250 – Post initialisation in data classes

Data classes have a mechanism that allows you to specify custom (post) initialisation code without having to override the dunder method __init__ that the decorator @dataclass creates.

Take a look at the class Student defined below:

from dataclasses import dataclass, field

@dataclass
class Student:
    name: str
    grades: list[int]
    avg: float = field(default=0, init=False)
    
    def __post_init__(self):
        self.avg = sum(self.grades) / len(self.grades)

The attribute avg is set to not be an attribute that’s specified when instantiating the class Student. Instead, the attribute is computed from the list grades during post initialisation.

The method __post_init__ is called automatically after the dunder method __init__ runs, so self.grades has been assigned yet but the user hasn’t had the chance to inspect the instance and realise the value of avg is not correct.

Here’s an example instantiation of the class Student:

john = Student("John", [3, 4, 4, 4])
print(john.avg)  # 3.75

Note how the average 3.75 was not passed to the call to Student and yet the attribute avg has the correct value.