This packages includes many decorators that will make you write cleaner Python code.
This package requires Python 3.11 or later. There are multiple options for installing this package.
Option 1: Installing with pip from Pypi
Run pip install pedantic.
Option 2: Installing with conda from conda-forge
Run conda install -c conda-forge pedantic
- Install Git if you don't have it already.
- Run
pip install git+https://github.com/LostInDarkMath/pedantic-python-decorators.git@master
- Download the latest release here by clicking on
pedantic-python-decorators-x.y.z-py-none-any.whl. - Execute
pip install pedantic-python-decorators-x.y.z-py3-none-any.whl.
The @pedantic decorator - Type checking at runtime
The @pedantic decorator does the following things:
- The decorated function can only be called by using keyword arguments. Positional arguments are not accepted.
- The decorated function must have type annotations.
- Each time the decorated function is called, pedantic checks that the passed arguments and the return value of the function matches the given type annotations.
As a consequence, the arguments are also checked for
None, becauseNoneis only a valid argument, if it is annotated viatyping.Optional.
In a nutshell:
@pedantic raises an PedanticException if one of the following happened:
- The decorated function is called with positional arguments.
- The function has no type annotation for their return type or one or more parameters do not have type annotations.
- A type annotation is incorrect.
- A type annotation misses type arguments, e.g.
typing.Listinstead oftyping.List[int].
from pedantic import pedantic
@pedantic
def get_sum_of(values: list[int | float]) -> int:
return sum(values)
get_sum_of(values=[0, 1.2, 3, 5.4]) # this raises the following runtime error:
# Type hint of return value is incorrect: Expected type <class 'int'> but 10.0 of type <class 'float'> was the return value which does not match.The @validate decorator
As the name suggests, with @validate you are able to validate the values that are passed to the decorated function.
That is done in a highly customizable way.
But the highest benefit of this decorator is that it makes it extremely easy to write decoupled easy testable, maintainable and scalable code.
The following example shows the decoupled implementation of a configurable algorithm with the help of @validate:
import os
from dataclasses import dataclass
from pedantic import validate, ExternalParameter, overrides, Validator, Parameter, Min, ReturnAs
@dataclass(frozen=True)
class Configuration:
iterations: int
max_error: float
class ConfigurationValidator(Validator):
@overrides(Validator)
def validate(self, value: Configuration) -> Configuration:
if value.iterations < 1 or value.max_error < 0:
self.raise_exception(msg=f'Invalid configuration: {value}', value=value)
return value
class ConfigFromEnvVar(ExternalParameter):
""" Reads the configuration from environment variables. """
@overrides(ExternalParameter)
def has_value(self) -> bool:
return 'iterations' in os.environ and 'max_error' in os.environ
@overrides(ExternalParameter)
def load_value(self) -> Configuration:
return Configuration(
iterations=int(os.environ['iterations']),
max_error=float(os.environ['max_error']),
)
class ConfigFromFile(ExternalParameter):
""" Reads the configuration from a config file. """
@overrides(ExternalParameter)
def has_value(self) -> bool:
return os.path.isfile('config.csv')
@overrides(ExternalParameter)
def load_value(self) -> Configuration:
with open(file='config.csv', mode='r') as file:
content = file.readlines()
return Configuration(
iterations=int(content[0].strip('\n')),
max_error=float(content[1]),
)
# choose your configuration source here:
@validate(ConfigFromEnvVar(name='config', validators=[ConfigurationValidator()]), strict=False, return_as=ReturnAs.KWARGS_WITH_NONE)
# @validate(ConfigFromFile(name='config', validators=[ConfigurationValidator()]), strict=False)
# with strict_mode = True (which is the default)
# you need to pass a Parameter for each parameter of the decorated function
# @validate(
# Parameter(name='value', validators=[Min(5, include_boundary=False)]),
# ConfigFromFile(name='config', validators=[ConfigurationValidator()]),
# )
def my_algorithm(value: float, config: Configuration) -> float:
"""
This method calculates something that depends on the given value with considering the configuration.
Note how well this small piece of code is designed:
- Fhe function my_algorithm() need a Configuration but has no knowledge where this come from.
- Furthermore, it doesn't care about parameter validation.
- The ConfigurationValidator doesn't know anything about the creation of the data.
- The @validate decorator is the only you need to change, if you want a different configuration source.
"""
print(value)
print(config)
return value
if __name__ == '__main__':
# we can call the function with a config like there is no decorator.
# This makes testing extremely easy: no config files, no environment variables or stuff like that
print(my_algorithm(value=2, config=Configuration(iterations=3, max_error=4.4)))
os.environ['iterations'] = '12'
os.environ['max_error'] = '3.1415'
# but we also can omit the config and load it implicitly by our custom Parameters
print(my_algorithm(value=42.0))- @count_calls
- @deprecated
- @does_same_as_function
- @frozen_dataclass
- @frozen_type_safe_dataclass
- @for_all_methods
- @in_subprocess
- @mock
- @overrides
- @pedantic
- @pedantic_class
- @rename_kwargs
- @require_kwargs
- @retry
- @timer
- @timer_class
- @trace
- @trace_class
- @trace_if_returns
- @unimplemented
- @validate
There are no hard dependencies. But if you want to use some advanced features you need to install the following packages:
- multiprocess if you want to use the
@in_subprocessdecorator - flask if you want to you the request validators which are designed for
Flask(see unit tests for examples):FlaskParameter(abstract class)FlaskJsonParameterFlaskFormParameterFlaskPathParameterFlaskGetParameterFlaskHeaderParameterGenericFlaskDeserializer
Feel free to contribute by submitting a pull request :)
The usage of decorators may affect the performance of your application.
For this reason, I would highly recommend you to disable the decorators if your code runs in a productive environment.
You can disable pedantic by set an environment variable:
export ENABLE_PEDANTIC=0
You can also disable or enable the environment variables in your project by calling a method:
from pedantic import enable_pedantic, disable_pedantic
enable_pedantic()This package is not compatible with compiled source code (e.g. with Nuitka).
That's because it uses the inspect module from the standard library which will raise errors like OSError: could not get source code in case of compiled source code.
Don't forget to check out the documentation. Happy coding!
![dependabot[bot] avatar](https://avatars.githubusercontent.com/in/29110?v=4 "dependabot[bot]")
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent