential file access using file streams. File
streams provide simple and portable file handling techniques.

379

380

■

CHAPTER 18

■

FUNDAMENTALS OF FILE INPUT AND OUTPUT

FILES

File operations

Main Memory

File Buffer

External Memory

Write
File
Read

FILES

■

381

When a program is terminated, the program data stored in main memory is lost. To store
data permanently, you need to write that data to a file on an external storage medium.

䊐 File Operations
Single characters or character strings can be written to text files just like they can be output on screen. However, it is common practice to store records in files. A record contains
data that forms a logical unit, such as the human resource information for a person. A
write operation stores a record in a file, that is, the existing record in the file is updated or
a new record is added. When you read a record, this record is taken from the file and
copied to the data structure of a program.
Objects can be put into permanent storage using similar techniques. However, this
normally involves more than just storing an object’s data. You also need to ensure that
the object can be correctly reconstructed when it is read, and this in turn involves storing type information and references to other objects.
External mass storage media, such as hard disks, are normally block-oriented—that is,
data is transferred in blocks whose size is a multiple of 512 bytes. Efficient and easy file
management thus implies putting the data you need to store into temporary storage in
main memory, in a so-called file buffer.

䊐 File Positions
From the viewpoint of a C++ program, a file is simply a long byte array. The structure of
the file, using records for example, is entirely the programmer’s responsibility, allowing
for a maximum degree of flexibility.
Every character in a file occupies a byte position. The first byte occupies position 0,
the second byte position 1, and so on. The current file position is the position of the byte
that will be read or written next. Each byte that is transferred automatically increases the
current file position by 1.
In the case of sequential access, the data is read or written byte by byte in a fixed order.
The first read operation starts at the beginning of the file. If you need access to some
piece of information in a file, you must read the file content from start to finish. Write
operations can create a new file, overwrite an existing file, or append new data to an
existing file.
Easy access to given data in a file implies being able to set the current file position as
required. This technique is known as random file access and will be discussed in one of the
following chapters.

382

■

CHAPTER 18

■

FUNDAMENTALS OF FILE INPUT AND OUTPUT

FILE STREAMS

Stream classes for file access

ios

istream

ostream

iostream

ifstream

ofstream

fstream

FILE STREAMS

■

383

C++ provides various standard classes for file management. These so-called file stream
classes allow for easy file handling. As a programmer you will not need to concern yourself with file buffer management or system specifics.
Since the file stream classes have been standardized, you can use them to develop
portable C++ programs. One program can thus process files on a Windows NT or UNIX
platform. You simply need to recompile the program for each platform you use.

䊐 The File Stream Classes in the iostream Library
The class hierarchy on the opposite page shows that the file stream classes contain the
stream classes, with which you are already familiar, as base classes:
■
■
■

the ifstream class derives from the istream class and allows file reading
the ofstream class derives from the ostream stream class and supports writing
to files
the fstream class derives from the iostream stream class. As you would
expect, it supports both read and write operations for files.

The file stream classes are declared in the fstream header file. An object that
belongs to a file stream class is known as a file stream.

䊐 Functionality
The file stream classes inherit the functionality of their base classes. Thus, the methods,
operators, and manipulators you have already used for cin and cout are also available
here. Thus every file stream has:
■
■
■
■

methods for non-formatted writing and reading of single characters and/or data
blocks
the operators << or >> for formatted reading and writing from or to files
methods and manipulators for formatting character sequences
methods for state queries.

File handling methods, particularly methods for opening and closing files, round off the
package.

384

■

CHAPTER 18

■

FUNDAMENTALS OF FILE INPUT AND OUTPUT

CREATING FILE STREAMS

Sample program
// showfile.cpp
// Reads a text file and outputs it in pages,
// i.e. 20 lines per page.
// Call: showfile filename
// ---------------------------------------------------#include 
#include 
using namespace std;
int main( int argc, char *argv[])
{
if( argc != 2 )
// File declared?
{
cerr << "Use: showfile filename" << endl;
return 1;
}
ifstream file( argv[1]);
// Create a file stream
// and open for reading.
if( !file )
// Get status.
{
cerr << "An error occurred when opening the file "
<< argv[1] << endl;
return 2;
}
char line[80];
int cnt = 0;
while( file.getline( line, 80)) // Copy the file
{
// to standard
cout << line << endl;
// output.
if( ++cnt == 20)
{
cnt = 0;
cout << "\n\t ----  to continue ---- "
<< endl;
cin.sync(); cin.get();
}
}
if( !file.eof() )
// End-of-file occurred?
{
cerr << "Error reading the file "
<< argv[1] << endl;
return 3;
}
return 0;
}

CREATING FILE STREAMS

■

385

䊐 Opening a File
You need to open a file before you can manipulate it. To do so, you can
■
■

state the file name, which can also contain a path
define a so-called file access mode.

If the path is not explicitly stated, the file must be in the current directory. The file
access mode specifically defines whether read and/or write access to the file is permitted.
Any files still open when a program terminates are automatically closed.

䊐 File Stream Definition
You can open a file when you create a file stream—you simply state the file name to do
so. In this case default values are used for the file access mode.

Example: ifstream myfile("test.fle");
The file name test.fle is passed to the constructor of the ifstream class, which
opens the file for reading. Since the path was not stated, the file must be in the current
directory. When a file is opened, the current file position is the beginning of the file.
If you create a file stream for write-only access, the file you state need not exist. In
this case a new file is created.

Example: ofstream yourfile("new.fle");
This statement creates a new file called new.fle and opens the file for writing. But be
careful! If the file already exists, it will be truncated to a length of zero bytes, or in other
words deleted.
You can create a file stream which does not reference a specific file and use the
open() method to open a file later.

Example: ofstream yourfile;
yourfile.open("new.fle");

This example has the same effect as the previous example. More specifically, open()
uses the same default values for file access when opening a file as the default constructor
for the class.
It rarely makes sense to use fixed file names. In the case of the sample program on the
opposite page, you state the file name in the command line when you launch the program. If no file name is supplied, the program issues an error message and terminates.
Using interactive user input is another possible way to define a file name.

386

■

CHAPTER 18

■

FUNDAMENTALS OF FILE INPUT AND OUTPUT

OPEN MODES

Flags for the open mode of a file
Flag

✓

Effects

ios::in

Opens an existing file for input.

ios::out

Opens a file for output. This flag implies
ios::trunc if it is not combined with one of the
flags ios::in or ios::app or ios::ate.

ios::app

Opens a file for output at the end-of-file.

ios::trunc

An existing file is truncated to zero length.

ios::ate

Open and seek to end immediately after opening.
Without this flag, the starting position after opening is
always at the beginning of the file.

ios::binary

Perform input and output in binary mode.

NOTE
1. These flags are defined in the baseclass ios, which is common to all stream classes, and
are of the ios::openmode type.
2. By default a file is opened as a text file in so-called text mode. When you read from or
write to a text file, control characters to indicate newlines or the end-of-file are interpreted separately and adapted to the current platform (so-called “cooked mode”).When
a file is opened in binary mode, the file contents are left unchanged (the so called “raw
mode”).

Default settings when opening a file
The constructor and the method open() of all stream classes use the following default
values:

Class

Flags

ifstream

ios::in

ofstream

ios::out | ios::trunc

fstream

ios::in | ios::out

OPEN MODES

■

387

To open a file in any but the default mode, you must supply both the file name and the
open mode. This is necessary, for example, to open an existing file for write access without deleting the file.

䊐 Open Mode Flags
In addition to the file name, you can pass a second argument for the open mode to the
constructors and the open() method. The open mode is determined by using flags. A
flag represents a single bit in a computer word. If the flag is raised, the bit in question will
contain the value 1, with 0 representing all other cases.
You can use the bit operator, |, to combine various flags. Either the flag ios::in or
ios::out must be stated in all cases. If the flag ios::in is raised, the file must already
exist. If the flag ios::in is not used, the file is created, if it does not already exist.

Example: fstream addresses("Address.fle", ios::out | ios::app);
This opens a file for writing at end-of-file. The file is created, if it does not already exist.
The file will automatically grow after every write operation.
You can use the default mode for the fstream class, that is, ios::in | ios::out,
to open an existing file for reading and writing. This so-called update mode is used for
updating the information in a file and is often seen in conjunction with random file
access.

䊐 Error Handling
Errors can occur when opening a file. A user may not have the required access privileges,
or the file you want to read may not exist. The state flag failbit of the ios base class
is raised in this case. The flag can either be queried directly using the fail() method,
or indirectly by querying the status of a file stream in an if condition.

Example: if( !myfile)

// or: if( myfile.fail())

The fail bit is also set if a read or write error occurs. If a read operation fails, the end
of the current file may have been reached. To distinguish this normal behavior from a
read error, you can use the eof() method (eof = end-of-file) to query the eof bit:

Example: if( myfile.eof())

// At end-of-file?

The eof bit is set if you try to carry on reading at the end of a file. The sample program
on the previous page illustrates the potential issues.

388

■

CHAPTER 18

■

FUNDAMENTALS OF FILE INPUT AND OUTPUT

CLOSING FILES

Sample program
// fcopy1.cpp : Copies files.
// Call: fcopy1 source [ destination ]
// ---------------------------------------------------#include 
#include 
using namespace std;
inline void openerror( const char *file)
{
cerr << "Error on opening the file " << file << endl;
exit(1);
// Ends program closing
}
// all opened files.
void copy( istream& is, ostream& os);
// Prototype
int main(int argc, char *argv[])
{
if( argc < 2 || argc > 3)
{ cerr << "Call: fcopy1 source [ destination ]"
<< endl;
return 1;
// or: exit(1);
}
ifstream infile(argv[1]);
// Open 1st file
if( !infile.is_open())
openerror( argv[1]);
if( argc == 2)
// Just one sourcefile.
copy( infile, cout);
else
// Source and destination
{
ofstream outfile(argv[2]);
// Open 2nd file
if( !outfile.is_open() )
openerror( argv[2]);
copy( infile, outfile);
outfile.close();
// Unnecessary.
}
infile.close();
// Unnecessary.
return 0;
}
void copy( istream& is, ostream& os) // Copy it to os.
{
char c;
while( is.get(c) )
os.put(c);
// or: os << c ;
}

CLOSING FILES

■

389

䊐 Motivation
After you have completed file manipulation, the file should always be closed for the following reasons:
■
■

data may be lost, if for some reason the program is not terminated correctly
there is a limit to the number of files that a program can open simultaneously.

A program that terminates correctly will automatically close any open files before exiting. A file stream destructor will also close a file referenced by a stream. However, if the
file is no longer in use before this point, you should close the file explicitly.

䊐 Methods close() and is_open()
Each of the file stream classes contains a definition of a void type method called
close(), which is used to close the file belonging to the stream.

Example: myfile.close();
However, the file stream continues to exist. It is therefore possible to use the stream
to open and manipulate another file.
If you are not sure whether a file stream is currently accessing a file, you can always
perform a test using the is_open() method . In the case of the myfile file stream, the
test is as follows:

Example: if( myfile.is_open() )
{ /* . . . */ }

// File is open

䊐 The exit() Function
Open files are also closed when you call the global function exit(). The actual reason
for using this function is to terminate a program in an orderly manner and return an error
code to the calling process.

Prototype: void exit( int status );
The calling process, to which the status error code is passed for evaluation, will
often be the command interpreter—a Unix shell, for example. Successful program execution normally produces the error code 0. The statement return n; is thus equivalent
to the statement exit(n); when used in the main() function.
The program on the opposite page copies a file stated in the command line. If the user
forgets to state a second (target) file, the source file is copied to standard output. In this
case, the source file will need to be a text file.

390

■

CHAPTER 18

■

FUNDAMENTALS OF FILE INPUT AND OUTPUT

READING AND WRITING BLOCKS

Sample program
// Pizza_W.cpp
// Demonstrating output of records block by block.
// --------------------------------------------------#include 
#include 
using namespace std;
char header[] =
"
* * * P I Z Z A P R O N T O * * *\n\n";
// Record structure:
struct Pizza { char name[32]; float price; };
const int MAXCNT = 10;
Pizza pizzaMenu[MAXCNT] =
{
{ "Pepperoni", 9.90F },
{ "White Pizza", 15.90F },
{ "Ham Pizza", 12.50F }, { "Calzone", 14.90F } };
int cnt = 4;
char pizzaFile[256] = "pizza.fle";
int main()
{
cout << header

// To write records.
<< endl;

// To write data into the file:
int exitCode = 0;
ofstream outFile( pizzaFile, ios::out|ios::binary );
if( !outFile)
{
cerr << "Error opening the file!" << endl;
exitCode = 1;
}
else
{
for( int i = 0; i < cnt; ++i)
if( !outFile.write( (char*)&pizzaMenu[i],
sizeof(Pizza)) )
{ cerr << "Error writing!" << endl;
exitCode = 2;
}
}
if( exitCode == 0)
cout << "\nData has been added to file "
<< pizzaFile << "\n" << endl;
return exitCode;
}

READING AND WRITING BLOCKS

■

391

The file stream classes can use all the public operations defined in their base classes.
This means you can write formatted or unformatted data to a file or read that data from
the file block by block or character by character.

䊐 Formatted and Unformatted Input and Output
The previous sample programs illustrated how to use the methods get(), getline(),
and put() to read or write data from or to text files. Formatted input and output of
numerical values, for example, requires the >> and << operators and appropriate manipulators or formatting methods.

Example: double price = 12.34;
ofstream textFile("Test.txt");
textFile << "Price: " << price << "Dollar" << endl;

The file Test.txt will contain a line of text, such as "Price ... " that exactly
matches the screen output.
Converting binary data to legible text is not practicable if you are dealing with large
amounts of data. It makes sense to write the data for a series of measurements to a binary
file in the order in which they occur in the program. To do so, you simply open the file
in binary mode and write the data to the file, or read it from the file, block by block.

䊐 Transferring Data Blocks
The ostream method write() transfers given number of bytes from main memory to a
file.

Prototype: ostream& write( const char *buf, int n);
Since write() returns a reference to the stream, you can check to ensure that the write
operation was successful.

Example: if( ! fileStream.write("An example ", 2) )
cerr << "Error in writing!" << endl;

A warning is issued if an error occurs while writing the characters "An". You can use the
read() method to read data blocks from the file. The method transfers a data block
from a file to a program buffer.

Prototype: istream& read( char *buf, int n);
The methods read() and write() are often used for files with fixed length records.
The block that needs to be transferred can contain one or more records. The buffer in
main memory is either a structure variable or an array with elements belonging to the
structure type. You need to cast the address of this memory area to (char *) as shown
in the example opposite.

392

■

CHAPTER 18

■

FUNDAMENTALS OF FILE INPUT AND OUTPUT

OBJECT PERSISTENCE

Class Account
// Class Account with methods read() and write()
// --------------------------------------------------class Account
{
private:
string name;
// Account holder
unsigned long nr;
// Account number
double balance;
// Balance of account
public:
. . .
// Constructors, destructor,
// access methods, ...
ostream& Account::write(ostream& os) const;
istream& Account::read(istream& is)
};

Implementing methods read() and write()
// write() outputs an account into the given stream os.
// Returns: The given stream.
ostream& Account::write(ostream& os) const
{
os << name << '\0';
// To write a string
os.write((char*)&nr, sizeof(nr) );
os.write((char*)&balance, sizeof(balance) );
return os;
}
// read() is the opposite function of write().
// read() inputs an account from the stream is
// and writes it into the members of the current object
istream& Account::read(istream& is)
{
getline( is, name, '\0');
// Read a string
is.read( (char*)&nr, sizeof(nr) );
is.read( (char*)&balance, sizeof(balance));
return is;
}

OBJECT PERSISTENCE

■

393

䊐 Storing Objects
Objects are created during program runtime and cleaned up before the program terminates. To avoid this volatility, you can make an object persistent, that is, you can store
the object in a file. However, you must ensure that the object can be reconstructed, as it
was, when read. This means dealing with the following issues:
■
■

Objects can contain other objects. You will generally not know how to store a
member object.
Objects can contain references to other objects. However, it does not make sense
to store pointer values in a file, as the memory addresses will change each time
you re-launch the program.

For example, the class Account on the opposite page contains the member object
name, which is a string type. As string type objects are used to handle variable
length strings, the object just contains a reference to the string. It therefore makes no
sense to save the memory content of size sizeof(name) occupied by the object name
in a file. Instead, you should write the string itself to a file.
One possible solution to this issue is to store the data to allow them to be passed to a
constructor for the class when read. Another solution involves providing methods to
allow the objects to write their own data members to files or read them from files. This
technique is normally preferable since the class can now handle data storage itself, allowing it to write internal status data while simultaneously preventing external access to
that data.

䊐 Storing Account Class Objects
The opposite page shows the Account class, with which you are already familiar. File
input and output methods have been added to the class. A file stream that references a
file opened in binary mode is passed as an argument to the methods read() and
write(). The return value is the stream in both cases, so the status can be queried
when the function is called.

Example: if( ! anAccount.write( outFile) )
cerr << "Error in writing!" << endl;

When you read an account, you can simultaneously create an empty object that the
read() method can access.

Example: if( ! anAccount.read( inFile) )
cerr << "Error in reading!" << endl;

The member object name is saved as a C string, that is, as a string terminated by the null
character, '\0'. The << operator and the function getline() are available for this
task.

■

CHAPTER 18

■

exercise s

394

FUNDAMENTALS OF FILE INPUT AND OUTPUT

EXERCISES

For exercise 1
Possible calls to the program

fcopy:

fcopy file1 file2

A file, file1, is copied to file2. If file2 already exists, it is overwritten.
fcopy file1

A file, file1, is copied to standard output, that is, to the screen if
standard output has not been redirected.
fcopy

For calls without arguments, the source and destination files are entered
in a user dialog.
More details on the istream class method read()
If is is a file stream that references a file opened for reading, the
following call
Example:

char buf[1024];
is.read(buf, 1024);

transfers the next 1024 bytes from file to the buffer buf. Provided that no
error occurs, no less than 1024 bytes will be copied unless end-of-file is
reached. In this case the fail and eof bits are set.The last block of bytes
to be read also has to be written to the destination file.The method
gcount() returns the number of bytes transferred by the last read
operation.
Example:

int nread = is.gcount();

// Number of bytes
// in last read op.

EXERCISES

■

395

Exercise 1
The sample program fcopy1, which copies a file to the screen or to a second
file, was introduced in this chapter.Write a program named fcopy to enhance
fcopy1 as follows:
■

■

■

■

If the program is launched without any arguments, it does not issue an
error message and terminate but requests user input for the names of the
source and target files. If an empty string is given as the name of the target file, that is, the Return key is pressed, the source file is displayed on
screen.
If the command line or the user dialog contains valid source and target
file names, a binary copy operation is performed.
Copy the data block by block with the read() and write() methods.
The default block size is 1024 bytes.
The copy() function returns false if an error occurs while copying and
true in all other cases.

Also refer to the notes on the opposite page.

Exercise 2
a. Modify the sample program Pizza_w.cpp in this chapter to allow the
user to add new pizza records to the four standard pizzas and store
these records on file.
b. Then write a program called Pizza_r.cpp, which displays the pizza
menu, that is, outputs the contents of the pizza file.
Exercise 3
Test the methods read() and write() in the Account class.To do so, write a
program called Account_rw.cpp that
■
■

initializes an array with account objects and stores the array in a file
reads the contents of the file to a second array and displays the accounts
in that array to allow you to check them.

Use binary mode for read or write access to the file.

396

■

CHAPTER 18

FUNDAMENTALS OF FILE INPUT AND OUTPUT

For Exercise 4
New members of class

TelList

New data members:
string filename;
bool dirty;

// File name
// true, if data is not
// stored yet.

New methods:
const string& getFilename() const;
bool setFilename( const string& fn);
bool isDirty() const;
bool load();
bool save();
bool saveAs();

// Read data from the file
// Save data.
// Save data as ...

Extended menu of the application program
* * * * *

Telephone List

* * * * *

S = Show all entries
F = Find a telephone number
A = Append an entry
D = Delete an entry
----------------------------------------O = Open a file
W = Save in the file
U = Save as ...
----------------------------------------Q = Quit the program
Your choice:

EXERCISES

■

397

Exercise 4
The program TelList, which was written as an exercise for Chapter 16, needs
to be modified to allow telephone lists to be saved in a file.
To allow this, first add the data members and methods detailed on the
opposite page to TelList.The string filename is used to store the name of
the file in use.The dirty flag is raised to indicate that the phone list has been
changed but not saved.You will need to modify the existing methods append()
and erase()to provide this functionality.
The strings in the phone list must be saved as C strings in a binary file,
allowing for entries that contain several lines.
Add the following items to the application program menu:
O = Open a file

Read a phone list previously stored in a file.
W = Save

Save the current phone list in a file.
U = Save as . . .

Save the current phone list in a new file.
Choosing one of these menu items calls one of the following methods as
applicable: load(), save() or saveAs().These methods return true for a
successful action and false otherwise.The user must be able to supply a file
name for the save() method, as the list may not have been read from a file
previously.
If the phone list has been modified but not saved, the user should be
prompted to save the current phone list before opening another file or
terminating the program.

■

CHAPTER 18

■

solutions

398

FUNDAMENTALS OF FILE INPUT AND OUTPUT

SOLUTIONS

Exercise 1
// ---------------------------------------------------// fcopy.cpp
// Copy files
// Call: fcopy [ source [ destination ] ]
// ---------------------------------------------------#include 
#include 
using namespace std;
char usage[] = "Call: fcopy [source [destination]}";
inline void openerror( const char *file)
{
cerr << "Error opening the file " << file << endl;
exit(1);
}
bool copy( istream& is, ostream& os),
ok = true;

// Prototype,
// ok flag.

int main(int argc, char *argv[])
{
char source[256] = "", dest[256] = "";
switch( argc )
{
case 1:

// No file declared
// ==> input file name.
cout << "Copying source file to "
"destination file!\n"
"Source file: ";
cin.getline( source, 256);
if( strlen(source) == 0)
{ cerr << "No source file declared!" << endl;
return 1;
}
cin.sync();
// No previous input
cout << "Destination file: ";
cin.getline( dest, 256);
break;
case 2:
// One file is declared.
strcpy( source, argv[1]);
break;
case 3:
// Source and destination files are declared.
strcpy( source, argv[1]);
strcpy( dest, argv[2]);
break;

SOLUTIONS

■

default:
// Invalid call to the program.
cerr << usage << endl;
return 2;
// or: exit(2);
}
if( strlen(dest) == 0)
// Only source file?
{
// yes ==> output to cout.
ifstream infile(source);
if( !infile )
openerror( source);
ok = copy( infile, cout);
// The file is closed by the ifstream destructor.
}
else
// Copy source to destination
{
// file in binary mode.
ifstream infile( source, ios::in | ios::binary);
if( !infile )
openerror( source);
else
{
ofstream outfile( dest, ios::out | ios::binary);
if( !outfile)
openerror( dest);
ok = copy( infile, outfile);
if( ok)
cerr << "File " << source << " to file "
<< dest <<" copied!"<< endl;
}
}
if(!ok)
{ cerr << "Error while copying!" << endl;
return 3;
}
return 0;
}
bool copy( istream& is, ostream& os) // To copy
{
// is to os.
const int BufSize = 1024;
char buf[BufSize];
do
{
is.read( buf, BufSize);
if( is.gcount() > 0)
os.write(buf, is.gcount());
}
while( !is.eof() && !is.fail() && !os.fail() );
if( !is.eof() ) return false;
else
return true;
}

399

400

■

CHAPTER 18

FUNDAMENTALS OF FILE INPUT AND OUTPUT

Exercise 2
// ---------------------------------------------------// Pizza.h
// Header file for Pizza_W.cpp and Pizza_R.cpp.
// ---------------------------------------------------#include 
#include 
#include 
using namespace std;
// Structure of a record:
struct Pizza { char name[32]; float price; };
#define MAXCNT
20
// Maximum number of pizzas
#define FILENAME "pizza.fle"
inline void header()
{
cout << "
* * *
<< endl;
}

P I Z Z A

P R O N T O

* * *\n\n"

// ---------------------------------------------------// Pizza_w.cpp
// Demonstrating blockwise writing of records.
// ---------------------------------------------------#include "Pizza.h"
Pizza pizzaMenu[MAXCNT] =
{
{ "Pepperoni", 9.90F }, { "White Pizza", 15.90F },
{ "Ham Pizza", 12.50F }, { "Calzone", 14.90F } };
int cnt = 4;
char pizzaFile[256] = FILENAME;
int main()
// Write records.
{
int i;
header();
cout << "\nOur standard offer:\n" << endl;
cout << fixed << setprecision(2);
for( i = 0; i < cnt; ++i)
cout << setw(20) << pizzaMenu[i].name
<< setw(10) << pizzaMenu[i].price << endl;
cout << "\n-----------------------------------------\n"
<< endl;

SOLUTIONS

// Input more pizzas via keyboard:
while( cnt < MAXCNT)
{
cin.sync(); cin.clear();
cout << "What pizza should be added "
<< "to the menu?\n\n" << "Name: ";
cin.getline( pizzaMenu[cnt].name, 32);
if( pizzaMenu[cnt].name[0] == '\0')
break;
cout << "Price: ";
cin >> pizzaMenu[cnt].price;
if( !cin)
cerr << "Invalid input!" << endl;
else
++cnt;
if( cnt < MAXCNT)
cout << "\n... and the next pizza!\n"
<< "Stop with .\n";
}
// Add data to the file:
int exitCode = 0;
ofstream outFile( pizzaFile, ios::out | ios::binary);
if( !outFile)
{
cerr << "Error opening the file!" << endl;
exitCode = 1;
}
else
{
for( int i = 0; i < cnt; ++i)
if( !outFile.write( (char*)&pizzaMenu[i],
sizeof(Pizza)) )
{
cerr << "Error writing to file!"
<< endl;
exitCode = 2;
}
}
if( exitCode == 0)
cout << "\nData added to file " << pizzaFile
<< ".\n" << endl;
return exitCode;
}

■

401

402

■

CHAPTER 18

//
//
//
//

FUNDAMENTALS OF FILE INPUT AND OUTPUT

---------------------------------------------------Pizza_r.cpp
Demonstrating block by block reading of records.
---------------------------------------------------

#include "Pizza.h"
char pizzaFile[256] = FILENAME;
int main()
{
header();

// Read and display records.

ifstream inFile( pizzaFile, ios::in | ios::binary);
if( !inFile)
{
cerr << "Pizza file does not exist!" << endl;
return 1;
}
Pizza onePizza;
int cnt = 0;
cout << "\n-------------------------------------------"
<< "\nThe available pizzas:\n" << endl;
cout << fixed << setprecision(2);
while( true)
if( !inFile.read( (char*)&onePizza, sizeof(Pizza)) )
break;
else
{
cout << setw(20) << onePizza.name
<< setw(10) << onePizza.price << endl;
++cnt;
}
cout << "\n------------------------------------------\n"
<< endl;
if( !inFile.eof())
{
cerr << "Error reading file!" << endl;
return 2;
}
else
cerr << "These are " << cnt << " pizzas!\n" << endl;
return 0;
}

SOLUTIONS

■

Exercise 3
//
//
//
//
//

------------------------------------------------------Account_rw.cpp
Writes an array with objects of class Account to
a file and feed the array into another array.
-------------------------------------------------------

#include "Account.h"
#include 
#include 
using namespace std;

// Definition of the class Account

Account AccTab1[3] =
{
Account("Lucky, Luke", 707070, -1200.99),
Account("Mickey, Mouse", 123000, 2500.0),
Account("Snoopy, Dog\n"
// String can contain more
"Cell #: 01771234567", 543001) // than one line.
};
Account AccTab2[3];

// Calls to default constructor

int cnt = 3;
char file[] = "account.fle";
int main()
{
int i = 0;
// --- Write accounts to file --ofstream outFile( file, ios::out | ios::binary );
if( ! outFile)
{
cerr << "Error opening file " << file
<< endl;
return 1;
}
for( i = 0; i < cnt; ++i)
if( !AccTab1[i].write(outFile) )
{
cerr << "Error writing to file " << file
<< endl;
return 2;
}
outFile.close();

403

404

■

CHAPTER 18

FUNDAMENTALS OF FILE INPUT AND OUTPUT

// --- Reads accounts from file --ifstream inFile( file, ios::out | ios::binary );
if( ! inFile)
{
cerr << "Error opening file " << file
<< endl;
return 3;
}
for( i = 0; i < cnt; ++i)
if( !AccTab2[i].read(inFile) )
{
cerr << "Error reading file " << file
<< endl;
return 4;
}
inFile.close();
// --- Displays the accounts read --cout << "The file " << file << " contains the "
<< "following accounts:" << endl;
for( i = 0; i < cnt; ++i)
AccTab2[i].display();
cout << endl;
return 0;
}

Exercise 4
//
//
//
//
//
//
//

------------------------------------------------------telList.h
A class TelList to represent a list
containing names and telephone numbers.
The methods load(), save(), and saveAs() serve for
loading and saving a telephone list.
--------- ---------------------------------------------

#ifndef _TelList_
#define _TelList_
#include 
using namespace std;
#define PSEUDO -1
#define MAX 100

// Pseudo position
// Maximum number of elements

SOLUTIONS

■

// Type of a list element:
struct Element { string name, telNr; };
class TelList
{
private:
Element v[MAX];
int count;
string filename;
bool dirty;

// The array and the actual
// number of elements.
// File name
// true if data has been changed
// but not yet saved.

public:
TelList() : count(0), filename(""), dirty(false)
{}
int getCount() { return count; }
Element *retrieve( int i )
{
return (i >= 0 && i < count)? &v[i] : NULL;
}
bool append( const Element& el )
{
return append( el.name, el.telNr);
}
bool append( const string& name, const string& telNr);
bool erase( const string& name);
int search( const string& name) const;
void print() const;
int print( const string& name) const;
int getNewEntries();
const string& getFilename() const { return filename; }
bool setFilename( const string& fn)
{ if( fn.empty()
return false;
else { filename = fn; dirty = true; return true; }
}
bool isDirty() const { return dirty; }
bool load();
bool save();
bool saveAs();
};
#endif

// _TelList_

// ------------------------------------------------------// TelList.cpp
// Implements the methods of class TelList.
// ------------------------------------------------------#include "telList.h"
// Definition of class TelList
#include 
#include 
#include 
using namespace std;

405

406

■

CHAPTER 18

FUNDAMENTALS OF FILE INPUT AND OUTPUT

bool TelList::append( const string&
const string&
{
if( count < MAX
&& name.length() > 1
&& search(name) == PSEUDO)
{
v[count].name = name;
v[count].telNr = telNr;
++count;
dirty = true;
return true;
}
return false;
}

name,
telNr)
// Any space
// minimum 2 characters
// does not exist

bool TelList::erase( const string& key )
{
int i = search(key);
if( i != PSEUDO )
{
if( i != count-1)
// Copy the last element
v[i] = v[count-1];
// to position i.
--count;
dirty = true;
return true;
}
return false;
}
// -------------------------------------------------// Methods search(), print(), getNewEntries()
// are unchanged (refer to solutions of chapter 16).
// -------------------------------------------------// Methods for loading and saving the telephone list.
bool TelList::load()
{
cout << "\n--- Load the telephone list "
<< "from a file. ---" << "\nFile: ";
string file;
// Input file name.
cin.sync(); cin.clear();
// No previous input
getline( cin, file);
if( file.empty())
{
cerr << "No filename declared!" << endl;
return false;
}

SOLUTIONS

// Open the file for reading:
ifstream infile( file.c_str(), ios::in | ios::binary);
if( !infile )
{
cerr << "File " << file
<< " could not be opened!" << endl;
return false;
}
int i = 0;
while( i < MAX)
{
getline( infile, v[i].name, '\0');
getline( infile, v[i].telNr, '\0');
if( !infile)
break;
else
++i;
}
if( i == MAX)
cerr << "Max capacity " << MAX
<< " has been reached!" << endl;
else if( !infile.eof())
{
cerr << "Error reading file " << file << endl;
return false;
}
count = i;
filename = file;
dirty = false;
return true;
}
bool TelList::saveAs()
{
cout << "-- Save the telephone list in a file. --"
<< "\nFile: ";
string file;
// Input file name.
cin.sync(); cin.clear();
// No previous input
getline( cin, file);
if( !setFilename(file))
{
cerr << "No file name declared!" << endl;
return false;
}
else
return save();
}

■

407

408

■

CHAPTER 18

FUNDAMENTALS OF FILE INPUT AND OUTPUT

bool TelList::save()
{
if( filename.empty())
return saveAs();
if( !dirty)
return true;

// Save the telephone list.

ofstream outfile( filename.c_str(),
ios::out | ios::binary);
if( !outfile )
{
cerr << "File " << filename
<< " could not be opened!" << endl;
return false;
}
int i = 0;
while( i < count)
{
outfile << v[i].name << '\0';
outfile << v[i].telNr << '\0';
if( !outfile)
break;
else
++i;
}
if( i < count)
{
cerr << "Error writing to file " << filename << endl;
return false;
}
dirty = false;
return true;
}
// ------------------------------------------------------// TelList_.cpp
// Organize a telephone list with class TelList.
// ------------------------------------------------------#include "telList.h"
// Definition of class TelList
#include 
#include 
#include 
using namespace std;
inline void cls()
{
cout << "\033[2J\n"; // If ANSI control characters are
}
// not available, output new-lines.

SOLUTIONS

■

inline void go_on()
{
cout << "\n\nGo on with return! ";
cin.sync(); cin.clear();
// No previous input
while( cin.get() != '\n')
;
}
int menu();
// Enter a command
char askForSave();
// Prompt user to save.
char header[] =
"\n\n
* * * * * Telephone List * * * * *\n\n";
TelList myFriends;
// A telephone list
int main()
{
int action = 0;
// Command
string name;
// Read a name
while( action != 'Q')
{
action = menu();
cls(); cout << header << endl;
switch( action)
{
// --------------------------------------------------//
case 'S': case 'F': case 'A': case 'D':
//
unchanged (refer to the solutions of chapter 16).
// --------------------------------------------------case 'O':
// To open a file
if(myFriends.isDirty() && askForSave() == 'y')
myFriends.save();
if( myFriends.load())
cout << "Telephone list read from file "
<< myFriends.getFilename() <<"!"
<< endl;
else
cerr << "Telephone list not read!"
<< endl;
go_on();
break;
case 'U':
// Save as ...
if( myFriends.saveAs())
cout << "Telephone list has been saved in file: "
<< myFriends.getFilename() << " !" <

>=

Logical operators

!

Assignment operators
(op is a binary arithmetic
or a binary bitwise operator)

=
op=

&

|

^
()

~

<<

[]

& * -> ,
new delete

✓

Relational operators

>>

Bitwise operators
Function call, subscript operator
Other operators

NOTE
The assignment operator =, the address operator &, and the comma operator, have a predefined
meaning for each built-in type. This meaning can be changed for classes by a definition of your own.

䊐 Rules
An operator is always overloaded in conjunction with a class. The definition scope of an
operator is simply extended—the characteristics of the operator remain unchanged. The
following rules apply:
■
■
■
■

You cannot create “new operators”—that is, you can only overload existing operators.
You cannot redefine the operators for fundamental types.
You cannot change the operands of an operator. A binary operator will always be
binary and a unary operator will always be unary.
The precedence and the order of grouping operators of the same precedence
remains unchanged.

GENERALS

■

413

䊐 Overloading
An operator is said to be overloaded if it is defined for multiple types. In other words,
overloading an operator means making the operator significant for a new type.
Most operators are already overloaded for fundamental types. In the case of the
expression:

Example: a / b
the operand type determines the machine code created by the compiler for the division
operator. If both operands are integral types, an integral division is performed; in all
other cases floating-point division occurs. Thus, different actions are performed depending on the operand types involved.

䊐 Operators for Classes
In addition to defining methods, C++ offers the interesting possibility of defining the
functionality of a class by means of operators. Thus, you can overload the + operator
instead of, or in addition to, using the add() method. For the objects x and y in this
class:
is equivalent to x.add(y)

x + y

Using the overloaded operators of a class expressions of this type can be as easily defined
as for fundamental types. Expressions using operators are often more intuitive, and thus
easier to understand than expressions containing function calls.
Many operators belonging to the C++ standard library classes are already overloaded.
This applies to the string class, with which you are already familiar.

Example: string str1("Hello "), str2("Eve");
str1 += str2;
if( str2 < "Alexa") ...
cout << str1;
str2[2] = 'i';

//
//
//
//

Operator +=
Operator <
Operator <<
Operators [] and =

The tables on the opposite page show those operators that can be overloaded. Some
operators cannot be overloaded, such as the cast operators, the sizeof operator, and
the following four operators:
. ::
?:

.*

member access and scope resolution operators
conditional operator

These operators either have a fixed significance in the classes for which they are defined,
or overloading the operator makes no sense.

414

■

CHAPTER 19

■

OVERLOADING OPERATORS

OPERATOR FUNCTIONS (1)

Operators < and ++ for class DayTime
// DayTime.h
// The class DayTime containing operators < and ++ .
// --------------------------------------------------#ifndef _DAYTIME_
#define _DAYTIME_
class DayTime
{
private:
short hour, minute, second;
bool overflow;
public:
DayTime( int h = 0, int m = 0, int s = 0);
bool setTime(int hour, int minute, int second = 0);
int getHour()
const { return hour;
}
int getMinute() const { return minute; }
int getSecond() const { return second; }
int asSeconds() const
// Daytime in seconds
{ return (60*60*hour + 60*minute + second); }
bool operator<( const DayTime& t) const // compare
{
// *this and t
return asSeconds() < t.asSeconds();
}
DayTime& operator++()
// Increment seconds
{
++second;
// and handle overflow.
return *this;
}
void print() const;
};
#endif
// _DAYTIME_

Calling the Operator <
#include "DayTime.h"
. . .
DayTime depart1( 11, 11, 11), depart2(12,0,0);
. . .
if( depart1 < depart2 )
cout << "\nThe 1st plane takes off earlier!" << endl;
. . .

OPERATOR FUNCTIONS (1)

■

415

䊐 Naming Operator Functions
To overload an operator, you just define an appropriate operator function. The operator
function describes the actions to be performed by the operator. The name of an operator
function must begin with the operator keyword followed by the operator symbol.

Example: operator+
This is the name of the operator function for the + operator.
An operator function can be defined as a global function or as a class method. Generally, operator functions are defined as methods, especially in the case of unary operators.
However, it can make sense to define an operator function globally. This point will be
illustrated later.

䊐 Operator Functions as Methods
If you define the operator function of a binary operator as a method, the left operand will
always be an object of the class in question. The operator function is called for this
object. The second, right operand is passed as an argument to the method. The method
thus has a single parameter.

Example: bool operator<( const DayTime& t) const;
In this case the lesser than operator is overloaded to compare two DayTime objects. It
replaces the method isLess(), which was formerly defined for this class.
The prefix operator ++ has been overloaded in the example on the opposite page to
illustrate overloading unary operators. The corresponding operator function in this class
has no parameters. The function is called if the object a in the expression ++a is an
object of class DayTime.

䊐 Calling an Operator Function
The example opposite compares two times of day:

Example: depart1 < depart2
The compiler will attempt to locate an applicable operator function for this expression
and then call the function. The expression is thus equivalent to
depart1.operator<( depart2)

Although somewhat uncommon, you can call an operator function explicitly. The previous function call is therefore technically correct.
Programs that use operators are easier to encode and read. However, you should be
aware of the fact that an operator function should perform a similar operation to the corresponding operator for the fundamental type. Any other use can lead to confusion.

416

■

CHAPTER 19

■

OVERLOADING OPERATORS

OPERATOR FUNCTIONS(2)

Class Euro
// Euro1.h : The class Euro containing arithmetic operators.
// -------------------------------------------------------#ifndef _EURO_H_
#define _EURO_H_
#include 
// The class stringstream
#include 
using namespace std;
class Euro
{
private:
long data;
// Euros * 100 + Cents
public:
Euro( int euro = 0, int cents = 0)
{
data = 100L * (long)euro + cents;
}
Euro( double x)
{
x *= 100.0;
// Rounding,
data = (long)(x>=0.0 ? x+0.5 : x-0.5); //ex. 9.7 -> 10
}
long getWholePart() const { return data/100; }
int getCents() const { return (int)(data%100); }
double asDouble() const { return (double)data/100.0; }
string asString() const;
// Euro as string.
void print( ostream& os) const // Output to stream os.
{ os << asString() << " Euro" << endl;
}
// ---- Operator functions ---Euro operator-() const
// Negation (unary minus))
{
Euro temp;
temp.data = -data;
return temp;
}
Euro operator+( const Euro& e2) const
// Addition.
{
Euro temp;
temp.data = data + e2.data;
return temp;
}
Euro operator-( const Euro& e2) const
// Subtraction.
{ /* Analog just as operator + */ }
Euro& operator+=( const Euro& e2)
// Add Euros.
{
data += e2.data;
return *this;
}
Euro& operator-=( const Euro& e2); // Subtract euros.
{ /* Just as operator += */ }
};
// Continued on the next double page.

OPERATOR FUNCTIONS(2)

■

417

䊐 Notes on the Sample Class Euro
The opposite page shows the Euro class, which represents the new European currency.
The member data stores a given amount of euros as an integer in the format:
(integer part)*100 + Cents.

Thus data/100 returns the number of euros and data%100 the number of cents. This
technique allows for easy implementation of the arithmetic operations needed for the
Euro class.
In addition to a constructor that is passed whole euros and cents as arguments, there is
a constructor that can process a double value of euros and a standard copy constructor.

Example: Euro

e1(9,50), e2(20.07), e3(-e1);

䊐 Negation, Addition, and Subtraction
The unary operator - does not change its operand. In the previous example, e3 is thus
assigned a value of -9,50 euro, but e1 remains unchanged. The operator function is
thus a const method that creates and returns a temporary object.
The binary operators + and - do not change their operands either. Thus, the operator
functions also create temporary objects and return them with the correct values.

Example: Euro

sum = e1 + e2;

The expression e1 + e2 results in e1.operator+(e2). The return value is used to
initialize the new object, sum.

䊐 The += and -= Operators
Although the operators + and - were overloaded for the Euro class, this does not automatically mean that the operators += and -= are overloaded. Both are distinct operators
that require separate definitions. Of course, you should overload the operators to ensure
that the statements

Example: sum += e3;

and

sum = sum + e3;

produce the same results.
The binary operators += and -= change the current object, that is, the left operand. A
temporary object is not required! The expression sum += e3 represents the current
object after modification. Thus, the operator function returns a reference to *this.

418

■

CHAPTER 19

■

OVERLOADING OPERATORS

USING OVERLOADED OPERATORS

File Euro1.h continued
// Continues file Euro1.h
// -------------------------------------------------------inline string Euro::asString() const // Euro as string
{
stringstream strStream;
// Stream for conversion
long temp = data;
if( temp < 0) { strStream << '-'; temp = -temp; }
strStream << temp/100 << ','
<< setfill('0') << setw(2) << temp%100;
return strStream.str();
}
#endif
// _EURO_H_

Sample program
// Euro1_t.cpp
// Tests the operators of class Euro.
// -------------------------------------------------------#include "Euro1.h"
// Definition of the class
#include 
using namespace std;
int main()
{
cout << "* * * Testing the class Euro * * *\n" << endl;
Euro wholesale( 20,50), retail;
retail = wholesale;
// Standard assignment
retail += 9.49;
// += (Euro)9.49
cout << "Wholesale price: ";
wholesale.print(cout);
cout << "Retail price: ";
retail.print(cout);
Euro discount( 2.10);
// double-constructor
retail -= discount;
cout << "\nRetail price including discount: ";
retail.print(cout);
wholesale = 34.10;
cout << "\nNew wholesale price: ";
wholesale.print(cout);
Euro profit( retail - wholesale);
// Subtraction and
// copy constructor
cout << "\nThe profit: ";
profit.print(cout);
// Negative!
return 0;
}

USING OVERLOADED OPERATORS

■

419

䊐 Calling Operator Functions
The following expressions are valid for the operators in the Euro class.

Example: Euro

wholesale(15,30), retail,
profit(7,50), discount(1,75);
retail = wholesale + profit;
// Call: wholesale.operator+( profit)
retail -= discount;
// Call: retail.operator-=( discount)
retail += Euro( 1.49);
// Call: retail.operator+=( Euro(1.49))

These expressions contain only Euro type objects, for which operator functions have
been defined. However, you can also add or subtract int or double types. This is made
possible by the Euro constructors, which create Euro objects from int or double
types. This allows a function that expects a Euro value as argument to process int or
double values.
As the program opposite shows, the statement

Example: retail += 9.49;
is valid. The compiler attempts to locate an operator function that is defined for both the
Euro object and the double type for +=. Since there is no operator function with
these characteristics, the compiler converts the double value to Euro and calls the
existing operator function for euros.

䊐 Symmetry of Operands
The available constructors also allow you to call the operator functions of + and – with
int or double type arguments.

Example: retail = wholesale + 10;

// ok
// ok

wholesale = retail - 7.99;

The first statement is equivalent to
retail = wholesale.operator+( Euro(10));

But the following statement is invalid!

Example: retail = 10 + wholesale;

// wrong!

Since the operator function was defined as a method, the left operand must be a class
object. Thus, you cannot simply exchange the operands of the operator +. However, if
you want to convert both operands, you will need global definitions for the operator
functions.

420

■

CHAPTER 19

■

OVERLOADING OPERATORS

GLOBAL OPERATOR FUNCTIONS

Operators overloadable by methods only
The operator functions of the following operators have to be methods:
Operators

=

✓

Meaning
Assignment operator

()

Function call

[]

Subscript operator

->

Class member access

NOTE
The function call operator () is used to represent operations for objects like function calls. The
overloaded operator -> enables the use of objects in the same way as pointers.

The new Euro class
// Euro.h
// The class Euro represents a Euro with
// global operator functions implemented for + and -.
// --------------------------------------------------#ifndef _EURO_H_
#define _EURO_H_
// ....
class Euro
{
// Without operator functions for + and -.
// Otherwise unchanged, specifically with regard to
// the operator functions implemented for += and -=.
};
// ---------------------------------------------------// Global operator functions (inline)
// Addition:
inline Euro operator+( const Euro& e1, const Euro& e2)
{
Euro temp(e1);
temp += e2;
return temp;
}
// Subtraction:
inline Euro operator-( const Euro& e1, const Euro& e2)
{
Euro temp(e1);
temp -= e2;
return temp;
}
#endif
// _EURO_H_

GLOBAL OPERATOR FUNCTIONS

■

421

䊐 Operator Functions: Global or Method?
You can define an operator function as a global function instead of a method. The four
operators listed opposite are the only exceptions.
+=

-=

*=

/=

%=

These operators always require a so-called l-value as their left operand, that is, they
require an object with an address in memory.
Global operator functions are generally preferable if one of the following situations
applies:
■
■

the operator is binary and both operands are symmetrical, e.g. the arithmetic
operators + or *
the operator is to be overloaded for another class without changing that class, e.g.
the << operator for the ostream class.

䊐 Defining Global Operator Functions
The operands for a global operator function are passed as arguments to that function.
The operator function of a unary operator thus possesses a single parameter, whereas the
operator function of a binary operator has two.
The Euro class has been modified to provide a global definition of the operator functions for the operators + and -.

Example: Euro operator+(const Euro& e1, const Euro& e2);
Both operands are now peers. More specifically, conversion of int or double to Euro
is performed for both operands now. Given a Euro object net, the following expressions are valid and equivalent:

Example: net + 1.20

and

1.20 + net

They cause the following function calls:
operator+( net, 1.20)
operator+( 1.20, net)

and

However, a global function cannot access the private members of the class. The function operator+() shown opposite therefore uses the += operator, whose operator
function is defined as a method.
A global operator function can be declared as a “friend” of the class to allow it access
to the private members of that class.

422

■

CHAPTER 19

■

OVERLOADING OPERATORS

FRIEND FUNCTIONS

Class Euro with friend functions
// Euro.h
// The class Euro with operator functions
// declared as friend functions.
// --------------------------------------------------#ifndef _EURO_H_
#define _EURO_H_
// ....
class Euro
{
private:
long data;
// Euros * 100 + Cents
public:
// Constructors and other methods as before.
// Operators -(unary), +=, -= as before.
// Division Euro / double :
Euro operator/( double x)
// Division *this/x
{
// = *this * (1/x)
return (*this * (1.0/x));
}
// Global friend functions
friend Euro operator+( const Euro& e1, const Euro& e2);
friend Euro operator-( const Euro& e1, const Euro& e2);
friend Euro operator*( const Euro& e, double x)
{
Euro temp( ((double)e.data/100.0) * x) ;
return temp;
}
friend Euro operator*( double x, const Euro& e)
{
return e * x;
}
};
// Addition:
inline Euro operator+( const Euro& e1, const Euro& e2)
{
Euro temp;
temp.data = e1.data + e2.data;
return temp;
}
// Subtraction:
inline Euro operator-( const Euro& e1, const Euro& e2)
{
Euro temp;
temp.data = e1.data - e2.data;
return temp;
}
#endif
// _EURO_H_

FRIEND FUNCTIONS

■

423

䊐 The Friend Concept
If functions or individual classes are used in conjunction with another class, you may
want to grant them access to the private members of that class. This is made possible
by a friend declaration, which eliminates data encapsulation in certain cases.
Imagine you need to write a global function that accesses the elements of a numerical
array class. If you need to call the access methods of the class each time, and if these
methods perform range checking, the function runtime will increase considerably. However, special permission to access the private data members of the class can dramatically
improve the function’s response.

䊐 Declaring Friend Functions
A class can grant any function a special permit for direct access to its private members.
This is achieved by declaring the function as a friend. The friend keyword must precede the function prototype in the class definition.

Example: class A
{ // . . .
friend void globFunc( A* objPtr);
friend int B::elFunc( const A& objRef);
};

Here the global function globFunc() and the method elFunc() of class B are
declared as friend functions of class A. This allows them direct access to the private
members of class A. Since these functions are not methods of class A, the this pointer is
not available to them. To resolve this issue, you will generally pass the object the function needs to process as an argument.
It is important to note that the class itself determines who its friends are. If this were
not so, data encapsulation could easily be undermined.

䊐 Overloading Operators with Friend Functions
The operator functions for + and - in the Euro class are now defined as friend functions, allowing them direct access to the private member data.
In order to compute interest, it is necessary to multiply and divide euros by double
values. Since both the expression Euro*num and num*Euro are possible, friend functions are implemented to perform multiplications. As the example shows, friend functions can also be defined inline in a class.

424

■

CHAPTER 19

■

OVERLOADING OPERATORS

FRIEND CLASSES

Class Result
// Result.h
// The class Result to represent a measurement
// and the time the measurement was taken.
// --------------------------------------------------#ifndef _RESULT_
#define _RESULT_
#include "DayTime.h"
// Class DayTime
class Result
{
private:
double val;
DayTime time;
public:
// Constructor and access methods
friend class ControlPoint; // All methods of
};
// ControlPoint are friends.

