## Summary
I only noticed this in the ecosystem report of
https://github.com/astral-sh/ruff/pull/22213 after merging it. The
change to displaying `Top[]` wrapper around the entire signature instead
of just the parameters had the side effect of not showing it at all when
displaying a top ParamSpec specialization. This PR fixes that.
Marking internal since this is a fixup of a not-released PR.
## Test Plan
Added mdtest that fails without this PR.
## Summary
A couple things I noticed when taking another look at the callable type
materializations.
1) Previously we wrongly ignored the return type when
bottom-materializing a callable with gradual signature, and always
changed it to `Never`.
2) We weren't correctly handling overloads that included a gradual
signature. Rather than separately materializing each overload, we would
just mark the entire callable as "top" or replace the entire callable
with the bottom signature.
Really, "top parameters" is something that belongs on the `Parameters`,
not on the entire `CallableType`. Conveniently, we already have
`ParametersKind` where we can track this, right next to where we already
track `ParametersKind::Gradual`. This saves a bit of memory, fixes the
two bugs above, and simplifies the implementation considerably (net
removal of 100+ LOC, a bunch of places that shouldn't need to care about
topness of a callable no longer need to.)
One user-visible change from this is that I now display the "top
callable" as `(Top[...]) -> object` instead of `Top[(...) -> object]`. I
think this is a (minor) improvement, because it wraps exactly the part
in `Top` that needs to be, rather than misleadingly wrapping the entire
callable type, including the return type (which has already been
separately materialized). I think the prior display would be
particularly confusing if the return type also has its own `Top` in it:
previously we could have e.g. `Top[(...) -> Top[list[Unknown]]]`, which
I think is less clear than the new `(Top[...]) -> Top[list[Unknown]]`.
## Test Plan
Added mdtests that failed before this PR and pass after it.
### Ecosystem
The changed diagnostics are all either the change to `Top` display, or
else known non-deterministic output. The added diagnostics are all true
positives:
The added diagnostic at
aa35ca1965/torchvision/transforms/v2/_utils.py (L149)
is a true positive that wasn't caught by the previous version. `str` is
not assignable to `Callable[[Any], Any]` (strings are not callable), nor
is the top callable (top callable includes callables that do not take a
single required positional argument.)
The added diagnostic at
081535ad9b/starlette/routing.py (L67)
is also a (pedantic) true positive. It's the same case as #1567 -- the
code assumes that it is impossible for a subclass of `Response` to
implement `__await__` (yielding something other than a `Response`).
The pytest added diagnostics are also both similar true positives: they
make the assumption that an object cannot simultaneously be a `Sequence`
and callable, or an `Iterable` and callable.
## Summary
Resolve https://github.com/astral-sh/ty/issues/2226
We need to add a special case in `apply_type_mapping` instead of
directly in `promote_literals_impl` because we do not reach this with
non generic non tuple nominal instances. We still ensure we apply the
normal mapping if we do not see `float` or `complex` instances.
## Test Plan
Update existing mdtest and add a new case to `literal_promotion.md`
## Summary
Other type checkers allow you to access all `FunctionType` attributes on
any object with a `Callable` type. ty does not, because this is
demonstrably unsound, but this is often a source of confusion for users.
And there were lots of diagnostics in the ecosystem report for
https://github.com/astral-sh/ruff/pull/22145 that were complaining that
"Object of type `(...) -> Unknown` has no attribute `__name__`", for
example.
The discrepancy between what ty does here and what other type checkers
do is discussed a bit in https://github.com/astral-sh/ty/issues/1495.
You can see that there have been lots of issues closed as duplicates of
that issue; we should probably also add an FAQ entry for it.
Anyway, this PR adds a subdiagnostic to help users out when they hit
this diagnostic. Unfortunately something I did meant that rustfmt
increased the indentation of the whole of this huge closure, so this PR
is best reviewed with the "No whitespace" option selected for viewing
the diff.
