PYTHON APPLICATION PROGRAMMING [15CS664]

MODULE – 4
4.1 CLASSES AND OBJECTS
Python is an object-oriented programming language, and class is a basis for any object
oriented programming language. Class is a user-defined data type which binds data and
functions together into single entity. Class is just a prototype (or a logical entity/blue print)
which will not consume any memory. An object is an instance of a class and it has physical
existence. One can create any number of objects for a class. A class can have a set of
variables (also known as attributes, member variables) and member functions (also known
as methods).

(Overview of general OOP concepts is given at the end of this module as an extra topic.
Those who are new to OOP concepts, it is suggested to have a glance and then continue
reading).

4.1.1 Programmer-defined Types

A class in Python can be created using a keyword class. Here, we are creating an empty
class without any members by just using the keyword pass within it.

class Point:
pass

print(Point)

The output would be –

<class ' main .Point'>

The term main indicates that the class Point is in the main scope of the current
module. In other words, this class is at the top level while executing the program.

Now, a user-defined data type Point got created, and this can be used to create any number
of objects of this class. Observe the following statements –

p=Point()

Now, a reference (for easy understanding, treat reference as a pointer) to Point object is
created and is returned. This returned reference is assigned to the object p. The process of
creating a new object is called as instantiation and the object is instance of a class. When
we print an object, Python tells which class it belongs to and where it is stored in the memory.
print(p)
 PYTHON APPLICATION PROGRAMMING [15CS664]

The output would be –

< main.Point object at 0x003C1BF0>

The output displays the address (in hexadecimal format) of the object in the memory. It is
now clear that, the object occupies the physical space, whereas the class does not.

4.1.2 Attributes
An object can contain named elements known as attributes. One can assign values to these
attributes using dot operator. For example, keeping coordinate points in mind, we can assign
two attributes x and y for the object p of a class Point as below –

p.x =10.0
p.y =20.0

A state diagram that shows an object and its attributes is called as object diagram. For the
object p, the object diagram is shown in Figure 4.1.

Point
x 10.0
p
y 20.0

Figure 4.1 Object Diagram

The diagram indicates that a variable (i.e. object) p refers to a Point object, which
contains two attributes. Each attributes refers to a floating point number.

One can access attributes of an object as shown –

>>> print(p.x)
10.0
>>> print(p.y)
20.0

Here, p.x means “Go to the object p refers to and get the value of x”. Attributes of an object
can be assigned to other variables –

>>> x= p.x
>>> print(x)
10.0
 PYTHON APPLICATION PROGRAMMING [15CS664]

Here, the variable x is nothing to do with attribute x. There will not be any name conflict
between normal program variable and attributes of an object.

A complete program: Write a class Point representing a point on coordinate system.

Implement following functions –
• A function read_point() to receive x and y attributes of a Point object as user input.
• A function distance() which takes two objects of Point class as arguments and
computes the Euclidean distance between them.
• A function print_point() to display one point in the form of ordered-pair.

import math
class Point:
""" This is a class Point representing a
coordinate point
"""
def read_point(p):
p.x=float(input("x coordinate:"))
p.y=float(input("y coordinate:"))
def print_point(p):
print("(%g,%g)"%(p.x, p.y))
def distance(p1,p2):
d=math.sqrt((p1.x-p2.x)**2+(p1.y-p2.y)**2)
return d

p1=Point() #create first object

print("Enter First point:")
read_point(p1) #read x and y for p1

p2=Point() #create second object

print("Enter Second point:")
read_point(p2) #read x and y for p2

dist=distance(p1,p2) #compute

distance print("First point is:")

print_point(p1) #print p1

print("Second point is:")

print_point(p2) #print p2

print("Distance is: %g" %(distance(p1,p2))) #print d

 PYTHON APPLICATION PROGRAMMING [15CS664]

The sample output of above program would be –

Enter First point:
x coordinate:10
y coordinate:20
Enter Second point:
x coordinate:3
y coordinate:5
First point is: (10,20)
Second point is:(3,5)
Distance is: 16.5529
Let us discuss the working of above program thoroughly –
• The class Point contains a string enclosed within 3 double-quotes. This is known as
docstring. Usually, a string literal is written within 3 consecutive double-quotes inside
a class, module or function definition. It is an important part of documentation and is
to help someone to understand the purpose of the said class/module/function. The
docstring becomes a value for the special attribute viz. doc available for any
class (and objects of that class). To get the value of docstring associated with a class,
one can use the statements like –

>>> print(Point. doc )

This is a class Point representing a coordinate point
>>> print(p1. doc )
This is a class Point representing a coordinate point

Note that, you need to type two underscores, then the word doc and again two
underscores.

In the above program, there is no need of docstring and we would have just used
pass to indicate an empty class. But, it is better to understand the professional way
of writing user-defined types and hence, introduced docstring.

• The function read_point() take one argument of type Point object. When we use
the statements like,
read_point(p1)

the parameter p of this function will act as an alias for the argument p1. Hence, the
modification done to the alias p reflects the original argument p1. With the help of this
function, we are instructing Python that the object p1 has two attributes x and y.

• The function print_point() also takes one argument and with the help of
formatstrings, we are printing the attributes x and y of the Point object as an ordered-
pair (x,y).
 PYTHON APPLICATION PROGRAMMING [15CS664]

• As we know, the Euclidean distance between two points (x1,y1) and (x2,y2) is
2 2
x1 x2 y1 y2

In this program, we have Point objects as (p1.x, p1.y) and (p2.x, p2.y). Apply the
formula on these points by passing objects p1 and p2 as parameters to the function
distance(). And then return the result.

Thus, the above program gives an idea of defining a class, instantiating objects, creating
attributes, defining functions that takes objects as arguments and finally, calling (or invoking)
such functions whenever and wherever necessary.

NOTE: User-defined classes in Python have two types of attributes viz. class attributes and
instance attributes. Class attributes are defined inside the class (usually, immediately after
class header). They are common to all the objects of that class. That is, they are shared by
all the objects created from that class. But, instance attributes defined for individual objects.
They are available only for that instance (or object). Attributes of one instance are not
available for another instance of the same class. For example, consider the class Point as
discussed earlier –

class Point:
pass
p1= Point() #first object of the class
p1.x=10.0 #attributes for p1
p1.y=20.0
print(p1.x, p1.y) #prints 10.0 20.0
p2= Point() #second object of the class
print(p2.x) #displays error as below
AttributeError: 'Point' object has no attribute 'x'

This clearly indicates that the attributes x and y created are available only for the object p1,
but not for p2. Thus, x and y are instance attributes but not class attributes.

We will discuss class attributes late in-detail. But, for the understanding purpose, observe the
following example –
class Point:
x=2
y=3
p1=Point() #first object of the class
print(p1.x, p1.y) # prints 2 3

p2=Point() #second object of the class

print(p2.x, p2.y) # prints 2 3
 PYTHON APPLICATION PROGRAMMING [15CS664]

Here, the attributes x and y are defined inside the definition of the class Point itself. Hence,
they are available to all the objects of that class.

4.1.3 Rectangles
It is possible to make an object of one class as an attribute to other class. To illustrate this,
consider an example of creating a class called as Rectangle. A rectangle can be created
using any of the following data –
• By knowing width and height of a rectangle and one corner point (ideally, a bottom left
corner) in a coordinate system
• By knowing two opposite corner points
Let us consider the first technique and implement the task: Write a class Rectangle containing
numeric attributes width and height. This class should contain another attribute corner which
is an instance of another class Point. Implement following functions –
• A function to print corner point as an ordered-pair
• A function find_center() to compute center point of the rectangle
• A function resize() to modify the size of rectangle

The program is as given below –

class Point:
""" This is a class Point
representing coordinate point
"""
class Rectangle:
""" This is a class Rectangle.
Attributes: width, height and Corner Point
"""
def find_center(rect):
p=Point()
p.x = rect.corner.x + rect.width/2
p.y = rect.corner.y + rect.height/2
return p
def resize(rect, w, h):
rect.width +=w
rect.height +=h
def print_point(p):
print("(%g,%g)"%(p.x, p.y))

box=Rectangle() #create Rectangle object

box.corner=Point() #define an attribute corner for box
box.width=100 #set attribute width to box
box.height=200 #set attribute height to box
box.corner.x=0 #corner itself has two attributes x and y
 PYTHON APPLICATION PROGRAMMING [15CS664]

box.corner.y=0 #initialize x and y to 0

print("Original Rectangle is:")
print("width=%g, height=%g"%(box.width, box.height))
center=find_center(box)
print("The center of rectangle is:")
print_point(center)
resize(box,50,70) print("Rectangle after resize:")
print("width=%g, height=%g"%(box.width, box.height))
center=find_center(box)
print("The center of resized rectangle is:")
print_point(center)