Class ControlPoint
#include Result.h
class ControlPoint
{
private:
string name;
// Name of control point
Result measure[100];
// Table with results
// . . .
public:
// Constructor and the other methods
// . . .
// Compute static values of measurement results
// (average, deviation from mean, ...).
bool statistic(); // Can access the private
// members of measure[i].
};

FRIEND CLASSES

■

425

䊐 Declaring Friend Classes
In addition to declaring individual friend functions, you can also make entire classes
“friends” of another class. All the methods in this “friendly” class automatically become
friend functions in the class containing the friend declaration.
This technique is useful if a class is used in such close conjunction with another class
that all the methods in that class need access to the private members of the other class.
For example, the class ControlPoint uses objects of the Result class. Calculations with individual measurements are performed repeatedly. In this case, it makes sense
to declare the ControlPoint class as a friend of the Result class.

Example: class Result
{
// . . .
friend class ControlPoint;
};

It is important to note that the ControlPoint class has no influence over the fact that
it is a friend of the Result class. The Result class itself decides who its friends are and
who has access to its private members.
It does not matter whether a friend declaration occurs in the private or public
section of a class. However, you can regard a friend declaration as an extension of the
public interface. For this reason, it is preferable to place a friend declaration in the
public area of a class.

䊐 Using Friend Functions and Classes
Using friend functions and friend classes helps you to create efficient programs.
More specifically, you can utilize global friend functions where methods are not suited
to the task in hand. Some common uses are global operator functions declared as friend
functions.
However, extensive use of friend techniques diffuses the concept of data encapsulation. Allowing external functions to manipulate internal data can lead to inconsistency,
especially if a class is modified or extended in a later version. For this reason, you should
take special care when using friend techniques.

426

■

CHAPTER 19

■

OVERLOADING OPERATORS

OVERLOADING SUBSCRIPT OPERATORS

A class representing arrays
// Array_t.cpp
// A simple class to represent an array
// with range checking.
// -------------------------------------------------#include 
#include 
// For exit()
using namespace std;
#define MAX 100
class FloatArr
{
private:
float v[MAX];
// The array
public:
float& operator[](int i);
static int MaxIndex(){ return MAX-1; }
};
float& FloatArr::operator[]( int i )
{
if( i < 0 || i >= MAX )
{ cerr << "\nFloatArr: Outside of range!" << endl;
exit(1);
}
return v[i];
// Reference to i-th element.
}
int main()
{
cout << "\n An array with range checking!\n"
<< endl;
FloatArr random;
// Create array.
int i;
// An index.
// Fill with random euros:
for( i=0; i <= FloatArr::MaxIndex(); ++i)
random[i] = (rand() - RAND_MAX/2) / 100.0F;
cout << "\nEnter indices between 0 and "
<< FloatArr::MaxIndex() << "!"
<< "\n (Quit by entering invalid input)"
<< endl;
while( cout << "\nIndex: " && cin >> i )
cout << i << ". element: " << random[i];
return 0;
}

OVERLOADING SUBSCRIPT OPERATORS

■

427

䊐 Subscript Operator
The subscript operator [] is normally used to access a single array element. It is a binary
operator and thus has two operands. Given an expression such as v[i], the array name
v will always be the left operand, whereas the index i will be the right operand.
The subscript operator for arrays implies background pointer arithmetic, for example,
v[i] is equivalent to *(v+i). Thus, the following restrictions apply to non-overloaded
index operators:
■
■

an operand must be a pointer—an array name, for example
the other operand must be an integral expression.

䊐 Usage in Classes
These restrictions do not apply if the index operator is overloaded for a class. You should
note, however, that the operator function is always a class method with a parameter for
the right operand. The following therefore applies:
■
■
■

the left operand must be a class object
the right operand can be any valid type
the result type is not defined.

This allows for considerable flexibility. However, your overloading should always reflect
the normal use of arrays. More specifically, the return value should be a reference to an
object.
Since an index can be of any valid type, the possibilities are unlimited. For example,
you could easily define associative arrays, that is, arrays whose elements are referenced by
strings.

䊐 Notes on the Sample Program
Range checking is not performed when you access the elements of a normal array. An
invalid index can thus lead to abnormal termination of an application program. However, you can address this issue by defining your own array classes, although this may
impact the speed of your programs.
The opposite page shows a simple array class definition for float values. The subscript operator [] has been overloaded to return a reference to the i-th array element.
However, when the array is accessed, range checking is performed to ensure that the
index falls within given boundaries. If an invalid index is found, the program issues an
error message and terminates.
The class FloatArr array has a fixed length. As we will see, variable lengths are possible using dynamic memory allocation.

428

■

CHAPTER 19

■

OVERLOADING OPERATORS

OVERLOADING SHIFT-OPERATORS FOR I/O

Declaration of the operator functions
// Euro.h : Class Euro to represent an Euro
// --------------------------------------------------#ifndef _EURO_H_
#define _EURO_H_
// ....
class Euro
{ // The class is left unchanged.
// The print() method is now superfluous.
};
// ---------------------------------------------------// Declaration of shift operators:
ostream& operator<<(ostream& os, const Euro& e);
istream& operator>>(istream& is, Euro& e);
#endif
// _EURO_H_

Definition of operator functions
// Euro_io.cpp
// Overload the shift operators
// for input/output of Euro type objects.
// --------------------------------------------------#include "Euro.h"
#include 
using namespace std;
// Output to stream os.
ostream& operator<<(ostream& os, const Euro& e)
{
os << e.asString() << " Euro";
return os;
}
// Input from stream is.
istream& operator>>(istream& is, Euro& e)
{
cout << "Euro amount (Format ...x,xx): ";
int euro = 0, cents = 0; char c = 0;
if( !(is >> euro >> c >> cents)) // Input.
return is;
if( (c != ',' && c != '.')
|| cents>=100)
// Error?
is.setstate( ios::failbit);
// Yes => Set
else
// fail bit.
e = Euro( euro, cents);
// No => Accept
return is;
// value.
}

OVERLOADING SHIFT-OPERATORS FOR I/O

■

429

When outputting a Euro class object, price, on screen, the following output statement
causes a compiler error:

Example: cout << price;
cout can only send objects to standard output if an output function has been defined for
the type in question—and this, of course, is not the case for user-defined classes.
However, the compiler can process the previous statement if it can locate a suitable
operator function, operator<<(). To allow for the previous statement, you therefore
need to define a corresponding function.

䊐 Overloading the << Operator
In the previous example, the left operand of << is the object cout, which belongs to the
ostream class. Since the standard class ostream should not be modified, it is necessary
to define a global operator function with two parameters. The right operand is a Euro
class object. Thus the following prototype applies for the operator function:

Prototype: ostream& operator<<(ostream& os, const Euro& e);
The return value of the operator function is a reference to ostream. This allows for normal concatenation of operators.

Example: cout << price << endl;

䊐 Overloading the >> Operator
The >> operator is overloaded for input to allow for the following statements.

Example: cout << "Enter the price in Euros: "
cin >> price;

The second statement causes the following call:
operator>>( cin, price);

As cin is an object of the standard istream class, the first parameter of the operator
function is declared as a reference to istream. The second parameter is again a reference to Euro.
The header file Euro.h contains only the declarations of << and >>. To allow these
functions to access the private members of the Euro class, you can add a friend declaration within the class. However, this is not necessary for the current example.

■

CHAPTER 19

■

exercise s

430

OVERLOADING OPERATORS

EXERCISES

Prefix and postfix increment
To distinguish the postfix increment operator from the prefix increment
operator, the postfix operator function has an additional parameter of type int.
Expression

✓

Operator Function Call

++obj

(Prefix)

obj.operator++()

obj++

(Postfix)

obj.operator++(0)

NOTE
The expression obj++ represents a copy of obj before incrementing.
The prefix and postfix decrement operators -- are distinguished in the same
manner.

For exercise 2: Calculating with fractions

Addition

Subtraction

Multiplication

Division

✓

a
b

+

a
b

-

a
b

*

a
b

/

c
d

=

c
d

=

c
d

=

c
d

=

a*d + b*c
b*d

a*d - b*c
b*d
a * c
b * d

a * d
b * c

NOTE
Optimized error handling for the Fraction class will be discussed in Chapter
28, “Exception Handling”

EXERCISES

■

431

Exercise 1
The < and ++ operators for the sample class DayTime were overloaded at the
beginning of this chapter. Now modify the class as follows:
■

■

■

Overload the relational operators
< > <= >= == and !=
and the shift operators
>> and << for input and output
using global operator functions.You can define these inline in the
header file.
Then overload both the prefix and postfix versions of the ++ and -operators.The operator functions are methods of the class.The -- operator decrements the time by one second.The time is not decremented
after reaching 0:0:0.
Write a main function that executes all the overloaded operators and displays their results.

Exercise 2
You are to develop a class that represents fractions and performs typical
arithmetic operations with them.
■

■

■

Use a header file called fraction.h to define the Fraction class with a
numerator and a denominator of type long.The constructor has two
parameters of type long: the first parameter (numerator) contains the
default value 0, and the second parameter (denominator) contains the
value 1. Declare operator functions as methods for - (unary), ++ and -(prefix only), +=, -=, *=, and /=.The operator functions of the binary
operators +, -, *, / and the input / output operators <<, >> are to be
declared as friend functions of the Fraction class.
Implement the constructor for the Fraction class to obtain a positive
value for the denominator at all times. If the denominator assumes a value
of 0, issue an error message and terminate the program.Then write the
operator functions.The formulae for arithmetic operations are shown
opposite.
Then write a main function that calls all the operators in the Fraction
class as a test application. Output both the operands and the results.

■

CHAPTER 19

■

solutions

432

OVERLOADING OPERATORS

SOLUTIONS

Exercise 1
//
//
//
//
//
//

---------------------------------------------------DayTime.h
Class DayTime with all relational operators,
the operators ++ and -- (prefix and postfix),
such as the operators << and >> for input/output.
----------------------------------------------------

#ifndef _DAYTIME_
#define _DAYTIME_
#include 
#include 
using namespace std;
class DayTime
{
private:
short hour, minute, second;
bool overflow, underflow;
void inc()
// private function for ++
{
++second;
if( second >= 60)
// handle overflow.
second = 0, ++minute;
if( minute >= 60)
minute = 0, ++hour;
if( hour >= 24)
hour = 0, overflow = true;
}
void dec()
// private function for -{
--second;
if( second < 0)
// handle underflow.
second = 59, --minute;
if( minute < 0)
minute = 59, --hour;
if( hour < 0)
hour = 0, underflow = true;
}
public:
DayTime( int h = 0, int m = 0, int s = 0)
{
overflow = underflow = false;
if( !setTime( h, m, s))
hour = minute = second = 0;
}

SOLUTIONS

bool setTime(int hour, int minute, int second = 0)
{
if(
hour
>= 0 && hour < 24
&& minute >= 0 && minute < 60
&& second >= 0 && second < 60 )
{
this->hour
= (short)hour;
this->minute = (short)minute;
this->second = (short)second;
return true;
}
else
return false;
}
int getHour()
const { return hour;
}
int getMinute() const { return minute; };
int getSecond() const { return second; };
int asSeconds() const
// daytime in seconds
{
return (60*60*hour + 60*minute + second);
}
DayTime& operator++()
{
inc();
return *this;
}
DayTime operator++(int)
{
DayTime temp(*this);
inc();
return temp;
}
DayTime& operator- -()
{
dec();
return *this;
}
DayTime operator- -(int)
{
DayTime temp(*this);
dec();
return temp;
}

// ++Seconds

// Seconds++

// --Seconds

// Seconds--

};
// --- Relational operators --// t1 < t2
inline bool operator<( const DayTime& t1,
const DayTime& t2)
{
return t1.asSeconds() < t2.asSeconds(); }

■

433

434

■

CHAPTER 19

OVERLOADING OPERATORS

// t1 <= t2
inline bool operator<=( const DayTime& t1,
const DayTime& t2)
{
return t1.asSeconds() <= t2.asSeconds(); }
// t1 == t2
inline bool operator==( const DayTime& t1,
const DayTime& t2)
{
return t1.asSeconds() == t2.asSeconds(); }
// t1 != t2
inline bool operator!=( const DayTime& t1,
const DayTime& t2)
{
return !(t1 == t2); }
// t1 > t2
inline bool operator>( const DayTime& t1,
const DayTime& t2)
{
return (t2 < t1); }
// t1 >= t2
inline bool operator>=(const DayTime& t1,const DayTime& t2)
{
return !(t1 < t2); }
// --- Input and Output --ostream& operator<<( ostream& os, const DayTime& t)
{
os << setfill('0')
<< setw(2) << t.getHour()
<< ':'
<< setw(2) << t.getMinute() << ':'
<< setw(2) << t.getSecond() << " Time";
os << setfill(' ');
return os;
}
istream& operator>>( istream& is, DayTime& t)
{
cout << "Enter daytime in hh:mm:ss format: ";
int hr = 0, min = 0, sec = 0;
char c1 = 0, c2 = 0;
if( !(is >> hr >> c1 >> min >> c2 >> sec))
return is;
if( c1 != ':' || c2 != ':' || ! t.setTime(hr,min,sec))
is.setstate( ios::failbit);
// Error!
// => Set fail bit.
return is;
}
#endif
// _DAYTIME_

SOLUTIONS

// ---------------------------------------------------// DayTim_t.cpp
// Testing the operators of class DayTime.
// ---------------------------------------------------#include "DayTime.h"
// Definition of the class
#include 
using namespace std;
int main()
{
DayTime cinema( 20,30);
cout << "\nThe movie starts at " << cinema << endl;
DayTime now;
cout << "What time is it now?" << endl;
if( !(cin >> now) )
cerr << "Invalid input!" << endl;
else
cout << "\nThe time is now" << now << endl;
cout << "\nThe movie has ";
if( cinema < now)
cout << "already begun!\n" << endl;
else
cout << "not yet begun!\n" << endl;
cout << "Now it is
" << now++ << endl;
cout << "After 2 seconds: " << ++now << endl;
DayTime depart(16,0);
cout << "Let's go at " << --depart << endl;
if( depart >= now )
cout << "You can ride with us!" << endl;
else
cout << "We don't have room!" << endl;
return 0;
}

■

435

436

■

CHAPTER 19

OVERLOADING OPERATORS

Exercise 2
// -----------------------------------------------------// Fraction.h
// A numerical class to represent fractions
// -----------------------------------------------------#ifndef _FRACTION_
#define _FRACTION_
#include 
#include 
using namespace std;
class Fraction
{
private:
long numerator, denominator;
public:
Fraction(long n = 0, long d = 1);
Fraction operator-() const
{
return Fraction(-numerator, denominator);
}
Fraction& operator+=(const Fraction& a)
{
numerator = a.numerator * denominator
+ numerator * a.denominator;
denominator *= a.denominator;
return *this;
}
Fraction& operator-=(const Fraction& a)
{
*this += (-a);
return *this;
}
Fraction& operator++()
{
numerator += denominator;
return *this;
}
Fraction& operator--()
{
numerator -= denominator;
return *this;
}

SOLUTIONS

friend
friend
friend
friend
friend
friend
};
#endif

Fraction
Fraction
Fraction
Fraction
ostream&
istream&

■

437

operator+(const Fraction&, const Fraction&);
operator-(const Fraction&, const Fraction&);
operator*(const Fraction&, const Fraction&);
operator/(const Fraction&, const Fraction&);
operator<< (ostream& os, const Fraction& a);
operator>> (istream& is, Fraction& a);

// ------------------------------------------------------// Fraction.cpp
// Defines methods and friend functions.
// ------------------------------------------------------#include "Fraction.h"
// Constructor:
Fraction::Fraction(long n, long d)
{
if(d == 0)
{ cerr << "\nError: Division by zero!\n";
exit(1);
}
if( n < 0 ) n = -n, d = -d;
numerator = n;
denominator = d;
}
Fraction operator+(const Fraction& a, const Fraction& b)
{
Fraction temp;
temp.denominator = a.denominator * b.denominator;
temp.numerator = a.numerator*b.denominator
+ b.numerator * a.denominator;
return temp;
}
Fraction operator-(const Fraction& a, const Fraction& b )
{
Fraction temp = a;
temp += (-b);
return temp;
}
Fraction operator*(const Fraction& a, const Fraction& b )
{
Fraction temp;
temp.numerator = a.numerator * b.numerator;
temp.denominator = a.denominator * b.denominator;
return temp;
}

438

■

CHAPTER 19

OVERLOADING OPERATORS

Fraction operator/(const Fraction& a, const Fraction& b )
{
if( b.numerator == 0)
{
cerr << "\nError: Division by zero!\n";
exit(1);
}
// To multiply a by the inverse of b:
Fraction temp;
temp.numerator = a.numerator * b.denominator;
temp.denominator = a.denominator * b.numerator;
if( temp.denominator < 0 )
temp.numerator = -temp.numerator,
temp.denominator = -temp.denominator;
return temp;
}
ostream& operator<<(ostream& os, const Fraction& a)
{
os << a.numerator << "/" << a.denominator;
return os;
}
istream& operator>>(istream& is, Fraction& a)
{
cout << "Enter a fraction:\n"
" Numerator:
";
is >> a.numerator;
cout << " Denominator != 0: ";
is >> a.denominator;
if( !is) return is;
if( a.denominator == 0)
{
cout << "\nError: The denominator is 0\n"
" New denominator != 0: ";
is >> a.denominator;
if( a.denominator == 0)
{
cerr << "\nError: Division by zero!\n"; exit(1);
}
}
if( a.denominator < 0 )
a.numerator = -a.numerator,
a.denominator= -a.denominator;
return is;
}

SOLUTIONS

//
//
//
//
//

------------------------------------------------------Fract_t.cpp
Testing the class Fraction.
Modules: Fract_t.cpp Fraction.cpp
-------------------------------------------------------

#include "Fraction.h"
int main()
{
Fraction a(1,3), b(4);
cout << "\nSome test results:\n\n";
cout << " a = " << a << endl;
cout << " b = " << b << endl;
cout
cout
cout
cout

<<
<<
<<
<<

"
"
"
"

cout << "
cout << "

a
a
a
a

+
*
/

b
b
b
b

=
=
=
=

"
"
"
"

<<
<<
<<
<<

--a = " <<
++a = " <<

(a
(a
(a
(a

+
*
/

b)
b)
b)
b)

<<
<<
<<
<<

endl;
endl;
endl;
endl;

--a << endl;
++a << endl;

a += Fraction(1,2);
cout << " a+= 1/2; a = " << a << endl;
a -= Fraction(1,2);
cout << " a-= 1/2; a = " << a << endl;
cout << "-b = " << -b << endl;
cout << "\nAnd now an input\n";
cin >> a;
cout << "\nYour input: " << a << endl;
return 0;
}

■

439

This page intentionally left blank

chapter

20

Type Conversion for
Classes
Implicit type conversion occurs in C++ when an expression cannot be
compiled directly but can be compiled after applying a conversion rule.
The programmer can stipulate how the compiler will perform implicit
type conversion for classes by defining conversion constructors and
functions.
Finally, we discuss ambiguity occurring due to type conversion and
how to avoid it.

441

442

■

CHAPTER 20

■

TYPE CONVERSION FOR CLASSES

CONVERSION CONSTRUCTORS

Possible conversions

Current
Class

Converting
Constructor

Another
Type

ConvertingFunction

Converting constructors of class Euro
// The class Euro defined in the last chapter
// contains the following conversion constructors:
Euro::Euro( int );
// int -> Euro
Euro::Euro( double );
// double -> Euro
// The following declarations are now possible:
// Conversion constructors
Euro my(100),
// int-> Euro,
your(321.41);
// double -> Euro.
my = 987.12;

// Implicit conversion:
// double -> Euro

your += 20;

// Implicit conversion:
// int -> Euro

your = Euro(999.99);

// Explicit conversion
// (constructor style)

my = (Euro)123.45;

// Explicit conversion
// (cast style)

your = my;

✓

// No conversion

NOTE
When the copy constructor performs a type conversion, a temporary object is first created and this
object is used in the assignment. The temporary object is cleaned up later.

CONVERSION CONSTRUCTORS

■

443

䊐 Possible Type Conversions
Implicit and explicit type conversion is also performed for classes in C++. As a programmer, you decide what kind of conversion is permissible. You can allow type conversion
between different classes or between classes and fundamental types.
Any type conversion involving a class is defined either
■
■

by a conversion constructor or
by a conversion function.

A conversion constructor performs type conversion by converting any given type to the
type of the current class. A conversion function performs conversion in the opposite
direction, that is, it converts an object of the current class to another type—a standard
type, for example.

䊐 Conversion Constructors
A constructor with a single parameter determines how to form an object of the new class
from the argument passed to it. For this reason, a constructor with only one parameter is
referred to as a conversion constructor. The copy constructor is an exception to this rule: it
creates an object of the same class and does not perform type conversion.
Each conversion constructor is placed by the compiler on a list of possible conversions. The standard string class contains a constructor that creates a string object
from a C string, for example.

Example: string::string( const char* );
This allows you to supply a C string as an argument wherever a string object is
required.

䊐 Calling a Conversion Constructor
Conversion constructors have already been used in several examples; for example, in the
Euro class. The compiler uses them to perform implicit and explicit type conversion.

Examples: Euro salary(8765.30);
salary += (Euro)897.1;
salary += 897.1;

// explicit
// implicit

The last statement initially causes a type mismatch. Addition is not defined for a euro
and a double value. The compiler therefore activates the conversion constructor to create a temporary Euro type object from the double value. This object is then added to
the value of the salary object.

444

■

CHAPTER 20

■

TYPE CONVERSION FOR CLASSES

CONVERSION FUNCTIONS

A converting function for the Euro class
// Euro.h : The class Euro represents a euro.
// ----------------------------------------------------// . . .
class Euro
{
private:
long data;
// Euros * 100 + Cents
public:
Euro( int euro = 0, int cents = 0);
Euro( double x);
// For conversion from Euro to double:
operator double() const { return (double)data/100.0; }
// . . . other methods as before.
};

Testing conversions
// Euro_t.cpp : Testing conversions of class Euro.
// ----------------------------------------------------#include "Euro.h"
// Definition of the class
#include 
using namespace std;
int main()
{
cout << " * * * Testing Conversions * * * \n" << endl;
Euro salary( 8888,80);
double x(0.0);
salary += 1000;
// implicit int -> Euro
salary += 0.10;
// implicit double -> Euro
x = salary;
// implicit Euro -> double
x = (double)salary;
// explicit Euro -> double
x = salary.operator double();
// also possible!
// Constructor style is also safe for built-in types:
x = double(salary);
int i = salary;
// Euro -> double -> int
// Output:
cout << " salary = " << salary << endl; // 9888,90 Euro
cout << "
x = " << x << endl;
// 9888.9
cout << "
i = " << i << endl;
// 9888
return 0;
}

CONVERSION FUNCTIONS

■

445

If you need to convert an object of the current class to another type, you must define a
conversion function to do so. This is an operator function that defines how conversion is
performed. Conversion functions are also automatically used by the compiler to perform
implicit and explicit type conversion.

䊐 Defining Conversion Functions
A conversion function is always implemented as a method of the current class. Its name
is made up of the operator keyword and the target type to convert to.

Example: operator int(void) const;
The previous statement declares a conversion function where the target type is int. You
may have noticed that the declaration of a conversion function does not contain a return
type. This is because the return type is implicitly defined by the target type in the name
of the conversion function. The target type can contain multiple keywords, such as
unsigned short or const float*.
Thus, conversion functions must be written to construct a target type object from the
current object,*this, and return the target object.
The Euro shown opposite contains a conversion function with a double target type.
In other words, the function converts a Euro type object to a floating-point number.

Example: double x = oneEuro;

// implicit

䊐 Conversion Function versus Conversion Constructor
The target type of a conversion function can also be a class. In this case, you must decide
whether it is preferable to use a conversion constructor in the target class.
If you do not want to modify the target class—perhaps because it is a standard class—
a conversion function will perform the task well.

䊐 Standard Type Conversion
In addition to user-definable type conversions, the compiler also performs standard type
conversions. In the previous example, an int variable is assigned to a euro object by this
method.

Example: int wholePart = oneEuro;
This first converts a Euro object to double and then to int, that is, the cents are truncated.

446

■

CHAPTER 20

■

TYPE CONVERSION FOR CLASSES

AMBIGUITIES OF TYPE CONVERSIONS

Explicit type conversion for class Euro
// Euro.h : The class Euro represents a euro.
// ----------------------------------------------------// . . .
class Euro
{
private:
long data;
// Euros * 100 + Cents
public:
explicit Euro( int euro = 0, int cents = 0);
explicit Euro( double x);
// Converting Euro to double:
double asDouble() const { return (double)data/100.0;}
// No conversion function operator double(),
// or as previously seen.
};

Testing explicit conversions
// Euro_E_t.cpp
// Tests explicit conversion of class Euro.
// --------------------------------------------------#include "Euro_Ex.h"
// Class definition
#include 
using namespace std;
int main()
{
Euro salary( 8888.8);
double x(0.0);
/* Now impossible:
salary += 1000;
salary += 0.10;
salary = 7777.77;
x = salary;
x = (double)salary;

// double constructor

// implicit int -> Euro
// implicit double -> Euro

// implicit Euro -> double
// There is no method
// operator double().
// The following conversions are ok:
salary = Euro( 7777.77);
// explicit double -> Euro
salary += Euro(1000.10);
x = salary.asDouble();
// explicit by method
// Euro -> double
int i = salary.asDouble(); // Euro -> double -> int
return 0;

}

AMBIGUITIES OF TYPE CONVERSIONS

■

447

䊐 Type Conversion Failure
Defining a conversion function or conversion constructor can prevent you from compiling a program that is otherwise unchanged.
The Euro class contains a conversion constructor that converts a double value to
euros. This means that the following statement is valid for two objects, wholesale and
retail, of the Euro type.

Example: retail = wholesale + 46.9;
If you now additionally implement the conversion function
operator double()

that converts a euro to a double value, the previous statement can no longer be compiled. Since both conversion types double -> Euro and Euro -> double are
defined, two possible conversions could be performed:
prov2 + Euro(546.9)

// To add euros

and
double(prov2) + 546.9;

// To add values
// of type double

However, the compiler can only perform implicit type conversion if the technique is not
ambiguous. If more than one choice is available, the compiler issues an error message.

䊐 Avoiding Implicit Type Conversion
You can prevent ambiguities by stating any desired conversions explicitly. This also has
the advantage of highlighting type conversions in your source code. Moreover, undesirable type conversion, which can occur when classes are extended at a later date, can be
avoided.
In order to ensure that some kinds of type conversion are only performed explicitly,
you can use the following techniques:
■

you can use an explicit declaration for the conversion constructor. As the
example on the opposite page shows, only explicit calls to the constructor are
possible in this case.

Example: wholesale + Euro(46.9)
■

// ok

implicit type conversions by conversion functions can be prevented by not defining the function, of course. Instead you can use a method of an appropriate name,
for example asType(). Type conversion can only be performed by calling this
function explicitly.

■

CHAPTER 20

■

exercise

448

TYPE CONVERSION FOR CLASSES

EXERCISE

Method simplify() of class Fraction
// Fraction.cpp
// . . .
// To simplify fractions:
void Fraction::simplify()
{
// Divide the numerator and denominator by
// the greatest common divisor.
if( numerator == 0)
{
denominator = 1;
return;
}
// Calculating the greatest common divisor
// using an algorithm by Euclid.
long a = (numerator < 0) ? -numerator : numerator,
b = denominator,
help;
while( b != 0)
{
help = a % b; a = b; b = help;
}
// a is the greatest common divisor
numerator /= a;
denominator /= a;
}

EXERCISE

■

449

Exercise
Enhance the numerical class Fraction, which you know from the last chapter, to
convert both double values to fractions and fractions to double. In addition,
fractions should be rounded after arithmetic operations.
■

■

■

First declare the simplify() method for the Fraction class and insert
the definition on the opposite page in your source code.The method
computes the largest common divisor of numerator and denominator.
The numerator and the denominator are then divided by this value.
Add an appropriate call to the simplify() function to all operator functions (except ++ and --).
Then add a conversion constructor with a double type parameter to the
class.
Example: Fraction b(0.5);

■

■

// yields the fraction 1/2

Double values should be converted to fractions with an accuracy of three
decimal places.The following technique should suffice for numbers below
one million. Multiply the double value by 1000 and add 0.5 for rounding.
Assign the result to the numerator. Set the value of the denominator to
1000.Then proceed to simplify the fraction.
You now have a conversion constructor for long and double types.To
allow for conversion of int values to fractions, you must write your own
conversion constructor for int!
Now modify the class to allow conversion of a fraction to a double type
number. Define the appropriate conversion function inline.

Use the function main() to test various type conversions. More specifically, use
assignments and arithmetic functions to do so.Also compute the sum of a
fraction and a floating-point number.
Output the operands and the results on screen.

■

CHAPTER 20

■

solution

450

TYPE CONVERSION FOR CLASSES

SOLUTION

// ------------------------------------------------------// Fraction.h
// A numerical class to represent fractions.
// The class converts Fraction <--> double
// and simplifies fractions.
// ------------------------------------------------------#ifndef _FRACTION_
#define _FRACTION_
#include 
#include 
class Fraction
{
private: long numerator, denominator;
public:
Fraction(long z, long n);
Fraction(double x);
// double-constructor
// Default long- and int-constructor:
Fraction(long z=0) : numerator(z), denominator(1) {}
Fraction(int z)
: numerator(z), denominator(1) {}
void simplify();
operator double()
// Fraction -> double
{
return (double)numerator / (double)denominator;
}
Fraction operator-() const
{ return Fraction(-numerator, denominator);
}
Fraction& operator+=(const Fraction& a)
{
numerator = a.numerator * denominator
+ numerator * a.denominator;
denominator *= a.denominator;
simplify();
return *this;
}
Fraction& operator-=(const Fraction& a)
{
*this += (-a);
simplify();
return *this;
}
// The rest of the class including methods
//
operator++()
and
operator--()
// and friend declarations are unchanged.
};
#endif

SOLUTION

//
//
//
//
//

■

-------------------------------------------------------Fraction.cpp
Defines methods and friend functions
that are not inline.
--------------------------------------------------------

#include 
#include 
#include "Fraction.h"
// Constructors:
Fraction::Fraction(long z, long n)
{
// Unchanged! Same as in Chapter 19.
}
Fraction::Fraction( double x)
{
x *= 1000.0;
x += (x>=0.0) ? 0.5 : -0.5;
numerator = (long)x;
denominator = 1000;
simplify();
}

// Round the 4th digit.

Fraction operator+(const Fraction& a, const Fraction& b )
{
Fraction temp;
temp.denominator = a.denominator * b.denominator;
temp.numerator = a.numerator*b.denominator
+ b.numerator * a.denominator;
temp.simplify();
return temp;
}
// The functions
// operator-()
operator<<()
// are left unchanged.

operator>>()

// The functions
// operator*()
and
operator/()
// are completed by a call to temp.simplify()
// just like the function operator+().
//
// The code of method Fraction::simplify(), as
// specified in the exercise, should be here.

451

452

■

CHAPTER 20

TYPE CONVERSION FOR CLASSES

// ------------------------------------------------------// Fract_t.cpp
// Tests the class Fraction with type conversions.
// ------------------------------------------------------#include 
#include "Fraction.h"
int main()
{
Fraction a, b(-1,5), c(2.25);
double x = 0.5, y;
a = x;
// double -> Fraction
cout << "\nSome test results:\n" << endl;
cout << " a = " << a << endl;
cout << " b = " << b << endl;
cout << " c = " << c << endl;
cout << "\nThe fractions as double values:\n" << endl;
// Fraction -> double:
cout << " a = " << (double)a << endl;
cout << " b = " << (double)b << endl;
cout << " c = " << (double)c << endl;
cout
cout
cout
cout
cout

<<
<<
<<
<<
<<

"\nAnd calculate with:\n"
" a + b = " << (a + b) <<
" a - b = " << (a - b) <<
" a * b = " << (a * b) <<
" a / b = " << (a / b) <<

<< endl;
endl;
endl;
endl;
endl;

cin >> a;
// Enter a fraction.
cout << "\nYour input:
" << a << endl;
a.simplify();
cout << "\nSimplified:
" << a << endl;
cout << "\nAs double value: " << (double)a << endl;
cout << "\nEnter a floating point value: "; cin >> x;
cout << "\nThis is in fraction form:
"
<< (Fraction)x << endl;
// To calculate the sum b + x :
cout << " b = " << b << endl;
cout << " x = " << x << endl;
// a = b + x;
// Error: ambiguous!
a = b + Fraction(x);
// ok! To compute fractions.
y = (double)b + x;
// ok! To compute doubles.
cout << " b + x as fraction:
" << a << endl;
cout << " b + x as double:
" << y << endl;
return 0;
}

chapter

21

Dynamic Memory
Allocation
This chapter describes how a program can allocate and release memory
dynamically in line with current memory requirements.
Dynamic memory allocation is an important factor in many C++
programs and the following chapters will contain several additional case
studies to help you review the subject.

453

454

■

CHAPTER 21

■

DYNAMIC MEMORY ALLOCATION

THE OPERATOR new

Sample calls to new
// Dynamic objects of type long and double
// -----------------------------------------------------long *ptr_long;
ptr_long = new long;
// No initialization
// of the long object.
*ptr_long = 1234567;
// Assign a value
double *ptr_double;
double z = 1.9;
ptr_double = new double(z);

// With initialization

++(*ptr_double);
*ptr_double += *ptr_long;

// Increment the value
// ok to add long value

ptr_long = new double(2.7);

// Error: ptr_long not
// pointing to double!

On the heap
Heap

ptr_long
1234567

ptr_double
1.9

THE OPERATOR new

■

455

䊐 Dynamic Memory Allocation
When a program is compiled, the size of the data the program will need to handle is
often an unknown factor; in other words there is no way to estimate the memory requirements of the program. In cases like this you will need to allocate memory dynamically,
that is, while the program is running.
Dynamically allocated memory can be released to continually optimize memory usage
with respect to current requirements. This in turn provides a high level of flexibility,
allowing a programmer to represent dynamic data structures, such as trees and linked
lists.
Programs can access a large space of free memory known as the heap. Depending on
the operating system (and how the OS is configured), the heap can also occupy large
amounts of unused space on the hard disk by swapping memory to disk.
C++ uses the new and delete operators to allocate and release memory, and this
means that objects of any type can be created and destroyed. Let’s look at the scenario
for fundamental types first.

䊐 Calling new for Fundamental Types
The new operator is an operator that expects the type of object to be created as an argument. In its simplest form, a call to new follows this syntax

Syntax:

ptr = new type;

Where ptr is a pointer to type. The new operator creates an object of the specified
type and returns the address of that object. The address is normally assigned to a pointer
variable. If the pointer belongs to a wrong type, the compiler will issue an error message.

Example: long double *pld = new long double;
This statement allocates memory for a long double type object, that is,
sizeof(long double) bytes.
The previous call to new does not define an initial value for the new object, however,
you can supply a value in parentheses to initialize the object.

Example: pld = new long double(10000.99);
Following this statement pld points to a memory address containing a long double
type with a value of 10000.99. The statement
cout << *pld << endl;

will output this value.

456

■

CHAPTER 21

■

DYNAMIC MEMORY ALLOCATION

THE OPERATOR delete

Sample program
// DynStd.cpp
// The operators new and delete for built-in types.
// The program contains errors!
// ==> Save all data before starting.
// --------------------------------------------------#include 
using namespace std;
int main()
{
cout << "\nTesting dynamic storage allocation! "
<< endl;
// To allocate storage:
double width = 23.78;
double* ptrWidth = &width;
double* ptrLength = new double(32.54);
double* ptrArea = new double;
// To work with ptrWidth, ptrLength, and ptrArea:
*ptrArea = *ptrWidth * *ptrLength;
delete ptrLength;
// Error: The object is still
// in use!
cout << "\nWidth
: " << *ptrWidth
<< "\nLength
: " << *ptrLength
<< "\nArea
: " << *ptrArea << endl;
// To free storage:
delete ptrWidth;
delete ptrLength;
delete ptrArea;
delete ptrLength;

//
//
//
//

Error: The object has not
been dynamically reserved
ok
ok

// Error: Pointer doesn't
// address any object.

ptrLength = new double(19.45);
// ok
// To give a name to a dynamic object:
double& length = *ptrLength;
// Reference
cout << "\nNew length
: " << length
<< "\nCircumference
: " << 2 * width * length
<< endl;
return 0;
// On terminating the program
}
// allocated storage will be freed.

THE OPERATOR delete

■

457

A program should make careful use of available memory and always release memory that
is no longer needed. Failure to do so can impact the performance of your computer system. Memory that is released is available for further calls to new.

䊐 Calling

delete

Memory that has been allocated by a call to new can be released using the delete operator. A call to delete follows this syntax

Syntax:

delete ptr;

The operand ptr addresses the memory space to be released. But make sure that this
memory space was dynamically allocated by a call to new!

Example: long *pl = new long(2000000);
. . . .
delete pl;

// to work with *pl.

If you do not call delete, the dynamically allocated memory space is not released until
the program terminates.
You can pass a NULL pointer to delete when you call the operator. In this case
nothing happens and delete just returns, so you do not need to check for NULL pointers when releasing memory.
A delete expression is always a void type, so you cannot check whether memory
has been successfully released.
As the sample program illustrates, misuse of delete can be disastrous. More specifically
■
■

do not call delete twice for the same object
do not use delete to release statically allocated memory.

䊐 Error Handling for new
If there is not enough memory available, the so-called new handler is called. The new
handler is a function designed for central error handling. Thus, you do not need to design
your own error handling routines each time you call new.
The new handler is activated by default and throws an exception. Exceptions can be
caught by the program, allowing the error condition to be remedied (refer to Chapter 28,
Exception Handling). Any exception that is not caught will terminate the program,
however, you can install your own new handler.
If you are working with an older compiler, please note that new returns a NULL
pointer if not enough memory is available.

458

■

CHAPTER 21

■

DYNAMIC MEMORY ALLOCATION

DYNAMIC STORAGE ALLOCATION FOR CLASSES

Sample program
// DynObj.cpp
// The operators new and delete for classes.
// --------------------------------------------------#include "account.h"
#include 
using namespace std;
Account *clone( const Account* pK);

// Create a copy
// dynamically.

int main()
{
cout << "Dynamically created objects.\n" << endl;
// To allocate storage:
Account *ptrA1, *ptrA2, *ptrA3;
ptrA1 = new Account;
ptrA1->display();

// With default constructor
// Show default values.

ptrA1->setNr(302010);
ptrA1->setName("Tang, Ming");
ptrA1->setStand(2345.87);
ptrA1->display();

//
//
//
//

Set the other
values by access
methods.
Show new values.

// Use the constructor with three arguments:
ptrA2 = new Account("Xiang, Zhang", 7531357, 999.99);
ptrA2->display();
// Display new account.
ptrA3 = clone( ptrA1);

// Pointer to a dyna// mically created copy.
cout << "Copy of the first account: " << endl;
ptrA3->display();
// Display the copy.
delete ptrA1;
delete ptrA2;
delete ptrA3;

// Release memory

return 0;
}
Account *clone( const Account* pK)
{
return new Account(*pK);
}

// Create a copy
// dynamically.

DYNAMIC STORAGE ALLOCATION FOR CLASSES

■

459

The operators new and delete were designed to create and destroy instances of a class
dynamically. In this case, in addition to allocating memory, a suitable constructor must
be called. Before releasing memory, the destructor must be called to perform cleaning up
tasks. However, the operators new and delete ensure that this happens.

䊐 Calling

new

with a Default Constructor

A call to new for a class is not much different from a call for a fundamental type. Unless
explicitly initialized, the default constructor is called for each new object, but you must
make sure that a default constructor exists!

Example: Euro* pEuro = new Euro;
This statement allocates memory for an object of the Euro class. If enough memory is
available, the default constructor for Euro is executed and the address of a new object
returned.

䊐 Explicit Initialization
To initialize an object explicitly, you can state its initial values in parentheses when you
call new.

Syntax:

Type *ptr = new Type(initializing_list);

The values in the initialization list are passed as arguments to the constructor. If the
compiler is unable to locate a suitable constructor, an error message occurs.

Example: Euro *pE = new Euro( -123,77);
This statement assigns the address of a new Euro class object to the pointer pE. The
object is initialized using the supplied values. The expression *pE thus represents the
entire object.

Example: *pE += 200;

// To add 200 euros.

The public members are referred to via the member access operator ->.

Example: cout << pE->getCents() << endl;

// 33

䊐 Releasing Memory
When an object that was created dynamically is destroyed, the delete operator makes
sure that the object is cleaned up. The destructor is first called, and only then is the
memory space released.
As previously discussed in the section on fundamental types, when you call delete
you must ensure that the pointer is addressing a dynamic object or that you are dealing
with a NULL pointer.

460

■

CHAPTER 21

■

DYNAMIC MEMORY ALLOCATION

DYNAMIC STORAGE ALLOCATION FOR ARRAYS

Sample program
// DynArr.cpp
// Operators new[] and delete[] for dynamic arrays.
// ---------------------------------------------------#include 
#include 
using namespace std;
int main()
{
cout << "Using a dynamic array.\n" << endl;
int size = 0, cnt = 0, step = 10,
i;
float x, *pArr = NULL;
cout << "Enter some numbers!\n"
"End with q or another character " << endl;
while( cin >> x)
{
if( cnt >= size)
// Array too small?
{
// => enlarge it.
float *p = new float[size+step];
// Copy the numbers:
for( i = 0; i < size; ++i)
p[i] = pArr[i];
delete [] pArr;
// Release old array:
pArr = p; size += step;
}
pArr[cnt++] = x;
}
// Work with the numbers:
if( cnt == 0)
cout << "No invalid input!" << endl;
else
{
float sum = 0.0;
cout << "Your input: " << endl;
for( i = 0; i < cnt; i++)
// To output and
{
// add.
cout << setw(10) << pArr[i];
sum += pArr[i];
}
cout << "\nThe average: " << sum/cnt << endl;
}
delete [] pArr;
// To free the storage
return 0;
}

DYNAMIC STORAGE ALLOCATION FOR ARRAYS

■

461

Imagine you are compiling a program that will store an unknown quantity of elements in
an array. Your best option is to let the program create the array dynamically. An array of
this type is known as a dynamic array.

䊐 The new[ ] Operator
The new[ ] operator is available for creating dynamic arrays. When you call the operator, you must supply the type and quantity of the array elements.

Syntax:

vekPtr = new Type[cnt];

The pointer vekPtr will then reference the first of a total of cnt array elements.
vekPtr has to be a pointer to Type for this reason. Of course, Type can also be a class.

Example: Account *pk = new Account[256];
This statement allocates memory for 256 Account type objects and uses the default constructor to initialize them. Those objects are
pk[0],

pk[1],

. . . , pk[255],

or in pointer notation:
*pk, *(pk + 1), ....., *(pk + 255).

If the array elements are of a class type, the class must have a default constructor, since you
cannot supply an initialization list when calling new[]. Starting values for the array elements cannot be assigned until later.

䊐 The delete[ ] Operator
It is always a good idea to release the memory space occupied by a dynamic array, if the
array is no longer needed. To do so, simply call the delete[] operator. The braces []
tell the compiler to release the whole array, and not just a single array element.

Example: delete[] pk;
The operand for delete[]—the pointer pk in this case—must reference the place in
memory that was allocated by a call to new[]! The destructor belonging to the current
class is called for each array element. This shows the big difference to delete, which
would merely call the destructor for *pk, i.e. for the first array element.
The program on the opposite page stores numbers in a dynamic array. The size of the
array is adjusted as required. To do so, a newer bigger array is created, the data is copied
to the new array, and the memory occupied by the old array is released.

462

■

CHAPTER 21

■

DYNAMIC MEMORY ALLOCATION

APPLICATION: LINKED LISTS

A simple linked list

first

Info

last

Info

1st element

Info

2nd element

3rd element

Appending a list element

first

last

New last
element

Info

Info

1st element

2nd element

Info

Info
3rd element

Deleting a list element

first

Info
Removed
element

Info
New first
element

Info
Info
New second
element

APPLICATION: LINKED LISTS

■

463

䊐 Dynamic Data Structures
Now, let’s implement a linked list as a sample application. A linked list is a dynamic data
structure that allows easy insertion and deletion of data. A data structure defines how data
can be organized in units, stored, and manipulated—as arrays, lists, or trees, for example.
The type of data structure you choose has a far-reaching effect on the amount of
memory you need, the speed of access to the data involved, and the complexity (or simplicity) of the algorithms (data operations) you need.
In contract to a static data structure, whose size is known before a program is
launched, a dynamic data structure can change size while a program is running. One
example of this is an array whose size can be changed during runtime.

䊐 Defining a Linked List
Another example is a linked list that is stored in main memory and has the following
characteristics:
■
■

each list element contains a data store for the live data and a pointer to the next
element in the list
each list element—except the first and last elements—has exactly one predecessor and one successor. The first element in the list has no predecessor and the last
element no successor.

Some elementary operations are defined for linked lists, such as inserting and deleting list
elements, or searching for and retrieving the information stored in a list element.

䊐 Advantages
The storage used for the list elements need not be contiguous. The main advantage of
linked lists is:
■
■

memory for the list elements is only allocated when needed
you only need to move a pointer when inserting or deleting list elements.

When an array element is inserted or deleted, the other array elements have to be moved
to make room or fill up the “gap” in the array. If there is no room left, you need to allocate memory for a new array and copy the data to it before inserting a new element.

464

■

CHAPTER 21

■

DYNAMIC MEMORY ALLOCATION

REPRESENTING A LINKED LIST

Classes of header file List.h
// List.h
// Defines the classes ListEl and List.
// --------------------------------------------------#ifndef _LISTE_H_
#define _LISTE_H_
#include "Date.h"
// Class Date from Chapter 14
#include 
#include 
using namespace std;
class ListEl
{
private:
Date date;
double amount;
ListEl* next;

// A list element.

// Date
// Amount of money
// Pointer to successor

public:
ListEl( Date
d = Date(1,1,1), double b = 0.0,
ListEl* p = NULL)
: date(d), amount(b), next(p) {}
// Access methods:
// getDate(), setDate(), getAmount(), setAmount()
ListEl* getNext() const { return next; }
friend class List;
};
// ---------------------------------------------------// Defining the class List
class List
{
private:
ListEl* first, *last;
public:
List(){ first = last = NULL; } // Constructor
~List();
// Destructor
// Access to the first and last elements:
ListEl* front() const { return first; }
ListEl* back() const { return last; }
// Append a new element at the end of the list:
void pushBack(const Date& d, double b);
// Delete an element at the beginning of the list
void popFront();
};
#endif // _LIST_H_

REPRESENTING A LINKED LIST

■

465

䊐 Representing List Elements
You can use a recursive data structure to represent a linked list. A recursive data structure
is a data structure containing a pointer to a data structure of the same type. Of course,
the data structure cannot contain itself—that would be impossible—but it does contain a
pointer to itself.
Now let’s look at a linked list used to represent transactions in a bank account. A
transaction is characterized by a date, a sum of money, and the reason for the transaction. Thus, the list element needed to represent a transaction will contain the transaction data in its data store and a pointer to the next element in the list.
The class shown on the opposite page, ListEl, was designed to represent list elements. To keep things simple, the data store contains only the date and a sum of money.
The public declaration includes a constructor and access methods for the live data.
Later, we will be overloading the << operator in order to output the list.
It is common practice to let the pointer for the last element in the list point to NULL.
This also provides a termination criterion—the next pointer just needs to be queried for
NULL.

䊐 Representing a List
To identify a linked list, you just point a pointer at the first element in the list. You can
then use the pointer to the successor of each element to address any element in the list.
A pointer to the last element in the list is useful for appending new elements.
The opposite page shows the class definition for the List class. The private section
comprises two pointers, which reference the first and last list elements respectively. The
constructor has an easy job—it simply points both pointers to NULL, thus creating an
empty list. The destructor has a more complex task: it has to release the memory occupied
by the remaining list elements.
The pushBack() method is used to append a new element to the end of the list. To
do so, memory is allocated dynamically and the new element becomes the successor of
what was previously the last element and the last pointer is updated. In addition, the
method must deal with a special case, where the list is empty.
The popFront() method deletes the first element in the list. This involves turning
the pointer to the first element around to the second element and releasing the memory
occupied by the first element. The special case with an empty list also applies.

■

CHAPTER 21

■

exercise s

466

DYNAMIC MEMORY ALLOCATION

EXERCISES

Notes on exercise 1
Effects of the splice() function

Insert
Position
1st Array

7

3

5

9

6

2

2nd Array

Result

7

3

9

1

4

2

6

9

1

4

2

6

8

3

8

3

5

5

9

6

2

5

EXERCISES

■

467

Exercise 1
Write a global function called splice() that “splices” two int arrays together
by first allocating memory for a dynamic array with enough room for both int
arrays, and then copying the elements from both arrays to the new array, as
follows:
■
■
■

first, the elements of the first array are inserted up to a given position,
then the second array is inserted,
then the remainder of the first array is appended.

Arguments:
Return value:

The two int arrays, their length, and the position at which
they are to be spliced.
A pointer to the new array

Exercise 2
Write a global function called merge() that merges two sorted int arrays by
first allocating memory for a dynamic array with enough room for both int
arrays and then inserting the elements of both arrays into the new array in
sequence.
Arguments:
Return value:

The two int arrays and their length.
A pointer to the new array

To test the function, modify the program used to sort arrays in Exercise 4 of
Chapter 17.

Exercise 3
Complete and test the implementation of a linked list found in this chapter.
■

■

■

■

■

First define the access methods shown opposite.Then overload the <<
operator for the class ListEl to allow formatted output of the data in
the list elements.You can use the asString() in the date class to do so.
Then implement the destructor for the List class.The destructor will
release the memory used by all the remaining elements. Make sure that
you read the pointer to the successor of each element before destroying
it!
Implement the methods pushBack() and popFront() used for appending and deleting list elements.
Overload the operator << in the List class to output all the data stored
in the list.
Test the List class by inserting and deleting several list elements and
repeatedly outputting the list.

■

CHAPTER 21

■

solutions

468

DYNAMIC MEMORY ALLOCATION

SOLUTIONS

Exercise 1
// ---------------------------------------------------// Splice.cpp
// Implements the splice algorithm.
// ---------------------------------------------------#include 
#include 
#include 
// For srand() and rand()
#include 
// and for time().
using namespace std;
// Prototype:
int *splice( int v1[], int len1,
int v2[], int len2, int pos);
int main()
{
cout << "\n * * * Testing the splice function * * *\n"
<< endl;
int i, len1 = 10, len2 = 5;
int *a1 = new int[len1],
*a2 = new int[len2];
// Initialize the random number generator
// with the current time:
srand( (unsigned)time(NULL));
for( i=0; i < len1; ++i)
a1[i] = rand();
for( i=0; i < len2; ++i)
a2[i] = -rand();

// Initialize the arrays:
// with positive and
// negative numbers.

