meiga 0.1.3

A simple, typed and monad-based Result type for Python

meiga 🧙

A simple, typed and monad-based Result type for Python.

This package provides a new type for your Python applications, the Result[Type, Type]. This Result type allows to define two subtypes, giving us the option to create useful return types.

This package is based in another solutions from another modern languages as the swift-based Result implementation.

Installation

pip install meiga

Getting Started

meiga is a framework that give us a simpler, clearer way of handling errors in Python. Use it whenever a class method or a function has the possibility of failure.

Consider the following example of a function that tries to extract a String (str) for a given key from a Dict.

from meiga import Result, Error


class NoSuchKey(Error):
    pass


class TypeMismatch(Error):
    pass


def string_from_key(dictionary: dict, key: str) -> Result[str, Error]:
    if key not in dictionary.keys():
        return Result(failure=NoSuchKey())

    value = dictionary[key]
    if not isinstance(value, str):
        return Result(failure=TypeMismatch())

    return Result(success=value)

Result meiga type provides a robust wrapper around the functions. Rather than throw an exception, it returns a Result that either contains the String value for the given key, or an ErrorClass detailing what went wrong.

Result Type

Let's image we have a dictionary that represent a user info data

>>> user_info = {"first_name": "Rosalia", "last_name": "De Castro", "age": 60}

And we try to obtain first_name

>>> result = string_from_key(dictionary=user_info, key="first_name")
Result[status: success | value: Rosalia]

You can check the status of the result

>>> result.is_success
True
>>> result.is_failure
False

If the result is a success you can get the expected value

>>> result.value
Rosalia 

Otherwise, if we try to access an invalid key or a non string value, returned result will be a failure.

>>> result = string_from_key(dictionary=user_info, key="invalid_key")
Result[status: failure | value: NoSuchKey]
>>> result.is_failure
True
>>> result.value
NoSuchKey() // Error 

Or

>>> result = string_from_key(dictionary=user_info, key="age")
Result[status: failure | value: TypeMismatch]
>>> result.is_failure
True
>>> result.value
TypeMismatch() // Error 

Alias

Use meiga aliases to improve the semantics of your code.

For success result you can use:

result = Result(success="Rosalia")
result = Success("Rosalia") # bool value

If return value is a bool you can use:

result = Success()
result = Success(True)
result = isSuccess

For failure results:

class NoSuchKey(Error):
    pass

result = Result(failure=NoSuchKey())
result = Failure(NoSuchKey())

If you don't want to specify the error, you can use default value with:

result = Failure()
result = Failure(Error())
result = isFailure # Only valid for a failure result with a Error() value

Bringing previous example back. that is the way you can use the alias:

from meiga import Result, Error, Success, Failure,


class NoSuchKey(Error):
    pass


class TypeMismatch(Error):
    pass


def string_from_key(dictionary: dict, key: str) -> Result[str, Error]:
    if key not in dictionary.keys():
        return Failure(NoSuchKey())

    value = dictionary[key]
    if not isinstance(value, str):
        return Failure(TypeMismatch())

    return Success(value)

Furthermore, there is a available a useful alias: NotImplementedMethodError

Use it when define abstract method that returns Result type

from meiga import Result, Error, NotImplementedMethodError

from abc import ABCMeta, abstractmethod

class AuthService:

    __metaclass__ = ABCMeta

    @abstractmethod
    def __init__(self, base_url: str):
        self.base_url = base_url

    @abstractmethod
    def create_token(self, client: str, client_id: str) -> Result[str, Error]:
        return NotImplementedMethodError

Handle Result

This framework also allows a method for handling Result type

When the operations is executed with its happy path, handle function returns the success value, as with result.value.

>>> result = string_from_key(dictionary=user_info, key="first_name")
Result[status: success | value: Rosalia]
>>> first_name = result.handle()
Rosalia

In addition, you can call another function after evaluate the result. Use optional parameters success_handler and failure_handler (Callable functions).

def success_handler():
    print("Do my successful stuff here!")

def failure_handler():
     print("Do my failure stuff here!")


result = string_from_key(dictionary=user_info, key="first_name")

result.handle(success_handler=success_handler, failure_handler=failure_handler)

If you need to add some arguments as a parameters, use success_args and failure_args:

def success_handler(param_1):
    print(f"param_1: {param_1}")

def failure_handler(param_1, param_2):
    print(f"param_1: {param_1}")
    print(f"param_2: {param_2}")


result = string_from_key(dictionary=user_info, key="first_name")

result.handle(success_handler=success_handler, 
              failure_handler=failure_handler,
              success_args=1,
              failure_args=(1, 2))

On the other hand, if something wrong happens handle function will raise an Exception (ReturnErrorOnFailure). Meiga has available a decorator to allow to handle the exception in case of error and unwrap the value in case of success.

from meiga import Result, Error
from meiga.decorators import meiga

@meiga
def handling_result(key: str) -> Result:
    user_info = {"first_name": "Rosalia", "last_name": "De Castro", "age": 60}
    first_name = string_from_key(dictionary=user_info, key=key).handle() 
    # Do whatever with the name
    name = first_name.lower()
    return Result(success=name)

If key is valid success value would be returned. Otherwise, an Error would be returned.

Assertions

To help us on testing functions that returns Result, meiga provide us two functions: assert_success and access_failure.

Check the following pytest-based test: tests/unit/test_result_assertions.py

Developers

Install requirements

pip install -r requirements/dev.txt

Test

pip install -e . && pytest

Upload to PyPi

python setup.py sdist bdist_wheel
twine check dist/*
twine upload --repository-url https://upload.pypi.org/legacy/ dist/*