Structure your click arguments.
dataclass-click lets you move your user arguments from kwargs to dataclasses, keeping things self-contained.
Click is pretty simple to start with, but when your program gets complex, you find yourself repeating a lot of work. This is particularly true if you have command groups with shared arguments.
The idea of dataclass-click
is to move the @option
and @argument
decorators off into annotations on dataclasses
and pass dataclass objects instead of kwargs.
from dataclasses import dataclass
from pathlib import Path
from typing import Annotated
import click
from dataclass_click import argument, dataclass_click, option
@dataclass
class Config:
# Auto-inferred type for built-in click types
target: Annotated[Path, argument()]
# Auto-inferred option names
foo: Annotated[float, option(required=True)]
# Automatically map mismatched names
bar: Annotated[int | None, option("--other")]
@click.command()
@dataclass_click(Config)
def main(config: Config):
# All args neatly packaged.
print(config.target, config.foo, config.bar)
print(list(config.target.iterdir()))
The pythonic name for options and attributes is copied from the dataclass attribute name. Options can then be named whatever you want. If you don't specify an option name, it will be inferred for you:
from dataclasses import dataclass
from dataclass_click import option, argument
from typing import Annotated
@dataclass
class Config:
foo: Annotated[str, argument()]
bar: Annotated[str, option()]
baz: Annotated[str, option("--bonno")]
Usage: main.py [OPTIONS] FOO
Options:
--bonno TEXT
--bar TEXT
--help Show this message and exit.
data-class click can infer click parameter type (eg: option(type=click.INT)
) from the attribute type.
These types are static so if, like click.Path
, you need arguments, you can always specify the type yourself like
normal.
from dataclasses import dataclass
from datetime import datetime
from pathlib import Path
from typing import Annotated
from uuid import UUID
from dataclass_click import option
@dataclass
class Config:
a: Annotated[str, option()] # click.STRING
b: Annotated[bool, option()] # click.BOOL
c: Annotated[int, option()] # click.INT
d: Annotated[float, option()] # click.FLOAT
e: Annotated[UUID, option()] # click.UUID
f: Annotated[datetime, option()] # click.DateTime()
g: Annotated[Path, option()] # click.Path(path_type=Path)
Type inference is extensible so a program can add and even change the default click ParameterTypes associated with a
Python data type. Eg: if you want to add a Decimal
:
from decimal import Decimal
import click
import dataclass_click
class DecimalParameterType(click.ParamType):
name = "integer"
def convert(self, value, param, ctx):
if isinstance(value, Decimal):
return value
try:
return Decimal(value)
except ValueError:
self.fail(f"{value!r} is not a valid decimal number", param, ctx)
DECIMAL = DecimalParameterType()
# Extend type inference to add Decimal
dataclass_click.register_type_inference(Decimal, DECIMAL)
To avoid tricky mismatches between required options and optional attributes, dataclass-click will infer a field is
required if neither default=
nor required=
are explicitly set and the attribute type hint is not optional:
from dataclasses import dataclass
from typing import Annotated
from dataclass_click import option
@dataclass
class Config:
# Infer --foo is required because str is not optional
foo: Annotated[str, option()]
# Infer --bar is not required because str | None is optional
bar: Annotated[str | None, option()]
# No inference performed because required= is set.
baz: Annotated[str | None, option(required=True)]
# Infer not required because default= set.
bob: Annotated[str, option(default="no")]
# If you want to shoot yourself in the foot, you can!
# Not inferred because required=False.
# Click will pass None even though the attribute is not optional
bonno: Annotated[str, option(required=False)]
Usage: main.py [OPTIONS]
Options:
--bob TEXT
--baz TEXT [required]
--bar TEXT
--foo TEXT [required]
--bonno
--help Show this message and exit.