Functional programming sugar for Python
Functional programming sugar for Python.
A simple pip command should do the trick. (In some cases, you may need to use sudo)
pip install alakazam
The Alakazam library does not, as of now, have any additional dependencies and is designed for Python 2 and 3.
To use the stream functionality of Alakazam, import the alakazam package. It is recommended that you alias the package as something like zz for easier typing.
To use the Alakazam lambda syntax, import the placeholders from alakazam explicitly.
import alakazam as zz from alakazam import _1, _2, _3, _4, _5
This library aims to make functional-style, and specifically stream-oriented programming in Python prettier, more pleasant, and easier on the eyes. Python has been capable of many functional programming tasks, but it has always been a little bit awkward to use those features for anything nontrivial. For instance, suppose we had some list arr, and we wanted to square every element of the list and then keep only the even squares. This is how we might approach this problem using Python’s built-in functional tools.
list(filter(lambda x: x % 2 == 0, map(lambda x: x ** 2, arr)))
There’s a lot of cruft here, with having to explicitly declare that we’re using lambdas every time we need to make a function. Additionally, we have to read the code almost backward to understand what it’s doing. While this backward sequencing of operations is fine in a language like Haskell (where pointfree notation hides the messy bracketing), in Python it would make much more sense if we could read our code in the normal order. This is where Alakazam comes in.
zz.of(arr).map(_1 ** 2).filter(_1 % 2 == 0).list()
Now the code reads left-to-right, and the lambdas are not nearly as bulky. The zz.ZZ (which is an alias for the class alakazam.Alakazam) is the entry point to any stream-based operations you might want to perform with Alakazam. The of method wraps any iterable in an Alakazam instance. Then the map and filter do the same thing as their global function equivalents. Finally, the list method converts the Alakazam iterable into an ordinary Python list. The important thing is that, now, a cursory left-to-right reading of the code yields “arr -> map -> filter -> list”, which is the sequence of operations we’re actually performing.
The lambda syntax is also significantly shortened. In cases where your anonymous function merely uses operators, element access, and attribute access, you can shorten it by using the placeholder lambdas. The placeholder constants _1, _2, _3, _4, _5, based loosely on the C++ Boost placeholders with the same names, are each defined as callable objects which return their nth argument. If you need more than five arguments, you can use zz.arg(n) directly, which is how the placeholders are implemented in the first place. These placeholders can be used with (almost) any built-in Python operator, and they can also be subscripted and have arbitrary attributes accessed on them. All of these operations will be translated into an anonymous function that performs the corresponding operation on its arguments. So _1 ** 2 is a function that squares its first (and only) argument, _1 + _2 is a function that adds two arguments together, and _1.name == "Alakazam" is a function which checks whether its argument’s name attribute is equal to “Alakazam”.
The Alakazam Wiki lists the functions available to user
Please see LICENSE.txt for licensing information.
Release Notes (0.6.0)
- intersperse and intertwine transformers added.
- string reducer added.
- assign as a synonym for set.
- All classes are now new-style in Python 2 as well as Python 3.
Release Notes (0.5.0)
- accumulate can take an initial value now.
- accumulate, filterfalse, and zip_longest now work correctly on Python 2.
- sum now works on any type that has __add__, including strings.
- compose works correctly when the argument list is empty now.
- Several functions that used to raise Python errors now raise AlakazamError.
- Several bugfixes having to do with lazy evaluation in Python 2.
Release Notes (0.4.0)
- Trace functions trace, traceid, and tracestack, for debugging help.
- New convenience syntax for invoking bind.
- map can take multiple arguments now (Issue #1).
- withobject transformer method.
- zipup producer method.
- absorb and consume reduction methods.
- swap convenience function for tuples.
- Terminology change: “Generator” to “Producer” to avoid confusion with the Python “generator” concept.
Release Notes (0.3.0)
- New convenience function id.
- Boolean functions not, and_, or_, and xor.
- min and max methods on Alakazam objects.
- New flatten and iterate methods.
- length and null reduction methods.
- None is now permitted as an argument to some functions where its behavior would have caused issues before.
Release Notes (0.2.0)
- Alakazam now uses Python 3 semantics for division (from __future__ import division) for consistency.
- New functions setindex, getindex, and delindex, for subscripted access and manipulation.
- New utility function raise_.
- Assignment lambdas with set, binder lambdas with bind, and deletion lambdas with delete.
- Errors are reported through AlakazamError now.
- Changed zip_longest and cross_product argument order to better match the itertools equivalents.
- Static methods on the Alakazam class can now be called globally.
- Function of_dict provided to load dictionaries into Alakazam as lists of key-value 2-tuples.