// To output the array:
cout << "1. array: " << endl;
for( i = 0; i < len1; ++i)
cout << setw(12) << a1[i];
cout << endl;
cout << "2. array: " << endl;
for( i = 0; i < len2; ++i)
cout << setw(12) << a2[i];
cout << endl;
cout << "\n At what position do you want to insert "
"\n the 2nd array into 1st array?"
"\n Possible positions: 0, 1, ..., " << len1
<< " : ";
int pos;

cin >> pos;

SOLUTIONS

■

int *a3, len3 = len1 + len2;
a3
= splice( a1, len1, a2, len2, pos);
if( a3 == NULL)
cerr << "\n Invalid position!\n" << endl;
else
{
cout << " The new spliced array: " << endl;
for( i = 0; i < len3; ++i)
cout << setw(12) << a3[i];
cout << endl;
}
delete[] a1;
delete[] a2;
delete[] a3;
return 0;
}
// ------------------------------------------------------// Function splice() inserts the array v2 into v1
// starting at position pos.
int *splice( int v1[], int len1,
int v2[], int len2, int pos)
{
if( pos < 0 || pos > len1)
return NULL;
int i = 0, i1 = 0, i2 = 0;
int *v = new int[len1+len2];
for( i = 0, i1 = 0; i1 < pos;
v[i] = v1[i1];
for( i2 = 0; i2 < len2;
v[i] = v2[i2];
for( ; i1 < len1;
v[i] = v1[i1];
return v;
}

++i, ++i1) // 1st part

++i, ++i2)

++i, ++i1)

// 2nd part

// 3rd part

469

470

■

CHAPTER 21

DYNAMIC MEMORY ALLOCATION

Exercise 2
// ---------------------------------------------------// merge.cpp
// Implements the merge algorithm.
// ---------------------------------------------------#include 
#include 
#include 
#include 
using namespace std;
// Prototypes:
void selectionSort( int arr[], int len);
int *merge( int v1[], int len1, int v2[], int len2);
int main()
{
cout << "\n * * * The Merge Algorithm * * *\n"
<< endl;
int i, len1 = 10, len2 = 20;
int *a1 = new int[len1],
*a2 = new int[len2];
// Initialize the random number generator
// with the current time:
srand( (unsigned)time(NULL));
for( i=0; i < len1; ++i)
// Initialized arrays:
a1[i] = rand();
for( i=0; i < len2; ++i)
a2[i] = rand();
selectionSort( a1, len1);
// To sort array a1.
selectionSort( a2, len2);
// To sort array a2.
// Output the arrays:
cout << "The sorted arrays:" << endl;
cout << "1st array: " << endl;
for( i = 0; i < len1; ++i)
cout << setw(12) << a1[i];
cout << endl;
cout << "2nd array: " << endl;
for( i = 0; i < len2; ++i)
cout << setw(12) << a2[i];
cout << endl;
int *a3, len3;
a3
= merge( a1, len1, a2, len2);
len3 = len1 + len2;

SOLUTIONS

■

cout << "The new merged array: " << endl;
for( i = 0; i < len3; ++i)
cout << setw(12) << a3[i];
cout << endl;
delete[] a1;
return 0;

delete[] a2;

delete[] a3;

}
// -----------------------------------------------------// Function selectionSort().
inline void swap( int& a, int& b)
// To swap.
{ int temp = a; a = b; b = temp; }
void selectionSort( int *arr, int len)
{
register int *p, *minp;
// Pointer to array elements,
int *last = arr + len-1; // Pointer to the last element
for( ; arr < last; ++arr)
{
minp = arr;
// Search minimum
for( p = arr+1; p <= last; ++p) // starting at
if( *minp > *p)
// position arr.
minp = p;
swap( *arr, *minp);
// To swap.
}
}
// -----------------------------------------------------// merge() : Merges two sorted arrays to create
//
a new sorted array.
int *merge( int v1[], int len1, int v2[], int len2)
{
int i = 0, i1 = 0, i2 = 0;
int *v = new int[len1+len2];
// New int array.
for( i=0; i1 < len1 && i2 < len2; ++i)
{
if( v1[i1] <= v2[i2])
v[i] = v1[i1++];
else
v[i] = v2[i2++];
}
if( i1 < len1)
// Copy the rest of v1 or v2.
while( i1 < len1)
v[i++] = v1[i1++];
else
while( i2 < len2)
v[i++] = v2[i2++];
return v;
}

471

472

■

CHAPTER 21

DYNAMIC MEMORY ALLOCATION

Exercise 3
//
//
//
//
//
//
//
//
//
//
//

---------------------------------------------------date.h :
Defines the class Date.
---------------------------------------------------date.cpp
Implements the methods of class Date,
which are not inline defined.
---------------------------------------------------These files are left unchanged
from Chapter 14 (solutions).

// ---------------------------------------------------// List.h
// Defines the classes ListEl and List
// to represent a linked list.
// ---------------------------------------------------#ifndef _LIST_H_
#define _LIST_H_
#include "Date.h"
#include 
#include 
using namespace std;
class ListEl
{
private:
Date date;
double amount;
ListEl* next;

// Date
// Amount of money
// Pointer to successor

public:
ListEl( Date
d = Date(1,1,1),
double b = 0.0,
ListEl* p = NULL)
: date(d), amount(b), next(p) {}
// Access methods
const Date& getDate() const { return date; }
void setDate()
// Sets the current date
{
date.setDate();
}
bool setDate( int day, int month, int year)
{
return setDate( day, month, year);
}

SOLUTIONS

double getAmount() const { return amount; }
void
setAmount(double a) { amount = a; }
ListEl* getNext() const { return next; }
friend class List;
};
// Output an element
ostream& operator<<( ostream& os, const ListEl& le);
// -----------------------------------------------------// Defines the List class
class List
{
private:
ListEl* first, *last;
public:
List(){ first = last = NULL; } // Constructor
~List();
// Destructor
// Access first and last elements:
ListEl* front() const { return first; }
ListEl* back() const { return last; }
// Appends a new element at the end of the list:
void pushBack(const Date& d, double b);
// Deletes an element at the beginning of the list.
void popFront();
};
// Outputs the list
ostream& operator<<( ostream& os, const List& le);
#endif // _LIST_H_
// ---------------------------------------------------// List.cpp
// Implements the methods of class List,
// which are not previously defined inline.
// ---------------------------------------------------#include "List.h"
// Destructor of the list:
List::~List()
{
ListEl *pEl = first, *next = NULL;
for( ; pEl != NULL; pEl = next)
{
next = pEl->next;
delete pEl;
}
}

■

473

474

■

CHAPTER 21

DYNAMIC MEMORY ALLOCATION

// Appends a new element at the end of the list:
void List::pushBack(const Date& d, double b)
{
ListEl *pEl = new ListEl( d, b);
if( last == NULL)
// List empty?
first = last = pEl;
else
last->next = pEl, last = pEl;
}
// Deletes an element from the beginning of the list.
void List::popFront()
{
if( first != NULL)
// Not empty?
{
ListEl *pEl = first;
// Save the first element.
first = first->next;
// Move to the next element.
delete pEl;
// Former first element
if( first == NULL)
// Empty now?
last = NULL;
}
}
// --- Global functions for output --// Outputs an element:
ostream& operator<<( ostream& os, const ListEl& le)
{
os << le.getDate().asString() << " Amount: ";
os << fixed << setprecision(2) << setw(10)
<< le.getAmount();
return os;
}
// Outputs the list:
ostream& operator<<( ostream& os, const List& List)
{
ListEl *pEl = List.front();
if( pEl == NULL)
os << "The list is empty!" << endl;
for( ; pEl != NULL; pEl = pEl->getNext() )
os << *pEl << endl;
return os;
}

SOLUTIONS

■

// ------------------------------------------------------// List_t.cpp
// Tests the List class.
// ------------------------------------------------------#include "List.h"
int main()
{
cout << "\n * * *
<< endl;
List

Testing the class list

aList;

cout << aList << endl;

* * *\n"

// A list
// List is still empty.

cout << "\nEnter account changes (Date and Amount)"
"\n(Type invalid input to quit, e.g. q):\n";
Date date;
int month, day, year;
char c;
double amount;
while( true)
{
cout << "Date format Month-Day-Year : ";
if( !(cin >> month >> c >> day >> c >> year)
|| ! date.setDate( month, day, year) )
break;
// Invalid date.
cout << "Account change: ";
if( !(cin >> amount) ) break;
aList.pushBack( date, amount);
}
cout << "\nContent of the list:\n";
cout << aList << endl;
cout << "\nRemoving the first element of the list:\n";
ListEl *ptrEl = ptrEl = aList.front();
if( ptrEl != NULL)
{
cout << "Deleting: " << *ptrEl << endl;
aList.popFront();
}
cout << "\nContent of the list:\n";
cout << aList << endl;
return 0;
}

475

This page intentionally left blank

chapter

22

Dynamic Members
This chapter describes how to implement classes containing pointers to
dynamically allocated memory.These include
■

your own copy constructor definition and

■

overloading the assignment operator.

A class designed to represent arrays of any given length is used as a
sample application.

477

478

■

CHAPTER 22

■

DYNAMIC MEMBERS

MEMBERS OF VARYING LENGTH

An object of class FloatArr in memory
Object
fArr
4.1 6.5 8.2 2.7

arrPtr

max:

10

cnt:

4

...

Data members of class FloatArr
// A class representing dynamic arrays of floats.
// --------------------------------------------------class FloatArr
{
private:
float* arrPtr;
// Dynamic member
int max;
// Maximum quantity without
// reallocating new storage.
int cnt;
// Number of present elements
public:
// Public methods here
};

MEMBERS OF VARYING LENGTH

■

479

䊐 Dynamic Members
You can exploit the potential of dynamic memory allocation to leverage existing classes
and create data members of variable length. Depending on the amount of data an application program really has to handle, memory is allocated as required while the application is running. In order to do this the class needs a pointer to the dynamically allocated
memory that contains the actual data. Data members of this kind are also known as
dynamic members of a class.
When compiling a program that contains arrays, you will probably not know how
many elements the array will need to store. A class designed to represent arrays should
take this point into consideration and allow for dynamically defined variable length
arrays.

䊐 Requirements
In the following section you will be developing a new version of the FloatArr class to
meet these requirements and additionally allow you to manipulate arrays as easy as fundamental types. For example, a simple assignment should be possible for two objects v1
and v2 in the new class.

Example: v2 = v1;
The object v2 itself—and not the programmer—will ensure that enough memory is
available to accommodate the array v1.
Just as in the case of fundamental types, it should also be possible to use an existing
object, v2, to initialize a new object, v3.

Example: FloatArr v3(v2);
Here the object v3 ensures that enough memory is available to accommodate the array
elements of v2.
When an object of the FloatArr is declared, the user should be able to define the
initial length of the array. The statement

Example: FloatArr fArr(100);
allocates memory for a maximum of 100 array elements.
The definition of the FloatArr class therefore comprises a member that addresses a
dynamically allocated array. In addition to this, two int variables are required to store
the maximum and current number of array elements.

480

■

CHAPTER 22

■

DYNAMIC MEMBERS

CLASSES WITH A DYNAMIC MEMBER

First version of class FloatArr
// floatArr.h : Dynamic array of floats.
// --------------------------------------------------#ifndef _FLOATARR_
#define _FLOATARR_
class FloatArr
{
private:
float* arrPtr;
// Dynamic member
int max;
// Maximum quantity without
// reallocation of new storage.
int cnt;
// Number of array elements
public:
FloatArr( int n = 256 );
// Constructor
FloatArr( int n, float val);
~FloatArr();
// Destructor
int length() const { return cnt; }
float& operator[](int i);
// Subscript operator.
float operator[](int i) const;
bool append(float val);
// Append value val.
bool remove(int pos);
// Delete position pos.
};
#endif
// _FLOATARR_

Creating objects with dynamic members
#include "floatArr.h"
#include 
using namespace std;
int main()
{
FloatArr v(10);
FloatArr w(20, 1.0F);

// Array v of 10 float values
// To initialize array w of
// 20 float values with 1.0.

v.append( 0.5F);
cout << " Current number of elements in v: "
<< v.length() << endl;
//
cout << " Current number of elements in w: "
<< w.length() << endl;
//
return 0;
}

1
20

CLASSES WITH A DYNAMIC MEMBER

■

481

The next question you need to ask when designing a class to represent arrays is what
methods are necessary and useful. You can enhance FloatArr class step by step by optimizing existing methods or adding new methods.
The first version of the FloatArr class comprises a few basic methods, which are
introduced and discussed in the following section.

䊐 Constructors
It should be possible to create an object of the FloatArr class with a given length and
store a float value in the object, if needed. A constructor that expects an int value as
an argument is declared for this purpose.
FloatArr(int n = 256);

The number 256 is the default argument for the length of the array. This provides for a
default constructor that creates an array with 256 empty array elements.
An additional constructor
FloatArr( int n, int val );

allows you to define an array where the given value is stored in each array element. In
this case you need to state the length of the array.

Example: FloatArr arr( 100, 0.0F));
This statement initializes the 100 elements in the array with a value of 0.0.

䊐 Additional Methods
The length() method allows you to query the number of elements in the array.
arr.length() returns a value of 100 for the array arr.
You can overload the subscript operator [] to access individual array elements.

Example: arr[i] = 15.0F;
The index i must lie within the range 0 to cnt-1.
The append() method can be used to append a value to the array. The number of
elements is then incremented by one.
When you call the remove() method it does exactly the opposite of append()—
deleting the element at the stated position. This reduces the current count by one, provided a valid position was stated.

482

■

CHAPTER 22

■

DYNAMIC MEMBERS

CREATING AND DESTROYING OBJECTS

Effects of the declaration FloatArr fArr( 10, 1.0F );
First, memory is allocated for the data members:
Object
fArr

arrPtr

max:

10

cnt:

10

Then storage is allocated for 10 array elements and the variables max and cnt are set to
10:
Object
fArr
?

arrPtr

max:

10

cnt:

10

?

?

?

?

?

?

?

?

?

Finally, a value of 1.0 is used to initialize the array elements:
Object
fArr
1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0

arrPtr

max:

10

cnt:

10

CREATING AND DESTROYING OBJECTS

■

483

The memory for the array elements is not contained in a FloatArr object and must be
allocated dynamically by the constructor. The object itself only occupies the memory
required for the data members arrPtr, max, and cnt. Thus, sizeof(FloatArr) is a
constant value that defaults to 12 bytes for 32 bit computers.
The additional dynamic memory allocation may need to be adjusted to meet new
requirements, for example, if an assignment is made. Finally, the memory has to be
released explicitly when an object is destroyed.

䊐 Constructing an Object
The first constructor in the FloatArr class is defined as follows:
FloatArr::FloatArr( int n )
{
max = n;
cnt = 0;
arrPtr = new float[max];
}

This allocates memory for n array elements. The current number of array elements is set
to 0.
The second constructor fills the array with the supplied value and is therefore defined
as follows:
FloatArr::FloatArr(int n, float val)
{
max = cnt = n;
arrPtr = new float[max];
for( int i=0; i < cnt; ++i)
arrPtr[i] = val;
}

The opposite page shows how memory is allocated for the object fArr and how this
object is initialized.

䊐 Destroying an Object
When an object is destroyed the dynamic memory the object occupies must be released.
Classes with dynamic members will always need a destructor to perform this task.
The FloatArr class contains a dynamic array, so memory can be released by a call to
the delete[] operator.
FloatArr::~FloatArr()
{
delete[] arrPtr;
}

484

■

CHAPTER 22

■

DYNAMIC MEMBERS

IMPLEMENTING METHODS

New version of class FloatArr
// FloatArr.cpp:
// Implementing the methods of class FloatArr.
// ----------------------------------------------------#include "floatArr.h"
#include 
using namespace std;
// Constructors and destructor as before.
// Subscript operator for objects that are not const:
float& FloatArr::operator[]( int i )
{
if( i < 0 || i >= cnt )
// Range checking
{
cerr << "\n class FloatArr: Out of range! ";
exit(1);
}
return arrPtr[i];
}
float FloatArr::operator[]( int i ) const
{
// Else as before.
}
bool FloatArr::append( float val)
{
if(cnt < max)
{
arrPtr[cnt++] = val; return true;
}
else
// Enlarge the array!
return false;
}
bool FloatArr::remove(int pos)
{
if( pos >= 0 && pos < cnt)
{
for( int i = pos; i < cnt-1; ++i)
arrPtr[i] = arrPtr[i+1];
--cnt;
return true;
}
else
return false;
}

IMPLEMENTING METHODS

■

485

䊐 Read and Write Access Using the Subscript Operator
The subscript operator can be overloaded to allow easy manipulation of array elements.

Example: FloatArr v(5, 0.0F);
v[2] = 2.2F;
for( int i=0; i < v.length(); ++i)
cout << v[i];

The operator allows both read and write access to the array elements and cannot be used
for constant objects for this reason. However, you will need to support read-only access
to constant objects.
The FloatArr class contains two versions of the operator function operator[]()
for this purpose. The first version returns a reference to the i-th array element and thus
supports write access. The second, read-only version only supports read access to the
array elements and is automatically called by the compiler when accessing constant
objects.
The implementation of these versions is identical. In both cases range checking is
performed for the index. If the index lies within the valid boundaries, an array element—
or simply a value in the case of the read-only version—is returned.

䊐 Appending and Deleting in Arrays
The FloatArr class comprises the methods append() and remove() for appending
and deleting array elements.
In the first version, the append() only works if there is at least one empty slot in the
array. In the exercises, append() is used to extend the array as required. This also
applies for a new method, insert(), which you will write as an exercise in this chapter.
When the remove() method is used to delete an element, the elements following
the deleted element move up one place, preserving the original order. The current count
is decremented by one. What was formerly the last element in the array is not deleted
but overwritten when a new element is inserted.
Another technique would be to copy the last element to the position of the element
that needs to be deleted, simply overwriting that element. Of course, this technique is
quicker and preferable for cases where the order of the elements is not significant.

486

■

CHAPTER 22

■

DYNAMIC MEMBERS

COPY CONSTRUCTOR

Effect of the standard copy constructor
FloatArr b(a);

// Creates a copy of a.

Object
a

arrPtr

max:

10

cnt:

4

4.1 6.5 8.2 2.7

Object
b

arrPtr

max:

10

cnt:

4

A self-defined copy constructor for class FloatArr
//floatArr.cpp: Implementing the methods.
// ---------------------------------------------------FloatArr::FloatArr(const FloatArr& src)
{
max = src.max;
cnt = src.cnt;
arrPtr = new float[max];
for( int i = 0; i < cnt; i++ )
arrPtr[i] = src.arrPtr[i];
}

COPY CONSTRUCTOR

■

487

䊐 Initializing with an Object
The next step is to ensure that an existing object can be used to initialize a new object.
Given an array, a, the following statement should be valid:

Example: FloatArr b(a);
The array b should now be the same length as the array a and the array elements in b
should contain the same values as in a.
The FloatArr class needs a copy constructor to perform this task. The constructor
has a reference to a constant array as a parameter.

Prototype: FloatArr( const FloatArr& );

䊐 Standard Copy Constructor
If a class does not contain a copy constructor, the compiler will automatically create a
minimal version, known as the standard copy constructor. This constructor copies the data
members of the object passed to it to corresponding data members of the new object.
A standard copy constructor is normally sufficient for a class. However, simply copying the data members would serve no useful purpose for objects containing dynamic
members. This would merely copy the pointers, meaning that the pointers of several different objects would reference the same place in memory. The diagram on the opposite
page illustrates this situation for two FloatArr class objects.
This scenario would obviously mean trouble. Imagine releasing memory allocated for
an object dynamically. The pointer for the second object would reference a memory area
that no longer existed!

䊐 Proprietary Version of the Copy Constructor
Clearly you will need to write a new copy constructor for classes with dynamic members,
ensuring that the live data and not just the pointers are copied from the dynamically
allocated memory.
The example on the opposite page shows the definition of the copy constructor for
the FloatArr class. Calling new[] creates a new array and the array elements of the
object passed to the method are then copied to that array.

488

■

CHAPTER 22

■

DYNAMIC MEMBERS

ASSIGNMENT

New declarations in class FloatArr
// FloatArr.h : Dynamic arrays of floats.
// --------------------------------------------------class FloatArr
{
private:
// . . . Data members as before
public:
// . . . Methods as before and
FloatArr(const FloatArr& src);
// Copy constructor
FloatArr& operator=( const FloatArr&); // Assignment
};

Defining the assignment
// In file floatArr.cpp
// The operator function implementing "=".
// --------------------------------------------------FloatArr& FloatArr::operator=( const FloatArr& src )
{
if( this != &src )
// No self assignments!
{
max = src.max;
cnt = src.cnt;
delete[] arrPtr;
// Release memory,
arrPtr = new float[max];
// reallocate and
for( int i=0; i < cnt; i++) // copy elements.
arrPtr[i] = src.arrPtr[i];
}
return *this;
}

Sample calls
#include "FloatArr.h"
int main()
{
FloatArr v;
FloatArr w(20, 1.0F);
const FloatArr kw(w);
v = w;
}

// Default constructor.
// Array w - 20 float values
// with initial value 1.0.
// Use copy constructor
// to create an object.
// Assignment.

ASSIGNMENT

■

489

Each class comprises four implicitly defined default methods, which you can replace with
your own definitions:
■
■

the default constructor and the destructor
the copy constructor and the standard assignment

In contrast to initialization by means of the copy constructor, which takes place when an
object is defined, an assignment always requires an existing object. Multiple assignments,
which modify an object, are possible.

䊐 Default Assignment
Given that v1 and v2 are two FloatArr class objects, the following assignment is
valid:

Example:

v1 = v2;

// Possible, but ok?

Default assignment is performed member by member. The data members of v2 are copied
to the corresponding data members of v1 just like the copy constructor would copy
them. However, this technique is not suitable for classes with dynamic members. This
would simply point the pointers belonging to different objects at the same dynamic allocated memory. In addition, memory previously addressed by a pointer of the target object
will be unreferenced after the assignment.

䊐 Overloading the Assignment Operator
In other words, you need to overload the default assignment for classes containing
dynamic members. Generally speaking, if you need to define a copy constructor, you will
also need to define an assignment.
The operator function for the assignment must perform the following tasks:
■
■

release the memory referenced by the dynamic members
allocate sufficient memory and copy the source object’s data to that memory.

The operator function is implemented as a class method and returns a reference to the
target object allowing multiple assignments. The prototype of the operator function for
the FloatArr class is thus defined as follows:
FloatArr& FloatArr::operator=( const FloatArr& src)

When implementing the operator function you must avoid self assignment, which would
read memory areas that have already been released.

■

CHAPTER 22

■

exercise s

490

DYNAMIC MEMBERS

EXERCISES

New methods of class List
// Copy constructor:
List::List(const List&);
// Assignment:
List& List::operator=( const List&);

New methods of class FloatArr
// Methods to append a float or an
// array of floats:
void append( float val);
void append( const FloatArr& v);
FloatArr& operator+=( float val);
FloatArr& operator+=( const FloatArr& v);
// Methods to insert a float or an
// array of floats:
bool insert( float val, int pos);
bool insert( const FloatArr& v, int pos );
// In any case, more memory space must be allocated
// to the array if the current capacity is
// insufficient.

EXERCISES

■

491

Exercise 1
Complete the definition of the List class, which represents a linked list and test
the class.
First, modify your test program to create a copy of a list. Call the default
assignment for the objects in the List class. Note how your program reacts.
A trial run of the program shows that the class is incomplete. Since the class
contains dynamic members, the following tasks must be performed:
■
■

Define a copy constructor for the List class.
Overload the assignment operator.

Exercise 2
Add the methods shown opposite to the FloatArr class. In contrast to the
existing method
bool append( float val);

the new method must be able to allocate more memory as required. As this
could also be necessary for other methods, write a private auxiliary function for
this purpose
void expand( int newMax );

The method must copy existing data to the newly allocated memory.
Overload the operator += so it can be used instead of calling the function
append().
The insert() method inserts a float value or a FloatArr object at
position pos. Any elements that follow pos must be pushed.
Also overload the shift operator << to output an array using the field width
originally defined to output the array elements.

✓

NOTE
The width() method in the ostream class returns the current field width, if you call the method
without any arguments.

Now add calls to the new methods to your test program and output the results
after each call.

■

CHAPTER 22

■

solutions

492

DYNAMIC MEMBERS

SOLUTIONS

Exercise 1
// ---------------------------------------------------// List.h
// Definition of classes ListEl and List
// representing a linked list.
// --------------------------------------------------#ifndef _LIST_H_
#define _LIST_H_
#include "Date.h"
#include 
#include 
using namespace std;
class ListEl
{
// Unchanged as in Chapter 21
};
// -----------------------------------------------------// Definition of class List
class List
{
private:
ListEl* first, *last;
public:
// New methods:
List(const List&);
// Copy constructor
List& operator=( const List&); // Assignment
// Otherwise unchanged from Chapter 21
};
#endif // _LIST_H_
// ---------------------------------------------------// List.cpp
// Implements those methods of class List,
// that are not defined inline.
// ---------------------------------------------------#include "List.h"
// Copy constructor:
List::List(const List& src)
{
// Appends the elements of src to an empty list.
first = last = NULL;
ListEl *pEl = src.first;
for( ; pEl != NULL; pEl = pEl->next )
pushBack( pEl->date, pEl->amount);
}

SOLUTIONS

■

// Assignment:
List& List::operator=( const List& src)
{
// Release memory for all elements:
ListEl *pEl = first,
*next = NULL;
for( ; pEl != NULL; pEl = next)
{
next = pEl->next;
delete pEl;
}
first = last = NULL;
// Appends the elements of src to an empty list.
pEl = src.first;
for( ; pEl != NULL; pEl = pEl->next )
pushBack( pEl->date, pEl->amount);
return *this;
}
// All other methods unchanged.

// ------------------------------------------------------// List_t.cpp
// Tests the class List with copy constructor and
// assignment.
// ------------------------------------------------------#include "List.h"
int main()
{
cout << "\n * * * Testing the class List * * *\n"
<< endl;
List list1;
// A list
cout << list1 << endl;
// The list is still empty.
Date date( 11,8,1999);
// Insert 3 elements.
double amount( +1234.56);
list1.pushBack( date, amount);
date.setDate( 1, 1, 2002);
amount = -1000.99;
list1.pushBack( date, amount);
date.setDate( 2, 29, 2000);
amount = +5000.11;
list1.pushBack( date, amount);

493

494

■

CHAPTER 22

DYNAMIC MEMBERS

cout << "\nThree elements have been inserted!"
"\nContent of the list:" << endl;
cout << list1 << endl;
cout << "\nPress return to continue! "; cin.get();
List list2( list1);
cout << "A copy of the 1st list has been created!\n"
"Contents of the copy:\n" << endl;
cout << list2 << endl;
cout << "\nRemove the first element from the list:\n";
ListEl *ptrEl = ptrEl = list1.front();
if( ptrEl != NULL)
{
cout << "To be deleted: " << *ptrEl << endl;
list1.popFront();
}
cout << "\nContent of the list:\n";
cout << list1 << endl;
list1 = list2;

// Reassign the copy.

cout << "The copy has been assigned to the 1st list!\n"
"Contents after assignment:\n" << endl;
cout << list1 << endl;
return 0;
}

SOLUTIONS

■

Exercise 2
// -----------------------------------------------------// floatArr.h : Dynamic arrays of floating-point numbers.
// -----------------------------------------------------#ifndef _FLOATARR_
#define _FLOATARR_
#include 
using namespace std;
class FloatArr
{
private:
float* arrPtr;
int max;
int cnt;

//
//
//
//

Dynamic member
Maximum quantity without
reallocating new storage.
Number of present array elements

void expand( int newMax);

// Helps enlarge the array

public:
// Constructors , destructor,
// assignment, subscript operator, and method length()
// as before in this chapter.
// Methods to append a floating-point number
// or an array of floating-point numbers:
void append( float val);
void append( const FloatArr& v);
FloatArr& operator+=( float val)
{
append( val);
return *this;
}
FloatArr& operator+=( const FloatArr& v)
{
append(v);
return *this;
}
// Methods to insert a floating-point number
// or an array of floating-point numbers:
bool insert( float val, int pos);
bool insert( const FloatArr& v, int pos );
bool remove(int pos);
// Delete at position pos.
// To output the array
friend ostream& operator<<( ostream& os,
const FloatArr& v);
};
#endif

// _FLOATARR_

495

496

■

CHAPTER 22

//
//
//
//

DYNAMIC MEMBERS

----------------------------------------------------FloatArr.cpp
Implements the methods of FloatArr.
-----------------------------------------------------

#include "floatArr.h"
//
//

Constructors, destructor, assignment,
and subscript operator unchanged.

//

---

The new functions

---

// Private auxiliary function to enlarge the array.
void FloatArr::expand( int new)
{
if( newMax == max)
return;
max = newMax;
if( newMax < cnt)
cnt = newMax;
float *temp = new float[newMax];
for( int i = 0; i < cnt; ++i)
temp[i] = arrPtr[i];
delete[] arrPtr;
arrPtr = temp;
}
// Append floating-point number or an array of floats.
void FloatArr::append( float val)
{
if( cnt+1 > max)
expand( cnt+1);
arrPtr[cnt++] = val;
}
void FloatArr::append( const FloatArr& v)
{
if( cnt + v.cnt > max)
expand( cnt + v.cnt);
int count = v.cnt;
for( int i=0; i < count; ++i)
arrPtr[cnt++] = v.arrPtr[i];
}

// Necessary if v == *this

SOLUTIONS

// Insert a float or an array of floats
bool FloatArr::insert( float val, int pos)
{
return insert( FloatArr(1,val), pos);
}
bool FloatArr::insert( const FloatArr& v, int pos )
{
if( pos < 0 || pos >= cnt)
return false;
// Invalid position
if( max < cnt + v.cnt)
expand(cnt + v.cnt);
int i;
for( i = cnt-1; i >= pos; --i)
arrPtr[i+v.cnt] = arrPtr[i];
for( i = 0; i < v.cnt; ++i)
arrPtr[i+pos] = v.arrPtr[i];
cnt = cnt + v.cnt;
return true;

// Shift up
// starting at pos
// Fill gap.

}
// To delete
bool FloatArr::remove(int pos)
{
if( pos >= 0 && pos < cnt)
{
for( int i = pos; i < cnt-1; ++i)
arrPtr[i] = arrPtr[i+1];
--cnt;
return true;
}
else
return false;
}
// Output the array
ostream& operator<<( ostream& os, const FloatArr& v)
{
int w = os.width();
// Save field width.
for( float *p = v.arrPtr; p < v.arrPtr + v.cnt; ++p)
{
os.width(w);
os << *p;
}
return os;
}

■

497

498

■

CHAPTER 22

DYNAMIC MEMBERS

// ----------------------------------------------------// FloatV_t.cpp
// Tests the class FloatArr.
// ----------------------------------------------------#include "FloatArr.h"
#include 
#include 
using namespace std;
int main()
{
FloatArr v(10);
FloatArr w(15, 1.0F);
cout <<
<<
cout <<
<<

// Array v for 10 float values
// Initialize the array w of
// 15 float values with 1.0.
" Current total of elements in v: "
v.length() << endl;
" Current total of elements in w: "
w.length() << endl;

float x = -5.0F;
for( ; x < 6 ; x += 1.7F)
v.append(x);

// Append values.

v += v;

// Also possible!

cout << "\nThe array elements after appending:"
<< endl;
cout << setw(5) << v << endl;
const FloatArr cv(v);

// Copy constructor
// creates const object.
cout << "\nThe copy of v has been created.\n";
cout << "\nThe array elements of the copy:\n"
<< setw(5) << v << endl;
w.remove(3);
w.append(10.0F);
w.append(20.0F);

//
//
//
//

Erase the element at
position 3.
Add a new element.
And once more!

v = w;
cout << "\nAssignment done.\n";
cout << "\nThe elements after assigning: \n"
<< setw(5) << v << endl;
v.insert( cv, 0);
cout << "\nThe elements after inserting "
" the copy at position 0: \n"
<< setw(5) << v << endl;
return 0;
}

chapter

23

Inheritance
This chapter describes how derived classes can be constructed from
existing classes by inheritance. Besides defining derived classes, we will
also discuss
■

how members are redefined

■

how objects are constructed and destroyed, and

■

how access control to base classes can be realized.

499

500

■

CHAPTER 23

■

INHERITANCE

CONCEPT OF INHERITANCE

Is relation

Car
Properties
and capacities
of class

Car

PassCar

Truck

Properties
and capacities
of class

Properties
and capacities
of class

Car

Car

Additional
properties and
capacities of class

Additional
properties and
capacities of class
Truck

PassCar

CONCEPT OF INHERITANCE

■

501

䊐 Base Classes and Derived Classes
Inheritance allows new classes to be constructed on the basis of existing classes. The new
derived class “inherits” the data and methods of the so-called base class. But you can add
more characteristics and functionality to the new class.
A fleet management program used by a car hire company needs to handle all kinds of
vehicles—automobiles, motorcycles, trucks, and so on. All of these vehicles have an
identification number that indicates the vehicle, the manufacturer, and the vehicle status, such as “hired,” “repair shop,” and so on. Additionally, operations such as “modify
status” are required for the class.
To differentiate between vehicle types, various classes are derived from the base class
Car, such as PassCar, which is used to represent passenger-carrying vehicles. This class
has additional attributes, such as the number of seats, type, sunroof (yes/no), and various
additional operations.

䊐 Is Relationship
An object of the PassCar type is a special object of the Car class. A passenger vehicle
is a special kind of car. In cases like this we can say that the derived class establishes an is
relationship to the base class.
We distinguish between this close relationship and a so-called has relationship. As
already mentioned, a has relationship occurs between two classes when an member of
one class has another class type. An Account object has a string object to represent
the name of the account holder, for example.

䊐 Data Abstraction and Reusability
Inheritance has a number of important benefits for software developers:
■

■

data abstraction: General characteristics and abilities can be handled by generic
(base) classes and specializations can be organized in hierarchical relationships by
means of derived classes. This makes it easier to manage complex situations and
relationships.
re-usability: Classes that you have defined and tested can be reused and adapted to
perform new tasks. The base class implementation need not be known for this
purpose: only the public interfaces are required.

502

■

CHAPTER 23

■

INHERITANCE

DERIVED CLASSES

Defining a derived class
class C : public B
{
private:
// Declaration of additional private
// data members and member functions
public:
// Declaration of additional public
// data members and member functions
};

Direct and indirect derivation

B

C

D

Base class B

B is a direct
base class

B is an indirect
base class

DERIVED CLASSES

■

503

When you define a derived class, the base class, the additional data members and methods, and the access control to the base class are defined.
The opposite page shows a schematic definition of a derived class, C. The C class
inherits the B class, which is defined in the public section following the colon. The
private and public sections contain additional members of the C class.

䊐 Access to Public Members in the Base Class
Access privileges to the base class B are designated by the public keyword that precedes the B. In other words,
■

all the public members in base class B are publicly available in the derived class
C.

This kind of inheritance ports the public interface of the base class to the derived
class where it is extended by additional declarations. Thus, objects of the derived class
can call the public methods of the base class. A public base class, therefore, implements the is relationship; this is quite common.
There are some less common cases where access to the members of the base class
needs to be restricted or prohibited. Only the methods of class C can still access the
public members of B, but not the users of that class. You can use private or protected derivation to achieve this (these techniques will be discussed later).

䊐 Access to Private Members of the Base Class
The private members of the base class are protected in all cases. That is,
■

the methods of the derived class cannot access the private members of the base
class.

Imagine the consequences if this were not so: you would be able to hack access to the
base class by simply defining a derived class, thus undermining any protection offered by
data encapsulation.

䊐 Direct and Indirect Base Classes
The derived class C can itself be a base class for a further class, D. This allows for class
hierarchies. Class B then becomes an indirect base class for class D.
In the graphic on the opposite page, the arrow ↑ means directly derived from. That is,
class D is a direct derivation of class C and an indirect derivation of B.

504

■

CHAPTER 23

■

INHERITANCE

MEMBERS OF DERIVED CLASSES

Base class Car and derived class PassCar
// car.h:
Definition of baseclass Car and
//
of the derived class PassCar
// -------------------------------------------------#include 
#include 
using namespace std;
class Car
// Base class
{
private:
long
nr;
string producer;
public:
// Constructor:
Car( long n = 0L, const string& prod = "");
// Access methods:
long getNr(void) const { return nr; }
void setNr( long n ) { nr = n; }
const string& getProd() const{ return producer; }
void setProd(const string& p){ producer = p; }
void display( void ) const;

// Display a car

};
class PassCar : public Car
// Derived class
{
private:
string passCarType;
bool
sunRoof;
public:
// Constructor:
PassCar( const string& tp, bool sd,
int n = 0 , const string& h = "");
// Access methods:
const string& getType() const{ return passCarType; }
void setType( const string s) { passCarType = s; }
bool getSunRoof() const
{ return sunRoof; }
void setSunRoof( bool b ) { sunRoof = b; }
void display() const;
};

MEMBERS OF DERIVED CLASSES

■

505

Let’s look at the example on the opposite page to illustrate how derived classes are
defined. The Car class and a derived class PassCar are defined in the example.

䊐 Data Members and Methods
The base class Car contains two data members, nr and producer, which are used to
represent an identification number and the name of the manufacturer. The derived class
PassCar inherits these data members. Thus, an object of the PassCar class also contains the data members nr and producer. The object includes a so-called base subobject of type Car.
The PassCar class additionally contains the data members passCarType and
sunRoof to represent a passenger vehicle with or without a sunroof. So a PassCar type
object has a total of four data members. For the sake of simplicity, we have omitted further data members, such as the number of seats, etc.
The base class Car contains a constructor, access methods, and the method
display(), which is used for screen output. The methods are also inherited by the
derived class PassCar.
In the PassCar class a constructor, additional access methods, and a second output
function also called display() are declared. The derived class thus inherits a method
called display() and declares a method with the same name. The display()
method is said to have been redefined.
Every member function and every data member in a derived class can be redefined.
The member assumes a new meaning for the derived class. The member inherited from
the base class is also available in the derived class and will retain its original meaning.
We will be looking at this point in more detail later.

䊐 Public Interface
Since the Car class is a public base class of the PassCar class, all the public members of the base class are available in the derived class. For example, you can call the
getNr() method for an object named cabrio in the PassCar class.

Example: cout << "Car number: "<< cabrio.getNr();
The public interface of the derived class thus comprises
■
■

the public members of the base class and
the public members additionally defined in the derived class.

506

■

CHAPTER 23

■

INHERITANCE

MEMBER ACCESS

Accessing members of base class Car

class Car
{
private:
long nr;
string producer;
public:
. . .
long getNr(void);
. . .

not ok

ok

}

class PassCar : public Car
{
private:
string passCarType;
bool
sunRoof;
public:
. . .
}

void PassCar::display( void) const
{
cout << "\nCar number: "
<< getNr();
cout << "\nProducer: "
<< getProd();
cout << "Type: "<< passCarType;
cout << "Type: "<< passCarTyp
if( sunRoof) cout << "yes";
else
cout << " no";
cout << endl;
}

ok

MEMBER ACCESS

■

507

䊐 Access to Additional Members
The methods of derived classes can access any member additionally defined in the
derived class.

Example: const string& getType() const
{

return passCarType;

}

The getType() method directly accesses the private data member passCarType in
the PassCar class in this example.

䊐 Access to Private Members of the Base Class
However, a private member of the base class is not directly accessible for the methods of
the derived class. The output function display() in the derived class PassCar, for
example, cannot contain the following statement:

Example: cout << "Producer: " << producer;
As producer is a private data member of the base class Car, the compiler would issue
an error message at this point.
Methods belonging to derived classes only have indirect access to the private data
members of the base class. They use access methods in the public declaration of the
base class for this purpose. The opposite page shows a version of the display() method
that calls the get methods in its base class Car.
When you call an access method, you do not need to state the method’s base class.
The base class is identified by the this pointer, which is passed implicitly as an argument. The call to getProd() on the opposite page is thus equivalent to:

Example: this->getProd();

䊐 Name Lookup
The following rules apply when searching for the name of a method:
■
■

the compiler looks for the name of the method called in the derived class first
if the name cannot be found, the compiler walks one step up the tree and looks
for a public method with that name.

The above example thus calls the getProd() in the base class Car, as the method is
not defined in the PassCar class.

508

■

CHAPTER 23

■

INHERITANCE

REDEFINING MEMBERS

New version of method display()
//
//
//
//

Within file Car.cpp
This version of method PassCar::display() calls
the method display() of the base class.
---------------------------------------------------

void PassCar::display( void) const
{
Car::display();
// Method in base class
cout << "Type:
cout << "\nSunroof:
if(sunRoof)
cout << "yes "<<
else
cout << "no " <<
}

" << passCarType;
";
endl;
endl;

REDEFINING MEMBERS

■

509

䊐 Redefinition
There are two options for the names of data members or methods in derived classes:
1. The name does not occur in the base class → no redefinition.
2. The name already exists in the base class → redefinition.
In the second case, the member of the same name continues to exist unchanged in the
base class. In other words, redefining members in a derived class has no effect on the base
class.
However, the name lookup rules for the compiler lead to the following scenario:
■

if a member is redefined in a derived class, it will mask the corresponding member in the base class.

This situation is similar to the one seen for local and global variables. A local variable
will mask a previously defined global variable with the same name.

䊐 Redefinition and Overloading
Normally, methods are redefined in derived classes. This adopts the methods to the new
features of the class. When a method is redefined, the signature and the return type of
the method can be changed. However, a redefinition does not overload functions since
the derived class has a different scope.
Redefining a method will always mask a method with the same name in the base class.
Of course, you can overload methods within the same class, and this means you can
repeatedly redefine a base class method for a derived class.

䊐 Access to the Members in the Base Class
If you redefine a method in a derived class, this does not alter the fact that the base class
method still exists. If the method was declared in the public section of the base class,
you can call it to redefine a method. The range :: operator is used to access the base
class method.
The new version of the display() method opposite illustrates this point. The
display() method defined in the base class is used to output the data members of the
base class. To do so, you must use the range operator with the name of the base class.
Otherwise the display() method in the derived class will call itself and head off into
an indefinite recursion.

510

■

CHAPTER 23

■

INHERITANCE

CONSTRUCTING AND DESTROYING DERIVED CLASSES

First version of the constructor of PassCar
// First version of the constructor of PassCar
// -------------------------------------------------PassCar::PassCar(const string& tp, bool sr, int n,
const string& hs)
{
setNr(n);
// Initial values for data
setProd(hs);
// members of the base class.
passCarType = tp;
sunRoof = sr;

// Initial values for data mem// bers of the derived class

}

Second version with base class initializer
// Second version of the constructors of PassCar
// ---------------------------------------------------PassCar::PassCar(const string& tp, bool sr, int n,
const string& hs) : Car( n, hs)
{
passCarType = tp;
// Initial values for data memsunRoof = sr;
// bers of the derived class
}

Third version with base class and member initializer
// Third version of the constructor of PassCar
// ---------------------------------------------------PassCar::PassCar(const string& tp, bool sr, int n,
const string& hs)
: Car( n, hs), passCarType( tp ), sunRoof( sr )
{
// There remains nothing to do
}

CONSTRUCTING AND DESTROYING DERIVED CLASSES

■

511

䊐 Constructor Calls
The constructor of a derived class is required to create an object of the derived class type.
As the derived class contains all the members of the base class, the base sub-object must
also be created and initialized. The base class constructor is called to perform this task.
Unless otherwise defined, this will be the default constructor.
The order in which the constructors are called is important. The base class constructor is called first, then the derived class constructor. The object is thus constructed from
its core outwards.
The first version of the constructor for PassCar, as shown opposite, sets initial values by calling the access methods of the base class. An implicit call to the default constructor of the base class occurs prior to this, and the base sub-object is initialized with
default values. This process has the same drawbacks as the technique of creating objects
with member objects. A default constructor must be available in the base class and initialization with incorrect values before assigning live values impacts the response of the
program.

䊐 Base Initializer
If the base class contains a constructor with parameters, it makes sense to call this constructor. This immediately initializes the data members with correct values. A base initializer for the constructor of the derived class can be defined for this purpose.
The second version of the constructor for PassCar contains a base initializer.

Example: Car( n, hs )
The syntax of the base initializer for base sub-objects is similar to that of the member initializer for member sub-objects. This means that you can state both the base and the
member initializer in a list separated by commas. The third version of the PassCar constructor illustrates this point.

䊐 Destroying Objects
When an object is destroyed, the destructor of the derived class is first called, followed by
the destructor of the base class. The reverse order of the constructor calls applies.
You need to define a destructor for a derived class if actions performed by the constructor need to be reversed. The base class destructor need not be called explicitly as it
is executed implicitly.

512

■

CHAPTER 23

■

INHERITANCE

OBJECTS OF DERIVED CLASSES

Sample program
// car_t.cpp: Testing the base class Car and
//
the derived class PassCar.
// ----------------------------------------------------#include "car.h"
int main()
{
const PassCar beetle("Beetle", false, 3421, "VW");
beetle.display();
cout << "\nAnd the passenger car number again: "
<< beetle.getNr() << endl;
PassCar cabrio("Carrera", true);
cabrio.setNr(1000);
cabrio.setProd("Porsche");
cabrio.display();
cout << "\nOnly data of the base class: ";
cabrio.Car::display();
return 0;
}

Screen output
--------------------------------------------Car number:
3421
Producer:
VW
Type:
Beetle
Sunroof:
no
And the passenger car number again: 3421
--------------------------------------------Car number:
1000
Producer:
Porsche
Type:
Carrera
Sunroof:
yes
Only data of the base class:
--------------------------------------------Car number:
1000
Producer:
Porsche

OBJECTS OF DERIVED CLASSES

■

513

䊐 Declaring Objects
The program opposite illustrates how objects of derived classes can be used. Two objects,
beetle and cabrio, of the derived class PassCar type are declared. As the PassCar
class does not contain a default constructor, both objects must be initialized. However, it
is sufficient to state a PassCar type with or without a sunroof as default values exist for
all other data members.
The object beetle is declared as const just to show that the get methods and the
display() method can also be called for constant objects since they were declared as
read-only methods.
However, the following call is invalid:

Example: beetle.setNr( 7564 );

// Error

This means you have to correctly define all the initial values for the object when you
declare it.

䊐 Calling Redefined Methods
When you call a redefined method, the object type determines what version of the
method will be executed. In the PassCar class the method display() has been redefined. The statement

Example: cabrio.display();
also outputs the additional data members passCarType and sunRoof. However, in
the case of the van object in the Car class, calling

Example: van.display();
will execute the method in the base class.

䊐 Calling Methods in the Base Class
You may be wondering if a base class method can be called for an object of a derived
class, if the method has been redefined in the derived class. This is possible using the
scope resolution operator,::.
If you want to display the basic data of the cabrio object, you can use a direct call to
the base class method display() to do so.

Example: cabrio.Car::display();
The name of the method is preceded by the name of the base class and the scope resolution operator in this case.

514

■

CHAPTER 23

■

INHERITANCE

PROTECTED MEMBERS

Sample classes
// safe.h : The classes Safe and Castle
// --------------------------------------------------#include 
using namespace std;
class Safe
{
private:
int topSecret;
protected:
int secret;
void setTopSecret( int n) { topSecret = n;}
int getTopSecret() const { return topSecret;}
void setSecret( int n){ secret = n;}
int getSecret() const { return secret;}
public:
int noSecret;
Safe()
{ topSecret = 100; secret = 10; noSecret = 0; }
};
class Castle : public Safe
{
public:
Castle()
{
// topSecret = 10;
setTopSecret(10);
secret = 1;
noSecret = 0;
}
void test()
{
// top.Secret = 200;
setTopSecret(200);
secret = 20;
noSecret = 2;
}
};

//
//
//
//

Error, because private
ok, because protected
ok, because protected
ok, because public

//
//
//
//

Error, because private
ok, because protected
ok, because protected
ok, because public

PROTECTED MEMBERS

■

515

䊐 Access to Sheltered Members
The private members of a base class are equally inaccessible for the methods and friend
functions of a derived class.
When you create a class hierarchy you may want require the methods and friend
functions of a derived class to communicate directly with the members of the base class.
This would be particularly necessary if the base class contained members for use as building blocks for derived classes but not for general purpose use.
For example, a class used to represent a window on screen could contain the dimensions and other characteristics of a general windows. The characteristics need protecting;
however, methods in derived classes will still need direct access.

䊐 Protected Members
To allow methods and friend functions access to the sheltered members of a base class,
let’s introduce an additional level of access control between private and public.
This is achieved by means of protected declarations.
A member declared protected is sheltered from external access just like a private member. That means, a protected member is inaccessible for base class objects
and any classes derived from the base class. However, in contrast to a private member,
methods and friend functions of derived classes can access the member.
The classes defined opposite, Safe and Castle, show that protected members of
the base class can be accessed directly in a derived class. In contrast to this, protected
members are inaccessible to users of these classes.

Example: Castle

treasure;
treasure.topSecret = 1;
treasure.secret = 2;
treasure.setTopSecret(5);
treasure.noSecret = 10;

//
//
//
//

Error: private
Error: protected
Error: protected
ok

Protected declarations should be used with caution. If you change the declaration of a
protected member, every class derived from this class must be examined to ascertain

whether additional modifications are necessary.

■

CHAPTER 23

■

exercise s

516

INHERITANCE

EXERCISES

For exercise 1
Class Truck being derived from class

Car

Additional data members:
Type
Number of axles
Load capacity

int
double

Additional methods:
void
int
void
void

setAxles( int a );
getAxles() const;
setCapacity( double cp );
getCapacity() const;

void display() const;

EXERCISES

■

517

Exercise 1
The classes Car and PassCar are to modify to allow objects to be created and
destroyed. In addition, the class Truck is to be added to the class hierarchy.
■

Change the classes Car and PassCar to make the constructor issue the
following message:
"Creating an object of type ... ."

■

Define a destructor for the Car and PassCar classes.The destructor
should issue the following message:
"Destroying an object of type ...."

■

■

■

■

Then define the class Truck, which is derived from Car, using the data
members shown opposite, a constructor, a destructor, and the additional
methods shown opposite.
Implement the constructor for the Truck class—the constructor should
again issue a suitable message. Use the base initializer to initialize the data
members of Car.
Define a destructor for Truck—the destructor should again issue a suitable message for trucks.
To test your class, create and display a Truck type object in your main
function. If required by the user, enable your program to create and display objects of the types PassCar and Car.

Observe how the various objects and member objects are created and
destroyed.

Exercise 2
Derive two classes, DepAcc and SavAcc, from the Account class, which was
defined in Chapter 14, in the section titled “Const Objects and Methods.”
Additionally define an overdraft limit and an interest rate for the DepAcc class.
The SavAcc contains the members of the base class and an interest rate.
■

■

For both classes, define constructors to provide default values for all
parameters, add access methods, and add a display() method for
screen output.
Test the new classes by initializing objects of the DepAcc and SavAcc
types in the object declarations and outputting them.Then modify both a
savings and a deposit account interactively and display the new values.

