from __future__ import annotations
from dataclasses import Field, fields, is_dataclass
from inspect import isclass
from typing import TYPE_CHECKING, Any, ClassVar, Iterable, Protocol, Type, final, runtime_checkable
if TYPE_CHECKING:
from typing import AbstractSet
from typing_extensions import TypeAlias, TypeGuard
[docs]@final
class Empty:
"""A sentinel class used as placeholder."""
EmptyType: TypeAlias = Type[Empty]
"""Type alias for the :class:`Empty` sentinel class."""
@runtime_checkable
class DataclassProtocol(Protocol):
"""Protocol for instance checking dataclasses"""
__dataclass_fields__: ClassVar[dict[str, Any]]
def extract_dataclass_fields(
dt: DataclassProtocol,
exclude_none: bool = False,
exclude_empty: bool = False,
include: AbstractSet[str] | None = None,
exclude: AbstractSet[str] | None = None,
) -> tuple[Field[Any], ...]:
"""Extract dataclass fields.
Args:
dt: A dataclass instance.
exclude_none: Whether to exclude None values.
exclude_empty: Whether to exclude Empty values.
include: An iterable of fields to include.
exclude: An iterable of fields to exclude.
Returns:
A tuple of dataclass fields.
"""
include = include or set()
exclude = exclude or set()
if common := (include & exclude):
msg = f"Fields {common} are both included and excluded."
raise ValueError(msg)
dataclass_fields: Iterable[Field[Any]] = fields(dt)
if exclude_none:
dataclass_fields = (field for field in dataclass_fields if getattr(dt, field.name) is not None)
if exclude_empty:
dataclass_fields = (field for field in dataclass_fields if getattr(dt, field.name) is not Empty)
if include:
dataclass_fields = (field for field in dataclass_fields if field.name in include)
if exclude:
dataclass_fields = (field for field in dataclass_fields if field.name not in exclude)
return tuple(dataclass_fields)
def extract_dataclass_items(
dt: DataclassProtocol,
exclude_none: bool = False,
exclude_empty: bool = False,
include: AbstractSet[str] | None = None,
exclude: AbstractSet[str] | None = None,
) -> tuple[tuple[str, Any], ...]:
"""Extract dataclass name, value pairs.
Unlike the 'asdict' method exports by the stdlib, this function does not pickle values.
Args:
dt: A dataclass instance.
exclude_none: Whether to exclude None values.
exclude_empty: Whether to exclude Empty values.
include: An iterable of fields to include.
exclude: An iterable of fields to exclude.
Returns:
A tuple of key/value pairs.
"""
dataclass_fields = extract_dataclass_fields(dt, exclude_none, exclude_empty, include, exclude)
return tuple((field.name, getattr(dt, field.name)) for field in dataclass_fields)
def simple_asdict(
obj: DataclassProtocol,
exclude_none: bool = False,
exclude_empty: bool = False,
convert_nested: bool = True,
exclude: set[str] | None = None,
) -> dict[str, Any]:
"""Convert a dataclass to a dictionary.
This method has important differences to the standard library version:
- it does not deepcopy values
- it does not recurse into collections
Args:
obj: A dataclass instance.
exclude_none: Whether to exclude None values.
exclude_empty: Whether to exclude Empty values.
convert_nested: Whether to recursively convert nested dataclasses.
exclude: An iterable of fields to exclude.
Returns:
A dictionary of key/value pairs.
"""
ret: dict[str, Any] = {}
for field in extract_dataclass_fields(obj, exclude_none, exclude_empty, exclude=exclude):
value = getattr(obj, field.name)
if is_dataclass_instance(value) and convert_nested:
ret[field.name] = simple_asdict(value, exclude_none, exclude_empty)
else:
ret[field.name] = getattr(obj, field.name)
return ret
def is_dataclass_instance(obj: Any) -> TypeGuard[DataclassProtocol]:
"""Check if an object is a dataclass instance.
Args:
obj: An object to check.
Returns:
True if the object is a dataclass instance.
"""
return hasattr(type(obj), "__dataclass_fields__") # pyright: ignore[reportUnknownArgumentType]
def is_dataclass_class(annotation: Any) -> TypeGuard[type[DataclassProtocol]]:
"""Wrap :func:`is_dataclass <dataclasses.is_dataclass>` in a :data:`typing.TypeGuard`.
Args:
annotation: tested to determine if instance or type of :class:`dataclasses.dataclass`.
Returns:
``True`` if instance or type of ``dataclass``.
"""
try:
return isclass(annotation) and is_dataclass(annotation)
except TypeError: # pragma: no cover
return False
advanced_alchemy