prefect.utilities.collections
Utilities for extensions of and operations on Python collections.
AutoEnum
An enum class that automatically generates value from variable names.
This guards against common errors where variable names are updated but values are not.
In addition, because AutoEnums inherit from str
, they are automatically
JSON-serializable.
See https://docs.python.org/3/library/enum.html#using-automatic-values
Examples:
class MyEnum(AutoEnum):
RED = AutoEnum.auto() # equivalent to RED = 'RED'
BLUE = AutoEnum.auto() # equivalent to BLUE = 'BLUE'
Source code in prefect/utilities/collections.py
class AutoEnum(str, Enum):
"""
An enum class that automatically generates value from variable names.
This guards against common errors where variable names are updated but values are
not.
In addition, because AutoEnums inherit from `str`, they are automatically
JSON-serializable.
See https://docs.python.org/3/library/enum.html#using-automatic-values
Example:
```python
class MyEnum(AutoEnum):
RED = AutoEnum.auto() # equivalent to RED = 'RED'
BLUE = AutoEnum.auto() # equivalent to BLUE = 'BLUE'
```
"""
def _generate_next_value_(name, start, count, last_values):
return name
@staticmethod
def auto():
"""
Exposes `enum.auto()` to avoid requiring a second import to use `AutoEnum`
"""
return auto()
def __repr__(self) -> str:
return f"{type(self).__name__}.{self.value}"
PartialModel
A utility for creating a Pydantic model in several steps.
Fields may be set at initialization, via attribute assignment, or at finalization when the concrete model is returned.
Pydantic validation does not occur until finalization.
Each field can only be set once and a ValueError
will be raised on assignment if
a field already has a value.
Examples:
>>> class MyModel(pydantic.BaseModel):
>>> x: int
>>> y: str
>>> z: float
>>>
>>> partial_model = PartialModel(MyModel, x=1)
>>> partial_model.y = "two"
>>> model = partial_model.finalize(z=3.0)
Source code in prefect/utilities/collections.py
class PartialModel(Generic[M]):
"""
A utility for creating a Pydantic model in several steps.
Fields may be set at initialization, via attribute assignment, or at finalization
when the concrete model is returned.
Pydantic validation does not occur until finalization.
Each field can only be set once and a `ValueError` will be raised on assignment if
a field already has a value.
Example:
>>> class MyModel(pydantic.BaseModel):
>>> x: int
>>> y: str
>>> z: float
>>>
>>> partial_model = PartialModel(MyModel, x=1)
>>> partial_model.y = "two"
>>> model = partial_model.finalize(z=3.0)
"""
def __init__(self, __model_cls: M, **kwargs: Any) -> None:
self.fields = kwargs
# Set fields first to avoid issues if `fields` is also set on the `model_cls`
# in our custom `setattr` implementation.
self.model_cls = __model_cls
for name in kwargs.keys():
self.raise_if_not_in_model(name)
def finalize(self, **kwargs: Any) -> T:
for name in kwargs.keys():
self.raise_if_already_set(name)
self.raise_if_not_in_model(name)
return self.model_cls(**self.fields, **kwargs)
def raise_if_already_set(self, name):
if name in self.fields:
raise ValueError(f"Field {name!r} has already been set.")
def raise_if_not_in_model(self, name):
if name not in self.model_cls.__fields__:
raise ValueError(f"Field {name!r} is not present in the model.")
def __setattr__(self, __name: str, __value: Any) -> None:
if __name in {"fields", "model_cls"}:
return super().__setattr__(__name, __value)
self.raise_if_already_set(__name)
self.raise_if_not_in_model(__name)
self.fields[__name] = __value
def __repr__(self) -> str:
dsp_fields = ", ".join(f"{key}={repr(value)}" for key, value in self.fields)
return f"PartialModel({self.model_cls.__name__}{dsp_fields})"
Quote
Simple wrapper to mark an expression as a different type so it will not be coerced by Prefect. For example, if you want to return a state from a flow without having the flow assume that state.
Source code in prefect/utilities/collections.py
class Quote(Generic[T]):
"""
Simple wrapper to mark an expression as a different type so it will not be coerced
by Prefect. For example, if you want to return a state from a flow without having
the flow assume that state.
