Skip to content

Narrow is with final types correctly #15646

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
wants to merge 5 commits into
base: master
Choose a base branch
from
Open
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
10 changes: 10 additions & 0 deletions mypy/subtypes.py
Original file line number Diff line number Diff line change
Expand Up @@ -1959,6 +1959,16 @@ def covers_at_runtime(item: Type, supertype: Type) -> bool:
item = get_proper_type(item)
supertype = get_proper_type(supertype)

# Left `Any` or `Type[Any]` type will never be covered at runtime:
if isinstance(item, AnyType) or (
isinstance(item, TypeType) and isinstance(item.item, AnyType)
):
return False

if isinstance(item, (CallableType, TypeType)) and item.is_singleton_type():

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

What if the RHS of the is operator is a specialized generic class like list[int]? The is conditional will always evaluate to False then.

if is_proper_subtype(item, supertype):
return True

# Since runtime type checks will ignore type arguments, erase the types.
supertype = erase_type(supertype)
if is_proper_subtype(
Expand Down
8 changes: 8 additions & 0 deletions mypy/types.py
Original file line number Diff line number Diff line change
Expand Up @@ -2167,6 +2167,9 @@ def with_unpacked_kwargs(self) -> NormalizedCallableType:
)
)

def is_singleton_type(self) -> bool:
return self.is_type_obj() and self.type_object().is_final

def with_normalized_var_args(self) -> Self:
var_arg = self.var_arg()
if not var_arg or not isinstance(var_arg.typ, UnpackType):
Expand Down Expand Up @@ -3111,6 +3114,11 @@ def __eq__(self, other: object) -> bool:
return NotImplemented
return self.item == other.item

def is_singleton_type(self) -> bool:
return (isinstance(self.item, Instance) and self.item.type.is_final) or isinstance(
self.item, NoneType
)

def serialize(self) -> JsonDict:
return {".class": "TypeType", "item": self.item.serialize()}

Expand Down
95 changes: 95 additions & 0 deletions test-data/unit/check-narrowing.test
Original file line number Diff line number Diff line change
Expand Up @@ -1374,7 +1374,102 @@ def g() -> None:
foo()
[builtins fixtures/dict.pyi]

[case testNarrowingIsFinalType]
from typing import Type, Union
from typing_extensions import final

@final
class Mark: ...

@final
class Other: ...

x: Union[Type[Mark], Type[Other], Type[None], int]

if x is Mark:

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

What about M = NewType("M", Mark")? This acts as a subtype of Mark (and is allowed even though Mark is marked @final), but x is Mark will evaluate to false if x is M. Maybe that's an edge case that can be ignored, but it does represent a small hole.

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I guess we can ignore this edge case. I'd be surprised if users are passing the NewType definition somewhere as a value, and even more in this particular situation combined with a @final. The benefit from this feature outweighs supporting the edge case.

And even then, isn't the issue only in the opposite case, when we have if x is not Marc:? A situation where at runtime we do enter the block, but we narrowed incorrectly.

Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Why can you create a NewType of a final type? That seems inconsistent - NewType creates a subtype, but final means that the type isn't allowed to have subclasses. I don't see this mentioned anywhere in the docs/PEPs. I guess it could mean that you have a subset of values, but they all will be the same type in reality. It's still a little contradictory though.

reveal_type(x) # N: Revealed type is "Type[__main__.Mark]"
else:
reveal_type(x) # N: Revealed type is "Union[Type[__main__.Other], Type[None], builtins.int]"

if x is not Mark:
reveal_type(x) # N: Revealed type is "Union[Type[__main__.Other], Type[None], builtins.int]"
else:
reveal_type(x) # N: Revealed type is "Type[__main__.Mark]"

y: Type[Mark] = Mark
if x is y:
reveal_type(x) # N: Revealed type is "Type[__main__.Mark]"
else:
reveal_type(x) # N: Revealed type is "Union[Type[__main__.Other], Type[None], builtins.int]"

if x is not y:
reveal_type(x) # N: Revealed type is "Union[Type[__main__.Other], Type[None], builtins.int]"
else:
reveal_type(x) # N: Revealed type is "Type[__main__.Mark]"

if x is type(None):
reveal_type(x) # N: Revealed type is "Type[None]"
else:
reveal_type(x) # N: Revealed type is "Union[Type[__main__.Mark], Type[__main__.Other], builtins.int]"

if x is not type(None):
reveal_type(x) # N: Revealed type is "Union[Type[__main__.Mark], Type[__main__.Other], builtins.int]"
else:
reveal_type(x) # N: Revealed type is "Type[None]"

z: Type[None]
if x is z:
reveal_type(x) # N: Revealed type is "Type[None]"
else:
reveal_type(x) # N: Revealed type is "Union[Type[__main__.Mark], Type[__main__.Other], builtins.int]"

if x is not z:
reveal_type(x) # N: Revealed type is "Union[Type[__main__.Mark], Type[__main__.Other], builtins.int]"
else:
reveal_type(x) # N: Revealed type is "Type[None]"
[builtins fixtures/isinstancelist.pyi]


[case testNarrowingAnyAgainstFinalType]
from typing import Type, Any
from typing_extensions import final

@final
class Mark: ...

x: Any
if x is Mark:
reveal_type(x) # N: Revealed type is "def () -> __main__.Mark"
else:
reveal_type(x) # N: Revealed type is "Any"

y: Type[Any]
if y is Mark:
reveal_type(y) # N: Revealed type is "Type[__main__.Mark]"
else:
reveal_type(y) # N: Revealed type is "Type[Any]"

[builtins fixtures/isinstancelist.pyi]


[case testNarrowingIsNonFinalType]
from typing import Type, Union

class Mark: ...

x: Union[Type[Mark], Type[None], int]

if x is Mark:
reveal_type(x) # N: Revealed type is "Union[Type[__main__.Mark], Type[None], builtins.int]"

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Why does this not narrow the type to Type[Mark] in the positive (if) case?

else:
reveal_type(x) # N: Revealed type is "Union[Type[__main__.Mark], Type[None], builtins.int]"

y: Type[Mark] = Mark
if x is y:
reveal_type(x) # N: Revealed type is "Union[Type[__main__.Mark], Type[None], builtins.int]"
else:
reveal_type(x) # N: Revealed type is "Union[Type[__main__.Mark], Type[None], builtins.int]"
[builtins fixtures/isinstancelist.pyi]
[case testNarrowingOptionalEqualsNone]
from typing import Optional

Expand Down
Loading