One of the most important things that programmers should keep in mind is "do not repeat yourself". This means that programmers should not repeat the same code again - in fact they must re-use the code. Programmers must look for an elegant solution when they faced any problem of creating highly repetitive code. In Python, this problem can be solved using the concept of meta-programming.
Meta-programming is the concept of building functions and classes whose main target is to manipulate code by modifying, wrapping or generating existing code.
The major features of meta-programming are:
There is also a varied range of useful topics that costs a huge demand when coming under the umbrella of meta-programming such as signature objects, check for internal classes and functions, execution of 'exec ()' codes etc. The default/base metaclass is called "type"; in some cases, programmers can gain control by manipulating or modifying the way classes are constructed - by adding extra accomplishment or injecting extra codes along with it. So, in these types of cases, programmers can use metaclass-programming some of the class-objects that already exist.
In other words, classes are just objects so programmers can introduce the concept of meta-classes into their practice to customize their creation.
Putting wrapper for a function:
Programmers can get the facility to add wrapper as a layer around a function to add extra processing capabilities such as timing, logging etc. The simple code to do this is:
import time from functools import wraps def karltime(func): \#Decorator for reporting the execution time. \@wraps(func) def wrapper(\*args, \*\*kwargs): start = time.time() result = func(\*args, \*\*kwargs) end = time.time() print(func.__name__, end-start) return result return wrapper
Here is a program (connected with the previous program) segment that is using simple decorator
\@karltime def countdown(n): while n \> 0: n -= 1
The decorator in Python's meta-programming is a special form of a function that takes functions as input and returns a new function as output.
There are some built-in decorators viz:
Working of Metaclasses:
A metaclass can be any class that gets inherited from "type". It usually overrides either _new_ or _init_ to offer customized behavior. A custom-metaclass can be placed explicitly on a class like this:
_metaclass_ = karlMeta
Metaclass creates class; normally when we build class by writing the word 'class', the default metaclass "type" gets automatically invoked to build that class. Classes are often referred to as types and are fairly sensible. Programmers can also add base-classes, fields, and methods. Program showing how they are used:
def karl(self, myName): print("RAY, " + myName) MyList = type('MyList', (list,), dict(x=62, karl=karl)) ml = MyList() ml.append("Mango") print(ml) print(ml.x) ml.karl("KARLOS") print(ml.__class__.__class__)
Until now, we have used the "type" metaclass directly; it involves hooking won operation into a creation of class objects and can be achieved by:
- Writing a sub-class of the metaclass.
- Inserting new metaclass using the 'metaclass hook'.