Introduction to C++

Data Abstraction w/ Classes
Topic #2

Lecture #1 plus Review
Abstract Data Types
Introduction to...Object Models
Introduction to...Data Abstraction
Using Data Abstraction in C++ ...an introduction to the class
Members of a Class
The class interface, using the class, the class interface versus implementation
Classes versus Structures
Constructors, Destructors
Dynamic Memory and Linked Lists

Programming Paradigms
The most important aspect of C++ is its ability to support many different programming paradigms
procedural abstraction
modular abstraction
data abstraction
object oriented programming (this is discussed later, once we learn about the concept of inheritance)

Procedural Abstraction
This is where you build a “fence” around program segments, preventing some parts of the program from “seeing” how tasks are being accomplished.
Any use of globals causes side effects that may not be predictable, reducing the viability of procedural abstraction

Modular Abstraction
With modular abstraction, we build a “screen” surrounding the internal structure of our program prohibiting programmers from accessing the data except through specified functions.
Many times data structures (e.g., structures) common to a module are placed in a header files along with prototypes (allows external references)

Modular Abstraction
The corresponding functions that manipulate the data are then placed in an implementation file.
Modules (files) can be compiled separately, allowing users access only to the object (.o) files
We progress one small step toward OOP by thinking about the actions that need to take place on data...

Modular Abstraction
We implement modular abstraction by separating out various functions/structures/classes into multiple .cpp and .h files.
.cpp files contain the implementation of our functions
.h files contain the prototypes, class and structure definitions.

Modular Abstraction
We then include the .h files in modules that need access to the prototypes, structures, or class declarations:
#include “myfile.h”
(Notice the double quotes!)
We then compile programs (on UNIX) by:
g++ main.cpp myfile.cpp
(Notice no .h file is listed on the above line)

Data Abstraction
Data Abstraction is one of the most powerful programming paradigms
It allows us to create our own user defined data types (using the class construct) and
then define variables (i.e., objects) of those new data types.

Data Abstraction
With data abstraction we think about what operations can be performed on a particular type of data and not how it does it
Here we are one step closer to object oriented programming

Data Abstraction
Data abstraction is used as a tool to increase the modularity of a program
It is used to build walls between a program and its data structures
what is a data structure?
talk about some examples of data structures
We use it to build new abstract data types

Data Abstraction
An abstract data type (ADT) is a data type that we create
consists of data and operations that can be performed on that data
Think about a char type
it consists of 1 byte of memory and operations such as assignment, input, output, arithmetic operations can be performed on the data

Data Abstraction
An abstract data type is any type you want to add to the language over and above the fundamental types
For example, you might want to add a new type called: list
which maintains a list of data
the data structure might be an array of structures
operations might be to add to, remove, display all, display some items in the list

Data Abstraction
Once defined, we can create lists without worrying about how the data is stored
We “hide” the data structure used for the data within the data type -- so it is transparent to the program using the data type
We call the program using this new data type: the client program (or client)

Data Abstraction
Once we have defined what data and operations make sense for a new data type, we can define them using the class construct in C++
Once you have defined a class, you can create as many instances of that class as you want
Each “instance” of the class is considered to be an “object” (variable)

Data Abstraction
Think of a class as similar to a data type
and an object as a variable
And, just as we can have zero or more variables of any data type...
we can have zero or more objects of a class!
Then, we can perform operations on an object in the same way that we can access members of a struct...

What is a Class?
Remember, we used a structure to group different types of data together under a common name
With a class, we can go the next step an actually define a new data type
In reality, structures and classes are 100% the same except for the default conditions
everything you can do with a class you can do with a structure!

What is a Class?
First, let’s talk about some terminology
Think of a class as the same as a data type
Think of an object as the same as a variable
An “object” is an instance of a class
Just like a “variable” is an instance of a specific data type
We can zero or more variables (or objects) in our programs

When do we used Classes?
I recommend using structures when you want to group different types of data together
and, to use a class when we are interested in building a new type of data into the language itself
to do this, I always recommend forming that data type such that it behaves in a consistently to how the fundamental data types work

But, What is a Data Type?
We’ve been working with fundamental data types this term, such as ints, floats, chars...
Whenever we define variables of these types,
memory is allocated to hold the data
a set of operations can now be performed on that data
different data types have different sets of operations that make sense (the mod operator doesn’t make sense for floats...)

