README.md

Serox

Rusty abstractions for Python

serox provides a suite of commonly-used Rust abstractions in a manner that is near-fully static-type-checker compliant, the exceptions being cases involving higher-kinded types (HKTs; e.g. Iterator.collect) as these are not currently supported by Python's type system.

The subset of abstractions most broadly-applicable are importable from serox.prelude.

Features

1. Iterator combinators that allow for the seamless chaining of operations over data with rayon-inspired functionality for effortless parallelism.

2. A Result pseudo-enum comprising Some and Null pseudo-variants. We say 'pseudo' as the Python analogue to Rust's tagged union is the union (A | B) type; since this type is not a data structure, we cannot implement methods on it directly and instead have to resort to some legerdemain.

3. An Option pseudo-enum. The T | None pattern is ubiquitous in Python yet, frustratingly, is not treated as a first-class citizen within the language; Option is a drop-in replacement that redresses this.

4. The qmark decorator emulates the '?' (error/null short-circuiting) operator, allowing for propagation of error and null values without disrupting the control flow. Without this, one has to resort to awkward pattern-matching to perform common operations such as unwrap_or (setting Null to a default value) or map (applying a function to the contained value if Some).

Example

Early exiting (in the fashion of Rust's ? operator) an Option/Result-returning function is enabled by the qmark ('question mark') decorator:

from serox.prelude import *

@qmark
def some_function(value: Option[int]) -> Option[float]:
    squared: int = value.map(lambda x: x ** 2).q
    # The above expands to the rather verbose:
    # match value:
    #     case Null():
    #         return Null[float]()
    #     case Some(x):
    #         squared = value ** 2

    return Some(1.0 / squared)

Requirements

Python version >=3.12.3 is required for typing purposes.

Installation

serox is available on PyPI and thus the latest version can be installed via pip with

pip install serox

or via uv with

uv add serox

Acknowledgements

Credit to result and rustshed for laying the groundwork for the Result and qmark implementations.