Click already had stubs but not for the `testing` submodule.
I used stubgen and then narrowed the types when it was easy. I left
`Any` where more work was necessary.
The change introduced in 395ab5abd1
broke the signature of `requests.post`, among others, since
`MutableSequence` is invariant, and plain `Dict[str, str]` values
were no longer valid for the `data` argument.
This changes the signature to have `Any` components as a
compromise. Adding more items to the union seems a bit too much, since
the error messages for invalid argument types are already pretty hard to
read.
Another option might be to use `Mapping` instead of `MutableMapping`
due to covariance, but I assume there's a reason why `MutableMapping`
is used here.
The previous definitions in `mock` caused many false positives in
internal Dropbox repositories.
One source of problems was `List[Mock]` not being compatible with
`List[SomeClass]`, since `list` is invariant.
The Python 2 version was slightly outdated compared to the Python 3
version.
The Python 2 version of Headers.__delitem__() had an optional argument
_index_operation that also exists in the implementation. But considering
this is internal use only, it was missing from the Python 3 version, and
it created problems with derived classes, I decided not to add it to
the merged stub. See also pallets/werkzeug#1051.
* Move google protobuf from 2 to 2and3
This should generally be ok. I ran the internal consistency
tests and they seemed to pass.
* Convert str to bytes
* repr to use str in google.protobuf.internal.containers
The dict stub was referring to an instance, not the type, leading to
__call__ being considered when using as a decorator, rather than
__init__.
mock is a backport of the stdlib module and should be defined the same.
This commit reorders any overloads where the first overload was
"shadowing" the second, preventing it from ever being matched by type
checkers that work by selecting the first matching overload alternative.
For example, the first overload alternative below is strictly broader
then the second, preventing it from ever being selected:
class Parent: pass
class Child(Parent): pass
@overload
def foo(x: *int) -> Parent: ...
@overload
def foo(x: int, y: int) -> Child: ...
The correct thing to do is to either delete the second overload or
rearrange them to look like this:
@overload
def foo(x: int, y: int) -> Child: ...
@overload
def foo(x: *int) -> Parent: ...
Rationale: I'm currently [working on a proposal][0] that would amend
PEP 484 to (a) mandate type checkers check overloads in order and
(b) prohibit overloads where an earlier alternative completely shadows
a later one.
[0]: https://github.com/python/typing/issues/253#issuecomment-389262904
This would prohibit overloads that look like the example below, where
the first alternative completely shadows the second.
I figured it would be a good idea to make these changes ahead of time:
if my proposal is accepted, it'd make the transition smoother. If not,
this is hopefully a relatively harmless change.
Note: I think some of these overloads could be simplified (e.g.
`reversed(...)`), but I mostly stuck with rearranging them in case I was
wrong. The only overload I actually changed was `hmac.compare_digest` --
I believe the Python 2 version actually accepts unicode.
A call like this:
```
from attr import attrib
from attr.validators import optional, instance_of
c = attrib(default=None, validator=optional(instance_of(bytes)))
```
Was returning no return value, because the first overload was T -> T.
The solution was to change the ordering so that the first overload is
None -> T and infer the type from the other arguments.
