Fluent Python

Chapter 9. A Pythonic Object

Never, ever use two leading underscores. This is annoyingly private.¹

Ian Bicking, Creator of pip, virtualenv, Paste and many other projects

Thanks to the Python data model, your user-defined types can behave as naturally as the built-in types. And this can be accomplished without inheritance, in the spirit of duck typing: you just implement the methods needed for your objects to behave as expected.

In previous chapters, we presented the structure and behavior of many built-in objects. We will now build user-defined classes that behave as real Python objects.

This chapter starts where Chapter 1 ended, by showing how to implement several special methods that are commonly seen in Python objects of many different types.

In this chapter, we will see how to:

• Support the built-in functions that produce alternative object representations (e.g., repr(), bytes(), etc).

• Implement an alternative constructor as a class method.

• Extend the format mini-language used by the format() built-in and the str.format() method.

• Provide read-only access to attributes.

• Make an object hashable for use in sets and as dict keys.

• Save memory with the use of __slots__.

We’ll do all that as we develop a simple two-dimensional Euclidean vector type.

The evolution of the example will be paused to discuss two conceptual topics:

• How and when to use the @classmethod and @staticmethod decorators.

• Private and protected attributes in Python: usage, conventions, and limitations. ...