## Test Plan
Snapshot added
## Summary
Provides a message like:
```
error[invalid-argument-type]: Cannot delete required key "name" from TypedDict `Movie`
--> test.py:15:7
|
15 | del m["name"]
| ^^^^^^
|
info: Field defined here
--> test.py:4:5
|
4 | name: str
| --------- `name` declared as required here; consider making it `NotRequired`
|
info: Only keys marked as `NotRequired` (or in a TypedDict with `total=False`) can be deleted
```
## Summary
TypedDict now synthesizes a proper `__delitem__` method that...
- ...allows deletion of `NotRequired` keys and keys in `total=False`
TypedDicts.
- ...rejects deletion of required keys (synthesizes `__delitem__(k:
Never)`).
## Summary
We already had `CallableTypeKind::ClassMethodLike` to track callables
that behave like `classmethods` (always bind the first argument). This
PR adds the symmetric `CallableTypeKind::StaticMethodLike` for callables
that behave like `staticmethods` (never bind `self`).
Closes https://github.com/astral-sh/ty/issues/2114.
## Summary
Previously, `del x[k]` incorrectly required the object to have a
`__getitem__` method. This was wrong because deletion only needs
`__delitem__`, which is independent of `__getitem__`.
Closes https://github.com/astral-sh/ty/issues/1799.
## Summary
This PR fixes https://github.com/astral-sh/ty/issues/1848.
```python
T = tuple[int, 'U']
class C(set['U']):
pass
type U = T | C
```
The reason why the fixed point iteration did not converge was because
the types stored in the implicit tuple type alias `Specialization`
changed each time.
```
1st: <class 'tuple[int, C]'>
2nd: <class 'tuple[int, tuple[int, C] | C]'>
3rd: <class 'tuple[int, tuple[int, tuple[int, C] | C] | C]'>
...
```
And this was because `UnionType::from_elements` was used when creating
union types for tuple operations, which causes type aliases inside to be
expanded.
This PR replaces these with `UnionType::from_elements_leave_aliases`.
## Test Plan
New corpus test
Identify and narrow cases like this:
```py
class Foo(TypedDict):
tag: Literal["foo"]
class Bar(TypedDict):
tag: Literal["bar"]
def _(union: Foo | Bar):
if union["tag"] == "foo":
reveal_type(union) # Foo
```
Fixes part of https://github.com/astral-sh/ty/issues/1479.
---------
Co-authored-by: Alex Waygood <Alex.Waygood@Gmail.com>
## Summary
This PR fixes https://github.com/astral-sh/ty/issues/2085.
Based on the reported code, the panicking MRE is:
```python
class Test:
def __init__(self, x: int):
self.left = x
self.right = x
def method(self):
self.left, self.right = self.right, self.left
if self.right:
self.right = self.right
```
The type inference (`implicit_attribute_inner`) for `self.right`
proceeds as follows:
```
0: Divergent(Id(6c07))
1: Unknown | int | (Divergent(Id(1c00)) & ~AlwaysFalsy)
2: Unknown | int | (Divergent(Id(6c07)) & ~AlwaysFalsy) | (Divergent(Id(1c00)) & ~AlwaysFalsy)
3: Unknown | int | (Divergent(Id(1c00)) & ~AlwaysFalsy) | (Divergent(Id(6c07)) & ~AlwaysFalsy)
4: Unknown | int | (Divergent(Id(6c07)) & ~AlwaysFalsy) | (Divergent(Id(1c00)) & ~AlwaysFalsy)
...
```
The problem is that the order of union types is not stable between
cycles. To solve this, when unioning the previous union type with the
current union type, we should use the previous type as the base and add
only the new elements in this cycle (In the current implementation, this
unioning order was reversed).
## Test Plan
New corpus test
## Summary
IIUC, tuples with a known structure (`tuple_spec`) use the standard
tuple `__eq__` which only returns `True` for other tuples, so they can
be safely excluded when disjoint from string literals or other non-tuple
types.