"""
def __init__(self, data: T) -> None:
self.data = data
def unquote(self) -> T:
return self.data
batched_iterable
Yield batches of a certain size from an iterable
Parameters:
Name | Description | Default |
---|---|---|
iterable |
An iterable Iterable |
required |
size |
The batch size to return int |
required |
Yields:
Type | Description |
---|---|
tuple |
A batch of the iterable |
Source code in prefect/utilities/collections.py
def batched_iterable(iterable: Iterable[T], size: int) -> Iterator[Tuple[T, ...]]:
"""
Yield batches of a certain size from an iterable
Args:
iterable (Iterable): An iterable
size (int): The batch size to return
Yields:
tuple: A batch of the iterable
"""
it = iter(iterable)
while True:
batch = tuple(itertools.islice(it, size))
if not batch:
break
yield batch
dict_to_flatdict
Converts a (nested) dictionary to a flattened representation.
Each key of the flat dict will be a CompoundKey tuple containing the "chain of keys" for the corresponding value.
Parameters:
Name | Description | Default |
---|---|---|
dct |
The dictionary to flatten dict |
required |
_parent |
The current parent for recursion Tuple |
None |
Returns:
Type | Description |
---|---|
Dict[Tuple[~KT, ...], Any] |
A flattened dict of the same type as dct |
Source code in prefect/utilities/collections.py
def dict_to_flatdict(
dct: Dict[KT, Union[Any, Dict[KT, Any]]], _parent: Tuple[KT, ...] = None
) -> Dict[Tuple[KT, ...], Any]:
"""Converts a (nested) dictionary to a flattened representation.
Each key of the flat dict will be a CompoundKey tuple containing the "chain of keys"
for the corresponding value.
Args:
dct (dict): The dictionary to flatten
_parent (Tuple, optional): The current parent for recursion
Returns:
A flattened dict of the same type as dct
"""
typ = cast(Type[Dict[Tuple[KT, ...], Any]], type(dct))
items: List[Tuple[Tuple[KT, ...], Any]] = []
parent = _parent or tuple()
for k, v in dct.items():
k_parent = tuple(parent + (k,))
if isinstance(v, dict):
items.extend(dict_to_flatdict(v, _parent=k_parent).items())
else:
items.append((k_parent, v))
return typ(items)
extract_instances
Extract objects from a file and returns a dict of type -> instances
Parameters:
Name | Description | Default |
---|---|---|
objects |
An iterable of objects Iterable |
required |
types |
A type or tuple of types to extract, defaults to all objects Union[Type[~T], Tuple[Type[~T], ...]] |
<class 'object'> |
Returns:
Type | Description |
---|---|
If a single type is given |
a list of instances of that type If a tuple of types is given: a mapping of type to a list of instances |
Source code in prefect/utilities/collections.py
def extract_instances(
objects: Iterable,
types: Union[Type[T], Tuple[Type[T], ...]] = object,
) -> Union[List[T], Dict[Type[T], T]]:
"""
Extract objects from a file and returns a dict of type -> instances
Args:
objects: An iterable of objects
types: A type or tuple of types to extract, defaults to all objects
Returns:
If a single type is given: a list of instances of that type
If a tuple of types is given: a mapping of type to a list of instances
"""
types = ensure_iterable(types)
# Create a mapping of type -> instance from the exec values
ret = defaultdict(list)
for o in objects:
# We iterate here so that the key is the passed type rather than type(o)
for type_ in types:
if isinstance(o, type_):
ret[type_].append(o)
if len(types) == 1:
return ret[types[0]]
return ret
flatdict_to_dict
Converts a flattened dictionary back to a nested dictionary.
Parameters:
Name | Description | Default |
---|---|---|
dct |
The dictionary to be nested. Each key should be a tuple of keys
as generated by dict |
required |
Returns A nested dict of the same type as dct
Source code in prefect/utilities/collections.py
def flatdict_to_dict(
dct: Dict[Tuple[KT, ...], VT]
) -> Dict[KT, Union[VT, Dict[KT, VT]]]:
"""Converts a flattened dictionary back to a nested dictionary.
