Python dataclass fields

Python dataclass fields (from the dataclasses module) let you declare class attributes declaratively with type hints, default values, and factories. The field() function customizes how attributes behave: mutable defaults via factories, excluding from __repr__, marking as init-only, or running post-initialization logic with __post_init__().

Use dataclasses with fields to reduce boilerplate and make class structure explicit.

This example shows dataclass field configuration and post-init hooks.

# run: python3 dataclass_fields.py
# description: dataclass field configuration and post-init
 
from dataclasses import dataclass, field
from typing import List
 
# Basic dataclass with defaults
@dataclass
class Point:
    x: float = 0.0
    y: float = 0.0
 
p = Point(3.0, 4.0)
print(f"Point: {p}")
 
# Mutable default with factory
@dataclass
class Team:
    name: str
    members: List[str] = field(default_factory=list)
 
team1 = Team("A")
team1.members.append("Alice")
team2 = Team("B")  # has separate empty list
print(f"Team 1: {team1}")
print(f"Team 2: {team2}")
 
# Field configuration
@dataclass
class Config:
    name: str
    password: str = field(repr=False)  # hidden in repr
    debug: bool = field(default=False, init=False)  # not in __init__
    tags: List[str] = field(default_factory=list, compare=False)  # excluded from comparisons
 
config = Config("myapp", "secret123")
print(f"Config: {config}")  # password not shown
print(f"Debug: {config.debug}")
 
# Post-init processing
@dataclass
class Circle:
    radius: float
    area: float = field(init=False)  # computed, not initialized
 
    def __post_init__(self):
        """Called after __init__ to compute derived attributes."""
        import math
        self.area = math.pi * self.radius ** 2
 
circle = Circle(5.0)
print(f"Circle: {circle}")
print(f"Area: {circle.area:.2f}")
 
# Field with default factory and post-init
@dataclass
class Database:
    host: str
    port: int = 5432
    username: str = "admin"
    password: str = field(repr=False, default="")
    _connection: object = field(init=False, repr=False, default=None)
 
    def __post_init__(self):
        """Initialize connection after dataclass init."""
        self._connection = f"postgres://{self.username}@{self.host}:{self.port}"
 
db = Database("localhost")
print(f"Database: {db}")
print(f"Connection: {db._connection}")
 
# Field with init=False and default_factory
@dataclass
class Cache:
    name: str
    _data: dict = field(init=False, default_factory=dict)
    _hits: int = field(init=False, default=0)
 
    def get(self, key):
        if key in self._data:
            self._hits += 1
        return self._data.get(key)
 
    def set(self, key, value):
        self._data[key] = value
 
cache = Cache("mycache")
cache.set("key1", "value1")
print(f"Cache: {cache}")
print(f"Hits: {cache._hits}")

field() parameters:

• default: default value
• default_factory: callable to generate default (for mutable types)
• init: include in __init__ (default True)
• repr: include in __repr__ (default True)
• compare: include in comparisons (default True)
• hash: include in __hash__ (default is compare value)
• metadata: arbitrary metadata dict for this field

Mutable defaults:

• Wrong: items: list = []: shared among all instances
• Right: items: list = field(default_factory=list): separate list per instance
• Also works for dict, set, or any callable

Post-initialization:

• __post_init__() called after __init__ completes
• Use for computed attributes, validation, setup
• Access all initialized attributes here

Init-only fields:

• field(init=False): set after init
• Useful for derived/computed attributes
• Won't appear in __init__ signature

Excluding from repr:

• field(repr=False): hidden from __repr__ output
• Good for secrets, large objects, internal state

Comparison control:

• field(compare=False): ignored in __eq__, __lt__, etc.
• Useful for caches, id fields not affecting equality
• By default all fields participate

Field ordering:

• Fields without defaults must come before fields with defaults
• field(default=..., ...) counts as “has default”
• Fields with init=False can go anywhere

Field metadata:

• field(metadata={'description': '...'}): store arbitrary info
• Accessible via fields() function
• Useful for generating docs, validators

Accessing field info:

• from dataclasses import fields
• fields(MyClass): returns tuple of Field objects
• Field.name, Field.type, Field.default, etc.

When to use fields vs simple defaults:

• Use field() for any mutable default
• Use field() for computed/derived attributes
• Simple immutable defaults can use attr: type = value
• Complex initialization use __post_init__()