Closes https://github.com/astral-sh/ty/issues/2140.
## Summary
The following snippet currently errors because we widen the inferred
type, even though `X` is covariant over `T`. If `T` was contravariant or
invariant, this would be fine, as it would lead to an assignability
error anyways.
```python
class X[T]:
def __init__(self: X[None]): ...
def pop(self) -> T:
raise NotImplementedError
# error: Argument to bound method `__init__` is incorrect: Expected `X[None]`, found `X[int | None]`
x: X[int | None] = X()
```
There are some cases where it is still helpful to prefer covariant
declared types, but this error seems hard to fix otherwise, and makes
our heuristics more consistent overall.
## Summary
This PR implements the strategy described in
https://github.com/astral-sh/ty/issues/1871: we iterate over the
positive types, resolve them, then intersect the results.
## Summary
This should make revealed types a little nicer, as well as avoid
confusing the constraint solver in some cases (which were showing up in
https://github.com/astral-sh/ruff/pull/21930).
## Summary
Add lockfiles for all mdtests which make use of external dependencies.
When running tests normally, we use this lockfile when creating the
temporary venv using `uv sync --locked`. A new
`MDTEST_UPGRADE_LOCKFILES` environment variable is used to switch to a
mode in which those lockfiles can be updated or regenerated. When using
the Python mdtest runner, this environment variable is automatically set
(because we use this command while developing, not to simulate exactly
what happens in CI). A command-line flag is provided to opt out of this.
## Test Plan
### Using the mdtest runner
#### Adding a new test (no lockfile yet)
* Removed `attrs.lock` to simulate this
* Ran `uv run crates/ty_python_semantic/mdtest.py -e external/`. The
lockfile is generated and the test succeeds.
#### Upgrading/downgrading a dependency
* Changed pydantic requirement from `pydantic==2.12.2` to
`pydantic==2.12.5` (also tested with `2.12.0`)
* Ran `uv run crates/ty_python_semantic/mdtest.py -e external/`. The
lockfile is updated and the test succeeds.
### Using cargo
#### Adding a new test (no lockfile yet)
* Removed `attrs.lock` to simulate this
* Ran `MDTEST_EXTERNAL=1 cargo test -p ty_python_semantic --test mdtest
mdtest__external` "naively", which outputs:
> Failed to setup in-memory virtual environment with dependencies:
Lockfile not found at
'/home/shark/ruff/crates/ty_python_semantic/resources/mdtest/external/attrs.lock'.
Run with `MDTEST_UPGRADE_LOCKFILES=1` to generate it.
* Ran `MDTEST_UPGRADE_LOCKFILES=1 MDTEST_EXTERNAL=1 cargo test -p
ty_python_semantic --test mdtest mdtest__external`. The lockfile is
updated and the test succeeds.
#### Upgrading/downgrading a dependency
* Changed pydantic requirement from `pydantic==2.12.2` to
`pydantic==2.12.5` (also tested with `2.12.0`)
* Ran `MDTEST_EXTERNAL=1 cargo test -p ty_python_semantic --test mdtest
mdtest__external` "naively", which outputs a similar error message as
above.
* Ran the command suggested in the error message (`MDTEST_EXTERNAL=1
MDTEST_UPGRADE_LOCKFILES=1 cargo test -p ty_python_semantic --test
mdtest mdtest__external`). The lockfile is updated and the test
succeeds.