A sample output would be:

Original Rectangle is: width=100, height=200

The center of rectangle is: (50,100)
Rectangle after resize: width=150, height=270
The center of resized rectangle is: (75,135)

The working of above program is explained in detail here –

• Two classes Point and Rectangle have been created with suitable docstrings. As
of now, they do not contain any class-level attributes.
• The following statement instantiates an object of Rectangle class.
box=Rectangle()

The statement

box.corner=Point()
indicates that corner is an attribute for the object box and this attribute itself is an
object of the class Point. The following statements indicate that the object box has
two more attributes –

box.width=100 #give any numeric value

box.height=200 #give any numeric value

In this program, we are treating the corner point as the origin in coordinate system and
hence the following assignments –

box.corner.x=0
box.corner.y=0

(Note that, instead of origin, any other location in the coordinate system can be given
as corner point.)
 PYTHON APPLICATION PROGRAMMING [15CS664]

Based on all above statements, an object diagram can be drawn as –

Rectangle

box width 100 Point

height 200
x 0
corner y 0

The expression box.corner.x means, “Go to the object box refers to and select the
attribute named corner; then go to that object and select the attribute named x.”

• The function find_center() takes an object rect as an argument. So, when a

call is made using the statement –

center=find_center(box)

the object rect acts as an alias for the argument box.

A local object p of type Point has been created inside this function. The attributes of
p are x and y, which takes the values as the coordinates of center point of rectangle.
Center of a rectangle can be computed with the help of following diagram.

Half of height
(x+ half of width, y+ half of height)

(x,y)
Half of width

The function find_center() returns the computed center point. Note that, the return
value of a function here is an instance of some class. That is, one can have an
instance as return values from a function.

• The function resize() takes three arguments: rect – an instance of Rectangle

class and two numeric variables w and h. The values w and h are added to existing
attributes width and height. This clearly shows that objects are mutable. State of
an object can be changed by modifying any of its attributes. When this function is
called with a statement –
resize(box,50,70)
 PYTHON APPLICATION PROGRAMMING [15CS664]

the rect acts as an alias for box. Hence, width and height modified within the function
will reflect the original object box.

Thus, the above program illustrates the concepts: object of one class is made as attribute
for object of another class, returning objects from functions and objects are mutable.

4.1.4 Copying
An object will be aliased whenever there an object is assigned to another object of same
class. This may happen in following situations – Direct object assignment (like p2=p1)
• When an object is passed as an argument to a function
• When an object is returned from a function

The last two cases have been understood from the two programs in previous sections. Let
us understand the concept of aliasing more in detail using the following program –
>>> class Point:
pass

>>> p1=Point()
>>> p1.x=10
>>> p1.y=20
>>> p2=p1
>>> print(p1)
< main .Point object at 0x01581BF0>
>>> print(p2)
< main .Point object at 0x01581BF0>

Observe that both p1 and p2 objects have same physical memory. It is clear now that the
object p2 is an alias for p1. So, we can draw the object diagram as below –

x 10
p1 p2
y 20

Hence, if we check for equality and identity of these two objects, we will get following result.

>>> p1 is p2
True
>>> p1==p2 True

But, the aliasing is not good always. For example, we may need to create a new object using
an existing object such that – the new object should have a different physical memory, but it
must have same attribute (and their values) as that of existing object. Diagrammatically, we
need something as below –
 PYTHON APPLICATION PROGRAMMING [15CS664]

x 10 x 10
p1 p2
y 20 y 20

In short, we need a copy of an object, but not an alias. To do this, Python provides a
module called copy and a method called copy(). Consider the below given program to
understand the concept.