518

■

CHAPTER 23

INHERITANCE

Exercise 3

Product
Properties:
Barcode
Name
Methods:

setCode()
getCode()
...
scanner()
printer()

PrepackedFood
Properties:
Price per piece

FreshFood
Properties:
Weight
Price per pound

Methods:

getPrice()
setPrice()
...
scanner()
printer()

Methods:

setWght()
getWght()
...
scanner()
printer()

EXERCISES

■

519

Exercise 3
A supermarket chain has asked you to develop an automatic checkout system.
All products are identifiable by means of a barcode and the product name.
Groceries are either sold in packages or by weight. Packed goods have fixed
prices.The price of groceries sold by weight is calculated by multiplying the
weight by the current price per kilo.
Develop the classes needed to represent the products first and organize
them hierarchically.The Product class, which contains generic information on all
products (barcode, name, etc.), can be used as a base class.
■

■

■

The Product class contains two data members of type long used for
storing barcodes and the product name. Define a constructor with
parameters for both data members.Add default values for the parameters to provide a default constructor for the class. In addition to the
access methods setCode() and getCode(), also define the methods
scanner() and printer(). For test purposes, these methods will simply
output product data on screen or read the data of a product from the
keyboard.
The next step involves developing special cases of the Product class.
Define two classes derived from Product, PrepackedFood and FreshFood. In addition to the product data, the PrepackedFood class should
contain the unit price and the FreshFood class should contain a weight
and a price per kilo as data members.
In both classes define a constructor with parameters providing
default-values for all data members. Use both the base and member initializer.
Define the access methods needed for the new data members.Also
redefine the methods scanner() and printer() to take the new data
members into consideration.
Test the various classes in a main function that creates two objects each
of the types Product, PrepackedFood and FreshFood. One object of
each type is fully initialized in the object definition. Use the default constructor to create the other object.Test the get and set methods and
the scanner() method and display the products on screen.

■

CHAPTER 23

■

solutions

520

INHERITANCE

SOLUTIONS

Exercise 1
// ---------------------------------------------------// Car.h : Defines the base class Car and
//
the derived classes PassCar and Truck
// -------------------------------------------------------#ifndef _CAR_H_
#define _CAR_H_
#include 
#include 
using namespace std;
class Car
{
// See previous definition in this chapter
};
class PassCar : public Car
{
// See previous definition in this chapter
};
class Truck : public Car
{
private:
int
axles;
double tons;
public:
Truck( int a, double t, int n, const string& hs);
~Truck();
void
int
void
double

setAxles(int l){ axles = l;}
getAxles() const
{ return axles; }
setCapacity( double t) { tons = t;}
getCapacity() const
{ return tons; }

void display() const;
};
#endif

SOLUTIONS

■

// -----------------------------------------------------// car.cpp
// Implements the methods of Car, PassCar, and Truck
// -----------------------------------------------------#include "car.h"
// -----------------------------------------------------// The methods of base class Car:
Car::Car( long n, const string& prod)
{
cout << "Creating an object of type Car." << endl;
nr = n;
producer = prod;
}
Car::~Car()
{
cout << "Destroying an object of type Car" << endl;
}
void Car::display() const
{
cout << "\n---------------------------- "
<< "\nCar number:
" << nr
<< "\nProducer:
" << producer
<< endl;
}
// ------------------------------------------------------// The methods of the derived class PassCar:
PassCar::PassCar(const string& tp, bool sd, int n,
const string& hs)
: Car( n, hs), PassCarTyp( tp ), sunRoof( sd )
{
cout << "I create an object of type PassCar." << endl;
}
PassCar::~PassCar()
{
cout << "\nDestroying an object of type PassCar"
<< endl;
}
void PassCar::display( void) const
{
Car::display();
// Base class method
cout << "Type:
" << passCarType
<< "\nSunroof:
";
if(sunRoof)
cout << "yes "<< endl;
else
cout << "no " << endl;
}

521

522

■

CHAPTER 23

INHERITANCE

// ---------------------------------------------------// The methods of the derived class Truck:
Truck::Truck( int a, double t, int n, const string& hs)
: Car( n, hs), axles(a), tons(t)
{
cout << "Creating an object of type Truck." << endl;
}
Truck::~Truck()
{
cout << "\nDestroying an object of type Truck\n";
}
void Truck::display() const
{
Car::display();
cout <<
"Axles:
" << axles
<< "\nCapacity:
" << tons << " long tons\n";
}
// ----------------------------------------------------// Car_t.cpp : Tests the base class Car and
//
the derived classes PassCar and Truck.
// ----------------------------------------------------#include "car.h"
int main()
{
Truck toy(5, 7.5, 1111, "Volvo");
toy.display();
char c;
cout << "\nDo you want to create an object of type "
<< " PassCar? (y/n) "; cin >> c;
if( c == 'y' || c == 'Y')
{
const PassCar beetle("Beetle", false, 3421, "VW");
beetle.display();
}
cout << "\nDo you want to create an object "
<< " of type car? (y/n) ";
cin >> c;
if( c == 'y' || c == 'Y')
{
const Car oldy(3421, "Rolls Royce");
oldy.display();
}
return 0;
}

SOLUTIONS

■

Exercise 2
// ----------------------------------------------------// account.h:
// Defines the classes Account, DepAcc, and SavAcc.
// ----------------------------------------------------#ifndef _ACCOUNT_H
#define _ACCOUNT_H
#include 
#include 
#include 
using namespace std;
class Account
{
private:
string name;
unsigned long nr;
double balance;
public:
Account(const string& s="X", unsigned long n
= 1111111L, double st = 0.0)
: name(s), nr(n), balance(st)
{ }
const string& getName() const { return name; }
void setName(const string& n) { name = n;}
unsigned long getNr() const { return nr; }
void setNr(unsigned long n) { nr = n; }
double getBalance() const
{ return balance; }
void
setBalance(double st){ balance = st; }
void display()
{ cout << fixed << setprecision(2)
<< "----------------------------------------\n"
<< "Account holder:
" << name << endl
<< "Account number:
" << nr
<< endl
<< "Balance of the account:" << balance <> db;
mickey.setName(s);
mickey.setInterest(db);
mickey.display();
DepAcc dag("Donald Duck", 7654321,
-1245.56, 10000, 12.9);
dag.display();
cout << "New limit:
dag.setLimit(db);
dag.display();
return 0;
}

"; cin >> db;

525

526

■

CHAPTER 23

INHERITANCE

Exercise 3
// ---------------------------------------------------// product.h : Defines the classes
//
Product, PrepackedFood, and FreshFood
// ---------------------------------------------------#ifndef _PRODUCT_H
#define _PRODUCT_H
#include 
#include 
#include 
using namespace std;
class Product
{
private:
long
bar;
string name;
public:
Product(long b = 0L, const string& s = "")
: bar(b), name(s)
{ }
void setCode(long b) { bar = b; }
long getCode() const { return bar; }
void setName(const string& s){ name = s; }
const string& getName() const { return name; }
void scanner()
{
cout << "\nBarcode:
"; cin >> bar;
cout <<
"Name:
"; cin >> name;
cin.sync(); cin.clear();
}
void printer() const
{
cout << "\n-------------------------------"
<< "\nBarcode:
" << bar
<< "\nName:
" << name
<< endl;
}
};
class PrepackedFood : public Product
{
private:
double pce_price;

SOLUTIONS

■

public:
PrepackedFood(double p = 0.0,long b = 0L,
const string& s = "")
: Product(b, s), pce_price(p)
{}
void
setPrice(double p){ pce_price = p;}
double getPrice()const
{ return pce_price; }
void scanner()
{
Product::scanner();
cout << "Price per piece:
"; cin >> pce_price;
}
void printer() const
{ Product::printer();
cout << fixed << setprecision(2)
<< "Price per piece:
" << pce_price << endl;
}
};
class FreshFood : public Product
{
private:
double wght;
double lbs_price;
public:
FreshFood(double g = 0.0, double p = 0.0,
long b = 0L, const string& s = "")
: Product(b, s), wght(g), lbs_price(p) {}
void
setWght(double g) { wght = g;}
double getWght()const
{ return wght; }
void
setPrice(double p) { lbs_price = p;}
double getPrice()const
{ return lbs_price; }
void scanner()
{
Product::scanner();
cout << "Weight(lbs): "; cin >> wght;
cout << "Price/lbs:
"; cin >> lbs_price;
cin.sync(); cin.clear();
}
void printer() const
{
Product::printer();
cout << fixed << setprecision(2)
<< "Price per Lbs:
" << lbs_price
<< "\nWeight:
" << wght
<< "\nTotal:
" << lbs_price * wght
<< endl;
}
};
#endif

527

528

■

CHAPTER 23

//
//
//
//

INHERITANCE

-----------------------------------------------------product_t.cpp
Tests classes Product, PrepackedFood, and FreshFood.
------------------------------------------------------

#include "product.h"
int main()
{
Product p1(12345L, "Flour"), p2;
p1.printer();

// Output the first product

p2.setName("Sugar");
p2.setCode(543221);

// Set the data members

p2.printer();

// Output the second product

// Prepacked products:
PrepackedFood pf1(0.49, 23456, "Salt"), pf2;
pf1.printer();

//
//
cout << "\n Input data of a
pf2.scanner();
//
pf2.printer();
//

Output the first
prepacked product
prepacked product: ";
Input and output
data of 2nd product

FreshFood pu1(1.5, 1.69, 98765, "Grapes"), pu2;
pu1.printer();

//
//
cout <<"\n Input data for a
pu2.scanner();
//
pu2.printer();
//
cout <<
<<
<<
<<
<<
<<
<<
<<
<<

Output first item
fresh food
prepacked product: ";
Input and output
data of 2nd product.

"\n-------------------------------"
"\n-------------------------------"
"\nAgain in detail: \n"
fixed << setprecision(2)
"\nBarcode:
" << pu2.getCode()
"\nName:
" << pu2.getName()
"\nPrice per Lbs: " << pu2.getPrice()
"\nWeight:
" << pu2.getWght()
"\nEnd price:
" << pu2.getPrice()
* pu2.getWght()
<< endl;

return 0;
}

chapter

24

Type Conversion in Class
Hierarchies
This chapter describes implicit type conversion within class hierarchies,
which occurs in the context of assignments and function calls.
In addition, explicit type casting in class hierarchies is discussed, in
particular, upcasting and downcasting.

529

530

■

CHAPTER 24

■

TYPE CONVERSION IN CLASS HIERARCHIES

CONVERTING TO BASE CLASSES

Example for implicit conversion

#include "car.h"
bool compare( Car&, Car&);
int main()
{
PassCar beetle("New Beetle", false, 3421, "VW"),
miata( "Miata", true, 2512, "Mazda");
bool res = compare( beetle, miata);
// ...
}
//
//
//
//
//

ok!
Implicit conversion
to base class.
Car& a = beetle;
Car& b = miata;

bool compare( Car& a, Car& b)
{
//
//
//
//
}

Here a is the base part of beetle,
b is the base part of miata.
If this is inconvenient, an explicit
type cast to type PassCar has to be performed.

CONVERTING TO BASE CLASSES

■

531

䊐 Implicit Conversion
If a class is derived from another class by public inheritance, the derived class assumes
the characteristics and features of the base class. Objects of the derived class type then
become special objects of the base class, just like an automobile is a special type of vehicle.
You can utilize the is relationship when handling objects. It is possible to assign an
object of a derived class to an object of the base class. This causes an implicit type conversion to a base class type.
The base class thus becomes a generic term for multiple special cases. Given that the
classes PassCar and Truck were derived from the Car class, objects of the PassCar
or Truck type can always be managed like objects of Car type.

䊐 Assignments
Implicit type conversion in class hierarchies occurs in assignments to
■
■

base class objects
pointers or references to the base class.

䊐 Function Calls
Additionally, a similar kind of implicit type conversion takes place for the arguments of
function calls.
Given the function compare() with the following prototype

Example: bool compare( Car& , Car& );
and two objects of the derived PassCar class type, beetle and miata, the following
statement is valid

Example: compare( beetle, miata);
The compiler performs implicit type conversion for the arguments beetle and miata,
converting them to the parameter type, that is, to a reference to the base class Car.
Type conversion for arguments used in function calls is similar to the type conversion
that occurs in assignments, as shown in the following section.

532

■

CHAPTER 24

■

TYPE CONVERSION IN CLASS HIERARCHIES

TYPE CONVERSIONS IN ASSIGNMENTS

䊐 Effect of an assignment
Car
auto;
PassCar bmw("520i", true, 4325,
"Bayerische Motorenwerke");
auto = bmw;

auto

bmw

nr: 4325

nr: 4325

producer:
"Bayer...."

producer:
"Bayer...."

passCarType:
"520i"
sunRoof: true

TYPE CONVERSIONS IN ASSIGNMENTS

■

533

䊐 Assignment to a Base Class Object
An object belonging to a derived class type can be assigned to an object of a base class.

Example: Car

auto;
PassCar bmw("520i", true, 4325,
"Bayerische Motorenwerke");
auto = bmw;

The object bmw, which belongs to the derived class PassCar, contains all the data
members of the base class, Car, i.e. the vehicle id number and the manufacturer. During
an assignment the object bmw is copied to the data members of the object auto step by
step.
This makes the above statement equivalent to:
auto.nr
= bmw.nr;
auto.producer = bmw.producer;

The data members additionally defined in the derived class are not copied!
The following statement outputs the copied data members:

Example: auto.display();
The fact that you can assign an object belonging to a derived class to a base class object
assumes that more will always fill less. The object on the right of the assignment operator
will always contain a member object of the type on the left of the operator.

䊐 Assignments to Derived Class Objects
This is not the case when you attempt to assign a base class object to an object of a
derived class. The assignment

Example: bmw = auto;

// Error!

is therefore invalid, since the values for the additional data members passCarType and
sunRoof are unknown.
An assignment in reverse order is only possible if you have defined an assignment of
this type or a copy constructor with a parameter of the type “reference to base class.”
Both would be able to supply default values for the additional data members of the
derived classes.

534

■

CHAPTER 24

■

TYPE CONVERSION IN CLASS HIERARCHIES

CONVERTING REFERENCES AND POINTERS

䊐 Effect of a pointer assignment
PassCar cabrio("Spitfire", true, 1001, "Triumph");
Car* carPtr = &cabrio;
carPtr = &cabrio;

carPtr

cabrio
nr

1001

producer

"Triumph"

carType

"Spitfire"

sunRoof

true

CONVERTING REFERENCES AND POINTERS

■

535

䊐 Converting to Base Class Pointers
The is relationship between a derived class and a base class is also apparent when references and pointers are used. A pointer of the type “pointer to base class,” or base class
pointer for short, can reference an object of a derived class type.

Example: Car* carPtr = &cabrio;
In this case cabrio is an object of the class PassCar.
The following rule applies for access to the referenced object:
■

a base class pointer can only access the public interface of the base class.

The additional members defined in the derived class are therefore inaccessible. To make
this more clear:

Example: carPtr -> display();
calls the display() method in the base class Car. Although carPtr points to an
object of the PassCar class in this case, it is impossible to call any methods additionally
defined in the derived class.

Example: carPtr->setSunRoof(false);

// Error

The object *carPtr belongs to the Car class and only represents the generic part of
cabrio. Thus, the following assignment is also invalid

Example: PassCar auto;
auto = *carPtr;

// Error!

although carPtr is pointing at an object of the PassCar type in this case!

䊐 Conversions in References to Base Classes
A similar situation arises when you are working with references. A reference of the type
“reference to base class” can point to an object of a derived class. The reference will
address only the generic part of the object in this case.

Example: Car& carRef = cabrio;
carRef.display();
carRef.setSunRoof(true);
PassCar auto;
auto = carRef;

// ok
// Output base members
// Error
// Error

Although the reference carRef points to an object of the PassCar type, it is impossible to assign the PassCar type object auto to this object.

536

■

CHAPTER 24

■

TYPE CONVERSION IN CLASS HIERARCHIES

EXPLICIT TYPE CONVERSIONS

Downcast

Pointer to

carPtr
Base class

Car

Downcast

static_cast(carPtr)
Derived
class

PassCar

Upcast

Pointer to

static_cast(PassCarPtr)
Base class

Car

Upcast

PassCarPtr
Derived
class

PassCar

EXPLICIT TYPE CONVERSIONS

■

537

䊐 Upcasts and Downcasts
Type conversions that walk up a class hierarchy, or upcasts, are always possible and safe.
Upcasting is performed implicitly for this reason.
Type conversions that involve walking down the tree, or downcasts, can only be performed explicitly by means of a cast construction. The cast operator (type), which was
available in C, or the static_cast< > operator are available for this task, and are
equivalent in this case.

䊐 Explicit Cast Constructions
Given that cabrio is again an object of the derived class PassCar, the following statements

Example: Car* carPtr = &cabrio;
( (PassCar*) carPtr )->display();

first point the base class pointer carPtr to the cabrio object. carPtr is then cast as a
pointer to the derived class. This allows you to access the display() method of the
derived class PassCar via the pointer. Parentheses are necessary in this case as the
member access operator -> has a higher precedence than the cast operator (type).
The operator static_cast< > conforms to the following

Syntax:

static_cast(expression)

and converts the expression to the target type type. The previous example is thus equivalent to

Example: static_cast(carPtr)->display();
No parentheses are required here as the operators static_cast<> and -> are of equal
precedence. They are read from left to right.
After downcasting a pointer or a reference, the entire public interface of the derived
class is accessible.

䊐 Downcast Safety Issues
Type conversions from top to bottom need to be performed with great care. Downcasting
is only safe when the object referenced by the base class pointer really is a derived class
type. This also applies to references to base classes.
To allow safe downcasting C++ introduces the concept of dynamic casting. This technique is available for polymorphic classes and will be introduced in the next chapter.

■

CHAPTER 24

■

exercise

538

TYPE CONVERSION IN CLASS HIERARCHIES

EXERCISE

Class hierarchy of products in a supermarket

Product

PrepackedFood

FreshFood

EXERCISE

■

539

Exercise
The class hierarchy representing a supermarket chain’s checkout system
comprises the base class Product and the derived classes PrepackedFood and
FreshFood.Your job is to test various cast techniques for this class (see also
Exercise 3 in Chapter 23).
■

■

■

■

■

■

Define a global function isLowerCode() that determines which one of
two products has the lower barcode and returns a reference to the product with the lower barcode.
Define an array with three pointers to the base class Product. Dynamically create one object each of the types Product, PrepackedFood, and
FreshFood.The three objects are to be referenced by the array pointers.
Additionally define a pointer to the derived class FreshFood. Initialize the
pointer with the address of a dynamically allocated object of the same
class.
Now call the method printer() for all four objects.Which version of
printer() is executed?
Perform downcasting to execute the correct version of printer() in
every case. Display the pointer values before and after downcasting.
Use the pointer of the derived class FreshFood to call the base class version of printer(). Perform an appropriate upcast.
Test the function isLowerCode()by multiple calls to the function with
various arguments. Output the product with the lower barcode value in
each case.

■

CHAPTER 24

solution

540

TYPE CONVERSION IN CLASS HIERARCHIES

■

SOLUTION

//
//
//
//
//
//

---------------------------------------------------product.h : Defines the classes
Product, PrepackedFood, and FreshFood
----------------------------------------------------

//
//
//
//
//

-----------------------------------------------------produc_t.cpp
Tests up and down casts for the classes
Product, PrepackedFood, and FreshFood.
------------------------------------------------------

Unchanged! See the previous chapter's solutions.

#include "product.h"
const Product& isLowerCode(const Product& p1,
const Product& p2);
int main()
{
Product* pv[3];
FreshFood* pu;
pv[0] = new Product(12345L, "Flour");
pv[1] = new PrepackedFood(0.49, 23456, "Salt");
pv[2] = new FreshFood(1.5, 1.69, 98765, "Grapes");
pu =

new FreshFood(2.5, 2.69, 56789, "Peaches");

cout << "\nA fresh product: ";
pu->printer();
cout << "\nThe generic data of the other products:";
int i;
for(i=0; i < 3; ++i)
pv[i]->printer();
cin.get();
cout << "\nAnd now the downcast: " << endl;
static_cast(pv[1])->printer();
static_cast(pv[2])->printer();
cin.get();
cout << "\nAnd an upcast: " << endl;
static_cast(pu)->printer();

SOLUTION

cout << "\nNow compare the barcodes!" << endl;
cout << "\nIs barcode for flour or salt smaller?";
isLowerCode(*pv[0], *pv[1]).printer();
cout << "\nIs barcode for salt or grapes smaller?";
isLowerCode(*pv[1], *pv[2]).printer();
return 0;
}
const Product& isLowerCode(const Product& p1,
const Product& p2)
{
if(p1.getCode() < p2.getCode())
return p1;
else
return p2;
}

■

541

This page intentionally left blank

chapter

25

Polymorphism
This chapter describes how to develop and manage polymorphic classes.
In addition to defining virtual functions, dynamic downcasting in
polymorphic class hierarchies is introduced.

543

544

■

CHAPTER 25

■

POLYMORPHISM

CONCEPT OF POLYMORPHISM

Example

Classes with virtual methods:

Base:
base class with virtual method display().

Derived1 and Derived2:
derived from Base with its own redefinitions
of method display().

Base class pointer and objects:

Base*

basePtr;

Derived1 angular;

// Base class pointer
// Objects

Derived2 round;
Calling virtual methods:
When a virtual method is called, the corresponding version of the method is
executed for the object currently referenced.

basePtr = &angular;
basePtr->display();

// Calling
// Derived1::display()

basePtr = &round;
basePtr->display();

// Calling
// Derived2::display()

CONCEPT OF POLYMORPHISM

■

545

䊐 Issues
If the special features of derived class objects are insignificant, you can simply concern
yourself with the base members. This is the case when dynamic allocated objects are
inserted into a data structure or deleted from that structure.
It makes sense to use pointers or references to the base class in this case—no matter
what type of concrete object you are dealing with. However, you can only access the
common base members of these objects.
However, you should be able to activate the special features of a derived class when
■
■

the object is accessed by a pointer or reference to the base class and
the concrete object type will not be known until the program is executed.

Given a base class pointer, carPtr, the statement

Example: carPtr->display();
should output all the data members of the object currently being referenced.

䊐 Traditional Approach
Traditional programming languages solved this issue by adding a type field both to the
base class and to the derived classes. The type field stored the type of the current class. A
function that manages objects via the base class pointer could query the concrete type in
a switch statement and call the appropriate method.
This solution has a disadvantage; adding derived classes at a later stage also meant
adding a case label and recompiling.

䊐 Object-Oriented Approach
The approach adopted by object-oriented languages is polymorphism (Greek for multiform). In C++, virtual methods are used to implement polymorphic classes. Calling a virtual method makes the compiler execute a version of the method suitable for the object
in question, when the object is accessed by a pointer or a reference to the base class!

546

■

CHAPTER 25

■

POLYMORPHISM

VIRTUAL METHODS

Calling the virtual method display()
// virtual.cpp : Tests the virtual method display()
//
of the classes Car and PassCar.
// ---------------------------------------------------#include "car.h"
// The Car class with virtual method display():
// class Car
// {
//
...
//
virtual void display() const;
// };
int main()
{
Car* pCar[3];
// Three pointers to the base class.
int i = 0;
// Index
pCar[0] = new Car( 5634L, "Mercedes");
pCar[1] = new PassCar("Miata",true,3421,"Mazda");
pCar[2] = new Truck( 5, 7.5, 1234, "Ford");
while( true)
{
cout << "\nTo output an object of type "
"Car, PassCar or Truck!"
"\n 1 = Car, 2 = PassCar, 3 = Truck"
"\nYour input (break with 0): ";
cin >> i;
--i;
if( i < 0 || i > 2)
break;
pCar[i]->display();
}
return 0;
}

VIRTUAL METHODS

■

547

䊐 Declaring Virtual Methods
The virtual keyword is used to declare a virtual method in a base class.

Example: virtual void display() const;
The definition of a virtual method is no different from the definition of any other member function.
A virtual method does not need to be redefined in the derived class. The derived class
then inherits the virtual method from the base class.

䊐 Redefinition
However, it is common practice for the derived class to define its own version of the virtual method, which is thus modified to suit the special features of the derived class.
Creating a proprietary version of a virtual method means redefining that method. The
redefinition in the derived class must have
1. the same signature and
2. the same return type
as the virtual method in the base class.
The new version of a virtual method is automatically virtual itself. This means you
can omit the virtual keyword in the declaration.
When you redefine a virtual function, be aware of the following:
■

■
■

if the return type is a pointer or reference to the base class, the new version of the
virtual method can also return a pointer or reference to a derived class (Note:
Not all compilers support this option.)
constructors cannot have a virtual declaration
a base class method does not become virtual just because it is declared as virtual
in a derived class.

If you use a different signature or return type of a virtual base class method to define a
method in a derived class, this simply creates a new method with the same name. The
method will not necessarily be virtual!
However, the virtual method in the base class will be masked by the method in the
derived class. In other words, only the non-virtual version of the method is available for
a derived class object.

548

■

CHAPTER 25

■

POLYMORPHISM

DESTROYING DYNAMICALLY ALLOCATED OBJECTS

Sample program
// v_destr.cpp
// Base class with a virtual destructor.
// ----------------------------------------------------#include 
#include 
// For strcpy()
using namespace std;
class Base
{
public:
Base()
{ cout << "Constructor of class Base\n"; }
virtual ~Base()
{ cout << "Destructor of class Base\n"; }
};
class Data
{
private:
char *name;
public:
Data( const char *n)
{ cout << "Constructor of class Data\n";
name = new char[strlen(n)+1]; strcpy(name, n);
}
~Data()
{ cout << "Destructor of class Data for "
<< "object: " << name << endl;
delete [] name;
}
};
class Derived : public Base
{
private:
Data data;
public:
Derived( const char *n) : data(n)
{ cout << "Constructor of class Derived\n"; }
~Derived()
// implicit virtual
{ cout << "Destructor of class Derived\n"; }
};
int main()
{
Base *bPtr = new Derived("DEMO");
cout << "\nCall to the virtual Destructor!\n";
delete bPtr;
return 0;
}

DESTROYING DYNAMICALLY ALLOCATED OBJECTS

■

549

Dynamically created objects in a class hierarchy are normally handled by a base class
pointer. When such an object reaches the end of its lifetime, the memory occupied by
the object must be released by a delete statement.

Example: Car *carPtr;
carPtr = new PassCar("500",false,21,"Geo");
. . .
delete carPtr;

䊐 Destructor Calls
When memory is released, the destructor for an object is automatically called. If multiple
constructors were called to create the object, the corresponding destructors are called in
reverse order. What does this mean for objects in derived classes? The destructor of the
derived class is called first and then the destructor of the base class executed.
If you use a base class pointer to manage an object, the appropriate virtual methods of
the derived class are called. However, non-virtual methods will always execute the base
class version.
In the previous example, only the base class destructor for Car was executed. As the
PassCar destructor is not called, neither is the destructor called for the data member
passCarType, which is additionally defined in the derived class. The data member
passCarType is a string, however, and occupies dynamically allocated memory—
this memory will not be released.
If multiple objects are created dynamically in the derived class, a dangerous situation
occurs. More and more unreferenced memory blocks will clutter up the main memory
without you being able to reallocate them—this can seriously impact your program’s
response and even lead to external memory being swapped in.

䊐 Virtual Destructors
This issue can be solved simply by declaring virtual destructors. The opposite page shows
how you would define a virtual destructor for the Car class. Just like any other virtual
method, the appropriate version of the destructor will be executed. The destructors from
any direct or indirect base class then follow.
A class used as a base class for other classes should always have a virtual destructor
defined. Even if the base class does not need a destructor itself, it should at least contain
a dummy destructor, that is, a destructor with an empty function body.

550

■

CHAPTER 25

■

POLYMORPHISM

VIRTUAL METHOD TABLE

VMT for the classes Car and PassCar

Object of type Car
VMT of Car

VMT pointer

nr

12345

producer Audi

Address of Car::display()
Address of Car::~Car()

Object of type PassCar
VMT pointer

nr

54321

producer Geo
passCarType 500
sunRoof

true

VMT of PassCar
Address of

PassCar::display()
Address of

PassCar::~PassCar()
Object of type PassCar
VMT pointer

nr

98765

producer VW
passCarType GTI
sunRoof

false

VIRTUAL METHOD TABLE

■

551

䊐 Static Binding
When a non-virtual method is called, the address of the function is known at time of
compilation. The address is inserted directly into the machine code. This is also referred
to as static or early binding.
If a virtual method is called via an object’s name, the appropriate version of this
method is also known at time of compilation. So this is also a case of early binding.

䊐 Dynamic Binding
However, if a virtual method is called by a pointer or reference, the function that will be
executed when the program is run is unknown at time of compilation. The statement

Example: carPtr->display();
could execute different versions of the display() method, depending on the object
currently referenced by the pointer.
The compiler is therefore forced to create machine code that does not form an
association with a particular function until the program is run. This is referred to as late
or dynamic binding.

䊐 VMT
Dynamic binding is supported internally by virtual method tables (or VMT for short). A
VMT is created for each class with at least one virtual method—that is, an array with the
addresses of the virtual methods in the current class.
Each object in a polymorphic class contains a VMT pointer, that is, a hidden pointer
to the VMT of the corresponding class. Dynamic binding causes the virtual function call
to be executed in two steps:
1. The pointer to the VMT in the referenced object is read.
2. The address of the virtual method is read in the VMT.
In comparison with static binding, dynamic binding does have the disadvantage that
VMTs occupy memory. Moreover, program response can be impacted by indirect
addressing of virtual methods.
However, this is a small price to pay for the benefits. Dynamic binding allows you to
enhance compiled source code without having access to the source code. This is particularly important when you consider commercial class libraries, from which a user can
derive his or her own classes and virtual function versions.

552

■

CHAPTER 25

■

POLYMORPHISM

DYNAMIC CASTS

Using dynamic casts
// cast_t.cpp
// Dynamic casts in class hierarchies.
// ---------------------------------------------------#include "car.h"
bool inspect( PassCar* ),
// Inspection of
inspect(Truck* );
// car types.
bool separate(Car* );
// Separates cars
// for inspection.
int main()
{
Car* carPtr = new PassCar("520i", true, 3265, "BMW");
Truck* truckPtr = new Truck(8, 7.5, 5437, "Volvo");
// ... to test some casts and ...
separate(carPtr);
separate(truckPtr);
return 0;
}
bool separate( Car* carPtr)
{
PassCar* PassCarPtr = dynamic_cast(carPtr);
if( PassCarPtr != NULL)
return inspect( PassCarPtr);
Truck* truckPtr = dynamic_cast(carPtr);
if( truckPtr != NULL)
return inspect( truckPtr);
return false;
}
bool inspect(PassCar* PassCarPtr)
{
cout << "\nI inspect a passenger car!" << endl;
cout << "\nHere it is:";
PassCarPtr->display();
return true;
}
bool inspect(Truck* truckPtr)
{
cout << "\nI inspect a truck!" << endl;
cout << "\nHere it is:";
truckPtr->display();
return true;
}

✓

NOTE
The compiler’s option “Run Time Type Information (RTTI)” must be activated, for example, under
Project/Settings. The GNU compiler activates these options automatically.

DYNAMIC CASTS

■

553

䊐 Safety Issues in Downcasting
Downcasts in class hierarchies are unsafe if you use a C cast or the static cast operator. If
the referenced object does not correspond to the type of the derived class, fatal runtime
errors can occur.
Given that carPtr is a pointer to the base class Car, which is currently pointing to a
PassCar type, the statement

Example: Truck * truckPtr = static_cast(carPtr);
will

not

cause

a

compiler

error.

But

the

following

statement,

truckPtr->setAxles(10); could cause the program to crash.

䊐 The dynamic_cast<> Operator
You can use the cast operator dynamic_cast<> to perform safe downcasting in polymorphic classes. At runtime the operator checks whether the required conversion is
valid or not.

Syntax:

dynamic_cast(expression)

If so, the expression expression is converted to the target type type. The target type
must be a pointer or reference to a polymorphic class or a void pointer. If it is a pointer
type, expression must also be a pointer type. If the target type is a reference,
expression must identify an object in memory.

䊐 Examples
Given a pointer carPtr to the base class Car, the statement

Example: Truck* truckPtr = dynamic_cast(carPtr);
performs a downcast to the derived Truck class, provided the pointer carPtr really
identifies a Truck type object. If this is not so, the dynamic_cast operator
will return a NULL pointer.
Given that cabrio is a PassCar type object, the following statements

Example: Car& r_car = cabrio;
PassCar& r_passCar=dynamic_cast(r_car);

perform a dynamic cast to the “reference to PassCar” type. In any other case, that is, if
the reference r_car does not identify a PassCar type object, an exception of the
bad_cast type is thrown.
The dynamic cast can also be used for upcasting. The classes involved do not need to
be polymorphic in this case. However, type checking is not performed at runtime. An
erroneous upcast is recognized and reported by the compiler.

■

CHAPTER 25

■

exercise s

554

POLYMORPHISM

EXERCISES

Menu options

* * * Car Rental Management * * *

P = Add a passenger car
T = Add a truck
D = Display all cars
Q = Quit
Your choice:

Different versions of method insert()

Add a new passenger car:

bool insert(const string& tp, bool sr,
long n, const string& prod);
Add a new truck:

bool insert(int a, double t, long n,
const string& prod);

EXERCISES

■

555

Exercise 1
Modify the vehicle management program to allow an automobile rental company
to manage its fleet of automobiles. First, define a class called CityCar that
contains an array of pointers to the 100 objects in the Car class.This also allows
you to store pointers to objects of the derived class types PassCar and Truck.
The objects themselves will be created dynamically at runtime.
■

■

■

■

Define a class CityCar with an array of pointers to the Car class and an
int variable for the current number of elements in the array.
The constructor will set the current number of array elements to 0.
The destructor must release memory allocated dynamically for the
remaining objects. Make sure that you use a virtual destructor definition
in the base class Car to allow correct releasing of memory for trucks and
passenger vehicles.
Implement two versions of the insert() method using the prototype
shown opposite. Each version will allocate memory to an object of the
appropriate type—that is of the PassCar or Truck class—and use the
arguments passed to it for initialization.The method should return false
if it is impossible to enter another automobile (that is, if the array is full),
and true in all other cases.
The display() method outputs the data of all vehicles on screen.To
perform this task it calls the existing display() method for each object.
Create a new function called menu() and store this function in a new
source file.The function will display the menu shown opposite, read, and
return the user’s choice.
Additionally, write two functions, getPassCar() and getTruck(), which
read the data for a car or a truck from the keyboard and write the data
into the appropriate arguments.
Create an object of the CityCar type in your main function. Insert one
car and one truck.These will be the first vehicles of the company’s fleet.
If a user chooses “Add car” or “Add truck,” your program must read the
data supplied and call the appropriate version of insert().

556

■

CHAPTER 25

POLYMORPHISM

Dialog with the receptionist
In function record()

Please type next article?
0 = No more articles
1 = Fresh food
2 = Prepacked article
?

Loop of main()

Another customer (y/n)?
If yes

to record

EXERCISES

■

557

Exercise 2
An automatic checkout system for a supermarket chain needs to be completed.
■

■

■

Declare the virtual methods scanner() and printer() in the base class
Product.Also define a virtual destructor.
Write the record() function, which registers and lists products purchased in the store in a program loop.
The function creates an array of 100 pointers to the base class, Product.
The checkout assistant is prompted to state whether a prepacked or
fresh food item is to be scanned next. Memory for each product scanned
is allocated dynamically and referenced by the next pointer in the array.
After scanning all the available items, a sequential list is displayed.The
prices of all the items are added and the total is output at the end.
Now create an application program to simulate a supermarket checkout.
The checkout assistant is prompted in a loop to state whether to define a
new customer. If so, the record() function is called; if not, the program
terminates.

■

CHAPTER 25

■

solutions

558

POLYMORPHISM

SOLUTIONS

Exercise 1
// ---------------------------------------------------// car.h : Defines the base class Car and
//
the derived classes PassCar and Truck
// -------------------------------------------------------#ifndef _CAR_H_
#define _CAR_H_
#include 
#include 
using namespace std;
class Car
{
private:
long
nr;
string producer;
public:
Car( long n = 0L, const string& prod = "");
virtual ~Car() {}
// Virtual destructor.
// Access methods:
long getNr(void) const { return nr; }
void setNr( long n ) { nr = n; }
const string& getProd() const { return producer; }
void setProd(const string& p){ producer = p; }
virtual void display() const;

// Display a car

};
// The derived classes PassCar and Truck are unchanged
// (see Chapter 23).
#endif
//
//
//
//

-----------------------------------------------------car.cpp
Implements the methods of Car, PassCar, and Truck
------------------------------------------------------

// Unchanged (see Chapter 23).
//

SOLUTIONS

// -----------------------------------------------------// city.h : Defines the CityCar class
// -----------------------------------------------------#ifndef _CITY_H_
#define _CITY_H_
#include "car.h"
class CityCar
{
private:
Car* vp[100];
int cnt;
public:
CityCar(){ cnt = 0;}
~CityCar();
bool insert(const string&
long n, const
bool insert(int a, double
long n, const

tp, bool sr,
string& prod);
t,
string& prod);

void display() const;
};
#endif // _CITY_H
// -----------------------------------------------------// city.cpp : Methods of class CityCar
// -----------------------------------------------------#include "city.h"
CityCar::~CityCar()
{
for(int i=0; i < cnt; ++i)
delete vp[i];
}
// Insert a passenger car:
bool CityCar::insert(const string& tp, bool sr,
long n, const string& prod)
{
if( cnt < 100)
{
vp[cnt++] = new PassCar( tp, sr, n, prod);
return true;
}
else
return false;
}

■

559

560

■

CHAPTER 25

POLYMORPHISM

// Insert a truck:
bool CityCar::insert( int a, double t,
long n, const string& prod)
{
if( cnt < 100)
{
vp[cnt++] = new Truck( a, t, n, prod);
return true;
}
else
return false;
}
void CityCar::display() const
{
cin.sync(); cin.clear();
// No previous input
for(int i=0; i < cnt; ++i)
{
vp[i]->display();
if((i+1)%4 == 0) cin.get();
}
}
// -------------------------------------------------// city_t.cpp : Test the CityCar class
// -------------------------------------------------#include "city.h"
char menu(void);
void getPassCar(string&, bool&, long&, string&);
void getTruck(int&, double&, long&, string&);
int main()
{
CityCar carExpress;
string tp, prod; bool
int
a;
long
n;

sr;
double t;

// Two cars are already present:
carExpress.insert(6, 9.5, 54321, "Ford");
carExpress.insert("A-class", true, 54320, "Mercedes");
char choice;
do
{
choice = menu();
switch( choice )
{
case 'Q':
case 'q': cout << "Bye Bye!" << endl;
break;

SOLUTIONS

case 'P':
case 'p': getPassCar(tp, sr, n, prod);
carExpress.insert(tp, sr, n, prod
break;
case 'T':
case 't': getTruck(a, t, n, prod);
carExpress.insert(a, t, n, prod);
break;
case 'D':
case 'd': carExpress.display();
cin.get();
break;
default: cout << "\a";
// Beep
break;
}
}while( choice != 'Q'
return 0;

■

);

&& choice != 'q');

}
char menu()
// Input a command.
{
cout << "\n * * * Car Rental Management * * *\n\n"
char c;
cout <<
"\n
P = Add a passenger car "
<<
"\n
T = Add a truck "
<<
"\n
D = Display all cars "
<<
"\n
Q = Quit the program "
<< "\n\nYour choice: ";
cin >> c;
return c;
}
void getPassCar(string& tp, bool& sr, long& n,string& prod)
{
char c;
cin.sync(); cin.clear();
cout << "\nEnter data for passenger car:" << endl;
cout << "Car type:
"; getline(cin, tp);
cout << "Sun roof (y/n):
"; cin >> c;
if(c == 'y' || c == 'Y')
sr = true;
else
sr = false;
cout << "Car number:
"; cin >> n;
cin.sync();
cout << "Producer:
"; getline(cin, prod);
cin.sync(); cin.clear();
}

561

562

■

CHAPTER 25

POLYMORPHISM

void getTruck(int& a, double& t, long& n, string& prod)
{
cout << "\nInput data for truck:" << endl;
cout << "Number of axles:
"; cin >> a;
cout << "Weight in tons:
"; cin >> t;
cout << "Car number:
";
cin >> n;
cin.sync();
cout << "Producer:
"; getline(cin, prod);
cin.sync();
}

Exercise 2
//
//
//
//
//

---------------------------------------------------product.h : Defining the classes
Product, PrepackedFood, and FreshFood
---------------------------------------------------. . .

class Product
{
private:
long
bar;
string name;
public:
Product(long b = 0L, const string& s = "")
: bar(b), name(s)
{ }
//

Access methods as previously used.
virtual void scanner();
// Virtual now!
virtual void printer() const;

};
// The classes PrepackedFood and FreshFood are unchanged!
// Refer to the solutions in Chapter 23.

SOLUTIONS

// -----------------------------------------------------// counter.cpp : Simulates a checkout desk
// -----------------------------------------------------#include "product.h"
void record();
int main()
{
cout << "\nHere is a checkout desk!" << endl;
char c;
while(true)
{
cin.sync();
cout << "\nAnother customer (y/n)?
";
cin >> c;
if(c == 'y' || c == 'Y')
record();
else
break;
}
return 0;
}

// -----------------------------------------------------// record() : Records the articles bought by a customer
//
and the total price.
void record()
{
Product* v[100];
int x, i, cnt = 0;
double sum = 0.0;
for (i = 0; i < 100; i++)
{
cin.sync();
cout << "\nWhat is the next article?" << endl;
cout << " 0 = No more article\n"
<< " 1 = Fresh Food\n"
<< " 2 = Prepacked article\n"
<< "? " ;
cin >> x;
if( x <= 0 || x >= 3)
break;

■

563

564

■

CHAPTER 25

POLYMORPHISM

switch(x)
{
case 2:
v[i] = new PrepackedFood;
v[i]->scanner();
sum += ((PrepackedFood*)v[i])->getPrice();
break;
case 1:
v[i] = new FreshFood;
v[i]->scanner();
sum += ((FreshFood*)v[i])->getPrice()
* ((FreshFood*)v[i])->getWght();
break;
}
}
cnt = i;
for( i=0; i < cnt; i++)
v[i]->printer();

// Output

cout << "\n-----------------------------"
<< fixed << setprecision(2)
<< "\nTotal price:
" << sum << endl;
}

chapter

26

Abstract Classes
This chapter describes how abstract classes can be created by defining
pure virtual methods and how you can use abstract classes as a
polymorphic interface for derived classes.To illustrate this we will be
implementing an inhomogeneous list, that is, a linked list whose elements
can be of various class types.

565

566

■

CHAPTER 26

■

ABSTRACT CLASSES

PURE VIRTUAL METHODS

The base class Coworker
// Coworker.h: Defining the abstract class Coworker.
// ---------------------------------------------------#ifndef _COWORKER_H
#define _COWORKER_H
#include 
#include 
using namespace std;
class Coworker
{
private:
string name;
// more information
public:
Coworker( const string& s = ""){ name = s; }
virtual ~Coworker() {}
// Destructor
const string& getName() const{ return name; }
void setName( const string& n){ name = n; }
virtual void display() const;
virtual double income() const = 0;
virtual Coworker& operator=(const Coworker&);
};
#endif

✓

NOTE
The virtual operator function for the assignment will be described in the section on ”Virtual
Assignments.”

PURE VIRTUAL METHODS

■

567

䊐 Motivation
Virtual methods are declared in the base class to ensure that they are available in any
derived classes via the common class interface. It may happen that they rarely perform
any useful tasks in the base class. For example, a destructor in a base class does not need
to perform any explicit cleaning-up operations.
In this case, of course, you can define a virtual dummy method whose address is
entered in the VMT of the base class. However, this creates op-code for a function that
should never be called. It makes more sense not to define a function like this. And this is
where C++ steps in and gives you the opportunity of declaring pure virtual methods.

䊐 Declaration
When a pure virtual method is declared, the method is identified by adding the expression = 0.

Example: virtual void demo()=0;

// pure virtual

This informs the compiler that there is no definition of the demo() method in the class.
A NULL pointer is then entered in the virtual method table for the pure virtual method.

䊐 The Base Class Coworker
The opposite page shows a definition of the Coworker class, which was designed to represent human resources data for a company. The class is used as a base class for various
employees, blue-collar, white-collar, and freelancers.
To keep things simple the Coworker class has only a name as a data member. However, it could also contain the address of an employee, or the division where the
employee works.
The Coworker class does not comprise data members to represent an employee’s
salary. It makes more sense to store data like this in derived classes where the hourly
wage and number of hours worked by a blue-collar worker and the monthly salary for a
white-collar worker are also defined. The income() method is therefore not defined for
the base class and can be declared as a pure virtual method.

568

■

CHAPTER 26

■

ABSTRACT CLASSES

ABSTRACT AND CONCRETE CLASSES

The derived class Laborer
// coworker.h:
Extending the headerfile.
// -------------------------------------------------class Laborer : public Coworker
{
private:
double wages;
// Pay per hour
int
hr;
public:
Laborer(const string& s="", double w=0.0, int h=0)
: Coworker(s), wages(w), hr(h){ }
double getWages() const { return wages; }
void
setWages( double w ){ wages = w; }
int
void

getHr() const { return hr; }
setHr(int h ) { hr = h; }

void
display() const;
double income() const;
Laborer& operator=(const Coworker&);
Laborer& operator=(const Laborer&);
};

✓

NOTE
The operator functions for the assignments are discussed in the section ”Virtual Assignments.”

ABSTRACT AND CONCRETE CLASSES

■

569

䊐 Concrete or Abstract?
If a class comprises pure virtual methods, you cannot create objects of this class type.

Example: Coworker worker("Black , Michael");
The compiler will issue an error message here, as the Coworker class contains the
pure virtual method income(). This avoids calling a method for worker that still
needs to be defined.
A class that does not allow you to create any objects is referred to as an abstract class.

✓

NOTE
A class with pure virtual methods is an abstract class.

In contrast, a class that allows you to create objects is referred to as a concrete class.

䊐 Deriving Abstract Classes
When a class is derived from an abstract class, it inherits all the methods the base class
contains, particularly the pure virtual methods. If all of these pure virtual methods are
implemented in the derived class, you can then create an object of the derived class type.

✓

NOTE
A class derived from a class containing pure virtual methods is a concrete class, if it contains a definition
for each pure virtual function.

This means you need to implement the income() method in the Laborer class
shown opposite. Since the hourly wage and the number of hours worked are both defined
for blue-collar workers, it is possible to implement that method.

Example: double Laborer::income()
{
return ( wages * hr );
}

A class derived from a concrete class can again contain pure virtual methods, due to
additional definitions in the derived class. In other words, an abstract class can be
derived from a concrete class.
An abstract class does not necessarily need to contain pure virtual functions. If the
class contains a protected constructor, objects of the class type cannot be created.
The constructor can only be called then by methods in derived classes. A constructor of
this type normally acts as base initializer, when an object of a derived class type is created.

570

■

CHAPTER 26

■

ABSTRACT CLASSES

POINTERS AND REFERENCES TO ABSTRACT CLASSES

The derived class Employee
// coworker.h:
Extending the headerfile.
// -------------------------------------------------class Employee : public Coworker
{
private:
double salary;
// Pay per month
public:
Employee( const string& s="", double sa = 0.0)
: Coworker(s), salary(g){ }
double getSalary() const { return salary; }
void
setSalary( double sa){ salary = sa; }
void
display() const;
double income()const { return salary; }
Employee& operator=( const Coworker& );
Employee& operator=( const Employee& );
};

Sample program
// coworker_t.cpp : Using the Coworker classes.
// ------------------------------------------------#include "coworker.h"
int main()
{
Coworker* felPtr[2];
felPtr[0] = new Laborer("Young, Neil",45., 40);
felPtr[1] = new Employee("Smith, Eve", 3850.0);
for( int i = 0; i < 2; ++i)
{
felPtr[i]->display();
cout << "\nThe income of " << felPtr[i]->getName()
<< " : " << felPtr[i]->income() << endl;
}
delete felPtr[0]; delete felPtr[1];
return 0;
}

POINTERS AND REFERENCES TO ABSTRACT CLASSES

■

571

Although you cannot define objects for abstract classes, you can declare pointers and references to abstract classes.

Example: Coworker *felPtr, &coRef;
The pointer felPtr is a base class pointer that can address objects belonging to derived
concrete classes. The reference coRef can also address objects of this type.

䊐 References to Abstract Base Classes
References to base classes are often used as parameters in functions. The copy constructor in the Coworker class is just one of them.

Example: Coworker( const Coworker& );
The copy constructor expects an object belonging to a derived class, since the base class
is abstract.
The assignment in the Coworker class has a reference as a parameter and returns a
reference to the abstract class.

䊐 Pointers to Abstract Base Classes
Pointers to base classes are generally used to address dynamically allocated objects. If the
base class is abstract, you can only allocate memory for objects belonging to derived, concrete classes.

Example: Coworker* felPtr;
felPtr = new Laborer("Young, Neil",45.,40);
cout << felPtr->income();

Since the income()method is virtual, a corresponding function found in the derived
class Laborer is executed.

䊐 Polymorphic Interface
Defining pure virtual methods also determines interfaces for general operations, although
the interfaces still need to implemented in the derived classes. If a derived class contains
its own definition of a virtual method, this version will also be executed if an object is
referenced by a base class pointer or reference. Abstract classes are therefore also referred
to as polymorphic interfaces to derived classes.
The opposite page shows the definition of the Employee class, which was also
derived from the abstract class Coworker. The operator functions for the assignments
are discussed and implemented in the following section.

572

■

CHAPTER 26

■

ABSTRACT CLASSES

VIRTUAL ASSIGNMENT

Assignment for class Coworker
// Virtual Assignment in the base class
Coworker& operator=(const Coworker & m)
{
if( this != &m )
// No self assignment
name = m.name;
return *this;
}

Assignments for class Employee
// Redefinition: virtual
Employee& operator=(const Coworker& m)
{
if( this != &m )
// No self assignment
{
Coworker::operator=( m );
salary = 0.0;
}
return *this;
}
// Standard Assignment: not virtual
Employee& operator=(const Employee& a)
{
if( this != &a )
{
Coworker::operator=( a );
salary = a.salary;
}
return *this;
}

✓

NOTE
Redefining the virtual operator function operator=(), which returns a reference to the derived class,
is not yet supported by all compilers. In this case the return type must be a reference to the base class
Coworker.

VIRTUAL ASSIGNMENT

■

573

䊐 Virtual Operator Functions
Operator functions implemented as methods can also be virtual. In this case, you can
ensure that the right version of an operator function will be executed when using a
pointer or reference to a base class to address a derived class object.
One example of this is the operator function for an assignment. If the function declaration is not virtual, and if the function is called via a base class pointer, only the base
data of the object is overwritten. Any additional data members of the derived class
remain unchanged.

䊐 Using Virtual Assignments
The assignment was declared virtual in the Coworker base class. The derived classes
Laborer and Employee both contain their own versions. Thus, in the following

Example: void cpy(Coworker& a,const Coworker& b)
{ a = b; }

the assignment of the Employee class is executed, if an object of this class type is the
first argument passed to it. If the object is a Laborer type, the assignment of the
Laborer class is performed.
In the case of the cpy()function, you can therefore assign two objects of any class,
including classes derived at a later stage, without having to modify the function itself!
However, you definitely need to define a version of the assignment for each derived class.

䊐 Redefining the Standard Assignment
When you define a new version for a virtual method in a derived class, this implies using
the signature of the original method. Since the standard assignment of a derived class has
a signature of its own, it is not virtual. The standard assignment for the Laborer class
has the following prototype:

Example: Laborer& operator=(const Laborer&);
The type const Laborer& is different from the const Coworker& type of the
parameter in the virtual operator function of the base class. The standard assignment
thus masks the virtual assignment in the base class. This gives rise to two issues:
■
■

the virtual operator function for the assignment must be defined for every derived
class
to ensure that the standard assignment is also available, the standard assignment
must also be redefined in every derived class.

