Sorry for thinking outloud
It'd be nice if there was autodocumentation for the plugin api
6 replies
Basically, I was getting really annoyed with how a lot of python web developers just do ad hoc interfaces between services. And I wanted to see if I could create a metaprogrammed OpenAPI client; but one where I could also use MyPy to verify you are using the API correctly.
@altendky stubtest does check that default values are a subtype of the arg's type hint
maybe the issue is that your stub is decorated?
@hauntsaninja https://gist.github.com/altendky/3327ab219a04e27f628b92aa6c480d8a here's the minimal example. i can certainly file a ticket if you think it is appropriate.
also, this isn't a pressing issue. at this point i'm starting to just merge the stubs into the package so this is relevant primarily in the context of just improving (i think?
:]) stubtest.
@altendky if i change your repro to create a pyi file instead of a py file in package-stubs, i get:
error: package.f is inconsistent, runtime argument "p" has a default value of type None, which is incompatible with stub argument type builtins.int
Stub: at line 1
def (p: builtins.int =)
Runtime: at line 1 in file /Users/shantanu/tmp/lol/package/__init__.py
def (p=None)altendky @altendky sighs
i'll explore the difference between that and the actual situation. thanks.
Hi, can anyone help me understand what is wrong with my annotations on my code https://bpa.st/MGSOM . mypy thinks there is an error on L8 but I dont understand what is wrong. It looks like mypy correctly restricts to one type of a Union when it enters an if block but it throws that knowledge away when it enters a locally defined function (compare L5 & L7). Is this a mypy bug?
from typing import Union
def foo(X: Union[int, str]) -> bool:
if isinstance(X, str):
reveal_type(X) # str
def g(y: int) -> bool:
reveal_type(X) # Union[int, str]
return len(X) > y # <<<<< error: Argument 1 to "len" has incompatible type "Union[int, str]"; expected "Sized"
return g(10)
reveal_type(X) # int
return bool(X)
Anyone please
Hi folks, like was stated in the documentation, I'm trying to manually go over the stub file created by stubgen so as to provide explicit types where possible instead of using
Any. I'm currently running into a snag where I don't know how to deal with a NewType that I defined - what I'm trying to do results in a mypy error. Please find below a minimal reproduction of the problem. Does anyone have any advice or am I forced to leave it as Any?# test.py
from dataclasses import dataclass
from typing import NewType
A = NewType("A", str)
@dataclass
class Test:
a: A# stubgen'd test.pyi
from typing import Any
A: Any
class Test:
a: A
def __init__(self, a: Any) -> None: ...# trying to remove 'Any' as much as possible
from typing import Type
A: Type[str]
class Test:
a: A # mypy error: "test.A" is not valid as a type
def __init__(self, a: A) -> None: ... # mypy error: "test.A" is not valid as a type
Has anyone experiences with supervising a thesis that has a student working on mypy? I'm going to supervise a bachelors thesis starting in April that is focused on analysing the exhaustiveness of structural pattern matching in mypy (https://pp.ipd.kit.edu/thesis.php?id=323&lang=en), as a colleague of mine developed such an analysis for Haskell.
Any advice about dmypy and mypy giving different results when checking the same code?
% dmypy check -- --local-partial-types dtests/dtests.py dtests/tests
Success: no issues found in 430 source files
% mypy --local-partial-types dtests/dtests.py dtests/tests
dtests/dtestlib/scales/AbstractIterableComparator.py:36: error: Cannot determine type of '_result'
dtests/dtestlib/scales/AbstractIterableComparator.py:39: error: Need type annotation for '_diff' (hint: "_diff: List[<type>] = ...")