>>> class Point:

pass

>>> p1=Point()
>>> p1.x=10
>>> p1.y=20

>>> import copy #import module copy

>>> p3=copy.copy(p1) #use the method copy()
>>> print(p1)
< main .Point object at 0x01581BF0>
>>> print(p3)
< main .Point object at 0x02344A50>
>>> print(p3.x,p3.y) 10 20

Observe that the physical address of the objects p1 and p3 are now different. But, values of
attributes x and y are same. Now, use the following statements –

>>> p1 is p3 False
>>> p1 == p3
False

Here, the is operator gives the result as False for the obvious reason of p1 and p3 are
being two different entities on the memory. But, why == operator is generating False as the
result, though the contents of two objects are same? The reason is – p1 and p3 are the
objects of user-defined type. And, Python cannot understand the meaning of equality on the
new data type. The default behavior of equality (==) is identity (is operator) itself. Hence,
Python applies this default behavior on p1 == p3 and results in False.
 PYTHON APPLICATION PROGRAMMING [15CS664]

(NOTE: If we need to define the meaning of equality (==) operator explicitly on user-defined
data types (i.e. on class objects), then we need to override the method eq () inside the
class. This will be discussed later in detail.)
The copy() method of copy module duplicates the object. The content (i.e. attributes) of one
object is copied into another object as we have discussed till now. But, when an object itself
is an attribute inside another object, the duplication will result in a strange manner. To
understand this concept, try to copy Rectangle object (created in previous section) as given
below –

import copy
class Point:
""" This is a class Point
representing coordinate point
"""
class Rectangle:
""" This is a class Rectangle.
Attributes: width, height and Corner Point
"""
box1=Rectangle()
box1.corner=Point()
box1.width=100
box1.height=200
box1.corner.x=0
box1.corner.y=0

box2=copy.copy(box1)
print(box1 is box2) #prints False
print(box1.corner is box2.corner) #prints True

Now, the question is – why box1.corner and box2.corner are same objects, when box1
and box2 are different? Whenever the statement

box2=copy.copy(box1)

is executed, the contents of all the attributes of box1 object are copied into the respective
attributes of box2 object. That is, box1.width is copied into box2.width, box1.height
is copied into box2.height. Similarly, box1.corner is copied into box2.corner. Now,
recollect the fact that corner is not exactly the object itself, but it is a reference to the object
of type Point (Read the discussion done for Figure 4.1 at the beginning of this Chapter).
Hence, the value of reference (that is, the physical address) stored in box1.corner is
copied into box2.corner. Thus, the physical object to which box1.corner and
 PYTHON APPLICATION PROGRAMMING [15CS664]

box2.corner are pointing is only one. This type of copying the objects is known as shallow
copy. To understand this behavior, observe the following diagram –

Rectangle Rectangle

box1 width 100 width 100 box2

Point
height 200 height 200
x 0
corner y 0 corner

Now, the attributes width and height for two objects box1 and box2 are independent.
Whereas, the attribute corner is shared by both the objects. Thus, any modification done to
box1.corner will reflect box2.corner as well. Obviously, we don’t want this to happen,
whenever we create duplicate objects. That is, we want two independent physical objects.
Python provides a method deepcopy() for doing this task. This method copies not only the
object but also the objects it refers to, and the objects they refer to, and so on.

box3=copy.deepcopy(box1)
print(box1 is box3) #prints False
print(box1.corner is box3.corner) #prints False

4.1.5 Debugging
While dealing with classes and objects, we may encounter different types of errors. For
example, if we try to access an attribute which is not there for the object, we will get
AttributeError. For example –

>>> p= Point()
>>> p.x = 10
>>> p.y = 20
>>> print(p.z)
AttributeError: 'Point' object has no attribute 'z'

To avoid such error, it is better to enclose such codes within try/except as given below –
try:
z = p.x
except AttributeError:
z = 0
 PYTHON APPLICATION PROGRAMMING [15CS664]

When we are not sure, which type of object it is, then we can use type() as –
>>> type(box1)
<class ' main .Rectangle'>

Another method isinstance() helps to check whether an object is an instance of a particular

class –
>>> isinstance(box1,Rectangle) True

When we are not sure whether an object has a particular attribute or not, use a function
hasattr() –
>>> hasattr(box1, 'width') True

Observe the string notation for second argument of the function hasattr(). Though the
attribute width is basically numeric, while giving it as an argument to function hasattr(), it
must be enclosed within quotes.