When inferring a specialization of a `Callable` type, we use the new
constraint set implementation. In the example in
https://github.com/astral-sh/ty/issues/1968, we end up with a constraint
set that includes all of the following clauses:
```
U_co ≤ M1 | M2 | M3 | M4 | M5 | M6 | M7
M1 ≤ U_co ≤ M1 | M2 | M3 | M4 | M5 | M6 | M7
M2 ≤ U_co ≤ M1 | M2 | M3 | M4 | M5 | M6 | M7
M3 ≤ U_co ≤ M1 | M2 | M3 | M4 | M5 | M6 | M7
M4 ≤ U_co ≤ M1 | M2 | M3 | M4 | M5 | M6 | M7
M5 ≤ U_co ≤ M1 | M2 | M3 | M4 | M5 | M6 | M7
M6 ≤ U_co ≤ M1 | M2 | M3 | M4 | M5 | M6 | M7
M7 ≤ U_co ≤ M1 | M2 | M3 | M4 | M5 | M6 | M7
```
In general, we take the upper bounds of those constraints to get the
specialization. However, the upper bounds of those constraints are not
all guaranteed to be the same, and so first we need to intersect them
all together. In this case, the upper bounds are all identical, so their
intersection is trivial:
```
U_co = M1 | M2 | M3 | M4 | M5 | M6 | M7
```
But we were still doing the work of calculating that trivial
intersection 7 times. And each time we have to do 7^2 comparisons of the
`M*` classes, ending up with O(n^3) overall work.
This pattern is common enough that we can put in a quick heuristic to
prune identical copies of the same type before performing the
intersection.
Fixes https://github.com/astral-sh/ty/issues/1968
## Summary
This contains two bug fixes:
- [Handle field specifier functions that accept
`**kwargs`](ad6918d505)
- [Recognize metaclass-based transformers as instances of
`DataclassInstance`](1a8e29b23c)
closes https://github.com/astral-sh/ty/issues/1987
## Test Plan
* New Markdown tests
* Made sure that the example in 1987 checks without errors
## Summary
We're actually quite good at computing this but the main issue is just
that we compute it at the type-level and so wrap it in `Literal[...]`.
So just special-case the rendering of these to omit `Literal[...]` and
fallback to `...` in cases where the thing we'll show is probably
useless (i.e. `x: str = str`).
Fixes https://github.com/astral-sh/ty/issues/1882
This fixes a bug @zsol found running ty against pyx. His original repro
is:
```py
class Base:
def __init__(self) -> None: pass
class A(Base):
pass
def foo[T](callable: Callable[..., T]) -> T:
return callable()
a: A = foo(A)
```
The call at the bottom would fail, since we would infer `() -> Base` as
the callable type of `A`, when it should be `() -> A`.
The issue was how we add implicit annotations to `self` parameters.
Typically, we turn it into `self: Self`. But in cases where we don't
need to introduce a full typevar, we turn it into `self: [the class
itself]` — in this case, `self: Base`. Then, when turning the class
constructor into a callable, we would see this non-`Self` annotation and
think that it was important and load-bearing.
The fix is that we skip all implicit annotations when determining
whether the `self` annotation should take precedence in the callable's
return type.
This is a first stab at solving
https://github.com/astral-sh/ty/issues/500, at least in part, with the
old solver. We add a new `TypeRelation` that lets us opt into using
constraint sets to describe when a typevar is assignability to some
type, and then use that to calculate a constraint set that describes
when two callable types are assignable. If the callable types contain
typevars, that constraint set will describe their valid specializations.
We can then walk through all of the ways the constraint set can be
satisfied, and record a type mapping in the old solver for each one.
---------
Co-authored-by: Carl Meyer <carl@astral.sh>
Co-authored-by: Alex Waygood <alex.waygood@gmail.com>
Fixes https://github.com/astral-sh/ty/issues/1787
## Summary
Allow method decorators returning Callables to presumptively propagate
"classmethod-ness" in the same way that they already presumptively
propagate "function-like-ness". We can't actually be sure that this is
the case, based on the decorator's annotations, but (along with other
type checkers) we heuristically assume it to be the case for decorators
applied via decorator syntax.
## Test Plan
Added mdtest.
## Summary
Infer `Literal[True]` for `isinstance(x, C)` calls when `x: T` and `T`
has a bound `B` that satisfies the `isinstance` check against `C`.
Similar for constrained typevars.
closes https://github.com/astral-sh/ty/issues/1895
## Test Plan
* New Markdown tests
* Verified the the example in the linked ticket checks without errors