dtests/dtestlib/scales/AbstractIterableComparator.py:72: error: Incompatible types in assignment (expression has type "None", variable has type "List[Any]")
dtests/dtestlib/scales/TypeStrictComparator.py:28: error: Cannot determine type of '_result'
dtests/dtestlib/scales/TypeStrictComparator.py:29: error: Cannot determine type of '_result'
dtests/dtestlib/scales/TypeStrictComparator.py:29: error: unused 'type: ignore' comment
dtests/dtestlib/scales/DictComparator.py:34: error: Incompatible types in assignment (expression has type "Dict[<nothing>, <nothing>]", variable has type "List[Any]")
dtests/dtestlib/scales/DictComparator.py:53: error: Incompatible types in assignment (expression has type "None", variable has type "List[Any]")
dtests/dtestlib/scales/DictComparator.py:33: error: unused 'type: ignore' comment
dtests/dtestlib/scales/DictComparator.py:49: error: unused 'type: ignore' comment
dtests/dtestlib/scales/RelaxedComparator.py:28: error: Cannot determine type of '_result'
dtests/dtestlib/scales/RelaxedComparator.py:29: error: Cannot determine type of '_result'
dtests/dtestlib/scales/RelaxedComparator.py:35: error: Redundant cast to "bool"
dtests/dtestlib/scales/RelaxedComparator.py:29: error: unused 'type: ignore' comment
Hi, I'm running into an issue with typevars unifying into object and therefore allowing incorrect code. Example:
T = TypeVar("T")
class Field(Generic[T]):
dtype: Type[T]
default: T
foo = Field(int, "asdf") # I expect this to fail, but instead it typechecks since foo is a Field[object]I found this issue which seems similar but is closed: python/mypy#6559
Is what I'm looking for possible in mypy?
5 replies
A coworker of mine just asked “Is there a preferred way to return a non-value (null) from a function returning floats? Type checking doesnt like None being returned when the type is float.” Have we thought about having the “Incompatible types in assignment” error message suggest using
Optional when the expression type is None? I would be happy to write a PR.
I personally wouldn't return
Python is more resistant to this than other languages, as division by 0 raises an Error instead of returning
nan as a replacement for None, as it is possible to accidentally produce a nan value.Python is more resistant to this than other languages, as division by 0 raises an Error instead of returning
inf or nan, but it's still possible to get nan for example by calculating inf - inf
there's a possibility that the B that I think I've given it and the B that it received aren't the same - i can check that - but is the above a known failure mode? if so, what does it indicate?
nvm got it
it's mypy
How does one install library stubs for mypy? I tried just installed boto3-stubs, but mypy still complains boto3 doesn't have any typings. How could I check if mypy is finding it or not? Details here: https://stackoverflow.com/questions/65894133/mypy-not-picking-up-installed-stubs
I'm not sure -- it worked for me?
shell:
PS C:\Users\A5rocks\Documents\boto-test> python -m venv venv
...
PS C:\Users\A5rocks\Documents\boto-test> .\venv\Scripts\activate
...
(venv) PS C:\Users\A5rocks\Documents\boto-test> pip install mypy boto3-stubs boto3
...
<open editor and put my code for example.py into, well, example.py>
(venv) PS C:\Users\A5rocks\Documents\boto-test> mypy example.py
example.py:3: note: Revealed type is 'def (self: boto3.NullHandler, record: Any) -> Any'example.py:
import boto3
reveal_type(boto3.NullHandler.emit)(not sure if the Anys were what you were talking about...)
I have encountered a strange behavior with any(). This is a simplified example:
import typing
T = typing.TypeVar('T')
def children(_: T) -> typing.Iterable[T]: return []
def orig(node: str) -> typing.Optional[str]:
return any(orig(child) for child in children(node))
def no_errors(node: int) -> typing.Optional[int]:
return any(no_errors(child) for child in children(node))this is the output of mypy:
$ mypy mypyany.py
mypyany.py:6: error: Incompatible return value type (got "bool", expected "Optional[str]")
Found 1 error in 1 file (checked 1 source file)
$I don't get the error in the orig() function (I can "suppose" is because any() is "too dynamic" returning the first element of a sequence of anything). But, I don't get why, just changing the return annotation type (as in no_errors) the error goes away!
1 reply
def my_decorator(f: Callable[..., T]) -> Callable[..., Container[T]]:...
i.e. can I properly replace the Callables such that the resulting decorated function keeps its typed arguments?