4.2 CLASSES AND FUNCTIONS

Though Python is object oriented programming languages, it is possible to use it as functional
programming. There are two types of functions viz. pure functions and modifiers. A pure
function takes objects as arguments and does some work without modifying any of the
original argument. On the other hand, as the name suggests, modifier function modifies the
original argument.

In practical applications, the development of a program will follow a technique called as

prototype and patch. That is, solution to a complex problem starts with simple prototype
and incrementally dealing with the complications.

4.2.1 Pure Functions

To understand the concept of pure functions, let us consider an example of creating a class
called Time. An object of class Time contains hour, minutes and seconds as attributes. Write
a function to print time in HH:MM:SS format and another function to add two time objects.
Note that, adding two time objects should yield proper result and hence we need to check
whether number of seconds exceeds 60, minutes exceeds 60 etc, and take appropriate
action.

class Time:
"""Represents the time of a day
Attributes: hour, minute, second """
def printTime(t):
print("%.2d:%.2d:%.2d"%(t.hour,t.minute,t.second))
def add_time(t1,t2):
sum=Time()
 PYTHON APPLICATION PROGRAMMING [15CS664]

sum.hour = t1.hour + t2.hour

sum.minute = t1.minute + t2.minute
sum.second = t1.second + t2.second
if sum.second >= 60:
sum.second -= 60
sum.minute += 1
if sum.minute >= 60:
sum.minute -= 60
sum.hour += 1
return sum
t1=Time()
t1.hour=10
t1.minute=34
t1.second=25
print("Time1 is:")
printTime(t1)
t2=Time()
t2.hour=2
t2.minute=12
t2.second=41
print("Time2 is :")
printTime(t2)
t3=add_time(t1,t2)
print("After adding two time objects:")
printTime(t3)

The output of this program would be –

Time1 is: 10:34:25

Time2 is : 02:12:41
After adding two time objects: 12:47:06

Here, the function add_time() takes two arguments of type Time, and returns a Time
object, whereas, it is not modifying contents of its arguments t1 and t2. Such functions are
called as pure functions.

4.2.2 Modifiers
Sometimes, it is necessary to modify the underlying argument so as to reflect the caller. That
is, arguments have to be modified inside a function and these modifications should be
available to the caller. The functions that perform such modifications are known as modifier
function. Assume that, we need to add few seconds to a time object, and get a new time.
Then, we can write a function as below –
 PYTHON APPLICATION PROGRAMMING [15CS664]

def increment(t, seconds):

t.second += seconds
while t.second >= 60:
t.second -= 60
t.minute += 1
while t.minute >= 60:
t.minute -= 60
t.hour += 1

In this function, we will initially add the argument seconds to t.second. Now, there is a
chance that t.second is exceeding 60. So, we will increment minute counter till t.second
becomes lesser than 60. Similarly, till the t.minute becomes lesser than 60, we will
decrement minute counter. Note that, the modification is done on the argument t itself. Thus,
the above function is a modifier.

4.2.3 Prototyping v/s Planning

Whenever we do not know the complete problem statement, we may write the program
initially, and then keep of modifying it as and when requirement (problem definition) changes.
This methodology is known as prototype and patch. That is, first design the prototype based
on the information available and then perform patch-work as and when extra information is
gathered. But, this type of incremental development may end-up in unnecessary code, with
many special cases and it may be unreliable too.

An alternative is designed development, in which high-level insight into the problem can
make the programming much easier. For example, if we consider the problem of adding two
time objects, adding seconds to time object etc. as a problem involving numbers with base
60 (as every hour is 60 minutes and every minute is 60 seconds), then our code can be
improved. Such improved versions are discussed later in this chapter.

4.2.4 Debugging
In the program written in Section 4.2.1, we have treated time objects as valid values. But,
what if the attributes (second, minute, hour) of time object are given as wrong values like
negative number, or hours with value more than 24, minutes/seconds with more than 60 etc?
So, it is better to write error-conditions in such situations to verify the input. We can write a
function similar to as given below –

def valid_time(time):
if time.hour < 0 or time.minute < 0 or time.second < 0:
return False
if time.minute >= 60 or time.second >= 60:
return False
return True
 PYTHON APPLICATION PROGRAMMING [15CS664]

Now, at the beginning of add_time() function, we can put a condition as –

def add_time(t1, t2):

if not valid_time(t1) or not valid_time(t2):
raise ValueError('invalid Time object in add_time')

#remaining statements of add_time() functions

Python provides another debugging statement assert. When this keyword is used, Python
evaluates the statement following it. If the statement is True, further statements will be
evaluated sequentially. But, if the statement is False, then AssertionError exception is
raised. The usage of assert is shown here –

def add_time(t1, t2):

assert valid_time(t1) and valid_time(t2)
#remaining statements of add_time() functions
The assert statement clearly distinguishes the normal conditional statements as a part of the
logic of the program and the code that checks for errors.

4.3 CLASSES AND METHODS

The classes that have been considered till now were just empty classes without having any
definition. But, in a true object oriented programming, a class contains class-level attributes,
instance-level attributes, methods etc. There will be a tight relationship between the object of
the class and the function that operate on those objects. Hence, the object oriented nature
of Python classes will be discussed here.

4.3.1 Object-Oriented Features

As an object oriented programming language, Python possess following characteristics:
• Programs include class and method definitions.
• Most of the computation is expressed in terms of operations on objects.
• Objects often represent things in the real world, and methods often correspond to the
ways objects in the real world interact.

To establish relationship between the object of the class and a function, we must define a
function as a member of the class. A function which is associated with a particular class is
known as a method. Methods are semantically the same as functions, but there are two
syntactic differences:
• Methods are defined inside a class definition in order to make the relationship between
the class and the method explicit.
• The syntax for invoking a method is different from the syntax for calling a function.

Now onwards, we will discuss about classes and methods.

 PYTHON APPLICATION PROGRAMMING [15CS664]

4.3.2 The init () Method

(A method init () has to be written with two underscores before and after the word init)

Python provides a special method called as init () which is similar to constructor

method in other programming languages like C++/Java. The term init indicates initialization.
As the name suggests, this method is invoked automatically when the object of a class is
created. Consider the example given here –

import math
class Point:
def init (self,a,b):
self.x=a
self.y=b
def dist(self,p2):
d=math.sqrt((self.x-p2.x)**2 + (self.y-p2.y)**2)
return d
def str (self):
return "(%d,%d)"%(self.x, self.y)

p1=Point(10,20) # init () is called automatically

p2=Point(4,5) # init () is called automatically

print("P1 is:",p1) # str () is called automatically

print("P2 is:",p2) # str () is called automatically

d=p1.dist(p2) #explicit call for dist()

print("The distance is:",d)

The sample output is –

P1 is: (10,20)
P2 is: (4,5)
Distance is: 16.15549442140351

Let us understand the working of this program and the concepts involved:
• Keep in mind that every method of any class must have the first argument as self.
The argument self is a reference to the current object. That is, it is reference to the
object which invoked the method. (Those who know C++, can relate self with this
pointer). The object which invokes a method is also known as subject.
 PYTHON APPLICATION PROGRAMMING [15CS664]

• The method init () inside the class is an initialization method, which will be
invoked automatically when the object gets created. When the statement like –
p1=Point(10,20)

is used, the init () method will be called automatically. The internal meaning
of the above line is –
p1. init (10,20)

Here, p1 is the object which is invoking a method. Hence, reference to this object is
created and passed to init () as self. The values 10 and 20 are passed to
formal parameters a and b of init () method. Now, inside init ()
method, we have statements

self.x=10
self.y=20

This indicates, x and y are instance attributes. The value of x for the object p1 is 10
and, the value of y for the object p1 is 20.

When we create another object p2, it will have its own set of x and y. That is, memory
locations of instance attributes are different for every object.

Thus, state of the object can be understood by instance attributes.

• The method dist() is an ordinary member method of the class Point. As mentioned
earlier, its first argument must be self. Thus, when we make a call as –
d=p1.dist(p2)
a reference to the object p1 is passed as self to dist() method and p2 is
passed explicitly as a second argument. Now, inside the dist()method, we are
calculating distance between two point (Euclidian distance formula is used) objects.
Note that, in this method, we cannot use the name p1, instead we will use self
which is a reference (alias) to p1.