Args:
dct (dict): The dictionary to be nested. Each key should be a tuple of keys
as generated by `dict_to_flatdict`
Returns
A nested dict of the same type as dct
"""
typ = type(dct)
result = cast(Dict[KT, Union[VT, Dict[KT, VT]]], typ())
for key_tuple, value in dct.items():
current_dict = result
for prefix_key in key_tuple[:-1]:
# Build nested dictionaries up for the current key tuple
# Use `setdefault` in case the nested dict has already been created
current_dict = current_dict.setdefault(prefix_key, typ()) # type: ignore
# Set the value
current_dict[key_tuple[-1]] = value
return result
quote
Create a Quote
object
Examples:
>>> from prefect.utilities.collections import quote
>>> x = quote(1)
>>> x.unquote()
1
Source code in prefect/utilities/collections.py
def quote(expr: T) -> Quote[T]:
"""
Create a `Quote` object
Examples:
>>> from prefect.utilities.collections import quote
>>> x = quote(1)
>>> x.unquote()
1
"""
return Quote(expr)
visit_collection
async
This function visits every element of an arbitrary Python collection and
applies visit_fn
to each element. If return_data=True
, a copy of the
data structure containing the results of visit_fn
is returned. Note that
return_data=True
may be slower due to the need to copy every object.
Parameters:
Name | Description | Default |
---|---|---|
expr |
a Python object or expression Any |
required |
visit_fn |
an async function that will be applied to every non-collection element of expr. Callable[[Any], Awaitable[Any]] |
required |
return_data |
if bool |
False |
Source code in prefect/utilities/collections.py
async def visit_collection(
expr, visit_fn: Callable[[Any], Awaitable[Any]], return_data: bool = False
):
"""
This function visits every element of an arbitrary Python collection and
applies `visit_fn` to each element. If `return_data=True`, a copy of the
data structure containing the results of `visit_fn` is returned. Note that
`return_data=True` may be slower due to the need to copy every object.
Args:
expr (Any): a Python object or expression
visit_fn (Callable[[Any], Awaitable[Any]]): an async function that
will be applied to every non-collection element of expr.
return_data (bool): if `True`, a copy of `expr` containing data modified
by `visit_fn` will be returned. This is slower than `return_data=False`
(the default).
"""
def visit_nested(expr):
# Utility for a recursive call, preserving options.
# Returns a `partial` for use with `gather`.
return partial(
visit_collection, expr, visit_fn=visit_fn, return_data=return_data
)
# Get the expression type; treat iterators like lists
typ = list if isinstance(expr, IteratorABC) else type(expr)
typ = cast(type, typ) # mypy treats this as 'object' otherwise and complains
# do not visit mock objects
if isinstance(expr, Mock):
return expr if return_data else None
elif typ in (list, tuple, set):
result = await gather(*[visit_nested(o) for o in expr])
return typ(result) if return_data else None
elif typ in (dict, OrderedDict):
assert isinstance(expr, (dict, OrderedDict)) # typecheck assertion
keys, values = zip(*expr.items()) if expr else ([], [])
keys = await gather(*[visit_nested(k) for k in keys])
values = await gather(*[visit_nested(v) for v in values])
return typ(zip(keys, values)) if return_data else None
elif is_dataclass(expr) and not isinstance(expr, type):
values = await gather(
*[visit_nested(getattr(expr, f.name)) for f in fields(expr)]
)
result = {field.name: value for field, value in zip(fields(expr), values)}
return typ(**result) if return_data else None
elif (
# Recurse into Pydantic models but do _not_ do so for states/datadocs
isinstance(expr, pydantic.BaseModel)
and not isinstance(expr, prefect.orion.schemas.states.State)
and not isinstance(expr, prefect.orion.schemas.data.DataDocument)
):
values = await gather(
*[visit_nested(getattr(expr, field)) for field in expr.__fields__]
)
result = {field: value for field, value in zip(expr.__fields__, values)}
return typ(**result) if return_data else None
else:
result = await visit_fn(expr)
return result if return_data else None