Decorators

A decorator is a function that takes another function as input and returns a new function that wraps it - typically adding behavior before and/or after the original runs, without modifying the original function's own code. This directly builds on the closures chapter from the Functions module: a decorator is a closure whose "remembered" enclosing variable happens to be the function it is wrapping.

Before the `@` syntax, look at what a decorator actually is: a function that accepts a function and returns a new function.

Notice `wrapper` takes `*args, **kwargs` - exactly the "accept anything" tool from the earlier functions module - so `log_calls` can wrap a function with any signature at all, not just ones that take two arguments.

`@decorator_name` placed directly above a function definition is exactly equivalent to the manual reassignment above - `def add(...): ...` followed by `add = log_calls(add)` - just cleaner to read.

Another classic, genuinely useful decorator: measuring how long a function takes to run, using the `time` module.

Here is the payoff for everything in this chapter and the closures chapter before it. What if, instead of timing a function, a wrapper checked whether it had already computed the result for these exact arguments, and returned the cached answer instead of recomputing it?

• -A decorator is a function that takes a function and returns a new (usually wrapping) function.
• -`@decorator_name` above a `def` is sugar for `func = decorator_name(func)`.
• -A wrapper almost always accepts `*args, **kwargs` so it can wrap functions of any signature.
• -A caching decorator is a closure holding a dict, checked and updated on every call - this is memoization, which the next phase of this course (Algorithms) builds on directly.