Defining new Data Types...
Therefore, when we define a new data type with the class construct
we need to specify how much memory should be set aside for each variable (or object) of this type
and, we need to specify which operations make sense for this type of data (and then implement them!!)
and, what operators makes sense (do be discussed with operator overloading)

Defining a Class...
Once we have decided on how the new type of data should behave, we are ready to define a class:
class data_type_name {
public:
//operations go here
private:
//memory is reserved here
};

For Example, here is a Class Interface
class string {
public:
string();
int copy(char []);
int length();
int display();
private:
char str[20];
int len;
};

Then, the Class Implementation
string::string() {
str[0]=‘\0’;
    len = 0;
}
int string::copy(char s [])
[
if (strlen(s) < 20)
strcpy (str, s);
else {
for (int i = 0; i< 20; ++i)
str[i] = s[i];
str[20]=‘\0’;
len = strlen(str); return len;
}

More of the Class Implementation
int string::length()
{
return len;
}
int string::display()
{
cout <<str;
return len;
}

Defining Objects of this Class
Notice how similar defining objects of class is to defining variables of any data type:
string my_str; vs. int i;
Defining an object causes the “constructor” to be invoked; a constructor is the same named function as the class (string) and is used to initialize the memory set aside for this object
Think about how much memory is set aside?
What initial values should it take on?

Using Objects of this Class
Think about how you can use those objects
my_str.copy("hi! ");
cout << my_str.length();
We are limited to using only those operations that are defined within the public section of the class interface
The only “built-in” operation that can be used with objects of a class is the assignment operation, which does a memberwise copy (as we learned with structures)

Using Objects of this Class
Notice how similar the use of these operations is to the cin.get function.....
cin.get(ch);
This should be a clue. cin therefore is an object of the istream class.
The dot is the member access operator; it allows us to access a particular public member function defined within the istream class.
The function get is therefore defined within the public section of the istream class

Limitations...
But, there are limitations!
If our goal is to really be able to use my string objects in a way consistent with the fundamental data types,
then I would expect to be able to read strings using the extraction operator
and to display strings by directly using the insertion operator
and to concatenate strings using +

Limitations...
With the class as it is defined, none of these things can be done...
the only operations that can be performed are those specified within the public section of the class interface, and a memberwise copy with the assignment operator
No other operations are known
Therefore, to be consistent, we must revise our class to use operator overloading

For Example, here is a Class Interface
class string {
public:
string();
int length();
    friend ofstream & operator <<
(ofstream &, const string &);
     friend ifstream & operator >>
(ifstream &, string &);
private:
   char str[20];
    int len;
};

List Example
For a list of videos, we might start with a struct defining what a video is:
struct video {
char title[100];
char category[5];
int quantity;
};

List Example
For a list of videos data type:
class list {
   public:
      list();
      int add (const video &);
      int remove (char title[]);
      int display_all();
   private:
      video my_list[CONST_SIZE];  //for now...
int num_of_videos;
};

List Example
For a client to create a list object:
int main() {
list home_videos; //has an array of 100 videos
list kids_shows; //another 100 videos here...
•••
video out_of_site;
cin.get(out_of_site.title,100,’\n’);
cin.ignore(100,’\n’);
•••
home_videos.add(out_of_site); //use operation

"Data Hiding"
Data Hiding
and
Member Functions

Data Abstraction in C++
Terminology
Data Hiding
Class Constructors
Defining and using functions in classes
Where to place the class interface and implementation of the member functions

“class” Terminology
Class
think data type
Object
instance of a class, e.g., variable
Members
like structures, the data and functions declared in a class
called “data members” and “member functions”

“class” Terminology
A class could be a list, a string, a counter, a clock, a bank account, etc.
discuss a simple counter class on the board
An object is as real as a variable, and gets allocated and deallocated just like variables
discuss the similarities of:
int i; list j;

