Learn the magic of writing programs that monitor, alter and react to the execution of program code by responding to imports, changes to variables, calls to functions and invocations of the builtins. This talk goes beyond the static world of metaclasses and class decorators.

We'll cover how to slide a class underneath a module to intercept reads/writes, place automatic type checking over your object attributes and use stack peeking to make selected attributes private to their owning class. We'll cover import hacking, metaclasses, descriptors and decorators and graphically describe how they work internally. Source examples and color technical diagrams.

Table-of-Contents

• What is Metaprogramming?
• Tools At Our Disposal
• Orientation Diagram: What is Metaprogramming

First Third of Talk: Import Hooking

• Sample Problem #1: Subclassing an Embedded Class
• A Solution to #1: Post-Import Hooking
• A Solution to #1 (Packaged Up)
• Alternate Solution: Pre-Import Hooking
• What Does a Subclassed Module Look Like?
• Some Benefits of Subclassing Modules

2nd Third of Talk: Metaclasses

• Orientation Diagram: Instances, Classes and Metaclasses
• Facts About Metaclasses
• Example #2: Define a Class from an SQL Table Definition
• Example Problem #2 (cont'd)
• Metaclasses versus Class Decorators
• About Meta-Inheritance
• Example #3: Log the Arguments/Return Value of Method Calls
• Lull After Metaclasses, Before Descriptors

Last Third of Talk: Descriptors

• Python's Mechanism of Attribute Lookup
• When to Use Which Lookup Mechanism
• Example 4: Overriding getattr
• Example 4: Using a Descriptor Instead
• Python's Mechanism of Attribute Lookup (descriptors)
• So What is a descriptor again?
• Where are descriptors used?
• Example 5: Caching an Attribute Value
• Example 6: Declare an Attribute Private to a Class
• Example 7: Tracking Changes in a Value