pythonwrench

Python library with tools for typing, manipulating collections, and more!

Installation

With uv:

uv add pythonwrench

With pip:

pip install pythonwrench

This library has been tested on all Python versions 3.8 - 3.14, requires only typing_extensions>=4.10.0, and runs on Linux, Mac and Windows systems.

Examples

Typing

Check generic types with isinstance_generic :

>>> import pythonwrench as pw
>>>
>>> # Behaves like builtin isinstance() :
>>> pw.isinstance_generic({"a": 1, "b": 2}, dict)
... True
>>> # But works with generic types !
>>> pw.isinstance_generic({"a": 1, "b": 2}, dict[str, int])
... True
>>> pw.isinstance_generic({"a": 1, "b": 2}, dict[str, str])
... False

... or check specific methods with protocols classes beginning with Supports

>>> import pythonwrench as pw
>>>
>>> isinstance({"a": 1, "b": 2}, pw.SupportsIterLen)
... True
>>> isinstance({"a": 1, "b": 2}, pw.SupportsGetitemLen)
... True

Finally, you can also force argument type checking with check_args_types function :

>>> import pythonwrench as pw

>>> @pw.check_args_types
>>> def f(a: int, b: str) -> str:
>>>     return a * b

>>> f(1, "a")  # pass check
>>> f(1, 2)  # raises TypeError from decorator

Collections

Provides functions to facilitate iterables processing, like unzip :

>>> import pythonwrench as pw
>>>
>>> list_of_tuples = [(1, "a"), (2, "b"), (3, "c"), (4, "d")]
>>> pw.unzip(list_of_tuples)
... [1, 2, 3, 4], ["a", "b", "c", "d"]
>>> pw.flatten(list_of_tuples)
... [1, "a", 2, "b", 3, "c", 4, "d"]

... or mathematical functions like prod or argmax :

>>> import pythonwrench as pw
>>>
>>> values = [3, 1, 6, 4]
>>> pw.prod(values)
... 72
>>> pw.argmax(values)
... 2
>>> pw.is_sorted(values)
... False

Easely converts common python structures like list of dicts to dict of lists :

>>> import pythonwrench as pw
>>>
>>> list_of_dicts = [{"a": 1, "b": 2}, {"a": 3, "b": 4}]
>>> pw.list_dict_to_dict_list(list_of_dicts)
... {"a": [1, 3], "b": [2, 4]}

... or dict of dicts :

>>> import pythonwrench as pw
>>>
>>> dict_of_dicts = {"a": {"x": 1, "y": 2}, "b": {"x": 3, "y": 4}}
>>> pw.flat_dict_of_dict(dict_of_dicts)
... {"a.x": 1, "a.y": 2, "b.x": 3, "b.y": 4}

Disk caching (memoize)

>>> import pythonwrench as pw
>>>
>>> @pw.disk_cache_decorator
>>> def heavy_processing():
>>>     # Lot of stuff here
>>>     ...
>>>
>>> data1 = heavy_processing()  # first call function is called and the result is stored on disk
>>> data2 = heavy_processing()  # second call result is loaded from disk directly

Semantic versionning parsing

>>> import pythonwrench as pw
>>> version = pw.Version("1.12.2")
>>> version.to_tuple()
... (1, 12, 2)
>>> version = pw.Version("0.5.1-beta+linux")
>>> version.to_tuple()
... (0, 5, 1, "beta", "linux")

>>> Version("1.3.1") < Version("1.4.0")
... True

Serialization

>>> import pythonwrench as pw
>>>
>>> list_of_dicts = [{"a": 1, "b": 2}, {"a": 3, "b": 4}]
>>> pw.dump_csv(list_of_dicts, "data.csv")
>>> pw.dump_json(list_of_dicts, "data.json")
>>> pw.load_json("data.json") == list_of_dicts
... True

Lazy loading

To speed up this package loading, you can install it with uv add pythonwrench[lazy], so import pythonwrench as pw will be faster.

Contact

Maintainer:

• Étienne Labbé "Labbeti": labbeti.pub@gmail.com