• The next method inside the class is str (). It is a special method used for string
representation of user-defined object. Usually, print() is used for printing basic
types in Python. But, user-defined types (class objects) have their own meaning and
a way of representation. To display such types, we can write functions or methods like
print_point() as we did in Section 4.1.2. But, more polymorphic way is to use
str () so that, when we write just print() in the main part of the program, the
str () method will be invoked automatically. Thus, when we use the statement
like –
print("P1 is:",p1)
 PYTHON APPLICATION PROGRAMMING [15CS664]

the ordinary print() method will print the portion “P1 is:” and the remaining portion is
taken care by str () method. In fact, str () method will return the string
format what we have given inside it, and that string will be printed by print()
method.

4.3.3 Operator Overloading

Ability of an existing operator to work on user-defined data type (class) is known as
operator overloading. It is a polymorphic nature of any object oriented programming. Basic
operators like +, -, * etc. can be overloaded. To overload an operator, one needs to write a
method within user-defined class. Python provides a special set of methods which have to
be used for overloading required operator. The method should consist of the code what the
programmer is willing to do with the operator. Following table shows gives a list of operators
and their respective Python methods for overloading.

Operator Special Function Operator Special Function

in Python in Python

+ add () <= le ()

- sub () >= ge ()

V .com
T Pulse
*

/
mul

truediv
()

()
==

!=
eq

ne
()

()

% mod () in contains ()

< lt () len len ()

> gt () str str ()

Let us consider an example of Point class considered earlier. Using operator overloading, we
can try to add two point objects. Consider the program given below –

class Point:
def init (self,a=0,b=0):
self.x=a
self.y=b
def add (self, p2):
p3=Point()
p3.x=self.x+p2.x
 PYTHON APPLICATION PROGRAMMING [15CS664]

p3.y=self.y+p2.y
return p3
def str (self):
return "(%d,%d)"%(self.x, self.y)

p1=Point(10,20)
p2=Point(4,5)
print("P1 is:",p1)
print("P2 is:",p2)
p4=p1+p2 #call for add () method
print("Sum is:",p4)

The output would be –

P1 is: (10,20)
P2 is: (4,5)
Sum is: (14,25)

In the above program, when the statement p4 = p1+p2 is used, it invokes a special method
add () written inside the class. Because, internal meaning of this statement is–
p4 = p1. add (p4)
Here, p1 is the object invoking the method. Hence, self inside add () is the
reference (alias) of p1. And, p4 is passed as argument explicitly.

In the definition of add (), we are creating an object p3 with the statement –
p3=Point()

The object p3 is created without initialization. Whenever we need to create an object with
and without initialization in the same program, we must set arguments of init () for
some default values. Hence, in the above program arguments a and b of init () are
made as default arguments with values as zero. Thus, x and y attributes of p3 will be now
zero. In the add () method, we are adding respective attributes of self and p2 and
storing in p3.x and p3.y. Then the object p3 is returned. This returned object is received
as p4 and is printed.

NOTE that, in a program containing operator overloading, the overloaded operator behaves
in a normal way when basic types are given. That is, in the above program, if we use the
statements
m= 3+4
print(m)

it will be usual addition and gives the result as 7. But, when user-defined types are used as
operands, then the overloaded method is invoked.
 PYTHON APPLICATION PROGRAMMING [15CS664]

Let us consider a more complicated program involving overloading. Consider a problem of

creating a class called Time, adding two Time objects, adding a number to Time object etc.
that we had considered in previous section. Here is a complete program with more of OOP
concepts.
class Time:
def init (self, h=0,m=0,s=0):
self.hour=h
self.min=m
self.sec=s
def time_to_int(self):
minute=self.hour*60+self.min
seconds=minute*60+self.sec
return seconds
def int_to_time(self, seconds):
t=Time()
minutes, t.sec=divmod(seconds,60)
t.hour, t.min=divmod(minutes,60)
return t
def str (self):
return "%.2d:%.2d:%.2d"%(self.hour,self.min,self.sec)
def eq (self,t):
return self.hour==t.hour and self.min==t.min and self.sec==t.sec
def add (self,t):
if isinstance(t, Time):
return self.addTime(t)
else:
return self.increment(t)
def addTime(self, t):
seconds=self.time_to_int()+t.time_to_int()
return self.int_to_time(seconds)
def increment(self, seconds):
seconds += self.time_to_int()
return self.int_to_time(seconds)
def radd (self,t):
return self. add (t)

