[Python-tut] Classes: constructors, access, inheritance

[wtd]

Let's construct a simple Name class as a demonstration. The class should have a constructor which takes two arguments. These are the first and last names. These become instance variables.

code:

class Name:    def __init__(self, first, last):       self.first = first       self.last = last

You'll notice the extra "self" argument in the constructor. All methods in a Python class take this argument, and it refers back to the object itself. You can call it anything you'd like, but "self" is the most common convention.

As with everything Python, indentation denotes scope.

Now, we can create a Name object and manipulate its instance variables.

code:

>>> class Name: ...    def __init__(self, first, last): ...       self.first = first ...       self.last = last ... >>> bob = Name("Bob", "Smith") >>> bob.first 'Bob' >>> bob.last = "Johnson" >>> bob.last 'Johnson' >>>

Notice those last few lines. I was able to directly manipulate and change the last name quite easil from outside of the objecty. In object-oriented programming that's a big no-no. We generally want to encapsulate our data, and only allow the outside world to change it in very specific ways.

So, how do we make the instance variable private?

In Python this is accomplished by prepending the variable (or method) name with two underscores.

code:

>>> class Name: ...    def __init__(self, first, last): ...       self.__first = first ...       self.__last = last ... >>> bob = Name("Bob", "Smith") >>> bob.__first Traceback (most recent call last):   File "<stdin>", line 1, in ? AttributeError: Name instance has no attribute '__first' >>> bob.__last Traceback (most recent call last):   File "<stdin>", line 1, in ? AttributeError: Name instance has no attribute '__last' >>>

The last few lines are clearly showing that trying to access these instance variables from outside of the object is now impossible. Beautiful.

Now, let's add a method to get a representation of the name as a string.

code:

class Name:    def __init__(self, first, last):       self.first = first       self.last = last    def to_string(self):       return self.__first + " " + self.__last

The only problem with this is that it doesn't follow Python convention. So instead we'll write the following, and

code:

class Name:    def __init__(self, first, last):       self.__first = first       self.__last = last    def __str__(self):       return self.__first + " " + self.__last

code:

>>> bob = Name("Bob", "Smith") >>> str(bob) 'Bob Smith' >>> print bob Bob Smith >>>

Inheritance is also critical to object-oriented programming. First off, let's create a NameWithTitle class that inherits from Name and create a new constructor that uses the base class constructor.

code:

class NameWithTitle(Name):    def __init__(self, title, first, last):       Name.__init__(self, first, last)       self.__title = title

Now we can create a new NameWithTitle object.

code:

>>> bob = NameWithTitle("Dr.", "Bob", "Smith") >>> str(bob) 'Bob Smith' >>> print bob Bob Smith >>>

You'll notice that it doesn't look like anything different has happened. Well, you'd be right. We have to override the __str__ method for that to happen.

code:

class NameWithTitle(Name):    def __init__(self, title, first, last):       Name.__init__(self, first, last)       self.__title = title    def __str__(self):       return self.__title + " " + Name.__str__(self)

And now, using it.

code:

>>> bob = NameWithTitle("Dr.", "Bob", "Smith") >>> str(bob) 'Dr. Bob Smith' >>> print bob Dr. Bob Smith >>>

[wtd]

Determining at run-time if an object is of a particular type is quite easy. For this we use the isinstance function.

code:

>>> isinstance("hello", str) True >>> isinstance([1,2,3], list) True >>> isinstance({'f': 'g'}, list) False >>>

Further, we can get the name of an object's type with the type function.

code:

>>> type(42) <type 'int'> >>>