“class” Terminology
For the list of videos data type we used
class list { <--- the data type!!!
   public:
      list(); <--- the constructor
      int add (const video &);    3 member functions
      int remove (char title[]);
      int display_all();
   private:
      video my_list[CONST_SIZE];                 data members
   int num_of_videos;
}; <--- notice like structures we need a semicolon

“class” Terminology
If we examine the previous class,
notice that classes are really very similar to structures
a class is simply a generalized structure
in fact, even though we may not have used structures in this way...
Structures and Classes are 100% identical except for their default conditions...
by default, all members in a structure are available for use by clients (e.g., main programs); they are public

“class” Terminology
We have seen the use of structures in a more simple context,
as we examined with the video struct.
It had three members (data members)
called title, category, and quantity.
They are “public” by default,
so all functions that have objects of type video can directly access members by:
video object;
object.title    object.category object.quantity

“class” Terminology
This limited use of a structure was appropriate, because
it served the purpose of grouping different types of data together as a single unit
so, anytime we want to access a particular video -- we get all of the information pertaining to the video all at once

Structure Example
Remember, anything you can do in a struct you can do in a class.
It is up to your personal style how many structures versus classes you use to solve a problem.
Benefit: Using structures for simple “groupings” is compatible with C
struct video {
char title[100];
char category[5];
int quantity;
};

“class” Terminology
To accomplish data hiding and encapsulation
we usually turn towards classes
What is data hiding?
It is the ability to protect data from unauthorized use
Notice, with the video structure, any code that has an object of the structure can access or modify the title or other members

Data Hiding
With data hiding
accessing the data is restricted to authorized functions
“clients” (e.g., main program) can’t muck with the data directly
this is done by placing the data members in the private section
and, placing member functions to access & modify that data in the public section

Data Hiding
So, the public section
includes the data and operations that are visible, accessible, and useable by all of the clients that have objects of this class
this means that the information in the public section is “transparent”; therefore, all of the data and operations are accessible outside the scope of this class
by default, nothing in a class is public!

Data Hiding
The private section
includes the data and operations that are not visible to any other class or client
this means that the information in the private section is “opaque” and therefore is inaccessible outside the scope of this class
the client has no direct access to the data and must use the public member functions
this is where you should place all data to ensure the memory’s integrity

Data Hiding
The good news is that
member functions defined in the public section can use, return, or modify the contents of any of the data members, directly
it is best to assume that member functions are the only way to work with private data
(there are “friends” but don’t use them this term)
Think of the member functions and private data as working together as a team

“class” Terminology
Let’s see how “display_all” can access the data members:
list:: display_all() {
for (int i=0; i<num_of_videos; ++i)
   cout <<my_list[i].title <<‘\t’
<<my_list[i].category
<<‘\t’ <<my_list[i].quantity <<endl;
    }

Data Hiding
Notice, that the display_all function can access the private my_list and num_of_videos members, directly
without an object in front of them!!!
this is because the client calls the display_all function through an object
object.display_all();
so the object is implicitly available once we enter “class scope”

Where to place....
In reality, the previous example was misleading. We don’t place the implementation of functions with this this class interface
Instead, we place them in the class implementation, and separate this into its own file

Class Interface  (.h)
Class Interface: list.h
class list {
   public:
   int display_all()
•••
};
list.h can contain:
prototype statements
structure declarations and definitions
class interfaces and class declarations
include other files

Class Implementation
Class Implementation list.cpp
#include "list.h" notice the double quotes
   int list::display_all() {
for (int i=0; i<num_of_videos; ++i)
   cout <<my_list[i].title <<‘\t’
<<my_list[i].category
<<‘\t’ <<my_list[i].quantity <<endl;
    }
Notice, the code is the same
But, the function is prefaced with the class name and the scope resolution operator!
This places the function in class scope even though it is implemented in another file
Including the list.h file is a “must”

Constructors
Remember that when you define a local variable in C++, the memory is not automatically initialized for you
This could be a problem with classes and objects
If we define an object of our list class, we really need the “num_of_videos” data member to have the value zero
Uninitialized just wouldn’t work!

