Preface
Programming languages sometimes fit neatly into tidy categories like imperative and functional. Imperative languages might further subdivide into those that are procedural and those that include features for object-oriented programming. The Python language, however, contains aspects of all of these three language categories. Though Python is not a purely functional programming language, we can do a great deal of functional programming in Python.
Most importantly, we can leverage many design patterns and techniques from other functional languages and apply them to Python programming. These borrowed concepts can lead us to create succinct and elegant programs. Python's generator expressions, in particular, avoid the need to create large in-memory data structures, leading to programs which may execute more quickly because they use fewer resources.
We can't easily create purely functional programs in Python. Python lacks a number of features that would be required for this. For example, we don't have unlimited recursion, lazy evaluation of all expressions, and an optimizing compiler.
Generally, Python emphasizes strict evaluation rules. This means that statements are executed in order and expressions are evaluated from left to right. While this deviates from functional purity, it allows us to perform manual optimizations when writing in Python. We'll take a hybrid approach to functional programming using Python's functional features when they can add clarity or simplify the code and use ordinary imperative features for optimization.
There are several key features of functional programming languages that are available in Python. One of the most important is the idea that functions are first-class objects. In some languages, functions exist only as a source code construct: they don't exist as proper data structures at runtime. In Python, functions can use functions as arguments and return functions as results.
Python offers a number of higher-order functions. Functions like map(), filter(), and functools.reduce() are widely used in this role. Less obvious functions like sorted(), min(), and max() are also higher-order functions; they have a default function and, consequently, different syntax from the more common examples.
Functional programs often exploit immutable data structures. The emphasis on stateless objects permits flexible optimization. Python offers tuples and namedtuples as complex but immutable objects. We can leverage these structures to adapt some design practices from other functional programming languages.
Many functional languages emphasize recursion but exploit Tail-Call Optimization (TCO). Python tends to limit recursion to a relatively small number of stack frames. In many cases, we can think of a recursion as a generator function. We can then simply rewrite it to use a yield from statement, doing the tail-call optimization ourselves.
We'll look at the core features of functional programming from a Python point of view. Our objective is to borrow good ideas from functional programming languages, and use these ideas to create expressive and succinct applications in Python.