574

■

CHAPTER 26

■

ABSTRACT CLASSES

APPLICATION: INHOMOGENEOUS LISTS

The abstract base class Cell and derived classes

// cell.h: Defining the classes Cell, BaseEl, and DerivedEl
// -------------------------------------------------------#ifndef _CELL_
#define _CELL_
#include 
#include 
using namespace std;
class Cell
{
private:
Cell* next;
protected:
Cell(Cell* suc = NULL ){ next = suc; }
public:
virtual ~Cell(){ }
// Access methods to be declared here.
virtual void display() const = 0;
};
class BaseEl : public Cell
{
private:
string name;
public:
BaseEl( Cell* suc = NULL, const string& s = "")
: Cell(suc), name(s){}
// Access methods would be declared here.
void display() const;
};
class DerivedEl : public BaseEl
{
private:
string rem;
public:
DerivedEl(Cell* suc = NULL,const string& s="",
const string& b="")
: BaseEl(suc, s), rem(b) { }
// Access methods would be declared here.
void display() const;
};
#endif

APPLICATION: INHOMOGENEOUS LISTS

■

575

䊐 Terminology
Let’s look into implementing an application that uses an inhomogeneous list. An inhomogeneous list is a linear list whose elements can be of different types. If the data you
need to store consists of objects in a class hierarchy, one list element could contain an
object belonging to the base class, whereas another could contain an object of a derived
class.
Due to implicit type conversions in class hierarchies, you can use the base class pointers to manage the list elements, that is, you can manage the elements in a linked list.
The following graphic illustrates this scenario:

1st list element
Info

2nd list element

Info

Info
Pointer

3rd list element

Pointer

Pointer

䊐 Representing List Elements
To separate the management of list elements from the information contained in the list,
we have defined an abstract class called Cell as a base class for all list elements. The
class contains a pointer of type Cell* as the data member used to link list elements.
Since Cell type objects are not be created, the constructor in the Cell class has a
protected declaration.
The Cell class does not contain any data that might need to be output. However,
each class derived from Cell contains data that need to be displayed. For this reason,
Cell contains a declaration of the pure virtual method display(), which can be modified for multiple derivations.
The classes BaseEl and DerivedEl, which are derived from Cell, represent list
elements used for storing information. To keep things simple, the BaseEl class contains
only a name, and the DerivedEl class additionally contains a comment. The public
declaration section contains a constructor and access method declarations. In addition, a
suitable version of the display() method is defined. Both classes are thus concrete
classes.

576

■

CHAPTER 26

■

ABSTRACT CLASSES

IMPLEMENTING AN INHOMOGENEOUS LIST

Defining class InhomList
// List.h: Defining the class InhomList
// --------------------------------------------#ifndef _LIST_H
#define _LIST_H
#include "cell.h"
class InhomList
{ private:
Cell* first;
protected:
Cell* getPrev(const string& s);
void insertAfter(const string& s,Cell* prev);
void insertAfter(const string& s,
const string& b,Cell* prev);
public: // Constructor, Destructor etc....
void insert(const string& n);
void insert(const string& n, const string& b);
void displayAll() const;
};
#endif

Inserting a list element in the middle
New
element:

Info

Info

Info
Pointer

Info

Pointer

Pointer

Definition of insertAfter() version
void InhomList::insertAfter(const string& s, Cell* prev)
{
if( prev == NULL )
// Insert at the beginning,
first = new BaseEl( first, s);
else
// middle, or end.
{
Cell* p = new BaseEl(prev->getNext(), s);
prev->setNext(p);
}
}

IMPLEMENTING AN INHOMOGENEOUS LIST

■

577

䊐 The InhomList Class
The inhomogeneous list must allow sorted insertion of list elements. It is no longer sufficient to append elements to the end of the list; instead, the list must allow insertion at
any given point. A pointer to the first element in the list is all you need for this task. You
can then use the pointer to the next list element to access any given list element.
The definition of the InhomList class is shown opposite. A pointer to Cell has
been declared as a data member. The constructor has very little to do. It simply sets the
base class pointer to NULL, thus creating an empty list.
The list will be sorted by name. When inserting a new element into the list, the insertion point—that is the position of the element that will precede the new element—must
first be located. In our example, we first locate the immediate lexicographical predecessor. The getPrev() method, shown opposite, performs the search and returns either
the position of the predecessor or NULL if there is no predecessor. In this case, the new
list element is inserted as the first element in the list.

䊐 Inserting a New List Element
After finding the insertion position you can call the insertAfter() method that allocates memory for a new list element and inserts the element into the list. There are two
cases that need to be looked at:
1. If the new element has to be inserted at the start of the list, what was originally
the first element now becomes the second. The new element becomes the first
element. The first pointer thus needs updating.
2. If the new element is inserted at any other position in the list, the first pointer
is not affected. Instead, you have to modify two pointers. The pointer in the preceding list element must be pointed at the new element and the pointer in the
new element must be pointed at what was formerly the successor of the preceding
element. This situation also applies in cases where the successor was a NULL
pointer, in other words, when the new element is appended to the list.
Since the list contains objects of the BaseEl and DerivedEl types, the
insertAfter() method has been overloaded with two versions. They differ only in
different calls to the new operator.
The insert() method was overloaded for the same reason. Both versions first call
the getPrev() method and the corresponding version of the insertAfter()
method.

■

CHAPTER 26

■

exercise

578

ABSTRACT CLASSES

EXERCISE

The complete class InhomList
class InhomList
{
private:
Cell* first;
protected:
Cell* getPrev(const string& s);
Cell* getPos(const string& s);
void
void
void

insertAfter(const string& s, Cell* prev);
insertAfter(const string& s,const string& b,
Cell* prev);
erasePos(Cell* pos);

public:
InhomList(){ first = NULL; }
InhomList(const InhomList& src);
~InhomList();
InhomList& operator=( const InhomList& src);
void
void
void

insert(const string& n);
insert(const string& n, const string& b);
erase(const string& n);

void displayAll() const;
};

EXERCISE

■

579

Exercise
Modify and complete the definition of the class InhomList, which represents an
inhomogeneous list.
■

■

■

■

■

■

■

■

■

Write the destructor for the InhomList class.The destructor releases
the memory occupied by the remaining list elements.
Implement the getPrev() method and both versions of the insert()
and insertAfter() methods.The algorithm needed for inserting list elements was described in the section “Implementing an Inhomogeneous
List.”
Implement the displayAll() method, which walks through the list
sequentially, outputting each element.
Test insertion and output of list elements. Check whether the comments
on the objects are output, if present.
Define the getPos() method, which locates the position of an element
to be deleted. If the element is in the list, its address is returned. Otherwise a NULL pointer is returned.
Write the erasePos() method, which deletes a list element at a given
position. Pay attention to whether the element to be deleted is the first
or any other element in the list. Since the destructor for Cell was
declared virtual, only one version of the deletePos() method is necessary.
Define the erase() method, which deletes a list element with a given
name from the list.
Test deletion of list elements. Continually display the remaining elements
in the list to be certain.
Now implement the copy constructor and assignment. Use the insert()
to construct the list, calling the applicable version of the method.You can
call the typeid() operator to ascertain the type of the list element
currently to be inserted.The operator is declared in the header file
typeinfo.
Example: if( typeid(*ptr) == typeid(DerivedEl)) ...

■

The expression is true if ptr references a DerivedEl type object.
Then test the copy constructor and the assignment

■

CHAPTER 26

■

solution

580

ABSTRACT CLASSES

SOLUTION

// -----------------------------------------------------// cell.h
// Defines the classes Cell, BaseEl, and DerivedEl.
// -----------------------------------------------------#ifndef _CELL_
#define _CELL_
#include 
#include 
using namespace std;
class Cell
{
private:
Cell* next;
protected:
Cell(Cell* suc = NULL ){ next = suc; }
public:
virtual ~Cell(){ }
Cell* getNext() const { return next; }
void setNext(Cell* suc) { next = suc; }
virtual void display() const = 0;
};
class BaseEl : public Cell
{
private:
string name;
public:
BaseEl( Cell* suc = NULL, const string& s = "")
: Cell(suc), name(s){}
// Access methods:
void
setName(const string& s){ name = s; }
const string& getName() const { return name; }
void display() const
{
cout << "\n--------------------------------"
<< "\nName:
" << name << endl;
}
};

SOLUTION

class DerivedEl : public BaseEl
{
private:
string rem;
public:
DerivedEl(Cell* suc = NULL,
const string& s="", const string& b="")
: BaseEl(suc, s), rem(b){ }
// Access methods:
void
setRem(const string& b){ rem = b; }
const string& getRem() const { return rem; }
void display() const
{
BaseEl::display();
cout << "Remark:
" << rem << endl;
}
};
#endif
// -----------------------------------------------------// List.h : Defines the class InhomList
// -----------------------------------------------------#ifndef _LIST_H_
#define _LIST_H_
#include "cell.h"
class InhomList
{
private:
Cell* first;
protected:
Cell* getPrev(const string& s);
Cell* getPos( const string& s);
void insertAfter(const string& s, Cell* prev);
void insertAfter(const string& s,const string& b,
Cell* prev);
void erasePos(Cell* pos);
public:
InhomList(){ first = NULL; }
InhomList(const InhomList& src);
~InhomList();
InhomList& operator=( const InhomList& src);
void insert(const string& n);
void insert(const string& n, const string& b);
void erase(const string& s);
void displayAll() const;
};
#endif

■

581

582

■

CHAPTER 26

ABSTRACT CLASSES

// -----------------------------------------------------// List.cpp : The methods of class InhomList
// -----------------------------------------------------#include "List.h"
#include 
// Copy constructor:
InhomList::InhomList(const InhomList& src)
{
// Append the elements from src to the empty list.
first = NULL;
Cell *pEl = src.first;
for( ; pEl != NULL; pEl = pEl->getNext() )
if(typeid(*pEl) == typeid(DerivedEl))
insert(dynamic_cast(pEl)->getName(),
dynamic_cast(pEl)->getRem());
else
insert(dynamic_cast(pEl)->getName());
}
// Assignment:
InhomList& InhomList::operator=(const InhomList& src)
{
// To free storage for all elements:
Cell *pEl = first,
*next = NULL;
while( pEl != NULL )
{
next = pEl->getNext();
delete pEl;
pEl = next;
}
first = NULL;

// Empty list

// Copy the elements from src to the empty list.
pEl = src.first;
for( ; pEl != NULL; pEl = pEl->getNext() )
if(typeid(*pEl) == typeid(DerivedEl))
insert(dynamic_cast(pEl)->getName(),
dynamic_cast(pEl)->getRem());
else
insert(dynamic_cast(pEl)->getName());
return *this;
}

SOLUTION

// Destructor:
InhomList::~InhomList()
{
Cell *pEl = first,
*next = NULL;
while( pEl != NULL )
{
next = pEl->getNext();
delete pEl;
pEl = next;
}
}
Cell* InhomList::getPrev(const string& n)
{
Cell *pEl = first,
*prev = NULL;
while( pEl != NULL )
{
if( n > dynamic_cast(pEl)->getName() )
{
prev = pEl;
pEl = pEl->getNext();
}
else
return prev;
}
return prev;
}
Cell* InhomList::getPos( const string& n)
{
Cell *pEl = first;
while( pEl != NULL &&
(n != dynamic_cast(pEl)->getName()))
pEl = pEl->getNext();
if( pEl != NULL &&
n == dynamic_cast(pEl)->getName())
return pEl;
else
return NULL;
}
void InhomList::insertAfter(const string& s, Cell* prev)
{
if( prev == NULL )
// Insert at the beginning:
first = new BaseEl( first, s);
else
// In the middle or at the end:
{ Cell* p = new BaseEl(prev->getNext(), s);
prev->setNext(p);
}
}

■

583

584

■

CHAPTER 26

ABSTRACT CLASSES

void InhomList::insertAfter( const string& s,
const string& b, Cell* prev)
{
if( prev == NULL )
// Insert at the beginning:
first = new DerivedEl( first, s, b);
else
// In the middle or at the end:
{
Cell* p = new DerivedEl(prev->getNext(), s, b);
prev->setNext(p);
}
}
void InhomList::insert(const string& n)
{
Cell* pEl = getPrev(n);
insertAfter(n, pEl);
}
void InhomList::insert(const string& n, const string& b)
{
Cell* pEl = getPrev(n);
insertAfter(n, b, pEl);
}
void InhomList::erasePos(Cell* pos)
{
Cell* temp;
if( pos != NULL)
if( pos == first )
// Delete the first element
{
temp = first;
first = first->getNext();
delete temp;
}
else
// Delete from the middle or at the end
{
// Get the predecessor
temp = getPrev( dynamic_cast(pos)
->getName());
if(temp != NULL)
// and bend pointer.
temp->setNext(pos->getNext());
delete pos;
}
}
void InhomList::erase(const string& n)
{
erasePos( getPos(n));
}

SOLUTION

void InhomList::displayAll() const
{
Cell* pEl = first;
while(pEl != NULL)
{
pEl->display();
pEl = pEl->getNext();
}
}
// -----------------------------------------------------// List_t.cpp : Tests the sorted inhomogeneous list
// -----------------------------------------------------#include "List.h"
int main()
{
InhomList liste1;
cout << "\nTo test inserting. " << endl;
liste1.insert("Bully, Max");
liste1.insert("Cheers, Rita", "always merry");
liste1.insert("Quick, John", "topfit");
liste1.insert("Banderas, Antonio");
liste1.displayAll(); cin.get();
cout << "\nTo test deleting. " << endl;
liste1.erase("Banderas, Antonio");
liste1.erase("Quick, John");
liste1.erase("Cheers, Rita");
liste1.displayAll(); cin.get();
cout << "\n----------------------------------"
<< "\nGenerate a copy and insert an element. "
<< endl;
InhomList liste2(liste1),
liste3;

// Copy constructor
// and an empty list.

liste2.insert("Chipper, Peter", "in good temper");
liste3 = liste2;
// Assignment
cout << "\nAfter the assignment: " << endl;
liste3.displayAll();
return 0;
}

■

585

This page intentionally left blank

chapter

27

Multiple Inheritance
This chapter describes how new classes are created by multiple
inheritance and explains their uses. Besides introducing you to creating
and destroying objects in multiply-derived classes, virtual base classes are
depicted to avoid ambiguity in multiple inheritance.

587

588

■

CHAPTER 27

■

MULTIPLE INHERITANCE

MULTIPLY-DERIVED CLASSES

The multiply-derived class MotorHome

Car

Home

MotorHome

Definition scheme for class MotorHome
class MotorHome : public Car, public Home
{
private:
// Additional private members here
protected:
// Additional protected members here
public:
// Additional public members here
};

MULTIPLY-DERIVED CLASSES

■

589

A class can contain not just one but several different base classes. In this case the class is
derived from multiple base classes in a process known as multiple inheritance.

䊐 The Multiply-Derived Class MotorHome
This class Car is used to represent vehicles and the class Home contains characteristic
values for an apartment, such as floor space, number and type of rooms, and typical operations, such as building, selling, or renting.
Using these two classes you can then derive the MotorHome class. The opposite page
shows the inheritance and definition schemes for the new class. An object of the
MotorHome class contains both the members of Car and the members of Home. More
specifically, the object contains two base sub-objects of type Car and Home.

䊐 Accessibility of Base Classes
Since the MotorHome class has two public base classes, it assumes the public interfaces of both classes. A MotorHome type object not only allows access to the additional
public members but to all the public members of the base classes Car and Home.
When defining a multiply-derived class, the accessibility, private, protected, or
public, must be defined separately for each base class. The MotorHome class could
have the public base class Car and the protected base class Home.

Example: class MotorHome:public Car,protected Home
{

. . . };

If the keyword is omitted, the base class will default to private.

Example: class MotorHome : public Car, Home
{ . . . };

This statement defines the public base class Car and the private base class Home.
This makes all the public members in Home private members of the derived class.
In multiple inheritance each public base class establishes an is relationship. This is
similar to simple inheritance. If the MotorHome class inherits two public base classes,
a motor-home is a special kind of motor vehicle and a special kind of home.

590

■

CHAPTER 27

■

MULTIPLE INHERITANCE

MULTIPLE INDIRECT BASE CLASSES

The multiple indirect base class Car

Car

Car

PassCar

Van

SUV

Definition scheme of class SUV
class SUV : public PassCar, public Van
{
// Here are additional methods and data members
};

MULTIPLE INDIRECT BASE CLASSES

■

591

䊐 Multiple Identical Base Classes
When multiply-derived classes are defined, a direct base class cannot be inherited more
than once. The following statement

Example: class B : public A, public A

// Error

{ . . . };

causes the compiler to issue an error message.
A class can be derived from several classes that have a common base class, however.
This class is then referred to as a multiple indirect base class.
The inheritance graph on the opposite page shows the multiply-derived class SUV,
which was derived from the classes PassCar and Van. Both base classes were themselves derived from the Car class. This makes Car a multiple indirect base class of the
SUV class.

䊐 Ambiguity
An object of the SUV class thus contains the members of Car twice. Access to members
of the Car class results in ambiguity.

Example: SUV mySUV(. . .);
cout << mySUV.getProd();

// Error

Both the base classes PassCar and Van contain a method called getProd(), which
they both inherited from the Car class. In this case the compiler cannot decide which
method is meant.
Ambiguity in the context of multiple inheritance is also possible when several base
classes contain members with identical names. If both the Home class and the Car class
contain a method called getNr(), the getNr() method cannot be correctly identified
in the following statement.

Example: MotorHome motorHome( . . .);
motorHome.getNr();

To resolve ambiguity of this kind, you can use the scope resolution operator to determine
which base class is meant.

Example: cout << motorHome.Home::getNr();
cout << mySUV.PassCar::getProd();

The getNr() method in the Home class is called first, followed by the getProd()
method inherited by PassCar from the Car class.

592

■

CHAPTER 27

■

MULTIPLE INHERITANCE

VIRTUAL BASE CLASSES

The virtual base class Car

Car

PassCar

Van

SUV

Definition scheme
class PassCar : public virtual Car
{
// Here are additional members
// of class PassCar
};
class Van : public virtual Car
{
// Here are additional members
// of class Van
};

VIRTUAL BASE CLASSES

■

593

䊐 Issues
You will not normally want a class created by multiple inheritance to contain multiple
instances of an indirect base class. Why should a station wagon contain two versions of
the manufacturer’s name or the chassis number for example? So you might be asking
yourself whether you can define multiply-derived classes that will contain only one
instance of an indirect base class.
C++ uses virtual base classes to do this. An object in a multiply-derived class contains
only one instance of the members in a virtual base class. The inheritance graph on the
opposite page uses the SUV class to illustrate this situation.

䊐 Declaration
A direct base class is declared virtual when a derived class is defined. You can use the
virtual keyword, which directly precedes the name of the base class.
In the definition scheme shown opposite, the Car class becomes the virtual base class
of PassCar and Van. However, the fact that the base class Car is virtual has no significance at this point.
A virtual base class takes effect in cases of multiple inheritance. The following definition

Example: class SUV
: public PassCar, public Van
{ . . . };

ensures that the SUV class only contains one instance of the virtual base class Car. An
object my of the SUV class gets sufficient memory for only one Car class sub-object.
More specifically, the statement

Example: cout<<"Producer: " << mySUV.getProd();
does not cause ambiguity.
The following items are important with respect to virtual base classes:
■
■

a virtual base class stays virtual even if further derivations are built. Each class
derived from PassCar also has the Car class as a virtual base class.
you cannot change the declaration of an indirect base class to virtual.

You must therefore decide what classes are to be declared virtual when you design the
class hierarchy. Later modifications will require modifications to the source code of any
derived classes.

594

■

CHAPTER 27

■

MULTIPLE INHERITANCE

CONSTRUCTOR CALLS

䊐 Building an inheritance graph
Class Definition
class multiDerived : public Base1, public Base2,
public Base3
{
// Here are additional data members and methods
};

Inheritance graph

Base1

Base2

MultiDerived

Base3

CONSTRUCTOR CALLS

■

595

䊐 Initialization
When an object is created in a simply-derived class, the sub-objects of the base classes
are created first on all levels of the class hierarchy. The sub-object whose class is nearer
to the top of the inheritance graph is created first.
The order of the constructor calls is “top down” and follows the inheritance graph.
The activation order used for the constructors in simple inheritance has been generalized
for multiple inheritance.

䊐 Inheritance Graph
Again, the inheritance graph, also called sub-object lattice, has an important job to do.
When a derived class is defined, the following rules apply:
■

In cases of multiple inheritance, base classes are entered into the inheritance
graph from left to right in the order in which they were stated when the class was
defined. The graph opposite illustrates this point.

If the class hierarchy does not contain any virtual base classes, the following applies to
the activation order of the constructors.
■
■

The base class constructors are executed first, top-down and from left to right on
each level.
Finally, the constructor belonging to the current class, which is at the bottom of
the inheritance graph, is executed.

If we look at the example on the opposite page, this means that the sub-objects of the
base classes Base1, Base2, and Base3 are created in this order. Then the constructor
of MultiDerived is executed.

䊐 Base Initializers
The constructor for the class at the bottom end of the inheritance graph uses base initializers to pass the values to the direct and indirect base classes. If the base initializer definition is missing in a constructor definition, the default constructor of the base class is
automatically executed.
Initial values are thus passed to the base class constructors “bottom up.”

596

■

CHAPTER 27

■

MULTIPLE INHERITANCE

INITIALIZING VIRTUAL BASE CLASSES

Class SUV
class SUV : public PassCar, public Van
{
private:
// . . .
public:
SUV( ... ) : Car( ... )
{
// Initialize additional data members
}
void display() const
{
PassCar::display();
Van::display();
// Output additional data members
}
};

INITIALIZING VIRTUAL BASE CLASSES

■

597

䊐 Constructor Calls in Virtual Base Classes
When an object is created for a multiply-derived class, the constructors of the base
classes are called first. However, if there is one virtual base class in the class hierarchy,
the virtual base class constructor is executed before a constructor of a non-virtual base
class is called.

✓

NOTE
The constructors of the virtual base classes are called first, followed by the constructors of non-virtual
base classes in the order defined in the inheritance graph.

The constructor of the virtual base class nearest the top of the inheritance graph is
executed first. This does not necessarily mean the top level of the class hierarchy, since a
virtual base class can be derived from a non-virtual base class.
In our example with the multiply-derived class SUV (Sport Utility Vehicle) the constructor for the virtual base class Car is called first, followed by the direct base classes
PassCar and Van, and last but not least, the constructor of the SUV class.

䊐 Base Initializers
You may be wondering what arguments are used to call the constructor of a virtual base
class. A base initializer of the directly-derived class or any other derivation could be
responsible. The following applies:

✓

NOTE
The constructor of a virtual base class is called with the arguments stated for the base initializer of the
last class to be derived, i.e. class at the bottom end of the inheritance graph.

The example opposite shows SUV containing a constructor with one base initializer.
Its arguments are passed to the constructor of the virtual base class Car.
For the purpose of initialization, it does not matter whether a class derived directly
from Car contains a base initializer or not. Base initializers for virtual indirect base
classes defined in the constructor of a direct base class are ignored. If the base classes
PassCar and Van also contained base initializers for the virtual base class Car, these
would be ignored too.
If the constructor for the last derived class does not contain a base initializer, the
default constructor is executed for each virtual base class. Whatever happens, a default
constructor must then exist in every virtual base class! Thus, base initializers that happen
to exist in base classes are also ignored.

■

CHAPTER 27

■

exercise s

598

MULTIPLE INHERITANCE

EXERCISES

The multiply-derived class MotorHome

Car

Home

MotorHome

Additional members of class MotorHome
Data member:
cat

Type
CATEGORY

Methods:
MotorHome(CATEGORY, long, const string&, int, double );
void
setCategory(CATEGORY )
CATEGORY getCategory() const ;
void display() const;

EXERCISES

■

599

Exercise 1
The multiply-derived class MotorHome needs to be fully implemented and tested.
■

■

■

■

Define an enumeration type CATEGORY that represents the categories
“Luxury,” “First Class,” “Middle Class,” and “Economy”.
Develop a class called Home with two data members used to store the
number of rooms and the size in square meters. Supply default values for
your constructor definition to create a default constructor. In addition to
access methods, also define the method display(), which outputs the
data members of an apartment.
Define a class derived from the Car and Home classes called MotorHome,
which is used to represent motorhomes. Inheritance of public base
classes is used.The MotorHome class contains a new data member used to
store one value of the CATEGORY type. In addition to defining a constructor with default values, also define appropriate access methods and a
display() method for output.
Place your definitions of the Home and MotorHome classes in a separate
header file, which includes the existing header file car.h.
Write a main function that first fully initializes a MotorHome type object
and then outputs the object.
Then create a second instance of the MotorHome type without initial values and display the object on screen. Call all the set methods in the
MotorHome class and its base classes to set your own values for the
objects.Then output the object once more.

600

■

CHAPTER 27

MULTIPLE INHERITANCE

Class hierarchy for the multiply-derived class SUV

Car
Data members:
car number
producer

PassCar

Van

Data members:
car type
sun roof (y/n)

Data members:
capacity (lbs)

SUV
Data members:
number of seats

EXERCISES

■

601

Exercise 2
Now fully define the SUV class for testing virtual base classes.
■

■

■

■

Change the definition of the PassCar class in the car.h header file to
make Car a virtual base class of the PassCar class.
Then define the Van class using the Car class as a virtual base class.The
new class should contain an additional data member used to represent
the payload of the van in kilograms.A maximum of 750 kg applies to vans.
The constructor should use default values to initialize the data members
with defaults, thus providing a default constructor for the class.A maximum value of 750 applies for the payload. In addition to the access
method display(), you still need to define methods for screen output.
Create the class SUV, which is derived from PassCar and Van, to represent a station wagon. Store the number of seats available in the station
wagon as a data member.
The constructor in the SUV class should use the base initializer to set all
the values of a station wagon using default values for every data member.
Additionally, define access methods and a display() to define output.
Use a main function to test the SUV class; the function should create station wagons with and without default values and display them on screen.

■

CHAPTER 27

■

solutions

602

MULTIPLE INHERITANCE

SOLUTIONS

Exercise 1
//
//
//
//
//
//
//
//
//
//
//

-----------------------------------------------------car.h : Definition of base class Car and
derived classes PassCar and Truck
-----------------------------------------------------car.cpp
Implementing methods of Car, PassCar, and Truck
-----------------------------------------------------These files have been left unchanged
from Chapters 23 and 25.

// -----------------------------------------------------// motorHome.h : Definition of the class Home and the
//
multiply-derived class MotorHome
// -----------------------------------------------------#ifndef _MOTORHOME_H_
#define _MOTORHOME_H_
#include "car.h"
#include 
#include 
using namespace std;
enum CATEGORY {LUXURY, FIRSTCLASS, SECONDCLASS, ECONOMY};
class Home
{
private:
int room;
double ft2;
public:
Home(int r = 0, double m2 = 0.0)
{ room = r; ft2 = m2;}
void setRoom(int n){ room = n;}
int getRoom() const { return room; }
void
setSquareFeet(double m2){ ft2 = m2;}
double getSquareFeet() const { return ft2; }
void display() const
{
cout << "Number of rooms:
" << room
<< "\nSquare feet: "
<< fixed << setprecision(2) << ft2 << endl;
}
};

SOLUTIONS

■

class MotorHome : public Car, public Home
{
private:
CATEGORY cat;
public:
MotorHome(long n=0L, const string& prod="", int ro=0,
double m2=0.0, CATEGORY k=ECONOMY)
: Car(n, prod), Home(ro, m2), cat(k)
{}
void
CATEGORY

setCategory(CATEGORY c){cat = c;}
getCategory() const { return cat;}

void display() const
{
cout << "\nMotorHome:
";
Car::display();
Home::display();
cout << "Category:
";
switch(cat)
{
case LUXURY:
cout
break;
case FIRSTCLASS:
cout
break;
case SECONDCLASS:
cout
break;
case ECONOMY:
cout
break;
}
cout << endl;
}

<< "

Luxury";

<< "

First class";

<< "

Second class";

<< "

Economy";

};
#endif
// -----------------------------------------------------// motorHome_t.cpp
// Testing the multiply-derived class MotorHome
// -----------------------------------------------------#include "motorHome.h"
int main()
{
MotorHome rv(12345L, "Texaco", 2, 40.5, LUXURY);
rv.display();
MotorHome holiday;
holiday.display();
cin.get();

// Default values

603

604

■

CHAPTER 27

MULTIPLE INHERITANCE

holiday.setNr(54321);
holiday.setProd("VW");
holiday.setRoom(1);
holiday.setSquareFeet(11.5);
holiday.setCategory(SECONDCLASS);
holiday.display();
return 0;
}

Exercise 2
// -----------------------------------------------------// car.h : Definition of base class Car and
//
the derived classes PassCar and Truck
// -----------------------------------------------------// car.cpp
// Implementing the methods of Car, PassCar, and Truck
// -----------------------------------------------------//
// These files are carried over from Chapter 23 and 25,
// with the following changes:
//
// Class Car is a virtual base class now
class PassCar : public virtual Car
{
// ...
};
class Truck : public virtual Car
{
// ...
};

// -----------------------------------------------------// suv.h : Defines the class Van and
//
the multiply-derived class SUV
// -----------------------------------------------------#ifndef _SUV_H
#define _SUV_H
#include "car.h"
class Van : public virtual Car
{
private:
double capacity;

SOLUTIONS

public:
Van(long n=0L, const string& prod="",
double l=0.0)
: Car(n,prod)
{
if(l > 750) l = 750;
capacity = l;
}
void setCapacity(double l)
{
if(l > 750)
capacity= 750;
else
capacity = l;
}
double getCapacity() const { return capacity; }
void display() const
{
cout << "Capacity:
"
<< capacity << " kg" << endl;
}
};
class SUV : public PassCar, public Van
{
private:
int cnt;
// Number of seats
public:
SUV(const string& tp="without type", bool sb=false,
long n=0L, const string& prod=" none ",
double l=0.0, int z = 1)
: PassCar(tp,sb), Car(n,prod),
Van(n,prod,l), cnt(z)
{ }
void display() const
{
PassCar::display();
Van::display();
cout << "Number of seats:
}
};
#endif

" << cnt << endl;

■

605

606

■

CHAPTER 27

MULTIPLE INHERITANCE

// -----------------------------------------------------// suv_t.cpp : Tests the class SUV
// -----------------------------------------------------#include "suv.h"
int main()
{
SUV mobil("Bravada", true, 120345, "Oldsmobile",350,6);
mobil.display();
SUV trucky;
trucky.display();
trucky.setNr(543221);
trucky.setProd("Renault");
trucky.setCapacity(1000.);
trucky.display();
return 0;
}

chapter

28

Exception Handling
This chapter describes how a C++ program uses error-handling
techniques to resolve error conditions. In addition to throwing and
catching exceptions, we also examine how exception specifications are
declared and exception classes are defined, additionally looking into the
use of standard exception classes.

607

608

■

CHAPTER 28

■

EXCEPTION HANDLING

TRADITIONAL ERROR HANDLING

Error checking after leaving a function

First calling
function

Second calling
function

Third calling
function

. . .
{
. . .
if(func()>0)
// every// thing
// is ok
else
exit(-1);
. . .

. . .
{
. . .
if(func()<=0)
// All errors
// are
// handled
. . .

. . .
{
. .
x= func();
if(x == 0)
//1. error
else if
(x == -1)
//2. error
...
}

}
}

int func( void)
{
. . .
if(dilemma)
return 0;
. . .
if(catastrophe)
return -1;
. . .
}
Called function

TRADITIONAL ERROR HANDLING

■

609

䊐 Error Conditions
Errors that occur at program runtime can seriously interrupt the normal flow of a program. Some common causes of errors are
■
■
■
■
■

division by 0, or values that are too large or small for a type
no memory available for dynamic allocation
errors on file access, for example, file not found
attempt to access an invalid address in main memory
invalid user input

Anomalies like these lead to incorrect results and may cause a computer to crash. Both of
these cases can have fatal effects on your application.
One of the programmer’s most important tasks is to predict and handle errors. You
can judge a program’s quality by the way it uses error-handling techniques to counteract
any potential error, although this is by no means easy to achieve.

䊐 Traditional Error Handling
Traditional structured programming languages use normal syntax to handle errors:
■
■

errors in function calls are indicated by special return values
global error variables or flags are set when errors occur, and then checked again
later.

If a function uses its return value to indicate errors, the return value must be examined
whenever the function is called, even if no error has occurred.

Example: if( func()> 0 )
// Return value positive => o.k.
else
// Treat errors

Error variables and flags must also be checked after every corresponding action.
In other words, you need to continually check for errors while a program is executing.
If you do happen to forget to check for errors, the consequences may be fatal.

610

■

CHAPTER 28

■

EXCEPTION HANDLING

EXCEPTION HANDLING

Using the throw statement
// calc_err.cpp: Defining the function calc(),
//
which throws exceptions.
// ---------------------------------------------------class Error
{
// Infos about the error cause
};

double calc( int a, int b )
{
if ( b < 0 )
throw (string)"Denominator is negative!";
if( b == 0 )
{
Error errorObj;
throw errorObj;
}
return ((double)a/b);
}

EXCEPTION HANDLING

■

611

䊐 Exception Handling Concept
C++ introduces a new approach to error handling. Exception handling is based on keeping
the normal functionality of the program separate from error handling. The basic idea is
that errors occurring in one particular part of the program are reported to another part of
the program, known as the calling environment. The calling environment performs central
error handling.
An application program no longer needs to continually check for errors, because in
the case of an error, control is automatically transferred to the calling environment.
When reporting an error, specific information on the error cause can be added. This
information is evaluated by the error-handling routines in the calling environment.

䊐 The throw Statement
An exception that occurs is recorded to the calling environment by means of a throw
statement; this is why we also say that an exception has been thrown.

Syntax:

throw fault;

The expression fault is an exception object that is thrown to the calling environment. It
can belong to any type except void.

Example: throw "Fire!";
In this example, the exception object is a string that is thrown to the calling environment.

䊐 Exception Classes
Normally, you define your own exception classes to categorize exceptions. In this case
you use the throw statement to throw an object belonging to a specific exception class.
An exception class need not contain data members or methods. However, the type,
which is used by the calling environment to identify the error, is important. Generally,
the exception class will contain members that provide more specific information on the
cause of the error.
In the sample program on the opposite page, the calc() function throws exceptions
in two cases, where the numerator is negative or has a value of 0. In the first case, the
exception thrown is a string. In the second case, the exception is an Error exception
class object. Instead of creating a local exception object errorObj, a temporary object
can be created:

Example: throw Error();

// It is shorter

612

■

CHAPTER 28

■

EXCEPTION HANDLING

EXCEPTION HANDLERS

Syntax of try and catch blocks
try
{
// Exceptions thrown by this block will be
// caught by the exception handlers,
// which are defined next.
}
catch( Type1 exc1)
{
// Type1 exceptions are handled here.
}
[ catch( Type2 exc2)
{
// Type2 exceptions are handled here.
}
. . .
//etc.
]
[ catch( ... )
{
// All other exceptions are handled here.
}]

✓

NOTE
The brackets [...] in a syntax description indicate that the enclosed section is optional.

EXCEPTION HANDLERS

■

613

䊐 How Exception Handling Works
The part of a program that performs central error handling in the calling environment is
referred to as an exception handler. An exception handler catches the exception object
thrown to it and performs error handling. The exception object type determines which
handler will catch it and consequently be executed.
This means that you need to specify two things when implementing exception handling:
■
■

the part of the program that can throw exceptions
the exception handlers that will process the various exception types.

C++ provides language elements for this task, the keywords try and catch. Each keyword precedes a code block and thus they are often referred to as try and catch blocks.
Syntactically speaking, each try and catch block is a statement, however.

䊐 try and catch Blocks
A try block contains the program code in which errors can occur and exceptions can be
thrown. Normally, a try block will consist of a group of functions that can produce similar errors.
Each catch block defines an exception handler, where the exception declaration, which is
enclosed in parentheses, defines the type of exceptions the handler can catch. The
catch blocks immediately follow the try block. A minimum of one catch block is
required.
The exception handlers defined by the catch blocks catch the exceptions thrown
within the try block. If there is no handler defined for a particular exception type, the
program will not simply enter an undefined state but will be orderly terminated by a call
to the standard function terminate().
It is common practice to define specific handlers for certain types of errors and one
generic handler for all other errors. This functionality is provided by a special syntax in
the catch statement with an exception declaration consisting of just three dots.

Syntax:

catch( ... )
{ // General handler for
// all other exceptions
}

Since the application program decides what reaction is applicable for certain error conditions, the try and catch blocks are formulated in the application.

614

■

CHAPTER 28

■

EXCEPTION HANDLING

THROWING AND CATCHING EXCEPTIONS

Demonstration program
// calc_err.cpp: Tests the function calc(),
//
which throws exceptions.
// ---------------------------------------------------#include 
#include 
using namespace std;
double calc( int a, int b );
int main()
{
int x, y;
double res;
bool flag = false;
do
{
try
// try block
{
cout << "Enter two positive integers: ";
cin >> x >> y;
res = calc( x, y);
cout << x << "/" << y << " = " << res << endl;
flag = true;
// Then to leave the loop.
}
catch( string& s)
// 1st catch block
{
cerr << s << endl;
}
catch( Error& )
// 2nd catch block
{
cerr << "Division by 0! " << endl;
}
catch(...)
// 3rd catch block
{
cerr << "Unexpected exception! \n";
exit(1);
}
}while( !flag);
// continued ...
return 0;
}

✓

NOTE
As the Error class contains no data members, the corresponding catch block declares only the type
of exception, and no parameters. This avoids a compiler warning since the parameter is not used.

THROWING AND CATCHING EXCEPTIONS

■

615

䊐 Backing Out of an Error Condition
When the throw statement is executed, an exception object is thrown. That is, a temporary object of the same type and content as the throw expression is created.

Example: throw "Cyclone!";
This creates a string as an exception object and copies the string "Cyclone!" to it.
Thus, if the throw expression is a class type, the copy constructor is executed.
The exception object is then thrown and the program control leaves the try block.
Any changes to the stack that took place after entering the try block are unwound.
This specifically involves destroying any local, non-static objects. Unwinding the stack
allow you to back out of the normal program flow in an orderly manner.

䊐 Searching for Handlers
After leaving the try block, the program control is transferred to an matching handler
in the catch blocks that follow. This search operation is always performed sequentially
beginning with the first catch block and the exception declaration of the handler
determines whether the handler should be executed.
A handler is called when the type in the exception declaration is
■
■
■

identical to the exception type thrown or
a base class of the exception type or
a base class pointer and the exception is a pointer to a derived class.

This is why the general exception handler catch( ... ) always has to be defined last.
Since the first suitable handler will be executed, and any exception thrown will be
caught by the general handler, a handler defined after the general handler would never
be called.

䊐 Continuing the Program
After executing a handler, the program continues with the first statement following the
catch blocks, unless the handler throws another exception or terminates the program.
After completing exception handling, the exception object that was thrown is destroyed.
The first two catch blocks handle both exceptions that the calc() function can
throw. In both cases a message is displayed and the program carries on prompting for
input and computing values. If an unexpected exception occurs, a message is again displayed, but in this case the program then terminates.

616

■

CHAPTER 28

■

EXCEPTION HANDLING

NESTING EXCEPTION HANDLING

Nested try and catch blocks
try
{
// Type1 exceptions are thrown here.
try
{
// Type1 and Type2 exceptions are thrown here.
}
catch( Type2 e2)
{
// Type2 exceptions are pre-handled here
throw;
// and thrown again.
}
// Other Type1 exceptions
// can be thrown.
}
catch( Type1 e1)
{
// Type1 exceptions are handled here.
}
catch(...)
{
// All remaining exceptions are handled here,
// particularly Type2 exceptions.
}

✓

NOTE
This scenario assumes that the error classes Type1 and Type2 are not derived one from another. If
class Type2 is derived from class Type1, any Type2 exceptions thrown will be caught by the handler
for the base class Type1.

NESTING EXCEPTION HANDLING

■

617

䊐 Nested try and catch Blocks
A program will normally contain multiple try blocks with appropriate exception handlers. This allows for various error handling in different parts of the program.
However, a try block can contain additional try blocks. This allows you to use the
handlers in a nested try block for special purpose error handling, leaving the handlers in
the surrounding try block to deal with remaining errors. Handlers in a nested try block
can also pre-handle specific errors and then pass control to the try block wrapper for
final handling.

䊐 Re-throwing Exceptions
In the last of these cases an exception thrown by the nested try block has to be passed
to the try block wrapper. This is achieved using a throw statement that does not
expect an exception object.

Example: throw;

// in a catch block

This re-throws the pre-handled exception, which can then be processed by the handler
in the surrounding try block. The statement is only valid within a nested catch block
for this reason.

䊐 Exception Specifications
The exceptions that a function can throw are features of that function. The application
programmer must have knowledge of both the function prototype and the exceptions the
function can throw to ensure that he or she will be capable of programming correct function calls and taking appropriate action in case of errors.
The exceptions a function can throw can be stated in a so-called exception specification list when you declare a function.

Example: int func(int) throw(BadIndex, OutOfRange);
The list BadIndex, OutOfRange states the exceptions that the function func()can
throw. If the list is empty, that is, if the list contains only the throw() statement, no
exceptions are thrown. If the throw statement is also missing, there is no specific information about possible exceptions and any exception can be thrown.

618

■

CHAPTER 28

■

EXCEPTION HANDLING

DEFINING YOUR OWN ERROR CLASSES

Exception handling for numeric operations
// calc_new.cpp: New version of function calc(),
//
which throws exceptions of type
//
MathError.
// ---------------------------------------------------#include 
#include 
using namespace std;
class MathError
{
private:
string message;
public:
MathError( const string& s) : message(s) {}
const string& getMessage() const {return message;}
};
double calc( int a, int b ) throw(MathError);
int main()
{
int x, y; bool flag = false;
do
{
try
// try block
{
cout << "Enter two positive integers: ";
cin >> x >> y;
cout << x <<"/"<< y <<" = "<< calc( x, y) << '\n';
flag = true;
// To leave the loop.
}
catch( MathError& err)
// catch block
{
cerr << err.getMessage() << endl;
}
}while( !flag);
// continued ...
return 0;
}
double calc( int a, int b ) throw (MathError)
{ if ( b < 0 )
throw MathError("Denominator is negative!");
if( b == 0 )
throw MathError("Division by 0!");
return ((double)a/b);
}

DEFINING YOUR OWN ERROR CLASSES

■

619

䊐 Exception Class Members
When an exception is thrown, the exception object type determines which exception
handler will be executed. For this reason, an exception class does not need to have any
members.
However, an exception class can contain data members and methods—just like any
other class. This makes sense, as locally defined non-static objects are destroyed when an
exception has been thrown and the stack is unwound. Thus, the exception handler can
no longer access the previously existing objects.
You can use the data members of error classes to rescue data threatened by an error
condition. For example, you can store data important for exception handling in an
exception object.

䊐 The Exception Class MathError
The exception class MathError is defined opposite. The calc() function throws an
exception when a number input by a user is negative or 0. When an exception is thrown,

Example: throw MathError("Division by 0!");
the error message is stored in the exception object. The exception handler can then use
the getMessage() method to evaluate the message.

䊐 Exception Hierarchies
New exception classes can be derived from existing exception classes. A base class will
normally represent a general error type and specific errors will be represented by derived
classes.
Thus, the exception class MathError could be defined to represent general errors in
mathematical computations, but it would make sense to define derived exception classes
for special cases, such as “Division by 0” or “Arithmetic overflow.” You could call these
classes DivisionByZero and OverflowError, for example.
Be aware of the following rules for exception handlers in this context:
■
■

given that T is a derived exception class, special errors of this type are handled by
the exception handler
if T is a base class, the handler will also catch the exception objects thrown by
derived classes, thus providing similar handling for generic and specific errors.

620

■

CHAPTER 28

■

EXCEPTION HANDLING

STANDARD EXCEPTION CLASSES

Exception classes derived from logic_error

invalid_argument

Invalid argument

out_of_range

Argument value not in its expected range

length_error

Length exceeds maximum capacity

domain_error

Domain error reported by the implementation

Exception classes derived from runtime_error
range_error

Range error in internal computation

underflow_error

Arithmetic underflow error

overflow_error

Arithmetic overflow error

Using standard exception classes
// Subscript operator of class FloatArr throws
// exceptions with type of standard class out_of_range:
// --------------------------------------------------#include 
#include 
using namespace std;
double& FloatArr::operator[](int i) throw(out_of_range)
{ if( i < 0 || i >= anz )
throw out_of_range("Invalid index!");
else return arrPtr[i];
}
// -------------- Test Program -----------------int main()
{
try
{
// Uses arrays of type FloatArr
}
catch(out_of_range& err)
{
cerr << err.what() << endl;
}
// The program continues here.
}

STANDARD EXCEPTION CLASSES

■

621

䊐 Hierarchy of Standard Exception Classes
The C++ standard library contains various exception classes, which are used in the string
and container libraries, for example. However, the standard exception classes can be
used just like exception classes of your own. Their definitions are to be found in the
header file stdexcept.
The standard exception classes are organized in a hierarchy, the common base class
being the exception class. In addition to a default constructor, a copy constructor, and
an assignment, this class contains a virtual public method, what(), which returns a
message on the error cause as a C string.

䊐 Representing Logic Errors and Runtime Errors
The following exception classes are derived from the exception class:
logic_error

used to represent logic errors, caused by anomalies in the program’s
logic. These errors can be avoided.

runtime_error

used to represent runtime errors, such as under- or overflows occurring in internal computations. These errors are unpredictable.

The opposite page contains on overview of the exception classes derived from the
logic_error and runtime_error classes. For example, the method at() in the
string class throws an out_of_range type exception when an invalid string position
is passed to it. If a string cannot be displayed because of its exceptional length, an exception of the invalid_argument type is thrown.
An exception of the overflow_error or underflow_error type is thrown if the
value to be displayed is too large or too small for the type in use. The range_error
class shows range errors, which can occur during internal computations.
A constructor with a string as a parameter is defined in every class derived from
exception. This means you can initialize exceptions of these types with error messages.
The what() method returns this error message as a C string.

■

CHAPTER 28

■

exercise s

622

EXCEPTION HANDLING

EXERCISES

Exercise 1: Error messages of the exception handler

The first exception handler’s message:

Error in reading:
Invalid index: ...

The second exception handler’s message:

Error in writing:
Invalid index: ...

EXERCISES

■

623

Exercise 1
The FloatArr class needs exception handling for cases where an invalid index is
stated when accessing an array member.
■

■

■

Define the exception class BadIndex for this purpose and store the class
in the header file “floatArr.h”.The exception class must contain a data
member to store the invalid index.The constructor expects an index
that it will copy to the data member.The const access method
getBadIndex() returns the data member.
Both subscript operators should be able to throw BadIndex type exceptions.Add an exception specification to the declaration of the subscript
operators.
The methods that expect the position as an argument, such as insert()
and remove(), should also throw exceptions.Add appropriate exception
specifications to the definitions and change the return types from bool to
void.
Change the definitions of the methods and operator functions to allow a
BadIndex type exception to be thrown when the index passed to the
function is outside the valid range.
Write a main function where a constant vector is created and initialized
with fixed values. Exception handling is required for the following scenarios.The array elements are displayed and an index is read until an invalid
index value is input.The catch handler should output the information
shown opposite for each invalid index.
Then create a non-constant array.Add further exception handling to be
performed. Elements are appended or inserted within a try block.
Include an invalid element access attempt, which causes the catch handler to output the information shown opposite.Then finally output the
array elements outside the try and catch blocks.

624

■

CHAPTER 28

EXCEPTION HANDLING

Exercises
For Exercise 2: Error messages of the exception handlers

Messages of the exception handlers
for an exception object of type DivisionByZero:

Error in initializing:
The denominator is 0!

Error in division:
No division by zero!

Error: Denominator is 0!
New denominator != 0: ...

EXERCISES

■

625

Exercise 2
Implement exception handling for the Fraction class, which is used to
represent fractions (see Exercise 2, Chapter 19). Dividing by 0 throws an
exception that affects the constructor for the Fraction class and the operator
functions / and >>.
■

Define the exception class DivError, which has no data members, within
the Fraction class.The exception class is of the following type
Fraction::DivError

■

■
■

Add an appropriate exception specification to the declarations of the
constructor and the operator functions / and >>.
Change the definition of the constructor in the Fraction class. If the
value of the denominator is 0, a DivisionByZero type exception should
be thrown.
Similarly modify the operator functions.
Now write a main function to test the various exceptions.You will need
to arrange three different try and catch blocks sequentially.
The first try/catch block tests the constructor. Create several fractions,
including some with a numerator value of 0 and one with 0 as its
denominator.The exception handler should issue the error message
shown opposite.
The second try/catch block tests divisions. Use a statement to attempt
to divide by 0.The corresponding exception handler should send the
second error message shown opposite to your standard output.
The third try/catch block reads numerators and denominators of
fractions in dialog. If the value of the denominator is 0, the denominator is
read again. If the value is still 0, the third error message as shown
opposite is output and the program terminates.