Constructors
Luckily, with a constructor we can write a function to initialize our data members
and have it implicitly be invoked whenever a client creates an object of the class
The constructor is a strange function, as it has the same name as the class, and no return type (at all...not even void).

Constructor
The list constructor was:  (list.h)
class list {
  public:
      list(); <--- the constructor
      •••
};
The implementation is: (list.cpp)
list::list(){
   num_of_videos = 0;
}

Constructor
The constructor is implicitly invoked when an object of the class is formed:
int main() {
list fun_videos; implicitly calls the constructor
  list all_videos[10];  implicitly calls the
  constructor 10 times for
each of the 10 objects!!

Dynamic Memory w/ Classes
But, what if we didn’t want to waste memory for the title (100 characters may be way too big (Big, with Tom Hanks)
So, let’s change our video structure to include a dynamically allocated array:
struct video {
char * title;
char category[5];
int quantity;
};

Dynamic Memory w/ Classes
Let’s write a class that now allocates this list of videos dynamically, at run time
This way, we can wait until we run our program to find out how much memory should be allocated for our video array

Dynamic Memory w/ Classes
What changes in this case are the data members:
class list {
   public:
      list();   ~list();
      int add (const video &);
      int remove (char title[]);
      int display_all();
   private:
      video *my_list;
   int video_list_size;
   int num_of_videos;
};

Default Constructor
Now, let’s think about the implementation.
First, what should the constructor do?
initialize the data members
list::list() {
my_list = NULL;
video_list_size = 0;
num_of_videos = 0;
}

Another Constructor
Remember function overloading? We can have the same named function occur (in the same scope) if the argument lists are unique.
So, we can have another constructor take in a value as an argument of the number of videos
and go ahead and allocate the memory,  so that subsequent functions can use the array

2nd Constructor
list::list(int size) {
my_list = new video [size];
video_list_size = size;
num_of_videos = 0;
}
Notice, unlike arrays of characters, we don’t need to add one for the terminating nul!

Clients creating objects
The client can cause this 2nd constructor to be invoked by defining objects with initial values
list fun_videos(20);  //size is 20
If a size isn’t supplied, then no memory is allocated and nothing can be stored in the array....

Default Arguments
To fix this problem, we can merge the two constructors and replace them with a single constructor:
list::list(int size=100) {
my_list = new video [size];
video_list_size = size;
num_of_videos = 0;
}
(Remember, to change the prototype for the constructor in the class interface)

Destructor
Then, we can deallocate the memory when the lifetime of a list object is over
When is that?
Luckily, when the client’s object of the list class lifetime is over (at the end of the block in which it is defined) -- the destructor is implicitly invoked

Destructor
So, all we have to do is write a destructor to deallocate our dynamic memory.
list::~list() {
delete [] my_list;
my_list = NULL;
•••
}
(Notice the ~ in front of the function name)
(It can take NO arguments and has NO return type)
(This too must be in the class interface....)

Review of Classes
What is the difference between a class and a struct
What is a data member?
Where should a data member be placed in a class? (what section)
What is a member function?
Where should member functions be placed, if clients should use them?

Review of Classes
What is the difference between a member function and a regular-old C++ function?
What is the purpose of the constructor?
Why is it important to implement a constructor?
What is the difference between a class and an object?

Review of Classes
Show an example of how a client program defines an object of a list class
How then would the client program call the constructor? (trick question!)
How then would the client program call the display_all function?
Why are parens needed?

Review of Classes
Write a simple class interface (called number) that has the following members:
an integer private data member (containing a value)
a constructor
a set member function, that takes an integer as an argument and returns nothing
a display member function

Review of Classes
Now, let’s try our hand at implementing these functions
a constructor
a set member function, that takes an integer as an argument and returns nothing
a display member function

Review of Classes
What happens if we forgot to put the keyword public in the previous class interface?
Why is it necessary to place the class name, followed by the scope resolution operator (::) when we implement a member function outside of a class?