Functional Extensions

for Python

https://pypi.org/project/funext/

This repository aims to extend the Python programming language by adding some useful (functional) extensions to objects, functions and data structures. Some of those extensions include:

• being able to chain pure (and mutating) functions by calling them directly on objects. Most of those functions are known from the built-in functions.

• applying functions to objects and chaining them through a pipe-function

• composing functions

Usage

from functional_extensions import l_, s_, d_, t_, f_

regular_list = [1, 2, 3, 4]
extended_list = l_(1, 2, 3, 4)

# all extended classes have a simple initializer function. Containers can either
# take individual values as *args, or a regular collection
extended_list = l_([1, 2, 3, 4]) # or
extended_list = l_(1, 2, 3, 4)

extended_set = s_(1, 2, 3, 4) # or
extended_set = s_(set(1, 2, 3, 4))

extended_dict = d_({'key': 'value'})

extended_tuple = t_(1, 2, 3, 4) # or
extended_tuple = t_((1, 2, 3, 4))

def add(a, b): return a + b
extended_function = f_(add)

Available functions

Object

pipe_(self, function, *args, **kwargs)

Applies the object self to a function function with the arguments from *args and **kwarngs and returns the result.

---

copy_(self) 
deepcopy_(self)

Copies self, either shallowly or deeply, and returns the result.

---

type_(self)

Returns type(self).

Sequences and Containers

Iterable

to_list_(self)
to_set_(self)
to_tuple_(self)

Converts the iterable self into the desired object.

---

map_(self, function, *args, **kwargs)

Applies every element of self to function and returns an iterable of the new elements. Passes *args and **kwargs to the map-function.

Returns an object of the same type as that of self.

---

for_each_(self, apply, *args, **kwargs)

For every element in self, function is called with said element, as well as *args and **kwargs. The list is then returned.

---

min_(self)
max_(self)
sum_(self)
all_(self)
any_(self)

Returns min(self), max(self), sum(self), all(self) and any(self) respectively.

---

sort_(self, key=None, reverse=False)

Sorts all elements from the iterable self in a new list and returns it. Calls the sorted-function under the hood with key and reverse.

---

enumerate_(self, start=0)

Returns an enumerate object from the iterable self.

---

zip_(self, *iterables)

Iterates over self and all iterables in *iterables, producing tuples with an item from each one.

---

filter_(self, condition)
filterfalse_(self, condition)

filter_ returns a new instance of this iterable with only the elements that condition returns true.

filterfalse_returns a new instance of this iterable with only the elements that condition returns false.

Reversible

reverse_(self)

Returns a new instance of the reversed iterable self.

Sized

len_(self)

Returns len(self)

MutableSequence

map_inplace_(self, apply, *args, **kwargs)

Applies every element of self to function and overwrites this element with the new value.

List

sort_inplace_(self, key=None, reverse=False)

Sorts the list in-place and returns the sorted list

Functions

compose_(self, function)

Composes a function

---

and_(self, other_function)
or_(self, other_function)
not_(self)

Composes two predicates.

---

partial_(self, *args, **kwargs)

Partially applys *args, and **kwargs to self and returns the partial function.

---

curry_(self, n=EMPTY)

Curries self.

---

complement_(self)

Cunstructs a negation of self.

---

all_fn_(self, *fs)
any_fn_(self, *fs)
none_fn_(self, *fs)
one_fn_(self, *fs)

Construct a predicate returning True when all, any, none or exactly one of self and fs return True.

Examples

Most of those functions should not need additional examples, as they are a mere re-phrasing of some basic concepts and functions of the Python programming language and funcy.

If you need an example anyway, you should consider looking at the test classes, which cover every function.

The why

An excerpt of the "zen of python":

Beautiful is better than ugly

A simple exercise. Initialize a list with the numbers [4, 1, 2, 3], take the negative square of these numbers, sort the list and print every element individually.

Which one of the following code pieces is more beautiful?

input = [4, 1, 2, 3]
squared = [-(x ** 2) for x in input]
squared.sort()
for element in squared:
  print(x)

def negative_square(x): return - (x ** 2)

l_(4, 1, 2, 3) \
    .map_inplace_(negative_square) \
    .sort_() \
    .for_each_(print) 

There are 4 things happening. In the top example, all things are expressed via a slightly different syntax. Initialization with a list literal, mapping by a list comprehension, calling the sort-function on the list and then printing all elements

In the bottom example, every "thing" that is happening, is expressed in a coherent way - by calling a function.

Requirement	Corresponding Function
Initialize a list with the numbers [4, 1, 2, 3]	l_(4, 1, 2, 3)
Take the negative square	.map_inplace_(negative_square) 1
Sort the list	.sort_()
Print every element individually	.for_each_(print)

¹ we could also use a lambda-function in-place to avoid having to declare an own function for this case. However, the code might become slightly less expressive. Which brings us to the next "principle"

Explicit is better than implicit

This principle means, that there shouldn't be anything unexpected happening under the hood, which is often the case in high-level codebases. Metaprogramming, making use of inheritance or just not choosing proper variable names can make the code hard to read and understand, making it more easy to make mistakes.

These extensions might make things more complex at first glance, but the naming is chosen quite carefully. If a function name corresponds to a name of a built-in function, this functions will not do more or less than exactly that function.

All other functions are designed to be pure, and not cause anything to happen outside of their little scope.

All function that are not pure, are explicitly named like that, for example map_inplace_

All functions that have similar names than built-in functions, have an explicit "_"-postfix to denote that they do things differently, for example sort_, which adheres to the rule that functions are pure, unlike the sort()- function from the builtin list-class.

Simple is better than complex

and

Readability counts

One might argue, that any code which doesn't directly solves a certain use case is, by definition, not simple. It might be correct to define complexity like that. But you can also define complexity by how much time it takes you to read and understand code. This library tries to help reduce complexity of the code itself, on a less abstract layer than the use-case-specific solutions are written in. By that, hopefully those solutions also become less complex.

There should be one-- and preferably only one --obvious way to do it.

Yes, in an attempt to fulfill other principles, this principle is arguably broken. Now there is one more way to write a for-loop or call a map-function. At least I try to keep this rule fulfilled within this library. There should be one obvious way to do one thing by using functional extensions.