■

CHAPTER 28

■

solutions

626

EXCEPTION HANDLING

SOLUTIONS

Exercise 1
// -----------------------------------------------------// floatArr.h : Represents dynamic float arrays.
// Methods throw exceptions for an invalid index.
// -----------------------------------------------------#ifndef _FLOATARR_
#define _FLOATARR_
#include 
using namespace std;
class BadIndex
{
private:
int index;
public:
BadIndex(int i){index = i;}
int getBadIndex() const {return index;}
};
class FloatArr
{
private:
float* arrPtr;
int max;
int cnt;

//
//
//
//

Dynamic member
Maximum number without
reallocating more memory.
Current number of elements.

void expand( int newSize);

// Function to help
// enlarge the array.

public:
FloatArr( int n = 256 );
// Constructors
FloatArr( int n, float val);
FloatArr(const FloatArr& src);
~FloatArr();
// Destructor
FloatArr& operator=( const FloatArr&); // Assignment
int

length() const { return cnt; }

// Subscript operators:
float& operator[](int i) throw(BadIndex);
float operator[](int i) const throw(BadIndex);
// Methods to append a float value
// or an array of floats:
void append( float val);
void append( const FloatArr& v);

SOLUTIONS

■

FloatArr& operator+=( float val)
{
append( val);
return *this;
}
FloatArr& operator+=( const FloatArr& v)
{
append(v);
return *this;
}
// Methods to insert a float value
// or an array of float values:
void insert( float val, int pos) throw(BadIndex);
void insert(const FloatArr& v,int pos) throw(BadIndex);
void remove(int pos) throw(BadIndex); // Remove
// at pos.
// Output the array
friend ostream& operator<<( ostream& os,
const FloatArr& v)
{
int w = os.width();
// Save field width.
for( float *p = v.arrPtr; p < v.arrPtr + v.cnt; ++p)
{
os.width(w); os << *p;
}
return os;
}
};
#endif

// _FLOATARR_

// ----------------------------------------------------// floatArr.cpp
// Implementing the methods of FloatArr.
// ----------------------------------------------------#include "floatArr.h"
// --- Constructors --FloatArr::FloatArr( int n )
{
max = n;
cnt = 0;
arrPtr = new float[max];
}

// Sets max and cnt.
// Allocates memory

FloatArr::FloatArr(int n, float val)
{
max = cnt = n;
arrPtr = new float[max];
for( int i=0; i < cnt; ++i)
arrPtr[i] = val;
}

627

628

■

CHAPTER 28

EXCEPTION HANDLING

FloatArr::FloatArr(const FloatArr& src)
{
max = src.max;
cnt = src.cnt;
arrPtr = new float[max];
for( int i = 0; i < cnt; i++ )
arrPtr[i] = src.arrPtr[i];
}
// --- Destructor --FloatArr::~FloatArr()
{
delete[] arrPtr;
}
// Private functions to help enlarge the array.
void FloatArr::expand( int newSize)
{
if( newSize == max)
return;
max = newSize;
if( newSize < cnt)
cnt = newSize;
float *temp = new float[newSize];
for( int i = 0; i < cnt; ++i)
temp[i] = arrPtr[i];
delete[] arrPtr;
arrPtr = temp;
}
FloatArr& FloatArr::operator=( const FloatArr& src )
{
if( this != &src )
// No self assignment!
{
max = src.max;
cnt = src.cnt;
delete[] arrPtr;

// Release memory,

arrPtr = new float[max];

// reallocate,

for( int i=0; i < cnt; i++) // copy elements.
arrPtr[i] = src.arrPtr[i];
}
return *this;
}

SOLUTIONS

■

float& FloatArr::operator[]( int i ) throw(BadIndex)
{
if( i < 0 || i >= cnt ) throw BadIndex(i);
return arrPtr[i];
}
float FloatArr::operator[]( int i ) const throw(BadIndex)
{
if( i < 0 || i >= cnt ) throw BadIndex(i);
return arrPtr[i];
}
// Append a float value or an array of floats.
void FloatArr::append( float val)
{
if( cnt+1 > max)
expand( cnt+1);
arrPtr[cnt++] = val;
}
void FloatArr::append( const FloatArr& v)
{
if( cnt + v.cnt > max)
expand( cnt + v.cnt);
int count = v.cnt;
// Necessary if
// v == *this.
for( int i=0; i < count; ++i)
arrPtr[cnt++] = v.arrPtr[i];
}
// Inserts a float value or an array of floats.
void FloatArr::insert(float val, int pos) throw(BadIndex)
{
insert( FloatArr(1, val), pos);
}
void FloatArr::insert( const FloatArr& v, int pos )
throw( BadIndex )
{
if( pos < 0 || pos > cnt)
// Append is also possible.
throw BadIndex(pos);
if( max < cnt + v.cnt)
expand(cnt + v.cnt);
int i;
for( i = cnt-1; i >= pos; --i)
// Shift up from
arrPtr[i+v.cnt] = arrPtr[i];
// position pos.
for( i = 0; i < v.cnt; ++i)
// Fill the gap.
arrPtr[i+pos] = v.arrPtr[i];
cnt = cnt + v.cnt;
}

629

630

■

CHAPTER 28

EXCEPTION HANDLING

// To delete
void FloatArr::remove(int pos) throw(BadIndex)
{
if( pos >= 0 && pos < cnt)
{
for( int i = pos; i < cnt-1; ++i)
arrPtr[i] = arrPtr[i+1];
--cnt;
}
else
throw BadIndex(pos);
}
// ------------------------------------------------------// arr_h.cpp
// Tests exception handling for float arrays.
// ------------------------------------------------------#include 
#include 
using namespace std;
#include "floatArr.h"
int main()
{
const FloatArr v(10, 9.9f);
bool ok = false;
while( !ok)
{
try
{
cout << "Here is the constant array v: \n";
cout << setw(8) << v <> i;
cout << "\nThe value read: " <<
ok = true;

v[i] << endl;

}
catch(BadIndex& err)
{
cerr << "Error in reading.\n"
<< "\nInvalid index: "
<< err.getBadIndex() << endl;
}
}

SOLUTIONS

//
//
//

■

FloatArr w(20);
// Array w
try
{
w.insert(1.1F, 0);
// To write.
w.insert(2.2F, 1);
w.insert(3.3F, 3);
// Error!
w[10] = 5.0;
// Error!
w.remove(7);
// Error!
}
catch(BadIndex& err)
{
cerr << "\nError in writing! "
<< "\nInvalid index: "
<< err.getBadIndex() << endl;
}
cout << "\nHere is the array: \n";
cout << setw(5) << w << endl;
return 0;

}

Exercise 2
// -----------------------------------------------------// fraction.h
// A numeric class to represent fractions,
// exception handling is included.
// -----------------------------------------------------#ifndef _FRACTION_
#define _FRACTION_
#include 
#include 
using namespace std;
class Fraction
{
private:
long numerator, denominator;
public:
class DivisionByZero
{
// No data members
};
Fraction(long z = 0, long n = 1) throw(DivisionByZero);
Fraction operator-() const
{
return Fraction(-numerator, denominator);
}

631

632

■

CHAPTER 28

EXCEPTION HANDLING

Fraction& operator+=(const Fraction& a)
{
numerator = a.numerator * denominator
+ numerator * a.denominator;
denominator *= a.denominator;
return *this;
}
Fraction& operator-=(const Fraction& a)
{
*this += (-a);
return *this;
}
Fraction& operator++()
{
numerator += denominator;
return *this;
}
Fraction& operator--()
{
numerator -= denominator;
return *this;
}
friend Fraction operator+(const Fraction&,
const Fraction&);
friend Fraction operator-(const Fraction&,
const Fraction&);
friend Fraction operator*(const Fraction&,
const Fraction&);
friend Fraction operator/(const Fraction&,
const Fraction&)
throw(Fraction::DivisionByZero);
friend ostream& operator<<(ostream&, const Fraction&);
friend istream& operator>>(istream& is, Fraction& a)
throw(Fraction::DivisionByZero);
};
#endif

SOLUTIONS

■

// ------------------------------------------------------// fraction.cpp
// Defines methods and friend functions.
// ------------------------------------------------------#include 
#include 
using namespace std;
#include "fraction.h"
// Constructor:
Fraction::Fraction(long z, long n)
throw(Fraction::DivisionByZero)
{
if(n == 0) throw DivisionByZero();
if( n < 0) z = -z, n = -n;
numerator = z;
denominator = n;
}
Fraction operator+(const Fraction& a, const Fraction& b)
{
Fraction temp;
temp.denominator = a.denominator * b.denominator;
temp.numerator = a.numerator*b.denominator
+ b.numerator * a.denominator;
return temp;
}
Fraction operator-(const Fraction& a, const Fraction& b )
{
Fraction temp = a;
temp += (-b);
return temp;
}
Fraction operator*(const Fraction& a, const Fraction& b )
{
Fraction temp;
temp.numerator = a.numerator * b.numerator;
temp.denominator = a.denominator * b.denominator;
return temp;
}
Fraction operator/(const Fraction& a, const Fraction& b )
throw(Fraction::DivisionByZero)
{
if( b.numerator == 0) throw Fraction::DivisionByZero();
// Multiply a with the inverse of b:
Fraction temp;
temp.numerator = a.numerator * b.denominator;
temp.denominator = a.denominator * b.numerator;
if( temp.denominator < 0 )
temp.numerator = -temp.numerator,
temp.denominator = -temp.denominator;
return temp;
}

633

634

■

CHAPTER 28

EXCEPTION HANDLING

ostream& operator<<(ostream& os, const Fraction& a)
{
os << a.numerator << "/" << a.denominator;
return os;
}
istream& operator>>(istream& is, Fraction& a)
throw(Fraction::DivisionByZero)
{
cout << "Enter a fraction:\n"
" Numerator:
";
is >> a.numerator;
cout << " Denominator != 0: ";
is >> a.denominator;
if( !is) return is;
if( a.denominator == 0)
{
cout << "\nError: The denominator is 0\n"
<< "New Denominator != 0: "; is >> a.denominator;
}
if( a.denominator == 0)
throw Fraction::DivisionByZero();
if( a.denominator < 0 )
a.numerator = -a.numerator,
a.denominator = -a.denominator;
return is;
}

// ------------------------------------------------------// fract_t.cpp : Testing the class Fraction.
// Modules: fract_t.cpp fraction.cpp
// ------------------------------------------------------#include 
#include "fraction.h"
int main()
{
try
// Tests the exception of the constructor:
{
Fraction c(5,0);
}
catch(Fraction::DivisionByZero& )
{
cout << "\nError on initializing: "
<< "\nThe denominator is 0\n";
}

SOLUTIONS

Fraction a(1,3), b(3);
try
{
cout << "\nSome test output:\n\n";
cout << " a = " << a << endl;
cout << " b = " << b << endl;
cout <<
cout <<
cout <<
b = 0;
cout <<

" a + b = " << (a + b) << endl;
" a - b = " << (a - b) << endl;
" a * b = " << (a * b) << endl;

" a / b = " << (a / b) << endl;
}
catch(Fraction::DivisionByZero& )
{
cout << "\nError on dividing: "
<< "\nNo division by zero! 0\n";
}

cout << "
cout << "

--a = " <<
++a = " <<

// Error!

--a << endl;
++a << endl;

a += Fraction(1,2);
cout << " a+= 1/2; a = " << a << endl;
a -= Fraction(1,2);
cout << " a-= 1/2; a = " << a << endl;
cout << "-b = " << -b << endl;
cout << "\nAnd now your input: \n";
try
{
cin >> a;
}
catch(Fraction::DivisionByZero&)
{
cerr << "\nError: The denominator is 0\n";
exit(1);
}
cout << "\nYour input: " << a
return 0;
}

<< endl;

■

635

This page intentionally left blank

chapter

29

More about Files
This chapter describes
■

random access to files based on file streams

■

options for querying file state

■

exception handling for files.

We will also illustrate how to make objects in polymorphic classes
persistent, that is, how to save them in files.The applications introduced in
this chapter include simple index files and hash tables.

637

638

■

CHAPTER 29

■

MORE ABOUT FILES

OPENING A FILE FOR RANDOM ACCESS

Combined open modes for read and write access

Open mode

✓

Effects

Must the file
exist?

ios::in
| ios::out

To open the file for input and
output.

yes

ios::in
| ios::out
| ios::trunc

Opens the file for input and
output. If the file already exists,
it will be truncated.

no

ios::in
| ios::out
| ios::app

Opens the file for input and
output. If the file does not exist,
it will be created.
Before each writing access, a
seek to end is performed.

no

NOTE
1. If the flag ios::binary is additionally set, the file will be opened in binary mode.
2. If the flag ios::ate (“at end”) is additionally set, the current seek position is set to
end-of-file immediately after opening.

OPENING A FILE FOR RANDOM ACCESS

■

639

䊐 Random File Access
So far we have only looked at sequential file access. If you need access to specific information in such a file, you have to walk through the file from top to tail, and new records
are always appended at the end of the file.
Random file access gives you the option of reading and writing information directly at a
pre-defined position. To be able to do this, you need to change the current file position
explicitly, that is, you need to point the get/put pointer to the next byte to be manipulated. After pointing the pointer, you can revert to using the read and write operations
that you are already familiar with.

䊐 Open Modes
One prerequisite of random file access is that the position of the records in the file can be
precisely identified. This implies opening the file in binary mode to avoid having to
transfer additional escape characters to the file.

Example: ios::openmode mode = ios::in | ios::out |
ios::app | ios::binary;
fstream fstr("account.fle", mode);

This statement opens the file "Account.fle" in binary mode for reading and appending at end-of-file. The file will be created if it did not previously exist. Random read access
to the file is possible, but for write operations new records will be appended at the end of
the file.
To enable random read and write access to a file, the file can be opened as follows:

Example: ios::openmode mode = ios::in | ios::out |
ios::binary;
fstream fstr("account.fle", mode);

However, this technique can only be used for existing files. If the file does not exist, you
can use the ios::trunc flag to create it.
The section “File State” discusses your error handling options if a file, such as
"account.fle" cannot be found.

640

■

CHAPTER 29

■

MORE ABOUT FILES

POSITIONING FOR RANDOM ACCESS

The three access points in a file

Access point

Beginning of the file

Positioning flag

File

ios::beg
•
•
•

Current position

ios::cur
•
•
•
•

End of file

ios::end

POSITIONING FOR RANDOM ACCESS

■

641

䊐 Discovering and Changing the Current Position
The file stream classes comprise methods to discover and change the current position in
a file. The tellp() and tellg() methods return the current position of the put or
get pointers as a long value.

Example: long rpos = myfile.tellg();
This statement queries the current position of the read pointer in the myfile stream.
The current position is always returned as a byte offset relative to the beginning of the
file.
The current file position can be modified using the seekp() or seekg() method.
The position is stated as a byte offset, relative to either the beginning or end of the file,
or relative to the current position in the file.

䊐 Positioning Flags
Three ios::seekdir type positioning flags are defined in the ios class for this purpose; these are ios::beg, ios::cur, and ios::end.
Imagine you want to write the object acc to the file "account.fle" at offset pos.
You can use the following statements to do so:

Example: ofstream fstr("account.fle", ios::out |
ios::binary);
fstr.seekp(pos, ios::begin);
acc.write( fstr );

This calls the write() method in the Account class, which allows an object to write
its own data members to a file (see Chapter 18).
If you do not specify a positioning flag, the position will be assumed to be relative to
the beginning of the file. The statement used to position the write pointer in the last
example can therefore be formulated as follows:

Example: fstr.seekp(pos );
The byte offset can also be negative for calls to the methods seekp() and seekg().
However, you cannot position the read/write pointer before the beginning of the file.
In contrast, it is possible to place the pointer at a position after the end of the file and
then perform a write operation, which will create a gap with unknown content in the
file. This only makes sense if all the empty slots in the file are of an equal length, as they
can be overwritten later. This option is often used when programming hash tables.

642

■

CHAPTER 29

■

MORE ABOUT FILES

POSITIONING FOR RANDOM ACCESS (CONTINUED)

Representing an index entry
// index.h: Defines the class IndexEntry
// ---------------------------------------------------#ifndef _INDEX_H
#define _INDEX_H
#include 
#include 
using namespace std;
class IndexEntry
{
private:
long key;
// Key
long recNr;
// Offset
public:
IndexEntry(long k=0L, long n=0L) { key=k; recNr=n;}
// Access methods ... and:
int recordSize() const
{ return sizeof(key) + sizeof(recNr); }
fstream& write( fstream& ind) const;
fstream& read( fstream& ind);
fstream& write_at(fstream& ind, long pos) const;
fstream& read_at( fstream& ind, long pos);
};
#endif

The read_at() and write_at() methods
// index.cpp: Implements the methods
// --------------------------------------------------#include "index.h"
// . . .
fstream& IndexEntry::write_at( fstream& ind, long pos)
const
{
ind.seekp(pos);
ind.write((char*)&key, sizeof(key) );
ind.write((char*)&recNr, sizeof(recNr) );
return ind;
}
fstream& IndexEntry::read_at(fstream& ind, long pos)
{
ind.seekg(pos);
ind.read((char*)&key, sizeof(key) );
ind.read((char*)&recNr, sizeof(recNr));
return ind;
}

POSITIONING FOR RANDOM ACCESS (CONTINUED)

■

643

䊐 Using Positioning Methods
The following statements are commonly used for random positioning
seekg( 0L); and

seekp( 0L, ios::end );

They set the current position to the beginning or end of a file. You should be aware that
the first argument is 0L to indicate that long type is required.
If you need to determine the length of a file, you can point the get pointer to the end
of the file and then query the position of the pointer:

Example: fstr.seekg(0L, ios::end);
unsigned long count = fstr.tellg();

The count variable will then contain the number of bytes occupied by the file.
These positioning methods are useful for files opened in binary mode. However, it
does not make much sense to use them for text files or particularly for devices. In text
mode, conversions of LF <=> CR/LF prevent the methods from working correctly.

䊐 Determining Positions in a File
The position of records in a files is easy to compute if all the records in the file are the
same length. Given that size is the length of a record
0L,

size,

2*size, ...

are the positions of the first, second, third records, and so on.
If you are working with variable length records, you cannot exactly compute their
positions. To enable random access you therefore need to store the positions of the
records in a separate structure, a so-called index.
The index stores pairs of keys and record positions, so-called index entries in a file. A
key, a social security number, or customer id, for example, must uniquely identify a
record. If the index is sorted, the position that correlates to the required key can be
quickly found using the binary search algorithm.

䊐 The IndexEntry Class
The IndexEntry class, used to represent an index entry, is shown opposite. The class
comprises methods for reading and writing an index entry at the current file position or
at any given position. The appropriate file stream is passed as an argument to the
method.

644

■

CHAPTER 29

MORE ABOUT FILES

■

FILE STATE

Representing an index
// index.h: (continued)
// Adds the definition of an index
// -------------------------------------------------#include 
class IndexFile
{
private:
fstream index;
string name;

// Filename of index

public:
IndexFile(const string& s);
~IndexFile() { index.close(); }
void insert( long key, long pos);
long search( long key);
void retrieve(IndexEntry& entry, long pos );
};

Constructor of class IndexFile
// index.cpp: (continued)
// Adds methods of class IndexFile
// --------------------------------------------------IndexFile::IndexFile(const string& file)
{
ios::openmode mode = ios::in | ios::out
| ios::binary;
index.open(file.c_str(), mode);
if(!index)
// If the file does not exist
{
index.clear();
mode |= ios::trunc;
index.open(file.c_str(), mode);
if(!index)
return;
}
else
name = file;
}
//. . .

FILE STATE

■

645

䊐 State Flags
A file stream can assume various states, for example, when it reaches the end of a file and
cannot continue reading. A file operation can also fail if a file cannot be opened, or if a
block is not transferred correctly.
The ios class uses state flags to define the various states a file can assume. Each state
flag corresponds to a single bit in a status-word, which is represented by the iostate
type in the ios class. The following state flags exist:
■
■
■
■

ios::eofbit
ios::failbit
ios::badbit
ios::goodbit

end of file reached
last read or write operation failed
an irrecoverable error occurred
the stream is ok, e.g. no other state flag is set.

The “flag” ios::goodbit is an exception to the rule since it is not represented by a
single bit, but by the value 0 if no other flag has been set. In other words a status-word
has the value ios::goodbit if everything is fine!

䊐 Discovering and Changing the State
There are multiple methods for discovering and modifying the status of a stream. A
method exists for each state flag; these are eof(), fail(), bad(), and good(). They
return true when the corresponding flag has been raised. This means you can discover
the end of a file with the following statement:

Example: if( fstr.eof() ) ...
The status-word of a stream can be read using the rdstate() method. Individual flags
can then be queried by a simple comparison:

Example: if( myfile.rdstate() == ios::badbit ). . .
The clear() method is available for clearing the status-word. If you call clear()
without any arguments, all the state flags are cleared. An argument of the iostate type
passed to clear() automatically becomes the new status-word for the stream.

䊐 The IndexFile Class
The IndexFile class, which uses a file to represent an index, is defined opposite. The
constructor for this class uses the clear() method to reset the fail bit after an invalid
attempt to open a non-existent file. A new file can then be created.
The IndexFile class comprises methods for inserting, seeking, and retrieving index
entries, which we will be implementing later in this chapter.

646

■

CHAPTER 29

■

MORE ABOUT FILES

EXCEPTION HANDLING FOR FILES

Defining your own exception classes
// exceptio.h : Exception classes for file access
// ------------------------------------------------#ifndef _EXCEPTIO_H
#define _EXCEPTIO_H
#include 
#include 
using namespace std;
class FileError
{
private:
string filename;
public:
FileError( const string& file) : filename(file){ }
string getName() const{ return filename; }
};
class OpenError : public FileError
{
public:
OpenError( const string& file):FileError(file){ }
};
class ReadError : public FileError
{
public:
ReadError( const string& file):FileError(file){ }
};
class WriteError : public FileError
{
public:
WriteError(const string& file):FileError(file){ }
};
#endif

EXCEPTION HANDLING FOR FILES

■

647

䊐 Implementing Your Own Exception Handling
You can exploit the error tracking options that state flags give you to implement your
own exception handling for files. For example, a method that reads records from a file
can throw an exception when the state flag ios::eof is raised, that is, when the end of
the file is reached.
The opposite page shows typical exception classes organized in a hierarchy that can be
used to represent error conditions on opening, reading from, and writing to a file. In each
case the file name is saved for evaluation by the exception handler.

䊐 Standard Exception Handling for Streams
C++ also provides standard exception handling for streams. You can use the exceptions() method to specify the flags in the status-word of a stream that will cause
exceptions to be thrown.
The exceptions() method is defined in the ios stream base class. The method
expects one or multiple state flags separated by the | sign. An exception is then thrown
for the flags specified.

Example: ifstream ifstrm("account.fle");
fstrm.exceptions(ios::failbit | ios::badbit);

On accessing the fstrm stream an exception is thrown if either one of the flags
ios::failbit or ios::badbit is raised. The operation that caused the error is then
terminated and the state flags are cleared by a call to clear(rdstate());.
The exception thrown here is of a standard exception class, failure. This type is
defined as a public element in the ios base class and comprises the virtual method
what() that returns a C string containing the cause of the error. The exception handler
will normally send the string to standard error output.
You can call exceptions() without any arguments to discover the state flags in a
status-word of a stream that can cause an exception to be thrown. If a bit is set in the
return value of the exceptions() method, an appropriate exception will be thrown
whenever this error occurs.

Example: iostate except = fstrm.exceptions();
if( except & ios::eofbit) ...

This statement uses a bitwise AND operator to ascertain whether an exception is thrown
when end-of-file is reached.

648

■

CHAPTER 29

■

MORE ABOUT FILES

PERSISTENCE OF POLYMORPHIC OBJECTS

// account.h : Defines the classes
//
Account, DepAcc, and SavAcc
//
with virtual read and write methods.
// -------------------------------------------------// . . .
enum TypeId { ACCOUNT, DEP_ACC, SAV_ACC };
class Account
{
private:
// Data members: as previously defined.
public:
// Constructor, access methods ...
virtual TypeId getTypeId() const { return ACCOUNT;}
virtual ostream& write(ostream& fs) const;
virtual istream& read(istream& fs);
};
class DepAcc : public Account
{
// Data members, constructor, . . .
TypeId getTypeId() const { return DEP_ACC; }
ostream& write(ostream& fs) const;
istream& read(istream& fs);
};
class SavAcc: public Account
{
// Data members, constructor, . . .
TypeId getTypeId() const { return SAV_ACC; }
ostream& write(ostream& fs) const;
istream& read(istream& fs);
};

The methods read() and write() of class DepAcc
// account.cpp: Implements the methods.
// ---------------------------------------------------#include "account.h"
ostream& DepAcc::write(ostream& os) const
{
if(!Account::write(os))
return os;
os.write((char*)&limit, sizeof(limit) );
os.write((char*)&deb, sizeof(deb) );
return os;
}
istream& DepAcc::read(istream& is)
{
if(!Account::read(is))
return is;
is.read((char*)&limit, sizeof(limit) );
is.read((char*)&deb, sizeof(deb));
return is;
}
// . . .

PERSISTENCE OF POLYMORPHIC OBJECTS

■

649

䊐 Storing Polymorphic Objects
Imagine you want to make the objects of a polymorphic class hierarchy persistent, that is,
store them in a file. You need to ensure that an object can be reconstructed precisely
when it is read. This gives rise to the fact that objects in polymorphic class hierarchies
contain virtual methods. So it is not simply a case of making the data members of an
object into records and writing them to a file.

✓

NOTE
1. You must write both the type and the data members of the object to a file.
2. If the objects contain dynamic members, you must save the referenced objects themselves along with
information on the object type.

To allow the class to assume control over object storage, you need methods that allow
the object to write its own data members to a file. The methods can have a virtual definition within the class hierarchy. Thus, if pointers are used to reference objects, the appropriate read/write operation for each object will be called.

䊐 Storing Objects of the Account Hierarchy
The opposite page shows the Account class, with which you should already be familiar.
Virtual file I/O methods have now been added. The implementation of the read() and
write() methods was discussed earlier in Chapter 18, “Fundamentals of File Input and
Output,” and is unchanged.
The derived classes DepAcc and SavAcc also contain definitions of the read() and
write()methods that read only their “own” objects and write them to files. The implementation first calls the appropriate base class method. If no errors occur, it is simply a
question of transferring the additional data members of the derived class to or from a file.
At present, no type information will be written to file or read from file. This task will
be performed by a special class whose features are used for file management. The following section contains more details on this topic.

650

■

CHAPTER 29

■

MORE ABOUT FILES

PERSISTENCE OF POLYMORPHIC OBJECTS (CONTINUED)

// account.h : (continued)
// Add definition of AccFile class
// -------------------------------------------------// . . .
#include "exceptio.h"
class AccFile
{
private:
fstream f;
// Stream
string name;
// Filename
public:
AccFile(const string& s) throw(OpenError);
~AccFile(){ f.close(); }
long
append( Account& acc) throw(WriteError);
Account* retrieve( long pos ) throw(ReadError);
};

The append() method
// account.cpp: (continued)
// Implements methods of class AccFile.
// --------------------------------------------------long AccFile::append( Account& acc) throw( WriteError)
{
f.seekp(0L, ios::end);
// Seek to end,
long pos = f.tellp();
// save the position.
if( !f )
throw WriteError(name);
TypeId id = acc.getTypeId(); // To write the TypeId
f.write( (char*)&id, sizeof(id));
if(!f)
throw WriteError(name);
else
acc.write(f);
if(!f)
throw WriteError(name);
else
return pos;
}
// . . .

// To write an object
// to the file.

PERSISTENCE OF POLYMORPHIC OBJECTS (CONTINUED)

■

651

䊐 A Scenario
Imagine you want to save the data for various account types, including current and savings accounts to a file. Since the objects you need to save belong to different types, you
must save both the data members and the type of object. This is the only way to ensure
that an object will be correctly reconstructed when read.
The methods in the class you are going to define should throw exceptions if errors
occur. The exception type thrown by a method is stated in the exception specification.

䊐 The AccFile Class
The AccFile class, which is used for random access to a file with account data, is
defined opposited. The data members are an fstream type file stream and a string used
for storing the file name.
The constructor saves the file name and opens a given file for reading and appending
at end-of-file. If the file cannot be opened, the constructor throws an OpenError class
exception.
The append() method writes an account passed to it as an argument at end-of-file
and returns the position where the account was written into the file.
In order to get the current type of the argument, the virtual method getTypeid() is
called. Depending on this type the append() method will write the appropriate type
field to the file and then call the virtual method write(), allowing the current object
to write itself to the file. If a write error occurs, the method will throw a WriteError
type exception.
The retrieve() method first reads the type identifier and then determines the type
of object it needs to allocate memory for. The data from the file is then transferred to the
dynamically allocated object by a call to the virtual method read(). Here too, an
exception will be thrown if a stream access error occurs.

652

■

CHAPTER 29

■

MORE ABOUT FILES

APPLICATION: INDEX FILES

The insert() method of class IndexFile
// index.cpp: (continued)
// Implements the methods of class IndexFile.
// ---------------------------------------------------// . . .
void IndexFile::insert(long k, long n)
throw(ReadError, WriteError)
{
IndexEntry entry;
int size = entry.recordSize();
// Length of an
// index entry.
index.clear();
index.seekg(0, ios::end);
long nr = index.tellg();
// Get file length
// 0, if file is empty.
if(!index) throw ReadError(name);
nr -= size;
// Last entry.
bool found = false;
while( nr >= 0 && !found )
// Search for position
{
// to insert
if( !entry.read_at(index, nr))
throw ReadError(name);
if( k < entry.getKey())
// To shift.
{
entry.write_at(index, nr + size);
nr -= size;
}
else
{
found = true;
}
}
entry.setKey(k); entry.setPos(n);
entry.write_at(index, nr + size);
if(!index)
throw WriteError(name);
}

// insert

APPLICATION: INDEX FILES

■

653

It makes sense to organize data sequentially in files if you need to walk through all the
records regularly. This is particularly true for files used to store salary data or phone bills.
However, most applications need to provide quick access to specific data. For example, a user would definitely prefer to be able to locate an account quickly by reference to
an account number, rather than searching through a file from top to bottom. Index files
can mean a real performance boost in cases like this.

䊐 Index Files
An index file comprises a so-called primary file containing the live data, and an index. The
index consists of pairs of keys and record positions for the primary file. A key stored in the
index will identify a record in the primary file. This situation can be more easily
explained by the following graphic:

Index

Primary file

12345 |

The index is sorted by reference to the keys for speed of access, allowing you to perform a
binary search to locate the position of a record.

䊐 Inserting into the Index
We can use the IndexFile class definition to represent an index. The insert()
method, which correctly inserts a new record into the sorted index, is defined opposite.
The read pointer is first set to end-of-file for insertions. If the current position is 0,
that is, the file is empty, the entry is inserted at offset 0. In all other cases, any entries
whose keys are greater than the new key are shifted down to make room for the new
entry.

654

■

CHAPTER 29

■

MORE ABOUT FILES

IMPLEMENTING AN INDEX FILE

Representing the index file
// index.h:
Defines the class IndexFile
// --------------------------------------------------class IndexFileSystem : public AccFile, public IndexFile
{
private:
string name;
public:
IndexFileSystem(const string s)
: AccFile(s + ".prim"), IndexFile(s + ".ind")
{ name = s; }
void
insert ( Account& acc );
Account* retrieve( long key );
};

Methods insert() and retrieve() of class IndexFileSystem
// index.cpp:
Implementing the methods.
// --------------------------------------------------void IndexFileSystem::insert(Account& acc)
{
// No multiple entries:
if(search(acc.getNr()) == -1)
{
long pos = append(acc); // Insert in primary file
if(pos != -1)
// Insert in
IndexFile::insert(acc.getNr(), pos);
// index file.
}
}
Account* IndexFileSystem::retrieve(long key )
{ long pos = search(key);
// Get the record address:
if( pos == -1 )
// Account number found?
return NULL;
else
{ IndexEntry entry;
// Read the index entry:
IndexFile::retrieve(entry, pos);
// Read from primary file:
return( AccFile::retrieve( entry.getPos() ));
}
}

IMPLEMENTING AN INDEX FILE

■

655

䊐 Index File for Account Management
Since an index file consists of a primary file and an index, it makes sense to derive the
class used to represent an index file from the classes of the primary file and the index file.
Let’s now look at a sample index file, used for managing bank accounts.
The IndexFileSystem class, which is derived from the two previously defined
classes AccFile and IndexFile, is defined on the opposite page. The only data member is a string for the file name. The constructor expects a file name as an argument and
composes names for the primary file and the index by adding a suitable suffix. Base initializers are then used to open the corresponding files.
It is not necessary to define a destructor, since files are automatically closed when the
base class destructors are called.

䊐 Inserting and Retrieving Records
The insert() method was defined to insert new records. It first calls the search()
method to check whether the account number already exists in the index. If not, a new
record is appended to the end of the primary file using the append() method. Then the
key and the address of the record are inserted into the index.
The IndexFileSystem class also contains the retrieve() method, which is used
to retrieve records from the primary file. The key, key, which is passed to the method, is
used by the search() method to look up the address of the required record in the
index. Then the record is retrieved from the primary file by the AccFile class
retrieve() method.
Only the retrieve() methods for the IndexFile and AccFile classes and the
search() method, which performs a binary search in the index, are needed to complete the index file implementation. It’s your job to implement these three methods as
your next exercise!
Using a sorted file to implement an index has the disadvantage that records need to
be shifted to make room to insert new records. As shifting is time-consuming, an index is
normally represented by a tree, which needs less reorganization.

■

CHAPTER 29

■

exercise s

656

MORE ABOUT FILES

EXERCISES

Exercise 1
Class IndexFile
class IndexFile
{
private:
fstream index;
string name;

// Filename of the index

public:
IndexFile(const string s) throw(OpenError);
~IndexFile( ){ index.close(); }
void insert( long key, long pos)
throw(ReadError, WriteError);
long search( long key) throw(ReadError);
void retrieve(IndexEntry& entry, long pos )
throw( ReadError);
};

Class AccFile
enum TypeId { ACCOUNT, DEPOSIT, SAVINGS };
class AccFile
{
private:
fstream f;
string name;
// Filename of primary file
public:
AccFile(const string s) throw(OpenError);
~AccFile(){ f.close(); }
long
append( Account& acc) throw(WriteError);
Account* retrieve( long pos ) throw(ReadError);
};

EXERCISES

■

657

Exercise 1
Complete and test the implementation of the IndexFileSystem class.The
methods should throw exceptions of an appropriate FileError type if an error
occurs.
a. Complete the constructor of the IndexFile class in order to throw an
exception of the type OpenError if the file can not be opened.
b. Write the retrieve() method for the IndexFile class.The method
retrieves a record at a given position in the index.
c. Define the search() method, which looks up an index entry for an
account number passed to it as an argument. Base the method on the
binary search algorithm.
Return value: The position of the record found, or -1 if the account
number is not found in the index.
d. Then define the retrieve() method for the AccFile class, which first
evaluates the type field at a given position in the account file, then
dynamically allocates memory for an object of the appropriate type, and
finally reads the data for an account from the file.
e. Write a main() function that uses a try block to create an IndexFileSystem type object and to insert several accounts of various types
into the index file.The subsequent user dialog reads a key and displays
the corresponding record on screen.Write an exception handler to handle the various errors that could occur.The name of the file and the
cause of the error must be output in any case of error.

658

■

CHAPTER 29

MORE ABOUT FILES

Hash Files
Hash Files profit from random file access in localizing file records directly. Each
file record must have the same length and it must be identified by a unique key,
the so-called hash key.
Hash file

4713123434

Hash function

Key

The idea behind hashing is to provide a function, the hash function, which is
applied to the hash key of a record and yields the address of the file record. If
the file records are numerated and if the hash key equals the record number, a
simple hash function can be the identical function. However, hash keys, such as
account or insurance numbers, consist of a fixed number of digits that do not
start with 0.
The following example shows a frequently used hash function
Example: Hash(key) = key % b;
This hash function transforms the hash value key into a record number between
0 and b-1.The number 0 is the first record number and b-1 is the last record
number in the address space of the hash file. For a sufficiently large prime
number b, the function Hash()yields a good distribution of file records in the
address space.
However, the hash function maps the hash key values key, key + b, key + 2*b,
etc. to the same record number. In this case collisions may occur, for example,
when a new record being inserted hashes to an address that already contains a
different record.
To solve this conflict, the new record must be inserted at some other
position.The process of finding another position at which to insert the new
record is called collision resolution.
Linear solution is a common collision resolution technique: Starting at the
occupied position, subsequent positions are checked sequentially until an
available position is found.The new file record is then inserted at this position.

EXERCISES

■

659

Exercise 2
A hash file is required for speed of access to customer data.The concept of hash
files is explained on the opposite page.To keep things simple, each record in the
hash file contains only a customer id and a name.The customer id is the key
used by the hash function opposite to compute the address of the record. Use
linear solution as collision resolution technique.
Note: Linear solution provides for adequate access times if the address space is
sufficiently large and not too full. It is also important to distribute the record
numbers yielded by the hash function evenly throughout the available address
space.The hash function opposite will guarantee a good distribution if b is a
sufficiently large prime number.
■

■

■

Develop the HashEntry class used to represent customer data.You need
to store the customer id as an unsigned long value and the name of
the customer as a char array with a length of 30. Supply default values for
the constructor declaration and additionally declare the read_at() and
write_at() methods that read customer information at a given position
in a stream or write information at that position. Both methods expect
the position and the stream as arguments.
Define the HashFile class to represent a hash file.The private members of the class comprise an fstream type file stream, a string used to
store the file name, an int variable used to store the number b, and the
hash function shown opposite as a method.The public members comprise a constructor that expects a file name and a number b as arguments. It opens the corresponding file for read and write access.The
destructor closes the file.
Additionally declare the methods insert() and retrieve() to insert or
retrieve single records. Both methods use a call to the hash function to
compute the appropriate record number in the hash file. If a collision
occurs, the methods perform a sequential search to locate the next free
slot in the address space (mod of address space size), or the desired customer data.
Test the HashFile by writing a main function that creates hash file with a
small address space (e.g. b = 7).Add various customer information
records to the hash file and then retrieve this data. Deliberately provoke
collisions using the customer ids 5, 12, and 19, for example.

■

CHAPTER 29

■

solutions

660

MORE ABOUT FILES

SOLUTIONS

Exercise 1
// -----------------------------------------------------// exceptio.h : Error classes for file processing
// -----------------------------------------------------// Unchanged (cf. earlier in this chapter).

// -----------------------------------------------------// account.h :
// Defines the classes
//
Account, DepAcc, and SavAcc
// with virtual read and write methods as well as
// the class AccFile to represent account files.
// -----------------------------------------------------#ifndef _ACCOUNT_H
#define _ACCOUNT_H
#include 
#include 
#include 
#include 
using namespace std;
#include "exceptio.h"
enum TypeId { ACCOUNT, DEP_ACC, SAV_ACC };
class Account
{
private:
string name;
unsigned long nr;
double balance;

// Data elements:

public:
// Constructor:
Account( const string c_name = "X",
unsigned long c_nr
= 1111111L,
double c_balance
= 0.0)
: name(c_name), nr(c_nr), balance(c_balance)
{ }
virtual ~Account() {}

// Virtual destructor

SOLUTIONS

■

// Access methods here:
long getNr() const { return nr; }
void setNr(unsigned long n){ nr = n; }
// . . .
// The other methods:
virtual TypeId getTypeId() const { return ACCOUNT; }
virtual ostream& write(ostream& fs) const;
virtual istream& read(istream& fs);
virtual void display() const
{
cout << fixed << setprecision(2)
<< "----------------------------------\n"
<< "Account holder:
" << name
<< endl
<< "Account number:
" << nr
<< endl
<< "Balance of account:
" << balance << endl
<< "----------------------------------\n"
<< endl;
}
};
class DepAcc : public Account
{
private:
double limit;
// Overdrawn limit
double interest;
// Interest rate
public:
DepAcc( const string s = "X",
unsigned long n = 1111111L,
double bal = 0.0,
double li = 0.0,
double ir = 0.0)
: Account(s, n, bal), limit(li), interest(ir)
{ }
// Access methods:
// . . .
// The other methods are implicit virtual:
TypeId getTypeId() const { return DEP_ACC; }
ostream& write(ostream& fs) const;
istream& read(istream& fs);

661

662

■

CHAPTER 29

MORE ABOUT FILES

void display() const
{
Account::display();
cout << "Overdrawn limit:
" << limit << endl
<< "Competitive interest: " << interest
<< "\n----------------------------------\n"
<< endl;
}
};
class SavAcc: public Account
{
private:
double interest;

// Compound interest

public:
// Methods as in class DepAcc.
};
// -------------------------------------------------// The definition of class AccFile
class AccFile
{
private:
fstream f;
string name;

// Filename

public:
AccFile(const string& s) throw(OpenError);
~AccFile(){ f.close(); }
long append( Account& acc)
throw(WriteError);
Account* retrieve( long pos) throw(ReadError);
void display() throw( ReadError);
};
#endif

SOLUTIONS

// ---------------------------------------------------// account.cpp
// Implement methods of the classes
// Account, DepAcc, SavAcc, and AccFile.
// ---------------------------------------------------#include "account.h"
#include 
ostream& Account::write(ostream& os) const
{
os << name << '\0';
os.write((char*)&nr, sizeof(nr) );
os.write((char*)&balance, sizeof(balance) );
return os;
}
istream& Account::read(istream& is)
{
getline( is, name, '\0');
is.read((char*)&nr, sizeof(nr) );
is.read((char*) &balance, sizeof(balance));
return is;
}
ostream& DepAcc::write(ostream& os) const
{
if(!Account::write(os))
return os;
os.write((char*)&limit, sizeof(limit) );
os.write((char*)&interest, sizeof(interest) );
return os;
}
istream& DepAcc::read(istream& is)
{
if(!Account::read(is))
return is;
is.read((char*)&limit, sizeof(limit) );
is.read((char*)&interest, sizeof(interest));
return is;
}
// ostream& SavAcc::write(ostream& os) const
// istream& SavAcc::read(istream& is)
// as in class DepAcc.

■

663

664

■

CHAPTER 29

MORE ABOUT FILES

// ---- Methods of class AccFile ---AccFile::AccFile(const string& s) throw( OpenError)
{
ios::openmode mode = ios::in | ios::out | ios::app
| ios::binary;
f.open( s.c_str(), mode);
if(!f)
throw OpenError(s);
else
name = s;
}
void AccFile::display() throw(ReadError)
{
Account acc, *pAcc = NULL;
DepAcc depAcc;
SavAcc savAcc;
TypeId id;
if( !f.seekg(0L))
throw ReadError(name);

cout << "\nThe account file: " << endl;
while( f.read((char*)&id, sizeof(TypeId)) )
{
switch(id)
{
case ACCOUNT: pAcc = &acc;
break;
case DEP_ACC: pAcc = &depAcc;
break;
case SAV_ACC: pAcc = &savAcc;
break;
default: cerr << "Invalid flag in account file"
<< endl;
exit(1);
}
if(!pAcc->read(f))
break;
pAcc->display();
cin.get();
}

// Go on with return

SOLUTIONS

if( !f.eof())
throw ReadError(name);
f.clear();
}
long AccFile::append( Account& acc) throw( WriteError)
{
f.seekp(0L, ios::end);
// Seek to end,
long pos = f.tellp();
// save the position.
if( !f )
throw WriteError(name);
TypeId id = acc.getTypeId();
f.write( (char*)&id, sizeof(id));

// Write the TypeId

if(!f)
throw WriteError(name);
else
acc.write(f);
// Add an object to the file.
if(!f)
throw WriteError(name);
else
return pos;
}
Account* AccFile::retrieve( long pos) throw(ReadError)
{
f.clear();
f.seekg(pos);
// Set the get pointer
if( !f )
throw ReadError(name);
TypeId id;
f.read( (char*)&id, sizeof(id) );
if(!f)
throw ReadError(name);
Account* buf;
switch( id )
{
case ACCOUNT:
case SAV_ACC:

buf = new Account;
break;
buf = new SavAcc;
break;

// Get TypeId

■

665

666

■

CHAPTER 29

MORE ABOUT FILES

case DEP_ACC:

buf = new DepAcc;
break;

}
if( !(buf->read(f)))
throw ReadError(name);

// Get data

return buf;
}
// ------------------------------------------------------// index.h: Contains definitions of classes
//
IndexEntry representing an index entry,
//
Index
representing the index and
//
IndexFile representing an index file.
// ------------------------------------------------------#ifndef _INDEX_H
#define _INDEX_H
#include 
#include 
#include 
#include "account.h"
using namespace std;
class IndexEntry
{
private:
long key;
long recPos;

// Key
// Offset

public:
IndexEntry(long k=0L, long n=0L){ key=k; recPos=n; }

void
long
void
long

setKey(long k)
getKey() const
setPos(long p)
getPos() const

{
{
{
{

key = k; }
return key; }
recPos = p; }
return recPos; }

int recordSize() const
{ return sizeof(key) + sizeof(recPos); }
fstream& write( fstream& ind) const;
fstream& read( fstream& ind);
fstream& write_at(fstream& ind, long pos) const;
fstream& read_at( fstream& ind, long pos);

SOLUTIONS

■

void display() const
{
cout << "Account Nr: " << key
<< " Position: " << recPos << endl;
}
};
class IndexFile
{
private:
fstream index;
string name;

// Filename of index

public:
IndexFile( const string& s) throw (OpenError);
~IndexFile() { index.close(); }
void insert( long key, long pos)
throw(ReadError, WriteError);
long search( long key) throw(ReadError);
void retrieve(IndexEntry& entry, long pos )
throw(ReadError);
void display() throw(ReadError);
};
class IndexFileSystem : public AccFile, public IndexFile
{
private:
string name;
// Filename without suffix
public:
IndexFileSystem(const string& s)
: AccFile(s + ".prim"), IndexFile(s + ".ind")
{ name = s; }
bool
insert( Account& acc);
Account* retrieve( long key);
};
#endif

667

668

■

CHAPTER 29

MORE ABOUT FILES

// -----------------------------------------------------// index.cpp : Methods of the classes
//
IndexEntry, Index, and IndexFile
// -----------------------------------------------------#include "index.h"
fstream& IndexEntry::write_at(fstream& ind, long pos) const
{
ind.seekp(pos);
ind.write((char*)&key, sizeof(key) );
ind.write((char*)&recPos, sizeof(recPos) );
return ind;
}
fstream& IndexEntry::read_at(fstream& ind, long pos)
{
ind.seekg(pos);
ind.read((char*)&key, sizeof(key) );
ind.read((char*)&recPos, sizeof(recPos));
return ind;
}
fstream& IndexEntry::write(fstream& ind) const
{
ind.write((char*)&key, sizeof(key) );
ind.write((char*)&recPos, sizeof(recPos) );
return ind;
}
fstream& IndexEntry::read(fstream& ind)
{
ind.read((char*)&key, sizeof(key) );
ind.read((char*)&recPos, sizeof(recPos));
return ind;
}
// --------------------------------------------------// Methods of class IndexFile
IndexFile::IndexFile(const string& file) throw (OpenError)
{
ios::openmode mode = ios::in | ios::out | ios::binary;
// Open file if it already exists:
index.open( file.c_str(), mode);
if(!index)
// If the file doesn't exist
{
index.clear();
mode |= ios::trunc;
index.open( file.c_str(), mode);
if(!index)
throw OpenError(name);
}
name = file;
}

SOLUTIONS

■

void IndexFile::display() throw(ReadError)
{
IndexEntry entry;
index.seekg(0L);
if(!index)
throw ReadError("IndexFile: Setting the get pointer");
cout << endl << "The Index: " << endl;
while( true)
{
if( !entry.read(index))
break;
entry.display();
}
if( !index.eof())
throw ReadError(name);
index.clear();
}
long IndexFile::search(long k) throw(ReadError)
{
IndexEntry entry;
long key;
long mid, begin = 0, end;
// Number of file records.
int size = entry.recordSize();
// Length of an index
// entry.
index.clear();
index.seekg(0L, ios::end);
end = index.tellg() / size;
if(!index)
throw ReadError(name);
if( end == 0)
return -1;
end -= 1;

// Position of the last entry

while( begin < end )
{
mid = (begin + end +1)/2 ;
entry.read_at(index, mid*size);
if(!index)
throw ReadError(name);
key = entry.getKey();
if( k < key)
end = mid - 1;
else
begin = mid;
}

669

670

■

CHAPTER 29

MORE ABOUT FILES

entry.read_at(index, begin * size);
if(!index)
throw ReadError(name);
if( k == entry.getKey() )
return begin * size;
else return -1;

// Key found?

}
void IndexFile::insert(long k, long n)
throw(ReadError, WriteError)
{
IndexEntry entry;
int size = entry.recordSize(); // Length of an index
// entry.
index.clear();
index.seekg(0, ios::end);
long nr = index.tellg();
// Get file length
// 0, if file is empty.
if(!index) throw ReadError(name);
nr -= size;
bool found = false;
while( nr >= 0 && !found )
{
if(!entry.read_at(index, nr))
throw ReadError(name);

// Last entry.
// Search position
// to insert

if( k < entry.getKey())
// To shift.
{
entry.write_at(index, nr + size);
nr -= size;
}
else
{
found = true;
}
}
entry.setKey(k); entry.setPos(n);
entry.write_at(index, nr + size);
if(!index)
throw WriteError(name);
}

// Insert

SOLUTIONS

■

void IndexFile::retrieve( IndexEntry& entry, long pos)
throw(ReadError)
{
index.clear();
if(!entry.read_at(index, pos))
throw ReadError(name);
}
// --------------------------------------------------// Implementing the methods of class IndexFileSystem.
bool IndexFileSystem::insert( Account& acc)
throw(ReadError, WriteError)
{
if(search(acc.getNr()) == -1) // No multiple entries.
{
long pos = append(acc);
// Add to primary file.
IndexFile::insert(acc.getNr(), pos); // Add to Index
return true;
}
else
return false;
}
Account* IndexFileSystem::retrieve(long key )
{
// Get the record address from the index:
long pos = search(key);
// Byte offset of
// index entry.
if( pos == -1 )
// Account number doesn't exist.
return NULL;
else
// Account number does exist:
{
IndexEntry entry;
// To read the index eintry
IndexFile::retrieve( entry, pos);
// Get from primary file:
return( AccFile::retrieve( entry.getPos() ));
}
}
// -----------------------------------------------------// index_t.cpp : Testing the index file
// -----------------------------------------------------#include 
#include 
using namespace std;
#include "index.h"
#include "account.h"

671

672

■

CHAPTER 29

MORE ABOUT FILES

int main()
{
try
{
IndexFileSystem database("AccountTest");
Account acc1( "Vivi", 490UL, 12340.57);
database.insert( acc1 );
SavAcc acc2( "Ulla", 590UL, 4321.19, 2.5);
database.insert( acc2 );
DepAcc acc3( "Jeany", 390UL, 12340.20, 10000.0, 12.9);
database.insert( acc3 );
database.IndexFile::display();
cin.get();
database.AccFile::display();
unsigned long key;
cout << "Key? "; cin >> key;
if(database.search(key) != -1)
cout << "Key " << key << " found" << endl;
else
cout << "Key " << key << " not found" << endl;
Account* pAcc = database.retrieve(key);
if( pAcc != NULL )
{
pAcc->display();
delete pAcc;
pAcc = NULL;
}
else cout << "Retrieving failed" << endl;
}
catch(OpenError& err)
{
cerr << "Error on opening the file:" << err.getName()
<< endl;
exit(1);
}
catch(WriteError& err)
{
cerr << "Error on writing into the file: "
<< err.getName() << endl;
exit(1);
}

SOLUTIONS

■

catch(ReadError& err)
{
cerr << "Error on reading from the file: "
<< err.getName() << endl;
exit(1);
}
catch(...)
{
cerr << " Unhandled Exception" << endl;
exit(1);
}
return 0;
}

Exercise 2
// ------------------------------------------------------// exceptio.h : Error classes for file processing
// ------------------------------------------------------// As seen previously in this chapter.

