Data Class

A data class is a class decorated with @dataclass from the dataclasses module, introduced in Python 3.7. It utilizes PEP 526 variable annotations to automatically generate boilerplate special methods (dunder methods)—such as __init__(), __repr__(), and __eq__()—directly into the class namespace based on the defined attributes.

Basic Syntax

To define a data class, apply the @dataclass decorator and declare class variables with type hints.

from dataclasses import dataclass

@dataclass
class Point:
    x: float
    y: float
    z: float = 0.0  # Attribute with a default value

When the Python interpreter processes this decorator, it dynamically synthesizes the following methods:

1. __init__(self, x: float, y: float, z: float = 0.0): Initializes the instance variables.
2. __repr__(self): Returns a string representation, e.g., Point(x=1.0, y=2.0, z=0.0).
3. __eq__(self, other): Compares the class instances as if they were tuples of their fields.

Decorator Parameters

The @dataclass decorator accepts several boolean arguments to control which methods are generated and how the class behaves:

@dataclass(
    init=True,         # Generates __init__()
    repr=True,         # Generates __repr__()
    eq=True,           # Generates __eq__()
    order=False,       # Generates __lt__(), __le__(), __gt__(), __ge__()
    unsafe_hash=False, # Forces generation of __hash__()
    frozen=False,      # Makes instances immutable (read-only)
    match_args=True,   # Generates __match_args__ for structural pattern matching
    kw_only=False,     # Forces all fields to be keyword-only arguments (Python 3.10+)
    slots=False        # Generates __slots__ for memory optimization (Python 3.10+)
)
class Configuration:
    pass

Setting frozen=True simulates immutability by overriding __setattr__() and __delattr__() to raise a FrozenInstanceError if attribute modification is attempted after initialization. It also automatically generates a __hash__() method, making the instances hashable.

The field() Function

For granular control over individual attributes, the field() function is used. It allows you to exclude specific fields from generated methods or handle mutable default values.

from dataclasses import dataclass, field
from typing import List

@dataclass
class Node:
    id: int
    # default_factory prevents shared state across instances for mutable types
    children: List['Node'] = field(default_factory=list)
    
    # Exclude from __repr__ and __init__
    internal_cache: dict = field(default_factory=dict, repr=False, init=False)

Key parameters of field() include:

• default: The default value for the field.
• default_factory: A zero-argument callable called to initialize a field’s value (required for mutable types like list or dict).
• init: If False, the field is excluded from the generated __init__().
• repr: If False, the field is excluded from the generated __repr__().
• compare: If False, the field is excluded from generated equality and ordering methods.

Post-Initialization Processing

Because the @dataclass decorator overwrites the __init__() method, custom initialization logic cannot be placed there. Instead, the dataclasses module provides the __post_init__() hook.

If defined on the class, the generated __init__() method will automatically call __post_init__() immediately after all fields have been initialized.

@dataclass
class Rectangle:
    width: float
    height: float
    area: float = field(init=False)

    def __post_init__(self):
        # Executes after width and height are bound to self
        self.area = self.width * self.height

If the class is defined with frozen=True, __post_init__() must use object.__setattr__(self, 'name', value) to bypass the immutability constraints during the post-initialization phase.

Master Python with Deep Grasping Methodology!

Learn More