T1=Time(3,40)
T2=Time(5,45)
print("T1 is:",T1)
print("T2 is:",T2)
print("Whether T1 is same as T2?",T1==T2) #call for eq ()
T3=T1+T2 #call for add ()
print("T1+T2 is:",T3)
 PYTHON APPLICATION PROGRAMMING [15CS664]

T4=T1+75 #call for add ()

print("T1+75=",T4)

T5=130+T1 #call for radd ()

print("130+T1=",T5)

T6=sum([T1,T2,T3,T4])
print("Using sum([T1,T2,T3,T4]):",T6)

The output would be –

T1 is: 03:40:00
T2 is: 05:45:00
Whether T1 is same as T2? False
T1+T2 is: 09:25:00
T1+75= 03:41:15
130+T1= 03:42:10
Using sum([T1,T2,T3,T4]): 22:31:15

Working of above program is explained hereunder –

• The class Time has init () method for initialization of instance attributes hour,
min and sec. The default values of all these are being zero.
• The method time_to_int() is used convert a Time object (hours, min and sec) into
single integer representing time in number of seconds.
• The method int_to_time() is written to convert the argument seconds into time
object in the form of hours, min and sec. The built-in method divmod() gives the
quotient as well as remainder after dividing first argument by second argument given
to it.
• Special method eq () is for overloading equality (==) operator. We can say one
Time object is equal to the other Time object if underlying hours, minutes and seconds
are equal respectively. Thus, we are comparing these instance attributes individually
and returning either True of False.
• When we try to perform addition, there are 3 cases – o Adding two time objects like
T3=T1+T2.
o Adding integer to Time object like T4=T1+75 o
Adding Time object to an integer like T5=130+T1

Each of these cases requires different logic. When first two cases are considered, the
first argument will be T1 and hence self will be created and passed to add ()
method. Inside this method, we will check the type of second argument using
isinstance() method. If the second argument is Time object, then we call
addTime() method. In this method, we will first convert both Time objects to integer
(seconds) and then the resulting sum into Time object again. So, we make use
 PYTHON APPLICATION PROGRAMMING [15CS664]

time_to_int() and int_to_time() here. When the 2nd argument is an integer,

it is obvious that it is number of seconds. Hence, we need to call increment()
method.

Thus, based on the type of argument received in a method, we take appropriate action.
This is known as type-based dispatch.

In the 3rd case like T5=130+T1, Python tries to convert first argument 130 into self,
which is not possible. Hence, there will be an error. This indicates that for Python,
T1+5 is not same as 5+T1 (Commutative law doesn’t hold good!!). To avoid the
possible error, we need to implement right-side addition method radd ().
Inside this method, we can call overloaded method add ().
• The beauty of Python lies in surprising the programmer with more facilities!! As we
have implemented add () method (that is, overloading of + operator), the builtin
sum() will is capable of adding multiple objects given in a sequence. This is due to
Polymorphism in Python. Consider a list containing Time objects, and then call
sum() on that list as –

T6=sum([T1,T2,T3,T4])
The sum() internally calls add () method multiple times and hence gives the
appropriate result. Note down the square-brackets used to combine Time objects as
a list and then passing it to sum().

Thus, the program given here depicts many features of OOP concepts.

4.3.4 Debugging
We have seen earlier that hasattr() method can be used to check whether an object has
particular attribute. There is one more way of doing it using a method vars(). This method
maps attribute names and their values as a dictionary. For example, for the Point class
defined earlier, use the statements –
>>> p = Point(3, 4)
>>> vars(p) #output is {'y': 4, 'x': 3}

For purposes of debugging, you might find it useful to keep this function handy:

def print_attributes(obj):
for attr in vars(obj):
print(attr, getattr(obj, attr))

Here, print_attributes() traverses the dictionary and prints each attribute name and
its corresponding value. The built-in function getattr() takes an object and an attribute
name (as a string) and returns the attribute’s value.