typed-descriptors is a small library of descriptor classes featuring static and runtime typechecking, runtime validation, and other useful features.
Static typechecking is compatible with PEP 484 type hints, and runtime typechecking is performed by the typing-validation library.
Contents
You can install the latest release from PyPI as follows:
pip install --upgrade typed-descriptors
Classes from the typed_descriptors module can be used to create statically typechecked descriptors which implement the following features:
- attributes with runtime typechecking on write
- attributes with validation on write
- readonly attributes (set once)
- cached properties
Typechecking is compatible with PEP 484 type hints. Runtime typechecking is performed by the typing-validation library.
Below is a simple example displaying all features listed above:
from collections.abc import Sequence
import networkx as nx # type: ignore
from typed_descriptors import Attr, Prop
class LabelledKn:
r"""
A complete graph :math:`K_n` with readonly size and mutable labels,
where the NetworkX graph object is computed lazily and cached.
"""
n = Attr(int, lambda self, n: n >= 0, readonly=True)
# type ^^^ validation ^^^^^^ ^^^^^^^^^^^^^ attribute is readonly
labels = Attr(Sequence[str], lambda self, labels: len(labels) == self.n)
# type ^^^^^^^^^^^^^ validation ^^^^^^^^^^^^^^^^^^^^^
graph = Prop(nx.Graph, lambda self: nx.complete_graph(self.n))
# type ^^^^^^^^ prop value ^^^^^^^^^^^^^^^^^^^^^^^^^
def __init__(self, n: int, labels: Sequence[int]):
# Setters for Attr instances take care of runtime typechecking and validation
# for the arguments 'n' and 'labels' which have been passed to the constuctor.
self.n = n
self.labels = labels
myobj = LabelledKn(3, ["a", "b", "c"]) # OK
myobj.labels = ("x", "y", "z") # OK
print(myobj.graph.edges) # OK: EdgeView([(0, 1), (0, 2), (1, 2)])
myobj.x = 5 # AttributeError (readonly descriptor)
myobj.y = ["a", "b", "c", "d"] # ValueError (lenght of y is not 3)
myobj.y = 5 # TypeError (type of y is not 'Sequence')
myobj.y = [2, 3, 5] # TypeError (type of y is not 'Sequence[str]')For the full API documentation, see https://typed-descriptors.readthedocs.io/
Please see CONTRIBUTING.md.