// ------------------------------------------------------// hashFile.h
// Defines the classes
// HashEntry representing a record in a hash file and
// HashFile representing a hash file.
// ------------------------------------------------------#ifndef _HASH_H_
#define _HASH_H_
#include 
#include 
#include 
#include 
#include 
using namespace std;
#include "exceptio.h"
class HashEntry
{
private:
unsigned long nr;
char name[30];

673

674

■

CHAPTER 29

MORE ABOUT FILES

public:
HashEntry(unsigned long n = 0L, const string& s = "")
{
nr = n;
strncpy(name, s.c_str(), 29); name[30]='\0';
}
long
getNr() const { return nr; }
void
setNr(unsigned long n){ nr = n; }
string getName() const { return name; }
void
setName(const string& s)
{ strncpy(name, s.c_str(), 29); name[30]='\0'; }
int getSize() const
{ return(sizeof(long) + sizeof(name)); }
fstream& write(fstream& fs);
fstream& read(fstream& fs);
fstream& write_at(fstream& fs, unsigned long pos);
fstream& read_at(fstream& fs, unsigned long pos);
virtual void display()
{
cout << fixed << setprecision(2)
<< "----------------------------------\n"
<< "Client number:
" << nr
<< endl
<< "Client:
" << name << endl
<< "----------------------------------\n"
<< endl;
cin.get();
}
};
class HashFile
{
private:
fstream f;
string name;
unsigned long b;

// Filename
// Size of address space

protected:
unsigned long hash_func(unsigned long key)
{ return key%b; }
public:
HashFile(const string s, unsigned long n )
throw(OpenError);

SOLUTIONS

■

void insert( HashEntry& rec)
throw( ReadError, WriteError );
HashEntry& retrieve( unsigned long key )
throw( ReadError );
void display();
};
#endif
// ------------------------------------------------------// hashFile.cpp : Methods of classes HashEntry and HashFile
// ------------------------------------------------------#include "hashFile.h"
fstream& HashEntry::write(fstream& f)
{
f.write((char*)&nr, sizeof(nr) );
f.write( name, sizeof(name) );
return f;
}

fstream& HashEntry::read(fstream& f)
{
f.read((char*)&nr, sizeof(nr) );
f.read( name, sizeof(name));
return f;
}
fstream& HashEntry::write_at(fstream& f, unsigned long pos)
{
f.seekp(pos);
f.write((char*)&nr, sizeof(nr) );
f.write( name, sizeof(name) );
return f;
}
fstream& HashEntry::read_at(fstream& f, unsigned long pos)
{
f.seekg(pos);
f.read((char*)&nr, sizeof(nr) );
f.read( name, sizeof(name));
return f;
}

675

676

■

CHAPTER 29

MORE ABOUT FILES

HashFile::HashFile(const string file, unsigned long n)
throw(OpenError)
{
ios::openmode mode = ios::in | ios::out | ios::binary;
f.open(file.c_str(), mode);
// Open file if it
// already exists.
if(!f)
// If file doesn't exist:
{
f.clear();
mode |= ios::trunc;
f.open(file.c_str(), mode);
if(!f)
throw OpenError(name);
}
name = file;
b = n;
HashEntry rec(0L, "");
f.seekp(0L);
for( unsigned long i=0; i < b; i++) // Initialize
{
// the address space
rec.write(f);
if(!f)
throw WriteError(name);
}
}
void HashFile::insert( HashEntry& rec)
throw( ReadError, WriteError)
{
HashEntry temp;
int size = temp.getSize();
// Hash-Wert:
unsigned long pos = hash_func(rec.getNr());
temp.read_at(f, pos*size);
if(!f)
throw ReadError(name);
else
{
if(temp.getNr() == 0L)
rec.write_at(f, pos*size);
else
{
bool found = false;

// Read a slot.

//
//
//
//
//

Slot free?
Yes => Add
to the file.
No => Search for
a free slot.

unsigned long p = (pos*size + size)%(b*size);

SOLUTIONS

■

while( !found && p!= pos*size )
{
temp.read_at(f, p);
if(!f)
throw ReadError(name);
else
if(temp.getNr() == 0L) // Free slot
found = true;
// found.
else
// Proceed to the next slot:
p = (p + size)%(b*size);
}
if( p == pos*size )
// Address space full.
throw WriteError(name);
if ( found == true )
rec.write_at(f,p);

// Add to file.

}
if(!f)
throw WriteError(name);
}
}

HashEntry& HashFile::retrieve( unsigned long key )
throw(ReadError)
{
static HashEntry temp;
int size = temp.getSize();
unsigned long pos = hash_func(key);

// Hash value.

temp.read_at(f, pos*size);

// Read a slot.

if(!f) throw ReadError(name);
if(temp.getNr() == key)
// Found?
return temp;
// Yes => finish
else
// No
=> search
{
unsigned long p = (pos*size + size)%(b*size);
while( p!= pos *size )
{
temp.read_at(f, p);
if(!f)
throw ReadError(name);

677

678

■

CHAPTER 29

MORE ABOUT FILES

else
if(temp.getNr() == key)
return temp;
else
p = (p + size)%(b*size);
}
temp.setNr(0L); temp.setName("");

// Record found.

// Proceed to the
// next slot.
// Key doesn't
// exist.

return temp;
}
}
void HashFile::display()
{
HashEntry temp;
f.seekg(0L);
for(unsigned int i = 0; i < b; i++)
{
temp.read(f);
if(!f)
throw ReadError(name);
temp.display();
}
f.clear();
}
// ------------------------------------------------------// hash_t.cpp : Tests hash files
// ------------------------------------------------------#include 
#include 
#include "hashFile.h"
using namespace std;
int main()
{
try
{
HashFile hash("Client.fle", 7);
cout << "\nInsert: " << endl;
HashEntry kde( 3L, "Vivi");
hash.insert( kde );

// Address space
// of length 7

SOLUTIONS

kde.setNr(10L); kde.setName("Peter");
hash.insert( kde );
kde.setNr(17L); kde.setName("Alexa");
hash.insert( kde );
kde.setNr(21L); kde.setName("Peter");
hash.insert( kde );
kde.setNr(15L); kde.setName("Jeany");
hash.insert( kde );
cout << "\nInsertion complete: " << endl;
hash.display();
unsigned long key;
cout << "Key? "; cin >> key;
HashEntry temp = hash.retrieve(key);
if(temp.getNr() != 0L)
temp.display();
else
cout << "Key " << key
<< " not found" << endl;
}
catch(OpenError& err)
{
cerr << "Error in opening the file:"
<< err.getName() << endl;
exit(1);
}

catch(WriteError& err)
{
cerr << "Error writing to file: "
<< err.getName() << endl;
exit(1);
}
catch(ReadError& err)
{
cerr << "Error reading from file: "
<< err.getName() << endl;
exit(1);
}
return 0;
}

■

679

This page intentionally left blank

chapter

30

More about Pointers
This chapter describes advanced uses of pointers.These include pointers
to pointers, functions with a variable number of arguments, and pointers
to functions.
An application that defines a class used to represent dynamic
matrices is introduced.

681

682

■

CHAPTER 30

■

MORE ABOUT POINTERS

POINTER TO POINTERS

The function accSort()
// accSort.cpp: Sorts an array of pointers to accounts
//
according to the account numbers
// --------------------------------------------------#include "account.h"
void ptrSwap(Account**, Account** );
void accSort( Account** kptr, int n)
{
Account **temp, **minp, **lastp;
lastp = kptr + n - 1;
// Pointer to the last
// pointer in the array.
for( ; kptr < lastp; ++kptr )
{
minp = kptr;
for( temp = kptr + 1; temp <= lastp; ++temp )
{
if( (*temp)->getNr() < (*minp)->getNr() )
minp = temp;
}
ptrSwap( kptr, minp );
}
}
void ptrSwap( Account **p1, Account **p2 )
{
Account *help;
help = *p1; *p1 = *p2; *p2 = help;
}

POINTER TO POINTERS

■

683

䊐 Motivation
Pointer variables are objects that have an address in memory, and this means you can use
pointers to address them. It is thus possible to create pointers to pointers. This is necessary
if
■
■

an array of pointers is to be dynamically allocated, or
a function expects an array of pointers as an argument.

In both cases you need to declare a pointer variable that can access the first element in
the array. Since each element in the array is a pointer, this pointer variable must be a
pointer to a pointer.

䊐 Generating Pointer Arrays Dynamically
Now let’s look into creating a dynamic array of pointers to Account class objects.

Example: Account** ptr = new Account*[400];
The pointer ptr is now pointing at the first pointer in the array with a total of 400
Account* type pointers. The array elements can be addressed as follows:
*ptr
*(ptr + i)

and
and

ptr[0]
ptr[i]

(pointer to index 0)
(pointer to index i)

Access to objects managed by the array is achieved as follows:
**ptr
**(ptr+i)

and
and

*ptr[0]
*ptr[i]

(object addressed by pointer at index 0)
(object addressed by pointer at index i)

䊐 Pointer Arrays as Arguments
When you define a function that expects an array of pointers as an argument, you must
define parameters to match.

Example: void accSort( Account **kptr, int len);
You can use the kptr parameter to manipulate a pointer array whose length is stored in
the second parameter, len. After calling

Example: accSort( ptr, 100);
kptr points to the first pointer ptr[0] in the pointer array ptr. Instead of
Account **kptr you can also use the equivalent form Account *kptr[].
The opposite page shows an implementation of the function accSort(). The func-

tion uses the selection sort algorithm (which you have already worked with) for sorting.
In this case it is important not to sort the accounts itself, but to sort the pointers instead.
This saves time-consuming copying.

684

■

CHAPTER 30

■

MORE ABOUT POINTERS

VARIABLE NUMBER OF ARGUMENTS

Fixed and varying arguments on the stack

char *buffer
Fixed arguments

int max
First varying
argument

va_start(argp,max)
•

Varying arguments
•

•

Last varying
argument

Scheme of a function with varying arguments
#include 
int func( char *buffer, int max, ... )
{
va_list argptr;
// Declares argument pointer.
long arg3;
. . .
va_start( argptr, max);
// Initialization.
arg3 = va_arg( argptr, long ); // Read arguments.
// To use argument arg3.
. . . .
va_end(argptr);
}

// Set argument pointer to NULL.

VARIABLE NUMBER OF ARGUMENTS

■

685

C++ allows you to define functions that allow a variable number of arguments. One
example of a function of this type is the standard C function printf(), which requires
at least one argument, a format string. The printf() function uses the conversion
specifiers in the format string to compute the number and type of arguments that follow.

䊐 Obligatory and Optional Arguments
Functions with a variable number of arguments always expect a fixed number of obligatory arguments and a variable number of optional arguments. At least one obligatory argument is required.
As you would expect, you need to define an appropriate parameter for each obligatory
argument when you define a function of this type. The optional arguments are represented by three dots ... in the parameter list. The function shown opposite, func(),
expects two or more arguments. The prototype is, thus, as follows

Prototype: int func( char *buffer, int max, ...);
To allow functions with a variable number of arguments to be defined, C++ pushes the
last argument onto the stack first. After calling the sample function func(), the stack
looks like the diagram opposite.
The optional arguments are accessed via a pointer, the so-called argument pointer,
which is designated by argptr here. The header files cstdarg or stdarg.h contain
macros, which conform to ANSI standard, to manage the pointer and assure that the
source code will be portable.

䊐 Access to Arguments
The following steps are required to read the optional arguments:
1. The va_list type argument pointer argptr must be declared in addition to
other local variables. The type va_list is defined in the header file stdarg.h
as a typeless or char pointer.
2. The macro va_start() is then called to point the argument pointer argptr
to the first optional argument. va_start() expects two arguments: the name of
the argument pointer and the name of the last obligatory parameter.

Example: va_start( argptr, max );

686

■

CHAPTER 30

■

MORE ABOUT POINTERS

VARIABLE NUMBER OF ARGUMENTS (CONTINUED)

The function input()
// input.cpp: The function input() reads characters
//
from the keyboard and appends '\0'.
//
The input can be corrected with backspace.
// Arguments: 1. Pointer to the input buffer.
//
2. Maximum number of characters to be read
/
3. Optional arguments: Characters that
//
terminate the input.
//
This list has to end with CR = '\r'!
// Returns:
Character that breaks the input.
// --------------------------------------------------#include 
#include 
// For getch() and putch()
int input(char *buffer, int max,... )
{
int c, breakc; // Current character, character to
// break with.
int nc = 0;
// Number of characters read.
va_list argp;
// Pointer to the following arguments.
while(true)
{
*buffer = '\0';
if( ( c = getch()) == 0)
c = getch() + 256;

// Read a character.
// For special keys:
// Extended code + 256.
va_start(argp, max);
// Initialize argp.
do
// Compare with break characters:
if( c == (breakc = va_arg(argp,int)) )
return(breakc);
while( breakc != '\r');
va_end( argp);
if( c == '\b' && nc > 0)
// Backspace?
{
--nc, --buffer;
putch(c); putch(' '); putch(c);
}
else if( c >= 32 && c <= 255 && nc < max )
{
// Place character into the buffer
++nc, *buffer++ = c; putch(c); // and output.
else if( nc == max)
// Is end of buffer reached?
putch('\a');
// Beep.
}

}

VARIABLE NUMBER OF ARGUMENTS (CONTINUED)

■

687

3. When the macro va_arg() is called, the optional argument pointed to by
argptr is read from the stack. The arguments of va_arg() are the name of the
argument pointer and the type of the optional argument:

Example: arg3 = va_arg( argptr, long);
Each call to the macro va_arg() sets the argument pointer to the next optional
argument. The result of va_arg() has the type stated in the call. It must be
identical to the type of the corresponding optional argument.
There is no special terminating condition for the last optional argument. A specific value (such as NULL, ⫺1 or CR) can be used, or the current number of arguments can be defined by an obligatory argument.
4. After evaluating the arguments the argument pointer is set to NULL by the
va_end() macro:

Example:

va_end( argptr);

Optional arguments can also be read more than once. The procedure described above is
repeated beginning at Step 2, that is, with the macro va_start().

䊐 Notes on the Example Opposite
The sample function input() on the opposite page uses the getch() function to read
character input from the keyboard and store it in the buffer addressed by the first argument. The second argument defines the maximum number of characters to be read. All
other arguments are characters that can terminate keyboard input. The last argument
must be a return character ('\r')!

Example: #define ESC

27
// ESC key
#define F1
(256 + 59)
// F1 key
input( name, 20, ' ', ESC, F1, '\r');

This call to input() reads up to 20 characters and stores them in the array name. Input
can be terminated by pressing the space, ESC, F1, or return keys. The return value is the
corresponding character code. Non-printable characters are ignored unless stated as
optional arguments.
Special keys, such as the function keys, return a value of 0 for the first call to
getch() and the extended code for the second call. For function keys this code is within
the range 59–68. To distinguish extended codes from normal ASCII codes (0–255),
the value 256 is added to the extended code. A table of extended codes is available in
the Appendix.

688

■

CHAPTER 30

MORE ABOUT POINTERS

■

POINTERS TO FUNCTIONS

A jump table
// funcptr.cpp: Demonstrates the use of an array
//
of pointers to functions.
// -------------------------------------------------#include 
#include 
// Prototype of atoi()
#include 
// Macros toupper(), tolower()
using namespace std;
void

error_message(char *), message(char *),
message_up(char *), message_low(char *);

void (*functab[])(char *) = { error_message, message,
message_up, message_low };
char call[]="Input: 1,2, or 3";
int main()
{
int n = 0;
cout << "Which of the three functions "
<< "do you want call (1,2, or 3)?\n";
cin >> n;
if( n<1 || n>3)
(*functab[0])( call );
else
(*functab[n])("Hello, world\n");
return 0;
}
void error_message( char *s)
void message( char *s)

{
{

cerr << s << endl; }
cout << s << endl; }

void message_up( char *s)
{
int c;
for( ; *s != '\0';++s) c = toupper(*s),cout.put(c);
}
void message_low( char *s)
{
int c;
for( ; *s != '\0';++s) c = tolower(*s), cout.put(c);
}

POINTERS TO FUNCTIONS

■

689

䊐 Using Pointers to Functions
In C++ the name of a function is a constant pointer to that function. It addresses the
machine code for the function. This is a situation that we have already seen for arrays—
the array name is also a constant pointer to the first array element.
There are many uses for pointers to functions. You can save them in an array to form a
jump table. Individual functions are then accessible via an index.
A pointer to a function can also be passed as an argument to another function. This
makes sense if the function you are calling needs to work with different functions
depending on the current situation.
The standard function qsort() is an example of this. qsort() uses the quick sort
algorithm to sort an array. Depending on the type of the array elements and the sort criteria, the qsort() function will expect as argument another comparison function.

䊐 Declaring Pointers to Functions
A pointer to a function is declared as follows:

Syntax:

type (* funcptr)( parameter_list );

This defines the variable funcptr, which can store the address of a function. The function has the type type and the parameter list stated. The first pair of parentheses is also
important for the declaration. The statement type *funcptr(parameter_list);
would declare a function funcptr that returned a pointer.
Now let’s point funcptr to the function compare() and call compare() via the
pointer.

Example: bool compare(double, double); // Prototype
bool (*funcptr)(double, double);
funcptr = compare;
(*funcptr)(9.1, 7.2);

Calling (*funcptr)() is now equivalent to calling compare(). The declaration of
compare() is necessary to let the compiler know that compare is the name of a function.
In the program shown opposite, functab is an array with four pointers to functions
of the void type, each of which expects a C string as an argument. functab is initialized by the functions stated in its definition and thus functab[0] points to
error_message(), functab[1] to message(), etc. When the program is executed, the function with the specified index is called.

690

■

CHAPTER 30

■

MORE ABOUT POINTERS

COMPLEX DECLARATIONS

1st Example:

char (* strptr) [50]
3.

1.

0.

2.

0. strptr is a
1. pointer to
2. an array with 50 elements of type
3. char.

2nd Example:

long

*

(* func () ) []

5.

4.

2.

0.

1.

3.

0. func is a
1. function with return value of type
2. pointer to
3. an array with elements of type
4. pointer to
5. long.

3rd Example:

char *
6.

5.

(* (* funcptr ) () ) []
3.

1.

0.

0. funcptr is a
1. pointer to
2. a function with return value of type
3. pointer to
4. an array with elements of type
5. pointer to
6. char.

2.

4.

COMPLEX DECLARATIONS

■

691

䊐 Operators and Complex Declarations
In the declaration and definition of a function or a variable the same operators that you
find in expressions are used in addition to the base type and the name. These operators
are:

Operator

Significance

[]

Array with elements of type

()

Function with return value of type

*

Pointer to

&

Reference to

A complex declaration always uses more than one of these operators.

Example: char *strptr[50];
This declares strptr as an array of pointers to char. In a declaration, a combination of
the three operators is permissible, however, the following exceptions apply:
■
■

the elements of an array cannot be functions
a function cannot return a function or an array (but it can return a pointer to a
function or an array).

Operators have the same precedence in declarations as in expressions. You can use
parentheses to redefine the order of precedence.

䊐 Rules
When a complex declaration is evaluated, the following rules are applied:
0. Always start with the identifier being declared.
Then repeat the following steps until all the operators have been resolved:
1. If the parentheses/brackets () or [] are on the right, they are interpreted.
2. If there is nothing or just a right bracket on the right ), the asterisk on the left is
interpreted, if it exists.
At last the base type is interpreted.
This proceeding is demonstrated by the example opposite. The above rules apply to
both the function and each of its arguments.

692

■

CHAPTER 30

■

MORE ABOUT POINTERS

DEFINING TYPENAMES

1st Example:
typedef DayTime FREETIME;
FREETIME timeArr[100];

2nd Example:
typedef struct { double re, im; } COMPLEX;
COMPLEX z1, z2, *zp;

3rd Example:
typedef enum { Mo, Tu, We, Th, Fr } WORKDAY;
WORKDAY day;

4th Example:
typedef enum { Diamonds, Hearts,
Spades, Clubs } COLOR;
typedef enum { seven, eight, nine, ten ,
jack, queen, king, ace } VALUE;
typedef struct
{
COLOR f;
VALUE w;
} CARD;
typedef CARD[10] HAND;
HAND player1, player2, player3;

DEFINING TYPENAMES

■

693

䊐 The typedef Keyword
C++ allows you to give types a new name using the keyword typedef.

Example: typedef unsigned char BYTE;
This defines the type name BYTE, which can then be used as an abbreviation of the
unsigned char type. The statement

Example: BYTE array[100];
will then define an array array with 100 elements of the unsigned char type. Type
names are normally uppercase, although this is not mandatory.

Examples: typedef int* INTPTR;
typedef enum{ RED, AMBER, GREEN } Lights;

Here, INTPTR identifies the type “pointer to int” and Lights is an enumerated type.
The new type name always assumes the position of a variable name in a typedef
definition. Omitting the typedef prefix will define a variable name but not a new type
name.
Type definitions do not allocate memory and do not create a new type. They simply
introduce a new name for an existing type.

Example: typedef char* (*PTR_TO_FUNC)();
The type name PTR_TO_FUNC is an abbreviation for the type “pointer to a function that
returns a pointer to char.” The declaration

Example: PTR_TO_FUNC search;
is then equivalent to
char* (*search)();

䊐 Advantages
The major advantage of using typedef is that it improves the readability of your programs, especially when complex types are named.
One additional advantage is that you can isolate platform dependent types. When a
program is ported to another platform, you only need to change the platform dependent
type once in the typedef definition.

694

■

CHAPTER 30

■

MORE ABOUT POINTERS

APPLICATION: DYNAMIC MATRICES

Class Matrix
// matrix.h: Representing dynamic matrices.
// --------------------------------------------------#include 
#include 
using namespace std;
class Row
{
double *ro;
int size;
public:
Row( int s) { size = s; ro = new double[s]; }
~Row(){ delete[]ro; }
double& operator[](int i) throw(out_of_range)
{if(i < 0 || i > size)
throw out_of_range("Column index: Out of Range\n");
else
return ro[i];
}
};
class Matrix
{
Row **mat;
// Pointer to array of rows
int lines, cols;
// Number of rows and columns
public:
Matrix(int ro , int co )
{ lines = ro; cols = co;
mat = new Row*[lines];
for(int i=0; i < lines; i++)
mat[i] = new Row(cols);
}
~Matrix()
{
for(int i=0; i < lines; i++)
delete mat[i];
delete[] mat;
}
int getLines() const { return lines; }
int getCols() const { return cols; }
Row& operator[](int i) throw(out_of_range)
{ if(i < 0 || i > cols)
throw out_of_range("Row index: Out of Range\n");
else
return *mat[i];
}
};

APPLICATION: DYNAMIC MATRICES

■

695

Now let’s develop an application that uses a class to represent dynamic matrices. Matrices are used for computing vectors needed to move, rotate, or zoom images in graphics
programming, for example.

䊐 The Matrix Class
Memory is to be allocated dynamically to a matrix m at runtime. Additionally, it should
be possible to use the index operator to access the elements of the matrix.

Example: m[i][j]

// Element in row i, column j

The class will therefore need a dynamic member to reference the matrix. As you already
know, a matrix is a single-dimensional array whose elements are single-dimensional
arrays themselves.
The class Row, which can be used to represent single-dimensional arrays of double
values, is defined opposite. The index operator is overloaded for the Row class to allow
an exception of the out_of_range type to be thrown for invalid indices.
The Matrix class contains a dynamic member, mat, which can address an array of
pointers to Row objects. mat is thus a pointer to a pointer.

䊐 Constructor, Destructor, and Subscript Operator
The constructor in the Matrix class creates an array of lines pointers to objects of the
Row type. A loop is then used to allocate memory to the rows dynamically.
In contrast, the destructor releases the memory occupied by the line arrays first, before
releasing the space occupied by the pointer array mat.
The subscript operator in the Matrix class returns the line array i for a given index
i. When the following expression is evaluated

Example: m[2][3]
the first call is to the subscript operator of the Matrix class, which returns a line array to
index 2. Then the subscript operator of the Row class is called for the line array. It
returns a reference to the double value at index 3.
You will be enhancing the Matrix class in the exercises to this chapter by overloading the copy constructor and the assignment, for example.

■

CHAPTER 30

■

exercise s

696

MORE ABOUT POINTERS

EXERCISES

Listing for exercise 1
#include 
using namespace std;
char* color[] = {"WHITE", "PINK", "BLUE", "GREEN" };
int main()
{
cout << *color[1] << " "
<< *color << " "
<< *(color[3] + 3) << " "
<< color[2] + 1 << " "
<< *( *(color + 1) + 3)
<< endl;
return 0;
}

For exercise 3
The standard function

qsort()

#include 
void qsort( void* array, size_t n, size_t size,
int (*compare)(const void*, const void*));

The function qsort(), ”quick sort,” sorts an array of n elements whose first
element is pointed to by array. The size of each element is specified by size.
The comparison function pointed to by compare is used to sort the content
of the array in ascending order. qsort() calls the function with two arguments
that point to the elements being compared.
You will normally need to define the comparison function yourself.The
function must return an integer less than, equal to, or greater than zero if the
first argument is less than, equal to, or greater than the second.

✓

NOTE
Since the comparison function compare will be called as a C function, it should
be declared and defined as
extern "C" int compare(....);

Refer to the Appendix “Binding C Functions.”

EXERCISES

■

697

Exercise 1
What does the program opposite output on screen?
Exercise 2
Write a function min() that computes and returns the minimum of two positive
numbers.The function expects a variable number of unsigned int values as
arguments.The last argument must be the number 0.

✓

Exercise 3
Write a C++ program that compares the speed of the quick sort and selection
sort algorithms
Sort two identical sequences of random numbers of the type int to test the
sort algorithms. Read the maximum number of random numbers from the
keyboard and allocate the needed memory dynamically. Display the time in
seconds required for the sort operation on screen after each sort operation.
NOTE
You can use the SelectionSort() function defined in Exercise 4 of Chapter 17 as an algorithm for
sorting the int values in this exercise.
Use the standard function qsort(), whose prototype is defined opposite, as quick sort algorithm
for this exercise.

Exercise 4
Write additional methods to complete the Matrix class.
■

■

■

Add a const version of the subscript operator to the Row and Matrix
classes. Use an inline implementation.
Define a constructor for the Matrix class.The constructor dynamically
allocates a matrix with a given number of rows and columns, and initializes the matrix elements with a given value.Also write a copy constructor.
Overload the assignment operator = and the compound assignment operator +=.
Addition is defined for two n ⫻ n matrices, A and B, which have equal numbers of rows and columns.The sum C is a n ⫻ n matrix whose elements
are computed by adding elements as follows
C[i,j] = A[i,j] + B[i,j]

■

for

i, j = 0, ..., n-1

Test the Matrix with a suitable main function that calls all the methods.
Display the results of the calculations on screen.

■

CHAPTER 30

■

solutions

698

MORE ABOUT POINTERS

SOLUTIONS

Exercise 1
Screen output: P WHITE E LUE K
Exercise 2
// ------------------------------------------------------// minivar.cpp
// Defines and tests the function min(), which
// computes and returns the minimum of positive integers.
// The function expects a variable number of arguments
// with unsigned int types.
// The last argument must be 0!
// ------------------------------------------------------#include 
unsigned int min( unsigned int first, ... )
{
unsigned int minarg, arg;
va_list argptr; // Pointer to optional arguments
if( first == 0)
return 0;
va_start( argptr, first);
minarg = first;
while( (arg = va_arg(argptr, unsigned int) ) != 0)
if( arg < minarg)
minarg = arg;
va_end (argptr);
return minarg;
}
// ----- A small function main() for testing--------------#include 
using namespace std;
int main()
{
cout << "\nThe minimum of : 34 47 19 22 58 "
<< "is: " << min(34, 47, 19, 22, 58, 0)
<< endl;
return 0;
}

SOLUTIONS

■

Exercise 3
// -- ---------------------------------------------------// sort_t.cpp
// Compares the performances of sorting algorithms
//
quick sort and selection sort
// For this purpose, two identical arrays are dynamically
// generated and initialized with random numbers.
// The times needed for sorting are displayed.
// ------------------------------------------------------#include 
#include 
#include 
#include 
using namespace std;
void isort(int *v, int lenv);
// For qsort():
extern "C" int intcmp(const void*, const void*);
main()
{
unsigned int i, size;
int *numbers1, *numbers2;
long time1, time2;
cout <<
<<
<<
<<

"\n
The performance of the sorting algorithms"
"\n
quick sort and selection sort"
"\n
is being compared.\n\n"
"\nHow many numbers are to be sorted?
";

cin >> size;
numbers1 = new int[size];
numbers2 = new int[size];
cout << "\nThere are "
<< size << " random numbers to be generated.\n";
srand((unsigned)time(NULL)); // Initialize the
// random number generator.
for(i = 0 ; i < size ; ++i)
numbers1[i] = numbers2[i] = rand(); // Random numbers
cout << "\nSorting starts! Please wait.\n";
time(&time1);
// Length of time
// for quick sort.
qsort(numbers1, size, sizeof(int), intcmp);
time(&time2);
cout

<< "\nTime taken by the quick sort algorithm: "
<<
time2 - time1 << " seconds.\n";

699

700

■

CHAPTER 30

MORE ABOUT POINTERS

cout << "\nI am sorting again. Please wait!\n";
time(&time1);
isort(numbers2, size);
time(&time2);

// Length of time
// for selection sort

cout << "\nTime taken by the insertion sort algorithm: "
<< time2 - time1 << " seconds.\n"
<< "\nOutput sorted numbers? (y/n)\n\n";
char c;
if( c ==
for( i
cout

cin >> c;
'Y' || c == 'y')
= 0 ; i < size ; ++i)
<< setw(12) << numbers1[i];

cout << endl;
return 0;
}

extern "C" int intcmp( const void *a, const void *b)
{
return (*(int*)a - *(int*)b);
}
// ------------------------------------------------------// isort() sorts an array of int values
//
using the selection sort algorithm.
void isort( int *a, int len)
{
register int *b, *minp;
int *last, help;
last = a + len - 1;
for( ; a <= last; ++a)
{
minp = a;

// Sort the array a of
// length len in ascending
// order

// Points to the last element

//
//
//
//
for( b = a+1; b <= last;
if( *b < *minp )
minp = b;

Search for the smallest
element starting at a.
minp points to the "current"
smallest array element.
++b)
// Search for the
// minimum.

help = *a, *a = *minp, *minp = help;
}
}

// Swap.

SOLUTIONS

■

Exercise 4
// -- ---------------------------------------------------// matrix.h : Represents dynamic matrices
// ------------------------------------------------------#ifndef _MATRIX_H_
#define _MATRIX_H_
#include 
#include 
using namespace std;
class Row
{
double *ro;
int size;
public:
Row( int s) { size = s; z = new double[s]; }
~Row(){ delete[]ro; }
double& operator[](int i)
{
if(i < 0 || i > size)
throw out_of_range("Row index: Out of Range\n");
return ro[i];
}
const double& operator[](int i) const
{
if(i < 0 || i > size)
throw out_of_range("Row index: Out of Range\n");
return ro[i];
}
};
class Matrix
{
private:
Row **mat;
int lines, cols;

// Pointer to array of rows
// Number of rows and columns

public:
Matrix(int ro , int co)
{
lines = ro; cols = co;
mat = new Row*[lines];
for(int i=0; i < lines; i++)
mat[i] = new Row(cols);
}
Matrix:: Matrix( int z, int s, double val);

701

702

■

CHAPTER 30

MORE ABOUT POINTERS

Matrix( const Matrix& );
~Matrix()
{ for(int i=0; i < lines; i++)
delete mat[i];
delete[] mat;
}
int getLines() const { return lines; }
int getCols() const { return cols; }
Row& operator[](int i)
{
if(i < 0 || i > cols)
throw out_of_range("Row index: Out of Range\n");
return *mat[i];
}
const Row& operator[](int i) const
{
if(i < 0 || i > cols)
throw out_of_range("Row index: Out of Range\n");
return *mat[i];
}
// Assignments:
Matrix& operator=( const Matrix& );
Matrix& operator+=( const Matrix& );
};
#endif
// -----------------------------------------------------// matrix.cpp : Defines methods of class Matrix
// -----------------------------------------------------#include "matrix.h"
Matrix:: Matrix( int ro, int co, double val)
{
lines = ro; cols = co;
mat = new Row*[lines];
// Array of pointers to
// arrays of rows
int i, j;
for(i=0; i < lines; i++)
// Arrays of rows:
{
mat[i] = new Row(cols);
// Allocate memory
for(j = 0; j < cols; ++j)
(*this)[i][j] = val;
// and copy values.
}
}

SOLUTIONS

Matrix:: Matrix( const Matrix& m)
{
lines = m.lines; cols = m.cols;
mat = new Row*[lines];
int i, j;
for(i=0; i < lines; i++)
{
mat[i] = new Row(cols);

// Rows and columns
// Array of pointers
// to arrays of rows
// Arrays of rows:
// To allocate
// storage

for( j = 0; j < cols; ++j)
(*this)[i][j] = m[i][j]; // and copy values.
}
}
Matrix& Matrix::operator=(const Matrix& m)
{
int i, j;
// Free "old" storage:
for(i=0; i < lines; i++)
delete mat[i];
delete[] mat;
lines = m.lines; cols = m.cols;

// Rows, columns

mat = new Row*[lines];

// Array of pointers
// to arrays of rows
// Array of rows:

for(i=0; i < lines; ++i)
{
mat[i] = new Row(cols);
for( j = 0; j < cols; ++j)
(*this)[i][j] = m[i][j];

// Allocate space

// and copy values.

}
return *this;
}
Matrix& Matrix::operator+=( const Matrix& m)
{
int i, j;
if( cols == m.cols && lines == m.lines)
for( i=0; i < lines; ++i)
for( j=0; j < cols; ++j)
(*this)[i][j] += m[i][j];
return *this;
}

■

703

704

■

CHAPTER 30

MORE ABOUT POINTERS

// -----------------------------------------------------// matrix_t.cpp : Tests dynamic matrices
// -----------------------------------------------------#include "matrix.h"
void display( Matrix& m);
// Output a matrix.
int main()
{
Matrix m(4,5);
try
{
int i,j;
for( i=0; i < m.getLines(); i++)
for( j=0; j < m.getCols(); j++)
m[i][j] = (double)i + j/ 100.0;
cout << "Matrix created" << endl;
display(m);
Matrix cop(m);
cout << "Copy generated." << endl;
display(cop);
cop += m;
cout << "Compute the sum:" << endl;
display(cop);
Matrix m1(4, 5, 0.0);
cout << "Initializing a matrix with 0:" << endl;
display(m1);
m = m1;
cout << "Matrix assigned:" << endl;
display(m);
}
catch(out_of_range& err)
{ cerr << err.what() << endl;
return 0;

exit(1);

}
void display( Matrix& m)
{
for(int i=0; i < m.getLines(); i++)
{
for(int j=0; j < m.getCols(); j++)
cout << m[i][j] << " ";
cout << endl;
}
cin.get();
}

}

chapter

31

Manipulating Bits
This chapter describes bitwise operators and how to use bit masks.The
applications included demonstrate calculations with parity bits,
conversion of lowercase and capital letters, and converting binary
numbers. Finally, the definition of bit-fields is introduced.

705

706

■

CHAPTER 31

■

MANIPULATING BITS

BITWISE OPERATORS

“True or False” table for bitwise operators

Bitwise AND

Result

Bitwise inclusive OR

Result

0 & 0

0

0 | 0

0

0 & 1

0

0 | 1

1

1 & 0

0

1 | 0

1

1 & 1

1

1 | 1

1

0 ˆ 0

0

~0

1

0 ˆ 1

1

~1

0

1 ˆ 0

1

1 ˆ 1

0

Examples
unsigned int a, b, c;

Bit pattern

a = 5;

0 0 . . . . . . . 0 0 1 0 1

b = 12;

0 0 . . . . . . . 0 1 1 0 0

c = a & b;

0 0 . . . . . . . 0 0 1 0 0

c = a | b;

0 0 . . . . . . . 0 1 1 0 1

c = a ˆ b;

0 0 . . . . . . . 0 1 0 0 1

c = ~a;

1 1 . . . . . . . 1 1 0 1 0

BITWISE OPERATORS

■

707

䊐 Bit Coding Data
In cases where conservative use of memory is imperative, data often need to be bit coded,
a technique to represent information as individual bits. Some examples of bit coded data
can be found in file access rights or the status-word of a stream.
To access bit coded data, you need to be able to read or modify individual bits. C++
has six bitwise operators to perform these tasks:
■

Logical bitwise operators
&

^
■

AND
exclusive OR

|

∼

inclusive OR
NOT

Bitwise shift operators
<< Left shift

>> Right shift

Operands for bitwise operators must have integral types. Operands belonging to float
or double types are invalid.
The boolean tables on the opposite page show the effect of the logical bitwise operators for individual bits. If a bit is set, that is, if it has a value of 1, it will be interpreted as
true. If the bit is not set, and thus has a value of 0, it will be interpreted as false. Examples for each bitwise operator follow.
The result type of a bitwise operation will be the integral type defined by the operand
type. If, for example, both operands are int types, the result will also be of an int type.

䊐 Arithmetic Type Conversions and Precedence
If the operands of a bitwise operator are of different types, normal arithmetic type conversion will occur. If one operand type is an int and the other a long, the int value
will be converted to long before the operation is performed.
The logical bitwise operators & or | should not be confused with the logical && and
|| operators. The latter do not affect individual bits but interpret the whole value of
their operands as boolean, returning a boolean value. The expression 1 && 2 returns
the value true, whereas 1 & 2 has the value 0.
The precedence of the bitwise NOT operator ∼ is high, since ∼ is a unary operator. As
you can see from the table of precedence in the appendix, both the binary operators &, ^,
and | have low precedence. However, their precedence is higher than that of the logical
operators && and ||.

708

■

CHAPTER 31

■

MANIPULATING BITS

BITWISE SHIFT OPERATORS

Right and left shift
unsigned int a, b;

Bit pattern

a = 12;

0 0 . . . . 0 0 0 0 1 1 0 0

b = a << 3;

0 0 . . . . 0 1 1 0 0 0 0 0

b = a >> 2;

0 0 . . . . 0 0 0 0 0 0 1 1

Using shift operators
// getbin_t.cpp: Defines the function getbin(), which
//
reads a binary number (ex. 0101... )
//
from the standard input, and returns
//
a value of type unsigned int.
// -------------------------------------------------#include 
using namespace std;
unsigned int getbin()
{
char c;
unsigned int val = 0;
while ( (c = cin.get()) == ' ' || c == '\t' )
;
// Ignore leading blanks and tabs
while( c == '0' || c == '1' )
// Read and convert
{
// the binary number
val = (val << 1) | (c - '0');
c = cin.get();
}
return val;
}

BITWISE SHIFT OPERATORS

■

709

䊐 Left and Right Shift
The shift operators << and >> shift the bit pattern of their left operand a certain number
of bit positions. The number of positions is defined by the right operand. The examples
opposite illustrate this point.
In the case of a left shift, 0 bits are padded. The bits dropped on the left are lost.

Example: short x = 0xFF00;
x = x << 4;

// Result: 0xF000

In the case of a right shift, 0 bits are padded from the left if the left operand is an
unsigned type or has a positive value. In all other cases, the compiler determines
whether to pad an expression with 0 bits (logical shift) or with the sign bit (arithmetic
shift), although an arithmetic shift normally occurs.

Example: short x = 0xFF00;
x = x >> 4;

// Result: 0xFFF0

To ensure portable source code, you should use right shifts for positive values only.

䊐 Integral Promotion
Integral promotion is performed for the operands of a shift operator, that is char is
extended to int. The result type in a shift operation is then the same as the type of the
left operand after integral promotion.
The result of a shift operation is unpredictable if the value of the right operand is negative or larger than the length of the left operand expressed in bits.

Example: char x = 0xFF;
x = x >> 9;

// undefined result

䊐 Applications
Shift operators allow you to perform efficient multiplication and division with 2n. Shifting a number n places left (or right) is equivalent to a multiplication (or division) by 2n.

Examples: unsigned res,number = 5;
res = number << 3;
res = number >> 1;

// 5 * 23 = 40
// 5 / 21 = 2

710

■

CHAPTER 31

■

MANIPULATING BITS

BIT MASKS

Bit positions

Bit pattern

0

1

Bit position

n

n-1

•

•

•

•

1

0

1

0

1

4

3

2

1

0

higher bits

lower bits

Example
#define MASK 0x20
char c = 'A';
c = c | MASK;

Current bit patterns:

0 1 0 0 0 0 0 1
| 0 0 1 0 0 0 0 0

'A' = 0x41
MASK = 0x20

0 1 1 0 0 0 0 1

'a' = 0x61

BIT MASKS

■

711

䊐 Deleting Bits
The bitwise AND operator is normally used to delete specific bits. A so-called mask is used
to determine which bits to delete.

Example: c = c & 0x7F;
In the mask 0x7F the seven least significant bits are set to 1, and all significant bits are
set to 0. This means that all the bits in c, with the exception of the least significant bits,
are deleted. These bits are left unchanged.
The variable c can be of any integral type. If the variable occupies more than one
byte, the significant bits in the mask, 0x7F, are padded with 0 bits when integral promotion is performed.

䊐 Setting and Inverting Bits
You can use the bitwise OR operator | to set specific bits. The example on the opposite
page shows how to change the case of a letter. In ASCII code, the only difference
between a lowercase and an uppercase letter is the fifth bit.
Finally, you can use the bitwise exclusive OR operator ^ to invert specific bits. Each
0-bit is set to 1 and each 1-bit is deleted if the corresponding bit in the mask has a
value of 1.

Example: c = c ^ 0xAA;
The bit pattern for 0xAA is 10101010. Every second bit in the least significant eight
bits of c is therefore inverted.
It is worthy of note that you can perform double inversion using the same mask to
restore the original bit pattern, that is, (x ^ MASK) ^ MASK restores the value x.
The following overview demonstrates the effect of a statement for an integral expression x and any given mask, MASK:
■
■
■

x & MASK
x | MASK
x ^ MASK

deletes all bits that have a value of 0 in MASK
sets all bits that have a value of 1 in MASK
inverts all bits that have a value of 0 in MASK.

The other bits are left unchanged.

712

■

CHAPTER 31

■

MANIPULATING BITS

USING BIT MASKS

Computing the parity of an integer
// parity_t.cpp: Defines the function parity(), which
//
computes the parity of an unsigned
//
value.
// Returns:
0, if the number of 1-bits is even,
//
1 in all other cases.
// --------------------------------------------------inline unsigned int bit0( unsigned int x )
{
return (x & 1);
}
int parity( unsigned int n)
{
unsigned int par = 0;
for( ; n != 0; n >>=1 )
par ^= bit0(n);
return (par);
}

USING BIT MASKS

■

713

䊐 Creating Your Own Masks
You can use the bitwise operators to create your own bit masks.

Example: x = x & ∼3;
In the bit pattern of 3, only the bits at positions 0 and 1 are set. The mask ∼3 therefore
contains a whole bunch of 1-bits and only the two least significant bits are 0. The above
expression would thus delete the two least significant bits in x.
The mask ∼3 is independent of the word length of the computer and is thus preferable
to the mask 0xFFFC.
The next example shows masks that address exactly one bit in a word. They are created by left shifting 1.

Examples: x = x | (1 << 6);
x = x & ∼(1 << 6);
The first expression sets the sixth bit in x. The same bit is then deleted, as only the sixth
bit in the mask ∼(1 << 6) has a value of 0.
Of course, you can also use masks such as (1 << n) where n is a variable containing
the bit position.

Example: int setBit(int x, unsigned int n)
{
if( n < sizeof(int) )
return( x & (1 << n);
}

䊐 Bitwise Operators in Compound Assignments
The binary bitwise operators &, |, ^, <<, and >> can be used in compound assignments.

Examples: x >>= 1;
x ^= 1;

Both statements are equivalent to
x = x >> 1;
x = x ^ 1

The function parity() shown opposite includes compound assignments with bitwise
operators. Parity bit computation is used to perform error recognition in data communications.

714

■

CHAPTER 31

■

MANIPULATING BITS

BIT-FIELDS

Header of ATM cells

Byte
1

General Flow
Control

Virtual Path
Identifier

2

Virtual Path
Identifier

Virtual Channel
Identifier

3

4

5

Virtual Channel
Identifier
Virtual Channel
Identifier

Payload
Type

CLP

Header Error
Control

General Flow Control
Virtual Path/Channel Identifier
Payload Type

Controls the data stream
Address of the virtual path/channel
Distinguish between payload and control
data
Mark cells with high priority
Check sum for header

CLP (Cell Loss Priority)
Header Error Control

Representing an ATM cell
struct ATM_Cell
{
unsigned GFC : 4;
unsigned VPI : 8;
unsigned VCI : 16;
unsigned PT : 3;
unsigned CLP : 1
unsigned HEC : 8;
char payload[48];
};

//
//
//
//
//
//
//

General Flow Control
Virtual Path Identifier
Virtual Channel Identifier
Payload Type
Cell Loss Priority
Header Error Control
Payload

BIT-FIELDS

■

715

C++ lets you divide a computer word into bit-fields and give the bit-fields a name. Bitfields offer major advantages; their uncluttered structure makes them preferable and less
error prone than using masks and bitwise operators to manipulate individual bits.

䊐 Defining Bit-Fields
Bit-fields are defined as data members of a class. Each bit-field is of unsigned int type
with an optional name and width. The width is defined as the number of bits the bit-field
occupies in a computer word and is separated from the bit-field name by a colon.

Example: struct { unsigned bit0_4 : 5;
unsigned
unsigned bit15

: 10;
: 1; } word;

The member word.bit0_4 designates the 5 least significant bits in a computer word
and can store values in the range 0 to 31. The second data member does not have a
name and is used to create a gap of 10 bits. The member word.bit15 contains the
value at bit position 15.
You cannot reference nameless bit-fields. They are used to align the subsequent bitfields at specific bit positions.
The width of a bit-field cannot be greater than that of the computer word. The width
0 has a special significance; the subsequent bit-field is positioned on the next word
boundary, that is, it begins with the next computer word. If a bit-field will not fit in a
computer word, the following bit-field is also positioned on the next word boundary.
There are several special cases you need to consider when dealing with bit-fields:
■

■

you cannot use the address operator for bit-fields. You cannot create arrays of bitfields. Neither restriction applies to a class containing bit-field members, however.
the order bit-fields are positioned in depends on the machine being used. Some
computer architectures position bit-fields in reverse order. This is true of DEC
Alpha workstations, for example.

䊐 The Sample Program Opposite
The opposite page shows a class designed to represent ATM cells. Cells are used for data
transportation in ATM (Asynchronous Transfer Mode) networks. Each cell comprises a 5
byte header with addresses and a checksum for error checking and 48 byte data section
or payload. The header shown here is used to connect a computer to the network in the
User Network Interface.

■

CHAPTER 31

■

exercise s

716

MANIPULATING BITS

EXERCISES

Sample screen output for exercise 1
******

BITWISE OPERATORS

******

Please enter two integers.
1st Number
2nd Number
The
The
The
The
The

bit
bit
bit
bit
bit

--> 57
--> -3

pattern
pattern
pattern
pattern
pattern

of
of
of
of
of

57 = x :
-3 = y :
x & y :
x | y :
x ^ y :

0000
1111
0000
1111
1111

0000
1111
0000
1111
1111

0011
1111
0011
1111
1100

1001
1101
1001
1101
0100

How many bit positions is x to be shifted?
Count --> 4
The bit pattern of x << 4 : 0000 0011 1001 0000
The bit pattern of x >> 4 : 0000 0000 0000 0011
Repeat (y/n)?

EXERCISES

■

717

Exercise 1
a. Write the function putBits() that outputs the bit pattern of a number
as an unsigned int type. Only the 16 least significant bits are to be
output no matter what the size of the computer word.The number is
passed as an argument to the function, which has no return value.
b. Write a tutorial to demonstrate the effect of bitwise operators. First
read two decimal integers from the keyboard and store them in the variables x and y.Then use the function putBits() to output the bit patterns of x, x&y, x | y, x ^ y, and ∼x.
To demonstrate the shift operators, shift the value of x a given number of
bit positions right and left. Read the number of bit positions from keyboard input. Use the value 1 in case of invalid input.
The opposite page shows sample output from the program.

Exercise 2
Your task is to encrypt data to prevent spying during data communications.The
sender uses a filter to encrypt the data in question, and the receiver uses the
same filter to decrypt the transmission.
a. Define the function swapBits() that swaps two bits in an int value.The
int value and the positions of the bits to be swapped are passed as arguments to the function.The return value is the new int value. If one of
the positions passed to the function is invalid, the int value should be
returned unchanged.
b. Write a filter that swaps the bits at bit positions 5 and 6, 0 and 4, and 1
and 3 in all characters except control characters (defined as ASCII Code
>= 32).
Test the filter by writing the encrypted output to a file and then using the same
filter to output the new file.The output must comprise the original unencrypted
data.

■

CHAPTER 31

■

solutions

718

MANIPULATING BITS

SOLUTIONS

Exercise 1
// -----------------------------------------------------// bits_t.cpp
// Demonstrates bitwise operators.
// -----------------------------------------------------#include 
#include 
using namespace std;
void putbits( unsigned int n);
int main()
{
int x, y, count;
char yn;

// Prototype of putbits()

// Learning bitwise operations

do
{
cout << "\n\n

******

BITWISE OPERATIONS

******\n";

cout << "\nPlease enter two integers.\n\n"
<< "1st number --> ";
cin >> x;
cout << "2nd number --> ";
cin >> y;
cout << "\nThe bit pattern of "
<< setw(6) << x << " = x :
putbits(x);

";

cout << "\nThe bit pattern of "
<< setw(6) << y << " = y :
putbits(y);

";

cout << "\nThe bit pattern of
putbits(x&y);

x & y

:

";

cout << "\nThe bit pattern of
putbits(x|y);

x | y

:

";

cout << "\nThe bit pattern of
putbits(x^y);

x ^ y

:

";

cout << "\n\nHow many bit positions"
" is x to be shifted?"
<< "\nNumber --> ";
cin >> count;

SOLUTIONS

■

if( count < 0 || count > 15)
{
cout << "Invalid input!"
<< " Shifting by one bit position.\n";
count = 1;
}
cout << "\nThe bit pattern of
x << "
<< setw(2) << count << " :
";
putbits( x << count);
cout << "\nThe bit pattern of x >> "
<< setw(2) << count << " :
";
putbits( x >> count);
cout << "\nRepeat (y/n)? ";
cin >> yn;
while( (yn | 0x20) != 'y' &&
;

yn != 'n')

}while( yn == 'y');
return 0;
}
// -----------------------------------------------------// Output the bit pattern of n (only the 16 lower bits).
void putbits( unsigned int n )
{
int i;
for( i = 15; i >= 0 ; --i)
{
cout << (char)( ((n>>i) & 1) + '0');
// i-th bit
if( i % 4 == 0 && i > 0)
// and after 4 bits
cout << ' ';
// one blank.
}
}

Exercise 2
// ------------------------------------------------------// hide_t.cpp: Filter to encrypt data.
//
Swap bits in bit positions 5 and 6,
//
0 and 4, 1 and 3 for all characters
//
except control characters.
//
Modules: hide_t.cpp, swapbits.cpp
//
// Call:
hide_t [ < sourcefile ] [ > destfile ]
// -------------------------------------------------------

719

720

■

CHAPTER 31

MANIPULATING BITS

#include 
using namespace std;
int swapbits( int ch, int bitnr1, int bitnr2); // Prototype
int main()
{
int c;
while( (c = cin.get()) != EOF)
{
if( c >= 32)
{
c = swapbits(c, 5, 6);
c = swapbits(c, 0, 4);
c = swapbits(c, 1, 3);
}
cout << c;
}
return 0;
}
//
//
//
//
//
//

// Encrypt data

// Control character?
// Swap bits

------------------------------------------------------swapbits.cpp: The function swapbits() swaps two bits
within an integer.
Arguments:
The integer and two bit positions.
Returns:
The new value.
-------------------------------------------------------

int swapbits( int x, int bitnr1, int bitnr2)
{
// To swap two bits in x.
int newx, mask1, mask2;
int msb = 8 * sizeof(int) - 1; // Highest bit position
if( bitnr1 < 0 || bitnr1 > msb ||
bitnr2 < 0 || bitnr2 > msb)
return x;
// Return, if bit position is invalid
mask1 = (1 << bitnr1);
mask2 = (1 << bitnr2);

newx = x & ~(mask1 | mask2);

// Delete both bits

if( x & mask1 )
if( x & mask2 )

// Swap bits.

return( newx);
}

// Shift 1 to position bitnr1
// Shift 1 to position bitnr2

newx |= mask2;
newx |= mask1;

chapter

32

Templates
Templates allow you to construct both functions and classes based on
types that have not yet been stated.Thus, templates are a powerful tool
for automating program code generation.
This chapter describes how to define and use function and class
templates. In addition, special options, such as default arguments,
specialization, and explicit instantiation are discussed.

721

722

■

CHAPTER 32

■

TEMPLATES

FUNCTION AND CLASS TEMPLATES

Template and instantiation

Instantiation for
Type long
Template

Type int
Type char

FUNCTION AND CLASS TEMPLATES

■

723

䊐 Motivation
As a programmer you will often be faced with implementing multiple versions of similar
functions and classes, which are needed for various types.
A class used to represent an array of int values is very similar to a class representing
an array of double values, for example. The implementation varies only in the type of
elements that you need to represent. Operations performed with elements, such as search
and sort algorithms, must be defined separately for each type.
C++ allows you to define templates—parameterized families of related functions or
classes:
■
■

a function template defines a group of statements for a function using a parameter
instead of a concrete type
a class template specifies a class definition using a parameter instead of a concrete
type.

A class template can provide a generic definition that can be used to represent various
types of arrays, for example. During instantiation, that is, when a concrete type is defined,
an individual class is created based on the template definition.

䊐 Advantages of Templates
Templates are powerful programming tools.
■
■
■

A template need only be coded once. Individual functions or classes are automatically generated when needed.
A template offers a uniform solution for similar problems allowing type-independent code to be tested early in the development phase.
Errors caused by multiple encoding are avoided.

䊐 Templates in the Standard Library
The C++ standard library contains numerous class template definitions, such as the
stream classes for input and output, string, and container classes. The classes string,
istream, ostream, iostream, and so on are instantiations for the char type.
The standard library also includes an algorithm library, which comprises many search
and sort algorithms. The various algorithms are implemented as global function templates and can be used for any set of objects.

724

■

CHAPTER 32

■

TEMPLATES

DEFINING TEMPLATES

Class template Stack
// stack.h : The class template Stack with
//
methods push() and pop().
//----------------------------------------------------template
class Stack
{
private:
T* basePtr;
// Pointer to array
int tip;
// Stack tip
int max;
// Maximum number of elements
public:
Stack(int n){ basePtr = new T[n]; max = n; tip = 0;}
Stack( const Stack&);
~Stack(){ delete[] basePtr; }
Stack& operator=( const Stack& );
bool empty(){ return (tip == 0); }
bool push( const T& x);
bool pop(T& x);
};
template
bool Stack::push( const T& x)
{
if(tip < max - 1)
// If there is enough space
{
basePtr[tip++] = x; return true;
}
else return false;
}
template
bool Stack::pop( T& x)
{
if(tip > 0)
// If the stack is not empty
{
x = basePtr{--tip];
return true;
}
else return false;
}

DEFINING TEMPLATES

■

725

䊐 Defining Function Templates
The definition of a template is always prefixed by
template

where the parameter T is a type name used in the definition that follows. Although you
must state the class keyword, T can be any given type, such as an int or double.

Example: template 
void exchange(T& x, T&y)
{
T help(x); x = y; y = help;
}

This defines the function template exchange(). The parameter T represents the type
of variables, which are to interchange. The name T is common but not mandatory.

䊐 Defining Class Templates
Example: template 
class Demo
{
U elem;
};

. . .

// etc.

This defines the class template Demo. Both U and Demo are treated like normal
types in the class definition. You simply need to state the name of the template, Demo,
within the class scope.
The methods of a class template are also parameterized via the future type. Each
method in a class template is thus a function template. If the definition is external to the
class template, function template syntax is used. The method name is prefixed by the
class template type and the scope resolution operator.
The example on the opposite page illustrates this point by defining a stack template.
A stack is managed according to the last-in-first-out principle, lifo-principle for short; the
last element to be “pushed” onto the stack is the first to be removed, or “popped,” from
the stack.
The methods of a class template are normally defined in the same header file. This
ensures that the definition will be visible to the compiler, since it requires the definition
to generate machine code for concrete template arguments.

726

■

CHAPTER 32

■

TEMPLATES

TEMPLATE INSTANTIATION

Sample program
// stack_t.cpp: Testing a stack
// ---------------------------------------------------#include 
#include 
using namespace std;
#include "stack.h"
typedef Stack USTACK;

// Stack for elements
// of type unsigned.

void fill( USTACK& stk );
void clear( USTACK& stk );
int main()
{
USTACK ustk(256);
// Create and fill
fill( ustk);
// the original stack.
USTACK ostk(ustk);
// Copy.
cout << "The copy: " << endl;
clear( ostk);
// Output and clear the copy.
cout << "The original: " << endl;
clear( ustk );
// Output, clear the original.
return 0;
}
void fill( USTACK& stk )
{
unsigned x;
cout << "Enter positive integers (quit with 0):\n";
while( cin >> x && x != 0 )
if( !stk.push(x) )
{
cerr << "Stack is full!"; break;
}
}
void clear( USTACK& stk )
{
if(stk.empty())
cerr << "Stack is empty!" << endl;
else
{
unsigned x;
while( stk.pop(x))
cout << setw(8) << x << " ";
cout << endl;
}
}

TEMPLATE INSTANTIATION

■

727

Defining a template creates neither a concrete function nor a class. The machine code
for functions or methods is not generated until instantiation.

䊐 Instantiating Template Functions
A template function is instantiated when it is first called. The compiler determines the
parameter type of T by the function arguments.

Example: short a = 1, b = 7;
exchange( a, b );

The template is first used to generate the exchange() function machine code for the
short type. The template functions can be called after this step.
This allows you to generate an exchange() template function for any type. Given
that x and y are two double variables, the following statement

Example: exchange( x, y );
creates a second template function for the double type.

䊐 Instantiation of Template Classes
The instantiation of a template class is performed implicitly when the class is used for the
first time, for example, when an object of the template class is defined.

Example: Stack istack(256);

// implicit

This statement first creates the template class Stack, generating the machine
code of all methods for the int type. After this step has been completed, an istack
object of the Stack type can be constructed.
If a further template class, such as Stack is created, the machine code generated for the methods in this template class will be different from the machine code of
the Stack methods.
In other words, developing templates will not reduce the amount of machine code
required for a program. However, it does spare the programmer’s extra work required to
develop multiple versions of functions and classes.
Templates are double checked for errors by the compiler—once when the template
definition is compiled and again during instantiation. The first check recognizes errors
that are independent of the template parameters. Errors in parameterization cannot be
detected until instantiation if, for example, an operator for the template argument type
has not been defined.

728

■

CHAPTER 32

■

TEMPLATES

TEMPLATE PARAMETERS

The Stack template with two template parameters
// stackn.h: Class Template Stack
// --------------------------------------template 
class Stack
{
private:
T
arr[n];
// Array
int tip;
// Tip of stack
int max;
// Maximum number of elements
public:
Stack(){ max = n; tip = 0; };
bool empty(){ return (tip == 0); }
bool push( const T& x);
bool pop(T& x);
};
template
bool Stack::push( const T& x)
{
if(tip < max - 1)
{
arr[tip++] = x; return true;
}
else return false;
}
template
bool Stack::pop(T& x )
{
if(tip > 0)
{
x = arr{--tip]; return
}
else return false;
}

true;

TEMPLATE PARAMETERS

■

729

䊐 Multiple Template Parameters
You can also define templates with multiple parameters like the following class template,

Example: template 
class Demo
{
// . . .

};

which has two parameters U and V. A class Demo is defined for each pair of U, V
types.
A template parameter need not always be a type name. Normal function parameters
are also permissible, particularly pointers and references.

Example: template
class Stack{ . . . };

This defines the class template Stack that is parameterized with the type T and
an integer n.
The example on the opposite page uses the parameter n to specify the size of an array
used to represent a stack. The major advantage is that the number of array elements is
already known when an object of the template class is instantiated. Objects can then be
created without allocating dynamic storage.
This simplifies the definition of the stack template. The copy constructor, the assignment operator, and the destructor no longer need to be defined!

䊐 Restrictions
Two restrictions apply to template parameters other than type parameters:
■
■

they cannot be modified
they cannot be floating-point types.

The following expression would thus be invalid in the definition opposite:

Example: ++n; // Error: changing template parameter
Even though double type template parameters are not permissible,

Example: template

// Error!

class Demo { . . . };

pointers and references to floating-point types are:

Example: template
class Demo { . . . };

730

■

CHAPTER 32

■

TEMPLATES

TEMPLATE ARGUMENTS

Sample program
// mini_t.cpp: Passing arguments to
//
function templates
// ---------------------------------------------#include 
using namespace std;
template 
T min( T x, T y)
{
return( (x < y) ? x : y);
}
int main()
{
short x = 10, y = 2;
cout << "x = " << x << " y = " << y << endl;
cout << "The smaller value is: "
<< min(x, y) << endl;
// Call is ok.
double z1 = 2.2;
float z2 = 1.1F;
cout << "\nThe smaller value is: "
<< min(z1, z2) << endl;
// Not ok!
double z3 = 1.1;
cout << "\nz1 = " << z1
<< " z3 = " << z3 << endl;
cout << "The smaller value is: "
<< min(z1, z3) << endl;
// Call is ok.
return 0;
}

TEMPLATE ARGUMENTS

■

731

䊐 Passing Arguments
A template is instantiated when a template argument is passed to it. The argument types
must exactly match to the types of the template parameters.
Not even implicit type conversions, such as float to double, are performed. In the
case of the template function min() this means that both arguments must be of the
same type. The following call

Example: float x = 1.1; double y = 7.7;
min ( x , y );

would lead to an error message, since the template function cannot be defined by the

Prototype: void min( float , double );

䊐 Restrictions
There are several restrictions for template arguments other than type names:
■
■
■

if the template parameter is a reference, only a global or static object can be
passed as a template argument
if the template parameter is a pointer, only the address of an object or a function
with global scope can be stated
if the template parameter is neither a reference nor a pointer, only constant
expressions can be used as template arguments.

Example: int cnt = 256;

// Error:
typedef Stack ShortStack;

Since only an int constant is permitted as a template argument, this statement provokes
an error.
Strings, such as "Oktoberfest," are also invalid as template arguments, as their
scope is static and not global.

Example: template class Demo{...};
Only globally defined strings can be used for instantiation, for example
char str[] = "Oktoberfest";
Demo income;

// global
// ok

732

■

CHAPTER 32

■

TEMPLATES

SPECIALIZATION

Function template min()
template 
T min( T x, T y)
{
return( (x < y) ? x : y)
}

Specializing the function template for C strings
#include 
const char* min( const char* s1, const char* s2 )
{
return( (strcmp(s1, s2) < 0 ) ? s1: s2 );
}

䊐 ANSI specialization
The ANSI standard does not differ between template functions and “normal” functions.
The definition of a function template and a function with the same name, which can be
generated by the function template, causes the compiler to output an error message (ex.
“duplicate definition ...”).
That is why the ANSI standard provides its own syntax for defining specializations:
#include 
template<>
const char* min( const char* s1, const char* s2 )
{
return( (strcmp(s1, s2) < 0 ) ? s1: s2 );
}

SPECIALIZATION

■

733

䊐 Motivation
A template function can only be instantiated if all the statements contained in the function can be executed. If you call the template function exchange() with two objects of
the same class, the copy constructor and the assignment must be defined for this class.
More specifically, all operators used in a template function must be defined for the
current argument type. Thus, the function template min(), which determines the lesser
of two arguments, can only be instantiated if the operator < is defined for the argument
type.
Besides non-executable instructions there are other reasons to prevent a function
template being instantiated for a particular type:
■
■

the generic approach defined by the template does not return any useful results
for a given type
there are more efficient approaches for some types.

The statement

Example: minStr = min(, "VIVIAN", "vivian" );
only returns the lower of the two addresses at which the C strings are stored.

䊐 Defining Specialization
In cases like this, it makes sense to specialize the template function definition. To do so,
you use a function with a separate definition to overload the template function. This
technique is demonstrated on the opposite page using the function template min(),
where a specialization has been defined for the char* type, both for older and more
modern compilers that support the current ANSI standard.
If a template function is replaced by a specialization, the appropriate version must be
executed when a call to the function is made. The order the compiler looks up a function
guarantees that if both a function template and a specialization are defined for a specific
type, the specialization will be called.
This also applies to the methods of a class template, which are function templates, of
course. More specifically, a template class can only be created if all the methods in the
appropriate function template can be instantiated without error.

734

■

CHAPTER 32

■

TEMPLATES

DEFAULT ARGUMENTS OF TEMPLATES

A class template representing quadratic matrices
// quadMat.h: Defines the template QuadMatrix
//
to represent quadratic matrices
// ----------------------------------------------------#include 
#include 
using namespace std;
template 
class QuadMatrix
{
private:
T mat[cnt][cnt];
public:
int dim() const{ return cnt; }
T* operator[](int line) throw(out_of_range)
{
if( line < 0 || line >= cnt)
throw out_of_range("Matrix: Index out of range");
else
return mat[line];
}
const T*

operator[](int line) const
throw(out_of_range)

{
if( line < 0 || line >= cnt)
throw out_of_range("Matrix: Index out of range");
else
return mat[line];
}
friend QuadMatrix& operator+(const QuadMatrix&,
const QuadMatrix&);
// etc.
};

DEFAULT ARGUMENTS OF TEMPLATES

■

735

䊐 Setting Defaults
You can define default arguments for template parameters, just as for function parameters. If an argument required to instantiate a template is missing, the default value is then
used.
You can specify default values in the template definition or when you declare a template in a module.

䊐 The Class Template QuadMatrix
The class template defined opposite, QuadMatrix, represents quadratic matrices. The subscript operator is overloaded to allow you to access a matrix element
m[i][j] in a given matrix m. If the line index i is outside of the valid range, a standard
out_of_range type exception is thrown.
The default values are chosen to create a matrix m for int values with 10 rows and 10
columns following this definition:

Example: typedef QuadMatrix < > IntMat;
IntMat m;

You cannot omit the angled brackets since the QuadMatrix type does not exist.
QuadMatrix is merely the name of a template.
The following definition

Example: typedef QuadMatrix DoubleMat;
DoubleMat dm;

defines a matrix dm of double values with 10 rows and 10 columns.

䊐 Rules
The same rules apply to the default arguments of templates as to the default arguments of
functions:
■
■

if you declare a default argument for at least one parameter, you must define
default values for all the remaining parameters
if a template argument for which a default argument was declared is omitted during instantiation, all the remaining template arguments must be omitted.

736

■

CHAPTER 32

■

TEMPLATES

EXPLICIT INSTANTIATION

Sample program for the class template QuadMatrix
// expIns_t.cpp: Tests explicit instantiation
// ----------------------------------------------------#include 
#include 
using namespace std;
#include "quadMat.h"
// Explicit Instantiation:
template class QuadMatrix;
int main()
{
QuadMatrix m;
try
{
for(int k=0; k < m.dim(); k++)
{
for( int l = 0; l < m.dim(); l++)
{
m[k][l] = k*l;
cout << setw(2) << m[k][l] << " ";
}
cout << endl;
}
}
catch(out_of_range& err )
{
cerr << err.what() << endl;
}
return 0;
}

EXPLICIT INSTANTIATION

■

737

In addition to implicit instantiation of templates, which occurs when a template function is called, for example, explicit instantiation is also possible. This is important when
you design libraries that contain template functions and classes for application programs.

䊐 Syntax
Explicit instantiation can be achieved by the following

Syntax:

template declaration;

where declaration contains the name of the template and the template arguments.
Explicit instantiation for the class template Stack would be performed as follows:

Example: template class Stack;
This declaration creates the template class Stack with a maximum of 50 long double type elements.
Function templates can also be instantiated explicitly.

Example: template short min( short x, short y);
This creates a template function for the short type from the function template min().

䊐 ANSI Instantiation
The ANSI standard provides an additional technique for the explicit instantiation of
function templates. Template arguments are stated in the angled brackets that follow the
function name, when the function is first called.

Example: min(x, y);
In this case, a template function min() for the long type is generated. This advanced
syntax for function templates is not supported by all C++ compilers, however.
Explicit instantiation of function templates extends their possible usage:
■

■

function templates can be parameterized by types that cannot be derived from the
function arguments—more specifically, function templates can be defined without function parameters
function templates can be defined with function parameters that are not template
parameters themselves.

■

CHAPTER 32

■

exercise s

738

TEMPLATES

EXERCISES

Interpolation search
The elements of a numerical array are assumed to be unique and sorted in
ascending order.
The given value is compared with the array element at the position where the
value is “expected” to be. For example, if the value searched for is two-thirds of
the way from the lowest to the highest subarray element, the probe would be
made two-thirds from the lowest to the highest index of the subarray. If the
required value is lesser than that of the array element found at the expected
position, the search is continued in the left part of the subarray, just like a binary
search. Otherwise the search continues in the right part of the subarray.
The “expected” position exp in an array v can be calculated as follows: If key
is the required value, begin is the lowest, and end is the highest index of the
corresponding subarray, the following applies:
double temp = (double)(key-vp[begin]);
temp /= (vp[end]-vp[begin]);
temp = temp * (end - begin) + 0.5;
exp
= begin + (int)temp;

Insertion sort algorithm
The following technique is used to divide the array into a left, sorted part and a
right, unsorted part:
Each subsequent element in the unsorted part of the array is selected and
taken out from the array.As long as a greater array element can be found
starting from the end of the left subarray, the element is shifted up by one
position. If a smaller array element is found, the selected element is inserted at
the vacant position.

✓

NOTE
The left, sorted part originally consists of only one element, the first array
element.

Graphic:
30

50

70

40

20

↑ to take out
30

50

70
70 > 40?

30

50

20
yes
70

20

EXERCISES

■

739

Exercise 1
■ Define a function template interpolSearch() that looks up a given element in a sorted, numeric array.The array elements are of the same type
as the template parameter T.
The function template has three parameters—the value searched for of
type T, a pointer to the first array element, and the number of array elements.
The function template returns the index of the first element in the array
that corresponds to the searched for value, or –1 if the value cannot be
found in the array.
Implement the function template. Use the technique described opposite
as your algorithm for the interpolation search. Store the function template definition in the header file search.h.
■ Define a function template, insertionSort(), which sorts a numeric
array in ascending order.The array elements are of the type of the template parameter T.
The function template has two parameters—a pointer to the first array
element, and the number of array elements.There is no return value.
■ Define a function template display() to display a numeric array on
screen.
The function template has two parameters—a pointer to the first array
element and the number of array elements.There is no return value.
■ Use the function templates interpolSearch(), insertionSort(), and
display() to define template functions for double and short types.To
do so, define an array of double values and an array with short values.
Write a main function that creates and calls each template function
insertionSort() for the int and double types.Then display the
sorted arrays.
Add a call to each template function search() in your main function.
Call search() passing values that exist and do not exist in the array.

740

■

CHAPTER 32

TEMPLATES

Exercises
The class FloatArr (as defined in Chap. 28, Ex. 1)
class FloatArr
{
private:
float* arrPtr;
int max;
int cnt;

// Without conversion functions

//
//
//
//

Dynamic member
Maximum number, without having
to reallocate storage.
Current number of elements

void expand( int new Size);

// Function to help
// enlarge the array.

public:
FloatArr( int n = 256 );
FloatArr( int n, float val);
FloatArr(const FloatArr& src);
~FloatArr();
FloatArr& operator=( const FloatArr& );
int length() const { return cnt; }
float& operator[](int i) throw(BadIndex);
float operator[](int i) const throw(BadIndex);
void append( float val);
void append( const FloatArr& v);
FloatArr& operator+=( float val)
{
append( val);
return *this;
}
FloatArr& operator+=( const FloatArr& v)
{
append(v);
return *this;
}
void insert( float val, int pos) throw(BadIndex);
void insert( const FloatArr& v, int pos )
throw(BadIndex);
void remove(int pos) throw(BadIndex);
friend ostream& operator<<( ostream& os,
const FloatArr& v);
};

EXERCISES

■

741

Exercise 2
Define a class template Array to represent an array with a maximum of
n elements of type T.Attempting to address an array element with an invalid
index should lead to an exception of the BadIndex (defined previously) error
class being thrown. If there is no space left to insert an element in the array, an
exception of the OutOfRange type should be thrown.
■

■

■

■

First define the error class OutOfRange without any data members. Use
the error class BadIndex, which was defined in Exercise 1 of Chapter 28.
Change the existing FloatArr class into a class template, Array.
Use 255 as the default value for the parameter n of the class template.
This allocates memory for the array statically. Now define a default constructor and a constructor that initializes a given number of array elements with a given value.You do not need to define a copy constructor, a
destructor, or an assignment operator.
As access methods define the size() and the length() methods, the
size() method returns the maximum number of array elements, that
is, n; the length() method returns the current number of elements in
the array.
Also define methods for inserting and deleting elements like those
defined for the FloatArr class.The methods have a void return type and
can throw exceptions of type BadIndex and/or OutOfRange.
Additionally overload the index and shift operators <<.The subscript
operator throws BadIndex type exceptions.
Test the class template Array for double and int types first.
Define arrays of the appropriate types, then insert and delete elements in
the arrays. Output all the elements of the array.
Modify the test program by adding an array for objects of a class type.
Use the DayTime class from Exercise 1, Chapter 19 for this purpose.
Test the array template by defining an array with 5 DayTime class objects
and inserting a few objects.Then display all the objects on screen.

■

CHAPTER 32

■

solutions

742

TEMPLATES

SOLUTIONS

Exercise 1
// -----------------------------------------------------// interpol.cpp : Template function interpolSearch()
// -----------------------------------------------------#include 
using namespace std;
template 
long interpolSearch(const T& key, T* vp, int len)
{
int expect, begin = 0, end = len - 1;
double temp;
if( end < 0
// Array is empty or
|| key > vp[end]
// or key is out of
|| key < vp[begin] )
// range
return -1;
while( begin <= end )
{
if(key > vp[end] || key < vp[begin] ) // Key is not
return -1;
// in range.
temp = (double)(key - vp[begin])
/ (vp[end]-vp[begin]);
temp = temp * (end - begin) +0.5;
expect = begin + (int)temp;
if( vp[expect] == key )
// Key found?
return expect;
if( vp[expect] > key)
end = expect - 1;
else begin = expect+1;
}
return -1;
}
template 
void insertionSort( T* vp, int len)
{
T temp;
for( int i=0; i < len; i++)
{
temp = vp[i];
// Take element out.
int j;
// Shift greater elements up:
for( j = i-1; j >= 0 && vp[j] > temp; j--)
vp[j+1] = vp[j];
vp[j+1] = temp;
// Insert.
}
}

SOLUTIONS

template 
void display(T* vp, int len)
{
cout << "\n\nThe array: " << endl;
for(int i = 0; i < len; i++)
{
cout << vp[i] << " ";
if( (i+1)%10 == 0)
cout << endl;
}
cout << endl; cin.get();
}
// Two arrays for testing:
short sv[5] = { 7, 9, 2, 4, 1};
double dv[5] = { 5.7, 3.5, 2.1, 9.4, 4.3 };
int main()
{
cout << "\nInstantiation for type short: " << endl;
display(sv, 5);
insertionSort(sv, 5);
cout << "\nAfter sorting: ";
display(sv, 5);
short key;
cout << "\nArray element? "; cin >> key; cin.sync();
int pos = interpolSearch(key, sv, 5);
if( pos != -1)
cout << "\nfound!" << endl, cin.get();
else
cout << "\nnot found!" << endl, cin.get();
// ------------------------------------------------cout << "\nInstantiation for type double: " << endl;
display(dv, 5);
insertionSort(dv, 5);
cout << "\nAfter sorting: ";
display(dv, 5);
double dkey;
cout << "\nArray element? "; cin >> dkey; cin.sync();
pos = interpolSearch(dkey, dv, 5);
if( pos != -1)
cout << "\nfound!" << endl, cin.get();
else
cout << "\nnot found!" << endl, cin.get();
return 0;
}

■

743

744

■

CHAPTER 32

TEMPLATES

Exercise 2
// -----------------------------------------------------// array.h
// Use of class templates to represent arrays.
// -----------------------------------------------------#ifndef _ARRAY_H_
#define _ARRAY_H_
#include 
#include 
using namespace std;
class BadIndex
{
private:
int index;
public:
BadIndex(int i):index(i){}
int getBadIndex() const { return index; }
};
class OutOfRange {

/* Without data members*/ };

template 
class Array
{
private:
T
arr[n];
// The array
int cnt;
// Current number of elements
public:
Array( ){ cnt = 0;}
Array(int n, const T& val );
int
int

length() const { return cnt; }
size()
const { return n; }

T& operator[](int i) throw(BadIndex)
{
if( i < 0 || i >= cnt ) throw BadIndex(i);
return arr[i];
}
const T& operator[](int i) const throw(BadIndex)
{
if( i < 0 || i >= cnt ) throw BadIndex(i);
return arr[i];
}

SOLUTIONS

■

745

Array& operator+=( float val) throw(OutOfRange)
{
append( val);
return *this;
}
Array& operator+=(const Array& v)
{
append(v);
return *this;
}

throw(OutOfRange)

void append( T val) throw(OutOfRange);
void append( const Array& v) throw(OutOfRange);
void insert( T val, int pos)
throw(BadIndex, OutOfRange);
void insert( const Array& v, int pos )
throw(BadIndex, OutOfRange);
void remove(int pos) throw(BadIndex);
};
template 
Array::Array(int m, const T& val )
{
cnt = m;
for(int i=0; i < cnt; i++ )
arr[i] = val;
}
template 
void Array::append( T val) throw(OutOfRange)
{
if( cnt < n)
arr[cnt++] = val;
else
throw OutOfRange();
}
template 
void Array::append( const Array& v) throw(OutOfRange)
{
if( cnt + v.cnt > n)
// Not enough space.
throw OutOfRange();
int count = v.cnt;
for( int i=0; i < count; ++i)
arr[cnt++] = v.arr[i];
}

// Necessary if
// v == *this

746

■

CHAPTER 32

TEMPLATES

template 
void Array::insert( T val, int pos)
throw(BadIndex, OutOfRange)
{
insert( Array(1,val), pos);
}
template 
void Array::insert( const Array& v, int pos )
throw(BadIndex, OutOfRange)
{
if( pos < 0 || pos >= cnt)
throw BadIndex();
// Invalid position.
if( n < cnt + v.cnt)
throw OutOfRange();
int i;
for( i = cnt-1; i >= pos; --i)
arr[i+v.cnt] = arr[i];

// Shift up
// starting at pos.

for( i = 0; i < v.cnt; ++i)
arr[i+pos] = v.arr[i];
cnt = cnt + v.cnt;

// Fill the gap.

}
template 
void Array::remove(int pos) throw(BadIndex)
{
if( pos >= 0 && pos < cnt)
{
for( int i = pos; i < cnt-1; ++i)
arr[i] = arr[i+1];
--cnt;
}
else throw BadIndex(pos);
}
template 
ostream& operator<<(ostream& os, const Array& v)
{
int w = os.width();
// Save the field width
for( int i = 0; i < v.cnt; ++i)
{
os.width(w);
os << v.arr[i];
}
os << endl;
return os;
}
#endif

SOLUTIONS

//
//
//
//
//
//

■

---------------------------------------------------DayTime.h
Class DayTime with relational operators,
the operators ++ and -- (prefix and postfix),
and the operators << and >> for I/O.
----------------------------------------------------

// The same as in Chapter 19.

//
//
//
//

-----------------------------------------------------Array_t.cpp
Testing class templates Array.
------------------------------------------------------

#include "array.h"
#include "DayTime.h"
#include 
#include 
#include 
using namespace std;
typedef Array
typedef Array

IntArr;
DoubleArr;

typedef Array DayTimeArr;
int main()
{
try
{
const DoubleArr vd(10, 9.9);
DoubleArr kd;
cout << "\nThis is the constant array of doubles: \n";
cout << setw(8) << vd;
kd = vd;
cout << "\nAn array of doubles after the assignment: "
<< endl;
cout << setw(8) << kd;
kd.remove(3);
kd.append(10.0);
kd.append(20.0);

//
//
//
//

Delete the element at
position 3.
Add a new element.
And repeat!

cout << "\nThis is the modified array: "
<< endl;
cout << setw(8) << kd;

747

748

■

CHAPTER 32

TEMPLATES

IntArr vi;
int i;
for(i=0; i < 10; i++)
vi.append(rand()/100);
cout << "\nThis is the array of int values: \n";
cout << setw(12) << vi;
vi += vi;
cout << "\nAnd append: \n";
cout << setw(12) << vi;
IntArr ki(vi);
cout << "\nThis is the copy of the array: \n";
cout << setw(12) << ki;
DayTimeArr vt;
DayTime temp;

// Array of DayTime objects.

for(i=0; i < 3; i++)
{
if( !(cin >> temp))
break;
vt.append(temp);
}
cout << "\nThe array with objects of type DayTime:\n";
for(i=0; i < 3; i++)
cout << setw(20) << vt[i] << endl;
}
catch(BadIndex& err)
{
cerr << "\nIndex " << err.getBadIndex()
<< " invalid";
exit(1);
}
catch(OutOfRange& )
{
cerr << "\nArray is full!";
exit(2);
}
return 0;
}

chapter

33

Containers
This chapter describes standard class templates used to represent
containers for more efficient management of object collections.These
include
■

sequences, such as lists and double ended queues

■

container adapters, such as stacks, queues, and priority queues

■

associative containers, such as sets and maps, and

■

bitsets.

Besides discussing how to manage containers, we will also be looking at
sample applications, such as bitmaps for raster images, and routing
techniques.

749

750

■

CHAPTER 33

■

CONTAINERS

CONTAINER TYPES

Sequences and associative containers

Containers

Sequences

Associative
Containers

...
Arrays

Stacks

Queues
etc.

Sets

Maps

Bitsets

CONTAINER TYPES

■

751

䊐 What is a Container?
Containers are used to store objects of the same type and provide operations with which
these objects can be managed. These operations include object insertion, deletion, and
retrieval. Memory is allocated for containers dynamically at runtime. Containers thus
provide a safe and easy way to manage collections of objects.
The C++ standard library provides various class templates for container management
in the Containers Library. These classes can be categorized as follows:
■
■

sequential containers, or sequences, where the objects are arranged sequentially
and access to an object can either be direct or sequential
associative containers, where the objects are generally organized and managed in
a tree structure and can be referenced using keys.

䊐 Sequences
Sequential containers are distinguished by the operations defined for them, which are
either generic or restricted. Restricted operations, such as appending at the end of a container, have constant runtimes. That is, the runtime is proportional to a fixed period of
time and does not depend on the number of objects in the container.
The following are sequential containers:
■
■
■

arrays, which provide the same operations as C arrays but increase and decrease
in size dynamically, in contrast to C arrays
queues, which are managed on the FIFO (First In First Out) principle. The first
element to be inserted is also removed first
stacks, which are managed on the LIFO (Last In First Out) principle. The last
element to be inserted is removed first.

䊐 Associative Containers and Bitsets
Associative containers comprise sets, which allow quick access to objects via sortable
keys, and maps, which maintain efficient object/key pairs.
There are also so-called bitsets, which represent bit sequences of a given length and
provide bitwise operators, with which bits can be manipulated.

752

■

CHAPTER 33

■

CONTAINERS

SEQUENCES

Operations for sequences
Class Template

vector >

Time needed to insert or
remove an object

At the end: constant.
At the beginning or in the
middle: linear.

list >

In all positions: constant.

deque >

At the beginning or end:
constant.
In the middle: linear

Container adapters
Class Template

Insertion

Deletion

stack >

at the end

at the end

queue >

at the end

at the
beginning

priority
based

at the
beginning

priority_queue,
Compare=less >

Sequences and header files
Container

Header File

vector



list



deque



stack



queue



priority_queue

Container, Compare >

SEQUENCES

■

753

䊐 Representing Sequences
The Containers Library defines so-called container classes representing containers. These
are class templates parameterized by the type T of the objects to be managed.
Three basic class templates are defined for sequences.
■

■

■

The container class vector supports standard array operations, such as direct access to individual objects via the subscript operator [],
and quick appending and deletion at the end of the container. However, the runtime for insertion and deletion at the beginning or in the middle of the container
is linear, that is, proportional to the number of objects stored in the container.
The container class list provides functionality that is typical for double linked lists. This includes quick insertion and deletion at any given
position. General list operations, such as sorting and merging, are also defined.
The container class deque (double ended queue, pronounced
“deck”) provides direct access via a subscript operator, just like a normal array,
but offers optimized insertion and deletion at the beginning and end of the container. The same operations in the middle of a container have a linear runtime.

The second template parameter is used for any storage allocation to be performed. The
storage management is represented by a so-called allocator class, which is parameterized
by an object of type T. It enables dynamic memory allocation for objects of type T. The
default value of the template parameter is the standard allocator class allocator
that uses the new and delete operators to allocate and release memory.

䊐 Adapter Classes
The basic sequence classes are used to construct so-called adapter classes. An adapter class
expects a sequence as a template argument and stores the sequence in a protected
data member.
The opposite page shows various adapter classes. The priority_queue template
represents priority queues. The relationship between the keys used to manage the priorities is defined in the comparator class, Compare. The default value of the template
parameter is the predefined comparator class, less, which uses the lesser than operator < for type T.

754

■

CHAPTER 33

■

CONTAINERS

ITERATORS

Iterating lists
// Outputs a list containing integers.
// --------------------------------------------------#include 
#include 
using namespace std;
typedef list INTLIST;
// int list
int display(const INTLIST& c)
{
int z = 0;
list::const_iterator pos;

// Counter
// Iterator

for( pos = c.begin(); pos != c.end(); pos++, z++)
cout << *pos << endl;
cout << endl;
return z;
}

Iterating vectors
// iterat_t.cpp: Outputs an array of accounts.
// --------------------------------------------------#include 
#include 
using namespace std;
#include "account.h"
typedef vector AccVec;

// Account vector

void display(const AccVec& v)
{
AccVec::const_iterator pos;

// Iterator

for( pos = v.begin(); pos < v.end(); pos++)
pos->display();
cout << endl;
}

ITERATORS

■

755

䊐 Positioning and Iterating in Containers
Each object in a container occupies the specific position where it was stored. To allow
you to work with the objects in a container, the positions of the objects in the container
must be accessible. There must therefore be at least one mechanism that allows:
■
■

read and/or write access to the object at any given position and
moving from the position of one object to the position of the next object in the
container.

This situation should be familiar from your experience of working with pointers. Given
that i is the index of an element in an array v, (v+i) is its address, *(v+i) the array
element itself, and (v + (++i)) the address of the next array element.
Iterators were introduced in C++ to provide a uniform model for positioning and iteration in containers. An iterator can thus be regarded as an abstraction of a pointer.

䊐 Iterator Types
Two types of iterators are important in this context:
■

■

bidirectional iterators, which can be shifted up by the increment operator ++
and down with the decrement operator --, and use the operators * and -> to
provide write or read access to objects
random access iterators, which are bidirectional iterators that can additionally
perform random positioning. The subscript operator [] was overloaded for this
purpose, and the operations defined for pointer arithmetic, such as addition/subtraction of integers or comparison of iterators, are defined.

The container classes vector and deque have random access iterators and the
container class list has bidirectional iterators.

䊐 Iterator Classes
The types iterator and const_iterator are defined in all the above classes to represent iterators. An iterator belonging to one of these classes can reference constant or
non-constant objects.
The methods begin() and end() are also defined. The begin() method accesses
the first position and end() accesses the position after the last container object.
Containers that belong to adapter classes offer only restricted access to the beginning
or end. You cannot use iterators to walk through them.

756

■

CHAPTER 33

■

CONTAINERS

DECLARING SEQUENCES

The derived container class sortVec
// sortVec.h: The Class Template SortVec representing
//
a sorted vector.
//--------------------------------------------------#include 
// For class template vector
#include  // For comparator class less
using namespace std;
template  >
class SortVec : public vector
{
public:
SortVec() { }
SortVec(int n, const T& x = T());
void insert(const T& obj);
// in sorted order
int search(const T& obj);
void merge(const SortVec& v);
};

Using the container class sortVec
// sortv_t.cpp :
Tests the template SortVec.
//--------------------------------------------------#include "sortVec.h"
typedef SortVec IntSortVec;
int main()
{
IntSortVec v, w;

// Default constructor

v.insert(2);
v.insert(7); v.insert(1);
int n = v.search(7);
w.insert(3); w.insert(9);
v.merge(w);
return 0;
}
// The array v then contains the elements: 1 2 3 7 9

DECLARING SEQUENCES

■

757

䊐 Constructors of vector, list, and deque
The container classes vector, list, and deque define three constructors and a copy
constructor with which sequences can be created. Their functionality is similar for the
various classes and is discussed in the following section using the vector class as an
example.
The statement

Example: vector v;
declares an empty container v for objects of the Account type. You can then insert
individual objects into the container.
However, you can also declare a container and fill it with a predefined number of
object copies.

Example: Fraction x(1, 1);
vector cont(100, x);

This defines the container cont with 100 Fraction type objects, and fills it with the
object x. If the second argument is not supplied, each of the 100 objects is initialized by
the default constructor.
Finally, you can initialize a container with a part of another container. To do so, you
must state a range of iterators.

Example: vector v(first,last);
The arguments first and last are iterators in an existing container. The new container, v, is initialized using objects in the range [first, last): this includes all the
objects between the positions first (including first) and last (excluding last).

䊐 Constructors for Adapter Classes
Only a default constructor and the copy constructor are defined for adapter classes.
Given that wait is a predefined queue of the container class queue, the following statement

Example: queue w(wait);
creates a new queue, w, and uses the object wait to initialize it.
The opposite page shows the derived container class sortVec, which is used to represent sorted, dynamic arrays. The class is parameterized by the type T of array elements.
The second template parameter is a comparator class, which represents a comparison criterion for sorting.

758

■

CHAPTER 33

■

CONTAINERS

INSERTING IN SEQUENCES

Inserting methods

Method

void push_back(const T&x);

Effect

Adds x at the end of the
sequence.

void push_front(const T&x);

Adds x before the first
element of the sequence.

iterator

insert(iterator
pos, Inserts x after position pos
const T& x = T() ); and returns the position of
the newly inserted element.

size_type insert(iterator pos, Inserts n copies of x after
size_type n, const T& x) position pos and returns the
number of inserted elements.

void insert(iterator pos,
InputIterator first
InputIterator last)

Inserts all elements from
range [first,last) after
position pos into the
sequence.

Method insert() of the derived container class SortVec
// Method insert() adds a new object at the end
// of the vector and reorganizes in ascending order.
//--------------------------------------------------template 
void SortVec::insert(const T& obj)
{
SortVec::iterator pos, temp;
push_back(obj);
// Add at the end.
pos = end();

pos--;

// Last position

while (pos-- > begin())
// Sort:
{
if( obj < *pos)
// Swap:
{ temp = pos; *(++temp) = *pos; *pos = obj; }
else
break;
}
}

INSERTING IN SEQUENCES

■

759

䊐 Insertion Methods
The following methods are defined in the container classes vector, deque, and list
■
■

push_back()
insert()

insert at end
insert after a given position.

Additionally, the following method is available in the list and deque classes
■

push_front()

insert at beginning.

This method is not defined in the vector class.
The insert() method is overloaded in various versions, allowing you to insert a single object, multiple copies of an object, or object copies from another container. Given
two containers v and w the following

Example: w.insert(--w.begin(), v.begin(), v.end());
inserts the objects from container v in front of all the other objects in w. A container can
of course be assigned to another container of the same type. The assignment operator is
overloaded for containers to allow this operation.

䊐 Runtime Behavior
The push_back() and push_front() methods are preferable on account of their
constant runtime. Insertion of one object with the insert() method also has a constant runtime in the list class. However, this is linear in the vector and deque
classes, that is, the time increases proportionally to the number of objects in the container.
This dissimilar runtime behavior for methods can be ascribed to the implementation
of various container classes. Normally, containers of the list type are represented by
double linked lists in which each element possesses a pointer to the preceding and following element. This allows for extremely quick inserting at a given position.
The container classes vector and deque are represented as arrays. Inserting in the
middle means shifting the objects in the container to make place for the new object.
Therefore the runtime will increase proportionally with the number of objects the container holds.

䊐 Insertion in Adapter Classes
There is only one insertion method for adapter classes: push(). In stacks and queues,
push() appends an object with a constant runtime. Insertion of objects into priority
queues depends on the priority of the object and the runtime is linear.

760

■

CHAPTER 33

■

CONTAINERS

ACCESSING OBJECTS

Method search() of container class sortVec
// sortvec.h
// Method search() seeks an object obj in the vector
// using the binary search algorithms:
// The object obj is first compared to the element in
// the middle of the vector. If obj is smaller than the
// "middle" element, it must belong to the left half or
// else to the right half of the vector. We repeat this
// process comparing obj with the "middle" element in
// the section where it is known to be, repeatedly
// halving the size of the interval until the interval
// consists of a single point, which is where obj
// belongs.
// This algorithm has logarithmic time and thus
// is very fast.
// ---------------------------------------------------template 
int SortVec::search(const T& obj)
{
int first = 0, last = end() - begin() - 1, mid;
while( first < last )
{
mid = (first + last + 1)/ 2;
// Search the left half,
if( obj < (*this)[mid] )
last = mid - 1;
else first = mid;
// the right half.
}
if ( obj == (*this)[first] )
return first;
else return size();
}

// Found?
// Yes.
// Not found.

ACCESSING OBJECTS

䊐 The

front()

and

back()

■

761

Methods

Access to individual objects in the container classes vector, deque, and list can be
performed by the following methods
■
■

front()
back()

for access to the first element and
for access to the last element.

Both methods return a reference to the object in question.

Example: double z = v.front();
v.front() = 1.9;

This saves the first object in container v in the variable z and then overwrites the object
by 1.9.

䊐 Access via Indices
The subscript operator [] is overloaded in the vector and deque classes to permit the
use of indices for access to the objects in a container. An index is a positive integer of
the type size_type.

Example: v[20] = 11.2;

// the 21st object

Given that pos is an iterator that references the first object in a container v, the expression v[20] is equivalent to pos[20].
When you use the subscript operator, you must ensure that the index does not exceed
the valid range. You can use the access method at() to throw an exception if an index
is out of range.

Example: v.at(20) = 11.2;
The at() method throws an exception of the standard error class out_of_range if an
error occurs.
The subscript operator and the at() method are not defined in the list class.
Before you can manipulate the tenth object in the container, for example, you need to
walk through the container sequentially up to that position.

䊐 Access to Objects in Adapter Classes
The method top() is defined for access to the element with the highest priority, or the
element at the top of the stack, in the adapter classes priority_queue and stack.
The queue class comprises the front() method, which is used to access the first
element.

762

■

CHAPTER 33

■

CONTAINERS

LENGTH AND CAPACITY

Method merge() of container class SortVec
// sortvec.h: Method merge() merges the argument vector
//
with the vector *this.
// --------------------------------------------------template 
void SortVec::merge(const SortVec& v)
{
SortVec temp;
// Temporary vector
SortVec::iterator pos = begin(); // Iterator
int n1 = 0, n2 = 0;
// Copy the smaller object into vector temp:
while(n1 < size() && n2 < v.size())
if( pos[n1] <= v[n2] )
temp.push_back(pos[n1++]);
else
temp.push_back(v[n2++]);
// Append the rest:
while( n1 < size())
temp.push_back(pos[n1++]);
while( n2 < v.size())
temp.push_back(v[n2++]);
*this = temp;
}

LENGTH AND CAPACITY

■

763

The identifying features of a container are
■
■

its length, that is, the number of objects held in the container, and
the capacity, that is, the maximum number of objects the container can store.

The length of a container changes after every insertion or deletion—the capacity does
not.

䊐 Length of a Container
The length of a container is discovered by a call to the size() method. The method
returns an integer of the size_type type.

Example: Fraction x(1, 1);
vector v(100, x);
vector::size_type sz = v.size();

The variable sz contains the value 100 in this case.
The length of an empty container is always 0. You can also use the empty() method
to discover whether a container is empty. The method returns true in this case.

Example: while( !cont.empty() ) ...
The methods size() and empty() are defined for all container classes. You can use
the resize() method to change the length of a container.

Example: cont.resize( n, x);
The length is increased to n provided n > size() is true, or decreased for n <
size(). If n == size(), nothing happens.
If the length is increased, n – size() copies of the object x are appended to the
container. The second argument, x, can be omitted. In this case, the default constructor
for a type T object is called as often as necessary.

䊐 Capacity
The capacity of a container can be checked using the max_size() method.

Example: size_type k = cont.max_size();
The return value depends on the amount of memory available and the object size.
Only the size() and empty() methods are defined for adapter classes. You cannot
discover the capacity of an object, nor can you call resize() to change its length.

764

■

CHAPTER 33

■

CONTAINERS

DELETING IN SEQUENCES

A priority queue
// prior_t.cpp : Testing a priority queue
// ------------------------------------------------#include 
#include 
#include 
using namespace std;
class Parcel
{
private:
unsigned int prio;
// Priority
string info;
public:
Parcel(unsigned int p, const string& s)
:prio(p), info(s) {}
// Access methods, ... overloaded operators:
friend bool operator<(const Parcel& x,const Parcel& y)
{ return (x.prio < y.prio); }
friend ostream& operator<<(ostream& os,
const Parcel& x)
{ os << x.prio << " "<< x.info << endl; return os; }
};
int main()
{
priority_queue

pq;

pq.push(Parcel(7,"Bob"));
pq.push(Parcel(1,"Peter"));
pq.push(Parcel(4,"Susan"));

// Insert

while( !pq.empty() )
{
cout << pq.top() << endl; // Output
pq.pop();
// and delete
}
return 0;
}
// Output:
//
//

7
4
1

Bob
Susan
Peter

DELETING IN SEQUENCES

■

765

䊐 Deletion Methods
The following methods are available for deleting objects in the container classes vector, deque, and list:
■
■

pop_back()
erase()

■

clear()

deletes the last object in the container
deletes the object at a given position, or deletes all the objects in
a given range
deletes all the objects in a container.

The following method is additionally defined in the deque and list classes:
■

pop_front() deletes the first object in the container.

The method does not have a return value, just like the pop_back() method.
The pop_back() and pop_front() methods are preferable on account of their
constant runtimes. Using the erase() method to delete an object at the beginning or
in the middle of a container also provides a constant runtime in the container class
list. However, the runtime is linear in the vector and deque classes, since objects
must be shifted within the container to fill the gap left by the deletion.

䊐 Deleting Ranges of Objects
When you use the erase() method to delete the objects in a given range, the position
of the first element to be deleted and the position after the last object to be deleted are
required as arguments.

Example: cont.erase(cont.begin() + 10, cont.end());
This deletes all the remaining objects in the container, starting at position 11. The
erase() method returns the new position of the object immediately after the range of
objects deleted.

䊐 Deletion in Adapter Classes
There is only one method of deletion for adapter classes, namely pop(). Given that
wait is a queue of the queue type, the following statement

Example: wait.pop();
deletes the element at the beginning of the queue. In the case of a stack, pop() deletes
the element at the top of the stack, and for priority queues, the object with the highest
priority. The runtime is constant in all cases.

766

■

CHAPTER 33

■

CONTAINERS

LIST OPERATIONS

Sample program
// list_t.cpp: Tests list operations
// ---------------------------------------------------#include 
#include 
#include 
using namespace std;
typedef list INTLIST;
int display( const INTLIST& c);
int main()
{
INTLIST ls, sls;
int i;
for( i = 1; i <= 3; i++)
ls.push_back( rand()%10 );

// ex. 1 7 4

ls.push_back(ls.front());

// 1 7 4 1

ls.reverse();

// 1 4 7 1

ls.sort();

// 1 1 4 7

for( i = 1; i <= 3; i++)
sls.push_back( rand()%10 );

// ex. 0 9 4

// Insert first object of sls before the last in ls:
INTLIST::iterator pos = ls.end();
ls.splice(--pos, sls, sls.begin());

// 1 1 4 0 7

display(sls);

// 9 4

ls.sort();
sls.sort();
ls.merge(sls);
ls.unique();

//
//
//
//

return 0;
}

0
4
0
0

1 1 4 7
9
1 1 4 4 7 9
1 4 7 9

LIST OPERATIONS

■

767

The container class list comprises methods for list operations that are not defined in
other container classes. These are
■
■
■

sorting and inverting lists
merging two sorted lists
splicing lists.

䊐 Sorting, Inverting, and Splicing Lists
A container of the list type, or list container for short, can be sorted by a call to
sort(). This assumes that the operator < is defined in class T. A call to sort() sorts
the container in ascending order.
You can use the reverse() method to invert a list container, that is, to reverse the
order of the objects in the container. What was originally the first element in the container becomes the last, and the second element becomes the second to last, and so on.
The merge() method is used to merge two list containers. Given that ls1 and ls2
are two sorted list containers, the following call

Example: ls1.merge(ls2);
creates the sorted list ls1, whose objects comprise the original objects of ls1 and ls2.
The ls2 container is empty following this operation.

䊐 Splice Operations
Splice operations insert the objects from one list container at a given position in another
list container and remove them from the original container. You can transfer either a
whole container or just part of a container.

Example: ls1.splice(pos, ls2);
This inserts the whole of container ls2 in front of position pos in ls1. ls2 is emptied
by this statement. The following statement

Example: ls1.splice(pos1, ls2, pos2);
Inserts the element at position pos2 in ls2 before the element at position pos1 in ls1
and deletes it from ls2. If you want to transfer part of a container, the third and fourth
arguments must contain the starting and end position.
You cannot use a splice operation to insert at a position before begin() or after
end().

768

■

CHAPTER 33

■

CONTAINERS

ASSOCIATIVE CONTAINERS

Container classes

Container Class

Representing

set< class T,
class Compare = less,
class Allocator = allocator >

collections of objects with

multiset< class T,
class Compare = less,
class Allocator = allocator >

collections of objects with

unique keys

equivalent keys, i.e.
possibly multiple copies of
the same key value

map< class Key, class T,
class Compare = less,
class Allocator = allocator >

collections of objects/key

multimap< class Key, class T,
class Compare = less,
class Allocator = allocator >

collections of objects/key

pairs where the keys are
unique

pairs with possibly
equivalent keys

Associative containers and header files

Container Class

Header File



set< T, Compare, Allocator >
>



map< Key, T, Compare, Allocator >

File Type                       : PDF
File Type Extension             : pdf
MIME Type                       : application/pdf
PDF Version                     : 1.6
Linearized                      : Yes
Author                          : Kirch-Prinz, Ulla.; Prinz, Peter.
Create Date                     : 2008:05:28 01:29:18-07:00
Modify Date                     : 2008:09:23 18:39:48-07:00
Has XFA                         : No
XMP Toolkit                     : Adobe XMP Core 4.2.1-c041 52.342996, 2008/05/07-20:48:00
Metadata Date                   : 2008:09:23 18:39:48-07:00
Format                          : application/pdf
Title                           : A Complete Guide to Programming in C++
Creator                         : Kirch-Prinz, Ulla.; Prinz, Peter.
Document ID                     : uuid:d529af70-8d4a-4462-809c-395830c6dcea
Instance ID                     : uuid:8ecb258b-f548-4800-a574-c46b687b6698
Producer                        : Acrobat Distiller 5.0 (Windows)
Page Count                      : 846